The fms-related tyrosine kinase 3 (Flt3) and its ligand (Flt3lg) are important regulators of hematopoiesis and dendritic cell (DC) homeostasis with unsettled coevolution. Gene synteny and deduced amino acid sequence analyses identified conserved flt3 gene orthologs across all jawed vertebrates. In contrast, flt3lg orthologs were not retrieved in ray-finned fish, and the gene locus exhibited more variability among species. Interestingly, duplicated flt3/flt3lg genes were maintained in the allotetraploid Xenopus laevis. Comparison of modeled structures of X. laevis Flt3 and Flt3lg homoeologs with the related diploid Xenopus tropicalis and with humans indicated a higher conformational divergence between the homoeologous pairs than their respective counterparts. The distinctive developmental and tissue expression patterns of Flt3 and Flt3lg homoeologs in tadpoles and adult frogs suggest a subfunctionalization of these homoeologs. To characterize Flt3 cell surface expression, X. laevis–tagged rFlt3lg.S and rFlt3lg.L were produced. Both rFlt3lg.S and rFlt3lg.L bind in vitro Flt3.S and Flt3.L and can trigger Erk1/2 signaling, which is consistent with a partial overlapping function between homoeologs. In spleen, Flt3.S/L cell surface expression was detected on a fraction of B cells and a population of MHC class IIhigh/CD8+ leukocytes phenotypically similar to the recently described dual follicular/conventional DC-like XL cells. Our result suggests that 1) Flt3lg.S and Flt3lg.L are both involved in XL cell homeostasis and that 2) XL cells have hematopoietic origin. Furthermore, we detected surface expression of the macrophage/monocyte marker Csf1r.S on XL cells as in mammalian and chicken DCs, which points to a common evolutionary origin in vertebrate DCs.
Bone marrow–derived dendritic cells (DCs) are central to the mammalian immune system by bridging the adaptive and innate arms. By processing and presenting Ags in the context of the MHC, DCs trigger T cell–mediated immune responses by interacting with the TCR. TCR signaling, leading to T cell activation, depends on various factors, including the affinity of the TCR for the Ag–MHC complex, the expression of coreceptors (CD4 and CD8), as well as costimulation (e.g., CD28) and coinhibition signals (e.g., CTLA-4 and PD-1) provided by their respective ligands (CD80/86 and PDL1) on APCs (1, 2). In secondary lymphoid organs, migratory conventional DCs (cDCs) drive adaptive immune responses and maintain systemic homeostasis against invading pathogens and harmless Ags, respectively (3). Adaptive immunity characterized by MHC molecules and somatically diversified Ag receptors of the Ig superfamily (TCRs and BCRs) is thought to having arisen in the common ancestor of all jawed vertebrates (4). Whereas B and T cells are easily identified by their respective receptors or coreceptors in all jawed vertebrates, the identification of DCs has been more challenging because of their numerous common features with macrophages (5). To date, dendritic-like cells have been mostly described in Gnathostomata according to their morphology, high expression of MHC class II (MHC-II), and costimulatory molecules (CD83, CD80) (5–7). Functionally, dendritic-like cells in ray-fined fish and birds have the capacity to phagocytose and, importantly, to stimulate T cell proliferation in vitro (5, 6, 8). In Xenopus, single DC-like cells called XL cells have been identifies in the spleen to date (9, 10). XL cell features are reminiscent of both mammalian follicular and conventional DCs, because they express a high level of MHC-II, bind and retain native Ags, express genes related to both follicular DCs (FDCs) and cDCs (spi1, cxcl13, and ccl19), and are in close contact to B and T cells (10). Therefore, it was hypothesized that XL cells represent a primordial “double-duty” DC with FDC and cDC roles in Xenopus, and likely in all ectotherms that lack bona fide FDCs. Although increasing evidence suggests that dendritic-like cells are present in all ectotherm vertebrates, many aspects of their ontogeny, physiology, and function are unknown.
