Bidirectional regulation of neutrophil migration by mitogen-activated protein kinases.

To kill invading bacteria, neutrophils must interpret spatial cues, migrate and reach target sites. Although the initiation of chemotactic migration has been extensively studied, little is known about its termination. Here we found that two mitogen-activated protein kinases (MAPKs) had opposing roles in neutrophil trafficking. The extracellular signal-regulated kinase Erk potentiated activity of the G protein-coupled receptor kinase GRK2 and inhibited neutrophil migration, whereas the MAPK p38 acted as a noncanonical GRK that phosphorylated the formyl peptide receptor FPR1 and facilitated neutrophil migration by blocking GRK2 function. Therefore, the dynamic balance between Erk and p38 controlled neutrophil 'stop' and 'go' activity, which ensured that neutrophils reached their final destination as the first line of host defense.

Introduction

Chemotaxis, or directed cell migration of cells in response to a gradient of chemoattractant is essential for lymphocytes to find antigens and for neutrophils to find sites of infection and inflammation[1]. Chemotactic cells such as blood neutrophils and neutrophil-like, differentiated HL60 cells respond to chemoattractant, such as fMet-Leu-phe (fMLP), by adopting polarized morphology (polarization) and crawling up the gradient (directional sensing). Extensive studies have been conducted to understand the mechanisms for cell polarization and directional sensing[1-4]. Neutrophils, for example, utilize a self-organizing mechanism that diverges from the same attractant receptor through different trimeric G proteins to break symmetry and polarize[5]. Both Cdc42 and microtubule pathways are important for neutrophil directional sensing[6-9]. In addition to these mechanisms that promote chemotaxis, there are inhibitory mechanisms that instruct migrating cells to stop directional movement in presence of a single attractant, and to navigate when multiple attractants are present[10]. Thus, the inhibitory mechanisms ensure chemotactic cells to reach correct destinations where they perform bactericidal functions for phagocytes and antigen identification for homing lymphocytes. To date, little is known about the regulatory mechanisms that inhibit cell migration.

Besides activating the appropriate trimeric G proteins, attractants promote phosphorylation of their receptors by G protein-coupled receptor kinases (GRKs). This phosphorylation enables the receptors to bind arrestins, which in turn prevents the receptors from activating G proteins and terminates signaling, a process termed desensitization[11-14]. The receptor-arrestin complex is subsequently internalized via either clathrin-dependent or independent pathway[14]. Internalized receptors are sorted to either degradation or recycling compartments[14]. In neutrophils, both formyl peptide receptor 1 (FPR1) internalization and decreased G protein coupling are mediated through GRK2, but do not require arrestins[12, 13]. This GRK-dependent receptor desensitization decreases the number of potential active receptors on the cell surface, thereby reducing the internal signal generated in response to a given concentration of an attractant. However, previous studies reported a negligible chemotaxis defect of cells expressing receptors that cannot be desensitized[15], suggesting receptor desensitization per se is not required for chemotaxis. The role of receptor desensitization in cell migration hence remains unclear.

The MAP Kinases including Erk, Jnk and p38 are involved in inflammation, apoptosis and migration[16,17]. p38 MAPK has been shown to regulate neutrophil chemotaxis both in vivo and in vitro[18-21], though the underlying mechanisms remain unclear. The roles of Erk and Jnk in chemotaxis have not been fully elucidated. Our findings indicate that the two MAPKs, p38 and Erk, differentially regulated neutrophil migration, as p38 MAPK counteracted while Erk enhanced the GRK2-mediated receptor desensitization. Furthermore, we have identified p38 MAPK as a novel G protein-coupled receptor kinase, which bound and phosphorylated FPR1. This p38-mediated phosphorylation prevented GRK2 binding to the same receptor, thus inhibiting GRK2-mediated FPR1 desensitization. Therefore, the p38 MAPK and Erk-mediated signals respectively control the net chemotactic “go” and “stop” behaviors of migrating neutrophils and enable neutrophils to arrive efficiently at the site of infection to carry out their bactericidal functions.

