An endogenous chemorepellent directs cell movement by inhibiting pseudopods at one side of cells.

Eukaryotic chemoattraction signal transduction pathways, such as those used by Dictyostelium discoideum to move toward cAMP, use a G protein-coupled receptor to activate multiple conserved pathways such as PI3 kinase/Akt/PKB to induce actin polymerization and pseudopod formation at the front of a cell, and PTEN to localize myosin II to the rear of a cell. Relatively little is known about chemorepulsion. We previously found that AprA is a chemorepellent protein secreted by Dictyostelium cells. Here we used 29 cell lines with disruptions of cAMP and/or AprA signal transduction pathway components, and delineated the AprA chemorepulsion pathway. We find that AprA uses a subset of chemoattraction signal transduction pathways including Ras, protein kinase A, target of rapamycin (TOR), phospholipase A, and ERK1, but does not require the PI3 kinase/Akt/PKB and guanylyl cyclase pathways to induce chemorepulsion. Possibly as a result of not using the PI3 kinase/Akt/PKB pathway and guanylyl cyclases, AprA does not induce actin polymerization or increase the pseudopod formation rate, but rather appears to inhibit pseudopod formation at the side of cells closest to the source of AprA.

INTRODUCTION

Directed movement of eukaryotic cells by chemoattraction or chemorepulsion is critical during embryogenesis (Theveneau and Mayor, 2012), the trafficking of immune cells during inflammation (Sadik and Luster, 2012; Kolaczkowska and Kubes, 2013), and parasite pathogenicity (Zaki ). Although a considerable amount is known about the mechanisms of chemoattraction, relatively little is known about chemorepulsion. Studies on the movement of starved Dictyostelium discoideum cells toward cyclic adenosine monophosphate (cAMP) and folate have elucidated chemoattraction signal transduction pathways (Chen ; Kortholt ). In a cAMP gradient, Dictyostelium cells use a G protein–coupled cAMP receptor to induce F-actin–driven pseudopods and filopods to move toward the cAMP (Klein ; Kumagai ; Nichols ). Rather than a single pathway, at least six different signal transduction pathways act downstream from the cAMP receptor to mediate motility. These pathways include PI3K/PIP3, TorC2, phospholipase A, guanylyl cyclase, MAPK, and ElmoE (Ma ; Kortholt ; Yan ). For chemotaxis toward folate, Dictyostelium cells use a folate receptor, the G proteins Gα4, Gβ, and Gγ, and Ras (De Wit and Bulgakov, 1985; Hadwiger ; Lim ). Unlike chemoattraction to cAMP, cells moving toward folate do not require PI3K, TorC2, phospholipase A, or guanylyl cyclase signal transduction pathways (Kortholt ).

By activating the cAMP receptor in a different way compared with cAMP, a synthetic analogue of cAMP acts as a chemorepellent by activating the G protein Gα1 and inhibiting phospholipase C to cause an accumulation of PI(3,4,5)P3 at the side of the cell farthest from the source of the analogue of cAMP, and myosin contraction at the side of the cell closest to the source of the analogue of cAMP, which results in the cell moving away from the source of the cAMP analogue (Van Haastert ; Keizer-Gunnink ; Cramer ). AprA is an endogenous secreted protein that induces chemorepulsion of Dictyostelium cells (Phillips and Gomer, 2012). An orthologue of AprA, dipeptidyl peptidase IV (DPPIV), acts as a chemorepellent for neutrophils (Herlihy , 2015, 2017; White ). Unlike the repetitive pulses of cAMP that drive Dictyostelium aggregation during development, AprA appears to be continuously secreted by growing cells, and appears to help cells at the edge of a colony to move away from the colony to find new sources of food. Compared to cAMP pulses, endogenous AprA gradients thus have a negligible temporal component. Cells sense AprA using the G-protein–coupled receptor GrlH to induce chemorepulsion (Tang ). Although loss of phospholipase C or PI3 kinases 1 and 2 reduce chemoattraction to cAMP, cells lacking these components still move away from AprA (Phillips and Gomer, 2012). In this article, we show that in addition to not needing phospholipase C or PI3 kinases 1 and 2, AprA chemorepulsion also does not require the cAMP chemoattraction components protein kinase B, guanylyl cyclase, ElmoE, WasA, and NapA, and that AprA prevents pseudopod formation at the side of a cell closest to the source of AprA, resulting in a movement away from AprA.

RESULTS

AprA chemorepulsion uses a G-protein–coupled receptor, G proteins, and Ras

In cAMP-mediated chemoattraction, several pathways act downstream from the cAMP receptor (Kortholt ; Yan ). To compare the pathway(s) mediating AprA chemorepulsion to the cAMP chemoattraction pathways, we used an Insall chamber to determine whether AprA could induce chemorepulsion in cells lacking key components of the different pathways. A calculation of the theoretical amount of AprA at various distances from the center of a colony of cells indicated that a 0–300 ng/ml AprA gradient in the Insall chamber would reasonably approximate the natural AprA gradient at the edge of the colony (Supplemental Figure S1), and this gradient was used for the work in this article. The movement of cells toward or away from the AprA source in the Insall chamber was measured as the forward migration index (FMI), with a positive FMI indicating chemorepulsion. Defining the x-axis as being parallel to the gradient, FMI is the x-axis distance component of a cell’s movement over the course of the experiment divided by the integrated distance the cell moved along its path. As previously observed, AprA caused chemorepulsion of wild-type AX2 cells (Figure 1A; Phillips and Gomer, 2012). In addition, AprA caused chemorepulsion of wild-type NC4, and chemorepulsion of the axenic strains AX3, AX4, KAX3, JH8, and JH10 (Supplemental Figure S2A). The axenic strains (AX2, AX3, AX4, KAX3, JH8, and JH10) have a mutation in the axeB gene encoding neurofibromin (NF1), which causes them to have increased macropinocytosis compared with the nonaxenic NC4 cells (Veltman ; Bloomfield ). Unlike the axenic strains, NC4 cells were grown on bacteria, so axenic and nonaxenic cells, and cells grown in axenic media and on bacteria, responded to AprA.

FIGURE 1:

AprA uses some but not all components of the cAMP signal transduction pathways. (A) Cells of the indicated strains were imaged for 40 min in growth medium (control) or in an AprA gradient in growth medium in Insall chambers. A positive forward migration index (FMI) indicates chemorepulsion from the AprA and a negative FMI indicates chemoattraction. Values are mean ± SEM from at least 30 cells per strain from ≥3 independent experiments. At left, * indicates p < 0.05, ** indicates p < 0.01, and *** indicates p < 0.001 compared with control for that genotype (t tests). At right, * indicates p < 0.05, ** indicates p < 0.01, and *** indicates p < 0.001 compared with wild-type (WT; t tests). (B, C) The data analyzed for A were also analyzed for persistence of cell movement and cell speed.

