STAT3 haploinsufficiency is associated with autosomal dominant hyper-IgE syndrome

Abstract

The autosomal dominant hyper-IgE syndrome (AD-HIES) is a primary immunodeficiency, which originates from heterozygous missense mutations in the signal transducer and activator of transcription 3 (STAT3) gene. It is accepted that most STAT3 variants causing AD-HIES are dominant negative. Whether haploinsufficient mutations cause a phenotype in humans is still debated. We report on a family with a heterozygous STAT3 nonsense mutation that led to rapid decay of the mutant mRNA and protein, leading to haploinsufficiency. To explore STAT3 heterozygosity, we created a Stat3 haploinsufficient (Stat3^+/−) mouse model in which we found that Stat3^+/− mice had increased IgE serum levels, reduced T_H17 cell differentiation, and were susceptible to a cutaneous Staphylococcus aureus infection. Together, our findings provide mechanistic evidence for the impact of haploinsufficiency in STAT3 with residual protein expression as an important cause for immune deficiency. The implications extend to the diagnosis of immunodeficiency disorders and to the design of gene therapy in situations where gene dosage matters.

The heterozygous STAT3 nonsense mutation Y657* leads to haploinsufficiency and AD-HIES–related symptoms.

INTRODUCTION

Autosomal dominant hyper–immunoglobulin E (IgE) syndrome (AD-HIES) is a rare primary immunodeficiency arising, in most cases, as a consequence of impaired cytokine-induced signal transduction due to heterozygous mutations in STAT3, encoding the signal transducer and activator of transcription 3 (STAT3) (1, 2). AD-HIES is a multisystem disorder, characterized by both immunologic and nonimmunologic manifestations. High serum IgE levels, eosinophilia, eczema, and skin and lung infections constitute the immunologic profile of AD-HIES, whereas characteristic facial appearance, scoliosis, retained primary teeth, joint hyperextensibility, bone fractures following minimal trauma, and craniosynostosis are the main nonimmunologic manifestations (3, 4).

STAT3 is a central component of multiple Janus kinase (JAK)/STAT signaling pathways, acts downstream of many cytokine receptors which are key to lymphocyte function, and is crucial for the cytokine-mediated differentiation of T helper 17 (T_H17) cells (5–7). In AD-HIES, mutations in STAT3 affect RORγt up-regulation impairing T_H17 cell development (8–10). This explains in part the susceptibility of these patients to bacterial and fungal infections, since T_H17 cells have been found to be essential for the immune response against these pathogens (11).

Most of the STAT3 dominant-negative (DN) mutations are inherited or de novo missense mutations. Yet, in-frame deletions, small in-frame insertions, essential splice-site mutations resulting in small in-frame deletions, or exon skipping has also been described in patients with AD-HIES (1, 2, 12–14). Consistent with the observation that complete loss of STAT3 leads to early embryonic death in mice (15), no biallelic loss-of-function mutations have been described in humans. A large-scale functional analysis of all reported pathogenic STAT3 variants in vitro suggested that more than 95% of all variants exert a DN effect through the expression of aberrant proteins (caused by premature stop of translation, reinitiation of translation, or alternative transcripts) rather than by haploinsufficiency (HI) (16). Accordingly, this study proposed negative dominance to be the only mechanism underlying AD-HIES in patients heterozygous for pathogenic STAT3 variants.

Here, we describe biological and clinical consequences of a heterozygous STAT3 nonsense mutation (p.Y657*, located in the SH2 domain), which led to rapid decay of the mutant protein due to nonsense-mediated decay (NMD), and protein instability. We experimentally found that this mutation did not have DN effects, supporting the possibility of AD-HIES was caused by STAT3 HI. To evaluate in vivo whether STAT3 HI can cause AD-HIES phenotypic alterations, we created a mouse model of Stat3 HI (Stat3^+/−). These mice recapitulated many of the aspects observed in patients with AD-HIES, such as increased serum IgE levels and reduced capacity of naïve CD4⁺ T cells to differentiate to T_H17 cells. Moreover, Stat3^+/− mice were more susceptible to cutaneous Staphylococcus aureus infection, with diminished interleukin-17 (IL-17)–producing cells. Together, our model reveals important insights into the biological consequences of insufficient STAT3 in patients with AD-HIES.

RESULTS

The STAT3 nonsense mutation Y657* leads to NMD and STAT3 protein instability

We performed genetic analyses in a severely affected female with suspected AD-HIES, born to nonconsanguineous parents (Fig. 1A). The patient suffered from newborn skin rash starting at day 20 of life, which was followed by refractory eczema. She developed S. aureus skin abscesses and suffered from generalized skin itching. Both serum IgE levels and blood eosinophil counts were highly elevated (Table 1). Her father presented with an overall milder disease phenotype, with a history of skin lesions from infancy accompanied with infections until the age of 20 years (Table 1), while the mother and the parent’s second daughter were healthy. Sequencing of all coding STAT3 exons including the intron-exon boundaries in both affected family members revealed a heterozygous nonsense mutation in exon 21 (c.1971delT) which is predicted to lead to premature termination of translation (Y657*) and thus a truncation of the C-terminal part of the SH2 domain and the transactivation domain of STAT3 (Fig. 1B). The mutation was absent in both unaffected family members. We excluded the presence of mutations in other HIES-associated genes and in genes known to interfere with STAT signaling by whole-exome sequencing. We also excluded an additional STAT3 mutation in the promoter region of STAT3 in the father and affected daughter by sequence analysis. Therefore, we concluded that the unique heterozygous STAT3 nonsense mutation accounted for the phenotype observed in both affected family members. Since the truncation of the C-terminal 114 amino acids in STAT3-Y657* partially deletes the SH2 domain and the entire transactivation domain, we tested whether a truncated STAT3 protein was produced in cells harboring the heterozygous nonsense mutation. We measured protein levels with antibodies directed against the C- or N-terminal part of STAT3 in whole cell lysates from Epstein-Barr virus–transformed B (EBV-B) cells, derived from three different healthy donors (HDs), two patients with AD-HIES, each carrying one of the most frequent DN missense mutations, the R382W and the V637M mutation (17), and from the patient with Y657* mutation. As expected, STAT3 levels were detectable with both antibodies in EBV-B cells from HD, with comparable levels in lysates derived from EBV-B cells with the R382W mutation. On the contrary, STAT3 levels analyzed with the C-terminal or the N-terminal antibody in lysates from EBV-B cells from patient with Y657* mutation were significantly reduced in comparison to HD cells (Fig. 1, C and D). A similar expression level was observed in EBV-B cells with the V637M mutation (Fig. 1, C and D), as shown previously (12). These results are consistent with STAT3 expression exclusively derived from the nonmutant allele, suggesting that the mutant protein is either not at all produced from the mutant allele or alternatively undergoes rapid decay. To prove whether the absence of the mutant protein is reflected by NMD, we isolated cDNA from EBV-B cells from HD, with the Y657* mutation and also EBV-B cells with the R382W and V637M mutations and sequence the full-length transcript. Only wild-type (WT) sequences were present in cDNA samples of EBV-B cells with Y657* mutation, comparable to HD controls (fig. S1A). Of note, cDNA sequencing of known coexpressible forms of mutant and WT alleles (R382W and V637M) clearly demonstrated biallelic expression patterns (fig. S1A). These results suggest that nonexpressibility of mutant Y657* protein is affected by NMD. To confirm that Y657* mutation leads to NMD, we treated EBV-B HD and Y657* cells with the NMD inhibitor cycloheximide (CHX), isolated cDNA, and sequenced the exon 21 of STAT3. These results showed that the cDNA from CHX-treated EBV-B HD cells had normal sequence expression as the nontreated cells (Fig. 1E). On the contrary, CHX-treated EBV-B cells with Y657* mutation had abnormal splicing with the introduction of a premature stop codon, which would lead to NMD (Fig. 1F). These results lead us to evaluate STAT3 mRNA stability, by incubating EBV-B cells with actinomycin D, an inhibitor of de novo transcription, and measure STAT3 mRNA levels by quantitative reverse transcription polymerase chain reaction (qRT-PCR). EBV-B cells with the R382W and V637M mutation showed a significant reduction of STAT3 mRNA stability compared to EBV-B HD cells (fig. S1B). A similar result was also observed in EBV-B cells with Y657* mutation, but the reduced STAT3 mRNA stability on these cells was significantly higher (fig. S1B). We also analyzed STAT3 protein stability by measuring STAT3 protein levels over time after the addition of CHX, which inhibits new protein synthesis. The half-life (T_1/2) of STAT3 protein in EBV-B cells with R382W mutation was similar to EBV-B HD cells, while in EBV-B cells with V637M mutation, it was already reduced at 4 hours after CHX stimulation. In contrast, STAT3 T_1/2 in EBV-B cells containing the Y657* mutation was significantly reduced already 2 hours after CHX stimulation compared to EBV-B HD cells (Fig. 1, G and H). To further analyze the Y657* mutation, we stimulated EBV-B cells with interferon-α (IFN-α) at different time points and evaluated STAT3 phosphorylation (p-STAT3) by flow cytometry. As expected, EBV-B cells from HD had a fourfold increase of p-STAT3 after IFN-α stimulation. A similar result was observed for p-STAT3 levels in EBV-B cells with the R382W mutation (fig. S1, C and D), while the V637M mutation showed a significant reduction of p-STAT3 as compared to HD, an effect that was previously observed (12, 18). EBV-B cells with the Y657* mutation showed 50% less p-STAT3 as compared to cells from HD at all times points analyzed (fig. S1, C and D). Together, these results indicate that in this family, STAT3 was only expressed by the WT allele, the mutant transcript is affected by NMD, and that the STAT3 protein is therefore highly unstable, suggesting that the phenotype of the affected patients originated from a functional STAT3 HI.

