Defects in Emerin–Nucleoskeleton Binding Disrupt Nuclear Structure and Promote Breast Cancer Cell Motility and Metastasis

The ability for cells to move within their environment is crucial for numerous physiologic and pathologic processes. During development, cell migration contributes to shaping of the growing embryo and forming nascent tissues (1). In mature organisms, immune cells, such as neutrophils, are mobilized from the blood to enter sites of infection (1). Migration of epithelial cells and fibroblasts is vital for proper wound healing and tissue repair (2). Cancer cell migration is required for metastasis, which is responsible for approximately 90% of all cancer deaths (3). During metastasis, cancer cells leave the primary tumor by both pushing and pulling their way through surrounding tissues to enter and exit blood vessels and colonize distant sites (1, 4, 5).

Altered nuclear structure remains a hallmark for cancer diagnosis (6). Histologic features such as nuclear size, shape, number of nucleoli, and chromatin texture are considered clinically relevant, however the functional significance of these alterations in the context of cancer progression remains unclear (6). Both smaller and larger nuclei have been implicated in increased invasiveness during carcinogenesis, but one thing all of these cells share during metastatic transformation is an increase in nuclear deformability regardless of size (6, 7). The cytoplasm is very flexible and can undergo large deformations (6). The cytoskeleton can also actively remodel, allowing the cell to penetrate openings as small as 1 μm (8). The cell nucleus, on the other hand, is two to ten times stiffer than the surrounding cytoplasm and occupies about 10% of the cellular volume, with a typical diameter of 10 to 20 μm (9). The nucleus is larger than many of the openings encountered in the extracellular environment. Thus, substantial nuclear deformations (10) are required for cells to move through these small spaces.

Nuclear deformability is affected by several factors, including nuclear envelope protein expression and chromatin structure (4, 11, 12). The nuclear envelope is comprised of two lipid bilayers, the outer nuclear membrane and the inner nuclear membrane (INM). Underlying the INM is a network of type-V intermediate filament proteins called lamins. This network is composed of overlapping A- and B-type lamin filaments that form along the INM, which along with the integral INM proteins form the nuclear lamina (13). The relative expression of A- and B-type lamins differentially effects the shape and mechanical properties of the nucleus (14). Cells with low levels of A-type lamins have irregularly shaped nuclei, whereas those with high levels of lamin A/C have rounded nuclei (4, 15). Nuclei of cells with low levels of B-type lamins form blebs, whereas cells with high levels of lamin B form invaginations (15).

There are more than 130 predicted INM proteins, many of which show cell-type-specific expression patterns (16–18). The importance of the nuclear lamina in providing structural support to the nucleus and controlling nuclear size is well established (9, 19). In addition, several studies have reported altered distribution, expression, or both, of INM proteins in various human cancers including ovarian cancer, breast cancer, basal cell carcinoma, colorectal carcinoma, hepatocellular carcinoma, osteosarcoma, and prostate and thyroid cancers (20–26). Although the impact of lamins in cancer progression and metastasis is becoming clearer (9, 15, 19, 27–29), how INM proteins impact cancer progression and metastasis remains unclear.

Emerin is a ubiquitously expressed integral INM protein (30, 31) with roles in nuclear structure, chromatin architecture, genomic organization, cell signaling, and gene expression (32–36). Recent evidence supports a role for emerin in cancer (29). Emerin expression is decreased 7.5-fold (EMBL-EBI; GSE52914) in triple-negative breast cancer (37–39). Breast cancer prognosis is negatively correlated with emerin expression, as 5- and 10-year survival rates for emerin low expressors was 75% and 47%, respectively, compared with emerin high-expressors, which were 88% and 63%, respectively (Human Protein Atlas analysis of data from the Cancer Genome Atlas Network). The Cancer Genome Atlas Network also reports at least 102 EMD mutations linked to cancer, many of which are expected to affect emerin expression and protein folding.

Emerin is important for regulating nuclear structure through its interactions with lamins, nuclear actin, the LINC complex, and chromatin (25, 40, 41). How these functional interactions affect nuclear structure during cancer transformation to impact invasive cancer cell migration has not been investigated. Here we report the functional interactions of emerin with the nucleoskeleton is important for the nuclear structural changes occurring during metastatic transformation and establish a link between increased emerin function and metastasis inhibition.

