Golgi screen identifies the RhoGEF Solo as a novel regulator of RhoB and endocytic transport

The control of intracellular membrane trafficking by Rho GTPases is central to cellular homeostasis. How specific guanine nucleotide exchange factors and GTPase‐activating proteins locally balance GTPase activation in this process is nevertheless largely unclear. By performing a microscopy‐based RNAi screen, we here identify the RhoGEF protein Solo as a functional counterplayer of DLC3, a RhoGAP protein with established roles in membrane trafficking. Biochemical, imaging and optogenetics assays further uncover Solo as a novel regulator of endosomal RhoB. Remarkably, we find that Solo and DLC3 control not only the activity, but also total protein levels of RhoB in an antagonistic manner. Together, the results of our study uncover the first functionally connected RhoGAP‐RhoGEF pair at endomembranes, placing Solo and DLC3 at the core of endocytic trafficking.


| INTRODUCTION
Intracellular trafficking relies on a tight cooperation between cytoskeletal elements, proteins and lipids that come together to regulate cellular homeostasis. [1][2][3] While the contribution of the small GTPases of the Rab and Arf families to membrane trafficking is well established, the involvement of Rho GTPases in this process is less understood. Since aberrant membrane trafficking is known to be associated with various diseases including cancer, it is important to elucidate the molecular mechanisms underlying the control of Rho GTPase signalling at different target membranes.
Rho GTPases function as molecular switches cycling between an active GTP-bound state and an inactive GDP-bound state. RhoGEFs promote the exchange of bound GDP for GTP, leading to Rho activation and initiation of downstream signalling pathways. RhoGAPs promote the low intrinsic GTP hydrolysis activity of the Rho GTPase, thereby dampening downstream signalling. 4,5 In addition to the plasma membrane, active Rho GTPase pools are also found at endomembranes such as endosomes and the Golgi complex. 2,3 For example, RhoB is detected at endosomes from where it controls the recycling of cargoes such as the epidermal growth factor receptor (EGFR). 3,6,7 Notably, RhoB pools localized at the plasma membrane and endosomes appear to be functionally distinct, the latter playing the main role in EGFR recycling. 7 Additionally, RhoB and RhoA are important for the architecture of the Golgi complex, their hyperactivation causing severe fragmentation of this organelle. 8 With 145 members, the RhoGEF and RhoGAP proteins greatly exceed the 10 classical Rho family GTPase switch proteins. 2,[9][10][11] This complexity underscores the need for unbiased approaches to identify and characterize GEFs and GAPs co-regulating Rho GTPase signalling in distinct cellular contexts.
Among the RhoGAP proteins, the Deleted in Liver Cancer (DLC) family stands out, being deregulated in different cancer types more frequently than any other Rho regulator. 12,13 DLC3, also known as STARD8, is the least characterized member of the DLC family. 10 Work from our lab has uncovered a RhoGAP-specific function of DLC3 in the maintenance of the integrity of the Golgi complex and the endocytic recycling compartment. 14,15 Furthermore, in HeLa cells, DLC3 partially co-localized with endosomal RhoB and regulated trafficking in a RhoA/B dependent manner. 14 The RhoGEF counterpart of DLC3 responsible for Rho activation at endomembranes has nevertheless remained elusive to date.
By performing an image-based RNAi screen using the morphology of the Golgi complex as a readout, we here identify the RhoGEF protein Solo, also known as ARHGEF40 or Scambio, as an antagonist of cellular DLC3 activity. Subcellular fractionation, imaging and optogenetic recruitment experiments further revealed Solo to be present on a subset of RhoB-positive endosomes, where it can locally activate this GTPase. In addition to activity, we unexpectedly find that Solo and DLC3 co-regulate RhoB protein levels. Together, the results of our study uncover the first functionally connected RhoGAP-RhoGEF pair at endomembranes, placing Solo and DLC3 at the core of RhoB regulation and endocytic trafficking. supplemented with 10% foetal calf serum (FCS; PAA Laboratories). The cells were maintained in a humidified incubator at 37 C with 5% CO 2 . All cell lines were authenticated, tested negative for Mycoplasma (Lonza, LT07-318) and were kept in culture for no longer than 2 months. HRP-labelled secondary anti-mouse-IgG, anti-rabbit-IgG and antigoat-IgG were purchased from Dianova (Cat# 115-035-062, Cat# 111-035-144, Cat# 705-035-147). Alexa-Fluor ® -labelled secondary IgG antibodies, Alexa-Fluor ® -labelled phalloidin and TO-PRO-3 were obtained from Invitrogen. DAPI was obtained from Sigma-Aldrich.