Fig. 1. The heterozygous mutation STAT3 Y657* shows reduced STAT3 expression and undergoes rapid NMD.

(A) Pedigree of the AD-HIES family with the mildly affected father (I1; gray square) and severely affected daughter (II-1; black circle). The mother (I-2) and the second daughter (II-2) are unaffected (white circles). Genotypes are indicated. (B) Genomic Sanger sequencing showing the heterozygous c.1971delT variant in exon 21 of STAT3 which generates a stop codon and predicts the premature termination of translation (Y657*). (C) Western blot analysis of total extracts from the EBV-B cells of HDs (HD1, HD2, and HD3) and from patients with the mutations R382W, V637M, and Y657*. Blots were probed with antibodies against the C- or N-terminal part of the STAT3 protein. Glyceraldehyde phosphate dehydrogenase (GAPDH) was used as loading control. (D) A quantification of the STAT3 (C-ter) and STAT3 (N-ter)/loading control ratio after normalization is shown for independently performed experiments as in (C). The mean ± SD is plotted. The average of the three HD is shown. Sanger sequencing of EBV-B HD (E) and Y657* (F) cells untreated or treated for 4 hours with CHX (100 ng/ml). (G) Western blot analysis of STAT3 in total extracts from EBV-B cells of HD and patients after treatment with CHX at the indicated times. GAPDH was used as loading control. (H) A quantification of the STAT3 (N-ter)/loading control ratio after normalization is shown for independently performed experiments as in (G). The mean ± SD is plotted. STAT3 protein T_1/2 (horizontal line) was determined as time after CHX addition until STAT3 levels decreased to 50% of starting level, as indicated by vertical lines. Data are representative of three independent experiments. P values by two-way analysis of variance (ANOVA) followed by Tukey’s multiple comparisons test (D and H).

Table 1. Clinical phenotype of STAT3-deficient patients.

+, present; −, absent; CMC, chronic mucocutaneous candidiasis.

Subjects	Age† (y)	Sex	STAT3 mutation	Infections								Connective tissue signs					Immunological signs		Other findings	References
				Skin	Bacterial	Oral	CMC	Aspergillus	Abscesses	Eczema	Pneumonia	Tooth retention	Facial features	Fractures	Hyperextensibility	Vascular abnormalities	IgE (IU/ml)	Eosinophils (cells/μl)
P1	44	M	Y657*	+	−	+	−	−	+	−	−	−	−	−	−	−	600–700	−	Asthma, Chronic dysentery	This work
P2	6	F	Y657*	+	+	+	−	−	+	+	+	+	+	+	−	+	15850	4712	Allergic Rhinitis, Food allergies	This work
P3	37	M	c.1140-3C > G	−	−	+	−	+	−	−	−	−	−	−	−	−	1003	100–1090	Sino-Orbital Aspergillosis, Allergic Rhinitis	Natarajan et al. (45)
P4	35	M	ΔEx13	+	+	+	−	−	−	+	+	+	−	+	+	−	17028	2400	Asthma	Anolik et al. (19)
P5	10	F	ΔEx14	−	−	−	−	−	+	+	+	−	+	−	−	−	55400	1164	Alopecia totalis	Al Khatib et al. (13)
P6	45	M	F561del	−	−	−	−	−	−	+	−	−	−	−	−	−	2168	2900	Invasive dermatophyte infection	Simpson et al. (21)

†As described in each publication and at the moment of diagnosis for the patients described in this work.

The STAT3 Y657* mutation leads to STAT3 HI

Our results indicate that the truncated protein STAT3 Y657* is not expressed in viable cells and undergoes NMD. To additionally evaluate this mutation, we generated cytomegalovirus (CMV) promoter–driven vectors for ectopic expression of STAT3 variants in a cell culture model. Upon transient transfection of WT and the variants R382W and V637M in human embryonic kidney (HEK) 293T cells, STAT3 was readily detectable by Western blot. In contrast, STAT3 was almost absent in cells transfected with the Y657* variant (fig. S2, A and B). In vivo, the monoallelic STAT3 nonsense mutant Y657* was coexpressed with WT STAT3 from the nonmutant allele. We therefore studied the colocalization of both, the mutant and the WT STAT3, by cotransfecting vectors encoding for N-terminal fusions with two distinct fluorescent proteins [enhanced green fluorescent protein (EGFP) fused to WT, R382W, V637M, and Y657* and mCherry fused to WT STAT3] and analyzed STAT3 expression by confocal microscopy. We observed that in cells cotransfected with either the mutant R382W or V637M, STAT3 WT protein colocalized at equal levels if they were cotransfected with either R382W or V637M mutant; nevertheless when cells were cotransfected with the Y657* mutant, the colocalization with the STAT3 WT protein was significantly reduced (fig. S2, C and D). Moreover, the analysis of the fluorescence intensity of each individually expressed protein showed that the intensity of the Y657* mutant was significantly reduced compared to the WT protein. In addition, this analysis showed that the cotransfection with the Y657* mutant did not alter the expression of the WT protein, in comparison with the fluorescence intensity of the WT protein coexpressed with the other two mutant proteins (fig. S2E). These data suggest that the mutant Y657* does not affect the expression of the WT protein. We then asked whether additional AD-HIES–associated STAT3 mutations induced a similar expression pattern to the STAT3-Y657* mutant. We evaluated various mutant STAT3 variants by forced expression in HEK293T cells, and we focused on mutations inducing larger protein truncations, i.e., deletion mutations or splice-site mutations (fig. S2F), identified in individuals with characteristic features of AD-HIES (Table 1) (12, 13, 19–22). The WT STAT3, the two most frequently occurring DN missense mutations (R382W and V637M), and the Y657* mutation were included for comparison. The predicted changes included a single amino acid deletion F561del (linker domain) (21), an in-frame skipping of exon12 (DNA binding domain, ΔEx12) (12, 22), an out-of-frame skipping of exon 13 (C-terminal half) (ΔEx13) (19), an in-frame skipping of exon 14 (DNA binding domain) (ΔEx14) (13), an out-of-frame skipping of exon 22 (transactivation domain, ΔEx22) (22), an in-frame skipping of exon 22/23 (transactivation domain, ΔEx22-23) (20), a partial deletion of exon 21 (SH2 domain, ΔEx690-P699) (12), and the STAT3 β splice variant (transactivation domain) (fig. S2F). Whereas all truncated proteins were readily detectable, the variant ΔEx13 was weakly expressed, similarly to the Y657* variant. Two variants, the F561del and ΔEx14, showed significantly reduced expression in comparison to STAT3 WT transfected cells (fig. S2, G and H). These data suggest the possibility that more than one AD-HIES–associated STAT3 mutation could have a similar effect to the STAT3 Y657* mutant.

Together, our data suggest that the variant Y657* does not have a DN effect. To evaluate this hypothesis, we performed similar experiments using the STAT3-deficient colon cancer cell line A4 (STAT3^−/− A4) (23). Thereby, we avoided any effects of endogenous STAT3 expression, as present in HEK293T cells. First, we measured STAT3 protein levels by Western blot on STAT3^−/− A4 cells nontransfected or transfected with expression vectors for Myc-tagged WT STAT3 or Flag-tagged mutant STAT3 V637M or STAT3 Y657*. Transfection with either the WT STAT3 or the STAT3 V637M vector led to a clear detection of the STAT3 protein. In contrast, we could not detect any STAT3 expression in cells transfected with STAT3 Y657* vector (Fig. 2, A and B). Second, and despite the fact that EBV-B cells represent a more biological model, we also analyzed in transfected cells the stability of STAT3 protein by stimulating the cells with CHX and measuring STAT3 expression overtime. We observed that cells transfected with STAT3-V637M mutation showed a similar, but still reduced, STAT3 T_1/2 to cells transfected with STAT3 WT plasmid; however, we could not detect any STAT3 expression when cells were transfected with Y657* mutation (Fig. 2, C and D). Furthermore, to evaluate whether the Y657* mutant protein was degraded by the ubiquitin-dependent proteasome, we measured STAT3 protein levels by Western blot on STAT3^−/− A4 cells transfected with WT or mutant cDNA expression vectors, with or without pretreatment with the proteasome inhibitor MG-132. STAT3 protein levels were similar in cells transfected with WT STAT3 before and after MG-132 treatment, whereas STAT3 protein levels were increased in cells transfected with STAT3 V637M in the presence of MG-132. However, we could not detect any STAT3 protein after MG-132 treatment in cells transfected with STAT3 Y657* (Fig. 2, E and F), in line with a model that the Y657* mutant protein goes rapid proteasome-independent decay. Last, to evaluate whether the Y657* mutant protein had a DN effect, we cotransfected STAT3^−/− A4 cells with decreasing amounts (50, 25, 10, and 5 ng) of mutant STAT3 vector and a constant amount (50 ng) of WT STAT3 vector, before and after MG-132 treatment. Cotransfection with the STAT3 V637M vector was associated with dose-dependently reduced WT protein levels, confirming the DN effect of STAT3 V637M (Fig. 2G) (16). When we cotransfected the cells with the mutant STAT3 Y657* vector, we found stable WT protein levels, regardless of the mutant concentration (Fig. 2H). This suggested that STAT3 Y657* does not have a DN effect.

