https://orcid.org/0000-0003-2451-2986
Abstract
Growth hormone (GH) binding to GH receptor activates janus kinase 2 (JAK2)-signal transducer and activator of transcription 5b (STAT5b) pathway, which stimulates transcription of insulin-like growth factor-1 (IGF1), insulin-like growth factor binding protein 3 (IGFBP3) and insulin-like growth factor acid-labile subunit (IGFALS). Although STAT5B deficiency was established as an autosomal recessive disorder, heterozygous dominant-negative STAT5B variants have been reported in patients with less severe growth deficit and milder immune dysfunction. We developed an in vivo functional assay in zebrafish to characterize the pathogenicity of three human STAT5B variants (p.Ala630Pro, p.Gln474Arg and p.Lys632Asn). Overexpression of human wild-type (WT) STAT5B mRNA and its variants led to a significant reduction of body length together with developmental malformations in zebrafish embryos. Overexpression of p.Ala630Pro, p.Gln474Arg or p.Lys632Asn led to an increased number of embryos with pericardial edema, cyclopia and bent spine compared with WT STAT5B. Although co-injection of WT and p.Gln474Arg and WT and p.Lys632Asn STAT5B mRNA in zebrafish embryos partially or fully rescues the length and the developmental malformations in zebrafish embryos, co-injection of WT and p.Ala630Pro STAT5B mRNA leads to a greater number of embryos with developmental malformations and a reduction in body length of these embryos. These results suggest that these variants could interfere with endogenous stat5.1 signaling through different mechanisms. In situ hybridization of zebrafish embryos overexpressing p.Gln474Arg and p.Lys632Asn STAT5B mRNA shows a reduction in igf1 expression. In conclusion, our study reveals the pathogenicity of the STAT5B variants studied.
Introduction
Normal human growth depends, among other factors, on the growth hormone (GH) acting on its receptor (GHR) and activating the janus kinase 2 (JAK2)-signal transducer and activator of transcription 5b (STAT5b) signaling pathway, which transduces signals from circulating GH to target genes such as the insulin-like growth factor-1 (IGF1), the insulin-like growth factor binding protein 3 (IGFBP3) and the insulin-like growth factor acid-labile subunit (IGFALS) (1). IGF1 is highly conserved in vertebrates and plays a central role in pre- and post-natal growth (2,3). Briefly, GH binds to GHR which belongs to the Type 1 cytokine receptor superfamily. Members of this superfamily lack intrinsic kinase activities and require the recruitment and trans-activation of two molecules of associated JAK2 kinases, one per each GHR monomer, for initiating signal transduction (4,5). Activated JAK2 phosphorylates several of the 7 tyrosine residues located within the intracellular domain of GHR. This, in turn, leads to docking of STAT5b monomers through their Src homology 2 (SH2) domain. STAT5b monomers phosphorylated at Tyr699 homo-dimerize and translocate to the nucleus to recognize and bind DNA promoter regions regulating transcription of specific genes, such as IGF1, IGFBP3 and IGFALS.
STATs are pleiotropic proteins that function both as signal transducers and transcription factors. They are widely expressed in many tissues, are activated by many growth factors, such as GH and other cytokines such as interleukins, and are involved in many biological activities (6). STAT5A and STAT5B genes encode proteins that are ~95% identical in amino acid sequence (7). Knockout (KO) studies have shown that Stat5b-deficient mice have diminished post-natal growth compared with wildtype (WT) (20–30% smaller). On the other hand, double Stat5a/b KOs exhibit growth defects and immune dysregulation comparable to human STAT5B deficiency (8,9), evidencing some redundancy between these two highly related proteins in mice that is not observed in humans.
Mutations in many of the genes involved in the GHR-JAK2-STAT5b signaling pathway have been identified in part aided by massive parallel sequencing techniques such as whole exome sequencing (WES) and whole genome sequencing (WGS). At least eleven patients with molecular defects in the STAT5B gene have been reported and are associated with GH insensitivity, which is characterized by severe post-natal growth failure and low IGF-1 levels (10–12). In addition, patients with STAT5B defects have immune dysfunction with potential fatal pulmonary involvement (13). Since STAT5b is also required in the signaling of several cytokines such as interleukin-2 and γ-interferon, it seems likely that the immune defect is due to its inactivation.
Interestingly, haploinsufficiency for the STAT5B gene appears to affect growth, since heterozygous carriers are shorter than their WT relatives (14). Even though STAT5B deficiency (MIM 245590) has been established as an autosomal recessive disorder, where patients are either homozygous or compound heterozygous for STAT5B variants, recently, three heterozygous dominant-negative inactivating STAT5B variants were reported in patients with less severe growth deficits and milder immune dysfunction than those with autosomal recessive STAT5B deficiency (15). Further research is needed to better understand genotype–phenotype correlations of STAT5B variants and to determine if heterozygous variants are associated with a significant phenotype.
Recently, our group identified a novel heterozygous STAT5B variant in the SH2 domain c.1896G > T, (p.Lys632Asn) in a patient with short stature and mild immune dysregulation (11). Although we suspected this variant behaves as a dominant-negative variant, in vitro assays were inconclusive. Reconstitution studies showed that this variant was normally expressed but unable to become phosphorylated upon GH stimulation, did not translocate to the nucleus, and did not activate transcription. Interestingly, co-transfection of p.Lys632Asn and WT STAT5B also resulted in decreased basal transcriptional activity compared with WT alone. However, in response to GH, co-transfection of mutant and WT STAT5B in a 1:1 ratio resulted in significant transcriptional activity. In the end, these in vitro methods cannot account for the influence of genetic modifiers and diverse environmental factors that affect growth in vivo. The generation of in vivo models can overcome these limitations.
