1. Cell Biology
  2. Chromosomes and Gene Expression
Download icon

HP1α is a chromatin crosslinker that controls nuclear and mitotic chromosome mechanics

  1. Amy R Strom
  2. Ronald J Biggs
  3. Edward J Banigan
  4. Xiaotao Wang
  5. Katherine Chiu
  6. Cameron Herman
  7. Jimena Collado
  8. Feng Yue
  9. Joan C Ritland Politz
  10. Leah J Tait
  11. David Scalzo
  12. Agnes Telling
  13. Mark Groudine
  14. Clifford P Brangwynne
  15. John F Marko
  16. Andrew D Stephens  Is a corresponding author
  1. Howard Hughes Medical Institute, Department of Chemical and Biological Engineering, Princeton University, United States
  2. Department of Molecular Biosciences, Northwestern University, United States
  3. Institute for Medical Engineering and Science and Department of Physics, Massachusetts Institute of Technology, United States
  4. Department of Biochemistry and Molecular Genetics, Feinberg School of Medicine, Northwestern University, United States
  5. Biology Department, University of Massachusetts Amherst, United States
  6. Robert H. Lurie Comprehensive Cancer Center, Feinberg School of Medicine, Northwestern University, United States
  7. The Fred Hutchinson Cancer Research Center, United States
  8. Department of Physics and Astronomy, Northwestern University, United States
Research Article
  • Cited 1
  • Views 2,227
  • Annotations
Cite this article as: eLife 2021;10:e63972 doi: 10.7554/eLife.63972

Abstract

Chromatin, which consists of DNA and associated proteins, contains genetic information and is a mechanical component of the nucleus. Heterochromatic histone methylation controls nucleus and chromosome stiffness, but the contribution of heterochromatin protein HP1α (CBX5) is unknown. We used a novel HP1α auxin-inducible degron human cell line to rapidly degrade HP1α. Degradation did not alter transcription, local chromatin compaction, or histone methylation, but did decrease chromatin stiffness. Single-nucleus micromanipulation reveals that HP1α is essential to chromatin-based mechanics and maintains nuclear morphology, separate from histone methylation. Further experiments with dimerization-deficient HP1αI165E indicate that chromatin crosslinking via HP1α dimerization is critical, while polymer simulations demonstrate the importance of chromatin-chromatin crosslinkers in mechanics. In mitotic chromosomes, HP1α similarly bolsters stiffness while aiding in mitotic alignment and faithful segregation. HP1α is therefore a critical chromatin-crosslinking protein that provides mechanical strength to chromosomes and the nucleus throughout the cell cycle and supports cellular functions.

Introduction

Chromatin, which fills the nucleus, is a repository of information, but it is also a physical element that provides structure, mechanical rigidity, shape, and function to the nucleus. Heterochromatin is the stiff, compact, and gene-poor form of chromatin. Heterochromatin loss results in abnormal nuclear morphology, which is a hallmark of human disease (Stephens et al., 2019a; Uhler and Shivashankar, 2018). Increasing the amount of heterochromatin by elevating histone methylation levels can increase nuclear stiffness and restore nuclear shape and function in perturbed model cell lines and patient cells of human diseases (Liu et al., 2018; Stephens et al., 2019b; Stephens et al., 2018; Stephens et al., 2017). Chromatin stiffness also plays a key role during cell division, as mitotic chromosome mechanics are key to the proper segregation of the genome during mitosis (Batty and Gerlich, 2019; Ribeiro et al., 2009; Stephens et al., 2011; Sun et al., 2018). Recently, it has been reported that methylated histones/heterochromatin are also a mechanical component of mitotic chromosomes (Biggs et al., 2019). However, in addition to methylated histones, protein ‘readers’ of epigenetic marks play a key role in defining heterochromatin (and euchromatin). A key histone methylation reader, Heterochromatin Protein 1α (HP1α), remains poorly characterized as to its role in controlling the mechanical properties of heterochromatin. To what degree HP1α contributes to the mechanical resistive capabilities of chromatin, how this contribution is intertwined with histone methylation, and how these result in proper nuclear and mitotic mechanics and function, are all open questions.

HP1α is a major component of constitutive heterochromatin (James and Elgin, 1986; Singh et al., 1991; Wreggett et al., 1994). Functionally, HP1α is a homodimer that binds to both DNA and to H3K9me2,3 constitutive heterochromatin marks. The direct association of HP1α with H3K9me2,3 heterochromatin and its direct binding to Suv39h1/2, the histone methyltransferase that deposits H3K9me2,3, has led to reports that HP1α is necessary for either maintenance or establishment of histone methylation (Bannister et al., 2001; Krouwels et al., 2005).

Loss of HP1α could therefore indirectly alter chromatin mechanics by modulating histone methylation levels. Alternatively, HP1α homodimerization and/or higher-order oligomerization could directly impact mechanics through physical bridging of two chromatin fibers, resulting in crosslinking of DNA or H3K9me2,3-marked nucleosomes (Canzio et al., 2011; Cheutin et al., 2003; Machida et al., 2018). Consistent with this possibility, chromatin crosslinks have been suggested to be a key element of chromatin organization and mechanics (Banigan et al., 2017; Belaghzal et al., 2021; Lionetti et al., 2020; Stephens et al., 2017). The capacity of HP1α to drive liquid-liquid phase separation (Larson et al., 2017; Strom et al., 2017) could also contribute to altered chromatin organization and mechanics, given the emerging evidence for links between phase separation and nuclear mechanics (Shin et al., 2018). These mechanisms could also affect mechanics in mitotic chromosomes, where HP1α is also present (Akram et al., 2018; Serrano et al., 2009). Therefore, it is now critical to determine the role of HP1α in controlling chromatin mechanics during both interphase and mitosis, as well as the functions of HP1α-mediated chromatin mechanics.

Nuclear and mitotic chromosome micromanipulation force measurements have been critical to understanding the mechanical properties of chromatin, making these techniques ideal for probing the relative roles of histone modifications and chromatin-binding proteins. Nucleus micromanipulation force measurements provide a novel capability, allowing the separation of chromatin, which dominates the initial force-response regime, from the other major mechanical component, lamin A, which dictates strain stiffening in the long-extension regime (Stephens et al., 2017). This two-regime force response was recently verified by the AFM-SPIM force measurement technique (Hobson et al., 2020a). Chromatin-based nuclear mechanics are dictated by euchromatin and heterochromatin levels, particularly through post-translational modifications of histones by acetylation or methylation, respectively (Heo et al., 2016; Hobson and Stephens, 2020b; Krause et al., 2019; Liu et al., 2018; Nava et al., 2020; Stephens et al., 2019b; Stephens et al., 2018; Stephens et al., 2017). These changes in chromatin-based nuclear mechanics can, independently of lamins, cause abnormal nuclear morphology, which is a hallmark of human disease (Stephens et al., 2019a). A recent high-throughput screen revealed that many key chromatin proteins also contribute to nuclear shape (Tamashunas et al., 2020), raising the question of the relative roles of histone modifications versus chromatin proteins such as HP1α.

Recent experimental and modeling studies suggest chromatin proteins, like HP1α, may contribute to mechanics by acting as physical linkers. Experimental data for nuclear mechanical response can only be reconciled with models which contain chromatin (an interior polymer), a lamina (a peripheral meshwork), and chromatin-chromatin and chromatin-lamina linkages (Banigan et al., 2017; Hobson and Stephens, 2020b; Stephens et al., 2017). Further studies have suggested that these linkages may govern nuclear shape stability (Lionetti et al., 2020; Liu et al., 2021; Schreiner et al., 2015). Experimental studies have shown chromatin linkages to the nuclear periphery aid shape stability and mechanics (Schreiner et al., 2015). Furthermore, recent chromosome conformation capture (Hi-C) and mechanics experiments suggest that chromatin is physically linked about once per 15 kb, since chromatin organization and mechanical response are perturbed only upon extreme chromatin fragmentation by restriction enzymes (Belaghzal et al., 2021). Whether chromatin-binding proteins like HP1α provide mechanical and morphological stability to the nucleus and whether their function is to maintain histone modifications or act as physical linkers remains an open question.

Most studies of epigenetic modification of chromatin and nuclear mechanics have focused on the interphase nucleus. However, it is conceivable that some of the epigenetic marks involved in heterochromatin formation during interphase might survive and have effects during cell division. Consistent with this, recent work indicates that hypermethylation of histones can persist into metaphase and is correlated with increased stiffness of mitotic chromosomes/metaphase chromatin (Biggs et al., 2019). However, it remains unknown whether the readers of those marks, such as HP1α, contribute significantly to metaphase chromatin structure and mechanics and how important they are to ensuring the success of mitosis.

Here, we determine the mechanical role of heterochromatin protein HP1α and its independence from histone methylation. We created and characterized an auxin-inducible-degradation (AID, [Nishimura et al., 2009]) system for rapid depletion of endogenous HP1α in human U2OS cells. Using these novel CRISPR-derived HP1α-AID-sfGFP cells, we find that the transcriptional profile and chromatin organization are largely unchanged by rapid degradation of HP1α. However, rapid HP1α degradation causes decreased chromatin-based rigidity in both nuclei and mitotic chromosomes. Concurrently, we observe increases in aberrant nuclear morphology and incidence of mitotic errors, both of which are associated with disease. Increasing histone methylation rescues nuclear and mitotic chromosome mechanics associated with HP1α depletion, indicating that these factors contribute independently. Rescue experiments with a HP1α mutant protein reveal that its dimerization is essential for the maintenance of nuclear structure. Computational modeling supports the conclusion that HP1α’s contribution to nuclear mechanics follows primarily from its function as a chromatin-chromatin crosslinker, suggesting that constitutive heterochromatin may be thought of as a polymer gel (Colby et al., 1993). These findings contribute to our understanding of the role of histone methylation and heterochromatin levels in controlling nuclear organization, mechanics, and morphology in healthy and diseased cell states.

Results

Rapid degradation of HP1α using an auxin-inducible degron

We generated a novel endogenous HP1α auxin-inducible degron for rapid and reversible depletion of HP1α protein in the cell. This was accomplished using CRISPR (Doudna and Charpentier, 2014) to tag both endogenous copies of the CBX5 gene in U2OS cells with an auxin-inducible degron (AID, [Nishimura et al., 2009]) and reporter Superfolder Green Fluorescent Protein (sfGFP) at the C terminus (HP1α-AID-sfGFP). Immunostaining demonstrated that modification of the endogenous loci did not alter the HP1α protein localization pattern (Figure 1A), while PCR, western blotting, and flow cytometry showed that all endogenous CBX5 alleles were tagged and only modified protein was expressed (Figure 1C and D, Materials and methods). Modification of the endogenous HP1α allele did not alter transcription or H3K9me2,3 levels (see Source data 3, 76 upregulated and 56 downregulated transcripts, which represents a change in 0.8% of genes across >16,600; compare methylation levels in parental and tagged cells see Figure 2—figure supplement 1D). HP1α degradation was observed by fluorescence microscopy or flow cytometry of HP1α-AID-sfGFP cells and by western blot after 4 hr of treatment with 1 mM auxin (Indole-3-acetic acid, Figure 1B-D). These conditions consistently resulted in >90% degradation of HP1α. The degradation was reversible as protein levels recovered over 2 days after removal of auxin (Figure 1B-D). Thus, we report the novel generation of an endogenously tagged HP1α cell line, which has a fluorescent reporter and is capable of rapid, reversible degradation in hours.

Figure 1 with 1 supplement see all
Generation of an CRISPR endogenous HP1α-auxin-inducible-degron-sfGFP cell line.

(A) Example images of HP1α-AID-sfGFP relative to wild-type cells stained for HP1α via immunofluorescence along with Hoechst DNA stain and phase contrast images. Scale bar = 20 μm. (B) Example images of HP1α-AID-sfGFP before, after 4 hr of auxin treatment, and 2 days post auxin removal. Hoechst DNA stain aids labeling of nuclei. (C) Western blot and (D) Flow cytometer graph of GFP intensity of control (ctrl/untreated), auxin-treated for 4 hr, 2 days after removal of auxin, and wild-type (WT) showing short-term loss and recovery of HP1α-AID-sfGFP. (E) Graph of RNA-seq data showing that few genes change transcript levels as determined by q-value <0.05 (calculated via -Log10P) and absolute change of Log2 fold >1 (marked in orange), with expression of only 40 out of of 16,663 genes changing significantly comparing control/untreated versus 4 hr auxin-treated.

Previous studies have shown that disruption of HP1α binding and localization through RNAi knockdown of its binding partners results in chromatin decompaction and loss of transcriptional silencing (Frescas et al., 2008; Hahn et al., 2013; Shumaker et al., 2006). Because tethering of HP1α to specific sites is sufficient to induce chromatin compaction and transcriptional silencing (Li et al., 2003; Verschure et al., 2005), we sought to determine whether rapid depletion of HP1α by auxin treatment would significantly alter global transcription or chromatin organization. RNA-Seq data was acquired, mapped (STAR), and quantified (RSEM), and the differential gene expression analysis was performed using DESeq2 for greater than 16,500 genes (see Materials and methods). Transcription analysis of HP1α-AID-sfGFP control and 4 hr auxin-treated cells revealed that only three genes were downregulated and only 37 genes were upregulated (q-value <0.05 and fold-change >2, Figure 1E, Source data 1). Lack of transcriptional changes upon rapid degradation of HP1α was further supported by comparing control and 16 hr of auxin treatment, which yielded 15 downregulated and four upregulated genes (see Source data 2). These data are similar to previous reports that in mammalian systems, HP1 proteins are not required for maintenance of silencing (Maksakova et al., 2011). Furthermore, satellite derepression and other transcriptional changes previously reported after redistribution of HP1α may be unique to each organism and be indirect or dependent on secondary chromatin rearrangements. In addition, DAPI and Hoechst staining patterns showed similar dense regions of nuclear stain typical of heterochromatin in both treated and untreated cells (Figure 1 and Figure 1—figure supplement 1). Furthermore, histone density and distribution do not significantly change on a single-cell basis (Figure 1—figure supplement 1). These results indicate that no global change in transcription or global and local changes in histone density occurred after rapid HP1α degradation.

HP1α is a major mechanical component of the interphase nucleus that contributes to nuclear shape maintenance

We hypothesized that HP1α could aid nuclear mechanics due to its association with heterochromatin. To test this hypothesis, we perform single-nucleus micromanipulation force measurements on untreated and auxin-induced HP1α-degraded nuclei. Micromanipulation is an extensional force measurement technique capable of separating chromatin- and lamin-based nuclear mechanics (Stephens et al., 2017). First, a single nucleus is isolated from a living cell following treatment with latrunculin A to depolymerize actin and local lysis applied via micropipette spray (Figure 2A). The isolated nucleus is then loaded between two micropipettes. One micropipette is moved to extend the nucleus, while the other micropipette’s deflection, multiplied by the premeasured bending constant, measures force (Figure 2A). The force-extension relation is nonlinear, but can be decomposed into two linear slopes, which provide nuclear spring constants (nN/μm) for the short-extension regime (<3 µm), quantifying chromatin-based stiffness, and the long-extension regime (>3 µm), quantifying lamin-based strain stiffening (Stephens et al., 2019b; Stephens et al., 2018; Stephens et al., 2017; example Figure 2B).

Figure 2 with 3 supplements see all
HP1α is a mechanical component of the nucleus controlling nuclear shape, separately from histone methylation.

(A) Example images of a single isolated nucleus via transmitted light and HP1α-AID-sfGFP fluorescence and single nucleus micromanipulation force-extension measurement experiment. The pull pipette extends the nucleus while the bending of a premeasured force pipette provides the force measurement. Scale bar = 10 μm. (B) Example traces of micromanipulation force-extension for control (black) and auxin-induced degradation of HP1α (orange) provide a measure of nuclear spring constant from the slope (dotted lines). Initial slope provides chromatin-based nuclear spring constant while the second slope provides the lamin-based strain stiffening nuclear spring constant. (C and D) Graphs of average and single chromatin-based nuclear spring constant for (C) parental cell line control and 4–6 hr auxin treated and (D) HP1α-AID-sfGFP with and without auxin and/or methylstat treatment. n = 11–18 nuclei each. (E) Example images of cells treated with and without auxin and/or methylstat. (F) Quantified relative fluorescence of HP1α-AID-sfGFP and heterochromatin marker H3K9me2,3. (G) Quantified abnormal nuclear morphology determined as solidity value less than 0.96, statistics via chi-squared analysis. Another way to quantify abnormal nuclear morphology is via average nuclear curvature reported in Figure 2—figure supplement 2. Three biological experiments (shown as black dots) each consisting of n = 109, 102, 105 control; n = 137, 115, 165 auxin, n = 31, 34, 32 methylstat, and n = 102, 92, 78 auxin methylstat. Average measurements were similar for control, methystat, and auxin with methylstat 0.971 ± 0.0001 but different for auxin 0.969 ± 0.0015, p=0.005. p values reported as *<0.05, **<0.01, ***<0.001, no asterisk denotes no significance, p>0.05. Error bars represent standard error.

Micromanipulation force measurements reveal that degradation of HP1α affects nuclear mechanics. Parental unmodified U2OS cells, control or auxin-treated, show no change in the chromatin-based nuclear spring constant (0.35 vs. 0.34 ± 0.06 nN/μm, p=0.73, Figure 2C), and nuclear mechanics of cells with tagged HP1α (vs. 0.40 ± 0.03 nN/μm, p=0.42, Figure 2D) are not significantly different from parental control. This suggests that auxin treatment alone does not alter nuclear mechanics, and the addition of the AID-sfGFP tag to HP1α does not alter normal nuclear mechanical response. HP1α-AID-sfGFP cells were imaged before nucleus isolation to verify presence or absence of HP1α via sfGFP reporter. Auxin-induced HP1α degradation resulted in a 45% decrease in short-extension chromatin-based nuclear stiffness (0.40 vs 0.22 ± 0.03 nN/μm, p<0.001, Figure 2D). However, long-extension strain stiffening remained relatively unchanged (example, Figure 2B; Figure 2—figure supplement 1A, p=0.99) in agreement with the observation of no decrease in lamin A/C or B1 levels (Figure 2—figure supplement 1, B C). This data indicates that HP1α contributes to chromatin mechanics of the cell nucleus.

Previous work has shown that nuclear softening due to perturbations of chromatin and its mechanics, particularly the loss of heterochromatin, can induce abnormal nuclear morphology (Stephens et al., 2019b; Stephens et al., 2018). Consistent with these prior findings, we quantified nuclear shape by shape solidity (ratio of area to convex area) and found that HP1α-degraded cells displayed a statistically significant decrease in average solidity (0.971 control vs. 0.969 auxin, p=0.005). Strikingly, we observe a large increase in the fraction of nuclei that have low levels of solidity, which we refer to as abnormal nuclei. Abnormal nuclei increase from 10 ± 1% in untreated cells to 22 ± 5% upon HP1α loss, as quantified by nuclei below a specified solidity threshold (solidity <0.96, Figure 2G). Another way to quantify shape is average nuclear curvature, which increases when the nucleus deviates from its normal elliptical shape (see Materials and methods). Tracking average nuclear curvature of single nuclei over time post auxin treatment reveals a significant increase in nuclear curvature during interphase (0% control vs 36% HP1α-degraded single nuclei persistently exhibit curvature increased by >0.05 μm−1, Figure 2—figure supplement 2), coincident with HP1α loss and decreased nuclear stiffness (4 hr auxin, Figure 2). Loss of nuclear mechanics and shape has been shown to cause dysfunction via nuclear ruptures and increased DNA damage (Stephens et al., 2019b; Stephens et al., 2018; Xia et al., 2018). We found similar results upon degradation of HP1α, as we observed dispersal of NLS-RFP into the cytoplasm during loss of nuclear compartmentalization by ruptures and a doubling of DNA damage as measured by ϒH2AX foci (Figure 2—figure supplement 3). These results agree with siRNA knockdown of HP1α in U2OS cells, which have demonstrated accumulation of DNA damage foci in a previous report (Lee et al., 2013). These data establish that HP1α degradation results in nuclear softening, abnormal nuclear morphology, and nuclear dysfunction.

