Isthmin 1 (ism1) is required for normal hematopoiesis in developing zebrafish

Hematopoiesis is an essential and highly regulated biological process that begins with hematopoietic stem cells (HSCs). In healthy organisms, HSCs are responsible for generating a multitude of mature blood cells every day, yet the molecular pathways that instruct HSCs to self-renew and differentiate into post-mitotic blood cells are not fully known. To understand these molecular pathways, we investigated novel genes expressed in hematopoietic-supportive cell lines from the zebrafish (Danio rerio), a model system increasingly utilized to uncover molecular pathways important in the development of other vertebrate species. We performed RNA sequencing of the transcriptome of three stromal cell lines derived from different stages of embryonic and adult zebrafish and identified hundreds of highly expressed transcripts. For our studies, we focused on isthmin 1 (ism1) due to its shared synteny with its human gene ortholog and because it is a secreted protein. To characterize ism1, we performed loss-of-function experiments to identify if mature blood cell production was disrupted. Myeloid and erythroid lineages were visualized and scored with transgenic zebrafish expressing lineage-specific markers. ism1 knockdown led to reduced numbers of neutrophils, macrophages, and erythrocytes. Analysis of clonal methylcellulose assays from ism1 morphants also showed a reduction in total hematopoietic stem and progenitor cells (HSPCs). Overall, we demonstrate that ism1 is required for normal generation of HSPCs and their downstream progeny during zebrafish hematopoiesis. Further investigation into ism1 and its importance in hematopoiesis may elucidate evolutionarily conserved processes in blood formation that can be further investigated for potential clinical utility.


