|Home | About | Journals | Submit | Contact Us | Français|
Macrophage precursors originate from, and undergo lineage commitment within designated sites of hematopoiesis, such as the mammalian bone marrow. These cells subsequently differentiate in response to stimulation with macrophage colony-stimulating factor (CSF-1). The amphibian bone marrow, unlike that of mammals, has been overlooked as a source of leukocyte precursors in favor of the liver sub-capsular region, where hematopoiesis occurs in anurans. Here we report that the bone marrow rather than the liver periphery provides macrophage progenitors to the amphibian Xenopus laevis. We identified the amphibian CSF-1, examined its gene expression in developing and virally infected X. laevis and produce it in recombinant form (rXlCSF-1). This rXlCSF-1 did not bind or elicit proliferation/differentiation of sub-cortical liver cells. Surprisingly, a sub-population of bone marrow cells engaged this growth factor and formed rXlCSF-1-concentration-dependant colonies in semi-solid medium. Furthermore, rXlCSF-1-treated bone marrow (but not liver) cultures comprised of cells with characteristic macrophage morphology and high gene expression of the macrophage marker, colony stimulating factor-1 receptor (CSF-1R). Together, our findings indicate that in contrast to all other vertebrates studied to date, Xenopus committed macrophage precursors populations are not present in the central site of hematopoiesis, but reside in the bone marrow.
Macrophage-lineage cells are indispensable to vertebrate host immunity and homeostasis. While the developmental pathways governing the production of mononuclear phagocytes are complex and poorly understood, it is well established that the colony-stimulating factor-1 (CSF-1, macrophage colony-stimulating factor, M-CSF) is the principal molecule responsible for the development, differentiation, proliferation and survival of these cell lineages across all vertebrate species examined to date [1–4].
The functional form of CSF-1 is a homo-dimer, which ligates the c-fms proto-oncogene product, the CSF-1 receptor (CSF-1R) . The CSF-1R cell surface expression is restricted primarily to committed macrophage precursors and their progeny cells [6,7], dictating the specificity of CSF-1 function.
Committed macrophage precursors of vertebrate species originate and reside in designated hematopoietic sites. In birds and mammals these cells arise from the bone marrow pluripotent populations [4,8,9] whereas in bony fish they emerge in the head kidney [10,11]. In anurans such as Xenopus spp., the sub-cortical (peripheral) zone of the liver functions as the site of hematopoiesis [12–15], while the Xenopus bone marrow is relatively rudimentary and thought to serve only as the site of granulocyte differentiation/storage [13,14].
Here we report the first discovery of an anuran CSF-1, the production of a recombinant form of this X. laevis macrophage growth factor (rXlCSF-1) and the utilization of rXlCSF-1 to identify and differentiate Xenopus macrophage precursors. While the liver periphery serves as the central site of hematopoiesis in this species, it does not appear to possess cells capable of binding, proliferating or differentiating in response to rXlCSF-1. In contrast and to our surprise, the X. laevis bone marrow contained a cell population(s) that bound rXlCSF-1. Upon in vitro culture with the rXlCSF-1, X. laevis bone marrow cells formed colonies in semi-solid medium and differentiated into cells with classic macrophage morphology and high CSF-1R expression. Our findings indicate that in contrast to other vertebrates, committed Xenopus macrophage populations are not present in the primary hematopoietic site of the sub-capsular liver, but are found in the rudimentary bone marrow.
Out-bred pre-metamorphic (stage 54–56) tadpoles, metamorphic (stage 64) and adult (2 years old) frogs were obtained from our X. laevis research resource for immunology at the University of Rochester (http://www.urmc.rochester.edu/smd/mbi/xenopus/index.htm). All animals were handled under strict laboratory and UCAR regulations (Approval number 100577/2003-151).
Gene synteny analysis was performed using the NCBI server, map viewer option. The CSF-1 loci of the human (Homo sapiens, chromosome 1), mouse (Mus musculus, chromosome 3) and green anole (Anolis carolinensis, chromosome 4) were compared. The conserved genes syntenic to CSF-1 were located in the Xenopus (Silurana) tropicalis genome (gene scaffold NW_003163367.1) and found to flank a putative CSF-1 gene. Partial X. laevis CSF-1 cDNA was identified using primers against the X. tropicalis CSF-1. RACE PCR was performed in accordance with manufacturers’ directions (Clonetech) to identify the 5′ and 3′ regions of the X. laevis cDNA transcript (Acc. No.: JX418294). The partial CSF-1R cDNA (Acc. No.: JX418295) was identified by methods described for CSF-1. All sequences of primers used are listed in the online supplementary table 1.