HP1α and histone methylation contribute independently to nuclear mechanics and morphology

It is unclear exactly how the different components of heterochromatin work together to define its structure and function and how dependent they are on one another. For example, studies using genetic knockouts (Bosch-Presegué et al., 2017) and long-term depletion RNAi studies of heterochromatic components (Frescas et al., 2008; Hahn et al., 2013; Shumaker et al., 2006) have reported that HP1α not only binds to methylated histones, but also aids in histone methylation establishment and maintenance (Jacobs and Khorasanizadeh, 2002; Nielsen et al., 2002; Schotta et al., 2002). HP1α could simply alter levels of H3K9-methylated histones to affect nuclear mechanics and morphology. Levels of the constitutive heterochromatin mark H3K9me2,3 did not change significantly after 16 hr or 96 hr of HP1α depletion (Figure 2, E and F; Figure 2—figure supplement 2, D E with H3K9ac). Thus, while rapid reduction of HP1α levels affects nuclear mechanics and morphology, it does not cause significant changes in histone methylation.

Previous reports have shown that increased histone methylation stiffens the nucleus (Stephens et al., 2019b; Stephens et al., 2018; Stephens et al., 2017). Cells were treated with the broad histone demethylase inhibitor methylstat, which increases H3K9 methylation approximately three-fold over its normal levels in the HP1α-AID-sfGFP cell line (Figure 2, E and F). Micromanipulation force experiments with HP1α-AID-sfGFP cells treated with 1 μM methylstat for 48 hr measured a stiffer chromatin-based nuclear spring constant (control 0.40 vs. methylstat 0.56 ± 0.03 nN/μm, p=0.003, Figure 2D), similar to previously reported experiments on different cell lines (Stephens et al., 2018). Increased broad histone methylation via methylstat did not significantly increase HP1α-AID-sfGFP levels (Figure 2, E and F). Thus, chromatin-based nuclear mechanics can be modulated by changing either HP1α levels or methylated histone levels.

We reasoned that elevating levels of methylated histone in HP1α-degraded nuclei would reveal the relative contributions of histone methylation and HP1α to nuclear mechanics and shape. If chromatin mechanics is dictated entirely by HP1α, increasing histone methylation in auxin-treated cells should not change nuclear mechanics; in that case, the nuclear spring constant should match that of the HP1α-degraded cells. Alternatively, if the methylation state of histones contributes to chromatin stiffness independently of HP1α, methylstat-treated HP1α-degraded nuclei will have a larger spring constant than HP1α-degraded nuclei and may display rescued nuclear shape.

Experiments are consistent with the second scenario, where increasing histone methylation levels in HP1α-degraded cells resulted in rescued nuclear mechanics and shape. Micromanipulation force measurements reveal a larger nuclear spring constant for HP1α-degraded nuclei with increased histone methylation as compared to HP1α-degraded with normal levels of methylation, returning to a spring constant similar to wild-type levels (auxin 0.22 vs. auxin+methylstat 0.33 ± 0.03 nN/μm, p<0.001, Figure 2D). Alternatively, compared to normal levels of HP1α with increased histone methylation, loss of HP1α and increased histone methylation resulted in a decreased nuclear spring constant (auxin+methylstat 0.33 vs. methylstat 0.56 ± 0.03 nN/μm, p<0.001, Figure 2D). Strain stiffening in the lamin-dependent regime remained similar across all treatments (Figure 2—figure supplement 1). Consistent with the mechanical measurements, methylstat treatment rescues abnormal morphology associated with HP1α degradation from 22% abnormal to 13% (Figure 2G). Altogether, these results suggest that HP1α and methylated histone levels both contribute to chromatin-based nuclear mechanics and morphology. Moreover, the approximately additive nature of the changes in nuclear stiffness, along with the lack of interdependence between levels HP1α and histone methylation, suggest that these mechanisms contribute to mechanics independently.

Maintenance of nuclear morphology depends on HP1α dimerization

HP1α forms a homodimer that can bridge strands of chromatin by binding two H3K9me2,3 marks on different nucleosomes through its chromodomain (Jacobs and Khorasanizadeh, 2002; Machida et al., 2018; Nielsen et al., 2002) or two strands of DNA through a positively charged KRK patch in the hinge (Larson et al., 2017). We reasoned that the role of HP1α in determining nuclear shape and mechanics, independent of histone methylation levels, might be due to its ability to physically crosslink chromatin strands. This linking ability would be dependent on HP1α dimerization, which can be disrupted with a point mutant, HP1αI165E (Brasher et al., 2000; Lechner et al., 2005; Lechner et al., 2000; Thiru et al., 2004). To determine if dimerization is key to its mechanical and morphological contributions in vivo, we asked whether a non-dimerizing mutant (HP1αI165E) could rescue nuclear morphology when the endogenous protein was degraded.

HP1α-AID-sfGFP cells were infected with lentivirus to stably express either exogenous HP1αWT-mCherry (positive control) or HP1αI165E-mCherry (dimer mutant, [Brasher et al., 2000]) under an SFFV promoter. Two days post-infection, these cells stably expressing HP1αWT-mCherry or HP1αI165E-mCherry were treated with 1 mM auxin for 16 hr to degrade the endogenous HP1α-AID-sfGFP and assess the ability of the rescue construct to maintain and recover normal nuclear shape. These cells were fixed and immunostained for lamin A/C, and then the shape of the nucleus was quantified using average curvature of each nucleus (see Materials and methods).

We first measured nuclear curvature in parental U2OS, and control and auxin-treated HP1α-AID-sfGFP cells to determine normal and abnormal nuclear curvature, respectively. HP1α-AID-sfGFP cells have no difference in nuclear shape compared to parental U2OS cells (parental 0.114 ± 0.002 μm−1; HP1α-AID-sfGFP 0.118 ± 0.001 μm−1, p>0.05), and auxin-treatment of parental cells does not alter their nuclear morphology (parental - auxin 0.114 ± 0.002 μm−1; parental + auxin 0.117 ± 0.001 μm−1, p>0.05). Similar to our previous measurements above (Figure 2 and Figure 2—figure supplement 2), auxin-treated HP1α-degraded nuclei exhibited higher average nuclear curvature compared to control HP1α-AID-sfGFP nuclei (Figure 3, control 0.116 ± 0.003 μm−1 vs. HP1α-degraded with no rescue, 0.142 ± 0.001 μm−1, p=0.0002). Expression of exogenous HP1αWT-mCherry in auxin-treated HP1α-AID-sfGFP-degraded cells rescued nuclear morphology to near wild-type levels (p=0.99, control vs. HP1α-degraded with HP1αWT rescue, 0.123 ± 0.002 μm−1). However, the dimer mutant HP1αI165E-mCherry did not rescue nuclear morphology in auxin-treated cells (Figure 3, control vs. HP1α-degraded with HP1αI165E rescue, 0.151 ± 0.003 μm−1, p<0.0001). We observed that a subset of nuclei had abnormal nuclear shape (curvature greater than one standard deviation above control). Using this metric, cells in which HP1α-AID-sfGFP was degraded displayed a higher level of abnormally shaped nuclei compared to control (30% vs. 11%, Figure 3C). Expression of HP1αWT-mCherry recovered WT levels (13% abnormal), while expression of HP1αI165E-mCherry did not, leaving many abnormally shaped nuclei (31%). Together, these results indicate that HP1α dimerization is essential to its function in nuclear morphology and indicates that bridging or crosslinking of chromatin fibers is important in determining nuclear shape.

HP1α dimerization is essential for maintenance of nuclear shape.

(A) Example images of HP1α-AID-sfGFP cells control (-Aux) and auxin treated (+Aux) with and without exogenous HP1α wild-type (WT) or dimer mutant (I165E) rescue constructs tagged with mCherry. Scale bar = 10 μm. (B) Graph of average nuclear curvature measurements with individual trials as black dots. (C) Graph of percentage of abnormally shaped nuclei, determined as greater than 0.15 μm−1 curvature, which is the average untreated nucleus plus the standard deviation. Eight experimental biological replicates were measured for each condition (denoted as black dots) consisting of -auxin, n = 37, 45, 45, 57, 58, 57, 55, 54; +auxin, n = 60, 44, 41, 60, 60, 60, 60, 60; +auxin and WT exogenous rescue, n = 27, 30, 36, 60, 60, 60, 60, 19; +auxin and I165E exogenous rescue, n = 38, 40, 50, 59, 56, 58, 58, 51. p values reported as no asterisk >0.05, *<0.05, **<0.01, ***<0.001, calculated by one-way ANOVA. Error bars represent standard error.

Simulations of nuclear mechanics modulating chromatin crosslinking recapitulate experimental degradation of HP1α

To assess the role of HP1α in chromatin-based nuclear mechanical response, we performed Brownian dynamics simulations using a previously developed shell-polymer model (Banigan, 2021; Banigan et al., 2017; Stephens et al., 2017). In these simulations, chromatin is modeled as a crosslinked polymer that is physically linked to a peripheral polymeric lamin shell that encapsulates the polymer chromatin (see Materials and methods). In this model, each chromatin bead is 0.57 µm in diameter and represents a few Mbp of the genome. This coarse-grained model can capture the effects of alterations to histone modifications through the polymer spring constant and perturbations to lamin A/C through the lamin spring constant (light red data points in Figure 4, A and C; Stephens et al., 2017). In particular, varying the polymer spring constant models alterations to chromatin compaction via histone modifications; the short-extension nuclear force response is suppressed as the polymer spring constant is decreased (Stephens et al., 2017).

Simulations of nuclear mechanical response and experimental measurements of peripheral heterochromatin support a model with HP1α as a chromatin-chromatin crosslinker.

(A) Force-strain relationship for simulated nuclei with various levels of chromatin-lamina (shell) linkages. Colors indicate different percentages of chromatin segments linked to the lamina. Insets: Snapshots of simulations with a portion of the lamina (green) removed to reveal the interior chromatin (blue) for two different applied stretching forces, F. (B) Spring constants for short- and long-extension regimes for simulations with various levels of chromatin-lamina crosslinks, quantified by percentage of peripheral chromatin subunits linked to the shell (blue and red, respectively). (C) Force-strain relation for simulated nuclei with various levels of chromatin-chromatin crosslinks. (D) Spring constants for short- and long-extension with varying levels of chromatin-chromatin crosslinks (blue and red). Vertical dashed lines in (A) and (C) separate the short-extension and long-extension regimes. Each force-strain data point in (A) and (C) is an average that is computed from n ≥ 11 simulations. Short-extension spring constants in (B) and (D) are each computed from nshort ≥13 and 10 force-extension data points, respectively. Long-extension spring constants in (B) and (D) are each computed from nlong ≥19 and 15 force-extension data points, respectively. (E) Example images of HP1α-AID-sfGFP nuclei untreated or auxin-treated, analyzed for (F) enrichment measurements (peripheral/internal average signal) to determine peripheral enrichment of DNA (Hoechst), HP1α, and H3K9me2,3. p values denoted as n.s. >0.05 and ****<0.0001, calculated by one-way ANOVA. Error bars in (A)-(D) show standard error of the mean.

However, rapid depletion of HP1α does not alter histone methylation state or lamin expression levels (Figure 2, E and F; Figure 2—figure supplement 2), so we sought to identify a distinct physical role for HP1α within this framework. We hypothesized that HP1α might instead govern mechanics either by linking heterochromatin to the lamina via proteins such as PRR14 (Poleshko et al., 2013) and LBR (Polioudaki et al., 2001; Ye et al., 1997), or by binding and bridging nucleosomes (Azzaz et al., 2014; Canzio et al., 2011; Erdel et al., 2020; Machida et al., 2018). Thus, we explored whether HP1α might impact nuclear mechanical response by forming chromatin-chromatin crosslinks or by forming chromatin-lamina linkages.

We first investigated whether depletion of chromatin-lamina linkages in the simulations could generate the same mechanical effects as HP1α degradation in the experiments. In simulations, we varied the frequency of linkages between the chromatin and the lamina from zero up to ~50% of the chromatin subunits that reside near the shell. We found that the frequency of chromatin-lamina linkages affects the two-regime force response of the model nucleus (Figure 4A). The spring constants quantifying both the short- and long-extension force responses decrease as the number of chromatin-lamina linkages is decreased (Figure 4B). With fewer chromatin-lamina linkages, the mechanical coupling between the nuclear periphery and the interior is lost, which suppresses short-extension rigidity; simultaneously, the loss of these linkages uncouples the lamina from the stiff chromatin interior, which also decreases the long-extension stiffness. This result contrasts with measurements from the micromanipulation experiments (Figure 2D; Figure 2—figure supplement 1), which show that the short-extension spring constant, but not the long-extension spring constant, decreases after HP1α degradation. Thus, we conclude that the mechanical contributions of HP1α more likely arise from an alternative structural function.

We therefore investigated the effects of varying the levels of chromatin-chromatin crosslinkers in the simulation model. We varied crosslinking frequency from zero up to about one in three subunits crosslinked, above which the chromatin polymer is a percolated network and therefore solid-like. We found that the level of crosslinking markedly alters the force-strain relation (Figure 4C); increasing crosslinking stiffens the nucleus. However, in contrast to chromatin-lamina linkages, crosslinks govern stiffness of only the short-extension force response (Figure 4D). This is a signature of their specific effect in resisting deformations of the chromatin interior. These qualitative trends agree with the measurements from micromanipulation experiments (Figure 2, B and D). The simulation data also includes points that are in reasonable quantitative agreement with the experiments. These results are consistent with a model in which the HP1αI165E mutant abolishes crosslinking and thus decreases the short-extension nuclear spring constant (Figure 2, B and D), which may generate abnormal nuclear morphology. Altogether, the simulations support the conclusion that HP1α contributes to nuclear mechanical response by acting as a chromatin-chromatin crosslinking element.

HP1α degradation does not release heterochromatin from the nuclear periphery

To test our prediction from simulations that HP1α does not act mechanically as a chromatin-lamina linker, we experimentally assayed its location and peripheral heterochromatin tethering capabilities. Specifically, we investigated whether HP1α acts similarly to two known chromatin-lamina tethers, LBR and PRR14, which show enrichment at the periphery and maintain localization of peripheral H3K9-marked heterochromatin (Dunlevy et al., 2020; Giannios et al., 2017; Nikolakaki et al., 2017; Poleshko et al., 2013; Solovei et al., 2013). We measured peripheral enrichment ratios (average intensity at the periphery over the interior) of DNA (Hoechst), HP1α, and H3K9me2,3, using lamin B1 as a marker for the periphery. In untreated cells, both DNA and HP1α have enrichment ratios of about 1 (0.99 ± 0.01 vs. HP1α 0.86 ± 0.01, p=0.78, Figure 4, E and F), demonstrating a lack of peripheral enrichment, while H3K9me2,3 was somewhat enriched (1.48 ± 0.04, p<0.0001 vs. 0.99 DNA). Upon degradation of HP1α, peripheral enrichment of DNA and H3K9me2,3 do not change (0.99 vs 0.94, p=0.99 and 1.48 vs 1.34, respectively, p=0.49), whereas depletion of previously reported chromatin-lamina tethers result in a 50% or greater decrease in peripheral localization of H3K9me2,3 (Poleshko et al., 2013), which supports our conclusion that HP1α does not mechanically function as a chromatin-lamina linkage in this cell type.

HP1α provides mechanical strength to mitotic chromosomes and enhances mitotic fidelity

Given HP1α’s mechanical role in chromatin-based nuclear mechanics, we hypothesized that HP1α could also contribute to mitotic chromosome mechanics. As in interphase nuclear mechanical response, heterochromatin has recently been shown to govern mitotic chromosome mechanics (Biggs et al., 2019). It has previously been reported that most HP1α is removed from chromosomes during prophase by phosphorylation of H3S10, which is known to disrupt HP1α-H3K9me2,3 binding (Fischle et al., 2005; Hirota et al., 2005). However, some HP1α binding is maintained throughout mitosis (Serrano et al., 2009), suggesting a possible role for HP1α in mitotic chromosome mechanics.

We used fluorescence imaging and micropipette micromanipulation methods (Biggs et al., 2019; Sun et al., 2018) to assay the presence of HP1α-AID-sfGFP in prometaphase cells (identified by their round shape) and mitotic chromosomes without or with auxin treatment for 4 hr to degrade HP1α (Figure 5B). Prometaphase cells show both chromosome-bound and diffuse, cytoplasmic HP1α-AID-sfGFP signals. Both cytoplasmic and chromosomal HP1α-AID-sfGFP signals nearly completely disappear upon auxin-induced degradation (Figure 5B). To further verify the presence of HP1α on mitotic chromosomes, we isolated mitotic chromosome bundles from cells via gentle lysis and capture. Fluorescence imaging of these isolated bundles without the high background fluorescence of the cytoplasm allowed us to observe that HP1α is clearly present on mitotic chromosomes (Figure 5A). In addition to concentrated foci, HP1α-AID-sfGFP is also present on chromosome arms (Figure 5C and Figure 5—figure supplement 1, A B). Confocal imaging of live cells revealed that concentrated foci are located at the pericentromeric region (Figure 5—figure supplement 1C), in agreement with previously published reports (Akram et al., 2018; Fischle et al., 2005; Hirota et al., 2005; Serrano et al., 2009). By additional fluorescence imaging, we observed that HP1α-AID-sfGFP is lost upon auxin-induced degradation (Figure 5B C). Thus, we confirmed that endogenous HP1α-AID-sfGFP is associated with mitotic chromosome arms and pericentromeres, and it is degraded after 4 hr of auxin treatment.

Figure 5 with 1 supplement see all
HP1α is a mechanical component of the mitotic chromosome aiding proper segregation in mitosis.

(A) Example image of the steps to isolating a mitotic chromosome from a live cell using micropipettes. (B) Representative live mitotic cells and isolated mitotic chromosome bundles imaged via phase contrast and HP1α-AID-sfGFP fluorescence intensity across treatments. Values calculated by measuring the cell’s or chromosome bundle’s fluorescence minus the background fluorescence, normalized to the average intensity of the untreated cellular HP1α fluorescent intensity. p Values reported as ***<0.001, calculated by student’s t-test. (C) Example images of the endogenous HP1α-AID-sfGFP fluorescence of an isolated mitotic bundle outside of the lysed cell. Yellow box denotes the area where the graphed line scan was drawn. The line scan reveals HP1α on chromosome arms. (D) Example images of a force-extension experiment. The right pipette pulls away from the left pipette, which stretches the chromosome and causes the left pipette to deflect. The left ‘force’ pipette has a premeasured bending constant (in pN/um) to calculate force. Left graph, example traces of force-extension experiments for the different conditions. (E) Graph of average doubling force (100% strain) in picoNewtons for each condition, which is determined by slope of the force extension traces and the initial chromosomes length. For B-E, n = 20 for control and auxin treated, n = 16 for methylstat, and n = 14 auxin methylstat treated, p values calculated by student’s t-test. (F) Example images of abnormal mitotic segregation via anaphase bridge or nondisjunction. Graphs of percentage of mitotic cells displaying abnormal metaphase misalignment (black bars) and anaphase/telophase missegregation (white bars) via presence of anaphase bridges or nondisjunction/aneuploidy in control untreated cells (-) or auxin-treated (+) cells for 4 or 16 hr. Metaphase misalignment three to four biological replicate experiments (black dots) consisting of n = 16, 15, 20, 37 -aux, n = 33, 33, 24 +aux 4 hr, n = 22, 48, 58, 54 +aux 16 hr. Anaphase and telophase missegregation 3–4 experiments (black dots) consisting of n = 29, 23, 30, 30 -aux, n = 32, 29, 18 +aux 4 hr, n = 20, 35, 36, 45 +aux 16 hr. p values reported as *<0.05, **<0.01, ***<0.001, ****<0.0001, calculated by Student’s t-test. (G) HP1α-AID-sfGFP cells - auxin or +auxin for 24 hr were tracked through mitosis to determine if abnormal mitosis results in abnormally shaped daughter nuclei measured via nuclear curvature (parental 34 nuclei from 17 mitoses; -auxin 46 nuclei from 23 mitoses; +auxin 51 nuclei from 26 mitoses, p value from one-way ANOVA). Percentage of abnormal mitosis presented in Figure 5—figure supplement 1D. Error bars represent standard error. Scale bar in A-C = 10 μm and F = 20 μm.

The mechanical role of HP1α in mitotic chromosomes was investigated by micromanipulation force measurements. The isolated bundle of chromosomes was held by one micropipette while two additional micropipettes were used to capture and isolate a single chromosome (Figure 5D). The single mitotic chromosome is then extended with the stiff pull pipette, while deflection of the other, much less stiff force pipette provides a force measurement, in the same manner as our experiments on interphase nuclei (Figure 5D). For each isolated chromosome, we calculated a force versus extension plot (Figure 5D). Because each of the 23 human chromosomes is a unique length, we calculate a length-independent measurement by extrapolating the force-extension slope to determine the ‘doubling force’—the force at which the chromosome length would be doubled (i.e. force at 100% strain, Figure 5E). Since the pipettes hold opposite ends of the chromosome, tension is distributed across the whole chromosome (Figure 5D example images). Therefore, the resistive force measured includes contributions from chromatin, and thus HP1α, in both the chromosome arms and the pericentromeric region. We find that depletion of HP1α reduced mitotic chromosome doubling force by approximately 40%, from 262 ± 50 pN in control cells (spring constant 27 pN/μm) to 148 ± 12 pN in auxin-treated cells (16 pN/μm) (p=0.03, Figure 5E), indicating that HP1α significantly contributes to mitotic chromosome mechanics.

