Abstract
Maintaining the contractile phenotype of vascular smooth muscle cells
(VSMCs) is critical for vascular homeostasis. However, the role of the
3D chromatin architecture in regulating VSMC identity remains elusive.
A genome-scale CRISPR screen identifies LEMD3 as a potential regulator
to maintain VSMC identity. Lemd3 deficiency in VSMCs results in the
loss of the contractile phenotype and exacerbates intimal hyperplasia
in mice. Protein interactome analysis reveals that LEMD3 interacts with
CBX3, a principal reader of H3K9me2/3, subsequently anchoring
heterochromatin at the nuclear periphery. Employing the DNA polymer
model based on Hi-C data, whole-chromosome simulations demonstrate that
Lemd3 depletion disturbs the chromatin structure. Multi-omics analysis
further reveals that Lemd3 depletion alters the genome conformation as
the increase of inter-TAD (topologically associated domain)
interactions at the boundaries of A and B compartments, which
correlates with decreased chromatin accessibility and repressed
expression of VSMC contractile genes. This study reveals that LEMD3
organizes the 3D chromatin architecture by anchoring heterochromatin at
the nuclear periphery to maintain the VSMC contractile identity.
Subject terms: Nuclear organization, Cardiovascular diseases, Nucleus
__________________________________________________________________
Li et al. show that the inner nuclear membrane protein LEMD3 organizes
3D chromatin architecture by anchoring heterochromatin at the nuclear
periphery to maintain vascular smooth muscle cell contractile identity.
Introduction
Vascular smooth muscle cells (VSMCs) exhibit a quiescent, contractile
phenotype, as they highly express contractile markers, such as MYH11
(also known as smooth muscle myosin heavy chain 11), calponin,
transgelin (also known as SM22α), and α-smooth muscle actin, to
maintain vascular homeostasis under physiological conditions^[60]1,
while contractile VSMCs transiently dedifferentiate to acquire the
migratory and proliferative capabilities for vasculogenesis during the
physiological processes of vascular development and regeneration^[61]2.
Moreover, in response to abnormal cues in the vascular
microenvironment, such as alterations in mechanical stretching,
excessive production of growth factors, extracellular matrix
remodeling, and reactive oxygen species (ROS) accumulation^[62]3–[63]9,
VSMCs suffer from the loss of a contractile phenotype and subsequently
transform into a series of pathological phenotypes related to various
vascular diseases, including atherosclerosis, vascular calcification,
intimal hyperplasia, and aortic aneurysm^[64]10–[65]13. Thus,
maintaining the contractile identity of VSMCs is beneficial for
preventing VSMC phenotypic switching and related vascular diseases.
Although some critical transcription factors or repressor-driven gene
profiling transitions, as well as epigenetic modulations, including
noncoding RNA, DNA methylation and histone acetylation, have been
reported to orchestrate VSMC phenotypic switching^[66]14–[67]18, the
mechanism by which VSMCs maintain their contractile phenotype remains
poorly understood. Current knowledge is mainly limited to the fact that
serum response factor (SRF) and its coactivator myocardin exert their
transcriptional activities as well as epigenetic regulation on histone
variant H2A.Z and H3K4me2 modification to facilitate VSMC contractile
identity^[68]19–[69]22.
Compelling evidence has suggested that cell identity programs result
from the interaction between transcription factors and the chromatin
landscape that they encounter. Chromatin adopts a complex
three-dimensional (3D) conformation within the nucleus. Recent
high-throughput chromosome conformation capture (Hi-C) analyses of
comprehensive interaction maps over large regions or whole genomes have
indicated that the genome is hierarchically organized into chromosome
territories, A/B compartments, topologically associated domains (TADs),
and chromatin loops^[70]23. The intermingling of chromosome
territories, A-B compartment switching and dynamic changes in TADs are
involved in the alteration of the 3D chromatin architecture, which
further affects chromatin accessibility, thereby modulating
transcription factor-dependent gene expression^[71]24. Notably, TADs
are fundamental 3D genome units that engage in dynamic higher-order
inter-TAD interactions^[72]25. However, investigations of inter-TAD
interactions are relatively limited, and the modulation of inter-TAD
interactions remains poorly understood.
The 3D chromatin structure transitions are commonly recognized during
development and cell differentiation^[73]23,[74]26 and are often
dysregulated in disease processes^[75]27–[76]30. In the cardiovascular
system, dynamic changes in 3D chromatin structure have been reported to
be involved in cardiomyocyte maturation, cardiac remodeling, and
dilated cardiomyopathy^[77]27,[78]31,[79]32. However, to date, the role
of the 3D chromatin architecture in the vascular system is still
elusive. Here, we identified LEMD3, an inner nuclear membrane protein,
as a critical regulator that maintains the VSMC contractile phenotype
using a genome-scale CRISPR knockout screen. Further protein
interactome screening revealed that LEMD3 interacted with CBX3, a
principal reader of the heterochromatin-associated histone modification
H3K9me2/3, subsequently anchoring heterochromatin at the nuclear
periphery, whereas Lemd3 depletion caused the repositioning of
heterochromatin from the nuclear periphery toward the interior
validated via immunofluorescence staining. Through Hi-C technology and
whole-chromosome modeling, we further revealed that Lemd3 depletion
disturbed the 3D chromatin architecture, namely, enhanced inter-TAD
interactions at the boundaries of A and B compartments, which
correlated with the downregulation of contractile marker genes, as well
as the subsequent loss of the contractile phenotype. Thus, LEMD3
organized the 3D chromatin architecture to maintain the contractile
identity of VSMCs.
Results
A genome-wide CRISPR knockout screen identifies LEMD3 essential for
maintaining the vascular smooth muscle cell contractile phenotype
To identify genes critical for the maintenance of the VSMC contractile
phenotype, we designed a genome-scale CRISPR knockout screen in the
mouse vascular smooth muscle cell line MOVAS. First, we engineered a
MOVAS cell line expressing an endogenous fluorescent reporter using
CRISPR/Cas9 knock-in technology to insert a P2A-EGFP (enhanced green
fluorescent protein) cassette before the stop codon of the endogenous
contractile phenotype gene ACTA2 to yield the coexpression of the ACTA2
protein and EGFP (Fig. [80]1a). Next, we validated whether EGFP
expression was regulated consistently with the previously
well-recognized altered pattern of ACTA2 expression during the VSMC
phenotypic transition in the knock-in cell line. Flow cytometry
analysis revealed that the fluorescence intensity of EGFP was increased
in VSMCs with the transforming growth factor-β (TGF-β)-or serum
starvation-primed contractile phenotype but was significantly decreased
following platelet-derived growth factor BB (PDGF-BB)-induced VSMC
synthetic phenotypic switching (Fig. [81]1b), suggesting that EGFP
efficiently served as a sensitive reporter for the VSMC phenotypic
transition in the knock-in MOVAS cell line.
Fig. 1. A genome-wide CRISPR knockout screen identifies genes essential for
maintaining the contractile phenotype of VSMCs.
[82]Fig. 1
[83]Open in a new tab
a Schematic diagram showing the generation of an EGFP
reporter-expressing MOVAS cell line. P2A, a self-cleaving peptide. b
Representative flow cytometry and quantification of EGFP expression in
the reporter MOVAS cell line in response to 48 h of TGF-β treatment
(blue line), serum starvation (red line), or PDGF-BB treatment (green
line). WT, wild-type; TGF-β, transforming growth factor-β; PDGF-BB,
platelet-derived growth factor BB. n = 3 independent experiments. c
Schematic diagram of the genome-scale CRISPR knockout screen. d Flow
cytometry analysis of EGFP reporter knock-in MOVAS cells that were
transduced with or without the pooled lentivirus encoding a genome-wide
sgRNA library with two rounds of cell sorting. Wild-type (WT) MOVAS
cells were utilized as a negative control for EGFP fluorescence. e
Identification of the top candidate genes using the MAGeCK algorithm.
Pink circles represent the top candidate genes that have been
previously reported to regulate the VSMC phenotype. MAGeCK, Model-based
Analysis of Genome-wide CRISPR/Cas9 Knockout. RRA score, robust rank
aggregation score. f RT‒qPCR analysis of Acta2, Cnn1, Tagln, and Myh11
gene expression in the A7r5 rat VSMC cell line transfected with the
scrambled siRNA (20 nM) or siRNA targeting Lemd3 (20 nM) for 24 h. The
data are presented as the relative fold changes to siRNA[scramble]
(n = 5 independent experiments). Error bars indicate s.e.m. Statistical
analysis was performed using ordinary one-way ANOVA with Tukey’s
multiple comparisons test for panel b and two-sided unpaired Student’s
t test for (f).
We further generated stable Cas9 expression in the EGFP reporter cell
line via lentivirus transduction to apply CRISPR technology
(Supplementary Fig. [84]1a). Then, the EGFP reporter MOVAS cells with
stable Cas9 expression were targeted with a mouse genome-scale CRISPR
knockout (GeCKO) v2 library containing 6 sgRNAs per gene (total sgRNA
size of 130,209)^[85]33, followed by two successive rounds of flow
cytometry sorting. Since we aimed to target the genes whose knockout
caused loss of the contractile phenotype, we preferentially sorted the
cells with reduced EGFP fluorescence intensity (EGFP^low population)
(Fig. [86]1c). Consequently, the EGFP^low population with 18.2%
proportion was enriched following two rounds of sorting (Fig. [87]1d
and Supplementary Fig. [88]1b). Next-generation sequencing of the
population revealed sgRNA enrichment for 109 genes (MAGeCK^[89]34 sgRNA
enrichment score <10^−3) (Fig. [90]1e, and Supplementary Data [91]1).
Among the top 8 genes, Acta2 served as the positive control, while Nf1
and Nf2 have been previously reported to maintain VSMC contractile
identity^[92]35,[93]36, indicating the reliability of the genome-wide
CRISPR knockout screen (Fig. [94]1e). We subsequently focused on the
remaining 5 genes with unknown functions in VSMCs. Using specific small
interfering RNAs (siRNAs), we validated that knockdown of these 5
genes, including Cand1, Ajuba, Gdi1, Xpo6 and Lemd3, resulted in
significant downregulation of VSMC contractile markers, including
Acta2, Cnn1, and Tagln, suggesting the loss of the contractile
phenotype (Fig. [95]1f and Supplementary Fig. [96]2). Interestingly,
previous evidence has shown that Lemd3 deficiency in mice results in
abnormal vascular development and embryonic lethality^[97]37,
suggesting a potential role for LEMD3 in the vascular system, but
without further investigation thus far. Therefore, we preferentially
explored the role of Lemd3 in VSMC identity and vascular homeostasis.
LEMD3 maintains the contractile phenotype of vascular smooth muscle cells in
vitro
We firstly clarified the nuclear localization of LEMD3 in VSMCs. As
shown by the immunocytochemical staining on rat VSMCs, LEMD3 was
substantially expressed and mainly localized in the nuclei of VSMCs
(Fig. [98]2a). Meanwhile, we validated it in human arteries by
immunohistochemical staining, as evidenced by the obvious staining of
LEMD3 in the cellular nuclei of arterial media mainly containing VSMCs
(Fig. [99]2b). Following the transfection of the EGFP-LEMD3 plasmid
into the A7r5 VSMCs line, we observed that LEMD3 fused with EGFP was
localized mainly to the nuclear membrane in VSMCs (Fig. [100]2c),
consistent with the role of LEMD3 as an inner nuclear membrane
protein^[101]38. Next, we explored the expression alterations of LEMD3
during VSMC phenotypic switching. As results, serum starvation and
rapamycin priming VSMCs into contractile phenotype significantly
upregulated LEMD3 expression (Fig. [102]2d, e and Supplementary
Fig. [103]3a, b), whereas PDGF-BB causing VSMC dedifferentiation
markedly downregulated LEMD3 expression (Fig. [104]2f and Supplementary
Fig. [105]3c). Moreover, we validated the regulation of LEMD3
expression in vivo. As a consequence, LEMD3 expression was reduced at 3
days after wire injury in mouse carotid arteries compared to sham
arteries (Fig. [106]2g). These results suggested that LEMD3 expression
positively correlated with the contractile phenotype of VSMCs.
Fig. 2. LEMD3 maintains the contractile phenotype of VSMCs.
[107]Fig. 2
[108]Open in a new tab
a Representative images of immunocytochemical staining for LEMD3 in
primary rat VSMCs. Scale bar = 100 µm. n = 3 independent experiments. b
Representative images of immunohistochemical staining for LEMD3 in
human internal mammary arteries. The region between the dashed lines
corresponds to the medial area. Scale bar = 20 µm. n = 3 independent
samples. c Representative images of immunofluorescence staining of rat
VSMCs transfected with the pEGFP-N1-LEMD3 plasmid or control plasmid.
The nuclei were stained with DAPI (blue). Scale bar = 5 µm. This
experiment was repeated three times with similar results. d
Representative Western blotting and quantification of LEMD3, ACTA2,
CNN1, TAGLN in the A7r5 rat VSMC cell line under serum starvation
(48 h). GAPDH was used as an internal control. The data are presented
as the relative fold changes to control (Ctrl) group (n = 6 independent
experiments). e Representative Western blotting and quantification of
LEMD3, ACTA2, CNN1, TAGLN in the A7r5 rat VSMC cell line under
rapamycin treatment (100 nM, 48 h). GAPDH was used as an internal
control. The data are presented as the relative fold changes to vehicle
group (n = 6 independent experiments). f Representative Western
blotting and quantification of LEMD3, ACTA2, CNN1, TAGLN in the A7r5
rat VSMC cell line under PDGF-BB treatment (10 ng/ml, 48 h). GAPDH was
used as an internal control. The data are presented as the relative
fold changes to vehicle group (n = 6 independent experiments). g
Representative western blotting and quantification of LEMD3 in the
lysates of sham-operated and wire-injured mouse carotid arteries at 3
days after surgery. GAPDH was used as an internal control. Data were
presented as relative fold change to Sham (n = 4 independent mice). h
Representative Western blotting and quantification of LEMD3, ACTA2,
CNN1, and TAGLN in the A7r5 rat VSMC cell line transfected with the
scrambled siRNA (20 nM) or siRNA targeting Lemd3 (20 nM) for 12 h,
followed by serum starvation treatment (48 h). GAPDH was used as an
internal control. The data are presented as the relative fold changes
to siRNA[scramble] (n = 4 independent experiments). i Representative
Western blotting and quantification of LEMD3, ACTA2, CNN1, and TAGLN in
the A7r5 rat VSMC cell line transfected with the scrambled siRNA
(20 nM) or siRNA targeting Lemd3 (20 nM) for 12 h, followed by
rapamycin treatment (100 nM, 48 h). GAPDH was used as an internal
control. The data are presented as the relative fold changes to
siRNA[scramble] (n = 4 independent experiments). j Representative
images of phalloidin staining and quantification of F-actin (red) in
primary rat VSMCs transfected with the scrambled siRNA (20 nM) or siRNA
targeting Lemd3 (20 nM) for 48 h. DAPI staining (blue) indicates the
nucleus. Scale bar = 25 µm. n = 3 independent experiments. k Collagen
gel contraction assays using primary rat VSMCs transfected with the
scrambled siRNA (20 nM) or siRNA targeting Lemd3 (20 nM) for 48 h.
n = 3 independent experiments. l EdU incorporation assays of primary
rat VSMCs transfected with the scrambled siRNA (20 nM) or siRNA
targeting Lemd3 (20 nM) for 48 h. n = 6 independent experiments. Error
bars indicate s.e.m. Statistical analysis was performed using two-sided
unpaired Student’s t test for (d-g), two-way ANOVA followed by
Bonferroni’s multiple comparisons test for (h) and (i), the χ^2 test
for (j) and the Mann‒Whitney U test for (k) and (l).
