PLEKHG3 enhances polarized cell migration by activating actin filaments at the cell front

Significance

Polarized cell migration plays a pivotal role in the development and repair of tissues. PI3K, Rho GTPases, and actin filaments are known to be involved in a positive feedback loop that induces and maintains cell polarity. Here, we show that the pleckstrin homology and RhoGEF domain containing G3 (PLEKHG3) selectively binds to newly polymerized actin and that this interaction exerts a positive regulatory effect on PLEKHG3 activity that enhances and sustains the cell front. A lack of PLEKHG3 ablates cell polarity, resulting in a decrease in cell migration. These findings provide the missing link that explains how Ras-related C3 botulinum toxin substrate 1 (Rac1) and actin polymerization are coupled by a positive feedback loop to ensure the stability of cell polarity.

Abstract

Cells migrate by directing Ras-related C3 botulinum toxin substrate 1 (Rac1) and cell division control protein 42 (Cdc42) activities and by polymerizing actin toward the leading edge of the cell. Previous studies have proposed that this polarization process requires a local positive feedback in the leading edge involving Rac small GTPase and actin polymerization with PI3K likely playing a coordinating role. Here, we show that the pleckstrin homology and RhoGEF domain containing G3 (PLEKHG3) is a PI3K-regulated Rho guanine nucleotide exchange factor (RhoGEF) for Rac1 and Cdc42 that selectively binds to newly polymerized actin at the leading edge of migrating fibroblasts. Optogenetic inactivation of PLEKHG3 showed that PLEKHG3 is indispensable both for inducing and for maintaining cell polarity. By selectively binding to newly polymerized actin, PLEKHG3 promotes local Rac1/Cdc42 activation to induce more local actin polymerization, which in turn promotes the recruitment of more PLEKHG3 to induce and maintain cell front. Thus, autocatalytic reinforcement of PLEKHG3 localization to the leading edge of the cell provides a molecular basis for the proposed positive feedback loop that is required for cell polarization and directed migration.

Cell polarity is essential for many cellular processes: It allows neurons to form dendrites and axons, enables dividing cells to produce daughter cells, engenders fibroblasts with wound-healing activity, and gives leukocytes the ability to crawl to infection sites (1, 2). Cell polarity is modulated by signaling cascades that center around the action of phosphoinositide kinases, the activation of Rho small GTPases by phosphatidylinositol-3,4,5-Tris phosphate (PIP₃, a lipid product of PI3Ks), and the activity of F-actin (3, 4). Several studies have shown that F-actin generates positive feedback for the activation of PI3K, Ras-related C3 botulinum toxin (Rac), and/or cell division control protein 42 (Cdc42) (4–6). The feedback activation of PI3K is, in turn, regulated by multiple Rho family small GTPases via coupled positive feedback loops (7) in which the existing filaments generate new free barbed ends, and new F-actin filaments grow on the free ends (8). In fibroblasts, T cells, and macrophages, cell polarity and chemotaxis are abrogated when Rac or Cdc42 is inhibited (9, 10). Rac and Cdc42 regulate cell migration through the precise spatial and temporal coordination of dynamic actin-based structures found at the leading edge of migrating cells (11, 12).

GTPase activity, which controls various cellular functions, is regulated by guanine-nucleotide exchange factors (GEFs) and GTPase activation proteins (GAPs). GEFs activate signaling by catalyzing the exchange of G protein-bound GDP to GTP, whereas GAPs terminate signaling by inducing the hydrolysis of GTP (13). Sixty-nine identified GEFs belong to the Dbl family and are responsible for accelerating the intrinsic nucleotide exchange activity of Rho-family small GTPases (14, 15). In addition to possessing a Dbl homology (DH)–pleckstrin homology (PH) module, most GEFs contain additional functional domains, such as the Src homology 2 (SH2), Src homology 3 (SH3), Ras-GEF, and Ser/Thr or Tyr kinase domains. These regions are critical for GEFs’ ability to interact with other regulatory proteins and to activate them in response to cellular signals (16).

Here, we report that the pleckstrin homology and RhoGEF domain containing G3 (PLEKHG3) is a PI3K-regulated RhoGEF for Rac1 and Cdc42 that selectively binds to newly polymerized actin at the leading edge of migrating fibroblasts. We discovered the existence of a positive feedback loop from actin filaments to PLEKHG3 from the observation that PLEKHG3 accumulates at the site of newly polymerized actin at the leading edge when the cell moves forward. PLEKHG3 was recruited to the newly formed protrusion area when a photoactivatable Rac1 (PA-Rac1) was continuously activated. Thus, PLEKHG3 is indispensable for both inducing and maintaining cell polarity and directional motility. These findings reveal that PLEKHG3 generates a positive feedback loop that controls cell polarity and directional motility, explaining how Rac1 and actin polymerization are coupled by a positive feedback loop to ensure the stability of polarity.

Results

PLEKHG3 Localizes to the Leading Edge and Controls Cell Migration.

To identify new GEFs that control cytoskeletal dynamics during cell migration, we generated a library consisting of 63 human GEFs and analyzed their subcellular localizations in NIH 3T3 cells. The GEFs were classified into six groups based on their distinct subcellular localizations: one GEF was localized in the nucleus, one GEF was localized in microtubules, two GEFs were localized in actin filaments, six GEFs were localized in the plasma membrane (PM), six GEFs were distributed throughout the whole cell, and 47 GEFs were localized in the cytoplasm (Fig. S1 and Table S1). We selected nine cytoskeleton-related GEFs that localized at the PM, actin filaments, or microtubules and examined their involvement in controlling cell migration (Fig. 1A). Cells expressing seven of the nine selected GEFs migrated more rapidly in the presence of FBS than did control cells (Fig. 1B). Most notably, PLEKHG3 demonstrated a polarized subcellular localization with enrichment at the leading edge (Fig. 1C and Movie S1). To test whether the unique localization of PLEKHG3 at the leading edge was a general feature of cell lines other than NIH 3T3, PLEKHG3 was expressed in human umbilical vein endothelial cells (HUVECs) and MDA-MB-231 cells. Indeed, we observed the polarized subcellular localization of PLEKHG3 and the increased migration among HUVECs and MDA-MB-231 cells overexpressing this protein (Fig. S2 A–C). We confirmed the colocalization of PLEKHG3 and F-actin filaments by cytochalasin D treatment. A disruption pattern visualized in F-actin was observed in PLEKHG3 (Fig. S2D) but not in Ras-specific guanine nucleotide-releasing factor 1 (RALGPS1), a PM GEF (17).

Fig. S1.

Localization of the 63 human GEFs. Confocal images show the subcellular localization of 63 CFP-conjugated human GEFs in NIH 3T3 cells. The localizations were classified into six categories: one GEF was localized in the nucleus, one GEF was localized in microtubules, two GEFs were localized in actin filaments, six GEFs were localized in the PM, six GEFs were distributed throughout the whole cell, and 47 GEFs were localized in the cytoplasm. (Scale bar, 20 µm.)

Table S1.

