A homozygous STIM1 mutation impairs store-operated calcium entry and natural killer cell effector function without clinical immunodeficiency

To the Editor: 
 
Stromal interaction molecule 1 (STIM1) is a transmembrane protein pivotal to store-operated calcium entry (SOCE) that localizes to either the cell or endoplasmic reticulum (ER) membranes, with the N-terminus in either the extracellular space or the ER, respectively. Plasma membrane ORAI calcium release–activated calcium modulator 1 (ORAI1) Ca2+ channels are activated by STIM1. Families previously described with recessive STIM1 mutations (MIM #612783) had life-threatening viral, bacterial, and fungal infections; developmental myopathy; hypohidrosis; and amelogenesis imperfecta (AI; generalized developmental enamel abnormalities).1, 2, 3 We investigated a consanguineous family, segregating a novel syndrome of recessive AI and hypohidrosis by using autozygosity mapping and clonal sequencing. A homozygous rare missense mutation in STIM1 (p.L74P) in the EF-hand domain was identified (see the Methods and Results sections in this article's Online Repository at www.jacionline.org). 
 
The family was re-evaluated, with particular attention paid to features associated with recessive STIM1 mutations (Table I and see Table E1, Table E2, Table E3 in this article's Online Repository at www.jacionline.org). The 2 affected cousins (18 and 11 years old, respectively) did not have overt clinical immunodeficiency. Further evaluation of their immune systems showed a normal immunoglobulin profile with an adequate specific antibody response to both nonlive (pneumococcus, tetanus and, Hib) and live (mumps, measles, and rubella) vaccinations. In addition, both subjects had detectable IgG against varicella zoster virus after a previous uncomplicated primary infection. The younger cousin was also found to have IgG against EBV viral capsid antigen, suggesting previous exposure, but neither showed any evidence of acute infection or previous exposure to cytomegalovirus. 
 
 
 
Table I 
 
Summary of the main clinical and clinical immunologic features in subjects with either homozygous or heterozygous STIM1 c.221T>C mutations 
 
 
 
Lymphocyte studies showed stable CD8 T-cell depletion in the older affected subject only. Other lymphocyte subsets, including CD4 T, natural killer (NK), and B cells, were within the normal range (Table I). However, despite normal PHA and anti-CD3 simulation responses, T-lymphocyte and NK cell SOCE was grossly abnormal, which is consistent with disruption of the Ca2+-binding EF-hand and in keeping with previous reports for recessive STIM1 mutations (see Fig E1, A, in this article's Online Repository at www.jacionline.org).1, 2, 3 The defect in NK cell SOCE was associated with impaired NK cell effector function, as shown by assays of granule exocytosis and intracellular IFN-γ production in response to K562 tumor cells (see Fig E1, B). After recently published mouse studies, which confirmed the importance of STIM1 to neutrophil SOCE and associated functions,4 we also evaluated neutrophil function. This was found to be within normal limits. 
 
 
 
Fig E1 
 
Defective SOCE and impaired NK cell function in STIM1-Leu74Pro patients' cells. A, Calcium flux in lymphocytes after anti-CD3/anti-CD16, 1 μmol/L thapsigargin, or 500 nmol/L ionomycin administration. B, Granule exocytosis ... 
 
 
 
Despite abnormal immune system SOCE, the affected subjects in this case appear to be able to compensate for this deficit and avoid overt immunodeficiency. It is possible that the relative preservation of T-cell function might compensate for NK cell dysfunction. Neither might yet have encountered a pathogen that would expose this particular immune system limitation (see Table E2). An ability to mount a partial response to viral infections was reported in a family with clinical immunodeficiency and a history of viral infections caused by a homozygous missense R429C change affecting the STIM1 cytoplasmic domain.2 A mouse model characterized by conditional knockout of Stim1 and Stim2 in both CD4+ and CD8+ T cells has recently provided further insight into the importance of Stim1 in immune system development and virus-specific memory and recall responses, which prevent acute viral infections from becoming chronic.5 
 
Recessive STIM1 mutations can be associated with other immune dysregulations, including autoimmune disease. The older cousin had a transient episode of idiopathic thrombocytopenic purpura when 2 years old that might have been unrelated to the STIM1 mutations. There were no other clinical or serologic markers consistent with autoimmune disease, and regulatory T-cell numbers were normal. 
 