The fms-related tyrosine kinase 3 (Flt3), also known as CD135, belongs to a family of the class III receptor tyrosine kinases (RTKs) that includes Kit (CD117), CSF1 receptor (Csf1r; CD115), and platelet-derived growth factor receptors (Pdgfrs). Similar to other RTKs, Flt3 is a cell surface receptor that binds its cognate extracellular Flt3 ligand (Flt3lg). The high-affinity ligand binding leads to receptor activation of the intracellular tyrosine kinase domain, which initiates multiple intracellular signaling cascades (11, 12). In humans and mice, Flt3 is expressed by hematopoietic, lymphoid, and DC progenitors in the bone marrow as well as by mature DCs in lymph nodes and spleen. As such, Flt3 signaling regulates progenitor survival and proliferation, as well as DC homeostasis (13, 14). Flt3 overexpression or mutation is associated with hematological cancers, including acute myeloid leukemia and B and T cell acute lymphoblastic leukemia (11, 13). Similar to Csf1r and Kit, Flt3 is evolutionarily conserved in jawed vertebrates, including teleost fish and tetrapods (15, 16). The Flt3 conserved role in hematopoiesis in zebrafish is supported by impairment of hematopoiesis induced by gene silencing with morpholinos, as well as expansion of the myeloid compartment resulting from the expression of a constitutively activated human FLT3 gene (16). In contrast, the role of Flt3 in DC biology outside mammals is less clear. Recently, flt3 transcripts and Flt3 molecules have been detected in dendritic-like cells of the Atlantic cod and chicken, respectively, suggesting that Flt3 has an evolutionarily conserved role in DC development and homeostasis (8, 17–19).
In mammals, Flt3lg is required for DC differentiation and homeostasis (20–24). Hematopoietic growth factors such as Flt3lg are dimeric short-chain α-helical bundles (12). They are expressed as membrane forms at the cell surface and are secreted by proteolytic cleavage or alternative splicing (11). Both secreted and membrane forms are biologically active, although their potential distinct biological roles remain unclear (11). Flt3lg proteins are abundantly produced by various immune and nonimmune cells, including bone marrow, stromal cells (endothelial cells), T cells, and thymic fibroblasts (11, 25–28). Interestingly, unlike Csf1 and Kit ligand, which are conserved across most jawed vertebrates, Flt3lg gene orthologs have only been identified in tetrapods to date (29).
The evolution of immune genes in jawed vertebrates has been shaped by two rounds of whole-genome duplications, as well as by species-specific tandem gene duplications (30, 31). Duplicated genes often undergo negative selection by loss of function (pseudogenization or diploidization), which leads to singletons. Alternatively, gene duplication can trigger neo- or subfunctionalization (32). Such phenomena can provide a glimpse into functional evolution of complex immune factors such as Flt3lg/Flt3. Whereas Xenopus tropicalis is a diploid species, its relative Xenopus laevis is an allotetraploid species that arose by hybridization of two diploid species some 17–18 million years ago (MYA) (33). Interestingly, the two subgenomes evolved asymmetrically, with the short (S) chromosomes having undergone more interchromosomal rearrangements, gene losses, and silenced gene expression than the long (L) chromosomes (33). Notably, most genes involved in Ag receptor generation (e.g., TCR and BCR), Ag presentation (e.g., MHC class Ia and MHC-II), and complement and pathogen recognition have been lost in S homoeologous chromosomes, whereas multiple genes encoding proteins involved in costimulation (e.g., CD86), signaling (e.g., Lck), cytokines, and cytokine receptors have been retained presumably because of stoichiometrically controlled expression or subfunctionalization of the two homoeologous genes (32, 33). During our efforts to improve immune gene annotation in Xenopus genomes, we noted the occurrence of Flt3 and Flt3lg genes in the conserved genomic region of X. laevis L and S chromosomes. This motivated our interest to further investigate the expression profiles and functions of these homoeologous genes.
In the present work, we have investigated the evolution of Flt3/Flt3lg genes in gnathostomes. We found that although flt3 is well conserved across all jawed vertebrates, flt3lg was only retrieved in tetrapods and cartilaginous and lobe-finned fish but not in ray-finned fish. Furthermore, we provide evidence that duplicated X. laevis flt3 and flt3lg have been retained and likely subfunctionalized during evolution. Our study using recombinant tagged Csf1.S and Flt3lgs also substantiates the hematopoietic origin of XL cells suggesting a common origin of DCs in tetrapods.