The cAMP pathway requires the G-protein–coupled receptor cAR1 and the G proteins Gα2 and Gβ (Wu ; Parent and Devreotes, 1996). We previously observed that AprA binds to the G-protein–coupled receptor GrlH, requires GrlH and the G protein Gβ to inhibit proliferation and to induce chemorepulsion, and uses the G protein Gα8 to inhibit proliferation (Bakthavatsalam ; Phillips and Gomer, 2012; Tang ). We observed that compared with wild type, gα8 cells did not move away from AprA (Figure 1A). Although these cells appeared to move toward AprA, the effect was not statistically significant. Cells lacking either of the Ras proteins RasC or RasG show chemoattraction to cAMP, whereas cells lacking both RasC and RasG move randomly in cAMP gradients (Kortholt ). Although rasC and rasG cells moved away from AprA, rasC cells did not (Figure 1A). Ras activation can be assessed by examining the translocation to the cell cortex of a chimeric protein containing the Ras-binding domain (RBD) of the serine/threonine kinase Raf1 and green fluorescent protein (GFP; Kortholt and van Haastert, 2008), and cAMP induces this translocation (Kortholt and van Haastert, 2008). Compared to buffer treated wild-type cells, AprA also induced the translocation of RBDRaf1-GFP to the cell cortex (Figure 2, A and B). In addition, using a pull-down assay of active Ras with Raf-RBD affinity beads, we observed that compared with buffer treated wild-type cells, AprA activated Ras within 5 min (Supplemental Figure S3, A and B). The anti-Ras antibodies appeared to detect multiple bands, and these may be some of the 11 different Ras gene products in Dictyostelium (Kortholt and van Haastert, 2008). We did not observe RBDRaf1-GFP concentrated at the cortex in grlH cells, suggesting that RBDRaf1-GFP translocation to the membrane in vegetative cells requires GrlH (Supplemental Figure S4A). Together, these results indicate that cAMP and AprA chemotaxis both use G protein–coupled receptors, G proteins, and Ras, and that both involve receptor-­mediated activation of Ras.

FIGURE 2:

AprA causes activation of Ras. (A–C) Localization of the Ras-binding protein RBDRaf1-GFP in wild-type AX2 and pakD cells incubated with a uniform concentration of AprA or buffer (control) for 20 min, fixed and stained with DAPI (blue); bar is 20 µm. (B) Quantification of RBDRaf1-GFP translocation in cells. Images are representative of, and data are mean ± SEM of three independent experiments. * indicates p < 0.05 (two-way ANOVA; Fisher’s LSD test).

AprA chemorepulsion does not need, nor appear to activate, PI3 kinase pathway components such as Akt

Downstream from G proteins and Ras, PI3 kinase is part of one pathway that potentiates, but is not required for, cAMP chemoattraction (Andrew and Insall, 2007). We previously observed that PI3 kinases 1 and 2 are not required for AprA chemorepulsion (Phillips and Gomer, 2012). Other components of the PI3 kinase pathway include the Dock180-related RacGEF DockA, the Akt/PKB protein kinase PkbA, the SGK family protein kinase PkgB, the dagA product CRAC (cytosolic regulator of adenylyl cyclase, a pleckstrin homology [PH] domain–containing protein), and the cAMP-dependent protein kinase A catalytic subunit PkaC. Loss or mutation of these proteins causes partial defects in cAMP chemotaxis (Meili ; Comer ; Para ; Scavello ). Phosphatase and tensin homologue (PTEN) regulates PI3 kinase signaling by dephosphorylating the PI3 kinase product PI(3,4,5)P3 to PI(4,5)P2 (Maehama and Dixon, 1998). Loss of PTEN partially decreases chemotaxis to cAMP (Iijima and Devreotes, 2002). docA, pkbA, and pkbA/pkgB cells moved away from AprA, while dagA, pkaC, and pten cells did not (Figure 1A), indicating that some but not all components of the PI3 kinase pathway are necessary for AprA chemorepulsion, and that some PI3 kinase pathway proteins that help cAMP chemoattraction are not needed for AprA chemorepulsion.

cAMP-induced activation of PI3K causes activation of the kinase Akt/PKB, which results in cAMP-induced phosphorylation at Akt substrate sites of many proteins (Cai ). To determine whether AprA has a similar, and possible long-term, effect on Akt substrate phosphorylation, we incubated wild-type cells with AprA, and stained cells and Western blots of cell lysates for phosphorylated Akt substrates. Although AprA appeared to slightly increase levels of the phosphorylated form of a ∼19 kDa Akt substrate, the effect was not statistically significant, and although AprA gradients appeared to slightly increase levels of phosphorylated Akt substrates at the side of the cell away from the source of the AprA, this effect also was not statistically significant (Supplemental Figure S5, A–D). These results suggest that AprA does not significantly activate Akt.

AprA chemorepulsion uses TORC2, phospholipase A, a Ca 2+ channel, and protein kinase C

The target of rapamycin (TOR) is a protein complex that exists in two functionally different forms, TOR complex 1 (TORC1) and TORC2 (Zoncu ). TORC2 is a major regulator of the actin cytoskeleton (Schmidt ). The TORC2 components Lst8 and PiaA mediate cAMP chemoattraction downstream from G proteins and Ras, and loss of Lst8 or PiaA leads to a partial inhibition of chemotaxis in cAMP gradients (Lee ). lst8 and piaA cells did not move away from AprA (Figure 1A), suggesting that whereas TORC2 facilitates cAMP chemoattraction, TORC2 is necessary for AprA chemorepulsion.

Phospholipase A is part of another pathway that mediates cAMP chemoattraction downstream from G proteins and Ras (Chen ), and loss of phospholipase A (PlaA) causes cells to have reduced persistence of movement toward cAMP (Bosgraaf and Van Haastert, 2009). plaA cells did not move away from AprA (Figure 1A), suggesting that phospholipase A is necessary for AprA chemorepulsion.

cAMP causes a rise in intracellular Ca2+ levels by inducing Ca2+ influx (Traynor ). The plasma membrane Ca2+ channel IplA is homologous to the inositol trisphosphate receptors (InsP3Rs) of higher eukaryotes (Traynor ). Disruption of iplA in Dictyostelium blocks cAMP-stimulated Ca2+ influx, but does not appear to affect chemoattraction toward cAMP (Lusche ). iplA cells did not move away from AprA (Figure 1A). The protein kinase C orthologue PkcA is involved in both Ca2+-dependent and -independent signaling during cAMP chemotaxis, and pkcA cells show chemotaxis toward cAMP but with an abnormally high cell speed (Mohamed ). pkcA cells did not move away from AprA (Figure 1A). These results suggest that IplA and PkcA, although not necessary for chemoattraction toward cAMP, are necessary for chemorepulsion from AprA.

AprA chemorepulsion does not require guanylyl cyclases

In addition to the PI3 kinase pathway, the TorC2 pathway, and the phospholipase A pathway, guanylyl cyclases mediate cAMP chemoattraction downstream from G proteins and Ras (Bolourani ). Dictyostelium has two guanylyl cyclases: the membrane-bound guanylyl cyclase GcA and the soluble guanylyl cyclase SgcA (Roelofs and Van Haastert, 2002). cAMP stimulates the activation of guanylyl cyclase in wild-type cells (van Haastert and Kuwayama, 1997). gca cells, lacking both guanylyl cyclases, show defects in chemotaxis toward cAMP (Roelofs and Van Haastert, 2002). However, gca cells moved away from AprA (Figure 1A), suggesting that guanylyl cyclases are not necessary for AprA chemorepulsion.

AprA chemorepulsion requires the MAPK ERK1

Extracellular signal–regulated kinases (ERKs) are a conserved group of mitogen-activated kinases (MAPKs; Cargnello and Roux, 2011). Dictyostelium has two MAPKs, ERK1 and ERK2 (Hadwiger and Nguyen, 2011). In starved cells, cAMP causes a rapid increase in ERK2 phosphorylation, which is followed by an increase in ERK1 phosphorylation, and the ERK1 phosphorylation can be detected by an anti-phospho MAPK antibody that detects threonine phosphorylation at a conserved TXY motif (Schwebs and Hadwiger, 2015). Loss of ERK1 causes strong defects in directionality and speed of the cells toward cAMP (Sobko ). erk1 cells did not move away from AprA (Figure 1A), and actually appeared to move toward AprA. However, AprA did not significantly affect the levels of ERK1 phosphorylation at the conserved TXY motif (Supplemental Figure S6A), suggesting that ERK1, but not ERK1 phosphorylation at the TXY motif, is necessary for AprA chemorepulsion.

