Targeted Disruption of the Murine fps/fes Proto-Oncogene Reveals that Fps/Fes Kinase Activity Is Dispensable for Hematopoiesis

Abstract

The fps/fes proto-oncogene encodes a cytoplasmic protein-tyrosine kinase that is functionally implicated in the survival and terminal differentiation of myeloid progenitors and in signaling from several members of the cytokine receptor superfamily. To gain further insight into the physiological function of fps/fes, we targeted the mouse locus with a kinase-inactivating missense mutation. Mutant Fps/Fes protein was expressed at normal levels in these mice, but it lacked detectable kinase activity. Homozygous mutant animals were viable and fertile, and they showed no obvious defects. Flow cytometry analysis of bone marrow showed no statistically significant differences in the levels of myeloid, erythroid, or B-cell precursors. Subtle abnormalities observed in mutant mice included slightly elevated total leukocyte counts and splenomegaly. In bone marrow hematopoietic progenitor cell colony-forming assays, mutant mice gave slightly elevated numbers and variable sizes of CFU-granulocyte macrophage in response to interleukin-3 (IL-3) and granulocyte-macrophage colony-stimulating factor (GM-CSF). Tyrosine phosphorylation of Stat3 and Stat5A in bone marrow-derived macrophages was dramatically reduced in response to GM-CSF but not to IL-3 or IL-6. This suggests a distinct nonredundant role for Fps/Fes in signaling from the GM-CSF receptor that does not extend to the closely related IL-3 receptor. Lipopolysaccharide-induced Erk1/2 activation was also reduced in mutant macrophages. These subtle molecular phenotypes suggest a possible nonredundant role for Fps/Fes in myelopoiesis and immune responses.

The fps/fes proto-oncogene (hereafter referred to simply as fps) encodes a nonreceptor protein-tyrosine kinase (PTK) (2, 55, 62) which is abundantly expressed in cells of the myeloid lineage (13, 15, 21, 41), where it has been proposed to play an essential role in regulating survival and terminal differentiation (14, 68). Transfection of the K562 erythroleukemia cell line with fps potentiated phorbol ester-induced terminal myeloid differentiation (68), while fps antisense oligonucleotides blocked retinoic acid-induced granulocytic differentiation and led to apoptosis in acute promyelocytic leukemia cells (14). Although fps displays a limited tissue-specific expression pattern, the biological function of Fps cannot be restricted to a role in myelopoiesis, as it is also expressed in several other diverse cell lineages, including endothelial, epithelial, and neuronal cells (5, 22). In contrast, the closely related Fer kinase displays a widespread expression pattern (16, 23, 39). It is possible that, as the only two known members of this distinct subclass of PTKs, Fps and Fer perform redundant biochemical functions.

In addition to their carboxyl-terminal catalytic domains, Fps and Fer also contain central SH2 domains and long amino-terminal domains that include three putative coiled-coil motifs. The amino-terminal domains mediate homotypic oligomerization of Fps (52) and Fer (8, 35); however, heterotypic oligomers are not formed between Fps and Fer, and homotypic oligomerization is not required for Fer kinase activation (8). The SH2 domain may regulate Fps activity through intramolecular interactions (28, 36) or through intermolecular interactions with other tyrosine-phosphorylated proteins, including putative substrates (33). A phosphopeptide library screen using the Fps SH2 domain as an affinity matrix has identified a consensus-binding sequence (pYExV/I) which is present in several potential targets, including a number of other protein kinases, tyrosine phosphatases, cell surface antigens, Bcr, and γ-adaptin (58).

Oncogenic fps alleles were frequently isolated from avian (v-fps) and feline (v-fes) retroviruses. These encode Gag-Fps fusions proteins with unregulated kinase activities which can abrogate cytokine requirements and influence differentiation of hematopoietic progenitor cells (6, 34, 35) and reduce or eliminate growth factor requirements in transformed fibroblasts (56). Interestingly, Fps activation in fibroblasts correlated with down-regulation of the platelet derived growth factor β-chain (PDGFβ) receptor (3). Expression of Gag-Fps proteins in transgenic mice under the control of a minimal β-globin promoter caused tumors in lymphoid and mesenchymal tissues (66) and hypertrophic effects in the heart and trigeminal nerves (67). In contrast, tissue-specific overexpression of wild-type cellular fps in transgenic mice caused no overt phenotype (21), while mice expressing low levels of an activated fps allele developed vascular hyperplasia progressing to multifocal hemangiomas but exhibited no apparent hematopoietic defects or malignancies (20).

Increased tyrosine phosphorylation has been described for a number of cellular proteins in v-fps-transformed fibroblast cells, including Ras-specific GTPase-activating protein (Ras-GAP) and the associated Rho-GAP and Dok proteins (12, 36, 46), Bcr (42), Shc (44), phosphatidylinositol 3′-kinase (17), the SH2-containing signal transducers and activators of transcription [Stat]) protein Stat3 (18), the PDGFβ receptor (3), and connexin 43 (37). Putative substrates for the cellular Fps kinase include Bcr and Ras-GAP (27, 42), Stat3 (48), and Cas and Fyb (9, 33).

Potential molecular roles for cellular Fps have largely focused on signaling downstream from cytokine receptor superfamily members, including those for granulocyte-macrophage colony-stimulating factor (GM-CSF), interleukin 3 (IL-3), IL-4, IL-6, and IL-11, erythropoietin (Epo) oncostatin M, leukemia inhibitory factor, and ciliary neurotrophic factor. Fps or a closely related kinase becomes tyrosine phosphorylated upon stimulation of responsive cells with IL-3 (24), GM-CSF (24, 40), IL-4 (32), IL-6 (43), and Epo (25). Fps has also been detected in association with a number of cytokine receptors, including those for IL-4 (32), the shared β-chain of the human IL-3, IL-5, and GM-CSF receptors (4, 24, 51), as well as the common gp130 chain which is used by the receptors for IL-6, IL-11, leukemia inhibitory factor, oncostatin M, and ciliary neurotropic factor (43). Interestingly, overexpression of catalytically inactive Fps in cytokine-dependent TF-1 or 32D cells did not interfere with proliferative responses to GM-CSF and IL-3 or differentiation in response to G-CSF (64).

Despite evidence of cytokine-induced activation of Fps and association with cytokine receptors, its involvement in cytokine receptor signaling remains unclear. This is in contrast to more established roles for nonreceptor PTKs of the Jak (Janus), Syk, and Src families (7, 30, 38). To further explore the in vivo function of the Fps kinase and clarify its role in cytokine signaling and myelopoiesis, we have targeted the endogenous mouse fps locus with a kinase-inactivating missense mutation. We show that mice expressing only inactive Fps display normal levels of hematopoietic cell types in the periphery and bone marrow (BM); this demonstrates that Fps activity is not required for normal hematopoiesis. BM from these mice contain normal numbers of hematopoietic progenitors of the myeloid, erythroid, and lymphoid lineages, and BM-derived myeloid progenitor cells display normal colony-forming responses to a number of cytokines, including IL-3 and GM-CSF. Therefore, either Fps kinase activity is not involved in the cellular response to these cytokines, or the biochemical function that it provides is redundant. Interestingly, BM macrophages (BMM) from mutant mice displayed dramatically reduced tyrosine phosphorylation of Stat3 and Stat5A in response to GM-CSF but not IL-3. We also noticed a dose-dependent reduction in lipopolysaccharide (LPS)-induced tyrosine phosphorylation of Erk1 and Erk2 in BMMs, suggesting that Fps either plays a direct role in signaling downstream from the LPS receptor (CD14) or possibly has an indirect effect resulting from autocrine signaling caused by LPS-induced cytokine release. These subtle molecular phenotypes suggest a potential nonredundant role of Fps kinase activity in myeloid cell functions.