Zebrafish has become an excellent vertebrate model of human disease (16,17). Its advantages include its external fertilization, rapid development and optical transparency. The GH-IGF-1 system is highly conserved throughout vertebrates and all their major players have orthologs in zebrafish, including STAT5B. Zebrafish possess two stat5 genes (stat5.1 and stat5.2), which appear to result from a duplication event, independent of the relatively recent mammal-specific duplication event which generated STAT5A and STAT5B genes in mammals (18). Stat5.1 deficient zebrafish generated using crispr-cas9 exhibit a significant reduction of body length and body weight revealing that stat5.1 is the ortholog gene of mammalian STAT5B (19). On the other hand, morpholino (MO) knockdown of stat5.1 led to zebrafish embryos indistinguishable from uninjected controls except for the presence of large gaps in the matrix that separates myotomes in the somites along the anterior–posterior axis (20). Other studies using stat5.1 MOs revealed decreased expression of rag1, a marker of more mature lymphoid cells (21).
To date, no in vivo functional studies have been developed in zebrafish to evaluate the consequences of pathogenic STAT5B human variants in the context of the whole organism during development. In the present study we developed an overexpression assay using zebrafish as a biosensor to characterize the pathogenicity of three STAT5B human variants: c. 1888G > C (p.Ala630Pro), a homozygous autosomal recessive variant; c.1421A > G (p.Gln474Arg) a heterozygous dominant negative variant; and c.1896G > T, (p.Lys632Asn), a heterozygous variant with potential dominant negative behavior.
Results
MO knockdown of stat5.1 does not lead to a reduction of body length in zebrafish embryos
The overall domain structure of human STAT5b and the position of the variants studied are pictured in Figure 1. STAT5b is a latent transcription factor that contains five conserved domains: an N-terminal domain, a coiled-coiled domain, a DNA binding domain (DBD), a linker domain, an Src homology 2 (SH2) domain, and a transactivation domain (TAD) (22) (Fig. 1). Both p.Lys632Asn and p.Ala630Pro STAT5B variants are located within the SH2 domain, whereas variant p.Gln4747Arg is located within the DBD. The N-terminal and coiled-coiled domains mediate protein–protein interactions, whereas the SH2 domain is necessary for binding phosphorylated tyrosines, allowing STAT5b to dock on phospho-tyrosine residues of activated receptors and to homo-dimerize when STAT5b itself is phosphorylated by JAK2. In the nucleus, the DBD binds DNA elements, and the TAD drives transcriptional activity.
Schematic of the human STAT5b protein adapted from Ramirez et al., 2020. The three variants studied are indicated: p.Gln474Arg located in the DNA binding domain, and p.Ala630Pro and p.Lys632Asn located in the SH2 domain. NH2, amino acids 1–144; Coiled-coil, amino acids 144–330; DNA binding, amino acids 330–496; linker, amino acids 496–592; SH2, amino acids 592–685; TAD, amino acids 707–787. Tyrosine 699 (Y699) can be phosphorylated by JAK2 and other kinases.
In order to develop an in vivo assay to test the pathogenicity of human STAT5B variants, we first evaluated the phenotype obtained through the knockdown of stat5.1. To do this, we microinjected a stat5.1 MO and measured the body length of the morphant embryos. Despite the previous observation that stat5.1 zebrafish KO mutant obtained through crispr/cas9 technique results in a significant body length reduction of embryos (19), stat5.1 morphants did not result in a significant body length reduction compared with embryos injected with a control MO or uninjected embryos (P > 0.05, Fig. 2A and B). In addition, no morphological deformities were observed in morphant embryos.
Effect of stat5.1 MO knockdown on zebrafish embryos. (A) Lateral view of representative pictures of control MO and morphant embryos at 72 hpf. (B) Statistical analysis of body length (jaw to tail fin) in uninjected, control MO and stat5.1 MO injected embryos at 72 hpf. Data are shown as mean ± SEM. Mann–Whitney Test. ns = not significant, P > 0.05. N = number of embryos. mm = millimeters. Scale bar: 1 mm.
Overexpression of human STAT5B mRNAs in zebrafish embryos leads to a reduction of body length and developmental malformations
Since no phenotype was observed upon microinjection of stat5.1 MO, we decided to develop an overexpression assay to test the pathogenicity of human STAT5B variants by interference with endogenous stat5.1 signaling. For this purpose, we first microinjected human WT STAT5B mRNA into zebrafish embryos. Most of the zebrafish embryos overexpressing WT STAT5B were normal except for a small number of embryos that showed specific developmental malformations including pericardial edema (15%, 15/102 embryos), cyclopia (2%, 2/102 embryos) and bent spine (15%, 15/102 embryos) (Table 1). Unexpectedly, when measured, these embryos were slightly but significantly smaller (3.12 ± 0.01 mm, a 2.91% reduction) than uninjected controls (3.21 ± 0.01 mm) (P < 0.0001) (Fig. 3A and B).
