Differences in nuclear and nuclear body morphology and relative positioning among four cell types.

(A-B) Wide-field deconvolution light microscopy (A) and 3D solid models in XY and XZ orientations (B). Immunostaining of nuclear periphery (NP), nucleoli, and nuclear speckles using antibodies against RL1 (nuclear pore, purple), MKI67IP (nucleolar GC, orange), and SON (nuclear speckle, green) in H1, K562, HCT116, and HFF cells. (C) Comparing Volume (V), surface area (SA), and roundness (SA/V) of nuclei measured from NP 3D solid models in H1 (red), K562 (blue), HCT116 (green), and HFF (yellow) cells. (D) Comparison across cell lines of nucleolar and nuclear speckle (NS) numbers, average volumes, and summed volumes per nucleus. Individual NS are significantly larger in HCT116 cells, even though total NS volumes are similar across all four cell lines. (E) Pairwise average distances between locales (asymmetric, as defined in SFig. 1B). (F) Principal Component (PC) 1 (x-axis) versus PC2 (y-axis) scatterplot using PCA of 14 morphological features reveals adjacent clustering of H1 and K562 cells with HCT116 and HFF clusters closer to each than to H1 and K562 cells. Each scatterplot point represents an individual cell. Arrow lengths show the quality of representation and arrow direction show the loadings on PC1 and PC2.

Global changes in nuclear genome intranuclear positioning and their correlations with changes in gene expression and DNA replication timing.

(A) Chr1 left arm browser view suggests chromosome trajectory alternating between positioning of early replicating regions near nuclear speckles and late-replicating regions at either the nucleolar (H1, K562) or the nuclear periphery (HCT116, HFF); Top to bottom-nuclear speckle (SON, green), Nucleolus (MKI67IP, orange), and nuclear lamina (LMNB1, purple) TSA-seq, LMNB1 (purple) DamID, and 2-fraction Repli-seq (Replication timing, brown). Left to right-H1, K562, HCT116, and HFF cells. (B) Comparison of Pearson correlations of TSA-seq and DamID datasets between different cell lines (H1,K562,HCT116, HFF) reveals much higher conservation of genome positioning relative to nuclear speckles, which is largely conserved, versus nucleoli or nuclear lamina; (C) Left-Genes positioned closer (blue)/ further (red) to nuclear speckles show a bias towards increased / decreased gene expression in HFF versus H1 cells. Gene fraction (x-axis) versus log2 FKPM (HFF/H1) (y-axis); Right-Similar comparison for genes positioned closer (blue) / further (red) to nuclear lamina does not show such bias; (D-E) Changes in gene expression between H1 and HFF cells vary largely as function of changes in nuclear speckle rather than nuclear periphery or nucleolar relative positioning; (D) 2D histograms showing mean ratio changes in gene expression between H1 versus HFF cells (log2 ratios of FKPM, color-coded) for binned genes as function of their changes in z-normalized TSA-seq scores (left to right: LMNB1 (y) vs SON (x), MKI67IP (y) vs SON (x), MKI67IP (y) vs LMNB1 (x)); (E) Linear modeling of changes in gene expression versus z-normalized TSA-seq score changes reveals significantly larger coefficients (dependence) for SON (green) versus MKI67IP (orange) or LMNB1 (blue). (F-G) Similar comparison as in (D-E), but for replication timing (2-fraction Repli-seq), shows changes in DNA replication timing are function of both changes in SON and LMNB1 TSA-seq.

Varying gene composition, DNA replication timing, and speckle proximity of Type I versus Type II SON TSA-seq peaks which align with gene expression “hot-zones” and DNA replication initiation zones (IZs).

(A) SON TSA-seq Type I (red ticks) and Type II (blue ticks) local maxima (“peaks”) align with DNA early replication IZs identified in 16-fraction (S1 (early) – S16 (late)) Repli-seq. Top to bottom: K562 SON TSA-seq, HCT116 SON TSA-seq and Repli-seq, H1 SON TSA-seq and Repli-seq, HFF SON TSA-seq; (B-C) Histograms of distances (x-axis, nm) (B) and boxplots showing speckle association fractions (<250 nm) (C) of HFF Type I and II peaks located in IMR90 fibroblasts [43] from nuclear speckles show higher and unimodal (Type I) versus lower and bimodal (Type II) nuclear speckle attachment frequencies ; (D) Pileup plots showing SON TSA-seq (left), Repli-seq (2nd to left), FKPM RNA-seq (2nd from right), and Hi-C compartment scores (right) flanking (+/- 500 kb) Type I (red) versus Type II (blue) SON TSA-seq peaks; (E) Boxplots of Trep (timing of replication, left) and Twidth (variation in replication timing, right) in H1 (top) and HCT116 (bottom) show earlier and less variable DNA replication timing for Type I versus II peaks; (F) Pileup Repli-seq profiles for early, early-mid, late-mid, and late IZs (left, +/- 1 Mbp) show progressively later timing of replication correlating with the lower amplitude of SON TSA-seq peaks (right) centered at the IZ center (right).