Introduction
Hematopoiesis is an essential cellular process in which hematopoietic stem cells (HSCs) differentiate into the multitude of different cell lineages that comprise mature blood [1][2][3]. HSCs must self-renew and persist for an organism's lifespan to replenish the mature, post-mitotic blood cells that are constantly being recycled, ensuring that the system is never depleted [4,5]. The control of this recycling and replacement of blood cells is regulated by an intricate set of signaling molecules and molecular pathways, many of which are still enigmatic. Improper regulation of hematopoiesis can result in serious diseases such as anemia, thrombocytopenia, neutropenia, and leukemia, so understanding these signaling pathways is of clinical relevance.
HSPC differentiation is a developmentally restrictive process, controlled by a multitude of cytokines. These small, extracellular proteins influence HSPCs to self-renew and/or undergo stepwise differentiation into mature blood cell lineages and are secreted in hematopoietic niches, mainly by stromal cells that are found in hematopoietic-supportive tissues and organs (reviewed in [21][22][23]). These factors then bind to receptors on the surface of HSPCs to mediate a multitude of different downstream cellular responses. Identification and elucidation of the downstream molecular events activated by cytokines is of key interest due to their essential role in hematopoietic regulation; improper differentiation of HSPCs can lead to an accumulation of immature cells, causing the development of lymphoma and leukemia.
To study hematopoiesis and HSPC biology, many laboratories utilize Danio rerio (zebrafish), which have become a promising model system for many reasons (reviewed in [24,25]). First, they are the phylogenetically lowest vertebrate model system that has a similar circulatory and hematopoietic system to humans, including adaptive immunity (reviewed in [26]). Secondly, zebrafish are transparent and develop ex utero; within 48 hpf functional HSPCs are present [7,8,[27][28][29][30]. Due to the fact that a multitude of hematopoietic-specific fluorescent transgenic zebrafish lines currently exist (reviewed in [24,31]), HSPCs can be visualized, isolated, and studied in early embryos, a feat not possible in mammals. The fact that zebrafish are fecund, generating hundreds of embryos per clutch, allows large sample sizes and experimental replication. Finally, because zebrafish develop quickly and externally, they are excellent for mutagenesis studies [32][33][34][35][36] and screening compounds that have utility for treating human diseases [37][38][39][40][41]. For all of these reasons, the zebrafish has become a powerful model for studying normal hematopoiesis and its dysregulation during disease.
HSPCs, the cells mutated during the onset of hematopoietic disease, are not easily grown with traditional cell culture techniques because they require specific microenvironments and cytokine signals to keep them proliferating and to encourage their differentiation into mature blood cells. Studies performed in mice [42][43][44][45] and humans [46,47], isolating and growing HSPCs on stromal cell lines, were important for elucidating cytokines and other signaling molecules required for HSPC proliferation and differentiation. The recent generation of hematopoietic-supportive zebrafish stromal cell lines from sites of embryonic [48,49] and adult [50] hematopoiesis now allows functional testing of HSPCs in zebrafish. Putative hematopoieticsupportive factors expressed in zebrafish cell lines have been identified with RNA sequencing (RNA-seq), comparing the transcriptome of these cells to that of non-hematopoietic supportive zebrafish stroma [48]. Additionally, comparison of their transcriptome to mammalian hematopoietic-supportive stroma allows investigation of transcripts conserved throughout vertebrate evolution [51].
One interesting transcript uniquely upregulated in all hematopoietic-supportive zebrafish cell lines was isthmin 1 (ism1), a gene first identified in the midbrain-hindbrain boundary (MHB) of Xenopus [52]. ism1 is expressed in lymphocytes, [53] bone marrow [54], and in embryonic blood islands [52], and encodes a secreted 60 kDa protein containing a copy of the thrombospondin repeat (TSR) region, a domain involved in cell migration and tissue remodeling [55]. Importantly, it is also expressed in mouse and chicken lateral plate mesoderm, tissue which gives rise to blood in the developing embryo [54]. ism1 is also implicated in angiogenesis; addition of ISM1 protein into matrigel plugs with murine tumors resulted in decreased endothelial capillary networks and decreased overall tumor growth [55]. Additionally, ism1 morphant zebrafish exhibit decreased inter-segmental vessels (ISVs) [55]. Importantly, ism1 levels are increased in response to the upregulation of Wnt signaling [56], implicating its role in early embryonic processes such as cell fate specification, migration, and the beginning of definitive hematopoiesis. Finally, ism1 is co-expressed with fibroblast growth factor ligands that are essential for HSC specification during this developmental period [52]. ism1's high expression within hematopoietic-supportive stromal cell lines coupled with its expression during development in blood-forming tissues and co-expression with essential hematopoietic factors indicated that ism1 was potentially involved in the formation and modulation of HSPCs.
To understand ism1's role within developmental hematopoiesis, we performed loss-offunction experiments, which indicate that ism1 morphants have reduced mature erythroid and myeloid cells. Additionally, ism1 morphants show reduced numbers of HSPCs present in developing fish. Overall, these data indicate that ism1 is an important gene for the formation of the embryonic zebrafish hematopoietic system.

Zebrafish
Wildtype (AB) and transgenic zebrafish lines (gata1a:DsRed [57], mpx:EGFP [58], and kdrl: EGFP [59]) used in these studies were raised and maintained in accordance with California State University, Chico IACUC guidelines. All experiments were approved by the IACUC committee before being performed.

ism1 sequence read counts
All sequenced libraries were processed and analyzed as previously described [48].

Morpholino and ism1 mRNA injections
ism1 antisense morpholino (MO) was designed against the 5' untranslated region (UTR) and start codon to prevent translation of ism1 mRNA (Gene Tools, Philomath, OR). The MO sequence is as follows: 5'-CCAGACGCACCATCCTCTTCACC-3'. For microinjection into embryos, a mix of 8 μL of 7.6 mg/mL of ism1 MO was mixed with 0.6 μL of phenol red for a ism1's role in blood formation final concentration of 7.0 ng/nL of ism1 MO. 1 μL of this mix was loaded into a needle made with a PM102 micropipette puller (MicroData Instrument, Plainfield, NJ). Single-cell stage embryos were collected, placed onto a 1% agarose microinjection chamber plate with troughs, and injected with 1 nL (7.0 ng) of ism1 MO with a PM 1000 Cell Microinjector (MicroData Instrument, Plainfield, NJ). For rescue injections, phenol red was reduced to 0.2 μL, and 0.4 μL of 44.7 ng/μL ism1 mRNA was added. In this way, rescued embryos received 7.0 ng of ism1 MO and 17.88 ng of ism1 mRNA.

