LAP2βtransmits force to upregulate genes via chromatin domain stretching but not compression

Abstract

There is increasing evidence that force impacts almost every aspect of cells and tissues in physiology and disease including gene regulation. However, the molecular pathway of force transmission from the nuclear lamina to the chromatin remain largely elusive. Here we employ two different approaches of a local stress on cell apical surface via an RGD (Arg-Gly-Asp)-coated magnetic bead and whole cell deformation at cell basal surface via uniaxial or biaxial deformation of a fibronectin-coated flexible polydimethylsiloxane substrate. We find that nuclear protein LAP2β mediates force transmission from the nuclear lamina to the chromatin. Knocking down LAP2β increases spontaneous movements of the chromatin by reducing tethering of the chromatin and substantially inhibits the magnetic bead-stress or the substrate-deformation induced chromatin domain stretching and the ensuing dihydrofolate reductase (DHFR) gene upregulation. Analysis of DHFR gene-containing chromatin domain alignments along or perpendicular to the direction of the stretching/compressing reveals that the chromatin domain must be stretched and not compressed in order for the gene to be rapidly upregulated. Together these results suggest that external-load induced rapid transcription upregulation originates from chromatin domain stretching but not compressing and depends on the molecular force transmission pathway of LAP2β.

Graphical Abstract

1. Introduction

Increasing evidence shows that mechanical forces, together with soluble molecules, impact every aspect of living cells and tissues in physiology and disease. Since the effect of the forces at the cellular and subcellular levels is dominated by the contact forces, the structural linkages between cellular components and between structural molecules are crucial in transmitting and propagating forces from one site to another [1]. Of particular interest is the largest organelle in the cell-the nucleus, an important compartment for gene regulation. Over the last two decades, structural proteins of the LINC (linker of nucleoskeleton and cytoskeleton) complex that link the cytoskeleton with nuclear lamins have been identified [1, 2], providing a possible physical pathway for nuclear mechanotransduction when the load is applied at the cell surface. Intra-nuclear lamin A/C networks that link the nuclear envelope with the chromatin are shown to scale with tissue mechanics, and regulate differentiation [3], and control translocation of transcription factor MKL1 [4]. Patterning cell geometry that alters nuclear shape and mechanics modulates gene expression patterns [5] and facilitates transport of transcriptional factors via nuclear pores [6]. Lamin A mutations lead to myofiber dysfunction and death [7]. All these findings suggest that nuclear structural proteins and mechanics play important roles in nuclear functions such as gene expression. However, how forces and mechanics impact nuclear structure and function has largely been elusive. There is evidence that compressive forces induce YAP (Yes-associated protein) translocation via opening nuclear pores [8]. Recently it is shown that the nucleus of a progenitor stem cell acts as a spatial ruler that gauges compression-induced constraints to trigger biochemical signaling molecules when the cell migrates through 3D dense and stiff matrices [9]. This cell surface force-triggered biochemical signaling cascade is an important pathway to the nucleus to regulate its functions. However, the force also likely impacts the nucleus in a multifaceted manner. The experimental evidence comes when a local stress is applied via integrins, leading to direct chromatin stretching (without cytoplasmic biochemical cascades) that is associated with gene upregulation [10]. F-actin intactness and cytoskeletal prestress are necessary in force-induced direct chromatin stretching and anisotropic (i.e., force-direction dependent) upregulation of a transgene and endogenous genes [10–12]. Furthermore, besides the known important force-transmitting roles of nuclear Lamin A/C and Lamin B from the nuclear envelope to the nuclear interior [1–4, 10], BAF (barrier-to-autoregulation factor) and HP1 (heterochromatin protein 1) are shown to transmit force from the nuclear lamina to the chromatin [10]. However, it remains unclear if other nuclear structural proteins can transmit forces from the nuclear lamina to the chromatin. In addition, while it has been shown that compressive forces can trigger chondrocytes to respond [13], stimulate nuclear translocation of beta-catenin in Drosophila embryonic tissues [14], and induce convergent extension in Xenopus embryos [15], it is not clear at all if the compressive forces have direct deformation effects on chromatin structure and the ensuing gene regulation and/or they generate indirect effects on transcription via translocation of cytoplasmic molecules. The recent published findings using a local cyclic sinusoidal stress via the magnetic bead to directly deform the chromatin reveal gene upregulation [10–12] but the sinusoidal stress induces both stretching and compression of the same chromatin domain during one cycle of stress. Thus, the important question remains open if it is the stretching or compressing of the chromatin domain or both that induce gene upregulation.

While it is well known how forces are transmitted from the F-actin to the nuclear lamina via the LINC complex [10] that consists of the KASH-domain proteins (Nesprins) and SUNs (Sun1/2), it is less clear how forces are transmitted from the lamina to the chromatin. In a published report we have shown that BAF (barrier-to-autoregulation factor) is a structural protein that transmits stress from the nuclear lamina to the chromatin [10]. However, whether other protein(s) can also mediate force transmission inside the nucleus has been unclear. LAP2β is an intra-nuclear molecule that binds to Lamin B1 [16] whereas LAP2β’s LEM-domain at the C-terminal binds to BAF [17] and its ‘LEM-like’ domain in the N-terminus directly interacts with DNA [18]. Separately, it is reported that LAP2β interacts with HDAC3 [19]. LAP2β/HDAC3/cKrox complex tethers LADs (lamina-associated domains) to the nuclear periphery [19–21]. These structural interactions suggest that LAP2β is a putative molecule to physically link the nuclear lamina and the LADs to transmit force but the experimental evidence is lacking.

In this study we report that LAP2β transmits force from the nuclear lamina to the chromatin. Using two different strategies of loading: a local magnetic bead stress on the cell apical surface via integrins and a whole cell strain at the cell basal surface via substrate deformation through integrins, we show that it is chromatin domain stretching and not compression that induces rapid gene upregulation.

2. Materials and Methods

2.1. Cell culture and reagents

Ten copies of bacterial artificial chromosome (BAC), mouse DHFR transgenes and LacO repeats were inserted into the genome of CHO DG44 subclone (endogenous DHFR knockout) [22]. After stable expression of EGFP-LacI, cells were cultured in Ham’s F12 medium with 10% dialyzed fetal bovine serum (without thymidine and hypoxanthine, Cell Media Facility, The School of Chemical Sciences, University of Illinois at Urbana-Champaign). Cells were sub-cultured every 3 or 4 days by TrypLE (Thermo Fisher Scientific) treatment and were assigned to each experimental group randomly. Potential mycoplasma contamination was constantly monitored using DAPI-staining and visual inspection by imaging during the experiments and no mycoplasma contamination was observed.

The permeabilization buffer contains 20 mM HEPES, 110 mM potassium acetate, 5 mM sodium acetate, 1 mM ethylene glycol-bis(2-aminoethylether)-N,N,N’,N’-tetraacetic acid (EGTA), 1 μg/ml aprotinin, 1 μg/ml leupeptin, 1 μg/ml pepstatin, 2 mM DL-Dithiothreitol (DTT) and 250 μM adenosine triphosphate (ATP), pH 7.0. All the reagents mentioned above were purchased from Sigma, as well as digitonin, wheat germ agglutinin (WGA), 2-Aminoethoxydiphenyl borate (2-APB), Nifedipine, gadolinium (III) chloride (GdCl₃), Triptolide, Flavopiridol and Pitstop 2. ML-7 was purchased from Enzo Life Sciences, Inc.

2.2. Stretching and compressing whole cells on a stretchable substrate

PDMS (polydimethylsiloxane) chambers (STB-CH-04-XY Biaxial Chamber, Strex, San Diego, CA, 2.0 × 2.0 × 1.0 cm³) were sterilized by autoclave, coated with 30 μg/mL fibronectin (F4759, Sigma; fibronectin contained RGD (Arg-Gly-Asp) for binding to the integrins on the cell basal surface) in PBS at 37 °C for 12 hours, and seeded with cells at the density of 3 × 10⁵ cells/chamber. After cell culture for 36 h, the PDMS chamber was mounted onto the stretching system (STB-190-XY Microscope Mountable Biaxial Stretching System, Strex). The uniaxial stretching in a step-function (square-waves) and the stretching extents of 12% or 20% were selected. It took 0.5 second for the system to stretch the PDMS chamber and the stretch was held statically until the “stop” button was pressed that relaxed the PDMS chamber. For compressing the cells uniaxially or biaxially, a PDMS chamber (STB-CH-04-XY Biaxial Chamber) was mounted onto a metal stand, custom designed and manufactured by the machine shop at the Department of Mechanical Science and Engineering, UIUC, which held the chamber in fixed position with uniaxial (12% or 20%) or biaxial (12%) stretch. The pre-stretched PDMS chamber was then coated with fibronectin and cells were seeded and cultured inside the chamber. To initiate static compression of cells, the PDMS chamber was removed from the metal stand that released the pre-stretch. After cell compression for 2 minutes, the PDMS chamber was mounted back onto the metal stand manually to terminate cell compression. In most cases, the GFP spots within most cells aligned almost as a straight line [22]; in a few cases when they were not, they were excluded from data analysis because the induced strains were too complex to be analyzed for only chromatin domain stretching (i.e., tensile strain) or only compression.

