Mswi/SNF Interacts with the Ribosome and Its Inhibition/Mutations Alter Translation and Sensitize to Mtor/PI3K Inhibitors

bioRxiv preprint doi: https://doi.org/10.1101/2021.05.10.443459; this version posted May 10, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. mSWI/SNF interacts with the ribosome and its inhibition/mutations alter translation and sensitize to mTOR/PI3K inhibitors

Livia Ulicna1, Samuel C. Kimmey1,2, Christopher M. Weber1, Grace M. Allard1, Sean C. Bedall1, Gerald R. Crabtree1,2, Gregory R. Bean1, and Capucine Van Rechem1,3,*

1 Department of Pathology, Stanford University, Stanford, CA 94305, USA 2 Department of Developmental Biology, Stanford University, Stanford, CA 94305, USA 3 Lead Contact *Correspondence: [email protected]

Conflict of Interest G.R.C. if a founder and stockholder of Foghorn Therapeutics. Other authors declare no potential conflicts of interest.

Running Title mSWI/SNF regulates translation

Abstract

The chromatin remodelers mammalian SWItch/Sucrose Non-Fermentable (mSWI/SNF) subunits are mutated, deleted or amplified in more than 40% of cancers. Understanding their functions in normal cells and the consequences of cancer’s alterations will lead to path toward new targeted therapies. Canonically, mSWI/SNF complexes regulate the structure of chromatin, however they likely have additional functions which could be relevant in carcinogenesis. Here, we highlight the substantial alteration of mSWI/SNF subunits expression in both the nucleus and cytoplasm in breast cancer cases. We demonstrate mSWI/SNF cytoplasmic localization and interaction with the translation initiation machinery. Short-term inhibition and depletion of specific subunits alter protein synthesis, implicating a direct role for these factors in translation. Inhibition and depletion of specific subunits increase sensitivity to mTOR-PI3K inhibitors, suggesting a potential therapeutic opportunity for diseases harboring mutations in these complexes. Indeed, SMARCA4 pathogenic mutations decrease protein synthesis. Furthermore, taking advantage of the DepMap studies, we demonstrate cancer cells harboring mutations of specific mSWI/SNF subunits exhibit a genetic dependency on translation factors and are particularly sensitive to translation pathway inhibitors. In conclusion, we report an unexpected cytoplasmic role for mSWI/SNF in protein synthesis, suggesting potential new therapeutic opportunities for patients afflicted by cancers demonstrating alterations in its subunits.

Statement of significance

This study establishes direct functions for mSWI/SNF in protein synthesis. mSWI/SNF inhibition, depletion and cancer mutations alter translation and increase sensitivity to translation pathway inhibitors, illustrating the potential for new therapeutic strategies.

Introduction

The mammalian SWI/SNF (mSWI/SNF, SWItch/Sucrose Non-Fermentable) complexes are evolutionary conserved ATP-dependent chromatin remodelers that modify chromatin architecture and thus influence gene expression (1-3). In human, there are three main complexes: BRG1/BRM Associated Factors (BAF), Polybromo- associated BAF (PBAF), and non-canonical BAF (ncBAF) (4). These complexes are composed of both common and specific subunits which are altered in more than 40% of cancers (5). Indeed, genes coding for mSWI/SNF are not only mutated in 20% (1-3), but also exhibit copy number variation in 30% of cancers (5). Many mutations result in loss-of-function and have been shown to not only drive disease progression but also facilitate resistance to therapy (1-3, 6). Alteration rates of specific subunits are disease-specific, suggesting context-dependent functions (1-3, 5).

1 bioRxiv preprint doi: https://doi.org/10.1101/2021.05.10.443459; this version posted May 10, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

While mSWI/SNF functions are mostly studied in the nucleus in the context of nucleosome positioning through direct interaction between mSWI/SNF and the nucleosome, few studies investigate the possibility of non- chromatin substrates for these complexes. In yeast, SWI/SNF evicts Silent information regulator Sir3p from the nucleosome, requiring a direct interaction between these proteins (7), and the catalytic subunit Brg1 (SMARCA4) activates Mec1 kinase by direct protein-protein interaction (8). In humans, BAF directly evicts Polycomb Repressive Complex 1 (PRC1) from the chromatin (9). Furthermore, mSWI/SNF subunits are not only present in the cell nucleus but can also shuttle between nucleus and cytoplasm (10, 11). Differential localization could be relevant in disease. For example, SMARCA4 is cytoplasmic in a subset of corticotroph adenomas (12), SMARCA2 nuclear localization correlates with better prognosis in lung cancer (13), and cytoplasmic ARID1B promotes oncogenesis and correlates with advanced pancreatic cancer (14). Together, these studies provide evidence for possible disease-associated functions of mSWI/SNF complexes outside of their direct roles on the nucleosome. In this study, analysis of breast cancer patient tissues reveals a substantial alteration of nuclear and cytoplasmic expression of the ATPases SMARCA2-4 and the PBAF-specific subunits PB1, BRD7 and ARID2. We further demonstrate that cytoplasmic mSWI/SNF subunits interact with the translation machinery and affect protein synthesis. mSWI/SNF inhibition and depletion in combination with mTOR-PI3K inhibitors decrease cell viability, suggesting a possibility for protein synthesis targeted therapy in mSWI/SNF-perturbed diseases. Indeed, expression of SMARCA4 loss-of-function pathogenic mutations detected in cancers decrease translation. Finally, analysis of the DepMap database unveiled mSWI/SNF-mutated cancer cell lines’ genetic dependency on translation factors and sensitivity to translation pathway inhibitors. Together, our data support a direct role for mSWI/SNF subunits in protein synthesis with the potential to develop into new therapeutic strategies.

