Four p38 mitogen-activated protein kinases (p38α, β, γ, δ) have been described. To understand the role of p38 family members in inflammation, we determined their relative expression in cells that participate in the inflammatory process. Expression was measured at the level of mRNA by reverse-transcriptase PCR and protein by Western blot analysis. p38α was the dominant form of p38 in monocytes; expression of p38δ was low and p38β was undetected. In macrophages, p38α and p38δ were abundant, but p38β was undetected. p38α and p38δ were also expressed by neutrophils, CD4+ T cells, and endothelial cells. Again, p38β was not detected in neutrophils, although low amounts were present in CD4+ T cells. In contrast, p38β was abundant in endothelial cells. p38γ protein was not detected in any cell type, although p38γ mRNA was present in endothelial cells. Immunokinase assays showed a strong activation of p38α and a lesser activation of p38δ in LPS-stimulated macrophages. Abs specific for mono- and dual-phophorylated forms of p38 suggested that LPS induces dual phosphorylation of p38α, but primarily mono-phosphorylation of p38δ. IL-1β activated p38α and p38β in endothelial cells. However, p38α was the more activated form based on kinase assays and phosphorylation analysis. Expression and activation patterns of p38α in macrophages and endothelial cells suggest that p38α plays a major role in the inflammatory response. Additional studies will be needed to define the contribution of p38δ to macrophage, neutrophil, and T cell functions, and of p38β to signaling in endothelial cells and T cells.
In recent years, an effort has been mounted to delineate the intracellular signaling cascades in cells that mediate inflammation. Much attention has been given to the mitogen-activated protein kinase (MAPK)2 superfamily due to their consistent activation by pro-inflammatory cytokines, and to their role in nuclear signaling. This superfamily includes, among others, the extracellular signal response kinases, c-jun N-terminal kinases, and the p38 family of kinases. Interest in the p38 family has been particularly intense following the discovery that p38 inhibitors are antiinflammatory in vivo (1, 2, 3).
Four genes encode the known members of the p38 family, p38α (1), p38β (4, 5), p38γ (6, 7), and p38δ (8, 9). In addition to high homology, the family is defined by a TGY motif within kinase subdomain seven. Dual phosphorylation of this domain in p38α, β, γ, and δ is catalyzed by the kinase, MKK6, and is required for p38 activation (10, 11, 12). With the exception of p38β, each p38 may also be activated by MKK3 (10, 11, 12). p38α, β, γ, and δ each phosphorylate activating transcription factor 2 (ATF2) (4, 5, 6, 7, 8, 9, 10, 11, 12). Even so, distinct functions for the kinases are suggested by the ability of p38α and p38β, but not p38γ and p38δ, to phosphorylate MAPK-activated protein kinase-2 (MAPKAPK2) (8, 10, 12). Conversely, p38γ and p38δ, but not p38α and p38β, favor stathmin as a substrate (13), and an unidentified 12-kDa protein in muscle is a selective substrate for p38γ (7).
The widely used pyridinyl imidazole inhibitors of p38, SB203580 and SB202190, are nearly equipotent against p38α and p38β, but do not inhibit p38γ or p38δ (4, 5, 8, 9, 12). These and similar compounds reduce the severity of murine collagen-induced arthritis and rat adjuvant arthritis, as well as hyperangiogenesis associated with murine air pouch granuloma (2, 3). The antiinflammatory effects can be attributed, in part, to the ability of the inhibitors to suppress monocyte/macrophage production of TNF-α, IL-1β, and other cytokines (14, 15, 17). Numerous other antiinflammatory activities have been elucidated, including suppression of IL-1-induced prostaglandin H synthase-2 expression by endothelial cells (18), FMLP-induced neutrophil chemotaxis (19), and IL-2- and IL-7-induced lymphocyte proliferation (20).