| Generation of a polyclonal anti-Solo antibody by rabbit immunization
Protein epitope analyses including sequence comparisons were carried out to avoid cross-reactivity. The selected peptide (1015-1032aa of human ARHGEF40, isoform 1, UniProtKB Q8TER5-1) was synthesized and separated by HPLC for subsequent immunization of three rabbits (Pineda Antibody-Service). The pre-immune and resulting immunization sera were analysed by western blotting, periodically over a total time span of 3 months, which included monthly rounds of immunization. To compare the quality, specificity and affinity of the sera collected from the three animals, lysates of HeLa cells with either Solo overexpression or Solo knockdown were used (see Figure S2B). The IgG fraction of the antibody generated was separated by affinity chromatography (Pineda Antibody-Service) and aliquots were stored at À80 C for subsequent use.  Table S1. For plasmid transfections of HeLa and HEK293T cells, LipofectamineLTX (ThermoFisher Scientific) and  regents, respectively, were used according to manufacturer's instructions.

| RNAi Golgi screen and analysis
The RhoGEF screen was performed using a WiScan Hermes High content imaging system (Idea Biomedical). To this end, 5 Â 10 4 HeLa cells were seeded per well in CollagenR coated glass bottom 96-well plates (Greiner). The cells were subjected to reverse transfection using 2.5 pmol siRNA consisting of siDLC3sp and siRhoGEF or siNT control mixed at a 1: 1 ratio. 72 h post-transfection, the cells were washed with PBS, fixed and stained for the Golgi complex (Giantin, Alexa Fluor ® 488, green channel) and the nuclei (TO-PRO-3, far red channel) as described in the immunofluorescence staining section. Images were acquired with the high resolution 40Â 0.75 NA objective for green and red channels. Per well, a coverage of 85% and a field density of 20% was applied, resulting in 200 frames. Images were analysed with the implemented WiSoft Minerva analysis application development platform: In brief, nuclei and Golgi complexes were segmented using the corresponding channels. Cells were segmented related to nuclei and Golgi compartments using the cytoplasmic background signal of the Golgi staining. To increase the accuracy, cells touching the image edge, mitotic cells and cells with oversegmented nuclei were excluded from the analysis. The number and average perimeter of Golgi fragments were measured per cell in order to quantify the state of Golgi fragmentation. The screen was performed twice.
For statistical analysis, the number of Golgi vesicles and average perimeter of Golgi vesicles a log-transformation was applied, as the data show a Poisson distribution and strongly skewed normal distribution, respectively. The mean μ log and standard deviation σ log were calculated from the log-transformed data and back-transformed to obtain the mean and variance in normal space as μ ¼ e μ log þ0:5σ 2 log , following the Finney estimator approach. 16 The data were statistically analysed using a N-way ANOVA (NANOVA) method in Matlab (Mathworks, R2021a). The two p-values obtained from the two screenings were combined as: with the two weights where N is the relative number of observables at each screening.

| Quantitative real-time PCR
Total RNA was isolated using the NucleoSpin ® RNA Kit (Macherey-Nagel) according to manufacturer's instructions. 100 ng RNA were used for real-time PCR with the Power SYBR ® Green RNA-to-C T ™     Images were acquired at 37 C and 5% CO 2 every 20 s for a time interval of 10 min. Image processing was done with Zen 2.3 blue software.

| Setup of optogenetics experiments used to validate the OptoEndo-Solo-GEF tool
Samples were illuminated with 405 nm light in a custom-made box equipped with six equally spaced high-power 3W LEDs 700 mA. Illumination cycles were 5 sec on À 35 sec off. Following 90 cycles of illumination, cells were fixed with 4% (w/v) paraformaldehyde. This was followed by permeabilization for 5 min with 0.1% (v/v) Saponin in PBS before immunofluorescence staining for EEA1, as described above. Dark control samples were not exposed to light and were fixed directly after being taken out of the incubator.

| Setup and analysis of live cell imaging optogenetics experiments
HeLa cells were seeded onto collagen-coated 35

| Solo is required for the integrity of the Golgi complex in a RhoGEF-dependent manner
To validate Solo as a regulator of Golgi complex morphology, we depleted the transcript using two independent siRNAs and confirmed the efficient knockdown at the RNA ( Figure S1-figure supplement 3A) and protein level using a Solo-specific antibody generated inhouse ( Figure 1D; Figure Figure S1supplement 1-3 and Tables S1-S3 By contrast, the GEF inactive Solo was distributed diffusely throughout the cytoplasm ( Figure 1G, crop-ins). Since the Solo antibody did not show a specific signal in immunostainings of HeLa cells, we employed subcellular fractionation to monitor the distribution of endogenous Solo in the cytosolic vs. membrane fraction, containing e.g. endosomes and the plasma membrane 29 ( Figure 1I). Calnexin, which was used as a marker for the membrane fraction showed the expected extraction pattern and validated our approach. Interestingly, endogenous Solo was predominantly detected in the membrane fraction together with the small GTPase RhoB but not RhoA. We therefore investigated if RhoB contributes to the Golgi fragmentation phenotype induced by Solo. As shown in Figure 1J,K, depletion of RhoB was sufficient to revert the Golgi phenotype of Solo transfected cells. Together, these findings consolidate the results of the screen and suggest that Solo regulates the morphology of the Golgi through RhoB.