Microscopic visualization of the hematopoietic system
To discern hematopoietic phenotypes and to quantitate cell lineages, florescent microscopy was utilized. Transgenic zebrafish were visualized under a Leica M165C (Leica, Wetzlar, Germany) fluorescent dissecting microscope at time points correlated with the emergence of specific hematopoietic cell lineages. Erythrocytes were visualized at 48 hpf with gata1a:DsRed transgenic animals and further examined by flow cytometry. Neutrophils and macrophages were visualized and individual cells were counted under the microscope at 48 hpf with mpx: GFP transgenic animals. Zebrafish and their fluorescently labeled cells were imaged using a Leica FireCam camera (Leica, Wetzlar, Germany), scored, and enumerated. For myeloid quantitation, images were labeled by a reference number and the numbers of mpx:GFP + cells were counted in each animal by several undergraduate students to insure no bias in results.

Flow cytometry
To enumerate the percentage of fluorescent cells in an embryo, we used transgenic zebrafish in combination with flow cytometry. 72 hpf transgenic embryos were grouped in samples of three and washed 3x with E3 [60]. After the last wash, the E3 was removed, leaving 100 μL. 1 mL of 10 mM dithiothreitol (DTT) in E3 was added, and samples were incubated for 25 mins. Samples were than washed 3x with Dulbecco's phosphate-buffered saline (DPBS) containing Ca 2+ and Mg 2+ . After the last wash, 500 μL of DPBS and 5 μL of 5 mg/mL (26U/mL) Liberase TM (Roche, Upper Bavaria, Germany) were added. Samples were incubated at 37˚C on a horizontal orbital shaker at 180 rpm for 60 mins. Samples were than triturated with a P-1000 to ensure proper dissociation and transferred to a 5 mL polystyrene round bottom tube with cell strainer cap. 1 μL of SytoxRed (ThermoFisher Scientific, Waltham, MA) was added to each sample to mark dead cells. Samples were run through a BD Accuri C6 flow cytometer (BD Biosciences, San Jose, CA) and enumerated. Data were analyzed using FloJo software (FloJo LLC, Ashland, Oregon) to quantitate total percentage of positive fluorescent cells.

Quantitation of HSPCs in developing zebrafish embryos
HSPC isolation and culture was performed as previously described [61]. Samples were given carp serum, Gcsf, and Epo to stimulate myeloid and erythroid differentiation [62,63]. They were incubated at 32˚C and 5% CO 2 for 7-10 days and imaged with an Olympus IX53 inverted microscope (Olympus, Center Valley, PA) at 40x to enumerate colony forming units (CFUs). generated with the iScript cDNA synthesis kit (Biorad, Hercules, CA), and PCR was performed with Jumpstart ReadyMix Taq (Sigma-Aldrich, St. Louis, MO).

Quantitative RT-PCR (qRT-PCR)
mRNA was extracted from embryos at 48 and 72 hpf using a Qiagen RNAeasy kit (Qiagen, Hilden, Germany). cDNA was then generated with the iScript cDNA synthesis kit (Biorad, Hercules, CA), and PCR was performed with SsoFast SYBR Mastermix (Biorad, Hercules, CA). Fold expression was measured as ΔΔCT using ef1a [28] as a reference gene and whole kidney cDNA as a reference tissue.