Results

Opposite roles of Erk and p38 MAPK in cell migration

MAPKs have been implicated to regulate cell migration[16,17]. However, whether different MAPK isoforms play distinct roles in cell migration and the underlying mechanisms remain unclear. We treated differentiated HL60 cells with specific inhibitors for MAPKs or RNAi knockdown (), and characterized migratory behaviors of these cells in fMLP concentration gradients generated by the EZ-taxiscan device. While ~80% of control cells (n = 109) migrated up the fMLP (100 nM) gradient and reached the top (Fig. 1a,b; Supplementary Table 1; Supplementary Video 1), only ~20% of the cells treated with RNAi for p38α (the predominant isoform expressed in HL60 cells[22]) or p38 MAPK inhibitor, SB203580 migrated through and reached the top (n = 147, Fig. 1a,b; Supplementary Table 1; Supplementary Video 2). The remaining cells were able to polarize and migrate initially in the attractant gradient, but quickly lost directionality, wandered aimlessly without net forward locomotion, thus ceasing directional migration. The chemotaxis index (CI, the ratio of net migration in correct direction to total migration length[8]) was significantly lower in p38α RNAi and SB203580-treated cells as compared to controls (0.46 and 0.41 vs. 0.72, P < 0.001, ). Similar results were obtained in another chemotaxis assay, in in vivo transmigration assay, and by using human neutrophils ().

Figure 1

Erk and p38 MAPK play opposite roles in neutrophil chemotaxis

(a) Trajectories of control (Ctrl), SB203580 (SB, 10 μM), p38 RNAi, PD98059 (PD, 50 μM) and Erk RNAi-treated cells in 100 nM fMLP gradient. Each trace represents one individual cell trajectory. Three independent experiments were performed, each using > 30 cells per condition, one of representative experiments is shown. Bar, 100 μm. (b) Relative percentage of cells migrated through the entire gradient field (black bars) compared to cells failed to reach the top (open bars) in above five groups and SP600125 (SP, Jnk inhibitor at 10 μM)-treated cells. (c) Chemotaxis index (CI) of above six groups of cells. *, P < 0.001, compared to control, bars indicate mean ± SEM. (d) Migration speed of the six groups of cells. Bars indicate mean ± SEM.

To further demonstrate the regulatory effect of p38 MAPK on cell migration, we examined migratory behaviors in p38α deficient neutrophils obtained from p38α conditional knockout mice[23]. p38α deficient neutrophils also exhibited significantly decreased transmigration in vivo and chemotaxis in vitro (), consistent with the data we observed above in SB203580 or RNAi treated cells.

In contrast to inhibiting p38 MAPK, cells treated with the Erk inhibitor, PD98059 or Erk RNAi showed increased chemotaxis compared to controls, with more than 95% cells that had migrated through the entire gradient (n = 331, Fig. 1a,b; Supplementary Table 1; Supplementary Video 3). The CI of Erk RNAi or PD98059-treated cells was significantly higher than control (0.88 and 0.92 vs. 0.72, P < 0.001, ). Similar results were observed with another Erk inhibitor, U0126 (). However, inhibition of another MAPK, Jnk with SP600125 has little effects on cell migration (). Cell polarization, measured by actin polymerization and cell migration speed, measured before cells lost net forward locomotion, were not affected in those p38 or Erk-inhibited cells (; ). Thus, down-regulation of p38 MAPK inhibits directional cell migration in fMLP gradient, while inhibition of Erk enhances cell chemotaxis, suggesting that these two MAPKs have opposing functions in neutrophil chemotaxis.

Concentration-dependent cell behavioral switch

As high concentrations of chemtoattractants are known to inhibit neutrophil orientation[24], we next characterized chemotactic behaviors of differentiated HL60 cells in different fMLP concentration gradients. In gradients generated with lower concentrations of fMLP (50 and 100 nM), the majority of cells (>70%, n = 186) displayed continuous directional migration up the gradient field (). In contrast, under higher concentrations of fMLP (500 nM and 1000 nM) gradient, only ~10% of cells (n = 355) completed migration through the gradient (). The CI expectedly was significantly lower at higher fMLP concentration gradients (0.38 and 0.27 for 500 nM and 1000 nM, respectively, ) than that at lower fMLP concentration gradients (0.66 and 0.72 for 50 nM and 100 nM, respectively) (). Similar results were obtained in neutrophil transmigration into the peritoneal cavity of wild type mice (), using human neutrophils (), and with cells stimulated in uniform concentrations of fMLP (used for assessment of chemokinesis, ). Thus, cells migrate in response to higher concentration fMLP gradients (for example, 500 nM) exhibits a similar phenotype to cells treated with SB203580 or p38α RNAi migrating in response to lower concentration fMLP gradient (100 nM).

