1. Cell Biology
  2. Developmental Biology and Stem Cells
Download icon

HMMR acts in the PLK1-dependent spindle positioning pathway and supports neural development

  1. Marisa Connell
  2. Helen Chen
  3. Jihong Jiang
  4. Chia-Wei Kuan
  5. Abbas Fotovati
  6. Tony LH Chu
  7. Zhengcheng He
  8. Tess C Lengyell
  9. Huaibiao Li
  10. Torsten Kroll
  11. Amanda M Li
  12. Daniel Goldowitz
  13. Lucien Frappart
  14. Aspasia Ploubidou
  15. Millan S Patel
  16. Linda M Pilarski
  17. Elizabeth M Simpson
  18. Philipp F Lange
  19. Douglas W Allan
  20. Christopher A Maxwell Is a corresponding author
  1. University of British Columbia, Canada
  2. Leibniz Institute on Aging—Fritz Lipmann Institute, Germany
  3. University of Alberta, Canada
  4. BC Children’s Hospital, Canada
  5. Life Sciences Centre, University of British Columbia, Canada
Research Article
Cited
0
Views
623
Comments
0
Cite as: eLife 2017;6:e28672 doi: 10.7554/eLife.28672

Abstract

Oriented cell division is one mechanism progenitor cells use during development and to maintain tissue homeostasis. Common to most cell types is the asymmetric establishment and regulation of cortical NuMA-dynein complexes that position the mitotic spindle. Here, we discover that HMMR acts at centrosomes in a PLK1-dependent pathway that locates active Ran and modulates the cortical localization of NuMA-dynein complexes to correct mispositioned spindles. This pathway was discovered through the creation and analysis of Hmmr-knockout mice, which suffer neonatal lethality with defective neural development and pleiotropic phenotypes in multiple tissues. HMMR over-expression in immortalized cancer cells induces phenotypes consistent with an increase in active Ran including defects in spindle orientation. These data identify an essential role for HMMR in the PLK1-dependent regulatory pathway that orients progenitor cell division and supports neural development.

https://doi.org/10.7554/eLife.28672.001

eLife digest

As an embryo develops, its cells divide, grow and change into many different types of cells that eventually build our body. When cells divide, they first need to duplicate their genetic material. A structure called the spindle then distributes the two copies of the genetic information between the new cells. Cells must position their spindle precisely, and the way the spindle is oriented helps to determine what type of cell will develop. If the spindle fails to align properly, it can disrupt the development of specific tissues and organs and even lead to diseases such as cancer.

Numerous proteins help to position the spindle. For example, a protein called Ran-GTP ensures that motor proteins are anchored on opposite sides of the dividing cell, which tug on the spindle and position it between them. If the spindle gets pulled too closely to one side, a protein called PLK1 changes parts of the motor proteins to reduce the pulling force and to reposition the spindle towards the center.

Previous research has shown that non-motor proteins, such as a protein called HMMR are also part of the motor-protein complex. However, until now it was not known how HMMR was involved in repositioning the spindle during this process.

Now, Connell et al. have used mice that lacked HMMR to find out if it helps the cells in the brain to develop. The results show that without HMMR, very few mice were able to survive and many suffered from deformed and underdeveloped brains. In these mice, the orientation of the spindle changed and fewer cells of the correct type could be formed. Connell et al. then analyzed different types of cells grown in the laboratory to better understand how HMMR controls the position of the spindle. In all cases, HMMR formed a complex with Ran-GTP and was needed for the cells to orient their spindle correctly. When HMMR was absent, PLK1 could not work properly, and the spindle was positioned incorrectly.

This suggests that HMMR is essential for the spindle to align properly and is needed to help brain cells develop and become specialized. The next step will be to understand how HMMR, Ran and PLK1 work together during cell division. Studying mice that survive without HMMR offer an opportunity to examine how poorly aligned spindles affect their development.

https://doi.org/10.7554/eLife.28672.002

Introduction

The development and homeostasis of the brain require that neural stem cells, termed neuroepithelial progenitors (NPs), balance expansion with differentiation (Gönczy, 2008; Knoblich, 2008). During neurogenesis, NPs primarily undergo divisions that produce the various types of cells required for proper brain development. During these divisions, the orientation of the axis of division and mitotic spindle position determines the placement of the progeny cells within the tissue and, as a consequence, imbalances in the spindle positioning pathway can lead to defects in brain morphology (Lancaster and Knoblich, 2012).

Mitotic spindle position is controlled by forces applied on astral microtubules by dynein molecular motor complexes, which are anchored at the cell cortex by complexes of nuclear mitotic apparatus protein (NuMA)- G-protein signaling modulator 2 (GPSM2, aka LGN) (Konno et al., 2008; Morin et al., 2007; Peyre et al., 2011). In mammalian cultured cells, a gradient of Ran-GTP at chromosomes excludes LGN-NuMA complexes from lateral regions of the cell cortex to establish cortical asymmetry of these complexes and to position the spindle (Kiyomitsu and Cheeseman, 2012). The activity of dynein complexes anchored at polar regions of the cortex are regulated by the proximity of the spindle pole and its associated polo-like kinase 1 (PLK1) gradient (Kiyomitsu and Cheeseman, 2012) and complexes of dynein light chain 1 (DYNLL1)- hyaluronan mediated motility receptor (HMMR) (Dunsch et al., 2012). If one pole moves too close to the cortex, the centrosomal components of this pathway strip dynein from the cortex to reduce pulling forces and reposition the mitotic spindle. PLK1 activity, for example, stabilizes astral microtubules that are needed to strip dynein-dynactin complexes from the cortex (Zhu et al., 2013). HMMR is not proposed to be part of the NuMA-PLK1 spindle positioning pathway but rather to regulate cortical dynein through a complex containing CHICA (aka FAM83D) and DYNLL1 (Dunsch et al., 2012). However, HMMR and PLK1 activity are known to be interconnected during mitosis: during spindle assembly, PLK1 activity as measured by a kinetochore-localized, FRET-reporter construct is reduced following treatment of cells with siRNA targeting HMMR (Chen et al., 2014) and HMMR-threonine 703 is a putative substrate for PLK1 identified through mitotic phosphoproteome analysis (Nousiainen et al., 2006).

During cell division, HMMR acts downstream from Ran-GTP to localize targeting protein for XKLP2 (TPX2) and promote spindle assembly (Groen et al., 2004; Joukov et al., 2006; Chen et al., 2014; Scrofani et al., 2015). This action is reliant upon a carboxy-terminal basic leucine zipper (bZip) motif in HMMR, which targets it to the centrosome (Maxwell et al., 2003), and enables the activation of Aurora A by TPX2 (Chen et al., 2014) (Scrofani et al., 2015). Aurora A can directly phosphorylate PLK1 on Thr210 (Macůrek et al., 2008), which suggests that HMMR may impact the PLK1-dependent positioning pathway. Consistent with a potential role establishing the planar cell division exhibited in neural progenitors, HMMR has been shown to regulate apicobasal polarity in mammary epithelia (Maxwell et al., 2011), while expression of a truncated Hmmr (1–317 of 794 aa, termed Hmmr m/m) impairs planar cell division in ovarian follicle cells (Li et al., 2015), and germ cells in the testes (Li et al., 2016).

HMMR is expressed in the developing nervous system (Casini et al., 2010) and proliferative regions of the adult mouse brain (Lindwall et al., 2013), Recently, HMMR has been shown to be required for anterior neural tube closure and morphogenesis in Xenopus, where HMMR reduction leads to the absences of ventricular lamina and increased intraocular distance, olfactory bulb size, and forebrain width (Prager et al., 2017). The N-terminal microtubule binding region in HMMR is needed for neural tube morphogenesis in Xenopus (Prager et al., 2017) and the very terminal region is similar to that of Miranda (Chang et al., 2011), a regulator of asymmetric NP cell division in Drosophila (Ikeshima-Kataoka et al., 1997; Shen et al., 1997).

Hmmr mutant mice models are viable, including when central exons are targeted in Hmmr−/− mice (Tolg et al., 2003) and Hmmrm/m mice (Li et al., 2015), which result in the expression of truncated Hmmr transcript and protein (exons 1–7 or exons 1–10, respectively). Here, we studied the requirement of HMMR during oriented NP cell division and nervous system development through the creation of Hmmr-deficient mice with targeted disruption of Hmmr following exon 2. We find that HMMR is needed for neonatal survival and proper brain development. Our studies using cultured primary fibroblasts, directed differentiation of embryonic stem cells, and immortalized cancer cell lines, including neuroblastoma-like cells, uncovered a role for HMMR in the PLK1-dependent positioning pathway at mitotic spindle poles.

Results

Hmmrtm1a/tm1a neonates have reduced survival

We generated mice encoding a targeting construct following Hmmr exon 2, termed Hmmrtm1a(EUCOMM)Hmgu (hereafter Hmmrtm1a/tm1a, Figure 1A, Figure 1—figure supplement 1A–B). Western blot analysis of lysates from tissues known to have elevated levels of HMMR expression (spleen and testes) (Line et al., 2002) using antibodies targeting the N-terminal peptide in HMMR revealed that HMMR expression was completely lost in Hmmrtm1a/tm1a mice (Figure 1B). Adult Hmmrtm1a/tm1a mice were rare, and those Hmmrtm1a/tm1a mice that did survive were smaller than their wild-type (WT) littermates (Figure 1C). Similar to the phenotypes seen in Hmmrm/m mice attributed to misoriented germ cell divisions (Li et al., 2016), we observed atrophic seminiferous tubules and an increase in apoptosis in the testes as indicated by TUNEL staining in Hmmrtm1a/tm1a mice (Figure 1D–E). Additionally, Hmmrtm1a/tm1a mice were less fertile (fewer litters and fewer pups per litter) (Figure 1F–G). Few adult Hmmrtm1a/tm1a mice survived despite no evidence of embryonic lethality at E14.5 and E18.5 (Figure 1H). To identify when Hmmrtm1a/tm1a mice were dying, we monitored neonates for 2 days following birth. 12.5% of Hmmrtm1a/tm1a neonates were found dead within 3 hr of birth and 76.9% were found dead within the first 48 hr after birth (Figure 1I).

Figure 1 with 1 supplement see all
Hmmrtm1a/tm1a mice are smaller, exhibit fertility defects, and have decreased survival.

(A) Genotyping PCR confirmed insertion of the targeting vector between exon 2 and exon 4 in Hmmrtm1a/+ (Het) or Hmmrtm1a/tm1a (KO) but not in Hmmr+/+ (WT) mice. (B) HMMR expression in tissues extracted from WT, Het, or KO mice. Actin served as a loading control. (C) Weight at wean for WT and KO mice. Data are represented as mean ±SD (*p=0.028 (males), p=0.022 (females); for males, n = 10 (WT), 3 (KO); for females, n = 12 (WT), 4 (KO)). (D) Defects in seminiferous tubules are present in a KO male (*, atrophic tubules) relative to age-matched WT mouse stained with H&E. Scale bars, 200 µm. (E) Apoptosis (TUNEL staining) in KO male seminiferous tubules relative to age-matched WT mouse. Scale bars, 100 µm. (F) Number of litters per 6 months breeding time for matings of WT, Het, and KO mice. (*p<0.05; **p<0.01; n = 11 matings (WT x Het), 2 (Het x Het), 4 (WT x KO), 3 (Het x KO), 2 (KO x KO)). (G) Pups per litter for matings of WT, Het, and KO mice (See Figure 1F for n values). (H) Percentage of WT, Het, and KO pups at E14.5, E18.5 and weaning (~21 days) (***p<0.001; n = 64 (E14.5), 49 (E18.5), 133 (wean)). (I) Survival analysis for WT, Het or KO neonates during the first 48 hr following birth (n = 34 (WT), 69 (Het), 36 (KO)).

https://doi.org/10.7554/eLife.28672.003

Hmmrtm1a/tm1a mice display defects in brain structure, neural progenitor division, and differentiation

Necropsy samples from Hmmrtm1a/tm1a neonates (P0-1) demonstrated morphological defects in the brain, including defects in overall structure and size (Figure 2A). In multiple matching sections taken from WT or Hmmrtm1a/tm1a neonatal brains, we measured the area of the cerebrum and ventricles. We found large variation in the size of Hmmrtm1a/tm1a neonatal brains with three of the nine measuring two standard deviations smaller (microcephaly) than the mean brain size for age-matched (P0-1) WT littermates (Figure 2B). In addition, three of nine Hmmrtm1a/tm1a neonatal brains measured two standard deviations larger (megalencephaly) than the mean brain size for age-matched (P0-1) WT littermates, but these larger areas correlated with increases in the ventricle area in two of these three Hmmrtm1a/tm1a brains (Figure 2C). The putative reduced cortical area identified in Hmmrtm1a/tm1a relative to WT brains was not accompanied by increased levels of apoptosis, as measured by TUNEL staining, at either E18.5 or P0-1 (Figure 2—figure supplement 1A).

Figure 2 with 1 supplement see all
Hmmrtm1a/tm1a mice display defects in brain structure, neural progenitor division, and differentiation.

(A) Hmmrtm1a/tm1a (KO) neonate brains showed abnormalities compared to Hmmr+/+ (WT) brains (*, enlarged ventricles). Scale bars, 500 μm. (B) Quantification of brain size in WT and KO neonate (P0-1) brains. Dotted line indicates average of WT brains. The grey region indicates two standard deviations from the WT mean (n = 3 (WT), 9 (KO), p=0.512). (C) Correlation between brain size and ventricular area in WT and E14.5 P0-1 brains. (D) HMMR localization to the mitotic spindle (γ-tubulin) in neuroepithelial progenitor (NP) cells of WT and KO E14.5 brains. Scale bar, 10 μm. (E) Mitotic NP cells labeled with γ-tubulin (centrosomes) and DAPI in WT and KO E14.5 brains. Scale bar, 5 μm. (F) Division angle for NP cells in WT or KO E14.5 embryos (***p<0.001; n = 94 (WT), 96 (KO)). (G) Mitotic rate in Pax6+ NP cells in WT and KO E14.5 brains. Stars indicate dividing cells. Scale bar, 10 μm. (H) Quantification of mitotic rate in WT and KO E14.5 brains. Data are represented as mean ±SD (*p=0.018; n = 4(WT), 5(KO)). (I) Tbr2 localization in WT and KO E14.5 brains. Scale bars, 20 μm. (J) Quantification of Tbr2+ cells in E14.5 brains. Data are represented as mean ±SD (*p=0.019; n = 4 (E14.5, WT), 5 (E14.5, KO)).

https://doi.org/10.7554/eLife.28672.005

A putative reduction in cortical area and the potential manifestation of enlarged ventricles (hydrocephalus) in Hmmrtm1a/tm1a mice is consistent with phenotypes associated with defects in the orientation of NP cell division (Noatynska et al., 2012; Kielar et al., 2014). As HMMR is known to regulate the cell division axis in immortalized cancer cells (Dunsch et al., 2012), we next examined the localization of HMMR and the axis of NP cell division during neurogenesis in E14.5 embryos. In WT embryos, HMMR localized to spindle microtubules in dividing NP cells lining the ventricle surface (Figure 2D). The angle of NP cell divisions measured during anaphase was strongly biased to be planar with the ventricle surface (<30 degrees off the axis of centrosomes lining the surface) with an average spindle angle of 18.0 ± 19.5 degrees (Figure 2E–F). In WT brains, 60.6% or 82.9% of divisions were oriented within 15 degrees or 30 degrees of the ventricular surface, respectively. In Hmmrtm1a/tm1a embryos, however, HMMR was absent from mitotic spindles (Figure 2D) and the axis for division was altered with an average angle of 24.8 ± 19.5 degrees (p<0.05; unpaired t-test) with only 38.5% or 66.6% of divisions oriented within 15 degrees or 30 degrees of the ventricular surface, respectively (χ2 test, p<0.001) (Figure 2E–F). The degree of misorientation observed in Hmmrtm1a/tm1a NP cells compares to that observed in LGN-mutant NP cells, which have 37.1% or 60% of divisions oriented within 15 degrees or 30 degrees of the ventricular surface, respectively (Konno et al., 2008). We also observed a significant reduction in the number of mitotic Pax6+ radial glial cells at E14.5 in the brains of Hmmrtm1a/tm1a embryos (Figure 2G–H) and a consistent, significant reduction in Tbr2+ intermediate progenitor cells in Hmmrtm1a/tm1a brains (Figure 2I–J). However, we observed no significant difference between WT and Hmmrtm1a/tm1a brains in the levels of TUNEL-positive cells in the cortex, the localization of Par3 and ZO-1 in cells lining the ventricles, or the levels of centrosome amplification in dividing NP cells (Figure 2—figure supplement 1A–C). These data support the conclusion that the reduced cortical areas observed in Hmmrtm1a/tm1a neonatal mice, which are less able to successfully transition to extrauterine life, result from the reduced production of intermediate progenitor cells that correspond with impaired planar spindle orientation during NP cell division and an overall decreased mitotic rate of Pax6+ radial glial cells.