Data for 63 human GEFs

No.	Name	NCBI ID	GEF information	GEF family
1	ABR	NM_021962.3	673 aa: RasGEF(197–426)	Ras GEF
2	AKAP13	AF127481.1	893 aa: RhoGEF(75–269) PH(258–414)	Rho GEF
3	ARHGEF1	NM_198977.1	879 aa: RhoGEF(384–570) PH(585–730)	Rho GEF
4	ARHGEF4	NM_015320.2	690 aa: RhoGEF(285–460) PH(472–611)	Rho GEF
5	ARHGEF5	NM_005435.3	1,597 aa: RhoGEF(1175–1356) PH(1378–1504)	Rho GEF
6	ARHGEF6	NM_004840.2	776 aa: RhoGEF(242–419) PH(450–549)	Rho GEF
7	ARHGEF7	NM_001113513.1	646 aa: RhoGEF(94–271) PH(302–400)	Rho GEF
8	ARHGEF9	BC117406.1	516 aa: RhoGEF(104–285) PH(291–430)	Rho GEF
9	ARHGEF10	NM_014629.2	1,344 aa: RhoGEF(397–581) PH(614–678)	Rho GEF
10	ARHGEF15	BC036749.1	841 aa: RhoGEF(418–598) PH(622–742)	Rho GEF
11	ARHGEF16	BC002681.1	421 aa: RhoGEF(1–178) PH(201–334)	Rho GEF
12	ARHGEF18	NM_015318.3	1,015 aa: RhoGEF(102–296) PH(336–454)	Rho GEF
13	ASEF	AB042199.1	619 aa: RhoGEF(214–395) PH(401–540)	Rho GEF
14	BIG3	AF413080.1	1,770 aa: Sec7(185–391)	Arf GEF
15	C9orf100	BC033666.1	335 aa: RhoGEF(27–195) PH(228–331)	Rho GEF
16	CYTH1	BC038385.1	398 aa: Sec7(62–244) PH(259–379)	Arf GEF
17	CYTH2	BC004361.1	400 aa: Sec7(61–243) PH(258–378)	Arf GEF
18	CYTH3	BC028717.1	399 aa: Sec7(66–248) PH(263–382)	Arf GEF
19	CYTH4	BC041161.1	394 aa: Sec7(62–243) PH(258–376)	Arf GEF
20	FARP1	BC041595.1	1,045 aa: RhoGEF(542–728) PH(739–861) PH(916–1028)	Rho GEF
21	FARP2	NM_014808.2	1,054 aa: RhoGEF(537–720) PH(735–855) PH(913–1025)	Rho GEF
22	FGD1	BC034530.1	961 aa: RhoGEF(376–559) PH(592–698) PH(815–920)	Rho GEF
23	FGD3	AY211386.1	634 aa: RhoGEF(160–339) PH(372–478)	Rho GEF
24	FGD5	BC035364.1	540 aa: RhoGEF(27–160) PH(179–301) PH(444–533)	Rho GEF
25	GBF1	NM_004193.2	1,855 aa: 2xSec7(397–552 699–884)	Arf GEF
26	GEF–H1	NM_004723.3	958 aa: RhoGEF(208–402) PH(442–557)	Rho GEF
27	GEFT	BC012860.1	474 aa: RhoGEF(57–228) PH(233–375)	Rho GEF
28	LBC	XM_012429.3	581 aa: RhoGEF(244–431) PH(478–578)	Rho GEF
29	NET1	BC053553.1	596 aa: RhoGEF(175–354) PH(358–500)	Rho GEF
30	NGEF	BC031573.1	710 aa: RhoGEF(274–455) PH(477–603)	Rho GEF
31	PDZ–RhoGEF	BC057394.1	1,562 aa: RhoGEF(775–961) PH(979–1121)	Rho GEF
32	PLEKHG3	BC129953.1	1,219 aa: RhoGEF(96–270) PH(252–396)	Rho GEF
33	RABGEF1	NM_014504.2	491 aa: VPS9(270–370)	Rab GEF
34	RABIF	BC018488	123 aa: RabGEF(19–123)	Rab GEF
35	RALGPS1	NM_001190729.1	1,614 aa: RasGEF(46–284) PH(391–506)	Ras GEF
36	RAP1GDS1	BC098334.1	558 aa: ARM(43–162, 268–381, 390–458)	Rap GEF
37	RAPGEF2	NM_014247.2	1,499 aa: RasGEF(713–938)	Ras GEF
38	RAPGEF3	NM_001098532.2	881 aa: RasGEF(616–848)	Ras GEF
39	RAPGEF4	NM_007023.3	1,011 aa: RasGEF(768–1005)	Ras GEF
40	RAPGEF5	NM_012294.3	730 aa: RasGEF(491–725)	Ras GEF
41	RAPGEF6	NM_001164386.1	1,609 aa: RasGEF(857–1081)	Ras GEF
42	RAPGEFL1	NM_016339.3	456 aa: RasGEF(217–450)	Ras GEF
43	RASGEF1A	BC022548.1	481 aa: RasGEF(214–457)	Ras GEF
44	RASGEF1B	BC121003.1	472 aa: RasGEF(202–448)	Ras GEF
45	RASGEF1C	BC057759.1	238 aa: RasGEF(49–195)	Ras GEF
46	RASGRP1	BC109297.1	797 aa: RasGEF(201–430)	Ras GEF
47	RASGRP2	XM_045647.1	671 aa: RasGEF(212–445)	Ras GEF
48	RASGRP3	NM_015376.2	689 aa: RasGEF(148–378)	Ras GEF
49	RASGRP4	AF448437.1	673 aa: RasGEF(197–426)	Ras GEF
50	RGL	AF186779.1	768 aa: RasGEF(228–497)	Ras GEF
51	RGL1	NM_001297670.1	766 aa: RasGEF(226–495)	Ras GEF
52	RGL2	BC032681.1	777 aa: RasGEF(239–509)	Ras GEF
53	SH2D3C	BC032365.1	860 aa: RasGEF(585–754)	Ras GEF
54	SOS1	NM_005633.3	1,333 aa: RhoGEF(201–388) PH(439–545) RasGEF(776–1015)	Rho GEF
55	SOS2	NM_006939.2	1,332 aa: RhoGEF(199–386) PH(437–543) RasGEF(774–1013)	Rho GEF
56	TEM4	AF378754.1	2,063 aa: RhoGEF(1067–1252) PH(1284–1432)	Rho GEF
57	TIAM1	NM_003253.2	1,591 aa: RhoGEF(1041–1232) PH(1235–1406)	Rho GEF
58	TIAM2	NM_012454.3	1,701 aa: RhoGEF(1100–1291) PH(1294–1466)	Rho GEF
59	TRAD	AB011422	1,289 aa: RhoGEF(233–405) PH(410–544)	Rho GEF
60	VAV1	NM_005428.2	845 aa: RhoGEF(195–371) PH(385–508)	Rho GEF
61	VAV2	NM_003371	839 aa: RhoGEF(194–369) PH(384–506)	Rho GEF
62	VAV3	AF067817.1	847 aa: RhoGEF(193–369) PH(383–506)	Rho GEF
63	WBSCR16	NM_030798.3	464 aa: 6xRCC	Ran GEF

Fig. 1.