We next investigated whether histone methylation and the HP1α protein separately govern chromosome mechanics during mitosis, as they do during interphase. Increasing histone methylation via methylstat treatment has previously been shown to play a critical role in mechanical stiffness of mitotic chromosomes (Biggs et al., 2019). Furthermore, evidence exists for direct biochemical interactions between epigenetic marks on nucleosomes, independent of mark-reading proteins such as HP1α (Bilokapic et al., 2018; Zhiteneva et al., 2017). Thus, we aimed to determine whether histone methylation and HP1α contribute independently to mitotic chromosome stiffness.

We treated cells with the histone demethylase inhibitor methylstat to increase levels of methylated histones in cells with or without HP1α, controlled by the addition of auxin. Mitotic chromosomes isolated from cells treated with methylstat to increase methylated histone levels indeed show a significant, greater than 100% increase in doubling force from 262 ± 50 pN to 745 ± 164 pN (p=0.005, Figure 5, D and E), recapitulating previous results for HeLa cells (Biggs et al., 2019). Mitotic chromosomes isolated from cells treated with both methylstat to increase methylation and auxin to degrade HP1α have a doubling force comparable to those treated with methylstat alone, 452 ± 116 pN (p=0.18, Figure 5, D and E). Oppositely, mitotic chromosomes from cells that were treated only with auxin compared to both auxin and methylstat had significantly different doubling forces (148 vs. 452 pN, p=0.005). The data suggest that histone methylation stiffens mitotic chromosomes independently of HP1α and thus has a dominant role in determining mitotic chromosome mechanics. At the same time, we emphasize that HP1α clearly plays a major role in mitotic chromosome mechanics in wild-type cells.

HP1α depletion is known to lead to chromosomal instability, aberrant recombination, anaphase bridges, and lagging chromosomes (Chu et al., 2014). Therefore, HP1α’s role in metaphase chromosome mechanics may have functional importance during mitosis. To test this, we measured the percentage of mitotic cells with chromosome misalignment in metaphase or anaphase bridges during chromosome segregation in control, 4 hr, and 16 hr auxin-treated HP1α-degraded populations. HP1α depletion resulted in significant increases in both metaphase misalignment, from 8% in control to 28% in auxin 4 hr and 41% in auxin 16 hr treatments, and mis-segregation as measured by ana/telophase bridges, from 15% to 28% in auxin 4 hr and 50% in auxin 16 hr treatments (all p<0.05, Figure 5F). Thus, loss of HP1α disrupts chromosome mechanics and causes dysfunction in mitosis via chromosome misalignment and mis-segregation.

Abnormal mitosis has also been reported to disrupt nuclear morphology in the daughter cells (Gisselsson et al., 2001). Thus, we tracked cells treated without or with auxin for 24 hr through mitosis to determine if abnormal mitosis resulted in abnormal nuclear morphology after mitosis. Abnormal mitosis in parental or untreated HP1α-AID-sfGFP cells is rare, but it results in daughter cells with high nuclear curvatures (red dots, Figure 5G). Cells with HP1α degraded more frequently undergo abnormal mitosis (Figure 5, F and G). Interestingly, following both normal and abnormal mitosis, HP1α-degraded daughter cells exhibit increased average nuclear curvature in G1, 4 hr after mitosis (Figure 5G). This data suggests abnormal mitosis upon HP1α degradation may not be the primary cause of abnormal nuclear shape since normal mitosis under these conditions results in equally high curvature for daughter nuclei after mitosis (Figure 5G). Taken together, HP1α is necessary for proper mitotic chromosome mechanics and function, and its depletion results in abnormal mitosis and, independently, abnormally shaped daughter interphase nuclei.

Discussion

Constitutive heterochromatin comprises an essential nuclear compartment known to perform genome-stabilizing functions through its biochemical and mechanical properties. HP1α is an essential protein component of heterochromatin that orchestrates its structural and functional roles (Kumar and Kono, 2020). To directly characterize these roles, we developed a new tool for rapid and reversible depletion of endogenous HP1α protein through auxin-inducible degradation (Nishimura et al., 2009). Interestingly, rapid degradation of HP1α over 4 hr does not significantly alter large-scale transcriptional profile or chromatin organization (Figure 1). Nonetheless, rapid degradation of HP1α has significant effects on interphase and mitotic chromosome mechanics and morphology (Figures 24). Furthermore, HP1α’s role is dependent on its ability to dimerize (Figure 3). Together with polymer simulations of interphase nuclear mechanics (Figure 4), these results indicate that HP1α acts as a dynamic chromatin-chromatin crosslinker to provide mechanical strength to the nucleus, and that this function may persist through mitosis.

HP1α is not essential for transcription repression or heterochromatin maintenance on short time scales

Our data are the first to separate the direct and indirect roles of HP1α in heterochromatin and its major functions in maintaining heterochromatin and regulating transcription. Early studies of HP1α established its association with compacted regions (beta chromatin) (Bannister et al., 2001), transcriptional silencing in yeast (Fischer et al., 2009; Sadaie et al., 2008), and silencing in Drosophila and mammalian cells at specific sites (Li et al., 2003; Verschure et al., 2005). Recent studies have shown a capacity for HP1α to suppress transcription in HEK293 cells when overexpressed (Lee et al., 2019) and in MEF cells while recruited to a specific array (Erdel et al., 2020). In contrast, our studies assay global transcription after rapid loss of endogenous HP1α in human cells. We find that rapid HP1α degradation does not result in significant changes in gene transcription and local compaction (Figure 1 and Figure 1—figure supplement 1), suggesting that its presence is dispensable for maintenance of these heterochromatic features over timescales < 24 hr.

Chromatin compaction and transcriptional repression also depend on methylation, which promotes HP1α binding, which in turn may recruit the methyltransferases for further propagation of methylation (Bannister et al., 2001). However, we found that HP1α is not necessary for short-term (4 hr) or longer-term (16 hr) maintenance of histone methylation. In particular, after rapid degradation of HP1α by our endogenous auxin-induced degradation construct, there was no significant change in constitutive heterochromatin marker H3K9me3, commonly associated with transcriptional repression. The lack of widespread changes in transcription agrees with the lack of change in H3K9me3 levels (Figure 2). This is consistent with a previous report that genetic deletion of HP1α does not alter global H3K9me3, but rather, alters specific satellite H3K9me3 in parts of the genome with repetitive DNA sequences (Bosch-Presegué et al., 2017). Furthermore, our results are supported by the recent finding that heterochromatin foci sizes, compaction, and accessibilities are independent of HP1α binding (Erdel et al., 2020). Together, these results indicate an inability for the transcription machinery to function at heterochromatic loci regardless of whether or not HP1α is present. Our data also showed that increased histone methylation via methylstat did not result in a global increase in HP1α levels.

Altogether, these results are consistent with the existence of a heterochromatin compaction state that is insensitive to the presence or absence of HP1α (Erdel et al., 2020). Instead, the functional impact of HP1α may appear in other processes, such as DNA replication (Schwaiger et al., 2010), chromosome segregation (Abe et al., 2016), epigenetic imprinting and inheritance (Hathaway et al., 2012; Holla et al., 2020; Nakayama et al., 2000), or post-mitotic reformation of the nucleus (Liu and Pellman, 2020). Nonetheless, as we discuss below, despite its limited impact on global transcription and chromatin organization, HP1α serves an important function as a mechanical stabilizer of the genome and nucleus.

HP1α governs nuclear stiffness with a distinct and separate mechanical contribution from histone methylation

While rapid depletion of HP1α did not alter heterochromatin-specific properties and functions such as histone methylation levels or transcriptional repression over short time scales, it did significantly contribute to nuclear mechanics. Degradation of HP1α resulted in a drastic decrease in the short-extension rigidity of the nucleus, reducing the spring constant by 45% (Figure 2B,D). Lamin A levels and large-deformation nuclear stiffness, however, were unaffected by HP1α degradation (Figure 2B and Figure 2—figure supplement 1A). These results are consistent with prior experiments showing that chromatin dominates the mechanical response to small deformations, while lamins underlie strain stiffening to large deformations (Stephens et al., 2017). Similarly, HP1α has been shown to provide mechanical resistance for a single DNA fiber (Keenen et al., 2021). Furthermore, consistent with HP1α’s newfound role in chromatin-based mechanics, we find that HP1α degradation results in the loss of nuclear shape stability (Figure 2E,G), similar to the effects of other chromatin perturbations that soften the cell nucleus (Furusawa et al., 2015; Stephens et al., 2019a; Stephens et al., 2019b; Stephens et al., 2018; Wang et al., 2018). Thus, while rapid depletion of HP1α has little apparent effect on genome organization (Figure 1), HP1α is critical to maintaining the mechanical integrity of chromatin.

It is known that the mechanical contribution of chromatin to the short-extension force response of the nucleus depends on histone modification state (Heo et al., 2016; Hobson and Stephens, 2020b; Krause et al., 2019; Liu et al., 2018; Nava et al., 2020; Stephens et al., 2019b; Stephens et al., 2018; Stephens et al., 2017). We considered the possibility that histone methylation contributes to mechanics through its impact on HP1α binding to chromatin (Bannister et al., 2001; Erdel et al., 2020; Lachner et al., 2001; Nakayama et al., 2001). However, our experiments show that histone methylation has a distinct contribution to chromatin-based nuclear mechanical response that is largely separate from HP1α (Figure 2D,F,G). In particular, nuclear rigidity (and corresponding shape stability) lost by HP1α degradation can be recovered by hypermethylation of histones via methylstat treatment. Furthermore, HP1α has an additive effect with methylation on nuclear mechanical response: chromatin-based nuclear stiffness decreases after HP1α degradation with or without treatment with methylstat. Together, these results suggest that HP1α and histone methylation modulate separable mechanical responses within the cell nucleus. The methylation-based mechanical response may be due to direct interactions between histone marks (Bilokapic et al., 2018; Zhiteneva et al., 2017) or effects of other histone mark readers.

HP1α contributes to nuclear mechanical response by acting as a chromatin crosslinker

What is the separate mechanical role of HP1α in heterochromatin? HP1α is a homodimer capable of physically bridging chromatin fibers by binding methylated histones or DNA (Canzio et al., 2011; Cheutin et al., 2003; Machida et al., 2018). We found evidence that this capability supports a distinct mechanical function. HP1α’s dimerization is essential to its role in maintaining nuclear shape stability (Figure 3), which has been shown here (Figure 2) and previously (Stephens et al., 2019a; Stephens et al., 2018) to depend on chromatin-based nuclear stiffness. Thus, we conclude that HP1α’s ability to dimerize and crosslink chromatin is essential to HP1α’s contributions to chromatin-based nuclear stiffness (Figure 6).

HP1α is a mechanical element of interphase nuclei and mitotic chromosomes.

In wild-type (WT) nuclei, HP1α acts as a chromatin-chromatin crosslinker, resulting in stiffer nuclear mechanics. Other components that contribute to nuclear mechanics include the chromatin polymer (whose mechanical contribution is dictated by histone methylation), the lamina, and chromatin-lamina linkages. Nuclei with HP1α degraded have abnormal shape and softer chromatin-based short-extension mechanical response. Degradation of HP1α also leads to softer mitotic chromosomes and mitotic defects, including chromosome misalignment and anaphase bridges.

This interpretation is supported by coarse-grained polymer simulations of cell nuclear mechanical response. In our model, chromatin is modeled as a crosslinked polymer gel, while the nuclear lamina is modeled as a polymeric shell that is physically linked to the interior chromatin. This model previously recapitulated measurements from nucleus micromanipulation experiments, which observed the two-regime force-extension relationship, its dependence on histone modifications (altering chromatin fiber stiffness) and nuclear lamins (altering the polymeric shell meshwork), and the changes to the shape of the nucleus when it is stretched (Banigan et al., 2017; Stephens et al., 2017). Here, we showed that the short-extension stiffness, but not the long-extension stiffness, is highly sensitive to the number of chromatin-chromatin crosslinks (Figure 4C,D). Specifically, with few crosslinks, the short-extension spring constant is small (~50% of the WT simulation), which parallels the observed result for nuclei with HP1α degraded (Figure 2) and the expected mechanics for nuclei with the HP1α dimerization mutant (see above). Thus, the separate roles of histone methylation and HP1α can be modeled as altering the chromatin polymer fiber and chromatin-chromatin crosslinks, respectively (Figure 6). Interestingly, although we model HP1α as a permanent chromatin-chromatin crosslink, chromatin binding by HP1α in vivo is transient, with a typical exchange time of ~10 s (Cheutin et al., 2003; Festenstein et al., 2003; Kilic et al., 2015). Apparently, chromatin-bound HP1α is sufficiently abundant that crosslinks continuously percolate interphase chromatin to provide a robust mechanical response and thereby maintain nuclear shape. Simulations with transient crosslinks or experiments with HP1α chromatin-binding mutants (Nielsen et al., 2001) could further investigate this phenomenon and its implications for chromatin organization, chromatin-based nuclear mechanics, and nuclear morphology.

More broadly, the finding that HP1α acts as a chromatin crosslinker is consistent with other experimental data suggesting that chromatin organization and mechanics is supported by widespread physical crosslinking. Recent chromosome conformation capture (Hi-C) and micromanipulation experiments show that moderate fragmentation of chromatin does not alter genome organization and mechanics. On the basis of these experiments, it is hypothesized that chromatin may be physically linked as frequently as once per 10–25 kb (Belaghzal et al., 2021). Our data show that HP1α is one of the likely many possible chromatin crosslinking elements in the genome. There is a growing list of chromatin proteins and nuclear components contributing to maintenance of nuclear morphology, some of which have been identified by a genetic screen for effects on nuclear morphology (Tamashunas et al., 2020) and a variety of other experiments (reviewed in Stephens et al., 2019a). Other chromatin crosslinkers to be investigated include chromatin looping proteins and other components implicated by various experiments, such as cohesin, CTCF, mediator, and possibly RNA.

Crosslinking and gelation are intimately coupled to phase separation (Harmon et al., 2017). Therefore, HP1α may contribute to nuclear mechanics through a phase transition mechanism. In a phase transition model, HP1α dimers crosslinking certain regions of the chromatin polymer would lead to polymer-polymer or sol-gel transitions (Khanna et al., 2019; Tanaka, 2002) that contribute to the elastic modulus of the whole network (Colby and Rubinstein, 2003; Semenov and Rubinstein, 2002; Shivers et al., 2020). Furthermore, HP1α binding to methylated histones is known to alter the structure of the nucleosome core, which could promote nucleosome-nucleosome interactions, and induce polymer-polymer phase separation of the chromatin fiber (Sanulli et al., 2019). Additionally, purified HP1α protein in vitro exhibits liquid-liquid phase separation by itself, with naked DNA, and with nucleosome arrays (Larson et al., 2017; Shakya et al., 2020), and HP1 condensates bound to dsDNA in vitro can lend mechanical strength (Keenen et al., 2021). The material properties of these in vitro condensates varies depending on the chromatin content (Larson et al., 2017; Shakya et al., 2020). More generally, phase separation in an elastic network such as chromatin can be regulated by the local mechanical properties of the material (Shin et al., 2018; Style et al., 2018). Together, these observations suggest a complex physical picture that is dictated by both HP1α’s self-interaction and chromatin binding capabilities, in addition to length, concentration, and phase behavior of the chromatin itself (Gibson et al., 2019; Maeshima et al., 2021Strickfaden et al., 2020). The material state and categorization of the phase transition of HP1α-rich heterochromatin in vivo have been debated (Erdel et al., 2020; Larson et al., 2017; Strom et al., 2017; Williams et al., 2020), and the underlying chromatin may be ‘solid-like’ (Strickfaden et al., 2020), so further work is necessary to completely understand the interplay of these components in determining phase behavior and mechanics of the interphase nucleus.

HP1α is a mechanical element of mitotic chromosomes and is essential for proper mitosis

Mechanical components of interphase chromatin may remain attached to mitotic chromosomes in order to maintain the mechanical strength of chromosomes during mitosis. Recent work has shown that heterochromatin-based histone modifications/methylation also control the mechanical strength of chromosomes, while euchromatin-based histone acetylation does not (Biggs et al., 2019). That paper hypothesized that increased histone methylation could be aided by ‘histone reader’ heterochromatin-associated proteins, specifically HP1α. Our data reveal that, similar to HP1α in interphase nuclei, HP1α during mitosis is a significant mechanical component of the mitotic chromosome (Figure 5). HP1α degradation leads to more extensible mitotic chromosomes, but the stiffness can be recovered by hypermethylation via methylstat treatment. The fact that HP1α still provides mechanical stiffness in mitotic chromosomes, a chromatin-only system without lamins, further supports that HP1α mechanically functions as a chromatin crosslinker. Previous work has proposed that mitotic chromosomes are dense polymer gels based on their elastic response, which relies on the continuity of the DNA backbone (Poirier and Marko, 2002), topology (Kawamura et al., 2010), and the chromatin cross-bridging condensin protein complex (Sun et al., 2018). Our experiments implicating HP1α as a crosslinking element (in interphase) and measuring the mechanical contributions of HP1α in mitotic chromosomes further support this picture. Methylation could serve as an additional compaction agent by providing further crosslinking, stiffening the chromatin fiber itself, or generating poor solvent conditions that further compact mitotic chromosomes (Batty and Gerlich, 2019; Gibcus et al., 2018; Maeshima et al., 2018). Together, these components generate the rigidity necessary for robust mitotic chromosomes.

Loss of HP1α results in dysfunction, marked by improper chromosome alignment and segregation. Previous reports had noted that loss of HP1α and HP1γ, specifically at the centromere, causes increased incidence of chromatin bridges (Lee et al., 2013) and mitotic alignment errors (Yi et al., 2018), genetic deletion of HP1α increases merotelic and syntelic attachments (Bosch-Presegué et al., 2017), and mitosis is dependent on HP1α phosphorylation (Chakraborty et al., 2014). Our findings with rapid degradation of HP1α reveal a threefold increase in both misalignment and missegregation, which were mostly observed as anaphase bridges, which could be due, in part, to aberrant DNA damage repair (Chiolo et al., 2011; Peng and Karpen, 2007). Our results are in agreement with HP1α interacting with LRIF1 at the centromere, which when perturbed results in similar misalignment and missegregation (Akram et al., 2018). However, further work is required to determine if chromosome misalignment is due to a biochemical pathway or mechanical pathway where whole-chromosome mechanics controlled by HP1α influences proper segregation.

Conclusion

We have established that HP1α has consistent mechanical and functional implications for chromosomes throughout the cell cycle. While rapid degradation of HP1α has little effect on the global transcriptional profile, loss of HP1α strongly impairs interphase and mitotic chromosome mechanics. This leads to deleterious and potentially catastrophic effects, such as abnormal nuclear morphology and chromosome segregation defects. When present, HP1α is a crosslinking element, and it mechanically stabilizes interphase and mitotic chromosomes, suppressing abnormal nuclear deformations and mitotic defects. It remains unclear whether HP1α’s phase separation capability is important to this biophysical function. More broadly, our experiments demonstrate that mechanical softening of the nucleus due to loss of HP1α’s chromatin crosslinking ability, rather than transcriptional changes, could underlie defects in fundamental nuclear functions such as nuclear compartmentalization, DNA damage prevention and response, and migration, all of which have been shown to depend on nuclear mechanics (Gerlitz, 2020; Stephens, 2020; Xie et al., 2020). These mechanical changes could also have broad implications for human diseases, such as breast cancer, where increased invasiveness (migration ability) has been correlated with decreased HP1α levels (Vad-Nielsen and Nielsen, 2015) and inhibition of HP1α dimerization (Norwood et al., 2006). Overall, we have revealed a direct structural role for HP1α in whole-nucleus and mitotic chromosome mechanics that furthers our understanding of chromatin-based nuclear stiffness and has important cellular functional consequences.

