Screening of gene function in cell intoxication by CNF1 links Sec61 translocon to Rac1 GTPase activity

TL;DR

The study shows how a bacterial toxin affects cell signaling by linking protein production in the endoplasmic reticulum to the activity of a key cell protein called Rac1.

Contribution

The study reveals a novel connection between the Sec61 translocon and Rac1 GTPase activity through a genome-wide siRNA screen in CNF1-intoxicated cells.

Findings

siRNAs targeting Sec61A1, HACE1, and Lu/BCAM significantly inhibit CNF1-mediated depletion of Rac1.

Inhibition of Sec61A1 reduces GTP-loading and membrane association of Rac1 in intoxicated cells.

Blocking N-glycosylation in the ER prevents Rac1 GTP-loading in CNF1-treated cells.

Abstract

The CNF1 toxin from extraintestinal pathogenic Escherichia coli (ExPEC) deamidates glutamine 61 of Rac1 small GTPase, as well as its equivalents in RhoA and Cdc42 into glutamic acid. This post-translational modification of Rho proteins abrogates the hydrolysis of GTP into GDP, thereby enhancing signal transduction. Meanwhile, the sustained GTP-loading of Rac1 Q61E sensitizes it to ubiquitin-mediated proteasomal degradation catalyzed by the HACE1 E3 ligase rate-limiting factor, leading to a cellular depletion of Rac1 over time. We report data from a quantitative genome-wide screen of siRNAs inhibiting CNF1-mediated cellular depletion of Rac1 in primary human cells. As best hits, we identified a group of three siRNAs targeting the Sec61A1 subunit of the Sec61 translocon, as well as HACE1 and the Lu/BCAM host cell receptor of CNF1. We extend these findings by identifying a group of siRNAs targeting genes involved in ER and Golgi homeostasis and trafficking. Functional studies showed that both chemical and genetic inhibition of Sec61A1 dampens GTP-loading and membrane association of Rac1 in CNF1-intoxicated cells, while the proper deamidation of RhoA provides a control of CNF1 cytosolic action. Finally, we extend these findings by showing that inhibition of N-glycosylation of neo-synthesized proteins in the ER abrogates Rac1 GTP-loading in CNF1-treated cells. Collectively, these data point to a control of Rac1 signaling operated by protein biosynthesis and N-glycosylation in the ER.

The remarkable evolutionary convergence of bacterial effectors from pathogens toward the host small GTPase Rac1, the master regulator of the actin cytoskeleton, confers to these microbes an enhanced capacity to invade host cells and tissues. The CNF1 toxin, a colonization factor of the gastrointestinal tract produced by pathogenic strains of Escherichia coli, has been instrumental in deciphering the regulation and function of Rac1. By performing a whole-genome screen based on CNF1 action, we establish the key requirement of Sec61 translocon-dependent protein biosynthesis and N-glycosylation at the endoplasmic reticulum for proper activation of Rac1 in intoxicated cells. Our data connect the Sec61 translocon and N-glycosylation of neo-synthesized proteins at the endoplasmic reticulum in the control of the activity of Rac1 and other Rho GTPases.

Figures

High-throughput screening of siRNA inhibiting CNF1-induced Rac1 depletion. (A) Representative confocal images of endogenous Rac1 immunofluorescence signal intensities (left panels, green) in control and CNF1-intoxicated cells, as well as a heat map (right panels) showing the number of Rac1-positive cells expressed as fold to non-treated conditions (NT). HUVECs were treated with the indicated concentrations of CNF1 in nanomolar (nM) for 6 h, and images were acquired on an Opera Phenix automatic confocal microscope. The lower graph shows the percentage of Rac1-positive cells as a function of the log10 values of CNF1 concentrations. Dots show mean values of n = 3 independent experiments, n = 5 replicates. Linear regression shows the linear relationship (R2 = 0.62) between values of Rac1-positive cells and a range of CNF1 concentrations comprised between 2 and 0.5 nM, with a calculated EC50 value of 0.8 nM. (B) The upper panel shows the principle of the screen designed to identify siRNAs inhibiting CNF1-induced cellular depletion of Rac1. HUVEC transfected with siCtrl and treated with CNF1 set the minimal value of endogenous Rac1 signal (αRac1). HUVEC was transfected with siLu/BCAM or siHACE1, and then intoxicated with CNF1, resulting in maximal protection values of Rac1 signals. The lower panels show representative confocal images of HUVEC transfected with siCtrl, siLu/BCAM, or siHace1 and treated with CNF1 at 1 nM for 6 h, as well as siCtrl cells left untreated. Confocal images acquired on Opera Phenix at objective 20× show endogenous Rac1 immunofluorescence (green) and nuclei (blue). Scale bar, 50 µm. Values show the number of Rac1-positive cells (No. Rac1+cells), total cells (No. total cells), and the calculated ratio of Rac1+ cells. (C) Scatter plot showing the distribution of Rscr values of Rac1-positive cells transfected with 21 reordered siRNAs from the list of top-ranked siRNA (confirmatory screens), as a function of the calculated Z-score of cell number. Cells were treated with CNF1 at 1 nM for 6 h. Each dot corresponds to mean values from n = 3 independent experiments, n = 4 replicates. (D) Immunoblots show endogenous Rac1 signals from cells transfected with siCtrl, siHace1, and siSec61A1 that were left untreated or intoxicated with 1 nM of CNF1 for 6 h. GAPDH shows equal protein loading control. One representative experiment, n = 5. The lower graph shows intensities of Rac1 signals normalized to GAPDH and expressed as fold to the siCtrl condition. Data are mean ± SEM. Each dot corresponds to an individual value from n = 5 independent experiments. One-way ANOVA with Dunnett’s correction; siCtrl+ CNF1 vs siCtrl, P < 0.0001; siCtrl+ CNF1 vs siHace1+ CNF1, *P = 0.0008; siCtrl+ CNF1 vs siSec61A1 + CNF1, P = 0.0062. (E) Immunoblots show the upper shift that undergoes the deamidated form of RhoA (RhoA Q63E) in CNF1-intoxicated cells, versus wild-type RhoA (RhoA Q63) in Urea-SDS-PAGE condition. The cells were transfected with siCtrl, siSec61A1, siHace1, or siLu/BCAM and left untreated or intoxicated with CNF1 at 1 nM for 6 h. Immunoblot GAPDH shows equal protein loading. One representative experiment, n = 3. (F) Scatter plot showing the distribution of Rscr values of Rac1-positive cells transfected with six siRNAs targeting structural and regulatory subunits of the Sec61 translocon, as well as siLu/BCAM, as a function of the calculated Z-score of cell number. Cells were treated with CNF1 at 1 nM for 6 h. Each dot corresponds to mean values from n = 2 independent experiments, n = 6 replicates.

Active Sec61 translocon is required for CNF1-induced Rac1 depletion. (A) Immunoblots show endogenous Rac1 immunofluorescence signal intensities from cells treated with siCtrl, siSec61A1, or treated with mycolactone (Myco) at 100 nM for 24 h, and prior to treatment with 1 nM of CNF1 for 6 h. Immunoblot GAPDH shows equal protein loading. An immunoblot anti-beta-1 integrin was performed to control the inhibition of protein biosynthesis in the ER. One representative experiment, n = 3. (B) Immunoblots show the electromobility shift of RhoA Q63E versus RhoA Q63 in cells treated with mycolactone at 100 nM for 24 h (Myco) prior to treatment with 1 nM CNF1 for 6 h. Immunoblot GAPDH shows equal protein loading. One representative experiment, n = 3. (C) The graph shows intensities of Rac1 signals normalized to GAPDH and expressed as fold to the non-treated (NT) condition. Cells were treated for 24 h with mycolactone (Myco) at the indicated concentrations prior to treatment with 1 nM of CNF1 for 6 h. Bars show mean ± SEM. Each dot corresponds to an individual value from n = 3 independent experiments. One-way ANOVA with Dunnett’s correction; NT + CNF1 vs Myco 10 nM + CNF1, ns; NT + CNF1 vs Myco 25 nM + CNF1, *P = 0.035; NT + CNF1 vs Myco 50 nM + CNF1, P = 0.044; NT + CNF1 vs Myco 75 nM + CNF1, P = 0.079; NT + CNF1 vs Myco 100 nM + CNF1, P = 0.029. (D) The graph shows intensities of beta-1 integrin signals normalized to GAPDH and expressed as fold to the non-treated (NT) condition. Cells were treated for 24 h with mycolactone (Myco) at indicated concentrations prior to treatment with CNF1 at 1 nM for 6 h. Bars show mean ± SEM. Each dot corresponds to an individual value from n = 4 independent experiments. One-way ANOVA with Dunnett’s correction; Myco 10 vs NT, *P < 0.001; Myco 10 vs Myco 25, P = 0.0019; Myco 10 vs Myco 50, *P = 0.0005; Myco 10 vs Myco 100, *P = 0.0007.