Fig. 2. The STAT3 Y657* mutant is unstable and goes rapid proteasome-independent decay.

(A) Western blot of extracts from STAT3^−/− A4 cells nontransfected (NT), transfected with expression vectors for STAT3 WT (Myc-tagged), the STAT3 mutant V637M, or the mutant Y657* (both Flag tagged). (B) Mean ± SD of STAT3 (N-ter)/loading control ratio after normalization, for independently performed experiments. Total protein extracts from transfected STAT3^−/− A4 cells with either the WT STAT3 (Myc-tagged) plasmid or the STAT3 mutant V637M (Flag-tagged) (C) or the STAT3 mutant Y657* (Flag-tagged) (D) after treatment with CHX (100 μg/ml). Mean ± SD of STAT3 (N-ter)/loading control ratio after normalization, for independently performed experiments. STAT3 protein T_1/2 (horizontal line) was determined as time after CHX addition until STAT3 levels decreased to 50% of starting level, as indicated by vertical lines. Data are representative of three independent experiments. (E) Total protein extracts from NT or transfected STAT3^−/− A4 cells, after treatment (+) with 20 μM MG-132 for 3 hours. (F) Mean ± SD of STAT3/loading control ratio after normalization, for independently performed experiments. Total extracts of STAT3^−/− A4 cells transfected with either the WT STAT3 (Myc-tagged) plasmid or the STAT3 mutant V637M (Flag-tagged) (G) or the STAT3 mutant Y657* (Flag-tagged) (H), cotransfected with decreasing amounts (50, 25, 10, and 5 ng) of mutant STAT3 plasmid and with constant amounts of WT STAT3 plasmid (50 ng). The transfected cells were treated (+) with 20 μM MG-132 for 3 hours. In all experiments, extracts were probed with antibodies specific for the Myc tag, the Flag tag, and the N-terminal part of STAT3. GAPDH was used as loading control. Mean ± SD of STAT3 (N-ter)/loading control ratio after normalization, for independently performed experiments. Data are representative of three independent experiments. P values by two-way ANOVA followed by Tukey’s multiple comparisons test (G and H).

To address this point more directly, first, we performed the same cotransfection experiment as described above, with decreasing amounts (50, 25, 10, and 5 ng) of mutant STAT3 vector and a constant amount (50 ng) of WT STAT3 vector and stimulated the cells with IFN-α to analyze p-STAT3 expression. The results showed that there was an increase in p-STAT3 levels with reducing amounts of STAT3 V637M mutant protein (Fig. 3A), while the cotransfection with Y657* mutant resulted in equal p-STAT3 levels, independently of the amount of mutant protein (Fig. 3B). Second, we evaluated STAT3 transcriptional activity in a reporter assay, by cotransfecting cells with various amounts of mutant cDNA (100, 50, 20, and 10 ng) and a constant amount of WT STAT3 cDNA (100 ng), and later, we stimulated the cells with IFN-α, and the STAT3 transcriptional activity of the proteins produced was measured in a fluorescence-based reporter assay. The STAT3 transcriptional activity of cells transfected with the mutants V637M and Y657* was equally significantly reduced compared to cells transfected with WT STAT3. When cells were cotransfected with reducing amounts of V637M mutant, STAT3 activity increased significantly (Fig. 3C). On the contrary, STAT3 transcriptional function was not affected by the presence of the Y657* mutant (Fig. 3C). Together, these results suggest that there is no residual function from the Y657* mutant that could affect the WT protein and that the mutant Y657* does not have a DN effect, which therefore could lead to HI.

Fig. 3. The heterozygous mutation STAT3 Y657* leads to HI.

Total extracts of STAT3^−/− A4 cells transfected with either the WT STAT3 (Myc-tagged) plasmid or the STAT3 mutant V637M (Flag-tagged) (A) or the STAT3 mutant Y657* (Flag-tagged) (B), cotransfected with decreasing amounts (50, 25, 10, and 5 ng) of mutant STAT3 plasmid and with constant amounts of WT STAT3 plasmid (50 ng) and stimulated (+) with IFN-α (250 ng/ml) for 15 min. In all experiments, extracts were probed with antibodies specific for p-STAT3, the Myc tag, the Flag tag, and the N-terminal part of STAT3. GAPDH was used as loading control. A quantification of p-STAT3 (N-ter)/loading control ratio after normalization, for independently performed experiments as in (A) and (B). The mean ± SD is plotted. Data are representative of three independent experiments. (C) Fluorescence-based reporter assay in cells transiently cotransfected on collagenized 48-well plates with constant amounts of an expression vector encoding nonfused STAT3 WT (100 ng) and decreasing amounts (100, 50, 25, and 10 ng) of a vector construct encoding the nonfused STAT3 mutant V637M or Y657* as indicated. Each transfection sample contained 100 ng of the synthetic STAT3-responsive reporter construct (p3xSIE-CMVmin-tdTomato; red fluorescence) and a transfection control (pEGFP; green fluorescence). STAT3 activity and DNA amounts were equalized using an unrelated plasmid DNA (pBluescript). Nontransfected samples were used as background controls. After 24 hours, the cells were either left untreated or were stimulated with IFN-α (200 ng/ml). After overnight incubation, red fluorescence signals, indicating activity of the STAT3-responsive promoter, were recorded by imaging in life cells at various time points. Mean red fluorescence intensities were quantified with ImageJ and were normalized relative to the unstimulated condition. The mean ± SD is plotted. Data are representative of three independent experiments. P values by two-way ANOVA followed by Tukey’s multiple comparisons test.

Stat3^+/− mice exhibit reduced STAT3 protein expression, increased IgE serum levels, and IgE plasma cell differentiation

To evaluate the role of STAT3 HI in vivo, we created STAT3 heterozygous mice by crossing mice carrying floxed alleles of Stat3 with mice that express Cre recombinase under the control of the mouse SRY-box containing gene 2 promoter (Sox2), in which the Cre recombinase activity is detected in the epiblast cells at embryonic day 6.5 (24). In this way, we obtained Stat3^+/+Sox2^Cre mice referred as Stat3^+/+ (WT) mice and Stat3^+/pSox2^Cre referred as Stat3^+/− mice. The STAT3 protein was detected at significantly lower levels in splenic, thymic, and bone marrow cells from Stat3^+/− mice compared with cells from Stat3^+/+ mice (Fig. 4, A and B), while the levels of STAT1 protein in cells derived from Stat3^+/− mice were similar to Stat3^+/+ cells (fig. S3, A and B), confirming the specificity of STAT3 deletion. Mice were viable, and we did not observe significant differences in body weight between adult Stat3^+/+ and Stat3^+/− mice (fig. S3C). Moreover, there were no differences between the mature CD4⁺ T cells and B220⁺ B cell populations between Stat3^+/+ and Stat3^+/− mice (fig. S3, D and E), as described in previous Stat3 conditional mouse models (25, 26).

Fig. 4. Stat3^+/− mice have reduced STAT3 levels, increased IgE secretion, and in vitro IgE PC differentiation.

(A) STAT3 expression was evaluated by Western blot in Stat3^+/+ and Stat3^+/− cells derived from the spleen, thymus, and bone marrow (BM). Tubulin was used as loading control. Each lane represents one mouse. (B) A quantification of the STAT3/loading control ratio after normalization is shown for independently performed experiments as in (A). The mean ± SD is plotted. (C) IgE serum levels in Stat3^+/+ and Stat3^+/− mice were measured by ELISA. (D) Flow cytometry to identify IgG1⁺ and IgE⁺ cells from IgM⁻IgD⁻ cells 4 days after culturing Stat3^+/+ and Stat3^+/− B220⁺ splenocytes with IL-4 (12.5 ng/ml) and anti-CD40 antibody (62.5 ng/ml) in the absence or presence of IL-21 (25 ng/ml). Numbers in dot plots represent cell frequencies. (E) Mean (±SD) frequencies of the IgG1⁺ and IgE⁺ cells as gated in (D). (F) Levels of IgG1 and IgE in the supernatant of cell cultures as done in (D) were measured by ELISA. B220⁺ splenocytes from Stat3^+/+ and Stat3^+/− mice were stimulated for different time points with IL-21 (50 ng/ml) (G), IL-10 (50 ng/ml) (H), and IL-6 (50 ng/ml) (I), and the phosphorylation of STAT3 (p-STAT3) was evaluated by flow cytometry. Mean (±SD) of mean fluorescent intensity (MFI) is plotted. Data are representative of at least three independent experiments. P values by unpaired t test (C) or two-way ANOVA followed by Bonferroni’s or Tukey’s multiple comparisons test (B and E to I).