Materials and Methods
Premetamorphic tadpoles (stage 56) and adult frogs (1 y old) were obtained from the X. laevis research resource for immunology at the University of Rochester (https://www.urmc.rochester.edu/microbiology- immunology/xenopus-laevis.aspx). Animals were handled in accordance with the stringent laboratory and University Committee on Animal Research regulations (approval no. 100577/2003-151).
Phylogenetic and gene synteny analysis
Deduced Flt3 and Flt3lg amino acid sequences were obtain using public databases (https://www.ncbi.nlm.nih.gov/, https://www.xenbase.org/) and were aligned using the MUSCLE algorithm. The length of the predicted Flt3lg is highly different across the investigated genera, and thus only the domains predicted as a 4-helix cytokine–like core were used for protein alignment. Domain prediction was carried out using the online server InterPro (https://www.ebi.ac.uk/interpro/). Full Flt3 sequences were used for multiple alignment and phylogenetic analysis. Phylogeny was inferred using the neighbor-joining method (34) using MEGA X (35, 36). Confidence of branching was estimated by percent bootstrapping (1000 replicates).
Predicted functional domains of Flt3/Flt3lg from the X. laevis S and L chromosomes, X. tropicalis, and Homo sapiens were aligned using the online platform (http://multalin.toulouse.inra.fr/) (37). Functional Flt3 domains were identified based on the alignment with human FLT3 (29). Three-dimensional homology models of the extracellular Flt3 domains and 4-helix cytokine–like cores of Flt3lg from X. laevis S and L homoeologs and X. tropicalis were compared using SWISS-MODEL protein modeling server (38).
Relative gene expression
RNA extraction, cDNA synthesis, and quantitative reverse transcription–PCR (RT-qPCR) analysis were realized as previously described (39). The expression level of each gene was normalized to gapdh endogenous control and then normalized against the lowest observed expression among all tested tissues including both S and L homoeologous genes. Homoeologous-specific primers were designed using Primer-BLAST (https://www.ncbi.nlm.nih.gov/tools/primer-blast/). Primer specificities were assessed by analyzing PCR products on 2% agarose gel and RT-qPCR melting curves. All primers used are listed in Supplemental Table I.
Recombinant protein production
Tagged recombinant proteins were produced by cloning X. laevis extracellular Flt3lg.S and Flt3lg.L into pMIB/V5 His A vector (Thermo Fisher Scientific). Coding sequences were amplified using X. laevis cDNA by PCR (all primers used are listed in Supplemental Table I). HindIII and XhoI domains were inserted before double digestion (New England Biolabs) and T4 DNA ligase to insert in the plasmid vector (New England Biolabs). Sf9 insect cells were transfected using Cellfectin II (Invitrogen) and 1 µg of plasmid containing xlFlt3lg.S, xlFlt3lg.L, or xlCsf1.S (40). Large-scale protein production was performed in Sf-900 II serum-free medium supplemented with 10 µg/ml gentamicin. rFlt3lg.L and rFlt3lg.S were purified by Ni-NTA-agarose chromatography (Qiagen) as previously described (40). The recombinant enriched fractions supplemented with 0.02% NaN3 and cOmplete EDTA-free protease inhibitor cocktail (Roche) were stored at 4°C or −20°C for short- and long-term storage, respectively.
Cell isolation and flow cytometry
Splenocytes (200,000) harvested using a 70-µm nylon mesh were stained with 4 ng/µl enriched rFlt3lg.S or rFlt3lg.L or rCsf1.S in staining buffer for 1 h on ice as previously described (40, 41). Costaining was performed using Xenopus-specific biotinylated anti-CD8 (AM22), anti–MHC-II (AM20), and anti–IgM-Alexa Fluor 680 (10A9). Prior to flow cytometry measurement, stained cells were incubated with 15 pg of DAPI (BD Pharmingen) to stain and therefore exclude dead cells. Flow cytometry was performed on an LSR II flow cytometer (BD Biosciences). Analysis was carried on using the FCS Express 7 software (De Novo).