Because the available compounds inhibit both p38α and p38β, it has remained unclear which antiinflammatory activities can be attributed to p38α and which can be attributed to p38β. It is also possible that the function of the two kinases may be redundant in some cells. Furthermore, although the compounds do not block p38γ or p38δ, similarities in the activation of p38 family members suggest roles for p38γ and p38δ in inflammation. Specifically, MKK6 activates each p38, and when overexpressed by cell transfection, each p38 can be activated in cells treated with IL-1β and TNF-α (10, 11, 12). However, the contributions of p38γ and p38δ to inflammatory processes are largely unexplored.
The tissue distribution of mRNAs encoding each p38 family member has been evaluated by Northern and dot blot analysis (4, 5, 6, 7, 8, 9). Only p38γ exhibited a limited tissue distribution, with high expression of mRNA in skeletal muscle. p38α, p38β, and p38δ mRNAs are more widely distributed. However, analysis of tissue mRNA does not yield information on expression within specific cell lineages. Knowledge of kinase expression within a cell type is essential for understanding the function of a kinase, and the expression of the p38 family members within cell lineages has been largely unexplored. Furthermore, the activation of endogenous p38 family members by relevant environmental stimuli has not yet received much attention.
To investigate the role(s) of p38 family members in inflammation, we determined the expression profile of p38α, β, γ, and δ in several cell lineages that participate in the inflammatory response. Lineage- and differentiation-specific expression was observed. Furthermore, differential activation of p38 members was observed in endothelial cells and macrophages responding to inflammatory stimuli.
Materials and Methods
Isolation and culture of human cells
Neutrophils were isolated from blood of normal volunteers by density centrifugation using 1-Step Polymorphs gradient (Accurate Chemical, Westbury, NY). Monocytes were isolated by centrifugal elutriation, as described (17), and were ≥95% CD14 positive. To prepare CD4+ T cells, blood mononuclear cells were isolated by density-gradient centrifugation using Ficoll-Paque Plus (Pharmacia, Uppsala, Sweden). Mononuclear cells were resuspended to 5 × 107/ml in ice-cold PBS containing 0.5% BSA, 5 mM EDTA, and 50 μl/ml anti-CD4 Dynabeads (Miltenyi Biotec, Auburn, CA). After 15 min, the cells were washed and resuspended to 1 ml, and CD4+ T cells (>95% CD4+) were isolated using a Miltenyi VS+ Positive Selection Column (Miltenyi Biotec).
Macrophages were derived by culture of adherent monocytes. For this, blood mononuclear cells were resuspended to 5 × 106/ml in serum-free RPMI 1640 medium, and 17 ml was dispensed into TC75 Falcon flasks (Becton Dickinson, Franklin Lakes, NJ). Following 2 h of culture (6% CO2, 37°C) to allow the monocytes to adhere, nonadherent cells were removed by washing, and the media were replaced with RPMI 1640 containing 5% FCS, 5% human AB serum (Gemini BioProducts, Calabasas, CA), 50 nM calcitriol, 2 mM glutamine, 100 U/ml penicillin-G, 100 μg/ml streptomycin, and 0.6 μg/ml fungizone. Cells were fed with fresh media every 2 days, and in vitro differentiated macrophages were used for experimentation on day 7.
Primary HUVEC and human coronary artery endothelial cells purchased from Clonetics (San Diego, CA) were cultured in TC75 Falcon flasks in endothelial cell growth medium (Clonetics). Cells were used at passage four or five.
Analysis of p38α, β, γ, and δ mRNA levels by reverse- transcriptase PCR (RT-PCR)
Cells were lysed in RNAzol (Teltext, Friendswood, TX), and total RNA was isolated according to the manufacturer’s instructions. RNA integrity was verified by agarose gel electrophoresis and ethidium bromide staining. Relative levels of p38α, β, γ, and δ mRNA were determined using an RT-PCR approach based on the work of Holland et al. (21) and described in detail in Protocol Part Number 402823 (PE Applied Biosystems, Foster City, CA). Sample RNA (20 ng) was added to a 50 μl RT-PCR reaction mix (TaqMan PCR Core Reagent Kit; PE Applied Biosystems) that included specific forward and reversed primers and a specific probe sequence. The gene-specific primers provided templates for the RT reaction (48° × 30 min). Next, AmpliTaq Gold polymerase was heat activated (10 min × 95°C), and PCR was performed in the same reaction mix. Probes were complementary to a region amplified by the primer set and contained a 5′-reporter dye, 6-carboxyfluorescein (FAM), which was quenched by the presence of a 3′-quencher dye, 6-carboxytetramethylrhodamine (TAMRA). Taq polymerase has a 5′-nucleolytic activity and digests probe that binds cDNA during amplification. The released reporter dye fluoresces, and the accumulation of dye was measured in real time using a ABI Prism 7700 Sequence Detection System. The data were analyzed using Sequence Detector v1.6 software (PE Applied Biosystems).