An inverted Leica microscope (DMIRE2), equipped with a 63× oil objective (NA 1.32) and filter sets for DAPI, GFP and Quasar 570 (Chroma Technology Corp., Vermont), was used for monitoring GFP spots inside the cell nucleus and RNA FISH imaging. For imaging GFP spots before and after stretch or compression of PDMS substrate, the stretching unit in the system could also be used manually to deform the PDMS substrate for tracing the cell of interest with the microscope.

2.3. Magnetic twisting cytometry

Cells were plated onto the glass-bottomed petri dish (GBD00001-200, Cell E&G, San Diego, CA) coated with collagen I (16 μg ml⁻¹, 37°C for 1 h, Thermo Fisher Scientific) and cultured for 24 hours followed by incubation with RGD-coated magnetic beads for 15 min at 37°C. RGD-magnetic beads were added at 1.8×10⁵ beads per dish (30 μl of 1 mg bead per ml medium; ~6×10⁶ ferromagnetic beads (Fe₃O₄) per mg beads) so that there was approximately one bound bead per cell after unbound beads were washed away. Since these beads were monodisperse and had almost identical diameter (~4 μm in diameter) and the same magnetic moment, each bead under the same magnetic twisting field generated the same stress. The variabilities of chromatin deformation among different beads are largely due to the different positions and distances of various beads relative to the chromatin domains and the rolling direction of the bead relative to the long axis of the cell (i.e., the direction of the majority of the stress fibers of the cell). In a few cases there were two beads bound to a single cell. Because these beads generated complex stress fields and strain fields in the cell, making it difficult to analyze each bead’s contribution to the chromatin strains, these beads were excluded from analysis. About 50% of the cells in a dish were bound by a bead. Those cells without a bound bead in the same dish were used as stress-free controls. The petri dish with cells was placed into a chamber where surrounding coils were employed for application of magnetic fields. A strong magnetizing pulse (~1000 G, <1 ms) was used to magnetize the beads along the y direction. A twisting field in a sinusoidal waveform along the z direction was then applied. The magnitude and frequency of the magnetic field were controlled by a custom-made program [10].

Chromatin spontaneous movements were measured only in the lateral direction (i.e., the x-y plane), which would underestimate the total movement of the chromatin. However, since these cells were very spread, the z-height of the nucleus was small relative to the x-y dimensions. The chromatin domains (i.e., the chromatin GFP spots) rarely moved out of the focal plane during the experiments, suggesting that the impact might be small.

2.4. RNA FISH and data analysis

Custom RNA oligoes conjugated with fluorescent dyes (Quasar 570) for 5’-probe were designed by Stellaris® Probe Designer and produced by Biosearch Technologies Inc. For detection of the full length of mouse DHFR mRNA (https://www.ncbi.nlm.nih.gov/nuccore/NM_010049.3). Full mouse DHFR RNA FISH probe set was labeled with Quasar 670 Dye in some experiments for getting maximum difference of DHFR upregulation between the negative control group and other siRNA treatments in bead twisting experiments):

• gggaggcaaacggttctaag, cgcgcattctatttgtgtag, gtcaagtttggcgcgaaatc, gataaaatcctaccagcctt, gacgatgcagttcaatggtc, tccggttgttcaataagtct, accatgtctactttacttgc, taaacagaactgcctccgac, ctttcttctcgtagacttca, ttaggaggggagcagagaac, attatataggggctagggtt, accttgttagattactggga, tagctcctacttttatgagc, gctctgctgtttaaaacctg, gcagtatccattctcaattc, ttctctgacctgatgatctg, actaatttacaaggccaggc, atggacaccacactcacaag, agctctgaacagaccatttc, agcttacagacacaaggctg, ctgtgtttataccctgtatt, ctaggttttcagagtgcttt, ctactactgctgctttgtta, agtcatgggttctaaggaca, caatgcttcctagttggatt, catcaccagggagaaaagct, tgtcgtagccagatgacaag, gctcccaaaacaacacatct, gggcacccttaaaagtaact, acactggtggacaatgctta, ttcatctctttgttcatggt, tctagtgtacctggtcaatt, caaatgtctgttaaccccag, cattgtataacacacctcct, ctgccagggttacaaacata, ttgcaaatacactgccagtc, aatgctctataccctcattt, gcaagaaagcgctattgctt, tagctctgggagatgtcaac, ttttggcatgtatgaggtgt, acgacaatttccttgtgtct, ccttgaactaggtttcttgt, gggttgtcaatgggaatctg, cagtgtggaacatcgtgcaa, tttatgttagcagcttggga, cacaatgctttggtggaagc, atcatgtctccttttcagta, cctcgccttctaacacaaaa.

5’-end mouse DHFR RNA FISH probe set was labeled with Quasar 570 Dye:

• tctgcgggaagcctaagatc, gggaggcaaacggttctaag, ctcttctgcacactgcaatg, cgcgcattctatttgtgtag, catcctatttgtgcagctaa, catcctatttgtgcagctaa, gtcaagtttggcgcgaaatc, cagccttcacgctaggattg, gatggcagcggggataaaat, tgcagttcaatggtcgaacc, atattttgggacacggcgac, agtacttgaactcgttcctg, agaggttgtggtcattcttt, agattctgtttaccttccac, aggttttcctacccataatc, ggtcgattcttctcaggaat, agtcttaaggcatcatccaa, tgccaattccggttgttcaa, tccaaaccatgtctacttta, taaacagaactgcctccgac, ctggttgattcatggcttcc, ccttgtcacaaagagtctga, gtcactttcaaattcctgca, ttccccaaatcaatttctgg, gggtattctgggagaagttt, ctttcttctcgtagacttca, ttaggaggggagcagagaac, agtcccatggtcttataaaa, ctcatagatctaaagccagc, accttgttagattactggga, gacatttcttaaggcacttt, gcactgagacctttatagca, cacttgaggtctcatgggag, cacagtaccctgtgcatatg, tacatcactggggtctcttg, tacttttatgagcccacaca, ctttggacttacctgcctag, ctctgctgtttaaaacctgt, ttctttatagtctgagttcc, agctttgctgcaagtgtgat, gattttctgtctgagtgagc, ttacataatcttccacctgc, gcagacaatttcagtgtttc.

Cells were fixed by incubation with 3.7% paraformaldehyde (Electron Microscopy Sciences, Inc.) in PBS at room temperature for 12 minutes and washed twice with PBS followed by permeabilization with 70% ethanol at 4 °C for 8 hours. After cells were washed with wash buffer (2× SSC, 10% deionized formamide, Thermo Fisher Scientific) at room temperature for 5 minutes, RNA FISH probes in the hybridization buffer (150 μL, 250 nM) were added to each petri dish for hybridization at 37 °C for 12 h. Cells were then incubated with wash buffer at 37 °C for 30 minutes after removal of RNA FISH probes. For nucleus imaging, cells were incubated with wash buffer containing DAPI (Thermo Fisher Scientific) at 37 °C for another 30 minutes, washed twice with 2× SSC buffer at room temperature for 5 minutes, and imaged immediately. Data analysis was the same as previously reported [11]. In short, the fluorescence area was determined by tool of “freehand sections”, and its integrated density was measured as the level of gene transcription after the mean background was subtracted.

2.5. siRNA transfection

Cells were transfected with siRNAs for 48 hours to knockdown specific genes by using Lipofectamine 3000 (Thermo Fisher Scientific) according to the manufacturer’s protocol. All custom siRNAs were designed by siDirect Version 2.0 and were produced by Sigma. The siRNA sense sequences are:

• CCACUAAAGGGCAGAGCAA for LAP2β siRNA #1;

• GCUCUCUAAUGAAGACCUU for LAP2β siRNA #2;

• UUGAAGAAAAGCUUUUCGGGG for ZBTB7B siRNA #1;

• CCCAGUCUGUCACAAGAUAAU for ZBTB7B siRNA #2;

• AGUAAUUUCCAUAUUUGUGGA for HDAC3 siRNA #1;

• AGAAGAUGAUCGUCUUCAAGC for HDAC3 siRNA #2;

• GCUAAAAAGCAAAUGCAUUCC for LEMD2 siRNA #1;

• GAGUACAUCGCUAACGUGACC for LEMD2 siRNA #2;

• CCAAUAGUUCAGAUCCCUU for YAP siRNA #1;

• GCCACUCUCUGAGUCUGAA for YAP siRNA #2.