| Solo activates RhoB on endosomes
A characteristic of small GTPases is to bind to their GEFs in a GDPbound or nucleotide-free state. By contrast, effectors bind stronger to the GTP-loaded form. Since Solo has not been previously linked to RhoB, we first performed GFP pulldown assays using lysates from We then used RBD pulldowns to directly assess how endogenous RhoB activity is regulated by Solo and DLC3. Before lysis, the cells were stimulated with EGF, which is a potent RhoB activator. 32 These assays revealed a 40% reduction in the total cellular GTP-RhoB levels (normalized to GAPDH) upon Solo depletion. Intriguingly, this decrease in RhoB-GTP was paralleled by a significant drop in RhoB protein levels ( Figure 3D,E), which was not observed for RhoA or RhoC ( Figure S3-figure supplement 2A). We also measured RhoB protein levels in HeLa cells depleted of GEFH1, one of the few established GEFs for RhoB. 33 No changes in RhoB amounts were observed under these conditions, indicating that the downregulation of RhoB is specific to Solo and not a general response to global perturbations of RhoB GEF activity ( Figure S3-figure supplement 2C). Importantly, simultaneous knockdown of Solo and DLC3 rescued RhoB-GTP as well as total RhoB amounts to levels that were comparable to those of control cells ( Figure 3D,E). While normalization of RhoB-GTP to RhoB-GDP amounts showed no net changes in RhoB activity among the different knockdowns ( Figure 3F), these results clearly demonstrate that Solo and DLC3 co-regulate RhoB in HeLa cells. These effects were predominantly observed at the post-transcriptional level as qPCR analysis did not reveal significant changes in RhoB transcript levels in response to neither Solo nor DLC3 depletion ( Figure 3G).
An intact endosomal trafficking system was found to be required for proper RhoB subcellular localization and the regulation of RhoB stability. 30,34 When visualizing endogenous RhoB using a specific RhoB antibody ( Figure S3-figure supplement 2C), we noted that F I G U R E 2 Legend on next page.
RhoB was concentrated in the perinuclear area in control cells, while the stabilized RhoB protein also accumulated at the periphery in DLC3 knockdown cells ( Figure 4H). Importantly, simultaneous knockdown of Solo not only reduced the increase in RhoB levels observed upon DLC3 depletion, but also led to re-clustering of the GTPase in the perinuclear area as observed for the control cells. This result demonstrates that Solo and DLC3 are required for maintaining the subcellular distribution patterns of RhoB.
To obtain mechanistic insights into the regulation of RhoB protein levels, we next performed cycloheximide chase experiments ( Figure S3-figure supplement 3A and 3B). In agreement with measurements in other cell lines, 35,36 RhoB was found to be short-lived in In sum, these experiments show that Solo and DLC3 form a novel GEF-GAP pair that regulate endosomal RhoB in a complex manner, modulating not only its total cellular activity, but also its localization and turnover rates.