Mitotic spindle orientation requires the centrosome targeting bZip motif in HMMR

To better understand the mechanisms through which HMMR controls spindle orientation, we isolated mouse embryonic fibroblasts (MEFs) from WT or Hmmrtm1a/tm1a mice and imaged these cells through mitosis (Figure 3A). Hmmrtm1a/tm1a MEFs contained significantly higher levels of centrosome amplification (defined as cells containing greater than two centrosomes, which were quantified using colocalization of pericentrin and γ–tubulin) (Figure 3B–C), and were less able to orient the cell division axis, measured at anaphase relative to their long cell axis in interphase (Figure 3A,D–E). In addition, Hmmrtm1a/tm1a MEF progeny cells were more likely to display indications of genome instability, including increases in the frequency of mitotic death, binucleated progeny and micronuclei (Figure 3F). To ensure that the observed spindle orientation defects were not due to changes in cell shape that may accompany the loss of HMMR, we confirmed that the width to long axis ratio of the cell did not differ between WT or Hmmrtm1a/tm1a MEFs (Figure 3—figure supplement 1A). Thus, Hmmrtm1a/tm1a MEFs are less able to correctly orient the axis of cell division, which is similar to the defective spindle orientation that was observed following the treatment of HeLa cells with siRNA targeting HMMR (Dunsch et al., 2012).

Figure 3 with 1 supplement see all
The bZip domain of HMMR is required for spindle orientation.

(A) Cell division in mouse embryonic fibroblasts (MEFs) isolated from Hmmr+/+ (WT) or Hmmrtm1a/tm1a (KO) embryos. DNA is counterstained with Hoechst (red). Yellow line indicates long axis of interphase cell and red line indicates cell division axis. Scale bars, 20 μm. (B) Centrosome amplification in WT or KO MEFs indicated by colocalization of pericentrin and γ-tubulin. Arrow indicates centrosome. (C) Quantification of centrosome amplification in WT or KO MEFs. Amplification was counted as >2 centrosomes per cell as indicated by colocalization of pericentrin and γ-tubulin. Data are represented as mean ±SD (**p=0.0002; 3 replicates of ≥100 cells). (D) Schematic of spindle angle measurement in MEFs. (E) Spindle angle in WT or KO MEFs. KO MEFs were transduced with HMMR constructs truncated at the indicated amino acid. WT and KO average spindle angle are significantly different (**p<0.01; n = 117 (WT), 114 (KO), 50 (KO + FL), 56 (KO + 679), and 31 (KO + 623)). (F) Mitotic outcomes in WT or KO MEFs (n = 111 (WT) and 94 (KO)).

https://doi.org/10.7554/eLife.28672.007

To uncover the critical domain in HMMR needed to establish the cell division axis, we transduced GFP-HMMR (human) constructs into Hmmrtm1a/tm1a MEFs. We used truncations of the bZip domain, which is required to target HMMR to the centrosome (Maxwell et al., 2003), including GFP- HMMRFL (full-length, FL), HMMR1-679 (truncation following the b-Zip domain), and HMMR1-623 (truncation lacking the b-Zip domain). The transduction of HMMRFL or HMMR1-679, gene products that are known to locate at centrosomes (Figure 3—figure supplement 1B)(Maxwell et al., 2003), recovered the axis of cell division (Figure 3E). However, the expression of HMMR1-623 failed to rescue the axis of cell division (Figure 3E), implicating the centrosome targeting bZip motif (amino acids 624–679) as an essential domain for the orientation of cell division.

Fixed spindle position and asymmetric cortical NuMA-dynein localization requires HMMR

As the centrosome targeting domain was critical to HMMR function in spindle positioning, we next used time-lapse imaging of HeLa cells that stably expressed DHC-GFP and were previously treated with either scrambled control siRNA or siRNA targeting HMMR to determine whether the centrosome-localized, PLK1-dependent positioning pathway, which regulates the localization and activity of dynein complexes anchored at polar regions of the cortex (Kiyomitsu and Cheeseman, 2012), was functional in HMMR-silenced cells. As multipolar spindles occur at an elevated frequency in HMMR-silenced cells (Maxwell et al., 2005), we restricted our analysis to mitotic cells with phenotypically normal bipolar spindles. Immediately prior to anaphase in control-treated cells with an off-center mitotic spindle, DHC-GFP was asymmetrically localized to the far polar region of the cortex and absent from the cortex near the proximal pole leading to the correction of spindle position (Figure 4A, Video 1) as described previously (Kiyomitsu and Cheeseman, 2012). In HMMR-silenced cells with an off-center mitotic spindle, however, DHC-GFP remained on the polar cortex and its retention on the cortex predicted the direction of spindle rotation (Figure 4A, Video 2). We also measured the intensity of DHC-GFP recruited to the cortex relative to the cytoplasmic intensity and determined that overall DHC-GFP recruitment to crescents was not significantly different between scrambled control-treated and HMMR-silenced cells (Figure 4—figure supplement 1A). In addition, the spindle was maintained in an off-centered position in HMMR-silenced cells as determined by the ratio of the distance between the cortex and the location of chromosomes in anaphase cells (Figure 4B). A prolonged period of spindle rotation was frequently observed in HMMR-silenced HeLa cells expressing DHC-GFP (Figure 4C) and, as seen in separate experiments, in HMMR-silenced HeLa cells expressing mCherry histone H2B, eGFP-TUBA (Figure 4—figure supplement 1B–C; Videos 34).

Figure 4 with 1 supplement see all
HMMR regulates cortical NuMA/Dynein localization.

(A) Localization of Dynein heavy chain tagged with GFP (DHC-GFP) stably expressed in HeLa cells treated with scrambled (siScr) or HMMR (siHMMR) siRNA using time-lapse imaging (arrows indicate predicted direction of movement). Scale bar, 10 μm. (B) Quantification of off-center spindles at anaphase start in cells treated with siScr or siHMMR (n = 25 (siScr), 26 (siHMMR)). (C) Quantitation of spindle movement in cells treated with siScr or siHMMR. Back and forth movement is exemplified by Figure 5A (top panels) and rotation is exemplified by Figure 5A (middle and bottom panels) (p<0.001; n = 22 (siScr), 28 (siHMMR)). (D) Localization of NuMA and DHC-GFP in HeLa cells transfected with siScr or siHMMR with mispositioned spindles. Scale bar, 5 μm. (E) Quantification of NuMA and DHC-GFP intensities at the farthest polar cortex compared to the proximal polar cortex in HeLa cells transfected with siScr or siHMMR. Cells are pooled from experiments in panels 1E,F. (***p<0.0001; n = 39 (untreated), 32 (siScr), 53 (siHMMR)). (F) Localization of DHC-GFP and NuMA in HeLa cells transfected with siScr or siHMMR. Scale bar, 5 μm. (G) Quantification of NuMA and DHC-GFP at the poles compared to the midzone in HeLa cells transfected with siScr or siHMMR. All data are represented as mean ±SD (p<0.001; 3 replicates of ≥25 cells).

https://doi.org/10.7554/eLife.28672.009
Video 1
Dynein localization in HeLa DHC-GFP cells treated with scrambled siRNA that possess off-center spindles.

Images were captured every 1 min.

https://doi.org/10.7554/eLife.28672.011
Video 2
Dynein localization in HeLa DHC-GFP cells treated with siRNA targeting HMMR that possess off-center spindles.

Images were captured every 1 min.

https://doi.org/10.7554/eLife.28672.012
Video 3
Spindle movement in HeLa cells expressing mCherry-histone H2B, eGFP-TUBA treated with scrambled siRNA.

Images were captured every 15 min.

https://doi.org/10.7554/eLife.28672.013
Video 4
Spindle movement in HeLa cells expressing mCherry-histone H2B, eGFP-TUBA treated with siRNA targeting HMMR.

Images were captured every 15 min.

https://doi.org/10.7554/eLife.28672.014

We next examined the localization of the dynein anchoring protein NuMA in fixed HMMR-silenced cells by immunofluorescence. When the fluorescence intensity of DHC-GFP or NuMA was measured along the cortex in control-treated cells with an off-center spindle, the ratio of intensities along the far polar cortex was elevated compared to that along the proximal polar cortex (Figure 4D–E), as expected. However, NuMA was retained at the proximal polar cortex and the consequent ratio of intensities was significantly dampened in HMMR-silenced mitotic cells with an off-center metaphase spindle (Figure 4D–E). Yet, when the metaphase spindle was centered, the asymmetric cortical localization of NuMA and DHC-GFP was indistinguishable in control-treated and HMMR-silenced cells (Figure 4F–G). That is, the ratio of intensities along the polar cortex relative to the midzone cortex was elevated at the polar cortex for both NuMA and DHC-GFP, as reported (Kiyomitsu and Cheeseman, 2012). Our data indicate HMMR is dispensable for the loss of NuMA and DHC-GFP at the midzone cortical region, which is established by Ran-GTP gradient generated at chromosomes (Kiyomitsu and Cheeseman, 2012), but HMMR is needed for a centrosome-localized positioning pathway that strips NuMA and DHC-GFP when the pole is brought proximal to the cortex.

HMMR enables the PLK1-dependent spindle pole positioning pathway

We identified a requirement for HMMR in the removal of cortical NuMA-DHC-GFP, which is consistent with a prior report (Dunsch et al., 2012). However, we observed spindle rotation in HMMR-silenced cells rather than fixed and misoriented spindles (Dunsch et al., 2012). It was postulated that HMMR-DYNLL1 complexes at the centrosome functioned independent of the PLK1 intrinsic positioning pathway (Dunsch et al., 2012). That is, spindle pole-localized HMMR may create a local increase of DYNLL1 at the spindle/spindle poles that binds to dynein in competition with dynactin and displaces dynein from the cortex (Dunsch et al., 2012); however, Dyn2 (yeast DYNLL1 ortholog) promotes, rather than impedes, the incorporation of dynactin into the dynein motor complex (Stuchell-Brereton et al., 2011). So, we examined whether the loss of HMMR altered the composition of the dynein motor complex. HeLa cells expressing DHC-GFP were treated with control siRNA or siRNA targeting HMMR and DHC complexes were immunoprecipitated with antibodies recognizing GFP. In HMMR-silenced lysates, relative to control-treated lysates, the abundance of subunits for dynein (DHC, dynein intermediate chain (DIC), and DYNLL1) and dynactin (p150glued) were unchanged (Figure 5A). While the amount of DHC-GFP precipitated and the levels of DIC co-precipitated with DHC-GFP remained unaffected, the level of co-precipitated DYNLL1 was reduced in HMMR-silenced lysates (Figure 5A). Moreover, a corresponding reduction in p150glued co-precipitated with DHC-GFP was also observed in HMMR-silenced lysates (Figure 5A). We also noted that both FLAG-DYNLL1 and CHICA were retained at spindle poles, although absent from spindle fibers, in HMMR-silenced cells (Figure 5B–C). Therefore, HMMR appears dispensable for the spindle pole localization of CHICA and DYNLL1 and HMMR promotes, rather than restricts, the incorporation of p150glued into DHC complexes. As our data do not support the proposed PLK1-independent role for HMMR in spindle positioning, we investigated a putative role for HMMR in the PLK1-dependent positioning pathway.

HMMR enables the PLK1-dependent spindle pole positioning pathway.

(A) Western blot analysis of mitotic HeLa extracts stably expressing DHC-GFP treated with scrambled (siScr) siRNA or siRNA targeting HMMR (siHMMR) subjected to immunoprecipitation with GFP antibody or control IgG and blotted with the indicated antibodies, including GFP (GFP-DHC), dynein intermediate chain (DIC), p150glued, DYNLL1, HMMR or Actin (four replicates). (B) Localization of Flag-Dynein light chain (DYNLL1) in HeLa cells treated with siScr or siHMMR (three replicates). Scale bars, 10 μm. (C) Localization of CHICA in cells treated with siScr and siHMMR (three replicates). Scale bars, 10 μm. (D) Localization of phospho-PLK1 (pPLK1) in HeLa cells expressing siScr or siHMMR. Scale bars, 5 μm. (E) Quantification of pPLK1 intensity from pole to pole. Data represents mean (50 cells from three experiments). (F) Astral microtubules contact the cortex in cells treated with siScr, siHMMR, or BI2536 (Plk1 inhibitor). Yellow box indicates EB1 inset and white dotted line indicates region of quantification. Scale bar, 5 μm. (G) Quantification of EB1 at the cortex in cells treated with siScr, siHMMR, or BI2536. Data are represented as mean ±SD (***p<0.001; 2 replicates of >40 cells per treatment). (H) NuMA localization in HeLa cells treated with siScr or siHMMR and DMSO or BI2536 (PLK1 inhibitor, 20 nM). Yellow box indicates magnified region. Scale bar, 5 μm. (I) Quantification of NuMA localization at the cortex near to the centrosome at the region nearest the centrosome compared to those farther away. Data are represented as mean ±SD (**p<0.01; n = 29 (siScr + DMSO), 39 (siScr + BI2536), 25 (siHMMR + DMSO), 39 (siHMMR + BI2536) from three experiments).

https://doi.org/10.7554/eLife.28672.015

PLK1 activity at centrosomes and kinetochores enables the removal of LGN-NuMA-DHC complexes from the cortex (Kiyomitsu and Cheeseman, 2012; Zhu et al., 2013; Tame et al., 2016). As PLK1 activity is reduced at kinetochores in HMMR-silenced cells (Chen et al., 2014), we measured the levels of phosphorylated-PLK1, the active form of the kinase, by immunofluorescence in HMMR-silenced and control-treated mitotic cells. In HMMR-silenced mitotic cells, we observed a significant reduction in the fluorescence intensity of pPLK1 at spindle poles (Figure 5D–E). Consistent with a putative reduction in PLK1 activity at spindle poles, we observed a significant decrease in the number of EB1-positive microtubule plus ends contacting the cortex in HMMR-silenced or BI2536-treated mitotic cells (Figure 5F–G). As these data suggest HMMR enables PLK1-dependent processes, we tested if the concurrent inhibition of HMMR and PLK1 significantly altered the spindle positioning pathway when compared to the inhibition of either alone. We focused on the localization of NuMA to the cortical regions most proximal to the spindle pole in control treated, HMMR-silenced, BI2536-treated, or dual inhibited metaphase cells with off-center spindles (Figure 5H). In control-treated cells, NuMA was uniformly absent from the cortex, while NuMA was largely retained in HMMR-silenced cells with the exception of a discreet loss in the inner region directly proximal to the pole (Figure 5H–I). In mitotic cells treated with BI2536, the localization of NuMA mirrored that observed in HMMR-silenced cells and was not additively altered in dual inhibited cells (Figure 5H–I) consistent with a requirement for PLK1 and HMMR in a shared positioning pathway. Thus, our data supports the conclusion that astral microtubule density is dampened in HMMR-silenced mitotic cells in a PLK1-dependent manner, which disturbs the spindle pole-localized positioning pathway and results in spindle rotation due to the retention of NuMA-DHC complexes at the proximal cortex.