Both cousins were intolerant of warm environments and aware of their inability to sweat normally. This limited the older cousin's ability to participate in sport. There was no clinical or serologic evidence of myopathy. This is in contrast to other recessive STIM1 mutations and also to dominant STIM1 mutations affecting the EF-hand that cause tubular aggregate myopathy (MIM #160565).6 Hypomineralized AI affected the primary and secondary dentitions of both affected cousins (see Fig E2 in this article's Online Repository at www.jacionline.org), which is in keeping with reports of other recessive STIM1 mutations. The cousins were physically small (height, weight, and head circumference <0.4th percentile) when assessed at 18 years and 9 years, 10 months of age, respectively. Without comparable data from other subjects with recessive STIM1 mutations, it is unclear whether this is a cosegregating feature. 
 
 
 
Fig E2 
 
Hypomineralized AI as the presenting feature in a family with STIM1 L74P change. A, Pedigree of the consanguineous family investigated. The 2 affected cousins with AI and hypohidrosis are shaded black. Genotypes of the c.221T>C variant are indicated ... 
 
 
 
The L74P STIM1 change within the EF-hand domain precedes the first Ca2+-binding aspartate residue by 2 amino acids (see Fig E2) and therefore might be expected to distort the Ca2+-binding region of the protein. Therefore we compared the response of mutant YFP-STIM1 (L74P) with the depletion of Ca2+ stores after thapsigargin or cyclopiazonic acid (CPA) treatment with that of wild-type YFP-STIM1 and the previously published EF-hand mutant7 YFP-STIM1 (D76A, see Fig E3 in this article's Online Repository at www.jacionline.org). 
 
 
 
Fig E3 
 
STIM1 localization and Ca2+ flux in cells transfected with STIM1 constructs. A, TIRFM of HEK293 cells transfected with either wild-type (WT), D76 A mutant, or L74P mutant YFP-STIM1 after treatment with 2 μmol/L thapsigargin ... 
 
 
 
Using total internal reflection fluorescence microscopy (TIRFM), we replicated previous observations that wild-type YFP-STIM1 relocalizes to puncta proximal to the plasma membrane after treatment of transfected HEK293 cells with 2 μmol/L thapsigargin to deplete ER Ca2+ stores through sarcoendoplasmic reticulum calcium transport ATPase (SERCA) inhibition (see Fig E3, A). The EF-hand mutant YFP-STIM1 (D76 A) was present in these puncta before thapsigargin treatment, with no observable response to thapsigargin (see Fig E3, A). Similarly, mutant YFP-STIM1 (L74P) showed no response to thapsigargin but also appeared to form constitutive puncta, which was less distinct in appearance than that for the D76A mutant (see Fig E3). 
 
We compared Ca2+ fluctuations in HEK293 cells transfected with ORAI-CFP and either wild-type YFP-STIM1, mutant YFP-STIM1 (D76A), or mutant YFP-STIM1 (L74P; see Fig E3, B and C). Both YFP-STIM1 (D76A) and YFP-STIM1 (L74P) transfected cells had increased basal Ca2+ concentrations compared with wild-type YFP-STIM1 and reduced peak and integral responses to CPA-induced SERCA inhibition (see Fig E3, B and C). However, in contrast to the EF-hand mutant YFP-STIM1 (D76A), YFP-STIM1 (L74P) did not demonstrate reduced SOCE after CPA washout and Ca2+ restoration, suggesting that the previously reported desensitization of SOCE observed with the YFP-STIM1 (D76A) mutant does not occur with the YFP-STIM1 (L74P) mutant form. Therefore the L74P mutation appears to result in a distinct molecular phenotype compared with the loss of function observed in immunodeficient patients and the constitutive activation observed in patients with myopathy. 
 
This study is the first to report recessive STIM1 mutations in patients presenting with AI and hypohidrosis without overt clinical immunodeficiency or myopathy. Clinical immunologic investigations were consistent with abnormal NK cell and T-lymphocyte function that might be expected to be associated with ongoing clinical immunodeficiency. However, despite severely abnormal SOCE, this was not the case in these patients. Missense mutations affecting the EF-hand can have very different clinical phenotypes with respect to the immune system, muscle, sweating, and enamel formation. This has important implications for clinical evaluation, as well as understanding the biological functions of STIM1.

