Characterization of the MEK5-ERK5 Module in Human Neutrophils and Its Relationship to ERK1/ERK2 in the Chemotactic Response*

The role of the extracellular signal-regulated kinase (ERK) 1 and ERK2 in the neutrophil chemotactic response remains to be identified since a previously used specific inhibitor of MEK1 and MEK2, PD98059, that was used to provide evidence for a role of ERK1 and ERK2 in regulating chemotaxis, has recently been reported to also inhibit MEK5. This issue is made more critical by our present finding that human neutrophils express mitogen-activated protein (MAP) kinase/ERK kinase (MEK)5 and ERK5 (Big MAP kinase), and that their activities were stimulated by the bacterial tripeptide, formyl methionyl-leucyl-phenylalanine (fMLP). Dose response studies demonstrated a bell-shaped profile of fMLP-stimulated MEK5 and ERK5 activation, but this was left-shifted when compared with the profile of fMLP-stimulated chemotaxis. Kinetics studies demonstrated increases in kinase activity within 2 min, peaking at 3–5 min, and MEK5 activation was more persistent than that of ERK5. There were some similarities as well as differences in the pattern of activation between fMLP-stimulated ERK1 and ERK2, and MEK5-ERK5 activation. The up-regulation of MEK5-ERK5 activities was dependent on phosphatidylinositol 3-kinase. Studies with the recently described specific MEK inhibitor, PD184352, at concentrations that inhibited ERK1 and ERK2 but not ERK5 activity demonstrate that the ERK1 and ERK2 modules were involved in regulating fMLP-stimulated chemotaxis and chemokinesis. Our data suggest that the MEK5-ERK5 module is likely to regulate neutrophil responses at very low chemoattractant concentrations whereas at higher concentrations, a shift to the ERK1/ERK2 and p38 modules is apparent.

Neutrophils, while playing an important role in host defense by killing microbial pathogens, are also responsible for tissue destruction in inflammatory conditions such as rheumatoid arthritis (1) and cystic fibrosis (2). A crucial initial step in this is the recruitment of neutrophils to sites of infection and inflammation by a process known as chemotaxis. A number of studies have reported that the extracel-lular signal-regulated protein kinase (ERK) 1 1 and ERK2 modules are involved in regulating neutrophil chemotaxis (3)(4)(5)(6)(7). These studies showed that PD98059, a specific inhibitor of mitogen-activated protein (MAP) kinase/ERK kinase (MEK)1 and MEK2, the immediate upstream regulators of ERK1 and ERK2, inhibited chemotaxis in response to fMLP or IL8. However, PD98059 and another specific MEK1/MEK2 inhibitor, UO126, have recently been reported to also inhibit MEK5, the upstream regulator of ERK5 (8,9). This therefore creates some doubt as to whether the previous conclusion is valid and whether the MEK5-ERK5 module could also be involved in regulating the chemotactic response.
ERK5 or Big MAP kinase (BMK) is a recently described stress-activated MAP kinase (10,11). This kinase is similar in many aspects to ERK1 and ERK2, but has a unique loop 12 domain within the kinase region, which is followed by an unusually long C terminus. Although murine tissues have been reported to express three forms of ERK5 protein species, generated by alternative splicing (12), human tissues contain only one form of the protein, despite alternative splicing in the 5Ј-non-coding region (11). The activity of ERK5 is up-regulated through the phosphorylation of the TEY activation motif by MEK5. Alternative splicing at the 5Ј-end of MEK5 yields MEK5␣ and MEK5␤ and alternative splicing between the kinase subdomain IX-X region of each MEK5 species can potentially generate two further MEK5 species (13). Kinases such as MEK kinase (MEKK)2, MEKK3, tumor progression locus (Tpl)-2/cancer osaka thyroid (Cot) and mixed lineage kinase (MLK)like MAP triple kinase have all be been reported to serve as MAP kinase kinase kinases of the ERK5 module (8,10,14). Recent studies have identified the kinase WNK (with no lysine) 1 to be an upstream regulator of the MEK5-ERK5 module, acting via MEKK2 and MEKK3 (15).
ERK5 is activated by stresses such as H 2 O 2 and shear stress (8,10,16). However, unlike the p38 and JNK stress MAP kinases, ERK5 is not activated by UV irradiation or anisomycin (8). Depending on the cell type, the activity of ERK5 can be stimulated by agents such as serum, epidermal growth factor, 1,25 dihydroxyvitamin D 3 and phorbol 12-myristate 13-acetate (8,17). Substrates that have been identified for ERK5 include transcription factors such as Ets-1, MEF2C, and Sap1a (8,17,18). Recent studies have demonstrated that deletion of ERK5 results in angiogenic failure and death of the embryo (18). Both MEK5 and ERK5 have sequences that suggest that these kinases may interact with cytoskeletal elements such as actin (13,19).
The presence of MEK5 and ERK5 in neutrophils has not been described, and their roles in neutrophils remain unknown. The aims of this study were to characterize the expression and activation of MEK5 and ERK5 in human neutrophils, and to investigate whether this kinase module could regulate neutrophil migration. We have found for the first time, that human neutrophils express MEK5 and ERK5 and the activity of the MEK-ERK5 module was stimulated by fMLP. Activation of the MEK-ERK5 module was blocked by wortmannin, an inhibitor of phosphatidylinositol 3-kinase. Investigations with the recently described specific MEK inhibitor, PD184352, at concentrations that inhibited ERK1/ERK2 but not ERK5 activity, provided further evidence that ERK1 and ERK2 are involved in regulating the chemotactic response.