One of the main characteristics of patients with DN mutations in STAT3 is increased serum IgE levels, which highlights the role of STAT3 in mature B cells. Therefore, we evaluated Ig levels in the serum of Stat3^+/+ and Stat3^+/− mice. The analysis of IgM, IgG1, IgA, IgG2b, IgG2c, and IgG3 did not show any differences between Stat3^+/+ and Stat3^+/− mice (fig. S3F). Nevertheless, the levels of IgE were significantly increased in Stat3^+/− mice (Fig. 4C), which was consistent with a previous STAT3 AD-HIES mouse model (27). To further investigate the role of STAT3 HI on plasma cell (PC) development and Ig secretion, we cultured splenic B220⁺ cells from Stat3^+/+ and Stat3^+/− mice with anti-CD40 and IL-4 in the absence or presence of IL-21, previously shown to negatively regulate class switch recombination of IgE⁺ PCs (28). After 4 days of culture, we evaluated the presence of IgG1⁺ and IgE⁺ PCs by flow cytometry. We found that in the absence of IL-21, ~12% of Stat3^+/+ cells were IgG1⁺ cells, and 70% were IgE⁺ cells, with similar levels observed in Stat3^+/− cells (Fig. 4, D and E). On the contrary, the addition of IL-21 induced an expansion in the levels of IgG1⁺ cells, for both Stat3^+/+ and Stat3^+/− cells. However, the IgE⁺ cells were significantly increased if Stat3 was expressed on one allele only (Fig. 4, D and E). Moreover, when we analyzed IgG1 and IgE in the supernatant of these cultures, we detected that IgG1 levels were similar between Stat3^+/+ and Stat3^+/− cells, with an increase after the addition of IL-21 (Fig. 4F). In contrast, and consistent with the results observed by flow cytometry, the presence of IL-21 induced a significant increase in the secretion of IgE from Stat3^+/− cells, in comparison with their WT counterparts (Fig. 4F). These results confirmed previous data on IL-21 negatively regulating the differentiation to IgE⁺ cells through STAT3 signaling (28). We later stimulated splenic Stat3^+/− and Stat3^+/+ B220⁺ cells with IL-21 for different time points and measured p-STAT3 by flow cytometry. We observed that WT cells had a fourfold induction of p-STAT3 after 15 min of IL-21 stimulation (Fig. 4G). In contrast, Stat3^+/− cells had less than twofold of p-STAT3 expression upon 15 min of IL-21 stimulation, with a moderate increase over time, which remained significantly lower than the p-STAT3 expression in WT control cells (Fig. 4G). A similar result was observed when we stimulated B220⁺ cells with IL-10 (Fig. 4H) and IL-6 (Fig. 4I), with significant reduced p-STAT3 in Stat3^+/− cells after 30 and 45 min of cytokine stimulation (Fig. 4, H and I). Furthermore, STAT3 levels were reduced in unstimulated and stimulated Stat3^+/− cells, in comparison to WT cells in any of the stimulated conditions (fig. S3, G to I). As expected, when we normalized the p-STAT3 levels to their corresponding STAT3 levels, the p-STAT3 did not have significant differences between STAT3 WT and heterozygous cells, with any of the cytokines analyzed (fig. S3J). Together, these data show that STAT3 HI affects in vitro IgE secretion, IgE⁺ PC differentiation, and cytokine-induced p-STAT3.

STAT3 HI impairs in vitro T_H17 cell differentiation

Next, we investigated the role of STAT3 HI in T cell differentiation, considering that patients with DN mutations in STAT3 have reduced T_H17 cell numbers (8–10, 12). We isolated splenic CD4⁺ naïve T cells from Stat3^+/+ and Stat3^+/− mice and cultured them for 4 days with IL-12, IL-4, and transforming growth factor–β (TGF-β) to induce the differentiation of T_H1, T_H2, and inducible T regulatory (iT_reg) cells, respectively. The analysis by flow cytometry showed that the differentiation of T_H1 cells, as measured by the IFN-γ⁺ cells, was not affected by the Stat3 HI (fig. S4, A and B). At the same time, the frequency of IL-13⁺ cells, indicative of T_H2 cells, and the percentage of Foxp3⁺ iT_reg cells were comparable between Stat3^+/+ and Stat3^+/− cells (fig. S4, A and B). When CD4⁺ naïve WT cells were cultured with the cytokines IL-6 plus TGF-β or IL-6 plus TGF-β and IL-23, to induce T_H17 differentiation, we detected that ~35% of cells became T_H17 in any of the cytokine polarizing conditions (Fig. 5, A and B). However, CD4⁺ naïve Stat3^+/− cells only induced up to 15% of T_H17 cells when they were cultured either with IL-6 plus TGF-β or IL-6 plus TGF-β and IL-23 (Fig. 5, A and B). These data confirmed the relevance of STAT3 for T_H17 differentiation. Furthermore, since IL-6 is one of the cytokines inducing STAT3 activation in CD4⁺ T cells (29–31) and considering its role in T_H17 differentiation (32–34), we questioned whether Stat3 HI also affected p-STAT3 in IL-6–stimulated CD4⁺ cells. Therefore, we stimulated Stat3^+/+ and Stat3^+/− cells with IL-6 and after 15, 30, and 45 min, we measured p-STAT3 by flow cytometry. We found that Stat3^+/− cells had significantly reduced p-STAT3 compared with Stat3^+/+ cells, at all different time points analyzed (Fig. 5C). By contrast, the levels of p-STAT1 were similar between IL-6–stimulated Stat3^+/+ and Stat3^+/− cells (Fig. 5D), consistent with the specificity of STAT3 heterozygosity. Moreover, the analysis of STAT3 expression in IL-6–stimulated Stat3^+/− cells showed a significant reduction of total STAT3 in comparison with IL-6–stimulated Stat3 WT cells (fig. S4C) with no differences in the levels of STAT1 expression (fig. S4D). These results were confirmed by Western blot, where IL-6–stimulated CD4⁺ Stat3^+/− cells showed a significant reduction of p-STAT3 and STAT3 expression when compared to their WT counterparts, while the levels of p-STAT1 and STAT1 were similar to stimulated WT cells (fig. S4, E and F). Together, these data indicate that STAT3 HI impairs in vitro T_H17 differentiation and p-STAT3 expression in CD4⁺ T cells and confirms the relevance of STAT3 in the generation of T_H17 cells.

Fig. 5. Stat3^+/− CD4⁺ cells have impaired capacity to differentiate to T_H17 cells.

(A) Naive CD4⁺ cells from Stat3^+/+ and Stat3^+/− mice were cultured in vitro with T_H17 polarizing conditions and analyzed by flow cytometry on day 4 for the indicated cytokine expression. Representative dot plots show intracellular IFN-γ and IL-17A expression after restimulation of cells with PMA and ionomycin. Numbers in dot plots represent cell frequencies. (B) Mean (±SD) frequencies of the cells gated in (A) are shown. CD4⁺ splenocytes from Stat3^+/+ and Stat3^+/− mice were stimulated for different time points with IL-6 (50 ng/ml) and the phosphorylation of STAT3 (p-STAT3) (C) and STAT1 (p-STAT1) (D) were evaluated by flow cytometry. Mean (±SD) of mean fluorescent intensity (MFI) is plotted. Data are representative of five independent experiments (A and B) or at least three independent experiments (C and D). P values by two-way ANOVA followed by Bonferroni’s test.

STAT3 HI impairs the IL-17–secreting cells in the immune response against cutaneous S. aureus infection

Patients with DN mutations in STAT3 suffer from recurrent bacterial and fungal infections including mucocutaneous candidiasis and pulmonary aspergillosis (3, 17). Therefore, we sought to investigate how Stat3^+/− mice respond to different types of infections. First, we infected Stat3^+/+ and Stat3^+/− mice with Candida albicans intraperitoneally and evaluated the course of infection during 3 days (weight loss). We found that Stat3^+/− mice showed similar weight loss than Stat3^+/+ mice during infection (fig. S5A), indicating that gene dosage of Stat3^+/− did not affect a systemic C. albicans infection. When we evaluated C. albicans growth at day 3 postinfection (p.i.), we did not detect differences in the weight of kidney and spleen or in the load of C. albicans in these organs (fig. S5, B and C). These data indicate that Stat3^+/− mice can respond to a systemic C. albicans infection as efficient as their WT counterparts.

Next, we evaluated the susceptibility of Stat3^+/− mice to a strictly intradermal S. aureus infection, considering that most of the bacterial infections present in patients with DN mutations in STAT3 are skin infections caused by this pathogen (3). S. aureus infection intradermally induces a local inflammation followed by spontaneous healing of the skin at days 7 to 9 p.i., with a higher infiltration of polymorphonuclear leukocytes (PMLs) at the beginning of the infection (35). Therefore, we decided to evaluate Stat3^+/− mice 5 days after S. aureus infection. We observed that Stat3^+/− mice failed to clear the bacteria as effectively as WT mice did (Fig. 6A). Moreover, the frequency of PML (defined by Ly6G⁺ CD64⁻ cells) was significantly reduced in Stat3^+/− compared to Stat3^+/+ mice (Fig. 6, B and C). Dermal macrophages (Macs) have been found to be essential for the rapid resolution of S. aureus skin infection (35, 36), and dermal Macs differentiating from recently immigrated Ly6C^hi monocytes can be distinguished from bona fide resident CD206⁺ Macs. Therefore, we analyzed the different Macs subpopulations in Stat3^+/+ and Stat3^+/− mice 5 days p.i., and we found that the total numbers of CD64^hi Macs were similar between infected Stat3^+/+ and Stat3^+/− mice (fig. S5, D and E). Likewise, the analysis of Ly6C^hi and CD206⁺ Macs showed comparable population sizes in Stat3^+/+ and Stat3^+/− mice (fig. S5, F and G).

Fig. 6. Stat3^+/− mice are susceptible to intradermal S. aureus infection.