MATERIALS AND METHODS

Construction of the fps gene targeting vector.

The complete murine fps locus has been cloned and sequenced (unpublished data). PCR mutagenesis using Pfu thermal stable DNA polymerase (Stratagene) was used to convert the AAG codon for Fps residue lysine 588 to an AGA arginine codon. The template was a 2.5-kbp genomic XbaI fragment containing exon 14 cloned in pGEM-1 (Promega). PCRs were performed with a sense-strand mutagenic oligonucleotide (5′-GTG GCT GTG AGA TCT TGC CGA-3′) in combination with the T7 primer or an antisense mutagenic oligo (5′-TCG GCA AGA TCT CAC AGC CAC-3′) in combination with the reverse primer. Aliquots of these two purified PCR products were combined in a second PCR using only the external T7 and reverse primers. The resulting PCR product was digested with XbaI and subcloned in place of the corresponding wild-type XbaI fragment in the context of a plasmid containing the complete murine fps locus (pXNK4), giving the mutant version (pXNR24). The mutation was confirmed by DNA sequencing.

The targeting construct was produced in the context of a modified version of pPNT (61), called pPNT-NHS14, in which the cloning site upstream of the phosphoglycerine kinase (PGK)-neomycin phosphotransferase (neo) cassette was modified by digestion with XhoI and insertion of a linker containing SalI, HpaI, and NheI sites (unpublished data). A 450-bp EcoRI fragment located downstream of the last fps exon was first cloned into the EcoRI site of pPNT-NHS14 between the PGK-neo and PGK-tk (thymidine kinase) cassettes, giving pPNT450HNS14.2. The 9.0-kbp EcoRI fragment from pXNR24 was subcloned into pBluescriptIIKS−, recovered as a NotI-to-SalI fragment, and cloned between the corresponding sites in pPNT450NHS14.2, giving the final targeting vector.

ES cell culture and chimeric mouse production.

Mouse embryonic stem (ES) cells (R1; passage 8) were kindly provided by Andras Nagy (47). Propagation, electroporation, and selection of recombinant R1 clones were carried out essentially as described by Wurst and Joyner (63) except that the concentration of 2-mercaptoethanol in culture medium was 100 μM, and recombinant leukemia inhibitory factor was prepared in bacteria by using plasmid pGEX-2T-MLIF, kindly provided by L. Grey and J. Heath (54). R1 cells were expanded on embryonic fibroblast feeder layers; after trypsinization and preplating to remove embryonic fibroblasts, they were electroporated with the NotI-linearized targeting vector, using a Gene Pulser (Bio-Rad). Cells were then plated on gelatinized tissue culture plates in the absence of embryonic fibroblasts and selected by using 200 μg of active Geneticin (G418; GibcoBRL) per ml and 2 μM ganciclovir (Syntex Inc.). Drug-resistant clones were picked and replated onto gelatinized 24-well plates. After 2 days in culture, the clones were trypsinized, and 10% was taken for pooled PCR analysis, 80% was plated onto embryonic fibroblast feeders on 24-well plates, and the remaining 10% was maintained on gelatinized 24-well plates for subsequent PCR analysis of individual clones. PCR analysis (63) was carried out with a sense primer (p2/neo21; 5′-CCGCTTCCTCGTGCTTTACGG-3′), corresponding to sequences in the 3′ region of the PGK-neo cassette, and an antisense primer (p1/mfes1.6RC#2; 5′-GACAGGGTTTCCTGTCATGTG-3′), corresponding to genomic sequences immediately downstream from the 450-bp EcoRI fragment used as the short arm of homology. Southern blot analysis was carried out with the cloned 450-bp EcoRI fragment as a probe to confirm the identity of ethidium bromide-stained PCR products. Individual clones from positive pools were analyzed by the method described above. These clones were expanded on embryonic fibroblasts, and chimeric mice were produced by the “darning needle” method (47). Chimeric males were bred with albino CD1 females, and the fps genotype of agouti pups was determined by PCR or genomic Southern blotting analysis. Routine analysis of genotypes was carried out with total DNA from tail biopsy as templates in PCRs with an exon 13 sense primer (p3; 5′-GACAAGTGGGTTCTGAAGCACGAGG-3′) and an exon 15 antisense primer (p4; 5′-GACCCCGATGAGACGCACAATGTTGG-3′). The PCR product was subsequently digested with BglII and resolved on 1.2% agarose gels. Ethidium bromide staining revealed a single 1,028-bp PCR product from wild-type animals, 628- and 400-bp fragments from homozygous mutants, and equal molar amounts of all three fragments from heterozygous mice. Alternatively, genotypes were determined by Southern blotting analysis of BglII-digested DNA probed with a complete murine fps cDNA provided by Andrew Wilks (62).

Immune-complex kinase assays and immunoblotting analysis.