HMMR interacts with RanBP2 and regulates the centrosome localization of Ran

As PLK1 phosphorylates HMMR at threonine 703 (pHMMR) (Nousiainen et al., 2006), we used mass spectrometry to identify proteins that are co-precipitated with antibodies targeting pHMMR as a means to discover putative pathways through which centrosome-localized HMMR may regulate spindle position. In pHMMR immunoprecipitates from mitotic or G2-phase synchronized HeLa cell lysates, we identified a number of actin-, myosin-, and dynein-binding proteins (Figure 6—figure supplement 1; Figure 6—source data 1), including known interactors FAM83D/CHICA and DYNLL1 (Dunsch et al., 2012). We filtered out proteins that were also co-precipitated with antibodies targeting a non-phosphorylated N-terminal peptide in HMMR (unpublished results) and found that proteins related to small GTPases, such as ARHGAP17, RACGAP1, and RanBP2, were uniquely co-precipitated with pHMMR antibodies (Figure 6A). As Ran regulates cortical LGN-NuMA-dynein localization during mitosis (Kiyomitsu and Cheeseman, 2012), we focused on the putative pHMMR-RanBP2 interaction and confirmed this interaction by western blot analysis (Figure 6B).

Figure 6 with 1 supplement see all
Phosphorylated HMMR interacts with RanBP2 and regulates the centrosome localization of Ran.

(A) Mass spectrometry analysis of proteins co-precipitated from HeLa cell lysates with antibodies targeting HMMR phosphorylated at threonine 703 (pHMMR). Proteins, other than HMMR, co-precipitated with antibodies targeting non-phosphorylated HMMR have been removed. (B) Immunoblots of mitotic HeLa extracts subjected to immunoprecipitation with pHMMR antibody or control IgG and blotted with the indicated antibodies. (C) Ran activity assay (GTP-bound Ran is immunoprecipitated) in mitotic HeLa cells transfected with scrambled (Scr) or siRNA targeting HMMR (two replicates). (D) Immunoblots of mitotic HeLa extracts treated with the indicated concentration of non-hydrolysable GTP (GTPγS) and immunoprecipitated with HMMR antibody or control IgG and blotted with the indicated antibodies (three replicates). (E) Localization of constitutively-active Ran (RanQ69L; RanCA) in mitotic HeLa cells transfected with scrambled (Scr) or siRNA targeting HMMR. Yellow boxes indicate magnified region. Scale bar, 5 μm. (F) Quantification of the ratio of intensities for RanCA relative to γ-tubulin in mitotic HeLa cells transfected with scrambled (Scr) or siRNA targeting HMMR. Data are represented as mean ±SD (**p<0.01; 3 replicates of ≥50 cells). (G) Localization of constitutively-active Ran (RanQ69L; RanCA) in cells treated with DMSO or BI2536 (Plk1 inhibitor, 20 nM). Yellow boxes indicate magnified region. Scale bar, 5 μm. (H) Quantification of RanCA localization at the centrosome in cells treated with DMSO or BI2536 (20 nM). Data are represented as mean ±SD (***p=0.0006; 2 replicates of >30 cells).

https://doi.org/10.7554/eLife.28672.016

RanBP2 binds specifically to Ran-GTP (Vetter et al., 1999), so we postulated that HMMR may affect Ran-GTP levels or localization. To test this postulate, we first measured the levels of Ran-GTP in scrambled siRNA-treated or HMMR-silenced HeLa cell lysates with a commercially available assay that pulls-down the active form of Ran (Ran-GTP). In control experiments, scrambled siRNA-treated HeLa cell lysates were treated with either non-hydrolysable GTP (GTPγS) or with GDP to switch all Ran to Ran-GTP or Ran-GDP, respectively, prior to immunoprecipitation. As expected, the level of co-precipitated Ran was significantly augmented in cell lysates treated with GTPγS and lost in cell lysates treated with GDP (Figure 6C). Having verified our control conditions, we then compared the levels of Ran-GTP in control-treated and HMMR-silenced cell lysates. In HMMR-silenced cell lysates, the level of Ran-GTP precipitated was slightly reduced relative to scrambled control-treated cell lysates (Figure 6C). When we confirmed the knockdown efficacy of HMMR in these experiments, we noted that HMMR was also co-precipitated when lysates were pretreated with GTPγS suggesting HMMR may interact specifically with Ran-GTP, which is greatly increased by GTPγS treatment. To titrate a putative HMMR-Ran-GTP interaction, we precipitated HMMR from mitotic cell lysates previously treated with increasing amounts of GTPγS. While the levels of precipitated HMMR remained constant, we found that the level of Ran co-precipitated with HMMR was increased in a GTPγS dose-dependent manner (Figure 6D). Thus, HMMR interacts with Ran-GTPγS in cell lysates. We predicted that HMMR may affect the localization of active Ran in mitotic cells. To test this, we expressed constitutively active Ran (Ran Q69L, RanCA) (Kazgan et al., 2010) in HeLa cells treated with scrambled siRNA or siRNA targeting HMMR. In scrambled siRNA-treated cells, a fraction of RanCA colocalized with the spindle pole demarked by γ-tubulin (Figure 6E), consistent with the identification of Ran in the centrosome proteome (Andersen et al., 2003). However, the fraction of RanCA that colocalized with γ-tubulin, as measured by the ratio of intensities for RanCA and γ-tubulin, was significantly reduced in HMMR-silenced cells (Figure 6E–F). Similarly, inhibition of PLK1 activity, through treatment with a small-molecule inhibitor BI2536, also reduced the fraction of RanCA that colocalized with γ-tubulin (Figure 6G–H). This data shows that reducing PLK1 activity or silencing HMMR reduces the localization of constitutively active Ran at mitotic centrosomes and suggests the phosphorylation of HMMR (pHMMR) by PLK1 may promote pHMMR-RanBP2-Ran-GTP complexes at spindle poles.

HMMR localizes Ran and positions metaphase spindles in neural cells and tissues

To investigate the HMMR-Ran pathway in neural cells and tissues, we first utilized the neuroblastoma cell line, SHSY5Y, which is known to polarize NuMA during cell division (Izumi and Kaneko, 2012). Following the transduction of shRNA targeting HMMR or nonhairpin control shRNA, we confirmed the loss of HMMR immunofluorescence at mitotic spindles (Figure 7—figure supplement 1A). In HMMR-silenced cells compared to control-treated cells, we observed a decrease in the immunofluorescence intensity detected for endogenous Ran colocalized with pericentrin, a centrosome marker (Figure 7—figure supplement 1B–C). We then investigated mitotic spindle structure and position as well as the localization of Ran in sections derived from WT and Hmmrtm1a/tm1a E14.5 brains. In these sections stained for the spindle marker beta-tubulin (Figure 7A), we measured the position and length of the spindle in dividing neural progenitor cells that lined the ventricles. When compared to spindles within wild-type progenitor cells, spindles in Hmmrtm1a/tm1a neural progenitors were less centered (a/b measurement) and significantly shorter (c measurement) (Figure 7B). Additionally, the density of spindle fibers, as measured by the intensity of beta-tubulin fluorescence, was significantly lower in Hmmrtm1a/tm1a neural progenitors than those in WT cells (Figure 7C). Similar to our observations in HMMR-silenced HeLa or SH-SY5Y cells, we found a fraction of endogenous Ran, as measured by immunofluorescence, colocalized with a centrosome marker (γ-tubulin, Figure 7D); this fraction of Ran, as measured by the ratio of intensities for Ran and γ-tubulin, was significantly reduced in Hmmrtm1a/tm1a relative to WT neural progenitors (Figure 7E).

Figure 7 with 2 supplements see all
HMMR regulates centrosomal Ran and spindle centering in neural tissues.

(A) Spindle position, density, and length in neuroepithelial progenitor (NP) cells in Hmmr+/+ (WT) or Hmmrtm1a/tm1a (KO) E14.5 mice. Yellow dotted line indicates cell border and white dotted line indicates the ventricle edge. Scale bar, 5 μm. (B) Quantification of spindle position and length in WT and KO E14.5 mice. Data are represented as mean ±SD (**p=0.003 (a/b),***p=5.4e-7 (c); n = 57 (WT- a/b), 56 (KO- a/b), 56 (WT- c), 51 (KO- c)). (C) Quantification of spindle density in WT and KO E14.5 mice. Data are represented as mean ±SD (**p=0.003; n = 25 (WT), 20 (KO)). (D) Localization of Ran in NP cells in WT and KO E14.5 mice. Scale bar, 5 μm. (E) Quantification of Ran localization in WT and KO E14.5 brains. Data are represented as mean ±SD (**p=0.002; n = 29 (WT), 49 (KO)).

https://doi.org/10.7554/eLife.28672.019

Taken together, our data obtained from studies of cells, tissues and animals that are deficient for HMMR indicate a critical role for the protein in establishing the correct position of the mitotic spindle during cell division. Complete loss of HMMR is sufficient to alter the position and orientation of NP cell division and, in the case of Hmmrtm1a/tm1a mice, disturb brain development and impair the animal’s ability to successfully transition to extrauterine life. Our in vitro data support a model where HMMR supports PLK1 activity at the centrosome, which stabilizes astral microtubules and through the phosphorylation of HMMR localizes Ran to mitotic spindle poles. These PLK1-dependent spindle pole positioning processes are critical to reduce cortical localization of NuMA-DHC in off-center spindles and prevent spindle rotation (Figure 7—figure supplement 2A–B).

Ectopic HMMR alters Ran localization, cortical NuMA and spindle positioning

To compare and contrast the mitotic consequences that follow HMMR deletion to those that follow ectopic GFP-HMMR expression, we studied spindle positioning and NuMA localization using HeLa cells with doxycycline-inducible expression of GFP-HMMR (tet-HMMR) (Figure 8A), which express GFP-HMMR along the spindle microtubules in mitotic cells and undergo spindle tumbling when grown on L-shaped micropatterned substrates (He et al., 2017). In subconfluent cultures of induced tet-HMMR cells, metaphase spindles were more frequently off-center (Figure 8B). We next examined whether induced tet-HMMR expression altered the localization of co-expressed RanWT, RanCA, or RanDN constructs tagged with mCherry. In HeLa cells, these constructs localized as previously reported (Hutchins et al., 2009): RanWT was cytoplasmic, RanCA localized to the mitotic spindle, and RanDN localized to the chromosomes (Figure 8C). In induced tet-HMMR cells, however, RanWT localized to the spindle similar to that of RanCA (Figure 8C). Induction of GFP-HMMR expression also caused NuMA to be lost entirely from the cortex (Figure 8D–E). We observed a similar effect on cortical NuMA localization in HeLa cells overexpressing RanCA (Figure 8F–G). Taken together, these data indicate that expression of GFP-HMMR induces defects in the spindle positioning pathway that are consistent with an ectopic localization of Ran-GTP on mitotic spindles (Figure 7—figure supplement 2B).

HMMR over-expression affects spindle positioning through Ran activation.

(A) HMMR over-expression in induced tet-HMMR cells. HMMR is over-expressed under the control of a tet-inducible promoter (+dox). (B) HMMR over-expression leads to mispositioned mitotic spindles. Data are represented as mean ±SD (p=0.015; 3 replicates of ≥30 cells). (C) Localization of RanWT, RanCA, and RanDN constructs in tet-HMMR cells. Scale bars, 5 μm. (D) NuMA localization in induced and uninduced tet-HMMR cells. Lower panel shows magnified view of cortex near spindle pole. Scale bars, 5 μm. (E) Quantification of NuMA localization in induced and uninduced tet-HMMR cells. Data are represented as mean ±SD (***p=0.0004; 3 replicates of ≥30 cells). (F) Localization of NuMA in HeLa cells expressing RanCA constructs. Scale bar, 5 μm. (G) Quantification of NuMA localization in HeLa cells expressing RanCA constructs. Data are represented as mean ±SD (p=0.002; 3 replicates of ≥25 cells).

https://doi.org/10.7554/eLife.28672.022

Discussion

HMMR is classified as a non-motor spindle assembly factor (Manning and Compton, 2008) and has been shown to be a critical cell division gene product in immortalized cancer cells (Neumann et al., 2010). In order to study HMMR functions in vivo, we generated Hmmrtm1a/tm1a mice, which encode a targeting construct following Hmmr exon 2. Our approach is in contrast to previous published Hmmr mutant mice. An initial murine model, termed Hmmr−/−, was generated by targeted disruption at exon 8 (Tolg et al., 2003) while an alternative model, termed Hmmrm/m, was generated by targeted disruption at exon 10 (Li et al., 2015); thus, truncated N-terminal Hmmr transcript or protein is expressed in both of these published Hmmr models. Hmmr−/− animals exhibited fertility defects (Tolg et al., 2003), while Hmmrm/m mice demonstrated deficient spindle orientation during gametogenesis (Li et al., 2016; Li et al., 2015). However, Hmmrm/m and Hmmr−/− animals are both viable and phenotypically normal. This contrasts with the severely diminished survival we observed in Hmmrtm1a/tm1a neonates and, for the few animals that survived the transition, the reduced body sizes we observed in rare, adult Hmmrtm1a/tm1a mice.

We observed in Hmmrtm1a/tm1a mice previously unseen neural defects such as enlarged ventricles and microcephaly; both of these congenital conditions can arise from defects in spindle orientation and planar cell division within the progenitor population (Lancaster and Knoblich, 2012). In accordance, we observed a reduced proportion of mitotic Pax6+ cells that did not properly align their division axis in the subventricular zone (SVZ) and a decrease in the proportion of Tbr2+ progenitors at E14.5, suggesting an overall decreased neuronal production in these animals. These cellular phenotypes and reduced body size, are similar to those that accompany microcephalic brains observed in MagohMos2/+ mice (Silver et al., 2010), which are haploinsufficient for a regulator of the levels of expression of the dynein adaptor protein, Lis1. We also observed enlarged ventricles, which augmented the gross brain size in a proportion of Hmmrtm1a/tm1a mice. The etiology of these phenotypes may relate to impaired cerebrospinal fluid flow, as seen with mutation of dynein components and related to primary cilia dyskinesis (Ibañez-Tallon et al., 2002; Ostrowski et al., 2010). As HMMR is a non-motor adaptor for dynein, a more detailed analysis of the effect on CSF flow is Hmmrtm1a/tm1a mice is warranted.

