Impairment of Mycobacterial But Not Viral Immunity by a Germline Human STAT1 Mutation

See allHide authors and affiliations

Science  13 Jul 2001:
Vol. 293, Issue 5528, pp. 300-303
DOI: 10.1126/science.1061154


Interferons (IFN) α/β and γ induce the formation of two transcriptional activators: gamma-activating factor (GAF) and interferon-stimulated gamma factor 3 (ISGF3). We report a natural heterozygous germline STAT1 mutation associated with susceptibility to mycobacterial but not viral disease. This mutation causes a loss of GAF and ISGF3 activation but is dominant for one cellular phenotype and recessive for the other. It impairs the nuclear accumulation of GAF but not of ISGF3 in heterozygous cells stimulated by IFNs. Thus, the antimycobacterial, but not the antiviral, effects of human IFNs are principally mediated by GAF.

Mendelian susceptibility to mycobacterial disease is a rare syndrome (MIM 209950), leading to severe clinical infections with weakly virulent mycobacterial species, such as Bacillus Calmette-Guérin (BCG) vaccines (1) or environmental nontuberculous mycobacteria (2) and more virulentMycobacterium tuberculosis (3). Other types of microorganisms rarely cause severe clinical disease, except forSalmonella, which infects less than half of the patients. Null recessive mutations have been identified in IL12B(4), encoding the p40 subunit of interleukin-12 (IL-12), in IL12RB1 (5, 6), encoding the β1 chain of the IL-12 receptor, in IFNGR1(7, 8), and in IFNGR2 (9), encoding the two chains of the IFN-γ receptor (IFN-γR). Recessive and dominant mutations, associated with partial IFN-γR, deficiency, have been found in IFNGR1 (3, 10) and IFNGR2 (11). These studies established that human IL-12–dependent IFN-γ–mediated immunity is essential to control mycobacteria and provided means of molecular diagnosis and rational treatment based on pathophysiology. However, no clear genetic etiology has been identified for a number of patients.

We investigated two unrelated patients with unexplained mycobacterial disease. Proband 1 (P1) is a 33-year-old French woman who developed disseminated BCG infection in childhood (12). She had experienced many common viral infections, the clinical course of which was normal. Mutations in IL12B and IL12RB1 had been excluded (13). We characterized cellular responses to IFN-γ by electrophoretic mobility shift assay (EMSA) using Epstein-Barr virus (EBV)–transformed B (EBV-B) cells. The level of nuclear-protein binding to gamma interferon–activating sequences (GAS) in P1 cells stimulated with IFN-γ (Fig. 1A) was 25 ± 3% of that in identically treated control cells (14). The GAS-binding protein, designated gamma-activating factor (GAF), consisted of STAT-1/STAT-1 homodimers (13). This profile was similar to that observed in partial IFN-γR1 and IFN-γR2 deficiency, but mutations in IFNGR1 and IFNGR2 were excluded (13). Moreover, only 25 ± 2% of the GAF detected in control cells was detected in P1 cells stimulated with IFN-α (13). Thus, the binding of nuclear STAT-1 homodimers to GAS was affected equally by IFN-γ and IFN-α in P1 cells.

Figure 1

Cellular responses to IFN-γ and IFN-α. (A) GAS probe-binding nuclear proteins from EBV-B cells from a control (C) and the patient (P), in response to various concentrations of IFN-γ, as determined by EMSA. An excess of unlabeled probe is indicated by “E.” (B) Subcellular distribution of STAT-1 in SV40 fibroblasts from a control (C) and the patient (P), either not stimulated (NS) or stimulated with IFN-γ (105 IU/ml), as shown by indirect immunofluorescence. (C) Tyrosine 701 phosphorylation of STAT-1 in EBV-B cells from a control (C) and the patient (P), either not stimulated (NS) or stimulated with IFN-α or IFN-γ (105 IU/ml), as determined by STAT-1 immunoprecipitation followed by immunoblotting of tyrosine 701–phosphorylated STAT-1 or total STAT-1 molecules. (D) ISRE probe-binding nuclear proteins from EBV-B cells from a control (C) and the patient (P), in response to various concentrations of IFN-α, as determined by EMSA. (E) Subcellular distribution of STAT-2 in SV40 fibroblasts from a control (C) and the patient (P), either not stimulated (NS) or stimulated with IFN-α (105IU/ml), as detected by indirect immunofluorescence. (F) Levels of p48, MXA, and GADPH mRNAs in EBV-B cells from a control (C) and the patient (P), either not stimulated (NS) or stimulated with IFN-α (104 IU/ml) or IFN-γ (102 IU/ml), as detected by Northern blotting.

