Stable multilineage xenogeneic replacement of definitive hematopoiesis in adult zebrafish

Bony fishes are the most numerous and phenotypically diverse group of vertebrates inhabiting our planet, making them an ideal target for identifying general principles of tissue development and function. However, lack of suitable experimental platforms prevents the exploitation of this rich source of natural phenotypic variation. Here, we use a zebrafish strain lacking definitive hematopoiesis for interspecific analysis of hematopoietic cell development. Without conditioning prior to transplantation, hematopoietic progenitor cells from goldfish stably engraft in adult zebrafish homozygous for the c-mybI181N mutation. However, in competitive repopulation experiments, zebrafish hematopoietic cells exhibit an advantage over their goldfish counterparts, possibly owing to subtle species-specific functional differences in hematopoietic microenvironments resulting from over 100 million years of independent evolution. Thus, our unique animal model provides an unprecedented opportunity to genetically and functionally disentangle universal and species-specific contributions of the microenvironment to hematopoietic progenitor cell maintenance and development.

identifying the molecular pathways regulating the number and maintenance of HSCs (for a recent example, see 12 ). However, while the zebrafish is clearly suitable for carrying out large-scale screens, it is not clear whether potential biological differences between the hematopoietic cells of fish and mammals may hinder the direct translational application of such results. Hence, it is important to explore the possibility of carrying out transplantations of xenogeneic HSCs into a zebrafish background with a view to generating an in vivo complement of existing in vitro models. As a first step in this direction, we explored the possibility of establishing a xenogeneic hematopoietic system in zebrafish. Our results indicate that the hematopoietic environment of the zebrafish supports the maintenance of xenogenic hematopoietic progenitor cells, opening up unprecedented opportunities for comparative studies on vertebrate hematopoiesis.

Results
Transplantation of goldfish whole kidney marrow into c-myb mutant zebrafish. In a first set of experiments, we compared the extent of hematopoietic reconstitution after transplantation of zebrafish or goldfish cells over a period of several weeks (Fig. 1A). Following transplantation of approximately 4 × 10 5 zebrafish kidney marrow cells per 6-week-old adult c-myb mutant recipient 10 , hematopoietic reconstitution was readily observable after two weeks, as described previously 10 (see also Supplementary Fig. 4A). After transfer of approximately the same number of goldfish cells (3.3 ± 0.9 × 10 5 ; n = 7), evidence of reconstitution became apparent at a later time point (Fig. 1B); nonetheless, in week 4 after transplantation, c-myb mutant recipients exhibited a reddish complexion (Fig. 1C), had lost their heart edema as a sign of reversal of anemia, and, as a consequence of improved tissue oxygenation, could swim much faster than their unmanipulated mutant siblings. When the cellular composition of the head kidney marrow was analyzed, the first distinct signs of repopulation were also evident at 4 weeks, and reconstitution was essentially complete at 6 weeks after transplantation (Fig. 1B). Of note, a shift in the relative distributions of different cell types -identifiable by their distinct forward and side light scatter characteristics -was evident; the most obvious change was the paucity of cells in the myelomonocytic gate (marked in green in Fig. 1B) relative to the original cell composition (Fig. 1B). Pulse-labelling with the nucleoside analogue EdU five weeks after transplantation (Fig. 1D) indicated that vigorous proliferation of hematopoietic cells occurred in the c-myb mutant recipients after transplantation of both zebrafish and goldfish cells, but not in unmanipulated c-myb mutant animals (Fig. 1E). Proliferation occurred primarily in progenitor and lymphocyte-like cells.
No reactivation of endogenous hematopoiesis after transplantation. Next, we investigated whether the presence of wild type hematopoietic cells might stimulate c-myb mutant progenitor cells to undergo normal hematopoietic maturation, contributing to reconstitution after transplantation of goldfish hematopoietic cells. To exclude this possibility, we generated c-myb mutant fish additionally transgenic for gata1:DsRed 13 and ikaros:eGFP 14 reporters; the former transgene marks red blood cells, whereas the latter directs GFP expression in hematopoietic precursors and lymphoid cells. In gata1:DsRed; ikaros:eGFP double-transgenic c-myb wild type fish, numerous green fluorescent cells are clearly visible at 1 day after fertilization (dpf), whereas fewer fluorescent cells could be detected in c-myb mutants; this is compatible with the observation of failing definitive hematopoiesis in c-myb mutants 10 ( Supplementary Fig. 1A). At the same time point, red-fluorescent cells were detectable in both wild type and mutant fish, compatible with the presence of unimpaired primitive hematopoiesis 10 ( Supplementary Fig. 1A). In older fish, expression of gata1:DsRed and ikaros:eGFP reporters was confined to c-myb wild type fish, but absent in c-myb mutants; importantly, cellular fluorescence was not observed after transplantation of wild type zebrafish kidney marrow cells (Supplementary Fig. 1B and 1C). After the hematopoietic system had been fully reconstituted ( Supplementary Fig. 1D), we used flow cytometry to examine the whole kidney marrow for the presence of red or green (i.e. mutant host) cells; however, none could be detected (Supplementary Fig. 1E-G). In adult gata1:DsRed and ikaros:eGFP wild type fish, gata1 expression is particularly strong in the progenitor population, whereas ikaros expression predominates in cells with the light scatter characteristics of lymphocytes ( Supplementary Fig. 1E). As expected, these cell populations are absent in adult c-myb mutants ( Supplementary Fig. 1F). Similarly, no such cells could be found in c-myb mutants fully reconstituted with non-transgenic wild type kidney marrow cells ( Supplementary Fig. 1G). These observations indicate that the c-myb mutation results in cell-intrinsically defective hematopoietic stem cell (HSC) maintenance and/or differentiation, preventing their contribution to hematopoiesis even after transplantation with wild type donor tissue. mutant transplanted with goldfish whole kidney marrow cells (gf > c-myb −/− ). The different cell populations in zebrafish kidney are identified by their forward (size) and side (granularity) light scatter characteristics: red, erythrocytes; blue, lymphocytes; green, myelomonocytic cells; magenta, progenitor cells. (C) Macroscopic representation of kidney of wild type zebrafish (left), zebrafish mutants (middle), and zebrafish mutants reconstituted with goldfish hematopoietic cells (4 weeks after transplantation; right), all 10 weeks old. The locations of the head kidneys are indicated by ovals. (D) Schematic outline of proliferation assay. (E) Flow cytometric analysis of whole kidney marrow cells isolated from fish of the indicated genotypes pulse-labelled with EdU, creating a red-fluorescent signal in cells. No signal is observed in control mock-pulsed cells (two leftmost columns). Proliferating cells are detectable only in wild type zebrafish kidney marrow and in mutant zebrafish reconstituted with either zebrafish or goldfish cells, but not in unmanipulated zebrafish mutants (third column from the left). The light scatter characteristics of EdU-positive cells (red) are superimposed on EdUnegative cells (blue) (second and fourth columns from the left). Representative data of corresponding biological replicates are shown (zebrafish wild type, n = 7 From an experimental point of view, this observation indicates that reconstituted hematopoietic and immune systems will consist entirely of descendants of transplanted HSCs.