EXPERIMENTAL PROCEDURES
Materials-Myelin basic protein, protein A-Sepharose, lucigenin, and fMLP were purchased from Sigma-Aldrich Pty. Ltd. (Sydney, Australia). The anti-ERK5 (L-19) and anti ERK2 (C-14) antibodies were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). We also obtained an anti-ERK5 antibody from Sigma-Aldrich and this was used in the majority of our experiments. The anti-MEK5 antibody was purchased from Stressgen Biotechnologies (Victoria, Canada). [␥-32 P]ATP (4000 Ci/mmol) was obtained from Geneworks (Adelaide, Australia). M r markers were from MBI Fermentas (Hanover, MD) and Bio-Rad. pUC19/HpaII and SppI/EcoRI DNA markers were from Geneworks (Adelaide, Australia). PD98059 was purchased from New England Biolabs, Inc., (Beverly, MA), and PD184352 was obtained from Prof. P. Cohen, University of Dundee, Scotland, UK. Millicell-PCF 3-m culture inserts were obtained from Millipore Australia (Sydney, Australia). Renaissance Chemiluminescence Reagent Plus was obtained from PerkinElmer Life Science Products (Boston, MA). Reinforced nitrocellulose was purchased from Schleicher and Schuell (Dassel, Germany). The Western blot recycling kit was obtained from Alpha Diagnostic International (San Antonio, Texas). TRIzol® Reagent was purchased from Invitrogen Life Technologies (Mt Waverly, Australia). AMV reverse transcriptase and Expand High Fidelity polymerase were from Roche Diagnostics Australia (Castle Hill, Australia). Q Buffer was from Qiagen Pty. Ltd. (Clifton Hill, Australia).
Isolation of Neutrophils and Monocytes-Human neutrophils were isolated from the peripheral blood of healthy volunteers by the rapid single-step method (20). The neutrophil preparation was Ͼ98% pure and viable, and were left for 30 min at 37°C before they were stimulated. Monocytes were prepared from the mononuclear cell fraction by adherence to plasma-coated dishes (21).
Chemotaxis-Neutrophil chemotaxis was investigated by a chamber method and/or the under-agarose method as described previously (4). To determine the dose response relationship between chemotaxis and fMLP, we used the chamber method because the specific concentrations of a ligand cannot be determined in the under-agarose method because of diffusion of the ligand. For the chamber assay, neutrophils (10 6 in 500 l RPMI 1640 that had been supplemented with bovine serum albumin, 2%, w/v) were placed in filter chambers (Millicell culture plate inserts with 3-m pore size polycarbonate filters). The chambers were then placed into wells (24-well plate) that contained RPMI 1640-bovine serum albumin (0.5 ml) and either fMLP or Me 2 SO. After 45 min at 37°C in a humidified atmosphere of 5% CO 2 in air, the inserts were removed and neutrophils that had migrated across the filter and collected on the surface of the wells were harvested by gentle pipetting and counted. Previous studies on the kinetics of neutrophil migration in this system have demonstrated that the majority of neutrophils had migrated across the filter within this period (4).
For the under-agarose method, 6 ml of agarose (1% v/v) was allowed to set in culture dishes (60-mm diameter) and wells (sets of 3) were made. To assess the chemotactic response, fMLP (5 ϫ 10 Ϫ8 M), neutrophils (2 ϫ 10 5 ), and Me 2 SO (0.1% v/v), all in 5-l aliquots, were added to the outer, center and inner wells, respectively, and the dishes were placed in a humidified atmosphere of 5% CO 2 in air at 37°C for 90 min. In this assay, neutrophils continuously migrate out of the wells in every direction, albeit with the majority of the migrating cells exhibiting directional migration toward the chemoattractant-containing well (23). The leading neutrophils form a distinct migrating front. To quantify migration, we measured the distance between the edge of the central well and the migrating front moving toward either the fMLP (chemotaxis) or Me 2 SO (random migration) containing well using an inverted Leitz microscope fitted with a grid eyepiece graticule. The plates were sealed and stored at 4°C until photomicrographed.
To assess the effect of PD184352, neutrophils (10 6 ) were preincubated with the inhibitor or ME 2 SO for 1 h before being used in the above chemotaxis assays. The cells with the inhibitor or vehicle were placed in the chambers or wells as described above. Cell viability at the end of the pre-incubation period, as judged by the trypan blue exclusion test, was not affected by the inhibitor, with Ͼ99% of the cells being able to exclude the dye.
Preparation of Cell Lysates-The cells were lysed in buffer A (20 mM Hepes, pH 7.4, 0.5% Nonidet P-40 (v/v), 100 mM NaCl, 1 mM EDTA, 2 mM Na 3 VO 4 , 2 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride, and 10 g/ml each of leupeptin, aprotinin, pepstatin A, and benzamidine) for 2 h (4°C) with constant mixing (24). Cell debris was sedimented (12,000 ϫ g 30 s), and the protein content of the soluble fractions was determined by the Lowry's protein estimation method. Samples were stored at Ϫ20°C for up to 2 weeks with no apparent loss of kinase activity being detected within this period.
RNA Isolation and RT-PCR-Total RNA was isolated from normal human monocytes and neutrophils using TRIzol® reagent as per the manufacturer's instructions. cDNA was synthesized from total RNA with AMV reverse transcriptase using an oligo d(T) primer. PCR primers, analogous to those used by Yan et al. (12) to amplify murine ERK5, were designed to the human ERK5 sequence. These were: sense primer ACGAGTACGAGATCATCGAGACC, and antisense primer GGTCAC-CACATCGAAAGCATTAGG. These are designed to amplify a 125-bp fragment (nucleotides 238 -362) from the human ERK5 mRNA sequence (GenBank TM accession number HSU25278). The 5Ј primer is located 52 bp 5Ј of an exon-intron junction (GenBank TM accession number AC124066, base 17697), and the 3Ј-primer is located 25 bp 3Ј of the same intron (intron-exon junction at bp 17009 of GenBank accession number AC124066). These primers therefore also define a PCR product of 774 bp (GenBank TM accession number AC124066, base 17723-16950) that can be amplified from human genomic DNA. This intron corresponds to the murine sequence reported as being alternatively spliced by Yan et al. (12). PCR amplification was performed in 50-l reactions using 2 l of cDNA as template, 100 M of each dNTP, 1 g of each primer, and 3.5 units of Expand High Fidelity polymerase in a final 1ϫ concentration of the recommended buffer plus 1ϫ Q Buffer (Qiagen Pty. Ltd., Clifton Hill, Australia). Thirty cycles of amplification was performed with each cycle consisting of a denaturation of 94°C for 30 s, annealing at 60°C for 30 s and an extension of 68°C for 30 s. PCR products were resolved on a 2.5% agarose gel with pUC19/HpaII and SppI/EcoRI DNA markers (500 ng).