A recent study identified misoriented division planes in apical neuroprogenitors and postnatal granule cell precursors, without changes to the cell division rate, during brain development of Hmmrm/m mice expressing truncated N-terminal Hmmr protein (Li et al., 2017). Transient megalencephaly was observed in Hmmrm/m brains during the neuronal differentiation period, up to PND7 but normalized by PND14, and attributed to misoriented divisions in apical neuroprogenitors leading to an increased number of Tbr+ intermediate progenitors contributing to cerebral cortex enlargement (Li et al., 2017). In Hmmrtm1a/tm1amice, however, the predominant phenotype observed was microcephaly, which may be explained by a decreased mitotic rate for Pax6+ radial glial cells; consistent with the recent observations in Hmmrm/m brains, we did observe rare Hmmrtm1a/tm1a brains at E14.5 with cortex enlargement.

The different neural phenotypes between Hmmrm/m and Hmmrtm1a/tm1a mice imply the expression of truncated N-terminal Hmmr protein retains critical functions needed for neural development and neonatal survival. In support of this, the microtubule-binding domain of HMMR (Assmann et al., 1999; Maxwell et al., 2003) has recently been shown to be important for neural development in Xenopus (Prager et al., 2017). Microinjection of morpholinos targeting Hmmr in Xenopus laevis embryos resulted in gross alterations in forebrain morphogenesis and anterior neural tube closure and these defects were rescued by the expression a truncated HMMR construct that preserved the microtubule-binding domain (Prager et al., 2017). These defects were attributed to the loss of a non-mitotic role for HMMR in the promotion of neural cell polarization and radial intercalation concomitant with neural tube closure (Prager et al., 2017). While we observed no changes in cell morphology in Hmmrtm1a/tm1a MEFs and no defects in the localization of Par3 or ZO-1 in Hmmrtm1a/tm1a NSCs lining the ventricles, future studies should investigate the disruption of Hmmr function in non-mitotic Hmmrtm1a/tm1a tissues and cells, such as during directed cell migration needed for corticogenesis and neural tube closure (Marín and Rubenstein, 2003 and Ayala et al., 2007).

Directed smooth muscle cell migration following balloon catheter injury in rats is reliant upon HMMR (Silverman-Gavrila et al., 2011), and HMMR is expressed in cells within the SVZ and rostral migratory stream in the adult mouse brain, where neural progenitor cells persist throughout life (Lindwall et al., 2013). While a detailed investigation of non-mitotic processes are needed in Hmmrtm1a/tm1a mice, we focused on mitotic processes due to our analysis of Hmmr promoter activity and HMMR expression during the directed neural differentiation of mouse embryonic stem cells (ESC). When we derived neural rosette-like structures from a mouse mouse HmmrBB0166/+ ESC line (Figure 7—figure supplement 1D–E), in which β-geo is inserted following Hmmr exon 7 on one allele, or from WT ESCs (Figure 7—figure supplement 1F), we found Hmmr/HMMR expression was restricted to apical-positioned, cyclin B1-positive rosette-neural stem cells (r-NSCs); moreover, spindle position was disturbed in apical-positioned r-NSCs derived from HmmrBB0166/+ mESCs (Figure 7—figure supplement 1G–H). Thus, Hmmr shows cell cycle-restricted expression in G2 phase and mitotic cells during the process of neural differentiation, and our data are consistent with the defects we observed in Hmmrtm1a/tm1a mice being attributed to alterations in neural progenitor expansion and differentiation as determined by changes to the control of spindle orientation and cell division.

Hmmrtm1a/tm1a mice exhibited the following phenotypes that are known to arise from defects in spindle orientation and planar cell division: microcephaly (Gai et al., 2016), enlarged ventricles (Dietrich et al., 2009; Godin et al., 2010), and testicular atrophy and fertility defects (Li et al., 2016; Li et al., 2015). Moreover, in HeLa cells, HMMR is critical to spindle positioning (Dunsch et al., 2012). However, other mitotic or post-mitotic processes may also be disturbed in Hmmrtm1a/tm1a mice independent of spindle orientation, such as an increased rate of apoptosis, centrosome amplification (Marthiens et al., 2013), and decreased division rate (Caviness et al., 2003; Caviness et al., 1995). In our study, we focused our efforts on addressing many of these additional factors and specifically those related to cell division. TUNEL analysis revealed no increase in apoptosis in Hmmrtm1a/tm1a mice at both embryonic and neonate stage. An elevated occurrence of supernumerary centrosomes, as induced for example through overexpression of PLK4, results in embryonic lethality (Vitre et al., 2015). Our analysis revealed no evidence of multipolar spindles or centrosome amplification in neural tissues, but we did note an increased frequency of centrosome amplification in Hmmrtm1a/tm1a MEFs. In addition, Hmmrtm1a/tm1a MEFs require more time to complete the spindle assembly checkpoint when followed through cell division independent of spindle orientation (Unpublished results). However, whether such delays in cell division kinetics are present in Hmmrtm1a/tm1a mice, or if an associated reduced rate of cell division contributed to the observed neural defects remain unknown. A loss of control of the cell division axis, however, has been consistently observed in HMMR-silenced HeLa cells (Dunsch et al., 2012) and, as described here, in r-NSCs derived from HmmrBB1066/+ mESCs, Hmmrtm1a/tm1a MEFs, and neural progenitor cells lining the SVZ in brains isolated from Hmmrtm1a/tm1a mice.

The intrinsic spindle positioning pathway regulates components of the dynein motor complex on the cortex to generate forces that center spindles. It is well recognized that the PLK1 gradient at both centrosomes and kinetochores is essential in regulating the spindle positioning pathway (Kiyomitsu and Cheeseman, 2012; Tame et al., 2016). Here, we find HMMR is a critical component of a PLK1-dependent positioning pathway, which contrasts with the findings of a prior report that predicts HMMR-CHICA-DYNLL1 play a PLK1-independent role that regulates the composition of cortical dynein complexes (Dunsch et al., 2012). While DYNLL1 and CHICA were identified in our mass spectrometry analysis of pHMMR immunoprecipitates and we found that the incorporation of DYNLL1 into DHC-GFP complexes was dampened in HMMR-silenced cell lysates, our data are consistent with HMMR-DYNLL1 enabling the incorporation of p150-glued into DHC-GFP complexes. This role is similar to that reported for Dyn2 (yeast DYNLL1 ortholog) (Stuchell-Brereton et al., 2011) but contrary to the prediction that DYNLL1-bound form of dynein cannot bind to dynein adaptors required for cortical targeting (Dunsch et al., 2012).

HMMR acts downstream from Ran-GTP to localize TPX2 (Groen et al., 2004; Joukov et al., 2006; Chen et al., 2014; Scrofani et al., 2015), which relies upon a carboxy-terminal bZip motif in HMMR (Maxwell et al., 2003) and enables the activation of Aurora A by TPX2 (Chen et al., 2014) (Scrofani et al., 2015). As Aurora A can directly phosphorylate PLK1 on Thr210 (Macůrek et al., 2008), our observed reduction in pPLK1(T210) in HMMR-silenced cells can be mechanistically explained through reduced Aurora A activity. Consistent with reduced PLK1 activity in HMMR-silenced cells, we find that astral microtubules are reduced and cortical NuMA is retained, which is phenocopied and not additively augmented by PLK1 inhibition. Indeed, the growth rate of EB1-marked microtubule comets was previously shown to be dampened in HMMR-silenced cells although the conclusion of these experiments was that astral microtubule organization was similar to control-treated cells (Dunsch et al., 2012). We note that the density of astral microtubules is reduced in HMMR-silenced cells, which provides an alternative mechanistic explanation for the retention of cortical DHC-GFP in HMMR-silenced cells with off-center spindles.

Ran is a component of the centrosome proteome (Andersen et al., 2003), although FRET-based assessment of Ran activity has not been seen at the centrosome (Kaláb et al., 2006). Here, we find evidence for the location of Ran-GTP at mitotic centrosomes reliant upon PLK1 activity and HMMR. Mass spectrometry analysis discovered RanBP2 as a novel interactor with pHMMR and we show that HMMR precipitates Ran-GTPγS in a dosage-dependent manner and is needed to locate active Ran at mitotic centrosomes. Moreover, in HeLa cells induced to express GFP-HMMR we also observe ectopic localization of Ran to mitotic spindles and the loss of cortical NuMA localization similar to those induced by expression of RanCA. However, HMMR appears dispensable for the Ran-GTP gradient generated at chromosomes, which strips NuMA and DHC-GFP at the midzone cortical region (Kiyomitsu and Cheeseman, 2012). Therefore, our data supports a spatially restricted role for pHMMR in the localization of Ran-GTP at centrosomes and spindle fibers downstream of PLK1 activity. However, the relative contribution in the control of spindle positon for pHMMR function with Ran at centrosomes versus the role HMMR plays in promoting pPLK1(T210), potentially through modulation of Aurora A activity, warrants further study.

Taken together, the results of our study identify HMMR as a key component of the PLK1-dependent mitotic spindle positioning pathway that is needed for neural development and neonatal survival. We suggest that the loss of spindle positioning identified here, and by others (Dunsch et al., 2012; Li et al., 2016; Li et al., 2015) (Li et al., 2017), in HMMR-silenced and HMMR-deleted or mutated cells and tissues, the occurrence of pleiotropic phenotypes observed in Hmmrtm1a/tm1a and Hmmrm/m mice, and the requirement of HMMR during meiotic (Groen et al., 2004; Joukov et al., 2006; Scrofani et al., 2015) and mitotic spindle assembly (Maxwell et al., 2005; Maxwell et al., 2003; Dunsch et al., 2012; Chen et al., 2014) together indicate that the physiological function for HMMR is as a homeostasis, meiosis, and mitosis regulator.

Materials and methods

Key resources table
Reagent type (species) or resourceDesignationSource or referenceIdentifiers
genetic reagent (Mus musculus)Hmmrtm1a/tm1athis paper
cell line (Homo sapiens)HeLaATCC (Manassas, VA)CCL-2
cell line (M. musculus)Hmmr+/+this paper
cell line (M. musculus)Hmmrtm1a/tm1athis paper
cell line (H. sapiens)HeLa DHC-GFPMitocheck, PMID: 20360068
cell line (H. sapiens)HeLa tet-HMMRPMCID: PMC5464802
cell line (H. sapiens)HeLa eGFP-TUBA, mCherry-histone H2BGruneberg lab, PMID: 21187329
cell line (H. sapiens)SH-SY5YSigma (St. Louis, MO)94030304-1VL
cell line (M. musculus)ES cells Hmmr+/+Mutant Mouse Resource Regional Cente (MMRC)rE14TG2a
cell line (M. musculus)ES cells HmmrBB0166/+MMRC026467-UCD
cell line (M. musculus)ES cells Hmmrtm1a(EUCOMM)HmguEuropean Conditional Mouse Mutagenesis Program
transfected constructHMMR siRNA 5'ADharmacon (Lafayette, CO), PMID 24875404
transfected constructHMMR siRNA 5'BDharmacon, PMID 24875404
transfected constructHMMR siRNA 3'Dharmacon, PMID 24875404
antibodyActinSigmaa5060
antibodyalpha-tubulinAbcam (United Kingdom)ab4074
antibodybeta-tubulinSigmaT5293
antibodybeta tubulin-647Cell Signaling Technology (Danvers, MA)3624
antibodymCherryAbcamab167453
antibodyDICMillipore (Burlington, MA)mab1618
antibodyDynLL1Abcamab51603
antibodyEB1Abcamab53358
antibodygamma-tubulinSigmat6557
antibodyGFPAbcamab1218
antibodyHMMRAbcamab124729
antibodypHMMRPMID:22110403
antibodyNuMAAbcamab36999
antibodyNuMACell Signaling Technology3888
antibodyp150 gluedAbcamab151184
antibodyPar3Millipore1330
antibodyPax6Covance (Princeton, NJ)prb278p10
antibodyPericentrinCovanceprb432c
antibodypPLK1Abcamab189139
antibodyPLZFEMD Chemicalop128
antibodyRanCell Biolabs Inc (San Diego, CA)in STA-409 kit
antibodyRanAbcamab53775
antibodyRanBP2Abcamab197044
antibodyTbr2Abcamab183991
antibodyTPX2Novus Biologicals (Littleton, CO)mb500179
antibodyZO-1Invitrogen (Carlsbad, CA)339100
antibodyZO-1Invitrogen402200
recombinant DNA reagentpmCherry-C1-RanQ69LAddgene (Cambridge, MA), PMID: 20685962Addgene #30309
recombinant DNA reagentpcDNA-RanWT-mRFP1-polyAAddgene, PMID: 24908396Addgene #59750
recombinant DNA reagentmCherry-RanDNPMID: 23870127
recombinant DNA reagentHMMR-FLPMID: 12808028
recombinant DNA reagentHMMR 1–623PMID: 12808028
recombinant DNA reagentHMMR 1–679PMID: 12808028
commercial assay or kitRan Activation Assay kitCell Biolabs, IncSTA-409
chemical compound, drugBI2536Selleck Chemicals (Houston, TX)S1109

Generation of Hmmrtm1a/tm1a mice

Hmmrtm1a/+ ES cell strains with the L1L2_Bact_P cassette inserted after Hmmr exon 2 were purchased from The European Conditional Mouse Mutagenesis Program (HEPD0778_4_B11; EUCOMM). Hmmrtm1a/+ ES cells were introduced into blastocyst stage embryos by microinjection and resulting male chimera mice were bred to C57BL/6J female to obtain Hmmrtm1a/+ mice. Hmmrtm1a/+ mice were then intercrossed to generate Hmmrtm1a/+ mice on a C57BL/6J background. All mice were maintained in the pathogen-free Centre for Molecular Medicine and Therapeutics animal facility on a 6 am-to-8 pm light cycle, 20 ± 2°C, with 50 ± 5% relative humidity, and had food and water ad libitum. All procedures involving animals were in accordance with the Canadian Council on Animal Care (CCAC) and UBC Animal Care Committee (ACC) (Protocol no. A13-0168).

Genotyping

Tail clips from embryos and ear notches from weaned animals were lysed in GB buffer (100 mM Tris, pH8.8, 100 mM (NH4)2SO4, 100 mM MgCl2, 1% β-mercaptoethanol, 0.5% triton X-100 and 1.6 mg/ml protease K) at 50°C for 3–5 hr. Protease K was inactivated at 95°C for 10 min. PCR was performed using AccuStart II PCR mix (Quanta Biosciences, Beverly, MA). Primers used for PCR1 were: Hmmr-for, 5’-AGATACAACCTTGCTTGCTTCGGC-3’, loxR, 5’-TGAACTGATGGCGAGCTCAGACC-3’ (mutant 507 bp); Primers used for PCR2 were: Hmmr-5'-arm, 5’-CAGGCCTTAGAAGCTGACATGAGC-3’, Hmmr-3'-arm, 5’-TCCAAACTTCTCACTGCAGACAGC-3’, LAR3, 5’-CAACGGGTTCTTCTGTTAGTCC-3’ (WT 515 bp, mt 339 bp).

Immunoblotting

Mouse tissues were harvested, snap frozen, ground into powder, lysed in RIPA buffer supplemented with protease inhibitor (Roche, Switzerland) and sonicated. Lysates were clarified by centrifugation at 16,000xg for 20 min at 4°C, and concentration was determined by BCA protein assay kit (Thermo Scientific, Waltham, MA). Lysates were mixed with SDS sample buffer, separated by SDS-PAGE, and blotted with the indicated antibodies: Actin (rabbit (rb), Sigma, 1:2500) and HMMR (rb, Abcam, 1:500).