Relative mRNA levels were calculated using a variant of the Comparative CT Method (22). Using this technology, several PCR cycles are required before the specific fluorescent signal emerges above background fluorescence (see Fig. 2⇓ for examples). The number of PCR cycles (CT) required to generate a threshold of 0.03 fluorescence units was determined in triplicate for each test RNA sample. During the exponential phase of PCR, cDNA is amplified by 1 + AE per cycle, in which AE is the amplification efficiency. AE values were calculated as described elsewhere (23), and ranged from 0.96 to 1 for p38α, β, γ, δ, and 18S rRNA. Thus, each cycle results in an approximate doubling of cDNA, and so it can be estimated that a sample yielding a CT of 10 has about 8 times the mRNA of a sample with a CT of 13. More precisely, the amount of a mRNA transcript in a sample is proportional to 1/(1 + AE)CT. Relative 18S rRNA values were determined and used to normalize samples for RNA content. 18S rRNA was selected because the proportion of total RNA that is ribosomal RNA does not vary substantially between cell types (24). Sample calculations are provided in Table I⇓.
Primer Express (ABI) was used to design primers and probes. Primers and probes were as follows: 18S rRNA, (5′) CGC CGC TAG AGG TGA AAT TCT, (3′) CAT TCT TGG CAA ATG CTT TCG, (probe) 6-FAM-ACC GGC GCA AGA CGG ACC AGA-TAMRA; p38α, (5′) AAC CTG TCT CCA GTG GGC TCT, (3′) CGT AAC CCC GTT TTT GTG TCA, (probe) 6-FAM-CGC CTA TGG CTC TGT GTG TGC TGC TT-TAMRA; p38β, (5′) CAC CCA GCC CTG AGG TTC T, (3′) AGA TGC TGC TCA GGT CCT TCT, (probe) 6-FAM-ACA CGC CCG GAC ATA TAT CCA GTC CC-TAMRA; p38γ, (5′) CGC CTC CGG CTG AGT TT, (3′) GCT TGC ATT GGT CAG GAT AGA, (probe) 6-FAM-CAG AGC GAT GAG GCC AAG AAC TAC ATG AA-TAMRA; p38δ, (5′) TGC TCG GCC ATC GAC AA, (3′) TGG CGA AGA TCT CGG ACT GA, (probe) 6-FAM-AAG GTG GCC ATC AAG AAG CCG AGC-TAMRA.
Generation and purification of polyclonal Abs specific to p38 family members
Polyclonal antisera were prepared in rabbits to the following peptides: p38α (amino acids (aa) 341–360), p38β (aa 346–364), p38γ (aa 321–339), and p38δ (aa 262–280). Peptides were conjugated to keyhole limpet hemacyanin and injected with CFA. Rabbit polyclonal antisera were also generated to full-length recombinant p38α and to GST-flag-p38δ. For immunoprecipitation studies, IgG was purified from antiserum using the ImmunoPure IgG (protein A) Purification Kit (Pierce, Rockford, IL).