Results

Breast cancer specimens predominantly gain SMARCA4 and lose SMARCA2 expression Several mSWI/SNF subunits are observed in the cytoplasm in tumors and altered cellular localization could be useful markers (12-14). While mSWI/SNF subunits are primarily considered tumor suppressors in association with cancer, evidence has begun to shed light on their potential oncogenic roles. For example, SMARCA4 is unlikely to be a tumor suppressor in breast cancer (15). Because mis-expression or localization is not necessarily associated with genomic alteration, we evaluated mSWI/SNF subunits in breast cancer cases by immunohistochemistry. We utilized a tissue microarray composed of 46 cases including primary tumors and nodal and distant metastases. We first focused on the mutually exclusive ATPases. SMARCA4 was very minimally expressed in normal breast epithelium, with occasional cells being nuclear positive (Fig. 1A and Supplementary Fig. S1A). Hence, we only observed gain of SMARCA4 expression in breast cancer. Remarkably, 51% of cases presented a gain in nuclear and cytoplasmic SMARCA4 and 18% a weak cytoplasmic staining that could be considered as background (a potential limitation of IHC). Notably, five cases (11%) harbored a SMARCA4 mutation, three of which were associated with gain-of-expression by IHC (Supplementary Fig. S1B). SMARCA2 exhibited moderate nuclear expression in ~50% of normal breast epithelial cells (Fig. 1B and Supplementary Fig. S1C). 67% of cases demonstrated loss of nuclear SMARCA2 and 9% concomitantly gained cytoplasmic expression. Furthermore, 11% of cases showed increased nuclear and cytoplasmic SMARCA2 staining, and 7% exhibited gain of exclusive cytoplasmic staining. Of note, some cases presented both gains and losses of SMARCA2 expression within different samples. Taken together, SMARCA2 is the primary mSWI/SNF ATPase expressed in normal breast and is predominantly lost in breast cancers. On the contrary, SMARCA4 is relatively overexpressed. Increased cytoplasmic staining of both ATPases was observed in 17% of cases. Gain of SMARCA4 with concurrent loss of SMARCA2 was predominant, with 66% of cases demonstrating a pattern of mSWI/SNF complexes supportive of a potential swap involving one ATPase for the other.

PBAF-specific subunits expression is highly altered in breast cancers Mutations in the PBAF-specific subunits ARID2 and PB1 are reportedly less frequent in breast cancers than the BAF-specific ARID1A and ARID1B (2, 3). Therefore, we decided to further investigate expression levels of PBAF subunits. 2 bioRxiv preprint doi: https://doi.org/10.1101/2021.05.10.443459; this version posted May 10, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Figure 1 – mSWI/SNF subunits expression and subcellular localization are altered in breast cancers. Nuc A Normal Breast Cyto & Nuc Cyto ‘Background’ Nuc SMARCA4

51% 18% 22%

B Normal Breast SMARCA2 Cyto Nuc Nuc Cyto & Nuc SMARCA2

67% 9% 9% 11%

Cyt o C SMARCA2 SMARCA4 SMARCA2 & SMARCA4 Sample 1 Sample 2 Sample 3 Sample 4 SMARCA2 SMARCA4

66% 14% D No rma l B re a st Cyto & Nuc Nuc PB1

41% 30% Cyto & Nuc E Normal Breast Cyto BRD7

68% 16% F Normal Breast Cyto Cyto Cyto & Nuc ARID2

12% 26% 11% Figure 1. Legend on next page.

3 Figure 1 – mSWI/SNF subunits expression and subcellular localization are altered in breast cancers. Representative examples of immunohistochemistry from breast cancer cases for SMARCA4 (A), SMARCA2 (B), identical tumors from the same patients for SMARCA2 and SMARCA4 (C), PB1 (D), BRD7 (E) and ARID2 (F). 20X images. Cyto = cytoplasmic, Nuc = nuclear. PB1 presented various levels of exclusive nuclear staining in normal breast epithelium (Fig. 1D and Supplementary Fig. S2A). Within our breast cancer cases, while 30% demonstrated a loss in nuclear staining, 41% showed substantial cytoplasmic expression and 22% exhibited weak cytoplasmic staining. In normal breast, all cells showed cytoplasmic BRD7 with weak nuclear staining in less than 50% of the cells (Fig. 1E and Supplementary Fig. S2B). In breast cancer samples, BRD7 was prominently overexpressed: 68% of cases exhibited increased nuclear and cytoplasmic staining and 16% exhibited exclusively cytoplasmic gain. Finally, ARID2 resembled BRD7 in normal breast, being predominantly cytoplasmic with few cells exhibiting a weak nuclear staining (Fig. 1F and Supplementary Fig. S2C). In breast cancer cases, 31% demonstrated gain in nuclear ARID2 expression, 26% a gain of exclusively cytoplasmic staining, 11% a gain in both compartments and 12% a loss of cytoplasmic ARID2. All considered, while a substantial number of cases presented loss of expression, PBAF-specific subunits were primarily overexpressed in breast cancer cases. Of note, some patients’ tumors showed both gains and losses and/orFigure variation 2 – mSWI in/SNF compartmental subunits localize distribution in the cytoplasm across and different interact samples, with the translation highlighting machinery. the complexity of tumor and cancer evolution observed in primary, recurrent and metastatic samples.

