Dominant-negative STAT5B mutations cause growth hormone insensitivity with short stature and mild immune dysregulation

Growth hormone (GH) insensitivity syndrome (GHIS) is a rare clinical condition in which production of insulin-like growth factor 1 is blunted and, consequently, postnatal growth impaired. Autosomal-recessive mutations in signal transducer and activator of transcription (STAT5B), the key signal transducer for GH, cause severe GHIS with additional characteristics of immune and, often fatal, pulmonary complications. Here we report dominant-negative, inactivating STAT5B germline mutations in patients with growth failure, eczema, and elevated IgE but without severe immune and pulmonary problems. These STAT5B missense mutants are robustly tyrosine phosphorylated upon stimulation, but are unable to nuclear localize, or fail to bind canonical STAT5B DNA response elements. Importantly, each variant retains the ability to dimerize with wild-type STAT5B, disrupting the normal transcriptional functions of wild-type STAT5B. We conclude that these STAT5B variants exert dominant-negative effects through distinct pathomechanisms, manifesting in milder clinical GHIS with general sparing of the immune system. Severe growth hormone insensitivity syndrome (GHIS) with immunodeficiency is caused by autosomal recessive mutations in STAT5B. Here the authors report heterozygous STAT5B mutations with dominant-negative effects, causing mild GHIS without immune defects.

P atients diagnosed with growth hormone insensitivity syndrome (GHIS) share common clinical characteristics of impaired postnatal growth due to low or undetectable serum IGF1 concentrations despite normal or elevated growth hormone (GH) concentrations [1][2][3] . Autosomal-recessive (AR) mutations of the GH receptor, GHR ("Laron syndrome", MIM 262500) are the most prevalent molecular cause of GHIS, although a few autosomal-dominant (AD) cases have also been reported (MIM 604271) [4][5][6][7][8][9] . STAT5B deficiency (MIM 245590), a rare cause of GHIS with immunodeficiency, is an AR disorder, first described in a patient with severe short stature (height SDS −7.5), who was T-lymphopenic 10,11 and succumbed to a progressive pulmonary disease 12 . STAT5B, typical of STAT proteins, is composed of discrete protein modules including a 4-alpha helix coiled-coiled domain (CCD), a DNA-binding domain (DBD), an SH2 (src-homology 2) domain for docking to phosphorylated tyrosines, and a C-terminal transcriptional activation domain (TAD). All seven recessively inherited inactivating STAT5B mutations characterized to date lack functional SH2 and downstream TAD domains, and often the entire protein is immunologically undetectable 13 . One copy of wild-type (WT) STAT5B allele appears to be sufficient for normality as heterozygous relatives of affected patients are of normal height and without immunological or pulmonary complications 14 . Since STAT5B functions as a dimer when activated, it is conceivable that natural heterozygous STAT5B variants exist which disrupt dimeric functions. Recurrent somatic activating heterozygous missense STAT5B mutations in the SH2 or TAD domains, for example, were recently identified and reported to be causal of lymphomas [15][16][17] . Germline heterozygous STAT5B variants associated with impaired human growth and/or immunity have yet to be identified. Our previous functional evaluations of two rare heterozygous STAT5B missense variants identified in children with idiopathic short stature had demonstrated that the variants were unlikely to be the sole cause of growth failure 18,19 .
We now report the first germline heterozygous STAT5B variants with dominant-negative effects, identified by targeted and whole-exome sequencing (WES), in short-statured subjects from three unrelated families. Neither the index patients nor affected relatives suffer from severe immunological disturbances. The three missense mutations retain the capability to become robustly tyrosine phosphorylated upon GH stimulation and, subsequently, to form dimers with themselves or with the STAT5B WT protein. However, their capacity to act as a transcription factor is blunted since nuclear import is abrogated in STAT5B with a mutation mapping to the CCD domain (p.Gln177Pro) while STAT5B proteins with DBD mutations fail to bind canonical STAT5B DNA response elements (p.Gln474Arg, p.Ala478Val). Importantly, each mutant STAT5B protein interferes with the normal functions of the WT isoform. Altogether, we demonstrate that specific heterozygous STAT5B germline mutations exert dominant-negative effects resulting in STAT5B deficiency clinically characterized by significant postnatal growth impairment, mild GH insensitivity, eczema, and elevated IgE.

