Patients with Griscelli syndrome and normal pigmentation identify RAB27A mutations that selectively disrupt MUNC13-4 binding

Patient cohort

In 1989, a registry for hemophagocytic lymphohistiocytosis (HLH) was established,¹ with the aim of centralizing patient information and immunologic and genetic studies for diagnosis. HLH was defined by the diagnostic criteria established by the Histiocyte Society.⁹

Of 901 patients referred between 1989 and 2013, 795 (88%) were given a diagnosis of HLH. Of these, 592 underwent mutation analysis, which in 240 (40%) cases provided identification of a genetic diagnosis of FHL-related condition. Within this cohort, we selected all cases in which pigment dilution, identified by clinicians on the basis of hair color, was reported as part of the clinical picture. In addition, we retrospectively reviewed all patients in whom the final diagnosis was FHL but in whom a genetic marker could not be attributed. In such cases we decided to sequence the pigment dilution–related genes. Informed consent was obtained in all cases by the parents or legal guardian. The study was approved by the A.O.U. Meyer Institutional Review Board.

Perforin expression and degranulation assays

PBMCs from patients with FHL and healthy donors were isolated by using Ficoll gradient centrifugation. Perforin expression in NK cells (CD3⁻CD56⁺ cells of PBLs) was detected by means of intracellular staining (after fixation and permeabilization) with the BD PharMingen (San Jose, Calif) reagent set and cytofluorimetric analysis, as previously reported.³¹

Resting and activated NK cells were also tested by using the granule release assay (GRA), quantifying cell-surface CD107a expression on coculture with K562, as previously described.^11,32,33 Briefly, anti-CD107a–phycoerythrin mAb was added during the coculture for 3 hours at 37°C in 5% CO₂. Thereafter, the cells were stained with anti-CD56–allophycocyanin and anti-CD3–peridinin-chlorophyll-protein complex mAb and analyzed by using flow cytometry (FACSCalibur, Becton Dickinson). All reagents were from BD Biosciences. Surface expression of CD107a was assessed in CD3⁻CD56⁺ cells (NK cells). Results were evaluated as ΔCD107a (ie, Percentage of CD107a⁺ cells of stimulated − Percentage CD107a⁺ cells of unstimulated sample) and defined as defective when lower than the 10th percentile of values in healthy control subjects.

Cell culture

HEK 293 and P815 target cells were maintained in RPMI or Dulbecco modified Eagle medium, respectively (Gibco, Life Technologies, Grand Island, NY), supplemented with 10% FBS. Primary human CD8⁺ T cells were negatively selected from PHA blasts by using the Dynabeads Untouched Human CD8⁺ kit (Life Technologies), according to the manufacturer's instructions, and stimulated with irradiated (3000 rads) PBMCs and 1 μg/mL PHA (Oxoid, Hampshire, United Kingdom). Cells were grown in RPMI supplemented with 5% human serum, 100 U/mL recombinant IL-2, 1 mmol/L sodium pyruvate, 2 mmol/L l-glutamine, 0.075% sodium carbonate, and 50 μmol/L 2-mercaptoethanol to obtain CTLs. Polyclonal NK cell populations were also obtained after NK cell purification (RosetteSep Method; STEMCELL Technologies, Vancouver, British Columbia, Canada) and cultured, as previously described.¹¹

Cytotoxicity assays

CTLs and P815 target cells were resuspended in RPMI (without phenol red, Gibco), 2% FBS, and 0.5 μg/mL anti-CD3 antibody (clone UCHT1, BD Biosciences); plated at different effector/target ratios as shown; and incubated at 37°C for 4 hours. The CytoTox 96 kit (Promega, Madison, Wis) was used to assess target cell lysis per the manufacturer's instructions. Activated purified NK cells were tested for cytolytic activity against the 221 B-EBV target cell line in a standard ⁵¹Cr-release assay.¹¹

Biochemical studies

HEK 293 cells were lipofected (Lipofectamine 2000, Life Technologies) with full-length human Rab27a–green fluorescent protein (GFP) locked in its GTP-bound form by mutation Q78L with and without additional mutations A76V, Y159C, S163R, or p.R141_V142delinsI and either Munc13-4–GFP or Mlph-GFP.