AprA chemorepulsion requires RacC and SCAR

Movement of cells toward cAMP involves regulation of the actin cytoskeleton (Devreotes and Horwitz, 2015). A Gβγ effector, ElmoE (engulfment and cell motility E), links cAMP receptor signaling to the actin cytoskeleton during chemotaxis, and elmoE cells have reduced chemotaxis to cAMP (Yan ). Wiskott–Aldrich syndrome protein (WASP) is an actin nucleation–promoting factor that activates the Arp2/3 complex to form branched actin filaments for pseudopod extension (Myers ). Disruption of one of the copies of Dictyostelium WASP (wasA) causes a decrease in chemotaxis toward cAMP (Myers ). The Rho GTPase RacC regulates activation of WASP to potentiate polarized F-actin formation during cAMP chemotaxis (Han ). In contrast, loss of wasA does not affect chemotaxis toward folate (Davidson ). Another actin nucleation–promoting factor that activates the Arp2/3 complex is a WASP-related protein called suppressor of cAR (SCAR). scrA cells and cells lacking NapA, a component of the SCAR complex, show normal chemotaxis toward cAMP (Ibarra ). Although elmoE, wasA, and napA cells moved away from AprA, racC and scrA cells did not (Figure 1A). A cAMP gradient causes Ras to activate downstream proteins including Rho GTPases (Affolter and Weijer, 2005). In starved cells, cAMP causes phosphorylation of Ser71 of RacC as detected by an anti-Cdc42/Rac1-phospho-Ser71 antibody (Schwarz ). AprA did not have a significant effect on levels of phosphorylated RacC (Supplemental Figure S6B). Together, these data indicate that some but not all components of the machinery that regulates the actin cytoskeleton are necessary for AprA chemorepulsion, and that although AprA requires RacC to induce chemorepulsion, AprA does not affect RacC phosphorylation at the Ser71 site.

AprA does not affect persistence or speed of mutant cells

To further understand the mechanism by which AprA affects cell movement, we examined the persistence of cell movement using the data from Figure 1A. Persistence is the Euclidean distance between the start and end point of a cell’s movement divided by the integrated distance traveled by the cell. Cells have a greater persistence when they move directly from start to end points, while cells have a lower persistence when they move randomly. Persistence thus characterizes the straightness of cell movement. In the 0–300 ng/ml AprA gradients used in this report, neither wild-type cells nor any of the mutants examined showed a change in persistence of movement in an AprA gradient compared with no gradient (Figure 1B and Supplemental Figure S2B). This indicates that over a period of 40 min, for cells that did exhibit chemorepulsion from AprA, cells tended to move to an end point with roughly the same persistence (track straightness) whether or not there was an AprA gradient, but when there was an AprA gradient, the tracks tended to be directed away from AprA. As previously observed, some mutants appeared to have inherent changes in persistence of cell movement (Han ; Bosgraaf and Van Haastert, 2009; Para ; Yan ; Scavello ). Together, these results suggest that AprA mediates chemorepulsion without changing persistence of cell movement. The maximum concentration of accumulated AprA for cells in shaking culture is 300 ng/ml (Choe ). We previously observed a significant increase in the directional persistence of wild-type cells in a 0–2 µg/ml AprA gradient compared with cells in no gradient (Phillips and Gomer, 2012), suggesting that cells increase directional persistence if the AprA concentration exceeds physiological concentrations.

During cAMP-mediated chemoattraction, cell speeds tend to increase (Scavello ). To examine whether AprA mediates chemorepulsion by changing the speed of a cell, we used the data from Figure 1A to examine cell speeds. Speed is the path length traveled by a cell divided by the time taken to travel that path. As previously observed for wild-type AX2 (Phillips and Gomer, 2012), AprA did not significantly affect the speed of wild-type or mutant cells, although several mutants had decreased speed in the presence or absence of AprA (Figure 1C and Supplemental Figure S2C). Together, these results indicate that AprA mediates chemorepulsion without changing cell speed, and that as previously observed for some of the mutants where speed was measured for starved cells in a cAMP gradient (Myers ; Han ; Ibarra ; Bosgraaf and Van Haastert, 2009; Para ; Scavello ), some of the mutants have an inherent decrease in speed.

An AprA gradient shifts pseudopod formation from the rear to the front of a cell without changing the frequency of pseudopod formation

Depending on the nature of the gradient, there can be increased pseudopod and filopod formation at the front of a cell during chemoattraction (Heid ), or no increase in pseudopods (Bosgraaf and Van Haastert, 2009). To determine whether chemorepulsion from AprA causes increased pseudopod formation, we examined the morphology of wild-type AX2 and NC4 cells in an AprA gradient (Figure 3, A–D, and Supplemental Figure S7, A–C). Imagining the gradient going east-west, with the AprA source to the west, we considered the front of a cell to be the 90° sector extending from northeast to southeast, the back of the cell to be the sector extending from northwest to southwest, and the sides to be the combination of the remaining two 90° sectors. In the absence of a gradient, we used the same sectors to define an arbitrary front, back, and sides irrespective of the direction of cell movement. In the absence of a gradient, approximately half of the wild-type AX2 cells at any given time had a discernable pseudopod (Figure 3B). As expected for a random distribution of pseudopod directions, the percent of cells with pseudopods at the front was similar to the percent of cells with pseudopods at the back, and there were roughly twice as many cells with pseudopods at the sides. In AprA gradients, the percent of cells with pseudopods at the front was increased, the percent of cells with pseudopods at the back was sharply decreased, and the percent with pseudopods at the side was similar to the control (Figure 3B). Similar results were obtained using videomicroscopy to examine pseudopod protrusions over 20 min in wild-type AX2 and NC4 cells (Figure 3C and Supplemental Figure S7A). Despite changing the location of pseudopods, the presence of an AprA gradient did not significantly change the overall frequency of pseudopod projections of both wild-type AX2 and NC4 cells, but did cause an increase in the frequency of filopod projections of AX2 cells (Figure 3D and Supplemental Figure S7B). AprA gradients did not significantly affect the size of the pseudopods, the length of the filopods, the lifespan of the pseudopods, or the lifespan of the filopods, and in the presence or absence of an AprA gradient, there was a similar weak trend toward increased pseudopod lifetime with increased pseudopod length (Supplemental Figures S8, A–E, and S7C). Together, these results suggest that similar to the chemoattractant cAMP, AprA causes cells to decrease pseudopod formation at the back of a cell and increase pseudopod formation at the front of a cell. However, unlike steep cAMP gradients, which increase the overall pseudopod formation rate compared with no gradient, and like shallow cAMP gradients, which do not increase pseudopod formation (Bosgraaf and Van Haastert, 2009; Van Haastert, 2010), AprA gradients do not increase the rate of pseudopod formation. These results were observed for both axenic (AX2) and nonaxenic (NC4) cells.