We then analyzed the subcellular distribution of STAT-1 in simian virus 40 (SV40)–transformed fibroblasts (SV40 fibroblasts) by immunofluorescence (15). A smaller proportion of P1 STAT-1 accumulated in the nucleus upon IFN-γ stimulation than in control cells (Fig. 1B). This suggested that the smaller number of GAS-binding STAT-1 dimers was due to low levels of nuclear GAF and not to a low affinity of nuclear STAT-1 for GAS. Tyrosine 701 phosphorylation is essential for STAT-1 dissociation from IFN-γR1, homodimerization, and nuclear translocation (16). We therefore subjected EBV-B cells to immunoprecipitation with a STAT-1–specific mAb and immunoblotting with a STAT-1–phosphotyrosine 701–specific Ab (17). After stimulation with IFN-α or IFN-γ, the level of tyrosine 701 phosphorylation was lower in P1 than in control cells (Fig. 1C). Similar results were obtained by immunoblotting with a phosphotyrosine-specific monoclonal antibody (13). In contrast, similar levels of serine 727–phosphorylated STAT-1 were detected in P1 and control cells (18). This suggests that in P1 cells, STAT-1 molecules are poorly phosphorylated at tyrosine 701 in response to both IFN-α and IFN-γ. This probably accounts for the impaired nuclear accumulation and DNA binding of GAF.

We investigated the DNA binding activity of STAT-1/STAT-2/p48 trimers, known as interferon-stimulated gamma factor 3 (ISGF3) (14). Similar levels of IFN stimulated response element (ISRE)–binding complex were detected in control and P1 EBV-B cells after IFN-α stimulation (Fig. 1D). Supershift experiments confirmed that the ISRE-binding complex consisted of ISGF3 (13). The subcellular distribution of ISGF3 was then analyzed in SV40 fibroblasts with a STAT-2–specific mAb (15). IFN-α stimulation resulted in strong nuclear accumulation of STAT-2 molecules in both control and P1 cells (Fig. 1E). ISGF3 activation in response to IFN-γ (19) was not detected in our experimental conditions. Thus, GAF activation was impaired in P1 cells stimulated by IFN-α and -γ, whereas ISGF3 activation by IFN-α was not affected. We investigated IFN-α– and IFN-γ–inducible gene transcription by Northern blotting (20). P48 transcription was induced by IFN-γ in control but not P1 cells, whereas MXA transcription was induced by IFN-α in both control and P1 cells (Fig. 1F). The dissociation of IFN-induced signaling pathways mediated by GAF and ISGF3 is therefore functionally relevant for IFN-responsive target genes.

We searched for mutations in the two molecules known to be shared by the IFN-α/β and IFN-γ signaling pathways, JAK-1 and STAT-1. Sequencing of the P1 JAK1 coding region revealed no mutation (13). Sequencing of the STAT1 gene and cDNA revealed a heterozygous substitution (T → C) at nucleotide position 2116 of the coding region, resulting in a serine (S) for leucine (L) substitution at amino acid position 706 (L706S) (21) (Fig. 2A). The L706S mutation was not found in the patient's parents (Fig. 2B), suggesting that the mutational event occurred in one of the parental germlines. P1 transmitted the mutation to her daughter. Unlike all other family members, P1 and her daughter displayed the same abnormal cellular phenotype, with impaired activation of GAF, but not ISGF3, in response to both IFN-γ and IFN-α (13). The L706S mutation was not found in 50 unrelated healthy individuals examined. STAT1was also sequenced in 50 patients with mycobacterial disease. An unrelated patient (proband 2, P2) was found to be heterozygous for the L706S mutation (Fig. 2B). P2 is a 10-year-old American girl who developed Mycobacterium avium infection at 6 years of age (22). These findings strongly suggest a causal relationship between heterozygosity for the STAT1 L706S allele and vulnerability to mycobacteria.

Figure 2

STAT1 genotype and cellular and clinical phenotype. (A) STAT-1, with its coiled-coil (CC), DNA-binding (DNA-B), Src-homology 2 (SH2), tail segment (TS), and transactivation (TA) domains and the positions of tyrosine 701 (Y701), leucine 706 (affected by mutation L706S), and serine 727 (S727). (B) STAT1 genotype and cellular and clinical phenotype in the two kindreds. Healthy individuals with two wild-type copies of STAT1 are shown in white. L706S heterozygous P1 (A.II.2) with BCG infection is shown in black, and her heterozygous daughter, with a cellular but currently no clinical phenotype, is indicated by a vertical bar. Heterozygous P2 (B.II.1) with M. avium infection is shown in black.