PLEKHG3 localizes to the leading edge of the cell and controls cell migration. (A) The localization of nine GEFs at the PM, microtubules, and actin filaments in NIH 3T3 cells. (See Fig. S1 and Table S1.) (B) Cells overexpressing seven of the nine GEFs showed increases in the average velocity of cell migration in the FBS-containing medium compared with the control-expressing cells of these GEFs (n > 250). (C) Changes in cell morphology and the localization of CFP-C1 and CFP-PLEKHG3 during cell migration. (See Fig. S2 and Movie S1.) (D) A schematic for the generation of knockout (KO) hESC lines using the CRISPR/Cas9 system. (E) The endogenous PLEKHG3 localization of H9 and PLEKHG3^−/− cells. PLEKHG3 is located at the leading edge of migrating cells in H9 cells but not in PLEKHG3^−/− cells. The length (yellow line) is the longest distance between any two points along the boundary; the width (blue line) is the secondary axis of the best fit ellipse of the cells. (F) Quantification of the length/width ratio of H9 and PLEKHG3^−/− cells: PLEKHG3^−/− cells showed a longer morphology than the control cells (n >100). (G) The average velocity of cell migration decreased in PLEKHG3^−/− cells (n > 150). (See Figs. S3–S5.) The data represent mean ± SEM; *P < 0.1; **P < 0.01. (Scale bars, 20 μm.)

Fig. S2.

PLEKHG3 localizes to the leading edge of the cell and controls cell migration. (A) The localization of overexpressed PLEKHG3 in different cell lines. (B) The average velocity of cell migration was increased MDA-MB-231 cells and HUVECs overexpressing PLEKHG3 compared with control-expressing cells of PKHG3 (n > 70). (C) Cell morphologies and localization during cell migration in MDA-MB-231 and HUVEC cells overexpressing PLEKHG3. (D) The binding of PLEKHG3 to F-actin was examined in cells treated with cytochalasin D. PLEKHG3 showed similar patterns of dissociation from F-actin when cytochalasin D was added. In contrast, the localization of RALGPS1, which localizes to the PM, was not disrupted by cytochalasin D treatment. (E) A schematic for the adjacent region of PLEKHG3 exon2. ATG, start codon of PLEKHG3 CDS. F, forward primer-binding site. R, reverse primer-binding site. (F) Sequences of wild-type and CRISPR/Cas9-induced biallelic nonsense mutations in the targeted region of PLEKHG3^−/− clones. A single-nucleotide insertion (+1) leads to a frame shift (the introduction of a premature stop codon) and consequently to gene knockout. The asterisk indicates the insertion of nucleotide “A.” The data represent mean ± SEM; *P < 0.1; **P < 0.01. (Scale bars, 20 µm.)

To confirm the apparent involvement of PLEKHG3 in controlling cell migration, a fibroblast cell line was differentiated from the PLEKHG3-knockout human ES cells (hESCs) based on the CRISPR/Cas9 method (Fig. 1D and Fig. S2 E and F). The knockout cell line showed an ablation of the PLEKHG3 expression at the leading edge (Fig. 1E), a deviated morphology (Fig. 1F), and cell motility substantially decreased from that of control cells (Fig. 1G). The change in cell shape and cell motility was also recorded in cells of several cell lines treated with PLEKHG3 siRNA (Fig. S3).

Fig. S3.

The loss of PLEKHG3 results in elongated cell morphology and a decrease in cell motility. (A) The endogenous localization of PLEKHG3 and actin filaments in NIH 3T3 cells. PLEKHG3 localizes mostly at the leading edge of migrated cells. (B) Quantification of endogenous PLEKHG3 transfected with control or PLEKHG3 siRNAs. (C) The average shape of cells transfected with control (Ctrl) or PLEKHG3 siRNAs in different cell lines. PLEKHG3-depleted cells showed a longer morphology than control cells (n > 150). (D) The average velocity of cells transfected with control or PLEKHG3 siRNAs in different cell lines. The velocity of cell migration decreased in PLEKHG3-depleted cells (n > 230). The data represent the mean ± SEM; *P < 0.1; **P < 0.01. (Scale bar, 20 µm.)

PLEKHG3 Binds Directly to F-Actin Through an Actin-Binding Domain.

To elucidate the region of PLEKHG3 that is responsible for the colocalization with F-actin, we generated several truncated forms of PLEKHG3 and assessed their subcellular localizations in NIH 3T3 cells. Human PLEKHG3 [also known as ARHGEF43; National Center for Biotechnology Information (NCBI) no. BC129953] encodes a 1,219-amino acid protein with a predicted mass of 134 kDa. It contains a tandem DH–PH domain catalytic cassette in the N-terminal sequence and does not harbor any other known domain or motif (Fig. S4A). Our results revealed that the region encompassing residues 910–940 of PLEKHG3 exhibited a subcellular localization similar to that of F-actin and that smaller fragments of this region were unable to colocalize to actin filaments (Fig. S4 B and C). Furthermore, we found that both the DH–PH domain and actin-binding domain (ABD) of PLEKHG3 are required for the polarized subcellular localization of PLEKHG3 and increased cell migration (Fig. S5 A–C). The ability of PLEKHG3 (amino acids 1–950) to induce cell polarity was also observed in the PLEKHG3^−/− cells (Fig. S5 D and E).

Fig. S4.

PLEKHG3 binds directly to F-actin through an actin-binding domain. (A) A schematic of PLEKHG3 constructs. (B) The localization of the full-length (FL) and truncated forms of PLEKHG3. Amino acids 910–940 (designated the ABD) are required for binding to actin filaments. (C) Cells overexpressing PLEKHG3 (amino acids 910–940) were treated with cytochalasin D to confirm its ability to bind F-actin (n > 50). (D) The direct binding of PLEKHG3 (amino acids 890–950) and F-actin was examined using actin cosedimentation assays. After high-speed centrifugation, the supernatant (S) and pellet (P) fractions were resolved by SDS/PAGE and stained with Coomassie Blue. In the absence of F-actin, GST-PLEKHG3 (amino acids 890–950) was found in the supernatant. In the presence of F-actin, recombinant GST-PLEKHG3 (amino acids 890–950) was found predominantly in the F-actin–containing pellet. α-Actinin was used as the positive control. BSA was used as the negative control. (E) Multiple alignment of the ABDs from several GEFs known to bind to actin filaments, as performed using DNASTAR software: human PLEKHG3 (amino acids 910–940); human FLJ00018 (amino acids 158–213); human PDZ-RHOGEF (amino acids 564–600); human TEM4 (amino acids 81–135); and rat FRABIN (amino acids 19–69). (Scale bar, 20 µm.)

Fig. S5.

Both the DH–PH domain and the ABD of PLEKHG3 are required for the induction of cell polarity. (A) The localization of the full-length and truncated forms of PLEKHG3 in NIH 3T3 cells. (B) PLEKHG3 (amino acids 1–950), which contains both the DH–PH domain and the ABD, induces cell polarity during cell migration. (C) The average velocity of cell migration was increased in NIH 3T3 cells overexpressing PLEKHG3 (amino acids 1–950) as compared with control-expressing cells of PLEKHG3. (D) The localization of PLEKHG3 (amino acids 1–950) in PLEKHG3^−/− cells. (E) The average velocity of cell migration increased in PLEKHG3^−/− cells when PLEKHG3 (amino acids 1–950) was overexpressed. The data represent the mean ± SEM; **P < 0.01. (Scale bar, 20 µm.)