Western Blot-Proteins in the lysates were separated by 8 or 12% SDS-PAGE as appropriate and transferred to nitrocellulose membrane. Even transfer of proteins between the lanes was confirmed by staining with Ponceau S. Membranes were probed with anti-ERK2, ERK5, MEK5, or p38 antibody using standard procedures, and the immune complexes were detected by enhanced chemiluminescence (25). In some cases, the blots were stripped using a Western blot recycling kit and reprobed with another antibody.
Immunoprecipitation of ERK1, ERK2, ERK5, MEK5, and p38 -Lysates, normalized for protein content (1 mg), were precleared by incubation with protein A-Sepharose (15 l of a 1:1 slurry) for 20 min at 4°C with constant mixing. The supernatants were removed and incubated with an anti-ERK5, ERK2, MEK5, or p38 antibody (3 g each) for 2 h at 4°C with constant mixing. Protein A-Sepharose was added and after a further 30 min incubation at 4°C with constant mixing, the immunocomplexes were sedimented and were washed once with buffer A and once with assay buffer (20 mM Hepes, pH 7.2, 20 mM ␤-glycerophosphate, 3.8 mM p-nitrophenyl phosphate, 10 mM MgCl 2 , 1 mM dithiothreitol, 50 M Na 3 VO 4 , and 20 M ATP) at 4°C. The samples were then used in kinase activity assays, kinase quantitation, or Western blot analysis.
Kinase Activity Assay-The assay was started by adding 30 l of assay buffer (30°C) containing 10 Ci of [␥-32 P]ATP, 3.8 mM p-nitrophenyl phosphate, and 15 g of myelin basic protein. After 20 min, the assay was terminated by the addition of Laemmli buffer and boiling the samples for 5 min at 100°C. Phosphorylated myelin basic protein was resolved by 16% SDS-polyacrylamide gel electrophoresis and was detected and quantitated using an Instant Imager (Packard Instruments).
Statistical Analysis-Statistical analyses were performed using Student's unpaired t test or analysis of variance (ANOVA) followed by the Tukey-Kramer multiple comparisons test as appropriate.

H 2 O 2 Stimulated the Activity of ERK5-Since H 2 O 2 has
widely been reported to stimulate the activity of ERK5 (8,16,17), we first investigated whether H 2 O 2 was able to stimulate the ERK5 module in neutrophils. Neutrophils were incubated with H 2 O 2 , lysed, and lysates were subjected to immunoprecipitation with an anti-ERK5 antibody that was raised against residues 790 -803 in the C terminus of human ERK5 (Sigma-Aldrich Corp.) and had previously been used for this purpose (15,26,27). ERK5 kinase activity was assayed using myelin basic protein as a substrate. Consistent with the data in other cell-types, H 2 O 2 stimulated the kinase activity in the ERK5 immunoprecipitates ( Fig. 1). In contrast, H 2 O 2 did not stimulate the dual phosphorylation of ERK1/ERK2, suggesting that the kinase activity in the ERK5 immunoprecipitates was unlikely to have been caused by contaminating ERK1 or ERK2.
Expression of ERK5 and MEK5 in Neutrophils-Previously, we reported that ERK5 in COS-1 monkey kidney cells migrated in SDS gels with an M r of greater than the top M r marker of 113,000 (17). This is consistent with other reports that ERK5 migrates at 100 or higher (16,28,29). To characterize the expression of ERK5 in neutrophils, we first subjected neutrophil lysates to Western blotting. The anti-ERK5 antibody detected a number of bands that migrated in SDS gels with M r of ϳ97,000, ϳ 90,000 ϳ82,000 and ϳ45,000. On prolonged exposure of the film, a faint ϳ120-kDa band could also be seen in the neutrophils ( Fig. 2a and also Fig. 3b). In contrast, when the lysates were immunoprecipitated with the anti-ERK5 antibody prior to Western blotting, the blots showed predominantly 2 bands of M r of ϳ97,000 and ϳ120,000 (Fig. 2b). No bands could be seen if the immunoprecipitation was carried out with an anti-␤-actin ( Fig. 2c) or ERK2 (Fig. 2d) antibody. This implies that the ϳ97and ϳ120-kDa bands were specifically brought down by the anti-ERK5 antibody.
When lysates from human peripheral blood monocytes, human embryonic kidney HEK293T cells, and HL60 cells (not shown) were subjected to Western blotting with the anti-ERK5 antibody, the blots showed the presence of predominantly the ϳ120-kDa band (Fig. 2e), although some monocyte samples also contained faster migrating bands (see Fig. 3b). These data imply that the ϳ120-kDa band is ERK5. The presence of faster migrating anti-ERK5 immunoreactive bands in cell lysates is not unprecedented as others have previously reported that ERK5 blots sometimes contained multiple immunoreactive bands that had been ascribed to ERK5 degradation products (30). For comparison, lysates from HEK293T cells were also probed with an anti-ERK5 antibody from Santa Cruz Biotechnology. According to the manufacturer's specification sheet, this antibody detects a ϳ120-kDa ERK5 band. Akin to the antibody from Sigma-Aldrich, the Santa Cruz antibody also detected a ϳ120-kDa band in HEK293T cells (Fig. 2e, right lane). Taken together, our data imply that the ϳ120-kDa band is ERK5. The faster migrating species are likely to be degradation products (30).
