Strain-dependent modifiers exacerbate familial leukemia caused by GATA1-deficiency

Abstract

GATA1 plays a critical role in differentiation, proliferation, and apoptosis during erythropoiesis. We developed a Gata1 knock-down allele (Gata1.05) that results in GATA1 expression at 5% of endogenous level. In female mice heterozygous for both the Gata1.05 and wild-type alleles, we observed a predisposition to erythroblastic leukemia three to six months after birth. Since no male Gata1.05 progeny survive gestation, we originally maintained heterozygous females in a mixed genetic background of C57BL/6J and DBA/2 strains. Around 30% of these mice reproducibly develop leukemia, but the other subset did not develop leukemia, even though they harbor a high number of preleukemic erythroblasts. These observations prompted us to hypothesize that there may be potential influence of genetic determinants on the progression of Gata1.05-driven hematopoietic precursors to full-blown leukemia. In an initial examination of Gata1.05/X mice backcrossed into C3H/He, BALB/c, DBA/2, C57BL/6J and 129X1/SvJ strains, we discerned that the backgrounds of C57BL/6J and 129X1/SvJ significantly expedited leukemia onset in Gata1.05/X mice. Conversely, backgrounds of C3H/He, BALB/c and DBA/2 did not substantially modify the effect of the Gata1 mutation. This indicates the existence of genetic modifiers that accentuate Gata1.05 leukemogenesis. Subsequent cohort studies evaluated Gata1.05/X mice within mix backgrounds of BALB/c:129X1/SvJ and BALB/c:C57BL/6J. In these settings, Gata1.05-driven leukemia manifested in autosomal dominant patterns within the 129X1/SvJ background and in autosomal recessive patterns within C57BL/6J background. To the best of our knowledge, this study provides the inaugural evidence of genetic modifiers that can reshape the outcome based on leukemia-associated gene signatures.

To the editor,

GATA1 is a transcription factor that plays crucial roles in erythropoiesis, megakaryopoiesis, and other hematopoietic pathways [1,2,3,4]. We previously established a Gata1.05 allele, in which Gata1 gene expression is reduced into 5% of endogenous level [5]. Since the Gata1 gene is located on the X-chromosome, heterozygous female mice (G1^1.05) carry two types of erythroid progenitors due to random X-chromosome inactivation. Erythroid progenitors with an active Gata1.05 allele and an inactive wild-type X-chromosome struggle to differentiate into matured cells, leading to the accumulation of immature erythroid progenitors in the G1^1.05 [6,7,8,9]. These abnormal erythroid progenitors frequently undergo cancerous changes. Consequently, approximately 30% of G1^1.05 in a mixed background colony of C57BL/6J:DBA/2 strains are predisposed to develop erythroleukemia, which morphologically resembles human pure erythroid leukemia [6,7,8,9,10]. These findings have led us to hypothesize that genetic variations exist between the C57BL/6J and DBA/2 strains, which may act as modifiers of Gata1.05-driven leukemogenesis.

To investigate this hypothesis, we generated a mouse cohort of 369 G1^1.05. These mice were produced by backcrossing G1^1.05 in a mixed background to five inbred strains: C3H/He, BALB/c, DBA/2, C57BL/6J, and 129X1/SvJ. The resulting strains were named C3.G1^1.05, C.G1^1.05, D2.G1^1.05, B6.G1^1.05, 129.G1^1.05, respectively, followed by the number of backcross generations. Eventually, we used 44 mice of C.G1^1.05 (N2-N9), 20 mice of C3.G1^1.05 (N2-N6), 126 mice of B6.G1^1.05 (N2-N8), 40 mice of D2.G1^1.05 (N2-N6), and 139 mice of 129.G1^1.05 (N2-N9) for the survival time analysis. We then studied the development of leukemia in these mice. Our findings revealed a significantly elevated risk of early mortality in B6.G1^1.05 and 129.G1^1.05 when compared to the other three backgrounds (Fig. 1A). In both the B6.G1^1.05 and 129.G1^1.05 groups, the peak mortality occurred between 90 and 180 days of age. This aligns closely with a previous report on leukemogenesis in G1^1.05 in a mixed background [6].