Functional Sec61 translocon allows CNF1-induced translocation of Rac1 to membranes. (A) Immunoblots show the relative distribution of Rac1 and RhoA between cytosol (Cyt) and membrane (Mb) fractions in HUVEC intoxicated with CNF1. Cells were pre-treated with siCtrl, siSec61A1, 100 nM mycolactone (Myco), or 15 µM lovastatin (Lova) 24 h prior to intoxication with CNF1 at 1 nM for 2 h. Immunoblots anti-Rho GDI or anti-GAPDH show controls of membrane fractionation and equal loading of cytosolic proteins. Immunoblot anti-beta-1 integrin shows control of the inhibition of protein biosynthesis in the ER. Representative immunoblots of n = 3 independent experiments. Rac1 immunosignals were quantified and are expressed as the signal ratio between membrane and cytosol, together with membrane fractions normalized to NT (graph). Bars show means ± SEM. Each dot corresponds to an individual value from n = 3 independent experiments. One-way ANOVA with Dunnett’s correction, siCtrl vs siCtrl + CNF1, *P = 0.4966; siCtrl + CNF1 vs siSec61A1 + CNF1, *P = 0.0001; siCtrl + CNF1 vs siCtrl + Myco + CNF1, *P = 0.0006; siCtrl + CNF1 vs siCtrl + Lova + CNF1, *P = 0.0007. (B) Immunoblots show the distribution of Rac1 between aqueous (Aq) and detergent (Dt) phases after Triton X-114 nonionic detergent phase separation. HUVECs were transfected with siSec61A1 or siCtrl for 24 h prior to the extraction of prenylated-Rac1 in the Triton X-114 phase. As a control, siCtrl-transfected cells were treated with 15 µM of lovastatin for 24 h (Lova). Immunoblot anti-GAPDH shows control of phase separation. Immunoblot anti-beta-1 integrin shows control of Sec61 translocon inhibition. One representative experiment, n = 4. Rac1 signals were quantified and expressed as signal ratio between the detergent and aqueous phases, normalized to the siCtrl condition (graph). Bars show means ± SEM. Each dot corresponds to an individual value from n = 4 independent experiments. One-way ANOVA with Dunnett’s correction, siCtrl vs siSec61A1, ns; siCtrl vs siCtrl + Lova, P = 0.0058; siSec61A1 vs siCtrl + Lova, *P = 0.0452.

Active Sec61 translocon allows CNF1-mediated activation of Rac1. (A) Representative immunoblots show levels of GTP-bound Rac1 (Rac1-GTP) in cells transfected with siCtrl, siHace1, siLu/BCAM, or siSec61A1 prior to intoxication with CNF1 at 1 nM for 2 h. Levels of GTP-bound Rac1 were assessed by GST-PAK pulldown. Immunoblots of anti-GAPDH and anti-Rac1 on total cell lysates show equal protein amounts engaged in the GST-PAK pulldown. Graph: Rac1 immunosignals were quantified and are expressed as the ratio of GTP-bound Rac1 normalized to the total Rac1 signal. Bars show means ± SEM. Each dot corresponds to a value from n = 3 independent experiments. One-way ANOVA with Dunnett’s correction, siCtrl vs siLu/BCAM, *P = 0.0041; siCtrl vs siSec61A1, *P = 0.0019. (B) Graph shows the variation of GTP-bound Rac1 (Rac1-GTP) levels normalized to Rac1 total level (Rac1) in cells left untreated (NT) or treated for 24 hours with mycolactone (Myco) at indicated concentrations together with CNF1 at 1 nM. Data are expressed as fold variations compared with cells treated with CNF1. Levels of Rac1-GTP were assessed by GST-PAK pulldown and anti-Rac1 immunoblotting. Bars show means ± SEM. Each dot corresponds to a value from n = 2 independent experiments. One-way ANOVA with Dunnett’s correction, NT vs Myco 10, ns; NT vs Myco 25, P = 0.0089; NT vs Myco 50, P = 0.002; NT vs Myco 100, P = 0.0012.

ER stress induced by tunicamycin blocks CNF1-mediated activation of Rac1. (A) Representative immunoblots show GTP-bound Rac1 (Rac1-GTP) levels in cells treated with ER stressors. HUVECs were treated for 24 h with 2 µM thapsigargin (Tg), 2 µM tunicamycin (Tn), or 100 nM mycolactone (Myco) prior to the addition of CNF1 at 1 nM for 2 h. The fraction of GTP-bound Rac1 was purified by GST-PAK pulldown. Immunoblots anti-GAPDH and anti-Rac1 show equal protein amounts engaged in GST-PAK pulldowns. Immunoblots anti-beta-1 integrin and anti-ATF4 show Sec61 translocon inhibition and induction of PERK-dependent ER stress signaling, respectively. Representative immunoblots, n = 3 independent experiments. (B) Graph shows levels of GTP-bound Rac1 normalized to GAPDH and expressed as fold inhibition of Rac1 GTP compared with untreated cells. Bars show means ± SEM. Each dot corresponds to an individual value from n = 3 independent experiments. One-way ANOVA with Dunnett’s correction, NT vs Tg, ns; NT vs Tn *P = 0.0002; NT vs Myco *P = 0.0006.

Tables

TABLE 1. List of 21 siRNA-targeted genes ranked according to data of confirmatory screensa

No.	Gene	Screen name	Ref_Seq	Rscreenorm confirmation	Shift RhoA
1	Lu/BCAM	LU/BCAM	NM_001013257	0.81	N
2	HACE1	HACE1	NM_020771	0.78	Y
3	SEC61A1	SEC61A1	NM_013336	0.77	Y
4	USO1	VDP	NM_003715	0.54	NA
5	VSIG8	LOC391123	NM_001013661	0.44	Y
6	RGS7BP	LOC401190	NM_001029875	0.42	Y
7	SWSAP1	FLJ35119	NM_175871	0.40	Y
8	SLF1	ANKRD32	NM_032290	0.39	Y
9	SH3D19	SH3D19	NM_001009555	0.38	NA
10	SLC35E1	SLC35E1	NM_024881	0.36	NA
11	CYB561A3	CYB561A3	NM_153611	0.33	NA
12	TRUB2	TRUB2	NM_015679	0.32	NA
13	TBC1D22A	C22ORF4	NM_014346	0.31	NA
14	PTCD1	PTCD1	NM_015545	0.28	NA
15	TMEM43	TMEM43	NM_024334	0.26	NA
16	CFAP77	FLJ46082	NM_207417	0.25	NA
17	HNRNPUL2	HNRPUL2	NM_001079559	0.21	NA
18	ZNF598	ZNF598	NM_178167	0.20	NA
19	HNRNPA2B1	HNRPA2B1	NM_002137	0.20	NA
20	OR2T6	OR2T6	NM_001005471	0.15	NA
21	LIN9	LIN9	NM_173083	0.10	NA

TABLE 2. List of 22 siRNA-targeted genes ranked according to data of confirmatory screensa

No.	Gene	Ref_Seq	Rscr confirmation	Shift RhoA
1	SEC61A1	NM_013336	0.78	Y
2	PITPNB	NM_012399	0.60	Y
3	SNX4	NM_003794	0.58	Y
4	TMED9	NM_017510	0.50	Y
5	SEL1L	NM_005065	0.48	Y
6	GBF1	NM_004193	0.45	Y
7	RAB11A	NM_004663	0.42	Y
8	USO1/VDP	NM_003715	0.41	NA
9	COMMD8	NM_017845	0.35	NA
10	SLC35E1	NM_024881	0.30	Y
11	RAB41	NM_001032726	0.29	NA
12	HIPK2	NM_022740	0.27	NA
13	COMMD2	NM_016094	0.21	NA
14	CYB561A3	NM_153611	0.16	NA
15	COMMD5	NM_001081004	0.16	NA
16	PODXL	NM_005397	0.16	NA
17	FAM109A	NM_144671	0.13	NA
18	EXOC7	NM_015219	0.08	NA
19	GNL3	NM_014366	0.06	NA
20	AGAP1	NM_014914	0.01	NA
21	ARAP3	NM_022481	0.00	NA
22	HGS	NM_004712	−0.11	NA

Equations

Full text

INTRODUCTION

Rac1 small GTPase drives branched actin cytoskeleton polymerization at the interface of membranes downstream of numerous cell signaling pathways controlling cell adhesion, metabolism, division, and fate, as well as gene programs of alarm and defense against pathogen attacks (1–6). Consistent with the central role played by Rac1 in host-pathogen interactions, multiple pathogenic bacteria have evolved key effectors of pathogenicity and colonization targeting this small GTPase (3–6). Rac1 and other Rho proteins, including RhoA and Cdc42, are small GTPases able to switch between a guanosine diphosphate (GDP)-bound inactive form, sequestered in the cytosol in association with a Rho guanine dissociation inhibitor (Rho GDI), and a GTP-bound active form anchored at membranes where they transduce signals (2, 7, 8). How the regulation of Rac1 is integrated within larger signaling networks of cell adaptation to intrinsic stresses largely remains to be defined, considering the master regulatory function of Rac1 in cell fate and defenses against bacterial pathogens.