Stat3^+/+ and Stat3^+/− mice were intradermally inoculated with 10⁷ CFU S. aureus in both ears and analyzed after 5 days. One ear was used to quantify CFU per lesion (dilution series plated from homogenized skin tissue) (A), and the second ear was used to determine skin PML (Ly6G^hi skin cells gated on CD45^highCD11b^high cells) by flow cytometry (B and C). (B) Representative dot plots (B) and mean (±SD) frequencies (C) of the PML cells gated in (B) are shown. Draining LNs from Stat3^+/+ and Stat3^+/− S. aureus–infected mice were harvested 5 days p.i., and cells were cultured for 5 and 24 hours with PMA and ionomycin, and the levels of TNF-α, IL-6 (D), IL-17A, and IFN-γ (E) were measured in the supernatant of the cell cultures by LEGENDplex protein assays or ELISA, and intracellular IL-17A and IFN-γ were measured by flow cytometry (F to I). Representative dot plots show intracellular IFN-γ and IL-17A of CD3⁺γδ⁺ cells (F and G) and CD3⁺CD4⁺ cells (H and I). Numbers in dot plots represent cell frequencies. Mean (±SD) frequencies (G and I) of the cells gated in (F) and (H) are shown. Data were pooled from three independent experiments. Mann-Whitney test (A, C, G, and I) and two-way ANOVA followed by Bonferroni’s test (D and E) were used for statistical analysis.

To better understand the basis of the susceptibility of Stat3^+/− mice to S. aureus infection, we analyzed infiltrating dermal lymphocytes. The cytokine IL-17 has an important role in immune responses at epithelial sites and particularly in S. aureus skin infections by inducing neutrophil-attracting chemokines and granulopoiesis-inducing factors (37). Since IL-17 cytokines can be produced by CD4⁺ T_H17 cells and CD3⁺γδ⁺ T cells, which are also relevant in S. aureus skin infection (38, 39), we evaluated the frequency of dermal CD4⁺- and CD3⁺γδ⁺-producing IL-17A and IFN-γ, after they were stimulated ex vivo with phorbol 12-myristate 13-acetate (PMA) and ionomycin. Flow cytometry analysis showed a trend to a lower frequency of CD3⁺γδ⁺ T cells secreting IL-17A derived from Stat3^+/− infected mice in comparison with Stat3 WT cells; however, no statistical significances were observed, and a similar result was obtained for the IFN-γ–producing cells (fig. S5, H and I). However, we observed a significant reduction in the CD4⁺ IL-17A–secreting cells in Stat3^+/− cells compared to Stat3^+/+ cells, whereas the frequency of CD4⁺ IFN-γ⁺ cells was similar for both genotypes (fig. S5, J and K).

It has previously been demonstrated that the IL-17–producing cells that mediate the host defense against the S. aureus skin infection are recruited from draining lymph nodes (LNs) to the infected skin (40). Therefore, we analyzed the IL-17–producing cells derived from the draining LNs of Stat3^+/+- and Stat3^+/−-infected mice. First, we cultured LN-derived cells with PMA and ionomycin for 5 and 24 hours and analyzed the levels of secreted cytokines in the cell culture supernatants. We observed that the proinflammatory cytokines TNF-α and IL-6 were quantitatively similar between Stat3^+/+ and Stat3^+/− cells after 5 or 24 hours of stimulation (Fig. 6D). By contrast, Stat3^+/− cells produced significantly less IL-17A and significantly more IFN-γ than Stat3^+/+ cells at both time points (Fig. 6E). To investigate in depth the source of reduced IL-17A and increased IFN-γ in the supernatant of Stat3^+/− LN-derived cells, we evaluated IL-17A– and IFN-γ–producing cells by flow cytometry and found that ~35% of Stat3^+/+ CD3⁺γδ⁺ T cells produced IL-17A, while only 15% of Stat3^+/− CD3⁺γδ⁺ cells were positive for IL-17A⁺. In contrast, there was a significant expansion in the frequency of Stat3^+/− CD3⁺γδ⁺ cells secreting IFN-γ, in comparison with their WT counterparts (Fig. 6, F and G). In addition, we found that the frequency of CD4⁺IL-17A⁺ Stat3^+/− cells was reduced by ~50% as compared to WT cells, but the percentage of Stat3^+/− CD4⁺IFN-γ⁺ cells was significantly increased when compared with Stat3 WT CD4⁺IFN-γ⁺ cells (Fig. 6, H and I). Collectively, these data show that STAT3 HI impairs the immune response against dermal S. aureus infection by altering the balance between the IL-17A– and IFN-γ–producing cells. These experimental data support the model that STAT3 HI can generate symptoms classically associated with AD-HIES.

DISCUSSION

The DN effect of STAT3 mutations in AD-HIES has been primarily based on in vitro experiments in which the STAT3 variants generate aberrant proteins or have low STAT3 transcriptional activity (16). Nevertheless, the alternative option that mutations in STAT3 have a haploinsufficient effect leading to an AD-HIES phenotype has never been proven. In this study, we show that a STAT3 heterozygous mutation was sufficient to cause AD-HIES, which is supported by both in vitro functional studies and through a Stat3 heterozygous mouse model. We demonstrate that STAT3 HI resulting from a heterozygous nonsense mutation (Y657*) that goes through NMD leads to unstable STAT3 mRNA decay of the mutant protein. The incomplete phenotypic penetrance of the mutation is clearly in line with other immunological diseases associated with HI in a variety of human traits (41).

In contrast to the well-characterized STAT3 DN variant R382W, which leaves the STAT3 protein levels unaffected (12, 42, 43), the levels of STAT3 in EBV-B cells from patient with Y657* mutation were reduced to ~50% compared to EBV-B cells from HD. The same levels were observed in EBV-B cells with the V637M variant, and a similar effect was observed for the p-STAT3 after IFN-α stimulation. The fact that these two variants are located in the SH2 domain of STAT3, which is responsible for its phosphorylation, can explain this last observation. Moreover, the significantly reduced stability of the protein observed in EBV-B cells with these two variants could be the cause of their reduced STAT3 levels. Our results are consistent with a previous study (42) in which they classified different STAT3 mutations using five computer-based programs to predict their impact on protein stability. The STAT3-V637M mutation was classified as a “structural-functional” mutation, predicted to be both destabilizing and to directly impair a domain specific STAT3 function, while the STAT3-Y657* mutation was predicted to be a “structural” mutation which destabilizes and impairs the folding of STAT3, thereby indirectly affecting STAT3 functions such as DNA binding, phosphorylation, or dimerization.

Through our cotransfection experiments, we demonstrated that the V637M variant has a DN effect over the WT protein, in terms of p-STAT3 and STAT3 transcriptional activity, while the variant Y657* left the WT protein completely unaffected, which allow us to assume that this variant is haploinsufficient. These findings show that two variants having similar protein expression levels can have a different effect of dominance. This can be explained by the stability of monomers and homodimers, influenced by each specific variant. It has been shown that the V637M and Y657* mutations showed reduced ability to form homodimers, reduced global conformational stability of both their monomers and homodimers, and reduced DNA binding activity. The V637M mutation had less capacity to form homodimers in comparison to the Y657* mutant but still showed DNA binding activity, while this was not the case for the Y657* mutant, which shows complete loss of DNA binding activity (18). These data suggest that the formation, stability, and DNA binding activity of the homodimers are affected specifically by each mutation, influencing the dominance of each variant and therefore the differences observed in the clinical manifestations of the patients. Our data differ from a previous study, in which 143 STAT3 variants that have been identified as causing AD-HIES were analyzed for their effect of dominance through their STAT3 transcriptional activity. In this work, the variant Y657* was classified as hypomorphic and have a DN effect (16). One possible explanation for the discrepant interpretations may relate to the fact that dominant negativity was only measured by analyzing the STAT3 transcriptional activity induced by the single variant, without evaluating the effect of the mutant over the WT protein, as done by us and others (1, 43).

To experimentally address STAT3 HI in vivo, we created a mouse model in which the loss of one allele of Stat3 generated Stat3 heterozygous mice. Stat3 heterozygous mice were viable and fertile, in contrast to homozygous deletion of Stat3 which is embryonically lethal in mice (15). Stat3^+/− mice displayed main characteristics also present in patients with AD-HIES, such as reduced STAT3 expression, increased IgE levels in steady state, reduced in vitro T_H17 cell differentiation, and high susceptibility to skin S. aureus infection. Only one AD-HIES mouse model has been described, in which mice carried the STAT3 DN mutation V463 (27). In this model, mice also show increased IgE levels; nevertheless and contrary to our observed model, in vitro T_H17 differentiation of naïve CD4⁺ T cells was reduced to almost not detectable levels in mutant mice (9.8-fold reduction), with only twofold reduction observed in the Stat3^+/− mice analyzed here. These two models clearly show that HI and DN can have different biological effects, which most likely depend on the STAT3 dosage, but both can show characteristics present in patients with AD-HIES.

Stat3^+/− mice also showed increased IgE secretion and IgE⁺ B cell differentiation when B cells were cultured in the presence of the cytokine IL-21. The role of IL-21 in the differentiation of IgE PCs has been proposed as an explanation for the increased IgE levels observed in patients with AD-HIES (28). In this model, while the cytokine IL-4 promotes the differentiation of both IgE and IgG1 PCs, IL-21 through the IL-21 receptor and JAK/STAT3 signaling induce a block in the IgE PCs differentiation, favoring the development toward IgG1 PCs. In patients with STAT3 mutations and in cells from Stat3^+/− mice, where the IL-21–JAK/STAT3 signaling is disrupted and cannot longer induce IgG1 differentiation, the IgE PCs are developed, and consequently, the levels of IgE are also increased.