Results
Patients and variant identification. The clinical profiles of male index patients from families 1 and 2 were consistent with GHIS, with postnatal growth failure, serum IGF1 concentrations close to (Proband 1) or below the detection limit (Proband 2), whereas basal and stimulated GH serum concentrations were normal (Table 1). For Proband 1, low serum concentrations of acid-labile subunit (IGFALS) corroborated the diagnosis of GHIS 13 ; for Proband 2, the lack of response to exogenous GH in an extended stepwise IGF1 generation test (Supplementary Table 1) confirmed a state of GHIS. Moreover, affected siblings ( Fig. 1a; the monozygotic twin in family 1 and a brother in family 2) presented with comparable biochemistries and growth profiles (see "Detailed patient reports" in Supplementary Note 1 and Supplementary  Table 2). Initial targeted sequencing of key genes along the GH-IGF1 axis revealed a de novo heterozygous STAT5B variant in the twin brothers of family 1 (c.530A > C, exon 5, p.Gln177Pro; CCD, Fig. 1b) and a maternally inherited c.1433C > T variant (exon 12, p.Ala478Val, DBD) in family 2. Subsequent WES analysis, performed in both families, excluded additional pathogenic variants (for criteria used in the WES analysis pipeline, see Supplementary Note 1 and Supplementary Tables 3 and 4).
For Proband 3, family 3, the top candidate variant from WES analysis was a heterozygous STAT5B c.1421A > G variant (exon 12, p.Gln474Arg; DBD). A novel heterozygous JAK2 variant (c.2374C > T, p.Pro792Ser) was also identified but did not segregate with the clinical phenotype of the family (Supplementary Note 1 and Supplementary Table 4). Proband 3 had familial short stature accompanied by autoimmune thyroiditis, celiac disease, and poor growth response to rhGH treatment (Table 1). Interestingly, Proband 3 and her two siblings who had short stature and carried the variant, but not the affected father, also presented with microcephaly (Supplementary Note 1 and  Supplementary Table 2). Re-analysis of the exome data for potential variants that might contribute to the microcephalic phenotype, however, were unrevealing. Of note, probands from families 1 and 2 were not microcephalic.
None of our patients presented with symptoms of severe immune dysfunction normally associated with STAT5B deficiency 20 , although the majority of carriers of the identified STAT5B variants had eczema (Fig. 1a), Proband 3 had autoimmune thyroiditis and celiac disease which were successfully controlled, and Proband 1 had childhood bronchial asthma. Immunological evaluations performed for STAT5B variant carriers in the three families were normal, with the exception of elevated IgE concentrations in eight out of nine patients (Supplementary Table 2 and Supplementary Data 1).
Detailed patient phenotype and genetic analyses reports, and immunological profiles of the three families can be found in the Supplementary Information. These are the only three families in whom a dominant-negative STAT5B mutation has been identified. Over the past decade, our research groups have investigated a total of 164 children with marked short stature and overlapping phenotypes of GHIS accompanied by variable symptoms suggestive of immune dysregulation or in whom initial GHR sequencing did not reveal any pathogenic genomic aberration.
Mutated STAT5B proteins are phosphorylated and can dimerize. Each of the three non-synonymous STAT5B variants is private and not listed in the large-scale variant databases, with amino acid substitutions predicted to be pathogenic (Supplementary Table 5). To evaluate functional pathogenicity, Nterminally tagged STAT5B p.Gln177Pro, p.Ala478Val, and p. Gln474Arg variants were re-generated. Expression of each variant and GH-induced Tyr-phosphorylation (pSTAT5) were shown to be comparable to those of tagged WT STAT5B in reconstituted HEK293(hGHR) systems (Fig. 2a, upper panels). Coimmunoprecipitation (co-IP) experiments, furthermore, supported homo-dimerization capabilities for each variant, and, mimicking a heterozygous state, ability to hetero-dimerize with WT STAT5B (Fig. 2a, bottom panels). Interestingly, p.Gln177Pro demonstrated reproducible and robust, GH-induced phosphorylation which was markedly and time-dependently sustained (Fig. 2b). This enhanced STAT5 phosphorylation was corroborated in primary dermal fibroblasts (Fig. 2c, d) stimulated with GH or interferon gamma (IFNγ). Enhanced phosphorylation (Fig. 2d) was not due to changes in basal STAT5B or GHR expression patterns which were similar to those in control fibroblasts (Fig. 2c, e).