Lipofected HEK 293 cells were lysed with PBS plus 1% Triton X-100, 4-(2-aminoethyl) benzenesulfonyl fluoride hydrochloride (AEBSF), 1 mmol/L CaCl₂, and 1 mmol/L MgCl₂. Lysates were precleared with 100 μL of Protein G Bead slurry (GE Healthcare, Fairfield, Conn) with rotating at 4°C for 30 minutes and then split into 3 samples: no antibody, rabbit anti-Rab27a (produced in rabbits immunized with human Rab27a-GST and screened against lysates from healthy donor and Rab27A-deficient human and mouse cell lines), and either goat anti–Munc13-4 (Everest Biotechnology, Upper Heyford, United Kingdom) or rabbit anti-Mlph (Abcam, Cambridge, United Kingdom). Lysates and antibodies were incubated for 2 hours with rotating at 4°C, followed by addition of Protein G beads and incubation for 1 hour. Beads were washed 4 times with lysis buffer, resuspended with sample buffer, and boiled at 95°C for 5 minutes, and the eluate was then transferred to a fresh tube.

Samples were separated on a 10% acrylamide gel and transferred onto nitrocellulose. Membranes were blocked in PBS with 0.2% Tween 20 and 5% dried skim milk powder and incubated with primary antibodies (as used in IP plus mouse anti-Rab27a [Abnova, Taipei City, Taiwan] or rabbit anti-Mlph [Proteintech, Chicago, Ill]). Membranes were washed 3 times for 10 minutes with PBS plus 0.2% Tween 20 and incubated in secondary antibodies conjugated to horseradish peroxidase (HRP; anti-goat HRP, anti-mouse HRP, and anti-rabbit light chain–specific HRP; all from Jackson ImmunoResearch, West Grove, Pa), washed again as before with an additional 10-minute wash with PBS alone, and then developed with UptiLight (Cheshire Sciences, Chester, United Kingdom) and exposed on x-ray film (Kodak, Rochester, NY).

Pigmentation analysis

Hair samples were cleared in xylene, mounted with synthetic resin (Eukitt; Fluka, Sigma-Aldrich, St Louis, Mo), and observed under a Leitz (Wetzlar, Germany) light microscope equipped with a ProgRes C10plus camera and dedicated software (Jenoptik, Jena, Germany).

Mutation analysis

Genomic DNA was isolated from peripheral blood samples by using the BioRobot EZ1Workstation (Qiagen, Hilden, Germany). RAB27A, LYST, AP3B1, and PRF1 genes were analyzed by means of direct sequencing. The coding exons and exon-intron boundaries were amplified and directly sequenced in both directions with the BigDye Terminator Cycle Sequencing Ready Reaction Kit (Applied Biosystems, Foster City, Calif). Sequences obtained with an ABI Prism 3130XL Sequence Detection System (Applied Biosystems) were analyzed and compared with the reported gene structure by using the dedicated software SeqScape (Applied Biosystems). All mutations were confirmed in the parents.

In silico analysis

All variants of the sequence were searched in dbSNP (http://www.ncbi.nlm.nih.gov/snp/). For variants not reported, we used 3 bioinformatics tools to predict whether an amino acid substitution could be benign or deleterious: SIFT (Sorting Intolerant From Tolerant; http://sift.jcvi.org/), PolyPhen (Polymorphism Phenotyping; http://genetics.bwh.harvard.edu/pph/), and Pmut (http://mmb.pcb.ub.es/Pmut).³⁴

Patients with FHL with partial albinism with assigned genetic markers

A total of 21 children from 20 families were reported to have HLH and some degree of pigment dilution. They accounted for 2.5% of the total cases of HLH enrolled in the registry. In 12 patients a genetic defect associated with pigment dilution and immunodeficiency could be identified (Table I) and assigned as follows: GS2, 6 cases; Chediak-Higashi syndrome, 4 cases; and Hermansky-Pudlak syndrome type 2, 1 case. Unexpectedly, we found biallelic PRF1 mutations in 1 patient with albinism. The association between albinism and FHL type 2 was never reported and appears to be noncausal.³⁵

We identified 7 novel mutations in the RAB27A, LYST, and AP3B1 genes. Among these, 3 were missense mutations, and all were predicted by bioinformatics facilities as damaging protein function. In 2 cases we observed a single variation: patient UPN241 has a stop mutation in RAB27A, with protein study showing a protein defect, and patient UPN860 has a single mutation in LYST, an intronic defect that is predicted by the software Human Splicing Finder to create a new acceptor site. Full characterization of these patients has yet to be completed.