FIGURE 3:

AprA enhances pseudopod formation at the front and inhibits pseudopod formation at the back of cells. (A) A wild-type AX2 cell in an AprA gradient. Arrows indicate protruded filopods. The image was taken using a 1.4 NA 60× oil objective; bar is 5 µm. (B) From images of at least 30 cells per experiment, the percent of cells with a discernible pseudopod at the front (the side of the cell away from the AprA source), sides, or back (the side of the cell closest to the AprA source) in the presence or absence of an AprA gradient was calculated. Each point represents the average of the ≥30 cells for each of five independent experiments. Lines represent mean ± SEM of the five averages. A wild-type AX2 cell showing a pseudopod is shown; bar is 10 µm. (C) Approximately 30 cells per experiment were imaged for 140 s; pseudopods were scored as forming at the front, sides, or back; and the percent of pseudopods in these locations was then calculated. Each point represents the average of the 30 cells for each of seven independent experiments. Lines represent mean ± SEM of the seven averages. (D) Quantification of pseudopod and filopod projections per minute in the presence or absence of an AprA gradient, imaging at least 13 cells per experiment. The data represent mean ± SEM of all of the cells from three independent experiments. * indicates p < 0.05, ** indicates p < 0.01, and *** indicates p < 0.001 (two-way ANOVA; Fisher’s LSD test for B and C and a t test for D).

Unlike cAMP, AprA does not increase cytoskeletal actin and myosin II

cAMP induces starved wild-type cells to rapidly increase the amount of actin and myosin II in the detergent-insoluble cytoskeleton (Supplemental Figure S9, B and D; Dharmawardhane ). To determine whether AprA has a similar effect on the cytoskeleton of vegetative cells, we incubated wild-type cells with AprA and assayed for total actin and myosin II as well as actin and myosin II in the cytoskeleton for up to 12 h to examine the possible long-term effects of AprA (AprA appears to form a relatively time-invariant gradient compared with the rapid pulses of cAMP that induce chemoattraction). Like cAMP, AprA did not significantly affect levels of total actin and total myosin II, but unlike cAMP, AprA did not significantly affect levels of cytoskeletal actin or cytoskeletal myosin II (Figure 4, A–D, and Supplemental Figure S9, A–D).

FIGURE 4:

AprA does not significantly affect levels of polymerized actin or myosin II. Cells were incubated in growth medium with AprA or an equivalent amount of buffer for the indicated amounts of time, and whole cell lysates (A, C, and E) or detergent-insoluble cytoskeletons (B and D) were run on SDS–polyacrylamide gels. Gels were stained with Coomassie (A, B), or Western blots of the gels were stained with anti-myosin II antibodies (C, D), or anti-P-threonine antibodies (E). Densitometry was used to estimate levels of actin (A, B), myosin II (C, D), or phosphorylated myosin (E). For B, polymerized actin densitometry was normalized to the total actin (A) densitometry for each timepoint. For C and E, polymerized myosin II and phosphorylated myosin densitometry was normalized to the total myosin II (C) densitometry for each timepoint. Values are mean ± SEM for 3 (C), 4 (A, D, and E), or 5 (B) independent experiments. * indicates p < 0.05 and ** indicates p < 0.01 (t test with Welch’s correction). Images show representative staining for each experiment.

In cell movement, myosin II stabilizes the cytoskeleton by associating with the actin meshwork, and provides force on actin filaments (Levi ). Myosin II is active in its filamentous form, which is negatively regulated by phosphorylation (Liang ). During chemoattraction to cAMP, one report showed no significant change in the level of phosphorylated myosin (Berlot ), while another showed a 1.6-fold increase in myosin phosphorylation (Dembinsky ). In agreement with Berlot , we observed that cAMP did not change the level of phosphorylated myosin (Supplemental Figure S9E). In contrast, AprA caused a slight decrease in phosphorylated myosin II starting at 5 min (Figure 4E). These results indicate that unlike cAMP, AprA does not increase levels of cytoskeletal actin and myosin, and also unlike cAMP, AprA decreases levels of phosphorylated myosin II.

Cells that are unresponsive to AprA do not form actin-rich pseudopods at the front of the cell

To move toward cAMP, Dictyostelium cells rapidly increase the amount of polymerized actin (F-actin) at the leading edge by locally activating signaling cascades (Sasaki ). To determine whether mutants that are defective in AprA-induced chemorepulsion affect actin polymerization, we exposed cells to an AprA gradient and stained for F-actin. As before, we defined the leading edge of a cell in an AprA gradient as the 25% of the cell periphery on the side away from the source of AprA, and for cells in no gradient, we used the same sector of a cell, irrespective of the direction that the cell was moving in. As previously observed, AprA induced F-actin formation at the leading edge of wild-type cells (Figure 5A; Phillips and Gomer, 2012). Many of the mutants that moved away from AprA (rasG, docA, pkbA, pkbA/pkgB, gca/sgcA and elmoE) also showed a significant increase in the number of cells with F-actin at the leading edges in AprA gradients compared with the cells in the control (Figure 5B). In contrast, we did not observe any changes in localization of F-actin in those mutant cells that did not move away from AprA (Figures 1 and 5B). Together these data suggest that mutants that are not repelled by AprA (Figure 1) also do not show more F-actin in the sector of the cell facing away from AprA.

FIGURE 5:

AprA causes a localization of F-actin in Dictyostelium cells. (A) Localization of F-actin in wild-type cells incubated in an AprA gradient for 20 min, fixed, and stained with phalloidin 488 (green) for F-actin and the DNA dye DAPI (blue). DIC represents differential interference contrast. Image is representative of three independent experiments. * indicates that the source of the rAprA was above and to the right of the figure; bar is 10 µm. (B) Quantification of cells with F-actin at the leading edge in AprA gradients or no gradient (control), fixed and stained with phalloidin 488. For each experiment, at least 30 cells were examined and the percent of cells with F-actin localized at the edge away from the AprA source was calculated. Values are mean ± SEM of the percentages of cells from three independent experiments. * indicates p < 0.05, ** indicates p < 0.01, and *** indicates p < 0.001 compared with control for the indicated genotype (t tests).

AprA requires PakD and SCAR to regulate Ras

p21-Activated kinase D (PakD) regulates actin assembly, is necessary for cAMP chemoattraction, and localizes to cell protrusions in starved cells (Garcia ). We previously observed that PakD localizes to a punctum in a motile cell, and that loss of PakD abolishes AprA-mediated chemorepulsion (Phillips and Gomer, 2014). To determine whether AprA affects the location of PakD, we analyzed the PakD localization in pakD cells expressing PakD-GFP (Phillips and Gomer, 2014). Compared to controls, AprA induced a significant increase of cells with detectable PakD-GFP puncta, and these puncta did not colocalize with F-actin or myosin II (Figure 6, A and B). These data suggest that AprA affects PakD puncta.

FIGURE 6:

AprA induces translocation of PakD. (A) Localization of ectopically expressed PakD-GFP in pakD cells incubated in growth medium with AprA or an equivalent volume of buffer (control) for 20 min, fixed, and stained with phalloidin 555 for F-actin (red), anti-myosin II (cyan), and the DNA dye DAPI (blue). Arrows in A indicate PakD-GFP puncta within the cells. DIC represents differential interference contrast; bar is 10 µm. (B) Quantification of PakD-GFP puncta in those cells. Quantitation of PakD-GFP puncta was analyzed by observers blinded to whether cells were exposed to exogenous AprA or not. Images are representative of, and data are mean ± SEM of, four independent experiments. * indicates p < 0.05 (paired t test).