Facultative LADs transition most frequently to a partially repressed, middle-to-late replicating iLAD and less frequently to an active, early replicating facultative iLAD in different cell types.

(A) Chr1 region with examples of LADs (red rectangles) showing peaks in LMNB1 DamID in one cell type changing in other cell types to facultative iLADs (fiLADs) showing either valleys (“v”) (blue rectangles) or peak-within-valleys (“p-w-v”) (yellow-orange rectangles) in their LMNB1 DamID signals. Replication timing is late for fLADs, changes to early for v-fiLADs but remains late-to-middle for p-w-v fiLADs, despite Hi-C A compartment scores for both p-w-v and v-fiLADs. Top to bottom - LMNB1 DamID for H1, HCT116, K562, and HFF, Hi-C A/B Eigenvector score for H1 and HCT116, 16-fraction Repli-seq for H1 and HCT116, and 2-fraction Repli-Seq for K562 and HFF (H1-red; HCT116-green; K562-blue; HFF-brown) (some v-fiLADs and iLADs are p-w-v fiLADs classified as v-fiLADs or missed by our classification scheme); (B-C) Numbers (B) and mean sizes (C) of LADs (red), p-w-v fiLADs (yellow/orange), and v-fiLADs (blue) in the four cell types. HFF cells have smallest number of p-w-v and v fiLADs; (D) Genes in same domain type show similar expression mean levels but show increased expression in p-w-v fiLADs versus LADs or v fiLADs versus p-w-v fiLADs in H1 versus HCT116. Log2 (FKPM(H1)+1 / FKPM(HCT116)+1) for genes in one type of domain in H1 and the same or another type of domain in HCT116 (H1 domain type : HCT116 domain type); (E) Pileup plots for LMNB1 DamID (top), E/L 2-fraction Repli-seq (middle), and Hi-C compartment score (bottom) for LADs (red), p-w-v fiLADs (yellow/orange), v fiLADs (blue) in H1 (left) versus HCT116 (right) reveals overall trend towards higher LMNB1 DamID signal, less early replication timing, and lower Hi-C compartment score of p-w-v fiLADs versus v fiLADs which resemble flanking iLAD regions; (F) 16-fraction Repli-seq pileup plots of LADs (top), p-w-v fiLADs (middle), and v fiLADs (bottom) confirms late, middle, versus early DNA replication patterns, respectively, for these domains in both H1 (left) and HCT116 (right).

LADs shift towards nuclear interior but peak-within-valley (p-w-v) fiLADs shift towards nuclear lamina and replicate later in LMNA/LBR double knockout (DKO) K562 cells.

