A potential Rho GEF and Rac GAP for coupled Rac and Rho cycles during mesenchymal-to-epithelial-like transitions

ABSTRACT

The leading edge-to-cadherin contact transitions that occur during metazoan developmental processes and disease states require fine coordination of Rac and Rho pathways. Recently the elmo-mbc complex, a Rac GEF and RhoGAP19D, a Rho GAP were identified as key, conserved regulators that link Rac and Rho during these transitions. The corresponding Rho GEF and Rac GAP remain hidden amongst the large family of GEF and GAP proteins. Identification of these regulators is essential to understand GTPase coordination during these transitions. Here we find two candidates based on the mammalian literature and use RNAi to explore the fly ortholog effects on the dorsal closure epidermis. RhoGEF64C and RhoGAP92B are strong contenders to couple Rac and Rho during mesenchymal-to-epithelial-like transitions.

Metazoan cells transition between migratory and cell-cell adherent states, which is critical for several developmental processes including gastrulation, neural crest, heart valve formation, palatogenesis, myogenesis, tumor metastasis and wound healing [1–4]. For productive migration and cell-cell adhesion these transitions require precisely coordinated actin cytoskeletal rearrangements, which are regulated by finely-tuned and integrated Rho family GTPase signaling pathways [5].

Guanine nucleotide exchange factors (GEFs) and GTPase activating proteins (GAPs) control the GTP-GDP state of Rho proteins, but identifying specific GEFs and GAPs for different processes has proven to be a challenge given the large protein family complexity (~ 70 GEFs and ~ 50 GAPs in mammals) and GTPase promiscuity [6,7]. In addition Rho protein signaling pathways are often coupled, which further complicates data interpretation [5]. An understanding of how GEFs and GAPs coordinate and precise Rho protein signaling requires simple, well-defined processes, pathways and proteins that can be used to examine the complex relationships between them.

Recently, the Rho GAP, RhoGAP19D, and the atypical Rac GEF, the elmo-mbc complex, which lacks a dbl-homology domain, were identified to genetically link Rac and Rho cycles during mesenchymal-to-epithelial-like transitions (MET-like) of Drosophila dorsal closure (Figure 1A)⁸. The leading edge of epidermal cells during dorsal closure contains diverse and distinct Rho family protein-regulated actin structures: an actomyosin band, lamellipodia, and filopodia [9–12]. When opposite fronts align and interact, the cells transition to cadherin contacts and seal the dorsal side of the embryo [13]. This study revealed marked similarities between Drosophila and mammalian regarding Rho regulation during MET-like processes [14,15]. In the mammalian literature RhoGTPases and their regulators are best characterized (Figure 1B), but their complex genome duplications make it difficult to define simple, non-redundant rho regulator relationships. The lack of GEF and GAP genome duplications in Drosophila, accessibility of dorsal closure, and remarkable conserved regulators during MET-like transitions make this organism an effective system to broadly address Rho regulators and Rho signaling.

Figure 1.

(A) Current model of Rho regulation during MET-like transitions (B) Schematic of pros and cons of Mammalian (italics) and Drosophila (bold italics) studies (C) Rho regulator orthologies for Mammals (italics) and Drosophila (bold italics). Orthologies were identified using simple BlastP (NCBI) searches between mouse/human and Drosophila (D) Montages of maximal intensity Z-stack projections of embryos, which express endogenous DE-cadherin-GFP and ectodermal Gal4-driven UAS-Lifeact-RFP with indicated UAS-RNAi constructs. Scale bar = 25 µm. (E) Single focal plane of an epidermal leading edge in en2.4-GAL4 UAS-LifeAct-RFP embryos of indicated UAS-RNAi constructs. Scale bar = 5 µm. (F) Quantification of the number of lamellipodia detected per length of engrailed positive leading edge from still images in indicated embryo backgrounds

Identification of the Rac GEF and Rho GAP that function during MET-like processes remains only half the picture and is not sufficient to understand the Rac-Rho cycle coordination. We mined the mammalian literature for Rho GEF and Rac GAP candidates that could function during leading edge-to-cadherin contact transitions and complete the circuit. In mammals, the RhoGAP19D ortholog, Arhgap21, was described as an epithelial-to-mesenchymal transition (EMT) Rho GAP [15]. Similarly Net1, a dbl-homology containing Rho GEF, was independently found to function during EMT processes [16]. This association makes Net1 family proteins candidates to work in conjunction with Arhgap21-related Rho regulation, although the relationship between EMT and MET processes is unclear. The sole Drosophila ortholog of Net1 is RhoGEF64C (Figure 1C) and was chosen for analysis during MET-like transitions of dorsal closure.