Adult splenocytes (5 × 105) were incubated in amphibian phosphate saline buffer supplemented with 500 ng/ml rFlt3lg.S or rFlt3lg.L for 5, 15, 30 and 60 min at 27°C, then centrifuged for 1 min at 13,000 × g at 4°C. Cell pellets were lysed in 125 µl of immunoprecipitation lysis buffer (Thermo Scientific Pierce) supplemented with protease inhibitor mixture (cOmplete, Roche), phosphatase inhibitors II and III (Sigma-Aldrich), and 1% SDS. Cell debris was removed by centrifugation at 16,000 × g during 8 min at 4°C. After boiling the supernatant in 4× loading buffer with 2-ME for 10 min, the samples were loaded on 12% acrylamide gels for electrophoresis and then transferred onto polyvinylidene difluoride membranes. The membranes were blocked using EveryBlot blocking buffer (Bio-Rad) for 1 h at room temperature and probed overnight at 4°C with p-Erk1/2 (Thr202/Tyr204) Ab (Cell Signaling Technology, no. 9101) at 1:1000 dilution or Grp94 mAb (Thermo Fisher Scientific, clone 9G10.F8.2) at 1:200 dilution in blocking buffer. The membranes were washed three times with TBS supplemented with 0.1% Tween 20 (TBST) and incubated for 1 h at room temperature with goat anti-rabbit or anti-rat IgG (H+L) Abs conjugated with HRP (Thermo Fisher Scientific) at 1:1000 dilution in blocking buffer. After three washes with TBST, the membranes were incubated for 5 min with SuperSignal West Femto maximum sensitivity substrate (Thermo Fisher Scientific) and imaged using a Bio-Rad ChemiDoc instrument. Chemiluminescent signals were analyzed by measuring the peak area using ImageJ (National Institutes of Health, 1.53k).
Expression levels of flt3 and flt3lg homoeologs were analyzed using the Mann–Whitney U test. For gene expression, comparisons between the FACS-sorted cells the Kruskal–Wallis test followed by the multiple step-up method of Benjamini, Krieger, and Yekutieli were used. The results were considered significant at an α level of 5% (p < 0.05). All statistical analyses were carried out using GraphPad Prism 9 software (GraphPad Software).
Evolutionary relationships of flt3/flt3lg orthologs
We first compared gene syntenies of flt3 and flt3lg genomic regions in X. laevis with those of the related diploid X. tropicalis, as well as with Homo sapiens, Callorhinchus milii, Latimeria chalumnae, Danio rerio, and Takifugu rubripes. Although the gene synteny of flt3 loci has been preserved among all the species considered, including both X. laevis homoeologs L and S (Fig. 1), flt3lg loci revealed signs of segment recombination events (Fig. 2). Importantly, Flt3lg genes have conserved proximity with slc17a7, pih1d1, and aldh16a among tetrapods and elasmobranch species, but not teleost fish (e.g., Danio rerio, Takifugu rubripes, Oncorhynchus mykiss, Nothobranchius furzeri, Oreochromis niloticus, and Larimichthys crocea; (Fig. 2 and data not shown). The flt3 high degree of conservation is also supported by phylogenetic analysis of deduced coding amino acid sequences, where the phylogenetic tree topology matching the species evolutionary history is robust (e.g., high bootstrap support; Supplemental Fig. 1). In contrast, Flt3lg phylogeny appears to be less robust and subjected to more diversity (Supplemental Fig. 1).