Statistics
Relative fold change was done by setting the control as the standard. For triplicates, fold change per experiments were averaged and plotted with standard deviation. To discern statistical difference, data were analyzed using an unpaired two-tailed Student's T test. Ã = p<0.05, ÃÃ = p<0.0001, and N.S. = no significance.

Results
Zebrafish kidney stroma (ZKS) cells [50], zebrafish embryonic stromal trunk (ZEST) cells [48], and caudal hematopoietic embryonic stromal tissue (CHEST) cells [49] are hematopoieticsupportive stromal cell lines isolated and grown from adult and developing zebrafish. These stromal cell lines are derived from different physical locations of adult hematopoietic maintenance [64] (ZKS cells), embryonic HSC emergence [7,8] [50]), zebrafish embryonic stromal trunk (ZEST, derived from the embryonic trunk; blue [48]), and caudal hematopoietic embryonic stromal tissue (CHEST, derived from the CHT; green [49]). (B) Each stromal cell line's transcriptome was analyzed with RNA-seq and compared to a non-hematopoietic supportive stromal cell line (ZF4); transcripts were plotted as a Venn diagram demonstrating 447 conserved transcripts. (C) ism1 transcript levels for each cell line presented as reads per kilobase of transcript per million reads mapped (RPKM). See Table 1 Fig 1A) and support hematopoietic proliferation and differentiation of all blood cell types in culture [27, 48-50, 66, 67]. Therefore, we hypothesized that these cells would generate and secrete proteins required for hematopoiesis. Total RNA from each stromal cell line was isolated, sequenced, and compared to a non-supportive hematopoietic cell line (zebrafish fibroblast cells; ZF4 [68]) to remove housekeeping genes from our analysis. RNA-seq of the supportive hematopoietic stromal cell lines identified 447 shared transcripts ( Fig 1B; top genes involved in hematopoiesis are listed in Table 1). Transcripts were analyzed as reads per kilobase of transcript per million reads mapped (RPKM), which normalized and accounted for transcript length. From these transcripts, ism1 was chosen for further research due to its high RPKM levels across all three cell lines (Fig 1C), the fact that it is a secreted protein, and because it has a conserved ortholog in humans. To confirm that ism1 was expressed in ZKS, ZEST, and CHEST stromal cells, we performed RT-PCR for ism1, validating its expression in these tissues (S1 Fig). We also observed ism1 expression in whole kidney, the main site of hematopoiesis in the adult zebrafish [64]. To confirm that ism1 was only expressed in the stromal cells of the kidney, we fractionated cells from WKM using FACS; myeloid, lymphoid, and precursor cell populations can be easily separated based on their size and granularity [57]. RT-PCR analysis of these cells confirmed that ism1 is expressed in the supportive hematopoietic tissues, but not in hematopoietic cells themselves (S1 Fig).