• GCTATGAATCCAGAGCTTA for BANf1 siRNA

2.6. Immunofluorescence and western blotting

CHO cells were fixed by 3.7% paraformaldehyde in PBS at room temperature for 15 minutes. After washed twice with PBS, cells were permeabilized with 0.3% Triton X-100 (Sigma) at room temperature for 15 minutes. Cells were then treated with 5% normal donkey serum (Jackson ImmunoResearch Laboratories, Inc.) in PBS at room temperature for 4 h for blocking and incubated with primary antibodies (1:100 dilution, v/v) in PBS (with 1% BSA) at 4 °C overnight. Primary antibodies include rabbit monoclonal anti-BAF antibodies (Abcam, ab129184) and rabbit monoclonal anti-YAP/TAZ antibodies (Cell Signaling Technology, #8418). Cells were washed with PBS three times for 5 minutes each and incubated with secondary antibodies (Donkey Anti-Rabbit IgG H&L, Alexa Fluor® 555, preadsorbed, 1:200 dilution, v/v, Abcam, ab150062) in PBS at room temperature for 1 hour. For western blotting, we used YAP/TAZ (D24E4) Rabbit monoclonal antibody (1:1000 dilution, v/v, Cell Signaling, #8418), Lap2/TMPO Rabbit polyclonal antibody (1:5000 dilution, v/v, ABclonal, A2534), HDAC3 Rabbit monoclonal antibody (1: 500 dilution, v/v, ABclonal, A19537), LEMD2 Rabbit polyclonal antibody (1: 500 dilution, v/v, ABclonal, A18558), ZBTB7B polyclonal antibody (1:1000 dilution, v/v, Proteintech, 11341-1-AP), BANF1 polyclonal antibody (1:500 dilution, v/v, NOVUSBIO, #NBP1-76751), and GAPDH rabbit polyclonal antibody (1:1000 dilution, v/v, Sigma, G9545), and goat anti-rabbit IgG (HRP-linked, 1: 1000 dilution, v/v, Cell Signaling, #7074). The two-color infrared laser system (LI-COR) was used for imaging. For Poisson’s ratio determination, EBFP2-LaminB1-10 (addgene plasmid #55244) was transfected into the cells attached to the PDMS chamber. After one-day transfection, the fluorescent images of the same cell nucleus were captured before and after 12% uniaxial stretch of the PDMS substrate. The displacements of cell nucleus along (Δy) and vertical (Δx) to the substrate stretch direction were measured using ImageJ. After Δy scatter plot as a function of y (the length of the cell nucleus along the substrate stretch direction) and Δx scatter plot as a function of x (the length of the cell nucleus vertical to the substrate stretch direction), the Poisson ratio of cell nucleus is determined as the negative ratio of slopes of the two linear fits, and equals 0.37.

2.7. Chromatin immunoprecipitation

A commercial kit from Cell Signaling (#56383) was employed to conduct chromatin immunoprecipitation (ChIP). CHO cells were seeded onto PDMS chambers. After cells reached ~90% confluency (~1 × 10⁶ cells/chamber, whose size was 2.0 × 2.0 × 1.0 cm³), the PDMS substrate was compressed or stretched in a sinusoidal waveform at 0.5 Hz for 2 min. Cells were incubated with 1% formaldehyde for fixation at room temperature for 15 min, and further incubated with glycine for 5 min to quench the formaldehyde. Cells without any substrate compression or stretch were also fixed as the control. After washed by ice cold PBS, cells were collected into cold PBS containing Protease Inhibitor Cocktail (PIC) by cell scrapers (Fisher Scientific), pelleted by centrifuge and incubated twice (10 min each) with 1 ml ChIP Sonication Cell Lysis Buffer plus PIC on ice. After another centrifuge, cells were treated with 1 ml ChIP Sonication Nuclear Lysis Buffer on ice for 20 min and sonicated by using Covaris M220 (peak power 75, duty factor 5, cycles 200 for 40 min).

For each immunoprecipitation (IP), 100 μl chromatin was diluted by 400 μl 1× ChIP Buffer plus PIC. 10 μl of diluted sample was used as the 2% Input. Pol II S5p antibody (3 μg, Abcam, ab5131) or Pol II S2p antibody (3 μg, Abcam, ab5095) was mixed into the rest sample as one IP reaction followed by rotation at 4 °C overnight. Normal Rabbit IgG antibody (2 μg, Cell Signaling, #2729) was used as the negative control. Protein G magnetic beads (30 μL) were added to each IP reaction, which was rotated at 4 °C for another 2 h. Protein G magnetic beads in each IP were pelleted by a magnetic separation rack (Promega), washed by low salt solution three times with rotation at 4 °C (5 min each) and high salt solution once. Chromatin was eluted after gentle vortexing (1,200 rpm) at 65 °C for 30 min with Thermomixer^™ R (Eppendorf). NaCl and Proteinase K were added to the eluted chromatin and incubated at 65 °C for 2 hours for reverse cross-links. Spin columns were used for DNA purification.

Quantitative polymerase chain reaction (qPCR) was performed based on SsoAdvanced Universal SYBR Green Supermix (Bio-Rad, #1725271) by using CFX Connect Real-Time PCR Detection System (Bio-Rad). Primers were designed and produced by Sigma. The primer sequences are shown below:

Mouse DHFR promoter primer set
Forward primer		ACCTGTATCGGGAAGGTTGGA
Reverse primer		AGTCACCCCCAACAGCCTTT
CHO Ctgf primer set
Forward primer			TGTGTGATGAGCCCAAGGAT
Reverse primer			TGCTTCTCCAGTCTGCAGAA
CHO egr-1 promoter primer set
Forward primer		ACAGCTCCCGGGTCTTATGT
Reverse primer		CAAAACAAACGCTCTGCGCT
CHO FKBP5 promoter primer set
Forward primer		TCCAGTGCTTTGATCATCTGT
Reverse primer		GCTTGGGAACTTGTGTGAAGC
CHO DHFR non-promoter primer set
Forward primer	TTTTCCAGATACCCAGGCGT
Reverse primer	TAGGAGGGGAGCAGAGAACT

2.8. DAPI staining

CHO cells were added with 1 ml of 4% paraformaldehyde and fixed for 15 minutes. The cells were washed three times with 1 ml of washing solution (PBS containing 3% BSA Albumin Fraction V, Biofroxx, #9048-46-8) for 5 minutes. 1 ml of permeabilization solution (PBS containing 0.3% Triton X-100, sigma, #9002-93-1) was added and CHO cells were incubated at room temperature for 15 minutes. 1 ml of washing solution was used to wash 1–2 times, 5 minutes each time. After the washing solution was moved, DAPI (Biofroxx, #28718-90-3) was diluted with PBS (dilution ratio=1:1000) and added to cells at room temperature for 30 minutes in the dark. Cells were washed with washing solution for three times, five minutes each time, and imaged immediately.

2.9. Cell Counting Kit-8

CHO cells were cultured in a 96-well plate for pre-culture in the incubator (37°C, 5% CO₂). 10 microliters of CCK-8 solution was obliquely added into 100 microliters of culture medium each well. The culture plate was placed in the incubator for 2 hours, and the absorbance at 450 nm was measured with a microplate reader to calculate the cell viability.

2.10. Traction force microscopy

Cells were cultured on 8-kPa Polyacrylamide (PA) substrates coated with fibronectin at 30 μg per ml. PA gels were created on glass bottom dishes (Cell E&G, Catalog # GBD00001-200) by mixing 40% acrylamide (Biorad; Catalog #1610140) with 2% Bis-acrylamide (BioRad; Catalog #161-0142) using a previously established protocol [23]. PA gels were created with 0.2-μm fluorescent (FITC) nanoparticles near the cell basal surface to quantify displacement fields. Cells were cultured on PA gels for 24 hours for optimum spreading and then images were acquired before and after trypsinization with 0.25% Trypsin-EDTA (Gibco) to obtain with stress and without stress conditions. Results were analyzed using a custom-MATLAB code and an analysis that estimates cell tractions from displacement fields generated from relative movements of the fluorescent nanoparticles [24].