Protein sequence alignments were performed using the Clustal W software (http://www.ebi.ac.uk/clustalw/). Signal peptide regions were identified using the SignalP 3.0 Server (http://www.cbs.dtu.dk/services/SignalP/) and the transmembrane regions predicted using the TMHMM Server v. 2.0 (http://www.cbs.dtu.dk/services/TMHMM/). Phylogenetic analysis was performed by Clustal X software using the neighbor joining method and bootstrapped 10,000 times, with values expressed as percentages.
Fathead minnow cells (FHM; American Type Culture Collection, ATCC No.CCL-42) were maintained in DMEM (Invitrogen) supplemented with 10% fetal bovine serum (Invitrogen), penicillin (100 U/mL) and streptomycin (100 μg/mL) with 5% CO2 at 37°C. FV3 was grown by a single passage on FMH cells, purified by ultracentrifugation on a 30% sucrose cushion and quantified by plaque assay on FMH monolayer under an overlay of 1% methylcellulose .
Animals were infected by intra-peritoneal (ip) injections with 5×106 plaque forming units (PFU) of FV3 in 100 μl volumes. Three days after infection, frogs were euthanized by immersion in 0.5% tricaine methane sulfonate (MS-222), tissues removed and processed for RNA isolation.
Total RNA and DNA were extracted from frog tissues using the Trizol reagent following the manufacturer’s directions (Invitrogen). All cDNA synthesis was performed using the iScript cDNA synthesis kit according to manufacturers’ directions (Bio-Rad, Hercules, CA) using 500 ng of total DNAse treated (Ambion) RNA. One microliter of these respective synthesized cDNA samples or 50 ng of total isolated DNA were used as templates for RT-PCR analysis. PCR products were resolved on 1.5% agarose gels, visualized with ethidium bromide and compared against a 1kb plus DNA marker (Invitrogen).
Quantitative PCR gene expression analysis was performed using the delta^delta CT method using the ABI 7300 real-time PCR system and PerfeCTa® SYBR Green FastMix, ROX (Quanta). Expression analysis of the X. laevis CSF-1 and CSF-1R was performed relative to the GAPDH endogenous control and normalized against the lowest observed expression (tadpole muscle, N =3; fresh bone marrow cells, N=6, respectively). Q-PCR analysis of FV3 infected animals was performed using kidney and bone marrow tissues from 6 adult frogs and spleen and liver tissues from 3 animals (N=6 and 3, respectively) and normalized against the healthy kidney expression. Expression analysis was performed using ABI sequence detection system software (SDS). All primers were validated prior to use. Primer sequences are listed in the online supplementary table 1.
The portion of the X. laevis CSF-1 sequence representing the signal peptide-cleaved, extra-cellular fragment was ligated into the pMIB/V5 His A insect expression vector (Invitrogen) and introduced into Sf9 insect cells (cellfectin II, Invitrogen). Transfected Sf9 supernatants were confirmed to express the rXlCSF-1, positive transfectants were selected using 10 μg/mL blasticidin, scaled up into 500 mL liquid cultures and grown for 5 days. Cultures were pelleted, supernatants were removed and dialyzed overnight at 4°C against 150 mM sodium phosphate, concentrated against polyethylene glycol flakes (8 kDa) at 4°C, dialyzed overnight at 4°C against 150 mM sodium phosphate and passed through an Ni-NTA Agarose column (Qiagen) to bind the rXlCSF-1. The column was washed 2×10 volumes of high stringency wash buffer (0.5% Tween 20; 50 mM Sodium Phosphate; 500 mM Sodium Chloride; 100 mM Imidazole) and 5x with low stringency wash buffer (as above, but with 40 mM Imidazole). The rXlCSF-1 was eluted in fractions using 250 mM Imidazole. The purity of the eluted fractions was confirmed by silver-stain and the presence of rXlCSF-1 assessed by western blot against the V5 epitope on rXlCSF-1. Fractions containing rXlCSF-1 were pooled and the protein concentration determined by the Bradford Protein Assay (BioRad). A protease inhibitor cocktail (Roche) was added to the purified rXlCSF-1 and the protein aliquoted and stored at 4°C until use.
