Effects of p38 MAP Kinase Inhibitors on the Differentiation and Maturation of Erythroid Progenitors

Abstract

In rodents, p38 MAP kinase inhibitors (p38is) induce bone marrow hypocellularity and reduce reticulocyte and erythrocyte counts. To identify target cell populations affected, a differentiating primary liquid erythroid culture system using sca-1⁺cells from mouse bone marrow was developed and challenged with p38is SB-203580, SB-226882, and SB-267030. Drug-related alterations in genes involved at different stages of erythropoiesis, cell-surface antigen expression (CSAE), burst-forming unit erythroid (BFU-E) colony formation, and cellular morphology (CM), growth (CG), and viability were evaluated. CSAE, CM, and decreases in BFU-E formation indicated delayed maturation, while CG and viability were unaffected. Terminal differentiation was delayed until day 14 versus day 7 in controls. CSAE demonstrated higher percentages of sca-1⁺cells after day 2 and reduced percentages of ter119⁺ cells after day 7 in all treated cultures. Real-time reverse transcriptase polymerase chain reaction revealed a transient delay in expression of genes involved at early, intermediate, and late stages of erythropoiesis, followed by rebound expression at later time points. Results demonstrate p38is do not irreversibly inhibit erythrogenesis but induce a potency-dependent, transient delay in erythropoietic activity. The delay in activity is suggestive of effects on sca-1⁺bone marrow cells caused by alterations in expression of genes related to erythroid commitment and differentiation resulting in delayed maturation.

Introduction

The p38 mitogen-activated protein kinase (p38) is a family of serine/threonine kinases that consists of four isoforms, including p38α, p38β, p38γ and p38δ. Each isoform differs in cellular localization, regulation of kinase activation, and phosphorylation of downstream substrates (Dominguez, Powers, and Tamayo 2005). The alpha isoform of p38 plays a pivotal role in many physiological processes, including cytokine production (TNFα and interleukin-1) involved in inflammation (Dominguez, Powers, and Tamayo 2005; Lee 1994; Nagata et al. 1998; Tamura et al. 2000; Uddin et al. 2004). Multiple p38 inhibitors of the pyridinyl imidazole class have been developed, including SB-203580 [4-[4-Fluorophenyl)-2-(4-methylsulfinylphenyl)-5-(4-pyridyl)imidazole] (Bhagwat et al. 1999), SB-226882 [4-(4-(4-Fluorophenyl)-1-(4-piperidinyl)-1H-imidazol-5-yl]-N-methyl-2-pyrimidinamine] (Bhagwat et al. 1999), and SB-267030 [1-(4-Piperdinyl)-4(4-fluorophenyl)-5[(2-methylphenyl) amino]pyrimidin-4-yl]imidazole] (Metcalf and Dillon 2006). These pyridinyl imidazole–based p38α inhibitors have shown activity toward other p38 isoforms, including p38β and p38γ however, they are more selective for inhibition of p38α, and both selectivity and inhibition profiles have been previously reported (Fabian et al. 2005; Lee and Dominguez 2005). Inhibition of p38α has become an attractive yet challenging therapeutic target for treatment of chronic inflammatory disorders, such as rheumatoid arthritis, septic shock, restenosis, and inflammatory bowel syndrome (Dominguez, Powers, and Tamayo 2005), but few candidates have progressed to Phase II clinical trials due to unsuitable safety profiles noted preclinically and clinically (Goldstein and Gabriel 2005).

Administration of the pyridinyl imidazole–based p38is to rodents results in transient, potency-dependent bone marrow hypocellularity, reduced red cell mass, and decreased reticulocyte counts (Zhang et al. 2002). Although activation of p38α has been observed in many hematopoietic cells types, the basis for the predilection of these compounds to affect the erythropoietic process following in vivo administration is unclear. Several studies have been performed to characterize the action of p38 and its role during erythropoiesis. Tamura et al. (2000) demonstrated that most p38 homozygous knockout die during embryonic development, with survivors anemic as a consequence of failed definitive erythropoiesis. The mechanism may involve diminished erythropoietin (EPO) gene expression due to lack of p38 dependent stabilization of EPO mRNA (Tamura et al. 2000). Treatment of a human erythroid cell line with EPO resulted in gradual and sustained activation of p38 activity, which decreased upon EPO withdrawal (Jacobs-Helber, Ryan, and Sawyer 2000). It was hypothesized that p38is may modulate proliferation, commitment, differentiation, and/or maturation of uncommitted stem cell antigen positive (sca-1⁺) bone marrow cells via alterations in genes involved in erythroid lineage commitment.