The expression of ERK5 in neutrophils was also investigated at the mRNA level. Human placenta has previously been reported to contain three ERK5 transcripts (11). However, since these are generated by alternative splicing in the 5Ј-non-coding region, only one form of ERK5 is expressed at the protein level (11). In contrast, murine tissues have been reported to contain three isoforms of ERK5 protein, designated mERK5a, mERK5b and mERK5c, which are derived from differential splicing of the N terminus of mERK5 (12). The data in Fig. 3a demonstrate the presence of only one species of ERK5 mRNA in both peripheral blood monocytes and neutrophils (lanes 2 and 3, lower arrow). The presence of a PCR product corresponding to genomic DNA in the neutrophil cDNA sample (lane 3, upper arrow) demonstrates that the PCR would be capable of detecting larger, alternatively spliced, mRNA species analogous to those found in mice, if they were present. In the absence of reverse transcriptase RNA templates (lanes 4 and 5), only the product corresponding to genomic DNA was seen (lane 7) indicating some contamination of the RNA with DNA. In the absence of template no product was detected (lane 6). Although it is not possible to exclude the possibility that contaminating mononuclear leukocytes could have contributed to the PCR product in the neutrophil preparations, template dilution experiments with mononuclear leukocyte samples to 1/50 of the amount of RNA present in neutrophil samples, corresponding to the maximum level of mononuclear leukocyte contamination, suggest that this contamination would only have contributed to a proportion of the neutrophil PCR signal (not shown). More importantly and consistent with the data in Fig. 3a, when lysates from mononuclear leukocyte were Western-blotted at a protein load of 2 g (Fig. 3b, lane 3) as opposed to 100 g of neutrophil lysate protein (Fig. 3b, lane 1), no immunoreactive bands could be detected (see also kinase activity data in Fig.  5c). In contrast, the same mononuclear leukocyte fraction loaded at 100 g displayed a strong band at ϳ120-kDa band and a faster migrating band (lane 2). A very faint ϳ120-kDa band in addition to the ϳ97-kDa band could be detected in the neutrophil lysates. The above data imply that faint ϳ120-kDa and ϳ97-kDa bands seen in the neutrophils could not have come from contaminating mononuclear leukocytes.
We next investigated the expression of MEK5 in neutrophils. MEK5␣ and MEK5␤ migrate in SDS gels with M r of ϳ50,000 and 40,000 respectively (13). To confirm that neutrophils express MEK5, we probed neutrophil lysates with an antibody that was raised against residues 59 -74 of MEK5␣. MEK5␤ lacks this sequence (13). Fig. 4a shows that neutrophil lysates contained an immunoreactive protein that migrated on SDS gels with an M r of ϳ50,000. When the lysates were subjected to immunoprecipitation and Western blotting with the anti-MEK5␣ antibody, a band of the expected M r was also seen (Fig.  4b). Although interaction between MEK5 and ERK5 has been demonstrated in cells overexpressing epitope-tagged MEK5 and ERK5, such interaction has not been reported for the endogenous kinases. To investigate this, the above blot was stripped and reprobed with the anti-ERK5 antibody. MEK5 immunoprecipitates contained little or no ERK5 (Fig. 4c).
FIG. 2. Expression of ERK5 protein in neutrophils, monocytes, and HEK293T cells. Cells (3 ϫ 10 7 for neutrophils and monocytes, 1 ϫ 10 7 for kidney cells) were lysed, and aliquots containing equal amounts of proteins were subjected to either Western blotting (WB) or immunoprecipitation (IP) followed by Western blotting using the indicated antibodies. The anti-ERK5 antibody was from either Sigma Aldrich (a-e) or Santa Cruz (e). a, WB of neutrophil lysates after an overnight exposure; b, duplicate neutrophil lysates (1 mg of lysate protein each) immunoprecipitated and Western-blotted with the anti-ERK5 antibody; c, neutrophil lysates immunoprecipitated with a monoclonal anti-␤-actin antibody and Western-blotted with the anti-ERK5 antibody; d, neutrophil lysates immunoprecipitated with a goat anti-ERK2 antibody and Western-blotted with the anti-ERK5 antibody; e, monocyte and HEK293T cell lysates Western-blotted with the Sigma or Santa Cruz (HEK, right lane) anti-ERK5 antibody. ERK5 migrated in SDS gels with a M r of ϳ120,000. ERK5 immunoprecipitates from neutrophil samples contained predominantly the ϳ120and ϳ97-kDa bands. When neutrophil lysates were subjected to immunoprecipitation with the anti-␤-actin (c) or anti-ERK2 (d) antibody, no anti-ERK5 antibody reactive bands were detected. The blots (c and d) were stripped and reprobed with the anti-␤-actin (c) or anti-ERK2 (d) antibody to show that each antibody recognized its cognate antigen (right panels). MEK5 immunoprecipitates also did not contain ERK2 (Fig. 4d) or p38 (data not shown). In contrast, when the ERK5 immunoprecipitates were subjected to Western blot analysis using the anti-MEK5 antibody, a band that migrated with the expected M r of MEK5␣ was detected (Fig. 4e). As expected, the ERK5 immunoprecipitates did not contain ERK1 or ERK2 (Fig. 4f) or p38 (data not shown). MEK5␣ was not pulled down if the precipitation was conducted in the absence of the anti-ERK5 antibody (Fig. 4e). These data imply that MEK5 co-precipitated with ERK5 and that the MEK5-ERK5 interaction did not survive the immunoprecipitation step with the anti-MEK5 antibody. The data also validate the specificity of the antibodies and demonstrate the lack of cross-reaction with ERK1/ERK2, ERK5, and p38.