The vector control was produced by transfecting Sf9 cells in parallel to the rXlCSF-1 production, but with an empty expression vector, and the methodology described for the generation and isolation of the rXlCSF-1.
The ASF culture medium used in these studies has been previously described . All cell cultures were established using ASF supplemented with 10% fetal bovine serum, 2.5% heat-inactivated X. laevis serum, 20 μg/mL kanamycin and 100 U/mL penicillin/100 μg/mL streptomycin (Gibco). Amphibian PBS (APBS) has been previously described .
Immediately after euthanasia, frog livers were aseptically removed, the liver peripheries peeled off, and the cells isolated by gently passing through sterile nylon mesh. Liver cells were layered over 51% percoll (Sigma)/49% APBS and centrifuged at 400×g at 4°C for 25 minutes to separate out red blood cells and debris. The leukocyte-containing interfaces were removed and washed with ice-cold APBS prior to culture. Frog femurs were aseptically removed, flushed with 10 mL of ice cold APBS each (50 units/mL Heparin, Lancaster and 100 U/mL penicillin/100 μg/mL streptomycin, Gibco) and washed with ice-cold APBS prior to culture. Cultures were cyto-spun (Shandon Southern), Giemsa-stained and images derived using an Axiovert 200 inverted microscope and Infinity 2 digital camera (objective: 40/0.6, Zeiss). Digital images were analyzed using the Image-Pro Plus software.
For differentiation experiments, 5×104 of X. laevis liver periphery and bone marrow cells were incubated at 27 °C in individual wells of 48 well plates with rXlCSF-1 (1, 100 ng/mL) or equal volumes of vector control. Subsequently, cells were washed with, and re-suspended in, FACS staining buffer (APBS, 1%BSA, 0.01% sodium azide).
For MHC class II staining experiments, bone marrow cells were incubated on ice for 1h with anti-Xenopus MHC class II monoclonal antibody (AM20 ), washed and stained with a goat α-mouse FITC Ab (Sigma). Cells were again washed and analyzed by flow cytometry.
To assess rXlCSF-1 binding, freshly isolated bone marrow and liver periphery cells were incubated with vector control, 30 ng or 3 μg of rXlCSF-1 in culture medium on ice for 1h. Cells were washed and sequentially stained with α-V5 (Invitrogen) and goat α-mouse FITC (Sigma) Abs. Cells were washed and analyzed by flow cytometry.
All flow cytometry was performed on a FACSCanto II (BD) instrument using consistent instrument settings. Data analysis was performed using FlowJo software.
X. laevis liver periphery and bone marrow cells were incubated at 27 °C for 10 days in 35 × 10 mm sterile petri dishes in culture medium containing 0.8% methylcellulose (Spectrum) and either vector control, 1 or 100 ng/mL of rXlCSF-1. Colonies were enumerated using a grid and an inverted microscope (objectives: 25/0.40 and 40/0.6, Vista Vision, VWR).
One microgram of rXlCSF-1 was incubated in absence or presence of 2.5 mM disuccinimidyl suberate (DSS, final concentration, Therom Scientific) cross-liker for 30 min. Cross-linking was terminated for 15 min with 50 mM Tris (final concentration). The reactions were visualized using western blot against the V5 epitope and developed using ECL (Pierce) on X-ray film (Eastman Kodak Co.)
Statistical analysis was performed using a one-way analysis of variance (ANOVA) and Tukey’s post hoc test. Probability level of P<0.05 was considered significant. Vassar Stat was used for statistical computation (http://faculty.vassar.edu/lowry//anova1u.html).