The CNF1 toxin can enhance the capacities of extraintestinal pathogenic Escherichia coli (ExPEC) to colonize the gastrointestinal tract (9). CNF1 binds to the cell surface-exposed Lu/BCAM, and eventually the p67-laminin receptor precursor, to become endocytosed through clathrin- and caveola-independent processes (10–14). The enzymatic part of CNF1 is translocated through vesicular membranes to the cytosol once the toxin reaches the acidic environment of late endosomal compartments (13, 15). CNF1 catalyzes the deamidation of a critical glutamine residue Q61 of Rac1 and Cdc42, or the equivalent Q63 in RhoA, into a glutamic acid (16–18). The deamidation catalyzed by CNF1 onto Rho proteins abrogates their intrinsic and GTPase-activating protein (GAP)-stimulated GTPase activity (16, 17, 19). Consequently, Rho proteins remain locked into a GTP-bound active state, able to transduce signals at the interface of membranes. Whether endogenous GTPase-dead mutants of Rho proteins in CNF1-intoxicated cells undergo a spontaneous loading of guanosine triphosphate or whether the loading of GTP requires the activity of cellular factors remains an open question.

Remarkably, CNF1, by converting endogenous Rho proteins into dominant positive mutant forms, exacerbates a cryptic regulation of Rho protein stability that involves the ubiquitin-proteasomal system (UPS) (19). The degradative polyubiquitination of the GTP-bound form of Rac1 is catalyzed by the HECT-domain and ankyrin repeat-containing E3 ligase 1 (HACE1) (20–22). The activation and proteasomal degradation of Rac1 by the UPS contribute to the invasion of cells and tissues by ExPEC, where beta-1 integrin mechano-activation and signaling are hijacked by bacterial type I pili tipped with FimH (9, 23). Although several cellular targets of HACE1 have been identified, the control exerted by HACE1 on Rac1 is essential in the regulation of ROS production by NADPH oxidase, cyclin-D1 expression, and the tumor suppressor function of HACE1 (21, 24–26). HACE1 accumulates at the edge of membrane ruffles triggered by the ectopic expression of active mutants of Rac1 or in cells intoxicated by CNF1 (20, 27). Moreover, HACE1 displays a Rab1-dependent localization in Golgi stacks, where its E3 ligase activity controls the monoubiquitination of syntaxin-5 and post-mitotic fusion of membranes for Golgi reassembly (28, 29).

A few studies have started to pinpoint connections between Rac1 signaling and the secretory pathway. Notably, Rac1 interacts with the small GTPase Sar1 (secretion-associated Ras-superfamily gene 1) to enhance the formation of endoplasmic reticulum exit sites (ERES), which ensure a supply of lipids to the plasma membrane when under mechanical strains (30). Rac1 and its exchange factor beta-PIX also regulate the recruitment of Arp2/3 to the trans-Golgi network (TGN), where it drives the actin-dependent formation of clathrin-AP1-coated carrier vesicles on the way to the endocytic pathway (31, 32). Finally, it has been reported that expression of the ER-located inositol-requiring transmembrane kinase/endoribonuclease 1 alpha (IRE1 alpha), a sensor and transducer of the unfolded protein response (UPR), controls the levels of active GTP-bound Rac1 in mouse embryonic fibroblasts (MEFs) (33). Much remains to be established on how ER- and Golgi-associated components control Rac1 activity to promote cell homeostasis.

We developed an image-based method to quantify the cellular pool of Rac1, enabling us to assign each siRNA in a pan-human genome screen a score of protection against CNF1-induced depletion of Rac1. This quantitative approach allowed us to pinpoint the critical function of the Sec61A1 subunit of the Sec61 translocon in CNF1-mediated depletion of Rac1, together with a group of genes involved in ER and Golgi trafficking and homeostasis. We report that the blocking effect imposed by knocking down Sec61A1 on the depletion of Rac1 in CNF1-treated cells occurs at the stage of Rac1 activation. The Sec61 translocon forms a ribosome-binding and gated channel, which facilitates the co-translational translocation of nascent proteins in the ER membrane and lumen (34). Remarkably, Sec61 can be specifically inhibited by mycolactone, a polyketide produced by Mycobacterium ulcerans (35–37). Recent cryo-electron microscopy data, as well as complementary Sec61A1 mutant-based functional studies, showed that mycolactone docks to Sec61A1 inner channel, thereby locking the Sec translocon in an inactive state (36, 38, 39). As a result, mycolactone inhibits the production of type-I and II transmembrane proteins, as well as the production of most secreted proteins (40, 41). Taking advantage of mycolactone’s defined mode of action, we showed that chemical inhibition of the Sec61 translocon by mycolactone impairs the activation of Rac1 and RhoA in CNF1-treated cells, whereas RhoA is properly deamidated, thereby establishing the intracellular action of the toxin. This suggests that inhibition of the Sec61 translocon uncouples Rho deamidation from the exchange of GDP by GTP, pointing to intrinsic cell-based activation of Rho GTPases in CNF1-intoxicated cells. Together, these data provide evidence that the regulation of Rac1 activity integrates signals from both protein biosynthesis and N-glycosylation that emanate from the endoplasmic reticulum.

RESULTS

High content screening of siRNAs inhibiting CNF1-induced cellular depletion of Rac1