To better study the hematopoietic process, an in vitro method to evaluate effects of p38is in the mouse was developed. Differentiation and maturation of human CD34⁺ and/or CD36⁺ erythroid progenitor cells is dependent on a microenvironment that mimics human bone marrow (Freyssinier et al. 1999; Scicchitano et al. 2003). In contrast to human CD34⁺and CD36⁺cells, mouse hematopoietic progenitor cells are not commercially available. Furthermore, hematopoietic stem cells with the potential to expand into the erythroid lineage compose only approximately 1% of cells present in mouse bone marrow (Spangrude, Heimfeld, and Weissman 1988). Sca-1 is an integral membrane glycoprotein routinely used to purify uncommitted stem cells (Bugarski et al. 2006). The most conventional way to evaluate murine hematopoietic stem cells is using a colony-forming unit assay derived from mouse bone marrow (Gregory 1976). This assay involves growing hematopoietic stem cells in a semisolid medium, which results in the generation of erythroid lineage progenitors known as burst-forming unit erythroids (BFU-Es) (Gregory 1976). Due to limitations with this assay, including difficulty in obtaining viable cultured cells from semisolid media and insufficient cell number for conducting multiparameter assessment, we developed a short-term mouse liquid bone marrow culture system, suitable for expansion of erythroid progenitors from purified mouse bone marrow sca-1⁺cells. Liquid culture systems have previously been utilized to assess effects on the erythropoietic process using murine fetal liver erythroblast precursor cells (von Lindern et al. 2001), erythroblasts isolated from p53⁻/⁻ mice (Dolznig et al. 2001) and mouse embryonic stem cell lines generated by embroid body formation (Carotta et al. 2004). Unlike the cell populations in these assays, we utilized cells not yet committed to erythroid lineage for a short-term, primary murine liquid culture system capable of producing sufficient numbers for multiparameter assessment.

Sca-1⁺cells, isolated from mouse bone marrow, were induced to differentiate along the erythroid lineage using selective media for 14 days. To modulate p38 activity, sca-1⁺cells were cultured in the presence or absence of three selected p38is of varying potencies. Expression of early-, intermediate-, and late-stage genes involved in both early erythroid commitment and differentiation/maturation were assessed using TaqMan® real-time RT-PCR (reverse transcriptase polymerase chain reaction). Differentiation and maturation of hematopoietic progenitors into committed erythroid cells were assessed by flow cytometry using lineage-specific cell surface markers, and BFU-E colony formation was assessed using the colony-forming unit assay along with cell growth, viability, and morphology. This methodology allowed identification of target populations in mouse bone marrow affected by p38is.

Materials And Methods

Mice and Isolation of Bone Marrow Stem Cells

Forty eight- to ten-week-old male CD-1 mice (Charles River Laboratories, Raleigh, North Carolina) were used to provide bone marrow cells using a modification of published methods (Visser et al. 1984). All animal care was performed in accordance with guidelines set for in the Guide for the Care and Use of Laboratory Animals (National Research Council 1996). The GlaxoSmithKline institutional animal care and use committee reviewed and approved all animal procedures used in these experiments.

Briefly, mice were killed via carbon dioxide asphyxiation/exsanguination, and bone marrow cells from femurs and tibias were aseptically collected. Bone marrow cells flushed from shafts with ISCOVE’S Modified Dulbecco’s Media (IMDM, Invitrogen, Carlsbad, California) were supplemented with 10% fetal bovine serum (FBS) (Invitrogen) and 100 U/mL penicillin-streptomycin (Invitrogen) and passed through sterile 40 μm nylon cell strainers (BD Bioscience, San Jose, California) to generate single-cell suspensions. Cells were centrifuged, washed twice, and pellets resuspended in supplemented IMDM media and pooled. Cell count and viability were determined using Trypan Blue exclusion dye stain (Sigma-Aldrich, St. Louis, Missouri). The pooled cell suspension was divided equally between four T75 cell culture flasks (BD Bioscience) and stored overnight (6°C) prior to enrichment for sca-1⁺cells.

Enrichment of Mouse Sca-1⁺Bone Marrow Cells

Sca-1⁺cell enrichment from mouse bone marrow was performed twice using MACS Sca-1 Isolation System (Mitenyi Biotec, Auburn, California) and magnetic microbeads conjugated with monoclonal rat antimurine sca-1 antibodies (Mitenyi Biotec). Cell samples were centrifuged and resuspended in phosphate buffered saline (PBS) (Invitrogen) supplemented with 0.5% bovine serum albumin (Sigma-Aldrich) and 2 mM EDTA, pH 7.2 (Sigma-Aldrich) twice, then incubated with sca-1 Multi-Sort Micro-Beads (Mitenyi Biotec). Cells and beads were centrifuged, washed, and resuspended in PBS. Cell suspensions were loaded onto sca-1 selection columns, and eluted sca-1 negative fractions were removed, pooled, and saved for flow cytometric analysis. Sca-1⁺fractions were flushed with PBS, reseparated through freshly prepared columns, and pooled. Both sca-1⁺fractions were assessed, as described below, using flow cytometry. Prior to culture initiation, the sca-1⁺fraction was centrifuged and washed with supplemented IMDM medium, and the resulting pellet was resuspended in liquid culture bone marrow media (StemCell Technologies, Vancouver, British Columbia) supplemented as indicated below.