In contrast to the normal response of Stat3^+/− mice to the systemic C. albicans infection and consistent with the skin infections observed in the patient analyzed in this study, the immune response to the intradermal S. aureus infection in Stat3^+/− mice was significantly affected. At the site of infection, Stat3^+/− mice showed reduced IL-17A–producing CD4⁺ T cells and increased frequency of PML, while CD4⁺ and CD3⁺γδ⁺ IL-17A–producing cells derived from the draining LNs were significantly reduced in comparison to Stat3^+/+ mice. Although we cannot exclude the possibility that other factors are involved in the increased susceptibility to S. aureus infection in Stat3^+/− mice, it is plausible that the reduced PML frequency is associated to the phenotype observed in the infected mice, as well as the reduced IL-17A–producing cells present in their LNs. Further investigations will be needed to understand the mechanisms underlying these observations.

Despite the fact that investigating the phenotype of a mouse model containing the specific Y657* mutation would represent a more accurate model to the patients’ clinical characteristics, with the Stat3 heterozygous model, we show the relevance of the effect of deleting one STAT3 allele in the context of STAT3 immunodeficiencies associated to AD-HIES.

Since STAT3 activity is predicted to be reduced to only 50% (compared to the residual 25% observed in DN forms), HI is expected to cause a milder or different AD-HIES disease phenotype. Patients with AD-HIES with no detectable STAT3 mutation but with similar histories of elevated IgE levels have been shown to frequently suffer from nontypical AD-HIES, like food allergies and anaphylaxis, unlike patients with DN STAT3 mutations (44). Likewise, the patient carrying the de novo STAT3 splice acceptor site mutation in intron 12 (c.1140-3C > G) presented with elevated IgE levels, allergic rhinitis, eosinophilic esophagitis, and a fatal invasive aspergillosis; however, the patient did not exhibit the full extent of AD-HIES manifestations (45). Similarly, the patient with the heterozygous STAT3 mutation p.Phe561del (F561del), which has been associated with AD-HIES due to the presence of elevated IgE levels and eosinophils, presented with other characteristics that are not typical from AD-HIES, such as invasive dermatophyte infection (21). These data are consistent with a recent study in which disseminated coccidioidomycosis was present in two patients that have proposed haploinsufficient STAT3 mutations (46), although these patients did not exhibit AD-HIES characteristics. The STAT3 mutant p.Ser381Pfs*2 (ΔEx13) did not allow for sustained STAT3 protein expression in our in vitro analysis, as we observed for the STAT3 mutation Y657*. This patient had a similar phenotype to the one described in our study, with chronically elevated IgE levels, high eosinophils count, severe eczematous dermatitis, and recurrent staphylococcal skin and oral infections (19). On the other hand, the p.Thr412_Asp427del (ΔEx14) allele, in which the patient also presented AD-HIES manifestations (13) (Table 1), allowed for STAT3 expression, although at reduced levels compared to STAT3 WT. This indicates that variable amounts of STAT3 can generate variable clinical manifestations seen in AD-HIES. Further investigations are necessary to be able to correlate their STAT3 expression levels, with their effect of dominance and their phenotypic characteristics.

The full penetrance of STAT3 DN mutations has been postulated, although the clinical manifestations are also quite variable (47, 48), raising the question whether a genotype-phenotype correlation exists. A previous study in which they classified the mutations according to presence of the variants in the DNA binding domain or in the SH2 domain of STAT3 showed a trend for symptoms associated to the first group, such as increase scoliosis and high palate, and other symptoms, such as increased frequency of respiratory infections, to the second group. Nevertheless, the differences were not statistically significant, and therefore, this kind of analysis needs to be performed with an increased cohort and broader variants (47). Incomplete penetrance is frequently described in cases of dominance by HI (41), as it is the case for the patients with Y657* mutation, which shows no DN effect and the patients present quite variable clinical manifestations. Similarly, incomplete penetrance has been also considered for STAT3 gain-of-function mutation carriers. Patients with these mutations present with autoimmunity and lymphoproliferation between other disorders, such as diabetes and growth failure (49). A recent study found a correlation between the basal transcriptional activity of the mutations and their clinical manifestations. Nevertheless, the authors of this study also postulate that unknown factors such as additional mutations, epigenetic changes, or environmental factors (especially infections) could probably also influence penetrance (50). In a similar way, these factors could be involved in the variability of symptoms observed in patients with STAT3 AD-HIES. It has been proposed that, in terms of population genetics, STAT3 is under strong negative selection (51), which is consistent with the accepted AD inheritance of STAT3-HIES and with the low frequency of missense variants in the general population analyzed by the gnomAD (Genome Aggregation Database) (16). However, these prediction models have a high false-positive rate, and therefore, it is suggested that any new variant found in patients with HIES should be tested functionally for an accurate molecular diagnosis (16).

STAT3 is one of the seven STAT molecules belonging to the JAK/STAT signaling pathway. Loss- and gain-of-function mutations in STAT1, STAT2, STAT5b, and STAT6 have been reported, generating different immune dysregulations (52). So far, no studies have shown HI being the effect of dominance of these mutations; nonetheless, some reports have proposed HI in gen variants associated to the regulation of the JAK/STAT signaling pathway like Socs1 and Ptpn2 (53, 54), although experimental evidence was lacking to confirm this effect. To exclude HI as an effect of dominance in mutations in genes involved in the JAK/STAT signaling pathway, more experimental studies where the variant effect is measured over the WT protein still need to be performed.

We propose here that STAT3 variants can have a HI effect, in which the mechanism is still not known. A plausible explanation is that these mutations give rise to unstable proteins that are unable to form a stable dimer, and therefore, they are degraded or cleaved, leaving behind the stable WT protein. This interpretation gives an important role to the level of total protein that sets the threshold for available protein, which, together with the function of the tyrosine phosphorylation and DNA binding, defines the final STAT3 activity.

In summary, this study shows that STAT3 HI causes a phenotype reminiscent to the one observed in patients with DN STAT3 mutations. Furthermore, this study not only improves the understanding of STAT3 biology, but it is also relevant for the development of potential gene therapy approaches. Current therapies are typically prophylactic to prevent bacterial and fungal infections but do not provide a causal solution. Therefore, gene therapy approaches, which aim to edit genetically the patient’s cells in a highly specific way ex vivo to subsequently use them in an autologous transplantation, are under development (55, 56). The data presented here are relevant for these new therapeutic approaches, since it demonstrates that silencing a single allele of STAT3 is not sufficient to establish a normal STAT3 function and therefore improve the quality of life of patients carrying STAT3 mutations. Our findings uncover STAT3 HI as a possible cause for immune dysregulation and provide evidence that the development of gene editing procedures aiming at the inactivation of mutant STAT3 alleles should evaluate potentially associated risks.

METHODS

Sex as a biological variable

We included female and male animals in this study, and no sex bias has been observed; therefore, it was not considered as a biological variable.

PCR, sequencing, and qRT-PCR

Genomic DNA and total RNA were isolated from EBV-B patient–derived cells and from peripheral blood mononuclear cells using the QIAamp DNA Mini Kit (QIAGEN) and the Monarch Total RNA MiniPrep Kit (New England Biolabs), respectively, following the manufacturers’ protocols. RNA was reverse-transcribed into cDNA using Moloney murine leukemia virus (M-MLV) reverse transcriptase (Promega), according to the supplier’s instructions. Coding exons and cDNA sequences of STAT3 were amplified by PCR and subsequently purified using enzymatic cleanup with Multicore buffer (Promega), exonuclease I (Thermo Fisher Scientific), and FastAP thermosensitive alkaline phosphatase (AP; Thermo Fisher Scientific). Purified PCR products were prepared for Sanger sequencing using standard protocols and submitted to Eurofins Genomics. qRT-PCR was performed in triplicates using SYBR Green reagent (Thermo Fisher Scientific) on a StepOne real-time PCR system (Applied Biosystems). STAT3 mRNA relative expression to actin was calculated with the 2^-ΔΔCt method. In some experiments, EBV-B cells were treated with CHX (100 ng/ml) (Sigma-Aldrich) for 4 hours or with actinomycin D (10 μg/ml) (Sigma-Aldrich) for 4, 16, and 24 hours. Cells were harvested to isolate RNA and prepare cDNA as described.

Cell culture

EBV-B cells from the patients were cultured in RPMI 1640 medium supplemented with 10% heat-inactivated fetal bovine serum (FBS) and 1% penicillin-streptomycin (Gibco). HEK293T cells, derived from the human embryonic kidney 293 cell line, and STAT3^−/− A4 cells, derived from human DLD1 colon cancer cells by homologous recombination (23) and provided by S. Ehl, were cultured in Dulbecco’s modified Eagle’s medium supplemented with 10% FBS and 1% penicillin-streptomycin (Gibco). Cells were incubated at 37°C under an atmosphere containing 5% CO₂.

cDNA expression vectors, fluorescent protein fusion constructs, and transfection experiments

The cDNAs encoding the WT or mutant STAT3 (R382W, V637M, and Y657*) were cloned by RT-PCR from EBV-B cells derived from a HD or heterozygous mutation carriers, respectively. Restriction sites (Xho I and Hind III) were introduced with the PCR primers, and sequences were verified after cloning. The pEGFP-C1 vector (Clontech, Takara) was used to generate EGFP-fusion constructs. The mCherry vectors were generated by replacing EGFP with mCherry sequences. Nonfusion versions were obtained by subcloning into pEGFP-N1. Additional mutant cDNAs (F651del, ΔEx12, ΔEx13, ΔEx14, ΔEx22, ΔEx22-23, and ΔE690-P699) were generated by site-directed mutagenesis using overlap extension PCR with Q5 polymerase (New England Biolabs) according to the standard methods. Alternatively, pCR3_Flag vector was used to generate the mutants V637M and Y657* and Myc-tagged WT STAT3 plasmids. HEK293T cells were transfected with jetPEI (Polyplus) or XtremeGeneHP (Roche) and STAT3^−/− A4 cells with Lipofectamine 2000 kit (Thermo Fisher Scientific) transfection reagents, according to the manufacturer’s instructions. In some experiments, STAT3^−/− A4 cells were treated with the proteasome inhibitor MG-132 (20 μm) (Sigma-Aldrich) for the last 3 hours or with IFN-α (250 ng/ml) (Abcam) for the last 15 min of the transfection.