MDA-MB-231 (ATCC Catalog No.: HTB-26, RRID:CVCL_0062), MDA-157 (ATCC Catalog No.: HTB-24, RRID:CVCL_0618), MCF10A (ATCC Catalog No.: CRL-10317, RRID:CVCL_0598), and normal primary mammary epithelial cells (Catalog No.: PCS-600–010) were purchased from ATCC. The MDA-231 and MDA-157 cells were grown in DMEM with high glucose and GlutaMAX supplement (Gibco/Life Technologies, Catalog No.: 10566-016) with 10% FBS (Gibco/Life Technologies, Catalog No.: 16000-044) and 1% penicillin/streptomycin (Gibco/Life Technologies, Catalog No.: 5140-122). MCF10A cells were grown in Ham's F-12 media (Lonza, Catalog No.: 12-615F) with 5% horse serum (Gibco/Life Technologies, Catalog No.: 16050-130), 20 ng/mL EGF (Life Technologies, Catalog No.: PHG0313), 0.5 mg/mL hydrocortisone (Thermo Fisher Scientific, Catalog No.: AC35245-0010), 100 ng/mL cholera toxin (MilliporeSigma, Catalog No.: 227036), 10 μg/mL Insulin (Sigma-Aldrich, Catalog No.: 10516), and 1% penicillin/streptomycin. Primary mammary epithelial cells were grown in mammary epithelial cell basal medium (ATCC, Catalog No.: PCS-600-030) with all the components from the Mammary Epithelial Cell Growth Kit (ATCC, Catalog No.: PCS-600-040). All cell lines were maintained at 37°C in 5% CO₂. Cell lines are authenticated yearly using ATCC short tandem repeat (STR) profiling through their Cell Line Authentication Service. The latest cell line authentication was done in January 2021. Mycoplasma testing is done regularly by PCR using the Universal Mycoplasma Detection Kit (ATCC, Catalog No.: 30-1012K).

Cell-cycle analysis was done using a Guava Muse Cell Analyzer (Luminex, Catalog No.: 0500-3115) using the Muse Cell Cycle Kit (Luminex, Catalog No.: MCH100106) and Muse Annexin V & Dead Cell Kit (Luminex, Catalog No.: MCH100105), per manufacturer instructions. Proliferation was analyzed by plating 1 × 10⁵ cells of each cell line in a 12-well plate. Cells were trypsinized at 24, 48, 72, 96, and 120 hours after plating and counted using a Countess II FL Automated Cell Counter (Thermo Fisher Scientific).

Falcon trans-well inserts with 3.0 μm (Catalog No.: 353096) or 8.0 μm (Catalog No.: 353097) pores were used for the cell migration assays. For the cell invasion assays, 8.0 μm pores coated with Matrigel (Catalog No.: 354483) were placed in 24-well plates (Falcon, Catalog No.: 353504). Cells were plated at 2 × 10⁵ cells/insert in serum-free media. Complete growth media was placed in the well beneath the insert. Cells were allowed to migrate for 24 hours. Cells that did not migrate remained on top of the porous membrane. These cells were wiped off using PBS and a sterile cotton swab. The cells on the bottom of the membrane were fixed for 15 minutes using 3.7% formaldehyde in PBS, washed three times with PBS, and permeabilized with 0.5% Triton X-100 in PBS for 20 minutes. Membranes were cut away from the inserts and placed cell-side down on a drop of Prolong Diamond Antifade with DAPI (Life Technologies, Catalog No.: P36971) on a slide (VWR, Catalog No.: 48311-703). A coverslip (VWR, Catalog No.: 48366-205) was placed on top and allowed to solidify in the dark overnight.

Tissue protein lysates were obtained from Origene. Samples used were CP565563, CP531533, CP544778, CP552484, CP627491, CP529617, CP531866, CP537436, CP537799, CP541360, CP542781, CP603959, CP607640, CP626552, and CP8115065. Samples were suspended in NuPAGE LDS buffer (Life Technologies, Catalog No.: NP0008) containing reducing agent (Life Technologies, Catalog No.: NP0009). MDA-231 cells stably expressing GFP-emerin, GFP-M45A, GFP-M151, GFP-S54F, and GFP-M196 were collected and resuspended in NuPAGE LDS buffer containing reducing agent. Samples were resolved on by SDS-PAGE (Life Technolgies, Catalog No.: NP323BOX) and transferred to nitrocellulose membranes (GE Healthcare, Catalog No.: 10600004). Membranes were blocked using 3% BSA (Sigma-Aldrich, Catalog No.: A7906) in PBS + 0.1% Tween-20 (PBST) for 2 hours. Appropriate primary and secondary antibodies were diluted in 3% BSA in PBST and incubated on the membrane for 2 hours each. Signal intensity was measured using the ChemiDoc Imaging system (Bio-Rad). Densitometric assessment of bands was done using the Image Lab software (Bio-Rad). The content of each protein was normalized to the proper control band (either actin or γ-tubulin). Patient samples were normalized to the CP565563 control patient without cancer. MDA-231 cells expressing GFP-emerin or each GFP-emerin mutant were normalized to MDA-231 cells. Primary antibodies used for blotting included rabbit emerin polyclonal antibody (1:5,000 dilution; Proteintech, Catalog No.: 10351-I-AP), mouse γ-tubulin mAb (1:10,000 dilution; Sigma-Aldrich, Catalog No.: T6557, RRID:AB_532292), mouse actin mAb (1:2,500 dilution; Sigma-Aldrich, Catalog No.: A5441, RRID:AB_476744), rabbit H3K9me2 polyclonal antibody (1:5,000 dilution; Active Motif, Catalog No.: 39239, RRID:AB_2793199), rabbit H3K9me3 polyclonal antibody (1:10,000 dilution; Abcam, Catalog No.: ab8898, RRID:AB_306848), and rabbit H3K27me3 polyclonal antibody (1:10,000 dilution; Millipore Sigma, Catalog No.: 07-449, RRID:AB_310624). Secondary antibodies included horseradish peroxidase-conjugated rabbit (1:10,000 dilution; Invitrogen, Catalog No.: 31462) and mouse (1:10,000 dilution; Invitrogen, Catalog No.: 31432) antibodies.