Cell culture

HmmrBB0166/+ mouse ES cells (BB0166; MMRRC: 026467-UCD) and the parental control mouse ES cells (E14TG2a; MMRRC: 015890-UCD) were purchased from a Mutant Mouse Regional Resource Center (University of California, Davis). ES cells were maintained on mitomycin C-treated MEFs prepared as previously described (Conner, 2001). ES cells were cultured and neural induction was initiated as previously described (Barberi et al., 2003).

To generate Hmmrtm1a/tm1a MEFs, heterozygous mice were interbred as previously described (Johnson et al., 1995) with the following exceptions: embryos were collected at day 13.5 (day 1) and homogenized using an 18-gauge needle and a 10cc syringe, cultures were incubated for two days, trypsinized on day 3, and frozen once confluent.

HeLa cells (ATCC: CCL-2) were maintained as previously described (Chen et al., 2014). HeLa cells expressing mouse DHC-GFP were obtained from Mitocheck and maintained as described (Hutchins et al., 2010). HeLa cells expressing mCherry-Histone H2B, eGFP-TUBA were obtained from the Gruneberg lab (University of Liverpool) (Zeng et al., 2010) and maintained in media with 0.3 μg/ml puromycin (Invitrogen) and 0.5 μg/ml blasticidin S (Invitrogen). Live imaging was performed using Leibovitz’s L-15 media supplemented with 10% FBS. BI2536 (Selleck Chemicals) treatment was performed for 2 hr at a concentration of 20 nM.

HeLa cells with tet-on inducible expression of enhanced GFP fused in frame with full-length HMMR (GFP-HMMR) were obtained from Dr. LM Pilarski (University of Alberta) and produced as described (He et al., 2017). For experiments, expression of HMMR was induced using 2 μg/ml of doxycycline (Clontech, Mountain View, CA) for 10 hr.

SH-SY5Y (Sigma 94030304; ECACC validated prior to purchase) were maintained as previously described (Izumi and Kaneko, 2012).

Immunostaining

Neural rosettes and MEFs were grown on coverslips coated with 0.1% gelatin and fixed with 4% paraformaldehyde (PFA) for 15 min. For pericentrin, cells were fixed with 4% PFA followed by MeOH for 15 min at −20°C. HeLa cells were fixed in ice-cold MeOH for 15 min at −20 °C Cells were blocked with 0.3% triton X-100, 10% donkey serum, 0.1% BSA in PBS (rosettes) or 0.3% triton X-100, 1.0% BSA, in PBS. Cells were incubated with primary antibodies for 2 hr at RT or overnight at 4°C and secondary antibodies for 1 hr at RT.

Mouse tissues were fixed in 4% PFA overnight and stored in 70% EtOH. Tissues were paraffinized, embedded, and sectioned at 5 µm intervals. Deparaffinization and antigen-retrieval were performed and sections were processed for immunostaining as previously described (Li et al., 2015). Primary Antibodies Used: α-tubulin (rabbit (rb), Abcam, 1:1000); β-tubulin (mouse (ms), Sigma); β-tubulin-647 (rb, Cell Signaling, 1:500); EB1 (rat, Abcam, 1:500.); γ-tubulin (ms, Sigma, 1:2000); GFP (ms, Abcam, 1:500); HMMR (rb, Abcam, 1:500; [Li et al., 2015]); NuMA (rb, Abcam, 1:500–1000); Par3 (rb, Millipore, 1:100); Pax6 (rb, Covance, 1:300); Pericentrin (rb, Covance, 1:500); pPLK1(Thr210) (rb, Abcam,1:200); PLZF (ms, EMD chemical, 1:100); Ran (ms, Cell Biolabs, 1:500), Ran (rb, Abcam, 1:50); Tbr2 (rb, Abcam, 1:100); TPX2 (rb, Novus, 1:500); ZO-1 (rb, Invitrogen, 1:1000); ZO-1 (ms, Invitrogen, 1:100). Secondary Antibodies Used: AlexaFluor 488, AlexaFluor 549, and AlexaFluor 647 (Invitrogen).

For TUNEL, samples were stained with the In Situ cell death detection kit, Fluorescein (Roche) following the manufacturer’s instructions.

Coverslips were mounted with Prolong Gold antifade reagent with DAPI (Invitrogen) and images were acquired with confocal microscopy (FluoView Fv10i, Olympus (Japan) or Axiovert 200, Zeiss (Germany)). Image analysis was performed using ImageJ.

Live imaging

For DHC-GFP HeLa, imaging was performed using a Perkin Elmer Ultraview VOX spinning disk confocal microscope using a Leica DMI6000 inverted microscope equipped with a Hamamatsu 9100–02 camera. Images were taken at two intervals. MEFs, Tet-HeLa cells, and eGFP-TUBA HeLa cells were imaged using an ImageXpress Micro High Content Screening System (Molecular Devices, Inc., Sunnyvale, CA) for up to 24 hr at 15 min intervals. Prior to imaging cells were stained with Hoechst. Image analysis was performed using ImageJ.

Virus packaging and transduction

HMMR constructs were delivered and expressed using the Gateway system (Invitrogen). Briefly, HEK293FT cells (Invitrogen R7007) were transfected using Lipofectamine 2000 (Invitrogen) with vectors containing, HMMRFL, HMMR1-623, or HMMR1-679, for 72 hr. Supernatant was collected and concentrated using Lenti-X concentrator (Clontech). Virus was added to MEFs with polybrene (8 μg/ml). After 24 hr, the media was replaced and cells grown for 24 hr prior to imaging.

Transfection

On-target plus siRNA (Dharmacon) and scrambled siRNA as previously described (Chen et al., 2014). pmCherry-C1-RanQ69L was a gift from Jay Brenman (Addgene plasmid # 30309) (Kazgan et al., 2010). pcDNA-RanWT-mRFP1-polyA was a gift from Yi Zhang (Addgene plasmid # 59750) (Inoue and Zhang, 2014). The mCherry-RanDN construct was graciously provided by Dr. Iain Cheeseman (Kiyomitsu and Cheeseman, 2013). Transfection of DNA and siRNA used JetPrime (Polyplus Transfection, France) following the manufacturer’s protocols. Cells were harvested 96 hr post-transfection of siRNA.

Ran activity assay

HeLa cells were treated with siRNA and synced with a double thymidine block and treated with MG132 for 2 hr starting 8 hr post-release. Ran activity assays were performed using the Ran Activation Assay kit (Cell Biolabs, Inc) as per manufacturer’s protocols.

Co-immunoprecipitation

DHC-GFP HeLa cells treated with siRNA and synced with a double thymidine block were homogenized with lysis buffer (Dunsch et al., 2012) with a phosphatase inhibitor (Phosphostop (Roche)) and Protease Inhibitor cocktail (Roche). HeLa cells were treated with indicated plasmids and synced with a nocodazole block and then released into mg-132. Cells were lysed with lysis buffer (25 mM HEPES, pH 7.5, 150 mM NaCl, 1% NP-40, 10 mM MgCl2, 1 mM EDTA, 2% glycerol). GTPγS was loaded prior to immunoprecepitation by adding EDTA (10 µM) and GTPγS at the indicated concentration for 30 min at 30°C. The reaction was stopped with MgCl2 (65 µM). The supernatants were used for immunoprecipitation with IgG (ms and rb, Sigma), GFP (ms, Abcam), HMMR (rb, Abcam), or pHMMR (Maxwell et al., 2011), overnight at 4°C, followed by incubation with protein A/G beads overnight at 4°C (Santa Cruz Biotechnology). Protein A/G beads were washed with lysis buffer three times. Bound proteins were separated by SDS-PAGE and analysed by western blotting. Actin (rb, Sigma, 1:2500); mCherry (rb, Abcam, 1:1000); HMMR (rb, Abcam, 1:500); GFP (ms, Abcam, 1:1000); DIC (ms, Millipore, 1:500); DynLL1 (rb Abcam, 1:1000); p150Glued (ms, Abcam, 1:500); Ran (ms, Cell Biolabs Inc, 1:1000); and RanBP2 (Abcam, 1:5000).

Mass spectrometry of protein complexes

Cells were lysed at 0.5–1.0 × 107 cells/ml in immunoprecipitation buffer (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1 mM EDTA, 0.5% NP-40) supplemented with protease and phosphatase inhibitors (Roche). Cell lysates were clarified by centrifugation at 16,000 X g for 10 min at 4°C and protein concentration was determined using the BCA protein assay kit (Thermo Fisher). For immunoprecipitation, lysates were precleared with protein G or A/G PLUS-Agarose beads (Santa Cruz). Protein complexes were isolated by incubation with the indicated antibodies at 4°C on rotation, and then with protein G or A/G PLUS-Agarose beads for 6 hr at 4°C on rotation. Isolated complexes were washed four times with lysis buffer.

Following IP, protein samples on beads were eluted twice with 50 µL of 100 mM citric acid, pH 2.6 at 50°C for 10 min shaking at 1300 rpm, followed by centrifugation, collection of the supernatant and neutralization with 125 µL of 1 M HEPES, pH 8.5. Proteins were reduced by adding 5 µL of 200 mM DTT and incubating at 37°C for 60 min, followed by alkylation by adding 10 µL of 400 mM IAA and incubation at room temperature for 60 min in the dark. The reaction was quenched by adding 10 µL of 200 mM DTT. Proteins were digested with Trypsin/Lys-C mix (Promega, Madison, WI) at an enzyme:protein ratio of 1:100 at 37°C for 16 hr. For stable isotope labeling by reductive dimethylation formaldehyde and heavy formaldehyde (C13D2O) was added to 40 mM final concentration to IgG control and HMMR IP samples, respectively. Sodium cyanoborohydride was added to a final concentration of 20 mM immediately after to both samples and incubated at 21°C for 60 min. Both conditions were combined, acidified to pH 2.5 with TFA, and the peptides purified with C18-STAGE tips as described (Rappsilber et al., 2003).

Liquid chromatography tandem mass spectrometry analysis was performed with the Easy nLC ultra-high-pressure LC system (Thermo Fisher Scientific) couple to a Q Exactive HF mass spectrometer with an EASY-Spray source. An EASY-Spray C18 column (Thermo-Fisher, 50 cm long, 75 µm inner diameter) heated to a temperature of 50°C was used for separation. Dried Stage-tip eluates were resuspended in 10 µL buffer A (0.1% FA) and 2 µL was used for injection. The peptides were loaded at a back pressure of 550 bar and separated with a gradient of 3–25% buffer B (0.1% FA in 80% ACN) over 105 min followed by 25–40% buffer B over 20 min at a flow rate of 300 nL/min. The chromatography method ends with a ramp from 40 to 100% buffer B over 3 min then a hold at 100% buffer B for 12 min. A column equilibration step using 11 µL buffer A was included prior to the next sample loading step.

MS data were acquired using a data-dependent top 12 method with a dynamic exclusion of 20 s. MS1 was performed at a resolution of 60,000 at m/z 200 with an AGC target of 3E6 and a maximum ion injection time of 75 ms over the m/z range 400 to 1800. HCD fragmentation of peptides was performed with an isolation range of 1.4 m/z and normalized collision energy set to 28. A resolution of 15,000 at m/z 200, an AGC target of 5E4 and a maximum ion injection time of 50 ms was set for fragment spectra acquisition.

Acquired spectra from two separate experiments and multiple injections were searched using Proteome Discoverer 2.1 (Thermo Fisher Scientific). Database search was performed against the Homo sapiens reference proteome including isoforms downloaded from UniProt in June 2016. Main search parameters: enzyme: Trypsin (full); missed cleavages: 2; precursor mass tolerance: 10 ppm; fragment mass tolerance: 0.02 Da; static modifications: +57.021 Da on C; variable modifications: +15.995 Da on M, +28.031 or +34.063 on K and peptide N terminus, +42.011 Da on protein N terminus. Identifications were filtered for 1% FDR at the peptide and protein level. Differential abundance of proteins between pHMMR and control IP was calculated based on the area of heavy and light dimethyl precursor peaks. Common contaminants and decoy identifications were filtered out. To identify proteins commonly identified in affinity purification experiments the identified proteins were searched against the CRAPOME database (www.crapome.org, version 1.1, H. sapiens). Proteins that were found in less than 30% of reported experiments were classified as ‘rare’, the reminder as ‘common’. Proteins not matched in the CRAPOME database were classified as ‘unknown’. Full Proteome Discoverer results are available in Figure 6—source data 1 and raw data is available through the PRIDE Archive.

Quantification of mitotic spindle orientation

Neural rosettes were stained with TPX2, ZO-1, and TUBB and images were acquired using confocal microscopy. In rosettes, spindle orientation was measured as the angle of the cleavage plane (anaphase or telophase cells) relative to the apical surface. For analysis of the division orientation of apical NP cells, E14.5 brain sections were stained for pH3 and γ-Tubulin. The long spindle axis of anaphase cells, defined by a line bisecting the two centrosomes, was used to indicate the cell division plane. For each progenitor, the angle between the long spindle axis and the apical surface (defined by a line along the centrosomes of apically localized interphase cells) was determined. In cultured MEFs, spindle orientation was measured between the long axis of G2 phase cells (15 mins prior to mitosis) and the angle of the mitotic spindle (anaphase cells).

Statistical methods

All replicates were biological replicates. Statistical analysis was performed using GraphPad Prism v5.01 for Windows (Graphpad Software, La Jolla, CA). Pairwise comparisons were made using two-tailed, paired or unpaired student’s t-test. Comparisons of multiple groups were made using one-way analysis of variance (ANOVA) with a Bonferroni post-test.