To gather additional insights into structural relationships among the different molecules, we conducted in silico conformation modeling. The Flt3 architecture including the extracellular domain with five Ig-like domains (D1–D5), the transmembrane, and the tyrosine kinase domains is conserved in X. laevis Flt3.S and Flt3.L (Fig. 3). All of the cysteines involved in the Ig folding via disulfide bonds in D1, D2, D3, and D4 are conserved among X. laevis, humans, and other jawed vertebrate species (29). However, cysteine residues are not conserved on the D2 in different amphibian species, including X. tropicalis, and in fish. Additionally, D1 and D2 have been lost in two cartilaginous fish (Carcharodon carcharias and C. milii; (Fig. 3A). Differently from the other RTKIII/V humans, the combined D1–D4 domains form an open horseshoe ring with D2 leaning against D3. D2 and D3 interaction is mediated by hydrophobic amino acids, which are evolutionarily conserved in jawed vertebrates whose D1 and D2 domains are conserved (Fig. 3A). Additionally, the amino acids involved in D3 and D4 hydrophobic interaction, as well as the linker residues that form an elbow, are conserved. This elbow domain is conserved among all RTKIII/Vs and is likely important for homotypic receptor conformation. Whereas in humans, FLT3LG binds to D3 of FLT3, it is divergent in the Xenopus model and in other jawed vertebrates. As highlighted in (Fig. 3, a single residue of Flt3 at the interface with Flt3lg is conserved. Interestingly, lobed and ray-fined fish appear to have lost this evolutionarily conserved hydrophobic residue (F281 in humans) forming the ligand-binding epitope in humans.
Pairwise structure superpositions of the extracellular domains of the two X. laevis and the single X. tropicalis Flt3 suggest that D3, which represents the putative Flt3lg binding domain, is more divergent in Flt3.L than Flt3.S and X. tropicalis (xt)Flt3 (Fig. 3C). Interestingly, amino acid sequences of the extracellular Flt3.S and Flt3.L domains show a higher percentage of identity between X. laevis homoeologs than with the Flt3 of X. tropicalis (Table I). Xenopus extracellular Flt3 shows relatively low percentage of identities (∼35%) with the human and mouse counterparts.
|.||X. laevis_S .||X. laevis_L .||X. tropicalis .||H. sapiens .||M. musculus .|
|.||X. laevis_S .||X. laevis_L .||X. tropicalis .||H. sapiens .||M. musculus .|
Based on in silico analysis, the Xenopus Flt3lg molecules are composed of a four-helix bundle or short chain helical cytokine fold, a transmembrane, and an intracellular domain as in humans. Examination of amino acid residues, which are critical for the bioactivity, structure, and dimerization of the human FLT3LG, indicates that these residues are largely conserved in X. tropicalis and the two X. laevis homoeologs (Fig. 4A) (29, 42, 43). However, the Flt3 binding domain, which is well conserved across tetrapods, has become more divergent for the L homeolog with the insertion of 3 aa (Fig. 4A) (29). Furthermore, analysis by structure superposition of the “hot spot” identified in humans by amino acid substitution and in vitro bioactivity reveals that the structure of these domains is markedly divergent between Flt3lg.S and Flt3lg.L (Fig. 4). Interestingly, amino acid sequences of Flt3lg four-helix cytokine-like cores show a higher percentage of identity between the X. laevis homoeologs than with Flt3lg of X. tropicalis (Table I). Xenopus Flt3lg shows a relatively low percentage of identity (∼30%) with the human and mouse counterparts (Table I).
Differential expression of X. laevis flt3/flt3lg homoeologs
To obtain evidence of distinctive functions of the two pairs of flt3/flt3lg homoeologs, we determined their relative gene expression patterns in various organs of tadpoles and adult frogs (Fig. 5). In tadpoles, the relative flt3.L expression in spleen was significantly more elevated than flt3.S (p = 0.0022). In adult spleen as well as in liver, flt3.L transcript levels were significantly higher than those of flt3.S (p = 0.0079) and tended to be more abundant in the bone marrow (p = 0.2856). It is also noteworthy that the increased transcript levels of flt3.L over the flt3.S homeolog were 130-fold higher in tadpole spleen versus only 10-fold in adults. Further differential expression between flt3.L over flt3.S homoeologs was detected in kidney and in thymus of adults (p = 0.0152 and p = 0.0411, respectively). Inversely, flt3.S was more expressed than flt3.L in the brain and thymus of tadpoles (p = 0.0022 and p = 0.0411, respectively).