Materials and methods

Key resources table
Reagent type (species) or resourceDesignationSource or referenceIdentifiersAdditional information
Gene (H. sapiens)CBX5GenBank
Cell line (H. sapiens)U20SATCCATCC HTB-96RRID:CVCL_0042
Transfected construct (H. sapiens)3' HP1α-AID- sfGFP 2A PuroRAddgene127906RRID:Addgene_127906
Transfected construct (H. sapiens)Guide RNA/Cas9 plasmid pX330 human 3' HP1α gRNAAddgene127906RRID:Addgene_127906
Sequence-based reagentGuide RNA sequence, 5’- acagcaaagagctaaaggag −3'This paper
Transfected construct (H. sapiens)HP1α-mCherryThis paper, in pHR vectorHP1α PCR from Addgene_17652
Transfected construct (H. sapiens)HP1αI165E- mCherryThis paper, in pHR vectorPoint mutant made with quickchange
AntibodyAnti-HP1 alpha (rabbit monoclonal)Abcamab109028(1:250)
RRID:AB_10858495
Antibodyanti-H3K9me2/3 (mouse monoclonal)Cell Signaling5327(1:100)
RRID:AB_10695295
Antibodyanti-Lamin B1 (rabbit polyclonal)Abcamab16048(1:500)
RRID:AB_443298
Antibodyanti-Lamin A/C (mouse monoclonal)Active Motif39287(1:1000)
RRID:AB_2793218
Chemical compound, drugAuxin (NaIAA)SigmaI51481 mM
Chemical compound, drugMethylstatSigmaSML0343-5MG1 μM
Software, algorithmKappa, nuclear curvatureFIJI
Schindelin et al., 2012

Cell lines, cloning, and characterization of HP1α-AID-sfGFP degron clone

Request a detailed protocol

U2OS (ATCC HTB-96) were validated by STR. These cells were cultured in DMEM/FBS and co-transfected with two plasmids, human 3' HP1α-AID- sfGFP 2A PuroR (Addgene 127906) and a guide RNA/Cas9 plasmid pX330 human 3' HP1α gRNA (Addgene 127907) with Lipofectamine 2000 according to manufacturer’s instructions. The guide RNA sequence, 5’- acagcaaagagctaaaggag −3', flanked the stop site of the CBX5 gene and was destroyed upon successful in-frame insertion of the AID-GFP 2A PuroR cassette. Modified cells were selected with 10 µg/ml puromycin and single-cell sorted into 96-well plates with a BD FACS Aria III gated with FACSDiva software to sort only the top 10% brightest GFP-expressing cells. Expression of HP1α-AID-sfGFP was monitored by fluorescence microscopy as clones were expanded and subjected to quality control (QC; quality control, consisting of immunoblotting, PCR and live cell microscopy, see supplementary materials). A homozygous clone that passed all QC (U2OS HP1α 4) was co-transfected with the transposon vector pEF1a-OsTIR-IRES-NEO-pA-T2BH (Addgene 127910) and SB100X in pCAG globin pA (Addgene 127909). Forty-eight hr post-transfection, cells were selected with 400 µg/ml G418 for 10 days (media with fresh G418 replaced every 2–3 days) and then allowed to recover in DMEM/FBS for 1 week. GFP positive cells were again single cell sorted, expanded and subjected to QC. Degradation of HP1α-AID-sfGFP by OsTIR1 was evaluated by flow cytometry, immunoblotting and live cell microscopy after treatment with 1 mM auxin (NaIAA, Sigma #I5148) for 4–16 hr. A clone (U2OS HP1α 4–61) that by all QC measures demonstrated no detectable HP1α-AID-sfGFP after auxin treatment was chosen and expanded.

Validation by PCR

Request a detailed protocol

Genomic DNA was extracted using the PureLink Genomic DNA Mini Kit (catalog number K182001) and PCR was performed with oligos that flanked the insertion site, yielding 2 PCR products for heterozygous HP1α clones or a single larger PCR product for HP1α clones homozygous for the AID-GFP-Puro insertion.

Cell line validation microscopy

Request a detailed protocol

Live cells were plated into four chambered glass bottomed dishes (Greiner Bio One, #627975) and mounted in a temperature and CO2 controlled chamber (Okolab) for viewing using a Nikon Eclipse Ti inverted microscope with a 100X, 1.45NA phase objective and Spectra X (Lumencor) LED excitation at DAPI (395/25) and GFP (470/24) wavelengths (used at 5% power). Cells grown on glass coverslips, fixed in 4% paraformaldehyde (Polysciences, #18814) and mounted in Prolong Diamond to preserve GFP signal were also prepared. Images were captured using an Orca Flash 4 sCMOS camera and analyzed, cropped and contrast adjusted for display using either Elements or Imaris software. Cells were tested for mycoplasma via imaging using hoechst weekly.

Immunoblotting and immunostaining

Request a detailed protocol

Cell pellets from each clone were resuspended and incubated in RIPA buffer (Thermo Scientific # 89901) containing 2x Protease inhibitor (Thermo Scientific # A32955) for 1 hr on ice, and then incubated for 10 min at RT with 25U benzonase nuclease (Millipore Sigma 70746-10KUN)/50 µL sample. After BCA protein quantification (Pierce), samples were subjected to reducing SDS-PAGE and LI-COR Western blot analysis. Anti-HP1 alpha primary antibody (Abcam #ab109028) was used at 1:250 and IRDye 680CW secondary (LiCOR #925–6807) was diluted 1:15000. Blots were scanned on an Odyssey CLx. Immunostaining was carried out as previously described (Politz et al., 2002) using Abcam #ab109028 primary antibody at 1:250, and secondary antibody (Jackson labs 711-165-152) at 1:200, and coverslips were mounted in Prolong Gold. Images were captured as described above.

RNA-seq

Request a detailed protocol

RNA was isolated using the Qiagen RNeasy kit according to manufacturer’s instructions. Cells were homogenized with a QIAshredder (Qiagen #79654) with β-mercaptoethanol in the RTL buffer. DNA was digested with RNase-Free DNase (Qiagen #79254) and purified with RNeasy MinElute Cleanup Kit (Qiagen #74204). Purified RNA was quantitated with a Nanodrop spectrophotometer and quality was confirmed on a bioanalyzer with a TapeStation R6K assay. A sequencing library from RNA with a RIN >9.5 was prepared using the TruSeq stranded mRNA Library Prep and sequencing was performed using an Illumina HiSeq 2500 workstation. There were over 16,000 genes with one transcript per million reads for control compared to auxin 4 hr as well as control compared to auxin 16 hr, The RNA-Seq reads were mapped with STAR and then quantified by RSEM, and the differential gene expression analysis was performed using DESeq2.

HP1α rescue constructs

Request a detailed protocol

Full length HP1α was amplified by PCR (Addgene 17652), and cloned using InFusion kit into a pHR lentiviral vector under an SFFV promoter and tagged C-terminally with mCherry and sspB. A mutation was introduced to disrupt dimerization at amino acid 165 in the chromoshadow domain, changing the codon ATA (coding for Isoleucine, I) to GAG (coding for Glutamic acid, E) to result in HP1αI165E. This mutation has been previously characterized to disrupt homodimerization of HP1α (Brasher et al., 2000).

Lentiviral expression of HP1α rescue constructs

Request a detailed protocol

LentiX cells were transfected with transfer plasmids pCMV-dR8.91 and pMD2.G, as well as expression construct of interest in a 9:8:1 mass ratio into HEK293T cells using FuGENE HD Transfection Reagent (Promega) per manufacturer’s protocol. After 48 hr, media containing viral particles was collected and filtered using 0.45 micron filter (Pall Life Sciences), and either used immediately or stored at −80°C. HP1α-AID-sfGFP cells were plated at 15–20% confluency on glass-bottom 96-well plates (Cellvis) and infected with 10–50 µL of virus-containing media. After 24 hr, viral media was removed and replaced with fresh DMEM, and cells were fixed or imaged at 3–7 days post-infection.

Immunostain, microscopy, and morphological analysis of nuclear shape in fixed cells

Request a detailed protocol

HP1α-AID-sfGFP cells expressing mCherry-tagged HP1αWT or HP1αI165E were treated with control media (no auxin) or 1 mM auxin (NaIAA, Sigma #I5148) for 16 hr before being fixed in 4% paraformaldehyde for 10 min, washed three times in PBS, permeabilized with 1% triton X-100 in PBS for 1 hr at room temperature with rocking, blocked with 5% FBS in 0.25% PBST for 1 hr at room temperature with rocking, and incubated with anti-Lamin A/C antibody 1:1000 (Active Motif, 39287) in block overnight. Samples were washed again three times with PBS and incubated with Goat anti-mouse IgG secondary antibody conjugated to Alexa fluor 647 (Thermo Fisher, A-21236) for >2 hr, washed again and incubated with Hoechst 1:2000 in PBS for 30 min. Images of fixed and stained cells were obtained with a spinning-disk confocal microscope (Yokogawa CSU-X1) with 100X oil immersion Apo TIRF objective (NA 1.49) and Andor DU-897 EMCCD camera on a Nikon Eclipse Ti body. Live samples were maintained at 37°C and 5% CO2 by a 96-well plate incubation chamber (Okolab). 405, 488, 561, and 647 lasers were used for imaging Hoechst, sfGFP, mCherry or Alexa 568, and Alexa 647, respectively. Laser power and digital gain were consistent for imaging all samples across an experiment, allowing for quantitative comparison of fluorescent intensities. Morphological analysis was performed in FIJI using a plugin that measures curvature; Kappa (Schindelin et al., 2012), which was created originally by Kevan Lu and is now maintained by Hadrien Mary. Briefly, one z-slice of the Lamin A/C immunostain channel at the center of the height of the nucleus was loaded into the Kappa plugin, traced, and a closed curve was fit to the signal. Curvature along the nuclear envelope trace was calculated as the inverse radius of curvature with the plugin and an average value of curvature per nucleus was recorded.

Peripheral association of chromatin

Request a detailed protocol

HP1α-AID-sfGFP cells were treated with control media or 1 mM Auxin for 16 hr, then immunostained with α-H3K9me2/3 (Cell Signaling, mouse mAb #5327) and α-Lamin B1 (Abcam ab16048), as described above, and fixed-cell images obtained as above. Lamin B1 stain was used in FIJI to define a nuclear periphery mask, and enrichment was calculated as average intensity within the periphery mask divided by average intensity of the nuclear interior.

Microscopy and morphological analysis of nuclear shape in living cells

Request a detailed protocol

Control media (no auxin) or 1 mM auxin (NaIAA, Sigma #I5148) was added to HP1α-AID-sfGFP cells expressing an miRFP-tagged histone H2B plated in 96-well glass bottom plates (Cellvis). Twenty-five X-Y points were chosen in each of the control and experimental wells, and a z-stack ranging eight microns was collected at each point every 30 min for 12 hr (with auxin added to experimental wells at time 0 hr). Morphological analysis was again performed with the FIJI plugin Kappa, this time using the histone signal to delineate the edge of the nucleus.

Nuclear rupture and DNA damage analysis

Request a detailed protocol

HP1α-AID-sfGFP U2OS cells were grown in cell culture dishes containing glass coverslip bottoms (In Vitro Scientific). Cells were treated 1 or 2 days prior to imaging with Cell Light Nucleus-RFP (NLS-RFP Fisher Scientific). Cells were untreated or treated with auxin 4–6 hr prior to imaging. Cells were imaged with a 40 × 0.75 NA air objective on an environmental incubation (37°C and 5% CO2) at 2 min intervals for 3 hr. Nuclear ruptures by observing RFP spilling out of the nucleus and into the cytoplasm as outlined in Robijns et al., 2016. DNA damage foci were counted using the Immunostaining procedure above with ɣH2AX conjugated Alexa 657 antibody (CST 9720, 1: 300) and Elements Bright Spot detection to determine the number of foci.

Cell protocol for single nucleus and mitotic chromosome isolation

Request a detailed protocol

Micromanipulation experiments used U2OS parent or HP1α-AID-sfGFP cells maintained in DMEM (Corning) with 10% fetal bovine serum (FBS) (HyClone) and 1% 100x penicillin streptomycin (Corning). The cells were plated and allowed to recover 1–3 days before nucleus or chromosome isolation. 1 mM auxin (NaIAA, Sigma #I5148) was added 4–6 hr before nucleus and chromosome isolation in ‘+auxin’ and ‘+auxin+methylstat’ experiments, and 1 μM methylstat was added 30–38 hr before ‘+methylstat’ and ‘+auxin+methylstat’ experiments. All experiments were performed without synchronization.

Single nucleus and mitotic chromosome isolation

Request a detailed protocol

Single nucleus (Stephens et al., 2017) and chromosome isolation (Biggs et al., 2019) experiments were performed on an inverted microscope (IX-70; Olympus) with a 60 × 1.42 NA oil immersion objective with a ×1.5 magnification pullout. Nuclei and Chromosomes were isolated at room temperature and atmospheric CO2 levels in DMEM 10% FBS 1% pen strep media in 3 hr or less to ensure minimal damage to the cells and chromosomes. Before isolation all cells were imaged for the absence or presence of HP1α with or without auxin treatment, respectively. For nucleus isolation, cells were treated with 1 µg/mL latrunculin A (Enzo Life Sciences) for 45 min before isolation to depolymerize the actin cytoskeleton. Interphase cells were lysed with 0.05% Triton-X 100 in PBS. After lysis, micromanipulation pipettes filled with PBS were used to capture and position the single isolated nucleus. Isolation aimed for G1 nuclei determined by their size (10–15 µm along the major axis). For chromosomes, prometaphase mitotic cells were identified by eye and lysed with 0.05% Triton-X 100 in PBS. After lysis, the bundle of interconnected chromosomes fell out of the cell and stabilized with a PBS filled pipette by light aspiration. While the bundle was stabilized, one end of a loose chromosome was aspirated into an easily bendable (Kavg = 40 pN/µm) ‘force’ pipette, moved away from the bundle, where the other end of the chromosome was grabbed by a stiff pipette. The bundle was heavily aspirated into the stabilizing pipette and then removed, leaving an isolated chromosome to be manipulated.

Nucleus mechanics measurements

Request a detailed protocol

The isolated nucleus is suspended between a stiff pull pipette and a pre-calibrated force pipette for defined size (2.8–3.3 µm diameter) and bending constant (1.5–2.0 nN/µm). The pull pipette provides either 3 µm extension (short regime only) or 6 µm extension of the nucleus (long regime) at a rate of 0.05 µm/s. Bending of the force pipette relative to extension of the nucleus provides a measure of force. Data is transferred to Excel where the slope of the force-extension provides a nuclear spring constant for chromatin (short extension 0–3 µm extension) and a lamin-based strain-stiffening nuclear spring constant (long regime slope minus short regime slope).

Mitotic chromosome mechanics measurements

Request a detailed protocol

Once a mitotic chromosome was isolated, the stiff pipette was moved 6.0 µm at a rate of 0.20 µm/s with step sizes of 0.04 µm/step using a Labview program, while the force pipette (Fp) and stiff pipette (Sp) were visually tracked. A linear regression of the deflection vs stretch (Fp/(Sp-Fp)) slope was calculated, multiplied by the force pipette spring constant (calibrated after the experiment) to give the spring constant of the chromosome, and multiplied by the initial length of the chromosome, to give the doubling force of the chromosome in a custom Python script, which is publicly available on GitHub (https://github.com/ebanigan/doubling_force) (Shams and Biggs, 2021).

Mitotic chromosome fluorescence

Request a detailed protocol

Cells were imaged on an inverted microscope (IX-70; Olympus) with a 60 × 1.42 NA oil immersion objective with a ×1.5 magnification pullout. in the GFP channel once a mitotic cell was identified and the final isolated chromosome was imaged in the GFP channel for each experiment to determine if they contained HP1α. Periodically, the chromosome bundle was also imaged in the GFP channel.

Nuclear morphology solidity measurements

Request a detailed protocol

Nuclei were selected via intensity threshold in Hoechst channel and made into an object or ROI and reported for shape solidity, which is a ratio of area over convex area of the nucleus. The threshold of 0.96 solidity was used to determine normal versus abnormally shaped nuclei.

Brownian dynamics simulations

Request a detailed protocol

Brownian dynamics simulations of a polymeric shell linked to an interior crosslinked polymer were performed as described previously (Banigan et al., 2017; Stephens et al., 2017). Simulation source code is publicly available on GitHub (https://github.com/ebanigan/shell-polymer) (Banigan, 2021). A total of 1000 shell subunits with diameter as = 0.71 μm are randomly placed on a sphere of radius Ri = 10 μm, which is shrunk to R0 = 5 μm during the simulation initialization (Banigan et al., 2017). Each shell subunit is connected by springs to 4 ≤ z ≤ 8 nearest neighbor shell subunits (<z > ≈4.5). A linear polymeric chain of 552 subunits with diameter ap = 0.57 μm, connected by springs, is initialized in a random globular conformation within the shell. The polymer is randomly crosslinked with NC crosslinks, where NC = 55 (20% of all polymer subunits are crosslinked) unless noted. NL polymer subunits near the surface of the sphere are linked by springs to the nearest shell subunit; NL = 40 (i.e. 7.2% of all polymer subunits or 22% of all peripheral subunits, defined by contact with the shell subunits in the initial configuration, are linked to the shell) unless noted. Tensile force is exerted across the nucleus by exerting force F along the x-axis on a single shell subunit at each of the two poles.

Spring potentials governing interactions between subunits have the form Usp=(ksp/2)(rij-rij,0)2 for rij>rij,0, where rij is the distance between subunits i and j, rij,0 is the sum of the two subunit radii, and ksp is the spring constant, which depends on the type of potential. ks = 0.8 nN/μm for shell-shell springs, kp = 1.6 nN/μm for ‘polymer springs’ connecting subunits along the polymer backbone, kC = kp for ‘crosslink springs’ connecting polymer subunits, and kL = kp for springs linking the polymer to the shell. All subunits repel each other via soft-core excluded volume interactions, modeled as Uex = (kex,ij/2)(rij-rij,0)2 for rij<rij,0, where kij is the repulsive spring constant; kex,ij = ks if i and j are both shell subunits, kex,ij = kp if i and j are both polymer subunits, and kex,ij = 2kskp/(ks +kp) if one is a shell subunit and the other is a polymer subunit.

All subunits are subject to uncorrelated thermal noise (T = 300 K). The system obeys the overdamped Langevin equation, which is solved by an Euler algorithm (Allen and Tildesley, 1987) with timestep dt = 0.0005.

Data availability

We have provided the RNAseq data sets in the supplemental material as excel files.

References

  1. Book
    1. Colby RH
    2. Rubinstein M
    (2003)
    Polymer Physics
    Oxford University.
    1. Stephens AD
    (2020) Chromatin rigidity provides mechanical and genome protection
    Mutation Research/Fundamental and Molecular Mechanisms of Mutagenesis 821:111712.
    https://doi.org/10.1016/j.mrfmmm.2020.111712

Decision letter

  1. Geeta J Narlikar
    Reviewing Editor; University of California, San Francisco, United States
  2. Kevin Struhl
    Senior Editor; Harvard Medical School, United States
  3. Sy Redding
    Reviewer; University of California, San Francisco, United States

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Acceptance summary:

This manuscript describes a significant and exciting role for the heterochromatin protein HP1-α in chromatin mechanics and nuclear shape stabilization that is suggested to rely on HP1-α's ability to cross-link different regions of chromatin. The work will be of much interest to communities studying chromatin biology and nuclear mechanics.

Decision letter after peer review:

Thank you for submitting your article "HP1α is a chromatin crosslinker that controls nuclear and mitotic chromosome mechanics" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, and the evaluation has been overseen by a Reviewing Editor and Kevin Struhl as the Senior Editor. The following individual involved in review of your submission has agreed to reveal their identity: Sy Redding (Reviewer #2).

As you will see all reviewers found the work to be exciting and significant. Each of them also had some major concerns that need to be addressed before we can consider publication in eLife. Through discussion amongst the reviewers, we have outlined the essential revision recommendations below. When you submit your revision, please describe your responses to the essential revision recommendations in the context of your point-by-point responses to the individual comments by each reviewer.

Summary:

This manuscript describes a significant and exciting role for the heterochromatin protein HP1-α in chromatin mechanics and nuclear shape stabilization that may rely on the ability of the protein to cross-link different regions of chromatin. While the experiments clearly highlight the potential for such a role of HP1-α, additional work is needed to strengthen the statistical significance of the results, test the functional impact of the engineered HP1-α protein and test the role of HP1-α in chromosome mechanics during mitosis.

Essential Revisions:

1. The section of the manuscript focused on effects on mitosis is weak and should be removed or substantially improved as outlined in detail in the reviews.

2. Cell shape analyses should show actual measurement data (not phenotypic classification) and the dependency of the nuclear shape defect on mitosis should be addressed.