A homozygous STIM1 mutation impairs store-operated calcium entry and natural killer cell effector function without clinical immunodeficiency To the Editor: Stromal interaction molecule 1 (STIM1) is a transmembrane protein pivotal to store-operated calcium entry (SOCE) that localizes to either the cell or endoplasmic reticulum (ER) membranes, with the N-terminus in either the extracellular space or the ER, respectively. Plasma membrane ORAI calcium release-activated calcium modulator 1 (ORAI1) Ca 21 channels are activated by STIM1. Families previously described with recessive STIM1 mutations (MIM #612783) had life-threatening viral, bacterial, and fungal infections; developmental myopathy; hypohidrosis; and amelogenesis imperfecta (AI; generalized developmental enamel abnormalities). [1][2][3] We investigated a consanguineous family, segregating a novel syndrome of recessive AI and hypohidrosis by using autozygosity mapping and clonal sequencing. A homozygous rare missense mutation in STIM1 (p.L74P) in the EF-hand domain was identified (see the Methods and Results sections in this article's Online Repository at www.jacionline.org).
The family was re-evaluated, with particular attention paid to features associated with recessive STIM1 mutations (Table I and see  Tables E1-E3 in this article's Online Repository at www.jacionline. org). The 2 affected cousins (18 and 11 years old, respectively) did not have overt clinical immunodeficiency. Further evaluation of their immune systems showed a normal immunoglobulin profile with an adequate specific antibody response to both nonlive (pneumococcus, tetanus and, Hib) and live (mumps, measles, and rubella) vaccinations. In addition, both subjects had detectable IgG against varicella zoster virus after a previous uncomplicated primary infection. The younger cousin was also found to have IgG against EBV viral capsid antigen, suggesting previous exposure, but neither showed any evidence of acute infection or previous exposure to cytomegalovirus.
Lymphocyte studies showed stable CD8 T-cell depletion in the older affected subject only. Other lymphocyte subsets, including CD4 T, natural killer (NK), and B cells, were within the normal range (Table I). However, despite normal PHA and anti-CD3 simulation responses, T-lymphocyte and NK cell SOCE was grossly abnormal, which is consistent with disruption of the Ca 21 -binding EF-hand and in keeping with previous reports for recessive STIM1 mutations (see Fig E1, A, in this article's Online Repository at www.jacionline.org). [1][2][3] The defect in NK cell SOCE was associated with impaired NK cell effector function, as shown by assays of granule exocytosis and intracellular IFN-g production in response to K562 tumor cells (see Fig E1, B). After recently published mouse studies, which confirmed the importance of STIM1 to neutrophil SOCE and associated functions, 4 we also evaluated neutrophil function. This was found to be within normal limits. Despite abnormal immune system SOCE, the affected subjects in this case appear to be able to compensate for this deficit and avoid overt immunodeficiency. It is possible that the relative preservation of T-cell function might compensate for NK cell dysfunction. Neither might yet have encountered a pathogen that would expose this particular immune system limitation (see Table E2). An ability to mount a partial response to viral infections was reported in a family with clinical immunodeficiency and a history of viral infections caused by a homozygous missense R429C change affecting the STIM1 cytoplasmic domain. 2 A mouse model characterized by conditional knockout of Stim1 and Stim2 in both CD4 1 and CD8 1 T cells has recently provided further insight into the importance of Stim1 in immune system development and virus-specific memory and recall responses, which prevent acute viral infections from becoming chronic. 5 Recessive STIM1 mutations can be associated with other immune dysregulations, including autoimmune disease. The older cousin had a transient episode of idiopathic thrombocytopenic purpura when 2 years old that might have been unrelated to the STIM1 mutations. There were no other clinical or serologic markers consistent with autoimmune disease, and regulatory T-cell numbers were normal.