Patients with FHL with partial albinism with unassigned genetic markers

For 9 patients, a genetic marker in one of the pigment dilution–associated genes (or one of the FHL-related genes) could not be assigned (Table I). In 2 siblings (patients UPN42 and UPN43) the genetic study (as well as the functional studies) could not be performed because of lack of consent. Two patients had findings suggestive of an underlying genetic defect (ie, defective degranulation at GRA [patient UPN540] or defective cellular cytotoxicity [patient UPN244]). Another patient with no assigned marker (patient UPN299) had HLH at 2.5 years of age (with normal GRA and cytotoxicity but abnormal T-cell proliferation after CD3 stimulation), which resolved after initial treatment. This patient was later given a diagnosis of a familial depigmentation syndrome combined with recurrent infections. The family of patient UPN299 is currently under investigation for identification of a possible novel defect. The remaining 4 patients had either normal degranulation (n = 2) or normal cytotoxicity (n = 1) or were not investigated. The unassigned patients will be subject to exome sequencing in the near future.

Analysis of pigmentation dilution in patients with FHL with partial albinism

Microscopic analysis of hair for pigment dilution produced varied results among subjects (Table I). All pigment granules were roughly oval, with the major axis aligned to the hair length. In patient UPN396 there were fine pigment granules (up to about 1.5 × 0.8 μm in diameter), as in normal blond hair. In patient UPN313 there were fine pigment granules accompanied by some medium-sized granules (about 5 × 3 μm in diameter). In 2 patients (patients UPN775 and UPN907) there were large granules (about 10 × 3 to 10 × 6 μm in diameter) unevenly distributed along the hair. One of these patients (patient UPN775) also had some hairs with fine pigment granules and no large granules.

In 3 subjects (patients UPN299, UPN797, and UPN886) there were medium-sized pigment granules. In 2 of these subjects (patients UPN797 and UPN886), there were also occasional black spindle-shaped (up to about 10 × 1 or 2 μm) images close to the hair axis but not exactly axial. Given the shape and intensely black color, these images were interpreted as air-containing fissures within the hair cortex. We have not been able to define the correlation between the degree of hair depigmentation and specific RAB27A mutations.

Patients with GS2 but no evidence of pigment dilution

In addition to the patients with FHL and partial albinism (defined on the basis of the partial or complete absence of normal hair color), 45 additional patients with suspected FHL (caused by either familial disease or a confirmed severe functional defect in degranulation or cytotoxicity assays) and apparently normal pigmentation were identified, in whom we did not find biallelic mutations in the FHL-related genes PRF1, UNC13D, STX11, and STXBP2 (or SH2D1A for XLP1 and XIAP for XLP2 in male subjects). Thus we asked whether they might have mutations in RAB27A that could give rise to immune defects underlying HLH in the absence of pigment dilution by preserving binding to Mlph. Of them, 29 could be investigated by using mutation analysis. Analysis of the RAB27A gene in 6 patients from 4 unrelated families of southern European origin unexpectedly showed biallelic mutations in RAB27A (Table II). All of these patients carried at least 1 novel mutation. Remarkably, the p.R141_V142delinsI mutation was shared by 5 patients from 3 unrelated families with GS2 but normal hair pigmentation: 3 homozygous siblings (patients UPN154, UPN155, and UPN313), 1 patient in association with the novel missense mutation p.S163R (patient UPN324), and another patient (patient UPN226) in association with the known p.Q172Nfs mutation. Patient UPN396 carried 2 novel missense mutations (A76V and Y159C). All missense mutations were tested by using bioinformatics software (PolyPhen-2) and defined to be pathogenic. It is interesting to note that within the same family, 2 patients (patients UPN154 and 155) had a quite typical early onset of GS2, whereas the third case (patient UPN313) had the disease only after the age of 10 years. When reviewing the hair phenotype of these patients, we confirmed that none had silver-gray hair, and only 2 were reported to have some “light or blond hair.”

The results of analysis in patients UPN396 and UPN324 are shown in Fig 1, A, and Fig E1, which is available in this article's Online Repository at www.jacionline.org. Both exhibited a severely reduced ability to kill target cells (maximal 20% cytotoxicity) compared with a healthy donor (maximal 80% cytotoxicity). To determine whether these patients still expressed Rab27a protein, we analyzed lysates from patients' CTLs using immunoblotting. We noted that the mutations found both in patient UPN396 (A76V and Y159C) and patient UPN324 (S163R and p.R141_V142delinsI) all resulted in reduced Rab27a protein levels in lysates from patients' CTLs (Fig 1, B). Interestingly, the Munc13-4 protein level was unaffected (Fig 1, B).