References

  1. 1
  2. 2
    The intracellular hyaluronan receptor RHAMM/IHABP interacts with microtubules and actin filaments
    1. V Assmann
    2. D Jenkinson
    3. JF Marshall
    4. IR Hart
    (1999)
    Journal of Cell Science 112 (Pt 22):3943–3954.
  3. 3
  4. 4
  5. 5
  6. 6
  7. 7
  8. 8
  9. 9
  10. 10
  11. 11
  12. 12
  13. 13
  14. 14
  15. 15
  16. 16
  17. 17
  18. 18
  19. 19
  20. 20
  21. 21
  22. 22
  23. 23
  24. 24
  25. 25
  26. 26
  27. 27
  28. 28
  29. 29
  30. 30
  31. 31
  32. 32
  33. 33
  34. 34
  35. 35
    Spindle misorientation of cerebral and cerebellar progenitors is a mechanistic cause of megalencephaly
    1. H Li
    2. T Kroll
    3. J Moll
    4. L Frappart
    5. P Herrlich
    6. H Heuer
    7. A Ploubidou
    (2017)
    Stem Cell Reports, 10.1016/j.stemcr.2017.08.013, 28943256.
  36. 36
  37. 37
  38. 38
  39. 39
  40. 40
  41. 41
  42. 42
  43. 43
    Interplay between BRCA1 and RHAMM regulates epithelial apicobasal polarization and may influence risk of breast cancer
    1. CA Maxwell
    2. J Benítez
    3. L Gómez-Baldó
    4. A Osorio
    5. N Bonifaci
    6. R Fernández-Ramires
    7. SV Costes
    8. E Guinó
    9. H Chen
    10. GJ Evans
    11. P Mohan
    12. I Català
    13. A Petit
    14. H Aguilar
    15. A Villanueva
    16. A Aytes
    17. J Serra-Musach
    18. G Rennert
    19. F Lejbkowicz
    20. P Peterlongo
    21. S Manoukian
    22. B Peissel
    23. CB Ripamonti
    24. B Bonanni
    25. A Viel
    26. A Allavena
    27. L Bernard
    28. P Radice
    29. E Friedman
    30. B Kaufman
    31. Y Laitman
    32. M Dubrovsky
    33. R Milgrom
    34. A Jakubowska
    35. C Cybulski
    36. B Gorski
    37. K Jaworska
    38. K Durda
    39. G Sukiennicki
    40. J Lubiński
    41. YY Shugart
    42. SM Domchek
    43. R Letrero
    44. BL Weber
    45. FB Hogervorst
    46. MA Rookus
    47. JM Collee
    48. P Devilee
    49. MJ Ligtenberg
    50. RB Luijt
    51. CM Aalfs
    52. Q Waisfisz
    53. J Wijnen
    54. CE Roozendaal
    55. DF Easton
    56. S Peock
    57. M Cook
    58. C Oliver
    59. D Frost
    60. P Harrington
    61. DG Evans
    62. F Lalloo
    63. R Eeles
    64. L Izatt
    65. C Chu
    66. D Eccles
    67. F Douglas
    68. C Brewer
    69. H Nevanlinna
    70. T Heikkinen
    71. FJ Couch
    72. NM Lindor
    73. X Wang
    74. AK Godwin
    75. MA Caligo
    76. G Lombardi
    77. N Loman
    78. P Karlsson
    79. H Ehrencrona
    80. A Wachenfeldt
    81. RB Barkardottir
    82. U Hamann
    83. MU Rashid
    84. A Lasa
    85. T Caldés
    86. R Andrés
    87. M Schmitt
    88. V Assmann
    89. K Stevens
    90. K Offit
    91. J Curado
    92. H Tilgner
    93. R Guigó
    94. G Aiza
    95. J Brunet
    96. J Castellsagué
    97. G Martrat
    98. A Urruticoechea
    99. I Blanco
    100. L Tihomirova
    101. DE Goldgar
    102. S Buys
    103. EM John
    104. A Miron
    105. M Southey
    106. MB Daly
    107. RK Schmutzler
    108. B Wappenschmidt
    109. A Meindl
    110. N Arnold
    111. H Deissler
    112. R Varon-Mateeva
    113. C Sutter
    114. D Niederacher
    115. E Imyamitov
    116. OM Sinilnikova
    117. D Stoppa-Lyonne
    118. S Mazoyer
    119. C Verny-Pierre
    120. L Castera
    121. A de Pauw
    122. YJ Bignon
    123. N Uhrhammer
    124. JP Peyrat
    125. P Vennin
    126. S Fert Ferrer
    127. MA Collonge-Rame
    128. I Mortemousque
    129. AB Spurdle
    130. J Beesley
    131. X Chen
    132. S Healey
    133. MH Barcellos-Hoff
    134. M Vidal
    135. SB Gruber
    136. C Lázaro
    137. G Capellá
    138. L McGuffog
    139. KL Nathanson
    140. AC Antoniou
    141. G Chenevix-Trench
    142. MC Fleisch
    143. V Moreno
    144. MA Pujana
    145. HEBON
    146. EMBRACE
    147. SWE-BRCA
    148. BCFR
    149. GEMO Study Collaborators
    150. kConFab
    (2011)
    PLoS Biology 9:e1001199.
    https://doi.org/10.1371/journal.pbio.1001199
  44. 44
    Receptor for hyaluronan-mediated motility correlates with centrosome abnormalities in multiple myeloma and maintains mitotic integrity
    1. CA Maxwell
    2. JJ Keats
    3. AR Belch
    4. LM Pilarski
    5. T Reiman
    (2005)
    Cancer Research 65:850–860.
  45. 45
  46. 46
  47. 47
  48. 48
  49. 49
  50. 50
  51. 51
  52. 52
  53. 53
  54. 54
  55. 55
  56. 56
  57. 57
  58. 58
  59. 59
  60. 60
  61. 61
  62. 62
  63. 63
  64. 64

Decision letter

  1. Iain M Cheeseman
    Reviewing Editor; Whitehead Institute, United States

In the interests of transparency, eLife includes the editorial decision letter and accompanying author responses. A lightly edited version of the letter sent to the authors after peer review is shown, indicating the most substantive concerns; minor comments are not usually included.

Thank you for submitting your article "HMMR localizes active Ran at the centrosome to orient cell division and support neural development" for consideration by eLife. Your article has been reviewed by three peer reviewers, and the evaluation has been overseen by a Reviewing Editor and Anna Akhmanova as the Senior Editor. The following individual involved in review of your submission has agreed to reveal his identity: Iain M Cheeseman (Reviewer #1).

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

Summary:

In this paper, the authors investigate the role of the HMMR protein in mitotic spindle positioning and orientation. The authors use multiple systems (mutant mouse, ES derived rosettes, fibroblasts derived from mutant mice, HeLa cells, neuroblastoma cells and Drosophila) to demonstrate the function of HMMR in spindle orientation. Although the role of HMMR in spindle orientation is not new, the authors have identified mechanism by which HMMR controls spindle positioning in cells. They find that reducing or eliminating HMMR perturbs spindle positioning and alters the cortical localization of both dynein and NuMA. They provide evidence that this occurs by HMMR associating with Ran to alter this signaling pathway.

Essential revisions:

The reviewers are in agreement that this paper is interesting, but feel that there are several things that are necessary to improve the paper to the point where it would be acceptable for publication in eLife. The primary suggestions relate to strengthening the mechanistic aspects of the work.

1) Your paper suggests an important role for spindle pole-localized HMMR in signaling to the cell cortex to promote proper spindle positioning and regulating DHC and NuMA localization. Prior work has suggested that Plk1 also acts as a spindle-localized signal to control spindle position. Does this suggest that HMMR impacts Plk1 activity as well? It would be very useful to consider the relationship between these two pathways and compare the effects from individually or simultaneously targeting these pathways:

• It is important test whether HMMR affects Plk1 signaling. The reviewers recommend that you combine Plk1 inhibition or cortically tethering with HMMR perturbations for the experiments in Figures 5 and 6.

• The authors show that HMMR centrosome localization domain is important for its function in spindle orientation. This is a nice part of the study, and the authors use many different techniques to decipher this mechanism. However, as they admit in the Discussion, the relative importance of HMMR-Ran-centrosomal pathway with respect to others involving Plk1 and DYNLL1 is not determined; making less clear what is the relevance of these results. The authors may want to dissect whether Ran centrosome localization alone is important for spindle orientation by modulating centrosomal Ran levels independently of HMMR and by altering DYNLL1 and PlK1 expression and activity in the HMMR overexpression experiments.

2) It would be helpful to more carefully evaluate the role of Ran in this process using Ran mutants (constitutively active or dominant negative). For example, does a Ran constitutively active mutant phenocopy HMMR overexpression? Alternatively, does Ran inhibition suppress the effects of HMMR overexpression? For the work on Ran, it would provide a useful addition if you could provide data to clarify the interaction and nature of the relationship between HMMR and Ran (direct interaction, through RanBP1, does this affect Ran activity, etc.):

• siHMMR cells still deplete NuMA from the lateral cortex, suggesting this chromosomal Ran effect is functional. The authors might consider including additional siRNA conditions, such as against Rcc1 alone or in combination with HMMR, to compare HMMR to loss of activated Ran. A discussion of how HMMR discriminates potential subcellular pools of Ran as well as how it may or may not interfere with the Plk1-Dynein regulatory loop would be beneficial.

• To further establish the link between the HMMR-Ran centrosomal pathway described in vitro and in vivo observations (and to confirm that the HMMR-centrosomal Ran pathway controls spindle orientation to regulate neurogenesis), the authors should show localization of Ran to centrosomes in vivo (the fact that they study this in neuroblastoma cells is not sufficient).

• It is also important to consider alternative explanations for the observed phenotypes beyond altered Ran signaling.

- HMMR loss is shown to amplify centrosome number in MEFs (although a discussion of how this might occur is not presented). Have the authors considered that this centrosome amplification may contribute to spindle orientation defects (e.g. see Kulukian et al. PNAS 2015) independent of any HMMR-Ran connection?

- To what extent is the orientation pathway described in vitro responsible for orientation defects in vivo? Polarity defects could affect spindle orientation upstream of the force localizing machinery (which is different from the mechanism the authors propose). Spindle density, mitotic defects and centrosome amplification phenotypes should be also considered when analyzing /stating the mechanism controlling spindle orientation. Are Astral MT affected? Is mitosis length increased? In spindle pictures in vivo and in ES derived rosettes, spindle density and size seem affected. The authors should characterize these features (in vivo and /or in cells). Generally it would be useful to analyze polarity markers, centrosome amplification, spindle morphology in vivo (or at least in rosettes).

3) The reviewers also recommend that you trim aspects of your analysis to focus on the novel and substantive new insights. All three reviewers expressed confusion for the diverse data and systems, some of which was not directly relevant to the overall conclusions. The authors could do a better job of integrating this data and experimental systems. The reviewers recommend removing some of the data from the work in Drosophila cells and other experiments that are not directly related to the main points of the paper, although we leave this to your discretion.

• In its current version, it seems that this paper is constructed so that it can accommodate a large amount of data obtained in different systems. The authors try to squeeze together a neurodevelopmental biology study and a cell biology study, and the resulting message is confused and confusing. The paper is difficult to read, and it needs a major overhaul.

• The inclusion of the Drosophila experiments seems unnecessary as there is no complementation of the Miranda phenotype (perhaps not surprising given the evolutionary difference), and they only use this to see cortical blebbing. If the authors want to include this data, they could alter this section to minimize the analysis of the complementation (perhaps a single sentence).

4) The in vivo analysis with the mouse model would benefit from some additional consideration. For the revised paper, you should use caution in the interpretation of the mouse knockout phenotypes and avoiding redundancy with other recently published work to highlight your novel discoveries:

• Previous work by Li et al. also generated a HMMR mouse mutant. There are a number of similarities, but also differences between these two papers. In the previous case, the authors only eliminated the C-terminal region of the protein. Here, the authors argue that the lethality of their mutant (but not the previous mouse model) is related to functions of the N-terminus. However, the prior paper also analyzed spindle orientation with apparently similar observations. It would be helpful to further clarify the differences and similarities between the two papers.

• Although the systems are used somewhat interchangeably, in some cases they are not used to their full potential. The power of this paper to generate a potent knockout for HMMR in mice. However, they do not use cells derived from these mice for most of the subsequent cell biological experiments. Instead, they deplete the protein by RNAi in HeLa cells, which may be both less efficient and also prone to off target defects (no rescue is shown).

• The authors state that HMMR effect is "pleiotropic" and are careful not to conclude that the observed developmental defects derive strictly from spindle orientation defects. However a large part of the Introduction is based on the idea that spindle orientation dictates cell fates in the developing CNS, and somehow directs the reader towards this interpretation. This is actually a very dogmatic view: although many published studies claim a causal relationship, they only show a correlation between strong neurogenesis defects (that may be caused by defective migration, cell death, or many other factors) and often minor defects in spindle orientation. As the authors later show in different HMMR mutant contexts, there are actually many defects:

- increased cell death,

- probable aneuploidy,

- centrosome mispositioning (described in rosettes),

- centrosome overduplications (described in MEFs)

- mitotic cell death (in Figure 5)which are as many possible causes of microcephaly. If all these defects occur also in vivo (which they do not document), then one cannot conclude that neurogenic and brain formation defects are specifically generated by spindle misorientation itself. Indeed, centrosome amplification has been linked to microcephaly independently of spindle orientation (Marthiens et al., 2013). Centrosome positioning defects in rosettes suggest that polarity may be affected, and indeed the authors cite defects in a-b in mammary epithelia. Polarity defects can affect the architecture of the neuroepithelium, affecting indirectly neuroepithelium structure and neurogenesis, and it would be necessary to investigate a-b cell polarity in mutant brains.

[Editors' note: further revisions were requested prior to acceptance, as described below.]

Thank you for resubmitting your work entitled HMMR acts in the PLK1-dependent spindle positioning pathway and supports neural development" for further consideration at eLife. Your revised article has been favorably evaluated by Anna Akhmanova (Senior editor), a Reviewing editor, and three reviewers.

The manuscript has been much improved, but there are some remaining issues that need to be addressed before acceptance, as outlined below:

All three reviewers expressed an appreciation for the changes to the manuscript, including the addition of a variety of new experiments, the removal of some unnecessary or distracting data, and changes to the text. However, there are some issues that were not resolved to the reviewer's satisfaction, and other concerns that arose due to the new changes to the paper. For submitting a revised version, please focus on the following key points:

1) Rework of the text to provide a coherent model for HMMR function. For the first submission of the paper, the primary focus was on HMMR playing a role together with Ran at the spindle poles. Due to the new data presented in the revised version, the emphasis changed to propose a shared pathway for HMMR and Plk1. This is a quite significant change, and although you have changed the title, this has not yet achieved an integrated model or narrative in other sections of the paper. It would be very helpful to provide clarity on how you perceive the role and function of HMMR in this context throughout the paper. Although much of this could likely be addressed through changes to the text, experiments distinguishing between a function together with Ran vs. a function with Plk1 would further strengthen the paper.

2) Use appropriate caution in the interpretation of the presented experiments. There are several points at which you make strong and definitive statements that appear to go beyond the nature of the presented data. This includes the co-IP Western blotting experiments, the Ran/RanBP1 binding assays, and the Ran centrosome localization (as examples). Balancing your interpretation and conclusions may also help with point 1 above for creating an appropriate overall model.

3) An additional point that arose during the reviewer discussion was the removal of the data for the localization of GFP fusions to HMMR wt and mutant forms rather than providing improved images (localization was not clear in the prior version). If this is robust and reproducible, it would be good to include these images instead of referencing a prior paper. If this is not feasible, please provide an explanation.

The reviewer comments are included below for your reference and for additional suggestions.

Reviewer #1:

This paper has a couple of key appeals to me. First, it uses both genetic perturbations in mouse models and cell biology in human tissue culture cells to investigate spindle positioning in diverse contexts. Second, by analyzing the role of HMMR in the established spindle positioning pathways (including dynein/NuMA, Plk1, and Ran-based mechanisms), this can help provide insights into the overall control of this process. For this revised version, the authors have conducted substantial additional experimentation, and have also made a number of changes to the text and figures (for example, removing the Drosophila data as requested by the reviewers). There are some nice additions to the paper, including the analysis of Plk1. Although this work represents a step forward, I am left with the feeling that this story is not fully definitive or comprehensive. The overall take home messages have changed substantially since the first submission, necessitating a change to the title and some of the core findings. In this case, instead of proposing that HMMR is acting through Ran, now the authors are proposing Plk1 as a primary factor. At the same time, there is still confounding data arguing for an interaction with Ran (through RanBP1), but this is ill-defined. Although some of the individual datapoints are intriguing, I think that they are far away from being able to make definitive statements like "Thus, PLK1 phosphorylation of HMMR (pHMMR) enables an interaction with RanBP2." or "PLK1 activity and HMMR, which binds Ran-GTP, are required to locate constitutively active Ran at mitotic centrosomes."