A B INIT ELONG C INITIATION ELONGATION Cy Nu Ch 40S 60S80S EP LP 40S 60S 80S EP LP SMARCA2 SMARCA2 SMARCA4 SMARCA4 SMARCC1 SMARCC1 SMARCC2 SMARCC2 Common subunits ARID2 Common subunits ARID2

PHF10 PHF10 260nm Absorbance Top Bottom PB1 PB1 PBAF PBAF IP BRD7 BRD7 D SMARCB1 SMARCB1

ARID1A ARID1A Input (2%) RPS3 IgG SMARCA4 ARID1B ARID1B BAF

DPF2 BAF RPS3 DPF2 SS18 IP -RNAse IP +RNAse SS18 BRD9 ncBAF GLTSCR1 ncBAF BRD9 EIF4A1

LDH Input (2%) IgG PABP SMARCB1 IgG PABP SMARCB1 PABP SMARCB1 Histones (Ponceau) RPS6 PABP

Figure 2 – mSWI/SNF subunits localize in the cytoplasm and interact with the translation machinery. A. Immunoblot from fractionated HEK293T cells (Cy: cytoplasm, Nu: nucleus, Ch: chromatin). B. Immunoblots from HEK293T polysome profiles fractions (INIT: initiation, ELONG: elongation, EP: early polysomes, LP: late polysomes, as represented in C). C. Representative polysome profile from HEK293T cells. D. Coimmunoprecipitation from HEK293T cells. +RNAse: RNAse A was added in the lysates prior to immunoprecipitation. Cytoplasmic mSWI/SNF subunits interact with the translation machinery Because of the significant population of mSWI/SNF subunits in the cytoplasm in normal and breast cancer tissues, we further assessed their subcellular localization in cells harboring intact mSWI/SNF complexes (4). Biochemical fractionation from HEK293T revealed the substantial presence of BAF, PBAF and ncBAF subunits in the cytoplasmic compartment (Fig. 2A). Indeed, specific subunits are distributed equally in cytoplasmic and

4 nuclear compartments (SMARCC1 and SMARCC2) and several are enriched in the cytoplasm compared to nucleus and chromatin combined (ARID2, PHF10, BRD7, ARID1A, ARID1B, DPF2 and BRD9). Based on the previously published role for the lysine demethylase KDM4A in the cytoplasm (16), we reasoned that cytoplasmic mSWI/SNF might be involved in protein synthesis. We performed polysome profiling and detected proteins present in collected fractions by western blot. All tested mSWI/SNF subunits were enriched in the initiating fractions (Fig. 2B-C and Supplementary Fig. S3). We further confirmed mSWI/SNF interaction with translation factors by coimmunoprecipitations (Fig. 2D). Treatment of extracts with RNAse A indicate that RNAs are likely not necessary for such interactions (Fig. 2D). Together, these data demonstrate mSWI/SNF’s interaction with the translation machinery and suggest that these complexes could directly impact translation. mSWI/SNF inhibition alters global protein synthesis In order to investigate mSWI/SNF involvement in protein synthesis, we took advantage of available small molecule bromodomain inhibitors targeting BRD9 (iBRD9), BRD9 and BRD7 (TP472), and SMARCA2, SMARCA4 and PB1 (PFI-3) (Fig. 3A). We treated HEK293T cells with these inhibitors for 24 hours and assessed protein synthesis for the last 30 minutes with puromycin or hour with L-azidohomoalanine (AHA). These assays revealed a substantial decrease of total protein synthesis upon each drug treatment (Fig. 3B-C). We then wanted to decipher if the translational defects observed upon mSWI/SNF inhibition could be a consequence of direct roles in translation, as opposed to their nuclear roles on transcription. We shortened mSWI/SNF inhibitor treatments to 10 minutes before assessing de novo protein synthesis activity to capture any changes in translation before transcriptional changes would be observed at the protein level, as RNA transcription, export and translation take at least 20 minutes for small proteins (17, 18). Puromycin incorporation is reduced upon 10 minutes of inhibition to a similar extent as at 24 hours (Fig. 3D). To confirm these results at a single cell level, we performed Simultaneous Overview of tri-Molecule Biosynthesis (SOM3B) (19). This method allows for the simultaneous measurement of puromycin and 5-BromoUridine incorporation (BrU, assessing transcription) in addition to viability and other cellular features at a single-cell level. SOM3B further confirmed the observed decrease in protein synthesis in live cells upon 10 minutes of mSWI/SNF inhibition (Supplementary Fig. S4A-B). While we did not detect any changes in transcription (data not shown), 10 minutes is a short timeframe for BrU incorporation and as a consequence the detected signal was not reliable. Protein synthesis remains active during all interphase stages of the cell cycle and is reduced during mitosis (20). To identify if the observed inhibition is consistent as cells transit though the cell cycle, we assessed IdU incorporation (S phase) and expression of Cyclin B1 and phosphorylation of histone H3 on Serine 10 to assign each cell to its cell cycle phase (19). mSWI/SNF inhibition reduced puromycin incorporation across all stages of interphase and altered translation very modestly in mitotic cells (Supplementary Fig. S4C-D). We then wanted to assess translational defects by polysome profiling assays. While after 24 hours of treatment each drug increased the levels of the 80S monosome and decreased the amount of polysomes, typical of a translation initiation defect, 10 minutes of mSWI/SNF inhibition led to increased polysomes, which could be reflective of ribosome collision (Fig. 3E and Supplementary Fig. S4E-H). Together, these data demonstrate that mSWI/SNF inhibition decreases global protein synthesis and that these defects are consistent with a direct function for mSWI/SNF subunits in translation.