Both cousins were intolerant of warm environments and aware of their inability to sweat normally. This limited the older cousin's ability to participate in sport. There was no clinical or serologic evidence of myopathy. This is in contrast to other recessive STIM1 mutations and also to dominant STIM1 mutations affecting the EF-hand that cause tubular aggregate myopathy (MIM #160565). 6 Hypomineralized AI affected the primary and secondary dentitions of both affected cousins (see Fig E2 in this article's Online Repository at www.jacionline.org), which is in keeping with reports of other recessive STIM1 mutations. The cousins were physically small (height, weight, and head circumference <0.4th percentile) when assessed at 18 years and 9 years, 10 months of age, respectively. Without comparable data from other subjects with recessive STIM1 mutations, it is unclear whether this is a cosegregating feature.
The L74P STIM1 change within the EF-hand domain precedes the first Ca 21 -binding aspartate residue by 2 amino acids (see Fig  E2) and therefore might be expected to distort the Ca 21 -binding region of the protein. Therefore we compared the response of mutant YFP-STIM1 (L74P) with the depletion of Ca 21 stores after thapsigargin or cyclopiazonic acid (CPA) treatment with that of wild-type YFP-STIM1 and the previously published EF-hand mutant 7 YFP-STIM1 (D76A, see Fig E3 in this article's Online Repository at www.jacionline.org).
Using total internal reflection fluorescence microscopy (TIRFM), we replicated previous observations that wild-type YFP-STIM1 relocalizes to puncta proximal to the plasma membrane after treatment of transfected HEK293 cells with 2 mmol/L thapsigargin to deplete ER Ca 21 stores through sarcoendoplasmic reticulum calcium transport ATPase (SERCA) inhibition (see Fig E3, A). The EF-hand mutant YFP-STIM1 (D76 A) was present in these puncta before thapsigargin treatment, with no observable response to thapsigargin (see Fig E3, A). Similarly, mutant YFP-STIM1 (L74P) showed no response to thapsigargin but also appeared to form constitutive puncta, which was less distinct in appearance than that for the D76A mutant (see Fig E3). We compared Ca 21 fluctuations in HEK293 cells transfected with ORAI-CFP and either wild-type YFP-STIM1, mutant YFP-STIM1 (D76A), or mutant YFP-STIM1 (L74P; see Fig  E3, B and C). Both YFP-STIM1 (D76A) and YFP-STIM1 (L74P) transfected cells had increased basal Ca 21 concentrations compared with wild-type YFP-STIM1 and reduced peak and integral responses to CPA-induced SERCA inhibition (see Fig E3,B and C). However, in contrast to the EF-hand mutant YFP-STIM1 (D76A), YFP-STIM1 (L74P) did not demonstrate reduced SOCE after CPA washout and Ca 21 restoration, suggesting that the previously reported desensitization of SOCE observed with the YFP-STIM1 (D76A) mutant does not occur with the YFP-STIM1 (L74P) mutant form. Therefore the L74P mutation appears to result in a distinct molecular phenotype compared with the loss of function observed in immunodeficient patients and the constitutive activation observed in patients with myopathy.
This study is the first to report recessive STIM1 mutations in patients presenting with AI and hypohidrosis without overt clinical immunodeficiency or myopathy. Clinical immunologic investigations were consistent with abnormal NK cell and T-lymphocyte function that might be expected to be associated with ongoing clinical immunodeficiency. However, despite severely abnormal SOCE, this was not the case in these patients. Missense mutations affecting the EF-hand can have very different clinical phenotypes with respect to the immune system, muscle, sweating, and enamel formation. This has important implications for clinical evaluation, as well as understanding the biological functions of STIM1.
We thank the family for participating in this study. We thank Dr Gareth Howell for technical assistance with cell sorting and Dr Peter Baxter, Consultant Paediatric Neurologist at Sheffield Children's NHS Foundation Trust, for his comments. We thank the Exome Aggregation Consortium and the groups that provided exome variant data for comparison. A full list of contributing groups can be found at http://exac.