The CNF1 toxin induces a time- and dose-dependent depletion of the GTP-bound active form of Rac1 (14, 19, 20). We aimed to exploit this property of CNF1 to screen for new regulatory components of Rac1. We developed an image-based quantitative method to measure the cellular levels of endogenous Rac1. This method relies on the fluorescent immunolabeling of Rac1, coupled to an automated imaging algorithm that segments and enumerates Rac1-positive cells within the whole cell population (Fig. S1A). The CNF1 toxin was first applied to human umbilical vein endothelial cells (HUVECs) for 6 h in 2-fold serial dilutions from 10 nM to 0.125 nM. The percentage of Rac1-positive cells showed a linear decrease between 0.5 nM and 2 nM of CNF1, R^2^ = 0.62 (Fig. 1A). The above data prompted us to perform the screen with CNF1 at 1 nM, that is, within the dynamic range of endogenous Rac1 signal detection (Fig. 1A). To set up the assay for gene knockdown, the cells were transfected with small interfering RNAs (siRNAs) targeting Lu/BCAM, to block the endocytosis of CNF1, or HACE1, to block the ubiquitin-mediated degradation of Rac1 (Fig. 1B; Fig. S1B). In siLu/BCAM and siHACE1 conditions, we recorded 81% and 77% of Rac1-positive cells treated with CNF1, respectively, whereas the levels of Rac1-positive cells dropped to 12% in the siRNA control (siCtrl) and CNF1 condition (Fig. 1B). Given the robustness of these effects, siHace1 and siCtrl in the absence or presence of CNF1 were chosen as positive and negative controls, respectively. We then screened a genome-wide siRNA library targeting 18,125 human genes. As a quality control, we determined an SSMD value for each plate using internal controls (e.g., cells treated with siCtrl or siHace1 in the absence or presence of CNF1) (Fig. S1C). This allowed us to retain a total of 60 plates (n = 384 wells each), encompassing a total of 16,433 siRNAs, for further analysis. Due to the non-random organization of the siRNA library, we calculated a plate-based normalization score, Rscreenorm (Rscr), of Rac1-positive cells using a method previously described (42). Within the ranked siRNAs of the primary screen, we identified siRNAs targeting HACE1 (position 3, Rscr = 0.86) and Lu/BCAM (position 22, Rscr = 0.79), supporting the robustness of the screen ( Table S1). We reordered 21 siRNAs with Rscr scores ≥ 0.79 (including siLu/BCAM) to perform three independent confirmatory screens and ranked them according to the average of their protection scores (Table 1 and Fig. 1C). Remarkably, we identified a group of three best hits encompassing siLu/BCAM and siHace1 together with an siRNA targeting the Sec61A1 subunit of the Sec61 translocon (Rscr > 0.77) (Fig. 1C; Table 1). For the remaining 18 siRNAs, we measured lower Rscr scores comprised between 0.36 and 0.54 (Fig. 1C; Table 1). Immunoblotting, as an alternative method of analysis, confirmed the high level of protection conferred by siSec61A1 against CNF1-induced depletion of Rac1 (Fig. 1D). The decrease in Sec61A1 mRNA in siSec61A1-knockdown cells was confirmed by qPCR (Fig. S1D). To rule out a potential interference of siSec61A1 in the CNF1 intoxication process, we took advantage of the unique property of the deamidated form of RhoA (RhoA Q63E), compared with wild-type RhoA among Rho proteins, to exhibit an electrophoretic mobility shift on urea-SDS-PAGE (43). This analytical method enables quantitative monitoring of the intracellular activity of the CNF1 toxin. HUVECs were treated with siCtrl, siLu/BCAM, siHace1, or siSec61A1 before the addition of CNF1 at 1 nM for 6 h. In siLu/BCAM-treated cells, which are no longer susceptible to CNF1, we did not detect the electrophoretic mobility shift of RhoA (Fig. 1E). In contrast, we observed the shifted form of RhoA in cells co-treated with siSec61A1 and CNF1, thereby indicating productive intoxication (Fig. 1E). Finally, we went on to confirm the protection conferred by siRNAs targeting core and accessory components of the Sec61 translocon including Sec61A, Sec61B, Sec61G, Sec62, and Sec63, against CNF1-induced depletion of Rac1, yielding Rscr values between 0.28 and 0.72 (Fig. 1F). This genome-wide quantitative screen provides a list of gene candidates involved in CNF1-induced cellular depletion of Rac1, also assigning a key function to Sec61A1 channel subunit of the Sec61 translocon.

*High-throughput screening of siRNA inhibiting CNF1-induced Rac1 depletion. (A) Representative confocal images of endogenous Rac1 immunofluorescence signal intensities (left panels, green) in control and CNF1-intoxicated cells, as well as a heat map (right panels) showing the number of Rac1-positive cells expressed as fold to non-treated conditions (NT). HUVECs were treated with the indicated concentrations of CNF1 in nanomolar (nM) for 6 h, and images were acquired on an Opera Phenix automatic confocal microscope. The lower graph shows the percentage of Rac1-positive cells as a function of the log10 values of CNF1 concentrations. Dots show mean values of n = 3 independent experiments, n = 5 replicates. Linear regression shows the linear relationship (R2 = 0.62) between values of Rac1-positive cells and a range of CNF1 concentrations comprised between 2 and 0.5 nM, with a calculated EC50 value of 0.8 nM. (B) The upper panel shows the principle of the screen designed to identify siRNAs inhibiting CNF1-induced cellular depletion of Rac1. HUVEC transfected with siCtrl and treated with CNF1 set the minimal value of endogenous Rac1 signal (αRac1). HUVEC was transfected with siLu/BCAM or siHACE1, and then intoxicated with CNF1, resulting in maximal protection values of Rac1 signals. The lower panels show representative confocal images of HUVEC transfected with siCtrl, siLu/BCAM, or siHace1 and treated with CNF1 at 1 nM for 6 h, as well as siCtrl cells left untreated. Confocal images acquired on Opera Phenix at objective 20× show endogenous Rac1 immunofluorescence (green) and nuclei (blue). Scale bar, 50 µm. Values show the number of Rac1-positive cells (No. Rac1+cells), total cells (No. total cells), and the calculated ratio of Rac1+ cells. (C) Scatter plot showing the distribution of Rscr values of Rac1-positive cells transfected with 21 reordered siRNAs from the list of top-ranked siRNA (confirmatory screens), as a function of the calculated Z-score of cell number. Cells were treated with CNF1 at 1 nM for 6 h. Each dot corresponds to mean values from n = 3 independent experiments, n = 4 replicates. (D) Immunoblots show endogenous Rac1 signals from cells transfected with siCtrl, siHace1, and siSec61A1 that were left untreated or intoxicated with 1 nM of CNF1 for 6 h. GAPDH shows equal protein loading control. One representative experiment, n = 5. The lower graph shows intensities of Rac1 signals normalized to GAPDH and expressed as fold to the siCtrl condition. Data are mean ± SEM. Each dot corresponds to an individual value from n = 5 independent experiments. One-way ANOVA with Dunnett’s correction; siCtrl+ CNF1 vs siCtrl, ****P < 0.0001; siCtrl+ CNF1 vs siHace1+ CNF1, ***P = 0.0008; siCtrl+ CNF1 vs siSec61A1 + CNF1, *P = 0.0062. (E) Immunoblots show the upper shift that undergoes the deamidated form of RhoA (RhoA Q63E) in CNF1-intoxicated cells, versus wild-type RhoA (RhoA Q63) in Urea-SDS-PAGE condition. The cells were transfected with siCtrl, siSec61A1, siHace1, or siLu/BCAM and left untreated or intoxicated with CNF1 at 1 nM for 6 h. Immunoblot GAPDH shows equal protein loading. One representative experiment, n = 3. (F) Scatter plot showing the distribution of Rscr values of Rac1-positive cells transfected with six siRNAs targeting structural and regulatory subunits of the Sec61 translocon, as well as siLu/BCAM, as a function of the calculated Z-score of cell number. Cells were treated with CNF1 at 1 nM for 6 h. Each dot corresponds to mean values from n = 2 independent experiments, n = 6 replicates.

Chemical inhibition of the Sec61 translocon abrogates CNF1-mediated depletion of Rac1

As an orthogonal approach, we tested whether chemical inhibition of the Sec61 translocon with mycolactone had an inhibitory effect on the depletion of Rac1 in CNF1-treated cells. Cells were incubated with 100 nM mycolactone for 24 h before the addition of the toxin at 1 nM for 6 h. In cells co-treated with mycolactone and CNF1, the level of endogenous Rac1 remained high as opposed to the control condition (Fig. 2A). The proper inhibition of Sec61 translocon activity was monitored by immunoblotting the beta-1 integrin transmembrane protein (Fig. 2A). In parallel, we verified that the treatment of cells with mycolactone at 100 nM did not affect the capacity of CNF1 to deamidate RhoA (Fig. 2B). By conducting dose-dependent experiments, we recorded a significant stabilization of Rac1 signal with concentrations of mycolactone of 25 nM, and above, which efficiently decreased the cellular pool of beta-1 integrin (Fig. 2C and D). Together, these findings extend data from the genetic screen, highlighting the requirement of a functional Sec61 translocon for the proper depletion of Rac1 in CNF1-treated cells.

*Active Sec61 translocon is required for CNF1-induced Rac1 depletion. (A) Immunoblots show endogenous Rac1 immunofluorescence signal intensities from cells treated with siCtrl, siSec61A1, or treated with mycolactone (Myco) at 100 nM for 24 h, and prior to treatment with 1 nM of CNF1 for 6 h. Immunoblot GAPDH shows equal protein loading. An immunoblot anti-beta-1 integrin was performed to control the inhibition of protein biosynthesis in the ER. One representative experiment, n = 3. (B) Immunoblots show the electromobility shift of RhoA Q63E versus RhoA Q63 in cells treated with mycolactone at 100 nM for 24 h (Myco) prior to treatment with 1 nM CNF1 for 6 h. Immunoblot GAPDH shows equal protein loading. One representative experiment, n = 3. (C) The graph shows intensities of Rac1 signals normalized to GAPDH and expressed as fold to the non-treated (NT) condition. Cells were treated for 24 h with mycolactone (Myco) at the indicated concentrations prior to treatment with 1 nM of CNF1 for 6 h. Bars show mean ± SEM. Each dot corresponds to an individual value from n = 3 independent experiments. One-way ANOVA with Dunnett’s correction; NT + CNF1 vs Myco 10 nM + CNF1, ns; NT + CNF1 vs Myco 25 nM + CNF1, *P = 0.035; NT + CNF1 vs Myco 50 nM + CNF1, **P = 0.044; NT + CNF1 vs Myco 75 nM + CNF1, **P = 0.079; NT + CNF1 vs Myco 100 nM + CNF1, **P = 0.029. (D) The graph shows intensities of beta-1 integrin signals normalized to GAPDH and expressed as fold to the non-treated (NT) condition. Cells were treated for 24 h with mycolactone (Myco) at indicated concentrations prior to treatment with CNF1 at 1 nM for 6 h. Bars show mean ± SEM. Each dot corresponds to an individual value from n = 4 independent experiments. One-way ANOVA with Dunnett’s correction; Myco 10 vs NT, ***P < 0.001; Myco 10 vs Myco 25, **P = 0.0019; Myco 10 vs Myco 50, ***P = 0.0005; Myco 10 vs Myco 100, **P = 0.0007.