In mammalian cells, Pak kinases appear to regulate Ras, and Ras appears to regulate Paks (Tang ; Sun ). To determine whether PakD regulates Ras activation during chemorepulsion, we examined whether pakD cells exhibit translocation of RBDRaf1-GFP to the cell cortex. In the absence of AprA, the percent of pakD cells with cortical RBDRaf1-GFP was approximately the same as the percent of AprA-treated wild-type cells with cortical RBDRaf1-GFP, and AprA did not significantly affect RBDRaf1-GFP localization in pakD cells (Figure 2, B and C). These data suggest that PakD negatively regulates Ras activation.

During chemotaxis to cAMP, a feedback loop involving F-actin and PI(3,4,5)P3 affects Ras activation (Sasaki ). To examine whether the actin cytoskeleton plays a similar role in Ras activation during chemorepulsion, we examined the translocation of RBDRaf1-GFP in scrA cells. We did not observe cortical RBDRaf1-GFP in control or AprA-treated scrA cells (Supplemental Figure S4B). These results indicate that AprA requires SCAR to induce RBDRaf1-GFP translocation to the cell cortex and thus Ras activation.

CnrN maintains cell polarity during AprA chemorepulsion

CnrN is a PTEN-like phosphatase that regulates phosphatidylinositol 3,4,5-trisphosphate (PIP3) levels during chemotaxis toward cAMP (Tang and Gomer, 2008). In starved cells moving toward cAMP, PTEN accumulates at the rear of the cell and dephosphorylates PIP3, preventing pseudopod formation at the rear (Iijima and Devreotes, 2002; Wessels ). Loss of CnrN blocks AprA-induced chemorepulsion (Herlihy ). In cnrN cells, AprA gradients caused enrichment of F-actin at the cell cortex but did not cause an enrichment of F-actin or myosin II at one side or another of the cnrN cells (Figure 7A), and this response was rescued by expressing CnrN in the cnrN cells (Figure 7, B–D). Using Western blots to analyze crude cytosol and crude membrane/cytoskeleton fractions, we previously found that AprA causes an increase in membrane and/or cytoskeleton-associated myc-tagged CnrN expressed in cnrN cells (Herlihy ). We observed that the myc-CnrN localized to the rear of cnrN cells and approximately colocalized with myosin II in AprA gradients (Figure 7, B and C). These results suggest that AprA causes a translocation of CnrN to the rear of cells and that CnrN is necessary for the biased cell cortex enrichment of F-actin and/or myosin II during AprA-mediated chemorepulsion.

FIGURE 7:

AprA causes the PTEN-like protein CnrN to translocate to the back of cells. (A, B) cnrN and cnrN cells ectopically expressing myc-tagged CnrN (CnrN) were incubated in the presence or absence (control) of an AprA gradient for 20 min, fixed, and stained with phalloidin 555 for F-actin (red), anti-myosin II (cyan), or anti-myc (green). Images are representative of cells from three independent experiments. * indicates that the source of AprA was above and to right of the image and the direction of cell movement was down the AprA gradient (toward the lower left). DIC: differential interference contrast; bars are 10 µm. (C) Approximately 80 cells from B were scored for CnrN translocation to the rear of the cell, and the percent of cells with CnrN translocation was calculated. Scores were generated by blinded observers. Values are mean ± SEM of the percentages from three independent experiments. (D) The percent of CnrN cells with F-actin at the leading edge was scored as in C. Values are mean ± SEM of the percentages from three independent experiments. * indicates p < 0.05 (paired t test).

DISCUSSION

We used different Dictyostelium knockout strains to compare the signal transduction pathways utilized by the Dictyostelium chemorepellent AprA to the chemoattractant cAMP signal transduction pathways. Some genes that are involved in chemoattraction to cAMP are necessary for chemorepulsion from AprA, while other genes that are involved in chemoattraction to cAMP do not appear to be necessary for chemorepulsion from AprA (Figure 8). A possible reason that the AprA chemorepulsion mechanism does not need some of the genes that are necessary for chemoattraction to cAMP is that chemorepulsion may require less directional accuracy than attraction to a point. In steep but not shallow cAMP gradients, activation of the cAMP signal transduction pathways induces actin polymerization and the formation of new pseudopods. Possibly as a result of AprA not using several of these pathway components, AprA does not induce actin polymerization or the formation of new pseudopods, but instead simply inhibits pseudopod formation at the side of the cell closest to the source of AprA. Like chemorepulsion from AprA, chemoattraction to folate does not require many of the cAMP signaling components, suggesting that some basic signal transduction pathways are sufficient to induce cell migration in gradients of attractants and repellents.

FIGURE 8:

Working model for chemorepulsion and chemoattraction signal transduction pathways. (A) The chemorepellent AprA signals through the GrlH G protein–coupled receptor, and G proteins Gα8, Gβ, and Gγ. AprA requires GrlH to activate Ras, which possibly activates TorC2, Pla2/Ca2+, MAPK, and some components of PI3K signaling to regulate actin cytoskeleton network proteins such as RacC, PakD, CnrN, and SCAR/WAVE during cell movement without inducing F-actin and pseudopod formation. (B) cAMP pulse–induced signaling pathways, which include cAR1 G protein–coupled receptor, G proteins Gα2, Gβ, and Gγ. cAMP activates Ras and other signaling pathways such as PI3K/PIP3, TorC2, Pla2, cGMP, MAPK, and ElmoE and cause actin cytoskeleton proteins such as Rac1, PakA, N-WasP, and SCAR/WAVE to induce enhanced F-actin and pseudopods.

The beginning of the cAMP chemoattraction pathway involves a G protein–coupled receptor activating G proteins and Ras, and AprA uses the G protein–coupled receptor GrlH, and the G proteins Gα8 and Gβ (Phillips and Gomer, 2012; Tang ). The AprA-related neutrophil chemorepellent DPPIV also uses a G protein–coupled receptor to repel cells (White ). cAMP pulses cause Ras to be activated within 1 min, and deactivated within 2–3 min (Kortholt ). We found that Ras is also necessary for, and activated by, AprA signaling. Ras was activated at both 5 and 20 min after adding AprA, indicating a much slower, if any, deactivation process. Both axenic cells (with high macropinocytosis) and NC4 cells (with normal macropinocytosis) were repelled by AprA, indicating that abnormally high levels of macropinocytosis do not interfere with AprA-mediated chemorepulsion (Veltman ; Bloomfield ). PakD and SCAR affect the ability of AprA to activate Ras, indicating that either other external signals or internal conditions modulate chemorepulsion, or that feedback loops amplify or attenuate chemorepulsion.

cAMP chemoattraction uses at least six pathways downstream from the cAMP receptor (Ma ; Kortholt ; Yan ). AprA chemorepulsion appears to use components of the TorC2, phospholipase A/Ca2+, and MAP kinase pathways. Although AprA chemorepulsion requires the MAP kinase pathway component ERK1, AprA does not appear to affect ERK1 phosphorylation at the Thr202/Tyr204 site, suggesting either that AprA chemorepulsion simply requires a basal level of ERK1 activity or that AprA regulates ERK1 at a different site.

AprA chemorepulsion does not appear to use components of the cyclic GMP, ElmoE, and PI3 kinase pathways. Although a chemorepellent cAMP analogue requires PI3 kinase and phospholipase C, AprA chemorepulsion does not require PI3 kinases 1 and 2 or phospholipase C (Keizer-Gunnink ; Phillips and Gomer, 2012) and here we also found that other components of PI3 kinase signaling such as Akt/PKB homologues PkbA and PkgB, and Dock-like protein (DockA) are not necessary for AprA chemorepulsion. In addition to being activated by PI3 kinase activity, Akt/PKB is also activated by TorC2 (Kamimura ). However, AprA did not discernibly affect the phosphorylation of Akt/PKB substrates, suggesting that Akt/PKB and its downstream targets are not involved in the AprA signaling pathway.