Depletion of specific mSWI/SNF subunits alters translation To rule out any potential off-target effect of mSWI/SNF bromodomain inhibitors, we confirmed our findings using engineered mouse embryonic stem cells containing an auxin-dependent degron within Smarca4, Arid1A or Pb1 alleles, allowing a complete degradation of endogenous protein upon auxin treatment (21). Addition of auxin for 2 hours inhibits de novo protein synthesis for Smarca4 and Arid1A but not Pb1 (Fig. 2E), highlighting the involvement of specific subunits or complexes in protein synthesis. Depletion of Smarca4 decreases protein synthesis by 50% but equally up- and downregulates expression of specific RNAs (Fig. 3F and (21)), validating potential independent roles for mSWI/SNF in translation in non-transformed cells. In order to assess more mSWI/SNF subunits, we detected steady-state translational levels in HAP1 cells depleted for specific PBAF (BRD7, ARID2), BAF (ARID1A) and ncBAF (BRD9) subunits, as well as for the two ATPases (SMARCA2 and SMARCA4) (Supplementary Fig. S4I). mSWI/SNF subunits’ subcellular localization in the HAP1 wild-type (WT) cells are comparable to the localization observed in HEK293T cells (Supplementary Fig. S4J). Furthermore, treatment of HAP1-WT with iBRD9 and TP472 for 10 minutes decreased overall protein 5 Figure 3 – mSWI/SNF inhibition/depletion inhibits protein synthesis.

3 3 A B - C -

DMSO iBRD9 TP472 PFI DMSO iBRD9 TP472 PFI

BROMODOMAIN

TARGETS HRP INHIBITORS -

iBRD9 BRD9 Puromycin Strep

TP472 BRD7, BRD9 incorporation) (AHA SMARCA2-4, PFI-3 Actin Actin PB1 1 1 * 0.8 * 0.8 * * 0.6 * 0.6 * 3 - 0.4 0.4 D 0.2 0.2 DMSO iBRD9 TP472 PFI 0 0

Polysomes E 40S 60S 80S TP472 10min

Puromycin 1.2

Actin 0.8 1 * 0.4 0.8 * TP472 0.6 * 260nm Absorbance 0.4 24h 0.2 Top Bottom 0 F G H Target: Smarca4 Arid1A Pb1 KO: WT SMARCA4

AUX - - + - + - + KO: WT SMARCA4 SMARCA2 ARID2 BRD7 ARID1A BRD9 Construct: Ct A4 Ct A4 - + + + + + + DOX GFP FLAG SMARCA4 Target Actin Actin Puromycin HRP - Strep Puromycin (AHA incorporation) (AHA Actin

Actin 1 Actin * 1 0.8 # 1 0.8 * * 0.6 * 0.8 0.6 * 0.4 * * 0.6 * 0.4 0.4 0.2 0.2 0.2 0 0 0 Figure 3. Legend on next page. 6 Figure 3 – mSWI/SNF inhibition/depletion inhibits protein synthesis. A. mSWI/SNF bromodomain inhibitors and targets. B. Puromycin incorporation assays in HEK293T cells treated with 10µg/ml iBRD9, 5µM PFI-3, and 5µM TP472 for 24 hours. Puromycin was added for the last 30 minutes. Lower panel: quantification of three independent experiments. C. L-azidohomoalanine AHA incorporation assays in HEK293T cells treated as in A for 24 hours. AHA was added for the last hour. Lower panel: quantification of three independent experiments. D. Puromycin incorporation assays in HEK293T cells treated as in A for 10 minutes. Lower panel: quantification of three independent experiments. E. Representative polysome profiles from HEK293T cells treated with 5µM TP472 for 24 hours (dotted brown) and 10 minutes (brown). F. AHA incorporation in ES cells after 2 hours of auxin (AUX) inducible degradation of Smarca4, Arid1A and Pb1. Doxycycline (DOX) induces the expression of the Fbox degrading endogenous subunits tagged with FLAG and a degron interacting with the Fbox in the presence of auxin. Lower panel: quantification of three independent experiments. G. Puromycin incorporation assays in HAP1-WT and knock-out cells. Lower panel: quantification of three independent experiments. H. Puromycin incorporation assays in HAP1-WT and HAP knock-out for SMARCA4 after transient transfection of GFP-SMARCA4 or a control vector. Lower panel: quantification of three independent experiments. * two-tailed students’ t-test, p < 0.05.

synthesis (Supplementary Fig. S4K). Puromycin integration assays revealed that SMARCA4, ARID2, BRD7 or ARID1A-depleted cells displayed a 50-70% decrease in translation, while SMARCA2 or BRD9-depleted cells were less affected (Fig. 3G). In order to assess the type of alterations our mSWI/SNF subunits-depleted cells could exhibit, we performed polysome profiling assays. While each HAP1 knockout cell line demonstrated altered polysome profiles compared to WT cells, these profiles were distinct. SMARCA4, ARID2 or BRD7-depleted cells showed identical profiles with an increase in initiating and early polysomes and decrease in late polysomes (Supplementary Fig. S5A-C); SMARCA2 or ARID1A-depleted cells exhibited a decrease across the entire profile (Supplementary Fig. S5D-E); and BRD9-depleted cells, while being more variable, demonstrated a consistent increase in polysomes (Supplementary Fig. S5F-G). These distinct polysome profile alterations upon depletion of mSWI/SNF subunits from different complexes are consistent with a specificity of each complex in translation. As described in the nucleus, these results suggest different roles for BAF, PBAF and ncBAF in the cytoplasm. Finally, to demonstrate that altered translation is specific to the lack of SMARCA4 in SMARCA4-depleted cells we transiently transfected GFP-SMARCA4 in these cells. While the overexpression of SMARCA4 in HAP1-WT cells did not change levels of translation, expression in SMARCA4-depleted cells rescued the defects observed upon knock-out (Fig. 3H). Together, these results reveal that depletion of specific mSWI/SNF subunits decreases overall translation and that such roles are conserved in mice. mSWI/SNF inhibition and depletion sensitize cells to mTOR-PI3K inhibitors Because mSWI/SNF inhibition decreases protein synthesis (Fig. 3B-D), we hypothesized that mSWI/SNF inhibitors could increase cells’ sensitivity to translation inhibitors. To assess this possibility, we performed cell viability assays using AZD8055 and BEZ235, inhibiting the kinases mTOR and mTOR-PI3K, respectively, in combination with mSWI/SNF bromodomain inhibitors. While 72 hours of iBRD9, TP472 or PFI-3 treatment alone decrease HEK293T viability, iBRD9 and TP472 decreased further the viability of cells co-treated with AZD8055 or BEZ235 (Fig. 4A-B). Based on these results we further hypothesized that the depletion of mSWI/SNF subunits would sensitize cells to translation inhibition. Cell viability assays in knockout HAP1 cells revealed that the depletion of ARID2 and BRD7 increased sensitivity to a range of concentrations of AZD8055 and BEZ235 (Fig. 4C-D), while depletion of SMARCA2, SMARCA4, ARID1A or BRD9 did not (Supplementary Fig. S6A-D). Together, these data demonstrate that mSWI/SNF inhibition and depletion of specific subunits creates a vulnerability to mTOR-PI3K inhibition in a variety of cell types.