To determine whether the colocalization of PLEKHG3 and F-actin reflected a direct interaction, we used a high-speed actin cosedimentation assay to evaluate the binding ability of purified F-actin with purified recombinant GST-PLEKHG3(amino acids 890–950). Indeed, recombinant GST-PLEKHG3 (amino acids 890–950) was found predominantly in the F-actin–containing pellet (P) (Fig. S4D), demonstrating that PLEKHG3 binds directly to F-actin. Amino acids 910–940 of PLEKHG3 do not exhibit sequence similarity to any known ABD, such as Rho guanine nucleotide exchange factor 11 (PDZ-RhoGEF) (18), tumor endothelial marker 4 (TEM4) (19), FLJ00018 (20), or FYVE, RhoGEF, and PH domain-containing protein 4 (Frabin) (Fig. S4E) (21).

PLEKHG3 Enhances Cell Polarity by Activating Rac1 and Cdc42.

Based on our observation of PLEKHG3 localization during cell migration, we found that PLEKHG3 localizes at the leading edge of the cell where most of the actin filaments are found. We hypothesized that the cell directionality depends on the accumulation of PLEKHG3 to newly synthesized F-actin at the leading edge. We therefore attempted to change the localization of PLEKHG3 from a subcellular region with high PLEKHG3 localization to a new region with lower PLEKHG3 localization and to observe any changes in cell directionality using the light-mediated dimerizer system cryptochrome 2 (CRY2)–CRY-interacting bHLH 1 (CIB1) (Fig. S6A). We used Lifeact (which is known to bind both G-actin and F-actin biochemically) for targeting PLEKHG3 to F-actin because Lifeact is used as a marker to visualize F-actin in living cells and is known not to interfere with the actin dynamics in cells (22, 23). The illuminated regions showed the induction of lamellipodia, and the direction of the cells eventually changed because of the accumulation of PLEKHG3 (Fig. S6 B and D and Movie S2). To confirm that exogenous PLEKHG3 controls cell polarity and directionality during migration, we used an optogenetic method called “light-activated reversible inhibition by assembled trap” (LARIAT) to inhibit the function of exogenous PLEKHG3 (24). Upon light stimulation, the PLEKHG3-GFP proteins rapidly formed clusters. The cells shrank and lost polarity (Fig. S6 F and G). When the blue light was removed, the clusters disassembled, the cells returned to their original size, and cell polarity was regained (Fig. S6H). To exclude the effect of endogenous PLEKHG3, PLEKHG3^−/− cells were transfected with exogenous PLEKHG3. We observed a similar phenomenon in PLEKHG3^−/− cells when exogenous PLEKHG3 was inhibited by light stimulation (Fig. S6 I–K). To test the ability of PLEKHG3 to induce new polarity following disruption, the cells were locally illuminated at the leading edge. The illuminated region retracted rapidly and then reprotruded when the light was removed, changing the direction of cell migration (Fig. S6 C and E and Movie S3). Collectively, these data indicate that PLEKHG3 controls cell polarity.

Fig. S6.

Inhibition of PLEKHG3 disrupts cell polarity. (A) A schematic of the light-mediated dimerizer system, CRY2–CIB1, used to recruit PLEKHG3 to the F-actin. (B) NIH 3T3 cells were cotransfected with mCherry-PHR-PLEKHG3 and CIB1-EGFP-Lifeact. The illuminated regions exhibited an accumulation of PLEKHG3 that eventually changed the direction of the cells. The white dotted circle indicates the position stimulated by light. Time is shown in minutes. (See Movie S2.) (C) Cells were cotransfected with mCherry-Lifeact, PLEKHG3-GFP, SNAP-CRY2-V_HH(GFP), and CIB1-SNAP-MP. The PLEKHG3 signal retracted and repolarized in a different position upon light stimulation. (See Movie S3.) (D) Fluorescence intensity profiles representing the expression levels of mCherry-PHR-PLEKHG3 before and after the light stimulation shown in B (n > 30). (E) Fluorescence intensity profiles representing the expression levels of mCherry-Lifeact before and after the light stimulation shown in C (n > 30). (F) Upon light stimulation, cells overexpressing PLEKHG3-GFP shrank and lost their polarity. However, the cells regained polarity when the light was turned off. VAV2 did not induce polarity before or after light stimulation. Morphological changes were monitored using mCherry-Lifeact. (G) Upon light stimulation, cells were trapped and became rapidly visible as clusters. Time is shown in minutes. (H) Measurement of the cell area before, during, and after light stimulation (n > 50). The cell areas occupied by PLEKHG3 and VAV2 were strongly reduced upon light stimulation compared with the corresponding values in control cells. (I) PLEKHG3^−/− cells were transfected with CIB1-SNAP-MP, SNAP-CRY2-V_HH(GFP), mCherry-Lifeact, and PLEKHG3-GFP. The cells shrank when the light was turn on and regained polarity when the light was turned off. Morphological changes were monitored using mCherry-Lifeact. (J) Upon light stimulation, cells were trapped and rapidly became visible as clusters. (K) Measurement of the cell area before, during, and after light stimulation (n > 30). The areas occupied by PLEKHG3 were reduced upon light stimulation compared with the control cells. The data represent the mean ± SEM; *P < 0.1; **P < 0.01. (Scale bar, 20 µm.)

We examined the localization of 63 human GEFs and found two, PLEKHG3 and TEM4, which both localized to actin filaments but differed in their localization during cell migration. Assessment of migrating cells coexpressing exogenous TEM4 and PLEKHG3 confirmed that TEM4 is highly expressed at the trailing edge, whereas PLEKHG3 is highly expressed at the leading edge (Fig. S7 A and B). We speculated that this difference could reflect differences in the ABDs of PLEKHG3 and TEM4. However, the ABDs of PLEKHG3 and TEM4 both showed localization patterns similar to that of F-actin (Fig. S7C), suggesting that other domains in PLEKHG3 and TEM4 may confer the ability to bind different F-actin filaments. It is known that the Dbl family of most GEFs is responsible for accelerating the intrinsic nucleotide exchange activity of Rho-family small GTPases (14, 15). TEM4 has been reported to regulate the ability of Rho subfamily members (RhoA, B, and C) to promote the formation of actin stress fibers (19). To test which small GTPases are regulated by PLEKHG3, chemically induced heterodimerization of FK506 binding protein (FKBP) and FKBP-rapamycin-binding-domain (FRB) (25) was used to translocate PLEKHG3 (DH–PH) from the cytosol to the PM. We observed that PLEKHG3 (DH–PH) induced the formation of lamellipodia and filopodia upon its recruitment to the PM following rapamycin treatment (Fig. S7D). This finding suggests that PLEKHG3 might regulate Rac or Cdc42, two small GTPases that are known to control dynamic actin filaments at the leading edge (26). To confirm this finding, we performed FRET imaging using Ras and interacting protein chimeric unit (Raichu)-Rac1 and Cdc42 FRET biosensors (Fig. S7E). A significant increase was observed in the FRET/CFP signal intensity ratio, indicating the activation of a small GTPases (Rac1 and Cdc42) at the PM (Fig. S7 F–H). In addition, to monitor the activation of the small GTPases Rac1 and Cdc42, we performed a pull-down assay with GST-PAK1-PDB on lysates. We detected higher levels of the active forms of Rac1 and Cdc42 in cells expressing PLEKHG3 (Fig. S7I). Together, these findings indicate that PLEKHG3 activates Rac1 and Cdc42.