Although the Ca2+ channel IplA is not important for cAMP chemoattraction (Traynor ), IplA is necessary for AprA chemorepulsion, indicating that there exist pathways used for AprA chemorepulsion that are not used by cAMP.

PTEN and the PTEN-like phosphatase CnrN are necessary for chemorepulsion from AprA, and AprA causes CnrN to accumulate at the rear of cells. Together, these observations suggest that CnrN is part of the AprA signal transduction pathway. Both phosphatases decrease levels of PIP3 (Comer and Parent, 2002; Tang and Gomer, 2008). Although Akt/PKB is regulated by PIP3, we did not see a statistically significant effect of AprA on Akt/PKB activity as measured by phosphorylation of Akt/PKB substrates. One possible explanation for this is that AprA activates PTEN and CnrN, and the resulting effect of presumably decreased levels of PIP3 at the rear of cells is too subtle to be detected by, or is undetected by, our phosphorylation of Akt/PKB substrate assays. A second possible explanation is that PTEN and CnrN are used for some function in AprA chemorepulsion that does not involve dephosphorylating PIP3. A third possible explanation is that AprA activates PTEN and CnrN, decreasing PIP3 levels, and some other part of the pathway blocks Akt/PKB activity.

Chemotaxis involves regulation of the actin cytoskeleton. In Dictyostelium, the small GTPase Rac activates SCAR, which in turn activates the key actin nucleation factor Arp2/3 (Eden ). As previously observed for cAMP chemoattraction (Veltman ), we found that the loss of RacC or SCAR caused a defect in AprA chemorepulsion. AprA did not significantly affect RacC phosphorylation at the Ser71 site, suggesting that either the basal level activity of RacC is sufficient to activate actin polymerization or that AprA activates Rac at a different site (Schwebs and Hadwiger, 2015).

In starved cells, cAMP increases levels of polymerized actin and myosin II (Dharmawardhane ), and cAMP either increases or has no effect on phosphorylation of the myosin II heavy chain (MHC II; Berlot ; Dembinsky ). We observed that cAMP did not affect MHC II phosphorylation, suggesting that subtle changes in culture conditions among labs affects cAMP-induced MHC II phosphorylation in starved cells. In vegetative cells, AprA did not significantly affect levels of polymerized actin and myosin II, and decreased MHC II phosphorylation. Because dephosphorylation of MHC II causes it to polymerize (Ravid and Spudich, 1989), one possible explanation for decreased MHC II phosphorylation having no effect on MHC II polymerization is that the myosin is already mostly polymerized in vegetative cells. In agreement with the lack of an effect of AprA on actin and myosin II polymerization, there was no effect of AprA on the frequency of pseudopod formation, and AprA simply decreased pseudopod formation at the rear of cells to induce chemorepulsion. Because there exist chemorepellents in higher eukaryotes (Herlihy , 2015, 2017; de Oliveira ; Nourshargh ; White ), an intriguing possibility is that they also use a simple mechanism to induce directional cell movement.

MATERIALS AND METHODS

Cell strains and culture

Dictyostelium discoideum strains were obtained from the Dictyostelium Stock Center (Fey ) and were wild-type AX2, grlH (DBS0350226; Tang ), gα8 (DBS0236107; Wu ), gβ (DBS0236531; Lilly ; Wu ; Peracino ), rasC (DBS0236853; Lim ), rasG (DBS0236862; Bolourani ), rasC (DBS0236858; Bolourani ), plaA (DBS0238068; Chen ), iplA (DBS0236260; Traynor ), docA (DBS0252716; Santorelli ), elmoE (DBS0350065; Yan ), dagA (DBS0235559; Lilly and Devreotes, 1995), pkbA (DBS0349876; Tang ), pkbA (DBS0236785; Meili ), piaA (DBS0349879; Tang ), lst8 (DBS0236517; Lee ), pkcA (DBS0350916; Mohamed ), gca (DBS0302679; Roelofs ), scrA (DBS0236926; Blagg ), pkaC (DBS0236783; Mann and Firtel, 1991), wasA (a gift from Robert Insall, Beatson Institute for Cancer Research, Glasgow, UK; Davidson ), pakD (DBS0350281; Garcia ), cnrN (DBS0302655; Tang and Gomer, 2008), racC (Wang ), pakD (DBS0350395; Phillips and Gomer, 2014), cnrN (DBS0302656; Tang and Gomer, 2008), erk1 (DBS0350622; Nguyen ), and pten (DBS0236830; Iijima and Devreotes, 2002). Cells were grown at 21°C in shaking culture in HL5 medium and in SM/5 agar with lawn of Escherichia coli B/R20 (Dictyostelium Stock Center) and 100 µg/ml dihydrostreptomycin and 100 µg/ml ampicillin were used to kill E. coli in Dictyostelium cultures obtained from SM/5 agar (Brock and Gomer, 1999). Cells expressing a selectable marker were grown under selection with the appropriate antibiotics and supplements (5 µg/ml blasticidin, 5 µg/ml neomycin sulfate, 100 µg/ml thymidine, 20 µg/ml uracil, and/or 25 µg/ml hygromycin). The plasmid pDm115RafRBD (a kind gift from Wouter-Jan Rappel; Kortholt and van Haastert, 2008) encoding the RBD of Raf1 was used to transform wild-type AX2, pakD, and scrA cells by electroporation (Shaulsky ) to generate the strains AX2/RBDRaf1-GFP, pakD, scrA, and grlH. Cells were grown under constant selection and the expression of RBDRaf1 was confirmed by immunofluorescence imaging of the fixed cells as described below.

Diffusion calculation

The theoretical concentration of AprA at various distances from the center of a disk of cells was calculated following Yuen and Gomer (1994) for a 3.79-mm-radius disk of cells in a square grid with 10 µm center-to-center spacing (essentially confluent). The diffusion coefficient (3 × 107 cm2/s) was calculated for the 55 kDa AprA, adjusting for diffusion on moist dirt or on agar (Yuen and Gomer, 1994) using a secretion flux of 2.6 × 10−8 ng AprA/cell per min (Choe ).

Reverse transcription (RT-PCR) analysis

To verify the genotype of strains, total RNA was isolated from wild-type AX2 and mutant strains using a QuickRNA miniprep kit (Zymoresearch) and cDNA was generated using a Maxima H minus first strand synthesis kit (Thermo Fisher). PCR with gene-specific primers was then used to verify loss of the cDNA associated with a gene disruption (Supplemental Figure S10 and Table 1) using specific primer pairs for each gene, and a primer pair for glyceraldehyde 3-phosphate dehydrogenase (gpdA) as a positive control (Brock and Gomer, 1999).

