Genetic analysis was subsequently performed as described below. As controls, peripheral blood was obtained from the Karolinska University Hospital blood bank.

Genomic DNA was isolated from the peripheral blood according to standard procedures. HLH-associated genes, including LYST, were sequenced as described previously (21). Mutations identified by high-throughput sequencing were verified by Sanger sequencing. Primers, PCR conditions, and sequencing reaction conditions are available on request.

Peripheral blood was collected in heparin tubes and processed within 48 h of collection. Peripheral blood mononuclear cells (PBMC) were obtained by density gradient centrifugation (Lymphoprep; Axis-Shield) and resuspended in complete medium (RPMI 1640 medium with 10% FBS, l-glutamine, penicillin, and streptomycin; all Hyclone). For some experiments, NK cells were isolated from PBMC by negative selection (Miltenyi Biotec), maintained in complete medium, and used within 2 days of isolation. These cells were determined by flow cytometry to be more than 95% pure CD3⁻CD56⁺ NK cells. The human erythroleukemia cell line K562 and mouse mastocytoma cell line P815 (American Type Culture Collection) were maintained in complete medium.

For cell staining and flow cytometry, fluorochrome-conjugated antibodies to CD3 (clone S4.1; Invitrogen), CD4 (S3.5; Invitrogen), CD8 (RPA-T8; BioLegend), CD16 (3G8; BD Biosciences), CD45 (30-F11; Invitrogen), CD56 (NCAM16.2; BD Biosciences), CD57 (HCD57; BioLegend), CD107a (H4A3; BD Biosciences), perforin (δG9; BD Biosciences), granzyme A (CB9; BD Biosciences), and granzyme B (GB11; BD Biosciences), were used. Fluorochrome-conjugated IgG1 and IgG2b (MOPC-21 and 27-35; BD Biosciences) isotype control antibodies were also used in addition to a fixable live/dead cell marker (Invitrogen). Mouse anti-CD3 (S4.1) and purified anti-CD16 (3G8) mAbs were used for redirected ADCC. For confocal microscopy, mouse monoclonal antibodies to α-tubulin (236-10501; Invitrogen), mannose-6-phosphate receptor (MEM-238; Invitrogen), and perforin (δG9; BioLegend) mAbs were used. Rabbit polyclonal antibodies used were Rab27a, Stx11, and Munc13-4 (all Protein Technologies Group), EEA1 (Cell Signaling Technologies), WASP-interacting protein (WIP, Santa-Cruz), and Cathepsin D (Upstate). Secondary antibodies were donkey anti-mouse and donkey anti-rabbit (both Invitrogen). DNA was labeled with DAPI and actin with phalloidin (both Invitrogen).

Fifty microliters of fresh whole blood were stained in Trucount tubes (BD Biosciences) according to manufacturers instructions with fluorescent conjugated antibodies as listed above to distinguish various leukocyte and lymphocyte populations.

Cells were either evaluated without pre-stimulation or after 36–48 h of stimulation with 500 IU/ml interleukin (IL)-2 (Proleukin, Novartis). For quantification of cytotoxic granule exocytosis, PBMC were mixed with target cells, with 2.5 μg/ml activating mAb where indicated as published (22). Cells were incubated for 2 h at 37°C. For intracellular staining, PBMC were surface stained with fluorochrome-conjugated antibodies as indicated. Cells were then fixed (Cytofix; BD Biosciences), followed by intracellular staining (Cytoperm; BD Biosciences). Isotype control antibodies were used as negative control. For assessment of functional responses, freshly isolated or IL-2-stimulated PBMC were incubated alone, with K562 cells, P815 cells, or P815 cells with either anti-CD16 or anti-CD3 antibody. After stimulation, the lymphocytes were surface stained with antibodies as indicated. All flow cytometry data were acquired on an LSRFortessa (BD Biosciences) and the resulting data were analyzed with FlowJo v9.9 software (TreeStar) and Prism Version 5.0 software (GraphPad). For analyses of cytotoxic granule constituent expression, at least 100 gated cytotoxic lymphocytes were analyzed.

Standard 4-h ⁵¹Cr-release assays were performed for NK cells as previously described (23). Briefly, 4 × 10^{4 51}Cr-labeled K562 target cells were incubated with peripheral blood effector cells in 200 μl of complete medium in 96-well V-bottom plates at various concentrations. Experiments were performed in triplicate, and effector to target cell ratios ranged from 100 to 0.3. The supernatants were measured for ⁵¹Cr release on a gamma-counter (Cobra Auto-Gamma, Packard). Lytic units at 25% target cell lysis were calculated as previously described (24). Cytotoxic T cell cytotoxicity requiring purified CD3⁺CD8⁺CD57⁺ cells was not performed due to the limited volumes of blood obtained from patients (22).