Fig. S7.

PLEKHG3 activates Rac1 and Cdc42. (A) GFP-TEM4 and mCherry-PLEKHG3 showed different localization patterns in polarized coexpressing cells. PLEKHG3 localizes mostly to the leading edge, whereas TEM4 is observed predominantly at the trailing edge. (B) Intensity profiles for GFP-TEM4 and mCherry-PLEKHG3. (C) The ABD domains of both PLEKHG3 and TEM4 bind to F-actin but fail to induce cell polarity in the same manner as full-length PLEKHG3 and TEM4. (D) Lamellipodia and filopodia were induced when the PLEKHG3 (DH–PH domain) fragment was recruited into the PM following rapamycin treatment. An enlarged view of the boxed region is shown at the right. (Magnification: 1.5×.) (E) Schematic of the FRET biosensor strategy used to activate small GTPases. (F) Cells were cotransfected with Lyn-FRB and mCherry-FKBP-PLEKHG3 (DH–PH) together with the FRET biosensor and were serum-starved to reduce basal small GTPase activity. PLEKHG3 (DH–PH) then was activated by rapamycin-induced translocation to the PM. Quantitative analyses of rapamycin-induced ratio changes (FRET/CFP) show that PLEKHG3 activates Rac1 and Cdc42. mCherry-FKBP-VAV2 and mCherry-FKBP-C1 were used as positive and negative controls, respectively (n > 120). (G) The FRET/CFP ratio images of Raichu-Rac1 FRET biosensors indicate the recruitment of control (mCherry), PLEKHG3 (DH–PH domain), and VAV2 (DH–PH domain) in the PM. (H) The FRET/CFP ratio images of Raichu-Cdc42 FRET biosensors. (I) Cells overexpressing PLEKHG3 showed increased levels of the active forms of Rac1-GTP and Cdc42-GTP. VAV2 is used as the positive control. Cell lysates were treated with GTPγS (positive control) or GDP (negative control) to activate or inactivate Rac1 and Cdc42. (Scale bars, 20 µm.)

PLEKHG3 Enhances Polarized Cell Migration via a Positive Feedback Loop at the Cell Leading Edge.

We observed similar oscillations in the localization of PLEKHG3 and F-actin in migrating cells (Fig. S8A). Closer observation of the leading edge via super-resolution microscopy (SIM) showed that F-actin and PLEKHG3 are strongly colocalized at the filopodia of the leading edge (Fig. 2 A and B and Fig. S8 B and D). Moreover, PLEKHG3 accumulated near newly formed actin filaments on a characteristic time scale of ∼30 s (Fig. 2C and Movie S4). Based on these observations, we hypothesized that there could be a positive feedback loop from polymerized actin to PLEKHG3. To test the involvement of PLEKHG3 in this positive feedback loop, we used PA-Rac1 to perform specific local activation at the leading edge (Fig. 2D). Upon light stimulation, Rac1 was activated, lamellipodia were formed, and PLEKHG3 was detected in the area of the newly formed protrusion (Fig. 2 E–G and Movie S5). Haugh’s group observed the relocalization of PI3K signaling at the protrusion upon photoactivation of PA-Rac1 (27). To eliminate the involvement of PI3K, cells were treated with the PI3K inhibitor LY294002 (LY29). Upon light stimulation, the accumulation of PLEKHG3 in the protrusion area was observed with treatment with LY29 (Fig. S8 C and E). These findings are consistent with previous reports that the photoactivation of Rac at the leading edge of the cell can rescue the protrusion defects induced by PI3K inhibition (27, 28). PI3K is not required for the protrusion. Furthermore, to test the hypothesis that actin induces a positive feedback signal to PLEKHG3, cells were treated with cytochalasin D. PA-Rac1 failed to induce protrusion, and PLEKHG3 and F-actin were disrupted (Fig. S8 C and E). Based on these findings, we propose that a positive feedback loop connects actin filaments and PLEKHG3.

Fig. S8.

PLEKHG3 enhances polarized cell migration via a positive feedback loop at the leading edge of the cell. (A) The expressions of PLEKHG3 and F-actin show a similar oscillating pattern during cell migration (n > 75). (B) The localization of PLEKHG3-GFP and mCherry-Lifeact at the leading edge of the cell was observed for 15 min using a confocal microscope with a 100× objective lens. (C) Cells were transfected with PA-Rac1, mCherry-PLEKHG3, and iRFP-Lifeact and were stimulated with light in the presence of DMSO, LY29, or cytochalasin D (Cyto D). Upon light stimulation, PLEKHG3 accumulation was observed in the protrusion area in the presence of LY29 treatment. However, PA-Rac1 failed to induce protrusion, and PLEKHG3 and F-actin were disrupted by cytochalasin D treatment. mCherry-PLEKHG3 and iRFP-Lifeact are shown in a pseudocolored intensity image. The white dotted circles indicate the position stimulated by light. (D) Intensity profiles representing the expression levels of PLEKHG3 and F-actin in B. (E) Fluorescence intensity profiles representing the expression levels of mCherry-PLEKHG3 before and after light stimulation in the presence of DMSO, LY29, or cytochalasin D (n > 25). The data represent the mean ± SEM; *P < 0.1; **P < 0.01. (Scale bar, 20 µm.)

Fig. 2.

PLEKHG3 enhances polarized cell migration via a positive feedback loop at the leading edge of the cell. (A) A model of how polymerized actin filaments increase the accumulation of PLEKHG3 at the leading edge. Gray indicates existing F-actin, red indicates newly polymerized actin filaments, and green circles indicate PLEKHG3. (See Figs. S6–S8 and Movies S2 and S3.) (B) The localization of PLEKHG3-GFP and mCherry-Lifeact at the leading edge was observed via SIM microscopy with a 100× objective lens. Arrowheads show the colocalization of PLEKHG3 and F-actin at the tip of the leading edge of the cell. (C) PLEKHG3 accumulates near newly formed actin filaments. Arrowheads show the newly formed actin filaments. Time is shown in seconds. (Scale bar, 10 µm.) (See Movie S4.) (D) A model for the positive feedback loop that connects actin filaments to PLEKHG3, as assessed using PA-Rac1. (E) Cells were cotransfected with PA-Rac1, mCherry-PLEKHG3, and iRFP670-Lifeact and were illuminated by blue light. PLEKHG3 is recruited to the newly formed protrusion where Rac1 is repeatedly activated by light (R2). mCherry-PLEKHG3 and infrared RFP (iRFP)-Lifeact are shown in a pseudocolored intensity image. The white dotted circle indicates the position that was stimulated by light. (See Movie S5.) (F) Fluorescence intensity profiles representing the expression levels of PLEKHG3 and F-actin over time in region 1 (R1) and region 2 (R2), as assessed from the images presented in E. After light stimulation, the fluorescence intensity decreases in R1 but increases in R2. In R1, red indicates PLEKHG3, and cyan indicates F-actin. In R2, blue indicates PLEKHG3, and green indicates F-actin. (G) Fluorescence intensity profiles representing the expression levels of mCherry-C1 (control) and mCherry-PLEKHG3 before and after light stimulation (n > 50). The data represent the mean ± SEM; *P < 0.1; **P < 0.01. (Scale bars, 20 µm.)