Next, we investigated whether LEMD3 is required for maintenance of the
contractile phenotype of VSMCs. Lemd3 knockdown suppressed serum
starvation- or rapamycin-induced upregulation of contractile gene
expression at both mRNA and protein levels in rat and human VSMCs
(Fig. [109]2h-i and Supplementary Fig. [110]4). Moreover, F-actin
staining revealed that Lemd3 silencing resulted in VSMC morphological
changes from an elongated contractile pattern to a polygonal synthetic
pattern (Fig. [111]2j), while collagen gel contraction assays also
revealed that Lemd3 knockdown reduced the contractility of VSMCs
(Fig. [112]2k). These results indicated that Lemd3 silencing led to the
loss of the contractile phenotype in VSMCs. VSMC proliferation and
migration have been reported to be enhanced following phenotypic
switching^[113]39. Accordingly, we evaluated the proliferation and
migration of Lemd3-silenced VSMCs. EdU (5-ethynyl-2’-deoxyuridine)
incorporation assays suggested that Lemd3 silencing promoted VSMC
proliferation (Fig. [114]2l). However, Lemd3 knockdown did not affect
VSMC migration, as evidenced by the results of the scratch wound assay
and transwell migration assay (Supplementary Fig. [115]5a-b).
Conversely, Lemd3 overexpression reversed PDGF-BB-induced loss of VSMC
contractile phenotype (Supplementary Fig. [116]5c-d). Together, these
results collectively supported that LEMD3 maintains the contractile
identity of VSMCs.
Lemd3 deficiency causes the loss of the vascular smooth muscle cell
contractile phenotype in vivo
To further validate the role of LEMD3 in VSMCs in vivo, we constructed
tamoxifen-induced smooth muscle cell-specific Lemd3 deletion mice
(Lemd3^SMKO mice) by intercrossing Myh11-CreER^T2 mice with
Lemd3^flox/flox mice in which exon 2 was flanked with loxP sites
(Fig. [117]3a and Supplementary Fig. [118]6b). Tamoxifen was
administered to 8-week-old Myh11-CreER^T2 Lemd3^flox/flox mice to
achieve smooth muscle cell-confined depletion of Lemd3 compared with
Lemd3^WT control mice (littermate Myh11-CreER^T2 mice), which received
a comparable tamoxifen treatment (Supplementary Fig. [119]6a).
Lemd3^SMKO mice presented no significant differences in body weight or
systolic blood pressure (101.8 ± 4.1 mmHg vs. 103.2 ± 3.9 mmHg)
compared with Lemd3^WT control mice (Supplementary Table [120]1),
whereas ultrasonography of mouse aortas revealed a slight expansion in
segments of the ascending aorta and aortic arch in Lemd3^SMKO mice
(Supplementary Fig. [121]6c-d). As measured by western blotting and
immunohistochemistry, Lemd3 deficiency significantly downregulated the
expression of ACTA2, CNN1 and TAGLN in mouse aortas, suggesting a loss
of the contractile phenotype (Fig. [122]3b-c). Accordingly, aortic
rings and mesenteric resistance arteries isolated from Lemd3^SMKO mice
exhibited reduced contraction in response to phenylephrine compared
with those isolated from Lemd3^WT mice (Fig. [123]3d and Supplementary
Fig. [124]6e). Since the in vivo studies only included male mice due to
the insertion of Myh11-CreER^T2 transgene into Y chromosome, primary
aortic smooth muscle cells were alternatively isolated from
Lemd3^flox/flox female and male mice to clarify the potential
sex-specific difference in the role of LEMD3 in VSMCs. Adenovirus
encoding Cre recombinase was applied in the isolated mouse primary
VSMCs to induce Lemd3 knockout. As results, Lemd3 deficiency comparably
downregulated the contractile gene expression in both female and male
background VSMCs (Supplementary Fig. [125]7a-d), excluding the
potential sex-specific difference.
Fig. 3. Lemd3 depletion results in the loss of the contractile phenotype in
VSMCs in vivo.
[126]Fig. 3
[127]Open in a new tab
a Schematic view of the generation of tamoxifen-inducible smooth muscle
cell-specific Lemd3 knockout mice. LoxP sites are indicated by red
triangles. b Representative western blot and quantification of protein
expression in the aortas of 12-week-old Lemd3^WT mice and Lemd3^SMKO
mice. GAPDH was used as an internal control. The data are presented as
the relative fold changes compared with Lemd3^WT mice. n = 6 mice per
group. c Representative images and quantification of
immunohistochemical staining for ACTA2 and CNN1 in the thoracic aortas
from 12-week-old Lemd3^WT mice and Lemd3^SMKO mice. Scale bar = 100 µm.
n = 4 mice per group. d Contraction of aortic rings isolated from
12-week-old Lemd3^WT mice and Lemd3^SMKO mice in response to
phenylephrine (Phe). n = 5 mice per group. e Left, Hematoxylin and
eosin (H&E) staining of the carotid arteries of sham-operated and
wire-injured Lemd3^WT mice and Lemd3^SMKO mice on Day 28 after surgery.
The inner circles represent the neointimal areas, and the outer circles
represent the medial and neointimal areas. Scale bar = 100 µm. Right,
quantitative analysis of the intima areas, the media areas, intima to
media ratio, and external elastic lamina (EEL) circumference in
H&E-stained sections of the carotid arteries of wire-injured Lemd3^WT
mice and Lemd3^SMKO mice at 28 days after surgery. n = 6 mice per
group. f EdU staining of the carotid arteries of Lemd3^WT mice and
Lemd3^SMKO mice at 28 days after wire injury. The white arrows mark the
EdU-positive cells in the neointima and medial area. Mouse vascular
elastic lamina (EL) exhibits green autofluorescence. Scale bar =
100 µm. n = 6 mice per group. HE hematoxylin and eosin, EEL external
elastic lamina. Error bars indicate s.e.m. Statistical analysis was
performed using two-sided unpaired Student’s t test with Welch’s
correction for (b-d) and the Mann‒Whitney U test for (e) and (f).
Next, we explored the protective role of LEMD3 in VSMCs under
pathological conditions. We generated a carotid wire injury model in
12-week-old male tamoxifen-induced Lemd3^SMKO mice and Lemd3^WT mice to
induce VSMC phenotypic switching and neointima formation. At 28 days
postinjury, the mouse carotid arteries were harvested (Supplementary
Fig. [128]7e). H&E staining of wire-injured carotid arteries showed
that Lemd3 deficiency markedly exacerbated intimal hyperplasia, as
evidenced by the elevations in the neointima areas and the
neointima/media ratios in Lemd3^SMKO carotid arteries compared with
those in Lemd3^WT controls (Fig. [129]3e). In addition, EdU staining
revealed the greater cell proliferation within vascular wall in carotid
arteries from Lemd3^SMKO mice than those from control mice at 28 days
after wire injury (Fig. [130]3f). Since the major cell type in arterial
media and neointima is VSMCs as well as the increased cell
proliferation is also a hallmark of VSMC dedifferentiation^[131]40,
these in vivo data collectively suggested that Lemd3 depletion
exacerbates cell proliferation and intimal hyperplasia after vascular
injury, possibly due to the loss of the VSMC contractile phenotype.
LEMD3 interacts with CBX3
Previous studies have reported that LEMD3 interacts with SMAD2/3 and
represses TGF-β/SMAD2/3 signaling^[132]41,[133]42, while compelling
evidence indicates that TGF-β/SMAD2/3 signaling maintains the
contractile phenotype of VSMCs^[134]43. To further verify whether
TGF-β/SMAD2/3 signaling was involved in LEMD3-regulated VSMC phenotypic
switching, we explored the alteration of TGF-β/SMAD2/3 signaling upon
Lemd3 knockdown in rat VSMCs. As results, TGF-β treatment significantly
induced SMAD3 phosphorylation (Supplementary Fig. [135]8a, b), whereas
Lemd3 knockdown further enhanced TGF-β-induced SMAD3 phosphorylation
(Supplementary Fig. [136]8a, b). Of interest, although the silencing of
Lemd3 amplified TGF-β/SMAD3 signaling, Lemd3 knockdown conversely
abolished TGF-β-upregulated contractile gene expression in VSMCs
(Supplementary Fig. [137]8c, d). Furthermore, we applied SIS3, a
selective inhibitor of SMAD3, and found that SIS3 did not influence on
the suppression of Lemd3 knockdown on TGF-β-induced VSMC contractile
phenotype (Supplementary Fig. [138]8c, d), suggesting that the effect
of LEMD3 on VSMC phenotypic switching was independent on TGF-β/SMAD
signaling.
To explore the underlying mechanism by which LEMD3 maintains the VSMC
contractile phenotype, we performed an interactome analysis to screen
potential LEMD3-binding proteins. HEK293T cells were transfected with a
FLAG-tagged LEMD3 plasmid, and the LEMD3-binding proteins were enriched
via immunoprecipitation with an anti-FLAG antibody followed by liquid
chromatography‒tandem mass spectrometry (LC‒MS) identification
(Fig. [139]4a). As a result, 67 proteins were identified as potential
LEMD3-binding proteins (Supplementary Data [140]2). Among them, SMAD2/3
and CBX3 (chromobox protein homolog 3) have been reported to maintain
the VSMC contractile phenotype^[141]44. Since we excluded the
involvement of TGFβ/SMAD signaling in the role of LEMDs in VSMCs, we
next investigated the role of the potential interaction between LEMD3
and CBX3 in maintaining VSMC contractile identity.
Fig. 4. LEMD3 interacts with CBX3.
[142]Fig. 4
[143]Open in a new tab
a Schematic diagram of the LEMD3 interactome. b Coimmunoprecipitation
(Co-IP) assays of HEK293T cells cotransfected with LEMD3-FLAG and
HA-CBX3 plasmids for 48 h. The lysates were immunoprecipitated with a
control IgG antibody, anti-FLAG antibody, or anti-CBX3 antibody,
followed by immunoblotting with the indicated antibodies. c Co-IP
assays of endogenous LEMD3 and CBX3 in rat VSMCs. The lysates were
immunoprecipitated with a control IgG antibody, anti-LEMD3 antibody, or
anti-CBX3 antibody, followed by immunoblotting with the indicated
antibodies. d Schematic illustration of LEMD3 domain depletion
mutations used to evaluate the interaction with CBX3. The presence or
absence of the deletion mutant binding to CBX3 was defined as + or −,
respectively. e HEK293T cells were cotransfected with the CBX3 plasmid
and the full-length LEMD3-FLAG, ΔC-terminal-FLAG, or ΔRRM-FLAG plasmid
for 48 h. The cell lysates were immunoprecipitated with an anti-CBX3
antibody, and the precipitates were analyzed by immunoblotting with an
anti-FLAG antibody. f Coimmunoprecipitation (Co-IP) assays of HEK293T
cells cotransfected with RRM-FLAG and HA-CBX3 plasmids for 48 h. The
lysates were immunoprecipitated with a control IgG antibody, or
anti-FLAG antibody, followed by immunoblotting with the indicated
antibodies. g, h Representative western blotting and quantification of
VSMC contractile marker expression in the A7r5 rat VSMC cell line
transfected with the scrambled siRNA (20 nM) or siRNA targeting Lemd3
(20 nM) for 12 h, followed by transfection with the pcDNA3.1,
LEMD3-FLAG, or ΔRRM-FLAG plasmid for 48 h. GAPDH was used as an
internal control. The data are presented as the relative fold changes
compared with that of the pcDNA3.1+vehicle group (n = 5 independent
experiments). i, j Representative western blotting and quantification
of VSMC contractile marker expression in the A7r5 rat VSMC cell line
transfected with the pcDNA3.1 or LEMD3-FLAG plasmid for 12 h, followed
by transfection with the scrambled siRNA (20 nM) or siRNA targeting
Cbx3 (20 nM) for 48 h. GAPDH was used as an internal control. The data
are presented as the relative fold changes compared with that of the
pcDNA3.1+siRNA[scramble] group (n = 3 independent experiments). Data
shown in (b, c, e and f) are from one representative of three
independent experiments with similar results. Error bars indicate
s.e.m. Statistical analysis was performed using two-way ANOVA followed
by Bonferroni’s multiple comparisons test for (h) and (j).
We first validated that LEMD3 interacted with CBX3 in HEK293T cells
overexpressing CBX3 and FLAG-tagged LEMD3 via coimmunoprecipitation
assays (Fig. [144]4b). Furthermore, the endogenous interaction between
LEMD3 and CBX3 was also verified in rat VSMCs and mouse aortas
(Fig. [145]4c and Supplementary Fig. [146]9a). Moreover, PDGF-BB
treatment efficiently downregulated LEMD3 expression, thereby reducing
the interaction between LEMD3 and CBX3 in VSMCs (Supplementary
Fig. [147]9b), associated with the loss of contractile phenotype caused
by PDGF-BB stimulation. Then, we further explored the specific binding
motif of LEMD3 that interacts with CBX3 by subcloning serial deletion
mutants of LEMD3, including a C-terminal deletion mutation
(ΔC-terminal, aa 1–876) and an RNA recognition motif (RRM) deletion
mutation (ΔRRM, aa 1–784), into the pcDNA3.1-FLAG plasmid. HEK293T
cells were transfected with plasmids encoding CBX3 and full-length
LEMD3-FLAG, ΔC-terminal-FLAG, or ΔRRM-FLAG, followed by
coimmunoprecipitation analysis with an anti-CBX3 antibody. Full-length
LEMD3 and the ΔC-terminal mutant, but not the ΔRRM mutant, efficiently
immunoprecipitated with CBX3, suggesting that LEMD3 interacts with CBX3
via its RRM domain (Fig. [148]4d-e). Further NanoBit-based binding
assays consistently confirmed that the RRM domain was the binding motif
of LEMD3 that interacted with CBX3 (Supplementary Fig. [149]9c). We
subcloned RRM domain into the pcDNA3.1-FLAG plasmid and transfected
HEK293T cells with plasmids encoding CBX3 and RRM-FLAG.