Functional Sec61 translocon enables CNF1-mediated association of Rac1 to membranes

We investigated whether Sec61A1 plays a role in Rac1 recruitment to cellular membranes upon CNF1 stimulation. Indeed, in cells co-treated with siSec61A1 and CNF1, we observed that Rac1 remained diffusely distributed in the cytosol (Fig. S2). This prompted us to fractionate the cellular membranes from the cytosol and assess the relative distribution of Rac1 in membrane and cytosolic fractions. The efficient enrichment of membranes from cytosolic components was controlled by immunoblotting glyceraldehyde-3-phosphate dehydrogenase (GAPDH), a cytosolic protein, as well as the Rho GDI factor that sequesters the GDP-bound form of Rho proteins in the cytosol (Fig. 3A). Cells were treated with CNF1 at 1 nM for 2 h to promote the activation of Rac1 over its degradation and therefore raise the level of membrane-associated Rac1 (Fig. 3A). As a control, we treated cells with lovastatin, an inhibitor of HMG-CoA reductase, the rate-limiting enzyme in the mevalonate pathway. This treatment blocks the prenylation of Rho proteins, preventing their association with cellular membranes. Consistently, in the condition of lovastatin treatment, we no longer observed a presence of Rac1 in the membrane fraction of cells intoxicated by CNF1 (Fig. 3A). Finally, when CNF1-intoxicated cells were co-treated with siSec61A1 or mycolactone, we observed an inhibition of the targeting of Rac1, as well as RhoA, to membranes (Fig. 3A). We next tested whether inhibition of the Sec61 translocon might interfere with the prenylation of Rac1, considering that HMG-CoA reductase is an integral membrane protein of the endoplasmic reticulum (44). We proceeded with the extraction of the prenylated form of Rac1 from total cell extracts by phase separation with Triton X-114, as previously described (45). Although lovastatin treatment reduced the levels of Rac1 in the detergent fraction, we recorded similar signal intensities of Rac1 between aqueous and detergent fractions in siCtrl and siSec61A1 conditions (Fig. 3B). We concluded that siSec61A1 had no detectable impact on the efficiency of Rac1 prenylation. Altogether, these results establish that a functional Sec61 translocon is required for the targeting of prenylated Rac1 to membranes in CNF1-treated cells.

*Functional Sec61 translocon allows CNF1-induced translocation of Rac1 to membranes. (A) Immunoblots show the relative distribution of Rac1 and RhoA between cytosol (Cyt) and membrane (Mb) fractions in HUVEC intoxicated with CNF1. Cells were pre-treated with siCtrl, siSec61A1, 100 nM mycolactone (Myco), or 15 µM lovastatin (Lova) 24 h prior to intoxication with CNF1 at 1 nM for 2 h. Immunoblots anti-Rho GDI or anti-GAPDH show controls of membrane fractionation and equal loading of cytosolic proteins. Immunoblot anti-beta-1 integrin shows control of the inhibition of protein biosynthesis in the ER. Representative immunoblots of n = 3 independent experiments. Rac1 immunosignals were quantified and are expressed as the signal ratio between membrane and cytosol, together with membrane fractions normalized to NT (graph). Bars show means ± SEM. Each dot corresponds to an individual value from n = 3 independent experiments. One-way ANOVA with Dunnett’s correction, siCtrl vs siCtrl + CNF1, *P = 0.4966; siCtrl + CNF1 vs siSec61A1 + CNF1, ***P = 0.0001; siCtrl + CNF1 vs siCtrl + Myco + CNF1, ***P = 0.0006; siCtrl + CNF1 vs siCtrl + Lova + CNF1, ***P = 0.0007. (B) Immunoblots show the distribution of Rac1 between aqueous (Aq) and detergent (Dt) phases after Triton X-114 nonionic detergent phase separation. HUVECs were transfected with siSec61A1 or siCtrl for 24 h prior to the extraction of prenylated-Rac1 in the Triton X-114 phase. As a control, siCtrl-transfected cells were treated with 15 µM of lovastatin for 24 h (Lova). Immunoblot anti-GAPDH shows control of phase separation. Immunoblot anti-beta-1 integrin shows control of Sec61 translocon inhibition. One representative experiment, n = 4. Rac1 signals were quantified and expressed as signal ratio between the detergent and aqueous phases, normalized to the siCtrl condition (graph). Bars show means ± SEM. Each dot corresponds to an individual value from n = 4 independent experiments. One-way ANOVA with Dunnett’s correction, siCtrl vs siSec61A1, ns; siCtrl vs siCtrl + Lova, **P = 0.0058; siSec61A1 vs siCtrl + Lova, P = 0.0452.

Functional Sec61 translocon enables GTP-loading of Rac1, RhoA, and Cdc42 in CNF1-intoxicated cells

We then assessed whether a functional Sec61 translocon is required for GTP-loading of Rho proteins in cells treated with CNF1. To this aim, we first measured the cellular levels of GTP-bound Rac1 and Cdc42 by pulldown using the p21-PAK effector binding (CRIB) domain fused to GST as bait (46). We first knocked down Hace1 and Lu/BCAM to establish the maximum and minimum levels of activation of Rac1 in cells intoxicated with CNF1 at 1 nM for 2 h, respectively (Fig. 4A). Interestingly, we observed a sharp decrease in the cellular pool of GTP-bound Rac1 when cells were treated with siSec61A1 prior to CNF1 intoxication (Fig. 4A). In these experimental conditions, we also measured a decrease in the cellular pool of GTP-bound Cdc42 (Fig. S3A). Complementary to these findings, we recorded a dose-dependent decrease of GTP-bound Rac1 in CNF1-intoxicated cells treated simultaneously with different concentrations of mycolactone, thereby establishing an EC_50_ value around 25 nM (Fig. 4B). As a control, we recorded a massive inhibitory effect of mycolactone on the expression of beta-1 integrin at a concentration of 25 nM and above (Fig. S3B). We extended these findings by showing that the inhibition of the Sec61 translocon with siSec61A1 or mycolactone in CNF1-treated cells decreased the pool of GTP-bound RhoA (Fig. S3C). Finally, we verified that mycolactone treatment inhibited CNF1-induced Rac1 activation in MCF12A cells, confirming the requirement of Sec61 in a different cell type (Fig. S3D). Together, these data establish that Sec61 translocon activity is essential for increasing the cellular levels of GTP-bound Rac1 in CNF1-treated cells.

*Active Sec61 translocon allows CNF1-mediated activation of Rac1. (A) Representative immunoblots show levels of GTP-bound Rac1 (Rac1-GTP) in cells transfected with siCtrl, siHace1, siLu/BCAM, or siSec61A1 prior to intoxication with CNF1 at 1 nM for 2 h. Levels of GTP-bound Rac1 were assessed by GST-PAK pulldown. Immunoblots of anti-GAPDH and anti-Rac1 on total cell lysates show equal protein amounts engaged in the GST-PAK pulldown. Graph: Rac1 immunosignals were quantified and are expressed as the ratio of GTP-bound Rac1 normalized to the total Rac1 signal. Bars show means ± SEM. Each dot corresponds to a value from n = 3 independent experiments. One-way ANOVA with Dunnett’s correction, siCtrl vs siLu/BCAM, ***P = 0.0041; siCtrl vs siSec61A1, ***P = 0.0019. (B) Graph shows the variation of GTP-bound Rac1 (Rac1-GTP) levels normalized to Rac1 total level (Rac1) in cells left untreated (NT) or treated for 24 hours with mycolactone (Myco) at indicated concentrations together with CNF1 at 1 nM. Data are expressed as fold variations compared with cells treated with CNF1. Levels of Rac1-GTP were assessed by GST-PAK pulldown and anti-Rac1 immunoblotting. Bars show means ± SEM. Each dot corresponds to a value from n = 2 independent experiments. One-way ANOVA with Dunnett’s correction, NT vs Myco 10, ns; NT vs Myco 25, **P = 0.0089; NT vs Myco 50, **P = 0.002; NT vs Myco 100, *P = 0.0012.