To address whether p38 MAPK or Erk play roles in terminating directional cell migration in above studies, we next assessed the activation of both p38 MAPK and Erk at different concentrations of fMLP. We found that activation of p38 MAPK (measured by its phosphorylation) peaked at 100 nM fMLP and severely decreased at 500 and 1000 nM fMLP (). Activation of Erk also peaked at 100 nM fMLP but as fMLP concentration was further increased, similar amount of phosphorylated Erk was observed, indicating a plateau of Erk activation (). In above activation assays, cells were stimulated for 2 min, as both p-p38 and p-Erk peaked at 2 min for each concentration of fMLP (). Similar results were obtained with human neutrophils ().

To test whether decreased p38 MAPK activity and sustained Erk activity are responsible for impaired chemotaxis at high concentrations of fMLP, we treated cells with either a p38 MAPK activator, anisomycin ()[25] or the Erk inhibitor, PD98059 and tested their chemotactic migration in the 500 nM gradient. Importantly, unlike most control cells that ceased directional migration in the middle of the gradient field, ~70% of p38-activated cells migrated through the field (n = 95, ). The CI thus obtained was significantly higher than control cells (0.76 vs. 0.37, P <0.001), but was comparable to the CI of control cells migrating in 100 nM fMLP gradient (). Similar results were observed when Erk was inhibited with PD98059 (). The above observations indicate that p38 MAPK counteracts while Erk enhances the “stop” signal during cell chemotaxis.

MAPKs differentially regulate FPR1 internalization

We next explored the possible signal for cessation of directional cell migration induced by high concentrations of fMLP. Attractants not only activate the appropriate trimeric G proteins, but also promote receptor internalization and thus preventing receptor from activating G proteins[11-13]. We assessed whether the impaired cell migration at high concentrations of fMLP resulted from reduced receptor availability secondary to receptor internalization. We first found receptor internalization in HL60 cells or human neutrophils, measured by fluorescence-labeled fMLP, increased along with increasing fMLP concentrations (), consistent with a previous report[24]. Further analysis of time response curves of the receptor internalization at 100 and 500 nM fMLP also confirmed above observations ().

Furthermore, to prevent receptor internalization, we used a mutant FPR1, FPR1-ΔST, in which all C-terminal Ser and Thr residues were mutated to Ala or Gly, thus the receptor cannot be internalized[15]. As expected, cells expressing wild-type FPR1 quickly arrested in a uniform concentration of 500 nM fMLP (, top panels). The majority of cells (~70%, n = 33) lost net locomotion within the 15 min recording period (, left panel). In contrast, only ~13% of FPR1-ΔST expressing cells (n = 69) arrested in the same recording period (, right panel). These cells also showed negligible receptor internalization (, bottom panels). Based on trajectories of the two groups of cells (), the mean migration distance of the wild-type cells were significantly less than the FPR1-ΔST expressing cells (17.9 vs. 33.4 μm, P < 0.01, ). In conclusion, increasing concentrations of fMLP enhances receptor internalization and promotes migrating neutrophils to stop. Prevention of receptor internalization restores cell migration at high concentration fMLP, indicating that increased receptor internalization at high concentrations of chemoattractant acts as an essential “stop” signal for chemotactic neutrophils.

We next determined whether p38 MAPK and Erk have opposing effects on receptor internalization. To visualize receptor internalization, we expressed YFP-tagged FPR1 in HEK293 cells which lack endogenous FPR1. Receptor internalization became evident within 15 min after fMLP stimulation in control cells, but was seen as early as 3 min after fMLP in SB203580-treated cells (). In contrast, much of the FPR1-YFP signal remained at the plasma membrane in PD98059-treated cells after 15 min, indicating that there was little if any receptor internalization (). Similar results were obtained in HL60 cells () and human neutrophils (). These results collectively show that Erk and p38 MAPK differentially regulate FPR1 internalization.