PI3K Controls PLEKHG3 to Guide Directed Cell Migration.

Because PI3K is known to regulate cell polarization, migration, and chemotaxis (29), we asked whether PI3K regulates the ability of PLEKHG3 to induce cell polarity during cell motility (Fig. 3A). To address this question, we treated cells with PDGF, an activator of PI3K. Under serum starvation, PLEKHG3 failed to induce cell polarity. Upon PDGF treatment, PLEKHG3 was translocated to the leading and trailing edges, and cell polarity was induced (Fig. S9 A and D). Conversely, treatment with LY29 caused PLEKHG3-expressing cells to lose their polarity (Fig. S9 B and E). To test whether PLEKHG3 is regulated via PI3K, we induced the activation of PI3K by recruiting the inter-SH2 (iSH2) domain from the cytosol to the PM and examined the distribution of PLEKHG3 localization. When rapamycin was added, PLEKHG3 was translocated to the leading edge of the cell (Fig. 3 B and C). To test whether the recruitment of PLEKHG3 to the leading edge was caused by direct activation of PI3K or indirect activation through Rac-induced actin polymerization, cells were treated with cytochalasin D. In the presence of cytochalasin D, endogenous PI3K failed to induce cell polarity; however, redistribution of PLEKHG3 localization was observed (Fig. S9 C and F). To confirm that PI3K controls PLEKHG3 to guide directed cell migration, we applied light-inducible PI3K (Fig. 3D) (30). We observed consistent migration toward the illuminated area, accompanied by the accumulation of PLEKHG3 upon light stimulation of PI3K (Fig. 3 E–G and Movie S6). Collectively, these results indicate that PLEKHG3 guides directed cell migration via PI3K activation.

Fig. 3.

PI3K controls PLEKHG3 to guide directed cell migration. (A) Schematic of how PLEKHG3 is regulated by PI3K. If PI3K regulates PLEKHG3, then the activation of PI3K triggers cell polarization as a result of PLEKHG3 relocalization. (B) Cells were cotransfected with Lyn-FRB, YFP-FKBP-iSH2 (YF-iSH2), and CFP-PLEKHG3 and were serum starved. PLEKHG3 activity was monitored following the activation of PI3K. Upon rapamycin treatment, PLEKHG3 localization was relocated and was detected mostly at the leading edge of migrating cells. (See Fig. S9.) (C) Intensity profiles representing the expression levels of CFP-PLEKHG3 at the leading edge before and after rapamycin treatment. YFP-FKBP-C1 (YF) is the negative control. (D) A model of PLEKHG3 accumulating at the leading edge and guiding cell migration upon light stimulation. The green line indicates PLEKHG3 expression. (E) Cells coexpressing mCherry-PLEKHG3, mCitrine-PHR-iSH2, and Lyn-CIBN-mCerrulean were serum starved and were illuminated with blue light within 120 min. The cells consistently migrated toward the illuminated area with PLEKHG3 accumulating at the leading edge, whereas the cells in the field without light exposure moved in random directions. The white dotted rectangle indicates the position that was stimulated by light. (See Movie S6.) (F) The kymograph shows the PLEKHG3 intensity without light (R1) and with light (R2) stimulation over 120 min. (G) Fluorescence intensity profiles representing the expression levels of mCherry-PLEKHG3 in response to PI3K activation for 120 min of migration. (H) PLEKHG3^−/− cells were transfected with Lyn-FRB, YFP-FKBP-iSH2, and CFP-Lifeact to test the ability of cells to trigger cell polarization when PIP₃ is produced. The cells were able to induce cell polarity in the absence of PLEKHG3. (I) Fluorescence intensity profiles representing the expression levels of CFP-Lifeact in H9 and PLEKHG3^−/− cells upon rapamycin treatment (n > 25). The data represent the mean ± SEM; *P < 0.1; **P < 0.01. (Scale bars, 20 µm.)

Fig. S9.

PI3K is an upstream target of PLEKHG3. (A) Cells coexpressing CFP-PLEKHG3 and YFP-PH_Akt1 were serum starved for 6 h, PDGF was added to induce PI3K activation, and PI3K activity was monitored using YFP-PH_Akt1. Under serum starvation, PLEKHG3 failed to induce cell polarity. When PDGF was added, PLEKHG3 was translocated to the leading and trailing edges to induce cell polarity and migration. (B) Cells coexpressing CFP-PLEKHG3 and YFP-PH_Akt1 were treated with LY29 to inactivate PI3K. This treatment caused PLEKHG3 to lose its localization at the leading and trailing edges; as a result, cells lost their polarity. (C) The recruitment of PLEKHG3 to the leading edge was caused by the activation of PI3K. Cells were cotransfected with Lyn-FRB, YFP-FKBP-iSH2, and CFP-PLEKHG3 and were serum starved. In the presence of cytochalasin D, endogenous PI3K failed to induce cell polarity; however, the redistribution of PLEKHG3 localization (arrowheads) was observed. Time is shown in minutes. (D) Fluorescence intensity profiles representing the expression levels of CFP-PLEKHG3 upon PDGF treatment. PDGF treatment increased the intensity of the PLEKHG3 signal at the leading edge of the cell. (E) Fluorescence intensity profiles representing the expression levels of CFP-PLEKHG3 upon LY29 treatment. LY29 treatment decreased the intensity of the PLEKHG3 signal at the leading edge of the cell. (F) Fluorescence intensity profiles representing the expression levels of CFP-PLEKHG3 before and after rapamycin treatment in the presence of cytochalasin D (n > 20). The data represent the mean ± SEM; *P < 0.1; **P < 0.01. (Scale bars, 20 µm.)

Several studies have reported that PI3K, Rac, and polymerized actin are the three core components of a local positive feedback loop that triggers cell polarization and migration (3, 4, 31). To test the ability of cells to induce polarity in the absence of PLEKHG3, we activated endogenous PI3K activity in PLEKHG3^−/− cells. The PLEKHG3^−/− cells are able to induce cell polarity when PI3K is activated (Fig. 3 H and I); therefore PLEKHG3 is not the only RhoGEF that is activated by PI3K to induce cell polarity. In addition to PLEKHG3, PI3K might regulate other GEFs that can also activate Rac1, such as guanine nucleotide exchange factor VAV2 (32) or T-lymphoma invasion and metastasis 1 (TIAM1) (3, 33, 34), to induce cell polarity and migration.