To investigate the development of Xenopus macrophages, we sought to identify the amphibian ortholog of the central macrophage growth factor, colony-stimulating factor-1 (CSF-1). Since CSF-1 molecules share poor identity across vertebrate species [1,3] we utilized the fully sequenced X. tropicalis genome for gene synteny analysis to identify the X. tropicalis CSF-1. Accordingly, we assessed the loci of the human (Homo sapiens, chromosome 1), mouse (Mus musculus, chromosome 3) and green anole (Anolis carolinensis, chromosome 4) chromosomes, bearing the CSF-1 genes of respective species (fig. 1a). These CSF-1 loci exhibited highly conserved organization surrounding the respective CSF-1 genes, with FAM40A, ACHYL1 and EPS8L3 flanking the CSF-1 gene in each case (fig. 1a). We traced these to a scaffold (NW_003163367.1) in the X. tropicalis genome, where these orthologs surrounded a putative X. tropicalis CSF-1 gene (fig. 1a). This gene shared the greatest amino acid sequence identity with the chicken and zebrafinch CSF-1 proteins, while in silico analysis revealed the presence of a single globular domain, a single trans-membrane domain, and conserved cysteines, characteristic of all known vertebrate CSF-1 proteins (data not shown).
We designed primers against this X. tropicalis CSF-1 and, via conventional and RACE PCR, we successfully identified the full cDNA transcript of the X. laevis CSF-1 (online suppl. fig. 1). The X. laevis transcript encodes a 258 residue protein with a signal peptide, a hallmark globular CSF-1 domain, conserved cysteines and a single trans-membrane domain (online suppl. fig. 1). The X. laevis cDNA transcript exhibits a long 3′ untranslated region with 5 mRNA instability motifs (ATTTA) [19,20] and an unconventional polyadenylation signal (AATATA, (online suppl. fig. 1).
To examine the evolutionary relationships among vertebrate CSF-1 proteins, we performed phylogenetic analyses (fig. 1b). The CSF-1 of X. tropicalis and X. laevis branched closest together and independently of the other vertebrate CSF-1s (fig. 1b). The amphibian CSF-1 proteins were phylogenetically ancestral to the avian and mammalian CSF-1 molecules, albeit with a low bootstrap value (38%, fig. 1b). The teleost CSF-1s branched ancestral to the amphibian, avian and mammalian molecules while the related fms-like-tyrosine-3 ligand of X. tropicalis (flt3l) was used as an out-group (fig. 1b).
We performed quantitative gene expression analysis of X. laevis CSF-1 in tissues of tadpoles, metamorphic and adult frogs (fig. 2a). In comparison to adult frogs, tadpoles generally exhibited lower CSF-1 transcript levels and distinct expression patterns, whereas animals undergoing metamorphosis (stage 64) displayed relatively intermediate CSF-1 expression levels (fig. 2a). In tadpoles, CSF-1 mRNA levels were highest in the spleen tissues, and modestly increased with frog development (fig. 2a). While tadpoles possessed relatively low liver and lung CSF-1 transcript levels, adult frogs exhibited significantly increased CSF-1 mRNA levels in these tissues (fig. 2a). Interestingly, whereas tadpoles and adults displayed very low CSF-1 gene expression in skeletal muscle, frogs undergoing metamorphosis had elevated CSF-1 mRNA levels in regressing tail muscle tissues (fig. 2a), corroborating macrophage involvement in this process .
To assess possible changes of X. laevis CSF-1 gene expression following immunological challenge, we examined the expression of this gene in immuno-relevant tissues of adult frogs infected for three days with the amphibian viral pathogen, Frog Virus 3 (FV3, Ranavirus, Iridoviridae, fig. 2b). Consistent with our prior work, which strongly implicated the frog kidney as the primary site of ranaviral replication , FV3 infections significantly increased the CSF-1 transcript levels in kidney tissues of infected animals (fig. 2b). Surprisingly, CSF-1 gene expression was also significantly elevated in the bone marrow of virally infected frogs (fig. 2b). Notably, although the kidneys of infected animals exhibited high viral gene expression and the presence of large quantities of the FV3 genome, the bone marrow of infected frogs did not possess substantial viral infiltration or viral gene expression (online suppl. fig. 2). The CSF-1 gene expression in the liver and spleen tissues was not significantly altered by FV3 infections. (fig. 2b).
To identify X. laevis macrophage precursor population(s) and to characterize the differentiation of these cells in vitro, we produced a recombinant form of the extra-cellular region of X. laevis CSF-1 protein (rXlCSF-1) using an insect protein expression system. This rXlCSF-1 possessed His6 and V5 epitopes for efficient protein purification and western blot analysis, respectively.