Coimmunoprecipitation analysis revealed that RRM domain interacted with
CBX3 (Fig. [150]4f).
To further confirm whether the interaction with CBX3 mediated the
effect of LEMD3 to maintain the VSMC contractile phenotype, we
transfected rat VSMCs with the pcDNA3.1-FLAG vector, FLAG-tagged LEMD3,
or FLAG-tagged ΔRRM mutant plasmid followed by vehicle or PDGF-BB
treatment. Unlike full-length LEMD3, overexpression of the ΔRRM mutant
that is unable to bind to CBX3 did not reverse the PDGF-BB-induced loss
of the VSMC contractile phenotype at either the mRNA or protein level
(Supplementary Fig. [151]9d and Fig. [152]4g, h). To specifically
target endogenous LEMD3-CBX3 interaction, we transfected rat VSMCs with
plasmid encoding RRM-FLAG. RRM domain overexpression blocked the
endogenous LEMD3-CBX3 interaction and decreased the contractile protein
levels in VSMCs (Supplementary Fig. [153]9e, f). Furthermore, we
silenced Cbx3 or Lemd3 in VSMCs, with concomitant overexpression of
Lemd3 or Cbx3, respectively. As results, Cbx3 silencing efficiently
reversed Lemd3 overexpression-enhanced SMC contractile protein
expression (Fig. [154]4i, j), while knockdown of Lemd3 conversely
inhibited Cbx3 overexpression-upregulated contractile protein
expression in VSMCs as well (Supplementary Fig. [155]9g), indicating
the interdependent relationship between LEMD3 and CBX3 in the
maintenance of VSMC contractile phenotype. These results collectively
suggested that the interaction of LEMD3 with CBX3 plays a critical role
in maintaining the VSMC contractile phenotype.
LEMD3 anchors heterochromatin at the nuclear periphery and maintains the 3D
chromatin architecture
CBX3 belongs to the heterochromatin protein 1 (HP1) family, which
consists of three members: HP1α (encoded by Cbx5), HP1β (encoded by
Cbx1) and HP1γ (encoded by Cbx3)^[156]45. As the principal reader of
the heterochromatin-associated histone modification H3K9me2/3, the HP1
family is involved in heterochromatin organization and subsequently
influences the 3D chromatin structure^[157]46–[158]48. Considering the
interaction between LEMD3 and CBX3, we explored whether LEMD3 affects
heterochromatin organization. By performing immunofluorescence staining
for the repressive histone modification H3K9me3, as a potential
indicator of heterochromatin, we observed that the fluorescence
intensity along the nuclear diameter exhibited a significant shift in
the radial distribution of H3K9me3 from the nuclear periphery toward
the interior upon Lemd3 knockdown in rat VSMCs (Fig. [159]5a, b).
However, Lemd3 knockdown in rat VSMCs did not affect the total amount
of H3K9me3, as evidenced by western blot analysis (Fig. [160]5c). To
further explore the impact of Lemd3 knockdown on the global landscape
of H3K9me3, we performed ChIP sequencing of rat VSMCs transfected with
scrambled siRNA or siRNA targeting Lemd3. H3K9me3 signaling exhibited
no statistically changes through either genome-wide profiling or
locus-specific analysis upon Lemd3 knockdown (Fig. [161]5d and
Supplementary Fig. [162]10). Meanwhile, we further verified the similar
alteration in nuclear distribution of H3K9me2 in VSMCs upon Lemd3
knockdown through immunofluorescence staining (Supplementary
Fig. [163]11a, b). Overall, altered nuclear distribution of H3K9me2/3
suggested that Lemd3 silencing reduced heterochromatin perinuclear
anchoring.
Fig. 5. LEMD3 mediates heterochromatin perinuclear anchoring and maintains
the global chromatin organization.
[164]Fig. 5
[165]Open in a new tab
a Left, representative images of immunofluorescence staining for
H3K9me3 (green) and Lamin B1 (red) in rat VSMCs transfected with the
scrambled siRNA (20 nM) or siRNA targeting Lemd3 (20 nM) for 48 h. DAPI
staining (blue) indicates the nucleus. Scale bar = 10 µm. n = 3
independent experiments. Right, quantification of immunofluorescence
staining intensities for H3K9me3. Upper, the intensity ratio of nuclear
interior to periphery; Lower, total amount of nuclear intensities. Six
cells were randomly selected in each experiment. n = 18 cells. b
Fluorescence intensity along the white line in a single cell was
measured using ImageJ software for the H3K9me3 (green) channel. Scale
bar = 5 µm. This experiment was repeated three times with similar
results. c Representative western blotting and quantification of
H3K9me3 levels in rat VSMCs transfected with the scrambled siRNA
(20 nM) or siRNA targeting Lemd3 (20 nM) for 48 h. Histone H3 was used
as an internal control. The data are presented as the relative fold
changes to siRNA[scramble] (n = 3 independent experiments). d ChIP-seq
data (n = 2) showing the genomic occupancy of H3K9me3 in the A7r5 rat
VSMC cell line transfected with the scrambled siRNA (20 nM) or siRNA
targeting Lemd3 (20 nM) for 48 h. e, f Hi-C analysis (n = 2) of the
A7r5 rat VSMC line transfected with the scrambled siRNA (20 nM) or
siRNA targeting Lemd3 (20 nM) for 48 h. e The 3D whole-nucleus maximum
entropy model (line rendering) and plane model of the nucleus (bond
rendering) of WT and Lemd3 KD VSMCs. Purple represents the A
compartment, and green represents the B compartment. f Curve diagram
showing the ratio of A/B compartments from the center to the periphery
of the nuclei. Blue represents WT, and red represents Lemd3 KD. g Left,
representative electron microscopy images of aortic smooth muscle cells
from 12-week-old Lemd3^WT mice and Lemd3^SMKO mice. The yellow arrows
mark the perinuclear heterochromatin. E elastic lamina. Scale
bar = 2 μm. Right, Quantitative analysis of the width of perinuclear
heterochromatin of aortic smooth muscle cells from 12-week-old Lemd3^WT
mice and Lemd3^SMKO mice. Four vascular smooth muscle cells were
randomly selected from each mouse aortic section. n = 6 mice per group.
Error bars indicate s.e.m. Statistical analysis was performed using the
Mann‒Whitney U test for (a) and (g), and two-sided unpaired Student’s t
test for (c).
To further explore whether the reduction in heterochromatin perinuclear
anchoring caused by Lemd3 knockdown altered the global chromatin
architecture, we performed a high-throughput chromosome conformation
capture (Hi-C) assay. Based on principal component analysis (PCA) of
Hi-C data, the genome can be divided into A and B compartments^[166]49,
which primarily overlap euchromatin and heterochromatin,
respectively^[167]50. We used parameters based on the maximum entropy
principle to construct 3D polymer models at 100-kb resolution using
Hi-C data^[168]51. The parameters of the polymer model are encoded in
the energy function, and their values are determined iteratively. The
parameters are adjusted to match the experimentally determined contact
frequencies, ensuring compatibility with experimental data while
minimizing biases. Our polymer models positioned compartments in a
B-A-B order from the center to the periphery of the nuclei
(Fig. [169]5e). This result was supported by imaging experiments and 3D
genome structure calculations at single-cell resolution, which revealed
that the chromosomes packed together to form an outer B compartment
ring, an inner A compartment ring, and an internal B compartment region
around the hollow nucleoli in all cells, based on the previous
definition standards^[170]24,[171]52. From certain views of the model,
we observed that chromatin appeared to be compressed toward the core of
the nucleus upon Lemd3 knockdown (Fig. [172]5e). As expected, H3K9me3
was predominantly found in B compartments (Supplementary
Fig. [173]11c), which are reported to harbor inactive
chromatin^[174]53,[175]54. In the polymer models of Lemd3 KD VSMCs, the
perinuclear B compartment was skewed toward the nuclear interior
compared with that of WT VSMCs (Fig. [176]5f), which was consistent
with the alterations of nuclear distribution of the heterochromatin
mark H3K9me2/3 upon Lemd3 knockdown. Furthermore, we analyzed the
chromatin organization of the aortic smooth muscle cells in Lemd3^WT
mice and Lemd3^SMKO mice via electron microscopy. The electron-dense
heterochromatin was distributed along the nuclear envelope (NE) in the
aortic smooth muscle cells of Lemd3^WT control mice (Fig. [177]5g).
However, perinuclear heterochromatin was decreased in the aortic smooth
muscle cells of the Lemd3^SMKO mice (Fig. [178]5g). Taken together,
these findings indicate that Lemd3 depletion results in the
repositioning of heterochromatin and disturbances in the 3D chromatin
architecture.
To investigate whether LEMD3 anchored H3K9me3-modified heterochromatin
at the nuclear periphery dependently on the interaction with CBX3, we
firstly confirmed that Cbx3 silencing phenocopied the effect of Lemd3
knockdown to cause the redistribution of H3K9me3-modified
heterochromatin from the nuclear periphery toward the interior in VSMCs
(Supplementary Fig. [179]11d, e). Next, we transfected rat VSMCs with
plasmid encoding RRM-FLAG to block the endogenous LEMD3-CBX3
interaction (Supplementary Fig. [180]9e), and also found that
H3K9me3-modified heterochromatin redistributed from the nuclear
periphery toward the interior upon RRM domain overexpression
(Supplementary Fig. [181]11f, g), suggesting that the interaction of
LEMD3 with CBX3 might mediate the anchoring of H3K9me3-modified
heterochromatin at the nuclear periphery.
Lemd3 depletion increases inter-TAD interactions at the boundaries of A and B
compartments and decreases contractile-related gene expression
We further explored the alterations in the 3D chromatin architecture at
the scales of the compartments and topologically associated domains
(TADs) upon Lemd3 depletion. At the compartment level, the vast
majority of the genome remained unchanged in A/B compartment
organization (Supplementary Fig. [182]12a). Recent studies revealed
that long-range 3D aggregation of B compartment interactions
facilitates the exclusion of gene-rich regions from the B compartment
during cell development^[183]55, whereas interactions between A and B
compartments are increased in tumors^[184]51. Interestingly, we
discovered that long-range interactions of B-B compartments were
decreased and that short-range interactions between nearby A-B
compartments were more frequent in certain chromatin regions upon Lemd3
knockdown (Fig. [185]6a and Supplementary Fig. [186]12b). Furthermore,
in regions with a strong compartment pattern, knockdown of Lemd3
resulted in a relative increase in inter-compartment interactions
compared to intra-compartment interactions (Supplementary
Fig. [187]12c), suggesting that the change in inter-compartment
interactions may be concentrated within a smaller specific range of the
genome. Therefore, we deepened the resolution of the analysis and
identified TADs of the Hi-C normalized matrix at 40-kb resolution. We
identified 259 WT-specific TAD boundaries, 184 KD-specific TAD
boundaries and 4071 common TAD boundaries. The TAD boundaries exhibited
no significant difference between the Lemd3 KD and control groups or
between the biological replicates (Supplementary Fig. [188]12d). These
findings suggested that the knockdown of Lemd3 does not affect the
overall TAD structure in VSMCs. We then classified genome-wide
interactions into intra-TAD interactions, interactions between adjacent
TADs and interactions between non-adjacent TADs. Interestingly, we
found that Lemd3 knockdown resulted in obvious changes in inter-TAD
interactions (Fig. [189]6b). Overall, the contacts of non-adjacent TAD
pairs were much weaker than those of adjacent TAD pairs (Fig. [190]6c).
Consequently, our attention was directed toward interactions between
adjacent TADs, particularly those located at the boundaries of A and B
compartments. We found that 1845 of the 4323 adjacent TAD pairs
exhibited altered interactions upon Lemd3 knockdown (Fig. [191]6b).
Both the upregulation and downregulation of the interactions between
neighboring TADs within A or B compartments were observed in
Lemd3-silenced VSMCs, whereas adjacent TAD interactions at the
boundaries of A and B compartments were exclusively increased by Lemd3
knockdown (Fig. [192]6d-e). Examples of the contact map of a 3-Mb
region on chromosome 6 and 11 showed that neighboring TAD-TAD
interactions at the A-B boundaries were markedly increased
(Fig. [193]6f, g).
Fig. 6. The Hi-C analysis indicates that Lemd3 depletion changes inter-TAD
interactions.
[194]Fig. 6
[195]Open in a new tab
a Heatmap showing the alterations in long-range compartment
interactions in the region of 50–100 Mb on chromosome 8 upon Lemd3
depletion. The red box indicates increased interactions between A-B
compartments, and the black box indicates decreased interactions
between B-B compartments. The bin size is 100 kb. b Bar plots showing
the percent change in TAD-TAD interactions. c Box plot showing the
average contacts of adjacent and non-adjacent TAD pairs. Box plots are
defined in terms of the minimum values and maximum values (whiskers),
median (center line) and 25th and 75th percentiles (bounds of box).
Adjacent: n = 4323 TAD pairs; Non-adjacent: n = 259868 TAD pairs. d
Contour plots showing the distribution of adjacent TAD pairs with
decreased or increased interactions upon Lemd3 depletion. e Histogram
showing the proportion of adjacent TAD pairs with increased or
decreased interactions upon Lemd3 depletion. Heatmap showing the
alterations in adjacent TAD-TAD interactions at the A-B boundary in the
region of 21‒24 Mb on chromosome 6 (f) and the region of 74‒77 Mb on
chromosome 11 (g) upon Lemd3 depletion. The dotted boxes denote
inter-TAD interactions at the A-B boundary. The bin size is 40 kb.
Statistical analysis was performed using the Mann‒Whitney U test for
the data in (c).
Recent studies revealed close connections between TAD boundaries,
chromatin accessibility and gene transcription^[196]56–[197]59. Thus,
we additionally performed RNA-seq and ATAC-seq to assess alterations in
gene expression and chromatin accessibility upon Lemd3 knockdown
(Fig. [198]7a). RNA-seq revealed that Lemd3 silencing upregulated 1169
genes and downregulated 1431 genes (Supplementary Fig. [199]13a,
Supplementary Data [200]3), whereas ATAC-seq revealed that chromatin
accessibility was predominantly increased at intergenic regions and
decreased at promoter regions, with 29,202 increased peaks and 1467
decreased peaks upon Lemd3 knockdown (Supplementary Fig. [201]13b, c).