(A) Representative deconvolved widefield images showing wildtype (WT) and DKO K562 cells. Scale bar = 1 μm. DNA staining (DAPI, grey) shows increased numbers of condensed chromatin foci (arrowheads) at nucleolar (orange) periphery as well as within the nuclear interior in DKO (middle and bottom rows) versus WT (top row) nuclei. DKO nuclei also are rounder and less lobular than wt nuclei; (B) Browser views over Chr1 left arm showing differences in (top to bottom) LMNB1 DamID, LMNB1, MKI67IP (nucleolar), and SON (nuclear speckle) TSA-seq, and 2-fraction Repli-seq (Early (E) / Late (L)) for WT (black) versus DKO (red) cells. Whereas LADs shift towards the nuclear interior, p-w-v fiLADs shift towards the nuclear lamina, or become actual LADs, with slightly later replication timing; in contrast, there is no significant change in nuclear speckle positioning. (C) Chromosome shifts away from (towards) nuclear lamina are inversely correlated with shifts towards (away from) nucleoli: 2D-histogram showing changes (KO – wt) in LMNB1 (x-axis) versus MKI67IP (y-axis) TSA-seq. For each histogram bin (pixel), the number of overlapping 25 kb genomic bins is grey-scale coded. Only pixels with at least 10 overlapping genomic bins are plotted. Pearson correlation = -0.58; (D-E) H3K9me3-enriched regions shift away from nuclear periphery and towards nucleoli whereas H3K27me3-enriched regions shift towards nuclear periphery and away from nucleoli in DKO versus wt cells; 2D histograms from (C) overlaid with mean wt H3K9me3 (D) or H3K27me3 (E) ChIP-seq color-coded values. (F) Upset plot showing numbers of differentially expressed genes in each of the 3 KO cell lines and overlap of these differentially expressed genes between KO lines; (G) No trend in gene expression changes as a function of increased or decreased nuclear lamin association after DKO: Scatterplot of gene expression differences (log2-ratio FKPM, y-axis) versus differences in LMNB1 DamID (x-axis) (DKO-wt). Only active genes are shown (blue, upregulated; red, downregulated; grey, no change); (H) Domains that shift closer to nuclear lamina in DKO cells also shift to later DNA replication: Scatterplot comparing mean changes over chromosome domains in E/L 2-fraction Repli-seq (DKO – wt) (y-axis) versus changes in LMNB1 TSA-seq (DKO – wt) (x-axis). Domain data points are colored based on their transition class (see text) (wt LADs, p-w-v fiLADs, and v-fiLADs, and DKO fLADs).

Spatial segregation within nuclei of different heterochromatin types revealed by LMNB1 and SON TSA-seq.

(A) Genomic ROI (ROIs C1 and C2 in SFig. 6) in H1 cells with disproportionately reduced nuclear lamina interactions (orange) which deviate from otherwise inverse correlation (green) between SON and LMNB1 TSA-seq. H1 SON(x-axis) versus LMNB1 (y-axis) TSA-seq scatterplot; (B) Mean TSA-seq LMNB1, MKI67IP, SON values for H1 ROI in HFF, HCT116, K562, and H1 cells (top to bottom) reveals decreased nuclear lamina but increased nucleolar proximity in H1 cells; (C) Most H1 ROIs are located within several Mbp of centromeres; distance (Mbp) boxplots of H1 ROIs versus other regions; (D) Higher enrichment of H3K9me3 versus other histone marks (mean ChIP-seq percentile values) in H1 ROIs in H1, HCT116, HFF, K562 cell lines; (E) Subdivision of HCT116 LAD genomic bins (100 kb) into four, color-coded clusters (C1-4) based on their SON and LMNB1 TSA-seq scores; light green points are iLAD bins; (F) LAD bins in C1-4 clusters show different histone marks; mean percentile ChIP-seq values of indicated histone marks for the C1-4 LAD clusters and iLADs; (G) H3K9me3, H3K27me3, H3K9me2, and H2AFZ (H2A.Z) (left to right) enriched regions show differential nuclear localization relative to nuclear speckles and lamina roughly paralleling C1-C4 clusters; SON versus LMNB1 TSA-seq 2D histograms of LADs. The color-code represents the average histone modification levels in each bin (iLADs, green points); (H-I) C1-4 LAD clusters vary functionally. C1 especially but also C3 LAD bins have lower gene expression, later DNA replication timing (Trep), and more uniform replication timing (Twidth) than C2 and C4 LAD bins. iLAD bins have highest gene expression, earliest replication timing, but are less variable in replication timing than C2 and C4 LAD bins. Boxplot distributions for log2(FKPM) (H), Trep (I, left), and Twidth (I, right); (J) cLAD segregate spatially differentially from fLADs largely based on their greater distance to nuclear speckles (iLADs, green). Color-coded cLAD score (# cell lines out of 7 in which a HCT116 LAD bin maps within a LAD) superimposed on SON versus LMNB1 TSA-seq scatterplot.

Polarity of nuclear genome organization.