GRK2 mediates the “stop” signal for chemotaxis

GRK2 is known to promote FPR1 internalization through phosphorylation[13]. To determine whether GRK2 mediates the “stop” signal during chemotaxis, we knocked down GRK2 in HL60 cells using RNAi (), and recorded migration of these cells in the presence or absence of SB203580. As expected, inhibition of p38 MAPK reduced directional migration of control cells in the 100 nM fMLP gradient (). Knockdown of GRK2 reversed this effect and restored cell migration (), suggesting that SB203580-induced cell arrest was mediated through GRK2. This finding led us to test whether p38 MAPK and Erk could affect GRK2 functions. On examining membrane-associated GRK2 in control cells, we found that it was increased after fMLP stimulation (). Treatment of the cells with SB203580 further increased membrane association of GRK2 at both basal and stimulated states (), indicating that p38 MAPK antagonizes GRK2 membrane recruitment. Similar results were obtained in human neutrophils (). In contrast, PD98059 prevented GRK2 membrane recruitment (), suggesting that Erk enhances the effect of GRK2.

p38 MAPK acts as a novel GRK and blocks GRK2 function

Erk has been shown to prevent GRK2 degradation[26,27]. We observed a similar effect in HL60 cells, as cells treated with Erk RNAi showed markedly decreased GRK2 protein abundance before and after fMLP stimulation (). However, the mechanism of how p38 MAPK regulates GRK2 remains unclear. Using immunostaining, we examined subcellular localizations of p38 MAPK and FPR1 in HL60 cells. The active p38 MAPK (p-p38) appeared in pseudopods of polarized cells; while total p38 MAPK was more uniformly distributed (). FPR1 mainly localized to the cell membrane, at both the leading and trailing edges (). Thus, active p38 MAPK could potentially interact with FPR1 at pseudopods. We tested this possibility using immunoprecipitation, and found increased binding of p38 MAPK to FPR1 after fMLP stimulation, which was inhibited by SB203580 (). Similar results were also found in human neutrophils (). In contrast, active Erk (p-Erk) was uniformly distributed and no interaction was detected between Erk and FPR1 ().

Since p38 MAPK phosphorylates Ser/Thr residues[28], and there are 19 Ser/Thr residues in the intracellular domains including 11 in the carboxyl tail of FPR1 (total 350 amino acids), we examined whether p38 MAPK phosphorylates the receptor. The three intracellular loops and the C-terminal tail of FPR1 were individually expressed as GST-tagged proteins and purified for in vitro phosphorylation assay. After incubation with purified active p38 MAPK, only the C-terminal tail was phosphorylated (GST-C, ); this phosphorylation was inhibited by SB203580 (data not shown). To map the phosphorylation site(s), we generated two mutants: one with the mutations of the six Ser/Thr residues within amino acids 319-332 of the C-terminal tail to Ala or Gly (MUT-A, ) and the other with the mutations of the five Ser/Thr residues among amino acids 334-342 (MUT-B, ). p38 MAPK failed to phosphorylate the MUT-B mutant (). We next mutated each of the five Ser/Thr residues within amino acid 334-342 individually, and identified Ser342 (S342G, ) as the only phosphorylation site for p38 MAPK. Notably, the phosphorylation site for p38 MAPK (Ser342) is not the same as those for GRK2 (), which was previously reported based on site mutagenesis[13] and shown in green color in the figure. The p38 MAPK phosphorylation site was shown in red.

We next examined whether p38 MAPK prevents GRK2 binding to FPR1. In control cells, fMLP stimulation increased binding between FPR1 and GRK2. This binding was further increased in p38 MAPK knockdown cells (). To dissect whether phosphorylation of FPR1 by p38 MAPK prevents GRK2 binding, we constructed FLAG-tagged wild-type and two mutants of FPR1: FPR1-S342A, which prevents p38 MAPK phosphorylation; and FPR1-S342D, which mimics phosphorylation at this site. p38 MAPK bound to wild-type FPR1 and two mutants equally well (). In contrast, GRK2 exhibited markedly decreased binding to FPR1-S342D (the phosphomimetic mutant) comparing to wild-type or FPR1-S342A (). Furthermore, FPR1-S342D expressed HL60 cells migrated for longer distances than wild type FPR1 expressed cells in high concentration fMLP (500 nM, ). Together, these results indicate that p38 MAPK phosphorylation at Ser342 could prevent GRK2 interaction with FPR1. Thus, p38 MAPK functions as a non-canonical G protein-coupled receptor kinase, which counteracts the function of GRK2.