Fig. 1

Mortality of Gata1.05/X mice (G1^1.05) is affected by genetic backgrounds. A Kaplan-Meier curves depicting the overall survival of G1^1.05 in C57BL/6J (black line, n = 126), 129X1/SvJ (blue line, n = 139), BALB/c (red line, n = 44), DBA/2 (green line, n = 40) and C3H (purple line, n = 20) backgrounds. Mice that were backcrossed onto the indicated inbred strains twice to nine times were used in this experiment. The details of mice used in this experiment are provided in Supplementary Table 1. A summary of the Log-rank test (upper row) and Generalized Wilcoxson test (lower row) results between the indicated groups is shown in table. n.s; not significant. B Observed mortality of C.G1^1.05, C3.G1^1.05, B6.G1^1.05, D2.G1^1.05 and 129.G1^1.05, along with their necropsy results. C-E Leukemia phenotype in B6.G1^1.05 and 129.G1^1.05 that underwent necropsy. Representative images of the macroscopic (C), Hematoxylin-Eosin microscopic (D), and May-Giemsa-stained cytospin (E) analyses of spleens from G1^1.05 developing leukemia in 129X1/SvJ (left) and C57BL/6J (right) backgrounds. Note that leukemia phenotypes of mice in 129X1/SvJ and C57BL/6J backgrounds do not exhibit substantial differences

Necropsy analyses revealed a strong correlation between leukemogenesis and mortality in B6.G1^1.05 and 129.G1^1.05 (Fig. 1B). These analyses showed pronounced splenomegaly characterized by a destroyed splenic structure due to the extensive invasion of proerythroblast-like leukemic cells (Fig. 1C-E). The leukemic alterations in 129.G1^1.05 (Fig. 1C-E; left panels) and B6.G1^1.05 (right panels) were indistinguishable between and bore resemblance to those observed in mice in a mixed background [6]. In contrast, when we examined four C.G1^1.05, three C3.G1^1.05, and five D2.G1^1.05 animals euthanized within a year as a humane endpoint, only two mice were found to have developed leukemia (Fig. 1B). Thus, the penetrance of Gata1.05-driven leukemogenesis appears to be strongly influenced by the genetic backgrounds of the mice.

To investigate the leukemia susceptibility phenotype of 129.G1^1.05 and B6.G1^1.05, we established two validation mouse cohorts by crossing C.G1^1.05, which rarely develop leukemia, with wild-type 129X1/SvJ or C57BL/6J males. We then crossbred the resulting F1 generation mice (F1-C:129.G1^1.05 and F1-C:B6.G1^1.05, respectively) to produce the second generation (Fig. 2A,B). Interestingly, F1-C:129.G1^1.05 exhibited a high mortality rate, with over half not surviving past six months (Fig. 2C). Comparable survival curves were observed in both N2-129.G1^1.05 and F2-C:129.G1^1.05 (Fig. 2A). The former represents the second backcross generation produced by crossing F1-C:129.G1^1.05 with 129X1/SvJ inbred males, while the latter denote the second filial generation produced by brother and sister intercross of F1-C:129.G1^1.05. Notably, F2-C:129.G1^1.05 showed a modestly increased survival rate when compared to F1-C:129.G1^1.05 or N2-129.G1^1.05 (Fig. 2C). When comparing the survival of F1-C:129.G1^1.05, N2-129.G1^1.05 and F2-C:129.G1^1.05 with 129.G^1.05 used in the discovery cohort study, we found that the survival rates of F1-C:129.G1^1.05 and N2-129.G1^1.05 were almost similar to that of 129.G^1.05, while the survival rate of F2-C:129.G1^1.05 was significantly increased (Supplementary Fig. 1A). Given that a quarter of the F2-C:129.G1^1.05 cohort were BALB/c progeny, it seems plausible that these mice contributed to the slightly enhanced survival rate, especially as they infrequently developed leukemia.