For the elmo-dock complexes in mammals, Arhgap22 was paired as a Rac GAP with Dock3 during mesenchymal-to-amoeboid transitions [17]. As with Dock proteins, a number of Arhgap22 orthologs exist in mammals (Figure 1C), however only one ortholog exists in Drosophila, RhoGAP92B. Unfortunately, the Drosophila nomenclature identifies the general presence of a Rho GEF or GAP domain and chromosomal location, but offers no information on Rho protein specificity, which may confuse readers. In addition, the mammalian orthologs SH3BP and Arhgap25 were found to regulate Rac during phagocytosis, a process in which elmo-dock complexes are known to regulate Rac [18,19]. Therefore, RhoGAP92B is a strong candidate and its function was probed in the dorsal closure epidermis.

To investigate the role of RhoGEF64C and RhoGAP92B at leading edges UAS-driven RNAi constructs were expressed in the ectoderm (P{GawB}69B). Stage 14–15 embryos that expressed both the RNAi construct and LifeAct-RFP were imaged to view actin structures during dorsal closure. Actomyosin cables were present along the leading edge in wt and elmo mutants, which were diminished and fragmented upon RhoGAP19D depletion, as seen previously (Figure 1D)⁸. In RhoGEF64C RNAi embryos the actin cable was absent in large stretches of the leading edges (Figure 1D). Actin cables were present in RhoGAP92B-depleted cell fronts as in wildtype (wt) but were occasionally discontinuous, and no major differences were observed between the two available RhoGAP92B RNAi constructs (Figure 1D). Surprisingly the combination of RhoGAP19D and RhoGEF64C RNAi, which are predicted to affect Rho cycles resulted in enlargement of lamellipodia and filopodia structures (Figure 1D). The actin cable may also be unexpectedly restored in these embryos, however the enlarged leading edge structures prevent resolution. Attempts to ectodermally express RhoGEF64C RNAi in elmo mutants failed to produce viable embryos, which suggest a potential synthetic lethality and was not pursued further (data not shown).

Cell fronts were imaged in more detail in the epidermis with en-GAL4 to prevent contaminating signal from the amnioserosa cells. These results were consistent with ectodermal expression results (Figure 1D) with RhoGEF64C RNAi having a reduced to non-existent actin cable while RhoGAP92B-depleted embryos were not grossly different from wt (Figure 1E). RhoGEF64C and RhoGAP19D double depletions again had exaggerated lamellipodia and filopodia, which were not reliably quantifiable due to their size and spanning multiple focal planes (Figure 1E). Quantification of lamellipodia in the single depletions revealed an increase upon RhoGEF64C depletion compared to wt similar to previous RhoGAP19D mutant results (Figure 1F). RhoGAP92B RNAis in contrast, both showed a mild decrease in the number of lamellipodia when compared to wt, which was more similar to earlier elmo-dock mutant reports (Figure 1F). Together these results suggest that RhoGEF64C and RhoGAP92B both are involved in the epidermal leading edge during dorsal closure but with very different effects, which may indicate differential effects on Rac vs. Rho.

Analysis of dorsal closure in ectodermally-depleted embryos revealed that embryos are able to complete closure, although RhoGAP92B-depleted embryos showed terminal delays similar to elmo-dock mutants, but were always resolved unlike the Rac GEF mutants (data not shown) [8]. In wt the closure progressed at a relatively constant speed for both cell fronts until they merged as seen previously (Figure 2A)⁸. In contrast, RhoGEF64C-depleted embryos showed a uneven speed of closure in the two cell fronts as dorsal closure proceeded, which could be a result of uneven depletion (Figure 2A). The unevenness challenged accurate speed description from kymograph slopes and therefore were not determined. Both RhoGAP92B-depleted cell fronts were similar in speed to wt for their closure (Figure 2A).

Figure 2.