SMARCA4 pathogenic mutations decrease protein synthesis mSWI/SNF subunits are frequently mutated in cancers and most of these mutations are considered loss-of- function (1-3). Furthermore, some heterozygous missense mutations act in a dominant-negative fashion (22). Therefore, we hypothesized that missense mutations of mSWI/SNF subunits could alter translation. To test this hypothesis, we took advantage of two well characterized SMARCA4 pathogenic mutations: p.K785R, located in the ATP cleft and acting as a catalytic mutant, and p.E861K, located in the DNA groove and acting to disrupt DNA binding (22). We transiently overexpressed these mutations in HAP1-WT cells and assessed protein synthesis by puromycin integration assays. Overexpression of either mutation decreased overall translation (Fig. 4E). Therefore, these alterations act as dominant-negative mutants and SMARCA4’s catalytic activity and DNA groove are both required for normal translation.

7 Figure 4 – mSWI/SNF inhibition, depletion or pathogenic mutation sensitizes cells to translation pathway inhibitors.

1.2

A B C 1 1 1 1 * * 0.8 0.8 * 0.8 0.8 0.6 0.6 * * # # 0.6 0.6 * * 0.4 0.4 * # * Viability Cell WT BRD7 ARID2 * 0.2 * # 0.2 0 25 50 100 200 400 0.4 * 0.4 0 25 50 100 200 400

Cell Viability Cell Viability Cell AZD8055 (nM) 0.2 0.2 D 1.2 1 DMSO + + + - + - + - DMSO + + + - + - + - 1 0.8 AZD8055 - + - + - + - + BEZ235 - + - + - + - + 0.8 * 0.6 - - + + ------+ + - - - - 0.6 * iBRD9 iBRD9 0.4 *

TP472 - - - - + + - - TP472 - - - - + + - - 0.4 0.2 * Cell Viability Cell WT BRD7 ARID2 0.2 0 ------+ + ------+ + 0 25 50 100 200 400 PFI-3 PFI-3 0 25 50 100 200 400 BEZ235 (nM)

SMARCA4 SMARCA4 MUTANTS SMARCA2 MUTANTS E F G 10 12 9 785R Ct WT E861K K

) 8 RPL22L1 10 ) GFP 7 DDX27 8 6 PValue SMARCA4 PValue 6 RPL22L1 5 Actin 4 log10( log10( -

4 - 3 2 2 1 -0.2 -0.15 -0.1 -0.05 0 -0.3 -0.25 -0.2 -0.15 -0.1 -0.05 0 Effect size Effect size ARID1A MUTANTS ARID1B MUTANTS H I 7 14 DDX27

Puromycin 12 6 RPL22L1 ) ) 10 RPL22L1 5

8 4 SRSF1 PValue PValue DDX27 Actin 6 3 DHX9

SRSF1 log10( log10( -

- 4 1 DHX9 2 0.8 0.6 * 2 1 0.4 * 0.2 -0.25 -0.2 -0.15 -0.1 -0.05 0 -0.2 -0.15 -0.1 -0.05 0 0 Effect size Effect size ARID1A MUTANTS 8 ARID1B MUTANTS J 7 AZD5363 K AZD5363 GSK2110183 7 GDC-0068 6 CCT128930 GSK2110183 GDC-0068 6 ) 5 ) doxycycline 5 CCT128930 4 PValue PValue 4 AZD8055 Palomid-529 AT7867 3 3 cycloheximide log10( log10( - GDC-0980 - 2 2 PP242 AT7867 1 1 buparlisib

-0.4 -0.3 -0.2 -0.1 0 -0.4 -0.3 -0.2 -0.1 0 Effect size Effect size Figure 4. Legend on next page. 8 Figure 4 – mSWI/SNF inhibition, depletion or pathogenic mutation sensitizes cells to translation pathway inhibitors. A-B. Cell viability assays in HEK293T cells treated with 10µg/ml BRD9, 5µM PFI-3, 5µM TP472, and 50nM AZD8055 (mTOR inhibitor, A) or 50nM BEZ235 (dual mTOR-PI3K inhibitor, B) for 72h. Averages of eight independent experiments are represented. * compared to its respective mSWI/SNF inhibitor alone, # compared to AZD8055 or BEZ235. C-D. Cell viability assays in HAP1-WT and PBAF subunits knockout cells treated with 25-400nM AZD8055 (C) or 25-400nM BEZ235 (B) for 72h. Averages of six (AZD8055) and five (BEZ235) independent experiments are represented. * compared to WT. E. Puromycin incorporation assays in HAP1-WT cells transiently overexpressing GFP- SMARCA4 ATPase domain mutations. Lower panel: quantification of three independent experiments. F-I. Genetic dependency of SMARCA4 (F), SMARCA2 (G), ARID1A (H) and ARID1B (I) mutated cancer cells compared to all other cancer cells in the DepMap database (combined RNAi dataset). J-K. Drug sensitivity of ARID1A (J) and ARID1B (K) mutated cancer cells compared to all other cancer cells in the DepMap database (PRISM repurposing primary screen 19Q4 dataset). Orange dot: most prominent translation-related gene/drug. Blue dots: q value <= 0.05. * and # two-tailed students’ t-test, p < 0.05.