Nuclear import of activated STAT5B p.Gln177Pro is abrogated. We next asked whether all phosphorylated STAT5B species could translocate to the nucleus, a necessary step for transcriptional actions. Immunofluorescent staining and deconvolution microscopy revealed that the tagged p.Gln177Pro variant, despite robust GH-induced phosphorylation and in contrast to WT and the other two STAT5B variants, failed to enter the nucleus, remaining localized to the cytoplasm and accumulated at the nuclear membrane (Fig. 3a). Strikingly, when co-expressed with WT STAT5B, not only did the p.Gln177Pro variant remain cytoplasmic upon GH stimulation, but co-localization of the variant and WT STAT5B was detected only in the cytoplasm and by the nuclear membrane (Fig. 3b, bottom panels). Our results strongly suggest that the p.Gln177Pro variant was defective in nuclear localization and, when dimerized with WT STAT5B, prevented the WT from mobilizing into the nucleus. Conversely, WT STAT5B could not facilitate the nuclear localization of associated p.Gln177Pro, suggesting that motifs or regions within each of the associated monomers may both be required for appropriate nuclear import. STAT5B p.Gln474Arg and p.Ala478Val are DNA-binding defective. For the p.Gln474Arg and p.Ala478Val variants, located within the DBD module, we evaluated whether DNA binding might be impaired, employing standard gel-shift electrophoretic mobility shift assay (EMSA) analysis. WT STAT5B readily bound and gel-shifted the STAT5B-specific DNA probe (GHRE) only under GH-stimulated conditions (Fig. 3c). However, neither of the two variants could gel-shift the probe, clearly demonstrating loss of DNA-binding functions. When Myc-tagged STAT5B variants were co-expressed with WT FLAG-STAT5B, gel-shifting was observed under GH-stimulated conditions (Fig. 3d), but this was attributed to bound WT FLAG-STAT5B, as the complexes were further shifted (supershifted) by anti-FLAG antibody and not by anti-Myc antibody (Fig. 3d). These results were consistent with the absence of p.Gln474Arg or p.Ala478Val peptides from the GHRE probe/protein complexes, suggesting that the protein The failure of STAT5B variants p.Gln474Arg and p.Ala478Val to bind GHRE parallels the inability to drive transcriptional activities as demonstrated by STAT5B-mediated luciferase reporter assays. Neither of the two variants by themselves could drive expression of the luciferase reporter ( Fig. 3e) and when coexpressed with WT STAT5B, the transcriptional activities of WT STAT5B were significantly blunted compared to WT STAT5B alone. Altogether, the results provide strong evidence that these DBD variants ablated STAT5B DNA-binding capabilities and prevented associated WT STAT5B from binding DNA and functioning as a transcription factor.

Discussion
To date, the clinical condition of STAT5B deficiency has been described exclusively as an AR trait, with the majority of mutations causing early protein termination (nonsense, frameshift mutations) and total STAT5B deficiency. Only two homozygous STAT5B missense mutations have been reported, both mapping to the SH2 domain and affecting overall protein stability and/or functions 10,21 (Fig. 1b). Here we report the identification of three novel germline STAT5B missense variants, with demonstrable dominant-negative effects, associated with short stature and mild GHIS in three unrelated families. The two loss-of-function missense variants in the DBD module are four amino acids apart, while p.Gln177Pro is located in the CCD module ( Supplementary  Fig. 1a, b). Each of the variants interacted with WT STAT5B, exerting dominant-negative effects that, ultimately, reduced transcriptional activity.