Western blot analysis
Freshly isolated monocytes, CD4+ T cells, and neutrophils were resuspended at 5 × 107 cells/ml of lysis buffer (50 mM Tris, pH 7.5, 50 mM KCl, 1 mM EDTA, 0.5% Nonidet P-40, 1 mM Na3VO4, 50 nM calyculin A, 100 μM TPCK (l-1-tosylamido-2-phenylethyl chloromethyl ketone), 1 μg/ml pepstatin, and 1 mM PMSF). Confluent cultures of macrophages, HUVEC, and human coronary artery endothelial cells in TC75 flasks were washed twice with ice-cold PBS and resuspended with scraping into 2 ml of lysis buffer. All cell types were allowed to lyse 5 min on ice, and the soluble portion of the lysates was obtained by centrifugation 2 min × 14,000 g. Three volumes of lysate were transferred to fresh tubes containing 1 vol of 4× Laemmli buffer and heated to 95°C for 5 min. Portions of lysate were also used to determine protein content using the Pierce BCA assay. Samples (25 μg protein) were resolved using 10% SDS-PAGE (Novex, San Diego, CA), and transferred to Immobilon-P membranes (Millipore, Bedford, MA). Blots were blocked 1 h in wash buffer (20 mM Tris-HCl, pH 7.5, 137 mM NaCl, and 0.1% Tween-20) containing 2.5% gelatin and 5% milk protein. Blots were next incubated with primary Ab/antiserum diluted into wash buffer containing 5% milk protein. Abs included 0.5 μg/ml anti-p38α (C20) (Santa Cruz Biotechnology, Santa Cruz, CA), 1/1000 dilutions of rabbit anti-dual-phospho-p38 (lot 5) or anti-mono-phospho-p38 (lot 2) (New England Biolabs, Beverly, MA), or 1/2700 dilutions of antiserum raised against p38β, γ, or δ peptides. Blotting with anti-p38α (C20) or anti-phospho-p38 (dual or mono) was performed for 2 h at room temperature. Other primary Abs were blotted at 4°C overnight. Blots were then washed and incubated with a 1/3000 dilution of goat anti-rabbit horseradish peroxidase (Dako, Carpenteria, CA). Binding of secondary Ab was detected using enhanced chemoluminescence (Pierce). Positive controls for Western blot detection of p38 family members included recombinant p38α, flag-p38β, p38γ, and GST-flag-p38δ prepared in E. coli, as described previously (10). Molecular weights of proteins were estimated based on log-linear plots using SeaBlue protein standards (Novex).
Immune depletion assays
Macrophages and HUVEC lysates were prepared as described above for Western blotting. A total of 100 μg lysate protein adjusted to 250 μl with lysis buffer was supplemented with 5 μg of purified IgG. IgG included anti-full-length p38α, anti-p38β (peptide), anti-full-length p38δ, and preimmune IgG. Lysates were then rocked 6 h at 4°C. Immune complexes were bound by addition of 30 μl of protein A-agarose (Calbiochem-Novabiochem, La Jolla, CA). Lysates were rocked 2 h, and bound agarose was removed by centrifugation (15 s × 14,000 × g). The cleared supernatants were subjected to Western blot analysis, as described above.
p38 family members were immunodepleted from cell lysates, as described above. The bound agarose was washed twice in lysis buffer and twice in PBS and then resuspended to 50 μl in PBS adjusted to contain 10 mM MgCl2, 0.5 mM EGTA, 4 mM DTT, 2.5 μM cold ATP, 10 μCi [γ-32P]ATP, 6000 Ci/mmol (Amersham), and 2 μg of ATF2 (11-109). Reactions (28°C × 30 min) were terminated by addition of 4× Laemmli load buffer and boiled 5 min. Samples were resolved by 10% SDS-PAGE. The gels were dried, and phosphorylated ATF2 (11-109) was visualized by autoradiography and quantified using a PhosphorImager.
Validation of PCR primers and Abs used to investigate expression of p38 family members
Our objective was to measure the expression of p38α, β, γ, and δ in cell types important to inflammation. For this purpose, primer pairs were designed to measure the expression of each form of p38 mRNA by RT-PCR. The primer pairs were first tested in PCR reactions containing 1 ng of purified plasmids encoding p38α, β, γ, or δ (Fig. 1⇓). Each primer pair was shown to be absolutely specific for its targeted p38 family member.