Since all vertebrate CSF-1 proteins function as homo-dimers, we wanted to confirm that the insect-expressed rXlCSF-1 was capable of dimerization in solution. Accordingly, rXlCSF-1 was incubated in amphibian PBS (APBS) in the absence (lane 1) or presence (lane 2) of the chemical cross-linker DSS (fig. 3). The reactions were then resolved by SDS-PAGE, and visualized by western blot against the V5 epitope on rXlCSF-1. The non-cross-linked rXlCSF-1 resolved as a monomer of roughly 22 kDa, whereas the DSS cross-linked product exhibited an additional band, indicative of rXlCSF-1 dimerization (fig. 3, lanes 1 and 2, respectively).
Because the sub-capsular zone of the amphibian liver is believed to be the primary source of hematopoietic progenitor populations, we isolated these cells from adult frogs and examined (by flow cytometry) the capability of these cells to bind rXlCSF-1 (fig. 4a). Additionally, since more evolutionarily advanced vertebrates utilize bone marrow as the primary source of macrophage precursors and because we observed increased bone marrow CSF-1 gene expression following FV3 infections (fig. 2b), we also examined the capacity of healthy X. laevis adult bone marrow cells to bind the recombinant growth factor (fig. 4b). In absence of antibody specific against X. laevis CSF-1, we used an antibody against the V5 tag of rXlCSF-1. When freshly isolated peripheral liver cells from adult frogs were incubated with the rXlCSF-1, stained with α-V5 and goat α-mouse FITC Abs for the recombinant protein and analyzed by flow cytometry, no fluorescence above vector control levels was observed, regardless of the cell sub-population gated on and examined (fig. 4a). In marked contrast to the sub-cortical liver cells, a subpopulation of relatively large, and non-complex bone marrow cells (fig. 4b panel iii) clearly bound to the rXlCSF-1, as evidenced by increased fluorescence intensity. Furthermore, this binding was concentration-dependent (fig. 4b panel iii). These results are representative of two independent experiments, each using cells from three different frogs. The intensity of the positive signal, as well as its cell population (bone marrow but not liver cell) and concentration dependency, all strongly suggests that α-V5 detects the specific interaction rXlCSF-1 with its receptor.
To substantiate our observations that the X. laevis bone marrow, but not sub-capsular liver cells are capable of binding and presumably responding to rXlCSF-1, we performed a proliferation/colony formation assay using methylcellulose-based semi-solid medium (fig. 5). Cells were isolated from the liver sub-capsular zone and from the bone marrow of adult frogs and cultured in 0.8% methylcellulose semi-solid medium with the vector control, 1 or 100 ng/mL of rXlCSF-1. The above treatment conditions had no observable effects on the colony formation of the sub-capsular liver cells (fig. 5). By contrast, rXlCSF-1 treatments of bone marrow-derived cells elicited significant and concentration dependent increases in colony formation (fig. 5). These rXlCSF-1-induced bone marrow cell colonies were composed of large (>10μm), homotypic cells, suggesting that these colonies potentially comprised of macrophage lineage cells. The total cell counts of the sub-capsular liver and bone marrow liquid cell cultures corroborated the above findings, where, regardless of treatment total numbers of peripheral liver cells were unchanged whereas the numbers of cultured bone marrow cell markedly increased with increasing concentrations of rXlCSF-1 (online suppl. Fig. 3).
Mammalian macrophages require concentrations of around 1000 units/mL for maximal proliferation , which translate to high ng/mL concentrations . In accordance with this and the observations described above, we cultured the X. laevis sub-capsular liver cells and bone marrow cells with either vector control or 100 ng/mL of rXlCSF-1 and examined the effects of these treatments on culture cell compositions by flow cytometry (fig. 6a). Consistently and regardless of treatment, no observable differences were seen with sub-capsular liver cell cultures (fig. 6a top panels), whereas intermittent culture times also had no observable differences (data not shown). In contrast, flow cytometry analysis of control-treated bone marrow cell cultures revealed a population of relatively large cells, while the parallel, rXlCSF-1-treated cultures contained greater proportions of these cells (28.8% vector control; 45.2%, rXlCSF-1 cultures), which, in comparison to controls, were larger in size and exhibited increased internal complexity (fig. 6a bottom panels). These results are representative of two independent experiments, each using cells from three individual adult frogs. Interestingly, freshly isolated bone marrow cells (fig 4b) and rXlCSF-1 derived bone marrow cultures (fig. 6a) exhibited distinct cell population profiles. Whether the addition of rXlCSF-1 can cause an expansion of a subpopulation of cells present in low numbers in the initial primary cells isolated from the bone marrow is not known.