Malformations during development in embryos overexpressing WT STAT5B, p.Ala630Pro, p.Gln474Arg and p.Lys632Asn mRNA
| Phenotype | WT STAT5B | p.Ala630Pro STAT5B | p.Gln474Arg STAT5B | p.Lys632Asn STAT5B |
|---|---|---|---|---|
| Pericardial edema | 15/102 = 15% | 36/113 = 32% | 59/108 = 55% | 48/89 = 54% |
| Cyclopia | 2/102 = 2% | 15/113 = 13% | 32/108 = 30% | 27/89 = 30% |
| Bent spine | 15/102 = 15% | 20/113 = 18% | 66/108 = 61% | 24/89 = 27% |
| Phenotype | WT STAT5B | p.Ala630Pro STAT5B | p.Gln474Arg STAT5B | p.Lys632Asn STAT5B |
|---|---|---|---|---|
| Pericardial edema | 15/102 = 15% | 36/113 = 32% | 59/108 = 55% | 48/89 = 54% |
| Cyclopia | 2/102 = 2% | 15/113 = 13% | 32/108 = 30% | 27/89 = 30% |
| Bent spine | 15/102 = 15% | 20/113 = 18% | 66/108 = 61% | 24/89 = 27% |
Malformations during development in embryos overexpressing WT STAT5B, p.Ala630Pro, p.Gln474Arg and p.Lys632Asn mRNA
| Phenotype | WT STAT5B | p.Ala630Pro STAT5B | p.Gln474Arg STAT5B | p.Lys632Asn STAT5B |
|---|---|---|---|---|
| Pericardial edema | 15/102 = 15% | 36/113 = 32% | 59/108 = 55% | 48/89 = 54% |
| Cyclopia | 2/102 = 2% | 15/113 = 13% | 32/108 = 30% | 27/89 = 30% |
| Bent spine | 15/102 = 15% | 20/113 = 18% | 66/108 = 61% | 24/89 = 27% |
| Phenotype | WT STAT5B | p.Ala630Pro STAT5B | p.Gln474Arg STAT5B | p.Lys632Asn STAT5B |
|---|---|---|---|---|
| Pericardial edema | 15/102 = 15% | 36/113 = 32% | 59/108 = 55% | 48/89 = 54% |
| Cyclopia | 2/102 = 2% | 15/113 = 13% | 32/108 = 30% | 27/89 = 30% |
| Bent spine | 15/102 = 15% | 20/113 = 18% | 66/108 = 61% | 24/89 = 27% |
Effect of overexpression of WT, p.Ala630Pro, p.Gln474Arg and p.Lys632Asn STAT5B mRNA on zebrafish embryos. (A) Lateral view of representative pictures in uninjected and STAT5B WT, p.Ala630Pro, p.Gln474Arg and p.Lys632Asn mRNA injected embryos (B) Statistical analysis of body length (jaw to tail fin) in uninjected, and WT, p.Ala630Pro, p.Gln474Arg and p.Lys632Asn STAT5B mRNA injected embryos. Embryos were measured at 72 hpf. Mann–Whitney test, ***P < 0.0001, ns = not significant, P > 0.05. N = number of embryos. mm = millimeters. Scale bar: 1 mm.
Overexpression of p.Ala630Pro led to embryos with an increased number of developmental malformations: pericardial edema (36/113 = 32%), cyclopia (15/113 = 13%), and bent spine (20/113 = 18%). Interestingly, overexpression of p.Gln474Arg or p.Lys632Asn STAT5B mRNA resulted in an even higher number of embryos with pericardial edema (p.Gln474Arg 59/108 = 55%; p.Lys632Asn 48/89 = 54%), cyclopia (p.Gln474Arg 32/108 = 30%; p.Lys632Asn 27/89 = 30%) and bent spine (p.Gln474Arg 66/108 = 61%; p.Lys632Asn 24/89 = 27%) (Table 1). These developmental malformations were classified into three groups according to their severity and scored (1 = normal, 2 = mild and 3 = severe) (Fig. 4A–G).
Effect of overexpression of WT, p.Ala630Pro, p.Gln474Arg and p.Lys632Asn STAT5B mRNA on zebrafish embryos. (A, C, E) Representative pictures of zebrafish embryos with pericardial edema (A), cyclopia (B) and bent spine (E) during development. (B, D, F) Statistical analysis of pericardial edema score (B), cyclopia (D) and bent spine (F) in embryos overexpressing WT, p.Ala630Pro, p.Gln474Arg and p.Lys632Asn STAT5B mRNA. (G) Percentage of embryos showing pericardial edema, cyclopia, and bent spine in embryos overexpressing WT, p.Ala630Pro, p.Gln474Arg and p.Lys632Asn STAT5B mRNA. Embryos at 72 hpf were classified into three groups according to the severity of the malformation and scored (1 = normal, 2 = mild and 3 = severe). Mann–Whitney test, ***P < 0.0001, *P < 0.05. N = number of embryos across three independent biological replicates.