For Flt3lg gene homoeologs, flt3lg.L was significantly more expressed than flt3lg.S in the spleen of tadpoles (Fig. 5, p = 0.0022). Similarly, in adult spleen as well as in the bone marrow, flt3lg.L was significantly more expressed than flt3lg.S (p = 0.0159 and p = 0.0079, respectively). Also, flt3lg.L transcript levels were significantly higher than those of flt3lg.S in tadpole adipose tissues (p = 0.0159), whereas no statistical difference in relative expression was detected between the two genes in adult skin and lung (p = 0.1143 and p = 0.0952). In tadpoles, flt3lg.S was more expressed than flt3lg.L in liver and gills (p = 0.0043 and p = 0.0031).
Collectively, the differences in three-dimensional structure models among homeolog proteins and the changes in their developmental and tissue expression converge in suggesting a selective retention of this duplicated set for potential distinctive function.
Cellular characterization of X. laevis flt3 homoeologs
To investigate the cellular expression of the two X. laevis Flt3 homoeologs, we generated and produced tagged recombinant X. laevis Flt3lg.S and Flt3lg.L (rFlt3lg.S and rFlt3lg.L) in an insect expression system. Whereas partially purified rFlt3lg.L migrated as a unique band with an apparent molecular mass of 35 kDa on SDS-PAGE electrophoresis, rFlt3lg.S exhibited multiple bands of ∼35 kDa, suggesting differences in posttranslational modifications (Fig. 6A). Based on the typical strong binding affinity of these ligands to their respective receptors, we used the tagged recombinant molecules to characterize cells expressing Flt3 at the cell surface (Flt3+ cells) by flow cytometry and, after cell sorting, by RT-qPCR.
Flow cytometry analysis on X. laevis adult splenocytes with rFlt3lg.S and rFlt3lg.L, revealed that similar cell populations with comparable staining intensity were stained by each ligand (16.92 ± 9.33% for rFlt3lg.S versus 17.83 ± 10.24 for rFlt3lg.L+; (Fig. 6B). In comparison, the tagged recombinant Csf1r.S ligand XlrCsf1.S produced in the same insect expression system stained only a minor fraction of the splenocytes from the same frogs (3.68 ± 1.39%; (Fig. 6F, Supplemental Fig. 2). To substantiate these results, Flt3+ cells costained with the X. laevis–specific anti-CD8 mAb were sorted for RT-qPCR analysis (Fig. 6C). Confirming the specificity of the staining, sorted rFlt3lg.S+ and rFlt3lg.L+ cells were markedly enriched in cells expressing high levels of flt3.S and flt3.L transcripts (Fig. 6D). It is noteworthy that significantly higher flt3.L transcript levels than for flt3.S were detected in the different splenic cell populations (Supplemental Fig. 3A). The binding of rFlt3lg.S and rFlt3lg.L to Flt3 was further validated by the rapid and transient phosphorylation of Erk1/2 triggered by the incubation of X. laevis splenocytes with rFlt3lg.S and rFlt3lg.L (Fig. 7).