2.11. Focal adhesion quantification

CHO cells were cultured on glass-bottomed dishes coated with fibronectin at 30 μg per ml for 24 h. Cells were fixed using 4% Paraformaldehyde (PFA) for 10 min at room temperature. Following PFA fixing, cells were treated with 0.25% Triton-X solution in PBS (phosphate-buffered saline) for 10 min. For surface blocking, cells were treated with 3% BSA in PBS for 30 min. Rabbit anti-Paxillin antibody (abcam; Catalog # ab32084) was used to stain focal adhesions. Antibody solutions were prepared in 0.125% Triton-X and 1.5% BSA and incubated at room temperature for 1.5 h. Following primary antibody incubation, cells were treated with a secondary antibody (Donkey anti-rabbit, abcam; Catalog # ab150061) conjugated with GFP for 1 h. Cells were washed twice with PBS. After antibody conjugation, ProLong^™ diamond antifade mountant (Thermo-Fisher, Catalog # P36961) was added with a coverslip on top. Cells were imaged using a Leica DMIRE2 epifluorescence microscope.

Focal adhesions were quantified using ImageJ and associated plugins using a previously established technical protocol [25]. In short, images were subjected to Contrast Limited Adaptive Histogram Equalization (CLAHE) plugin to enhance contrast after subtracting background from the images. After brightness/contrast adjustment, Laplacian of Gaussian plugin (Log3D) was used to filter the image based on XY parameters. The area of focal adhesions was quantified using the “analyze particles” command in ImageJ.

2.12. Code availability

The code written in C language that controls the waveform of magnetic field, and the MATLAB code that obtains the position list of GFP spots at different time points, are fully available upon request.

2.13. Statistical analysis

Student’s t-test (two-tailed, unpaired) was conducted. One-way ANOVA with Bonferroni Correction was employed to compare multiple groups.

3. Results

3.1. LAP2β mediates force transmission to chromatin for gene upregulation

Although it is known that LAP2β is structurally connected with the nuclear lamina and the chromatin [16–19] (Fig. 1A), it remains elusive whether LAP2β and its associated proteins serve as signaling scaffolds, tether chromatin to the nuclear lamina, or effectively transmit force from the nuclear lamina to the chromatin. To examine what roles LAP2β plays in the nucleus, we depleted LAP2β and its associated molecules individually and measured spontaneous chromatin movements. Chromatin domain movements were measured by quantifying the GFP-chromatin spots that contain the multiple copies of mouse dihydrofolate reductase (DHFR) genes into the CHO (Chinese Hamster Ovary) cells using the BAC (bacterial artificial chromosome) strategy [22]. The CHO cell line, deficient in endogenous CHO DHFR genes, contains a multi-copy insertion of a BAC with a 170 kb mouse genomic insert containing the 34 kb DHFR gene. The cell clone DHFR D10 stably expresses EGFP-dimer lac repressor (GFP-LacI) and is tagged with a 256mer lac operator repeat (~10 kb) and thus enables visualizing and measuring movements of the BAC transgene [22]. Knocking down LAP2β, HDAC3, or ZBTB7B (cKrox) led to elevation of chromatin spontaneous movements while silencing LEMD2, a molecule that binds to the nuclear lamina but is not in the LAP2β/HDAC3/cKrox complex, had no effects (Fig. 1B, Fig. S1A and Fig. S2). These siRNA treatments had no effects on nuclear size (Fig. S3) or cell viability (Fig. S4). Compared with depletion of HDAC3 or ZBTB7B, depleting LAP2β resulted in the highest increase in spontaneous chromatin movements (Fig. 1B and Fig. S1A), suggesting that the LAP2β protein is a structural molecule in tethering the chromatin to the nuclear lamina. To apply a local stress to the apical cell surface, we attached an RGD (Arg-Gly-Asp)-containing peptide coated magnetic bead to the cell via integrins (Fig. 1C left). The magnetic bead was magnetized in the Y-direction and a homogenous weak magnetic field (that does not generate magnetic field gradients) was applied along the Z-direction to generate a local rotational complex stress (a sinusoidal waveform of 15 Pa stress at 0.3 Hz) to the cell surface [12] (Fig. 1C right). Remarkably, the applied local stress-induced chromatin stretching was decreased from ~10% (Negative Control) to 4% (a 60% reduction) when LAP2β was knocked down, more reduction in chromatin stretching than knocking down HDAC3 or ZBTB7B (gene for cKrox protein), while silencing LEMD2 had no effects [Fig. 1D and Fig. S1B). These data show that the LAP2β/HDAC3/cKrox tripartite complex mediate force transmission from the lamina to the chromatin. Depletion of LAP2β had little effects on individual focal adhesion area and focal adhesion number per cell (Fig. S5A–C) but decreased cell tractions and cell spreading area (Fig. S5D–F), consistent with the published reports that actomyosin-mediated contractile forces depend on force transmission to the nucleus [10, 12]. The effects on DHFR transcription upregulation follow the patterns of the chromatin stretching abrogation: silencing LAP2β almost completely inhibited stress-induced upregulation and silencing HDAC3 or ZBTB7B had some inhibitory effects on stress-induced gene upregulation whereas knocking down LEMD2 had no inhibitory effects (Fig. 1E and Fig. S1C; Fig. S6). In addition, a double depletion of BAF1 (gene for BAF protein) and ZBTB7B (cKros) (Fig. S7) recapitulated the effect of LAP2β depletion on chromatin spontaneous movements, chromatin stretching, and ensuing gene transcription (Fig. S8; compare these results with those in Fig. 1B, D, and E). Together these results suggest that the LAP2β/HDAC3/cKrox complex, in addition to its role in tethering the chromatin to the nuclear lamina, transmits force from nuclear lamina to chromatin to stretch the chromatin to upregulate transcription and LEMD2 is not a force-transmitting molecule.

Fig. 1. LAP2β mediates force transmission to chromatin for a local stress-induced gene upregulation.

(A) Schematics of known structural linkages of molecules from nuclear lamina to chromatin. (B) Normalized spontaneous chromatin movements pretreated with siRNAs for individual molecules (over negative control siRNAs). Mean+s.e.m; n= 125, 73, 90, 96 and 116 cells for negative control siRNA, LAP2β siRNA, HDAC3 siRNA, ZBTB7B (gene for cKrox protein) siRNA, and LEMD2 siRNA, respectively; *P=0.0294; ***, P<0.001; ns=not significantly different. (C) Schematic of local stress application to the apical surface of a cell using a RGD-coated magnetic bead (4 μm in diameter). (D) Bead stress -induced chromatin stretching in CHO cells pretreated with various siRNAs for 48 h. Percent stretch (% stretch) is defined as the distance between two GFP-chromatin spots at the applied peak bead stress minus the distance between the same two GFP-chromatin spots without the applied bead stress and then divided by the distance between the two GFP spots without the bead stress. The magnetic bead stress was 15 Pa at 0.3 Hz. Mean ± s.e.m; n= 48, 54, 46, 47 and 58 cells for each siRNA: negative control, LAP2β, HDAC3, ZBTB7B, and LEMD2, respectively; there was only one bead bound to a single cell; *P=0.0108; ***, P<0.001; ns=not significantly different. (E) RNA FISH analysis of DHFR transcription by the bead stress (15-Pa at 0.3 Hz for 10 min; full DHFR transcript probes were used here to increase the negative control group signal). RNA FISH signals were normalized by the same-siRNA-treated cells without the bead stress. Note that transcription upregulation is heavily dependent on loading duration¹⁰ and here only the early short time transcription response is assayed. Mean+s.e.m; n= 113, 100, 100, 106 and 95 cells without the bead stress; n= 90, 110, 100, 93 and 96 cells with the bead stress; *P=0.0208 between bead stress of negative control siRNA and HDAC3 siRNA, *P =0.0186 between no bead stress and bead stress of LAP2β siRNA, ***P<0.001, ns=not statistically different. Student’s two-tailed t-test.