Multilineage hematopoietic reconstitution by goldfish cells.
In order to determine the relative contribution of goldfish cells in cell suspensions prepared from whole kidney marrow, we developed a simple PCR-based DNA typing assay; a small fragment of the highly conserved activin beta A genes of zebrafish and goldfish could be amplified using the same primer pair and readily distinguished by presence (goldfish) or absence (zebrafish) of an AvaII restriction site ( Fig. 2A; Supplementary Fig. 2A). The results of this assay suggested the presence of goldfish cells in the repopulated zebrafish kidney marrow (Fig. 2B). We carried out detailed  Fig. 2A). Amplicons are shown before and after AvaII digestion of wild type zebrafish (zf wt), c-myb mutant zebrafish (zf −/− ), and goldfish (gf). The AvaII digest of an amplicon generated from a 1+ 1 mixture of wild type zebrafish and goldfish cells is also shown on the right for reference. (B) Genotyping of whole kidney marrow cells of c-myb mutant zebrafish reconstituted with goldfish whole kidney marrow cells (gf > c-myb −/− ) at several time points (in weeks) after transplantation. (C) Histological analysis of reconstituted c-myb mutant animals. Sections of the thymus (top panels), kidney (middle panels), and heart (bottom panels) are shown; insets represent 4× higher magnifications (haematoxylin/eosin staining) with the origin of selected regions indicated by broken squares (see also Supplementary Fig. 3). The fourth column indicates the results of RNA in situ hybridization using a rag1 probe, detecting lymphoid progenitor cells actively rearranging antigen receptor loci. Wild type zebrafish (zf wt), c-myb mutant (c-myb −/− ), and c-myb mutant transplanted with goldfish whole kidney marrow cells 6 weeks after transplantation (gf > c-myb −/− ). All fish are 12 weeks old. Anatomical landmarks are indicated (G, gills; T, thymus; K, kidney; H, heart; A, atrium of the heart; V, ventricle of the heart). Scale bars, 50 μ m.
histological analyses to provide further evidence for the presence of ongoing hematopoiesis in c-myb mutants after transplantation of goldfish cells ( Fig. 2C; Supplementary Fig. 3). Notably, our analyses demonstrated colonization of the thymus with strong rag1 expression indicative of ongoing T cell development; similarly, B cell poiesis (as revealed by rag1 expression in the kidney marrow) is readily detectable after transplantation (Fig. 2C). Finally, the pleomorphic appearance of peripheral blood cells (including red blood cells, lymphocytes and myeloid cells) was indicative of multilineage reconstitution ( Fig. 2C; Supplementary Fig. 3). Cytological analysis indicated the presence of cells with the characteristically larger size of goldfish cells (Supplementary Fig. 2B); this size difference is also evident in the flow cytometric profiles (see Fig. 1B). Collectively, these results strongly suggest that transplantation of goldfish kidney marrow cells results in multilineage hematopoietic reconstitution of adult c-myb mutant zebrafish.
Serial transplantation of xenogeneic hematopoietic progenitor cells. Next, we assessed whether the hematopoietic system in c-myb mutants originating from goldfish cells would also support reconstitution after a second transplantation. To this end, we transferred all kidney marrow cells from one primary recipient to a secondary host (Fig. 3A). We established that recovery of hematopoiesis in secondary recipients took place with similar kinetics to those during primary reconstitutions; notably, the proportions of different cell lineages in the kidney are indistinguishable from those of the recipients of primary transplants (Fig. 3B-D). Moreover, the repopulated thymi displayed ongoing T cell development, as indicated by rag1 expression (Fig. 3E,F). Collectively, multilineage reconstitution after secondary transplantation suggests that progenitor cells survive in the zebrafish kidney marrow environment.