3. The role of HP1-lamin interaction in nuclear mechanics should be addressed.

4. Rev#2 makes the important point of validating the function of the engineered protein, which relates to concerns of Reviewer#1 on the quality (abundance) of the data showing surprising absence of effects on heterochromatin and transcription. This should be experimentally addressed or extensively discussed, as Rev#2 suggests.

5. There is concern about the small sample sizes and the differences in sample size between experimental groups of individual experiments. This concern should be addressed by increasing the sample size and making the sample sets being compared similar in size.

Reviewer #1:

Strom and co-authors describe a new, very intriguing role for HP1alpha in chromatin mechanics and nuclear shape maintenance which is independent of known nuclear mechanoregulators: histone methylation or nuclear lamina, and which is instead driven by HP1alpha-mediated crosslinking of chromatin. While conceptually exciting, the central conclusions lack definitive experimental evidence. Further, key measurements lack true statistical power and the polymer model predictions are not stringently challenged. Finally, the functional relevance of this HP1-mediated mechanical regulation remains mostly unexplored, as the chromatin mis-segregation phenotype is only superficially analyzed and its connection with chromatin mechanics is elusive.

1. Exclusion of contribution of chromatin methylation and lack of changes in chromatin compaction are based on a single chromatin mark (H3K9me2,3) and one example image of H2B-GFP, respectively. To truly understand the consequence of HP1alpha loss they authors would have preferentially performed chromatin accessibility analyses such as ATAC seq and/or super resolution microscopy.

2. Similarly, conclusions of HP1alpha being essential for nuclear shape maintenance (Figure 3) were made based on just 25-30 nuclei – given the very elegant and efficient Auxin degron system and simple 2D model system, it is unclear why so few cells were analyzed. In additional, instead of expressing shape abnormalities by a descriptive classification (% of cells with abnormal nuclei) the authors should include a panel of detailed parameters to precisely quantify measure cell shape, such as solidity and the EFC ratio of nuclear shape (which some of the co-authors established in a recent publication).

3. Evidence for chromatin-chromatin crosslinks being affected in the I165E mutants should be experimentally demonstrated for example using super resolution

4. Mechanical measurements should be validated with an independent, higher throughput mechanical manipulation method, such as AFM or imaging-based methods

5. The validity of the polymer model that excludes a role for HP1-lamina or inner nuclear membrane interactions in nuclear mechanics is based solely on the short- vs long-extension stiffness data that the authors have previously validated for chromatin vs lamina-dependent mechanical regimes. The role of HP1 and lamina interactions through LBR or PRR14 should be experimentally excluded with mutants or LBR/PRR14 siRNA, in particular as these proteins have been shown to be important for nuclear shape maintenance.

6. Studies around the role of HP1alpha in mitotic segregation are not described clearly and the connection with chromosome mechanics remain unclear. The experiments are severely underpowered with only very few cells analyzed for mechanical measurements (just 8-14 nuclei) and 3-fold differences in sample size per condition (Figure 5G" "n=15-50 mitotic cells for each condition…") for metaphase misalignment measurements.

Reviewer #2:

The manuscript from Strom, Biggs, and Banigan, et al. is an outstanding effort and an important contribution to understanding the complex roles chromatin and protein effectors play in the nucleus. It certainly rises to the level of publication, and I found it to be generally very exciting and compelling work. I have one major criticism of the work below, but would be happy to accept the paper for publication.

The major criticism I have of the work is the validation of the HP1 protein after addition of the degron and sfGFP. Especially, given that there are a number of findings in this work are seemingly contrary to expectations, i.e no change in gene expression or H3K9 maintenance. From my reading, the only validation in the manuscript that I see that adding the degron and GFP onto the C-terminus of HP1a (more than tripling the MW right next to the protein-protein interaction hub of HP1) does not affect its biology is a localization measurement. And while I realize that it is not easy to validate a structural protein, it would be nice to have some more corroboration or comparison to experiments that have investigated the formation of heterochromatin by HP1 proteins.

For example, to reconcile these data with previous findings, one might predict that gene expression or H3K9 levels should eventually recapitulate previous knockdown/out studies if the auxin treatment were to persist, provided cell type and cell cycle considerations. While RNA seq experiments are fairly involved, I would think it possible to test the methylation levels in the auxin-HP1 line after a few days of depletion to see if levels do come down to previously reported levels, at least at specific satellite sequences. Alternatively, it might be easier to assay the AID-GFP-HP1 at exogenous sites as has been previously reported. Though, it is noted however that the dimerization mutant does perform as expected.

In addition to validation, it would also be helpful to more directly compare the timing of your experiments with those done elsewhere to give the reader a better idea of when changes (epigenetic, transcriptional) are likely to arise, etc. Though, discussion is somewhat confounded by the diversity of HP1 activity in different organisms. There is the distinction of >24 hours in the discussion, but there are several HP1 experiments that show faster changes than this, specifically compaction following recruitment and transcriptional repression at exogenous sites.

Reviewer #3:

The authors use state-of-the-art genome engineering and degron tagging approaches combined with transcriptional profiling and sophisticated biophysical measurements. Most of the results are clear and convincing, but the section on mitotic chromosomes has substantial weaknesses and should either be removed or extended to substantiate the key conclusions. The main part on HP1alpha's role in interphase nuclei, however, is interesting and well supported by the data.

1. Localization of HP1alpha on mitotic chromosomes. The authors claim that HP1alpha remains bound to mitotic chromosomes, but this is not visible in the data shown in the paper. Prior studies have in fact clearly shown that HP1alpha completely dissociates from chromosome arms, while only a small fraction remains bound to pericentric chromatin (Hirota et al., 2005; Fischle et al., 2005; Serrano et al., 2009; Akram et al., 2018). The authors reference these papers to provide a rationale for studying HP1alpha's potential role in mechanical rigidity of chromosome arms, but they neglect the fact that the previous studies showed that HP1alpha only localizes to a very small region surrounding centromeres. This very limited enrichment of HP1alpha at pericentromeric regions is inconsistent with a direct major mechanical role of HP1alpha in stiffening chromosome arms as the authors suggest.

Figure 5A shows HP1alpha-GFP fluorescence throughout the entire cell with a potentially very slight enrichment in the central region. Without counterstaining DNA or chromatin, it is not possible to assess whether any HP1alpha remains bound to chromosomes, as chromosome positions are not known. The authors should provide high-resolution figures of HP1alpha-GFP counterstained with a DNA dye, ideally in live cells. They should further image chromosome spreads as in the prior work by Serrano et al., 2009 to clarify where on chromosomes HP1alpha binds. Figure 5C shows a fuzzy fluorescence signal for HP1alpha, but the image quality is insufficient to assess whether this is on chromosome arms or on pericentromeric regions. Moreover, without a control cell that does not express GFP-tagged HP1alpha, it is not possible to assess to which extent the fuzzy fluorescence signal shown in Figure 5C represents background autofluorescence. Alternatively, as the mitotic chromosome stiffness measurements are also not very convincing (see next point), the authors might consider removing the entire section on mitotic chromosomes.

2. HP1alpha's contribution to mechanical rigidity of mitotic chromosomes. The authors claim that HP1alpha depletion "strongly impairs" mitotic chromosome mechanics. Their microneedle-based force extension experiments shown in Figure 5E, however, show a large variability between individual measurements but only a very slight difference between HP1alpha-AID-sfGFP cells treated with auxin compared to untreated cells. There is a star that might indicate statistical significance (this is not explained in the legend), but even if that were true then the effect size would still be so small that biological relevance is questionable. Given the concerns about HP1alpha localization, the overall evidence supporting a "major role" of HP1alpha in the stiffness of mitotic chromosomes is insufficient to support this conclusion. The authors might either perform additional experiments to substantiate the conclusions or remove the part on mitotic chromosomes.

3. HP1alpha's role in regulating nuclear shape. The authors suggest that the softening of interphase nuclei resulting from degradation of HP1alpha directly causes aberrant nuclear morphologies. However, they also show that depletion of HP1alpha leads to chromosome missegregation, which is known to lead to aberrant nuclear morphologies owing to perturbed nuclear assembly during mitotic exit. To assess to which extent the nuclear morphology defects in HP1alpha depleted cells are caused directly via nuclear softening or indirectly via chromosome missegregation, the authors should separately analyze nuclear morphology shape changes as shown in Supplementary Figure 2 for those cells that have passed through mitosis after HP1alpha degradation versus those cells that remained in interphase during the entire imaging duration.

https://doi.org/10.7554/eLife.63972.sa1

Author response

Essential Revisions:

1. The section of the manuscript focused on effects on mitosis is weak and should be removed or substantially improved as outlined in detail in the reviews.

In the revised manuscript, we have substantially improved the section of the manuscript focused on the role of HP1α in mitotic chromosomes and function by adding experiments to address the Reviewers’ critiques. Specifically, we have: (A) Increased the number of micromanipulation force measurements to improve the statistics; (B) Imaged HP1α-AID-sfGFP on isolated chromosome bundles (in addition to the original imaging in cells) to assay its presence on chromosome arms and pericentromeric regions; (C) Imaged HP1α-AID-sfGFP in live cells, in which we previously observed HP1α-AID-sfGFP at pericentromeric regions; (D) Provided more measurements of metaphase and anaphase fidelity with and without auxin treatment; (E) Tracked the effects of mitotic errors on subsequent interphase nuclear shapes. Further details are provided below. Overall, our new data strengthens our novel findings of the role of HP1α in maintaining mitotic function and fidelity.

A. We have performed more single isolated chromosome micromanipulation force measurements, as requested by the reviewers (6-8 new measurements each, bringing the total to 14-20 each; Figure 5E). Our new data further supports our initial conclusions that HP1α is a mechanical component of mitotic chromosomes, as its depletion decreases doubling force from

262 +/- 49 to 148 +/- 12 pN/μm (p = 0.03). Furthermore, the new measurements have improved the statistics of the measurements for chromosomes from auxin+methylstat-treated cells, which remain similar to methylstat-treated chromosomes (p = 0.17), but are significantly different from those with auxin treatment (p = 0.04). This data supports the conclusion that levels of histone methylation are dominant in mitotic chromosome mechanics as compared to HP1α. Interestingly, this finding differs from observations in interphase nuclei, where HP1α and histone methylation appear to contribute similarly to nuclear stiffness. Thus, the measurement of mitotic chromosome stiffness shows that some mechanical principles are the same as for interphase nuclear mechanics (HP1α is a mechanical element) and others are different (mitotic chromosome stiffness is dominated by histone methylation rather than having similar contributions from both HP1α and histone methylation).

B. We have added new data quantifying the level of HP1α on isolated mitotic chromosome bundles. In the original manuscript we reported levels of HP1α-AID-sfGFP fluorescence in live mitotic cells, which after 4-hour auxin treatment decreased on average by 89% (now Figure 5B). We have added fluorescence measurements of ex vivo isolated chromosome bundles (Figure 5B, C, and figure supplement 1). We report that bulk fluorescence indicates that HP1α-AID-sfGFP associates with the chromosome bundle, and it decreases by ~75% after 4-hour auxin degradation. We have now provided multiple examples showing HP1α-AID-sfGFP fluorescence on isolated chromosome bundles. Line scans reveal that HP1α-AID-sfGFP fluorescence is present on chromosome arms (Figure 5C, with more examples in Figure 5—figure supplement 1) as well as at pericentromeric foci, as previously reported. The novelty of this finding is that HP1α on the chromosome arms may only be detectable by imaging isolated mitotic chromosome bundles because isolation removes chromosomes from the high background fluorescence in the mitotic cell. Below, in point C we note that pericentromeric foci are detectable by confocal, in agreement with the previously published reports cited by Reviewer 3, and we have now added HP1α-AID-sfGFP and Hoechst-stained DNA images to the manuscript (Figure 5—figure supplement 1C). We also now note that HP1α on chromosomal arms and at the pericentromere both should be able to contribute to mitotic chromosome mechanics. Altogether, this data strengthens our previous conclusion that HP1α is present on mitotic chromosomes and extends it to include the chromosome arms.

C. We have added new data imaging HP1α-AID-sfGFP in living cells via confocal microscopy. This data shows HP1α at the centromere relative to DNA/chromatin staining/fluorescence (see Figure 5—figure supplement 1C), which is in agreement with our own isolated bundle images as well as many other published reports.

D. To further bolster the mitosis section, we also increased the number of events measured for abnormal metaphase alignment and anaphase/telophase segregation. We have included another experimental replicate of – auxin and + auxin at 16 hours to increase the total number of events. This new data agrees with our original measurements. We have also included three experimental replicates measuring metaphase and anaphase/telophase abnormal events after rapid 4-hours of auxin-induced HP1α degradation (Figure 5F). These measurements further verify that loss of HP1α results in more metaphase misalignment and anaphase/telophase missegregation. Furthermore, it reveals that longer-term HP1α loss results in more mitotic failures, as we observe a significant increase in abnormal events when comparing short-term degradation of HP1α (4 hour auxin treatment) vs. longer-term degradation (16 hour).

E. In response to comments by Reviewers 1 and 3, we have provided new data analyzing how mitotic abnormalities in HP1α degraded cells lead to abnormal interphase nuclear morphology.

i. In the original manuscript, by tracking 42 cells' nuclear curvature measurements over 1 to 12 hours after addition of auxin, we clearly showed that abnormalities in nuclear morphology emerge during interphase (see Figure 2—figure supplement 2). Thus, over the short time intervals of rapid HP1α degradation, it appears that abnormal nuclear morphology can arise, independent of mitosis.

ii. Our new data provide time-lapse imaging to reveal that control untreated cells successfully complete mitosis and produce normally shaped sister nuclei. However, after 24 hours of auxin treatment, nearly half of HP1α-degraded nuclei tracked through mitosis exhibit abnormal mitotic events (in agreement with Figure 5F). However, HP1α degraded cells resulting in normal vs abnormal mitosis both produce sister nuclei with higher nuclear curvature(Figure 5G). Thus, abnormal mitosis is not required to produce abnormally shaped/high curvature sister nuclei post mitosis. Overall, abnormal nuclear morphology can arise due to the loss of nuclear shape regulation by HP1α during interphase (over short timescales) or arise after mitosis resulting from HP1α degradation (over long timescales).

As we discuss in the more detailed reply below, our reporting of experimental replicates and the associated number of measured events was confusing in the original manuscript. This may be why Reviewer 1 (and possibly others) was unsure about how many actual measurements were taken. We have clarified this in the figure legends and we now list the number of measurements for each of the replicates. We have also clearly defined asterisks denoting statistical significance. We cover this point again in Editor point 5.

2. Cell shape analyses should show actual measurement data (not phenotypic classification) and the dependency of the nuclear shape defect on mitosis should be addressed.

The original manuscript contained data addressing this point. In the revised manuscript, we better highlight this original data, state the quantified data instead using a phenotypic term, and provide increased numbers of nuclei for these measurements and new experiments to further address these points. The data quantify nuclear shape by either shape solidity (ratio of area to convex area) or nuclear curvature (phenotypic classifications of nuclei were then made based on these measurable quantities). The data shows that HP1α is essential for maintaining nuclear shape. The data also reveals that HP1α degradation decreased nuclear shape solidity or increased nuclear curvature. Higher solidity and decreased curvature can be recovered by increasing levels of histone methylation via methylstat or by introducing exogenous wild type HP1α, but not by exogenous expression of the dimer mutant HP1αI165E. This reveals that overall H3K9 methylation levels function independently of HP1α for these functions over 4-16 hour timescales, and that HP1α’s nuclear shape maintenance function depends on HP1α dimerization.

More specifically, the original manuscript measured nuclear shape changes via (A) nuclear shape solidity (Figure 2G); (B) timelapse nuclear curvature measurements (Figure 2—figure supplement 2); and (C) population nuclear curvature measurements, which we provide new data increasing biological replicates (Figure 3). We have revised the manuscript to include average nuclear shape solidity measurements in the text section “HP1α is a major mechanical component of the interphase nucleus that contributes to nuclear shape maintenance” and Figure 2 figure legend. Furthermore, we clarify we are reporting an increase in the number of nuclei with low solidity levels (< 0.96), which we for reference refer to as abnormal or irregular. To further address the reviewers’ concerns, we now additionally include: (D) new data, essentially recapitulating the experimental measurements of solidity and finding the same results, significance, and conclusions as already included in Figure 2; (E) new data to address the effect of abnormal mitotic events on nuclear morphology, as requested by the reviewers. Details are listed below.

A. Nuclear shape solidity. Original data and new/more data provided – Figure 2G consists of: “3 experiments each (shown as black dots) each consisting of n = 109, 102,105 control ; n = 137, 115, 165 auxin, n = 31, 34, 32 methylstat, and n = 102, 92, 78 auxin methylstat”. In the figure legend we have added, as requested by the reviewers, that average solidity also has similar statistical significance: “ Average measurements were similar for control, methystat, and auxin with methylstat 0.971+/- 0.0001 but different for + auxin 0.969+/-0.0015, p=0.005.” The standard deviation increased over ten-fold in + auxin conditions signalling a drastic change in the distribution of solidity. Specifically, under these conditions, there is an increase in the number of nuclei with low solidity (< 0.96 solidity as shown in Figure 2G). Thus we graphed the % of nuclei with low solidity to show that nuclear shape had destabilized.

B. Time series of nuclear curvature. Original data – We tracked nuclear curvature for single interphase nuclei over time during auxin-induced HP1α degradation, which clearly shows loss of shape was independent of mitosis and coincident with HP1α loss (Figure 2—figure supplement 2). Here we do measurements of nuclear curvature and assay for the presence of HP1α-AID-sfGFP each hour for 12 hours for 42 individual nuclei of – or + auxin. This data clearly shows that + auxin results in increased nuclear curvature coincident with HP1α-AID-sfGFP degradation during interphase and independent of mitosis. These experiments provide actual measurement data and address the dependency of the nuclear shape defects on mitosis (also see point E below).

C. Population nuclear curvature measurements. New data and original data – In the revised manuscript we provide 5 new experimental replicates to go along with the 3 original experimental replicates (8 total). This data strongly supports the major conclusion of the paper that HP1α dictates nuclear morphology though its dimerization. Population nuclear curvature measurements of HP1α degradation and rescue with exogenous HP1α (Figure 3) recapitulate both of these major findings from Figure 2G and Figure 2 – supplemental figure 2. In this figure, we now clearly state that for this data: “ Eight experimental replicates were measured for each condition (denoted as black dots) consisting – auxin, n = 37, 45, 45,57, 58, 57, 55, 54; + auxin, n = 60, 44, 41, 60, 60, 60, 60, 60 ; + auxin and WT exogenous rescue, n = 27, 30, 36, 60, 60, 60, 60, 19; + auxin and I165E exogenous rescue, n = 38, 40, 50, 59, 56, 58, 58, 51. P values reported as n.s > 0.05, * < 0.05, **< 0.01, ***< 0.001, ****<0.0001.”

D. New experiments recapitulating previous results of Nuclear Shape Solidity. New data shown here in Author response image 1. As a further proof of the validity of our data, we have redone the solidity measurements in a different lab setting to increase the number of measurements and show the data can be recapitulated. Here, we show that average solidity and % of nuclei with a solidity less than 0.96 are similar to our original data. The slight change in absolute numbers is due to different microscopes (widefield vs. confocal). 3 experiments each (shown as dots), with each experiment consisting of n = 96, 96, 97 untreated control (unt); n = 109, 96, 115 auxin (aux); n = 47, 48, 49 methylstat (ms); and n = 87,101,79 auxin methylstat (aux ms). The average nuclear shape solidity of auxin-treated cells is statistically significantly different (p < 0.05), while the rest are statistically similar (0.955 auxin vs 0.959, 0.960, 0.961 +/- 0.001 for control, methylstat, and auxin + methylstat, respectively). While these absolute measurements of solidity are less than in the original manuscript, the % abnormally shaped nuclei (solidity < 0.96) increases similarly by ~10% for auxin-treated nuclei and the change is significant (p < 0.05). The aim of sharing this data here is to show that these results are reproducible, robust, and rigorous through multiple measurements, measurement types, modalities, and replicates.

Author response image 1

E. Abnormal mitotic events. New data – To directly address the question of the dependency of the nuclear shape defects on mitosis we have included new data in the manuscript. From point B above, we clearly show in the original manuscript that abnormal nuclear shape is independent of mitosis. To determine the effect of abnormal mitosis (originally reported in Figure 5), we tracked nuclei/cells through mitosis either – auxin or + auxin for 24 hours. We now report that both normal and abnormal mitosis in +auxin conditions results in daughter nuclei with higher nuclear curvature measurements than – auxin (Figure 5 G). This new data provides the interesting insight that abnormal nuclear morphology can be independent of mitosis after rapid (4 hours) HP1α degradation, but HP1α degradation results in higher nuclear curvature in sister nuclei post mitosis on longer time scales (> 24 hours).