Interestingly, ATAC-seq signals were negatively correlated with
inter-TAD interactions at the boundaries of the A and B compartments
(Fig. [202]7b). Gene set enrichment analysis (GSEA) and heatmap of the
downregulated genes identified via RNA-seq revealed that Lemd3
depletion inhibited smooth muscle contraction and differentiation
pathways (Supplementary Fig. [203]13d-e, Supplementary Data [204]4).
Pathway enrichment analysis of the genes assigned to the downregulated
peaks via ATAC-seq also revealed that Lemd3 depletion inhibited smooth
muscle contraction and differentiation pathways (Supplementary
Fig. [205]13f and Supplementary Data [206]5). Conversely, the genes
assigned to the upregulated ATAC-seq peaks were functionally associated
with positive regulation of smooth muscle cell proliferation
(Supplementary Fig. [207]13g and Supplementary Data [208]6). These
findings are consistent with the data shown in Fig. [209]2, suggesting
that Lemd3 silencing downregulates VSMC contractile gene expression and
promotes VSMC proliferation. The integrated analysis of ATAC-seq and
RNA-seq data identified 143 genes, including the contractile-related
genes Acta2, Cnn1, Tagln, Myh11, Col4a1, etc., with decreased chromatin
accessibility and downregulated mRNA expression (Fig. [210]7c,
Supplementary Fig. [211]14 and Supplementary Data [212]7), suggesting
that LEMD3 maintained the chromatin accessibility of these genes and
facilitated their transcription. Further bioinformatics analysis of the
potential transcription factor-binding motifs in the DNA sequences of
ATAC-seq peaks mapped to these 143 genes suggested that the
transcription factors MEF2A/C/D, SRF, and TEAD3/4 which have been
reported to orchestrate VSMC contractile identity^[213]60–[214]63,
might be involved in LEMD3-facilitated contractile marker gene
expression (Fig. [215]7d and Supplementary Data [216]8). In addition,
we found that among these 143 genes, 62 genes were located in the TADs
at A-B boundaries which were most significantly enriched in the muscle
cell differentiation pathway and smooth muscle contraction pathway
(Fig. [217]7e and Supplementary Data [218]9). Together, these data
suggested that Lemd3 depletion substantially increased inter-TAD
interactions at the boundaries of A and B compartments and decreased
the chromatin accessibility of possible MEF2/SRF/TEAD-binding DNA
motifs, thereby downregulating VSMC contractile-related gene
expression.
Fig. 7. Lemd3 depletion increases inter-TAD interactions at the boundaries of
A and B compartments and downregulates smooth muscle contractile gene
expression.
[219]Fig. 7
[220]Open in a new tab
a Schematic representation of topologically associating domains (TADs).
RNA-seq (n = 3) was used to measure the transcriptional output.
ATAC-seq (n = 3) was used to identify chromatin accessibility. b
Scatter plot showing the correlation between the fold change of
ATAC-seq signals and the fold change of Hi-C contact frequencies at the
boundaries of the A and B compartments upon Lemd3 depletion. PCC
Pearson’s correlation coefficients. c Venn diagram showing the overlap
between the genes whose expression was downregulated and the genes
around the closed ATAC peaks. d Transcription factor binding motif
analysis of the DNA sequences of closed ATAC peaks corresponding to the
genes downregulated upon Lemd3 depletion. e Bar plots showing the
enriched GOBP pathways of the genes located in the TADs at the A-B
boundaries with both decreased chromatin accessibility and
downregulated mRNA expression. Hi-C contact maps (n = 2), histone
ChIP-seq tracts (n = 2), ATAC-seq tracts (n = 3) and RNA-seq tracts
(n = 3) surrounding the gene loci of Tagln (f) and Acta2 (g). The black
lines delineate TAD boundaries. The bin size is 40 kb. h Scatter plot
showing the statistical results of LEMD3-modulated human homologous
genes. Three coordinate axes represent the distance to the nearest A-B
boundary of human gene locus, the gene sequence identity compared to
homologs in the rat genome, and the P value of SNPs associated with CAD
from the statistical results of GWAS catalog, respectively. Dark red
dots represented genes which were located at the A-B boundaries and
possessed SNPs associated with CAD. Light red dots represented genes
which were located at the A-B boundaries but did not possess SNPs
associated with CAD. Black dots represent genes which possessed SNPs
associated with CAD but were not located at the A-B boundaries. Gray
dots represent genes which were not located at the A-B boundaries and
did not possess SNPs associated with CAD. SNPs, single nucleotide
polymorphisms. CAD, coronary artery disease. i Hi-C contact maps,
ChIP-seq tracts for histone modification and regional association plot
surrounding the gene locus of COL4A1 in human genome. Hi-C data were
from human coronary artery smooth muscle cells, and histone
modification data were from human smooth muscle cells derived from the
H9 cell line. The SNPs were calculated by van der Harst P, et al. The
black lines delineate TAD boundaries and the bin size is 40 kb. The
black arrows indicated inter-TAD interactions at the A-B compartment
boundary near the COL4A1 gene locus. CAD, coronary artery disease.
To further confirm the relationship of inter-TAD interactions at the
A-B boundaries with the modulation of VSMC contractile gene expression,
we generated Hi-C contact maps, H3K9me3 ChIP-seq tracts, H3K27ac
ChIP-seq tracts (mapping active promoter and enhancer regions),
ATAC-seq tracts and RNA-seq tracts surrounding several representative
gene loci. The TADs in which the contractile-related gene loci,
including Tagln, Acta2, Col4a1, Myh11 and Myocd were located, had more
contacts with the neighboring TADs at the A-B boundaries upon Lemd3
knockdown (Fig. [221]7f, g and Supplementary Fig. [222]15a-c).
Increased interactions between the active and inactive chromatin
domains were correlated with decreased chromatin accessibility and/or
gene expression (Fig. [223]7f, g and Supplementary Fig. [224]15a-c).
Moreover, we performed nascent RNA capture and detected the
transcription-dependent RNA synthesis of Tagln and Acta2 upon Lemd3
depletion. As results, knockdown of Lemd3 caused the reduced RNA
synthesis of Tagln and Acta2 in VSMCs, supporting that these
contractile gene expression levels were directly regulated by chromatin
accessibility-dependent transcription (Supplementary Fig. [225]15d).
Since MYOCD functions as a master transcriptional regulator in
maintaining VSMC contractile identity, we accordingly verified whether
LEMD3 maintained VSMC contractile phenotype in a MYOCD-dependent
manner. Of interest, MYOCD overexpression slightly but not completely
reversed the downregulation of contractile gene expression upon Lemd3
knockdown in VSMCs (Supplementary Fig. [226]16), suggesting
MYOCD-independent mechanism, such as the direct regulation of chromatin
architecture and accessibility, substantially contributed to
LEMD3-modulated VSMC identity.
Next, we explored whether the modulation pattern of inter-TAD
interactions at the A-B boundaries on gene expression also potentially
existed in human VSMCs or related to vascular pathologies. In
accordance, we re-analyzed Hi-C data of human coronary artery smooth
muscle cells from Zhao Q et al.^[227]64. We mainly focused on 143 genes
whose expressions were maintained by LEMD3 in rat genome
(Fig. [228]7c), and found 139 homologous genes in human genome
(Supplementary Fig. [229]17a and Supplementary Data [230]7). Based on
human Hi-C data, 40 of these homologous genes were located in the TADs
at A and B compartment boundaries (Fig. [231]7h), even exhibiting
obvious inter-TAD interactions (black arrows, Fig. [232]7i), in human
VSMCs as well, as illustrated by the contractile-related genes COL4A1
and MYH11 (Fig. [233]7i and Supplementary Fig. [234]17b), suggesting
that these human genes might be regulated by inter-TAD interactions at
A-B boundaries similarly as their homologous genes in rat genome.
Furthermore, according to previously reported genome-wide association
studies (GWAS) of coronary artery disease (CAD)^[235]65–[236]67, we
identified that 5 genes (Fig. [237]7h), including COL4A1 and MYH11,
located in the TADs at A and B compartment boundaries in human genome
and possessed single nucleotide polymorphisms (SNPs) associated with
CAD (Fig. [238]7i and Supplementary Fig. [239]17b). Taken together,
Lemd3 depletion increases inter-TAD interactions at the boundaries of A
and B compartments and downregulates VSMC contractile-related gene
expression, which might correlate with human vascular diseases, such as
CAD.
Discussion
The current understanding of the regulatory mechanisms of the VSMC
phenotypic transition limitedly focuses on the transcriptional
regulation of different VSMC identity-specific transcription factors
and epigenetic regulatory mechanisms, including noncoding RNAs, DNA
methylation, and histone modifications^[240]16,[241]68–[242]70. In
addition to these mechanistic processes, whether other critical factors
and biological processes are also involved in VSMC identity modulation
has long fascinated vascular biologists. In the present study, we
utilized a genome-wide CRISPR knockout screen to identify candidates
potentially maintaining the VSMCs contractile phenotype. Furthermore,
we confirmed that the inner nuclear membrane protein LEMD3 is a
significant regulator that maintains the VSMC contractile phenotype.
LEMD3 interacts with CBX3, the reader of the H3K9me3 histone
modification in heterochromatin, and thus anchors heterochromatin at
the nuclear periphery to organize the 3D chromatin architecture and
facilitate contractile-related gene expression in VSMCs. Lemd3
deficiency in VSMCs results in subtle alterations in chromatin
organization at the TAD level and decreases chromatin accessibility and
the expression of VSMC contractile-related genes, some of which have
substantial GWAS trait-associated loci for coronary artery disease,
thereby resulting in the loss of the contractile phenotype and
exacerbating postinjury neointima formation in mice (Fig. [243]8). Our
study revealed a novel mechanism for the modulation of VSMC identity in
which organizing the 3D chromatin architecture maintains the VSMC
contractile phenotype and vascular homeostasis.
Fig. 8. Schematic illustration of the role of LEMD3 in organizing the 3D
chromatin architecture to maintain the contractile phenotype of vascular
smooth muscle cells (VSMCs).
Fig. 8
[244]Open in a new tab
Inner nuclear membrane protein LEMD3 binds to CBX3, a principal reader
of H3K9me3, subsequently anchoring heterochromatin at the nuclear
periphery. Lemd3 depletion causes the repositioning of heterochromatin
from the nuclear periphery toward the interior in VSMCs. Furthermore,
Lemd3 depletion alters the genome conformation as the increase of
inter-TAD interactions at the boundaries of A and B compartments, which
correlates with the repression of VSMC contractile gene expression.
Overall, LEMD3 organizes the 3D chromatin architecture by anchoring
heterochromatin at the nuclear periphery to maintain the VSMC
contractile identity.
Although the 3D chromatin architecture plays a critical role in
determining cell fate in development and cancer research, the role of
the 3D chromatin architecture in the cardiovascular system has not been
fully explored. In general, A/B compartment switching and TAD
reorganization substantially contribute to the modulation of the 3D
chromatin architecture. A/B compartment switching is often observed in
the differentiation and reprogramming of embryonic/pluripotent stem
cells, whereas intra-TAD interactions and loop formation mediated by
CTCF and cohesion generally indicate enhancer‒promoter communication,
which facilitates gene expression^[245]59,[246]60. Recently, several
studies have suggested that 9.8% of the genome undergoes A/B
compartment switching during cardiomyocyte differentiation^[247]31, and
chromatin loop-mediated enhancer‒promoter interactions are altered in
dilated cardiomyopathy^[248]27. To the best of our knowledge, the
current study is the first to reveal the role of the 3D chromatin
architecture in the regulation of VSMC identity and vascular
homeostasis. Intriguingly, we globally observed that the inter-TAD
interactions were changed more obviously, compared with A/B compartment
switching or alterations in intra-TAD interactions, upon Lemd3
knockdown using Hi-C analysis in the present study. Although inter-TAD
interactions are infrequently observed compared with intra-TAD
interactions^[249]71,[250]72, a few studies have revealed the
nonnegligible effects of inter-TAD interactions on regulating chromatin
accessibility and gene expression^[251]55. For example, the nuclear
matrix protein HNRNPU maintains the 3D genome architecture globally in
mouse hepatocytes, whereas the depletion of HNRNPU leads to decreased
TAD boundary strengths and increased inter-TAD interactions^[252]73. In
addition, loss of lamins causes altered interactions among TADs but not
the overall TAD structure in mouse embryonic stem cells, and gene
transcription changes are correlated with alterations in interactions
between active and inactive chromatin domains^[253]74. Our study
revealed that Lemd3 depletion increased inter-TAD interactions at the
boundaries of A and B compartments and reinforced the importance of
inter-TAD interactions in the regulation of contractile gene expression
and VSMC identity. SRF functions as the master transcription factor to
control VSMC identity. Of interest, we found that the chromatin region
in which Srf gene locus located exhibited reduced inter-TAD interaction
within A compartment upon Lemd3 knockdown in VSMCs (Supplementary
Fig. [254]18a). Since inter-TAD interactions within A compartment are
generally considered to facilitate chromatin accessibility and gene
expression^[255]75,[256]76, we accordingly observed the downregulated
accessibility and expression of Srf gene upon Lemd3 silencing in
ATAC-Seq and RNA-Seq (Supplementary Fig. [257]18a, b). Thus, the
inter-TAD interactions within A compartment might also be involved in
LEMD3-modulated chromatin architecture and VSMC contractile phenotype.
Our study revealed a pivotal role for LEMD3 in regulating VSMC
identity. To date, approximately 35 integral membrane proteins have
been shown to localize at the inner nuclear membrane of mammalian
cells, but none of these proteins have been explored in VSMCs. Through
an unbiased genome-wide CRISPR knockout screening strategy, we
identified LEMD3 as a potential regulator of the maintenance of the
VSMC contractile phenotype. LEMD3 is the longest member of the LEM
domain protein family characterized by the presence of a bihelical
motif called the LEM domain. Regarding its known modulatory effects,
LEMD3 interacts with the nuclear lamina (lamin A/C) at the nuclear
envelope–lamina–chromatin interface and functionally recruits
chromatin-modifying proteins such as HDAC3 and transcriptional
regulators such as BAF to suppress gene transcription^[258]77. In
addition, LEMD3 has been reported to interact with R-SMADs through its
carboxyl terminal RRM domain and subsequently antagonize both
TGF-β-SAMD2/3 and bone morphogenic protein (BMP)-SMAD1/5/8
signaling^[259]78. Consistently, our interactome screen also detected
the interaction of LEMD3 with R-SMADs, including TGF-β-responsive
(SMAD2/3) and BMP-responsive (SMAD1/5) SMADs. Since TGF-β signaling
facilitates smooth muscle contractile gene expression in VSMCs^[260]43,
the LEMD3-SMAD2/3 interaction, which suppresses TGF-β signaling, cannot
explain the function of LEMD3 in maintaining the VSMC contractile
phenotype well. Meanwhile, we also found the inhibition of TGF-β/SMAD
signaling did not affect VSMC alterations upon Lemd3 silencing,
excluding the potential involvement of LEMD3-SMAD2/3 interaction in
regulating VSMC phenotype switching. Interestingly, our ATAC-seq data
revealed that Lemd3 depletion upregulated BMP signaling and the bone
development pathway (Supplementary Fig. [261]13g), which could be
explained well by the loss of LEMD3 and the SMAD1/5 interaction. This
finding was consistent with clinical case reports that Lemd3
loss-of-function mutations lead to osteopoikilosis, Buschke–Ollendorff
syndrome and melorheostosis in humans^[262]79. However, the role of
LEMD3 in maintaining the contractile phenotype of VSMCs is not well
understood in the context of the upregulation of BMP signaling upon
Lemd3 depletion. Accordingly, we explored the critical role of the
LEMD3-CBX3 interaction in the maintenance of VSMC contractile identity.