Cancer cells harboring mSWI/SNF mutations exhibit a genetic dependency on translation factors We further hypothesized that mSWI/SNF mutated cells could be dependent on translation. In order to test this hypothesis, we took advantage of the genetic dependency study of the DepMap database reporting RNAi against 17,309 genes in 712 cell lines (23). Fourteen mSWI/SNF subunit pathogenic mutations were dependent on one or more translation factors when compared to all other cell lines in the dataset (Fig. 4F-I and Supplementary Fig. S6E-N). In accordance with decreased translation associated with SMARCA4 mutations (Fig. 4E), SMARCA4- mutated cancer cells demonstrated a genetic dependency on the ribosomal subunit RPL22L1 and the RNA helicase DDX27 (Fig. 4F). SMARCA2-mutated cancer cells also exhibited a dependency on RPL22L1 (Fig. 4G), as well as ARID1A and ARID1B-mutated cancer cells (Fig. 4H-I). Cancer cells harboring mutations in either of these two BAF subunits also showed dependency on the RNA helicases DDX27, DDX9 and SRSF1, a splicing factor with a direct role in translation (Fig. 4H-I). mSWI/SNF mutations sensitize cancer cells to translation pathway inhibitors The specific genetic dependency of SMARCA2, SMARCA4, ARID1A and ARID1B-mutant cancer cell lines led us to speculate that these mutations could sensitize cancer cells to translation inhibitors. We analyzed the PRISM repurposing primary drug sensitivity screen in the DepMap database testing 578 cell lines with 4,686 compounds (24). Cancer cells mutated for either ATPase did not present any strong sensitivity to specific inhibitors when compared to all other cells in the database (q value < 0.05) but did sensitize cells to some translation pathways inhibitors (Supplementary Fig. S7A-B). However, cancer cells harboring ARID1A or ARID1B mutations were primarily more sensitive to inhibitors targeting the translation pathway: AKT, mTOR and PI3K inhibitors, as well as drugs directly affecting translation such as doxycycline and cycloheximide (Fig. 4J-K).

Discussion

In the present study we demonstrate the substantial cytoplasmic presence of the mSWI/SNF subunits and their implications in protein synthesis. Decreased translation upon mSWI/SNF bromodomain inhibitor treatment suggests that these domains are necessary for mSWI/SNF targeting to the translation machinery. This is a strong possibility as bromodomains recognize acetylation, and the ribosome is highly acetylated. While both short and long-term mSWI/SNF inhibition leads to decreased translation to the same extent, polysome profile alterations are vastly different. These differences could reflect the mechanisms behind such inhibition: ribosome collision stalls elongation, leading to increased polysome density before triggering a global translational dysregulation by inhibiting initiation (25). Future studies will need to decipher if and how defects in mSWI/SNF subunits, which are extensively present within the initiating fractions, lead to global ribosome collisions. While the knockout of each tested subunit decreased translation, cell sensitivity to mTOR-PI3K inhibition was increased only upon depletion of PBAF subunits. This could reflect specific roles in translation for different mSWI/SNF complexes. Indeed, a global role in translation does not preclude an additional role on specific mRNAs. For example, upon stress, the translation initiation factor eIF2a reduces global protein synthesis and preferentially translates specific mRNAs (26). Confirming potential specific roles, PBAF-depleted cell lines presented identical alterations in polysome profiles. Of note, polysome profiles from SMARCA4 knockout resembled PBAF depletion and SMARCA2 knockout resembled BAF (ARID1A); it will be interesting to further determine if cytoplasmic complexes would involve specific ATPases, in contrast to the nucleus where SMARCA2 and SMARCA4 are not complex-specific (4). Analyses of the DepMap database revealed the genetic dependency of mSWI/SNF-mutated cancers and genetic dependency to translation factors and their specific sensitivity to translation pathway inhibitors. This could be 9 used as a clinical strategy, such as PARP inhibitors for BRCA-deficient cancers (27). However, some subunits such as BRD9 did not reveal such dependency even though iBRD9 did decrease translation. For such subunits, only specific mutations such as in the bromodomain could lead to increased sensitivity to translation pathway inhibitors. Therefore, such specificity would be missed if considering all mutations; bromodomain mutations are only a subset of BRD9 pathogenic mutations. A systematic study of the consequences of each subunit’s cancer alterations will be necessary to assess consequences on translation and sensitivity to translation pathway inhibitors. Large scale sequencing studies revealed the co-occurrence of activated mutations in PI3K and inactivating mutations of the mSWI/SNF complexes (1, 3). A cooperative mechanism between these two altered pathways has therefore been suggested (1, 3). While biologically this co-occurrence is not understood, the roles of mSWI/SNF in translation could be relevant. PI3K activation promotes protein synthesis and cellular proliferation (28). The decreased translation due to inactivating mutations in mSWI/SNF subunits could need the counteraction of increased translation through PI3K activating or PTEN inactivating mutations (3). Altering the translation of specific mRNAs is another route to enable a selective advantage to these cancer cells; moreover, these possibilities are not mutually exclusive. We do not observe any specific association between PI3K mutation and mSWI/SNF alteration in expression and localization in our breast cancer cases (data not shown). SMARCA4 was overexpressed in more than 90% of our cases, confirming the fact that SMARCA4 is likely an oncogene and that its low expression could be a favorable outcome in breast cancer (3, 15). On the contrary, SMARCA2 expression was extensively lost. It would be interesting to determine if such breast cancer cases would be more sensitive to translation inhibitors, especially when there is no SMARCA4 overexpression and/or when SMARCA4 is mutated. Further, a proportion of breast cancer cases exhibited loss of PBAF subunits expression which could sensitize to translation pathways inhibitors. However, we also observed several cases with exclusive gain in cytoplasmic expression or opposite changes within different compartments. The consequences of such differences in subcellular functions could have great potential for the discovery of therapeutically targetable events.