The p.Gln177Pro substitution, located towards the C-terminal end of the first α-helix in the CCD ( Supplementary Fig. 1e), ablated ability to nuclear localize despite robust expression and GH-induced phosphorylation. The predicted disruption of α-helix 1 by the proline substitution, surprisingly, did not destabilize the whole protein (unlike the STAT5B p.Ala630Pro 10 ). However, the functional integrity of the four α-helices that comprise the CCD was likely disrupted, as the helix bundle was recently shown to act in concert as an unconventional nuclear localization signal 22 . The inability of phosphorylated p.Gln177Pro to translocate to the nucleus, where resident nuclear phosphatases act on phosphorylated STAT proteins 23,24 , could explain the accumulation of cytosolic, phosphorylated STAT5B species. Interestingly, impaired nuclear dephosphorylation also has been reported for AD, gain-offunction, STAT1 CCD mutants that are associated with chronic mucocutaneous candidiasis (MIM 614162), but none were reported as defective in nuclear localization 23,25 . CCD loss-offunction missense mutations in STAT3 (ref. 26 ), STAT2 (ref. 27 ), and STAT6 (ref. 28 ) have yet to be reported. Hence STAT5B p. Gln177Pro is the first described STAT variant with a nuclear localization defect and distinct dominant-negative actions. The normal phosphorylation and nuclear trafficking of our dominant-negative STAT5B DBD mutants, p.Gln474Arg and p. Ala478Val, is remarkably similar to characterized, dominantnegative STAT3 DBD mutations 29-31 associated with hyper-IgE syndrome (MIM 147060) 31,32 . The affected STAT5B residues, Gln474 and Ala478, are conserved among all members of the STAT family ( Supplementary Fig. 1a). Gln474 resides within, and Ala478 close to, a sequence segment that occupies the major grove of bound DNA in all previously investigated STAT homologous sequences (STAT1 (ref. 33 ), STAT3 (ref. 34 ), STAT6 (ref. 35 ); Supplementary Fig. 1b-d). The STAT5B p.Gln474Arg, in particular, would be predicted to directly interfere with DNA response element recognition, possibly by disrupting direct hydrophobic and polar contacts between glutamine and thymines and phosphates on the DNA backbone 34 , as was suggested for the analogous heterozygous STAT3 p.Gln469His and p.Gln469Arg mutations ( Supplementary Fig. 1c, d). Altogether, heterozygous expressed missense mutations within the STAT DBD domains can negatively impact normal dimeric STAT transcriptional functions, resulting in clinical syndromes and disorders.
All three heterozygous loss-of-function STAT5B missense variants lead to postnatal growth restriction for the index patients (height ranged from −2.9 to −5.3 SDS). Growth impairment in our patients is likely due to the consequence of diminished IGF1 serum concentrations and, presumably, extrahepatic resistance to GH, e.g. at the growth plate 36 . Intra-and interfamilial variability of stature in the three families was comparable to that normally seen in AR GHIS 1 . Strikingly, the growth deficit was in the range of patients carrying dominant-negative GHR mutations [4][5][6][7][8][9] (Supplementary Fig. 2), although heights were above −2 SDS in 3 out of the 11 individuals carrying dominant STAT5B mutations, one of whom, II-6 of family 3, is quite young and will be closely monitored. Overall, however, the phenotypes of our patients are clearly distinct from those of heterozygous relatives of AR STAT5B-deficient patients who typically have heights within the low normal range 14 (Supplementary Fig. 2).  In Family 3, the father (I-1) also carried a private heterozygous JAK2 variant c.2374C > T (p.Pro792Ser), which was transmitted to five of his six children, one of whom (II-4) did not carry the STAT5B p.Gln474Arg mutation. The binding of JAK2 by the homo-dimeric GHR, which, like all Type I cytokine receptor, lack intrinsic kinase activities, is crucial for activating the GH-induced GHR signaling cascades. Since II-4, age 7.9 years, was of normal stature (height SDS of −1.1), consistent with his normal IGF1 serum concentrations, we concluded that the JAK2 variant itself is unlikely to contribute significantly to postnatal growth retardation. With a sample size of only one, however, a potential synergistic effect with the STAT5B p.Gln474Arg mutation cannot be entirely ruled out. Subject II-4, furthermore, was non-syndromic (healthy, no evidence of eczema, normal IgE concentrations [Supplementary note 1]), suggesting that the JAK2 variant, by itself, did