BM was recovered from dissected femurs as previously described (60). Cos-1 cells were transfected with Fps expression plasmids by using Lipofectamine reagent as instructed by the manufacturer (Life Technologies, Inc.). Cells were lysed into 0.7 ml of KLB (20 mM Tris-HCl [pH 7.5], 150 mM NaCl, 1 mM EDTA, 1% [vol/vol] Nonidet P-40, 0.5% [vol/vol] sodium deoxycholic acid, 10 μg of aprotinin per ml, 10 μg of leupeptin per ml, 100 μM sodium orthovanadate, 100 μM phenylmethylsulfonyl fluoride). Cell lysates were clarified by centrifugation at 14,000 × g for 20 min at 4°C. Aliquots of 0.1 ml were taken for immunoblotting analysis of soluble Fps proteins. The remaining 0.6 ml of lysate was added to 20 μl of 30% (vol/vol) protein A conjugated to Sepharose CL-4B and 5 μl of crude polyclonal rabbit antiserum (anti-Fps/Fer, also known as anti-FpsQE, or anti-FerLA [22]) or 2 μg of an affinity-purified anti-Fps antibody, which was raised against a TrpE fusion with human Fps amino acids L401 to Q446 and affinity purified against a glutathione S-transferase fusion of murine Fps amino acids Q381 to E563. After mixing on a nutator platform for 2 h at 4°C, immune complexes were collected by brief centrifugation and then washed five times with KLB and once with KRB (20 mM Tris-HCl [pH 7.5], 10 mM MnCl₂, 100 μM sodium orthovanadate). Kinase reactions were performed by resuspending the washed immune complex with 30 μl of KRB supplemented with 10 μCi of [γ-³²P]ATP and incubating the mixture for 20 min at 30°C. Reactions were terminated by addition of 30 μl of 2× sodium dodecyl sulfate (SDS) sample buffer and heating at 100°C for 5 min. Proteins were resolved on SDS-polyacrylamide gels and either dried and subjected to radioautography for detection of kinase activity or transferred to Immobilon-P membrane (Millipore) by using a semidry apparatus (Bio-Rad) for immunoblotting analysis. Membranes were blocked overnight at 4°C with BLOTTO (5% Carnation skim milk powder in 10 mM Tris-HCl [pH 7.5]–150 mM NaCl). Fps proteins were detected by incubation at room temperature (RT) for 2 h with 1/500 dilutions of rabbit polyclonal anti-FpsQE antibody or affinity-purified anti-Fps antibody. After washing with TBST (10 mM Tris-HCl [pH 7.5], 150 mM NaCl, 0.05% [vol/vol] Tween 20), membranes were incubated with a 1/10,000 dilution of horseradish peroxidase-conjugated goat anti-rabbit immunoglobulin G (Vector Laboratories) in TBST for 1 h at RT. After a wash with TBST, immune complexes were detected by using enhanced chemiluminescence (ECL) reagents (NEN Life Science Products).

Peripheral blood cell counts.

Mice between 8 and 20 weeks of age were deeply anesthetized with chloroform, their chest cavities were opened, and peripheral blood (PB) was collected by cardiac puncture, using a 1-ml syringe fitted with a 23-gauge needle. Following PB collection, the blood was quickly mixed with 10% (wt/vol) EDTA disodium salt to a final concentration of 1.5 ± 0.25 mg of EDTA/ml of blood to inhibit clotting. Samples were collected at the same time each day to avoid any differences in cell counts due to the circadian rhythm of the mice (1). Leukocyte (WBC) counts, erythrocyte counts, hemoglobin, hematocrit, mean corpuscular volume, mean corpuscular hemoglobin, mean corpuscular hemoglobin concentration, and platelet counts were measured in a Baker System 9000 Hematology Series Cell Counter (Baker Instruments Corporation, Allentown, Pa.). Neutrophil, lymphocyte, monocyte, eosinophil, and basophil levels were quantified with a CELL-DYN 3500 multiparameter hematology analyzer (Abbott Diagnostics, Santa Clara, Calif.).

Isolation of BM for colony assays.

Mice between 8 and 16 weeks of age were euthanized by cervical dislocation, and femurs were dissected free of muscle and connective tissue by aseptic technique. Posterior ends of the femurs were cut open to expose the BM, which was flushed into 5 ml of ice-cold macrophage starvation media consisting of Dulbecco modified Eagle medium (GibcoBRL Life Technologies Inc., Grand Island, N.Y.), 0.5% fetal calf serum FCS; HyClone, Logan, Utah), 50 μM α-monothioglycerol (Sigma), and 1% antibiotic-antimycotic (GibcoBRL Life Technologies), using a 5-ml syringe fitted with a 23-gauge needle. Large clumps of marrow were disrupted by gently passing the freshly flushed BM through the needle several times. Cell counts were performed in a Coulter ZBI cell counter (Coulter Electronics Inc., Hialeah, Fla.).

Hematopoietic progenitor cell colony-forming assay.

Methylcellulose clonal cultures were established to assess the levels of hematopoietic progenitor cells in the BM of mice. Briefly, 50,000 unfractionated BM cells were cultured in 1.2 ml of semisolid medium consisting of 1% methylcellulose (Fluka), 10% fetal bovine serum (FBS), 2% bovine serum albumin (BSA), 200 μg of human plasma transferrin (iron saturated; (Calbiochem) per ml, 50 μM α-monothioglycerol (Sigma), and either (i) a cytokine cocktail consisting of recombinant human Epo (1 U/ml; R&D Systems), recombinant murine IL-3 (5 ng/ml; (R&D Systems), recombinant murine Steel factor (SF) (50 ng/ml; R&D Systems), and recombinant murine IL-6 (10 ng/ml; R&D Systems); (ii) recombinant murine GM-CSF (5 ng/ml; R&D Systems) and recombinant human Epo (1 U/ml); or (iii) recombinant human Epo (1 U/ml) alone. All BM samples were cultured in duplicate at 37°C and 5% CO₂ in a humidified atmosphere. Colony counts were made by placing each 35-mm-diameter plate within a gridded 60-mm-diameter tissue culture plate, and the entire area was scored with a Leitz Labovert inverted microscope (Leitz Wetzlar, Wetzlar, Germany). Plates were scored as follows: CFU-erythroid (CFU-E) colonies consisting of one or two clusters of 4 to 32 hemoglobinized erythroblasts per cluster were counted 2 days postplating; burst-forming unit-erythroid (BFU-E) colonies consisting of three or more clusters of hemoglobinized erythroblasts, CFU-macrophage (CFU-M) colonies consisting of 50 or more cells, CFU–granulocyte-macrophage (CFU-GM) colonies consisting of 50 or more cells, and CFU–granulocyte-erythroid-macrophage-megakaryocyte (CFU-GEMM) colonies consisting of 50 or more erythroid, granulopoietic, and thrombopoietic cells combined were counted 8 days postplating.

The number of cells per CFU-GM colony were counted by randomly picking 20 colonies from a 1.2-ml methycellulose culture seeded with 50,000 nucleated BM cells 9 days postplating. Colonies were picked by using a P-200 pipetteman and combined in 1 ml of Iscove modified Dulbecco medium–(IMDM)–2% FBS. Samples were stained with ethidium bromide-acridine orange stain and counted with a hemacytometer on a light microscope under UV light.

BMM cultures.

The procedure used for culturing BMM was for the most part the same as that outlined by Tushinski et al. (60), with a few minor modifications. BM cells flushed from femurs of mice were seeded into tissue culture plates (Sarstedt) in complete macrophage medium consisting of IMDM, 20% L-cell conditioned medium, 15% FCS, 20 mM glutamine, 1% antibiotic-antimycotic, and 50 μM α-monothioglycerol at a concentration of 10⁶ cells/ml and a density of 2.9 × 10⁵ cells/cm². Following a 24-h incubation, nonadherent cells were collected, pelleted at 700 × g for 5 min at RT, resuspended in half the original volume of complete macrophage medium, and seeded into fresh tissue culture plates. Adherent cells were discarded. Following a 2-day incubation, the adherent cells were discarded, the nonadherent cells were collected and plated in 35-mm-diameter tissue culture plates at a density of 1.9 × 10⁴ cells/cm² and a concentration of 2 × 10⁵ cells/ml. Following a 3- to 6-day incubation, the medium was replaced with fresh complete macrophage medium and the adherent macrophages were allowed to grow to 80 to 90% confluency before use in any experiments.