Inhibition of protein N-glycosylation abrogates the CNF1-mediated activation of Rac1

The ER is a defined compartment that fulfills critical physiological functions as a site of membrane and secretory protein synthesis, and post-translational modifications, also including quality control mechanisms that ensure the exit of properly folded proteins. The unfolded protein response (UPR) sensors monitor the folding status of neosynthesized proteins in the ER to adjust the flux of protein synthesis and the fate of cells. The UPR encompasses three established pathways, for example, the protein kinase RNA-like ER kinase (PERK)/ATF4, the IRE1alpha/XBP1, and ATF6 (47). The ER-resident transmembrane proteins PERK, IRE1alpha, and ATF6 act as sensors of protein unfolding and transducers via the post-translational modifications of cytosolic proteins or induction of genetic programs in the nucleus. Mycolactone is a known ER stressor that launches adaptive cell reactions to UPR involving PERK and IRE1alpha (48–50). This prompted us to test whether different ER stressors might inhibit the activation of Rac1 in CNF1-treated cells. To this end, cells were treated with thapsigargin (Tg) or tunicamycin (Tn) that trigger ER stresses via inhibition of endoplasmic reticulum Ca^2+^-ATPase SERCAs or the inhibition of N-linked glycosylation of neosynthesized proteins in the ER, respectively (51, 52). HUVECs were treated for 24 h with mycolactone, Tn, or Tg, followed by intoxication with CNF1 at 1 nM for 2 h to raise the cellular pool of GTP-bound Rac1. The induction of ER stress responses was monitored by recording the expression of ATF4 (Fig. 5A). In parallel, we observed a decrease in the pool of GTP-bound Rac1 in CNF1-intoxicated cells co-treated with Tn or mycolactone (Fig. 5B). Conversely, the ER stressor Tg had no impact on the pool of GTP-bound Rac1 in CNF1-treated cells, whereas it induced a strong expression of ATF4 (Fig. 5A and B). We verified that CNF1 induced a proper deamidation of RhoA in cells treated with Tg, Tn, and mycolactone (Fig. S4). Taken collectively, our data point to the importance of N-glycosylation of the flux of neosynthesized protein in controlling the pool of GTP-bound Rac1 in CNF1-treated cells.

*ER stress induced by tunicamycin blocks CNF1-mediated activation of Rac1. (A) Representative immunoblots show GTP-bound Rac1 (Rac1-GTP) levels in cells treated with ER stressors. HUVECs were treated for 24 h with 2 µM thapsigargin (Tg), 2 µM tunicamycin (Tn), or 100 nM mycolactone (Myco) prior to the addition of CNF1 at 1 nM for 2 h. The fraction of GTP-bound Rac1 was purified by GST-PAK pulldown. Immunoblots anti-GAPDH and anti-Rac1 show equal protein amounts engaged in GST-PAK pulldowns. Immunoblots anti-beta-1 integrin and anti-ATF4 show Sec61 translocon inhibition and induction of PERK-dependent ER stress signaling, respectively. Representative immunoblots, n = 3 independent experiments. (B) Graph shows levels of GTP-bound Rac1 normalized to GAPDH and expressed as fold inhibition of Rac1 GTP compared with untreated cells. Bars show means ± SEM. Each dot corresponds to an individual value from n = 3 independent experiments. One-way ANOVA with Dunnett’s correction, NT vs Tg, ns; NT vs Tn ***P = 0.0002; NT vs Myco **P = 0.0006.

Crosstalk between Rac1 Q61E depletion and regulators of ER and Golgi vesicular trafficking

The above data prompted us to search for a possible interference of components of the ER and Golgi secretory pathway in CNF1-mediated depletion of Rac1. To this aim, we performed a second-pass reanalysis of the best-ranked 260 siRNA hits of the primary screen (Rscr > 0.6) for their annotated function in vesicular trafficking, including components of the ER and Golgi secretory pathway (Materials and Methods). This ontology-based analysis led to selecting 22 siRNAs. These siRNAs were reordered and ranked according to the average of Rscr values defined in three independent experiments (Table 2). In total, 77.2% of the siRNA (n = 17) passed the confirmatory tests with recorded values of Rscr above 0.1 (Table 2). Note that assigned values of protection defined in this secondary screen further confirmed the major inhibitory effect triggered by siSec61A1 on the CNF1-induced cellular depletion of Rac1. The first seven siRNAs showed no impact on CNF1-induced deamidation of RhoA, pointing to an absence of effect on CNF1 action. Consistent with our hypothesis of an interplay between the flux of protein biosynthesis in the ER and Rac1 signaling, we found five siRNAs among the first seven hits that displayed a function assigned to ER and Golgi trafficking and homeostasis (Table 2). In conclusion, we identified siRNAs targeting genes having a function assigned to ER and Golgi homeostasis and trafficking for their inhibitory effect on CNF1-induced depletion of Rac1.

DISCUSSION

By conducting a genome-wide screen, we highlight gene candidates involved in CNF1-induced cellular depletion of Rac1. Most notably, the implementation of a quantitative approach highlights the critical role of the Sec61 translocon in CNF1-mediated activation of Rac1. The importance of the Sec61 translocon machinery as a whole is supported by the recorded inhibitory impact on CNF1-mediated depletion of Rac1 of siRNAs targeting core and regulatory components of the Sec61 translocon. Moreover, we establish with mycolactone the requirement of Sec61 translocon activity on CNF1-mediated depletion of Rac1. We did not identify the precise molecular circuit integrating the activity of Sec61 translocon in Rac1 signaling. Nevertheless, we show that both the knockdown and chemical inhibition of Sec61A1 inhibit the GTP-loading of Rac1 and its targeting to membranes downstream of Rac1 prenylation. Moreover, we report that the activation of Rac1 in CNF1-treated cells requires the proper N-glycosylation of neosynthesized proteins consistent with the idea that the GTP-loading of Rac1 is subjected to a control by the flow of protein biosynthesis in the ER. This idea is enforced by our findings of a panel of genes involved in ER/Golgi trafficking and homeostasis that interfere with the CNF1-mediated depletion of Rac1. Collectively, these data open new avenues in studying the crosstalk between the regulation of Rac1 and compartmentalized biosynthesis of integral membrane and secreted proteins at the endoplasmic reticulum.

The CNF1 toxin represents an example of an environmental factor that catalyzes a somatic alteration of endogenous Rho proteins that freeze them into a GTP-bound state able to transduce signals. Our data strongly suggest that the activity of the Sec61 translocon, a gatekeeper of transmembrane protein synthesis, controls the activation of deamidated forms of Rho proteins. Indeed, we show that the inhibition of Sec61 translocon uncouples the deamidation of RhoA catalyzed by CNF1 from its GTP-loading and membrane anchoring. This suggests that activation of endogenous RhoA Q63E, generated by the intoxication of cells with CNF1, involves GEF factors that could modulate the effects of the toxin according to cell types and metabolic status. We provide evidence of a broad control operated by Sec61A1 on the activation of all three Rho proteins. The Rho GDP-dissociation inhibitor 1 (Rho GDI) is a shared regulator between Rac1, Cdc42, and RhoA that sequesters them as GDP-bound forms in the cytosol, thus protecting them against the UPS (53). Here, we found no detectable alteration of the prenylation of Rac1 in siSec61A1-treated cells when intoxicated by CNF1, thereby pointing to the possible interplay between Rho GDI and Sec61 translocon. Although this hypothesis is not supported by the absence of an increase in Rho GDI expression in cells, where the Sec61 translocon is inhibited, we cannot rule out the possible involvement of posttranslational modifications of Rho GDI in this process that would block the displacement of Rho from Rho GDI (54). Our results establish a hierarchical control of the activity of Rac1 GTPase by the Sec61 translocon, which points to a possible involvement of ER stress. Although both Tn and Tg induce ER stress, we show that only tunicamycin treatment blocks CNF1-induced depletion of Rac1. Thus, the control exerted by Sec61 translocon on Rac1 signaling most likely occurs independently of a general impact on ER stress.