In turn, we measured embryos overexpressing the three STAT5B variants and found that overexpression of p.Ala630Pro did not lead to a reduction of body length compared with WT STAT5B mRNA injected embryos (3.08 ± 0.01 mm) (Fig. 3A and B). Meanwhile, overexpression of p.Gln474Arg and p.Lys632Asn variants led to embryos with a significant reduction of body length compared with WT STAT5B mRNA overexpressing embryos (2.58 ± 0.05 mm, a 17.34% reduction and 2.72 ± 0.05 mm, a 12.68% reduction respectively, significantly shorter than WT STAT5B mRNA overexpressing embryos, P < 0.0001) (Fig. 3A and B). This clearly shows a different mechanism of action between variant p.Ala630Pro, and variants p.Gln474Arg and p.Lys632Asn.
Co-expression of WT and p.Gln474Arg STAT5B mRNA partially rescues the length of zebrafish embryos and rescues or partially rescues the developmental malformations in zebrafish embryos
To determine the underlying mechanism involved in the phenotype observed when the STAT5B variants were overexpressed, we co-injected WT and p.Gln474Arg STAT5B mRNAs in a 1:1 ratio into zebrafish embryos and evaluated the ability of WT STAT5B mRNA to rescue both the body length and developmental malformations. Co-injection of WT STAT5B mRNA partially rescued body length reduction (P < 0.0001 WT STAT5B versus p.Gln474Arg/WT, Fig. 5A), and the severity of bent spine malformation (WT STAT5B versus p.Gln474Arg/WT; P < 0.0001, Fig. 5B and C). In addition, pericardial edema and cyclopia were fully rescued (WT STAT5B versus p.Gln474Arg/WT; P > 0.05, Fig. 5B and C).
Effect of co-injecting p.Gln474Arg together with WT STAT5B mRNA in a 1:1 ratio in zebrafish embryos. (A) Statistical analysis of body length in embryos overexpressing WT, p.Gln474Arg and p.Gln474Arg/WT STAT5B mRNA. (B) Statistical analysis of pericardial edema, cyclopia and bent spine score in embryos overexpressing WT, p.Gln474Arg and p.Gln474Arg/WT STAT5B mRNA. (C) Percentage of embryos showing pericardial edema, cyclopia and bent spine in embryos overexpressing WT, p.Gln474Arg and p.Gln474Arg/WT STAT5B mRNA. Embryos were scored at 72 hpf and classified into three groups according to the severity of the malformation (1 = normal, 2 = mild, and 3 = severe. Mann–Whitney test, ***P < 0.0001, *P < 0.05, ns = not significant, P > 0.05. N = number of embryos. mm = millimeters.
Co-expression of WT and p.Lys632Asn STAT5B rescues the length of zebrafish embryos but partially rescues developmental malformations in zebrafish embryos
In contrast to p.Gln474Arg, co-injection of WT together with p.Lys632Asn STAT5B mRNA completely rescued body length in zebrafish embryos (P > 0.05 WT STAT5B versus p.Lys632Asn/WT) (Fig. 6A). Regarding developmental malformations, co-injection of p.Lys632Asn/WT led to a partial rescue in the severity of each of the malformations: pericardial edema, cyclopia and bent spine (WT STAT5B versus p.Lys632Asn/WT; P = 0.01, P < 0.0001, P = 0.007, respectively, Fig. 6B and C).
Effect of co-injecting p.Lys632Asn together with WT STAT5B mRNA in a 1:1 ratio in zebrafish embryos. (A) Statistical analysis of body length in embryos overexpressing WT, p.Lys632Asn and p.Lys632Asn/WT STAT5B mRNA. (B) Statistical analysis of pericardial edema, cyclopia and bent spine score in embryos overexpressing WT, p.Lys632Asn and p.Lys632Asn/WT STAT5B mRNA. (C) Percentage of embryos showing pericardial edema, cyclopia and bent spine in embryos overexpressing WT, p.Lys632Asn and p.Lys632Asn/WT STAT5B mRNA. Embryos were scored at 72 hpf and classified into three groups according to the severity of the malformation (1 = normal, 2 = mild and 3 = severe. Mann–Whitney test, ***P < 0.0001, *P < 0.05, ns = not significant, P > 0.05. N = number of embryos. mm = millimeters.
Co-expression of WT and p.Ala630Pro STAT5B in zebrafish embryos leads to a greater number of embryos with developmental malformations and a reduction in body length of these embryos
Variant p.Ala630Pro, a loss-of-function variant, was the only one that did not lead to a reduction of body length compared with WT STAT5B mRNA overexpressing embryos. Nevertheless, we evaluated the consequences of co-injecting it with WT STAT5B in a 1:1 ratio. Surprisingly, co-injection of p.Ala630Pro with WT STAT5B mRNA led to a significant reduction in body length compared with either p.A630P or to WT STAT5B alone (2.37 ± 0.07 mm, a 23.03% reduction and 24.01% reduction, respectively, WT STAT5B versus p.Ala630Pro/WT, P < 0.0001, Fig. 7A). In addition, co-injection of WT and p.Ala630Pro STAT5B mRNA in a 1:1 ratio led to embryos with more severe malformations (WT STAT5B versus p.Ala630Pro/WT pericardial edema, cyclopia and bent spine P < 0.0001) (Fig. 7B), and to a greater proportion of embryos showing these defects (Fig. 7C).