Stimulation of the MEK5-ERK5 Module by fMLP-Previous studies have demonstrated that the MEK1/MEK2 inhibitor, PD98059, inhibited fMLP-stimulated neutrophil chemotaxis (3-7). As discussed above, this and the structurally unrelated MEK inhibitor, U0126, have now been found to also inhibit MEK5. Thus, it is imperative that we investigate whether some of the effects of the MEK inhibitors on neutrophil responses could be due to inhibition of the MEK5-ERK5 module (8,15). We first compared the activation of the MEK-ERK5 and ERK1/ ERK2 modules by fMLP over a concentration range of 10 Ϫ10 to 10 Ϫ6 M. The data in Fig. 5 demonstrate that incubation of neutrophils (3 ϫ 10 7 /sample) with fMLP resulted in enhanced MEK5 (a) and ERK5 (b) activities that reached their peaks at around 10 Ϫ9 M. The threshold was at around 10 Ϫ10 M. Kinase activities showed a downward trend as the concentration of fMLP was increased further. Any contribution from contaminating mononuclear leukocytes can be excluded on the basis that fMLP (10 Ϫ9 M) did not significantly alter ERK5 activity in samples prepared from an equivalent number (3 ϫ 10 7 ) of mononuclear leukocytes (Fig. 5c). This implies that the kinase activities seen in the immunoprecipitates (Fig. 5, a and b) could only have originated from neutrophils. fMLP also caused a retardation in the electrophoretic mobility of the ϳ120-kDa ERK5 band in neutrophils (Fig. 5c). Because the MEK5 immunoprecipitates contained little or no ERK5 and no ERK1/ERK2 or p38, the data in Fig 5a represent MEK5 activity even though MEK5 could contribute to the kinase activity in the ERK5 immunoprecipitates (but see also Fig. 6).
In contrast, fMLP-stimulated activation of ERK1 and ERK2 as assessed by Western blot followed a different profile (Fig.  5d). Whereas phosphorylation of ERK1/ERK2 was just detectable at 10 Ϫ9 M, the degree of phosphorylation increased dosedependently to 10 Ϫ7 M, and this was maintained at 10 Ϫ6 M fMLP. Since the phospho-ERK data only suggested activation, we also investigated ERK1 and ERK2 activation by immunoprecipitation followed by enzymatic assay. Surprisingly, ERK1/ FIG. 5. Dose-dependent stimulation of MEK5, ERK5, ERK1/ERK2, and p38 activities, and chemotaxis by fMLP. Cells were incubated with fMLP for 3 (a-c), 1 (d and e), 5 (f), or 40 (g) minutes. Kinase activation (a-f) and chemotaxis (g) were then determined. MEK5 (a), ERK5 (b and c) and p38 (f) activities were assayed after immunoprecipitation as described under "Experimental Procedures." Western blots of replicate samples of the immunoprecipitates demonstrate that similar amounts of the ϳ120-kDa band were present in each of the immunoprecipitates. A representative anti-ERK5 blot of immunoprecipitated ERK5 from unstimulated and fMLP-stimulated neutrophils is shown in c. fMLP caused a slight retardation in the electrophoretic mobility of the ϳ120-kDa band (c). To assay for ERK1 and ERK2 activation, the lysates were split so that one aliquot was subjected to Western blotting using a phosphospecific antibody (d, upper panel shows phosphorylated ERK1/ERK2, and lower panel shows protein loading) whereas the other aliquot was subjected to immunoprecipitation and followed by enzymatic assay (e). Chemotaxis (g) was investigated using 3-m polycarbonate filter chambers and the number of neutrophils that had migrated across the membrane over a 40-min incubation period and collected at the bottom of the wells were harvested and counted. In c, activation of ERK5 by fMLP (10 Ϫ9 M) was compared between equal numbers (3 ϫ 10 7 ) of neutrophils and mononuclear leukocytes. Kinase activity was stimulated in neutrophils but not in mononuclear leukocytes, exclude the possibility that the ERK5 activity seen in the neutrophil samples was due to contaminating mononuclear leukocytes. Control cells received vehicle at an amount that was equivalent to that present at 10 Ϫ6 M fMLP. Results shown are means Ϯ S.E. of 3-9 experiments. Significance of difference between control and fMLP-stimulated cells (a, b, e-g): *, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.005 (Student's unpaired t test or Tukey-Kramer multiple comparisons test as appropriate).
ERK2 kinase activity (Fig. 5e) showed a totally different dose response profile from that revealed by the Western blots. Thus, an increase in the activity of ERK1/ERK2 was detectable at 10 Ϫ10 M, and kinase activity reached a plateau at 10 Ϫ9 M fMLP. Activation was maintained at this level as the concentration was increased to 10 Ϫ7 M but declined as the concentration of fMLP was increased to 10 Ϫ6 M. In these experiments, each lysate was split into two aliquots so that the two different types of assays could be conducted on the same lysates. The possibility that ERK5 or MEK5 could have contributed to the kinase activity in the ERK2 immunoprecipitates was excluded on the basis that neither ERK5 nor MEK5 could be detected in the ERK2 immunoprecipitates (data not shown). Furthermore, immunoprecipitation of ERK1/ERK2 under more stringent conditions in the presence of SDS (0.1%) and sodium deoxycholate (1%) in addition to Nonidet P-40 did not alter the dose response profile of the kinase activity assays (data not shown). Given that the Western blot analysis does not measure enzymatic activity, the data from the immunoprecipitated ERK1/ERK2 are likely to reflect more accurately the status of cellular ERK1/ERK2 activity.