Based on the findings described above, we propose a tentative model for how PLEKHG3 pathways enhance cell polarity and cell migration. PLEKHG3 is a protein that binds directly to actin filaments. During cell migration, PLEKHG3 first redistributes to the leading and trailing edges (partial polarization); then, when the cell moves forward, it becomes localized at the leading edge (full polarization) (Fig. 4A). PLEKHG3 activity could be activated by PI3K-dependent or PI3K-independent pathways through Rac-induced actin polymerization (Fig. 4B). The activation of actin filaments generates a positive feedback loop involving PLEKHG3 and accounts for the localization of PLEKHG3 at the leading edge of the migrating cell. In conclusion, we show here that PLEKHG3 plays a role in controlling cell polarity and cell motility by selectively binding newly polymerized actin and activating Rac1 and Cdc42 to enhance local actin polymerization and subsequently further promote the recruitment of PLEKHG3 to induce and maintain the leading edge of the cell.

Fig. 4.

A tentative model for how PLEKHG3 pathways enhance cell polarity and cell migration. (A) PLEKHG3 does not induce polarity in nonmigratory cells. In contrast, in migratory cells, it first localizes to the leading and trailing edges to induce cell polarity (partial polarization) and then, when the cell moves forward, localizes mostly to the leading edge (full polarization). (B) PLEKHG3 activity could be activated by PI3K-dependent or PI3K-independent pathways through Rac-induced actin polymerization. The activation of actin filaments generates a positive feedback loop involving PLEKHG3 that accounts for the localization of PLEKHG3 at the leading edge of the migrating cell.

Discussion

We show here that the RhoGEF member PLEKHG3 contributes to the regulation of cell polarity and cell motility through its ability to bind F-actin filaments and to modulate the activities of Rac1 and Cdc42. When PLEKHG3 was recruited to different subcellular locations, new protrusions were induced, and the cell direction was changed upon light stimulation. PLEKHG3 binds directly to dynamic actin at the leading edge through an ABD that spans amino acids 910–940. Sequence comparison of the ABD of PLEKHG3 with those of other proteins known to bind to the actin cytoskeleton failed to identify any sequence similarity within known ABD proteins. The activity of PLEKHG3 depends on the presence of both the DH–PH domain and the ABD; deletion of either of these domains disrupts the localization of PLEKHG3 to the leading edge and abrogates cell polarity. In analyzing PLEKHG3 localization during cell migration, we found that PLEKHG3 relocates mostly to the leading edge of the migrating cell to help the cell move forward. Knockout PLEKHG3 cells showed a substantial decrease in cell motility compared with control cells. These findings suggested that PLEKHG3 has an important role in the regulation of cell migration. This effect was similar to that of cells lacking Coronin 1B, an F-actin–binding protein that is localized to the leading edge of migrating fibroblasts (35). Cells that lacked PLEKHG3 had a longer shape, similar to that observed in cells lacking other well-known proteins, such as Cofilin (36) or WISp39 (37), that are localized to the leading edge. There was no significant difference in the actin structures of the H9 and PLEKHG3^−/− cells. The elongated cell shape in the PLEKHG3^−/− cells might result from the enrichment of myosin II at the trailing edge of cells, which leads to antagonistic protrusion activity (38, 39), or from an asymmetric localization of the Arp2/3 complex to the leading edge of the elongated cell (36).

Here, we observed that PA-Rac1 led to the formation of lamellipodia, and PLEKHG3 was detected in the newly formed protrusion area on a characteristic time scale of ∼30 s. Closer observation showed the recruitment of PLEKHG3 at the newly formed actin filaments at the leading edge. Our finding is similar to those of previous studies involving the PI3K signaling pathway. PI3K recruitment and its lipid products accumulate within 1 min after lamellipodium induction (40). The accumulation of PLEKHG3 in the protrusion area was observed in the presence of LY29 treatment. This finding provides evidence that a positive feedback loop consisting of PLEKHG3, Rac1/Cdc42, and actin filaments, is involved in regulating the cell migration machinery at the leading edge of the cell. It also explains why PLEKHG3 is recruited rapidly to the leading edge. These findings suggest that PLEKHG3 is involved in two positive feedback loops, one that involves PI3K, Rac1, and actin filaments (4, 6, 31, 41), and another involving Rac1 and actin filaments. Recruitment of PLEKHG3 to the leading edge by PI3K activation during cell migration triggers the local activation of Rac1 and Cdc42, thereby inducing further actin nucleation and branching. This effect appears to explain why PLEKHG3 is specifically localized at the leading edge, whereas other well-known actin-binding GEFs, such as Frabin and FLJ00018, were not detected at that location (20, 21). Some reports have indicated that several actin-binding RhoGEFs, such as PDZ-RhoGEF (42) and FLJ000018 (20), have negative regulatory effects on GEF activity. Here, in contrast, we found that the binding of actin filaments to PLEKHG3 has a positive regulatory effect on GEF activity.

Our findings suggest that the interaction of PLEKHG3 with actin filaments at the leading edge of the cell is an important aspect of cell polarity and cell motility. It explains how Rac1 and actin polymerization are coupled by positive feedback, thereby stabilizing cell polarity.

Materials and Methods

The NIH 3T3 cells were transfected using a Neon transfection system (Invitrogen). The PLEKHG3^−/− cells were transfected using Lipofectamine LTX with Plus Reagent (Invitrogen). Live-cell imaging was conducted using a Nikon A1R confocal microscopy. Detailed experimental procedures are provided in SI Materials and Methods.

SI Materials and Methods

Plasmid Construction.

Sixty-three human GEF cDNAs were generated from human brain tissue mRNAs, and the Gateway system (Invitrogen) was used to construct a library of CFP-conjugated GEF plasmids. The mCherry-Lifeact vector was constructed by inserting the F-actin peptide–encoding sequence into the mCherry-C1 vector (22). Raichu-Rac1 and Raichu-Cdc42 were described previously (43). PTriEx-mVenus-PA-Rac1 (Addgene plasmid #22007) was a gift from K.M.H. GFP-TEM4 was generously provided by Natalia Mitin at the University of North Carolina at Chapel Hill, Chapel Hill, NC (19). Lyn-CIBN-mCerulean was cloned by inserting the myristoylation and CIBN sequences into the mCerulean-N1 vector. The mCitrine-PHR_CRY2-iSH2 construct was cloned by inserting PHR_CRY2 and the iSH2 domain of p85beta (amino acids 420–615) into the pmCitrine-C1 vector (Clontech) by fusing an (SG₇)₃ linker between the PHR_CRY2 and iSH2 domain sequences (30).

Cell Culture and Transfection.

The NIH 3T3 cells were maintained in DMEM containing 10% (vol/vol) FBS (Life Technologies), 100 U/mL penicillin, and 100 U/ mL streptomycin in humidified air (10% CO₂) at 37 °C. Cells were transfected using a Neon transfection system (Invitrogen) under the following conditions: voltage, 1,280 V; pulse, 2; width, 20 ms. A 96-well glass-bottomed black plate (89626; Ibidi) was precoated with fibronectin (1:500) (Invitrogen), and the cells were imaged 20–24 h posttransfection. The PLEKHG3^−/− cells were transfected using Lipofectamine LTX with Plus Reagent (1:1:1) (Invitrogen).

Generation of the PLEKHG3-Knockout Cell Line.

Gene targeting.