To investigate protein expression, polyclonal rabbit antisera were generated against unique p38β, γ, and δ peptides. An excellent IgG against p38α was already available (Santa Cruz Biotechnology). The anti-p38α IgG and p38β, γ, and δ antisera were first tested on Western blots containing 4 ng of recombinant p38α, β, γ, and δ (Fig. 1⇑). Each antiserum/IgG was absolutely specific for its target p38.
p38α, β, γ, and δ mRNA levels in inflammatory cell lineages
To measure relative p38α, β, γ, and δ mRNA expression, we used a recently developed RT-PCR approach that utilizes the 5′-exonuclease activity of Taq polymerase (21). The exonuclease activity cleaves a reporter dye from an oligonucleotide probe that hybridizes to the amplified cDNA sequence (23). The release of the dye is catalyzed coincident with cDNA amplification and can be followed in real time by fluorometric measurement. Primers and PCR conditions are optimized to yield amplification efficiencies approaching 100%. For this reason, it is possible to determine the relative expression of different mRNAs within the same sample. Fig. 2⇓ illustrates the fluorescence generated versus cycle number for RT-PCR reactions with p38α, β, γ, and δ primers and probes and macrophage RNA. Expression of 18S rRNA was also measured to normalize for RNA input. Using this technology, several PCR cycles are required before the specific fluorescent signal emerges above background fluorescence. The p38δ amplification plot was the first p38 plot to emerge above background and was closely followed by p38α. The p38β and p38γ amplification plots lagged behind the p38δ plot by >9 and >11 cycles, respectively. The data indicated that p38δ mRNA was the most abundant p38 mRNA in this macrophage sample. p38α mRNA was about 30% as abundant as p38δ mRNA, and p38β and p38γ mRNAs were expressed at very low levels. Table I⇑ shows the calculations used to convert amplification plots to relative expression values.
Expression of p38α, β, γ, and δ mRNAs was next evaluated in RNA from three to four independent isolates/cultures of monocytes, macrophages, endothelial cells, CD4+ T cells, and neutrophils (Table II⇓). p38α mRNA was by far the dominant form of p38 mRNA expressed by monocytes. p38γ mRNA was undetectable, p38β mRNA was negligible, and p38δ mRNA was expressed at 10% the level of p38α mRNA. In striking contrast to monocytes, p38δ as well as p38α mRNA was abundant in macrophages, neutrophils, and CD4+ T cells. Neutrophils resembled monocytes and macrophages in very low or absent expression of p38β and p38γ mRNA. CD4+ T cells expressed somewhat higher amounts of p38β mRNA. p38 expression in endothelial cells was markedly different from that found in leukocytes. Distinguishing endothelial cells was the robust expression of p38β mRNA and the readily detectable expression of p38γ mRNA. p38α and p38δ mRNAs were also present in endothelial cells. Expression of p38 family mRNAs was similar in umbilical vein and coronary artery endothelial cells (data not shown).
Protein levels of p38 family members in inflammatory cell lineages
Western blot analysis was used to investigate the expression of p38 protein in the five cell types (Fig. 3⇓). p38 protein levels (ng p38/μg cell protein) were estimated by comparing lysates against dilutions of recombinant p38 proteins (dilution series not shown). The correlation between protein and mRNA expression was good. For example, p38α was the dominant form of p38 protein in monocytes and was estimated at 0.5 ng/μg protein. p38δ expression in monocytes was estimated at 0.05 ng/μg, and p38β was undetectable (<0.015 ng/μg). The faint, low m.w. bands observed when monocyte lysates were probed with p38β antiserum were nonspecific, i.e., these bands were still detected when blots were probed with p38β antiserum in the presence of excess β-peptide (not shown). In macrophages, p38α and p38δ were both expressed strongly, but, again, p38β was undetectable. p38α was abundant in neutrophils and CD4+ T cells with somewhat less expression in endothelial cells. CD4+ T cells and neutrophils also expressed p38δ. In neutrophils, p38δ antiserum detected a doublet. The minor (lower) band, but not the major (upper) band, proved to be p38δ based on specific competition with δ peptide (not shown). p38β expression was low in CD4+ T cells and undetectable in neutrophils. In contrast to the leukocyte lineages, p38β protein was abundant (∼0.3 ng/μg) in both umbilical vein and coronary artery endothelial cells. p38δ was also detectable upon prolonged exposures in both types of endothelial cells. Although p38γ expressed in lysates of murine muscle was readily detected using the p38γ antiserum, p38γ was undetectable (<0.005 ng/μg) in the five cell types (data not shown). These data contrasted with the expression of p38γ mRNA in endothelial cells.