The nature of the advance and such definitive statements is further complicated by the fact that many of the data and images show relatively modest effects (such as the change in the Ran-GTP interaction with RanBP1 when HMMR is depleted – looks like a blotting issue – or the effect on RanCA localization in Figure 4E/G and Ran in Figure 7D – quantification looks like a modest change at best). Similarly, I don't understand mechanistically why HMMR depletion would lead to a reduction of DYNLL1 or dynactin with DHC. These type of co-IP Western experiments are quite sensitive and need to be carefully controlled. In this case, there is not a realistic explanation for this effect. The effect in cells looks more like a secondary effect on spindle structure to me rather than a direct effect on dynein interactions (they don't show dynein spindle localization, for example).

As much as I like the overall concepts and ideas in this paper, I am concerned that the substantive advances in this paper will be compromised by distracting data that does not rise to this level and that their model and conclusions currently being complicated by trying to include all of this regardless of significance. I would also strongly urge the authors to reconsider their overall model and to use caution in making strong and incontrovertible statements.

Reviewer #2:

The new version of the manuscript is much improved. The new focus on the Plk1/HMMR/Ran axis is much clearer, and the paper, now stripped from too much "satellite" data, reads much better.

Reviewer #3:

The revised manuscript by Connell, et al. largely resolves the concerns raised with the initial submission. I would now recommend for publication.

https://doi.org/10.7554/eLife.28672.026

Author response

Essential revisions:

The reviewers are in agreement that this paper is interesting, but feel that there are several things that are necessary to improve the paper to the point where it would be acceptable for publication in eLife. The primary suggestions relate to strengthening the mechanistic aspects of the work.

1) Your paper suggests an important role for spindle pole-localized HMMR in signaling to the cell cortex to promote proper spindle positioning and regulating DHC and NuMA localization. Prior work has suggested that Plk1 also acts as a spindle-localized signal to control spindle position. Does this suggest that HMMR impacts Plk1 activity as well? It would be very useful to consider the relationship between these two pathways and compare the effects from individually or simultaneously targeting these pathways:

• It is important test whether HMMR affects Plk1 signaling. The reviewers recommend that you combine Plk1 inhibition or cortically tethering with HMMR perturbations for the experiments in Figures 5 and 6.

HMMR does impact PLK1 activity. In the Introduction for the revised manuscript, we have referenced published studies linking HMMR with PLK1 activity, including that silencing HMMR reduces PLK1 activity at kinetochores during spindle assembly (Chen et al., Cell Cycle 2014) and that HMMR-T703 is a substrate for PLK1 as identified in a mitotic phosphoproteome (Nousiainen et al. PNAS, 2006).

These published findings, and the reviewer’s recommendations, founded our new MS analysis of proteins immunoprecipitated with antibodies against phosphorylated HMMR-T703 (pHMMR), which discovered RANBP2 (Figure 4A; Figure 4—figure supplement 1; subsection “HMMR interacts with RanBP2 and regulates the centrosome localization of Ran”, first paragraph). We have performed multiple new experiments and we found that silencing HMMR dampens the intensity of phosphorylated PLK1 at spindle poles (Figure 6D-E) and that small-molecule inhibition of PLK1 results in similar phenotypes as observed in HMMR-silenced cells, including a reduction in Ran co-localization with TUBG1 (Figure 4G-H), reduced astral microtubule density (Figure 6F-G), and reduced cortical NuMA localization (Figure 6H-I). We tested if the concurrent inhibition of HMMR and PLK1 significantly altered the spindle positioning pathway when compared to the inhibition of either alone and found that the localization of NuMA was not additively altered in dual inhibited cells (Figure 6H-I). We now conclude that HMMR acts in the PLK1-dependent spindle positioning pathway, and have modified the title of our article accordingly.

• The authors show that HMMR centrosome localization domain is important for its function in spindle orientation. This is a nice part of the study, and the authors use many different techniques to decipher this mechanism. However, as they admit in the Discussion, the relative importance of HMMR-Ran-centrosomal pathway with respect to others involving Plk1 and DYNLL1 is not determined; making less clear what is the relevance of these results. The authors may want to dissect whether Ran centrosome localization alone is important for spindle orientation by modulating centrosomal Ran levels independently of HMMR and by altering DYNLL1 and PlK1 expression and activity in the HMMR overexpression experiments.

We have performed new experiments to dissect the importance of the HMMR-Ran-centrosomal pathway with respect to others involving PLK1 or DYNLL1. Our new pHMMR IP/MS data discovered RANBP2 as a putative interaction that potentially placed the HMMR-RANBP2-Ran localization downstream of PLK1 activity. Consistently, the deficits in spindle positioning that followed PLK1 inhibition are similar to those that follow HMMR silencing and concurrent disruption of both did not have an additive effect. As outlined in our first response to comment 1 (above) and as we now summarize in Figure 7—figure supplement 2A, we believe HMMR acts with PLK1 to locate Ran and modulate the centrosome-localized, spindle positioning pathway.

HMMR-CHICA-DYNLL1 is proposed to strip cortical NuMA-DHC in a PLK1-independent manner however a definitive mechanism was not described (Dunsch et al., 2012). It was proposed that HMMR-CHICA anchored DYNLL1 at the spindle pole and the consequent gradient of DYNLL1 brought proximal to the cortex with an off-center spindle pole altered the composition of cortical dynein complexes by competing with dynactin, which is needed for retention on the cortex (Dunsch et al., 2012). In a set of new experiments (Figure 6A-C), we tested two predictions derived from this putative mechanism: (1) We IP’d DHC-GFP complexes from HMMR-silenced cells and found that co-precipitated DYNLL1 was reduced, which is consistent with the putative mechanism. However, we found that co-precipitated p150glued was also reduced. (2) We localized CHICA and DYNLL1 in HMMR-silenced spindles, and found their localization to spindle fibers was reduced but both were localized at spindle poles. Thus, we conclude that silencing HMMR does alter the composition of DHC-GFP complexes and reduces the incorporation of DYNLL1; however, the incorporation of p150glued is also reduced and thus HMMR-DYNLL1 may not impede dynein-dynactin complex formation. HMMR also appears dispensable for DYNLL1 and CHICA localization to the spindle poles although it does impact their localization to spindle fibers. Together, these data support a conclusion that HMMR acts in the PLK1-dependent spindle positioning pathway.

2) It would be helpful to more carefully evaluate the role of Ran in this process using Ran mutants (constitutively active or dominant negative). For example, does a Ran constitutively active mutant phenocopy HMMR overexpression? Alternatively, does Ran inhibition suppress the effects of HMMR overexpression? For the work on Ran, it would provide a useful addition if you could provide data to clarify the interaction and nature of the relationship between HMMR and Ran (direct interaction, through RanBP1, does this affect Ran activity, etc.):

Results from a new experiment demonstrate that the expression of a Ran constitutively active mutant does phenocopy the effect of HMMR overexpression on the cortical localization of NuMA (Figure 8 F-G). As recommended in Essential revision 3 (see below), we have removed the work in Drosophila to focus the study. The removal of the Drosophila work also led to the removal of the observation of cortical blebbing resulting from the ectopic expression of HMMR in neuroblasts; thus, we removed our analysis of cortical blebbing in tet-HMMR cells, as well. So, we did not examine if Ran inhibition suppresses the effects of HMMR overexpression in tet-HMMR cells. In new IP- MS analysis, we discover and validate a pHMMR-RANBP2 interaction to clarify the nature of the relationship between HMMR and active Ran (Figure 4A).

• siHMMR cells still deplete NuMA from the lateral cortex, suggesting this chromosomal Ran effect is functional. The authors might consider including additional siRNA conditions, such as against Rcc1 alone or in combination with HMMR, to compare HMMR to loss of activated Ran. A discussion of how HMMR discriminates potential subcellular pools of Ran as well as how it may or may not interfere with the Plk1-Dynein regulatory loop would be beneficial.

In the revised manuscript, we have discussed the role of HMMR in the PLK1-dynein regulatory loop and contrasted that role with the putative mechanism of action published concerning the HMMR-CHICA-DYNLL1 pathway (Discussion, seventh, eighth paragraphs). We have discussed the assignment of Ran in the centrosome proteome, the role of HMMR in co-localizing Ran with TUBG1, and our observation that HMMR is dispensable for the Ran-GTP gradient generated at chromosomes (Discussion, ninth paragraph). A connection between HMMR and RCC1 in the control of the generation of Ran-GTP at chromosomes is beyond the scope of this manuscript but is a provocative concept given the central role that HMMR plays in chromosome-dependent spindle assembly (Groen et al., 2004) (Joukov et al. 2006) (Chen et al., 2014) (Scrofani et al., 2015), which is RCC1-Ran-GTP dependent.

• To further establish the link between the HMMR-Ran centrosomal pathway described in vitro and in vivo observations (and to confirm that the HMMR-centrosomal Ran pathway controls spindle orientation to regulate neurogenesis), the authors should show localization of Ran to centrosomes in vivo (the fact that they study this in neuroblastoma cells is not sufficient).

In new studies, we examined spindle position, spindle fiber density, and Ran co-localization with TUBG1 in neuroblasts lining the ventricles in Hmmrtm1a/tm1a or Hmmr+/+E14.5 brains (Figure 7A-E).

• It is also important to consider alternative explanations for the observed phenotypes beyond altered Ran signaling.

- HMMR loss is shown to amplify centrosome number in MEFs (although a discussion of how this might occur is not presented). Have the authors considered that this centrosome amplification may contribute to spindle orientation defects (e.g. see Kulukian et al. PNAS 2015) independent of any HMMR-Ran connection?

- To what extent is the orientation pathway described in vitro responsible for orientation defects in vivo? Polarity defects could affect spindle orientation upstream of the force localizing machinery (which is different from the mechanism the authors propose). Spindle density, mitotic defects and centrosome amplification phenotypes should be also considered when analyzing /stating the mechanism controlling spindle orientation. Are Astral MT affected? Is mitosis length increased? In spindle pictures in vivo and in ES derived rosettes, spindle density and size seem affected. The authors should characterize these features (in vivo and /or in cells). Generally it would be useful to analyze polarity markers, centrosome amplification, spindle morphology in vivo (or at least in rosettes).

We thank the reviewers for this series of excellent observations. In the revised manuscript, we discuss alternate mitotic or post mitotic processes that are disturbed in HMMR-silenced cells (usually cancer cells) as additional explanations for observed phenotypes in Hmmrtm1a/tm1a mice (Discussion, third paragraph). As discussed (Discussion, second paragraph), we focused our studies on mitotic processes, rather than non-mitotic processes, given our observations that HMMR is expressed in cyclin-positive, apically positioned r-NSCs during the process of neural differentiation (Figure 7—figure supplement 1D-I). We now include the findings from many additional experiments to address each of the reviewer’s comments in the following ways:

- Polarity markers appear unaffected in vivo (Figure 2—figure supplement 1B).

- Centrosome amplification: We assessed centrosome numbers, and the presence/absence of multipolar spindles, in WT and KO NP cells and found no significant difference (Figure 2—figure supplement 1C).

- Spindle morphology in vivo: we examined spindle position, spindle length and spindle fiber density in neuroblasts lining the ventricles in Hmmrtm1a/tm1a or Hmmr+/+E14.5 brains (Figure 7A-E).

- Astral microtubules are affected in HMMR-silenced cells (Figure 6F-G). We were not able to resolve astral microtubule density in vivo.

- The length of mitosis is augmented in Hmmrtm1a/tm1a MEFs and HMMR-silenced cells, which is noted in the s paragraph of the Discussion as unpublished data. These unpublished data are described in a separate manuscript, entitled Chen et al. The non-motor adaptor HMMR dampens Eg5-mediated forces to preserve the kinetics and integrity of chromosome segregation (submitted, Mol Biol Cell).

3) The reviewers also recommend that you trim aspects of your analysis to focus on the novel and substantive new insights. All three reviewers expressed confusion for the diverse data and systems, some of which was not directly relevant to the overall conclusions. The authors could do a better job of integrating this data and experimental systems. The reviewers recommend removing some of the data from the work in Drosophila cells and other experiments that are not directly related to the main points of the paper, although we leave this to your discretion.

• In its current version, it seems that this paper is constructed so that it can accommodate a large amount of data obtained in different systems. The authors try to squeeze together a neurodevelopmental biology study and a cell biology study, and the resulting message is confused and confusing. The paper is difficult to read, and it needs a major overhaul.

• The inclusion of the Drosophila experiments seems unnecessary as there is no complementation of the Miranda phenotype (perhaps not surprising given the evolutionary difference), and they only use this to see cortical blebbing. If the authors want to include this data, they could alter this section to minimize the analysis of the complementation (perhaps a single sentence).

We have constructed a more cohesive story by removing the Drosophila data and the related data addressing cortical blebbing and daughter cell size in tet-Hmmr Hela cells. We have also removed many aspects of the neural rosette data. Finally, we have focused on the novel aspects of our work by contrasting our findings from our new mouse model with the findings obtained in published Hmmr mutant mouse models (Li et al., 2015, 2016) (Tolg et al., 2003) and recent work from Xenopus embryos treated with HMMR morpholinos (Prgaer et al., 2017).

4) The in vivo analysis with the mouse model would benefit from some additional consideration. For the revised paper, you should use caution in the interpretation of the mouse knockout phenotypes and avoiding redundancy with other recently published work to highlight your novel discoveries:

• Previous work by Li et al. also generated a HMMR mouse mutant. There are a number of similarities, but also differences between these two papers. In the previous case, the authors only eliminated the C-terminal region of the protein. Here, the authors argue that the lethality of their mutant (but not the previous mouse model) is related to functions of the N-terminus. However, the prior paper also analyzed spindle orientation with apparently similar observations. It would be helpful to further clarify the differences and similarities between the two papers.

As is noted by the reviewer in the next comment, the Hmmrtm1a/tm1amouse model is the more potent knockout for Hmmr in mice so we do not believe there is redundancy between the findings from our work and the published Hmmr mice. We do note the work from Li et al. (2015, 2016) in the Introduction and we discuss the similarities and differences in the Discussion. In particular, we highlight in the discussion the recently published work in Xenopus HMMR morphants that indicate a critical role for the N-terminus of the protein in neural development.

• Although the systems are used somewhat interchangeably, in some cases they are not used to their full potential. The power of this paper to generate a potent knockout for HMMR in mice. However, they do not use cells derived from these mice for most of the subsequent cell biological experiments. Instead, they deplete the protein by RNAi in HeLa cells, which may be both less efficient and also prone to off target defects (no rescue is shown).

We performed many studies in siHMMR-treated HeLa cells, rather than Hmmr KO MEFs, due to: (1) MEFs have a low mitotic rate, which makes them suboptimal for mechanistic studies; and, (2) many of the seminal observations describing the spindle positioning pathway in mammalian cells were obtained using HeLa cells making this model an optimal one for comparison with the published literature.

With respect to rescue in siHMMR-treated HeLa cells, we have previously published that the reduced kinetics of spindle assembly observed in siHMMR-treated HeLa cells can be rescued through the expression of GFP-HMMR (Chen et al., 2014), which argues against the effects observed here (using the same constructs) being the result of off-target effects. In addition, our findings describe consistent phenotypes (changes to ran localization, spindle structure and spindle position) in redundant models for HMMR-deletion (mice, Mefs, HMMR-silenced HeLa).