G proteins differentially regulate Erk and p38 MAPK

As shown above Erk and p38 MAPK play opposite roles in cell migration, raising the possibility that Erk and p38 MAPK antagonize each other. However, we found p38 MAPK activation was not altered in Erk RNAi cells, and vice versa (), indicating these two MAPKs are independently regulated. This contention is further supported by findings that p38 MAPK and Erk exhibit distinct activation patterns in response to increasing concentrations of fMLP (). In contrast to continuously activation of Erk, p38 MAPK displays a bell-shaped activation curve. To dissect how Erk and p38 MAPK are differentially regulated, we examined whether different heterotrimeric G proteins activated by FPR mediate Erk and p38 MAPK activation. Inactivation of Gi by pertussis toxin (PTX) abolished p38 MAPK activation, but only partially inhibited Erk activity (~70% inhibition, ), indicating Gi-independent activation of Erk. Since Erk and p38 MAPK exhibit different activation patterns with increasing fMLP concentrations, we performed concentration dependent activation of Erk and p38 MAPK in the presence of PTX. PTX blocked p38 MAPK activation at all fMLP concentrations tested, but only partially inhibited ERK activation at each fMLP concentration (). To screen other possible G proteins that mediate Erk activation, we used RNAi to knock down G proteins (). We found knocking down Gq significantly reduced fMLP induced Erk activation, but did not affect p38 MAPK activation ().

We further examined whether MAPK phosphatase may play a role in the differential regulation of Erk and p38 MAPK. We first tested one MAPK phosphatase, WIP1 that belongs to the protein phosphatase family 2C (PP2C) and is specific for p38 MAPK but not Erk[29]. Cells treated with okadaic acid (1nM, 30 min), a PP2 inhibitor, or RNAi knockdown of WIP1 (), showed a delayed decline of p-p38 (); however, neither treatment altered the bell-shaped activation curve of p38 MAPK (). We next tested two other dual-specificity MAPK phosphatases: MKP1 and MKP5, which both de-phosphorylate Erk and p38 MAPK[30]. Knockout of either MKP1 or MKP5 increased p-p38 abundance upon stimulation with chemoattractants, but still showed a bell-shaped activation curve (unpublished observations). Thus, these phosphatases do not appear to be required for the differential activation of Erk and p38 MAPK under our experimental conditions. The possibility however still remains of differential activation of heterotrimeric G proteins by one or more FPRs, which could be responsible for the observed differences in activation of Erk and p38 MAPK.

Discussion

Understanding how chemotaxis is dynamically and precisely regulated is of great significance due to its vital roles in inflammatory cell infiltration, lymphocyte homing, embryonic development, axon guidance, and tumor invasion. In addition to triggering polarization and directional sensing which are important for initiation of neutrophil chemotaxis, chemoattractant stimulation gradually evokes a distinct stop mechanism that negatively regulates directional cell migration and brings chemotactic cells to a state similar to that of unstimulated cells. This mechanism is represented by the increment of membrane-associated GRK2 and FPR1 internalization. Our results demonstrate that, to achieve efficient and precise directional migration, the stop mechanism is differentially controlled by Erk and p38 MAPK. Inhibiting p38 MAPK enhances the “stop” signal and makes migrating cells arrest, while enhancing p38 MAPK or inhibiting Erk activity appears to overcome the “stop” signal and ensure directed cell migration even at high concentration of fMLP that normally induces termination of cell chemotaxis. Our results not only reveal a mechanism (Erk-GRK2) for termination of chemotaxis, but also demonstrate that sustained chemotaxis requires constant suppression (provided by p38 MAPK) of the stop mechanism before migrating cells can reach their final destinations.