Cytokine stimulation of BMM cultures.

BMM cultures were incubated in macrophage starvation medium for 24 h, which was then replaced with fresh macrophage starvation medium for an additional 24 h prior to exposure of the cells to any cytokines. On the day of the experiment, macrophages were rinsed once with prewarmed phosphate-buffered saline (PBS) and then incubated in IMDM for 2 h. After a rinse with prewarmed IMDM, the cells were incubated with either 30 ng of recombinant murine GM-CSF, IL-3, or IL-6 (all from R&D Systems) per ml diluted in prewarmed IMDM at 37°C for 15 min. In the case of stimulation with LPS from Escherichia coli serotype O55:B5 (Sigma), cells were incubated with either 10, 100, or 1,000 ng of LPS per ml in prewarmed IMDM at 37°C for 30 min. Plates were put on ice immediately following exposure to the various treatments, the medium was quickly aspirated, and plates were rinsed with ice-cold PBS containing 100 μM sodium orthovanadate. Cell lysates from 35-mm-diameter tissue culture plates were scraped into 200 μl of 2× SDS protein sample buffer by using a rubber policeman, passed through a P-200 pipette tip several times to shear high-molecular-weight DNA, heated to 100°C for 10 min, centrifuged briefly at 14,000 × g, and then either run out on an SDS–7.5 or 11% polyacrylamide protein gel or stored at −20°C. Proteins were transferred by semidry blotting to Immobilon-P membrane, blocked with either 5% BSA TBST, 5% milk powder in TBST, or 3% milk powder in PBS and probed with the following antibodies: rabbit anti-rat Erk (clone K-23; Santa Cruz), mouse anti-human phospho-ERK (pErk (clone E-4; Santa Cruz), rabbit anti-mouse Stat3 (New England Biolabs), mouse anti-human pStat3 (clone 9E12; Upstate Biotechnology), rabbit anti-human Stat5A (Upstate Biotechnology), mouse anti-human pStat5A/B (clone 8-5 2; Upstate Biotechnology), rabbit anti-mouse Jak2 (Upstate Biotechnology), and rabbit anti-pJak2 (QCB Biosciences International). The secondary antibody used to detect the proteins was either peroxidase-conjugated goat anti-rabbit (Boehringer Mannheim) or peroxidase-conjugated sheep anti-mouse (Amersham Life Sciences), depending on the primary antibody used. Membranes were then exposed to ECL reagents (NEN Life Science Products) and then to film.

Flow cytometry.

Single-cell suspensions of BM cells were isolated from mouse femurs as described above; BM was flushed into PBS (pH 7.4) containing EDTA (1.5 mg/ml) at RT. Following extraction, BM cells were centrifuged at 300 × g for 5 min at RT and resuspended at a concentration of 20 × 10⁶ cells/ml in PBS–0.5% BSA–0.1% sodium azide (pH 7.2) (PAB). For staining of lineage-specific surface antigens, 2 × 10⁶ cells in 100 μl of PAB were initially incubated with rat anti-mouse CD16/CD32 (PharMingen Canada, Mississauga, Ontario, Canada) for 5 min at 4°C, to block Fc II and Fc III receptors. Cells were then incubated for 15 min at 4°C with the following monoclonal antibodies (MAbs) diluted in 10 μl of PAB: phycoerythrin (PE)-conjugated rat anti-mouse Ly-6G, fluorescein isothiocyanate (FITC)-conjugated rat anti-mouse CD11b, PE-conjugated rat anti-mouse TER-119, FITC-conjugated rat anti-mouse CD44, and PE-conjugated rat anti-mouse CD45R/B220 (all purchased from PharMingen Canada).

Erythrocytes were lysed by adding 2 ml of ice-cold lysis buffer (154 mM ammonium chloride, 10 mM potassium bicarbonate, 100 μM EDTA disodium salt) and incubated for 5 min at 4°C with constant motion. Samples were then centrifuged at 300 × g for 5 min at RT, washed once with 4 ml of PAB, and resuspended in 1 ml of 1% paraformaldehyde in cacodylate buffer (pH 7.2) for analysis on the flow cytometer.

Statistical analysis.

The means and standard deviations (SD) of all numeric data were calculated. Data were analyzed by Student’s t test, where appropriate. Comparisons of data sets yielding P values of greater than 0.05 were regarded as not statistically different.

RESULTS

Targeting of the murine fps gene.

To investigate the in vivo function of the fps proto-oncogene, gene targeting was used to generate transgenic mice expressing catalytically inactive Fps from the endogenous loci. The targeting vector (Fig. 1) consisted of (i) the Neo positive selectable marker flanked by a 450-bp short arm of homology corresponding to sequences in the 3′ flanking region of the fps locus and (ii) a 9-kbp long arm of homology containing exons 4 through 19 of fps. The herpes simplex virus thymidine kinase gene was included downstream of the short arm of homology to provide negative selection for nonhomologous recombination events. A dinucleotide mutation was introduced into exon 14, which is located midway through the long arm of homology. This mutation generated a novel BglII site and converted the lysine 588 codon to an arginine codon. Lysine 588 corresponds to a completely conserved residue in the protein kinase family which is located in kinase subdomain II and is known to be essential for catalytic function. We also engineered a cDNA version of this mutation in a human fps expression plasmid and showed that the encoded kinase lacked detectable catalytic activity in transfected Cos-1 cells (see Fig. 3A).

FIG. 1.

Structure of the murine fps locus and targeting strategy. The top line illustrates a restriction map of the murine fps locus; the 19 exons are indicated by black rectangles, and the promoter is indicated with a horizontal arrow. The targeting vector is shown on the middle line, with the two arms of homology from the 9- and 0.45-kbp EcoRI fragments shown as thick lines on either side of the PGK-neo cassette. The position of the mutation and novel BglII site in exon 14 is indicated with a vertical arrow. The PGK-tk cassette and flanking vector sequences are located to the right of the short arm of homology. The intersecting dotted lines indicate the presumptive regions of homologous recombination leading to the targeted locus shown on the bottom line. The positions of primers used for PCR screening (p1,p2) and analysis of genotypes (p3,p4) are indicated with arrowheads. The lengths of BglII fragments are indicated below the targeted locus. Restriction sites for BglII, EcoRI, and XbaI are indicated as Bg, R, and X, respectively. The polymorphic BglII site located in the 5′ flanking region is indicated in parentheses.

FIG. 3.

Analysis of Fps expression and kinase activity. (A) Lysates from Cos-1 cells (control) or Cos-1 cells transfected with cDNA expression plasmids encoding either wild type (Fps) or mutant (FpsK588R) proteins or empty vector (control) were subjected to immune-complex kinase assays (top) and immunoblotting (bottom). (B) Similar analysis of BM from wild-type (wt) animals or mice which were heterozygous (het) or homozygous (hom) for the fps^K588R allele. (C) Immune-complex kinase assay of bone marrow from animals of the three genotypes, using an antibody either specific for the related Fer kinase (α-Fer) or cross-reactive for Fps and Fer (α-Fps/Fer). The positions of p94 Fer and p92 Fps are indicated.