Total RNA from tumors and normal breast tissue controls were purchased from Origene Technologies, Inc. The RNAs purchased were matched to protein samples. 1-step qPCR was performed on 10 ng of total RNA from each sample using the Verso SYBR Green 1-Step RT-qPCR ROX Mix (Thermo Fisher Scientific) per manufacturer instructions, except that 500 nm of each primer were added, using the ABI StepOne Plus Real Time PCR Machine (Thermo Fisher Scientific). Relative expression was determined by comparing each experimental to the normal control breast RNA samples. Primer sequences used were purchased from Integrated DNA Technologies and are: GAPDH, 5′-ACATCGCTCAGACACCATG-3′ and 5′-TGTAGTTGAGGTCAATGAAGGG-3′; and EMD, 5′-AGCTCTCTTCATAGTAGTCGTCA and 5′-CTCCTCTTATAGCTTCTCTGACTTG.

This study was conducted according to the guidelines of the NIH and the University of the Sciences' Institutional Animal Care and Use Committee (IACUC). Six- to 8-week-old athymic NCR-nu/nu female nude mice were obtained from Charles River Laboratories. Mice were injected intraperitoneally with 0.1 to 0.6 mg/kg buprenorphine 1 hour prior to surgery. Mice were anesthetized with 2% to 5% isoflurane by inhalation. Using the orthotopic injection technique as described previously (42), a ventral incision was performed to expose the fourth inguinal right mammary gland. MDA-231 cells (5×10⁶) stably expressing infrared fluorescent protein 713 (iRFP713) and GFP-emerin or each GFP-emerin mutant were injected into the right mammary fat pad using a sterile syringe with a 26-gauge needle (n = 14–25 for each group). The incision was subsequently closed with four to five surgical staples and mice recovered from surgery on a 37°C heated pad and were closely monitored postsurgery for the entire protocol.

Mice were anesthetized with 2% to 5% isoflurane and imaged with the LICOR Pearl Trilogy small animal system (LICOR). Images were analyzed using the LICOR Image Studio software auto-selection tool with 5.5 SDs. The background reference circle was standardized for all image analyses and was placed on the contralateral mammary fat pad that was not injected. Eight weeks after fat pad injections, mice were euthanized by CO₂ inhalation and tumors were harvested, weighed, measured, and imaged. The tumors collected were cut in half and each half was either frozen in liquid nitrogen or fixed in 4% paraformaldehyde (PFA) for Western blotting and IHC analyses, respectively. Lungs were inflated with PFA, excised, and imaged on the LICOR Pearl Trilogy small animal system. Lungs were preserved in PFA for IHC analyses.

Wells were coated with poly-L-lysine (Advanced Biomatrix, Catalog No.: 5048) prior to seeding cells. Cells were seeded at confluency and pretreated with 10 μg/mL polybrene (Vector Builder). Cells were transduced using a multiplicity of infection of 100 with LV-iRFP713-P2A-Puro (Imanis Life Sciences, Catalog No.: LV032-L). Media was changed 4, 24, and 48 hours after treatment with the lentivirus. Selection with 4 ng/mL of puromycin dihydrochloride (Corning, Catalog No.: 61-385-RA) started 72 hours after transduction.

Wells were coated with poly-L-lysine prior to seeding cells. MDA-231 or MCF10A cells were plated at 70% confluency. The Lipofectamine3000 (Invitrogen, Catalog No.: L3000-015) protocol was used to transfect cells with GFP-emerin or each GFP-emerin mutant plasmid. Media was changed 24 hours after transfection. Selection with 3.6 mg/mL of G418 (VWR, Catalog No.: J847) was started 72 hours after transfection.