(A) GFP-channel kymographs generated from (1 frame/min) movies of embryos that express endogenous DE-cadherin-GFP and ectodermal Gal4-driven UAS-Lifeact-RFP with indicated UAS-RNAi constructs. Schematic shows region of kymograph selected. Brightest signal indicates cadherin signal at epidermal leading edges that convergence over time. The closure speeds calculated for each condition. (B) Maximal intensity Z-stack projection of DE-cadherin-GFP at epidermal seams in indicated embryos. Scale bar = 5 µm. Yellow arrow indicates newest cadherin contact. Yellow dashed lines show regions with constricted cells. (C) Schematic that indicates cadherin dimensions measured in the zone of epidermal cadherin contact formation. Box plot of measurements of indicated cadherin dimensions for wt and depleted cells. Measurements pooled from 3–5 embryos and include anterior and posterior seams. (D) Model for potential Rho regulation during MET-like transitions

Cadherin contact formation was analyzed next. Cadherin along the leading edges is concentrated at the vertices along the leading edge and spread along the new contact as they are brought together at the seam in wt and both depleted embryos (Figure 2B)¹³. RhoGEF64C-depleted embryos appeared qualitatively similar to wt (Figure 2B). In contrast, for RhoGAP92B RNAi embryos the new cadherin contacts appeared elongated along the axis of the seam (Figure 2B). The Rac GEF and Rho GAP mutants have major effects on the cadherin dimensions on the newly formed contacts [8], therefore cell dimensions were quantified in the depletions of the Rho regulator candidates. Quantification revealed significant differences between RhoGEF64C depletions and wt both before and after cadherin contact formation (Figure 2C), which may be due to the presence of occasional constricted cell fronts and contacts in the RNAi embryos (Figure 2B, yellow dashed lines). Since both leading edge (w1) and new cell-cell contacts (w2) widths are similarly affected, the RhoGEF64C-related effects likely occur before the transition and may be a result of the reduced actin cable or increased lamellipodia. An expansion of new cadherin contacts occurred in RhoGAP92B RNAi lines, similar to elmo-dock mutants (Figure 2C)⁸. Together these results suggest both the putative Rho GEF and Rac GAP have a role during MET-like processes, but affect the processes differently and is consistent with them acting on different Rho GTPases.

This study identifies two promising Rho protein regulator candidates that are involved in leading edge-to-cadherin contact transitions based on their orthology and effects on the epidermal cells during dorsal closure. Since it is difficult to assess expression and RNAi efficiency, further studies that use direct mutants and Rho protein biosensors are necessary to definitively demonstrate how these proteins are regulated, integrated and function with elmo-mbc and RhoGAP19D to control Rac and Rho cycles during MET-like transitions.

Unexpectedly the Rho GAP, RhoGAP19D and the presumed Rho GEF, RhoGEF64C share many phenotypic effects that include reduced actin cables, increased lamellipodia number, altered closure speeds and synthetic interactions with elmo-dock [8]. It could be expected that the GEF and GAP would have opposing activities based on their predicted biochemical activities. However, it is also possible that the coupling of Rac and Rho cycles may challenge this perception. The loss of regular Rho cycling may in fact similarly affect Rac cycles. Curiously, RhoGAP19D and RhoGEF64C double depletion greatly enhanced leading edge structures. This may be due to a dramatic expansion of Rac-dependent lamellipodia and Cdc42-dependent filopodia in the absence of Rho cycling. However, an unexplained restoration of the actomyosin cable may also occur. More work is needed to resolve this complex interaction and provocative phenotype, such as whether or not Myosin II is still present at the leading edges in these conditions. Another point in the potential pairing of RhoGEF64C and RhoGAP19D is the localization of the proteins. In addition to cell-cell contacts RhoGAP19D and its mammalian ortholog, Arhgap21, both have reported nuclear localizations [8,20]. Unique among GEFs, nuclear sequestration is an established means of regulating Net1 [16,21]. One intriguing possibility is that RhoGEF64C and RhoGAP19D are co-regulated by similar mechanisms.

During dorsal closure RhoGAP92B RNAi embryos resulted in phenotypes more similar to the elmo-mbc complex that included reduced lamellipodia number and expanded cell-cell contacts. This may again be a result of Rac and Rho coupling, which results in excessive Rho activity in the absence of Rac cycles. Direct analysis of in vitro activities combined with in vivo biosensors are necessary to distinguish between these connections. An attractive possibility maybe that a recruitment and inhibition occurs between the presumed Rac GAP and the Rho GEF (Figure 2D) similar to what is observed for the Rho GAP and Rac GEF [8]. Together this arrangement would tightly tether Rac and Rho cycles and if spatially-defined would create tight boundaries between Rac and Rho activities within the cell. Furthermore, it could explain many of the observations during MET-like transitions. For instance, loss of the candidate Rho GEF and the Rho GAP would result in a lack of inhibition of the Rac GEF and a failure to recruit the presumed Rac GAP, both of which could explain excessive Rac-dependent lamellipodia found in the double deletion. However, such an arrangement requires meticulously comprehensive analysis of rho regulators and raises a number of challenges interpreting in vivo and in vitro data in isolated Rho regulator studies.