Activation of p38 family members in macrophages
When overexpressed in transfected cells, each p38 family member can be activated by TNF-α or IL-1β (12). Therefore, we investigated the activation of endogenously expressed p38 family members in response to inflammatory stimuli. For this, macrophages were stimulated with LPS, TNF-α, or GM-CSF for timed intervals, lysed, and subjected to Western blot analysis using an Ab raised against p38α residues 171–186 in which Y182 was phosphorylated (Fig. 4⇓). A 41- and a 43.5-kDa form of p38 was phosphorylated following LPS and TNF-α stimulation, but not following GM-CSF stimulation. Similar blots were stripped and reprobed with an Ab raised against p38 residues 171–186 in which both T180 and Y182 were phosphorylated (Fig. 5⇓). Remarkably, this dual-phospho-p38 Ab recognized the 41-kDa and a very faint 44-kDa band, but it did not detect the 43.5-kDa band that was detected using the mono-phospho-p38 Ab. When the phosphatase inhibitor calyculin A was included as stimuli, the 44-kDa phospho-p38 was enhanced markedly. The reduced mobility exhibited by the 44-kDa band and the ability of phosphatase inhibitor to increase its intensity were consistent with the 43.5- and 44-kDa proteins representing mono- and dual-phosphorylated proteins, respectively.
The 41- and 43.5/44-kDa phosphoproteins comigrated with p38α and p38δ, respectively (data not shown). The identity of the 41- and 43.5/44-kDa phosphoproteins was confirmed by selective immunodepletion analysis. Ab raised against full-length p38α efficiently immunodepleted all p38α and all 41-kDa phospho-p38 from macrophage lysates (Fig. 6⇓). Ab raised against full-length p38δ depleted most of the p38δ present in the macrophage lysates along with most of the 43.5-kDa phosphoprotein. Similar experiments confirmed that the 44-kDa phosphoprotein induced in the presence of calyculin A was p38δ (data not shown).
Immunokinase assays were used to determine whether the phosphorylation of p38α and p38δ correlated with their activation. Macrophages were treated with or without LPS for 10 min and lysed, and p38α and p38δ were selectively immunoprecipitated from the lysates. The p38 bound in the immunoprecipitates was used to phosphorylate ATF2 (Fig. 7⇓). LPS induced p38α activity >12-fold and p38δ activity >8-fold above background. Induced p38α activity was ∼3.5-fold higher than induced p38δ activity.
Activation of p38 family members in endothelial cells
Activation of p38 family members in endothelial cells was of particular interest due to the relative abundance of p38β in this cell lineage. HUVECs were treated with IL-1β for timed intervals and lysates were evaluated for phosphorylation of p38 members (Fig. 8⇓A). A 41-kDa form of p38 was phosphorylated in response to IL-1β. Immunodepletion experiments identified the phosphoprotein as p38α (Fig. 8⇓B). Anti-full-length p38α selectively eliminated p38α from the lysates and removed all of the 41-kDa phosphoprotein. The p38β peptide Ab and the anti-full-length p38δ Ab selectively eliminated p38β and p38δ, respectively, but did not reduce the major 41-kDa phosphoprotein.
p38α, β, and δ were next immunoprecipitated from lysates of IL-1β-stimulated HUVEC and analyzed by immunokinase assay using ATF2 as substrate. Fig. 9⇓A confirms a strong activation (>10-fold) of p38α by IL-1β. However, p38β was also activated, although to a lesser extent than p38α. When ATF2 was used as substrate, p38α was >6-fold more active than p38β in IL-1β-stimulated HUVEC. The authenticity of p38β activity was confirmed. Neutralization of p38β antiserum with excess β peptide reduced the kinase activity measured in anti-p38β immunoprecipitates to prestimulation levels (Fig. 9⇓B). Immunokinase assays did not detect activation of p38δ following IL-1β treatment of HUVEC (data not shown).