3. The role of HP1-lamin interaction in nuclear mechanics should be addressed.

The revised manuscript addresses the possible role of HP1α in linking chromatin to lamins/lamina or otherwise interacting with lamins through new experiments that investigate HP1α’s role in maintaining peripheral H3K9me2,3 heterochromatin. We note that the experiments requested by Reviewer 1 (modulating LBR and PRR14) are a significant undertaking that would not directly address the main points of the manuscript. As we describe below, our new experiment more directly addresses the possibility of HP1α as a chromatin-lamina linker and/or lamin interactor.

First, in the original manuscript Figure 4, computational simulations showed that altering the level of chromatin-chromatin crosslinking changes nuclear mechanics differently as compared to altering the amount of chromatin-lamin linkers. These simulations test the possible mechanical roles of HP1α as either a chromatin-chromatin crosslinker or chromatin-lamin linker. The simulations show that loss of chromatin crosslinkers decreases the short-extension spring constant while having little effect on the long-extension spring constant. Furthermore, the simulations show that the loss of chromatin-lamin linkers results in decreased nuclear spring constant for both the short and long extensions. The results for the simulations with different levels of chromatin crosslinking match experimental measurements of nuclear mechanics upon auxin-induced degradation of HP1α. Together, these results show that loss of HP1α perturbs nuclear mechanical response in a manner consistent with a decrease in crosslinking within the chromatin polymer gel, but inconsistent with changes to the frequency of mechanically stable chromatin-lamina links. Based on these results, combined with our HP1 dimerization experiments and HP1α’s established ability to bridge nucleosomes (Machida et al. Mol Cell 2018), we conclude that the likely primary mechanical function of HP1α is to crosslink chromatin.

Second, we have provided new data to address possible HP1α-lamin interactions. Reviewer 1 notes that there are known chromatin-lamin linkers, including LBR and PRR14. As shown by previous work (Dunlevy et al., 2020; Giannios et al., 2017; Nikolakaki et al., 2017; Poleshko et al., 2013; Solovei et al., 2013), depletion of these known chromatin-lamin interacting proteins results in a significant decrease in peripheral H3K9me2,3 heterochromatin. To test if HP1α has a role as a chromatin-lamina linker, we measured both the level of HP1α at the nuclear periphery and the density of H3K9me2,3 heterochromatin at the nuclear periphery before and after HP1α degradation. Known chromatin-lamin linkers PRR14 and LBR are both themselves enriched at the nuclear periphery AND upon loss of them the peripheral enrichment of H3K9me2,3 is significantly decreased (Poleshko et al., 2013; Solovei et al., 2013). Unlike H3K9me2,3, which is enriched at the periphery (1.5 average periphery/average internal signal), HP1α peripheral signal/average (0.86) is similar to DNA staining (0.99), with both showing no peripheral enrichment. Rather, both HP1α and DNA appear to be equally partitioned between the nuclear periphery and the nuclear interior (average periphery/interior ratio~1). Thus, in this cell line, HP1α does not exhibit peripheral enrichment that is characteristic of known chromatin-lamin tethering proteins such as PRR14 and LBR. Furthermore, upon degradation of HP1α, the peripheral enrichment of H3K9me2,3 does not change significantly. Thus, unlike known chromatin-lamin linkers LBR and PRR14, we find that HP1α does not have a major role in maintaining peripheral localization of H3K9me2,3 in this cell line. Taken together this new data supports our conclusion that the mechanical contributions of HP1α are not due to associated chromatin-lamin interactions.

4. Rev#2 makes the important point of validating the function of the engineered protein, which relates to concerns of Reviewer#1 on the quality (abundance) of the data showing surprising absence of effects on heterochromatin and transcription. This should be experimentally addressed or extensively discussed, as Rev#2 suggests.

To address this point, we have added new experiments validating the function of the engineered protein. We have also added more discussion of how the lack of changes in heterochromatin organization and gene transcription, while apparently surprising, is actually in agreement with previous studies of mammalian cells. We provide several key points and new experiments to support this assertion:

A. Firstly, we modify the endogenous HP1α loci rather than using exogenous HP1α, which can result in overexpression due to not being under the control of the endogenous promoter. Thus, our modified HP1α is under native promoter control.

B. We have provided new RNA-seq data showing that the transcription profile does not change significantly for the modified HP1α-AID-sfGFP cell line as compared to the parent cell line. RNA-seq reveals 76 upregulated and 56 downregulated transcripts, which represents just 0.8% of the >16,600 genes. GO analysis reveals no change to the nucleus or chromatin proteins. GO analysis of cellular function returns only extracellular changes for upregulated genes (matrix, region, space, exosome, and vesicle) and no significant GO term for downregulated genes. This new data is added as a Data Supplement 3.

C. The main measurables in the manuscript for HP1α function are similar for parental and the AID-sfGFP modified cell lines, while they differ significantly when HP1α is degraded via auxin. These measurables include: (1) general HP1α distribution, Figure 1 (2) H3K9me2,3 levels, Figure 2—figure supplement 1D (3) interphase nuclear mechanics, Figure 2 (4) nuclear shape, Figures 2 and 3 (5) mitotic fidelity, Figure 5—figure supplement 1D.

D. We note that C-terminal tagging of HP1α has been shown to not disrupt function through rescue experiments in previous studies (Dialynas et al. 2007 J Cell Sci) as well as in our studies. Here in Author response image 2, we also supply supporting images of endogenous subnuclear localization of HP1α in parental U2OS cells stained with anti-HP1α (A), which have similar localization to the nucleolar periphery as an exogenous C-terminally tagged construct (B). C-terminally tagged endogenous HP1α-GFP-AID also localizes to nucleolar periphery (C) and directly overlaps with both N-terminally tagged (D) and C-terminally tagged (E) exogenous mCherry constructs, suggesting that the presence and orientation of the tag does not disrupt HP1α chromatin binding or subnuclear localization. Also, exogenous wild-type HP1α rescued nuclear shape while the dimer mutant did not, which Reviewer 2 noted was a significant finding to support the functionality of the endogenous modified HP1α.

Author response image 2
A.

Antibody stain with anti-HP1α in parental U2OS showing endogenous localization to heterochromatic areas and around nucleoli. B. Antibody stain with anti-HP1α in parental U2OS expressing exogenous C-terminally tagged HP1α-mCherry, showing similar localization around nucleoli and in heterochromatic areas. C. Antibody stain with anti-HP1α in HP1 α-GFP-AID C-terminally tagged cell-line, showing colocalization between antibody and GFP tag, and normal localization around nucleoli. D. Co-expression of N-terminally tagged exogenous mCherry-HP1α and endogenously tagged HP1 α-GFP-AID showing no difference in protein localization between N-terminal and C-terminal tagged populations. E. Co-expression of C-terminally tagged exogenous HP1α-mCherry and endogenously tagged HP1α=GFP-AID showing no difference in protein localization between endogenous and exogenous C-terminally tagged proteins.

E. The seemingly surprising absence of effects of HP1α depletion can be understood by considering the now well documented differences between fly and mammalian HP1α. In particular, previous studies in Drosophila show that HP1α depletion can modify position effect variegation (Eissenberg et al. Genetics 1992) and modulate levels of transposon and satellite RNA expression (Sienski et al. Cell 2012) as well as cell cycle regulators (De Lucia et al. Nucleic Acids Res 2005). In murine cells, HP1α is involved in repression of major and minor satellite RNA (Eissenberg and Elgin TrendsGenet 2014) and also helps regulate olfactory receptor expression (Clowney et al. Cell 2012). However, as we note in the manuscript, published studies of HP1α depletion or loss in human cells are in agreement with our findings of lack of changes in transcription and histone methylation (Zeng et al. Epigenetics 2010) (Maksakova et al., 2011), but a significant change in mitotic fidelity(Levine, Vander Wende, and Malik 2015; Abe et al. 2016) Furthermore, studies finding that HP1α acts as a repressor of transcription typically focused on a single or small set of sites to which HP1α was highly recruited (Erdel et al. Mol. Cell 2020; Li et al. Development 2013).

5. There is concern about the small sample sizes and the differences in sample size between experimental groups of individual experiments. This concern should be addressed by increasing the sample size and making the sample sets being compared similar in size.

We have revised the manuscript to better communicate the number of replicate experiments and the number of measurements for each experiment. We have also added new data to increase our sample sizes where requested, especially to address the above Editorial points 1, 2, and 3. The two most important things to note are that none of the data changed significance upon increasing n’s and we have, in most places, doubled the quantity of data by adding new experiments. Below we detail the sample sizes for the specified data.

Figure 1 – unchanged, but new RNAseq data set

i. Revised reporting of Figure 1—figure supplement 1 shows measurements for 3 replicate experiments with 20, 20, 20 nuclei each.

ii. RNAseq data supplements for parental vs. HP1α-AID-sfGFP modified (new addition), HP1α-AID-sfGFP untreated vs. auxin 4 hours, and HP1α-AID-sfGFP untreated vs. auxin 16 hours

Figure 2 – revised sample size reporting for clarity and provided new related data within this response

i. Panels F (protein levels) and G (nuclear shape solidity) updated in figure legend: “3 experiments each (shown as black dots), each consisting of n = 109, 102, 105 control ; n = 137, 115, 165 auxin, n = 31, 34, 32 methylstat, and n = 102, 92, 78 auxin methylstat”

ii. New supporting data in the reply (see above Editor point 2)- Nuclear shape solidity was measured again in new experiments and we observed the same outcome; please see data above in response to Editor point 2 D: “3 experiments each (shown as dots) each consisting of n = 96, 96, 97 control ; n = 109, 96, 115 auxin, n = 47, 48, 49 methylstat, and n = 87,101,79 auxin methylstat. Auxin is statistically significantly different p < 0.05 while the rest are statistically similar. While these absolute measurements of solidity are less than in the original manuscript, % abnormally shaped nuclei (solidity < 0.96) increases similarly by ~10%.”

iii. New data Figure 2—figure supplement 1 – “Normalized to parental control, immunofluorescence signal of H3K9me2,3 for parental and HP1α-AID-sfGFP modified cell lines without or with treatment of auxin for 16 hours and/or methylstat for 48 hours. Three experimental replicates parental -/- n = 216, 128, 229; +/- n = 307, 324, 215; -/+ n = 188, 115, 155; +/+ n = 184, 258, 284; HP1α-AID-sfGFP -/- n = 108, 101, 104; +/- n = 136, 114, 164; -/+ n = 30, 33, 31; +/+ n = 102, 91, 77. (E) Normalized immunofluorescence for H3K9me2,3 and H3K9ac in HP1α-AID-sfGFP – or + auxin for 4 days (96 hours) for six experimental replicates – auxin (218, 246, 263, 186, 238, 265) and + auxin (188, 184, 174, 208, 199, 187). P values reported as no asterisk > 0.05, * < 0.05, **< 0.01, ***< 0.001.”

Figure 3 – New data to increase sample size in all cases (5 new experimental replicates were added to the 3 original experimental replicates)

i. Nuclear curvature measurements: “Eight experimental replicates were measured for each condition (denoted as black dots) consisting of – auxin, n = 37, 45, 45, 57, 58, 57, 55, 54; + auxin, n = 60, 44, 41, 60, 60, 60, 60, 60 ; + auxin and WT exogenous rescue, n = 27, 30, 36, 60, 60, 60, 60, 19; + auxin and I165E exogenous rescue, n = 38, 40, 50, 59, 56, 58, 58, 51. P values reported as n.s. > 0.05, * < 0.05, **< 0.01, ***< 0.001, ****<0.0001.”

ii. New data in the reply – parental nuclear curvature measurements are similar (- or + auxin, 5 replicate experiments) to untreated HP1α-AID-sfGFP modified. See Reviewer 2 major point 1. Also added to the revised manuscript in the main text.

Figure 4

i. Additional simulations have been performed to improve the statistics. We have considered Reviewer 2’s comment in reporting the number of simulations (further addressed below) and clarified the wording.

ii. New experimental data on the fraction and enrichment of H3K9me2,3 and HP1α at the nuclear periphery: “One experiment measuring – auxin n = 93 and + auxin n = 149.”

Figure 5 – new data to increase sample sizes in all cases

i. Panels B (measured HP1α in isolated chromosome bundles) and E (micromanipulation force measurements). In general, we nearly doubled sample size by adding 6-8 measurements for each : “For B-E n = 20 for control and auxin treated, n = 16 for methylstat and n = 14 auxin methylstat treated.”

ii. Panel F abnormal metaphase and ana/telophase events, of which we add one new experiment for control and 16 hours auxin while we added a new condition – 4 hours auxin – with three new experimental replicates : “Metaphase misalignment 3-4 experiments (black dots) consisting of n = 16, 15, 20, 37 -aux, n = 33, 33, 24 +aux 4 hours, n = 22, 48, 58, 54 +aux 16 hours. Anaphase missegregation 3-4 experiments (black dots) consisting of n = 29, 23, 30, 30 -aux, n = 32, 29, 18 +aux 4 hours, n = 20, 35, 36, 45 +aux 16 hours.”

iii. All new data, Panel G abnormal mitoses and nuclear curvature for sister nuclei. “(G) HP1αAID-sfGFP cells – auxin or + auxin for 24 hours were tracked through mitosis to determine if abnormal mitosis results in abnormally shaped daughter nuclei measured via nuclear curvature (parental 34 nuclei from 17 mitoses; – auxin 46 nuclei from 23 mitoses; + auxin 51 nuclei from 26 mitoses, p = 0.00001). Percentage of abnormal mitosis presented in Figure 5—figure supplement 1D.”

Reviewer #1:

Strom and co-authors describe a new, very intriguing role for HP1alpha in chromatin mechanics and nuclear shape maintenance which is independent of known nuclear mechanoregulators: histone methylation or nuclear lamina, and which is instead driven by HP1alpha-mediated crosslinking of chromatin. While conceptually exciting, the central conclusions lack definitive experimental evidence. Further, key measurements lack true statistical power and the polymer model predictions are not stringently challenged. Finally, the functional relevance of this HP1-mediated mechanical regulation remains mostly unexplored, as the chromatin mis-segregation phenotype is only superficially analyzed and its connection with chromatin mechanics is elusive.

1. Exclusion of contribution of chromatin methylation and lack of changes in chromatin compaction are based on a single chromatin mark (H3K9me2,3) and one example image of H2B-GFP, respectively. To truly understand the consequence of HP1alpha loss they authors would have preferentially performed chromatin accessibility analyses such as ATAC seq and/or super resolution microscopy.

The Reviewer raises concerns over (1) the use of one methylated histone mark and (2) the use of H2B and (3) they ask for further data to support the result that chromatin compaction does not change. In the original manuscript we look at (1) H3K9me2,3 because it is the histone mark that HP1α binds and is thus the most relevant and (2) H2B because it is widely used to measure chromatin density due to its prevalence throughout chromatin. (3) We have revised the manuscript to remove any mention of changes in global chromatin compaction, which was not measured and is not necessary for the main conclusions of the paper.

1. The conclusions we make in the manuscript are supported by the data provided. In the original manuscript, we assayed changes in the most relevant constitutive heterochromatin maker, H3K9me2,3, which HP1α binds, and we found no change upon HP1α degradation via auxin treatment for 4 hours (Figure 2, E and F). In the revised manuscript, we include new data that auxin-induced degradation of HP1α does not change H3K9me2,3 histone methylation over 16 hours or 96 hours (Figure 2—figure supplement 1, E and E). The 4 day (96 hour) auxin-induced degradation also shows no change in euchromatin marker H3K9ac.

2. The Reviewer is incorrect that “one example image of H2B-GFP” is used to conclude that there is no change in chromatin compaction. In the original manuscript, we measured colocalization of dense H2B-miRFP and HP1α-sfGFP-AID, which showed that histonedense foci remained upon degradation of HP1α in three experimental replicates of 20 nuclei each (60 total). We have clarified these points in the revised manuscript, specifically stating that local chromatin compaction is not altered upon HP1α loss.

3. The Reviewer suggested that chromatin compaction and accessibility could be further investigated through ATAC-seq or super-resolution. Indeed, these would be interesting experiments, but they would also represent significant undertakings that would not directly support the central conclusions of this manuscript, which primarily concern the architectural role of HP1α in nuclear organization and function. We have revised the manuscript to remove any statement suggesting global changes in chromatin compaction, as we specifically measure (1) relevant histone methylation and (2) local chromatin foci compaction.

To further measure heterochromatin H3K9me2,3 we have added new data and analysis of enrichment at the periphery of the nucleus in 93-149 cells. This analysis shows that HP1α degradation does not change H3K9me2,3 heterochromatin enrichment at the periphery or DNA partitioning at the periphery (New Data Figure 4 E and F). Taken together, our data reveals that on the measured timescales, HP1α has no significant role in changing the level of its binding target H3K9me2,3 methylation or the euchromatic marker H3K9ac, or the enrichment of its binding target H3K9me2,3 at the periphery, or local compaction of chromatin.

2. Similarly, conclusions of HP1alpha being essential for nuclear shape maintenance (Figure 3) were made based on just 25-30 nuclei – given the very elegant and efficient Auxin degron system and simple 2D model system, it is unclear why so few cells were analyzed. In additional, instead of expressing shape abnormalities by a descriptive classification (% of cells with abnormal nuclei) the authors should include a panel of detailed parameters to precisely quantify measure cell shape, such as solidity and the EFC ratio of nuclear shape (which some of the co-authors established in a recent publication).

The Reviewer requested (i) more data and (ii) specific measurements. (i) First, we have more than doubled the data for Figure 3 (nuclear curvature) with 5 new experimental replicates. In addition, the Figure 2 measurements were completely recapitulated in new experiments, as shown in Editor point 2D above. (ii) Figure 3B and Figure 2—figure supplement 2 report nuclear curvature, and Figure 2 measures nuclear shape solidity. We have revised the manuscript to include average nuclear shape solidity reported in the text, which supports the major conclusions of the manuscript.

The original manuscript's reporting of the number of nuclei was confusing to multiple reviewers. While it may have been unclear, we measured > 25 nuclei for each of 3 experimental replicates for nuclear curvature. The fewest nuclei were measured in the original Figure 3, which still reported measurements of 83 nuclei (27, 30, 26 nuclei for each replicate). In the revised manuscript we have significantly increased sample size for nuclear curvature measurements (now Figure 3 has 8 experimental replicates and Figure 5G measures nuclear curvature post normal and abnormal mitosis). We have revised the manuscript to more clearly indicate the total number of cells measured for each replicated.

Below we share how we now clarify and fully report n for all nuclear shape measurements in the revised manuscript to address (i) quantity of data and (ii) how we quantify the data. The revised manuscript now clearly communicates the quantitative measurements we used to determine maintenance or loss of nuclear shape, instead of using phenotypic wording. Specifically, we have revised the manuscript and figures to replace “abnormal/irregular” with specific measurements of (A) nuclear curvature, (B) individual nucleus curvatures over time, (C) and solidity.

A. New data has been added to Figure 3 which measures nuclear curvature. The revised manuscript now includes 8 biological replicates (3 original and 5 newly added) and states:

“Eight experimental replicates were measured for each condition (denoted as black dots) consisting – auxin, n = 37, 45, 45,57, 58, 57, 55, 54; + auxin, n = 60, 44, 41, 60, 60, 60, 60, 60 ; + auxin and WT exogenous rescue, n = 27, 30, 36, 60, 60, 60, 60, 19; + auxin and I165E exogenous rescue, n = 38, 40, 50, 59, 56, 58, 58, 51. P values reported as n.s. > 0.05, * < 0.05, **< 0.01, ***< 0.001, ****<0.0001.”

B. Figure 2—figure supplement 2 tracks nuclear curvature for 42 nuclei every hour over 12 hours for each untreated control and auxin-induced HP1α degradation. We would like to point out that this is a rigorous quantitative measurement of nuclear shape, and it is taken during HP1α-AID-sfGFP degradation and entirely during interphase (independent of mitosis). This data strongly supports the data in Figure 3 and provides a measure of time. This time measurement, along with the population average of nuclear curvature, shows a significant increase at the time point 5 hours after auxin addition, which coincides clearly with the time interval of HP1α-AID-sfGFP degradation (Figure 1 and 2).

C. New data panel Figure 5G measures nuclear curvature after mitosis for parental, HP1α- AID-sfGFP without auxin, and HP1α- AID-sfGFP with auxin. This data further confirms that sister nuclei after mitosis have higher nuclear curvature upon HP1α degradation.