Antigen-presenting epithelial cells can play a pivotal role in airway allergy
To the Editor: Professional antigen-presenting cells (APCs; ie, dendritic cells, macrophages, and B cells) react against exogenous antigens and initiate an adaptive immune response by presenting antigen peptides in the groove of the MHC class II molecules. During inflammation, ectopic expression of MHC class II has been reported on cells from multiple tissues, including the nasal mucosa, suggesting an antigen-presenting capacity of epithelial cells (ECs). [1][2][3][4] The present investigation was designed to examine the contribution of nasal epithelial cells (NECs) to the allergic inflammatory process. The abilities of NECs to take up antigen, express MHC class II and costimulatory molecules, and stimulate antigen-specific activation and proliferation of CD4 1 T cells were investigated by using a human mucosal specimen (see the Methods section in this article's Online Repository at www. jacionline.org).
First, the cell-surface expression of MHC class II and costimulatory molecules on human and mouse nasal epithelial cells (MNECs) was confirmed (see Figs E1 and E2 in this article's Online Repository at www.jacionline.org). Then the ability of MNECs to present the antigen ovalbumin (OVA) to naive T cells was demonstrated. MNECs from sensitized mice displayed an enhanced MHC class II expression on coculture

Participating family
A consanguineous family of Pakistani heritage was reviewed in the clinical genetics clinic with regard to intolerance to warm environments and generalized dental enamel defects of both dentitions. Sample collection was performed after obtaining informed consent from the patients according to the principles of the Declaration of Helsinki and after local ethics approval. Detailed clinical evaluation was undertaken in appropriate clinical settings.

Genetic mapping
DNA was extracted from blood by using standard procedures. DNA from the 2 affected subjects was genotyped with Affymetrix 6.0 SNP microarrays (Affymetrix, High Wycombe, United Kingdom), and regions of homozygosity were identified by using AutoSNPa software. E1 Linkage was confirmed by means of analysis with fluorescence-labeled polymorphic microsatellite markers on a genetic analyzer (3130xlGenetic Analyzer; Applied Biosystems, Warrington, United Kingdom) using genotyping software (GeneMapper version 4.0; Applied Biosystems). Linkage analyses were performed with LINKMAP and MLINK from the FASTLINK software package. E2

Clonal and Sanger sequencing
We designed a SureSelect Target Enrichment Reagent (Agilent Technologies, Edinburgh, United Kingdom) targeting coding exons within the disease interval in parallel with the capture of disease intervals for 7 other unrelated disorders. The affected subject IV:N was sequenced with 80-nt reads on an Illumina (San Diego, Calif) GAIIx sequencer. Raw data were processed with the Illumina pipeline (version 1.3.4), and reads were aligned to the human reference sequence (hg19/GRCh37) by using Novoalign software (Novocraft Technologies, Selangor, Malaysia). Alignments were processed in the SAM/ BAM format E3 with Picard and the Genome Analysis Toolkit (GATK) E4,E5 to correct alignments around indel sites and to mark potential PCR duplicates. Variants were called in the Variant Call Format by using the Unified Genotyper function of GATK. Filtering of common variation and prediction of functional consequences of variants were performed by using in-house scripts.
PCR products for STIM1 exon 2 and STK33 exon 3 were amplified and sequenced by using the primer pairs shown in Table E1. PCR product cleanup was performed with ExoSAP-IT (Affymetrix) before Sanger sequencing with the BigDye Terminator Cycle Sequencing Kit, version 3.1 (Applied Biosystems) and analysis on an ABI 3130XL DNA analyzer (Applied Biosystems).

Flow cytometric analysis of calcium flux
PBMCs were labeled with Dulbecco modified Eagle medium containing 5 mmol/L Indo-1 for 45 minutes at 378C and then washed and cooled on ice. Cells were incubated for 20 minutes on ice with 5 mg each of unconjugated CD16 (3G8) and CD3-PerCP (OKT3; BD Biosciences, San Jose, Calif) antibodies and costained for gating markers CD19 (SJ25C1) and CD56 (NCAM16.2; BD Biosciences). Cells were washed and resuspended in cold HBSS without calcium. Samples were warmed to 378C and immediately collected on a UV laser equipped LSRII flow cytometer for 90 seconds and then spiked during collection with 1:100 goat anti-mouse antibody for a further 60 seconds (Jackson Laboratory, Bar Harbor, Me), followed by a 1:100 dilution of 200 mmol/L CaCl 2 in PBS solution, and collected for a further 9 minutes. Alternatively, samples were stimulated with the calcium ionophore ionomycin at 500 ng/mL (Sigma-Aldrich, St Louis, Mo) or 1 mmol/L thapsigargin (Sigma-Aldrich) to deplete ER stores of calcium, thereby triggering SOCE and an intracellular calcium ([Ca 21 ] i ) flux. Analysis was performed with FlowJo software (TreeStar, Ashland, Ore), calculating the ratio of calcium-bound to free Indo-1.