Sorted rFlt3lg+/CD8+ splenocytes were found to express high levels of DC-SCRIPT transcript (znf366.L), as well as xcr1.L, spi1.S/L, cxcl13.S/L, and ccl19.S transcripts (Fig. 6D). These gene products are hallmarks of DC-like cells in Xenopus and chickens, as well as cDCs and FDCs in mammals (10, 18, 44–46). Furthermore, sorted rFlt3lg+ cells expressed low transcript levels of the T cell marker cd3g.S (Supplemental Fig. 3B). Multiparametric flow cytometry analysis using different X. laevis–specific mAbs with rFlt3lg ligands identified a cell subset corresponding to a recently described dual FDC/DC subset named XL cells in X. laevis (10). This subset was characterized MHC-IIbright (Fig. 6E) with a higher size and complexity (i.e., forward scatter and side scatter) as reported for XL cells (Supplemental Fig. 2). This cell subset was also positively stained by anti-IgM and IgY mAbs as previously reported for XL cells (Fig. 6F, Supplemental Fig. 2). CD8 costaining further indicated that this putative XL cell population was CD8+ but at a lower level than CD8 T cells (Fig. 6F, Supplemental Fig. 2). RT-qPCR analysis did not reveal significant differences of cd8a.L or cd8b.L transcript levels between rFlt3lg+CD8+ and rFlt3lg+ sorted cells (Supplemental Fig. 3B). To further characterize these putative XL cells and distinguish them from macrophages, splenocytes were costained with rCsf1.S, which stains Csf1r.S+ macrophages (39). We detected a subpopulation (0.65 ± 0.27% SD) of total splenocytes that was Csf1r.S+, MHC-II+ intermediate and CD8dim. Finally, Flt3 ligands also stained a fraction of B cells as shown by high ighm.L transcripts levels in sorted rFlt3lg.S and rFlt3lg.L+ cells, as well as cell surface-positive IgM costaining (Fig. 6D, 6E). The frequency of IgM+ B cells costained with rFlt3lg.S and rFlt3lg.L varied considerably among the frogs tested (59.04 ± 18.85% and 57.30 ± 20.38%, respectively).
In the present study, we were able to extend the identification of flt3lg orthologs in conserved syntenic genomic regions to cartilaginous fish and lobed finned fish (Sarcopterygii), but not ray-finned fish (Actinopterygii). This suggests that this gene and the genomic locus have undergone substantial alteration during fish evolution. The multiple inversions, a sign of recombination events, observed in the flt3lg syntenic regions as well as the amino acid sequence divergences across the different taxa examined are consistent with this idea (47). Additionally, we observed that the putative binding domain of Flt3lg on Flt3 considerably diverged across vertebrates, especially in lobed and ray-finned fish. The additional whole-genome duplication that occurred during the emergence of ray-finned fish 350 MYA probably increased the evolution rate of the Flt3lg gene and the genomic locus (recombination and mutation), preventing the identification of any putative Flt3lg-like gene homologs in these taxa using current BLAST- and synteny-based searches.
In contrast to teleost fish, two sets of Flt3/Flt3lg genes have been retained in X. laevis subsequently to the whole-genome duplication that occurred 17–18 MYA. The corresponding syntenic regions of L and S chromosomes of X. laevis, and of X. tropicalis, are conserved with humans. Comparison of the deduced primary structures indicates that X. laevis gene homoeologs are more closely related than with the respective X. tropicalis homologs. This is consistent with the fact that the speciation divergence of X. laevis and X. tropicalis is estimated to have occurred 48 MYA, whereas the speciation resulting in L and S progenitors is estimated to date back 34 MYA (33). The latter event was followed by a hybridization step resulting in X. laevis allotetraploidization 17–18 MYA (33). Because of their more recent emergence, the primary structures of the S and L homoeologs are more closely related than with the primary structure of X. tropicalis. However, analysis of the predicted tertiary structures suggests that the conformation of proteins encoded by genes on the S chromosome are more related to the X. tropicalis orthologs, which would imply that allotetraploidization has allowed flt3.L and flt3lg.L to diverge, coevolve, and subfunctionalize from their S counterpart. In this regard, only the Flt3lg.S homeolog shows a different degree of posttranslational modification, which is likely important for protein conformation and function. In comparison, the recombinant human Flt3lg shows signs of N- and O-glycosylation (48). Considering the Flt3/Flt3lg cell signaling pathway, the apparent higher capacity of Flt3lg.S to trigger p-Erk1/2 might be due to different expression levels in the insect expression system or different purities. From tadpoles to frogs, the major site for hematopoiesis is located in the subcapsular region of the liver where erythropoiesis