3.2. Chromatin domain stretching but not compression by whole cell strains upregulates genes

The local bead stress is applied using a cyclic sinusoidal wave function that induces both stretching and compression of the chromatin domain within one loading cycle. Therefore, it is challenging to differentiate whether it is stretching or compression of the chromatin that leads to gene upregulation. In addition, the local bead stress generates a complex stress field inside the cytoplasm and the nucleus and induces both normal strains and shear strains on the chromatin domains [12] and hence this approach cannot be used to definitively determine how chromatin domain alignments affect chromatin-deformation mediated transcription. To determine if the chromatin domain alignment relative to the substrate deformation direction is important in gene transcription, we plated the cells on a fibronectin-coated flexible PDMS (poly-dimethyl-siloxane) substrate to apply a step-function uniaxial stretch (12% or 20%) (Fig. 2A top) or seeded the cells on a pre-stretched PDMS substrate and quickly released the pre-stretch to generate compression (12% or 20%) for a short time of 2 min (Fig. 2A bottom). It is known that GFP spots in the chromatin domain of most cells are aligned in a straight line [22] and thus we were able to use the chromatin domain alignment to determine the potential effects of stretching and compression on gene transcription. As expected, the GFP-labeled chromatin domain that aligned along the uniaxial stretch direction (0°) was stretched (Fig. 2B left panel) and the chromatin domain that aligned perpendicular to the stretch direction (90°) was compressed (Fig. 2B mid-left panel), due to the effect of Poisson’s ratio of the nucleus that was determined to be 0.37 (Fig. S9). Also as expected, the GFP-labeled chromatin domain that aligned along the uniaxial compressing direction (0°) was compressed (Fig. 2B mid-right panel) and the chromatin domain that aligned perpendicular to the compress direction (90°) was stretched (Fig. 2B right panel). The summarized data presented in Fig. 2C and Fig. S10 show that there was substrate-deformation (stretch or compress) magnitude dependent chromatin deformation. When no stretching or no compressing was applied, the chromatin domains were under “no applied deformation” condition and exhibited spontaneous movements, with strains of peak amplitudes of 1% (Fig. 2C and Fig. S11), much lower than the substrate-deformation induced chromatin deformation. The chromatin domains were tethered by the action of the myosin-II dependent cytoskeletal prestress as inhibition of myosin light chain kinase elevated the spontaneous movements of the chromatin domain and thus spontaneous strains (Fig. S11). Interestingly, as the chromatin domains were aligned along the uniaxial stretching direction or perpendicular to the uniaxial compressing direction, DHFR transcription was upregulated (Fig. 2D; Figs. S12 and 13) whereas as the chromatin domains were aligned perpendicular to the stretching direction or along the compressing direction, there was no DHFR transcription upregulation (Fig. 2D; Figs. S12 and 13), suggesting that transcription upregulation is closely associated with the local stretching of the chromatin domains and not with the local compression of the chromatin domains. When the substrate was stretched 12% and 20%, the chromatin domains aligned at 0° were stretched ~9% and 13% respectively (Fig. 2C) and DHFR transcription was upregulated to a similar level of ~30% (Fig. 2D), possibly because the 9% stretch already reached maximum unfolding of the chromatin domains for gene upregulation. On the other hand, When the substrate was compressed 12% and 20%, the chromatin domains aligned at 90° were stretched ~5% and 7% respectively and DHFR transcription was upregulated by a ~14% and 25% respectively (Fig. 2D), exhibiting stretch-magnitude dependent upregulation. Together these results show that it is rapid stretching but not rapid compressing of chromatin domains that quickly upregulate transcription of DHFR.

Fig. 2. Chromatin stretching but not compressing by a whole cell strain via uniaxial substrate deformation upregulates DHFR transcription.

(A) Schematic of whole cell strain (stretching or compression) at the basal surface via uniaxial PDMS (polydimethylsiloxane) substrate deformation. (B) Representative fluorescence images of nucleus (Hoechst 33342 staining) overlaid with GFP-chromatin spots (scale bars, 5 μm) and enlarged GFP-chromatin spots (scale bars, 1 μm) before and after a step-function substrate stretching/compressing of 12% or 20%. The GFP-chromatin spots were 1.5–2 μm from the cell basal surface and so GFP spots deformation was much lower than the substrate deformation. (C) Summarized data of chromatin deformation induced by a uniaxial stretch or compression of the PDMS substrate. Positive values of chromatin deformation indicate that that the distance of the GFP-chromatin spots becomes larger than the unperturbed (original) distance, i.e., the chromatin domain is stretched. Negative values of chromatin deformation indicate that the distance of the GFP-chromatin spots becomes smaller than the unperturbed (original) distance, i.e., the chromatin domain was compressed. Chromatin spontaneous movements without externally applied deformation were used as control (0% stretch/compression). n= 65, 65 and 41 chromatin alignments (each chromatin alignment contains 2–4 GFP spots) along the stretch direction (0°) for stretch at 0%, 12% and 20%; n= 42, 42 and 38 chromatin alignments vertical to the stretch direction (90°) for stretch at 0%, 12% and 20%; n= 52, 49 and 52 chromatin alignments along the compression direction for compression at 0%, 12% and 20%; n= 48, 42 and 48 chromatin alignments vertical to the compression direction for compression at 0%, 12% and 20%; ***P<0.001. (D) Normalized DHFR transcription in CHO cells after a step-function stretch or compression of the PDMS substrate for 2 min. 5’-end DHFR RNA FISH probes were used. n= 130 and 148 cells for stretch at 12% and 20% with chromatin alignment along the stretch direction; n= 115 and 135 cells for stretch at 12% and 20% with chromatin alignment vertical to the stretch direction; n= 101 and 194 and cells for compression at 12% and 20% with chromatin alignment along the compression direction; n= 89 and 191 cells for compression at 12% and 20% with chromatin alignment vertical to the compression direction; n= 397 and 743 cells without substrate deformation (No applied deformation) as control of the substrate stretch and compression respectively. Mean + s.e.m.; n>5 separate experiments; *, P=0.0427; **, P=0.0011; ***, P<0.001; ns=not statistically different.

If LAP2β is a critical structural molecule that mediates force transmission from the nuclear lamina to the chromatin, depletion of LAP2β should also impact substrate-deformation dependent chromatin deformation and the ensuing gene transcription. As such, we applied a step-function uniaxial substrate stretching of 12% or 20%. Similar to what was observed with the local bead stress, chromatin domain stretching was inhibited by more than 50% when LAP2β was depleted (Fig. 3A and Fig. S13A). The chromatin domain stretching was decreased from 8.5% to 3.7% for the 12% substrate stretching and from 12.4% to 6.9% for the 20% substrate stretching (Fig. 3A). After LAP2β was silenced, stretch-induced DHFR upregulation was completely inhibited for the 12% substrate stretch and was substantially inhibited for the 20% substrate stretch (Fig. 3B and Fig. S13B). From Fig. 3A & B, it appears that 4% chromatin domain stretching is the threshold for inducing DHFR upregulation since after LAP2β depletion the gene upregulation occurs when the chromatin stretching was 6.9% but not 3.7%. Furthermore, silencing LAP2β did not affect DHFR transcription of the chromatin domains that aligned perpendicular to the stretching direction (Fig. 3B and Fig. S13B). Knocking down LAP2β also reduced uniaxial 20% compressing-induced chromatin compression for the chromatin domains that aligned along the compressing direction and chromatin stretching for the chromatin domains that aligned perpendicular to the compressing direction (Fig. 3C and Fig. S13C). Chromatin domains were stretched 6.8% in the perpendicular directions, which was reduced to 3.9% after LAP2β knockdown (Fig. 3C). Importantly, knocking down LAP2β completely inhibited stretch-induced DHFR upregulation for chromatin domains that aligned perpendicular to the uniaxial compressing direction (Fig. 3D and Fig. S13D), showing again when the chromatin domain stretching was reduced to less than 4%, there was no gene upregulation, consistent with the substrate stretching results that 4% chromatin stretching is the threshold for transcription upregulation. Together these findings are consistent with those results obtained from a local applied stress and demonstrate that LAP2β is a bona fide force-transmitting molecule connecting the nuclear lamina to the chromatin to mediate force-induced gene upregulation.

Fig. 3. LAP2β mediates substrate deformation-induced force transmission to stretch chromatin and to upregulate transcription.