TABLE 1:

Oligonucleotides for genotyping of mutants by PCR.

grlH Forward: 5′-TCTAGCCAAGGTCATACTGTCGC-3′

grlH Reverse: 5′-AGATCGTTGGTGGCTCTTTCGG-3′

rasC Forward: 5′-AGCCGGTCAAGAAGAGTATAGCG-3′

rasC Reverse: 5′-CCCCTTTTCTTTGGTGGGAGCA-3′

rasG Forward: 5′-GTTGGTGGTGGTGGTGTCGG-3′

rasG Reverse: 5′-ACACAAAGGAAACCTTGACCAGTTC-3′

plaA Forward: 5′-GACGGTGGTGGCACAAAAGG-3′

plaA Reverse: 5′-GGGTATCAGCTCCATCTGGTGA-3′

iplA Forward: 5′-CGATACCATTGCATTTTGGTGCG-3′

iplA Reverse: 5′-CGACAGACAGAGCAAACGCAG-3′

docA Forward: 5′-AGGTTGGCCAATAGTTTGTCGT-3′

docA Reverse: 5′-CATAAGAGTCACGGCCGCCT-3′

elmoE Forward: 5′-CCATCATATTGGGGGCGGCG-3′

elmoE Reverse: 5′-GGGGCCACAACAGCTTGAGA-3′

dagA Forward: 5′-CAGCTGTCACTGAAGAGGCAATC-3′

dagA Reverse: 5′-CACCACCAGTGGCACCAACTC-3′

pkbA Forward: 5′-GGCGCAGAGATCGTATTGGCA-3′

pkbA Reverse: 5′-CCAAGAGGGAACGAGCATCTGG-3′

pkbR1 Forward: 5′-GGTAAACCCGCCAAAGCAGG-3′

pkbR1 Reverse: 5′-GGTGAATCTGGTGAACTTGATGCT-3′

piaA Forward: 5′-CGTACTGCTTGTGTAGCATTGGA-3′

piaA Reverse: 5′-CCCTACCAGGCAATTCAGCCA-3′

lst8 Forward: 5′-GCAGCAGCAGGTAACCCACA-3′

lst8 Reverse: 5′-TGCAACCACTAAACCACCATCACT-3′

pkcA Forward: 5′-TGGCGTTTCCAGATTCGTAGAGG-3′

pkcA Reverse: 5′-CAGGTGGTGGAGTTAGGAGGG-3′

gcA Forward: 5′-TGGATTCACTGAAATGGCCGCA-3′

gcA Reverse: 5′-CCCACGACACCAGCAATAACAGG-3′

sgcA Forward: 5′-GGCTCTAACAGTGGAAGGTTGTGC-3′

sgcA Reverse: 5′-GGGGACAGTGACAGCAGTAGCA-3′

napA Forward: 5′-CAACCACCCGCTCCAACCTC-3′

napA Reverse: 5′-TCAGCATGTTCCTCGGCACG-3′

scrA Forward: 5′-ACGTGAGCGTAGAGAAGCACG-3′

scrA Reverse: 5′-AGAGGATGGTGAAGGGGTGC-3′

pkaC Forward: 5′-CGCAGGCGGTGAGGTGTTTAC-3′

pkaC Reverse: 5′-GTGCCCACCAATCGACCGC-3′

wasA Forward: 5′-ACAAATGGTTGGTGGTTCAGCTCC-3′

wasA Reverse: 5′-TCATTTCTTCCACCGCCGCT-3′

pakD Forward: 5′-TCTTCTTCAACGCCACCTCCC-3′

pakD Reverse: 5′-GCACATGGTGGTAATGGTGGAG-3′

cnrN Forward: 5′-ACAGGCTTAGAAGCAAGTTGGAGA-3′

cnrN Reverse: 5′-ACGTTGTTGTGAAGGTTGAGTTACA-3′

gα8 Forward: 5′-CTGACCCCGAAACAAAGAAAAGAGC-3′

gα8 Reverse: 5′-TGACCACCAACATCAACAACACGG-3′

gβForward: 5′-CCAACAAAGTCCACGCTATTCCA-3′

gβ Reverse: 5′-GCGGTAGCATCACAAGCACC-3′

racC Forward: 5′-TGCAAACAATCGTTTCCCAGAAGA-3′

racC Reverse: 5′-GGCTAAATCGTTACCTTGTTCGGT-3′

gpdA Forward: 5′-ACCGTTCACGCCATCACTGCC-3′

gpdA Reverse: 5′-GACGGACGGTTAAATCGACGACTG-3′

pten Forward: 5′-AGTTGCAGTCTCTAAACAAAAGAG-3′

pten Reverse: 5′-GGTGCGTCTGATGCTACAAC-3′

erk1 Forward: 5′-GCAACTCGTTGGTATCGTGC-3′

erk1 Reverse: 5′-TGGATGTGCCAAAGCTTCCT-3′

Recombinant AprA, proliferation inhibition, and chemorepulsion assays

Recombinant AprA (rAprA) was expressed in E. coli, purified, and checked for purity as described previously (Brock and Gomer, 2005). The rAprA was resuspended in 20 mM NaPO4, pH 7.4, to typically 250 µg/ml. The rAprA was tested for proliferation inhibition and chemorepulsion activity before use for further experiments. Proliferation inhibition was measured as previously described (Bakthavatsalam ). Chemorepulsion activity in terms of FMI and speed was measured in Insall chambers as previously described (Muinonen-Martin ; Phillips and Gomer, 2012), with HL5 in the center control well and 300 ng/ml rAprA (diluted from the stock in 20 mM NaPO4) in HL5 in the outer well (White ). At least 10 cells per experiment were tracked by a blinded observer every 15 s for 40 min. As a control, cells were tracked as above but with a volume (equal to the volume of AprA added to the HL5 for the gradient experiment) of 20 mM NaPO4, pH 7.4, added to the HL5 in the outer well. The persistence of cell movement was measured as the directedness of cell movement in the AprA gradient irrespective of the direction of gradient and was calculated as previously described for directionality (Phillips and Gomer, 2012). The chemorepulsion assay for each strain (in the presence and absence of an AprA gradient) was repeated at least two times for a total of at least three independent replicates.

Imaging of cells in AprA gradients

Dictyostelium cells at 8 × 104 cells/ml in HL5 were grown overnight in type 354118 eight-well slides (Corning, Big Flat, NY) with 300 µl/well in a metal egg poacher filled with water to generate an approximately isothermal environment to minimize convection currents. Recombinant AprA (1.8 μl of 50 µg/ml) was added very gently to the top right corner of each well. After 20 min, the medium was gently removed using a pipette tip at the top right corner of the well. Cells were fixed with 4% wt/vol of paraformaldehyde (Electron Microscopy Sciences, Hatfield, PA) in phosphate-buffered saline (PBS) for 10 min, washed once with PBS, and permeabilized with 0.1% Triton X-100 (Alfa Aesar, Tewksbury, MA) in PBS for 5 min. Cells were washed with PBS twice, blocked with 1 mg/ml type 0332 bovine serum albumin (VWR, Solon, OH) in PBS for 1 h and washed once with PBS. Cells were incubated with streptavidin blocking solution (Vector Laboratories, Burlingame, CA) for 15 min. Cells were washed once with PBS with 0.1% Tween 20 (PBST; Fisher Scientific, Pittsburgh, PA), incubated with biotin blocking solution (Vector Laboratories, Burlingame, CA), and washed three times with PBST. Primary antibody solution in PBST was added (1:100 mouse monoclonal 3E6cE8 anti-myosin II [a kind gift from David Knecht, University of Connecticut, Storrs, CT]; Knecht and Loomis, 1987), 1:500 rabbit monoclonal anti-Myc tag (Cell Signaling Technology, Danvers, MA), or 1:500 rabbit monoclonal anti-phospho-Akt substrate [RXXS*/T*] (Cell Signaling Technology) and incubated at 4°C overnight. Cells were washed three times with PBST, incubated with 1:500 biotinylated donkey anti-mouse (Jackson ImmunoResearch, West Grove, PA) in PBST for 1 h, washed three times with PBST, incubated with 1:500 Alexa 647 streptavidin (Cell Signaling Technology) or 1:500 Alexa 488 anti-rabbit (Jackson) in PBST for 1 h, and stained with 165 nM of Alexa 488 phalloidin (Thermo Fisher Scientific, Waltham, MA) or 1:2000 ifluor 555 phalloidin (Abcam, Cambridge, MA) in PBS in the dark for 30 min. Cells were washed three times with PBS and dried for 15–20 min, and mounted in Vectashield hard set mounting medium with 4’,6-diamidino-2-­phenylindole (DAPI) (Vector) following the manufacturer’s directions. Each washing was done for 5 min and all the steps were performed at room temperature unless specified otherwise. For cells that were stained with only phalloidin, the permeabilization step was followed by washing three times with PBS and then incubation with Alexa 488 or ifluor 555 phalloidin in PBS. Images of cells were taken with a 40× objective on a FV1000 confocal microscope (Olympus, Center Valley, PA), processed with Olympus Fluoview Ver.4.2a software, ImageJ, and CorelDRAW X8. Fluorescence intensity was analyzed by Plot Profile in ImageJ software and translocation of RBDRaf1-GFP to the cell cortex was measured as a percent of total cells with cortical RBDRaf1-GFP. Quantification of F-actin, the presence or absence of PakD puncta, and the presence or absence of CnrN localization, were analyzed by blinded observers. In type 354118 eight-well slides (Corning, Big Flat, NY), the front of a cell was defined as the side of the cell facing away from the top right corner, where AprA was added to create an AprA gradient. The back of a cell was defined as the side of the cell facing toward the top right corner. For control wells, where buffer was added to the top right corner, irrespective of the direction of cell movement, the front of a cell was defined as the side of a cell facing away from the top right corner and the back was defined as the side facing toward the top right corner.