Cell Culture Conditions and p38i Treatment

Sca-1⁺bone marrow cells were cultured, in duplicate, under erythroid conditions in liquid culture bone marrow media (StemCell Technologies) containing IMDM, 15% heat-inactivated FBS, 100 U/mL penicillin-streptomycin, 1% bovine serum albumin, 10 μg/mL bovine pancreatic insulin, 200 μg/mL iron saturated human transferrin, 10⁻⁴ M 2-mercaptoethanol, 2 mM L-glutamine, 50 ng/mL recombinant murine (rm) stem cell factor, 10 ng/mL rm interleukin-3, 10 ng/mL recombinant human (rh) interleukin-6, and 3 U/mL rh erythropoietin at 37°C with 5% CO₂ and 1% O₂ in a humidified atmosphere. Cells were cultured for three days until logarithmic phase growth was achieved. Cells were cultured for an additional fourteen days in the presence or absence of noncytotoxic concentrations of three p38is, including SB-203580 [p38iA, 4-(4-Fluorophenyl)-2-(4-methylsulfinylphenyl)-5-(4-pyridyl)imidazole] (Bhagwat et al. 1999), SB-226882 [p38iB, 4-(4-(4-Fluorophenyl)-1-(4-piperidinyl)-1H-imidazol-5-yl]-N-methyl-2-pyrimidinamine] (Bhagwat et al. 1999), and SB-267030 [p38iC, 1-(4-Piperdinyl)-4(4-fluorophenyl)-5[(2-methylphenyl)amino]pyrimidin-4-yl]imidazole] (Metcalf and Dillon 2006) ranked in order of potency. Two independent experiments were performed. The p38i were provided by GlaxoSmithKline (GSK), King of Prussia, Pennsylvania.

Concentrations used and IC50 binding to human p38 are as follows: p38iA = 50 μM, 48 nM; p38iB = 1 μM, 27 nM; and p38iC = 0.1 μM, 0.4 nM, respectively (Adams et al. 2001; Fabian et al. 2005). Concentrations selected were based on p38i potency (Adams et al. 2001; Fabian et al. 2005), and pilot experiments conducted with these inhibitors (data not shown) and results from experiments demonstrating that treatment of human bone marrow stromal cells with similar concentrations of SB-203580, SB-226882, and SB-267030 did not produce cytotoxicity (Scicchitano et al. 2008). Dimethyl sulfoxide (DMSO, Sigma-Aldrich) was used as the vehicle control and for preparation of p38i stock solutions. Final test solutions of p38is contained 0.1% DMSO. Culture media was replenished on alternate days with fresh media containing 0.1% DMSO, 50 μM SB-203580, 1 μM SB-226882, or 0.1 μM SB267030 to maintain both drug and cell concentrations between 1 × 10⁵ to 5 × 10⁵ cells/mL.

Cell Growth and Viability Assessment

Cell growth and viability were assessed daily in each culture (n = 2) using Trypan Blue exclusion dye (Sigma-Aldrich). On days 0, 2, 5, 7, 11, and 14, a fraction of cells from each culture was harvested for expression profiling using real-time RT-PCR, assessment of lineage markers using flow cytometry, BFU-E colony formation, and morphological analysis. Data are represented as mean ± standard deviation.

Real-time RT-PCR Analysis

Total RNA was isolated from cell pellets using Trizol Reagent (Invitrogen) as previously described (Scicchitano et al. 2003). Genomic DNA was removed from each RNA sample using the DNA-free Kit (Ambion, Austin, Texas). This was followed by synthesis of first-strand cDNA using Superscript II^™ reverse transcriptase and random hexamers (Invitrogen) to prime cDNA synthesis.

Primer Express software (Applied Biosystems, Foster City, California) was used to design primers and fluorogenic probe sets for early (SCA-1, stem cell leukemia [SCL], GATA-2), intermediate (GATA-1, erythroid krüppel–like factor [EKLF], NFE2, EPO-Receptor [EPO-R], FOG [friend of GATA-1], FOG2 [friend of GATA-2]) and late-stage (β-GLOBIN, GLYCOPHORIN A, and PROTEIN 4.2) genes. A primer and probe set for 18S ribosomal RNA (18S) was included and used as a normalizing gene. Primer and probe sets were chosen based on published GenBank gene sequences (www.ncbi.com). Gene-specific primer (Department of Gene Expression Sciences, GSK) and fluorogenic probe (BioSource International Camarillo, California) sets are listed in Table 1. Real-time RT-PCR Core Reagent Kit (Applied Biosystems) was used for real-time RT-PCR.