D. Figure 2 solidity < 0.96 is used to determine abnormally shaped nuclei. The revised manuscript includes clear statements of number of measurements and now includes the average solidity that follows the same trends of significance: “ 3 experiments each (shown as black dots) each consisting of n = 109, 102,105 control ; n = 137, 115, 165 auxin, n = 31, 34, 32 methylstat, and n = 102, 92, 78 auxin methylstat. Average measurements were similar for control, methystat, and auxin with methylstat 0.971+/- 0.0001 but different for auxin 0.969+/-0.0015, p=0.005.”

We would also like to point out that we redid the solidity measurements and observed a similar decrease in average solidity with auxin-induced HP1α degradation relative to the other conditions as well as increase in number of nuclei with solidity less than 0.96 (see Editorial point 2D above).

Overall, we have provided new data and clarified the reporting of our quantitative measurements. Taken together, this data clearly shows loss of nuclear morphology upon degradation of HP1α that can be rescued by increased histone methylation (methylstat) or exogenous HP1α rescue, but not the dimer mutant HP1α I165E.

3. Evidence for chromatin-chromatin crosslinks being affected in the I165E mutants should be experimentally demonstrated for example using super resolution

The Reviewer requested new validation of the loss of chromatin-chromatin crosslinking via dimerization disruption in the HP1α I165E mutant. The HP1α I165E mutant has already been reported to lose its ability to dimerize; indeed single point mutations to the chromoshadow domain can disrupt dimerization (Brasher et al. EMBO J 2000, Lechner et al. Mol. Cell Biol. 2000, Thiru et al. EMBO J 2004, Lechner et al. Biochem Biophys Res Commun 2005). Additionally, human HP1α promotes intra- and inter-strand crosslinking of nucleosome arrays in vitro, and this ability is abolished by disrupting dimerization (Azzaz et al. 2014 J Bio Chem). Moreover, this mutant is widely used in the literature and shown by Number and Brightness analysis to exist as a monomer in living mammalian cells (Hinde et al., Sci Reports 2015; Machida et al., Cell 2018). Because this mutation has already been well established to disrupt dimerization of HP1α by other publications, we believe further validation is an unnecessary technical challenge. Super-resolution light microscopy techniques do not have a high enough resolution to directly visualize chromatin-chromatin crosslinks or their subsequent loss upon I165E dimerization. However, confirmation of HP1α as a di-nucleosome crosslinker has been accomplished with single particle electron microscopy (Machida et al., 2018 Cell), and recent methodological developments have allowed visualization of chromatin regions and found increased mesh density in heterochromatic as compared to euchromatic areas (Ou et al. Science 2017).

4. Mechanical measurements should be validated with an independent, higher throughput mechanical manipulation method, such as AFM or imaging-based methods

The Reviewer requests new validation of the force measurements. We note that micromanipulation is a well established technique for measuring the mechanical properties of both mitotic chromosomes and cell nuclei. Micromanipulation of mitotic chromosomes via micropipettes has been used to directly probe chromosome elasticity for over two decades (Houchmandzadeh et al. J Cell Biol 1997, Poirier et al. Mol Biol Cell 2000). Whole-nucleus micromanipulation was recently established by some of us (Stephens et al. Mol Biol Cell 2017), and the measurements were further validated by subsequent experiments, in part, by one of us (Hobson et al. Mol Biol Cell 2020) and another group (Shimamoto et al. Mol Biol Cell 2017). Therefore, we do not believe an independent validation is necessary.

Furthermore, we note a secondary, validating technique is extraneous for two additional reasons: (A) Micromanipulation force measurements provide unique insights into the identified two regimes of nuclear mechanical response, which are central to this paper and which standard AFM cannot provide; and (B) Micromanipulation force measurements have several internal experimental controls, which we describe below. This further bolsters confidence in the measurements included in the original manuscript, which show a clear statistically significant change in the chromatin dominated nuclear spring constant for HP1α degradation as compared to controls.

A. Micromanipulation has particular advantages over other techniques, such as AFM. In particular, standard AFM compression measurements of cell nuclei are incapable of separating the chromatin- and lamin-based force response regimes of the nucleus. In addition, standard AFM has to model force-compression as a Hertzian spring compressing an immovable substrate, and thus it lacks the ability to track both deformation and force accurately. In contrast, our micromanipulation extensional force measurement technique does not rely on the coarse estimations of AFM, but it instead allows unhindered, highly controlled, and fine tracking of extension vs. force. This allows micromanipulation to separate these mechanical regimes and observe that chromatin controls short extension nuclear force response while lamin A controls strain stiffening at longer extensions (> 3 μm, Stephens et al., 2017 MBoC). The technique and its novel findings were recently confirmed via a new more advanced Single-Plane Illumination Microscopy AFM technique (Hobson et al., 2020 MBoC). This first-of-its-kind combined SPIM AFM provides the needed fine measurements of deformation vs. force response necessary to separate the two force response regimes, further proving that standard AFM is incapable of these measurements. However, we do not do SPIM AFM because it has already been established to provide insights similar to those of micromanipulation, and it is technically demanding. The ability to measure and separate these two regimes is vital to the paper as we use it along with modeling to conclude that HP1α is a chromatin crosslinker more than a chromatin-lamin linker in Figure 4. Therefore, further validations of nuclear force response measurements are unnecessary.

For mitotic chromosome measurements, micromanipulation provides novel measurements of single isolated mitotic chromosome mechanics that have yet to be accomplished by other forms of force measurements.

B. The presented measurements demonstrate clear statistically significant differences between WT and HP1α-depleted cells. Our measurements show clear statistical significance change in the chromatin-dominated short extension nuclear spring constant for HP1α degradation in CRISPR HP1α-AID-sfGFP cells (Figure 2, untreated n=13 0.40±0.03 nN/μm vs. auxin n=18 0.22±0.03 nN/μm, t-test p = 0.0002). Oppositely, the strain stiffening lamin-A-dominated nuclear spring constant does not change (Figure 2—figure supplement 1, untreated 0.19±0.05 nN/μm vs. auxin 0.19±0.05 nN/μm, t-test p = 0.998). These measurements (one which changes and one that does not) are taken from the same nucleus force-extension measurements providing strong internal controls. The control parental cell line untreated vs auxin-treated shows an insignificant change (Figure 2 C, untreated n=8 0.35±0.04 nN/μm vs auxin n=11 0.34±0.02 nN/μm, t-test p = 0.72). Also parental U2OS vs. modified HP1α-AID-sfGFP untreated shows an insignificant change (parental untreated n=8 0.35±0.04 nN/μm vs. n=13 0.40±0.03 nN/μm, p = 0.42). Taken together, our data is both unique and rigorously shown to be statistically significant for changes in the chromatin-based nuclear force response regime due to HP1α degradation.

5. The validity of the polymer model that excludes a role for HP1-lamina or inner nuclear membrane interactions in nuclear mechanics is based solely on the short- vs long-extension stiffness data that the authors have previously validated for chromatin vs lamina-dependent mechanical regimes. The role of HP1 and lamina interactions through LBR or PRR14 should be experimentally excluded with mutants or LBR/PRR14 siRNA, in particular as these proteins have been shown to be important for nuclear shape maintenance.

The Reviewer questions the conclusions surrounding the polymer simulations model specific to ruling out HP1α-lamin interactions. The Reviewer is correct that the polymer model uses changes in short vs. long extension to determine whether HP1α degradation or loss of dimerization disrupt chromatin-chromatin or chromatin-lamin interactions. First of all, we would like to point out that the main purpose of the simulations is to investigate the possible mechanical roles of HP1α (as opposed to solely being descriptive of the experimental results). We have updated the manuscript to soften the conclusion drawn from the polymer simulations that vary chromatin-lamina linkages; indeed, the model alone does not rule out HP1α-lamina interactions. However, the model suggests that the changes in mechanical response due to loss of HP1α are accurately modeled as a loss of chromatin-chromatin crosslinkers from a crosslinked chromatin polymer gel. The conclusion of the model is that long-extension strain stiffening should be altered if HP1α has a significant role in the mechanical contribution of chromatin-lamina connections. The experimental data shows no change in the long-extension regime (see Figure 2—figure supplement 1A). Thus, our conclusions that HP1α and its dimer function are primarily functioning mechanically as a chromatin-chromatin crosslinker and not primarily as a chromatin-lamin linker are well supported.

Nonetheless, to address the Reviewer’s comment directly, we considered how HP1α might mediate chromatin-lamina interactions. Previously, PRR14 and LBR have both been shown to interact with HP1α, but neither of which has been shown to affect HP1α localization or behavior. Loss of PRR14 or LBR specifically disrupts peripheral localization of heterochromatin, which is measured by loss of the percentage of H3K9me2,3 chromatin at the periphery proximal to the lamina (Poleshko et al., 2013; Dunlevy et al., 2019 biorxiv; Solovei et al., 2013). If loss of HP1α disrupts chromatin-lamina linkages OR either of these proteins, it should phenocopy loss of PRR14 and LBR proteins, and we would expect peripheral heterochromatin enrichment to be disrupted.

We therefore measured localization of HP1α and its substrate heterochromatin near the nuclear periphery. Our new experiments added to the updated manuscript reveal no peripheral enrichment of HP1α and no change in peripheral H3K9me2,3 heterochromatin enrichment upon HP1α degradation (Figure 4, E and F). This data suggests that (1) HP1α is not enriched at the periphery like known chromatin-lamina tethers should be [0.86+/- 0.01 peripheral/total average signal], (2) HP1α degradation does not disrupt H3K9me2,3 peripheral enrichment [1.48 +/- 0.04 vs 1.34 +/- 0.03, p> 0.05], which suggests it is not a significant chromatin-lamina tether in this cell line, AND that its loss also does not disrupt LBR or PRR14 needed for maintenance of peripheral heterochromatin, and finally, (3) supports our previous assertion that HP1α’s main mechanical function is best recapitulated in simulations as a chromatin-chromatin crosslinker and not a chromatin-lamina tether.

6. Studies around the role of HP1alpha in mitotic segregation are not described clearly and the connection with chromosome mechanics remain unclear. The experiments are severely underpowered with only very few cells analyzed for mechanical measurements (just 8-14 nuclei) and 3-fold differences in sample size per condition (Figure 5G" "n=15-50 mitotic cells for each condition…") for metaphase misalignment measurements.

We have revised the text to include a clearer explanation of the connection between nuclear mechanics and mitotic segregation. In short, heterochromatin has been shown to have a mechanical role in nuclear mechanics, and, more recently, also a role in mitotic chromosome mechanics (Biggs et al., 2019 MBoC). We launched into similar questions of the differential contributions of HP1α and histone methylation in mitotic chromosome mechanics, with a similar approach as we performed on interphase nuclei. Furthermore, we find that HP1α has a functional role in mitosis, since HP1α degradation disrupts proper metaphase alignment and anaphase segregation. Thus, while a direct link between mechanics and mitotic function remains elusive, we report a close association between perturbations to HP1α-dependent mechanics and function.

Additionally, the Reviewer is concerned about low numbers of measurements. First, we again note that the style of reporting sample size (n) was confusing and we have revised the manuscript to share all the sample sizes of each replicate. To fully address this comment we have new data for both (A) single-chromosome micromanipulation force measurements as well as (B) metaphase and anaphase fidelity measurements. The new measurements further support the original data and conclusions of the paper. Furthermore, we have provided new experiments to quantify HP1αAID-sfGFP presence in isolated chromosome bundles; report HP1α-AID-sfGFP binding mitotic chromosomes in live cells at pericentromeric regions; and measure the effect of abnormal mitosis on interphase nuclear shape of resulting sister nuclei.

A. In the updated manuscript, we now include more single-chromosome micromanipulation force measurements along with measurements of HP1α-AID-sfGFP levels in isolated chromosome bundles and single chromosomes. We have added 6-8 new force measurements of each condition to increase our total micromanipulation chromosome force measurements to 14-20 (Figure 5E).

B. In the updated manuscript we provide new experimental measurements of metaphase alignment and anaphase segregation to bolster the number of events. The revised manuscript provides one more experiment for control (- auxin) and HP1α degradation (+ auxin 16 hours). Furthermore, we added three experiments for rapid degradation of HP1α via + auxin 4 hours, which show an increase of mitotic abnormalities in HP1α-depleted cells, but less than in 16 hours of degradation of HP1α. These data support the main conclusions that HP1α degradation results in mitotic errors, specifically metaphase misalignment and anaphase missegregation.

In the original manuscript, we had provided data from 3 experiments for each condition without or with auxin, where each experiment had at least 15 nuclei. We understand that the original manuscript wording was confusing concerning the true number of measurements. We have now clarified in the figure legends as to the number of experiments and list the number of measurements of each. Overall, this confusion underrepresented the number of measurements.

In the revised version, we have clearly written out all the numbers of cells measured. “Metaphase misalignment 3-4 experiments (black dots) consisting of n = 16, 15, 20, 37 -aux, n = 33, 33, 24 +aux 4 hours, n = 22, 48, 58, 54 +aux 16 hours. Anaphase missegregation 3-4 experiments (black dots) consisting of n = 29, 23, 30, 30 -aux, n = 32, 29, 18 +aux 4 hours, n = 20, 35, 36, 45 +aux 16 hours.” (Figure 5 legend)

Reviewer #2:

The manuscript from Strom, Biggs, and Banigan, et al. is an outstanding effort and an important contribution to understanding the complex roles chromatin and protein effectors play in the nucleus. It certainly rises to the level of publication, and I found it to be generally very exciting and compelling work. I have one major criticism of the work below, but would be happy to accept the paper for publication.

The major criticism I have of the work is the validation of the HP1 protein after addition of the degron and sfGFP. Especially, given that there are a number of findings in this work are seemingly contrary to expectations, i.e no change in gene expression or H3K9 maintenance. From my reading, the only validation in the manuscript that I see that adding the degron and GFP onto the C-terminus of HP1a (more than tripling the MW right next to the protein-protein interaction hub of HP1) does not affect its biology is a localization measurement. And while I realize that it is not easy to validate a structural protein, it would be nice to have some more corroboration or comparison to experiments that have investigated the formation of heterochromatin by HP1 proteins.

The Reviewer requests more data to support that HP1α-AID-sfGFP functions similarly to unmodified HP1α. We have addressed this through (A) RNA seq; (B) hetero/euchromatin levels; (C) nuclear force measurements (data from the original manuscript) and shape measurements; and (D) mitosis. Oppositely, auxin-induced HP1α degradation does result in changes to nuclear mechanics, shape, and mitotic fidelity.

A. In the revised manuscript, we include RNA seq data showing that there are no significant differences in transcription profiles between parental and AID-sfGFP modified cell line (data supplement). This results in 76 upregulated and 56 downregulated transcripts which represents a 0.8% change across >16,600 genes. GO analysis reveals no change to the nucleus or chromatin proteins suggesting that HP1α-AID-sfGFP does not cause abnormalities associated with chromatin. GO analysis of cellular function returns only extracellular changes for upregulated genes (matrix, region, space, exosome, and vesicle) and no significant enrichment for downregulated genes.

B. Heterochromatin (H3K9me2,3) levels were found to be similar between the parental and modified cell lines for immunofluorescence (Figure 2—figure supplement 1D).

C. In the original manuscript, the parental cell line has a nuclear spring constant similar to that of the modified cell line (see Figure 2, p > 0.1) further suggesting that chromatin structure is unaltered by the HP1α-AID-sfGFP modification. In agreement with these results, HP1α-AIDsfGFP untreated nuclei maintain normal nuclear shape that is not significantly different from parental nuclear shape (Parental 0.114 +/- 0.002 μm-1; HP1α-AID-sfGFP 0.118 +/- 0.001 μm-1, p > 0.05), until HP1α is degraded by auxin (HP1α-AID-sfGFP 0.136 +/- 0.008 μm-1, p < 0.0001). Additionally, auxin treatment of the parental line has no effect on nuclear shape in 8 trials (Parental – Auxin 0.114 +/- 0.002 μm-1; Parental + Auxin 0.117 +/- 0.001 μm-1, p > 0.05). Abnormal nuclear shape upon HP1α degradation is rescued by expression of an exogenous wildtype HP1α (see Figure 3 in main text). Taken together, modified HP1α-AID-sfGFP maintains chromatin-based nuclear mechanics and shape, similar to wild-type HP1α.

Author response image 3

D. In new experiments, we verify that parental mitosis fidelity is similar to the modified cell line (Figure 5G). Upon degradation of HP1α-AID-sfGFP mitotic failures increase.Altogether, these different experiments show that the nuclear functions that are disrupted by auxin-induced degradation of HP1α-AID-sfGFP are not perturbed by the presence of HP1α-AIDsfGFP in place of WT HP1α. This supports the functionality of the HP1α-AID-sfGFP protein construct.

For example, to reconcile these data with previous findings, one might predict that gene expression or H3K9 levels should eventually recapitulate previous knockdown/out studies if the auxin treatment were to persist, provided cell type and cell cycle considerations. While RNA seq experiments are fairly involved, I would think it possible to test the methylation levels in the auxin-HP1 line after a few days of depletion to see if levels do come down to previously reported levels, at least at specific satellite sequences. Alternatively, it might be easier to assay the AID-GFP-HP1 at exogenous sites as has been previously reported. Though, it is noted however that the dimerization mutant does perform as expected.

In addition to validation, it would also be helpful to more directly compare the timing of your experiments with those done elsewhere to give the reader a better idea of when changes (epigenetic, transcriptional) are likely to arise, etc. Though, discussion is somewhat confounded by the diversity of HP1 activity in different organisms. There is the distinction of >24 hours in the discussion, but there are several HP1 experiments that show faster changes than this, specifically compaction following recruitment and transcriptional repression at exogenous sites.

The Reviewer requests that as an additional validation, we do long-term loss studies in addition to short-term loss (4 hours of auxin treatment), which we focused on in the original manuscript. This is indeed an interesting question, and we have provided new data on this point. In the revised manuscript, we show that histone methylation levels are not changed at 16 hours of auxin treatment and 96 hours (Figure 2—figure supplement 1, D and E). Additionally, in the original manuscript, we provided the raw data showing lack of transcriptional change after 16 hours of auxin treatment. In the revised manuscript, we note the lack of change in transcription and have attached the RNAseq data as a data supplement. We were unable to do longer term loss experiments (>24 hours) for transcription and satellite analysis due to the technical challenges. Finally, we have revised the manuscript to further discuss how HP1α functions differently in different organisms (Li et al. 2003; Verschure et al. 2005; Fischer et al. 2009; Sadaie et al. 2008; Lee et al. 2019; Levine et al. 2015; James and Elgin 1986; Schwaiger et al. 2010; Singh et al. 1991; Wreggett et al. 1994)

Reviewer #3:

The authors use state-of-the-art genome engineering and degron tagging approaches combined with transcriptional profiling and sophisticated biophysical measurements. Most of the results are clear and convincing, but the section on mitotic chromosomes has substantial weaknesses and should either be removed or extended to substantiate the key conclusions. The main part on HP1alpha's role in interphase nuclei, however, is interesting and well supported by the data.

We appreciate that the Reviewer feels that the work on interphase nuclear mechanics and shape is “clear and convincing” and “interesting and well supported”. We have worked to revise the mitotic section (Figure 5) to rise to this level. We believe that our substantial amount of new experiments address these three major concerns raised by the Reviewer. We believe that the improved mitosis data will be equally interesting to the broad readership of eLife.

1. Localization of HP1alpha on mitotic chromosomes. The authors claim that HP1alpha remains bound to mitotic chromosomes, but this is not visible in the data shown in the paper. Prior studies have in fact clearly shown that HP1alpha completely dissociates from chromosome arms, while only a small fraction remains bound to pericentric chromatin (Hirota et al., 2005; Fischle et al., 2005; Serrano et al., 2009; Akram et al., 2018). The authors reference these papers to provide a rationale for studying HP1alpha's potential role in mechanical rigidity of chromosome arms, but they neglect the fact that the previous studies showed that HP1alpha only localizes to a very small region surrounding centromeres. This very limited enrichment of HP1alpha at pericentromeric regions is inconsistent with a direct major mechanical role of HP1alpha in stiffening chromosome arms as the authors suggest.

Figure 5A shows HP1alpha-GFP fluorescence throughout the entire cell with a potentially very slight enrichment in the central region. Without counterstaining DNA or chromatin, it is not possible to assess whether any HP1alpha remains bound to chromosomes, as chromosome positions are not known. The authors should provide high-resolution figures of HP1alpha-GFP counterstained with a DNA dye, ideally in live cells. They should further image chromosome spreads as in the prior work by Serrano et al., 2009 to clarify where on chromosomes HP1alpha binds. Figure 5C shows a fuzzy fluorescence signal for HP1alpha, but the image quality is insufficient to assess whether this is on chromosome arms or on pericentromeric regions. Moreover, without a control cell that does not express GFP-tagged HP1alpha, it is not possible to assess to which extent the fuzzy fluorescence signal shown in Figure 5C represents background autofluorescence. Alternatively, as the mitotic chromosome stiffness measurements are also not very convincing (see next point), the authors might consider removing the entire section on mitotic chromosomes.