Freshly isolated NK cells or CTL, resting or stimulated with 100 ng/ml phorbol 12-myristate 13-acetate (PMA) (Calbiochem) and 0.5 mM ionomycin (Sigma-Aldrich), or co-incubated with K562 cells were spun down in glass-bottom plates (Matrical, Brooks). After 15 min, cells were fixed with 4% paraformaldehyde in PBS, permeabilized with 0.5% saponin in PBS, and blocked in PBS with 5% FBS, 0.1% BSA-c (Aurion), and 2% normal donkey serum (Jackson ImmunoResearch). Images were acquired on a confocal microscope (Nikon A1R) with a 63× oil objective, using NIS Elements Software, and analyzed using ImageJ (Research Service Branch, National Institutes of Health, Bethesda, MD) or PAD (Digital Cell Imaging Laboratories, Belgium). To quantify the size and number of lytic granules, mean fluorescence intensity, as well as distances between granules, immune synapses (IS) and the microtubule-organizing center (MTOC), images of fixed cells with were analyzed using the PAD software. Object detection based on perforin or CD107a labeling was performed automatically and checked manually. MTOC XY co-ordinates were selected manually based on α-tubulin labeling, and the center of the IS was entered based on phalloidin labeling and phase contrast images. Mean and standard deviation (SD) per patient were then graphed using Prism.

Statistical analysis was performed using Prism software (version 5, GraphPad Software, Inc.), as specified.

In characterizing this cohort of CHS patients, we identified biallelic mutations in LYST in a total of 21 patients from 14 unrelated families (Table 1). While the majority of identified LYST mutations have been previously described (20, 21, 25–30), LYST c.3938delA (p.Q1313RfsX4) in family 9, and c.265insA (p.R886TfsX5) and c.5601delA (p.K1867NfsX11) in family 12 are novel. With the exception of LYST p.A1454N (patient 19) and p.G408R (families 17 and 18), which were present in compound heterozygous form, all mutations were nonsense or frameshift mutations predicted to result in truncated LYST protein. Clinically, 20 out of 21 patients had oculocutaneous albinism, 11 out of 21 patients developed HLH, and 12 out of 21 patients suffered neurological sequelae, including nystagmus, convulsions, and abnormal MRI, The six adult patients who had not developed HLH displayed albinism and neurological symptoms. Only three patients were transplanted, all successfully.

Peripheral blood from the 21 CHS patients with defined mutations in LYST were analyzed with respect to numbers and also stratified according to age at onset of HLH (Table 1). To grossly assess whether the CHS patient cytotoxic lymphocytes developed and differentiated normally, leukocyte subsets were enumerated in whole blood from 13 patients. In this respect, we examined PBMC counts of patients and compared them with healthy related (family) and unrelated (transport control) samples. CHS patient leukocyte numbers were similar to numbers observed in healthy individuals (Figure 1A). More specifically, neutrophil numbers were significantly decreased in CHS patients (Figure 1B), whereas monocyte cell numbers tended to be higher in CHS patients (Figure 1C), and lymphocyte numbers were significantly elevated in CHS patients as compared to healthy controls (Figure 1D). Overall B cell and T cell numbers were elevated above both control groups (Figures 1E–G), while the CD8⁺ T cell subset was not specifically elevated (Figure 1H). With respect to cytotoxic lymphocytes, CTL (CD8⁺CD57⁺ T cells) and NK cell numbers were not significantly elevated in CHS patients (Figures 1I,J), in contrast with a previous report that indicated somewhat elevated HNK-1⁺ (CD57⁺) cell numbers in CHS patients (31). The severity of lymphocytosis, consisting of elevated B cell and CD4⁺ T cell numbers, and neutropenia was most severe in patients with early onset HLH, and least severe in patients with no HLH (data not shown).

To examine whether CHS patient cytotoxic lymphocytes contained sufficient cytotoxic cargo proteins to induce target cell killing, we labeled freshly isolated PBMC with fluorochrome-conjugated antibodies to surface lineage and differentiation markers as well as to intracellular cytotoxic granule proteins and analyzed cells by flow cytometry (Figure 2A). The median fluorescence intensity of CD107a (lysosomal-associated membrane protein 1), perforin, granzyme A, and granzyme B was measured in CD3⁺CD8⁺CD57⁺ CTL (Figures 2B–E) and CD3⁻CD56^dim NK cell (Figures 2F–I) subsets in 18 and 20 patients, respectively. Values from patients, as well as familial and transport controls were normalized against local controls. Intracellular CD107a expression was slightly reduced, whereas perforin expression in CTL and NK cells varied considerably among patients. Granzyme A and B expression was elevated in patient CTL and NK cells. No correlation between expression of cytotoxic granule constituents and HLH diagnosis or age was observed (Figure 2). Thus, consistent with published analyses on 14-day cultured CHS CTL clones that indicated normal perforin and granzyme expression and processing (32), freshly isolated CHS patient cytotoxic lymphocytes expressed normal levels of cytotoxic granule constituents. If anything, levels of perforin and granzymes A and B were elevated in CHS patient lymphocytes, as observed in other familial HLH patients with mutations in genes required for lymphocyte cytotoxic granule exocytosis (23, 33). Of note, granzymes play pleiotropic roles in immune responses, not restricted to induction of target cell death (34).