NK cell responses
PBMCs were isolated from diluted blood by means of Ficoll separation, followed by NK cell purification by means of negative selection (with immunomagnetic reagents from Miltenyi Biotec, Bergicsch Gladbach, Germany). Isolated NK cells were stimulated with K562 tumor cells alone or in combination with 20 ng/mL IL-12/IL-18 (PeproTech, Rocky Hill, NJ; to maximize IFN-g by tumor-stimulated cells) and incubated for 6 hours at 378C with both GolgiStop and GolgiPlug (BD Biosciences). Cells were stained for the surface markers CD107a (clone; H4A3), CD56 (NCAM16.2), and CD3 (OKT3; BD Biosciences) before fixation for 15 minutes and permeabilization for 30 minutes with the AbD Serotec (Oxford, United Kingdom) intracellular staining kit. Cells were stained with anti-IFN-g (B27) and collected on an LSR II flow cytometer and analyzed in DIVA software (BD Biosciences).

STIM1 constructs for transfection studies
YFP-STIM1 (Addgene plasmid 18857) and the EF-hand mutant YFP-STIM1 (D76 A; Addgene plasmid 18859) constructs were provided by Tobias Meyer through Addgene (Cambridge, Mass). The ORAI1-CFP construct was provided by Anjana Rao (Addgene plasmid 19757). The L74P mutant YFP-STIM1 was produced by means of site-directed mutagenesis of the wild-type YFP-STIM1 plasmid by using the QuikChange II kit (Agilent Technologies, Santa Clara, Calif) per the manufacturer's instructions. The sequences of all 4 constructs were confirmed by means of Sanger sequencing, as above. working distance, 0.12 mm; Nikon, Tokyo, Japan). Cells were maintained at 378C and perfused with standard bath solution; ER store depletion was induced by 2 mmol/L thapsigargin. The plasma membrane was illuminated by using TIRF with a 488-nm argon laser (Prairie Technologies, Middleton, Wis), which was projected onto the specimen through the lens. Images were collected on an electron-multiplying CCD camera (DQC-FS, Nikon) by using NIS Elements imaging software, version 3.2 (Nikon), which was also used for analysis. Fluorescence intensities were background subtracted after acquisition and normalized to the initial intensity (F 0 ).

Calcium measurements in overexpressing cells
HEK293 cells were doubly transfected with ORAI1-CFP and either wildtype YFP-STIM1, mutant YFP-STIM1 (D76A), or mutant YFP-STIM1 (L74P). Twenty-four hours after transfection, cells expressing both CFP and YFP constructs were selected by using a Becton Dickinson FACSAria II cell sorter (BD Biosciences) and plated on glass coverslips. In each case basal [Ca 21 ] i levels were recorded, after which Ca 21 was removed from the perfusate (replaced with 1 mmol/L ethyleneglycol-bis-(b-aminoethylether)-N,N,N9,N9-tetra-acetic acid), and new basal levels of [Ca 21 ] i were determined. Cells were then exposed to CPA (100 mmol/L), and the resultant transient increases in [Ca 21 ] i levels were measured for peak amplitude and integral. After washout of CPA, Ca 21 (2.5 mmol/L) was readmitted to the perfusate, and capacitative Ca 21 entry was quantified as the maximal increase in [Ca 21 ] i observed. Data are presented as representative examples (see Fig  E1, B) and mean 6 SEM values (see Fig E1, C) determined from 12 control recordings, 12 recordings of D76A expressing mutants, and 13 recordings of L74P expressing mutants. Statistical significance was determined by means of ANOVA.