The Neon Transfection System (Invitrogen, Catalog No.: MPK1096) Kit was used to electroporate cells. MDA-231, MDA-157, or MCF10A cells (5 × 10⁶ cells/mL) were resuspended in the 100 μL tip. Cells were electroporated using protocols per manufacturer instructions. The MDA-231 and MDA-157 parameters used were 1,350 V, 20 milliseconds, 2 pulses. The MCF10A parameters used were 1,250 V, 20 milliseconds, 2 pulses. Media was changed 24 hours after electroporation. Selection with 3.6 mg/mL of G418 started 72 hours after electroporation.

Invasive breast cancer cells lines containing GFP-emerin or each GFP-emerin mutant were plated on coverslips at equal densities. Cells were rinsed with PBS, fixed with 3.7% formaldehyde in PBS, washed three times with PBS, treated with wheat germ agglutinin 594 conjugate (Thermo Fisher Scientific, Catalog No.: W11262) at 5.0 μg/mL for 10 minutes, and permeabilized with a 0.5% Triton X-100 in PBS for 20 minutes. Cells were rinsed three times with PBS and mounted with Prolong Diamond Antifade with DAPI (Life Technologies, Catalog No.: P36971) and visualized using the DAPI fluorescent light cube (Thermo Fisher Scientific, EVOS, Catalog No.: amep4650), Texas Red fluorescent light cube (Thermo Fisher Scientific, EVOS, Catalog No.: amep4655), and GFP fluorescent light cube (Thermo Fisher Scientific, EVOS, Catalog No.: amep4651) on the EVOS FL Auto using a 40× objective. Nuclear area was analyzed with the NII_Plugin on ImageJ (ImageJ, RRID:SCR_003070) using these DAPI images (43). Nuclei intensity segmentation threshold was standardized at 30 pixels and a minimum nuclear area was set at 200 pixels for all images. Images were manually checked to ensure one nucleus was selected in each box and debris or apoptosing nuclei were avoided. Cytoplasmic area was analyzed with the NII_Plugin on Image J using the Texas Red images. The intensity segmentation threshold was standardized at 30 pixels and a minimum area was set at 300 pixels for all images. The cytoplasmic and nuclear area was converted from pixels to microns prior to analysis. Volume analysis was performed using 60× images from a Nikon Ti2-E confocal microscope using the DAPI and GFP lasers (Nikon). The NIS Elements AR system was used to measure volume.

To examine if emerin expression is correlated with breast cancer invasiveness, we measured the protein expression levels of emerin in two invasive breast cancer cell lines. MDA-231 and MDA-157 cells had 1.5- to 1.8-fold less emerin protein expression compared with primary mammary epithelial cells and MCF10A cells (Fig. 1A and B). MDA-231 (109.5 μm² ± 2.9 μm²) and MDA-157 (115.6 μm² ± 2.5 μm²) nuclei each had a 1.6- to 1.7-fold decrease in nuclear area when compared with the MCF10A nuclei (183.5 μm² ± 5.9 μm², Fig. 1A and C). The DAPI images of cells plated at the same density highlight the extensive nuclear deformations seen in the invasive breast cancer cell lines when compared with the size, regularity, and circularity of the MCF10A nuclei (Fig. 1D; Supplementary Fig. S1A). The circularity of the nuclei was analyzed with 1.0 being a perfect, uniform circle. The MDA-231 (0.89 ± 0.08) and MDA-157 (0.87 ± 0.07) cells had more irregularly shaped nuclei when compared with the MCF10A cells (0.96 ± 0.06; Fig. 1E). Similar results were seen in emerin-null myogenic progenitor cells (Supplementary Figs. S1B and S1C).

GFP-emerin was expressed in MDA-231 cells to investigate whether emerin regulates nuclear morphology in invasive breast cancer cells. The nuclear area of cells expressing GFP or GFP-emerin was measured and compared with untransfected MDA-231 cells. There was a significant increase in nuclear area of GFP-emerin-expressing cells compared with untransfected cells (Fig. 2A and B). GFP-emerin was expressed at similar levels as endogenous emerin, which increases total emerin expression to levels similar to endogenous emerin levels in MCF10A cells (Fig. 3A and B). Similar results were seen in MDA-157 cells (Supplementary Figs. S2A and S2B). GFP-emerin expression in MCF10A cells had no significant effect on nuclear area (Fig. 2C and D).