Lysosomes from all cell types are enlarged in CHS patients and mouse models carrying Lyst mutations (35). We assessed lysosomal size and number in freshly isolated NK cells and CTL from our cohort of CHS patients by immunofluorescence staining of perforin and CD107a and analyses by confocal microscopy. The number of perforin (Figures 3A,B) and CD107a (Figure 3C) objects and their diameters (Figures 3D–F) among CTL (defined as CD3⁺perforin⁺ cells) or NK cell in patients and healthy controls (both familial and unrelated transport samples combined) were determined, providing estimates of perforin dense core numbers and diameters, as well as diameters of the outer, CD107a-coated cytotoxic granule limiting membrane (36). It should be noted that it has previously been reported that CD57⁺ NK cells show a particularly striking phenotype, with a single large cytotoxic granule (31), but lysosomal size and numbers have not been systematically compared between CTL and NK cells. In our cohort, the three patients (patients 1, 2, and 3) with the most 5′ LYST nonsense mutations had just one, enlarged perforin-containing vesicle in 81, 98, and 100% of NK cells, respectively. NK cells from 17 patients in total were examined and found to have 1–2 perforin-containing vesicles per cell (1.3 vesicles on average), whereas CTLs from seven patients had 1–4 perforin-containing vesicles per cell (3.6 vesicles on average). As such, relative to NK cells, T cells from CHS patients had significantly more cytotoxic granules (Figure 3A). In contrast, both NK cells and CTL from healthy controls averaged approximately seven perforin-containing vesicles per cell (Figure 3A). By comparison, CHS patients showed more numerous CD107a-delimited vesicles in NK cells compared to CTL (Figure 3C). Reflecting enlarged cytotoxic granules, in the three patients with the earliest LYST nonsense mutations, the mean NK cell perforin object diameter was 1.1 μm (0.81 ± 0.20, mean ± SD, for all 17 patients examined), with a CD107a object diameter of up to 1.3 μm (0.82 ± 0.23 μm), significantly larger than perforin (0.33 ± 0.04 μm) and CD107a (0.49 ± 0.13 μm) object diameters in normal NK cells (Figures 3D–F). CTL from the same patients contained more than one perforin object of a smaller size (0.50 ± 0.15 μm) relative to CD107a objects (0.54 ± 0.12 μm), albeit significantly larger than control CTL perforin (0.31 ± 0.03 μm) and CD107a (0.42 ± 0.72 μm) object diameters. The larger diameter of CD107a relative to perforin objects likely reflects staining of perforin in dense cores, whereas CD107a is confined to outer, delimiting membranes (36). In several patients, lysosomal morphology was also measured following 20 min of stimulation with target cells or PMA and ionomycin without any discernable difference observed in size or number of cytotoxic granules per cell (data not shown). Altogether, in spite of overall perforin expression being higher in NK cells relative to CTL (22), CHS patient NK cells most often had one giant perforin-containing cytotoxic granule as opposed to smaller and more numerous cytotoxic granules in CTL.

For release of cytotoxic granule content and target cell killing, perforin-containing granules traverse the actin cytoskeleton (37, 38). Smaller cytotoxic granules in CTL might, therefore, enable more efficient traversation and exocytosis. Based on this notion and previous findings of more granule exocytosis by in vitro stimulated CTL as compared to freshly isolated NK cells (20), we thus assessed whether freshly isolated or stimulated CTL from CHS patients might display more cytotoxic granule exocytosis than NK cell counterparts. Lysis of K562 target cells by freshly isolated PBMC was defective compared with controls and below the limit set as pathologic for a diagnosis of HLH in all except patients 6 and 15 (Figure 4A) (39). Culturing healthy control PBMC in IL-2 for 36–48 h increased NK cytotoxicity against K562 cells. CHS patient NK cell cytotoxicity also improved with IL-2, as previously shown (40), but not to the level seen in familial controls (Figure 4B), or patients with STX11 mutations (14, 23).