Effect of co-injecting p.Ala630Pro together with WT STAT5B mRNA in a 1:1 ratio in zebrafish embryos. (A) Statistical analysis of body length in embryos overexpressing WT, p.Ala630Pro and p.Ala630Pro/WT STAT5B mRNA. (B) Statistical analysis of pericardial edema, cyclopia and bent spine score in embryos overexpressing WT, p.Ala630Pro and p.Ala630Pro/WT STAT5B mRNA. (C) Percentage of embryos showing pericardial edema, cyclopia and bent spine in embryos overexpressing WT, p.Ala630Pro and p.Ala630Pro/WT STAT5B mRNA. Embryos were scored at 72 hpf and classified into three groups according to the severity of the malformation (1 = normal, 2 = mild and 3 = severe. Mann–Whitney test, ***P < 0.0001, *P < 0.05, ns = not significant, P > 0.05. N = number of embryos. mm = millimeters.
The expression of igf1 mRNA is reduced in embryos overexpressing p.Gln474Arg, and p.Lys632Asn STAT5B mRNA
To further characterize the effects of the variants causing a body length reduction we evaluated the integrity of the GH-IGF-1 signaling pathway, by performing in situ hybridization at 48 hpf using probes against gh1, stat5.1, igfals and igf1 mRNAs. Figure 8 shows that embryos overexpressing p.Gln474Arg, and p.Lys632Asn STAT5B mRNA but not WT or p.Ala630Pro STAT5B mRNA exhibited a reduction in igf1 expression. In addition, embryos overexpressing WT STAT5B mRNA showed reduced levels of igfals compared with uninjected controls.
WISH analysis of gh1, stat5.1, igfals and igf1 mRNA in representative zebrafish embryos at 48 hpf. For gh1 expression, embryos are the head region of the embryo, ventral view with anterior to the left. For stat5.1 expression, embryos are whole embryos, lateral view, with anterior to the left and dorsal up. For igfals expression, embryos are the anterior half of the embryo, lateral view with anterior to the left and dorsal up. For igf1 expression, embryos are whole embryos, lateral view, with anterior to the left and dorsal up. Total number of embryos analyzed are shown in Supplementary Material, Table S1.
Discussion
To our knowledge, this is the first in vivo functional assay using zebrafish as a biosensor to evaluate the pathogenic nature of human STAT5B variants. In this study, we show that overexpression of human WT STAT5B and STAT5B variants lead to a reduction of body length and developmental malformations in zebrafish embryos. These developmental malformations include pericardial edema, cyclopia and bent spine of different severity and incidence depending on the STAT5B mRNA overexpressed.
These results suggest that all variants (including WT STAT5B) capable of being expressed can potentially interfere with endogenous stat5.1 signaling resulting in altered zebrafish embryos, including a reduction of body length and specific developmental malformations. This seems to depend on but is not limited to, the ability of human STAT5B variants to form hetero-dimers STAT5b/stat5.1 with a potentially reduced transcriptional activity or other interfering mechanisms such as delayed phosphorylation or impaired nuclear transport/DNA binding affinity. STATs can also form several different homo- and hetero-dimers and STATs can form hetero-dimers with other proteins even without tyrosine phosphorylation (23).
Based on the results published by Xiong et al. (19), we hypothesized that dominant negative STAT5B variants might interfere with endogenous stat5.1 ultimately affecting embryo body length. The mechanism through which overexpression of WT STAT5B mRNA leads to a small effect on growth compared with uninjected embryos is probably the result of a subtle interference of endogenous stat5.1 signaling through the formation of STAT5b/stat5.1 hetero-dimers with reduced transcriptional activity compared with endogenous stat5.1 homo-dimers.
Both human STAT5B deficient patients, as well as Stat5b-KO mice, present normal size at birth. This is in agreement with the evidence that although fetal growth is highly dependent on IGF1, during the prenatal period its expression is independent of GH, and consequently it does not require Stat5b. However, loss-of-function mutants in zebrafish stat5.1 resulted in a significant reduction of body length compared with WT larvae at 96 hpf, suggesting that stat5.1 is required for normal somatic growth of zebrafish embryos/larvae (19).
In the present study, MO-mediated stat5.1 knockdown did not result in a significant reduction of body length, in agreement with previous studies (20). Interestingly, stat5.1 knockdown using Crispr-Cas9 did lead to a significant reduction in body length probably as a result of a more profound and prolonged stat5.1 knockdown obtained by this technique (19). In this study, we have observed that overexpression of some human STAT5B variants resulted in this phenomenon showing interference with endogenous stat5.1 signaling, possibly at the protein level. The absence of a phenotype using MO could be explained by a partial knockdown by MO, which can result in the complete retention of the normal physiological function of the gene product (24). In addition, there are many examples in the literature of the reverse situation, genes that show morphant phenotypes using MOs whereas the mutant alleles do not (25). This is predominantly owing to genetic compensation (26).