Previous studies have reported that CBX3 maintains the VSMC contractile
phenotype and inhibits postinjury neointima formation by increasing SRF
transcriptional activity and repressing Notch3 expression,
respectively^[263]44,[264]80. However, the exact mechanism and direct
role of CBX3 in regulating VSMCs transcription factor activity and gene
expression are poorly understood. CBX3 is the principal reader of
H3K9me2/3, one of the major repressive histone modifications, which
frequently exist in the inactive domain of chromatin, such as
heterochromatin^[265]81. The binders of these repressive histone
modifications are generally located at the nuclear periphery through
interactions with nuclear envelope proteins, thereby determining the
perinuclear anchoring of heterochromatin. For example, HP1α, which
belongs to the same protein family as CBX3, recognizes H3K9me3 in
heterochromatin, whereas the inner nuclear membrane protein lamin B
receptor (LBR) interacts with H3K9me3-bound HP1α and functions as an
anchoring site of heterochromatin for perinuclear
distribution^[266]82,[267]83. In the present study, we found that the
inner nuclear membrane protein LEMD3 also functions as an anchoring
site and interacts with H3K9me3-bound CBX3 to anchor heterochromatin at
the nuclear periphery. The interaction between CBX3 and LEMD3 organized
the 3D genome in VSMCs, which might be not only involved in the
maintenance of the VSMC contractile phenotype but also essential for
embryonic vascular development, since both CBX3 knockout and LEMD3
knockout cause abnormal vascular development^[268]37,[269]44, possibly
due to the dysfunction of VSMC differentiation, although further
investigation is needed. Of interest, RRM domain, the binding domain of
LEMD3 with CBX3, has also been reported to mediate the interaction of
other proteins (e.g., R-SMAD, transcription regulators GCL and BTF)
with LEMD3^[270]77,[271]84. Considering overexpression of RRM domain
may also influence on the binding of other proteins to LEMD3 in VSMCs,
we could not completely exclude that LEMD3 anchors heterochromatin at
the nuclear periphery and maintains VSMC contractile phenotype also
dependently on other interacting proteins, although we demonstrated
that the effect of LEMD3 on VSMC phenotype modulation was independent
on TGF-β/SMAD signaling (Supplementary Fig. [272]8). To further clarify
the specific role of LEMD3-CBX3 interaction in heterochromatin
anchoring as well as the maintenance of VSMC contractile phenotype, the
specific amino acid residues mediating LEMD3-CBX3 interaction need to
be identified, through unbiased RRM-interacting protein profiling and
subsequent site-directed mutagenesis assays of RRM domain to compare
the differences among the interactions with all RRM-interacting
proteins, which requires further investigation in future studies.
In addition, we did not explore the role of VSMC-specific LEMD3 in
female mice, because Myh11-CreER^T2 transgene is inserted into the Y
chromosome. Alternatively, we excluded the potential sex-specific
difference using primary aortic smooth muscle cells isolated from male
and female Lemd3^flox/flox mice, as evidenced by the loss of
contractile phenotype in both male and female background VSMCs upon
Cre-overexpressed adenovirus-mediated Lemd3 knockout, whereas
additional in vivo evidence might require further investigation on
other mouse line including female VSMC-specific Lemd3 knockout mice. Of
interest, SMC-targeting Cre lines driven by the Myh11 promoter also
display high activity in visceral SMCs^[273]85. Thus, loss of Lemd3 in
intestinal SMCs might impair intestinal contraction and consequently
cause intestinal obstructions. However, we did not observe the
significant phenotype of intestinal obstructions in Lemd3^SMKO mice
till 16-week age under basal or physiological conditions. Further
investigation might be required to explore the role of LEMD3 in
visceral SMCs, especially under some intestinal pathological
conditions.
Taken together, our study revealed that the inner nuclear membrane
protein LEMD3 organizes the 3D chromatin architecture by anchoring
heterochromatin at the nuclear periphery through its interaction with
CBX3, a principal reader of the heterochromatin-associated histone
modification H3K9me3. Lemd3 depletion caused the repositioning of
heterochromatin from the nuclear periphery toward the interior, as well
as the enhancement of inter-TAD interactions at the boundaries of A and
B compartments in chromatin, which correlated with the downregulation
of contractile-related genes and the subsequent loss of the contractile
phenotype. Since LEMD3-modulated genes involve some GWAS
trait-associated loci for coronary artery disease, LEMD3 dysfunction
might correlate with the pathogenesis of coronary artery disease. Thus,
LEMD3 functions as a vital regulator to maintain the VSMC contractile
identity and prevent vascular disease by organizing the 3D chromatin
architecture.
Methods
Experimental materials
An antibody against LEMD3 (ab121854) for western blot analysis
(dilution 1:1000) and immunohistochemical staining (dilution 1:100) was
purchased from Abcam (Cambridge, UK). An antibody against LEMD3
(orb107113) for co-immunoprecipitation (Co-IP) assays was purchased
from Biorbyt (Cambridge, UK). An antibody against Lamin B1 (66095-1-Ig,
dilution 1:200) for immunofluorescence staining was purchased from
Proteintech Group, Inc. (Wuhan, China). Antibodies against ACTA2
(ab5694), CNN1 (ab46794), and TAGLN (ab14106) for western blotting
(dilution 1:1000) and immunohistochemical staining (dilution 1:200)
were purchased from Abcam (Cambridge, UK). Antibodies against SRF
(16821-1-AP, dilution 1:1000) and GAPDH (60004-1-Ig, dilution 1:5000)
for western blotting were purchased from Proteintech Group, Inc.
(Wuhan, China). An antibody against CBX3 (ab217999) for western
blotting (dilution 1:1000) and Co-IP was purchased from Abcam
(Cambridge, UK). The anti-FLAG-M2 (F3165) antibody used for the Co-IP
assay was purchased from Sigma‒Aldrich (St. Louis, MO, USA). An
antibody against H3K9me3 (ab176916) for immunofluorescence staining
(dilution 1:2000), western blotting (dilution 1:1000), and ChIP-Seq was
purchased from Abcam (Cambridge, UK). An antibody against H3K9me2
(ab176882, dilution 1:500) for immunofluorescence staining was
purchased from Abcam (Cambridge, UK). An antibody against histone H3
(ab1791, dilution 1:1000) for western blotting was purchased from Abcam
(Cambridge, UK). An antibody against H3K27ac (ab4729) for ChIP-Seq was
purchased from Abcam (Cambridge, UK). IRDye-conjugated secondary
antibodies (dilution 1:10000) for western blot analysis were purchased
from Rockland, Inc. (Gilbertsville, USA). Normal mouse IgG (sc-2025)
used as a negative control for the co-IP assay and immunostaining was
obtained from Santa Cruz Biotechnology, Inc. (Dallas, USA). Normal
rabbit IgG (2729S) used for the co-IP and immunostaining assays was
obtained from Cell Signaling Technology (Boston, USA). YF633-phalloidin
(BN10053) was purchased from Biorigin, Inc. (Beijing, China).
4,6-Diamidino-2-phenylindole (DAPI) (62248), Alexa Fluor 488-conjugated
goat anti-rabbit IgG (A-11008, dilution 1:1000), and Alexa Fluor
555-conjugated goat anti-mouse IgG (A-21422, dilution 1:1000) for
immunofluorescence staining were purchased from Thermo Fisher
Scientific (Rochester, USA). PDGF-BB (100-14B-50) and TGF-β
(100-621C-2) were purchased from PeproTech, Inc. (Rocky Hill, USA).
Rapamycin (HY-10219) and selective SMAD3 inhibitor SIS3 (HY-13013) were
purchased from MCE (Shanghai, China). The Amaxa^® Cell Line
Nucleofector^® Kit V (VCA-1003) used for electrotransfection was
purchased from Lonza (USA). The Lipofectamine^® RNAiMAX reagent
(13778075) used for siRNA transfection was purchased from Invitrogen
(Carlsbad, USA). The JetPEI Transfection reagent used for plasmid
transfection was purchased from Polyplus-transfection SA (Strasbourg,
France). BrdU (5-bromo-2-deoxyuridine) (19–160) was purchased from
Sigma–Aldrich (St. Louis, MO, USA). A SanPrep column DNA gel extraction
kit used for the purification of DNA fragments from PCR or agarose gels
was obtained from Sangon Biotech, Inc. (Shanghai, CN). The ClonExpress
II One-Step Cloning Kit was purchased from Vazyme Biotech Co. (Nanjing,
CN). T1/Mach1-T1 phage-resistant chemically competent E. coli (ZC102-2)
was purchased from ZOMANBIO, Inc. (Beijing, China). The NucleoBond Xtra
Midi Plus Kit used for plasmid isolation from E. coli was purchased
from MACHEREY-NAGEL, Inc. (Bethlehem, PA, USA). The NanoBiT^®
protein–protein interaction assay kit (N2014) was purchased from
Promega Corporation (Madison, WI, USA).
Cell lines
Primary rat vascular smooth muscle cells (VSMCs) were isolated from the
thoracic aortas of 150–180 g male Sprague‒Dawley rats via collagenase
digestion^[274]86. Primary rat VSMCs were cultured in low-glucose
Dulbecco’s modified Eagle’s medium (DMEM) (Gibco, USA) supplemented
with 10–20% fetal bovine serum (FBS), and cells at passages 3–6 were
used in subsequent experiments. Primary human artery smooth muscle
cells (SMCs) isolated from arteries of the human umbilical cord were
maintained in Nutrient Mixture F12 Ham Kaighn’s Modification (F12K,
N6760-10X1L, Sigma Aldrich) supplemented with 10% FBS, 10% SMC Growth
Medium (Cell Applications), 100 units/ml penicillin and 100 μg/ml
streptomycin. Primary human artery SMCs at passages 4 to 6 were used in
subsequent experiments. The mouse aortic smooth muscle cell line MOVAS
(CRL-2797), rat aortic smooth muscle cell line A7r5 (CRL-1444) and
HEK293T cells (CRL-3216) were purchased from the American Type Culture
Collection (ATCC). MOVAS, A7r5 and HEK293T cells were cultured in
high-glucose DMEM supplemented with 10% FBS, 100 units/ml penicillin
and 100 μg/ml streptomycin. All the cells were maintained in a
water-saturated 5% CO[2] incubator at 37 °C.
Generation of the EGFP reporter-expressing MOVAS cell line
A custom-designed CRISPR guide oligo was cloned and inserted into the
PX458 plasmid, which contains a cassette for the transient expression
of SpCas9. The upstream and downstream homologous arms of the Acta2
gene and P2A-EGFP expression cassette were amplified via PCR and
subsequently cloned and inserted into the pUC19 plasmid to obtain the
homology template. MOVAS cells were transfected with homology templates
and PX458 CRISPR‒Cas9 plasmids using the Amaxa^® Cell Line
Nucleofector^® Kit V (Lonza, USA). Ninety-six hours after transfection,
EGFP-positive cells were sorted via flow cytometry to obtain
single-cell clones. Single-cell clones containing the endogenous
Acta2-P2A-EGFP gene sequence were further verified by sequencing
analysis.
The following sgRNAs were used in this study:Acta2-sgRNA-1:
5’-GGACTTAGAAGCATTTGCGG-3’; Acta2-sgRNA-2: 5’-AACAGGAATACGACGAAGCT-3’.
Lentivirus production
We used the lentiCas9-Blast plasmid (Addgene 52962) and the mouse
CRISPR Gecko v2 pooled library (Addgene 1000000052) to generate
lentiviruses. For each 10 cm dish, HEK293T cells at 90% confluence were
transfected with 6.8 μg of the plasmids of interest mentioned above,
5.2 μg of psPAX2 (Addgene 12260), and 3.4 μg of pMD2.G (Addgene 12259)
using 35 μl of Lipofectamine 3000 (Thermo Fisher Scientific L3000150),
30 μl of P3000 Enhancer (Thermo Fisher Scientific L3000150), and 2.5 ml
of Opti-MEM (Thermo Fisher Scientific 31985070). The media were changed
6 h after transfection. The supernatant containing the lentivirus was
harvested at 48 h posttransfection, filtered with a 0.45-μm PVDF
filter, and stored at −80 °C.
Genome-wide CRISPR knockout screening
The EGFP reporter-expressing MOVAS cell line was transduced with
lentiCas9-Blast. Blasticidin (10 µg/ml) was added to the cells at 72 h
after infection, and the mixture was maintained in culture for 7 days
to screen the cells stably overexpressing Cas9, which was further
verified by western blot analysis. A total of 240 million MOVAS cells
that stably expressed EGFP and Cas9 were subsequently seeded into
twenty 10 cm plates and infected with the pooled GeCKO v2 mouse
lentiviral library (over 300 × coverage) at an MOI of 0.3 to ensure
that most cells took up only one sgRNA. Puromycin (10 µg/ml) was added
to the cells at 72 h after infection, and the cells were maintained in
culture for 7 days. The cells were subsequently sorted via two rounds
of flow cytometry to enrich the EGFP^low population.
Genomic DNA was purified from the EGFP^low population, as well as from
the initial infected population, using the Gentra Purogene Kit
(Qiagen). Integrated sgRNAs were enriched by PCR amplification, with
eight replicate PCRs each with 1 µg of template per reaction, to
maintain complex sampling of the cellular population. Adapter sequences
and per-sample barcodes were added to the libraries via a second round
of PCR. The libraries were pooled and sequenced on an Illumina MiSeq
instrument using 150-bp single-end reads. The enriched genes were
identified from the sgRNA sequencing results using Model-based Analysis
of Genome-wide CRISPR-Cas9 Knockout (MAGeCK). The MAGeCK algorithm can
prioritize enriched genes by comparing the sgRNAs in EGFP^low cells to
those in unsorted cells. Briefly, the read counts of each sgRNA from
different samples were normalized to adjust for the effects of the
library size and read count distribution. The enriched genes were
subsequently identified by searching for genes whose sgRNAs ranked
consistently higher using robust rank aggregation (RRA). Genes with
smaller RRA values ranked higher in the knockout screen.