Identification of a novel homozygous missense p.L74P change in STIM1
Autozygosity mapping identified a single region of homozygosity shared by both affected cousins on chromosome 11 between rs11606404 and rs3815045 (chr11 :2,241,215-61,669,946, hg19). Multipoint linkage analysis of markers D11S921, D11S899, D11S915, and D11S4949 against disease by using LINKMAP results in a maximum LOD score of 3.06 at marker D11S899. On merging of overlapping exon intervals, the disease interval contained 3,838 RefSeq coding regions comprising 751,450 bp, 3,784 (739,189 bp or 98.4%) of which could be targeted while avoiding designing baits over repeat masked regions. After target enrichment, sequencing, alignment, and postprocessing, 94.6% of targeted bases were covered by 5 or more nonduplicate reads with a minimum Phred-like base quality score of 17 and minimum read mapping quality of 20.
A total of 526 variants passing standard GATK filters were identified within 20 bp of a coding exon within the disease locus. Variants were removed if present in dbSNP129 or in later versions of dbSNP with a minor allele frequency of 1% or greater, if present in other samples sequenced locally (n 5 31), or, in the case of missense variants, if predicted to be benign by using PolyPhen-2. E6 After these filtering steps, only 3 homozygous variants remained that might be predicted to alter gene function. The first of these (NM_152316: c.3G>C) was considered unlikely to be pathogenic despite altering the initiation codon of ARL14EP because of the presence of another in-frame initiation codon immediately adjacent to the mutated codon and lack of conservation of the first of these ATG codons in mammals. Of the remaining 2 changes, a missense mutation in STK33 (NM_030906: c.146G>A; p.G49D) was found in 4 of 96 ethnically matched control samples, whereas a missense mutation in STIM1 (NM_003156: c.221T>C; p.L74P) was excluded in 192 ethnically matched control samples and found to segregate as expected for a recessively inherited disease within the family. Subsequent interrogation of the Exome Aggregation Consortium database showed that although the STK33 variant was present at a frequency of 1.49% in subjects of South Asian ancestry, the STIM1 variant was not detected at all in the cohort of 60,706 subjects (Exome Aggregation Consortium, Cambridge, Mass; http://exac.broadinstitute.org; accessed February 2015).
Accordingly, the homozygous c.221T>C; p.L74P mutation identified in STIM1 was therefore considered to be the cause of the observed phenotype based on genetic data and the phenotypic overlap with previously reported recessive STIM1 and ORAI1 mutations.  Representative electropherograms are shown alongside the pedigree. B, The hypomineralized AI was characterized by opaque discolored enamel on clinical examination, with radiographs of unerupted teeth consistent with a near-normal volume of enamel and a clear difference in radiodensity between enamel and dentine. *Teeth that have been restored. C, Schematic illustration of STIM1 protein showing the domain structure. Positions of the AI and hypohidrosis-associated L74P mutation (red), dominant TAM or Stormorken syndrome mutations (grey), and recessive syndromic immunodeficiency mutations (black) are indicated above the protein. E-rich, Glutamate-rich region; K, lysine-rich region; MLS, microtubule tip localization signal; P, proline/serine-rich region; SAM, sterile a-motif domain; SOAR, STIM1 Orai1activating region; TM, transmembrane domain. D, Alignment of STIM1 EF-hand orthologous protein sequences. Although p.L74 is conserved in mammals, it is not as strongly conserved as amino acids mutated in dominant TAM. E, NMR structure of STIM1. E7 L74 is shown in red, TAM mutations are shown in dark gray, and Ca 21 binding residues, mutation of which cause constitutive STIM1 activation, are shown in yellow. Substitution of leucine 74 for proline is anticipated to distort the EF-hand loop, interfering with conformational changes in the presence/absence of Ca 21 .   Fuchs et al, 2012 E10 Wang et al 2014 E11 Schaballie et al, 2015 E12 This study Bohm et al, 2013 E13 Morin et al, 2014 E14 Individual (AR) or diagnosis (AD) P r 1, P r 2, and P r 3* P r 4 P r 5 and P r 6 P r 7 P r 8 and P r 9 V2 and V3 Tubular aggregate myopathyà NA P r 5 alive (HSCT) P r 7 lost to follow-up at 5 y P r 8 and P r 9 alive V2 and V3 alive All alive All alive AIHA, Autoimmune hemolytic anemia; AD, autosomal dominant; AR, autosomal recessive; CK, creatine kinase; HSCT, hematopoietic stem cell transplantation; ITP, idiopathic thrombocytopenic purpura; NA, not applicable; NC, no comment made; NR, comment made but feature not recognized. *Mutation confirmed in P r 1 and P r 3; no DNA sample available for P r 2. Mutation identified after death. àA missense change reported in tubular aggregate myopathy and the missense change reported as the cause of Stormorken syndrome have also been identified as the causes of York platelet syndrome, which is characterized by myopathy and platelet abnormalities (Markello et al, 2015 E15 ).