Both STAT5B variants p.Ala630Pro and p.Lys632Asn are located in the SH2 domain and are 2 amino acids apart. However, they behave very differently, both in the clinical phenotype of the patients carrying these variants as well as in the zebrafish overexpressing these variants. Interestingly, recent studies suggest that gain-of-function and dominant-negative mutations have much milder effects on protein structure than loss-of-function mutations and are thus less well predicted by nearly all computational variant effect predictors (27). The SH2 region is essential for docking of STATs to phospho-tyrosines on activated receptors, STATs dimerization and stabilization of phospho-STAT-DNA interactions. Previous in vitro assays show that variant p.Ala630Pro has an impaired expression compared with WT, is weakly phosphorylated upon GH stimulation and lacks transcriptional activity (28,29). In addition, p.Ala630Pro has aberrant folding and diminished solubility triggered by a misfolded SH2 domain, suggesting that it can produce additional defects through inhibition of proteasome function by aggregation and formation of cytoplasmic inclusions (29). Our results show that it behaves exactly like WT STAT5B in terms of body length reduction, suggesting that this variant behaves as either benign or loss-of-function (16). Nonetheless, the fact that this variant can introduce developmental malformations in a higher number of embryos and of greater severity compared with WT STAT5B, evidences its pathogenic nature. Owing to the nature of the missense variant and its position in the SH2 domain, its ability to interfere with endogenous stat5.1 signaling probably occurs through the formation of hetero-dimers with endogenous stat5.1 despite it being less able to become phosphorylated.
On the other hand, variant p.Gln474Arg which has been reported in heterozygosis in a patient with a less severe phenotype both in terms of growth and immune dysfunction was normally expressed in vitro, was able to become phosphorylated upon GH stimulation, retained its ability to form homo- and hetero-dimers with WT STAT5b and translocated to the nucleus but could not bind DNA lacking transcriptional activity (15). In addition, it had reduced transcriptional activity when co-expressed with WT STAT5B, and thus was considered to act as a dominant negative variant. Our results show that overexpression of p.Gln474Arg led to a reduction of body length of zebrafish embryos compared with WT STAT5B. It is possible that p.Gln474Arg forms non-functional hetero-dimers with endogenous stat5.1, and once they translocate to the nucleus, result in reduced DNA binding affinity and diminished transcriptional activity (Fig. 9).
Schematic of the human GH-induced activation of STAT5b. Human STAT5b protein variants p.Ala630Pro, p.Gln474Arg and p.Lys632Asn are shown in bold and the potential mechanisms of interference in the endogenous stat5.1 intracellular signaling cascade are shown in boxes. All variants, including WT, p.Ala630Pro, p.Gln474Arg and p.Lys632Asn STAT5b can interfere with endogenous stat5.1 through the formation of less functional STAT5b/stat5.1 hetero-dimers. In addition, variant p.Ala630Pro can interfere with endogenous stat5.1 by aggregation and through the formation of cytoplasmic inclusions. Variant p.Gln4747Arg can interfere by reduced DNA-binding affinity of the STAT5b/stat5.1 hetero-dimers. Finally, variant p.Lys632Asn can interfere by its increased affinity of SH2 docking to activated GHR.
Previous in vitro assays showed variant p.Lys632Asn was well expressed but incapable of being phosphorylated suggesting a reduced ability to form homo- or hetero-dimers with endogenous stat5.1. If not through the formation of hetero-dimers, variant p.Lys632Asn may dock to the activated GHR with higher affinity thus blocking access for endogenous stat5.1 monomers (Fig. 9). The change of Lys632 for Asn in the SH2 domain could promote destabilization of this domain altering its ability to interact with other proteins (11). Nonetheless, SH2-independent dimerization mechanisms were evidenced demonstrating that non-phosphorylated STAT4 formed dimers through its N terminal domain (30). Whether this mechanism occurs in vivo for p.Lys632Asn or p.Ala630Pro STAT5b remains to be established. In addition, the transcriptional activity observed in vitro in cells co-transfected with p.Lys632Asn and WT STAT5b under GH stimulation is contrasted with our in vivo results. These differences can be attributed to the intrinsic limitations of reconstitution in vitro studies where the conditions differ greatly from in vivo studies performed in the context of whole organisms.
Taken together, our in vivo results in zebrafish show that although all three variants appear to be pathogenic, they act differently according to the nature of the missense mutation. Considering both the ability of variants p.Gln474Arg and p.Lys632Asn to induce a reduction of body length in zebrafish embryos when overexpressed as well as the ability of WT STAT5B of rescuing or partially rescuing this phenotype when co-injected with either variant, it seems that they behave as dominant negative mutations. The addition of WT STAT5B appears to be able to ameliorate the effects caused by the interference of endogenous stat5.1 signaling. On the contrary, the fact that co-injection of p.Ala630Pro/WT leads to a higher number of embryos with a greater severity of developmental malformations and a dramatic body length reduction, suggests that when these two variants are overexpressed together, they could impact different molecular levels of the STATs signaling pathway in a synergistic way rather than by the simple sum of the isolated expression of each variant. Although the underlying mechanism of the effects of overexpressing variant p.Ala630Pro together with WT STAT5B is not understood, it is clear that p.Ala630Pro behaves differently from the other studied variants and not through a dominant negative mechanism.