The linearized targeting vector was electroporated into the R1 murine ES cell line (47), and combined G418-gancyclovir selection was applied. A total of 647 clones were screened by PCR across the short arm of homology, using primers p1 and p2 (Fig. 1). We identified 14 PCR-positive clones, which corresponded to a targeting frequency of approximately 2%. Homologous recombination could have occurred either upstream or downstream from the engineered mutation in the long arm of homology. We therefore performed a genomic Southern blotting analysis of BglII-digested ES cell DNA and determined that 7 of 14 targeted ES cell clones had incorporated the novel BglII site (data not shown). Three of these cell lines were used to produce chimeric mice by using the darning needle aggregation method (47), and germ line transmission of the fps^K588R allele was established. No obvious phenotype was apparent in heterozygous mutant mice, and when these were interbred, wild type, heterozygous, and homozygous mutant mice were produced in the expected Mendelian ratios (Table 1). Furthermore, homozygous mutant males and females were both fertile. Genotypes were routinely determined by PCR amplification of an exon 14-containing genomic sequence, using primers p3 and p4 (Fig. 1), followed by digestion with BglII (Fig. 2B). Alternatively, genotypes were determined by Southern blotting analysis of total genomic DNA digested with BglII and probed with the complete fps cDNA (Fig. 2A). A 6.8-kbp fragment encompassing most of the gene is reduced to 4.8- and 2.0-kbp fragments in the targeted allele. However, this analysis was complicated by the existence of an additional BglII polymorphism in the 5′ flanking region of the fps locus. The BglII site shown in parentheses in Fig. 1 is present in the 129/Sv but not the 129/SvJ genetic background. Therefore, depending on which alleles are present, the 5′ end of the locus appears on BglII fragments of either 2.1 kbp or approximately 10 kbp. This polymorphism was confirmed by probing BglII-digested DNA from 129/Sv and 129/SvJ mice with the cloned 2.1-kbp BglII fragment (data not shown). The R1 ES cell line used in these experiments was derived from (129/Sv × 129/SvJ)F₁ hybrid embryos, while the genomic library used to clone fps was prepared from 129/Sv DNA. Interestingly, although the targeting vector was produced from the cloned 129/Sv gene, which has this 5′ BglII site, the 129/SvJ allele was targeted in the ES cell clone used to establish these mice.

TABLE 1.

Genotypic analysis of offspring from heterozygous × heterozygous breeding pairs

Gender	No. (%) of indicated genotypea
	Wild type	Heterozygous	Homozygous
Female	55 (24)	132 (57)	43 (19)
Male	57 (22)	140 (55)	58 (23)
Combined	112 (23)	272 (56)	101 (21)

^aCumulative data collected from 20 separate heterozygous × heterozygous breeding pairs active over continuous periods of 3 to 6 months. 

FIG. 2.

Genotype analysis by Southern blotting and PCR. (A) Southern blot analysis of BglII-digested genomic DNA from animals that are wild type (wt) or either heterozygous (het) or homozygous (hom) for the mutant fps^K588R allele. Hybridization with a complete cDNA probe detected BglII fragments of 2.4 and 1.6 kbp in all samples, while the 6.8-kbp fragment in the wild-type allele is reduced to 4.8- and 2.0-kbp BglII fragments in the mutant allele. Two additional fragments of 2.1 or approximately 10 kbp are diagnostic of a polymorphic BglII site in the fps 5′ flanking region (Fig. 1). (B) PCR products from mouse genomic DNA using primers p3 and p4, which flank the mutation point (Fig. 1). The 1,028-bp PCR product is seen in all samples before BglII digestion (−), and this is partially or completely reduced by BglII digestion (+) to 628 and 400 bp in samples which are heterozygous or homozygous for the mutant allele, respectively.

Mice homozygous for the fps^K588R allele express inactive Fps.

We next examined Fps expression and kinase activity in BM, which is the major myeloid compartment. Fps was immunoprecipitated from isolated BM cells, and immune-complex kinase assays were performed to assess the intrinsic autophosphorylation activity. Although animals from all three genotypes expressed equal amounts of Fps protein, kinase activity was not detected in the homozygous mutant animals, and reduced activity was clearly apparent in the heterozygous mutant animals (Fig. 3B). We next examined BM for expression of the Fps-related kinase, Fer. Immune-complex kinase assays were repeated with an antibody which is cross-reactive for Fps and Fer in parallel with an antibody specific for Fer. This analysis showed that the slower-migrating Fer kinase was present in all three genotypes, while the faster-migrating Fps kinase displayed reduced activity relative to Fer in the heterozygous mutant animals and was completely inactive in the homozygous mutant animals (Fig. 3C). In several independent experiments of this type, we were unable to find evidence for changes in Fer expression or kinase activity in BM from the mutant animals.

Viability and fertility of mice expressing catalytically inactive Fps.

Of the 485 pups born to 20 different heterozygous × heterozygous breeding pairs, 112 (23%) were wild type, 272 (56%) were heterozygous for the fps^K588R allele, and 101 (21%) were homozygous for the mutant fps^K588R allele (Table 1). This clearly demonstrated that Fps kinase activity was not required for development or maturation. Furthermore, similar percentages of the three possible genotypes were seen in male and female offspring, indicating that the absence of Fps activity did not compromise the development of either gender. These results suggest that catalytically active Fps is not required for normal mouse development.

Mice heterozygous or homozygous for the fps^K588R allele appeared healthy, developed normally and did not display any impairment of reproductive capacity and neonatal survival. Both heterozygous × heterozygous and homozygous × homozygous breeding pairs produced litters similar in size to those produced by wild type × wild type breeding pairs (Table 2), demonstrating that catalytically inactive Fps does not hinder the fertility of male or female mice.

TABLE 2.

Litter sizes

Genotype	No. of breeding pairs	Total no. of litters born	No. of pups born/litter (mean ± SD)
Wild type × wild type	5	10	7 ± 3
Heterozygous × heterozygous	20	61	8 ± 3
Homozygous × homozygous	7	20	8 ± 3

Hematologic analysis of PB.

PB was collected from mice of all three genotypes and analyzed for hematologic abnormalities. Both male and female mice between 8 and 20 weeks of age were included in this study. Mice were excluded if they had either previously been set up in a breeding pair or were clearly injured due to fighting with other mice, both of which may have skewed the data due to physiologic changes associated with pregnancy or trauma. Despite high levels of Fps expression in cells of the myeloid/monocytic lineage (13, 15, 21, 41) and in vitro evidence suggesting that Fps plays a role in myeloid cell differentiation (14, 65, 68), no statistically significant differences were found between wild-type and homozygous mutant mice in the levels of any of the hematopoietic cell types quantified in the peripheral circulation of these mice (Table 3). All 14 of the hematologic parameters measured were within published normal ranges for laboratory mice of various strains (57). Substantial intermouse variability was observed in total WBC and levels of neutrophils, lymphocytes, monocytes, eosinophils, and basophils in mice of all three genotypes, which may have masked any subtle differences between the genotypes.

