(A) Phenylhydrazine treatment induces hemolytic anemia by lysis of erythrocytes. Gross visible signs of anemia can be seen in the fish at the time of sacrifice, as described in the text. Recovery of the anemia occurs over a 20 day time-course, documented visibly and by flow cytometry. (B) A group of wild-type fish (n=11) were treated with phenylhydrazine at day 0 and followed over a 20 day time course. Fish were sacrificed at days 2, 5, 12 and 20 following treatment and WKM prepared for flow cytometry. Data from matched sets of untreated fish was used to generate the numbers indicated by day 0. Erythrocytes and precursor populations were quantified at each time point. The line graph shows each as a percent of WKM (erythrocytes and precursors). As the erythrocyte count falls there is a reciprocal up-regulation of precursor production to compensate for the induced anemia.

WKM was collected from adult wild-type, pgy/+ and sbn/+ fish on the gata1:gfp background. Samples were analyzed by flow cytometry for FSC, SSC and GFP; dead cells were gated out by PI staining. Two erythroid populations were quantified: proerythroblasts and polychromatophilic erythroblasts. For each genotype, the mean percent of progenitors within WKM was graphed for erythrocytes. For proerythroblasts and polychromatophilic erythroblasts the mean percent graphed, indicated by Gata1+ cells on the axis, refers to the percent GFP positive in the “precursor” or “lymphocyte” gate respectively, see Fig. 1 for details. P values were calculated by two-tailed student's T-test by comparing wild-type to smad5 mutants for each cell type. Sample numbers for each were: wild-type (n=8), pgy/+ (n=5), sbn/+ (n=3). *P<0.02.

(A) The diagram represents the construct used to generate the gata1:Δbr-gfp transgenic line. A 5.4 kb promoter sequence was cloned upstream of a fusion construct encoding a truncated BMP receptor with a C-terminal GFP tag. (B) A representative example of a gata1:Δbr-gfp embryo at 22 hpf viewed under fluorescence. The embryo shows GFP expression (arrow) in the intermediate cell mass (ICM), which recapitulates the endogenous gata1 expression pattern. GFP expression is significantly down-regulated as the cells differentiate (not shown). (C) WKM from homozygous adult gata1:Δbr-gfp fish was analyzed by FACS along with WKM from wild-type (non-transgenic) fish. Samples show three distinct patterns for GFP expression. An overlay histogram of GFP expression is shown and the corresponding cell types listed, wild-type (blue), and gata1:Δbr-gfp (red). Wild-type WKM cells have no GFP expression. The gata1:Δbr-gfp WKM shows a different profile of GFP expression in mature erythrocytes compared to gata1:gfp adults, because the fusion protein is down-regulated as erythroid cells differentiate. GFP is expressed at a low level in the progenitor cells of gata1:Δbr-gfp, while very low levels of GFP are expressed in the mature erythrocytes. These cell types were confirmed by sorting and cytospins (data not shown). Although not shown here (see Fig. 1B) WKM from gata1:gfp adults shows high GFP expression in mature erythrocytes and low GFP expression in the progenitors.

Gata1 (GFP+) progenitors were isolated from wild-type (gata1:gfp) adult WKM, and a corresponding population of Gata1+ progenitors was isolated from gata1:Δbr-gfp adult WKM by FACS, with DAPI-based exclusion of dead cells. Total RNA was isolated from these cells and cDNA synthesized. QPCR was performed for analyzing expression levels of RNA from c/ebpα, pu.1 and gata1. The data was processed using the 2^-ΔΔCT method [26]. The median of each sample was normalized to wild-type (gata1:gfp) and the median average was graphed as fold change in RNA expression. Triplicates were performed in each of 4 independent experiments. P values were determined by T-test, * P<0.001, ** P<0.02.

We applied general approaches described by Traver and colleagues [16,17] to purify cellular subsets from wild-type fish in our system, which are of distinct genetic backgrounds compared to those used in the published protocols. (A) Whole kidney marrow (WKM) was separated by flow cytometry (FACS). Dead cells were gated out by propidium iodide (PI) staining and the live cells analyzed by forward and side scatter profiles. The contour plot demonstrates that distinct populations can be identified: erythrocytes, myelo-monocytes, lymphocytes, and precursors. The average percentage of WKM for the populations are shown (n=11): erythrocytes: 35% ±9.2, lymphocytes: 14% ±3.1, precursors: 5.5% ±2.2, myelo-monocytes: 23% ±5.8. In addition, there is a minor eosinophil gate component that gates with higher relative side scatter from the myelo-monocyte group, approximately 3%. (B) FACS analysis of WKM from gata1:gfp reporter fish. A histogram representing relative GFP expression levels in cells from WKM is shown. Three distinct populations can be seen. A GFP negative peak, corresponding to the leukocytes, a bright GFP+ (high) peak corresponding to mature erythrocytes and a GFP+ low population that contains the erythroid progenitors. (C) The erythroid progenitors can be further discriminated using the gating strategy shown, similar to that described [17]. WKM was sorted by forward scatter (FSC), side scatter (SSC) and GFP, then contour plots of FSC vs. SSC were generated, and lymphocyte and precursor gates drawn. From each a GFP histogram was generated, and a GFP+ (low) gate drawn. The relative percent of GFP+ (low) cells was determined and used to track the erythroid progenitor populations. As shown in the box on the far right, these protocols allow accurate counts of adult erythroid cells at different developmental stages. (D) Shown are representative cytospin preparations of sorted cell types, as indicated. Cells were stained with Geimsa and May-Grünwald and photographed with a 100× oil objective. (i) peripheral erythrocytes, (ii) erythrocytes from WKM, (iii) myelo-monocytes, (iv) lymphocytes, (v) precursors, (vi) proerythroblasts, (vii) polychromatophilic erythroblasts.