We transiently transfected a mouse fibroblast cell line deficient in STAT-1 with wild-type or L706S human STAT1 alleles or an insert-less vector (23). No STAT-1 was detected in cells transfected with insert-less vector (13). In IFN-γ–stimulated cells, wild-type STAT-1 accumulated in the nucleus, whereas L706S STAT-1 remained cytoplasmic (Fig. 3A). Similarly, stimulation by IFN-α led to the nuclear accumulation of both STAT-1 and STAT-2 in cells transfected with the wild-type STAT1 allele, but not in cells transfected with the L706S allele (Fig. 3B). This demonstrates a loss of function for the L706S STAT1 allele, in the nuclear accumulation of GAF and ISGF3 in response to both IFN-γ and IFN-α. Staining with a STAT-1–phosphotyrosine 701–specific antibody demonstrated the phosphorylation of nuclear wild-type STAT-1, but not of cytoplasmic L706S STAT-1 upon IFN-γ stimulation (Fig. 3C). This is consistent with previous experiments in heterozygous cells (Fig. 1C) and strongly suggests that L706S is a loss-of-function mutation, principally because it severely impairs the phosphorylation of tyrosine 701.

Figure 3

Characterization of the mutant STAT1allele. (A) Subcellular distribution of human STAT-1 in mouse STAT-1–deficient fibroblasts, either not stimulated (NS) or stimulated with murine IFN-γ (105 IU/ml), after transfection with wild-type (WT) or mutant (L706S) humanSTAT1 allele, as detected by indirect immunofluorescence. (B) Subcellular localization of human STAT-1 and mouse STAT-2 in mouse STAT-1–deficient fibroblasts, either not stimulated (NS) or stimulated with murine IFN-α (105 IU/ml), after transfection with the wild-type (WT) or mutant (L706S) human STAT1 allele, as detected by indirect immunofluorescence with a confocal microscope. (C) Tyrosine 701 phosphorylation of human STAT-1 in mouse STAT-1–deficient fibroblasts, either not stimulated (NS) or stimulated with IFN-γ (105 IU/ml), after transfection with the wild-type (WT) or mutant (L706S) human STAT1allele, as detected by indirect immunofluorescence.

Transient transfections were carried out with various ratios of wild-type and L706S STAT1 alleles. The nuclear accumulation of STAT-1 upon IFN-γ stimulation was impaired with a 7:3 ratio and completely inhibited with a 3:7 ratio (13), indicating that the L706S STAT1 null allele exerts a dominant-negative effect over the wild-type allele for GAF activation. Three of the four possible STAT-1 combinations recruited to the IFN receptor in wild-type/L706S cells are probably nonfunctional because at least one component cannot be phosphorylated on tyrosine 701. One functional combination (wild type/wild type) is responsible for the 25% residual activity. L706S STAT-1 binding to IFN-γR1 has not been assessed. Although the L706S STAT1allele also causes a loss of function for ISGF3 activation, it is recessive for this phenotype. Crystallographic studies established the crucial role of leucine 706 for STAT-1 homodimerization (24), and STAT-2 has been shown to be recruited first by the IFN-α receptor (25). Thus, L706S STAT-1 is unlikely to dimerize with IFN-αR–bound phosphorylated STAT-2, and sufficient wild-type STAT-1 is probably available in heterozygous cells to form functional ISGF3 complexes. The dimerization of L706S STAT-1 with STAT-2 has not been assessed.

The clinical phenotype of these patients is similar to that of patients with partial IFN-γR deficiency (3, 10,11). Thus, human IFN-mediated antimycobacterial immunity involves principally STAT-1–dependent, GAF-dependent, IFN-γ–stimulated pathways. IFN-γR knockout mice are vulnerable to mycobacteria (26), but the status of IFN-αR (27, 28) and STAT-1 (29, 30) knockout mice is not known. Heterozygous L706S STAT1 mutation does not compromise antiviral immunity, as the three STAT-1–deficient individuals reported here were resistant to all viruses they had been exposed to. IFN-αR (27, 28, 31), STAT-1 (29,30, 31), and STAT-2 (32) knockout mice are susceptible to all viruses tested, suggesting that mouse IFN-α–mediated antiviral immunity is in part ISGF3-dependent. STAT-1–independent effects of IFN-α also contribute to immunity against viruses in mice (31). Thus, either small amounts of GAF are sufficient to protect the patients against viruses, or, more probably, GAF plays no major role in IFN-α–mediated antiviral immunity in humans. IFN-α–stimulated ISGF3 activation, which is preserved in patients carrying the L706S STAT1 mutation, or STAT-1–independent IFN-α–stimulated pathways, presumably intact as well, are probably essential for human antiviral immunity.

  • * To whom correspondence should be addressed. E-mail: casanova{at}


View Abstract

Navigate This Article