Real-time RT-PCR was performed for each gene (n = 3) using 100ng single-stranded cDNA. Negative controls included no template and minus reverse transcriptase template. Normal murine bone marrow cDNA was used to generate standard curves for each profiled gene. Real-time RT-PCR cycling conditions were as follows: 50°C for two minutes, 95ºC for ten minutes, and forty cycles at 95ºC for fifteen minutes and at 58ºC for one minute. Data were collected and quantitatively analyzed on an ABI Prism 7700 Sequence Detection System (Applied Biosystems). Standard curves were generated to ensure efficiency of each primer and probe set (data not shown), and transcript values were calculated relative to a cDNA dilution series as described in Applied Biosystems Sequence Detection System User Bulletin 2: ABI Prism 700 Sequence Detection System.

Target and normalizing gene (18S) copy numbers were calculated using cycle threshold values and respective standard curves. Target quantities were normalized to the 18S, and average normalized target gene copy numbers were calculated and expressed as mean target copy number per 1 million copies of 18S ± standard deviation for each gene analyzed (n = 3).

Cell Surface Phenotype Analysis

Using flow cytometry, initial cell samples collected immediately prior to and following sca-1⁺enrichment and those harvested throughout the culture period on days 0, 2, 5, 7, 11, and 14 were evaluated for expression of erythroid differentiation and/or maturation cell surface markers using fluorochrome-conjugated antibodies to murine stem cell marker sca-1 [fluorescein isothiocyanate (FITC)] (BD Bioscience) and erythroid lineage marker ter119 [phycoerythrin (PE)] (BD Bioscience). Isotype-matched FITC and PE-conjugated control antibodies were used as controls (BD Bioscience).

Flow cytometric analyses were performed in triplicate. Briefly, a volume containing 2 × 10⁵ cells from each sample was distributed into 12 × 75 mm flow cytometry tubes (BD Bioscience), centrifuged, and pellets resuspended in PBS (Invitrogen). Cells were labeled for fifteen minutes on ice with rat antimurine CD16/CD32 (BD Bioscience) to block nonspecific binding, and then incubated with appropriate antibodies for thirty minutes at 4°C protected from light. After labeling, cells from each sample were washed with cold PBS (Invitrogen), centrifuged at 200 × g for five minutes and resuspended in 500 μL cold PBS (Invitrogen). Dead cells were excluded by 7-Amino-Actinomycin D staining (BD Bioscience). Analysis was performed in a BD Bioscience FACSscan flow cytometer using CellQuest acquisition and analysis software (BD Bioscience) after gating on viable cells. Ten thousand events per sample were recorded and analyzed.

Morphological Characterization

Morphological evaluation was performed on cells generated from harvested culture samples collected at each time point (n = 2) in two independent experiments. Cells from each sample were fixed in Saccamanno fluid (Fisher Scientific, Leicestershire, United Kingdom), centrifuged, washed, and resuspended. Using a Cytospin 2 cytocentrifuge (Shandon, Pittsburgh, Pennsylvania), resuspended cells were centrifuged onto Superfrost Plus Microslides (VWR Scientific, West Chester, Pennsylvania), air dried, stained with a Wright-Giemsa stain (Sigma-Aldrich), and examined microscopically (Lilli 1965).

BFU-E Colony-Forming Unit Assay

On day 0, prior to treatment, and days 2, 5, and 7, following treatment in liquid culture, fifteen thousand cells from vehicle-and p38iA-, p38iB-, and p38iC-treated cultures were transferred from the bone marrow liquid culture media to methylcellulose-based media containing similar components. Cells from each group were plated in triplicate and cultured at 37° C with 5% CO₂ and 1% humidified O₂ for fourteen days. BFU-E colonies were scored based on the presence of an orange-red color (synthesized hemoglobin) and multiclustered morphology using an inverted microscope on days 2 through 7 postplating.

Statistical Analysis

Statistical analysis was performed, where appropriate, using the Student’s t test. Differences versus control on each day of study with p ≤ were considered statistically significant.

Results

Enrichment of Mouse Sca-1⁺Bone Marrow Cells

Immunophenotypic characterization of pooled bone marrow prior to separation revealed sca-1⁺purity to be 5.6% (Figure 1A). Following two rounds of enrichment, sca-1⁺purity used for subsequent culture was 94% (Figure 1B).

Cell Growth and Viability

No significant differences in growth rates or viability were observed between untreated controls and p38i-treated cultures (Figure 2). Logarithmic phase growth was achieved three days following sca-1⁺cell separation and continued through day 14 (Figure 2). Cell counts in both p38i-treated and untreated control cultures increased exponentially from approximately day 1 to day 14 (Figure 2). Cell viability was not affected by p38i treatment (data not shown).