also takes place (49). In contrast to mammals, the X. laevis bone marrow, which develops after metamorphosis, appears to mainly harbor the final steps of myelopoiesis (40, 50). Considering lymphopoiesis, although the role of thymus in T cell development is well known, the site of common lymphoid progenitors and B cell lymphopoiesis remained to be fully investigated in adults and tadpoles (51, 52). Although histological studies in adults have shown that bone marrow hematopoietic cells are mostly represented by myeloid progenitors, low levels of rag1 and rag2 transcripts have been detected (53–55). In tadpoles, B cells have been suggested to have developed in the liver (51). Gene expression analysis revealed that in adult frogs flt3lg.L is significantly more expressed by nonstromal bone marrow cells than its counterpart from the S chromosome, whereas flt3.L is more expressed than the S homoeologs in the liver of adults. In contrast, in tadpoles flt3lg.S is more expressed than the L homeolog in the liver. We interpret these data as evidence of subfunctionalization where Flt3/Flt3lg.L gene homoeologs may play a more important role in adult hematopoiesis, whereas Flt3/Flt3lg.S gene homoeologs may have a preponderant role in the larval hematopoiesis. Additionally, the higher expression of Flt3/Flt3lg.L gene homoeologs in the tadpole and adult spleens as well as in adult splenocytes suggest a preponderant role in immune cell homeostasis. Interestingly, Flt3 and its ligand Flt3lg are also involved in mammalian neuron function and may play a role in the development of the nervous system (56, 57). Although Flt3 and Flt3lg genes are also expressed in the brain of X. laevis, the higher flt3.S transcript levels compared with the L counterpart in the brain of X. laevis tadpoles could imply a preferential role of this L homeolog during neural development in X. laevis. Although the distinctive differential expression of the two sets of Flt3/Flt3lg genes is consistent with a subfunctionalization, it is noteworthy that in vitro, rFlt3lg.S and rFlt3lg.L bind to both Flt3.S and Flt3.L, which implies that these homeolog genes still perform a partial overlapping function.
Outside of mammals, little is still known about Flt3lg/Flt3 function, particularly regarding DC biology. To fill this knowledge gap, we used X. laevis recombinant tagged Flt3lgs as reagents to identify and characterize immune cell population expressing Flt3 at their surface. Notably, we found that XL cells were brightly stained with rFlt3lg.S and rFlt3lg.L, indicating high levels of Flt3 surface expression by these cells, which is reminiscent of the mammalian cDCs and chicken DCs (19, 58). In addition to the high expression of surface MHC-II, expression of Flt3 on XL cells was confirmed by the enrichment with the Flt3lg staining of cells highly expressing XL cell markers, that is, mammalian FDC marker (cxcl13) or cDC markers (spi1, ccl19). The additional detection of surface Csf1r.S expression by these cells reinforces previous work suggesting the dual hematopoietic and myeloid lineage of XL cells (10). It is noteworthy that mice and chicken cDCs are also Csf1r+ (19, 59). The staining of XL cells with anti-IgM is likely due to X. laevis IgM bound to Fcamr, which is expressed at the XL cell surface, although the Fcr responsible for the IgY binding has not been identified to date (60). Our gene expression analysis suggests that XL cells express high levels of DC-SCRIPT (znf366.L) as in human cDC1s, mouse cDC1s, and fish DC-like cells (17, 61–63), as well as xcr1.L also produced by human and mouse cDC1s and chicken DC-like cells (18, 19, 64). Finally, our flow cytometry analysis shows that XL cells are CD8+ similar to cDC1s in mice (64). Based on these data we propose that XL cells and jawed vertebrate DCs are evolutionarily related and that ancient tetrapod DCs share common features of FDCs and cDCs.
Finally, the high expression of ighm.L transcript by Flt3+ and Flt3− splenocytes suggests that Flt3 is expressed by a subpopulation of IgM+ B cells in X. laevis as in mammals (65). In addition to its role in pro–B cell differentiation, Flt3 is a marker of B cell activation and proliferation (65–67). More specifically, Flt3 has been found to be required for B cell maturation including class-switch recombination to IgG (65). Nevertheless, the role of Flt3 in B cell activation and differentiation remains to be fully characterized.
We thank Tina Martin for the expert animal husbandry and Ruth Serra-Moreno for precious advice on the Western blot experiment.
This work was supported by National Institute of Allergy and Infectious Diseases Grants R21AI139718 and R24AI059830 from National Science Foundation Grant IOS-1754274.
The online version of this article contains supplemental material.
The authors have no financial conflicts of interest.