(A-C) LMNB1 TSA-seq provides a readout of nuclear genome polarity in flat nuclei due to the diffusion radius of tyramide free-radicals, identifying a LAD subset preferentially localizing at the nuclear equatorial plane periphery. (A) nuclear pore immunostaining from HFF nucleus convolved with the 3D exponential decay function of TSA staining (pseudo-colored intensity) predicts higher biotin-labeling of lamina-associated chromatin lying in nuclear equatorial plane versus top or bottom of nucleus. Top (x-y cross section); bottom (x-z cross section). Pixel size = 80 nm; (B) 2D color-coded histogram showing number of LAD genomic bins with given SON (x-axis) and LMNB1 (y-axis) TSA-seq mean values in HFF. Rectangular boxes show selection of Cluster 1, with low to moderate LMNB1 and low SON TSA-seq, and Cluster 2, with the highest LMNB1 TSA-seq values, LAD bins; (C) Cluster 1 LADs preferentially localize to the equatorial plane of the nuclear periphery. Heat map showing FISH probe location density over many fibroblast (IMR90) nuclei (data from [43]) from Cluster 1 (left) versus Cluster 2 (right) LADs superimposed on normalized nuclear shape. Top: x-y projection; bottom: x-z projection; (D) Distal chromosome arms also preferentially localize to equatorial plane. Same as (C) but using all the FISH probes mapped to Chr4 and color-coded by distance (averaged over probes) to centromere in Mbp; (E) Distance to IMR90 nucleus x-y center also varies with chromosome distance from centromeres: IMR90 multiplex FISH probes distance (y-axis) [43] from nuclear center (projected x-y plane distances) as a function of Chr5 position (x-axis). Grey-interquartile range of probe distances; blue-mean probe distance; red-smoothed mean probe distance. Bottom track-red rectangle marks centromere position, green rectangles mark Cluster 1 LADs.; (F and G) Overall nuclear genome polarization in fibroblasts (IMR90); (F) Boxplots of mean distance to nuclear center (projected x-y plane distance) for all probes (NA), or Type I (red) or Type II (blue) SON TSA-seq peaks; (G) Boxplots of mean distance to equatorial plane for SON TSA-seq peaks, facultative versus constitutive LADs, and SPIN states. Strong versus moderate bias of Type I (red) versus Type II (blue) HFF SON TSA-seq peaks, respectively, to locate close to x-y nuclear center (F) and near to the equatorial plane of nuclear interior (G, left panel). Facultative LADs (low cLAD scores) localize closer to equatorial plane than constitutive LADs (high cLAD scores) (G, middle panel). SPIN states show progressive trend of increased mean distances to equatorial plane from “speckle” to” interior active” to “interior repressed” and “near lamina 1”, to “near lamina 2” and “lamina” SPIN states (Wang, 2021) (G, right panel).

New insights into nuclear genome organization revealed by nuclear locale genome mapping.

Human nucleus schematic (middle) shows nuclear speckles (NS, green), nucleolus (orange), and nuclear lamina (NL, purple). (A) Changes in the transcriptional activity of genes (pink to purple gradient) primarily correlates with changes in their distance to nuclear speckles (green), rather than nuclear lamina or nucleolus, whereas changes in DNA replication timing (pink-purple-yellow gradient) correlates with distance to all three nuclear locales. Regions close to nuclear speckles replicate early, while regions close to nucleolus and/or nuclear lamina replicate late during S-phase. (B) Two types of speckle association domains (SPADs), with similarly elevated levels of gene expression but genes of different lengths and relative intron/exon fraction, associate with NS. Type-I SPADs, with shorter, more exon-rich genes, have higher NS association frequencies compared to Type-II SPADs, with longer, more intron-rich genes. (C) fLADs can assume at least three different chromatin states across different cell types: i) a B-compartment LAD (positive lamina DamID signal) with high NL association, low gene expression levels and late DNA replication (cell type 1); ii) an A-compartment p-w-v-fiLAD (peak-within-valley lamina DamID signal) with low NL association, intermediate gene expression and middle-to late DNA replication (cell type 2); iii) an A-compartment v-fiLAD (valley lamina DamID signal) with no NL association, high gene expression and early DNA replication (cell type 3). (D) In cells with flat nuclei, the genome is differentially organized relative to the nuclear equatorial plane: facultative heterochromatin regions are enriched near the NL within the equatorial plane, while constitutive heterochromatin regions are enriched near the NL towards the top or bottom of the nucleus. Nuclear speckles localize towards the nuclear equatorial plane and towards the nuclear center, with Type I SPADs closer than Type II SPADs to the equatorial plane and nuclear center. (E) Type I and II SPADs align with DNA replication initiation zones with the earliest-firing DNA replication initiation zones mapping to Type I SPADs and the earliest DNA replication foci appearing adjacent to NS. (F) LADs and p-w-v fiLADs compete for NL versus nucleoli association: in a LMNA/LBR double knockout cell line, LADs shift from the NL towards nucleoli and the nuclear interior while p-w-v fiLADs shift towards the NL.