Selected GFP-emerin mutants (Fig. 3C) were overexpressed in MDA-231 cells to examine the functional interactions important for emerin modulating nuclear size. Each emerin mutant disrupts the binding of only one or two binding partners (Fig. 3C). GFP-M45A (negative control) disrupts binding to all emerin partners. We successfully created MDA-231 cells stably expressing each GFP-emerin mutant at wild-type levels without altering the expression of endogenous emerin in these cells. This resulted in a 2-fold increase of total emerin expression (Fig. 3A and B), which equals emerin expression in MCF10A cells (Figs. 1A, 3A and B) and normal primary mammary epithelial cells (Fig. 1A and B). GFP-emerin rescued nuclear area of MDA-231 cells (180.8 ± 8.5 μm²; Fig. 3D and E), as it was comparable to MCF10A nuclear area (183.5 ± 2.8 μm²; Fig. 2C). GFP-M45A failed to rescue nuclear area (88.8 ± 3.9 μm²), as expected. GFP-S54F (103.2 ± 5.5 μm²) and GFP-M151 (90.7 ± 2.6 μm²) also failed to rescue nuclear area. GFP-M24 (160.8 ± 7.8 μm²) and GFP-M196 (182.4 ± 11.7 μm²) rescued nuclear area (Fig. 3D and E). Nuclear volume was measured in MDA-231 cells expressing each GFP-emerin mutant to confirm changes in nuclear area correlated with changes in nuclear volume. The nuclear volume of MDA-231 cells was 250.7 ± 14.57 μm³ and the nuclear volume of MCF10A cells was 314.9 ± 12.72 μm³ (Fig. 3F and G). GFP-emerin rescued nuclear volume (327 ± 10.24 μm³; Fig. 3F and G). GFP-M45A failed to rescue nuclear volume (186 ± 10.65 μm³), as expected. GFP-S54F (221.2 ± 18.91 μm³) and GFP-M151 (176.3 ± 6.45 μm³) also failed to rescue nuclear volume. GFP-M196 (338.1 ± 23.71 μm³) rescued nuclear volume (Fig. 3F and G).

Nuclear circularity was measured in MDA-231 cells expressing GFP-emerin and each GFP-emerin mutant. GFP-emerin rescued nuclear circularity (0.94 ± 0.08, Fig. 3H), as it was comparable to MCF10A cells (0.96 ± 0.05, Fig. 1E). GFP-M45A (0.81 ± 0.05), GFP-S54F (0.82 ± 0.03), and GFP-M151 (0.87 ± 0.05) failed to rescue nuclear circularity (Fig. 3H). GFP-M24 (0.96 ± 0.05) and GFP-M196 (0.94 ± 0.06) rescued nuclear circularity (Fig. 3H). Nuclear area was compared with cytoplasmic area to test if changes in nuclear size were due to changes in cell size. MDA-231 cells are known to be smaller than MCF10A cells, so it was no surprise the cytoplasmic area of MDA-231 cells was 50% smaller than MCF10A cells (Supplementary Fig. S2C). Importantly, nuclear size was reduced even further than expected, as the nuclear to cytoplasmic ratio of MDA-231 cells was reduced by 25% (0.476 ± 0.022) compared with MCF10A cells (0.631 ± 0.016, Fig. 3I). Expression of GFP-emerin rescued both the cytoplasmic area of MDA-231 cells (Supplementary Fig. S2A) and the nuclear to cytoplasmic ratio (0.578 ± 0.021; Fig. 3I). GFP-M45A (0.456 ± 0.022) failed to rescue the nuclear to cytoplasmic ratio, as expected. GFP-S54F (0.466 ± 0.027) and GFP-M151 (0.480 ± 0.024) also failed to rescue the nuclear to cytoplasmic ratio. Only GFP-M196 (0.573 ± 0.017) rescued the nuclear to cytoplasmic ratio (Fig. 3I) and cytoplasmic area of MDA-231 cells. Thus, although expression of emerin alters cytoplasmic area, its expression has a larger effect on nuclear size. The result of this inappropriate scaling is a much smaller nucleus in the absence of emerin. Expression of GFP-emerin or each GFP emerin mutant in MCF10A cells caused little change in their nuclear area (Fig. 3J).

Recent research has shown variations in chromatin compaction plays an important role in nuclear mechanics for small (1 to 2 μm) deformations, whereas lamins are more important for larger deformations (12). To better understand the effects of emerin expression on heterochromatin formation, the levels of histone posttranslational modifications indicative of heterochromatin were analyzed (Fig. 4A and B). The levels of H3K9me2, H3K9me3, and H3K27me3 were unchanged upon expression of GFP-emerin, GFP-M45A, GFP-S54F, GFP-M151, and GFP-M196 (Fig. 4A and B).