Methods

Cell culture and drug treatments HEK293T cells were obtained from ATCC. HAP1 and HAP1 SMARCA2 (HZGHC004055c012), SMARCA4 (HZGHC002878c004), ARID2 (HZGHC000907c009), ARID1A (HZGHC000618c010), BRD7 (HZGHC000923c010), BRD9 (HZGHC000934c010) knockout cell lines were obtained from and characterized by Horizon Discovery. Culturing media and media supplements were used based on supplier’s instructions. HAP1 cell lines were used under 30 passages. Mouse embryonic stem cells genome editing and culture are described in (21). BEZ235 (Apexbio), AZD8055 (AChemBlock) iBRD9 (Tocris), TP472 (Tocris) and PFI-3 (Tocris) inhibitors were used at the indicated concentrations on cells seeded since 24 to 72 hours.

Antibodies LDH (Santa Cruz, sc-133123), Streptavidin-HRP (Invitrogen SA100-01), Puromycin (Sigma, MABE343), Flag- tag (Sigma, F3165-.2MG), Veriblot IgG (Abcam, ab131366), Normal rabbit IgG (Cell signaling, 2729S), eIF3A (Abcam, ab118357), eIF4A1 (Abcam, ab31217), PABP (Abcam, ab21060), RPS6 (Abcam, ab58350), RPS3 (Abcam, ab140676), Beta actin (Abcam, ab8226), SMARCC2 (Abcam, ab71907), SMARCA4 (Abcam, ab110641; IP,WB), SMARCA4 (Abcam, ab108318; IHC), SMARCC1 (Abcam, ab172638), SMARCA2 (Abcam, ab15597; WB), SMARCA2 (Sigma HPA029981; IHC), PHF10 (Abcam, ab154637), PB1 (Abcam, ab196022; WB, IHC), BRD7 (Abcam, ab56036; WB), BRD7 (Protein tech 51009-9-AP; IHC), DPF2 (Abcam, ab134942), ARID2 (Abcam, ab51019; WB), ARID2 (Abcam ab113283; IHC), ARID1B (Abcam, ab84461), ARID1A (Cell signaling, 12354S), BRD9 (Bethyl laboratories, A303-781A), GLTSCR1 (Santa Cruz, sc-515086), SS18 (Cell signaling, 21792S), SMARCB1 (Abcam, ab192864), Cyclin B1 (Fisher Scientific, BDB554177), Phospho-Ser28- Histone3 (BioLegend, 641002).

Cell Fractionation Cell fractionation was performed as in (29). For the western-blot analysis a comparable fraction of each compartment was loaded on a gel.

10 Immunohistochemistry SMARCA4 was performed on a Roche Ventana Benchmark Ultra instrument (Tucson, AZ) using CC1 retrieval and antibody dilution at 1:25. SMARCA2 (1:300), BRD7(1:400) and PB1 (:2000) were performed manually using Tris-EDTA antigen retrieval (pH 9). ARID2 (1:250) was performed manually using citrate antigen retrieval (pH 6).

Immunoprecipitation Immunoprecipitation was performed as in (16).

Translation assays L-azidohomoalanine (AHA) incorporation were performed as in (16). Puromycin incorporation assays were performed as in (30). Antibodies intensities from western blot membranes were analyzed using ImageJ. Because of the string smear obtained for these western blots, loading controls (actin) were ran on separate membranes.

Polysome profiling Polysome profiling were performed as in (16). Because of the high number of proteins tested and size restriction, identical fractions from polysome profiles were assessed with several gels.

Viability assays Viability assays were performed following supplier’s instructions from the CellTiter-Glo 2.0 Cell Viability Assay (Promega). Luminescence reading was performed using GloMax® Discover Microplate Reader (Promega).

Simultaneous Overview of tri-Molecule Biosynthesis (SOM3B) SOM3B was performed as described previously in (19). Live cells were treated with inhibitors for 10 minutes and spiked with IdU, BrU. Single cell mass-cytometry datasets generated and presented are available in flowrepository.org with the following identifier: public identifier will be generated at time of publication.

Quantification and Statistical analysis Please refer to the figure legends or the experimental details for description of sample size (n) and statistical details. Cell culture assays were performed at least three independent times. Data are expressed as mean ± SEM (Standard Error of the Mean). Differences were analyzed by two-tailed Student’s t test, one-way ANOVA, P-values ≤ 0.05 were considered significant and annotated with *.

Acknowledgments The authors would like to thank Drs. Joshua Back and Johnathan Whetstine for their helpful discussions of the data, and Ellen Gomulia and Ivy Mangonon for assistance with SMARCA4 immunohistochemistry. G.R.C. is supported by the Howard Hughes Medical Institute, grants from the NIH, CA163915 and NS046789 and a grant from the Department of Defense, Breast Cancer Research fund. C.M.W. was supported by a grant from GSK, Sir James Black Fellowship. S.C.K and S.C.B received support from the NIH: T32GM007276 (S.C.K), 1DP2OD022550-01 (S.C.B.), 1R01AG056287-01 (S.C.B.), 1R01AG057915-01 (S.C.B.) and 1U24CA224309-01 (S.C.B.).