This work is the first characterization of p38 family member expression in purified primary cell types. We show that the expression of p38 family members is not ubiquitous, but is controlled during cell differentiation and in a lineage-specific fashion. Differential expression was most striking in monocytes that strongly expressed p38α, but did not express p38β or p38γ. Expression of p38δ was low, but detectable. However, the low expression of p38δ was probably overestimated since monocytes purified by elutriation contain ∼5% with lymphocytes, and lymphocytes express high levels of p38δ. We found that p38β expression was under strong lineage-specific control. With the exception of low expression of p38β in CD4+ T cells, the p38β gene was essentially silent in leukocytes. In contrast, p38β was expressed strongly in endothelial cells. p38γ was also silent in leukocytes, and although p38γ mRNA was expressed by endothelial cells, p38γ protein was undetectable, consistent with previous reports that p38γ is a muscle-specific protein (6, 7). Interestingly, p38δ expression was regulated during cellular differentiation within the myeloid lineage. As blood monocytes differentiated into macrophages, a striking induction of p38δ mRNA and protein occurred so that p38δ protein became at least as abundant as p38α.
When p38α, β, γ, and δ are overexpressed in transfected cells, each p38 family member can be activated by cotransfecting cells with MKK6 or by stimulating the cells with TNF-α or IL-1β (7, 8, 9, 10, 11, 12). These data suggest that the activation of p38 family members may be regulated coordinately. However, to date, only a few papers have dealt with the activation of endogenous p38β, δ, or γ (9, 25), and one report showed the activation of p38β to be independent of p38α activation in a neuronal cell line (25).
Herein, immunokinase assays were used to demonstrate that both p38α and p38δ are activated in macrophages stimulated with LPS. However, p38α was activated to a greater extent, and phosphorylation analysis suggested that p38δ and p38α were regulated distinctly. Following LPS activation of macrophages, p38α and p38δ were both recognized by Ab against p38α (171–186), in which only Y182 was phosphorylated. Following LPS activation, p38α was also recognized by Ab against p38 (171–186), in which both T180 and Y182 were phosphorylated, whereas p38δ was poorly recognized by this Ab. Thus, p38δ appeared to be preferentially phosphorylated on tyrosine versus threonine in LPS-activated macrophages. Differential phosphorylation of tyrosine may reflect the bimolecular interactions known to occur between MKK and MAPK level kinases. Phosphorylation occurs by an ordered addition of phosphate first to tyrosine and then to threonine with MKK dissociation from MAPK between each step (26). Our data suggest phosphorylation of p38δ T180 may be relatively inefficient in LPS-activated macrophages, or, conversely, dephosphorylation of p38δ T180 may be relatively efficient. We speculate that p38δ T180 represents a site of regulation by stimuli that may prime macrophages for full activation of p38δ. p38δ activity was increased 25-fold in macrophages treated with LPS and calyculin A versus LPS alone (C. Manthey, unpublished results), underscoring the potential for signals that down-regulate phosphatase activity to synergize with LPS for p38δ activation.
We report herein that endothelial cells express comparable levels of p38α and p38β. Immunokinase assays were used to demonstrate that both forms are activated in endothelial cells stimulated with IL-1β. However, p38α activity was >6-fold more than p38β activity. When lysates of IL-1β-treated cells were evaluated by Western blot, the major phospho-p38 was 41 kDa and was identified as p38α by immunodepletion analysis. Upon prolonged exposures, a 42.5-kDa phospho-p38 was also observed that comigrated with p38β (C. Manthey, unpublished). Attempts to identify this faint band by immunodepletion were unsuccessful because it was no longer detectable in lysates after immunodepletion even when Ab was omitted (presumably due to dephosphorylation during the extended procedure).