We examined, by microscopy, the cell morphology of bone marrow cultures treated with vector control or rXlCSF-1 and stained with Giemsa. Cultures incubated with the vector control (fig. 6b top panel) or in medium alone (data not shown) consisted primarily of mononuclear cells with large cytoplasms and very characteristic phagocyte morphology. Strikingly, cultures that had been incubated with rXlCSF-1, comprised of much larger cells, exhibiting extensive vacuolation and membrane ruffling (fig. 6b bottom panel), consistent with the hallmark characteristics of CSF-1-differentiated macrophages .
To confirm that the rXlCSF-1-derived bone marrow cultures contained macrophage lineage cells, we cloned a fragment of the X. laevis CSF-1 receptor (CSF-1R) cDNA and compared its gene expression in freshly isolated and rXlCSF-1-incubated bone marrow cell cultures (fig. 6c). The rXlCSF-1-derived cultures displayed significantly increased expression of the CSF-1R (fig. 6c), corroborating the responsiveness of these cells to the cognate CSF-1 ligand.
To further assess the rXlCFS-1-elicited differentiation of bone marrow cultures, we examined, by flow cytometry, the MHC class II surface expression on these cells. Bone marrow cells from the control cultures were comprised of a large and complex population that stained low for class II, and a population of large but relatively less-complex cells with high class II staining (online suppl. fig. 4 left panels). Interestingly, the rXlCFS-1-derived cultures were comprised of a larger, more complex, class II low population as well as an increased proportion of large cells with class II high staining, some, but not all of which exhibited increased granularity (online suppl. fig. 4 right panels).
This is the first report detailing macrophage development of an anuran, as well as the first identification of an amphibian CSF-1. The synteny surrounding this vertebrate gene has been extremely well conserved, permitting the identification of the X. tropicalis CSF-1. It should be noted that while birds and mammals possess a single, alternatively spliced CSF-1 [24,25], fish possess two distinct CSF-1 genes that do not appear to be alternatively spliced . It is unknown whether the distinct fish CSF-1s assume the roles conferred by the mammalian CSF-1 variants. Although the two zebrafish CSF-1 genes are located on distinct chromosomes, each flanked by some of the genes syntenic to the single mammalian CSF-1, we traced several of these segregated genes to a single X. tropicalis gene scaffold, negating the existence of an additional anuran CSF-1 (data not shown). In addition, we were unable to identify alternatively spliced X. laevis CSF-1 transcripts through either conventional or RACE-PCR, using a range of cDNA templates from healthy and immunologically challenged tadpoles and adults (data not shown). Although we cannot rule out the possibility of either an additional or alternatively spliced Xenopus CSF-1, our results suggest the absence of both. Notably, the N-terminal 150 residues of mammalian CSF-1s are sufficient for activity [26,27], while amphibians are evolutionarily intermediate between teleosts and mammals, which exhibit two un-spliced and one alternatively spliced CSF-1, respectively. Our results suggest that the Xenopus macrophage development strategies are different to those seen in other vertebrates. Possibly, these distinctions dictate the requirement for a single, membrane-bound Xenopus CSF-1, with more stringent enzymatic release into extra-cellular milieu.
Our gene expression studies revealed greatest CSF-1 transcript levels in adult liver and lung tissues and elevated kidney and bone marrow CSF-1 mRNA levels following FV3 infections. Perhaps the increased bone marrow and kidney CSF-1 expression evince increased production of macrophages and their recruitment to infection sites, respectively. Mammalian CSF-1 is responsible for the maintenance of alveolar macrophages  whereas Kupffer cells’ viability and liver integrity are interdependent of CSF-1 . Presumably, the above is reflected in the high CSF-1 expression in X. laevis liver and lung tissues.