The mechanisms behind Rho regulator coupling during leading edge-to-cadherin contact transitions is largely open and must be defined. The rho regulators discussed here have no known direct interactions and offer few clues apart from a PDZ domain in RhoGAP19D, some proline-rich regions in RhoGEF64C and respective proline-rich region and SH3 domain of elmo and mbc, which are known to interact [19,22]. Scaffolding proteins likely play a major role in the process and are known to coordinate other Rho signaling pathways [23]. The elmo-mbc complex, which is best studied, interacts with several scaffolding proteins p130cas, crk, rols and in mammals Acf7, the shot ortholog [19,24–26]. However, it is unknown how any these scaffolding proteins could directly link to the Rho regulators presented here. One likely scaffolding protein is the LIM and PDZ domain protein, zasp52 which was recently shown to be necessary for actin cable formation of dorsal closure [11]. Given the actin cable phenotypes upon loss of RhoGAP19D or RhoGEF64C, a zasp52 connection warrants investigation. In addition, both zasp52 and rols were identified in the original screen that uncovered the cadherin function of elmo-mbc and RhoGAP19D, which furthers the likelihood of involvement in the process [22]. Future work that investigates Rho regulator scaffolds during MET-like transitions will help build a framework to understand the mechanisms that control Rho protein coupling.

Methods

Fly strains

The following lines were generously shared in this study or obtained from the BDSC (source or BDSC stock number): ced-12 [19,3] (SM Abmayr) shg::GFP (60,584), Gal4-69B (1774) [27], en-GAL4 (8165) [28], UAS-LifeAct::RFP (58,362) [29], UAS-rhogap92B^RNAi (6444 – RhoGAP92B-1 and 33,391 – RhoGAP92B-2) [30,31]. UAS-rhoGEF64C^RNAi(31,130) [30].

Fly pushing and staging

All Drosophila work was carried out at 25°C. To visualize dorsal closure embryos were collected 12–14 hr after egg laying, dechorionated for 1 min in bleach, aligned and mounted on a coverslip in Halocarbon 200 oil (25,073, Polysciences, Inc).

Microscopy and imaging

Embryos were imaged on a Nikon Ti-E inverted microscope equipped with a Yokogawa CSU-X1 spinning disk and an EM-CCD Camera (Photometrics Evolve 512). A 20× air 40 × 1.25 N.A. water-immersion and 100 × 1.4 N.A. oil-immersion objectives were used. Image acquisition was performed with MetaMorph. For 20× images, 60 planes (1.2 µm) were captured that spanned the dorsal side of the embryo. For 100× images 60 planes (0.1 µm) were captured that spanned the epidermis leading edge or dorsal closure seam. GFP and RFP channels were captured sequentially. Laser power kept constant between experiments.

Image analysis

All images were analyzed with ImageJ (http://rsb.info.nih.gov/ij/). Cell dimensions were measured as follows. Cell widths before cadherin contact formation (wl) was as the distance between terminal cadherin foci at the leading edge in the 25 cells preceding the newest cadherin contact. Cell width of the new contact (w2) was measured as the distance between the two vertexes adjacent to the new contact in the 10 cells following the newest cadherin contact. Cell lengths (l) were measured from the midpoint of the two vertexes adjacent to the newest cadherin contact to the maximal distance obtainable within a cell in the 10 cells following the newest cadherin contact. Lamellipodia counts were obtained from single plane, still images captured as seen in Figure 1E. The numbers of lammellipodia webs were counted and divided by the length of the actin cable/segment length. Cell front speeds were calculated from the slope of the leading edges from kymographs taken from six measurements (left and right side) of three embryos for each condition.

Box plots and statistics

For all box and whisker plots, the ends of the box mark the upper and lower quartiles, the central horizontal line indicates the median, and the whiskers indicate the maximum and minimum values. For graphs (**) and (***) indicate P < 0.05 and P < 0.005 (unpaired T-test), respectively.

Funding Statement

This work was supported by the INFORM/LABEX [ANR-11-LABX-0054];

Disclosure statement

No potential conflict of interest was reported by the authors.