Another MAP kinase that regulates neutrophil responses such as chemotaxis is p38 (31,32). Since p38 is activated by fMLP, we also compared the dose-response characteristics of fMLP-stimulated p38 activation with that of the MEK5-ERK5 module. The data in Fig. 5f demonstrate that while 10 Ϫ9 M fMLP was ineffective at stimulating p38 activity, higher concentrations caused a 2 fold increase in kinase activity which peaked at 10 Ϫ7 M and decreased at 10 Ϫ6 M.
Because the above dose response curves all showed a bellshaped profile that had been reported for fMLP-stimulated chemotaxis (31), it is important to also establish a dose response profile of fMLP-stimulated chemotaxis under our conditions to permit a direct comparison between the profiles using the same batch of fMLP. The data in Fig. 5g, obtained using the chamber assay, demonstrate that the profile for fMLP-stimulated chemotaxis most closely resembled that for p38, followed by ERK1/ ERK2 and least by the MEK5-ERK5 module.
The kinetics of fMLP-stimulated MEK5 and ERK5 activation was investigated next. At 2 min, MEK5 activity (Fig. 6a) increased 2.2-fold whereas ERK5 activity (Fig. 6b) increased 1.9-fold. The increase in ERK5 activity was transient, reaching a peak at around 5 min after stimulation, whereas the increase in MEK5 activity reached a maximum between 2 and 3 min and persisted for the duration of the experiment. This difference suggests that MEK5 might probably only have contributed to a minor degree to the overall kinase activity seen in the ERK5 immunoprecipitates and that MEK5 and ERK5 have different inactivation mechanisms. For example, phosphorylation of MEK5 by ERK5 following the prior activation of ERK5 by MEK5 (30,33) could contribute to the difference in kinetics. In contrast, fMLP-stimulated ERK2 activation reached a peak at around 1 min and declined thereafter (Fig. 6c).
The above data demonstrate that fMLP stimulated the activity of the MEK5-ERK5 module. Since the interaction between MEK5 and ERK5 (Fig. 4, c-f) were investigated in resting neutrophils, we also investigated whether activation affected the degree of interaction between the kinases. Neutrophils were incubated with fMLP (10 Ϫ9 M) for 3 min, and the lysates were subjected to immunoprecipitation and Western blot analyses as described in Fig. 4. The data (not shown) demonstrate that fMLP did not affect the degree or pattern of interaction between MEK5 and ERK5 or cause ERK1 and ERK2 to interact with ERK5 or MEK5.
Mechanism of fMLP-stimulated Activation of the MEK5-ERK5 Module: Role of Phosphatidylinositol 3-Kinase-Although H 2 O 2 stimulated the activity of ERK5 in neutrophils, fMLP-stimulated activation of ERK5 did not require H 2 O 2 because fMLP generally has little or no effect on the NADPH oxidase at 10 Ϫ10 and 10 Ϫ9 M. We previously reported that p21 ras regulate the activation of ERK5 (17) and recently, p21 ras was reported to also regulate the activity of PI 3-kinase (34). Since fMLP is known to stimulate p21 ras (35) and PI 3-kinase (36), and fMLP-stimulated p38 activation is dependent on PI 3-kinase (37), it is possible that PI 3-kinase may lie upstream of the MEK5-ERK5 module. Although it is customary to investigate the effect of PI 3-kinase by using wortmannin and LY294002, two PI 3-kinase inhibitors that act via different mechanisms (38,39), we focused primarily on wortmannin. This was because LY294002 per se had been reported to stimulate the activity of p38 in human neutrophils without affecting fMLP-stimulated p38 activity, thereby making data interpretation difficult (37). In contrast, wortmannin, while having no effect on the activity of p38 in the absence of fMLP, partially blocked fMLP-stimulated p38 activity (37).
The data in Fig. 7 demonstrate that the ability of fMLP to stimulate MEK5 (a) and ERK5 (b) activities was dose-depend-FIG. 6. Comparison of the kinetics of fMLP-stimulated MEK5, ERK5, and ERK1/ERK2 activation. Neutrophils were incubated with fMLP (10 Ϫ9 and 10 Ϫ7 M for MEK5/ERK5 and ERK1/ERK2 activities, respectively) or Me 2 SO (0.1% v/v) for the indicated times, and kinase activities were assayed. Similar amounts of kinase were present in each of the immunoprecipitates. Results shown are means ϩ S.E. of three experiments each. Significance of difference between control and fMLP-stimulated cells: *, p Ͻ 0.05; **, p Ͻ 0.01 (Tukey-Kramer multiple comparisons test). ently blocked by wortmannin at concentrations that inhibit PI 3-kinase activity. In preliminary studies, we have also found that whereas LY294002 (20 M) did not affect the ability of fMLP to stimulate the activity of ERK5, this compound caused a 2-fold increase in ERK5 activity in the absence of fMLP (data not shown). It is possible that the unusual behavior of LY2094002 is directed preferentially toward stress-activated MAP kinases such as p38 and ERK5 in neutrophils, independently of PI 3-kinase since the inhibitor did not stimulate the activity of ERK1/ERK2 but inhibited fMLP-stimulated activation of ERK1/ERK2 in these cells (37). It is clear that LY294002 should be used with caution when assessing the roles of PI 3-kinase in neutrophils (37).