Flow cytometry
For the analysis of the EGFP fluorescence intensity during screening,
the cells were suspended in PBS supplemented with 2% FBS for direct
detection. The cells were then analyzed on an ARIA-SORP sorter (Becton
Dickinson, USA). The EGFP fluorescence intensity was detected using the
FITC channel. The data were analyzed using FlowJo software.
Human and animal samples
Human internal mammary artery samples were obtained from Peking
University People’s Hospital, Beijing, China. The study involving human
tissue was approved by the Medical Ethics Committee of Peking
University People’s Hospital. Internal mammary arteries were obtained
from patients that underwent coronary artery bypass/valve replacement.
The informed consent form was signed by the donors before collection.
All sample collection and experimental procedures followed national
laws and international ethical and technical guidelines. The entire
procedure was examined and approved by the Ethics Board at the
Institute of Peking University People’s Hospital (approval no.
2023PHB170-001). Human internal mammary artery samples were immediately
placed in 4% formalin, embedded in paraffin, and made into serial
sections for subsequent immunohistochemical staining.
Lemd3^flox/flox mice in which exon 2 was flanked with LoxP sites were
generated from C57BL/6J mice using the CRISPR/Cas9 system. Genotyping
was performed by PCR using the following two primers:
forward-5’-CGGACAGTGAGCGAGGCATT-3’ and
reverse-5’-AATGTAAACAAACGAGGTAAGAAGC-3’. The PCR program was as
follows: 94 °C, 2 min; 98 °C, 10 s; 60 °C, 30 s; 68 °C, 30 s (35
cycles); 68 °C, 10 min; and 4 °C, hold. The length of the PCR product
was 217 bp (+) or 183 bp (-). The Myh11-CreER^T2 mice were kindly
provided by Prof. Wei Li from Peking University People’s
Hospital^[275]87. Since the bacterial artificial chromosome containing
Myh11-CreER^T2 was inserted into the Y chromosome, only male mice were
used. We intercrossed Myh11-CreER^T2 Lemd3^flox/wt male mice with
Lemd3^flox/wt female mice to obtain offspring Myh11-CreER^T2
Lemd3^flox/flox male mice as well as the littermate Myh11-CreER^T2 male
mice. Eight-week-old male Myh11-CreER^T2 Lemd3^flox/flox mice and
Myh11-CreER^T2 mice were treated with 75 mg/kg tamoxifen for 5
consecutive days to generate smooth muscle cell-specific Lemd3 knockout
(Lemd3^SMKO) mice and Lemd3^WT control mice, respectively. The primers
used for Myh11-CreER^T2 genotyping were as follows:
SMWT1-5’-TGACCCCATCTCTTCACTCC-3’, SMWT2-5’-AACTCCACGACCACCTCATC-3’ and
phCREAS1-5’-AGTCCCTCACATCCTCAGGTT-3’. The PCR program was as follows:
95 °C for 1 min; 95 °C for 10 s, 52 °C for 30 s, and 72 °C for 45 s (35
cycles); 72 °C for 5 min; and hold at 4 °C. The length of the PCR
product was 287 bp (+). All experimental animals were maintained in
standard cages in an SPF environment with a 12-h light/dark cycle in
the Department of Laboratory Animal Science, Peking University Health
Science Center. All studies were conducted in compliance with the
guidelines of the Animal Care and Use Committee of Peking University.
Primary aortic smooth muscle cell isolation
VSMCs were isolated from the thoracic aortas from 6- to 10-week-old
Lemd3^flox/flox mice as described before^[276]88, with minor
modifications. Briefly, we separated the thoracic aortas from the mice.
The isolated aortas were washed twice in ice-cold PBS and cultured in
1 mL 0.2% collagenase I solution in Ham’s F12 medium at 37 °C for
30 min. The adventitia was stripped away from the aorta using forceps
under microscopic guidance. The aortas were opened longitudinally and
the endothelial cells were gently scraped off. The aortas were then
divided into small pieces, placed at the bottom of the culture dish,
and cultured for several days in DMEM/F-12 medium containing 20% FBS at
37 °C in a humidified atmosphere with 5% CO[2]. Cells that had migrated
from the explants were collected and maintained in the growing medium.
VSMCs at passages 3–6 were used for further experiments. Purity of
VSMCs was confirmed via positive staining for ACTA2.
Real-time quantitative polymerase chain reaction (RT‒qPCR)
Total RNA was extracted using TRIzol reagent (Vazyme Biotech Co.,
Nanjing, China), and equal amounts (1 μg) were reverse-transcribed into
cDNA using a reverse transcription kit (Vazyme Biotech Co., Nanjing,
China). Quantitative real-time PCR (RT‒qPCR) was performed using 2 ×
SYBR Green PCR mix (Vazyme Biotech Co., Nanjing, China) according to
the manufacturer’s instructions. All amplification reactions were
conducted in 40 cycles and were performed in triplicate. The data were
analyzed via the ΔΔCT method. All samples were normalized to β-actin.
The primers used for RT‒qPCR are presented in the Supplementary
Table [277]2.
Immunohistochemical and immunofluorescence staining
For immunohistochemical staining, cultured cells were fixed with 4%
paraformaldehyde for 15 min and then permeabilized with 0.5% Triton
X-100 for 20 min, while tissue sections were directly permeabilized
with 0.5% Triton X-100 for 20 min. Cells or sections were incubated
with primary antibodies at 4 °C overnight and then with secondary
antibodies before being stained with a DAB kit (ZSGB-BIO, Beijing,
China). The nuclei were counterstained with hematoxylin. Cells or
sections incubated with species-matched IgG were used as negative
controls. Images were obtained by using a Zeiss microscope.
Immunohistochemical images were quantified by using ImageJ software. We
first selected the positive area according to the same criteria, and
then use the average brightness multiplied by the area as its
expression intensity, and results were expressed as the ratio of
relative intensity compared with the controls.
For immunofluorescence staining, cultured cells were fixed with 4%
paraformaldehyde for 15 min and then permeabilized with 0.5% Triton
X-100 for 20 min. The sections were directly permeabilized with 0.5%
Triton X-100 for 20 min. The cells or sections were incubated with
primary antibodies at 4 °C overnight, followed by an incubation with a
secondary Alexa Fluor 488-conjugated goat anti-rabbit IgG at a 1:1000
dilution and secondary Alexa Fluor 555-conjugated goat anti-mouse IgG
(Thermo Fisher Scientific, Rochester, USA) at a 1:1000 dilution for
1 hour at room temperature. The nuclei were stained with DAPI at a
1:1000 dilution for 10 min at room temperature. For F-actin staining,
the cells were fixed with 4% paraformaldehyde for 15 min and then
permeabilized with 0.5% Triton X-100 for 20 min. Next, the cells were
incubated with YF633-phalloidin at a 1:100 dilution for 1 h at room
temperature. The nuclei were stained with DAPI at a 1:1000 dilution for
10 min at room temperature. The fluorescence signals were monitored
using a confocal laser scanning microscope (Leica Microsystems,
Wetzlar, Germany).
Western blotting
Cells or mouse tissues were lysed in RIPA buffer (Beyotime, Shanghai,
China), and whole-cell protein samples were extracted. Protein
concentrations were evaluated using a BCA protein assay kit (Beyotime,
Shanghai, China). Equal amounts of total protein were resolved on 8–12%
SDS‒PAGE gels and subsequently transferred onto nitrocellulose
membranes (PALL Inc., Wilmington, DE, USA). The membranes were blocked
with 5% milk or BSA in TBST. After being incubated with primary
antibodies at 4 °C overnight, the membranes were incubated with
IRDye-conjugated secondary antibodies (Rockland, Inc., Gilbertsville,
PA, USA) for 1 hour at room temperature. The immunofluorescence signals
were subsequently detected using an Odyssey infrared imaging system
(LI-COR Biosciences, Lincoln, NE, USA).
siRNA transfection
Small interfering RNAs (siRNAs) targeting rat Lemd3, Cand1, Ajuba, Gdi1
or Xpo6, and scrambled siRNAs were designed and synthesized by
GenePharma (Suzhou, China). Rat VSMCs were transfected with siRNAs
(20 nM) using RNAiMax reagent (Invitrogen, CA, USA) according to the
manufacturer’s protocol. The list of siRNA sequences is presented in
the Supplementary Table [278]3.
Collagen gel contraction assay
Primary rat VSMCs transfected with scrambled siRNAs (20 nM) or siRNAs
targeting Lemd3 (20 nM) were resuspended in DMEM containing 10% FBS at
a density of 5 × 10^5 cells/ml. A collagen lattice was prepared by
mixing the cell suspension and an ice-cold collagen gel solution at a
volume ratio of 1:4. Then, 0.5 ml of the mixture was added to a well of
a 24-well plate, followed by an incubation at 37 °C for 1 h. After
collagen polymerization, 1 ml of DMEM was added on top of each collagen
gel lattice. Twenty-four hours later, the plates were scanned, and the
area of the gel in each well was analyzed using ImageJ software.
Wire myography
Wire myography was performed to assess the contractility of the aortic
rings and mesenteric resistance arteries from 12-week-old Lemd3^WT mice
and Lemd3^SMKO mice. Briefly, the aortic rings and mesenteric
resistance arteries were isolated and removed from the mice following
euthanasia. The dissected vessels were immediately placed in
Krebs‒Henseleit buffer. After the removal of additional connective
tissues, the aortic rings and mesenteric resistance arteries were
sectioned to approximately 1.5 mm in length and subsequently subjected
to vascular tension measurements. The cumulative dose response to
phenylephrine (Phe) was obtained to characterize vasocontraction. The
data are presented as percentages of the maximal contraction induced by
60 mM KCl.
In vitro EdU incorporation assay
A BeyoClick™ EdU Cell Proliferation Detection Kit (Beyotime, Shanghai,
China) was used to detect cell proliferation in accordance with the
manufacturer’s instructions. Primary rat VSMCs were transfected with
the scrambled siRNA (20 nM) or siRNA targeting Lemd3 (20 nM). 48 h
after transfection, 20 μΜ EdU was added, and the cells were
subsequently incubated for another 3 h. Following fixation with 4%
paraformaldehyde for 15 min and permeabilization with 0.5% Triton X-100
for 20 min, the cells were incubated with Click reaction cocktail for
30 min and then stained with DAPI at a 1:1000 dilution for 10 min. The
cells were observed and photographed with an inverted fluorescence
microscope (Olympus, Japan).
Migration assay
Cell migration was determined by performing wound healing and transwell
migration assays. For the wound healing assay, primary rat VSMCs
transfected with the scrambled siRNA or siRNA targeting Lemd3 were
seeded in six-well plates and wounded by manually scraping the cells
with a 200 µl pipette tip. The medium was replaced with FBS-free DMEM
supplemented with or without PDGF-BB (20 ng/ml). The migration areas
were monitored at 0 h and 12 h postwounding under a microscope. For the
transwell migration assay, primary rat VSMCs transfected with the
scrambled siRNA or siRNA targeting Lemd3 were seeded in the top
chambers of 8-µm pore size transwells in plates (Corning Costar,
Lowell, MA) in FBS-free DMEM. In addition, 0.6 ml of DMEM supplemented
with 20% FBS was added to the well of the plate (lower compartment).
After a 12-h incubation at 37 °C, the cells on the upper surface of the
membrane were removed via gentle abrasion with a cotton bud, and the
cells that migrated to the lower surface were fixed with 4%
paraformaldehyde for 15 min and stained with 0.1% crystal violet. The
cells were subsequently washed with PBS several times to remove the
excess dye. The mean number of cells on the lower side of the surface
was counted in 4 randomly chosen high-power fields (HPFs) under a light
microscope. ImageJ software was used to measure the migration area and
cell number.
Mouse carotid artery wire injury
A carotid artery wire injury model was generated in 12-week-old male
Lemd3^WT mice and Lemd3^SMKO mice. Briefly, after isolation of the left
common carotid artery, carotid clamps were used to block the blood flow
of the left common carotid artery and the left internal carotid artery.
Next, a curved flexible wire (0.38 mm) was inserted into the left
external carotid artery to induce carotid artery injury. The wire was
subsequently removed, and the left external carotid artery was tied
proximally to the aortic arch. The mouse carotid arteries were
harvested at 28 days after wire injury. The mouse carotid arteries were
collected for hematoxylin‒eosin staining and immunofluorescence
staining.
Quantification of neointima formation
The mouse carotid arteries were harvested and embedded in Tissue-Tek
OCT (Sakura Finetek, Staufen, Germany). Every other section (7 μm each)
was collected within a standardized distance from the bifurcation point
(1000 to 2000 μm, the segments marked with curly brackets in
Supplementary Fig. [279]7e) and these sections were divided into 20
slides, so that every slide contained eight sections at 140 μm
intervals^[280]89,[281]90. Following H&E staining on the
cross-sections, the neointima was defined as the region between the
lumen and internal elastic lamina. The medial wall was defined as the
region between the internal and external elastic laminas. The areas of
the neointima and media on cross-section of H&E-stained artery segments
were measured with a computerized image analysis system (Image-Pro Plus
6.0 software) by two independent investigators in a blinded manner. For
quantitative analysis of neointima formation, the mean of eight
sections on one slide was taken for each mouse.
Measurement of blood pressure
Blood pressure was measured in Lemd3^WT mice and Lemd3^SMKO mice before
and after surgery using a CODA Mouse & Rat Tail-Cuff Blood Pressure
System (Kent Scientific Co., Connecticut, USA) according to the
scientific statement from the American Heart Association^[282]91. The
equipment was kept clean and free from foreign scents and blood odors.
The investigator was blinded with respect to the experimental groups to
perform the measurements, while the mice were tested in a randomized
manner. The mice underwent 7 consecutive days of training sessions from
1 to 5 P.M. each day to become accustomed to the tail-cuff procedure.
The measurements were then recorded by a single investigator at the
same time on 3 consecutive days. Five measurements were performed daily
on each mouse, and the average of 15 measurements represented the
systolic blood pressure of each mouse.
In vivo EdU incorporation assay
A carotid artery wire injury model was generated in 12-week-old male
Lemd3^WT mice and Lemd3^SMKO mice as described above. The mice were
sacrificed at 28 days after surgery. The mice were intraperitoneally
injected with 50 mg/kg 5-ethynyl-2’-deoxyuridine (EdU) 18 h prior to
sacrifice. Arterial sections were washed with PBS and incubated with
0.5% Triton X-100 for 20 min to permeabilize the cell membrane. The
sections were incubated with Click reaction cocktail for 30 min and
then stained with DAPI at a 1:1000 dilution for 10 min. The sections
were observed and photographed with an inverted fluorescence microscope
(Olympus, Japan).