not contribute to the other symptoms observed in affected family members. To date, loss-of-function JAK2 mutations have not been reported while somatic and germline dominantly inherited gain-of-function JAK2 variants, particularly the recurrent p.Val617Phe, are well established causes of hematologic disorders (thrombocythemia-3, thrombocytopenia, polycythemia vera) [37][38][39] . In family 3, there was a lack of symptoms indicative of similar hematologic abnormalities, suggesting JAK2 p.Pro792Ser is not a dominant gain-of-function mutation. Altogether, the heterozygous JAK2 p.Pro792Ser is unlikely to be a main contributor towards the clinical symptoms Our patients lacked the severe immune deficiency and pathological autoimmunity typically associated with total STAT5B deficiency. The immune complications of STAT5B deficiency are attributed to T-lymphopenia, most notably impairment in numbers and functions of T-regulatory cells (Treg) that express the transcription factor FOXP3, Treg + FOXP3 + , which are critical for maintenance of T-cell homeostasis 11,40 . Specifically, in STAT5B deficiency, it has been noted that peripheral, naïve Treg subsets were low, while dividing memory Treg subsets (CD45RO + Ki67 + ) were abnormally elevated and had reduced suppressive functionality, thus likely contributed to the autoimmune phenotypes of STAT5B-deficient patients 40,41 . Immunological profiles (B, NK, T-lymphocytes) of our present patients, in contrast, were unremarkable, with T-lymphocytes and subsets (Treg, Treg + FOXP3 + , naïve, and memory cells), in particular, to be within normal ranges (Supplementary Data 1).
An unexpected phenotypic feature in Family 3 was microcephaly in Proband 3 and two of her affected siblings but not in the carrier father. Whether the reduced head circumference was a consequence of the AD STAT5B mutation or, more likely, caused by additional genetic variants, remains unclear, but no obvious recessive or dominant candidate variants were revealed from available exome data. It is notable that the relatively normal head circumferences, albeit below the mean, of patients from families 1 and 2 were comparable to those observed in patients with GHIS 42 (Supplementary Fig. 3).
Additional major features frequently seen in AR STAT5B deficiency such as hyperprolactinemia 13 were detected in only a few of our AD STAT5B subjects (mother and aunt of family 2; Proband 3), and extrapulmonary autoimmune disease 41 was diagnosed only in Proband 3 (thyroiditis and celiac disease). Interestingly, elevated IgE concentrations manifested in only four of seven tested AR STAT5B-deficient patients 12,20,41 whereas eight of our nine AD STAT5B-deficient patients presented elevated IgE concentrations with the high penetrance, suggesting strong correlations between dominant actions of our STAT5B mutant proteins and IgE production. Mild eczema was reported for the majority of carriers of dominant-negative STAT5B mutations. Altogether, the clinical manifestations (i.e. growth impairment, elevated IgE, eczema) suggest variability in penetrance of the mutant genotypes and are consistent with the milder phenotype of AD STAT5B loss-of-function mutations (Supplementary Table 2 and Supplementary Fig. 2).
The prevalence of inactivating STAT5B mutations, either AR or AD, remains difficult to accurately determine from studies performed to date. Previous screenings for mutations or copynumber variants performed in short-statured patients (total n = 377), for example, did not reveal any pathogenic STAT5B variants 18,[43][44][45][46] , whereas in our present study (164 patients), three dominant-negative STAT5B mutations were identified. Future screenings of GHIS patients will validate whether the additional features of eczema and elevated IgE may be indications of potential inactivating AD STAT5B mutations. Intriguingly, each of the described loss-of-function germline STAT5B mutation is private, while reported somatic, heterozygous, STAT5B gain-of-function mutations associated with large granular lymphocytic leukemia or distinct types of T-or NK-lymphomas are clustered, often with the same mutation found in more than one individual [15][16][17][47][48][49] . The observation that STAT5B mutations, both germline loss-of-function and somatic gain-of-function, lead to clinical manifestations associated with dysfunctional T-lymphocytes, support critical roles of STAT5B in T-lymphocyte biology which cannot be compensated by the closely related STAT5A.