Morphological Characterization

Morphological evaluation was performed on days 0, 2, 5, 7, 11, and 14. Initially, the majority of cultured control cells consisted of a relatively uniform population of blasts with large central nuclei and minimal dark blue cytoplasm. Reduced cell size, condensed nuclei, and increased cytoplasm were evident in control cultures over time, which is consistent with erythroid differentiation (Allen and Dexter 1982; Fibach and Rachmilewitz 1993). By day 7, control cultures were composed primarily of intermediate-to late-stage erythroid cells, including some metarubricytes, nucleated red blood cells (nRBCs), reticulocytes, and apoptotic cells and occasional mature red cells (Figure 3A). Cellular morphology of cultures treated with the least potent inhibitor (p38iA) was similar to control cultures with some evidence of terminal differentiation on day 7 (Figures 3A, 3B). Cells treated with the two more potent inhibitors were less differentiated throughout the culture period and contained larger numbers of blast cells with increased nuclear/cytoplasmic ratios (Figures 3B through 3D). There were a greater percentage of immature cells in cultures treated with the most potent inhibitor (p38iC), although differences in cellular morphology between p38iB and p38iC were slight. By day 14, cultures treated with p38iB and p38iC exhibited morphologic features of terminal differentiation (as opposed to day 7 in controls).

BFU-E Colony-Forming Unit Assay

The effect of p38i on BFU-E colony formation is shown in Figure 4. A dose-dependent decrease in the number of BFU-E colonies was observed on day 2 following exposure to all three p38is as compared to control cultures. While numbers of BFU-E colonies in control and p38iA-and p38iB-treated cultures decreased significantly by day 7, colony numbers in cultures treated with the most potent inhibitor (p38iC) remained relatively unchanged.

Real-time RT-PCR Analysis

Control Cultures

In control cultures, real-time RT-PCR of SCA-1, SCL, and GATA-2 genes involved in early stages of erythroid development revealed distinct patterns of gene expression throughout the 14-day culture period (Figures 5A through 5C). Maximal SCA-1 gene expression was observed on day 0 (Figure 5A). SCA-1 expression decreased during stages of differentiation/maturation, and expression was virtually undetectable on day 14 (Figure 5A). SCL expression increased dramatically on days 2, 5, and 7 and then decreased on days 11 and 14 (Figure 5B). GATA-2 expression, in controls, was stable or generally decreased over the culture period (Figure 5C).

Genes expressed during intermediate stages of erythroid development (GATA-1, EKLF, EPO-R, and FOG) displayed similar biphasic patterns of gene expression (Figures 6A through 6D). Expression of these genes was low or undetectable on days 0, 2, and 5; peaked at day 7; decreased dramatically on day 11; and rebounded slightly on day 14. Expression patterns for NFE2 and FOG2 differed from other genes expressed during the intermediate phase of erythrogenesis (Figures 6E, 6F). NFE2 expression decreased progressively in control cultures from day 0 to day 14, whereas FOG2 expression was virtually undetectable on day 0, followed by a gradual increase, which peaked on day 5.

The most dramatic changes, in terms of magnitude, were observed in genes involved in later stages of erythroid development (Figures 7A through 7C). β-GLOBIN, GLYCOPHORIN A, and PROTEIN 4.2 gene expression in controls was virtually undetectable until day 7. On day 7, in control cultures, peak gene expression of β-GLOBIN, GLYCOPHORIN A, and PROTEIN 4.2 were increased 84; 750; and 1,855 fold, respectively, compared to day 0 values (Figures 8A through 8C).

p38i-treated Cultures

In p38i-treated cells, expression of genes involved in early stages of erythroid development (SCA-1, SCL, and GATA-2) differed from controls (Figures 5A through 5C). Maximal expression of SCA-1 was observed on day 0 (Figure 5A). In contrast to control cultures, p38i-treated cultures displayed a potency-dependent reduction in SCA-1 expression on days 2 and 5, but by day 7, expression approximated control values and, by day 11, rebounded in p38iB- and p38iC-treated cultures (Figure 5A). Expression of SCL in p38i-treated cultures was decreased through day 7 but variably increased by day 14 (Figure 5B). In contrast to controls, GATA-2 expression in p38i-treated cultures increased between days 11 and 14 in a generally potency-related pattern (Figure 6C).

Genes expressed during intermediate stages of erythroid development (GATA-1, EKLF, EPO-R, and FOG) displayed similar biphasic patterns of gene expression (Figures 6A through 6D) in p38i-treated cultures as controls. However, expression of intermediate stage genes in p38i-treated cultures was blunted on day 7 in a p38i potency-dependent manner. Rebound expression of intermediate stage genes was noted on day 14, which was generally greater than expression noted in controls. Potency-dependent reductions in FOG expression were noted in p38i-treated cultures on days 2 and 5 (Figure 6D). Expression patterns for NFE2 and FOG2 differed from other genes expressed during the intermediate phase of erythrogenesis (Figures 6E, 6F). In contrast to controls, NFE2 expression was decreased, followed by an increase in expression observed in p38i-treated cultures (Figure 6E). The maximal loss of NFE2 expression correlated temporally with inhibitor potency occurring maximally on day 5 for p38iC, day 7 for p38iB, and day 11 for p38iA. Suppression of FOG2 was noted for all p38i-treated cultures. However, a potency-dependent effect throughout the culture period was not observed (Figure 6F).