We have developed a quantitative approach to measure the percentage of Rac1-positive cells in the whole cell population, allowing us to rank siRNAs as a function of the level of protection they confer against CNF1-mediated cellular depletion of Rac1. The reported lists of confirmed hits, including the 21 top-ranked hits and the best 22 hits having an annotated function in vesicular trafficking, provide information for future studies aimed at assessing the central role played by Rac1 and HACE1 in cell biology, cancer, and host-pathogen interactions. These data also pinpoint the importance of a group of factors involved in ER and Golgi trafficking and homeostasis, encompassing PITPNB, TMED9, SEL1L, SLC35E1, and GBF1 (Table 2). Whether these factors control the activation of Rac1 or its proteasomal degradation remains to be clarified. Nevertheless, we show that none of these hits had an impact on the deamidation of RhoA, e.g. able to impair CNF1 endocytosis and/or translocation to the cytosol. The shortlist of hits involved in vesicular trafficking also encompasses SNX4 (sorting nexin 4) and the small GTPase Rab11a, thereby pointing to the importance of vesicular recycling in CNF1-induced degradation of Rac1. Indeed, Rab11a can associate with HACE1 and is a known substrate of the E3 ligase HACE1 in the regulation of beta-adrenergic receptor recycling (28, 55). This raises the question of whether aberrant relocation of GTPase-impaired GTP-bound Rac1 Q61E, in recycling compartments, for example, accounts for its increased susceptibility to UPS. The list of confirmed hits also includes the component of endoplasmic reticulum-associated degradation (ERAD) SEL1L (suppressor/enhancer of Lin-12-like), which supports the dislocation of misfolded proteins from the luminal part of the ER to the cytosol for degradation. We cannot formally exclude the involvement of the ERAD machinery in CNF1-induced depletion of Rac1. Nevertheless, such regulation of Rac1 would involve a Sec61-independent pathway since the inhibition of Sec61 has no impact on ERAD (56). Alternatively, SEL1L may control Rac1 signaling in CNF1-treated cells via its reported influence on ER homeostasis and stress (57). Finally, we failed to identify regulatory components of the endolysosomal pathway in the list of best hits, except for Lu/BCAM. This points to technical limits in the screening method, although we had to exclude data from six plates that did not meet the quality control criteria. Thus, hits related to vesicular trafficking of CNF1 might be identified by performing this type of screen more than once. On the other hand, we found that the number of cells remaining after siRNA transfection, performed in the absence of serum in our study, had an impact on defining protection scores. This resulted in a counter-selection of siRNAs promoting cell rounding and detachment. Finally, in a similar screening method, we have previously isolated a small chemical inhibitor of the sorting of CNF1, which acts at the level of matured early endosomes (14). This raises the hypothesis that the absence of detection of siRNAs targeting components of the endolysosomal transport of CNF1 here reflects a redundancy in the activity of regulatory factors controlling the intracellular trafficking of the toxin.

Our data point to the importance of an underappreciated connection between protein biosynthesis in the ER and the regulation of Rac1 small GTPase.

MATERIALS AND METHODS

Cell culture, reagents, and antibodies

Primary human umbilical vein endothelial cells (HUVEC) up to six passages (PromoCell) were routinely cultured on gelatin 0.2% (Sigma) pre-coated dishes at 37°C, 5% CO_2_ in human endothelial SFM medium (Invitrogen) supplemented (SFMc) with 20% fetal bovine serum (FBS), 20 ng/mL fibroblast growth factor-2 (FGF-2) (PeproTech), 10 ng/mL epidermal growth factor (EGF) (PeproTech), and 1 mg/mL heparin (Sigma-Aldrich). MCF-12A mammary gland epithelial cells (CRL-10782, ATCC) were cultured in DMEM/F12 supplemented with 5% horse serum (Biowest), recombinant human epidermal growth factor (20 ng/mL) (Peprotech), human recombinant insulin (10  µg/mL) (Life Technologies), hydrocortisone (0.5  µg/mL) (Sigma-Aldrich), and cholera toxin (100 ng/mL) (Sigma-Aldrich). The primary and secondary screens were conducted with HUVEC at passage three in EGM-2 supplemented medium (Lonza) lacking VEGF and antibiotics, which otherwise interfere with the transfection efficiency. Recombinant CNF1 toxin was produced in E. coli and purified, as described in (58). Cells were intoxicated with CNF1 at the indicated time and concentrations in SFMc medium at 37°C, 5% CO_2_. The mycolactone was prepared from Mycobacterium ulcerans strain 1615 ATCC35840, as described in references 50, 59. Mycolactone was stored in DMSO at −20°C in amber glass tubes. The toxin was diluted extemporaneously in SFMc. Chemical inhibitors used in this study comprise 15 µM lovastatin (Enzo Life Sciences, #BML-G226), 2 µM thapsigargin (Enzo Life Sciences, #BML-PE180), and 2 µM tunicamycin (Tocris Bioscience, #3516). For immunoblots, the primary antibodies used were anti-GAPDH clone 411 (Santa Cruz, sc-47724, 1:5,000), anti-Rac1 clone 102 (BD Bioscience, #610651, 1:2,000), anti-RhoA clone 26c4 (Bertoglio laboratory, 1:2,500), anti-Cdc42 (Santa Cruz, SC-8401), anti-beta-1 integrin clone 18 /CD29 (BD Transduction Laboratories, #610468, 1:1000), anti-Rho GDI clone G-2 (Santa Cruz, sc-373724, 1:15000), and anti-ATF4 clone D4B8 (Cell Signaling, #11815, 1:1,000). The 4',6-diamidino-2-phenylindole (DAPI) was purchased from Sigma-Aldrich (#B9542).

High content screening

In total, 384-well μclear black plates (Greiner Bio-One, #781091) were coated with fibronectin at 15 µg/mL (BD Bioscience, #356008) for 2 h at 37°C and dried overnight. Volumes of 250 nL of siRNA at 10 µM from siGENOME SMARTpool siRNA (Dharmacon, G-004675-E2 Lot 11157) or control siRNA (siCtrl) (Eurogentec, SR-CL000-005) and siHACE1 (Santa Cruz, sc-95301) were dispensed in a total of 66 plates using Echo550 Liquid Handler (Labcyte). The screen was performed in three runs of 22 plates each. A volume of 10 µL of an Opti-MEM Reduced Serum Medium (Gibco) containing 0.325 µL of HiPerfect transfection reagent (Qiagen) was added to each siRNA-containing well using a Microplate Washer Dispenser EL406 (Biotek). After 20 min at room temperature (RT), a volume of 40 µL of EGM-2 medium (Lonza) containing 2,000 HUVEC was added to each well using the same dispenser. Plates were then incubated at 37°C in 5% CO_2_ for 48 h. The next steps were conducted on the laboratory automation platform, Biocel System, operated with VWorks software (Agilent Technologies). The medium was removed before intoxicating cells with CNF1 at 1 nM in a volume of 50 µL of SFMc for 6 h. Cells were washed once with 50 µL of PBS before fixation in PBS containing 4% paraformaldehyde for 20 min at RT. Cells were washed in PBS and stained as follows. Briefly, cells were permeabilized in 50 µL of 0.5% Triton X-100 in PBS containing 0.1% BSA (Sigma) for 1 h. Each washing step was performed with 50 µL of PBS. Cells were incubated for 1 h with 30 µL of anti-Rac1 monoclonal antibodies (1:500). Cells were washed twice in PBS and incubated with secondary donkey anti-mouse antibodies coupled to AlexaFluor 488 (Alexa-488) (Invitrogen) (1:800) together with 1 µg/mL of 4',6-diamidino-2-phenylindole (DAPI). The plates were washed twice with PBS. Confocal images were taken using the automated confocal microscope IN Cell Analyzer 6000 (GE Healthcare) equipped with a dry objective 20×. Images of four fields were acquired for each well (n = 650 cells per well approximately). DAPI-stained nuclei were detected using 405 nm laser lines with a 450/50 nm emission filter. Endogenous Rac1 signal intensities were recorded using 488 nm laser lines with 540/75 nm emission filters. Images were analyzed by a multi-parameter approach using the Columbus image analysis software (Perkin Elmer, version 2.3.1) and the algorithm reported in Figure S4.