• The authors state that HMMR effect is "pleiotropic" and are careful not to conclude that the observed developmental defects derive strictly from spindle orientation defects. However a large part of the Introduction is based on the idea that spindle orientation dictates cell fates in the developing CNS, and somehow directs the reader towards this interpretation. This is actually a very dogmatic view: although many published studies claim a causal relationship, they only show a correlation between strong neurogenesis defects (that may be caused by defective migration, cell death, or many other factors) and often minor defects in spindle orientation. As the authors later show in different HMMR mutant contexts, there are actually many defects:

- increased cell death,

- probable aneuploidy,

- centrosome mispositioning (described in rosettes),

- centrosome overduplications (described in MEFs)

- mitotic cell death (in Figure 5)which are as many possible causes of microcephaly. If all these defects occur also in vivo (which they do not document), then one cannot conclude that neurogenic and brain formation defects are specifically generated by spindle misorientation itself. Indeed, centrosome amplification has been linked to microcephaly independently of spindle orientation (Marthiens et al., 2013). Centrosome positioning defects in rosettes suggest that polarity may be affected, and indeed the authors cite defects in a-b in mammary epithelia. Polarity defects can affect the architecture of the neuroepithelium, affecting indirectly neuroepithelium structure and neurogenesis, and it would be necessary to investigate a-b cell polarity in mutant brains.

These are excellent suggestions that overlap with aspects of those in the last criticism of point 2. As outlined in our response to that criticism, we have performed additional experiments to investigate these putative effects. With specific note to polarity defects, we looked at Par3 and ZO-1 as apical markers and their localization appears to be unaffected in vivo (Figure 2—figure supplement 1B).

[Editors' note: further revisions were requested prior to acceptance, as described below.]

The manuscript has been much improved, but there are some remaining issues that need to be addressed before acceptance, as outlined below:

All three reviewers expressed an appreciation for the changes to the manuscript, including the addition of a variety of new experiments, the removal of some unnecessary or distracting data, and changes to the text. However, there are some issues that were not resolved to the reviewer's satisfaction, and other concerns that arose due to the new changes to the paper. For submitting a revised version, please focus on the following key points:

1) Rework of the text to provide a coherent model for HMMR function. For the first submission of the paper, the primary focus was on HMMR playing a role together with Ran at the spindle poles. Due to the new data presented in the revised version, the emphasis changed to propose a shared pathway for HMMR and Plk1. This is a quite significant change, and although you have changed the title, this has not yet achieved an integrated model or narrative in other sections of the paper. It would be very helpful to provide clarity on how you perceive the role and function of HMMR in this context throughout the paper. Although much of this could likely be addressed through changes to the text, experiments distinguishing between a function together with Ran vs. a function with Plk1 would further strengthen the paper.

The Abstract, the presentation and discussion of our findings, and the model have been edited to clarify that silencing HMMR results in dampened PLK1 activity and a disturbed PLK1-dependent positioning pathway. We moved the “HMMR interacts with RanBP2 and regulates the centrosome localization of Ran”so that it follows the “HMMR enables the PLK1-dependent spindle pole positioning pathway” section to emphasize that HMMR acts through the PLK1 positioning pathway. As a putative mechanistic explanation for how silencing HMMR may dampen the PLK1-dependent spindle positioning pathway, we discuss the literature that demonstrates HMMR enables the formation of activating Aurora A-TPX2 complexes (Chen et al., Cell Cycle 2014) (Vernos and colleagues, Curr Biol 2015) and that active Aurora A directly phosphorylates PLK1-T210 (Medema and colleagues, Nature 2008). We modified the model to clarify the mechanism.

The presentation and discussion of our findings and the model have been edited to clarify that pHMMR locates active Ran at centrosomes in a PLK1-dependent manner. We highlight in the concluding paragraph of our Discussion that the relative contribution of pHMMR function with Ran vs. a function with active PLK1, potentially through HMMR-mediated modulation of Aurora A, warrants further study.

2) Use appropriate caution in the interpretation of the presented experiments. There are several points at which you make strong and definitive statements that appear to go beyond the nature of the presented data. This includes the co-IP Western blotting experiments, the Ran/RanBP1 binding assays, and the Ran centrosome localization (as examples). Balancing your interpretation and conclusions may also help with point 1 above for creating an appropriate overall model.

We appreciate the significance of this criticism. In our revised manuscript, we adjusted the interpretations and conclusions drawn from our findings, especially as they relate to the manner and significance of HMMR-Ran complexes at mitotic spindle poles.

3) An additional point that arose during the reviewer discussion was the removal of the data for the localization of GFP fusions to HMMR wt and mutant forms rather than providing improved images (localization was not clear in the prior version). If this is robust and reproducible, it would be good to include these images instead of referencing a prior paper. If this is not feasible, please provide an explanation.

We added improved images to Figure 3—figure supplement 1.

The reviewer comments are included below for your reference and for additional suggestions.

Reviewer #1:

[…] Although this work represents a step forward, I am left with the feeling that this story is not fully definitive or comprehensive. The overall take home messages have changed substantially since the first submission, necessitating a change to the title and some of the core findings. In this case, instead of proposing that HMMR is acting through Ran, now the authors are proposing Plk1 as a primary factor. At the same time, there is still confounding data arguing for an interaction with Ran (through RanBP1), but this is ill-defined. Although some of the individual datapoints are intriguing, I think that they are far away from being able to make definitive statements like "Thus, PLK1 phosphorylation of HMMR (pHMMR) enables an interaction with RanBP2." or "PLK1 activity and HMMR, which binds Ran-GTP, are required to locate constitutively active Ran at mitotic centrosomes."

We altered these statements to more precisely interpret our presented findings.

The nature of the advance and such definitive statements is further complicated by the fact that many of the data and images show relatively modest effects (such as the change in the Ran-GTP interaction with RanBP1 when HMMR is depleted – looks like a blotting issue – or the effect on RanCA localization in Figure 4E/G and Ran in Figure 7D – quantification looks like a modest change at best).

We noted the effects are modest in the text. The effects, while modest, were reproducible (not a blotting issue) and we include immunoblots from our replicate experiments (Author response image 1, corresponds to Figure 6C in revision #2).

Author response image 1

Similarly, I don't understand mechanistically why HMMR depletion would lead to a reduction of DYNLL1 or dynactin with DHC. These type of co-IP Western experiments are quite sensitive and need to be carefully controlled. In this case, there is not a realistic explanation for this effect. The effect in cells looks more like a secondary effect on spindle structure to me rather than a direct effect on dynein interactions (they don't show dynein spindle localization, for example).

We clarified the proposed mechanism in the text. Briefly, Barr and colleagues, J Cell Biol 2012 demonstrated that a HMMR-CHICA complex at the spindle/poles contributed to the positioning pathway, potentially in a PLK1-independent manner. HMMR-CHICA was shown to recruit DYNLL1 to the spindle/ poles and proposed to create a local increase of DYNLL1 activity. Consistently, we identified DYNLL1 and CHICA in pHMMR IPs by MS analysis and we found that the DYNLL1-DHC interaction was reduced in HMMR-silenced cells.

Barr and colleagues further speculated that the local increase of DYNLL1 caused DYNLL1- dynein interactions that resulted in loss of dynein from the cell cortex by competing with dynactin: “a prediction of this model is that the DYNLL1-bound form of dynein cannot bind to dynein adaptors required for cortical targeting” and “Therefore, DYNLL1 may compete with dynactin for binding sites on the dynein intermediate chain, and thereby displace dynein complexes from the cell cortex.” A prediction of this model would be that cortical dynein is retained in HMMR-silenced cells because dynactin is incorporated and not displaced by DYNLL1. While we found the DYNLL1-DHC interaction was reduced in HMMR-silenced cells, we do not find evidence for augmented dynactin-DHC complexes. In fact, we find that dynactin-DHC complexes are reduced. We agree that these co-IP experiments are quite sensitive. For this reason, we had 2 technicians independently verify the effect (Author response image 2, corresponds to Figure 5A in rev. #2).

Author response image 2

As much as I like the overall concepts and ideas in this paper, I am concerned that the substantive advances in this paper will be compromised by distracting data that does not rise to this level and that their model and conclusions currently being complicated by trying to include all of this regardless of significance. I would also strongly urge the authors to reconsider their overall model and to use caution in making strong and incontrovertible statements.

We have modified the model, and the text throughout, to highlight that HMMR feeds into the intrinsic PLK1 positioning pathway. We have also highlighted the fact that while we localize active Ran at spindle poles, and that its localization is reduced by the silencing of HMMR or the inhibition of PLK1, the relative contribution of active Ran at the spindle pole to the regulation of cortical NuMA localization is not yet clear.

https://doi.org/10.7554/eLife.28672.027

Article and author information

Author details

  1. Marisa Connell

    Department of Paediatrics, University of British Columbia, Vancouver, Canada
    Contribution
    Conceptualization, Formal analysis, Investigation, Writing—original draft, Writing—review and editing
    Contributed equally with
    Helen Chen
    Competing interests
    No competing interests declared
    ORCID icon 0000-0002-7594-2137
  2. Helen Chen

    Department of Paediatrics, University of British Columbia, Vancouver, Canada
    Contribution
    Conceptualization, Formal analysis, Investigation, Writing—review and editing
    Contributed equally with
    Marisa Connell
    Competing interests
    No competing interests declared
  3. Jihong Jiang

    Department of Paediatrics, University of British Columbia, Vancouver, Canada
    Contribution
    Formal analysis, Investigation, Writing—original draft, Writing—review and editing
    Competing interests
    No competing interests declared
  4. Chia-Wei Kuan

    Department of Pathology and Laboratory Medicine, University of British Columbia, Vancouver, Canada
    Contribution
    Investigation, Writing—review and editing
    Competing interests
    No competing interests declared
  5. Abbas Fotovati

    Department of Paediatrics, University of British Columbia, Vancouver, Canada
    Contribution
    Investigation, Writing—review and editing
    Competing interests
    No competing interests declared
  6. Tony LH Chu

    Department of Paediatrics, University of British Columbia, Vancouver, Canada
    Contribution
    Investigation, Writing—review and editing
    Competing interests
    No competing interests declared
  7. Zhengcheng He

    Department of Paediatrics, University of British Columbia, Vancouver, Canada
    Contribution
    Investigation, Writing—review and editing
    Competing interests
    No competing interests declared
  8. Tess C Lengyell

    Centre for Molecular Medicine and Therapeutics, University of British Columbia, Vancouver, Canada
    Contribution
    Investigation, Writing—review and editing
    Competing interests
    No competing interests declared
  9. Huaibiao Li

    Leibniz Institute on Aging—Fritz Lipmann Institute, Beutenbergstrasse, Germany
    Contribution
    Investigation, Writing—review and editing
    Competing interests
    No competing interests declared
    ORCID icon 0000-0003-4086-3321
  10. Torsten Kroll

    Leibniz Institute on Aging—Fritz Lipmann Institute, Beutenbergstrasse, Germany
    Contribution
    Investigation, Writing—review and editing
    Competing interests
    No competing interests declared
  11. Amanda M Li

    Department of Paediatrics, University of British Columbia, Vancouver, Canada
    Contribution
    Investigation, Writing—review and editing
    Competing interests
    No competing interests declared
  12. Daniel Goldowitz

    1. Centre for Molecular Medicine and Therapeutics, University of British Columbia, Vancouver, Canada
    2. Department of Medical Genetics, University of British Columbia, Vancouver, Canada
    Contribution
    Investigation, Writing—review and editing
    Competing interests
    No competing interests declared
  13. Lucien Frappart

    Leibniz Institute on Aging—Fritz Lipmann Institute, Beutenbergstrasse, Germany
    Contribution
    Investigation, Writing—review and editing
    Competing interests
    No competing interests declared
  14. Aspasia Ploubidou

    Leibniz Institute on Aging—Fritz Lipmann Institute, Beutenbergstrasse, Germany
    Contribution
    Supervision, Investigation, Writing—review and editing
    Competing interests
    No competing interests declared
  15. Millan S Patel

    Department of Medical Genetics, University of British Columbia, Vancouver, Canada
    Contribution
    Investigation, Writing—review and editing
    Competing interests
    No competing interests declared
  16. Linda M Pilarski

    Cross Cancer Institute, Department of Oncology, University of Alberta, Edmonton, Canada
    Contribution
    Resources, Writing—review and editing
    Competing interests
    No competing interests declared
  17. Elizabeth M Simpson

    1. Centre for Molecular Medicine and Therapeutics, University of British Columbia, Vancouver, Canada
    2. Department of Medical Genetics, University of British Columbia, Vancouver, Canada
    Contribution
    Resources, Investigation, Writing—review and editing
    Competing interests
    No competing interests declared
  18. Philipp F Lange

    1. Department of Pathology and Laboratory Medicine, University of British Columbia, Vancouver, Canada
    2. Michael Cuccione Childhood Cancer Research Program, BC Children’s Hospital, Vancouver, Canada
    Contribution
    Supervision, Investigation, Writing—review and editing
    Competing interests
    No competing interests declared
  19. Douglas W Allan

    Department of Cellular and Physiological Sciences, Life Sciences Centre, University of British Columbia, Vancouver, Canada
    Contribution
    Resources, Supervision, Writing—review and editing
    Competing interests
    No competing interests declared
  20. Christopher A Maxwell

    1. Department of Paediatrics, University of British Columbia, Vancouver, Canada
    2. Michael Cuccione Childhood Cancer Research Program, BC Children’s Hospital, Vancouver, Canada
    Contribution
    Conceptualization, Resources, Supervision, Funding acquisition, Writing—original draft, Writing—review and editing
    For correspondence
    cmaxwell@bcchr.ubc.ca
    Competing interests
    No competing interests declared
    ORCID icon 0000-0002-0860-4031

Funding

Michael Cuccione Foundation

  • Marisa Connell
  • Helen Chen
  • Christopher A Maxwell
  • Philipp F Lange

Canadian Breast Cancer Foundation

  • Tony Chu

BC Children's Hospital Research Institute

  • Zhengcheng He
  • Philipp F Lange
  • Christopher A Maxwell

Canada Research Chairs

  • Philipp F Lange

Michael Smith Foundation for Health Research

  • Philipp F Lange

Canadian Institutes of Health Research (OBC 134038)

  • Christopher A Maxwell

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

Acknowledgements

We thank Dr. Iain Cheeseman (MIT) for reagents, the UBC Bioimaging Facility, The Centre for Phenogenomics (Toronto, Canada), and the FLI Histology and Imaging Core Facilities. This work was funded by: Michael Cuccione Foundation (studentship, fellowship, or salary awards to HC, MC, PFL, or CAM), Canadian Breast Cancer Foundation (PhD fellowship to TC), BC Children’s Hospital Research Institute (studentship to ZH, and salary awards to PFL and CAM), Canada Research Chairs (salary award to PFL), Michael Smith Foundation for Health Research (salary award to PFL), and Canadian Institutes of Health Research (OBC 134036 to CAM; salary award to CAM).

Ethics

Animal experimentation: All procedures involving animals were in accordance with the Canadian Council on Animal Care (CCAC) and UBC Animal Care Committee (ACC) (Protocol no. A13-0168).

Reviewing Editor

  1. Iain M Cheeseman, Reviewing Editor, Whitehead Institute, United States

Publication history

  1. Received: May 16, 2017
  2. Accepted: October 5, 2017
  3. Accepted Manuscript published: October 10, 2017 (version 1)
  4. Version of Record published: November 10, 2017 (version 2)

Copyright

© 2017, Connell 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

  • 623
    Page views
  • 138
    Downloads
  • 0
    Citations

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

Comments

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)