(A) Analysis of heterozygous adults. WKM was collected from adult wild-type, pgy/+ and sbn/+ fish. Samples were analyzed by flow cytometry for FSC and SSC; dead cells were gated out by PI staining. Four populations were resolved: erythrocytes, myelo-monocytes, lymphocytes and precursors. For each genotype, the mean percent of WKM was graphed for erythrocytes, lymphocytes, precursors and myelo-monocytes. P values were calculated by two-tailed student's T-test by comparing wild-type to smad5 mutants for each cell type. Sample numbers for each were: wild-type (n=11), pgy/+ (n=5), sbn/+ (n=5) *P<0.02, **P<0.001. (B) Homozygous mutants rescued from embryonic lethality to adulthood by injection of mRNA encoding Smad5. Two mutant alleles that could be rescued are tc227 (believed to be a null allele) and m169. Analysis was as described above in (A). Sample numbers for each were: wild-type (n=4), tc227/tc227 (n=5), m169/m169 (n=2) *P<0.07, **P<0.03. The rescue of m169/m169 proved very challenging, which precluded generation of sufficient numbers of fish to test rigorously statistical significance, while the trend toward anemia is evident. Meanwhile, the tc227/tc227 mutants represent a stronger allele and shows clear anemia and additional alterations not seen in other mutants, for example in lymphocyte numbers.

(A) Scheme of erythroid development after commitment to the erythroid lineage. These three populations can be isolated using the gata1:gfp transgenic reporter line, as indicated in Fig. 1. (B-D) A cohort of wild-type, sbn/+ and pgy/+ adults carrying the gata1:gfp reporter transgene were treated with phenylhydrazine at day 0. On days 2, 5, and 12, adults were sacrificed, WKM prepared and samples analyzed by FACS. Dead cells were gated out using PI staining. The live cells were analyzed by forward scatter, side scatter, and GFP. Data from matched sets of untreated fish was used to generate the numbers indicated by day 0. The percent of each sub-population of erythroid progenitors was determined by gating on either the lymphoid or precursor fraction and then determining the number of GFP low positive cells in that fraction (see Fig. 1). The percentages were then graphed and P values were determined by two-tailed student's T-Test. For each time point, treated smad5 mutants were compared to age-matched sibling and treated wild-types. P values were as follows: polychromatophilic erythroblasts (day 2), for pgy/+ P<0.001, for sbn/+ P<0.02. For each time point, wild-type n=9, pgy/+ n=6 and sbn/+ n=3. As indicated in the boxed insert: ●=wild-type, ▲=pgy/+ ■=sbn/+. In B and C, the percent of Gata1+ cells refers to the “leukocyte” or “precursor” gate, respectively.

(A) A contour plot from FSC vs. SSC analysis of gata1:Δbr-gfp adult WKM is shown. While the cell numbers of the four major sub-populations are relatively normal, a fifth side scatter high (very granular) population of cells can be discriminated in these fish, representing eosinophils. (B) Cohorts of wild-type (gata1:gfp) and gata1:Δbr-gfp adults were treated with phenylhydrazine at day 0. On days 2, 4, 12 and 19 adults were sacrificed and WKM prepared and analyzed by FACS. Dead cells were gated out using DAPI staining, and the live cells were analyzed by forward scatter, side scatter and GFP. The percent of each sub-population of erythroid progenitors was determined by gating on either the lymphoid or precursor fraction and then determining the number of GFP(low) positive in that fraction. No significant changes were seen when comparing wild-type and gata1:Δbr-gfp fish. For each time point n=4. As indicated in the insert: ●=wild-type ■=gata1:Δbr-gfp. (C) Comparison of WKM sorts with wild-type (gata1:gfp) shows a relative increase in the number of eosinophils in WKM of gata1:Δbr-gfp adults; for each n=16, P<0.007. (A′) This population of cells was collected, centrifuged onto slides and stained with Giemsa and May-Grünwald to confirm their identification as eosinophils.