The effect of changes in nuclear area on migration of MDA-231 cells through 3.0 μm pores was studied using transwell assays. A 3.0 μm pore size was chosen because this pore size requires significant nuclear deformation for cells to successfully migrate. Transwell assays were performed on MDA-231 cells stably expressing GFP-emerin or each GFP-emerin mutant. GFP-emerin expression decreased migration by 71 ± 3.2% (Fig. 4C and D). GFP-M45A (10 ± 6.3% decrease), GFP-S54F (4.3 ± 8.6% decrease), and GFP-M151 (16 ± 4.4% decrease) failed to impair migration (Fig. 4C and D). GFP-M24 (66 ± 4.5% decrease) and GFP-M196 (57 ± 2.3% decrease) both impaired migration (Fig. 4C and D). Similar results were seen in transwell inserts with 8.0 μm pores (Supplementary Fig. S4), which is the size routinely used for studying impeded migration of breast cancer cells. Thus, GFP-emerin mutants that rescued nuclear size also impaired migration through small pores.

Cell invasion assays were done using 8.0 μm transwell inserts coated with Matrigel as the extracellular matrix. GFP-emerin expression decreased invasion by 51 ± 4.2% (Fig. 4E and F). GFP-M45A (1.0 ± 6.3% increase), GFP-S54F (39 ± 9.2% increase), and GFP-M151 (13 ± 6.0% increase) failed to impair invasion (Fig. 4E and F). GFP-M196 impaired invasion (56 ± 4.3% decrease). Thus, GFP-emerin mutants that rescued migration also rescued invasion to a similar extent. To ensure these affects were not caused by a general migration failure or an inability to respond to a signaling cue, migration was examined using scratch-wound healing assays. The scratch-wound closed at similar rates in MDA-231 cells and MDA-231 cells expressing GFP-emerin and each GFP-emerin mutant over 2-, 4-, 12-, and 24-hour periods (Fig. 4G and H).

Emerin inhibition of cancer cell migration in vitro was predicted to affect metastasis in vivo. Whether emerin expression impaired metastasis was tested in orthotopic mouse models of breast cancer. MDA-231 cells expressing iRFP713 (to follow implanted cells) and GFP-emerin or each GFP-emerin mutant were injected into the mammary fat pad of nude mice. Mice were imaged every week for 8 weeks. Mice were euthanized after 8 weeks and the tumors and lungs were removed and imaged. There was a 61 ± 12% decrease in primary tumor size and a 50 ± 9% decrease in tumor growth rate of GFP-emerin cells [66,637 fluorescent units (FU), 8,610 FU/week; Fig. 5A and B] compared with MDA-231 cells expressing only iRFP713 (control cells; 171,822 FU, 17,205 FU/week). GFP-M45A (184,376 FU, 22,722 FU/week), GFP-S54F (157,809 FU, 14,806 FU/week), and GFP-M151 (240,951 FU, 26,731 FU/week) failed to rescue primary tumor size and growth rate (Fig. 5A and B). GFP-M196 also significantly decreased primary tumor size and tumor growth rate (93,166 FU, 10,590 FU/week; Fig. 5A and B). Surprisingly, there was no correlation between tumor growth and the proliferation rate of MDA-231 cells or MDA-231 cells expressing GFP-emerin or each GFP-emerin mutant (Supplementary Fig. S3D). Neither cell-cycle dynamics nor apoptosis was affected by expression of emerin or any of the emerin mutants (Supplementary Figs. S3A–S3C).

The volume of excised tumors was measured after 8 weeks. There was an 88 ± 4.0% decrease in the primary tumor volume of GFP-emerin cells (0.26 cm³ ± 0.09 cm³) compared with MDA-231 cells expressing only iRFP713 (2.15 cm³ ± 0.4 cm³, Fig. 5C and D). GFP-M45A (2.08 cm³ ± 0.38 cm³), GFP-S54F (1.49 cm³ ± 0.32 cm³), and GFP-M151 (1.53 cm³ ± 0.32 cm³) failed to rescue primary tumor volume (Fig. 5C and D). GFP-M196 also significantly decreased primary tumor volume (0.31 cm³ ± 0.11 cm³; Fig. 5C and D).

GFP-emerin expression completely blocked lung metastasis (21.7 ± 5.3 FU) as compared with iRFP713 control cells (250 ± 59.8 FU; Fig. 5E and G). GFP-M196 also blocked lung metastasis (32.6 ± 5.9 FU; Fig. 5E and G). GFP-M45A (503 ± 186.2 FU), GFP-S54F (343 ± 98.1 FU), and GFP-M151 (679 ± 233.7 FU) failed to rescue lung metastasis. Only lungs from mice whose primary tumors grew over 50,000 FU were then analyzed to confirm the failure to metastasize in GFP-emerin and GFP-M196 cells was not caused by reduced tumor burden. 50,000 FU was selected because primary tumors >50,000 FU metastasized to the lungs in 100% of control cases. Using this threshold, GFP-emerin (15.4 ± 5.5 FU) and GFP-M196 (34.8 ± 7.3 FU) still eliminated lung metastasis (Fig. 5F). iRFP713 MDA-231 cells had 370 ± 55.2 FU in the lungs using this threshold (Fig. 5F). These demonstrate the function of emerin in metastasis formation is independent of its role in primary tumor growth.