ism1 morphants exhibit reduced erythrocyte production
To elucidate ism1's role in erythroid cell production, gata1a:DsRed transgenic zebrafish were used; Gata1 is a transcription factor expressed in erythrocytes, allowing for real time visualization of red blood cells in developing embryos. ism1 knockdown with a specific MO caused slowed circulation and caused blood to pool in gata1a:DsRed + zebrafish (Fig 2A) when examined at 48 hpf. The control group expressed no abnormalities with respect to blood circulation, compared to 38% of the ism1 morphants. The hearts of embryos injected with ism1 MO were pumping and appeared morphologically normal. Additionally, no blood was observed leaking from blood vessels. Injection of ism1 mRNA reduced the amount of pooled blood seen in ism1 morphants and increased the speed of erythrocyte circulation (Fig 2B).
Regardless of their circulation phenotypes, we speculated that the morphant embryos also had decreased numbers of erythrocytes. To quantitate total red blood cell numbers, flow cytometry was utilized. All ism1 MO injected embryos were analyzed without grouping them based on their circulation phenotype. Three individual 72 hpf gata1a:DsRed + ism1 MO injected zebrafish were randomly grouped together, enzymatically dissociated in a sample tube, and analyzed on a flow cytometer to quantitate gata1a:DsRed + cells. Percentages of gata1a:DsRed + cells were normalized to uninjected control embryos and expressed as fold change over control. ism1 knockdown caused an approximate 2-fold decrease in gata1:DsRed + cells compared to control ( Fig 2C) and the rescued group displayed erythrocyte levels similar to control groups (Fig 2C), indicating that ism1 depletion was the cause of the erythrocyte defect. Additionally, ism1 MO injected whole embryos were pooled into groups of 10, digested, and qRT-PCR was used to analyze alpha globin (hbaa) mRNA levels, a specific marker of differentiated erythrocytes. qRT-PCR demonstrated a 29-fold reduction of hbaa mRNA levels in ism1 MO injected animals ( Fig 2D). Taken together, these data indicate that ism1 reduction causes a decrease in embryonic erythrocytes and a decrease in hbaa mRNA levels in the whole organism.
While intersegmental vessels (ISVs) are not involved in blood development, previous studies indicated that ism1 was necessary for proper development of ISVs in zebrafish [55]. To confirm that ism1 was not necessary for the formation of the dorsal aorta, the source of HSC  [55]. However, the dorsal aorta was not negatively affected in morphants (S2 Fig), also in agreement with previous studies [55]. Although not previously described, the shortened ISVs recovered by 48 hpf without addition of exogenous ism1. In conclusion, ism1 morphants experience a reduction in ISVs that recovers by 48 hpf but have a normal dorsal aorta.

ism1 morphants exhibit reduced myeloid cells
Myeloid cells such as neutrophils and macrophages are cellular components critical for innate immune responses. To investigate ism1's role in myeloid cell biology, mpx:EGFP transgenic zebrafish were utilized. mpx:EGFP zebrafish embryos were injected with ism1 MO; all injected embryos were visualized and total mpx:EGFP + cells were individually enumerated at 48 hpf ( Fig 3A). ism1 MO injection resulted in a 20% reduction of mpx:EGFP + cells compared to control groups ( Fig 3B) and rescue with ism1 mRNA showed a recovery of mpx:EGFP + cells compared to ism1 MO injected embryos. qRT-PCR was performed on pools of 10 injected embryos for colony stimulating factor 3 receptor (csf3r; also called granulocyte colony stimulating factor receptor), and colony stimulating factor 1 receptor (csf1r; also called macrophage colony stimulating factor receptor), two markers for neutrophils and macrophages, respectively. qRT-PCR indicated a 11-fold reduction in csf3r mRNA levels and a 30% reduction of csf1r mRNA levels in ism1 MO injected embryos (Fig 3C). Overall, these data indicate that ism1 knockdown causes a reduction in mpx:EGFP + myeloid cells.

HSPCs are reduced in ism1 morphants
To confirm that ism1 loss was causing a reduction in HSPCs, groups of 10 MO-injected embryos at 48 hpf were placed in a tube, enzymatically digested, and plated in methylcellulose with exogenous hematopoietic-supportive growth factors [61]. After plating, the embryos were incubated for 7 days, allowing HSPCs to proliferate and differentiate (Fig 4A). Fish were not phenotypically selected before digestion; all injected fish were randomly pooled into groups of 10 and plated in methylcellulose. ism1 MO injected and rescued embryos generated hematopoietic colonies, which were enumerated. Colony forming units (CFUs) for ism1 MO injected zebrafish were reduced by 35% compared to control and rescue groups (Fig 4B). There was no statistical difference between control and rescue groups, indicating that ism1 is necessary for the generation of the proper numbers of HSPCs in developing zebrafish embryos. qRT-PCR on 48 hpf embryos was also performed to quantitate HSPC-related transcription factors. Data showed an overall 20-fold reduction in runx1, and a 10-fold reduction in cmyb mRNA levels between controls and ism1 MO injected embryos (Fig 4C). Overall, these data indicate that HSPC formation is reduced in ism1 morphants.