The guide RNA (gRNA) target sequence was subcloned into hCas9-2A-eGFP expression vector (pX458; Addgene plasmid no. 48138). The oligonucleotides for gRNA were as follows: PLEKHG3-gRNA, forward 5′-CACCcgctgcccggctgttgaacg-3′, reverse 5′-AAACcgttcaacagccgggcagcg-3′. H9 hESCs were nucleofected using Nucleofector 2b (Lonza). Forty-eight hours later, the EGFP-expressing cells were FACS purified and plated on mouse embryonic fibroblasts (MEFs). The colonies were picked ∼10–14 d later and were passaged twice before genomic DNA isolation. The PLEKHG3 genomic region flanking the gRNA-binding site was PCR amplified (forward primer: 5′-ACCTCTACCACCTCCTCGTC-3′, reverse primer: 5′-GCACAGCCAGGAAACAACAG-3′). The purified PCR products were subjected to a reannealing process to enable heteroduplex formation and were treated with SURVEYOR nuclease and SURVEYOR enhancer S (Integrated DNA Technologies). Simultaneously, the targeted region of the PLEKHG3 gene was PCR amplified and cloned into pCR2.1-TOPO vector (Invitrogen). The insertion sequence was verified by DNA sequencing to ensure that both alleles (from each hESC colony) were represented. The clones with biallelic nonsense mutations were expanded and differentiated for follow-up assays.

hESC culture and fibroblast differentiation.

The undifferentiated H9 hESC line was cultured on mitotically inactivated MEFs (Applied StemCell, Inc.) in a medium containing DMEM/F12, 20% (vol/vol) knockout serum replacement, 0.1 mM Eagle’s minimum essential medium-nonessential amino acids (MEM-NEAA), 1 mM l-glutamine, 55 μM β-mercaptoethanol (Life Technologies), and 4 ng/mL FGF2 (R&D Systems) (hESC medium) in 5% CO₂ at 37 °C (44). For fibroblast differentiation, the culture medium was changed gradually from hESC medium to a medium containing MEM α (GlutaMAX supplement, no nucleosides), 10% FBS for 2 wk. These cells were maintained further for at least 4 wk in a medium containing DMEM, 10% FBS, and 1 mM l-glutamine. Cells were coated in 0.1% gelatin from porcine skin (Sigma) before plating on the flask. Cell medium was changed every 24 h.

siRNA Transfection and Real-Time PCR.

The NIH 3T3 cells were transfected with 25 nM mouse siRNA-PLEKHG3 (SC-152313; Santa Cruz). The MDA-MB-231 and HUVEC cells were transfected with 10 nM of human PLEKHG3-siRNA (SR308671; OriGene). Cells were cultured for 30 h after transfection. To analyze the expression of PLEKHG3 mRNAs, total RNA was isolated using TRIzol (Life Technologies) and reverse-transcribed to cDNAs using SuperScript III (Invitrogen). The generated cDNA was amplified using a 2× real-time PCR smart kit containing EvaGreen (SolGent). The reaction was run at 95 °C for 10 min, followed by 40 cycles of 95 °C for 20 s, 55 °C for 30 s, and 72 °C for 30 s, on a CFX96 Real-Time system (Bio-Rad). All PCRs were performed in duplicate, and the relative transcript expression levels were measured by quantitative real-time PCR using the SYBR Green-based method. The average fold changes were calculated based on between-sample differences in the threshold cycle (Ct).

Reagents and Antibodies.

Rapamycin (Calbiochem) was applied at 0.5 µM for 15 min. LY294002 (Sigma) was applied at 50 µM for 60 min. PDGF-BB (PeproTech) was applied at 10 nM for 30 min. Cytochalasin D (Sigma) was applied at 4 µM for 30 min. Antibodies used were anti-mouse PLEKHG3 antibody (1:1,000) (X3-D3ZGY7; Abmart); anti-human PLEKHG3 antibody (1:500) (AP11167b; AbGent); and Alexa Fluor 594-conjugated phalloidin (1:500) (A12381; Molecular Probes).

Immunofluorescence.

Cells were fixed in 4% (vol/vol) formaldehyde for 10 min at room temperature. Cells were washed with PBS and blocked in PBS and Tween 20 containing 1% BSA. To determine the subcellular localization of PLEKHG3, cells were incubated with anti-mouse PLEKHG3 antibody for 1 h at room temperature. Cells were incubated with Alexa Fluor 488-conjugated anti-mouse IgG antibody (A21202; Invitrogen) (1:500) and Alexa Fluor 594-conjugated phalloidin.

F-Actin Cosedimentation Assay.

To evaluate the direct association of F-actin with PLEKHG3, high-speed actin cosedimentation was performed according to the manufacturer’s instructions (Cytoskeleton Inc.). Purified GST-PLEKHG3 (amino acids 890–950) protein was incubated with F-actin at room temperature for 45 min and centrifuged in an ultracentrifuge (Optima TLX; Beckman Coulter) at 270,500 × g for 1.5 h at 24 °C. The supernatant and pellet fractions were solubilized in SDS sample buffer, resolved by SDS/PAGE [4–12% (vol/vol)] (Invitrogen), and stained with Coomassie Blue.

LARIAT.

LARIAT, which is a method for inhibiting protein function, uses two modules: a multimeric protein (MP) and a light-mediated heterodimerizer. Here, the MP was a version of Ca²⁺/calmodulin-dependent protein kinase IIα (CaMKIIα) that lacked Ca²⁺-responsive catalytic activity. This MP was fused to CIB1, which binds to CRY2 upon blue light stimulation. A single-domain GFP-binding antibody, V_HH(GFP), was fused to CRY2 to recruit GFP-PLEKHG3 to CIB1-MP clusters (24).

Rac1-GTP and Cdc42-GTP Pull-Down Assay.

Measurement of GTP-bound Rac1 and Cdc42 was performed using active Rac1 and Cdc42 pull-down kits (16118 and 16119; Thermo Scientific). Briefly, GST-Pak1-PBD was used to affinity-precipitate GTP-bound Rac1 and Cdc42 from lysed cells. The protein extracts from the HEK293 cell line were separated on a Bolt 12% (vol/vol) Bis-Tris Plus Gel (BG00125BOX; Life Technologies), and the protein was transferred to a PVDF membrane (Life Technologies). The membrane was blocked in blocking solution (1:1) (LI-COR). Precipitated Rac1-GTP and Cdc42-GTP were detected by immunoblot analysis, using mouse monoclonal anti-Rac1 and anti-Cdc42 antibodies (Thermo Scientific). Then samples were labeled with 800-nm channel dye (anti-mouse; 1:3,000; LI-COR Biosciences). Western blotting was analyzed using the Odyssey Imaging System (Odyssey CLx; LI-COR Biosciences).

Microscopic and Imaging Analyses.

All live images were captured using an A1R confocal (Nikon) or an SIM (Nikon) microscopy system. Migration velocity, fluorescence intensity, kymographs, quantitative analysis, and videos were performed with Nikon imaging software (NIS-element) and MetaMorph 7.7 (Molecular Devices). The graph was drawn by an Excel program.

Statistical Methods.

Statistical significance was evaluated by a two-tailed unpaired Student’s t test: *P < 0.1 and **P < 0.01. All values represent the means of three independent experiments, and error bars where shown represent ± SEM.