Concentration-dependent Inhibitory Effects of PD184352 on fMLP-stimulated ERK5 and ERK1/ERK2 Activity-Whereas PD98059 inhibits MEK1/MEK2 and MEK5 at similar concentrations (40), studies with a recently described specific MEK inhibitor, PD184352, have demonstrated that this agent is able to distinguish between MEK5 and MEK1/MEK2 (40,41). Thus, whereas PD184352 totally inhibited the activation of ERK1 and ERK2 at around 0.5-1 M (40, 41), near total inhibition of ERK5 was observed at 10 M, although a small inhibitory effect was seen at 3 M (40). With respect to 23 other kinases tested, PD814352 and PD98059 shared similar specificities in that the PD compounds did not affect the activities of these kinases (9). We therefore investigated whether low concentrations of PD184352 were able to distinguish between the activation of ERK1/ERK2 and activation of ERK5 by fMLP in neutrophils. Consistent with previous data, PD184352 caused a marked inhibition (greater than 80%) of fMLP-stimulated activation of ERK1 and ERK2 at 0.5 M (Fig. 8, upper panel). At this concentration, PD184352 did not inhibit ERK5 activation (lower panel). The effect on ERK1/ERK2 activity was greater at 2 M and at this concentration, PD184352 caused a partial inhibition of fMLP-stimulated ERK5 activation. These data confirm that PD184352 can distinguish between the MEK5-ERK5 and MEK1/MEK2-ERK1/ERK2 modules if used at 0.5 M.
Inhibition of fMLP-stimulated Chemotaxis by PD184352-The above data imply that PD184352 could be used as an agent to define the roles ERK1 and ERK2 play in fMLP-stimulated chemotaxis. We first investigated the effect of PD184352 on chemotaxis using the chamber method. The data in Fig. 9a demonstrate that PD184352 inhibited chemotaxis in a dose-dependent manner. At 0.5 M, the chemotactic response was inhibited by ϳ25%. This result implies that the ability of fMLP to stimulate neutrophil migration was partially inhibited as a consequence of ERK1 and ERK2 inhibition and hence strengthens the conclusion that ERK1 and ERK2 are involved, at least in part, in regulating the chemotactic response. At 2 M, chemotaxis was inhibited by ϳ40% and at 10 M, the inhibitor caused a near total inhibition of neutrophil migration (Fig. 9a). However, data interpretation becomes difficult because 2 M PD184352 also caused a partial inhibition of ERK5 activity, in conjunction with a near total inhibition of the activity of ERK1/ ERK2 (Fig. 8). At 10 M PD184352, it was likely that the activities of ERK1/ERK2 and ERK5 were totally inhibited (40).
While the above data demonstrate that PD184352 reduced the number of neutrophils that were capable of migrating across the filter chamber, the data do not provide any information on the distance traversed by the neutrophils within the migration zone, defined as the area over which the neutrophils migrate, or the density of neutrophils within this zone. Thus, we also investigated the effect of PD184352 on chemotaxis using the underagarose method. This approach not only permitted the visualization of migrating neutrophils but also allowed another perspective to be obtained through the measurement of how far the neutrophils have migrated. This distance was measured from the edge of the central well containing the neutrophils to a migrating front formed by the leading neutrophils.
At 0.5 M, PD184352 reduced fMLP-stimulated chemotaxis by ϳ25% (Fig. 9b), without significantly affecting random migration (Fig. 9c). A greater inhibitory effect was evident when the density of neutrophils within the migratory zone was analyzed. To enumerate the number of migrating neutrophils, a square zone covering an area of 0.6 mm ϫ 0.6 mm, abutting the edge of the central well and sitting centrally over a perpendicular line between the centers of the neutrophil-and fMLPcontaining wells, was selected. This was representative of the cell density covering the migration zone up to the migrating front. Fig. 9d shows a representative photomicrograph of the central portion of this area. PD184352 reduced the number of neutrophils within this zone by ϳ50% (Fig. 9d). Because the viability of neutrophils was not affected by the inhibitor at the concentrations tested (maximum of 10 M), the above data imply that the reduction in chemotactic response was not due to cell death but to fewer neutrophils being able to respond to fMLP following inhibitor treatment. This is consistent with the reduction in the number of neutrophils that migrated across the filter in the chamber assay. Interestingly, the inhibition of chemotaxis by 10 M PD184352 was greater in the chamber method than in the under agarose method. This probably reflected a reduction in the ambient inhibitor concentration as the neutrophils migrated away from the central well in the under agarose method, resulting in the attenuation of inhibitor effect. PD184352 also inhibited fMLP-stimulated chemokinesis (data not shown). DISCUSSION Our data demonstrate that not only were MEK5 and ERK5 expressed in human neutrophils but that the activities of these kinases were also stimulated by fMLP. Western blot analyses using anti-ERK5 antibodies from two different sources demonstrate that ERK5 migrated in SDS gels with an M r of ϳ120,000. This is consistent with data from other laboratories where the M r markers were shown on their blots. In these studies, ERK5 migrates with an M r of 110,000 and above (16,17,28,29). Furthermore, the data sheet from Santa Cruz Biotechnology states that the expected M r of ERK5 in SDS gels is 120,000. Although this ϳ120-kDa band was detectable in neutrophil samples, especially in the anti-ERK5 immunoprecipitates, the ratio of this band to total protein content of the lysates was less than that seen in other cell types. This may be the consequence of ERK5 degradation in neutrophil samples. Thus, the neutrophil lysates contained a number of faster migrating anti-ERK5 antibody reactive species that were not seen or were seen infrequently in human peripheral blood monocytes, HL60, and HEK293T cells. Increasing the level of protease inhibitors in the lysis buffer did not abolish the faster migrating species in neutrophil samples, suggesting that degradation could have occurred prior to lysis. Interestingly, a previous study on the expression of MEK kinase (MEKK)1 in human neutrophils and macrophages has also observed multiple anti-MEKK1 antibody reactive species and it was concluded that a proteolytic product might represent the active form of MEKK1 (42). The presence of faster migrating bands in ERK5 immunoblots is not unprecedented because Cobb and co-workers (30) have also found faster migrating species in their ERK5 blots and these are attributed to ERK5 degradation products. The presence of ERK5 in neutrophils was also confirmed by PCR analysis that showed the presence of one PCR product. Our data and those of others (30) also raise the possibility that the physically larger ERK5 is more labile than the other MAP kinases because the anti-ERK1/ERK2, p38, and JNK blots did not have faster migrating species (22).