Discussion
In this study, we sought to elucidate molecular requirements for HSPC generation, support, and expansion. Utilizing previously described zebrafish hematopoietic-supportive stromal cell ism1's role in blood formation lines [48][49][50] we identified highly expressed transcripts determined by RNA-seq. ism1 was chosen from this list due to its involvement in angiogenesis [55,69], its presence in embryonic [52,54,55] and adult [53,54] hematopoietic tissue, the fact that it is a secreted protein [52,53,55], and because it shares synteny with human ISM1. Our study findings indicate that ism1 is required for proper embryonic hematopoiesis.
To elucidate ism1's role in hematopoiesis we performed loss-of-function experiments using MOs to generate ism1 morphants. Other MOs have been utilized to study ism1 [55], and our MO was directed towards a similar location in the genome 8bp 3' of Xiang et. al's translation- blocking MO; the location we chose was suggested by modeling software to be a more effective site for blocking ism1 translation. The MO we designed binds 10bp 5' to ism1's start site; as two splice variants are predicted for ism1, we wanted to ensure that the MO blocked all versions of Ism1 so we would not have multiple splice isoforms potentially confounding our results. Importantly, our MO and Xiang et. al's have the same phenotypic effect on ISV formation early in development. The dorsal aorta, the site of HSPC generation between 36-52 hpf [7,8], was unaffected (our data and [55]). We also observed that ISVs were only affected between 24-48 hpf, after which ISV branching was normal. To observe hematopoietic defects, we injected ism1 MOs into previously characterized transgenic fish lines that have fluorescent proteins expressed in specific hematopoietic cell lineages. ism1 morphants did not express any morphological defects such as bent tails, spinal deformation, or stunted growth (data not shown). However, they did have phenotypes in their hematopoietic tissues. Coupled with the fact that these phenotypes were rescued by the addition of exogenous ism1 mRNA indicates that Ism1 is a key protein involved in normal zebrafish hematopoiesis.
It is important to note that we are utilizing zebrafish as a model system to investigate the evolution of vertebrate hematopoiesis. While mammals are derived from lobe-finned fish and zebrafish are ray-finned fish (reviewed in [70]), zebrafish have proven to be an excellent model of hematopoiesis; their molecular control of blood formation and differentiation is well conserved with mammals [24][25][26]71]. Further evidence of this is that drugs identified in zebrafish are now being utilized in the clinic to treat human blood diseases [72]. One concern with using zebrafish as a model is that the MO may not be specific for only one copy of ism1. While the genome was duplicated early in teleost evolution [73,74], there is only one described copy of ism1 in the zebrafish genome; we found no additional copies in our scanning of genomic data and RNA-seq results, in agreement with others [55]. This one copy of zebrafish ism1 shares synteny with human ISM1; JAG1 (jag1b), BTBD3 (btbd3b), and SPTLC3 (sptlc3) are 5' to ISM1 (ism1), while TASP1 (tasp1), KIF16B (kif16bb), and FLRT3 (flrt3) are 3'. The only other genes in the zebrafish ism1 family are ism2a and ism2b. These two isthmin paralogs are located on different chromosomes (ism1 is on chromosome 13, ism2a is on 17, and ism2b is on 20) and share synteny with only the genes on the 5' side of human ISM2. Little is known about ism2a or ism2b's roles/functions in zebrafish. These genes share no synteny with ISM1 or ism1, indicating that ism2a and ism2b are likely different genes as opposed to duplicates of ism1. There are 9 ism1 orthologs in other fish species like cod, amazon molly, fugu, medaka, platyfish, spotted gar, stickleback, tetraodon, and tilapia, all of which only have one identified copy of ism1 in their genome. Overall, it appears that ism1 only has one identifiable copy in the zebrafish genome, is not mislabeled, and is similar to human ISM1.