Data formatting and normalization

We implemented a quantitative method of immunofluorescence signal processing to rank the protective effect of each siRNA on CNF1-induced depletion of Rac1. Cell segmentation was achieved based on the detection of DAPI signal intensities between nuclei (high intensity) and the cytoplasm (low intensity) using an algorithm implemented on Columbus software (Fig. S5). We next extracted quantitative information on Rac1 signal per cell corresponding to Alexa-488 signal intensities in segmented cell cytosol to define the number of Rac1-positive cells within the cell population. Thresholds set to define Rac1-positive cells for each siRNA correspond to mean (a.u. > 800) and minimum (a.u. > 300) Rac1 immunofluorescence intensities in siCtrl-transfected cells left untreated versus treated with CNF1. The quality control of the screen used the strictly standardized mean difference (SSMD) parameter based on positive (Posctrl, siCtrl, n = 10 wells) and negative (Negctrl, siCtrl + CNF1, n = 10 wells) controls per plate. In total, we removed from data analysis six plates with SSMD values >−5. Percentages of Rac1-positive cells defined in each condition were normalized plate by plate, giving a Rscreenorm score (Rscr), as described in (42). All siRNAs were ranked according to their values of Rscr, using the open-source R software (https://www.r-project.org):

$$[eqn]$$

Confirmatory screens

A total of 21 siRNAs from the list of top-ranked siRNA were re-ordered as a pool of 4 siRNAs (Dharmacon) and quantified for their efficacy against CNF1-mediated degradation of Rac1, as described in the primary screen. We have also short-listed 22 siRNAs targeting genes among the first 260 hits with scores above >0.6 and having an annotated function in vesicular trafficking. These siRNAs were re-ordered and screened for their efficacy against cellular depletion of Rac1 induced by CNF1, as described in the primary screen. For confirmatory screens, each siRNA confirmation was performed at least in triplicate in 384-well plates from three independent experiments.

Transfection procedure

HUVECs were seeded at sub-confluence. The next day, transfection was carried out with 100 nM of siRNAs prepared in a mix of serum-free OptiMEM (Gibco) and PolyMag reagent (OZ Biosciences). The mix was incubated at RT for 20 min and then added to the cells. The dishes were then placed on a magnetic plate for 15 min at 37°C. After incubation time, the cells were left in the incubator for 1 h; then, the medium was exchanged for SFMc. Cells were next incubated at 37°C for 48 h. The list of the siRNAs used in this study is described in Table S2. The reverse transfection in 384-well plates used conditions of the high-throughput screen.

Western blot analysis

Cells were lysed in 2× Laemmli buffer (Sigma-Aldrich) and boiled for 10 min at 100°C. Samples were resolved on 4–12% SDS-polyacrylamide gels and transferred to 0.45 µm PVDF-nitrocellulose membranes (GE Healthcare). Proper protein transfer was monitored with Ponceau S (Biorad) prior to blocking with 5% milk in Tris-buffer saline (TBS), Tween-20 at 0.005% (TBS-T) (Euromedex) for 1 h. Membranes were washed three times in TBS-T and incubated for 2 h with primary antibodies. Membranes were washed with TBS-T and incubated with anti-mouse or anti-rabbit horseradish peroxidase (HRP)-conjugated secondary antibodies (Dako) for 1 h. Signals were revealed using Immobilon Western Chemiluminescent HRP Substrate (Merck) on iBright CL1500 (Thermofisher). Protein signals were quantified using Image J software V2.3.1.

Pulldown of active Rho GTPases

Pulldowns of the GTP-bound active form of Rho GTPases were conducted as described in (58). HUVECs treated with siRNA or pharmacological inhibitors were intoxicated with CNF1 toxin for 2 h. Dishes were washed in PBS and lysed in the Rac lysis buffer (25 mM HEPES, 150 mM NaCl, 5 mM MgCl_2_, 0.5% Triton X-100, 4% glycerol, 20 mM beta-glycerophosphate, and 10 mM NaF, pH 7.4) supplemented with Pierce EDTA-free protease inhibitors (Thermo Scientific, A32965). Cells were lysed through a 25G needle, and lysates were centrifuged at 12,000 × g for 10 min at 4°C to remove debris. Cell lysates were incubated with 30 µg of recombinant GST-PAK^70-106^ (46)-coated agarose beads for 40 min at 4°C. Beads were washed three times in Rac lysis buffer by centrifugation at 1,500 × g, 4°C for 5 min. Beads were resuspended in 50 µL of 1× Laemmli buffer. Proteins were resolved on a 4%–12% SDS-PAGE, and Rac1 was visualized by anti-Rac1 immunoblotting. The same protocol was followed to prepare cell lysates for active RhoA pulldown with the following modifications: cells were lysed in RhoA lysis buffer (50 mM Tris-HCl, 500 mM NaCl, 10 mM MgCl_2_, 1% Triton X-100, pH 7.4) freshly supplemented with 0.5% sodium deoxycholate, 0.1% SDS, and Pierce EDTA-free protease inhibitors. Cell lysates were incubated with 100 µg of recombinant GST-Rhotekin RBD-coated agarose beads (60).

Membrane cytosol fractionation

A total of 4 × 10^6^ HUVECs were seeded in 100 mm plastic dishes the day before the experiment. Cells were scraped in 5 mL cold PBS, pelleted, and resuspended in 200 µL cold SI (250 mM sucrose, 3 mM imidazole, pH 7.4) buffer. Cells were lysed on ice by passing them 40 times through a 25G needle. Nuclei and cell debris were removed by centrifugation at 12,000 × g for 10 min at 4°C. The volume of each post-nuclear supernatant (PNS) was normalized if needed for ultracentrifugation at 100,000 × g for 1 h at 4°C. The supernatants were collected, while pellets corresponding to membrane fractions were homogenized in an equal volume of SI buffer. An equal volume of 20 µL of each fraction was resolved on SDS-PAGE and immunoblotted for Rac1, RhoA, Rho GDI, GAPDH, and beta-1 integrin.

Triton X-114 phase separation

The prenylation of Rho GTPases was analyzed by Triton X-114 partition, as described in (45). Briefly, 3.5 × 10^6^ HUVEC were washed twice in PBS and detached in PBS EDTA 3 mM. After centrifugation, the cell pellet was resuspended in 0.4 mL PBS supplemented with 2% Triton X-114 buffer and incubated for 30 min at 4°C. Nuclei were removed by centrifugation at 16,000 × g, 4°C for 2 min. Supernatants were next incubated at 37°C for 10 min before centrifugation at 22,000 × g for 10 min at RT to allow efficient phase separation. The upper aqueous phase was collected. The lower detergent phase was washed with equal volumes of 0.1% Triton X-114 PBS lysis buffer in three cycles of phase separation, as described above. Both fractions were precipitated by adding methanol/chloroform/H_2_O (4:1:3). Air-dried pellets were then resuspended in 60 µL of 1× Laemmli buffer and protein resolved on 4–12% SDS-PAGE.

RNA extraction and qRT-PCR

RNAs were extracted from HUVEC using the RNeasy Mini Kit (Qiagen, 74104) according to the manufacturer’s procedure. Reverse transcription was carried out with 1 µg of total RNA material and SuperScript IV Reverse Transcriptase (Invitrogen, #18090050). Quantitative PCR was performed with Master Mix PCR Power SYBR Green (Applied Biosystems, #4367659) and primers listed below. Conditions of qPCR were set as follows: 2 min at 50°C, 10 min at 95°C, 15 s at 95°C (40 cycles), and 1 min at 60°C on QuantStudio 7 Real-Time PCR (Thermo Fisher Scientific). The relative quantification was calculated by the 2^−ΔΔCt^ method with GAPDH as the housekeeping gene. Oligonucleotides used in this study were purchased from Sigma: for SEC61A1, 5′-GGACCGCTATCACCCTCTTTA and 5′-TCCATCAATGTGCCTCTGTTAGA; for GAPDH, 5′-AGGTGAAGGTCGGAGTCAACG and 5′- AGGGGTCATTGATGGCAACA.

Statistical analysis

Statistical tests were performed using GraphPad Software Prism 9.2. All data are presented as the mean ± SEM. Detailed information on the number of replicates is provided in the figure legends. P values were calculated using one-way ANOVA with corrections for multiple comparisons or paired two-tailed t-test, as indicated in the figure legends. Significance levels are indicated as follows: not significant (ns) for P >  0.05, *P  ≤  0.05, **P  ≤  0.01, ***P  ≤  0.001, ****P  ≤  0.0001. Exact P values, when possible, are provided in the figure legends.