Both the clinical presentation of the patients and functional studies support residual normal STAT5B activities when in the heterozygous state. Since STAT5B actions require dimerization, we hypothesize that mutant and WT monomers captured in WT: mutant heterodimers are functionally inactive and only the remaining predicted 25% WT:WT homodimers are transcriptionally competent. While 25% of dimeric WT STAT5B is clearly not adequate for full GH responsiveness, cytokine signaling appears to be less sensitive to diminished concentrations of active STAT5B, thus explaining the absence of severe immunologic or pulmonary problems in our patients. Interestingly, in mouse models, severe growth restriction and immune deficiency were observed only in the knock-out of both the Stat5b and Stat5a genes, but expression of normal Stat5a/b as low as~10% significantly reduced the severity of immunodeficiencies 50,51 . Finally, in our patients, the lack of severe immune problems and presence of residual WT STAT5B suggested that recombinant human GH or rhIGF1 therapy might be effective for normalizing growth, possibly dependent on the type of mutation. Effectiveness of either therapy or a combination of both treatments remains to be fully evaluated.
In conclusion, we report the first germline dominant-negative STAT5B mutations observed in patients who demonstrated milder but significant postnatal growth impairment, mild GH insensitivity, eczema, and elevated IgE. The STAT5B p.Gln177Pro variant is the first naturally occurring STAT mutation with a defect in nuclear localization, while the p.Gln474Arg and p. Ala478Val variants are DNA-binding deficient. These novel STAT5B mutations, acting through distinct pathophysiological mechanisms, manifested as milder clinical GHIS with general sparing of the immune system, thus broadening the clinical spectrum of STAT5B deficiency and GHI syndrome.

Methods
Patient consent and regulatory compliance. Written informed consent to participate in the study was obtained from all adult subjects and parents of minors. The study protocol including skin biopsy sampling from control individuals was in accordance with the principles of the Declaration of Helsinki and was approved by the Ethical Review Committee of the University of Leipzig, the Institutional Regulatory Board of the Cincinnati Children's Hospital Medical Center, and the Institutional Regulatory Board of the UCL Great Ormond Street Hospital Institute of Child Health, London.
WES and Sanger DNA Sequencing analyses. Genomic DNA extracted from whole blood was used for Sanger dideoxy-sequencing and WES. All coding exons of the genes indicated in the "Detailed patient reports" (Supplementary Note 1), including STAT5B were PCR amplified and subjected to direct sequencing (ABI PRISM 3100 and 310 Genetic Analyzers; Thermo Fisher Scientific, Waltham, MA). Sequences were compared to public references (for accession numbers see Supplementary Table 5). Primer sequences can be provided on request.