Interspecific competition of hematopoietic cells. The above experiments clearly indicate that under
non-competitive conditions, goldfish hematopoiesis can be maintained in the zebrafish kidney marrow environment. However, the perturbed ratios of different cell types (i.e., lymphocyte vs. myelomonocytic cells) indicate possible species-specific differences in microenvironmental cues influencing the quantitative outcome of transplantation. If so, then such differences should become more pronounced in situations of competitive repopulation. To illustrate this potential use of the c-myb transplantation model, we carried out reconstitution experiments using transplants consisting of different ratios of zebrafish and goldfish cells. To monitor the kinetics of the reconstitution process, we employed zebrafish cells carrying the ikaros:eGFP transgenic marker, which is expressed in lymphoid and myelomonocytic cells. When we used a 1+ 1 mixture of zebrafish and goldfish whole kidney marrow cells, we established that hematopoietic reconstitution was eventually dominated by descendants of zebrafish HSCs, as indicated by the diagnostic proportion of GFP-positive lymphocytes (Supplementary Fig. 4). However, it is evident from a comparison of pure zebrafish transplants (Supplementary Fig. 4A) with the 1+ 1 mixtures (Supplementary Fig. 4B) that zebrafish hematopoiesis becomes established at a later time point and that goldfish myelomonocytic cells are at a disadvantage as compared to their zebrafish counterparts. Whereas zebrafish myelomonocytic cells had essentially outcompeted their goldfish counterparts 4 weeks after transplantation, lymphopoiesis was dominated by zebrafish cells only after 6 weeks ( Supplementary Fig. 4C). When a 1+ 9 mixture of zebrafish and goldfish kidney marrow cells was used, reconstitution by zebrafish cells was even more delayed (Fig. 4A,B). However, goldfish cells proliferate alongside zebrafish cells in the initial phases of reconstitution, as indicated by the composition of the respective genomic DNA of kidney marrow cells (Fig. 4C). This observation argues against an immune-mediated depletion of either cell type and rather supports the notion that direct competition for niche space and/or growth and inductive factor activities contribute to the distorted outcome of chimeric reconstitution.