LEMD3 interactome
LEMD3-binding proteins were identified via liquid chromatography tandem
mass spectrometry (LC‒MS/MS)^[283]92. Briefly, HEK293T cells were
transfected with the pcDNA3.1 or LEMD3-FLAG plasmid. Forty-eight hours
after transfection, the cells were lysed in lysis buffer (50 mM
Tris-HCl [pH=7.4], 150 mM NaCl, 2 mM EDTA, and 0.3% NP-40). The cell
lysate samples were incubated with anti-FLAG M2 affinity gel (A2220,
Sigma‒Aldrich) according to the manufacturer’s instructions. The
anti-FLAG M2 affinity gel-bound proteins were eluted with FLAG peptide
(F3290, Sigma). The final eluent was collected and resolved on an
SDS-PAGE gel. The gel was fixed with a 30% ethanol:10% acetic acid
solution, washed with 10% ethanol, and subjected to standard reduction,
alkylation, and in-gel tryptic digestion. The resulting peptides were
lyophilized and reconstituted in 0.1% (v/v) aqueous formic acid.
Peptide separation was performed using an Easy-nanoLC system coupled
online to a LTQ Orbitrap Velos Pro mass spectrometer. Chromatography
utilized a reversed-phase column with mobile phase A (0.1% formic acid
in water) and mobile phase B (0.1% formic acid in acetonitrile) at a
flow rate of 300 nL/min. The elution gradient was programmed as
follows: 2% to 40% B over 70 min, 40% to 95% B over 5 min (70–75 min),
followed by a 5-min hold at 95% B. Eluting peptides were ionized via a
nano-electrospray ionization source (spray voltage: 2.2 kV; capillary
temperature: 300 °C) and analyzed by the LTQ Orbitrap Velos Pro.
Full-scan MS spectra (m/z 350–2000) were acquired in the Orbitrap
analyzer. In data-dependent acquisition mode, the 15 most intense
precursor ions from each full scan were sequentially selected for
CID-MS/MS fragmentation in the linear ion trap, using a normalized
collision energy of 35%. Dynamic exclusion was enabled with a duration
of 30 s to prevent repeated sequencing of abundant ions. Protein
identification was performed using Proteome Discoverer software
(version 1.4) with database searches against the UniProt Human Proteome
database. Search parameters included a precursor mass tolerance of 10
ppm, a fragment mass tolerance of 0.6 Da, and trypsin specificity
allowing up to two missed cleavages. Fixed modification was set as
carbamidomethylation of cysteine (+57.021 Da), while oxidation of
methionine (+15.995 Da) was specified as a variable modification. All
peptide-spectrum matches (PSMs) were filtered at a strict false
discovery rate (FDR) threshold of ≤ 1%. Two independent biological
replicates of LC‒MS were performed. Each replicate included an
experimental sample and a negative control sample. Samples from HEK293T
cells transfected with the pcDNA3.1 plasmid were used as the negative
controls.
Plasmid construction and transfection
The cDNA fragments encoding FLAG-tagged full-length human LEMD3 (NCBI
reference sequence: [284]NM_014319.5, LEMD3-FLAG, aa 1–911), LEMD3
ΔC-terminal mutant aa 1–876 and LEMD3 ΔRRM mutant aa 1–784, as well as
human CBX3 (NCBI reference sequence: [285]NM_016587.4), were
subsequently cloned and inserted into the pcDNA3.1 plasmid. Cells
cultured in 6-well plates were transfected with 2 μg of plasmid per
well using jetPEI (Polyplus-transfection SA, Strasbourg, France).
Coimmunoprecipitation (Co-IP)
HEK293T cells transfected with LEMD3-FLAG and CBX3 plasmids were lysed
with immunoprecipitation (IP) buffer (50 mM Tris-HCl [pH=7.4], 150 mM
NaCl, 2 mM EDTA and 0.3% NP-40) supplemented with a protease inhibitor
cocktail to analyze the binding of LEMD3 and CBX3. The lysates were
incubated with either anti-FLAG M2 antibody (2 μg), anti-CBX3 antibody
(2 μg) or negative control IgG antibody (2 μg) at 4 °C overnight. Then,
Protein A/G agarose beads (Santa Cruz, CA, USA) were added and
incubated for another 4 h at 4 °C. The precipitated proteins were
eluted from the beads by resuspending the beads in 2× SDS‒PAGE sample
loading buffer. After boiling for 10 min at 95 °C, the protein samples
were analyzed via western blotting.
Cultured rat VSMCs and mouse aortic tissues were lysed with IP buffer
supplemented with a protease inhibitor cocktail to analyze the
endogenous binding of LEMD3 and CBX3. The lysates were incubated with
either an anti-LEMD3 antibody (2 μg), anti-CBX3 antibody (2 μg) or
negative control IgG antibody (2 μg) at 4 °C overnight. Then, Protein
A/G agarose beads (Santa Cruz, CA, USA) were added and incubated for
another 4 h at 4 °C. The precipitated proteins were eluted from the
beads by resuspending the beads in 2 × SDS‒PAGE sample loading buffer.
After boiling for 10 min at 95 °C, the protein samples were analyzed
via western blotting.
NanoBiT-based binding assay
The NanoBiT^® protein–protein interaction assay kit (N2014) was
purchased from Promega Corporation (Madison, WI, USA). SmBit-CBX3
expression constructs were generated by cloning CBX3 into the pBiT2.1-N
vector containing the N-terminal SmBit. LgBit-LEMD3, LgBit-ΔC-terminal,
and LgBit-ΔRRM expression constructs were generated by cloning LEMD3,
the ΔC-terminal mutant, and the ΔRRM mutant into the pBiT1.1-N vector
containing the N-terminal LgBit, respectively. HEK293T cells were
transfected with empty vector or SmBit-CBX3 and LgBit-LEMD3,
LgBit-ΔC-terminal, or LgBit-ΔRRM plasmids in white 96-well cell culture
plates. Thirty-six hours after transfection, the culture media were
removed and replaced with HBSS containing 1 mg/ml bovine serum albumin
(BSA). The substrate furimazine was then added to each well. The plate
was incubated for 30 min, and luminescence was measured using a Thermo
Scientific Varioskan Flash reader.
Transmission electron microscopy (TEM)
The descending aortas of the Lemd3^WT mice and Lemd3^SMKO mice were
isolated, cut into 3 mm rings, and fixed with 3% glutaraldehyde in
0.1 M sodium cacodylate buffer (pH 7.4) for 3 h. After being washed in
cacodylate buffer, the tissues were postfixed with 1% OsO[4],
dehydrated through an ethanol series, infiltrated with propylene oxide
and embedded in Epon 812. Ultrathin sections were cut on an
ultramicrotome (LKB Huxley, Stockholm, Sweden) and counterstained with
uranyl acetate and lead citrate. Images were acquired using a
transmission electron microscope (HITACHI H-7000, Tokyo, Japan). We
randomly selected four vascular smooth muscle cells from each mouse
aortic section to measure the width of perinuclear heterochromatin. The
width of perinuclear heterochromatin was quantified by measuring the
width of the binarized chromatin from 1D intensity profiles along the
nuclear perimeter, sampled at every 10 perimeter pixels^[286]93, by two
independent investigators in a blinded manner.
Bulk RNA-seq library preparation
The A7r5 rat VSMC line was transfected with the scrambled siRNA (20 nM)
or siRNA targeting Lemd3 (20 nM). RNA was extracted from rat VSMCs
using the Qiagen RNeasy Mini Kit according to the manufacturer’s
instructions. The RNA-seq libraries were prepared using the NEBNext
Ultra RNA Library Pre Kit for Illumina and sequenced on the Illumina
NextSeq platform.
ChIP-seq library preparation
A7r5 rat VSMCs transfected with the scrambled siRNA or siRNA targeting
Lemd3 were crosslinked with 1% formaldehyde for 10 min at room
temperature and quenched by an incubation with glycine added at a final
concentration of 125 mM for 5 min. The fixed cells were resuspended in
lysis buffer (1% SDS, 5 mM EDTA, and 50 mM Tris-HCl, pH 8.1) containing
protease inhibitors and then subjected to 30 cycles (30 s on and 30 s
off) of sonication (Bioruptor, Diagenode) to generate chromatin
fragments of ~300 bp in length. Lysates were diluted in buffer (1%
Triton X-100, 2 mM EDTA, 150 mM NaCl, and 20 mM Tris-HCl, pH 8.1)
containing protease inhibitors. For immunoprecipitation, the diluted
chromatin was incubated with normal IgG (control), H3K9me3, or H3K27ac
antibodies (3 μg) overnight at 4 °C with constant rotation, followed by
an incubation with 50 μl of 50% (v/v) Protein A/G Sepharose beads for
an additional 2 h. The beads were washed with the following buffers:
TSE I (0.1% SDS, 1% Triton X-100, 2 mM EDTA, 150 mM NaCl, and 20 mM
Tris-HCl, pH 8.0); TSE II (0.1% SDS, 1% Triton X-100, 2 mM EDTA, 500 mM
NaCl, and 20 mM Tris-HCl, pH 8.0); and TSE III (0.25 M LiCl, 1% NP-40,
1% sodium deoxycholate, 1 mM EDTA, and 10 mM Tris-HCl, pH 8.0). The
pulled-down chromatin complexes were eluted with TE (1 mM EDTA and
10 mM Tris-HCl, pH 8.0), and the inputs were decrosslinked at 55 °C for
12 h in elution buffer (1% SDS and 0.1 M NaHCO[3]). The DNA was
purified with a QIAquick PCR Purification Kit (QIAGEN). ChIP-seq
libraries were prepared using the VAHTS^® Universal Pro DNA Library
Prep Kit for Illumina (Vazyme Biotech Co., Nanjing, China) and
sequenced on the Illumina NextSeq platform.
ATAC-seq library preparation
ATAC-seq libraries were prepared according to the standard ATAC-seq
protocol with minor modifications^[287]94. Briefly, 50,000 cells
transfected with the scrambled siRNA (20 nM) or siRNA targeting Lemd3
(20 nM) were lysed on ice in a tube containing 50 μl of cold lysis
buffer. The nuclei were collected and subjected to a transposition
reaction in a preheated metal bath at 37 °C. Then, the DNA fragments
were extracted and purified with a PCR purification kit. The DNA was
then resuspended in 29 μl of ddH[2]O. ATAC-seq libraries were then
prepared using the High-Sensitivity Open Chromatin Profile Kit for
Illumina (N248, Novoprotein) and sequenced on the Illumina NextSeq
platform.
Hi-C library preparation
The experiment was performed according to the in situ Hi-C
protocol^[288]95. Briefly, A7r5 rat VSMCs transfected with the
scrambled siRNA (20 nM) or siRNA targeting Lemd3 (20 nM) were washed
with PBS and crosslinked with 1% formaldehyde, and the reaction was
quenched with 200 mM glycine. The cells were lysed, the DNA was then
cut with DpnII, and the overhangs were filled with a biotinylated base.
The free ends were then ligated together in situ. The crosslinks were
reversed, the DNA was sheared to 300–500 bp, and then the biotinylated
ligation junctions were recovered with streptavidin beads. The
sequencing libraries were generated using a standard Illumina library
construction protocol. Briefly, ends of sheared DNA were repaired, and
the blunt ends were added to an “A” base to ligate with Illumina’s
adapters that have a single “T” base overhang. The DNA was subsequently
amplified via PCR for 8‒12 cycles. Products were purified using the
AMPure XP system and sequenced on the Illumina HiSeq X Ten platform.
Bulk RNA-seq analysis
The raw RNA-seq reads were mapped to the rat genome version rn7 using
TopHat (v2.1.1)^[289]96, and the expression value (number of raw reads)
for each gene was determined via the software HTSeq2 (v2.2.1).
Normalized (using the upper quartile method) expression values and
differentially expressed genes were determined using edgeR. Genes with
|logFC | >1 and adjusted P value < 0.05 were considered differentially
expressed genes (DEGs). We used the function bamCoverage in deepTools
(3.5.1)^[290]97 and normalized the bam files by a scaling factor. We
used Reads Per Kilobase per Million mapped reads (RPKM) to generate a
BigWig file that contained the RNA-seq signal (read density) at each
base pair across the genome.
ChIP-seq analysis
Paired ChIP-seq raw reads were trimmed with TrimGalore (v0.32), then
aligned to the rn7 genome using bwa-mem2 (v2.0pre2), and converted to
SAM files. The SAM files were then sorted and converted to bam format
using SAMtools (v1.15.1). PCR duplicates were removed using gatk
(v4.2.6.1) MarkDuplicates. Finally, bam files were filtered using
samtools view with “-f 2 -q 24” parameters. We used danpos
(v2.4)^[291]98 to call peaks and generate normalized wig files with the
parameters “danpos.py dregion --smooth_width 0 -c 25000000 -u 1 --frsz
200 --extend 200 --region_dis 3000 --pheight 1e-10” for the ChIP-seq
data. We used the wigToBigWig (v4) tool to convert wig files to BigWig
files. We used the top 10,000 widest H3K9me3 peaks in the
siRNA[scramble] group to plot a heatmap of the average signal in both
groups using deepTools (3.5.1)^[292]97 with the functions
“computeMatrix reference-point” and “plotHeatmap”.
ATAC-seq analysis
Paired raw ATAC-seq reads were trimmed with TrimGalore (v0.32), aligned
to the rn7 genome using bwa-mem2 (v2.0pre2), and then converted to SAM
files. The SAM files were then sorted and converted to bam format using
SAMtools (v1.15.1). PCR duplicates were removed using gatk (v4.2.6.1)
MarkDuplicates. Finally, bam files were filtered using samtools view
with “-f 2 -q 24” parameters. We merged three biological replicates
using “samtools merge”. We used danpos (v2.4)^[293]98 to call peaks and
generate normalized wig files with the parameters “danpos.py dpeak -c
25000000 -u 1 --frsz 200 -m 1 --smooth_width 0 --pheight 1e-20” for
ATAC-seq. We used the wigToBigWig (v4) tool to convert wig files to
BigWig files. We used the csaw R package^[294]99 to obtain different
ATAC-seq peaks using sliding windows at 300 bp with a cutoff (adjusted
P value < 0.05). Enrichment for known TF-binding sites of ATAC peaks
was performed using hypergeometric optimization of motif enrichment
(HOMER)^[295]100. We first lifted the ATAC peak position from the rn7
version to the rn6 version of the genome. Furthermore,
findMotifsGenome.pl was run on peak regions with the parameters “rn6
-size 200 -mask -p 24 -S 100 -len 6,8”.