Fluorescence-based reporter assay

To assess the transcription activating function of WT and mutant STAT3, we used synthetic promoter reporter vector constructs with the red fluorescent protein tdTomato (Clontech, Takara) under the control of a STAT3-responsive promoter, either composed of three copies of the sis-inducing element (SIE) sequence 5′-GTCGACATTTCCCGTAAATCGTCGA-3′ or nine copies of a STAT3-binding element 5′-TGCTTCCCGAACGT-3′ fused to the CMV minimal promoter. Cells grown on collagenized 48-well plates were transiently cotransfected with the indicated STAT3 constructs together with the reporter (100 ng per well) and a transfection control (pEGFP). Twenty-four hours after transfection, cells were stimulated with IFN-α (200 ng/ml) or left untreated, and fluorescence intensities were visualized and recorded in live cells using a FluoroSpot Analyzer (Cellular Technology Limited ImmunoSpot) at various time points. Mean fluorescence intensities were quantified with ImageJ software.

CHX treatment

EBV-B cells and transfected cells were treated with 100 μM and CHX (100 μg/ml; Sigma-Aldrich), respectively, and harvested at specific time points (0, 2, 4, 16, and 24 hours). Cell pellets were lysed for Western blot analysis as described below.

Mice

Stat3^+/− and Sox2^Cre mice were on a C57BL/6J genetic background and purchased from the Jackson Laboratory (USA). Mice were bred under specific pathogen–free conditions in the animal facilities of the University of Freiburg and the University of Munich and housed in groups of up to five mice. Mice were kept in 12-hour light/dark cycles, and food and water were provided ad libitum. Experiments were performed with males and females of 6 to 12 weeks old. All animal experiments were approved by the Federal Ministry for Nature, Environment and Consumer’s protection of the state of Baden-Wuerttemberg (Germany) and by the government of Upper Bavaria (Regierung von Oberbayern, Munich, Germany).

Flow cytometry of mouse T and B cell populations

Single-cell suspensions from spleens and bone marrows were resuspended in phosphate-buffered saline (PBS) containing 2% FBS [fluorescence-activated cell sorting (FACS) buffer] and stained for flow cytometry analysis. Data were acquired with a LSR Fortessa (BD Biosciences) and analyzed using FlowJo software version V10.4. Nonviable cells were excluded by labeling with fixable viability dye eFluor780 (eBioscience). All samples were gated on FSC-A versus SSC-A, followed by FSC-W versus FSC-H and then SSC-W versus SSC-H gates to exclude doublets. Antibodies against the following molecules were used: CD19 (RM4-5), B220 (RA3-6B2), CD4 (6D5), CD25 (PC61), CD62L (MEL-14), and CD44 (IM7) were from BioLegend, and IgM (II/41) and IgD (11-26C) were from eBioscience.

ELISA of serum Igs and cell culture supernatants

Blood samples were taken from 6- to 12-week-old mice, and serum was collected using blood collection tubes (BD Biosciences). Cell culture supernatants were collected at the indicated time points. Specific Igs were detected by enzyme-linked immunosorbent assay (ELISA) using the capture antibodies IgM, IgG2b, IgG2c, IgG3, and IgA and AP-conjugated antibodies (5 μg/ml) for detection (all from Southern Biotech). Mouse sera were serially diluted in duplicates. ELISA plates were developed with AP-streptavidin (Sigma-Aldrich) and phosphorylated nitrophenyl phosphate (Sigma-Aldrich). Absorbance at 405 nm was determined with a Multiskan FC plate reader (Thermo Fisher Scientific). ELISA for IgG1 and IgE was performed using the Mouse IgG1 and IgE Uncoated ELISA Kit, respectively (eBioscience), according to the manufacturer’s guidelines.

In vitro PC differentiation

For IgG1 and IgE PC differentiation, splenic B220⁺ cells were purified using anti-B220 magnetic beads (Miltenyi Biotec). Cells were cultured in complete B cell medium [RPMI with 10% FBS, 10 mM Hepes, GlutaMAX, penicillin-streptomycin (Gibco), and 50 μM β-mercaptoethanol (Sigma-Aldrich), supplemented with IL-4 (12.5 ng/ml; PeproTech) and anti-CD40 antibody (62.5 ng/ml; Miltenyi Biotec), in the absence or presence of IL-21 (25 ng/ml; PeproTech)]. Cells were cultured for 4 days, and cell culture supernatants were collected to measure IgG1 and IgE by ELISA. IgG1⁺ and IgE⁺ cell differentiation was measured by flow cytometry.

In vitro T cell polarization

Splenic CD4⁺ cells were purified using anti-CD4 magnetic beads (Miltenyi Biotec). Naïve CD4⁺ T cells were sorted to at least 98% purity using the following markers: CD4⁺CD25⁻CD62L⁺CD44⁻. Cells were sorted with a BD FACSAria III (BD Biosciences) and seeded at a density of 100,000 cells per well in a 96-well U-bottom plate. Cells were activated with anti-CD3 (BM10-37) and anti-CD28 (37.51) antibodies (both from BD Biosciences). For T_H1 polarization, cells were cultured on anti-CD3 (2 μg/ml)– and anti-CD28 (2 μg/ml)–coated plates with anti–IL-4 antibody and the recombinant cytokine IL-12 (20 ng/ml). For T_H2 polarization, cells were cultured on anti-CD3 (1 μg/ml)– and anti-CD28 (1 μg/ml)–coated plates with anti–IFN-γ antibody and the recombinant cytokine IL-4 (10 ng/ml). For iT_reg polarization, cells were cultured on anti-CD3 (1 μg/ml)– and anti-CD28 (1 μg/ml)–coated plates with anti–IFN-γ and anti–IL-4 antibodies and the recombinant cytokine TGF-β (15 ng/ml). For T_H17 polarization, cells were cultured on anti-CD3 (5 μg/ml)–coated plates and soluble anti-CD28 (1 μg/ml) with anti–IFN-γ and anti–IL-4 antibodies and the recombinant cytokines IL-6 (10 ng/ml), IL-23 (50 ng/ml), and TGF-β (5 ng/ml). Cells were cultured in complete medium [RPMI supplemented with 10% fetal calf serum (FCS), penicillin/streptomycin (Invitrogen), and 50 μM β-mercaptoethanol (Sigma-Aldrich)] for 4 days. All recombinant cytokines were from PeproTech, and the neutralizing antibodies anti–IL-4 (11B11) and anti–IFN-γ (XMG1.2) were used at 5 μg/ml and were both from BioLegend.

Cytokine and transcription factor analysis

For intracellular cytokine staining, CD4⁺-polarized cells were restimulated for 4 hours with PMA (50 ng/ml), ionomycine (500 ng/ml) (both from Sigma-Aldrich), and monensin (1×) (BioLegend). The cells were fixed and permeabilized with cytofix/cytoperm (BD Biosciences) according to the manufacturer’s instructions. For detection of Foxp3⁺ cells, cells were fixed and permeabilized using a Foxp3 staining kit (Thermo Fisher Scientific). The cells were stained with antibodies specific for IL-17A (TC11-18H10.1) and IFN-γ (XMG1.2) (both from BioLegend), IL-13 (eBio13A) (Thermo Fisher Scientific), and with antibody against the transcription factor Foxp3 (MF23) (BD Biosciences) and were analyzed with a BD LSRFortessa flow cytometer (BD Biosciences). Flow cytometry data were analyzed with FlowJo (version 10.4) software.

Intracellular p-STAT3 and p-STAT1 staining

For the intracellular detection of p-STAT3 in EBV-B cells, cells were starved in RPMI medium containing 2% FCS for 20 hours at 37°C. Cells were stimulated with IFN-α (250 ng/ml) (PeproTech) at 37°C for the indicated time points. For the intracellular detection of STAT1 and p-STAT3 in purified mouse B220⁺ and CD4⁺ cells, cells were cultured in Iscove’s modified Dulbecco’s medium (IMDM) containing 2% FCS for 30 min at 37°C, in a U-bottom 96-well plate. B220⁺ cells were stimulated with IL-21 (50 ng/ml), IL-10 (50 ng/ml), and IL-6 (50 ng/ml) and CD4⁺ cells with IL-6 (50 ng/ml) (all from PeproTech) at 37°C for the indicated time points. Stimulation was stopped by the addition of 100 μl of cold PBS and incubation on ice for 10 min. Cells were stained with viability dye eFluor780 (eBioscience) for 20 min on ice, washed with FACS buffer, fixed with cytofix fixation Buffer (BD Biosciences) for 20 min on ice, washed again, and permeabilized with Phosflow permeabilization buffer (BD Biosciences) for 30 min on ice. Cells were incubated with the following antibodies: anti–p-STAT3 (pY705) Alexa Fluor 647 (BD Biosciences), anti-STAT3 fluorescein isothiocyanate (FITC; R&D Systems), anti–p-STAT1 (pY701) Alexa Fluor 647 (BD Biosciences), and anti-STAT1 rabbit (Cell Signaling Technology) followed by anti-rabbit FITC (Invitrogen), for 30 min on ice, washed, and resuspended in FACS buffer. Cells were analyzed with a BD LSRFortessa flow cytometer (BD Biosciences). Flow cytometry data were analyzed with FlowJo (version 10.4) software.