Ras activation assay

Ras activity in AX2 wild-type cells was measured using the pull-down assay kit (Cat. #BK008-S; Cytoskeleton, Denver, CO). In brief, 1 × 106 cells/ml were treated with 300 ng/ml rAprA in 20 mM NaPO4 or an equal volume of buffer for 5 min, 5 × 106 cells were collected by centrifugation at 500 × g for 3 min and washed once with ice-cold Sørensen buffer (Gerisch ), and the pellet was lysed with lysis buffer (provided in the kit). The lysate was clarified by centrifugation at 10,000 × g for 1 min at 4°C. Raf-RBD affinity beads were added to the lysate and allowed to bind to the active form of Ras for 1 h at 4°C. Active Ras bound to beads was collected by centrifugation at 5000 × g for 1 min at 4°C. Ras bound beads were eluted in 2X SDS sample buffer with protease inhibitors (Thermo Fisher). SDS–PAGE and Coomassie staining of total lysate, and Western blot using anti-pan Ras (provided in the kit) were performed as described below. For quantitation of the immunoblot, bands were normalized to the Ponceau S–stained blot.

Filopod and pseudopod analysis

Filopod and pseudopod projections were analyzed in AprA gradients in an Insall chamber. After using a 10× objective to verify that cells were moving away from the AprA, cells were observed with a 60×/1.4 oil objective on a Nikon Microphot-FX. Images were recorded for 140 s at 2 s intervals. The sizes, locations, frequencies, and lifespans of filopod and pseudopod projections were analyzed in the video images. The Insall chamber has a reservoir A and a reservoir B. Cells on the platform between reservoirs A and B were imaged. If reservoir A had AprA, and B had HL5 medium, the front of a cell was defined as the side of the cell facing reservoir B, and the back was defined as the side of the cell facing reservoir A. For the control, reservoirs A and B both contained HL5, and irrespective of the direction of cell movement, the front was defined as the side of a cell facing toward reservoir B and the back was defined as the side facing reservoir A.

Actin and myosin polymerization, myosin phosphorylation, Akt substrates phosphorylation, and Western blots

Crude cytoskeletons were prepared using a modification of previously described methods (Reines and Clarke, 1985; Dharmawardhane ). Cells were grown to 2 × 106 cells/ml in HL5 in shaking culture, collected by centrifugation at 500 × g for 3 min, resuspended in HL5, collected by centrifugation, and resuspended to 1 × 107 cells/ml. rAprA (300 ng/ml) in 20 mM NaPO4 or an equal volume of 20 mM NaPO4 was added to 200 µl of cells. For whole cell lysates, at various times after addition, cells were mixed with an equal volume of 2X SDS sample buffer with protease and phosphatase inhibitors (Cell Signaling Technology), heated to 95°C for 3 min, and stored at −70°C. For crude cytoskeletons, at various times after addition, the cells were gently mixed with 1 ml of lysis buffer (100 mM PIPES, pH 6.8, 1 mM MgCl2, 2.5 mM ethylene glycol tetraacetic acid [EGTA], 10 mM TAME [N-α-p-tosyl-l-arginine methyl ester hydrochloride], 0.5% Triton X-100, and a 50-ml Pierce Protease Inhibitor Tablet, EDTA-free [Thermo Fisher Scientific, Waltham, MA] in each 50 ml of buffer). Samples were kept on ice for 2 min, and cytoskeletons were collected by centrifugation at 15,000 × g for 5 min at 4°C. The pellet was resuspended in 0.5 ml of lysis buffer, and cytoskeletons were again collected by centrifugation at 4°C. The pellet was resuspended in 200 µl of 1X SDS sample buffer with protease inhibitors (Thermo Fisher) and heated to 95°C for 3 min and stored at −70°C. Total actin and filamentous actin were detected by electrophoresis of the whole cell lysate and crude cytoskeletons, respectively, on Tris-glycine 4–20% polyacrylamide gels (Lonza, Rockland, ME) and staining with Coomassie Brilliant Blue. Western blots of similar gels were stained for 1 h at room temperature with 1:5000 diluted 3E6cE8 anti-myosin II mouse monoclonal antibody following Suess to detect total myosin II and cytoskeleton-associated myosin II. Myosin phosphorylation was assayed following Tang by staining Western blots of whole cell lysates with 1:1000 42H4 mouse anti-phospho-threonine (Cell Signaling Technology) following the manufacturer’s directions. Western blots of whole cell lysates were also stained with 1:1000 rabbit anti-phospho-Akt substrate [RXXS*/T*], anti-phospho-Rac1/Cdc42 [Ser71], or anti-phospho-p44/42 MAPK [ERK1/2] [Thr202/Tyr204] (all from Cell Signaling Technology, following the manufacturer’s directions). Bound antibody was detected with ECL Western blotting kits (Thermo Fisher). Band intensities on Western blots were quantified using Image Lab software. For total actin, total myosin, phospho-Akt substrates, phospho-ERK1, and phospho-RacC, for each independent experiment the band intensities were normalized to the band intensity at t = 0, and the mean ± SEM for each timepoint from the independent experiments was then calculated. For polymerized actin, for each timepoint in each independent experiment, the polymerized actin band intensity was normalized to the total actin band intensity for that timepoint, and then these values were normalized to the t = 0 value. The mean ± SEM of the t = 0 normalized values from the independent experiments was then calculated for each timepoint. Polymerized myosin II and phosphorylated myosin II values were calculated as for the polymerized actin.

Statistical analysis

Prism (GraphPad, San Diego, CA) and Microsoft Excel were used for data analysis using t tests or analysis of variance (ANOVA) with appropriate posttests. Data are shown as mean ± SEM. Statistical significance was defined as p < 0.05.

Click here for additional data file.