The most dramatic changes, in terms of magnitude, were observed in genes involved in later stages of erythroid development (Figures 7A through 7C). In comparison to control cultures, expression of β-GLOBIN, GLYCOPHORIN A, and PROTEIN 4.2 were suppressed in all p38i-treated cultures. Rebound increases in expression of these genes were observed by day 14. Since time points beyond day 14 were not included, maximal response for p38i-treated cultures may not have been observed.

Cell Surface Phenotype Analysis

Sca-1, a glycoprotein expressed on hematopoietic stem cells, and ter119, a glycoprotein expressed by erythroid lineage cells, were used for phenotypic characterization of cultures. Initially, sca-1⁺ hematopoietic progenitor cells were ter119^negative. As expected, the percentage of sca-1⁺ cells decreased and the percentage of ter119 cells increased over time, confirming cell maturation of erythroid lineage. Phenotypic characterization of sca-1⁺ cells by flow cytometric analyses is illustrated in Figure 8. Percentage of sca-1⁺ hematopoietic progenitor cells (sca-1⁺ter119^negative) was approximately 90% on day ⁻1. A 10% reduction in sca-1 expression was noted on day 0 in all cultures. Higher percentages of sca-1⁺ cells were detected in all p38i-treated cultures beginning on day 2 when compared to controls. However, the greatest effect was observed in cultures treated with the two more potent inhibitors (p38iB, p38iC). Characterization of ter119⁺ cells is depicted in Figure 9. Percentage of ter119⁺cells was initially very low in all cultures but progressively increased until day 11. However, at or after day 5, reduced percentages of ter119⁺ cells were noted in all p38i-treated cultures with the greatest effect observed in cultures treated with the two more potent inhibitors (p38iB, p38iC).

DISCUSSION

Previous in vivo and in vitro data have suggested the p38 MAP kinase-signaling pathway plays a pivotal role in erythropoiesis (Carotta et al. 2004; Dolznig et al. 2001; Foltz et al. 1997; Jacobs-Helber, Ryan, and Sawyer 2000; Nagata et al. 1997, 1998; Tamura et al. 2000; von Lindern et al. 2001). Various inhibitors of p38 MAPK have been shown to cause bone marrow hypocellularity and decreased reticulocyte and erythrocyte counts in rodents; however, the specific cellular target and pathophysiologic mechanism remains unclear (Zhang et al. 2002). In previous toxicologic studies from our laboratory (GSK unpublished data), mice given 300 mg/kg/day p38iB orally for one, four, or ten days had minimal (Figures 10A, 10B) to marked bone marrow hypocellularity and decreased red cell counts and hemoglobin and hematocrit values (up to 13% compared to control). Leukocyte and platelet counts were also decreased (up to 50%) in mice given 300 mg/kg/day compared to control. Mice given 100 mg/kg/day p38iB for ten days had similar reductions in red cell parameters; however, bone marrow hypocellularity was not observed, leukocyte counts were unchanged, and mean platelet counts were increased 22% compared to control. Similar findings were noted in rats given ten daily doses of 100 mg/kg p38iB, with mild to moderate bone marrow hypocellularity; decreases in mean femoral erythroid counts (67%), reticulocytes (95%), and peripheral blood cell counts (20%–41%); but with decreased leukocyte counts (97%) and platelet counts (66%–99%). In the rat, reversibility of bone marrow and peripheral blood findings was evident within approximately one week. In rats given p38iB for thirty days, bone marrow cellularity was normal and femoral erythroid and circulating reticulocyte counts were increased approximately twofold while mean red cell mass was decreased by only 25% to 29%, indicating some transiency or recovery in the response even with continued treatment. Bone marrow hypocellularity in the monkey was similar to changes observed in mice and rats. Monkeys given 30 mg/kg/day p38iB for ten days had mild to moderate bone marrow hypocellularity and decreased red cell count and hemoglobin and hematocrit values (up to 16% compared to predrug values). Leukocyte counts were slightly increased (36%–56% compared to predrug values) and platelets were unchanged. Similar hematologic findings were noted in toxicologic studies in rats with p38iA and p38iC.

This in vivo pattern of transient erythroid depression mimics that noted in this in vitro investigation and suggests that potential mechanisms can be translated preclinically. Despite the fact that we have noted hematologic effects with all p38 inhibitors we have tested toxicologically at relevant doses and that anecdotal evidence from other companies and institutions support these findings, it is still possible that these effects may reflect a structurally based or off-target effect of these compounds rather than a true pharmacologic effect of p38-signaling inhibition. However, unless and until a p38i reaches registration and lacks these effects, the hypothesis that the erythroid depression has a pharmacologic basis must be considered more likely.