| Solo regulates EGFR trafficking and signalling
We previously found DLC3 to control EGFR degradation, a receptor tyrosine kinase that is rapidly endocytosed upon EGF ligand binding and is transported to the lysosomes in a RhoB-dependent manner. 14,39,40 Considering the functional connection between Solo and DLC3, we explored whether Solo is also required for the regulation of  Table S2 these conditions, presumably resulting from EGFR mislocalization and the absence of respective downstream effector molecules. RhoA activation and stress fibre formation. 28,41,42 Here, we uncover a previously unknown function of Solo in maintaining the integrity of the Golgi complex. Both the actin and microtubule cytoskeleton play a major role in the structure, function and positioning of the Golgi complex. 25,26 In addition, a few studies also suggest that an intact keratin network is required for the maintenance of Golgi structure. [42][43][44] While our work did not focus on the organization of cytokeratin filaments, we did find that Solo depletion reduces stress fibre formation in HeLa cells ( Figure S1-figure supplement 1E), indicating that the Golgi phenotype we observe here may be a consequence of F I G U R E 4 Solo knockdown delays epidermal growth factor receptor (EGFR) trafficking and dampens the EGF signalling response. (A) HeLa cells were transiently transfected with control (siNT) or Solo-specific (siSolo ss#2) siRNAs. Following a 48 h incubation period, the cells were serum starved overnight and, prior to fixation, stimulated with 10 ng/mL EGF for the indicated times. The cells were then stained with an antibody recognizing the extracellular domain of EGFR (surface EGFR). This was followed by permeabilization and immunostaining with an antibody recognizing the C-terminal part of EGFR (total EGFR). Shown are representative maximum intensity projections of five confocal sections. All images were acquired and displayed using identical settings. Scale bar: 10 μm. (B) Following knockdown and EGF stimulation as described in (A), HeLa cells were lysed, followed by immunoblotting with the indicated antibodies. The shown experiment is representative of three independent biological repeats. (C) Densitometric quantification of total EGFR levels in siNT and siSolo cells of the EGF stimulation experiment representatively shown in (B). EGFR levels of the different time points were normalized to GAPDH and subsequently divided by the signal measured for the unstimulated siNT cells, which was set to 1. (D) Densitometric quantification of pEGFR levels in siNT and siSolo cells of the EGF stimulation experiment representatively shown in (B). pEGFR levels of the different time points were normalized to GAPDH and subsequently divided by the signal measured for the siNT cells at 5 min post EGF stimulation, which was set to 1. (E) Densitometric quantification of the pAKT-Thr308 levels in siNT and siSolo cells of the EGF stimulation experiment representatively shown in (B). The phosphorylation signal was normalized to the one obtained for the unphosphorylated protein and GAPDH and subsequently divided by the signal measured for the siNT cells at 5 min post EGF stimulation. The value obtained for the siNT cells was further normalized to 1. (F) Densitometric quantification of the pERK (Thr202/Tyr204) levels in siNT and siSolo cells of the EGF stimulation experiment representatively shown in (B). The phosphorylation signal was normalized to the one obtained for the unphosphorylated protein and GAPDH and subsequently divided by the signal measured for the siNT cells at 5 min post EGF stimulation. The value obtained for the siNT cells was further normalized to 1. (C-F) The bar diagrams display the mean values of three independent biological repeats, cumulating two experiments using siSolo ss#2 and one experiment using siSolo sp#4. Error bars: SEM. See also Figure S4-figure supplement 1. For statistical testing data, see Table S2 cytoskeletal rearrangements downstream of altered Rho activity.
Importantly, out of the 23 RhoGEFs we tested in our imaging-based screen, Solo stood out for its ability to antagonize the fragmented Golgi phenotype caused by DLC3 depletion. This provides strong evidence that the effect of Golgi morphology is specifically downstream of the Solo-DLC3 interplay and not an unspecific readout of globally perturbed RhoGEF-RhoGAP balance.
In addition to the actin cytoskeleton, the morphology of the Golgi complex is also influenced by membrane trafficking homeostasis.
Among GTPases, in particular, RhoA, RhoB, RhoD, Rac and Cdc42 have been shown to affect various steps of membrane trafficking.
Interestingly, while the GEF activity was important for the Golgi compaction phenotype, Solo was not enriched at this organelle under steady state. The localization of Solo on a subset of RhoB positive endosomes rather suggests that the effects on Golgi morphology could also stem from altered membrane trafficking. Notably, the GEF activity of Solo appeared to be important for the vesicular localization of Solo since the GEF inactive mutant displayed a diffuse cytoplasmic distribution. This hints at a positive feedback loop between Solo activation and membrane enrichment. Such a mechanism was previously reported for the Sec7 Arf-GEF at the TGN. 45 This GEF-dependent change in the vesicular localization of Solo was also observed in HUVEC cells, 28 suggesting that the involvement of Solo in endocytic trafficking is conserved in other cellular systems as well. In line with this, Solo was contained in the endosomal mass spectrometry dataset of MEF cells obtained by Ivaska and co-workers. 46 While these studies do not allow to discriminate whether Solo is an endosomal cargo or an activator of RhoB, our optogenetic approach enabled us to address endosomal RhoB activation by Solo-GEF in a localized manner. To our knowledge, this is the first report of a direct readout of endogenous endosomal RhoB activation. Although we cannot exclude that Solo impinges on Golgi complex structure by regulation of RhoA as well, our study clearly demonstrates a role of Solo in the regulation of RhoB. Since most of the endogenous Solo is membrane attached, for the future it will be interesting to explore the cellular cues that regulate the activation of full-length Solo on endomembranes.
We further uncover a multifactorial involvement of Solo and DLC3 in the regulation of RhoB, which extends beyond the regulation of its Rho GTPase activity. Specifically, we find that the interplay between Solo and DLC3 controls the activity, the subcellular distribution as well as the total protein levels of RhoB. Although we were not able to pinpoint the mechanistic details responsible for RhoB protein level regulation, an intact endosomal trafficking system was recently found to be required for the homeostasis of cellular RhoB amounts and protein localization. 30,34 This indicates that Solo and DLC3 may not only control RhoB activity directly, but also indirectly by impinging on the endocytic recycling compartment.
Motivated by the interplay between Solo and RhoB, we also investigated the role of Solo in EGFR signalling and trafficking. Mirroring the effects of DLC3 depletion, 14