To examine the role of ism1 in hematopoiesis, we first examined erythrocytes due to their abundance and importance during development; they are also the first hematopoietic cells to arise. gata1a:DsRed + ism1 morphants exhibited pooled blood and we observed what appeared to be slowed circulation. While some "pooled blood" phenotypes have been attributed to functional cardiac defects [75], no such phenotype was visualized in our experiments. This suggested that the morphant's vasculature might be disrupted, but our data, in agreement with prior studies [55], did not see a disruption of the dorsal aorta after ism1 knockdown. To further confirm that erythrocytes were specifically reduced, we enumerated them with flow cytometry and found a significant decrease in erythrocytes in ism1 morphants. Reduction of erythrocytes was further validated by qRT-PCR data showing reduction of the hbaa transcript, a hemoglobin gene. In conclusion, ism1 knockdown caused an overall reduction of erythrocytes and the phenotype was rescued by the addition of exogenous ism1 mRNA, indicating that ism1 is required for normal erythropoiesis. mpx:GFP zebrafish express GFP in both neutrophils and macrophages [58,63]; both are differentiated myeloid cell lineages. ism1 morphants had a 20% reduction in these myeloid cells, suggesting that myeloid cell lineages were not properly forming. We also saw a reduction in csf3r transcript, the signaling receptor necessary for granulocyte formation and differentiation, measured by qRT-PCR. While ism1 morphants exhibited an overall decrease in myeloid cells, these cell lineages were still generated, suggesting that ism1 isn't required for myeloid differentiation but is necessary for the generation or proliferation of myeloid HSPCs. Together, the reduction in both myeloid and erythroid cell lineages suggested a reduction of a common upstream progenitor cell. Therefore, ism1 is likely involved in the proliferation or differentiation of a multipotent progenitor such as the CMP or HSC.
Together, decreased production of differentiated erythroid and myeloid cell lineages indicated that ism1 knockdown somehow decreased the generation or proliferation of HSPCs during development. To investigate this possibility, we developed a surrogate assay to assess HSPC numbers in morphant zebrafish embryos [61]. This assay involved plating dissociated whole zebrafish embryos and adding exogenous factors that promote only HSPC differentiation and proliferation in vitro [62,63,76]. Plating HSPCs generates CFUs that arise from a single progenitor cell. Therefore, the number of CFUs counted after 7 days in culture corresponds to the number of total progenitor cells present in the organism at the time they were assayed. These results suggest that ism1 is transcribed, translated, and secreted by stromal cells during a window of HSPC formation in the embryo; reducing ism1 expression during this time reduced the number of HSPCs in the developing zebrafish. This non-cell-autonomous effect of ism1 is seen with the clonal methylcellulose assay-because less Ism1 is secreted by the hematopoietic stroma during development, less HSPCs are generated in the embryo. 48 hpf was chosen for these assays due to the fact that functional HSPCs are generated by this time in development [7,8,[27][28][29][30]. Our data, in agreement with previous studies [54], indicates that ism1 is expressed in the hematopoietic stroma, but not in hematopoietic cells themselves. These results reinforce that Ism1 is a secreted protein made in the stromal niche, and not produced by HSPCs themselves; HSPCs are present in the "precursor" [62,63,76,77] and "lymphoid" [29,57,62,63,76,77] fractions of blood cells analyzed in this study. Importantly, this was not a developmental delay; plating dissociated ism1 morphant embryos at later time points did not rescue the number of HSPC-derived colonies present (data not shown). Overall, the fact that HSPCs are reduced in ism1 morphants likely accounts for the decreased counts of differentiated erythroid and myeloid lineages, suggesting that ism1 is needed for normal levels of HSPC formation and differentiation during embryonic development. ism1 is extremely enigmatic, and many seemingly conflicting roles for this gene exist in the literature. ism1 was originally identified as a secreted protein expressed along with fibroblast growth factor (Fgf) 8 in the Xenopus midbrain-hindbrain organizer [52], and it has recently been described in craniofacial patterning