Assessment of p38i-treated cultures from the present study indicates that neither cell growth nor cell viability was affected, suggesting that the mechanism of p38i action on erythroid progenitors does not involve impairment of cell proliferation or a direct cytotoxic effect by decreasing cell survival. Instead, the results suggest specific effects of p38 inhibition on erythroid commitment, resulting in delayed differentiation and maturation. In the erythroid lineage, the earliest committed erythroid progenitors identified ex vivo are BFU-Es (Gregory 1976). In control cultures, total BFU-E colony numbers decreased from day 0 to day 7, indicating differentiation. Lower numbers of BFU-E colonies were observed in all p38i-treated cultures by day 2 as compared to controls, suggesting a p38i-induced delay in erythroid lineage commitment. While BFU-E colony formation in cultures treated with the two less potent p38is recovered to control levels by day 5, cells exposed to the most potent p38i did not recover until day 7. These data paralleled changes in both gene expression and phenotypic analysis, indicating that p38is induce a transient delay in erythroid cell commitment following exposure to p38i, resulting in delayed differentiation and maturation. In controls, morphologic evidence of terminal erythroid differentiation was first observed on day 7, which correlated with upregulation of genes involved during intermediate (GATA-1, EKLF, EPO-R, FOG) and late (β-GLOBIN, GLYCOPHORIN A, and PROTEIN 4.2) stages of erythroid development (Figures 6A through 6D and 7A through 7C), increased ter119 protein expression (Figure 10), and downregulation of sca-1 gene (Figure 5A) and protein expression (Figure 8). In contrast, in cultures treated with the two most potent p38 inhibitors, cells were morphologically less differentiated as compared to controls, and sca-1⁺ phenotypes were retained as early as the second day of culture. Gene transcripts associated with erythroid commitment and ifferentiation/maturation were markedly altered. This indicates that p38 inhibition is affecting uncommitted sca-1⁺ bone marrow cells. The resulting delay in erythroid differentiation (especially with the most potent p38i) manifested in delayed maturation and terminal differentiation as demonstrated by retention of sca-1⁺ phenotype, decreased ter119 expression at the time of erythroid maturation, and alterations in expression patterns of early-, intermediate-, and late-stage transcripts involved in erythroid commitment, maturation, and terminal differentiation.

Hematopoietic stem cell maturation into erythrocytes is controlled by a balance between expression of lineage-specific transcription factors, structural proteins, and growth factor receptors (Baron and Farrington 1994; Bungert and Engel 1996; Labbaye et al. 1995; Prasad et al. 1995; Shivdasani and Orkin 1996; Tsang et al. 1997; Ziegler et al. 1999). In control cultures, expression of genes involved in early stages of erythroid development was highest at culture initiation and subsequently decreased with time. Expression patterns of SCL and GATA-2 in the mouse were consistent with previous reports using a two-phase liquid culture system where human bone marrow cells were expanded and demonstrated downregulation of early genes during erythroid differentiation (Pope et al. 2000; Scicchitano et al. 2003). In the presence of p38is, both magnitude and temporal sequence of expression of these genes were markedly affected. The initial decrease in expression of SCA-1 and SCL suggests the changes in gene expression are evidence of alterations in early erythropoietic dynamics. Because GATA-2 is known to be required for proliferation and survival of early hematopoietic progenitors (Cheng at al. 1996; Ikonomi et al. 2000; Kehrl 1995; Labbaye et al. 1995; Migliaccio and Migliaccio 1998; Murrell et al. 1995; Pope et al. 2000), its upregulation following p38i treatment provides further evidence that the early erythroid cell is targeted or at least affected by perturbation of p38 signaling.

In control cultures, intermediate genes regulating globin gene expression (GATA-1, EKLF, EPO-R, FOG, and FOG2) (Baron and Farrington 1994; Labbaye et al. 1995; Migliaccio and Migliaccio 1998; Moroni et al. 1997; Ni, Yang, and Stoeckert 1999; Orlic et al. 1995; Pope et al. 2000; Tsang et al. 1997; Wada et al. 1999) were expressed at relatively low levels during early stages of culture and peaked by day 5 for FOG2 and day 7 for GATA-1, EKLF, EPO-R, and FOG. These data matched expression patterns previously observed using human liquid culture systems (Pope et al. 2000; Scicchitano et al. 2003). The explanation for the sustained increase in FOG2 expression in cultures treated with the most potent inhibitor (p38iC) as compared to FOG2 expression in cultures treated with p38iA and p38iB was not identified in this study and therefore requires further investigation. Peak expression of the intermediate gene, NFE2, in control cultures in this study was observed at culture initiation, followed by a progressive decline through day 14. These results differ from other reports using expanded erythroid precursor cells where NFE2 expression was first noted on day 4, peaked on day 10, and then declined as the cultures matured (Pope et al. 2000) and a second study, with expanded human CD36+ erythroid progenitor cells where NFE2 expression was consistent throughout the culture period (Scicchitano et al. 2003). The reason for the discrepancy in NFE2 expression was not identified in this study, and therefore, further investigations comparing NFE2 expression in human and murine liquid culture are required. In cultures treated with p38 inhibitors, the intermediate-stage genes GATA-1, EKLF, EPO-R, and FOG were downregulated as cells progressed to terminal differentiation, with potency-dependent suppression of peak expression on day 7. However, p38i-treated cultures demonstrated rebound expression of GATA-1, EKLF, NFE2, and EPO-R at later time points, indicating that expression of these genes may be only delayed temporally rather than irreversibly inhibited and that intermediate acting genes are likely not the primary target of p38 inhibitors.