(A) Chromatin deformation under 12% and 20% uniaxial substrate stretch with LAP2β siRNA #1 treatment of CHO cells (data of LAP2β siRNA #2 treatment are shown in SI Appendix, Fig. S5). Negative values of chromatin deformation indicated that the distance of the GFP spots became smaller than the unperturbed (original) distance, i.e., the chromatin domain was compressed. Mean+s.e.m.; ***P<0.001; chromatin alignments along stretch direction (0°, n=40 and 39 chromatin domains with 12% uniaxial substrate stretch for negative control siRNA and LAP2β siRNA respectively, n=40 and 35 chromatin domains with 20% uniaxial substrate stretch for negative control siRNA and LAP2β siRNA respectively); chromatin alignments perpendicular to stretch direction (90°, n=37 and 35 chromatin domains with 12% uniaxial substrate stretch for negative control siRNA and LAP2β siRNA respectively; n=40 and 36 chromatin domains with 20% uniaxial substrate stretch for negative control siRNA and LAP2β siRNA respectively). (B) Normalized DHFR transcription in cells with 12% and 20% uniaxial substrate stretch. 5’-end DHFR RNA FISH probes were used for quantification of DHFR transcription. Mean+s.e.m.; ***P<0.001; ns=not statistically different. For negative control siRNA, n=230 and 275 cells with 12% uniaxial substrate stretch for chromatin domains along and perpendicular to the stretch direction; n=106 and 104 cells with 20% uniaxial substrate stretch for chromatin domains along and vertical to stretch direction. For LAP2β siRNA, n=217 and 217 cells with 12% uniaxial substrate stretch for chromatin alignments along and perpendicular to stretch direction; n=129 and 108 cells with 20% uniaxial substrate stretch for chromatin alignments along and perpendicular to stretch direction. For 12% uniaxial substrate stretch, the controls were 240 and 519 cells without substrate deformation (No applied deformation) for negative control siRNA and LAP2β siRNA; for 20% uniaxial substrate stretch, the controls were 137 and 148 cells without substrate deformation (No applied deformation) for negative control siRNA and LAP2β siRNA. **P=0.0031 between negative control siRNA and LAP2β siRNA with 20% uniaxial stretch when chromatin aligned along the substrate stretch direction. (C) With 20% uniaxial substrate compression, LAP2β siRNA treatment of CHO cells led to significant decrease in compression of chromatin alignments along compression direction (0°, n=47 and 38 chromatin domains for negative control siRNA and LAP2β siRNA respectively) and decrease in stretch of chromatin alignments perpendicular to compression direction (90°, n=40 and 37 chromatin domains for negative control siRNA and LAP2β siRNA respectively). (D) The decrease in stretch of chromatin alignments perpendicular to compression (20%) direction caused by LAP2β siRNA treatment failed to upregulate DHFR transcription. For negative control siRNA, n=108 and 105 cells for chromatin alignments along and perpendicular to compression direction respectively with 20% uniaxial substrate compression, and signals were normalized by 153 cells without substrate deformation. For LAP2β siRNA, n=141 and 143 cells for chromatin alignments along and perpendicular to compression direction respectively with 20% uniaxial substrate compression, and signals were normalized by 304 cells without substrate deformation. Mean+s.e.m.; ***P<0.001, ns=not statistically different; Student’s t-test.

The above results obtained using uniaxial substrate deformation suggest that chromatin alignment in the nucleus relative to the deformation direction is the key factor in determining local chromatin domain stretching, which is necessary for upregulation of the gene located within that chromatin domain. To further test if this interpretation is valid, we employed biaxial (i.e., both x and y directions) stretching or biaxial compression of the flexible substrate. As expected, the chromatin domains that aligned along either direction of the 12% biaxial stretching (chromatin alignment 0°) were stretched by ~9% and the chromatin domains that aligned diagonally (chromatin alignment 45°, the maximum stretching direction) were stretched by ~12.5% (Fig. 4A). Depletion of LAP2β led to more than 50% reduction in the extent of the chromatin domain stretching for either chromatin alignment (Fig. 4A and Fig. S14A). DHFR transcription was increased by ~30% as a result of the stretch and there was no difference between chromatin alignment 0° and 45° in gene upregulation (Fig. 4B and Fig. S14A). Depletion of LAP2β substantially inhibited stretch-induced gene upregulation (Fig. 4B and Fig. S14B). Application of biaxial 12% substrate compression resulted in ~9% and 12% chromatin domain compression for 0° and 45° domain alignments, respectively (Fig. 4C and Fig. S14C). Importantly, there was no DHFR upregulation for any chromatin domains aligned in any direction of the x-y plane (Fig. 4D and Fig. S14D), completely consistent with the findings of the uniaxial substrate deformation. Collectively, these results show that for a short duration (~2 min) of the whole cell deformation from the basal surface via the substrate deformation, as long as the local chromatin domain is stretched, the gene within that domain can be rapidly upregulated.

Fig. 4. Biaxial substrate stretching but not compression of the whole cells upregulates gene transcription.

(A) Chromatin deformation under 12% biaxial substrate stretch for 2 min with LAP2β siRNA #1 treatment of CHO cells (data of LAP2β siRNA #2 treatment are shown in SI Appendix, Fig. S5). Mean+s.e.m; ***P<0.001; chromatin alignments along one of the two stretch directions (0°, n=46 and 44 chromatin domains for negative control siRNA and LAP2β siRNA, respectively); chromatin alignments 45° relative to the biaxial stretch directions (n=38 and 36 chromatin domains for negative control siRNA and LAP2β siRNA, respectively). (B) Normalized DHFR transcription under 12% biaxial substrate stretch for 2 min. Mean+s.e.m, ***P<0.001; ns=not statistically different. For negative control siRNA, n=175 and 150 cells for chromatin alignments along one of the stretch directions and 45° to the two stretch directions, respectively. For LAP2β siRNA, n=151 and 138 cells for chromatin alignments along one of the two stretch directions and 45° to the biaxial stretch directions. The controls were 211 and 254 cells without substrate deformation (No deformation) for negative control siRNA and LAP2β siRNA, respectively. **P=0.0065 between chromatin aligned 45° to the biaxial stretch directions pretreated with negative control siRNA and LAP2β siRNA. *P=0.0399 between LAP2β siRNA 0° and LAP2β siRNA 45°. (C) Chromatin deformation under 12% biaxial substrate compression for ~2 min with LAP2β siRNA #1 treatment of CHO cells. Negative values of chromatin deformation indicated that the distance between the GFP spots became smaller than the unperturbed (original) distance, i.e., the chromatin domain was compressed. Mean+s.e.m; ***P<0.001; chromatin alignments along one of the two compression directions (0°, n=67 and 35 for negative control siRNA and LAP2β siRNA, respectively); chromatin alignments 45° to the compression directions (n=42 and 33 for negative control siRNA and LAP2β siRNA, respectively). (D) Normalized DHFR transcription under 12% biaxial substrate compression for 2 min. Mean+s.e.m, ns=not statistically different. For negative control siRNA, n=191 and 192 cells for chromatin alignments along one of the compression directions and 45° to the two stretch directions, respectively. For LAP2β siRNA, n=96 and 95 cells for chromatin alignments along one of the two compression directions and 45° to the two compression directions. The controls were 132 and 128 cells without substrate deformation (No applied deformation) for negative control siRNA and LAP2β siRNA, respectively. For B and D, 5’-end DHFR RNA FISH probes were used for quantification of DHFR gene transcription. From A through D, at least 3 independent experiments were performed per condition.

3.3. Mechanism of stretching chromatin in gene upregulation

We have reported that RNA polymerase II (Pol II) recruitments to the promoter sites of the genes is necessary for stress-induced transcription upregulation [10, 11]. To determine how stretching could upregulate gene transcription, we applied a constant biaxial substrate stretch at 12% for 2 min. Chromatin immunoprecipitation (ChIP) assays were performed to assay Pol II phosphorylated serine 5 (S5p) at the promoter site of DHFR, a measure of Pol II recruitment to the DHFR promoter site during transcription initiation, and phosphorylated serine 2 (S2p) at the promoter site of DHFR, a measure of Pol II elongation. Pol II S5p was increased after the biaxial stretch and the elevation of Pol II S5p was substantially inhibited by depletion of LAP2β (Fig. 5A and Fig. S15A). Furthermore, Pol II S2p was increased after the biaxial stretch and the elevation of Pol II S2p was substantially inhibited by depletion of LAP2β (Fig. 5B and Fig. S15B). Since the biaxial stretch was applied for only 2 min and it took at least 10 min to complete the whole transcript of DHFR gene [22], the observed increase in Pol II S2p suggest that some paused Pol IIs were activated by the biaxial stretch. In contrast, application of 12% biaxial constant compression for 2 min did not alter Pol II S5p or Pol II S2p without or with LAP2β depletion (Fig. 5C, D and Fig. S15C, D). As expected, the controls such as the non-promoter regions of DHFR gene (Figs. S16 and S17) or the promoter regions of mechano-non-responsive gene FKBP5 (Figs. S18 and S19) [11] did not exhibit any change in Pol II S2p or Pol II S5p in response to 12% biaxial stretching. Importantly, DHFR gene upregulation is positively associated with the extent of chromatin stretching when all the data from the local bead stress and whole cell stretching are plotted (Fig. 6). These findings suggest that chromatin stretching is a key determinant in gene upregulation. Taken together, these results suggest that LAP2β-dependent stretching but not compressing the chromatin domain upregulates gene transcription because of increased Pol II recruitments to the DHFR gene promoter site to initiate transcription and the activation of paused Pol II at the DHFR gene locus.