TABLE 3.

Peripheral blood analysis

Hematologic parameter	Mean ± SD (sample size)
	wt	het	hom
WBC (103/μl)	8.49 ± 2.82 (35)	9.02 ± 3.54 (75)	8.79 ± 3.57 (51)
Neutrophils (%)	18.12 ± 9.47 (24)	27.00 ± 16.21 (9)	16.76 ± 5.62 (19)
Lymphocytes (%)	77.36 ± 10.50 (24)	67.77 ± 16.57 (9)	80.00 ± 6.24 (19)
Monocytes (%)	2.59 ± 2.03 (24)	3.34 ± 1.85 (9)	1.88 ± 1.46 (19)
Eosinophils (%)	0.75 ± 1.33 (24)	0.79 ± 0.74 (9)	0.53 ± 0.93 (19)
Basophils (%)	1.18 ± 0.82 (24)	1.14 ± 0.50 (9)	0.78 ± 0.51 (19)
Erythrocytes (106/μl)	9.60 ± 0.53 (35)	9.75 ± 0.66 (75)	9.53 ± 0.51 (51)
Hemoglobin (g/dl)	14.89 ± 0.62 (35)	15.07 ± 0.88 (75)	15.44 ± 0.63 (51)
Hematocrit (%)	43.42 ± 1.76 (35)	44.00 ± 2.60 (75)	44.79 ± 2.12 (51)
Mean corpuscular vol (fl)	45.33 ± 1.84 (35)	45.20 ± 1.98 (75)	47.04 ± 1.80 (51)
Mean corpuscular hemoglobulin (pg)	15.53 ± 0.70 (35)	15.48 ± 0.84 (75)	16.22 ± 0.70 (51)
Mean corpuscular hemoglobin concn (μg/μl)	3.43 ± 0.08 (35)	3.43 ± 0.11 (75)	3.45 ± 0.09 (51)
Platelets (103/μl)	945 ± 197 (35)	1,074 ± 275 (75)	962 ± 210 (51)

Spleen weights.

Spleens were removed from mice following PB collection and weighed as an indicator of possible changes in the physiologic functions of this organ. A skewing toward heavier spleens was seen in heterozygous and homozygous mutant mice compared with wild-type mice (Fig. 4), and broader ranges in spleen weights were seen for both homozygous and heterozygous mutant animals. However, a P value of 0.057 was calculated by Student t test, which indicated that the observed differences in weights are not statistically significant. Histological analysis did not reveal any substantial differences in cellularity or morphology in the spleen or other tissues.

FIG. 4.

Spleen weights. Normalized spleen weights of wild-type (wt) mice (n = 35) and animals either heterozygous (het; n = 79) or homozygous (hom; n = 55) for the fps^K588R allele. Spleen weights are presented as percentages of body weight. The dots below and above each box plot represent the 5th and 95th percentiles, respectively. The whiskers below and above each box plot extend to the 10th and 90th percentiles, respectively. Each box extends from the 25th to the 75th percentiles; the solid line within each box represents the median, and the dashed line represents the mean. Mean normalized spleen weight percentages ± SD): wt, 0.30% ± 0.08%; het, 0.36% ± 0.21%; hom, 0.39% ± 0.27%. Means of actual spleen weights ± SD (range) wt, 107 ± 28 mg (54 to 181 mg); het, 132 ± 76 mg (57 to 639 mg); hom, 133 ± 96 mg (85 to 780 mg). Comparison of wt versus hom data sets of normalized spleen weights by Student’s t test yielded a P value of 0.057.

BM hematopoietic progenitors.

Hematopoietic progenitors were cultured ex vivo in semisolid methylcellulose medium in the presence of either a cytokine cocktail (consisting of IL-3, IL-6, SF, and Epo) as a positive control, GM-CSF and Epo, or Epo alone. We observed no statistically significant differences in the numbers, types, or morphologies of the resulting hematopoietic colonies from mice of the different fps genotypes under any of these conditions (Fig. 5A to C). However, elevated numbers of CFU-GM-derived colonies and larger colony sizes were occasionally observed with some homozygous mutant mice in the presence of GM-CSF and Epo. Consequently, we analyzed the cellularity of CFU-GM colonies by randomly picking 20 colonies from each plate, pooling the cells, and counting them as an indicator of the proliferative capacity of the fps mutant CFU-GM progenitors relative to the wild-type progenitors (Fig. 5D). However, this did not reveal any statistically significant differences. Other stimuli tested included GM-CSF alone, IL-3 alone, M-CSF alone, or IL-3 with Epo, none of which revealed any significant differences (data not shown). These results demonstrate that catalytically active Fps tyrosine kinase is not necessary for BM hematopoietic progenitor cells to proliferate and differentiate in response to stimulation with either a cytokine cocktail, GM-CSF and Epo, IL-3 and Epo, Epo alone, or M-CSF alone.

FIG. 5.

BM hematopoietic progenitor cell colony assays. (A) BM cells were grown in the presence of a cocktail of cytokines, consisting of IL-3 (5 ng/ml), IL-6 (10 ng/ml), SF (50 ng/ml), and Epo (1 U/ml). (B) BM cells were grown in the presence of GM-CSF (5 ng/ml) and Epo (1 U/ml). Bars in panels A and B: white, total number of colonies; hatched, CFU-GM and CFU-M colonies combined; stippled, BFU-E colonies; black, CFU-GEMM colonies. Colony counts were made on day 8 postplating. In panel C, CFU-E colonies were grown from BM cells under the following conditions: the cytokine cocktail (white bars), GM-CSF and Epo (hatched bars), and Epo alone (black bars). CFU-E colony counts were made 2 days postplating. The sample sizes were six mice per genotype for each panel. The height of each bar represents the average number of colonies for the six mice analyzed, and the error bars represent SD. In panel D, 20 CFU-GM colonies were randomly picked from 1.2-ml methylcellulose cultures which had been seeded with 50,000 nucleated BM cells. Colonies were picked 9 days postplating, and cells were pooled and counted. The height of each bar represents the average number of cells per CFU-GM colony, and error bars indicate SD. The sample sizes were four mice of each genotype.

Flow cytometry analysis of BM hematopoietic cells.

Flow cytometry was used to determine whether myelopoiesis, erythropoiesis, and B-cell precursor production and/or development were in any way impaired in the BM of mice lacking catalytically active Fps. No significant differences were observed between mice of the different fps genotypes in the levels of Ly-6G⁺ CD11b⁺ myeloid cells (wild type, 49.7%; heterozygous mutants, 47.4%; homozygous mutants, 51.4%), TER-119⁺, CD44^lo erythroid precursors after the CFU-E stage (wild type, 17.8%; heterozygous mutants, 14.9%; homozygous mutants, 18.1%), or B220⁺ B-cell precursors (wild type, 30.4%; heterozygous mutants, 32.5%; homozygous mutants, 26.0%) (Fig. 6). Consistent with the results from the hematopoietic progenitor cell colony-forming assays and PB analysis presented above, these data demonstrate that the levels of lineage-specific hematopoietic precursors in BM of mice are not affected by the absence of catalytically active Fps.