Comparisons of nuclear and nuclear body morphologies and the intranuclear relative positioning of nuclear bodies across four cell types. (A) Nuclear periphery segmentation was achieved by first segmenting the anti-RL1 nuclear pore signal, which approximated point sources, followed by fitting alpha-shapes to define the convex hull fitting this staining. We chose nuclear pore rather than nuclear lamina staining to mitigate the “missing cone” of spatial frequency along the optical axis which has the effect of attenuating signals from flat surfaces such as the nuclear lamina top and bottom surfaces [51]; (B) Schematic illustrating asymmetric distance metric: for example, NP (nuclear periphery) to NS (nuclear speckle) distance measurements refer to the nearest distance to a nuclear speckle from each NP surface point, whereas NS to NP distance measurements refer to the nearest distance to the NP from NS; (C) Inter-locale distance distributions-pairwise distances between NP, NS, and NUC (nucleolus)-compared across cell types; (D) Scatterplot comparisons between cell types (H1, red; K562, blue; HCT116, green, HFF, yellow) of different morphological metrics (Volume, vol; Surface Area, SA; Distance, dist; Flatness; Nucleolus, Nuc.; Nuclear speckle, NS; Nuclear periphery, NP); each scatterplot dot represents measurements from a single nucleus. (E) Percentage of explained variances (y-axis) for PC1-6 (x-axis) using Principal Component Analysis (PCA) of 14 morphological features from the immunofluorescence (IF) staining; (F) Principal Component (PC) 1 (x-axis) versus PC3 (y-axis) scatterplot reveals overall morphological similarity of H1 (red) and K562 (blue) nuclei with extensive overlap in both PC1 and PC3, with HCT116 and HFF nuclei varying from H1 and K562 in their lower PC1 values and HCT116 nuclei varying in their higher PC3 values from the nuclei in the other three cell lines. Each point in the scatterplot represents a cell in principal component space. Arrow length shows the quality of representation and arrow direction show the loadings on PC1 and PC2.

Global changes in nuclear genome intranuclear positioning and their correlations with changes in gene expression and DNA replication timing.

(A) Chr1 left arm (0-60 Mbp) showing heterochromatin chromatin domains (orange highlights) which associate near nucleoli and away from the nuclear lamina in H1 and K562 cells but near the nuclear lamina in HCT116 and HFF cells. These heterochromatin regions are flanked by speckle-associated regions (green highlights) in all cell lines, suggesting a Chr1 left arm trajectory alternating between speckles and nucleoli in H1 and K562 cells versus between speckles and the lamina in HCT116 and HFF cells. Top to bottom: SON (green), MKI67IP (brown), and LMNB1 (blue) TSA-seq repeated for H1, K562, HCT116, and HFF cells; (B) Nuclear speckle-associated regions align with nucleolar TSA-seq peaks in K562 cells but nucleolar TSA-seq valleys in HCT116 cells. Top to bottom: SON (green), MKI67IP (brown), and LMNB1 (blue) TSA-seq repeated for K562 versus HCT116 cells; (C) Chromosome-wide pattern of increased LMNB1 TSA-seq peak amplitudes of LADs towards the ends of long chromosome arms (orange highlights) versus decreased peak amplitudes for centromere-proximal LADs (blue highlights); these differences are enhanced in HCT116 and HFF versus H1 and K562 cells. Top to bottom: Chr2 LMNB1 TSA-seq for H1 (red), K562 (blue), HCT116 (green), HFF (orange) over Chr2; (D) Regions shifted closer to nuclear speckles in HFF versus H1 cells on average are shifted away from the nuclear lamina, but not vice versa. Top: Peak in the summed SON TSA-seq signals over chromosome regions with significantly higher SON TSA-seq values in HFF versus H1 cells correlates with a valley in the summed LMNB1 TSA-seq signals for these same regions. Bottom: peak in summed LMNB1 TSA-seq for regions significantly closer to the nuclear lamina in HFF versus H1 cells but little change in summed SON TSA-seq for these same regions; (E) Scatterplots showing overall inverse relationship between changes in SON versus LMNB1 TSA-seq in H1 versus HFF cells for chromosome regions with different positions relative to nuclear speckles (top) but small changes in SON TSA-seq for chromosome regions with different positions relative to nuclear lamina (bottom). Top-percentile-normalized SON versus LMNB1 TSA-seq for genomic bins from chromosome regions shifting closer to nuclear speckles in HFF cells. Despite an overall inverse correlation trend, there is a widespread in changed LMNB1 scores. Additionally, some bins show positive correlation between SON and LMNB1 TSA-seq. Bottom-Corresponding scatterplot for regions that shift significantly closer to nuclear lamina in HFF cells shows small changes in SON TSA-seq; (F) Linear modeling of changes in gene expression versus z-normalized TSA-seq score changes reveals significantly larger coefficients (dependence) for changes in SON (green) versus MKI67IP (orange) or LMNB1 (blue) TSA-seq, as modeled for genes in all genomic regions, in LADs only, in iLADs only, or in Type I versus II SON TSA-seq local maxima. “True” versus “False” tests for statistical significance are at a p=0.05 threshold.