WES was performed on genomic DNA samples from individuals indicated in Supplementary Note 1 ("Detailed Patient Reports"). One microgram of dsDNA was sheared by sonication to an average size of 200 base pair (bp) on a Covaris S2 instrument. Library construction was performed in an automated fashion on an IntegenX Apollo324. After nine cycles of PCR amplification using the Clonetech Advantage II kit, 1 µg of genomic library was recovered for exome enrichment using the NimbleGen EZ Exome V2 kit. Libraries were sequenced on an Illumina HiSeq2500, generating approximately 30 million paired end reads 125 bases long each. Reads were aligned to the human reference genome version 19 (GRCh37) with the Burrows Wheeler Aligner 52 . Our analysis methods utilized the Broad Institute's Genome Analysis Toolkit (GATK) and followed the pipeline described in DePristo et al. 53 along with the modifications listed in the "Best Practices" document on their website [http://www.broadinstitute.org/gatk/]. Stringent filtering was applied for rare, non-synonymous variants that passed quality filters and had a read depth of >10 reads. Further filtering was performed depending on the inheritance mode of short stature and GHIS characteristics in each family, specifically de novo dominant, homozygous recessive, compound heterozygous, or X-linked form of inheritance in family 1; in family 2, a dominant mode of inheritance from the mother's side of the family, X-linked recessive inheritance, homozygous autosomal recessive, or a compound heterozygous pattern as well as dominant inheritance; and dominant inheritance in family 3. For dominant modes of transmission, given the severity of the phenotype, we excluded all variants Immunofluorescent microscopy. HEK293(hGHR) cells, 5 × 10 4 per well, were seeded on poly-lysine-coated eight-well chamber slides (Thermo Fisher Scientific), and transfected with a total of 0.5 μg plasmid DNA. For co-transfections, a 1:1 ratio of plasmid DNA was employed. Following GH treatment (100 ng ml −1 , 30 min), cells were washed in cold PBS and fixed with cold methanol, 5 min at −20°C. For immunofluorescent staining, cells were blocked with 5% goat serum and exposed overnight to primary mouse-anti-FLAG (1:500) and/or rabbit-anti-Myc (1:50) antibodies prior to treatment with secondary goat-anti-mouse-FITC (1:500; 626511; Thermo Fisher Scientific, Carlsbad, CA) and/or goat anti-rabbit-A555 antibodies (1:500; #A-21428; Life Technologies) for 4 h. Hoechst 33342 (Thermo Fisher Scientific) was added (15 min) and coverslips mounted with ProLong Gold (Thermo Fisher Scientific). Fluorescent images were obtained using a wide field deconvolution system (GE Healthcare) consisting of an inverted Nikon TE 200 Eclipse microscope, a Kodak CH350 CCD camera, and the Deltavision operating system. Images were acquired using a ×60 objective in a 1024 × 1024 format and deconvolved with nine iteration using SoftWoRx software (Applied Precision, GE Healthcare).
Luciferase reporter assays. HEK293(hGHR) cells (six-well plates) were transiently co-transfected 21 as described above with 1.0 μg pGL2: 8xGHRE (growth hormone response element) luciferase reporter plasmid and a total of 1.0 µg STAT5B WT and/or mutant cDNA plasmids or empty vector (each 0.5 μg) as indicated. Transfected cells were incubated overnight prior to an 8 h serum-free media treatment. Cells were stimulated with GH (100 ng ml −1 , 24 h) and cell lysates were collected and assayed using Promega's Luciferase Assay System (Promega, Madison, WI). Luciferase activity was measured on a Veritas Luminometer (Promega).
Electrophoretic mobility shift assay. 5′-Cy5.5-tagged rat spi2.1 GHRE (growth hormone response element; 5′-ACGCTTCTACTAATCCATGTTCTGAGAAA TCATCCAGTCTGCCCA-3′) was employed as probe for EMSAs 56,57 . Briefly, 50 fmol of the duplex DNA probe were incubated with 2 μg (standard EMSA) or 3 μg (supershift assays) nuclear extracts (see above) at room temperature for 20 min, protected from light, prior to size fractionation on a native 5% acrylamide gel (in standard tris-borate-EDTA buffer). Bands were visualized by scanning the gel on Li-Cor Odyssey CLx (Li-Cor Biotechnology, Lincoln, NE). For supershift EMSA, 2 μg of anti-FLAG antibody or 3 μg of anti-Myc antibody were included in the reaction mix and incubated at room temperature for an additional 10 min prior to size fractionation of the DNA-protein-antibody complexes.
Statistical analyses. Data were analyzed by one-way ANOVA with post hoc Tukey test. Statistical analysis of the data was performed using GraphPad Prism 5 software (GraphPad Software, La Jolla, CA).
Data availability. The authors declare that all the data supporting the findings of this study are included in the article (or the Supplementary material) and available from the corresponding author (V.H.) upon reasonable request. Data of pathogenic mutations reported within this study have been deposited in ClinVar with the accession codes SCV000681436 (p.Gln177Pro), SCV000680478 (p.Gln474Arg), and SCV000693651 (p.Ala478Val). WES data that support the findings are not publicly available due to information that could compromise the research participants privacy/consent.