An unexpected finding in our investigation is the discrepant dose response profiles of ERK1/ERK2 activation when two different assays of activation were compared. This difference was seen despite the fact that the same samples were used in the two types of assays. In preliminary investigations, we have also observed a similar discrepancy when we compared the dose response profiles of phorbol 12-myristate 13-acetate stimulated activation of ERK1 and ERK2 using the two types of assays. Although the basis for the discrepant results is not known, the kinase activity data is likely to more closely reflect the status of kinase activation in intact cells than the data obtained using the anti-phosphospecific antibody. Our finding therefore serves as a caution to others to confirm that results obtained from Western blots correlate with those from kinase activity assays.
The fMLP dose response profiles of MAP kinase activation show some degrees of similarity as well as difference between the different MAP kinases. Thus, kinase activity increased with in- creases in the fMLP concentration and declined when higher concentrations were attained, especially at 10 Ϫ6 M. They differed in that the MEK5 and ERK5 profiles were left-shifted, the ERK1/ ERK2 profile was more normally distributed whereas the p38 profile was right-shifted. Human neutrophils express two classes of fMLP receptors, FPR and FPRL1 that represent the high and low affinity fMLP receptors, respectively (43). In neutrophils, FPR has a K d for fMLP of around 10 Ϫ9 M whereas the K d of FPRL1 is several hundred-fold higher (43). It remains to be determined whether the dose response profiles reflect differential use of the two classes of fMLP receptors.
Previous studies have found that fMLP-stimulated p38 and ERK1/ERK2 activation in neutrophils required PI 3-kinase (37,42). The data in the present study demonstrate that this dependence on PI 3-kinase can be extended to fMLP-stimulated activation of the MEK5-ERK5 module. Thus, the ability of fMLP to stimulate MEK5 and ERK5 activities were dose-dependently inhibited by wortmannin. A relationship between PI 3-kinase and ERK5 has not been established previously.
Evidence from human neutrophils has demonstrated that fMLP-stimulated chemotaxis is dependent on PI 3-kinase (44,45), 2 possibly via its regulation of rac2, CDC42, and PAK1 (45)(46)(47), and deletion of PI 3-kinase ␥ in mice results in a severe defect in fMLP-stimulated chemotaxis in the neutrophils (48). Since p38, ERK1, ERK2, and ERK5 are downstream of PI 3-kinase (37, 42, and present data), it is possible that these MAP kinases collectively serve as effectors of PI 3-kinase in the chemotactic response. Although there is some controversy regarding the effect of PD98059 on neutrophil chemotaxis (3-7, 32, 49), the recent demonstration that PD98059 inhibited fMLP-and IL8-stimulated neutrophil adherence (50) lends further support for the ERKs being involved in regulating neutrophil migration. At 50 M, PD98059 totally inhibited fMLP-stimulated activation of ERK1/ERK2 and partially inhibited (50%) fMLP-stimulated chemotaxis (4). At this concentration, PD98059 also caused a partial inhibition of fMLPstimulated ERK5 activation. 2 Thus, it is not possible to determine whether ERK1 and ERK2 or ERK5 is involved in regulating fMLP-stimulated chemotaxis. This problem is circumvented by the use of PD184352 at concentrations that inhibited ERK1 and ERK2 but not ERK5 activity. Our data with 0.5 M PD184352 therefore provide the strongest piece of evidence to date that ERK1 and ERK2 are involved in regulating the chemotactic response. At higher concentrations, PD184352 also inhibited the MEK5-ERK5 module, and this was associated with a greater degree of inhibition of the chemotactic response. Although it is tempting to speculate from this data that the MEK5-ERK5 module could also be involved in regulating the chemotactic response, our current data suggest that this module is unlikely to play a significant role in regulating this response. Thus, the fMLP dose-response profile of MEK5 and ERK5 activation, being left-shifted is poorly correlated with the profile of fMLP-stimulated chemotaxis. Since previous data have also shown PD98059 to inhibit neutrophil superoxide production stimulated by fMLP and other stimuli (4,51), it would be important to investigate whether PD184352 (0.5 M) also inhibits this response. However, since the respiratory burst activity in response to fMLP is usually over by 3 min but maximal stimulation of ERK5 occurred at 5 min, this difference suggests that ERK5 does not play a role in regulating the respiratory burst activity. Thus, the role of the MEK5-ERK5 module remains to be defined.
The demonstration that ERK1 and ERK2 are involved in regulating the chemotactic response also provides strong sup-port to our contention that the status of ERK1 and ERK2 activity in intact cells is better represented by the enzymatic data than the anti-phospho-ERK blots. Consistent with this suggestion, the dose-response profile of ERK1/ERK2 activity more closely reflected that of fMLP-stimulated chemotaxis. Both reached their high points at 10 Ϫ8 and 10 Ϫ7 M and declined at 10 Ϫ6 M fMLP whereas the level of phosphorylated ERK1 and ERK2 showed very poor or no correlation with the chemotactic response.
Based on our results, we propose that there is a hierarchy in the utilization of MAP kinase modules by chemoattractants such as fMLP which also stimulates neutrophil responses pertinent to microbial killing. At low chemoattractant concentrations, MAP kinase activation is predominantly restricted to the MEK5-ERK5 and ERK1/ERK2 modules. As the concentration of the chemoattractant increases, there is a gradual shift from MEK5-ERK5 activation to p38 activation. Thus, the activity of the MEK5-ERK5 module declines as the activity of the p38 module is up-regulated by the chemoattractant. Further increases in the chemoattractant concentration eventually cause the activities of ERK1, ERK2 and p38 to also decline. The MAP kinase modules therefore provide the neutrophils with one family of signaling pathways that cover a range of chemoattractant concentrations.