To determine if emerin expression correlates with cancer aggressiveness in patients, emerin protein expression was measured in patient tissues. A significant decrease in emerin expression was seen in 14 different patients with a variety of receptor expression profiles, stages, and metastases (Fig. 6A, B, and D). qPCR analysis of RNA samples matching 10 of the protein samples was done to determine emerin mRNA expression; RNA was unavailable for the other four patient samples. There was no discernable pattern between emerin protein expression and emerin mRNA expression (Fig. 6C and D). Interestingly, there is a correlation between HER2-negative cancers having significantly less emerin when compared with healthy normal controls (Fig. 6D). Emerin expression was similar in multiple normal controls (Supplementary Fig. S5).

In this study, we showed emerin plays a direct and significant role in breast cancer migration and metastasis. The dynamics of cell migration and adhesion are particularly important for the process of metastasis, in which cancer cells escape from the primary tumor and enter the vasculature (intravasation); travel to a distant site; and extravasate to invade a new niche (5). The deformability of the nucleus limits the cell's ability to pass through tight spaces, with decreased compliance reducing or even stalling migration as the pore size decreases (15, 27). Alterations in nuclear structure and mechanics directly contributes to cancer cell progression and metastasis (9, 15, 27, 44).

The importance of the nuclear lamina in providing structural support to the nucleus and controlling nuclear size is well established (9, 19, 27, 44). In addition, several studies have reported altered distribution, expression, or both, of INM proteins in various human cancers (20–26). Many types of cancer, including breast cancer, present with lamina alterations (23). Aberrant localization and reduced expression of A-type lamins frequently correlate with cancer subtypes and cancer aggressiveness, proliferative capacity, and differentiation state (9, 15, 19, 27–29). Previous experiments showed nuclei lacking lamin A/C are significantly more compliant than wild-type controls (4, 45). Nuclei in cells lacking lamins A and C deform significantly more under strain application, whereas lamin B1-deficient cells have normal nuclear mechanics (4). This increased compliance in the absence of lamin A/C and other INM proteins, including emerin, is predicted to increase confined cell migration, cell invasion, and metastasis (Fig. 7). The nuclear lamina also participates in cytoskeletal and nucleoskeleton coupling, so mutations in lamins or other nuclear envelope proteins could alter force transmission between the nucleus and the cytoskeleton to affect nuclear deformation under externally applied loads (9, 46).

In addition to lamins, other nuclear envelope proteins have recently been implicated in a variety of cancers. Emerin expression is altered in osteosarcoma (25), which scales with cancer aggressiveness. Emerin was also recently implicated in prostate cancer aggressiveness, as treatment of prostate cancer cells with shRNA to downregulate emerin increased nuclear deformation and migration (41). The data presented here is the first to show emerin expression is decreased in invasive breast cancer cells and that increasing emerin expression rescued their nuclear size and structure and decreased their migration, invasion, and metastasis in mice.

There is ample evidence for emerin maintaining nuclear architecture. These include biophysical studies showing emerin-null cells have decreased elasticity (47) and the nuclear envelope is more compliant (47, 48). These cells also have altered nuclear morphology, increased nuclear deformability, impaired viability under mechanical strain, and defective mechanotransduction (45, 48). Emerin binds actin and caps the pointed end of actin filaments to stabilize these short nuclear actin polymers (33). Emerin co-purifies with nuclear protein 4.1R, which binds both nuclear spectrin and nuclear actin (49). Thus, emerin is predicted to anchor cortical nuclear actin networks near the nuclear envelope to increase nuclear rigidity. Our results support the role of emerin stabilization of the nucleoskeleton in maintaining proper nuclear rigidity, because disruption of emerin's interactions with the nucleoskeleton results in cells with smaller, deformed nuclei that migrate and invade more easily and metastasize more readily.