FIG. 6.

Flow cytometry analysis. BM from wild-type (wt; n = 7) animals or mice homozygous (hom; n = 6) or heterozygous (het; n = 7) for the fps^K588R allele were isolated and subjected to flow cytometry analysis. (A) Myeloid precursors were stained with PE anti-Ly6G and FITC anti-CD11b, both of which are specific for cells of the myeloid/monocytic lineage. (B) Erythroid precursors were stained with the erythroid-specific MAb PE anti-TER119 and FITC anti-CD44, which recognizes the majority of hematopoietic cells. (C) B cell precursors were stained with PE anti-B220, which is B cell specific. The percentage in the upper right hand corner of each quadrant represents the average percentage of positively labeled cells out of 20,000 BM cells counted after lysis of erythrocytes.

Cytokine and LPS stimulation of cultured BMM.

To analyze any perturbations in signaling pathways in these mice, we focused on BMM, which normally express high levels of Fps and the cytokine receptors with which Fps is thought to interact. BMM were cultured for approximately 10 days, then starved in 0.5% FBS for 48 h, followed by stimulation with the cytokines GM-CSF, IL-6, IL-3 (data not shown), or LPS (Fig. 7). Immunoblotting analysis following stimulation with GM-CSF revealed that Stat3 and Stat5A did not undergo activating tyrosine phosphorylation to the same extent in BMM homozygous for the fps^K588R mutation as in wild-type cells (Fig. 7A and B), suggesting that Fps is somehow involved in regulating the tyrosine phosphorylation of these two signaling proteins downstream from the GM-CSF receptor. A time course of the phosphorylation of Stat3 following GM-CSF stimulation revealed that the difference in phosphorylation status was greatest at 15 min after cytokine exposure (data not shown). Stat3 and Stat5A were tyrosine phosphorylated to the same degree following IL-6 (Fig. 7A and B) or IL-3 (data not shown) stimulation of wild-type and fps^K588R cells, suggesting that Fps is not involved in signaling from the IL-3 or IL-6 receptor in the same way as it is in GM-CSF receptor signaling.

FIG. 7.

Signaling in cultured fps^K588R homozygous mutant BMM. BMM from wild-type (+/+) or homozygous mutant (−/−) mice were starved for 48 h in 0.5% FBS prior to stimulation with either GM-CSF (30 ng/ml) or IL-6 (30 ng/ml). Cells were exposed to each cytokine for 15 min at 37°C, scraped in 2× SDS sample buffer, run out on an SDS–7.5% polyacrylamide gel, transferred to Immobilon-P membrane, and then probed successively with the indicated antibodies: (A) anti-pStat3 (top) followed by anti-Stat3 (bottom); or (B) anti-pStat5A/B (top) followed by anti-Stat5A (bottom). Stat5B (90 to 92 kDa) is constitutively phosphorylated in these BMM cultures, even under starvation conditions, whereas Stat5A (95 kDa) is phosphorylated only following GM-CSF stimulation. (C) BMM were starved for 48 h in 0.5% FBS prior to exposure to various doses of LPS for 30 min at 37°C. Whole-cell lysates were run out on SDS–11% polyacrylamide gels, transferred to Immobilon-P membrane, and then probed with anti-Erk antibody (bottom) followed by anti-pErk antibody (top).

As Jaks are presumed to be directly responsible for Stat phosphorylation downstream of cytokine stimulation, we also checked for activation of Jak2 as the most likely candidate to be mediating Stat3 and Stat5A phosphorylation following GM-CSF stimulation. Interestingly, Jak2 was tyrosine phosphorylated to a comparable extent in wild-type and homozygous mutant BMM following GM-CSF stimulation (data not shown). This suggested that Fps may act somewhere downstream of Jak2 in the activation of Stat3 and Stat5A, possibly by phosphorylating a Stat docking site on the GM-CSF receptor β chain, or by phosphorylation of Stat3 and Stat5A directly.

We also analyzed the activation of Erk1 (p44) and Erk2 (p42) following treatment of cultured BMM with LPS. Our results revealed a dose-dependent reduction in Erk1/2 activation, which was apparent at an LPS dose of 100 ng/ml (Fig. 7C). This difference in Erk activation between wild-type and homozygous mutant BMM was overcome with a dose of LPS an order of magnitude greater, suggesting that compensatory mechanisms were being activated at this higher LPS dose, which circumvented the block resulting from the catalytically dead Fps.

DISCUSSION

The results of this study demonstrate that Fps kinase activity is not required for normal mouse development or the production of mature hematopoietic cells. Interbreeding of heterozygous mutant mice produced animals of all three genotypes in the expected Mendelian ratios, showing that Fps kinase activity is not essential for any developmental process in spite of its widely distributed expression pattern during embryonic and fetal development (5, 22). Homozygous mutant males and females were both fertile, producing normal litter sizes either when bred to wild-type animals or when interbred, showing that Fps activity is not essential for gametogenesis, fertilization, or any other reproductive functions. Histological analysis of all major organs and tissues has not revealed gross phenotypes in mutant animals (data not shown). We have occasionally observed cellular hyperplasia in the lung, which is reminiscent of the pathology seen in the GM-CSF or GM-CSF receptor knockout mice (11, 53, 59); however, this was not a consistent finding. We also saw elevated levels of peripheral WBC and splenomegaly in some mutant animals, but these differences did not reach statistical significance. BM-derived progenitor cells from fps mutant mice and controls gave similar in vitro colony growth responses to IL-3 and GM-CSF alone, or in combination with Epo, as well as M-CSF (data not shown). We did note marginally elevated numbers of CFU-GM colonies and larger variability in colony numbers from the mutant animals in the presence of GM-CSF and IL-3. Furthermore, CFU-GM from some homozygous mutant mice were substantially larger and displayed greater cellular density. These observations initially suggested increased numbers of myeloid progenitors in the BM and perhaps an increased proliferative capacity of these mutant progenitors. However, when these assays were repeated with larger numbers of animals, the differences observed were not statistically significant. The slight differences we observed in peripheral WBC, spleen weights, and in vitro colony growth might have reflected intermouse variation. Alternatively, these differences could have been due to differences in immune status, although this seems unlikely, as the cohorts of animals used in these experiments were housed together in the same cages. LPS challenge did provoke greater increases in peripheral WBC and splenomegaly in mutant mice; however, these differences were not consistently observed and did not reach statistical significance (data not shown). We also observed reduced sensitivity to LPS-induced activation of Erk1 and Erk2 in mutant macrophage cultures, which is again consistent with subtle defects in immune response. Challenges with specific pathogens are being carried out to pursue this line of investigation. Preliminary analysis of phagocytosis also indicated that BMM from mutant animals were fully capable of ingesting bacteria and mounting a typical oxidative burst response (data not shown).