in humans [78]. Importantly, it has been identified in multiple embryonic tissues, many of which are involved in various regions of embryonic development. Of interest for blood formation, ism1 was identified in Xenopus blood islands [52] and the lateral plate mesoderm of chicks and mice [54], which is tissue that gives rise to blood. It is also found in bone marrow [54], the site of adult hematopoiesis in mammals. While ism1 was identified in the supportive hematopoietic stroma and not actual blood cells (our data and [54]), other studies indicate Ism1 is expressed in specific subsets of mouse lymphoid cells, including T and natural killer (NK) cells [53]. Our data does not correlate with this; zebrafish T and NK cells are present in the "lymphoid" population of WKM [57], and we detected no ism1 in this fraction, indicating that this may be a difference between fish and mammals. Another difference across species is the description of ism1 as an angiogenesis inhibitor; in mice exogenous ISM1 suppressed tumor angiogenesis [55,69], but in zebrafish the knockdown of ism1 reduced the growth of intersegmental vessels [55]. While Ism1 plays a role in apoptosis, its role in this process is also complicated. It specifically reduces endothelial cell (EC) growth by inducing apoptosis [55,69], while it has little to no effect on the proliferation or survival of glioma cells [69], NIH3T3 fibroblasts [55], Swiss 3T3 fibroblasts [55], B16 melanoma cells [55], and hepatocellular carcinoma cells [55]. This differential apoptotic capacity is not just different across cell types; it is linked to the solubility of ISM1. Soluble ISM1 causes apoptosis, while extracellular matrix (ECM)-immobilized ISM1 promotes EC adhesion, migration, and survival [79]. Unfortunately, no zebrafish-specific Ism1 antibody exists, preventing analysis of this in our experiments. It is important to note that the role of Ism1-induced apoptosis of hematopoietic cells is a possibility that could also explain a reduction in blood cells; further investigation may shed light on this subject.
Overall, these findings demonstrate a new role for ism1; its involvement in de novo hematopoietic cell formation. ism1 reduction negatively impacted the number of HSPCs formed and their ability to properly generate mature erythrocytes, neutrophils, and macrophages. This could be due to ism1 affecting a variety of signaling pathways; we speculate that ism1 is involved in molecular pathways that instruct HSPC formation and differentiation, as its knockdown decreases various mature blood cell lineages. This is of keen interest, because ism1 is secreted by our hematopoietic-supportive cell lines and affects various cell lineages, yet it has only been identified in a handful of tissues [52][53][54][55]78] and lymphoid cells [53]. ism1 expression is present in Xenopus blood islands suggesting its role in generating primitive blood [52]. This expression also implicates its importance in definitive hematopoiesis, as primitive erythropoiesis expresses many of the same molecular factors as its definitive counterpart [80]. ism1 is also expressed at the same time and in the same location as Xenopus Fgf8 in the notochord [52]; it is also present in the zebrafish notochord [55]. FGF signaling is involved in hematopoiesis [81][82][83][84][85][86][87] as it is a mediator of Notch and Wnt signaling [83]. It is also modulated by BMP signaling [82]. All of these pathways are required for HSC production. Interestingly, ism1 is a Wnt target [56] that is expressed (and likely secreted from) the notochord [52,55]; this is the first time a secreted signal from the notochord has been implicated in hematopoietic development. Overall, these previous studies and our results indicate ism1 is signaling to induce formation and functionality of hematopoietic tissues in the developing zebrafish. Further investigation of ism1's role in this process may shed light on an important signaling pathway for blood formation and regulation.
Supporting information S1 Fig. ism1 is expressed in hematopoietic-supportive stromal cells but not in hematopoietic cells. RT-PCR was performed for ism1 from ZKS, ZEST, CHEST, and kidney mRNA. Myeloid, lymphoid, and precursor cell mRNA isolated from zebrafish kidney was also interrogated for ism1 transcripts.