These nuclear size and structural changes may also result from disruption of its interactions with the linker of nucleoskeleton and cytoskeleton (LINC) complex. The LINC complex physically connects the cytoskeleton to the nucleus (50). LINC complex proteins and lamin A/C directly transmit mechanical force from the plasma membrane and cytoplasm to the nucleus (48, 50). Application of force to the plasma membrane or cytoskeleton deforms the nucleus and activates mechanosensitive genes (51). Emerin and lamin A directly bind nesprins and SUN-domain proteins (52), which likely serves to relay the cytoskeletal signals to the nucleoskeleton and chromatin. Thus, nucleoskeleton organization and the structure of the LINC complex are key factors in the transmission of mechanical stress to the nucleus (53). As wild-type nuclei undergo mechanical force, the nuclear envelope adapts by increasing in stiffness. Force strongly induces the phosphorylation of emerin, mediated by the SRC family kinases (SFK) (53). Interestingly, emerin depletion prevented nuclear adaptation to force. Nuclei expressing emerin mutants that prevented phosphorylation, failed to adapt to force (53). These results indicate application of force to nesprin 1 activates SFKs, which phosphorylate emerin to mediate the mechanical adaptation of isolated nuclei (53).

Besides the nuclear lamina, chromatin is an important contributor to nuclear stiffness (9, 11, 12, 15, 27, 54). Unlike the loadbearing, elastic shell of the nuclear lamina, chromatin exhibits a more viscoelastic behavior (i.e., it flows when subjected to forces and undergoes plastic deformations) (14). Chromatin exists in two configurations: euchromatin, which is less compact and typically transcriptionally active, and heterochromatin, which is more tightly compacted and associated with transcriptionally silent genes (55). A large portion of heterochromatin interacts with the nuclear lamina at the nuclear periphery (55). Treatment of mouse embryonic fibroblasts (MEF) with a histone methyltransferases inhibitor (HMTi) caused ∼20% larger nuclei with increased compliance and increased blebbing (12). These increases were attributed to decreased heterochromatin formation, as H3K9me2/3 and H3K27me3 levels were decreased. However, we showed emerin-null myoblasts have decreased H3K9me2/3 and H3K27me3 and increased H3K9ac and H4K5ac (32), yet have smaller nuclei (Supplementary Fig. S1). These suggest heterochromatic marks may affect nuclear size and structure differently in a cell-type dependent manner. The results reported here show there is no change in heterochromatin formation upon expression of GFP-emerin or GFP-emerin mutants, as defined by H3K9me2/3 and H3K27me3 levels (Fig. 4A and B). The reduction of emerin by 50% in invasive breast cancer cells, as opposed to its complete loss, may also explain this apparent inconsistency. It is also possible that the many mutations accompanying the transformation to invasive breast cancer cells may have affected chromatin architecture independent of emerin or other nuclear envelope proteins (56). It is also possible the functional interaction of emerin with HDAC3 may affect chromatin compaction independent of histone methylation. Regardless of the mechanism, our data show chromatin minimally contributes to the decreased nuclear size and increased compliance in invasive breast cancer cells. We propose the functional interaction of emerin with the nucleoskeleton is the major contributor to the nuclear structural changes seen in invasive breast cancer cells.

Collectively, our data show the functional interaction of emerin with the nucleoskeleton is important for the nuclear structural changes seen in invasive breast cancer cells. These smaller, softer nuclei are predicted to migrate more easily and to metastasize more readily (Fig. 7).

Loss of emerin protein expression during metastatic transformation would have similar effects to the loss of nucleoskeletal proteins, including lamins (1, 6, 9, 15, 27). Thus, we predict loss of emerin expression is an important driver of metastatic transformation.

J.M. Holaska reports grants from NIH and W.W. Smith Charitable Trust during the conduct of the study. No disclosures were reported by the other authors.

A.G. Liddane: Data curation, formal analysis, validation, investigation, visualization, methodology, writing–original draft, writing–review and editing. C.A. McNamara: Data curation, formal analysis, investigation. M.C. Campbell: Formal analysis, investigation. I. Mercier: Conceptualization, resources, supervision, validation, investigation, methodology. J.M. Holaska: Conceptualization, resources, data curation, formal analysis, supervision, funding acquisition, validation, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing.

We thank the Department of Pharmaceutical Sciences at the University of the Sciences and the Department of Biomedical Sciences at Cooper Medical School of Rowan University for providing laboratory space and funding. This work was supported by a predoctoral grant from the New Jersey Commission on Cancer Research (to A.G. Liddane), a grant from the W.W. Smith Charitable Trust (to J.M. Holaska), and a grant from the National Institute of Arthritis, and Musculoskeletal and Skin Diseases (R15AR069935 to J.M. Holaska). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health, the W.W. Smith Charitable Trust or the New Jersey Commission on Cancer Research. This work was also supported by Rowan University under the Camden Health Research Initiative. We thank Chris Janetopoulos, Darren Boehning, and Goutham Kodakandla for their helpful insights with volumetric analysis. We thank the members of the Holaska laboratory for the many helpful discussions regarding these studies and preparation of the manuscript.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.