Varying gene composition, DNA replication timing, and speckle proximity of Type I and II SON TSA-seq peaks which align both with gene expression “hot-zones” and DNA replication initiation zones (IZs).

(A) Boxplots of p-value of multimodality for Type I and II peaks (The Hartigan’s dip test) supports the unimodal versus bimodal distance distributions (Fig. 3B) of Type I versus II SON TSA-seq peaks, respectively, from nuclear speckles; (B) Boxplots showing different gene composition for Type I versus II SON TSA-seq peaks. Left to right: Gene length, exon length, intron length, and intro length/ exon length ratio; (C) Jaccard index shows greater conservation for Type I (red) versus II (blue) SON TSA-Seq peaks across the four cell lines; (D) Fraction of gene alleles transcriptionally active (x-axis) versus mean distance to nuclear speckles (y-axis) for Type I (red) versus II (blue) peaks, identified in HFF cells, from IMR90 cell FISH data [43] shows higher “ON” fraction for Type I genes closer to nuclear speckles; (E) Repli-seq fraction distribution of Type I (top) versus II (bottom) SON TSA-seq peaks in H1 (left) and HCT116 (right); (F) Live-cell imaging reveals earliest PCNA replication foci cluster near nuclear speckles in HCT116 cells, consistent with Type I SON TSA-seq peaks corresponding to IZs with the earliest replication timing. Left: Mid-nuclear optical z-sections at varying times (mins) after S-phase initiation defined by first appearance of PCNA foci showing PCNA foci (red, mCherry-PCNA) versus nuclear speckles (green, EGFP-SON); Right: Boxplots showing PCNA foci distance (y-axis) distributions from nuclear speckles at different times (mins) after S-phase began (x-axis). Top: statistics for nucleus 1; Bottom: statistics for foci from 5 nuclei. “X”= median. Scale bar = 2 μm.

Facultative LADs transition most frequently to a partially repressed, middle-to-late replicating iLAD and less frequently to an active, early replicating iLAD in different cell types.

(A) Chr1 distal left arm in H1 (red tracks) or HCT116 (green tracks) with examples of LADs (red rectangles) with peaks in LMNB1 DamID signals (top) in one cell type changing in other cell type to facultative iLADs (fiLADs) with either valley (“v”) (blue rectangles) or peak-within-valley (“p-w-v”) (yellow/orange rectangles) LMNB1 DamID signal (top). p-w-v fiLADs appear either as peak-within-valleys or valleys in the LMNB1 TSA-seq (2nd from top). LADs appear as peak-within-valleys whereas p-w-v fiLADs appear frequently as peaks in MKI67IP (nucleolar) TSA-seq (3rd from top). SON TSA-seq (bottom) over these regions show local minimums which often are deeper for LADs versus p-w-v fiLADs; (B) Local differences in LMNB1 DamID (y-axis) over chromosome domain versus flanking regions are shown for fiLADs (left) versus LADs (right, red); a cutoff corresponding to the lowest 5% of LADs was used to segment v-fiLADs (blue) (below this cutoff) from p-w-v fiLADs (olive); (C) LAD numbers (y-axis) and their transitions to p-w-v fiLADs (olive) or v-fiLADs (blue) between H1, K562, HCT116, and HFF cell types; (D) Comparisons of gene expression levels in different domain types in the same and different cell lines. LADs and p-w-v fiLADs show similar ranges of expression levels, with median expression actually slightly lower in p-w-v fiLADs versus LADs in H1 cells. Compared to p-w-v fiLAD genes, genes in v-fiLADs show a higher range of gene expression levels which in turn is lower than the range of gene expression levels in iLADs. (E) Boxplots for ratios of gene expression levels (log2((FKPM(cell type 1) +1)/FKPM(cell type 2)+1)) for genes in one type of chromosome domain in one cell type versus another type of chromosome domain in a second cell type; (F) Heat map summary of DamID, TSA-seq, ChIP-seq, and Repli-seq signals over different chromatin domain types (left) and the relative signal differences comparing domains with their flanking regions.