Cytotoxic lymphocyte exocytosis was assessed in PBMC co-incubated with either K562 or P815 cells added anti-CD16 antibody to trigger NK cells, or P815 cells added anti-CD3 to trigger CTL (Figures 4C–J). Exocytosis induced by K562 cells was defective (ΔCD107a⁺ < 5%) in CD3⁻CD56⁺ NK cells (ΔCD107a⁺ 3.5 ± 2.8%; mean ± SD) in 15 of 20 CHS patients (Figures 4C,G). Exocytosis induced by P815 cells added anti-CD16 was reduced in CD3⁻CD56⁺ NK cells from CHS patients (ΔCD107a⁺ 9.6 ± 5.3%), relative to healthy controls (ΔCD107a⁺ 45.2 ± 14.4%), though higher than against K562 cells (Figures 4C,G). Exocytosis induced by P815 cells added anti-CD3 was defective in CD3⁺CD8⁺CD57⁺ CTL (ΔCD107a⁺ 9.2 ± 4.6%), relative to healthy controls (ΔCD107a⁺ 54.7 ± 19.9%, Figures 4D,H). No correlation was observed between age of onset of HLH and the degree of impairment of fresh or short-term stimulated NK cell or CTL degranulation (not shown). Neither did the severity of the mutation correlate with exocytosis in freshly isolated nor IL-2 activated NK or T cells (Figures 4E,F,I,J), as was previously observed (20). Thus, contrary to our expectations, our assays did not uncover evidence supporting the notion of differential impairment of cytotoxic granule exocytosis in CTL versus NK cells from CHS patients.

In order to assess cytotoxic IS formation and granule polarization, freshly isolated NK cells from two CHS patients were mixed with K562 targets and allowed to form conjugates, which were then analyzed by microscopy. Healthy control NK cells conjugated to K562 target cells polarized perforin granules and accumulated actin and WIP, a key mediator of cytotoxic granule polarization and cellular cytotoxicity (41), at the mature IS (Figure 5A). NK cells from CHS patients were able to form conjugates with K562 targets to a similar extent as healthy controls and showed accumulated actin and WIP at the synapse, demonstrating that a mature IS was formed (Figures 5B–D). Strikingly, CHS NK cells failed to polarize their perforin granules to the mature IS (Figures 5E–G). The distance between the perforin granule and the IS in CHS NK cells was markedly increased compared with healthy donors (Figure 5E; 3.7 ± 0.42 μm, mean ± SD, versus 2.4 ± 0.86 μm, respectively). Similarly, the average distance of the MTOC to the synapse also was increased in LYST-deficient NK cells relative to controls (Figure 5F; 3.9 ± 0.38 μm in CHS patients versus 1.7 ± 0.69 μm in controls). Furthermore, the distance of the granules to the MTOC was also increased in LYST-deficient NK cells relative to controls (Figure 5G; 2.9 ± 0.15 versus 1.6 ± 0.97 μm, respectively). This indicates that in freshly isolated, LYST-deficient NK cells, cytotoxic granule coalescence and MTOC trafficking, early events in cytotoxic granule secretion, are severely impaired. Together, these results substantiate observations in IL-2 cultured NK cells and T cells from CHS patients, where normal IS formation yet impaired cytotoxic granule trafficking has been reported (25, 42).

In order to further characterize the enlarged granules in CHS patient cells, freshly isolated NK cells were labeled for proteins to demarcate cytotoxic granules, late endosomes and early endosomes (Figure 6). Perforin and/or CD107a were used to label the giant lysosomal compartment. A hole was sometimes observed in the center of the perforin positive organelle in CHS NK cells, as previously noted in CTL (Figure 6D) (42, 43). The late endosomal/lysosomal enzyme cathepsin D localized to the lumen of the enlarged cytotoxic granules in CHS patient NK cells (Figure 6A) and has previously been observed to be processed and localized correctly in CHS cultured CTLs, as expected from the largely normal lysosomal function in most cell types of CHS patients (32, 43). However, NK cell cytotoxic granules from CHS patients were negative for mannose-6-phosphate receptor (M6PR, Figure 6A), which transfers some lysosomal enzymes, including granzymes and serglycin, from the trans-Golgi network to late endosomes (44, 45). M6PR partially colocalized with perforin granules in healthy freshly isolated NK cells. Notably, the cytotoxic granule exocytosis regulators Rab27a and Munc13-4 were found to decorate the cytotoxic granule limiting membrane in freshly isolated, unstimulated CHS NK cells (Figures 6B,C), contrasting healthy donor cells where Rab27a and Munc13-4 co-localization is specifically induced by signals from activating receptors (46, 47). Moreover, intriguingly, freshly isolated CHS NK cells and CD3⁺perforin⁺ cells showed EEA1 labeling of the enlarged lytic granules in all 13 patients examined (Figures 6E–H). Taken together, the present results extend earlier findings of mislocalized early and late endosomal markers on CHS lysosomes in cultured NK and T cells to freshly isolated cytotoxic lymphocytes.