Mutations on one face of Rab27a can disrupt binding to Munc13-4 but not Mlph

The identification of missense mutations in Rab27a that affected only CTL effector function but not melanocytes offered a unique opportunity to compare the molecular function of Rab27a in cytotoxicity versus pigmentation. Because the interaction of Rab27a with Munc13-4 is required for granule secretion from CTLs and the interaction between Rab27a and Mlph is required for melanosome secretion, we hypothesized that this might be caused by mutations in Rab27a that selectively disrupted binding with Munc13-4 but not Mlph. To investigate this hypothesis, we used the structure of mouse Rab27a in complex with human synaptotagmin-like protein 2a (Slp2a) to examine how mutations A76V, Y159C, S163R, and p.R141_V142delinsI might affect potential interaction sites or the stability of the protein.³⁶ The mutation A76V lies adjacent to the active site of Rab27a, where the Mg²⁺ ion binds (green sphere, Fig 2) and GTP hydrolysis occurs, suggesting that the catalytic site might be impaired by this mutation. This is consistent with previous findings that A76V is a deleterious mutation found in patients with GS2 with immunodeficiency and albinism.^37,38 Interestingly, residues Y159, S163, and R141 all lie close to each other in the 3-dimensional structure on an outer face of Rab27a, and the structure provides insights into why mutations of these residues might disrupt a possible interaction with Munc13-4. The structure shows that Y159 forms a hydrogen bond to the carboxy oxygen of glutamate 161. When mutated to a cysteine, this bond would be lost, and the conformation of the protein would be altered. Similarly, S163 can form a hydrogen bond to the carboxy oxygen of aspartate 136, and this would be abolished by mutation to arginine (Fig 2, second inset in second row). The deletion of 3 bases in the RAB27A gene (c.422_424delGAG) is found in 5 of the 6 novel patients with GS2 and results in deletion of R141 and replacement of V142 by isoleucine. Interestingly, residue R141 can also form hydrogen bonds with the carboxy group of glutamate 161 (Fig 2, third inset in second row), and these bonds would no longer exist when this residue is deleted.

Although active (GTP-bound) Rab27a in CTLs and NK cells binds to the priming factor Munc13-4, in melanocytes the key effector protein of Rab27a is the synaptotagmin-like protein Mlph. Therefore we investigated the effect of these mutations on the ability of active Rab27a to bind to either Munc13-4 or Mlph to determine whether these mutations might disrupt the binding of Munc13-4 but not Mlph and explain the loss of cytotoxicity without the loss of pigmentation. We coexpressed a point mutant of Rab27a (Q78L) that lacks GTP hydrolysis activity^36,39,40 and each of the 4 mutations (A76V, Y159C, S163R, and delR141) found in patients (Table I), together with Mlph or Munc13-4 in HEK 293 cells, and examined whether the Rab27a mutations identified affected binding to Munc13-4 and Mlph.

When Munc13-4 immunoprecipitates were probed with anti-Rab27a antibodies, neither the S163R (Fig 3, A, lane 4) nor the p.R141_V142delinsI (Fig 3, A, lane 6) mutant forms of Rab27a were able to coprecipitate with Munc13-4 as effectively as native Rab27a (Fig 3, A, lane 2). Immunoprecipitation with Mlph showed that Rab27a S163R also showed greatly reduced binding to Mlph (Fig 3, B, lane 4). However, Rab27a p.R141_V142delinsI was able to bind Mlph (Fig 3, B, lane 6). Therefore the combination of these alleles would result in a specific loss of Rab27a interaction with Munc13-4 but not Mlph. These results provide a possible molecular explanation for the atypical GS2 phenotype seen in patients with severe immunodeficiency but lack of a pigmentation defect in patients UPN154, UPN155, and UPN313, who were all homozygous for the mutation at R141, as well as patient UPN324, in whom the R141 mutation is found on one allele and the S163R mutation is found on the other allele. The selective loss of Munc13-4 but not Mlph binding of the p.R141_V142delinsI mutation also provides a molecular explanation for patient UPN226 because p.Q172Nfs predicts production of a truncated protein after Q172 that would likely be degraded. Because no patients' cells were available, this could not be tested.

The mutations in patient UPN396 differed, with A76V lying in the catalytic site of Rab27a and Y159C lying close to S163, which was mutated in patient UPN324. Analyzing the ability of A76V and Y159C to interact with Munc13-4 and Mlph, we found that neither A76V nor Y159C could coprecipitate Munc13-4 as effectively as active Rab27a (Fig 3, C), although both could coprecipitate Mlph as effectively as active Rab27a (Fig 3, D). These findings explain why there is a loss of cytotoxicity in UPN396 (Fig 1, A) without a loss of pigmentation.