Although p38 MAPK has been shown to play an important role in neutrophil chemotaxis[18-21], the underlying mechanisms are not clear. Here we demonstrate that p38 MAPK acts as a novel GRK, which phosphorylates a chemoattractant receptor and blocks the function of the classical GRK2. Different phosphorylation patterns on GPCR may instruct different signals for downstream partners to perform different functions[31]. Our observation on FPR1 supports this concept. We demonstrated that p38 MAPK and GRK2 phosphorylate the C-terminal tail of FPR1 at distinct non-overlapping sites. Phosphorylation at Ser342 of FPR1 by p38 MAPK prevents the same receptor from interacting with GRK2, thereby blocking the GRK2-mediated “stop” signals and ensuring sustained cell migration. Thus, different phosphorylation patterns on FPR1 by p38 MAPK and GRK2 execute opposite functions downstream of the receptor, indicating the subtlety of p38 regulation of neutrophil migration.

Previous studies reported negligible chemotaxis defects of cells expressing receptors that cannot be desensitized, leading to the concept that receptor desensitization is not required for chemotaxis[15]. Our findings however show that GRK2-mediated receptor internalization and desensitization play an essential role in regulating cell migration and is responsible for the termination of cell migration seen at high concentrations of attractant. Thus, it is the enhancement of receptor desensitization but not its inhibition that blocks neutrophil chemotaxis. Moreover, our findings revealed that sufficient protection of receptor against desensitization is required for sustained cell migration, and over-protection of receptor from desensitization leads to non-stop migration. Therefore to achieve precise navigation of migrating cells, it is critical to control the balance of the acceleration and deceleration of receptor desensitization, which are mediated by GRK2 and p38 MAPK, respectively. In other words, signals that control the receptor desensitization play a central role in determining the final destination of migrating neutrophils.

Our results show that different heterotrimeric G protein signals downstream of the formyl peptide receptor are responsible for the differential regulation of Erk and p38 MAPK. Activation of p38 is dependent on Gi, while both Gi and Gq signals activate Erk. In neutrophils, there are two isoforms of FPRs: FPR1 is the high affinity receptor for fMLP (Kd ~10 nM), which activates Gi; FPR2 is the low affinity receptor for fMLP (Kd ~1 μm), which also activates Gq[32]. Thus, at lower concentrations of fMLP, FPR1 activates Gi, leading to the activation of both p38 and Erk; when fMLP concentration increases, FPR2 is activated, thus activating Gq, which is responsible for sustained Erk activation. At the same time, p38 activity decreases as more FPR1 internalizes when exposed to high concentrations of fMLP. We consider that this differential regulation of two MAPKs provide bidirectional control on GRK2 function at different stages during directional cell migration, thus facilitating migrating cells to accurately reach their destinations.

To reach sites of infection or inflammation, circulating neutrophils must first attach to the blood vessel lining endothelial cells, then transmigrate into tissue and reach sites of infection. The mechanisms responsible for the initial arrest of neutrophils to the endothelium are predominantly mediated by β2 integrins interacting with their endothelial ligand, ICAM-1(refs. [33, 34]). Whether this initial arrest mechanism also plays a role in terminating neutrophil migration remains unclear. Likewise, it is also unknown whether MAPKs or GRKs terminate directional cell migration through regulating integrin activation. Animals such as fruit flies display concentration-dependent behavioral switch in response to odors stimulation, which is mediated by the differential regulation of neural circuits[35]. Such concentration-dependent behavioral switch can also be observed at the cellular level as a bell-shaped dose response curve during cell migration. Whereas initial increase of the concentration of an attractant leads to increased directed cell migration, it eventually peaks off and further increase of the attractant concentration causes gradual decrease of chemotaxis. Therefore studies on the cellular level may provide further molecular mechanisms and insights for understanding behavioral switch in more complex organisms. Our results suggest a model for neutrophil migration in which the dynamic balance among GRK2 and two MAPKs regulates the “go” and “stop” behaviors at the receptor level. This model provides a mechanism for concentration-dependent switch for the cell to continue its movement or stop during directional cell migration. The model of the “stop” and “go” signals as in the present study provides the opportunity for pharmacological intervention of one or more of the specific pathways. Such an intervention would enable appropriate phagocyte infiltration into inflammatory sites while minimizing neutrophil-mediated tissue injury.