The mechanism through which all three STAT5B variants (including WT STAT5B) cause the developmental malformations observed remains to be elucidated. It is well known that STATs are pleiotropic proteins that have been implicated in a wide range of physiological and pathophysiological processes (6,31). Spinal deformity may be due to the reduction of sarcomere and/or myosin formation necessary for the development of a healthy musculoskeletal system (32), and stat5.1 is required for myogenesis and muscle morphogenesis in zebrafish (20). In addition, it is well known that STAT5b is widely expressed and has pleiotropic effects in different tissues during development. It remains to be studied if STAT5b has a specific role in kidney development since the pericardial edema observed in embryos overexpressing these variants is a pathognomonic sign of glomerular filtration defects (33). These developmental malformations may also be the result of excess STAT5b in tissues where normally endogenous stat5.1 is not expressed or is expressed in lower levels. Interestingly, recent studies showed that several collagen genes including col8a2, col14a1a, col28a2a and col5a1 were detected as targets of stat5.1 and the expression of most collagen genes was reduced during the initial occurrence of spine defects in stat3 mutants (19,34). These data implied that STAT family members might have critical roles in animal morphogenesis and maintenance of normal body shape. Furthermore, a recent study showed that zebrafish embryos exposed to N-(1,3-dimethylbutyl)-N′-phenyl-p-phenylenediamine (6PPD) resulted in a reduction of body length and deformities during development including bent spine and pericardial edema suggesting these malformations are related to changes at the level of the GH-IGF-1 axis (35). Moreover, increasing concentrations of 6PPD led to reduced levels of gh, ghra, ghrb, igf1, igf1ra and igf1rb in these embryos. Finally, a study using microRNAs showed that overexpression of miR-141/429a dramatically reduced body length when compared with control mimic-injected embryos or uninjected embryos while its ectopic expression led to pericardial edema in the embryos evidencing another example involving body length reduction and this specific malformation in the context of the GH-IGF-1 axis (36).
It is important to remark that there is not a correlation between the severity of the human phenotype caused by each one of the variants studied and their effect on growth and development in the zebrafish embryos. An example is variant p.Ala630Pro, a loss-of-function variant that results in a more severe clinical phenotype, but due to the fact that is less well expressed and results in an inactive STAT5b variant, appears to have a much less ability to interfere with the endogenous stat5.1 signaling pathway, resulting in minimal effects on growth and development in zebrafish embryos.
In order to understand the consequences of body length reduction of these embryos at the molecular level we investigated some of the key molecules of the GH-IGF-1 axis by in situ hybridization: gh1, stat5.1, igfals and igf1. Interestingly, loss of stat5.1 results in the downregulation of gh1 expression, likely reflecting the positive reciprocal feedback regulation between gh1 and stat5.1 in zebrafish, where stat5.1 protein can directly affect gh1 expression by binding to its promoter (19). It remains to be studied if this positive feedback loop exists in mammals. In spite of this reciprocal feedback regulation between gh1 and stat5.1 in zebrafish, our in situ results show that gh1 levels were not affected in embryos overexpressing WT STAT5B nor the three variants in the study. Embryonic, fetal, and early postnatal growth is GH-independent in humans and zebrafish. Accordingly, the vizzini gh1 mutant has no growth defects during early development (37–39). The reduction of expression of igf1 in embryos overexpressing p.Gln474Arg and p.Lys632Asn STAT5B mRNA probably accounts for the reduction of body length observed in these embryos (17.34% and 12.68% reduction, respectively). On the other hand, overexpression of WT or p.Ala630Pro led to a minimal reduction in body length compared with uninjected controls (2.91% reduction) with no apparent reduction of igf1 expression. This minimal reduction in body length, which is statistically significant owing to the large number of embryos used in the study, could result in a marginal reduction of igf1 expression levels which cannot be evidenced using in situ hybridization, a semi-quantitative technique. The reduction of igf1 mRNA levels appears to be independent from stat5.1 mRNA expression and is probably owing to the interference of human STAT5B mRNA at the protein level at different levels of the intracellular GH signaling pathways. Surprisingly, only overexpression of WT STAT5B led to a reduction of igfals expression at 48 hpf in the central nervous system (CNS). In previous studies from our group, we have shown that, during development, igfals is expressed in the CNS in zebrafish from 24 to 120 hpf (40). Although it has been clearly demonstrated in mammals that STAT5b is crucial for the transcription of the IGFALS gene in the liver (41), it is still unknown whether igfals expression is regulated by stat5.1 early during development in the CNS in zebrafish embryos. In addition, it is also unknown if als is capable of ternary complex formation together with igf1 and igfbp-3 in zebrafish. The molecular mechanism underlying the reduction of igfals expression in the CNS by WT STAT5B but not by the other variants will require further investigation.
Materials and Methods
Animals
WT zebrafish (Danio rerio) adults were maintained at 28°C on a 14 h light/10 h dark cycle and fed twice daily. Embryos were obtained by natural crossings. The staging was performed according to (42). All experiments were conducted following the National Institutes of Health guide for the care and use of laboratory animals and approved by the Comité Institucional para el Cuidado y Uso de Animales de Laboratorio or CICUAL (Institutional Committee for the Care and Use of Laboratory Animals) of the School of Medicine, University of Buenos Aires, Argentina.
Morpholinos
A splice-blocking MO targeting the first exon/intron junction of the zebrafish stat5.1 gene was provided by Gene Tools, LLC. The stat5.1 MO sequence is: 5’GTGAACTTGTGACTTACCAGAGTTG3`, 1 mM (20,21). The control MO is a standard control MO: 5’CCTCTTACCTCAGTTACAATTTATA3`. Stock MOs were diluted in water to 1 mM and working solutions were diluted in KCl 100 mM and Phenol red 0.25% (1 nl of 0.6 mM working concentration) (20).