The Reviewer requests data supporting the presence or absence of HP1α at both the centromere and on chromosome arms and how HP1α location relates to mechanics. New data show that HP1α is located throughout the chromosome and concentrated at the pericentromeric region. Specifically, to address this comment we have included: (A) Imaging of live ex vivo isolated mitotic chromosome bundles by widefield imaging that shows HP1α fluorescence as concentrated pericentromeric foci as well as a weaker signal on chromosome arms, see Figure 5 B, C, —figure supplement A and B; (B) Live-cell imaging of metaphase chromosomes by confocal, which recapitulates HP1α’s enriched localization to pericentromeric foci, whereas Hoechst-stained DNA is equally distributed, see Figure 5—figure supplement C; (C) Revisions to the text to clarify that HP1α is largely located at the centromere in our experiments, but can nonetheless contribute to whole-chromosome mechanics, and moreover, that micromanipulation force measurements are not solely a measure of chromosome arms.

A. We have revised the manuscript and Figure 5 to include several examples of ex vivo isolated mitotic chromosome bundles that clearly show HP1α-AID-sfGFP on the mitotic chromosomes. These images show dense foci, likely at the centromere, as well as a lesser, but still detectable, presence on the chromosome arms (Figure 5 B and C). We have also provided controls for autofluorescence and HP1α auxin-induced degradation (see Figure 5—figure supplement 1). We note in the revised manuscript that the cellular background is bright, possibly due to some amount of unbound HP1α, which could disrupt the ability to see low levels of HP1α on mitotic chromosome arms. By removing the chromosome bundle from the cell this decreases the background, making HP1α-AIDsfGFP visible on chromosomes throughout their arms. Thus, our ability to remove cellular background through chromosome bundle isolation is sufficient to reveal HP1α on the chromosome arms. We would also like to note a few points about the mentioned references:

– Hirota et al., 2005 imaging in cells – subject to background fluorescence

– Fischle et al., 2005 imaging in cells – subject to background fluorescence

– Serrano et al., 2009 chromosome spreads of HeLa cells show more than pericentromeric foci; they also show multiple foci across the chromosome (Figure 1). Thus, this work agrees with our findings that there is HP1α on the chromosome arms (although there is clearly less of it). We thank that Reviewer for pushing us to more carefully consider the literature and to do better fluorescence measurements. Specifically, the Reviewer suggested chromosome spreads. While we chose to use a different method, Serrano et al. provide initial data that chromosome spreads show HP1α at both the pericentromeric region and on the arms. This data is in agreement with our findings using chromosome bundle isolation.

– Akram et al., 2018 imaging in cells – subject to background fluorescence, though not all HP1α fluorescence appears to colocalize with pericentromeric labeling via ACA (see Figure 6).

– Overall these cited publications show that Aurora B facilitates HP1α removal from chromosome arms, suggesting that HP1α does have the capacity to bind to both the pericentromeric and chromosome arms. However, a measurable level of remaining HP1α may be obscured inside cells given strong cellular background fluorescence. In Figure 5 and Figure 5—figure supplement 1, A and B, we address the Reviewer’s worries about levels of background fluorescence and now show that isolated chromosome bundles have less background fluorescence, allowing us to better visualize and measure HP1α presence and localization.

B. Live-cell imaging using confocal microscopy shows HP1α-AID-sfGFP as foci corresponding to centromeres relative to DNA staining of whole chromosomes (Figure 5—figure supplement 1C). This data is in agreement with cited papers in the original manuscript as well as the papers referenced by the Reviewer.

C. We have clarified in the manuscript that micromanipulation force measurements of an isolated mitotic chromosome include the whole chromosome, arms and the centromere, and does not solely measure chromosome arm mechanics. The revised manuscript now states: “The resistive force measured includes both HP1α on the arms and at the pericentromere since tension is distributed across the whole chromosome from the pipettes holding opposite ends of the chromosome (Figure 5D example images).” Additionally, recently published data of in vitro work with HP1α demonstrates that in an isolated system, accumulation of HP1α in small puncta on DNA is sufficient to contribute to the mechanics of the entire DNA fiber (Keneen et al., eLife 2021). Thus, even beyond our finding of HP1α on arms, this suggests that accumulation of HP1α at pericentromeric sequences could contribute to mechanics of the entire chromosome.

We have novel evidence for HP1α presence on chromosome arms via chromosome bundle isolation and imaging. Furthermore, our data agrees with many other publications that HP1α remains at the pericentromere during mitosis. Micromanipulation of an isolated chromosome stretches the chromosome from end-to-end while measuring force vs. extension throughout the chromosome from one arm through the centromere to the other arm. Thus, HP1α located in the arm as well as the pericentromere would allow it to resist stretching in our micromanipulation force measurement experiments. Previous studies imaging condensin along the chromosome during micromanipulation stretching provide preliminary data that the centromere stretches less than chromosome arms (Sun et al., Chromosome Res 2018). Thus, it may be possible that the changes in mechanics are dependent on HP1α’s mechanical role at the centromere. However, our newly added data of HP1α-AID-sfGFP chromosome bundle imaging (see point A above) supports that HP1α is present at chromosome arms as well. Further studies will be needed to untangle the mechanical and structural role of the chromosome arms vs. the pericentromere, but our novel data provides new and exciting insights.

2. HP1alpha's contribution to mechanical rigidity of mitotic chromosomes. The authors claim that HP1alpha depletion "strongly impairs" mitotic chromosome mechanics. Their microneedle-based force extension experiments shown in Figure 5E, however, show a large variability between individual measurements but only a very slight difference between HP1alpha-AID-sfGFP cells treated with auxin compared to untreated cells. There is a star that might indicate statistical significance (this is not explained in the legend), but even if that were true then the effect size would still be so small that biological relevance is questionable. Given the concerns about HP1alpha localization, the overall evidence supporting a "major role" of HP1alpha in the stiffness of mitotic chromosomes is insufficient to support this conclusion. The authors might either perform additional experiments to substantiate the conclusions or remove the part on mitotic chromosomes.

The Reviewer requests clarification and more data to support the major conclusion that HP1α is a mechanical component of mitotic chromosomes. To directly address this comment, we have revised the text to use more measured language, specifically removing “major” and instead saying “a mechanical component of the mitotic chromosome”. Furthermore, as requested by the Reviewer, we have provided 6-8 additional force measurements to increase the total number measurements to 14-20 for each scenario. These new measurements provide more data to support the conclusion that HP1α is a mechanical component of mitotic chromosomes as auxin-degradation decreases the doubling force from 262 +/- 50 to 148 +/- 12 pN which is statistically significant (n = 20 each, p = 0.03) and represents an effect of size of ~40%. We have also revised the figure legends to clearly indicate “P values reported as * < 0.05, **< 0.01, ***< 0.001.”

The Reviewer pointed out that the force measurements show large variation between individual chromosome measurements. Variability likely stems from the inability to select a specific chromosome (random selection of 1 of 23 chromosomes), isolation of the chromosome, holding of the chromosome, and inherent variability of the mechanical measurement technique. Recent advancements in the technique have aided isolation efforts of a single chromosome by using a third micropipette to hold the bundle of chromosomes while the other two hold the ends of the single chromosome, allowing the ends to be pulled away from the bundle together. This technique aids maintaining the single chromosomes’s size and shape while it is pulled away from the bundle. Furthermore, we use doubling force to account for variability in which chromosome is grabbed from the bundle and where precisely the chromosome is grabbed at either end. This mitages issues of how length (and thickness) of the chromosome can affect strength, since a longer spring is weaker than a shorter spring given a set extension, and normalizes for these differences.

Overall, we believe our data is consistent, which is further aided by newly added data of 6-8 more measurements that ultimately supports that loss of HP1α results in a weaker mitotic chromosome. Furthermore, the new data help support that increased histone methylation levels via methylstat are dominant in chromosome mechanical stiffness as compared to HP1α since methylstat + auxin treatment does not significantly differ from methylstat treatment alone (auxin and methylstat 452 ± 116 pN vs. methylstat 745 ± 164 pN, p = 0.17).

3. HP1alpha's role in regulating nuclear shape. The authors suggest that the softening of interphase nuclei resulting from degradation of HP1alpha directly causes aberrant nuclear morphologies. However, they also show that depletion of HP1alpha leads to chromosome missegregation, which is known to lead to aberrant nuclear morphologies owing to perturbed nuclear assembly during mitotic exit. To assess to which extent the nuclear morphology defects in HP1alpha depleted cells are caused directly via nuclear softening or indirectly via chromosome missegregation, the authors should separately analyze nuclear morphology shape changes as shown in Supplementary Figure 2 for those cells that have passed through mitosis after HP1alpha degradation versus those cells that remained in interphase during the entire imaging duration.

The Reviewer requests experiments to address how abnormal mitosis affects nuclear shape in interphase. We have original manuscript data as well as new experiments to clearly address this important question.

First, we would first like to point out that in the original manuscript, increased nuclear curvature in interphase was shown to proceed coincident with HP1α degradation and independent of mitosis (see Figure 2—figure supplement 2). This original data addresses the issue of whether cells with abnormal nuclei have passed through mitosis “versus those cells that remained in interphase during the entire imaging duration”. Cells and their nuclei were tracked over 12 hours in interphase for nuclear curvature in auxin -/+ treated. This tracking shows loss of nuclear shape, measured as increased nuclear curvature over the entire tracked population (panel E), coincident with loss of HP1α-AID-sfGFP fluorescence signal at around 4-6 hours. Histograms also reveal the major shift in individual nucleus curvatures occurs within the 4-hour time window (panels D). This is the same time interval in which we measure nuclear mechanical softening (Figure 2D).

To directly address this Reviewer's comment we provide new data in the revised manuscript tracking cells through mitosis and into interphase after 24-hour pre-treatment in auxin. Our new experiments reveal that upon auxin-induced HP1α degradation both normal and abnormal mitosis result in daughter cells that have increased nuclear curvature (Figure 5G). In cells with degraded HP1α, abnormal mitoses are much more common, as we reported earlier, and these abnormal mitoses also lead to mostly abnormally shaped daughter nuclei. However, even cells that go through normal mitoses have higher average curvature than those in the “- Auxin” case, suggesting that abnormal mitosis is not required for abnormal daughter cell shape. Taken together, this data supports that rapid degradation of HP1α results in abnormal nuclear morphology (high nuclear curvature) both during interphase independent of mitosis AND over longer time intervals (16 hours) sister nuclei post mitosis independent of normal vs abnormal mitosis.

https://doi.org/10.7554/eLife.63972.sa2

Article and author information

Author details

  1. Amy R Strom

    Howard Hughes Medical Institute, Department of Chemical and Biological Engineering, Princeton University, Princeton, United States
    Contribution
    Conceptualization, Formal analysis, Validation, Investigation, Visualization, Methodology, Writing - original draft, Writing - review and editing
    Contributed equally with
    Ronald J Biggs and Edward J Banigan
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-1674-3242
  2. Ronald J Biggs

    Department of Molecular Biosciences, Northwestern University, Evanston, United States
    Contribution
    Conceptualization, Formal analysis, Investigation, Visualization, Methodology, Writing - review and editing
    Contributed equally with
    Amy R Strom and Edward J Banigan
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-9965-6346
  3. Edward J Banigan

    Institute for Medical Engineering and Science and Department of Physics, Massachusetts Institute of Technology, Cambridge, United States
    Contribution
    Conceptualization, Formal analysis, Investigation, Methodology, Writing - original draft, Writing - review and editing
    Contributed equally with
    Amy R Strom and Ronald J Biggs
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-5478-7425
  4. Xiaotao Wang

    Department of Biochemistry and Molecular Genetics, Feinberg School of Medicine, Northwestern University, Chicago, United States
    Contribution
    Data curation, Formal analysis, Investigation
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-3531-2157
  5. Katherine Chiu

    Biology Department, University of Massachusetts Amherst, Amherst, United States
    Contribution
    Formal analysis, Investigation, Visualization
    Competing interests
    No competing interests declared
  6. Cameron Herman

    Department of Molecular Biosciences, Northwestern University, Evanston, United States
    Contribution
    Formal analysis, Validation, Visualization
    Competing interests
    No competing interests declared
  7. Jimena Collado

    Department of Molecular Biosciences, Northwestern University, Evanston, United States
    Contribution
    Formal analysis, Validation, Visualization
    Competing interests
    No competing interests declared
  8. Feng Yue

    1. Department of Biochemistry and Molecular Genetics, Feinberg School of Medicine, Northwestern University, Chicago, United States
    2. Robert H. Lurie Comprehensive Cancer Center, Feinberg School of Medicine, Northwestern University, Chicago, United States
    Contribution
    Supervision, Funding acquisition
    Competing interests
    No competing interests declared
  9. Joan C Ritland Politz

    The Fred Hutchinson Cancer Research Center, Seattle, United States
    Contribution
    Conceptualization, Resources, Data curation, Formal analysis, Validation, Methodology, Writing - review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-5229-0087
  10. Leah J Tait

    The Fred Hutchinson Cancer Research Center, Seattle, United States
    Contribution
    Resources, Validation, Investigation
    Competing interests
    No competing interests declared
  11. David Scalzo

    The Fred Hutchinson Cancer Research Center, Seattle, United States
    Contribution
    Resources, Data curation, Validation
    Competing interests
    No competing interests declared
  12. Agnes Telling

    The Fred Hutchinson Cancer Research Center, Seattle, United States
    Contribution
    Resources, Validation, Investigation
    Competing interests
    No competing interests declared
  13. Mark Groudine

    The Fred Hutchinson Cancer Research Center, Seattle, United States
    Contribution
    Supervision, Funding acquisition
    Competing interests
    No competing interests declared
  14. Clifford P Brangwynne

    Howard Hughes Medical Institute, Department of Chemical and Biological Engineering, Princeton University, Princeton, United States
    Contribution
    Supervision, Funding acquisition, Writing - review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-1350-9960
  15. John F Marko

    1. Department of Molecular Biosciences, Northwestern University, Evanston, United States
    2. Department of Physics and Astronomy, Northwestern University, Evanston, United States
    Contribution
    Supervision, Funding acquisition, Writing - review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-4151-9530
  16. Andrew D Stephens

    Biology Department, University of Massachusetts Amherst, Amherst, United States
    Contribution
    Conceptualization, Data curation, Formal analysis, Supervision, Funding acquisition, Validation, Investigation, Visualization, Methodology, Writing - original draft, Project administration, Writing - review and editing
    For correspondence
    Andrew.stephens@umass.edu
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-5474-7845

Funding

Mark Foundation For Cancer Research (Life science research foundation Postdoctoral Fellowship)

  • Amy R Strom

Mark Foundation For Cancer Research (AWD1006303)

  • Amy R Strom

National Institutes of Health (U01 DA040601)

  • Clifford P Brangwynne

National Institutes of Health (GM114190)

  • Edward J Banigan

National Institutes of Health (U54DK107980)

  • John F Marko

National Institutes of Health (U54CA193419)

  • John F Marko

National Institutes of Health (R24DK106766)

  • Feng Yue

National Institutes of Health (1R35GM124820)

  • Feng Yue

National Institutes of Health (R01HG009906)

  • Feng Yue

National Institutes of Health (U01CA200060)

  • Feng Yue

National Institutes of Health (U01DA040583)

  • Mark Groudine

National Institutes of Health (1UM1HG011536)

  • John F Marko
  • Andrew D Stephens

National Institutes of Health (R00GM123195)

  • Andrew D Stephens

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

This work was supported by the NIH Center for 3D Structure and Physics of the Genome of the 4DN Consortium (U54DK107980 and 1UM1HG011536) and the NIH Physical Sciences-Oncology Center (U54CA193419). CPB and ARS were supported by the Howard Hughes Medical Institute, and grants from the NIH 4D Nucleome Program (U01 DA040601); ARS is supported by the LSRF Fellowship from Mark Foundation For Cancer Research. We thank Daniel S.W. Lee for experimental discussion and support, and Yiche Chang for the generous gift of pHR-HP1α-mCherry plasmids. We thank Daniel Shams for helping write a custom script for analyzing mitotic chromosome micromanipulation force measurements. EJB was supported by the NIH Center for 3D Structure and Physics of the Genome of the 4DN Consortium (U54DK107980), the NIH Physical Sciences-Oncology Center (U54CA193419), and NIH grant GM114190. XW and FY are supported by 1R35GM124820, R01HG009906, U01CA200060 and R24DK106766. JP, DS, LT, AT, and MG were funded by 4DN (U01DA040583). KC and ADS are supported by the Pathway to Independence Award (R00GM123195) and 4D Nucleome two center grant (1UM1HG011536).

Senior Editor

  1. Kevin Struhl, Harvard Medical School, United States

Reviewing Editor

  1. Geeta J Narlikar, University of California, San Francisco, United States

Reviewer

  1. Sy Redding, University of California, San Francisco, United States

Publication history

  1. Received: October 12, 2020
  2. Accepted: June 8, 2021
  3. Accepted Manuscript published: June 9, 2021 (version 1)
  4. Version of Record published: June 25, 2021 (version 2)

Copyright

© 2021, Strom et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 2,227
    Page views
  • 380
    Downloads
  • 1
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)

Further reading

    1. Cell Biology
    Lisa M Strong et al.
    Research Article Updated

    Autophagy is a cellular process that degrades cytoplasmic cargo by engulfing it in a double-membrane vesicle, known as the autophagosome, and delivering it to the lysosome. The ATG12–5–16L1 complex is responsible for conjugating members of the ubiquitin-like ATG8 protein family to phosphatidylethanolamine in the growing autophagosomal membrane, known as the phagophore. ATG12–5–16L1 is recruited to the phagophore by a subset of the phosphatidylinositol 3-phosphate-binding seven-bladedß -propeller WIPI proteins. We determined the crystal structure of WIPI2d in complex with the WIPI2 interacting region (W2IR) of ATG16L1 comprising residues 207–230 at 1.85 Å resolution. The structure shows that the ATG16L1 W2IR adopts an alpha helical conformation and binds in an electropositive and hydrophobic groove between WIPI2 ß-propeller blades 2 and 3. Mutation of residues at the interface reduces or blocks the recruitment of ATG12–5–16 L1 and the conjugation of the ATG8 protein LC3B to synthetic membranes. Interface mutants show a decrease in starvation-induced autophagy. Comparisons across the four human WIPIs suggest that WIPI1 and 2 belong to a W2IR-binding subclass responsible for localizing ATG12–5–16 L1 and driving ATG8 lipidation, whilst WIPI3 and 4 belong to a second W34IR-binding subclass responsible for localizing ATG2, and so directing lipid supply to the nascent phagophore. The structure provides a framework for understanding the regulatory node connecting two central events in autophagy initiation, the action of the autophagic PI 3-kinase complex on the one hand and ATG8 lipidation on the other.

    1. Cell Biology
    Laura Le Pelletier et al.
    Research Article

    Aging is associated with central fat redistribution and insulin resistance. To identify age-related adipose features, we evaluated the senescence and adipogenic potential of adipose-derived-stromal cells (ASCs) from abdominal subcutaneous fat obtained from healthy normal-weight young (<25y) or older women (>60y). Increased cell passages of young-donor ASCs (in vitro aging), resulted in senescence but not oxidative stress. ASC-derived adipocytes presented impaired adipogenesis but no early mitochondrial dysfunction. Conversely, aged-donor ASCs at early passages displayed oxidative stress and mild senescence. ASC-derived adipocytes exhibited oxidative stress, and early mitochondrial dysfunction but adipogenesis was preserved. In vitro aging of aged-donor ASCs resulted in further increased senescence, mitochondrial dysfunction, oxidative stress and severe adipocyte dysfunction. When in vitro aged young-donor ASCs were treated with metformin, no alteration was alleviated. Conversely, metformin treatment of aged-donor ASCs decreased oxidative stress and mitochondrial dysfunction resulting in decreased senescence. Metformin's prevention of oxidative stress and of the resulting senescence improved the cells' adipogenic capacity and insulin sensitivity. This effect was mediated by the activation of AMP-activated-protein-kinase as revealed by its specific inhibition and activation. Overall, aging ASC-derived adipocytes presented impaired adipogenesis and insulin sensitivity. Targeting stress-induced senescence of ASCs with metformin may improve age-related adipose tissue dysfunction.