Fig. 2

Inherited susceptibility to leukemia in Gata1.05/X mice (G1^1.05) in 129X1/SvJ and C57BL/6J backgrounds. A Schematic illustration of mouse mating strategy. N6 or deeper generation C.G1^1.05 mice were prepared by backcrossing G1^1.05 in a mixed background with BALB/c mice six and more generations. These C.G1^1.05 were crossed with 129X1/SvJ wild-type males, and subsequently F1-C:129.G1^1.05 (shown in gray) were generated. Note that F1-C:129.G1^1.05 display heterozygosity at the Gata1 locus (i.e., Gata1.05 and wild-type alleles), with equal contributions from the BALB/c and 129X1/SvJ backgrounds on all autosomes. G1^1.05 generated by crossing F1-C:129.G1^1.05 with wild-type 129X1/SvJ males are defined as N2-129.G1^1.05 (shown in blue). We also generated F2-C:129.G1^1.05 (shown in purple) through brother-sister crossing of F1-C:129.G1^1.05 and F1-C:129 males. Thus, N2-129.G1^1.05 (blue) harbor either a homozygous 129X1/SvJ background or a heterozygous BALB/c:129X1/SvJ background. Similarly, F2-C:129.G1^1.05 (purple) harbor a homozygous BALB/c background, a homozygous 129X1/SvJ background, or a heterozygous BALB/c:129X1/SvJ background. Details of the mice examined are provided in Supplementary Table 2B Schematic illustration of mouse mating strategy. N6 or deeper generation C.G1^1.05 mice were used. G1^1.05 obtained by crossing C.G1^1.05 with wild-type C57BL/6J males are defined as F1-C:B6.G1^1.05 (shown in turquois). Crossing F1-C:B6.G1^1.05 with wild-type BALB/c and C57BL/6J males generated N2-C.G1^1.05 (shown in red) and N2-B6.G1^1.05 (shown in black), respectively. Note that F1-C:B6.G1^1.05 display heterozygosity at the Gata1 locus (i.e., Gata1.05 and wild-type alleles), with equal contributions from the BALB/c and C57BL/6J backgrounds on all autosomes. N2-C.G1^1.05 (red) harbor a BALB/c:C57BL/6J heterozygous background or a BALB/c homozygous background, and N2-B6.G1^1.05 (black) harbor a BALB/c:C57BL/6J heterozygous background or a C57BL/6J homozygous background. Details of the mice examined are provided in Supplementary Table 3C Kaplan-Meier curves of the overall survival of F1 C:129.G1^1.05 (gray line, n = 35), N2 129.G1^1.05 (blue line, n = 9), and F2C:129.G1^1.05 (purple line, n = 156). D Kaplan-Meier curves of the overall survival of F1-C:B6.G1^1.05 (turquois line, n = 30), N2-C.G1^1.05 (red line, n = 57), and N2-B6.G1^1.05 (black line, n = 264). Results of Log-rank tests and Generalized Wilcoxson tests in comparison of N2-B6.G1^1.05 vs. F1-B6:C.G1^1.05 and N2-B6.G1^1.05 vs. N2-C.G1^1.05 are indicated. E Observed mortality data for mice used in C and necropsy results. F Observed mortality data for mice used in D and necropsy results

In stark contrast, over 80% of F1-C:B6.G1^1.05 survived for 10 months (Fig. 2D), aligning closely with the outcomes observed in C3.G1^1.05, C.G1^1.05, and D2.G1^1.05 (Fig. 1A). When analyzing N2-B6.G1^1.05 and N2-C.G1^1.05 that were generated by crossbreeding C:B6.G1^1.05 with C57BL/6J and BALB/c inbred males, respectively, we observed that N2-B6.G1^1.05, but not N2-C.G1^1.05, had a notably earlier mortality than F1-C:B6.G1^1.05 (Fig. 2D). Meanwhile, N2-B6.G1^1.05 showed a significantly higher survival rate compared with B6.G^1.05 used in the discovery cohort study (Supplementary Fig. 1B). We postulate that the earlier mortality observed in N2-B6.G1^1.05 is attributable to the genetic characteristics inherent in loci being homozygous for C57BL/6J inbred strain. The mortality rates in these cohort studies showed strong correlations with leukemogenesis (Fig. 2E,F). Taken together, it appears that the 129X1/SvJ strain possesses autosomal dominant traits while the C57BL/6J strain exhibits autosomal recessive traits, both of which seem to expedite the development of Gata1.05-driven leukemia.

The next aspect we investigated was whether strain variations influence leukemogenesis. We determined that the expression levels of GATA1 in the bone-marrow cells of wild-type female mice across BALB/c, C3H/He, and DBA/2 (with low incident), C57BL/6J and 129X1/SvJ (with high incident) inbred strains were nearly equivalent. An exception was a modest increase in C3H/He mice (Supplementary Fig. 2A). Gata1 cDNA sequences across these strains, accessible from the Ensemble Genome Browser, were identical. We found no strong correlation between typical hematopoietic indices in wild-type female mice and the incidence rate of Gata1.05-driven leukemia (Supplementary Fig. 2B, C). Given these findings, we conclude that there are yet unidentified genetic modifiers influencing GATA1-deficiency and its role in Gata1.05-driven leukemogenesis (Supplementary Fig. 3).

This study offers persuasive evidence that inherent genetic backgrounds interact with disease-causing mutations, like the Gata1.05 mutation, to influence the progression of leukemia. Considering the working model that supports a stepwise malignant transformation during carcinogenesis, we surmise that genetic variations relevant to the oncogenic process may give rise to synergistic or antagonistic effects on the clinical phenotype caused by primary oncogenic mutations. Our findings illuminate that among the group of polymorphisms typically considered silent, there are variations that can modify susceptibility to disease onset when influenced by genetic mutations associated with leukemia.