The late-stage transcripts β-GLOBIN, GLYCOPHORIN A, and PROTEIN 4.2 mRNA have been shown to be upregulated as cells undergo terminal erythroid differentiation (Migliaccio and Migliaccio 1998; Ni, Yang, and Stoeckert 1999; Pope et al. 2000; Scicchitano et al. 2003). Expression of late-stage genes was lower in p38i-treated cultures than in controls, especially with the two more potent inhibitors. However, since gene expression in controls peaked at day 7 and p38i treatment resulted in late-stage genes peaking by day 14 (gene expression analysis was not performed after day 14), results suggest transcription was only delayed in these late-stage erythropoietic genes rather than inhibited or directly downregulated. Thus, these results support the hypothesis that p38 inhibition is targeting erythroid commitment rather than differentiation and/or maturation of committed erythroid progenitor populations.

The cell surface antigens sca-1 and ter119 are linked to uncommitted hematopoietic cells and erythroid maturation, respectively (Kina et al. 2000; Okumura, Tsuji, and Nakahata 1992; Vaisman, Konijn, and Fibach 1999). In both control and treated cultures, expression of sca-1 was highest at culture initiation and progressively decreased as differentiation progressed, similar to patterns previously reported (Heimfeld et al. 1991). However, higher percentages of sca-1⁺ cells were detected in p38i-treated cultures at later time points, suggesting suppression of erythroid differentiation. Expression of ter119 in control cultures was lowest at culture initiation and subsequently increased as cultures proceeded. The p38i treatment transiently decreased expression of ter119, indicating a delay in the commitment process.

Most p38 homozygous knockout mice die during embryonic development, and surviving mice are anemic as a consequence of failed erythropoiesis (Tamura et al. 2000). In contrast to results from knockout experiments, results from this study suggest that p38 inhibition delays, but does not irreversibly inhibit, erythropoiesis. This confirms results from toxicologic studies in which anemias, associated with p38i, are mild and reversible (Zhang et al. 2002). The delay in maturation and terminal differentiation associated with p38 inhibition should not be considered evidence for targeting of intermediate or late stages of erythroid development. The sca-1⁺bone marrow cells appear to be directly affected, and later stages (metarubricytes, nRBCs, reticulocytes, etc.) seem to develop normally, but differentiation and maturation were delayed as compared to controls. This is supported by the pattern of gene changes that were noted, which at the intermediate and late stages appeared delayed temporally rather than inhibited completely. Although SB-203580 has the potential to produce off-target kinase activity at concentrations >10μM (Fabian et al. 2005; M. R. Lee and Dominguez 2005), the effects observed on erythropoiesis are considered to be related to inhibition of p38 MAPK, given results from previous experiments (Scicchitano et al. 2008; Zhang et al. 2002) and the fact that we have demonstrated similar and specific effects on erythroid cells using three chemically distinct p38is of varying potencies: SB-203580 (50 μM), SB-226883 (1 μM), and SB-267030 (0.1 μM).

The method utilized here for murine erythroid culture and assessment has not been previously reported. The concentrations of p38is utilized were normalized to counter effects due to differences in potency, but these adjustments made to normalize doses could not exactly equilibrate doses, and hence some potency-dependent differences were still observed. The use of a short-term liquid bone marrow culture system allowed multiparameter analysis to be performed on erythroid progenitors. Since sca-1⁺ cells only represent ~6% of murine bone marrow progenitors, purification was needed to expand and differentiate sca-1⁺ cells along the erythroid lineage. The eventual highly purified culture system retained phenotypic and molecular features of erythropoietic cells and provided a population of erythroid progenitor cells of sufficient quality and number for multiparameter genomic and proteomic analysis. Beginning with uncommitted sca-1⁺ cells, the murine liquid culture bone marrow model partially recapitulates molecular and cellular aspects of in vivo erythropoiesis and appears to be a useful tool for investigating effects of drug-induced or disease-related erythropoietic abnormalities in the mouse. By utilizing this culture system, the data have shown that p38is SB-203580, SB-226882, and SB-267030 induce a potency-dependent, transient delay of murine erythropoiesis via effects on commitment and differentiation of sca-1⁺ bone marrow cells without affecting proliferation or cell viability, resulting in delayed maturation and altered expression patterns of genes involved in the erythropoietic process.