Immunoblot analysis

Whole cell lysates from EBV-B cells and from HEK293T and STAT3^−/− A4 cells nontransfected or transfected with the indicated cDNA expression vectors were obtained by incubating the cells with lysis buffer [20 mM Hepes (pH 7.6), 2 mM MgCl₂, 150 mM NaCl, 10% glycerol, 0.1% NP-40, 1 mM Na₃VO₄, and 1 mM phenylmethylsulfonyl fluoride] and protease inhibitor mix (Sigma-Aldrich), for 30 min on ice. Lysates were centrifuged at 14,000g for 10 min at 4°C. The supernatants were processed directly for Western blot, and the following antibodies to the C-terminal (Cell Signaling Technology) and N-terminal (R&D Systems) part of the STAT3 protein, p-STAT3 (Tyr⁷⁰⁵; Cell Signaling Technology), Flag (Sigma-Aldrich), Myc (Cell Signaling Technology), tubulin–horseradish peroxidase (HRP; Proteintech), glyceraldehyde phosphate dehydrogenase (GAPDH)–HRP (Cell Signaling Technology), and β-actin (Sigma-Aldrich) were used.

Cells from the spleen, thymus, and bone marrow of mice were washed with cold PBS and lysed for immunoblot analysis with lysis buffer and protease inhibitor mix (Sigma-Aldrich). For detection of phosphorylated STAT1 and STAT3, CD4⁺ cells were cultured in IMDM medium containing 2% FCS for 30 min at 37°C, in a U-bottom 96-well plate, and were stimulated with IL-6 (50 ng/ml) (PeproTech) at 37°C for the indicated time points. Stimulation was stopped by the addition of 100 μl of cold PBS. After washing with cold PBS, cells were incubated with PhoshpoSafe extraction buffer (Merck Millipore) for 30 min on ice. Cells were centrifuged 14,000g for 10 min at 4°C, and the supernatant was collected for immunoblot analysis. Protein extracts corresponding to equal cell numbers were loaded onto the SDS–polyacrylamide gel electrophoresis gel. The samples were then blotted with the following antibodies to p-STAT3 (Tyr⁷⁰⁵; mouse monoclonal), p-STAT1 (Tyr⁷⁰¹; rabbit monoclonal), STAT3 (mouse monoclonal), STAT1 (rabbit monoclonal) (all from Cell Signaling Technology), and α-tubulin (mouse monoclonal) (Proteintech). Immunoblots were developed with Fusion Fx (Vilber Lourmat) and analyzed with ImageJ software.

Fluorescence staining and confocal imaging

Glass coverslips with transfected cells expressing WT or mutant STAT3 (with N-terminally fused fluorescent proteins) were rinsed with PBS, and nuclei were stained with Hoechst 33342 (Sigma-Aldrich). Samples were mounted onto glass slides with fluorescence mounting medium (Dako) for microscopy. Confocal fluorescence images were taken on a Zeiss laser scanning microscope, LSM710, with an Axiovert 200 M equipped with 63x water and 63x oil immersion objectives (Carl Zeiss). Images were evaluated and processed with the Zeiss ZEN software. Cells were processed for analyses 48 h post transfection. ImageJ software was used to measure Pearson’s coefficient for colocalization analysis and for fluorescence intensity analysis.

C. albicans infection

The C. albicans strain SC5314 (American Type Culture Collection) was maintained as glycerol stocks and grown on BBL CHROMagar plates (BD Biosciences). Single colonies were inoculated into liquid YPD medium (1% yeast extract, 2% peptone, and 2% glucose) and grown overnight at 30°C with horizontal shaking at 180 rpm. For infection experiments, C. albicans cultures were washed twice with PBS and counted using a Neubauer chamber. A total of 10⁶ colony-forming units (CFU)/g body weight were then injected into recipient mice intraperitoneally. Mice were weighed every day until the end of the experiment (day 3).

Fungal burden of organs from C. albicans infection

Mice were euthanized 3 days after C. albicans infection. Spleens and kidneys were weighed and later homogenized in myeloperoxidase buffer [200 mM NaCl, 5 mM EDTA, 10 mM tris (pH 8), 10% glycerol, and 1:100 protease inhibitor] using the gentleMACS Octo Dissociator (Miltenyi Biotec). Homogenates were then serially diluted and plated onto BBL CHROMagar plates (BD Biosciences). CFU were counted manually after 4 days of incubation at 30°C.

S. aureus skin infection

S. aureus strain Newman was used for the skin infection mouse model. Bacteria were grown in LB medium to exponential growth phase, washed with PBS, and resuspended in Dulbecco’s PBS. Bacterial concentrations were determined with an optical spectrophotometer (optical density at 600 nm). CFU of the inoculum were verified by serial dilutions on blood agar plates (Columbia agar with 5% sheep blood; bioMérieux). Approximately 10⁷ CFU of S. aureus in 10 μl of PBS were intradermally injected (30-gauge needle, U-100 insulin syringe) into the ear pinna of anesthetized mice.

Quantification of S. aureus CFU

Mice were euthanized 5 days after S. aureus infection; the ears were cut off at the hairline and homogenized using a tissue lyser (QIAGEN). CFU in the homogenized skin tissue were determined by serial dilutions on blood agar plates (Columbia agar with 5% sheep blood; bioMérieux).

Immune cell phenotyping of skin tissue

Mouse ears were collected at day 5 p.i. and subjected to enzymatic digestion by dispase (1 mg/ml; STEMCELL Technologies), collagenase II (2 mg/ml; PAA), and deoxyribonuclease I (0.8 mg/ml, Roche) in Hanks’ balanced salt solution for 2 hours at 1400 rpm shaking and 37°C. After digestion, the samples were filtered with a 40-μm cell strainer (BD Biosciences), washed with PBS, and stained with the indicated antibodies. The following antibodies were used: anti-mouse CD45 eFluor450, anti-mouse CD45 Peridinin Chlorophyll Protein-Cyanine 5.5 (PerCP-Cy5.5), and anti-mouse CD11b phycoerythrin (PE)–Cy7 (all from eBioscience); anti-mouse Ly6G FITC, anti-mouse Ly6C PerCP-Cy5.5, and anti-mouse CD11c FITC were from BD Biosciences. Anti-mouse CD64 PerCP-Cy5.5 was from BioLegend, and anti-mouse CD206 PE was from Invitrogen. Anti-CD16/32 (93; BD Biosciences) was used to block nonspecific binding. Cell samples were analyzed with a 10-laser flow cytometer (Gallios; Beckman Coulter), and data were analyzed with Kaluza software (version 1.2; Beckman Coulter).

Intracellular phenotypic analysis of S. aureus–infected mice

Cells derived from digested ears and from draining LNs (cervical) were resuspended in PBS containing 2% FCS and counted, and 1 × 10⁶ cells were cultured in a U-bottom 96-well plate, in complete medium [RPMI supplemented with 10% FCS and penicillin-streptomycin (Gibco) and stimulated with PMA (100 ng/ml), ionomycine (1 μg/ml) (both from Sigma-Aldrich), and monensin (1×) (BioLegend)] for 5 hours at 37°C. Cells were washed once before staining with fixable viability dye eFluor780 (eBioscience) and stained for surface markers using antibodies against CD3 (145-2C11), CD4 (GK1.5) (both from BioLegend), and TCRγδ (REA633) (Miltenyi Biotec) for 20 min at 4°C. Cells were fixed and permeabilized with cytofix/cytoperm (BD Biosciences) according to the manufacturer’s instructions. The cells were stained with antibodies specific for IL-17A (TC11-18H10.1) and IFN-γ (XMG1.2) (both from BioLegend) and were analyzed with a BD LSRFortessa flow cytometer (BD Biosciences). Flow cytometry data were analyzed with FlowJo (version 10.4) software.

Cytokine protein levels in cell culture supernatant

Protein levels of IL-17A, TNF-α, and IL-6 were measured from cell culture supernatants of LN cells that were cultured for 5 or 24 hours with PMA (100 ng/ml) (Sigma-Aldrich) and ionomycine (1 μg/ml) using LEGENDplex protein assays for mouse inflammation panel, and protein levels of IFN-γ were measured by ELISA, according to the manufacturer’s recommendations (both from BioLegend).

Statistics

Data are expressed as mean ± SD. Data were analyzed by Mann-Whitney test, two-tailed Student’s t test, or two-way analysis of variance (ANOVA) as appropriate, using the GraphPad Prism program. P values of less than 0.05, 0.01, and 0.001 were considered significant.

Supplementary Materials

This PDF file includes:

Figs. S1 to S5