References 1. C. Kadoch, G. R. Crabtree, Mammalian SWI/SNF chromatin remodeling complexes and cancer: Mechanistic insights gained from human genomics. Science Advances 1, e1500447 (2015). 2. C. Hodges, J. G. Kirkland, G. R. Crabtree, The Many Roles of BAF (mSWI/SNF) and PBAF Complexes in Cancer. Cold Spring Harbor Perspectives in Medicine 6, a026930 (2016). 3. P. Mittal, C. W. M. Roberts, The SWI/SNF complex in cancer - biology, biomarkers and therapy. Nat Rev Clin Oncol 17, 435-448 (2020). 4. N. Mashtalir et al., Modular Organization and Assembly of SWI/SNF Family Chromatin Remodeling Complexes. Cell 175, 1272–1288 (2018). 5. X. Sima et al., The genetic alteration spectrum of the SWI/SNF complex: The oncogenic roles of BRD9 and ACTL6A. PLOS ONE 14, e0222305 (2019).

11 6. R. Agarwal et al., Dynamic molecular monitoring reveals that SWI-SNF mutations mediate resistance to ibrutinib plus venetoclax in mantle cell lymphoma. Nat Med 25, 119-129 (2019). 7. B. J. Manning, C. L. Peterson, Direct interactions promote eviction of the Sir3 heterochromatin protein by the SWI/SNF chromatin remodeling enzyme. Proceedings of the National Academy of Sciences 111, 17827-17832 (2014). 8. P. Kapoor et al., Regulation of Mec1 kinase activity by the SWI/SNF chromatin remodeling complex. Genes & development 29, 591-602 (2015). 9. B. Z. Stanton et al., Smarca4 ATPase mutations disrupt direct eviction of PRC1 from chromatin. Nature Genetics 49, ng.3735 (2016). 10. Sang W. Park et al., BRD7 Regulates XBP1s’ Activity and Glucose Homeostasis through Its Interaction with the Regulatory Subunits of PI3K. Cell Metabolism 20, 1088 (2014). 11. J. H. Lee et al., Cytoplasmic localization and nucleo-cytoplasmic shuttling of BAF53, a component of chromatin-modifying complexes. Molecules and cells 16, 78-83 (2003). 12. S. Bilodeau et al., Role of Brg1 and HDAC2 in GR trans-repression of the pituitary POMC gene and misexpression in Cushing disease. Genes & Development 20, 2871-2886 (2006). 13. J. Fukuoka et al., Chromatin Remodeling Factors and BRM/BRG1 Expression as Prognostic Indicators in Non-Small Cell Lung Cancer. Clinical Cancer Research 10, 4314-4324 (2004). 14. S. Animireddy et al., Aberrant cytoplasmic localization of ARID1B activates ERK signaling and promotes oncogenesis. J Cell Sci 134, (2021). 15. Q. Wu et al., The BRG1 ATPase of human SWI/SNF chromatin remodeling enzymes as a driver of cancer. Epigenomics 9, 919-931 (2017). 16. C. Van Rechem et al., Lysine demethylase KDM4A associates with translation machinery and regulates protein synthesis. Cancer Discov 5, 255-263 (2015). 17. G. Fuchs et al., 4sUDRB-seq: measuring genomewide transcriptional elongation rates and initiation frequencies within cells. Genome Biol 15, R69 (2014). 18. S. O. Olofsson et al., Structure and biosynthesis of apolipoprotein B. Am Heart J 113, 446-452 (1987). 19. S. C. Kimmey, L. Borges, R. Baskar, S. C. Bendall, Parallel analysis of tri-molecular biosynthesis with cell identity and function in single cells. Nature Communications 10, 1185 (2019). 20. S. Pyronnet, J. Dostie, N. Sonenberg, Suppression of cap-dependent translation in mitosis. Genes Dev 15, 2083-2093 (2001). 21. C. M. Weber et al., mSWI/SNF promotes polycomb repression both directly and through genome-wide redistribution. bioRxiv, (2020). 22. H. C. Hodges et al., Dominant-negative SMARCA4 mutants alter the accessibility landscape of tissue- unrestricted enhancers. Nature Structural & Molecular Biology 25, 61-72 (2018). 23. J. M. McFarland et al., Improved estimation of cancer dependencies from large-scale RNAi screens using model-based normalization and data integration. Nat Commun 9, 4610 (2018). 24. C. Yu et al., High-throughput identification of genotype-specific cancer vulnerabilities in mixtures of barcoded tumor cell lines. Nat Biotechnol 34, 419-423 (2016). 25. N. K. Sinha et al., EDF1 coordinates cellular responses to ribosome collisions. Elife 9, (2020). 26. R. C. Wek, Role of eIF2alpha Kinases in Translational Control and Adaptation to Cellular Stress. Cold Spring Harb Perspect Biol 10, (2018). 27. K. N. Dziadkowiec, E. Gasiorowska, E. Nowak-Markwitz, A. Jankowska, PARP inhibitors: review of mechanisms of action and BRCA1/2 mutation targeting. Prz Menopauzalny 15, 215-219 (2016). 28. B. A. Hemmings, D. F. Restuccia, PI3K-PKB/Akt pathway. Cold Spring Harb Perspect Biol 4, a011189 (2012). 29. C. Van Rechem et al., The lysine demethylase KDM4A controls the cell-cycle expression of replicative canonical histone genes. Biochim Biophys Acta Gene Regul Mech 1863, 194624 (2020). 30. E. K. Schmidt, G. Clavarino, M. Ceppi, P. Pierre, SUnSET, a nonradioactive method to monitor protein synthesis. Nature Methods 6, 275-277 (2009).