Fig. 5. Lap2β mediates substrate stretch-induced recruitment and elongation of Pol II.

(A-D) ChIP assays were performed using normal rabbit IgG (negative control, 2 μg/IP), Pol II S5p antibody (3 μg/IP), or Pol II S2p antibody (3 μg/IP) on sheared chromatin from 2 million CHO cells transfected with scramble control siRNA (Neg Ctr) or LAP2β siRNA#1 with and without 12% biaxial constant substrate stretch for 2 min (A, B) or 12% biaxial constant substrate compression for 2 min (C, D). IP DNA relative to input DNA on DHFR promoter was determined by qPCR. Mean+s.e.m; n=4 independent experiments; ns, not statistically different; for Pol II S5p ChIP, P=0.0137 between biaxial stretch with negative control siRNA and LAP2β siRNA; P=0.0364 between no applied deformation and biaxial stretch with LAP2β siRNA; for Pol II S2p, P=0.0298 between biaxial stretch with negative control siRNA and LAP2β siRNA; P=0.044 between no applied deformation and biaxial stretch with LAP2β siRNA; **P<0.01; one-way ANOVA with Bonferroni test.

Fig. 6. DHFR upregulation is positively associated with the extent of chromatin stretching.

Mean+/−s.e.m. Each data point represents mean upregulation of DHFR transcription under mean chromatin stretching of CHO cells from at least 5 separate experiments. Note that there is no statistical difference between the no-stress baseline and 2–4% chromatin stretching condition for gene upregulation. ns, not statistically different. *P<0.05; one-way ANOVA with Bonferroni test.

3.4. Compressing chromatin for 60 min results in gene downregulation

We wondered why biaxial chromatin domain compression that should condense all domains of the chromatin in the x-y plane did not decrease DHFR transcription. We speculated that it was possibly due to the short duration (only 2 min) of the compression application such that the bonds of the higher structures of chromatins were not reformed yet. To test this idea, we extended the duration of compression experiments to 60 min. Indeed, as we expected, biaxial 12% step-function compression for 60 min downregulated DHFR transcription (Fig. S20). These results suggest that 60-min biaxial compression was long enough for the chromatins to form non-covalent bonds to condense and to shield gene promoters from Pol II binding for gene activation initiation. Trichostatin A is a histone deacetylase inhibitor that induces hyperacetylation of histone tails to decondense the chromatin. As expected, pretreating the cells with trichostatin A substantially decreased biaxial substrate-compression induced chromatin domain compression since the chromatins are already decondensed by trichostatin A (Fig. S21).

3.5. YAP or calcium is not involved in force-induced rapid gene upregulation

To determine if YAP was involved in force-induced gene upregulation, a local magnetic bead stress was applied to the CHO cell at the apical surface for 2 min. There was no translocation of YAP into the nucleus and silencing YAP did not inhibit local stress-induced DHFR upregulation (Figs. S22 and S23). To further determine the role of YAP, a 12% uniaxial substrate-stretch was applied to the attached cells. Translocation of YAP occurred only after 60 min of substrate stretching and transcription upregulation of the YAP regulated gene ctgf was inhibited by knocking down YAP (Fig. S24). These results suggest that YAP translocation was not involved in the local stress-induced or the whole-cell strain-induced DHFR upregulation in our experiments when mechanical loading was applied for a short period of 2–10 min.

Living cells have mechanosensitive ion channels that could be activated for calcium entry into the cytoplasm. When these ion channels were blocked with pre-treatment of gadolinium or when external calcium was removed using calcium-free medium [26], there was no inhibition of the local bead stress-induced DHFR upregulation (Fig. S25). Similarly, when endoplasmic reticulum calcium store was inhibited with nifedipine or when both external and internal calcium were blocked [32], there was no inhibition of local bead stress-induced gene upregulation (Fig. S25). These findings are consistent with an earlier study that mechanosensitive ion channels are not opened when the applied local magnetic bead stress was less than 200 Pa [27]. Together these results suggest that calcium is not involved in the local stress-induced gene upregulation in the present study.

3.6. Transport from cytoplasm to nucleus is not required for force-induced rapid gene upregulation

To further determine if transport of molecules from the cytoplasm to the nucleus is necessary for force-induced gene upregulation, we employed molecules that could alter nuclear pore complex. It is known that BAF (~10 kD) is a force-transmitting molecule between the nuclear lamina and the chromatin [10]. Pre-treating the cells with Pitstop 2, a disrupter of the nuclear pore complex permeability barrier [28], caused BAF to leak out of the nucleus (Fig. S26A, B) but had no effect in nuclear YAP/TAZ (Fig. S26C). Pitstop 2 treatment did not elevate spontaneous chromatin movement much but inhibited stress-induced chromatin stretching (Fig. S27). Importantly Pitstop 2 treatment inhibited local bead stress-induced DHFR upregulation (Fig. S26D), consistent with the role of BAF in mediating stress transfer to the chromatin where 50% knockdown of BAF inhibits bead stress-induced DHFR upregulation [10]. Since the stress-induced gene upregulation is completely abolished after Pitstop2 treatment (Fig. S26D) even though the chromatin was still stretched by ~7% (Fig. S27), these data suggest that some other molecules that are necessary for gene upregulation might be lost or inactivated in the nucleus. Pretreating the cells with WGA (wheat germ agglutinin) that specifically blocks the nuclear pores [29] had no inhibitory effects on the local stress-induced DHFR upregulation (Fig. S26D); pretreating the cells with Pitstop 2 and WGA simultaneously did not inhibit local stress-induced gene upregulation either (Fig. S26D), suggesting that Pitstop 2 disrupting effect on the nuclear pore complex is blocked by WGA. Together with the findings that blocking calcium entry into the cytoplasm and depleting YAP to inhibit its translocation have no impact on stress-induced gene upregulation, these results suggest that transport of molecules from the cytoplasm to the nucleus may not be necessary for local stress-induced rapid gene upregulation.

3.6. Initiation and/or release of paused Pol II are important in force-induced rapid gene upregulation

To further examine whether transcript initiation and/or release of paused Pol II are important in the local stress-induced transcription, we treated the cells with either Triptolide, an inhibitor of TFIIH helicase activity and hence transcription initiation [30] or Flavopiridol, an inhibitor of P-TEFb and thus of paused Pol II-dependent transcript elongation [31], or both drugs. We found that either drug or both inhibited stress-angle dependent DHFR upregulation and stress-induced DHFR upregulation (Fig. S28). ChIP assays show that TRP inhibits force-induced Pol II S5p and FP inhibits force-induced Pol II S2p (Fig. S29), consistent with their specific inhibitory effects on force-induced gene upregulation. These results suggest that both transcription initiation and transcript elongation are activated by the applied local bead stress. In the present study, we applied a step loading for PDMS substrate deformation which may bring about creep in the chromatins. To examine this possibility, we applied a step-function constant uniaxial 12% stretch. We found that there was no difference in chromatin domain stretching between 2 min and 10 min (Fig. S30), indicating that there is no creep in the chromatin domains, suggesting that the chromatin domains behave like an elastic solid during the 10-min stretching.

4. Discussion

In the current study, we find that LAP2β functions as a force-transmitting molecule that transfer stress from the nuclear lamina to the chromatin and as a tethering molecule that is part of the LAP2β/HDAC3/cKrox tripartite complex to tether the chromatin to the nuclear lamina and thus to the nuclear envelope. The fact that LAP2β is more dominant than HDAC3 or cKrox in tethering chromatin or transmitting stresses suggests that the structural pathway of LAP2β and BAF is more important than that of the LAP2β/HDAC3/cKrox complex (see Fig. 1A) in force-transmitting to chromatin and chromatin-tethering. This interpretation is supported by the previous finding that knocking down BAF completely inhibits force-induced DHFR upregulation [10]. We note, however, the current finding on the tethering and force-transmitting roles of LAP2β does not exclude its potential role as a signaling scaffold and/or in performing other nuclear functions.