LADs shift towards nuclear interior but peak-within-valley (p-w-v) fiLADs shift towards the nuclear lamina and replicate later after LMNA/LBR double knockout (DKO). (A)

Comparisons over Chr1 left arm between: (Top to bottom) LMNB1 DamID signals in wildtype (wt) K562 cells, parental clone expressing CRISPR Cas9 fused to a destabilization domain, LBR knockout (KO) clone, LMNA KO clone, and LBR/LMNA DKO clone. Black box at right end of plots shows centromere region. (B) Comparisons between wt (blue tracks) and DKO (orange tracks) over left arm Chr1 for: (Top to bottom) LMNB1 DamID, 2-fraction Repli-seq (Early / Late (E/L) ratio, and LMNB1, MKI67IP, and SON TSA-seq. Rectangles highlight specific domain classes in wt K562 defined by comparison of cell types (see text and Fig. 4) (LADs, red; p-w-v fiLADs, orange; v-fiLADs, blue). Also shown are “DKO fLADs” (green rectangles), a domain class defined as wt iLADs, not previously defined by cell type comparisons, which become LADs in the DKO; (C) Numbers of LADs in DKO that arise from wt LADs, p-w-v fiLADs, and v-fiLADs, and “DKO fLADs” which arise from wt iLADs; (D) Fraction of wt LADs, p-w-v fiLADs, and v-fiLADs that are LADs in DKO; (E) No trend in LAD gene expression changes as a function of increased or decreased nuclear lamin association: scatterplot of gene expression differences (log2-ratio FKPM, y-axis) versus differences in LMNB1 DamID (x-axis). Only active LAD genes are shown (blue, upregulated; red, downregulated; grey, no change); (F) Generation of a matched set of iLAD genes. LAD genes are expressed at lower expression levels than iLAD genes (left panel). To compare LAD and iLAD genes, we selected a matching set of iLAD genes with similar gene expression to the set of LAD genes (right panel). This was repeated 100 times for a robust result; (G) Fractions (y-axis) of differentially expressed genes in iLADs (grey) versus LADs (black) (defined in wt cells) in LBR KO (left), LMNA KO (middle), or DKO (right); (H) (Left) Differences in E/L Repli-seq ratios (y-axis) versus changes in LMNB1 TSA-seq for individual genomic bins corresponding to LADs in wt and/or DKO cells. Curve fitting through data points reveals a trend for later DNA replication timing for LAD genomic bins which shift closer to nuclear lamina in DKO; (Right) Scatterplot showing log2-ratios of gene expression levels (y-axis) versus changes in LMNB1 TSA-seq for individual genes in LADs within wt and/or DKO cells. No consistent trend for changes in gene expression in DKO versus wt cells is observed; (I) Scatterplot comparing mean changes over chromosome domains in E/L 2-fraction Repli-seq (DKO – wt) (y-axis) versus changes in LMNB1 Dam-ID (DKO – wt) (x-axis) further supports trend towards later DNA replication timing for LADs, p-w-v fiLADs, and v-fiLADs which show increased association with nuclear lamina in DKO (similar to Fig. 5H comparing E/L Repli-seq to LMNB1 TSA-seq).

LMNB1 and SON TSA-seq reveal differential intranuclear spatial segregation of centromeric and pericentromeric regions and LADs in different cell types.