The surface expression of the CSF-1R on progenitor cell populations is considered the hallmark of commitment to the macrophage lineage  where the mRNA and protein levels of CSF-1R increase to detectable levels from macrophage-colony forming units through macrophage development and maturation [31–33]. Prior to this commitment, CSF-1 stimulation is insufficient to drive progenitors down macrophage differentiation pathways, where CSF-1 acts in synergism with factors like kit ligand and IL-3 to influence pluripotent progenitors towards the mononuclear phagocyte lineage [8,9]. Subsequent to commitment and in coordination with increased CSF-1R expression, these cells become responsive to CSF-1 stimulation and differentiate along macrophage pathways [34,35]. Pluripotent stem cells and cell commitment occur in hematopoietic organs such as the avian and mammalian bone marrow [4,22] and the teleost head kidney [10,11]. Interestingly, although the Xenopus sub-cortical liver clearly functions as the primary site of hematopoiesis [13,14,36], our findings strongly suggest that committed macrophage precursor populations are located in the Xenopus bone marrow. Notably, where macrophage lineage commitment is defined by responsiveness to CSF-1, rXlCSF-1-stimulated bone marrow cells proliferated, formed colonies, and differentiated into cells with classic macrophage morphology and high CFS-1R expression. In contrast, rXlCSF-1-treatment of sub-capsular liver cells had no observable effects. Furthermore, whereas a distinct bone marrow cell population exhibited rXlCSF-1 binding, no such interactions were seen with cells derived form the sub-capsular liver.
It is worth reiterating that while the Xenopus bone marrow is relatively rudimentary, it has been confirmed as the site of granulocyte differentiation/storage [13,14]. Indeed, we have observed a large number of polymorphonuclear granulocytes amongst the freshly isolated bone marrow cells (data not shown). However, subsequent to several days of culture, we observed substantially fewer-to-no polymorphonuclear cells in the bone marrow cultures (regardless of rXlCSF-1 treatment, data not shown), reflected by the decreased total numbers of bone marrow cells recovered with culture time. Possibly, as in mammals, short-lived mature granulocytes are also stored in the Xenopus bone marrow. Presumably and for physiological reasons beyond our speculation, Xenopus mononuclear phagocytes may originate from sub-cortical liver-derived pluripotent stem cells and subsequently mature and reside in the amphibian bone marrow. Since freshly isolated and rXlCSF-1 derived bone marrow cultures exhibited distinct cell population profiles, it is possible that the bone marrow macrophage precursors are initially present in low proportions but are longer lived than the granulocytes and are otherwise expanded by the rXlCSF-1. The relatively small proportion of bone marrow cells staining positive for bound rXlCSF-1 is consistent with this possibility.
While macrophages and granulocytes display partially overlapping gene expression profiles including transcriptional regulation of CSF-1R, only mononuclear phagocytes but not granulocytes possess surface protein expression of this receptor . Unfortunately, the present work has been limited by the absence of an anti-Xenopus CSF-1R antibody, preventing us from confirming CSF-1R protein expression in the rXlCSF-1-derived macrophage cultures. However, by means of several distinct assays we have observed that a population of Xenopus bone marrow cells respond to the rXlCSF-1 and differentiate into cultures of primarily large mononuclear cells with very characteristic macrophage morphology. Thus, although we don’t have direct evidence that the increased expression of CSF-1R in rXlCSF-1-derived bone marrow cultures is paralleled by surface protein expression of this receptor, we are fairly confident that the resulting cell types are mainly mononuclear lineage phagocytes.
Our findings regarding the source of Xenopus committed macrophage precursors are notable in comparison to what has been seen in all other vertebrate species studied to date. The distinct physiology and/or ecological pressures of this organism likely dictate the Xenopus macrophage development pathways. Notably, the more terrestrial amphibians such as the Rana genus utilize the bone marrow for erythropoiesis  whereas this process undeniably occurs in the Xenopus liver periphery . Possibly, the use of the bone marrow for blood cell development has occurred in stages throughout evolution. Further investigations into the macrophage lineage developmental strategies utilized by different amphibian species could well lead to a better understanding of the evolution and the complexities of these pathways in all vertebrates.
We thank Tina Martin and David Albright for animal husbandry. We thank Dr. Nicholas Cohen and Dr. Eva-Stina Eldhom for their critical review of this manuscript. This research was supported by R24-AI-059830 and IOB-074271 grants from NIH and NSF, respectively. LG was supported by NSERC-PDF.