By understanding the expression and activation patterns of p38 members, insights into their functions are obtained. For example, the p38 inhibitors, SB202190 and SB203580, can block TNF-α production by monocytes by >95% (14, 15, and C. Manthey, unpublished data). These compounds block p38α and p38β with nearly equal potency, but do not inhibit p38γ or p38δ (8, 9, 12). Thus, either p38α or p38β might mediate TNF-α production. However, we show herein that p38α protein was expressed abundantly in leukocytes. In contrast, p38β protein expression was undetectable in monocytes, macrophages, and neutrophils, and a highly sensitive RT-PCR approach confirmed that p38β mRNA was absent or present in extremely low amounts in these cells. Thus, p38α, but not p38β, appears to contribute to TNF-α production by these lineages. Similarly, SB203580 was shown recently to block FMLP-induced neutrophil chemotaxis and superoxide production (19). Because expression of p38α is abundant in neutrophils and p38β was undetectable, it is likely that p38α also mediates these functions.
Inhibitors of p38 block IL-1β-induced endothelial cell expression of IL-6, and prostaglandin H synthase-2 (18). Although IL-1β activated p38α to a greater extent than p38β, our data do not exclude the possibility that p38β yet plays a role in these responses. Recently, p38 inhibitors were shown to block vascular endothelial cell growth factor-induced actin filament formation and migration (27), and more work is needed to investigate the role of p38β in responses to this and other stimuli. CD4+ T cells also expressed p38β. Although we did not investigate activation of p38β in this lineage, a contribution of p38β should be considered in lymphocyte responses suppressed by SB203580, e.g., IL-2- and IL-7-induced lymphocyte proliferation.
The role for p38δ in macrophage function is not known. The increase in p38δ upon macrophage differentiation suggests a role in functions gained as monocytes mature into macrophages, such as increased phagocytic activity. In this regard, it is notable that p38δ, but not p38α, phosphorylates stathmin (13), a ubiquitous 18-kDa cytosolic protein involved in microtubule dynamics (28). Also, we have found that SB202190 inhibits maximally ∼65% of LPS-induced TNF-α production by macrophages (C. Manthey, unpublished data). Since SB202190 does not inhibit p38δ (8, 9), we speculate that p38δ may mediate the ∼35% of TNF-α production that is insensitive to SB202190. A role for p38δ in inflammation was suggested recently by Jiang et al. (9), who reported the activation of renal p38δ during glomerulonephritis in rats.
In the brief time since the discovery of p38, a considerable effort has been mounted to develop p38 inhibitors for inflammatory diseases. Compounds described in the literature inhibit both p38α and p38β. As reported herein, p38α is the major p38 expressed by monocytes, and p38α is the most active p38 in LPS-treated macrophages and IL-1β-treated endothelial cells. The data suggest that p38α may be more important relative to p38β in the inflammatory response. Development of inhibitors with greater selectivity for p38α will permit the testing of this hypothesis and may result in compounds with more precise and focused in vivo pharmacology.
We thank George Keesler for recombinant p38α, β, δ, and γ protein and ATF2 (11-109); Tom Zamborelli for p38α, β, δ, and γ peptides; David Andrew for purified CD4+ T cells; and Stephen Kinney for purified neutrophils.
↵1 Address correspondence and reprint requests to Dr. Carl L. Manthey, 3-Dimensional Pharmaceuticals, 665 Stockton Dr., Exton, PA 19341.
↵2 Abbreviations used in this paper: MAPK, mitogen-activated protein kinase; AE, amplification efficiency; ATF2, activating transcription factor-2; CT, polymerase chain reaction cycle to threshold fluorescence; FAM, 6-carboxyfluorescein; GM-CSF, granulocyte-macrophage CSF; MKK, MAPK kinase; RT, reverse transcriptase; TAMRA, 6-carboxytetramethylrhodamine.
- Received October 23, 1998.
- Accepted January 6, 1999.
- Copyright © 1999 by The American Association of Immunologists