Further dividing LMNB1 versus SON TSA-seq scatterplot off-diagonal genomic bins into two Region-of-Interests (ROIs) reveals two subsets of genomic regions with low SON TSA-seq which lose nuclear lamina interactions specifically in H1 cells-the furthest off-diagonal ROI-C1 bins correspond to pericentric chromosome regions, particularly enriched on NOR-containing chromosomes, while ROI-C2 bins correspond to more distal chromosome regions flanking centromere regions. (A) An overall inverse relationship between LMNB1 and SON TSA-seq exists in both H1 and K562 cells creating a diagonal pattern in the LMNB (y-axis) versus SON (x-axis) TSA-seq scatterplot. Specifically in H1 cells, off-diagonal bins with low SON TSA-seq but also low LMNB1 TSA-seq appear. These were further divided into two H1 ROIs (C1, orange; C2, blue) relative to other genomic regions (green). These H1 ROI-C1 and ROI-C2 regions map to local clusters that lie within the overall LMNB1 versus SON scatterplot patterns in K562, HCT116, and HFF cells; (B-C) H1 C1 and C2 ROIs associate with both nucleoli and nuclear lamina in other cell lines but lose nuclear lamina association while gaining stronger nucleolar association in H1 cells: LMNB1 (B) and MKI67IP (nucleolus) (C) TSA-seq boxplots for H1 ROI-C1 and C2 (orange) bins versus other genomic regions (green); (D) (Top) The LMNB1 (y-axis) versus SON (x-axis) TSA-seq scatterplot on-diagonal positions of H1 C1 and C2 ROIs in wt K562 cells (left) shift closer to the off-diagonal distribution visualized in H1 cells after LBR, LMNA double-knockout (KO) (right); (Bottom) MKI67IP (y-axis) versus SON (x-axis) TSA-seq scatterplots reveal these H1 C1 ROIs show high nucleolar associations in both wt and double knockout K562 cells; (E) Centromeric and pericentromeric locations of H1 ROI C1 (red), C2 (blue) segmented regions (bottom 2 tracks) shown in H1 cells for non-NOR containing Chr7 (left) and NOR-containing Chr13 (right) regions with MKI67IP (brown), SON (green), LMNB1 (blue) TSA-seq (top 3 tracks) showing high nucleolar and low nuclear speckle and lamina interactions; (F) Majority of LAD bins have high LMNB1 TSA-seq scores in H1 and K562 (round cells), but medium scores in HCT116 and HFF (flat cells). Distribution of LAD (blue) versus iLAD (green) genomic bins with LAD count superimposed over corresponding bins (color-coded) in LMNB1 (y-axis) and SON (x-axis) TSA-seq scatterplots for H1, K562, HCT116, and HFF cells (left to right).

Polarity of nuclear genome organization. (A)

HFF Cluster 1 LAD bins (green), defined by their lowest SON TSA-seq scores, show low to moderate LMNB1 TSA-seq scores in both HFF and HCT116. In contrast, HFF Cluster 2 LAD bins, defined as LAD regions with the highest LMNB1 TSA-seq scores in HFF, while also showing high LMNB1 TSA-scores in HCT116 instead show lower LMNB1 TSA-seq scores relative to HFF Cluster 1 LAD bins in both K562 and H1 cells: SON (x-axis) versus LMNB1 (y-axis) TSA-seq scatterplots color-coded by HFF Cluster 1 and 2 LADs shown for HFF, HCT116, H1, and K562 cells; (B) Chromosome ideograms showing location of HFF1 Cluster 1 LAD regions, which show a bias towards the equatorial plane in IMR90 cells, towards the ends of long chromosome arms; (C-D) Distances to IMR90 nucleus center (C) and equatorial plane (D) vary with chromosome position, particularly distance from centromeres (dashed vertical lines): IMR90 multiplex FISH probe distances [43] from center and equatorial plane along each chromosome. Grey-interquartile range of probe positions; blue-mean probe position distance; red-smoothed mean probe distance. Ends of long chromosome arms, and chromosome arm regions containing Cluster 1 LADs, show trend of localizing away from nuclear center but close to equatorial plane; (E) Projection of IMR90 FISH probe distances (color-coded) from nuclear speckles (left), nucleoli (middle), and nuclear periphery (right) from a single nucleus onto the x-y plane (top) or x-z plane (bottom); (F) Scatterplots of mean distance from equatorial plane (y-axis) and from nuclear center (x-axis) of FISH probes in IMR90 nuclei with the mean distance in IMR90 nuclei (top) or mean TSA-seq scores in HFF nuclei of same FISH probe regions relative to the nuclear lamina (left), nucleolus (middle), or nuclear speckle (right) (FISH probe distances from [43]); (G-I) Scatterplots similar to (F) but color-coded with the genomic classification of each FISH probe: type I/II SPADs (G), LADs, p-w-v fILADs, v-fiLADs, iLADs (H, left panel), SPIN states (I). (H, middle and right panels) Boxplots for IMR90 mean FISH probe distances to nuclear center (middle panel) or nuclear equatorial plane (right panel) for HFF LADs, p-w-v fiLADs, v-fiLADs, and iLAD regions.