Overexpression of human STAT5B mRNAs
The NCBI reference sequences used were NG_007271.1 (gene), NM_012448.3 (mRNA) and NP_036580.2 (protein). All STAT5B variants, p.Gln474Arg, p.Ala630Pro and p.Lys632Asn, were submitted to the ClinVar database (https://www.ncbi.nlm.nih.gov/clinvar). Human WT STAT5B cDNA was commercially obtained (pCMV6-Entry, RC209429, Origene, Rockville, MD, USA), whereas STAT5B variant p.Lys632Asn was obtained by site-directed mutagenesis as previously described (11). For in vitro synthesis of mRNA, pCMV6-Entry plasmids were linearized with XhoI, purified and transcribed using mMESSAGE mMACHINE T7 Transcription Kit (Thermofisher Scientific) followed by Poly (A) Tailing Kit (Thermofisher Scientific) according to manufacturers` instructions. Human STAT5B cDNA variants p.Ala630Pro and p.Gln474Arg subcloned into pcDNA3.1-FLAG-STAT5B were a kind gift from Dr Vivian Hwa (University of Cincinnati, OH, USA). In order to remove FLAG tag, these variants were PCR amplified (2.4 kb) with Q5 DNA Polymerase (NEB) and the following primers: F_T7_hSTAT5B 5’TAATACGACTCACTATAGGGAGAATGGCTGTGT GGATACAAGCTCAG3′ and R_hSTAT5B 5’TCACGATTGTGCGTGCG3`. Forward primer included T7 RNA polymerase promoter sequence (italics). Following purification, in vitro mRNA was obtained as mentioned before.
Microinjections
The MOs and mRNA solutions were microinjected into zebrafish embryos at the 1–2 cell stage. The injection volume was 1–2 nl. Injection was driven by compressed N2, under the control of a PV830 Pneumatic PicoPump (World Precision Instruments, Sarasota, FL). A total of 200 pg of WT or variants STAT5B mRNA was injected into each zebrafish embryo. Optimization experiments (not shown) were performed using different amounts of mRNA. We decided to use 200 pg as this was the lowest dose able to induce a phenotype with a minimal mortality rate. For co-injection of WT/variants STAT5B mRNA in a 1:1 ratio, 100 pg of each mRNA were used for a total of 200 pg mRNA per embryo. Following injection, embryos were reared in E3 medium (5 mM NaCL, 0.17 mM KCL, 0.33 mM CaCl2, 0.3 mM MgSO4 and 0.1% methylene blue) at 28°C until 72 hours post fertilization (hpf).
Whole-mount in situ hybridization
For probe synthesis, cDNA from 48 hpf embryos was used as a template for PCR amplification. To generate cDNA, RNA was extracted from 48 hpf embryos with Trizol (Thermofisher Scientific) followed by DNAse treatment to eliminate DNA contamination and 1 μg of RNA was reversely transcribed using MMLV enzyme (Invitrogen) and random primers (Promega). Probes used were: igfals, gh1, stat5.1 and igf1. Probe igfals was generated as previously described (40). All probes were PCR amplified using cDNA as template and primers: gh1 F_gh1 5’ATGGCTAGAGCATTGGTGCTGT3` and R_gh1 5’CTACAGGGTACAGTTGGAATCCAG3` (633 bp) (43), stat5.1 F_stat5.1 5’TGAACACTTGACTCAGCAGCTGCC3` and R_stat5.1 5’TATCAGGCACCACAAATGGCACT3` (617 bp), igf1 F_igf1 5’TAGAGGACAGCGGGAGGAAT3´ and Igf1_R 5’GGGGACGATTAAATTCTCCA3` (428 bp). All probes were cloned into pGEM-T Easy (Promega); plasmid vectors were linearized with PstI and antisense digoxigenin riboprobe were generated using T7 RNA transcriptase and DIG RNA labeling kit (Roche) according to manufacturers` instructions. Embryos were dechorionated and fixed overnight in 4% (w/v) paraformaldehyde at 4°C before whole-mount in situ hybridization (WISH) with DIG-labeled anti-sense probes. WISH was performed as previously described (44). Zebrafish embryos at 48 hpf were cleared in 2:1 Benzyl Benzoate/Benzyl Alcohol and mounted in Canada balsam for photography in a Nikon Eclipse E200 microscope coupled to a Nikon Y-TV55 camera.
Data collection and analysis
Live embryos were anesthetized with 0.04% MS-222 (Sigma Aldrich, St. Louis, MO), mounted in 3% methylcellulose and photographed with a Nikon Eclipse E200 microscope coupled to a Nikon YTV55 camera using Micrometrics SE Premium software. All experiments were repeated at least three times. Mean ± Standard Error of the Mean (SEM) of results from multiple experiments of the same study are reported. Data were analyzed with Mann–Whitney test for non-parametric data using GraphPad (Prism) software (GraphPad Software, San Diego, CA) and significance was set at P < 0.05.
Acknowledgements
We would like to thank Mariana Cruz for fish care and maintenance and Dr Vivian Hwa for providing human STAT5b variant plasmids (p.Ala630Pro and p.Gln474Arg).
Conflict of Interest statement. None declared.
Funding
Consejo Nacional de Investigaciones Científicas y Tecnológicas (CONICET); Proyecto Unidad Ejecutora, Argentina (131); Agencia Nacional de Promoción Científica y Tecnológica; Préstamo BID PICT 2018, Argentina (579).
Data availability
All data generated or analysed during this study are included in this published article [and its supplementary information files].