Hi-C data analysis
For all Hi-C datasets of rat VSMCs, we used the HiCExplorer
(v3.7.2)^[296]101 analysis pipeline. The read pairs were aligned to the
rn7 genomes using the bwa-mem2 algorithm. The aligned read pairs were
assigned to DpnII restriction fragments. Invalid pairs with dangling
ends were excluded. The genome was divided into bins of specific sizes,
and valid pairs were counted per bin. We used 100, 40, 20, and 10 kb
bin sizes to generate raw and iterative correction and eigenvector
decomposition (ICE) normalized matrices. For the Hi-C analysis of human
samples, we downloaded the data in hic format from the GEO database,
which contains Hi-C contact maps with resolutions of 10, 25, 100 kb,
and converted them into 40 kb-resolution contact maps and saved them in
h5 format by using the “hicConvertFormat” and”hicMergeMatrixBins”. A
100 kb matrix was used to identify compartments via the juicer_tools
(v1.6)^[297]102 eigenvector, and a 40 kb matrix was used to define TAD
boundaries by calculating the insulation score using “hicexplorer
hicFindTADs”.
For TAD-TAD contacts, we classified genome-wide interactions into
intra-TAD interactions, interactions between neighboring TADs and
interactions between non-neighboring TADs (long-range interactions),
and counted the contact values within each region. Differences in
TAD-TAD contacts between the WT group and the Lemd3 KD group were
compared with edgeR. We utilized “bedtools intersect” from bedtools
(v2.18) with the parameters “F = 1e-9, f = 0.5” to retrieve the genes
located within each of these three types of TAD pairs based on the
chromatin locus. The surrounding TADs were divided into three types,
A-A, B-B, and A-B (also the boundaries of A/B compartments), based on
the eigenvector of the TAD position. The ATAC signal peaks in the TADs
at the boundaries of A and B compartments were sorted according to the
fold change in Hi-C contacts of their TADs after knockdown of Lemd3. We
calculated the averages of the fold change of ATAC signal and the fold
change of Hi-C contacts corresponding to every 200 ATAC peaks, and then
calculated the Pearson’s correlation coefficients between the two
factors.
We pooled two Hi-C samples from the same group. The log2Ratio matrix
was obtained with “hicexplorer hicCompareMatrices”. Visualization of
the pooled sample was implemented using “hicexplorer hicPlotMatrix” and
“hicexplorer hicPlotTADs”.
DNA polymer modeling
We applied a computational approach to derive polymer models and 3D
structures for pooled samples^[298]51. We studied the genome
organization of 100-kb genomic segments that represent each bead. This
model provides a more detailed representation of the genome and enables
us to characterize the spatial localization of the A and B
compartments. In our simulations, we modeled a total of 42 chromosomes,
representing the full complement of chromosomes in a diploid mammalian
nucleus. These include 21 homologous pairs, with each pair consisting
of two identical chromosomes to represent sister chromatids. Due to the
limitations of Hi-C data in distinguishing between sister chromatids,
we assumed that both copies within each pair share the same A/B
compartment profile and spatial organization. Each chromosome is
represented as a coarse-grained polymer chain, with beads corresponding
to 100 kb genomic segments, and interactions are derived from
experimentally inferred contact probabilities. This polymer model uses
an ensemble of structures, instead of a single, unique conformation, to
reproduce Hi-C data. The ensemble of structures is assumed to follow a
Boltzmann distribution with a potential energy function
[MATH:
UME(
r) :MATH]
, the expression of which can be derived according to the maximum
entropy principle and is provided below. The parameters of the polymer
model are solely encoded in the energy function. Molecular dynamics
simulations were conducted to collect structures consistent with the
energy function and the Boltzmann distribution.
The energy function of the chromosome model adopts the following form:
[MATH:
UME(
r)=Ur+Uidealr+Ucompt(r)
:MATH]
where
[MATH: r :MATH]
represents the 3D conformation of the entire genome. By definition,
[MATH: Ur :MATH]
and
[MATH:
Uidea<
/mi>lr :MATH]
are generic potentials shared by all chromosomes, and
[MATH:
Ucomp<
/mi>t(r
) :MATH]
describes compartment-type-specific interactions within the same
chromosome and between different chromosomes.
Specifically,
[MATH: Ur :MATH]
is the energy function for a confined homopolymer and consists of four
terms:
[MATH:
Ubond<
/mi> :MATH]
,
[MATH:
Uangl<
/mi>e :MATH]
,
[MATH:
Usc
:MATH]
and
[MATH: Uc
:MATH]
.
[MATH:
Ubond<
/mi> :MATH]
is the bonding potential between neighboring beads.
[MATH:
Uangl<
/mi>e :MATH]
is the angular potential applied among every three neighboring beads to
define the persistence length of the polymer.
[MATH:
Usc
:MATH]
is a soft-core potential applied to all the nonbonded pairs to enforce
the excluded volume effect among genomic loci.
[MATH: Uc
:MATH]
models a spherical boundary and is introduced to mimic the confinement
effect applied by the nuclear envelop onto chromosomes.
[MATH: I :MATH]
is an index over different chromosomes, and
[MATH: i :MATH]
and
[MATH: j :MATH]
are indices over all pairs of nonbonded chromatin beads from chromosome
[MATH: I :MATH]
.
[MATH: Ur=∑i∈I
[Ubondri,<
mi>i+1+<
msub>Uangleri,<
mi>i+1,r<
/mi>i+1,r
+2+Ucri]+∑i,j∈I,j>iUsc
(rij
) :MATH]
[MATH:
Uidea<
/mi>lrI :MATH]
measures the average contact probability for genomic loci separated at
a given sequence length of
[MATH: ∣j−i∣ :MATH]
. It is independent of the compartment type.
[MATH:
Uidea<
/mi>lrI=∑i,j∈Iαideal∣j<
/mi>−i∣f(rij
) :MATH]
[MATH:
Ucomp<
/mi>t(r
) :MATH]
describes compartment-type-specific interactions within the same
chromosome
[MATH:
Uintr<
/mi>a(r
) :MATH]
and between different chromosomes
[MATH:
Uinte<
/mi>r(r
) :MATH]
as follows:
[MATH:
Ucomp<
/mi>tr=Uintrar+Uinter(r)
:MATH]
[MATH:
Uintr<
/mi>ar=∑i,j∈Iaintra(
Ti
I,T
jI)f(r<
mi>ij) :MATH]
[MATH:
Uinte<
/mi>rr=∑i,j∈Iainter(
Ti
I,T
jI)f(r<
mi>ij) :MATH]
where
[MATH:
f(r
ij) :MATH]
determines the contact probability of a genomic pair with a spatial
distance of
[MATH:
rij
:MATH]
.
[MATH: fr :MATH]
was defined as follows:
[MATH: fr=121+tanhσrc−r,ifr≤r<
/mi>c12(<
mi>rcr)4,ifr>r<
/mi>c :MATH]
where
[MATH: rc
:MATH]
= 2 and
[MATH: σ :MATH]
= 2. The parameters
[MATH:
αidea<
/mi>l :MATH]
,
[MATH:
aintr<
/mi>a :MATH]
and
[MATH:
ainte<
/mi>r :MATH]
are entirely encoded in the energy function, their values were
determined iteratively so that the simulated structures reproduce Hi-C
contact maps to match the experimentally determined contact
frequencies.
[MATH: UME(r)
:MATH]
can be shown as the least biased potential to reproduce these
experimental constraints following the maximum entropy principle. These
ensure that the following ensemble averages determined with the
simulated genome conformation match the corresponding experimental
constraints calculated using Hi-C data; we used the parameters from Qi
and Zhang^[299]103.
Details of the molecular dynamic simulation and analysis
Molecular dynamics simulations were performed using the software
package LAMMPS (v2018). The simulations were maintained at a constant
temperature of
[MATH: T :MATH]
= 1 via Langevin dynamics with a damping coefficient of
[MATH: γ :MATH]
= 10.
We first placed all the chromosomes consecutively on a cubic lattice
with an edge length of
[MATH: 0.9R/3 :MATH]
, where
[MATH: R :MATH]
is the radius of the spherical confinement introduced to ensure a
volume fraction of 0.1, to generate an initial configuration for these
simulations. This configuration was subsequently equilibrated during a
100,000-step-long simulation under a time step of
[MATH: dt :MATH]
= 0.0001 and a 10,000-step-long simulation under a time step of
[MATH: dt :MATH]
= 0.001 to relax both the topology and energy of the polymer
structures. A 4,000,000-step-long simulation with a time step of
[MATH: dt :MATH]
= 0.01 was subsequently performed. The genome conformations were saved
every 2000 timesteps to calculate the ensemble averages. The last state
from this equilibration trajectory was then used to visualize the
genome structure with VMD (194a57).
We divided the radius of the simulated nucleoplasmic space from the
core to the perinuclear space into 50 equal parts of equal length, and
the whole space was divided into 50 parts of unequal volume from inside
to outside to compare the ratio of the A/B compartments from the core
to the perinuclear region. We calculated the ensemble averages from the
last saved 500 genome conformations, defined the number of compartments
per unit volume as the density of the compartments, and counted the A/B
compartments in each layer of the nucleoplasmic space in the ensemble
averages.
Functional enrichment analysis
For the pathway enrichment analysis, P values were obtained using an
empirical phenotype-based permutation test, and pathways with P
value < 0.05 were selected. The P values were corrected for multiple
hypotheses using the Benjamini–Hochberg method. Pathways associated
with the Gene Ontology biological process (GOBP) with adjusted P
value < 0.05 were included. The GO gene sets were obtained using the
msigdbr R package. We analyzed the genes around different peak signals
identified via ATAC-seq and the genes on the boundary of A/B
compartments through pathway enrichment analysis on GOBP.
We performed gene set enrichment analysis (GSEA) on GOBP gene sets
using the GSEA function with the parameters “minGSSize = 5, maxGSSize =
1200, p value Cutoff = 0.05, pAdjustMethod = “BH”, and by = “fgsea””.
GSEA of the DEGs identified via RNA-seq was performed with the
clusterProfiler R package (v4.4.4)^[300]104.
Nascent RNA capture
A STAR-Click Nascent RNA Capture Prep kit (Foreverstar Biotech,
Beijing, China) was used to detect RNA synthesis in accordance with the
manufacturer’s instructions. The rat VSMCs were transfected with the
scrambled siRNA (20 nM) or siRNA targeting Lemd3 (20 nM). 24 hours
after transfection, 1 mΜ of 5-ethynyl uridine (EU, an alkyne-modified
nucleoside) was added, and the cells were subsequently incubated for
another 2 h. After the incubation, total RNA or mRNA labeled with EU
were isolated and used in a copper catalyzed click reaction with an
azide-modified biotin, which created a biotin-based handle for
capturing nascent RNA transcripts on streptavidin magnetic beads. The
captured transcripts were used as a template for reverse
transcriptase-mediated cDNA synthesis for subsequent analysis using
qPCR.
Analysis of homologous genes in the rat and human genome
We found homologous genes in the human genome corresponding to rat
genes from the Ensembl database and homologene R package. The R package
is a wrapper for the homologene database by the National Center for
Biotechnology Information (NCBI)^[301]105. It allows searching for gene
homologs across species. We use the python module biopython^[302]106 to
obtain the homologous gene sequences in the NCBI Entrez database, store
them in fasta format respectively, and, via a step forward, use the
blastn command line tool^[303]107 to perform pairwise sequence
alignment of the homologous sequences, with the parameter set as
follows “-max_target_seqs 100 -word_size 11 -gapopen 5 -gapextend 2”.
For each pair of homologous genes, we weighted the gene sequence
identity of the sequence fragments aligned according to the sequence
length.
Identity of single nucleotide polymorphisms (SNPs) associated with disease
We first downloaded the statistics of reported SNPs associated with
coronary artery disease from the GWAS catalog^[304]108. We mapped these
SNPs to the human genome to find their homologs of LEMD3-modulated rat
genes. In addition, we downloaded the computational results of the
corresponding GWAS studies from the IEU OpenGWAS project^[305]109 based
on the PMID (PubMed ID: 29212778)^[306]67 of the papers reporting these
SNPs. These results were stored in vcf format, and we read the vcf
format files by vcfR R package^[307]110 and used the CMplots R
package^[308]111 to visualize the SNPs near the COL4A1 and MYH11 gene
loci. The SNPs are presented in the Supplementary Data [309]10.
Statistical analysis
The data are presented as the means ± standard errors of the means
(s.e.m.) and were analyzed using GraphPad Prism 8.0 software (GraphPad
Software, San Diego, CA, USA). The Shapiro‒Wilk normality test was used
to determine whether the data were normally distributed. For two-group
comparisons of normally distributed data, we applied two-sided unpaired
Student’s t test for data with similar variances or with Welch’s
correction if equal standard deviations were not assumed through an F
test. In addition, for comparisons of more than two groups of normally
distributed data, we applied ordinary one-way ANOVA with Tukey’s
multiple comparisons test for data with similar variances or
Brown–Forsythe and Welch’s ANOVA with Games–Howell’s multiple
comparisons test if equal standard deviations were not assumed. The
Mann‒Whitney U test was used for two groups or the Kruskal‒Wallis test
with Dunn’s multiple comparisons test were applied for more than two
groups of data that were not normally distributed. The detailed
statistical analysis applied to each experiment is presented in the
corresponding figure legends.
Reporting summary
Further information on research design is available in the [310]Nature
Portfolio Reporting Summary linked to this article.
Supplementary information
[311]Supplementary Information^ (8MB, pdf)
[312]41467_2025_63876_MOESM2_ESM.pdf^ (96.5KB, pdf)
Description of Additional Supplementary Information
[313]Supplementary Data 1^ (1.1MB, xlsx)
[314]Supplementary Data 2^ (13.7KB, xlsx)
[315]Supplementary Data 3^ (1.7MB, xlsx)
[316]Supplementary Data 4^ (74.5KB, xlsx)
[317]Supplementary Data 5^ (50.8KB, xlsx)
[318]Supplementary Data 6^ (702.4KB, xlsx)
[319]Supplementary Data 7^ (12.4KB, xlsx)
[320]Supplementary Data 8^ (47.2KB, xlsx)
[321]Supplementary Data 9^ (23.5KB, xlsx)
[322]Supplementary Data 10^ (11KB, xlsx)
[323]Reporting Summary^ (128.9KB, pdf)
[324]Transparent Peer Review file^ (9MB, pdf)
Source data
[325]Source Data^ (29.3MB, xlsx)
Acknowledgements