Discussion
Maintenance of HSCs and their differentiation into the various blood cell lineages is regulated by a plethora of extrinsic cues such as growth factors and cytokines, some of which are known to be particularly species-specific in their action. In this context, it is surprising to find that hematopoietic reconstitution from transferred goldfish cells occurs in the zebrafish microenvironment. However, the relative paucity of myeloid cells in the xenotransplants relative to their frequency in the goldfish kidney marrow is one indication that the conditions for differentiation and/or survival of goldfish myeloid cells provided by the zebrafish microenvironment are not optimal. Nonetheless, our results demonstrate a surprising degree of functional conservation of hematopoietic supportive factors in species that have evolved independently for more than 100 Mya and that inhabit very different ecological niches. More specifically, our results predict that the homeostatic chemokines and cytokines supporting niche occupancy, colonization of primary hematopoietic (kidney) and lymphoid (kidney, thymus) organs and also differentiation of the major cell lineages are functionally compatible with the respective goldfish receptors. Our findings therefore provide important information concerning the molecular evolution of intercellular communication systems once the appropriate genomic and transcriptomic resources have been developed for the goldfish. Collectively, our observations suggest that the c-myb I181N transplantation model described here could be developed for the study of hematopoietic systems of other fish species, facilitating comparative analyses of their cellular and functional characteristics. Our success bodes well for future attempts at transplanting into c-myb I181N zebrafish hematopoietic cells from more distantly related organisms, such as jawless fishes and mammals, including humans.
So far, transplantation of non-tumorigenic xenogeneic cells into zebrafish was mostly performed using embryonic hosts to avoid the complications arising from immunological incompatibility (for instance, see 15 ). Alternatively, hosts with compromised immune functions were used (for instance, see 16,17 ). By contrast, the c-myb I181N transplantation model does not require conditioning prior to transplantation. It is possible that the unmanipulated microenvironment contributes to the robust and stable engraftment of allogeneic 9 and xenogeneic (this paper) hematopoietic cells. One interesting potential application of our model is the definition of the relative contribution of individual microenvironmental factors underlying the species-specific differences observed here. This could be done for instance by replacing zebrafish genes, individually or in combination, with orthologous sequences from other fish species or even more distantly related vertebrates; a second, but not mutually exclusive, alternative is the use of pharmacological or genetic screens to examine possible epistatic effects of the many different hematopoietic factors. Moreover, it should be possible, without having to consider interference by hematopoietic cells of the host, to examine the relative fitness of co-transplanted hematopoietic cell types, each differing in particular functionalities. Similarly to what was previously demonstrated for immunocompromised zebrafish hosts, our model could also be used for the transplantation of allogeneic and xenogeneic 16,18,19 tumour cells to facilitate studies on their general biological properties and drug sensitivity profiles.

Methods
Animals. Zebrafish (D. rerio) wild type strains (TLEK and TL), the transgenic ikaros:eGFP 14 , the transgenic gata1:DsRed 13 (a generous gift of Dr. T. Schwerte, University of Innsbruck, Austria) and the mutant c-myb t25127 10 lines are bred at the Max Planck Institute of Immunobiology and Epigenetics. All procedures were conducted in accordance with institutional guidelines under a license (Az 35-9185.81/G-12/85) and approved by the local government (Regierungspräsidium Freiburg, Germany). In the past, homozygous c-myb mutants were housed in groups of 20 fish in separate tanks; recently, we have found that they can be maintained more easily by co-housing them with sentinel wild type fish, such as slc45a2 albino fish 20 . In some experiments, transparent 21 c-myb mutants were used (a generous gift of Dr. L. Zon, Harvard University).
Transplantation procedure. Whole kidney marrow (WKM) cells from adult fish were harvested as described 9 and finally resuspended at a concentration of 20,000 to 40,000 cells/μ l in 0.9 × PBS supplemented with 5%FCS; cells were injected retro-orbitally in a volume of 10 μ l per animal. After transplantation, zebrafish were kept at a water temperature of 22 °C (room temperature) instead of the usual 28.5 °C.
Flow cytometric analysis. Flow cytometric analysis was carried out as described; the identities of cell populations were assigned according to their light scatter characteristics 22 . Profiles were recorded with an Aria II instrument, with the exception of the data shown in Supplementary Fig. 1; in this case an AriaFusion instrument was used.
Fluorescence microscopy. Fluorescence microscopy was carried out as described 9,14 . Proliferation assay. To detect proliferating cells, 7 μ g of 5-ethynyl-2′ -deoxyuridine (EdU) in 10μ l PBS solution (Click-iT ® EdU Flow Cytometry Assay Kits, Invitrogen Molecular Probes ® ) was injected retro-orbitally into wild type zebrafish, c-myb mutants and c-myb mutants carrying zebrafish and goldfish transplants, 4 weeks after primary transplantation. The analysis was performed 14-16 hours after EdU treatment. Staining was based on a click reaction using the Alexa Fluor 647 ® dye and performed according to the manufacturer's instructions.
Genotyping. Genotyping for the activin beta A genes was carried out using primers 5′ -GGATTATTTTGGACTGGCTCC and 5′ -TTCAGATGCAGCATGTTGAGG; these bind to sequences that are identical for both zebrafish and goldfish orthologs and amplify fragments of almost identical size (zebrafish, nucleotides 248-598 in Genbank accession number BC129208; goldfish, nucleotides 134-479 in Genbank accession number AF169032) (Supplementary Fig. 2A). Amplicons were digested with AvaII (New England Biolabs) according to the manufacturer's instructions and the products resolved on agarose gels. Genotyping for the presence of the gata1:DsRed transgene was carried out with primers 5′ -CAAGGAGTTCATGCGCTTCA and 5′ -TTCACGCCGATGAACTTCAC (amplicon size 360 bp); genotyping for the presence of the ikaros:eGFP transgene was carried out with primers 5′ -TGGTGCCCATCCTGGTCGAG and 5′ -GTCCTCGATGTTGTGGCGGA (amplicon size 488 bp).