Mechanical loading in general can produce normal (tensile and compressive) stresses and/or shear stresses on the cell surface such as blood flow-induced shear stresses on the endothelial cells and blood pressure oscillation-induced stretching (tensile) and/or compression of vessel cells [32, 33]. While it is known that tensile stresses can exert their effects on the cells via stretching and unfolding proteins [34, 35] and chromatins [10–12], compressive stresses have been shown in causing cellular responses via translocation of cytoplasmic molecules [8, 9, 14]. It has remained unclear if the compressive forces can exert their effects on gene transcription via directly altering chromatin structures. In the present study, we find that only chromatin domain stretching but not compressing can rapidly upregulate DHFR transcription. From the results of the uniaxial substrate compression experiments, it is shown that the cell surface compression can only upregulate gene transcription within the chromatin domains that are aligned perpendicularly from the compression direction, which are, de facto, stretched because of the Poisson’s ratio effect. These findings are confirmed by the results obtained with biaxial stretching or compression, where all chromatin domains in the x-y plane are either stretched under biaxial substrate stretching or compressed under biaxial substrate compression. A study that applies a static compressive stress of 1 kPa for 60 min on the whole cell along the z-direction demonstrates chromatin condensation as a result of translocation of histone deacetylase 3 (HDAC3) into the nucleus [36]. The compression modes in the two studies are quite different: our study of the x-y plane biaxial compression that would lead to z-direction chromatin domain stretching and that study of the z-direction compression that would lead to x-y plane chromatin domain stretching, suggesting different chromatin domain deformation and different modulation of gene transcription. Nevertheless, we cannot rule out the role of translocation of molecules from the cytoplasm in regulating gene transcription because the compression duration was relatively long (60 min) in Fig. S20, but for most of our experiments, the stretching or compressing duration was only 2 min, not long enough for translocated molecules to affect gene transcription. In our study, when chromatin domain deformation is quantified, only the X-Y plane displacements are measured, which is a 2D simplification of the 3D displacement field that includes displacements in the Z direction. However, since all CHO cells are highly polarized and spread, most GFP-chromatin spots did not move out of the view field under mechanical loading, suggesting that the displacements of the chromatin in the Z direction are small relative to those in the x-y plane. Nevertheless, the uniaxial or biaxial substrate compression may still cause alignments of chromatin domains along the Z direction to be stretched and the genes within those domains to be activated. This possibility could be explored in the future. One important issue is the question of what is actually deformed when the chromatin domain is stretched. From the length and the diameter (~2 nm) of the B-DNA and the diameter (~100 nm) of the chromatin, it is apparent that the chromatin is a highly packed structure that is partially unpacked or decondensed when it is under mechanical loading [11]. The magnitude of stretching of the chromatin domain (>4% stretching) is likely enough for the Pol II to bind to the promoter site of the DHFR gene to initiate transcription and for the paused Pol II to be re-activated to elongate the transcripts. The histone 2B (H2B)-GFP deformation map in the nucleus shows that distribution of the extent of the chromatin stretching or compression is heterogeneous under the local magnetic bead stress at the cell surface [10, 37], suggesting that chromatins in the nucleus are not equally deformed and chromatin domain deformation depends on its location. Indeed, recently we find that chromatin domains near the nuclear periphery (<1.5 μm) (heterochromatin) are stiffer than the chromatin domains away (>1.5 μm) from the nuclear membrane [11]. Moreover, the chromatin deformation is throughout the whole nucleus when the local stress is applied outside on one side of the nucleus [10, 37], consistent with a recent finding that chromatins in the nucleus of a cell behave like a solid [38] and the published report that chromatin domain stretching does not vary with loading frequency from 0.3 to 6 Hz [11]. Importantly, it is shown recently that the chromatin domains must be demethylated at H3K9 for the gene to be activated or upregulated in addition to be stretched [11]. This finding provides an explanation of why some genes are activated and some are not activated even when the chromatins are deformed by mechanical stresses [11]. There is evidence that FET-family transcriptional regulators phase separate within living cells and regulate localized transcription [39]. It remains to be determined if phase separation of condensates and demixing in the nucleus can regulate external stress-induced transcription. A recent study shows that chromatin distributes only near the nuclear lamina and separates from the nuclear interior in Drosophila larvae and tissues as well as human effector T cells and neutrophils [40]; it will be interesting to determine what impact external force might have on chromatin deformation and gene regulation in the nuclear peripheral chromatin in these cells.

Our results that YAP is not involved in force-induced rapid gene upregulation are consistent with a published report that YAP molecules translocate only after four-hour cyclic stretching (1%–15%) of the flexible substrate (for the duration of 1 to 6 h) [41]. However, another study shows that when a cell nucleus is compressed, the relaying molecule YAP in the cytoplasm starts to translocate into the nucleus within several minutes of loading [8]. The reason for the difference is not clear at this time but it might be that the local compressive stress was applied directly over the nucleus and the magnitude of the local compressive stress (>45 Pa) was higher than the 15-Pa local stress used in our experiments.

Previously we have shown that myosin II dependent cytoskeletal prestress regulates chromatin compaction: elevating the prestress by treating the cells with Lysophosphatidic Acid (LPA) to activate RhoA decreases chromatin condensation and lowering the prestress by inhibiting ROCK with Y-27632 or inhibiting myosin light chain kinase with ML-7 increases chromatin condensation [10]. In addition, elevating the prestress increases endogenous DHFR transcription and lowering the prestress inhibits endogenous DHFR transcription [10]. Furthermore, inhibiting myosin II with blebbistatin substantially inhibited stress-angle and stress-mode dependent DHFR upregulation [12], suggesting that the cytoskeletal prestress is important in propagating applied stresses to stretch chromatin for gene upregulation. In the current study, we find that myosin-dependent prestress is also important in controlling the chromatin spontaneous movements as inhibiting myosin light chain kinase with ML-7 increases the chromatin endogenous displacements and strains (Fig. S11). These findings are consistent with the published reports [10, 12] and are in line with a report that myosin II inhibition rescues nuclear membrane rupture by decreasing large nuclear membrane deformation [42] when the cell forces its nucleus through a highly constricted pore of 3 μm. Together these findings suggest that the cytoskeletal prestress regulates chromatin condensation and compaction as well as chromatin endogenous movements and external stress transmission. Mechanosensitive phosphorylation and turnover of lamin A has been reported [3], which is followed by the finding that lamin A phosphorylation and turnover feedbacks to myosin-II mediated cell tension [43] and by the revelation that substrate rigidity driven actomyosin contractility tenses the nucleus to favor lamin-A/C accumulation and suppress soft tissue phenotypes [44]. The finding in the current study that Lap2β depletion decreases cell tractions (i.e., myosin II dependent prestress) is consistent with the published results on the effect of lamins on actomyosin contractility.

One question is whether the observed chromatin stretching and the ensuing gene upregulation is due to force sensitive elements at the cell surface of the anisotropic cytoskeleton responding to forces in a direction- and amplitude-dependent manner by releasing or modifying signaling molecules such as phosphorylating some transcription factors that translocate to the nucleus. The finding that nuclear protein BAF depletion [10] and Lap2β depletion can completely abolish the effects of force directions on chromatin deformation and the ensuing gene upregulation suggests that these intranuclear molecules transmit forces, independent of the force-sensitive elements at the cell surface. At 100 Hz, stress-induced chromatin stretching is observed within 1–2 milliseconds [11]. It would be extremely difficult to interpret these results by the biochemical cascades and the signaling molecules in the cytoplasm initiated by the force sensitive elements at the cell surface. However, it is still possible that some force-sensitive elements such as Piezo1 at the endoplasmic reticulum [45] and other stretch-activated molecules [9] at the nuclear envelope might mediate ultra-fast diffusion- or translocation-dependent mechanism might trigger the indirect mechanism of chromatin deformation. The finding from the current study that local chromatin domain alignment along the stretching direction but not perpendicular to the stretching direction dictates the chromatin-stretching dependent gene upregulation appears to rule out the contribution from Piezo1 or other stretch-sensitive ion channels for chromatin deformation since molecules from these ion channels should lead to a large-scale chromatin deformation that does not depend on chromatin domain alignment. Furthermore, since the level of gene upregulation is positively associated with the extent of chromatin stretching, it strongly supports the notion that the almost instantaneous chromatin stretching is due to a direct force effect on the chromatin, rather than an indirect effect as a result of ion channel openings at the nuclear envelope or the perinuclear endoplasmic reticulum that are expected to trigger an all-or-none response of gene activation and/or upregulation.

5. Conclusions

In summary our current study reveals that Lap2β behaves as a force-transmitting molecule that transmits stress from the nuclear lamina to the chromatin to deform chromatin to regulate gene transcription and as a tether to anchor the chromatin to the nuclear lamina. We demonstrate that chromatin stretching but not compression upregulates rapid DHFR transcription in response to a local stress applied to the apical surface of the cell and substrate deformation applied to the basal surface of the cell via integrins.