The fps proto-oncogene is most prominently expressed in the myeloid lineage, and several observations have suggested an essential biological role for fps in the survival and differentiation of myeloid progenitors (14, 64, 68). The evidence for a molecular function of the Fps kinase in the regulation of myelopoiesis comes largely from reports of its activation upon stimulation of responsive cells with a number of cytokines which play prominent roles in regulating the differentiation of hematopoietic progenitors along the myeloid and erythroid lineages (24, 25, 32, 40, 43) and association with several members of the cytokine receptor superfamily, including those for IL-3, GM-CSF, and Epo (4, 24, 25, 32, 51). Activation of these receptors leads to the tyrosine phosphorylation and dimerization-induced activation of the Stats (10). We therefore examined the tyrosine phosphorylation status of Stat3 and Stat5A in BM-derived monocytes from the fps mutant mice after stimulation with the cytokines IL-3, GM-CSF, and IL-6. All three cytokines induced activation of Stat3, and GM-CSF and IL-3 induced Stat5A activation in wild-type cells. In contrast, tyrosine phosphorylation of Stat3 and Stat5A was dramatically reduced in homozygous mutant cells after GM-CSF treatment. These results strongly argue for an involvement of Fps in GM-CSF signaling, which does not extend to signaling from IL-3 or IL-6 receptors. While in humans, the receptors for IL-3, IL-5, and GM-CSF employ a common β-chain (26), the mouse genome appears to encode two closely related β chains, one of which is specific to the IL-3 receptor (31), while the other appears to be parologous with the human β chain (19). As activation of Stat3 and Stat5A appears to be compromised downstream of GM-CSF but not IL-3 in these fps mutant mice, Fps may be required for signaling from the common murine β chain, but it may not be necessary for signaling from the IL-3-specific receptor. As this common murine β chain is employed by the IL-5 receptor as well, it will be interesting to see if IL-5 signaling is also compromised in these animals. If this is the case, it is possible that in humans, Fps plays an important role in signaling downstream from IL-3, IL-5, and GM-CSF.

Ectopic expression studies have recently revealed an intrinsic ability of Fps to phosphorylate and activate Stat3 (48) but not Stat5A (38). It has also been suggested that GM-CSF stimulation may induce a complex formation between Fps and Stat3 (50). Stat3 is also activated downstream from several other cytokines, including IL-6, IL-11, oncostatin M, ciliary neurotrophic factor, and leukemia inhibitory factor, all of which employ receptors containing a common signaling chain called gp130. Indeed, Fps was shown to associate with this common receptor subunit (43). Although we have not looked extensively at signaling from all of these cytokines, we did see normal IL-6-induced activation of Stat3 in BMM from mutant fps mice; this suggests that Fps kinase activity may not be required for Stat activation downstream of cytokines using the common gp130 receptor subunit.

Members of the Jak family of PTKs are acknowledged to play prominent roles in transmitting signals from all members of the cytokine receptor superfamily, and their role in Stat phosphorylation is well documented (29, 38). The recently reported mouse knockout model of Jak2 clearly demonstrated an essential role for this kinase in signaling from receptors from IL-3, IL-5, GM-CSF, Epo, and thrombopoietin (49). In contrast, the involvement of Fps in signaling from these cytokine receptors remains controversial. We found that activation of Jak2 in response to GM-CSF, IL-3, and IL-6 was not affected by loss of Fps kinase activity (data not shown). This suggests that Fps kinase activity is not required for activation of Jak2. The work reported here argues that Fps may indeed play an critical role in signaling from the GM-CSF receptor and perhaps a redundant role in signaling from other cytokine receptors. We cannot be certain if Fps is directly responsible for tyrosine phosphorylation of Stat3 and Stat5A, but this is certainly a possibility. This study provides the first in vivo evidence for such a role for the Fps kinase. Other alternatives which must be considered include an indirect role, such as phosphorylation of tyrosine residues on the GM-CSF receptor β chain, which might act as recruitment sites for Stat3 or Stat5A, or a kinase-dependent role of Fps as an adaptor protein involved in recruitment of Stats to the activated receptor. In the absence of Fps kinase activity, recruitment of Stat proteins to the receptor and subsequent phosphorylation by Jak2 might be greatly compromised.

Normal numbers of mature myeloid cells in mice expressing only catalytically inactive Fps indicates that either Fps kinase activity is not involved in regulating myelopoiesis or its function is redundant with that of other kinases. It is unlikely that Fps function is independent of kinase activity; however, the generation and analysis of Fps-null mice will be required to confirm this. Assuming Fps does play a role in myelopoiesis, a possible explanation for the lack of a substantial myeloid phenotype in mutant fps mice is functional redundancy with some other kinase. The most likely candidate is the widely expressed Fer kinase, which is the only other known kinase with a structure closely related to that of Fps. Interestingly, both Fps and Fer have been shown to oligomerize, and this is mediated by the conserved coiled-coil motifs in their N-terminal domains (8, 35, 52). Although this finding raises the possibility of dominant-negative effects involving either homotypic (Fps-Fps) or heterotypic (Fps-Fer) interactions, we have recently shown that Fps and Fer do not interact (8). This would argue that kinase-dead Fps is also unlikely to have a potent dominant-negative effect on Fer. Furthermore, coexpression of wild-type and kinase-dead Fer at equimolar levels did not result in any significant inhibition of Fer kinase activity (8). In contrast, Fps kinase activity was shown to be inhibited by either an N-terminal fragment of Fps or a full-length kinase-dead mutant (52). However, these studies suggested that substantial inhibition would not be seen at equimolar ratios. This is substantiated by the normal GM-CSF-induced Stat3 and Stat5A phosphorylation we observed in macrophages which are heterozygous for the fps^K588R allele unpublished data.

The expression patterns of Fps and Fer are quite distinct. Fps is more restricted, with relatively high levels seen in a limited subset of cell types including myeloid cells, vascular endothelial cells, chondrocytes, and some epithelial and neuronal cells (22). On the other hand Fer is very widely expressed (25, 39), and levels comparable to that of Fps in myeloid cells are seen in most tissues. We have recently cloned the murine fer locus and have found that the exon structure is closely related to that of mammalian fps (unpublished data). The close structural relationship between fps and fer and their encoded kinases suggests that they may engage in similar biological functions. The observed association of Fps and Fer with a variety of different cytokine and growth factor receptors suggests these kinases may serve a general role in cell signaling and that fps may have evolved more recently to provide this function in more specialized cell types, such as those of the myeloid lineage or the vascular endothelium. Interestingly, transgenic mice engineered to express an activated fps allele displayed a vascular hyperplasia but no apparent myeloid phenotype (20). Thus far we have not detected substantial defects in the myeloid or vascular lineages in mutant fps mice. We have recently targeted the murine fer gene with a kinase-inactivating missense mutation (8). Through the generation of compound mutant fps and fer mice and performance of genetic rescue experiments, we intend to establish whether these two related kinases perform essential yet redundant functions in cytokine and growth factor signaling.