A Heterozygous RAB27A Mutation Associated with Delayed Cytolytic Granule Polarization and Hemophagocytic Lymphohistiocytosis

Abstract

Frequently fatal, primary hemophagocytic lymphohistiocytosis (HLH) occurs in infancy resulting from homozygous mutations in NK and CD8 T cell cytolytic pathway genes. Secondary HLH presents after infancy and may be associated with heterozygous mutations in HLH genes. We report two unrelated teenagers with HLH and an identical heterozygous RAB27A mutation (c.259G→C). We explore the contribution of this Rab27A missense (p.A87P) mutation on NK cell cytolytic function by cloning it into a lentiviral expression vector prior to introduction into the human NK-92 cell line. NK cell degranulation (CD107a expression), target cell conjugation, and K562 target cell lysis was compared between mutant- and wild-type-transduced NK-92 cells. Polarization of granzyme B to the immunologic synapse and interaction of mutant Rab27A (p.A87P) with Munc13-4 were explored by confocal microscopy and proximity ligation assay, respectively. Overexpression of the RAB27A mutation had no effect on cell conjugate formation between the NK and target cells but decreased NK cell cytolytic activity and degranulation. Moreover, the mutant Rab27A protein decreased binding to Munc13-4 and delayed granzyme B polarization toward the immunologic synapse. This heterozygous RAB27A mutation blurs the genetic distinction between primary and secondary HLH by contributing to HLH via a partial dominant-negative effect.

Figure 1. Rapid clinical response of sHLH to immunosuppression with high dose CS, CsA, and anakinra

Serum ferritin (circles) and AST (squares) levels, and circulating platelet counts (triangles) are graphed pre- (day 0) and post-immunosuppression (methylprednisolone, CsA, and anakinra/rIL-1Ra) for the American sHLH patient with the heterozygous Rab27A p.A87P missense mutation.

Figure 2. Rab27A p.A87P mutation decreases NK cell cytolytic function

(A) eFluor450-labeled K562 target cells were mixed at increasing E:T ratios with NK-92 cells transduced with lentiviruses expressing empty vector control (left column), Rab27A WT (middle), or Rab27A p.A87P (right) and were incubated for 4 hours prior to FCM analysis of K562 cell death. FCM plots were gated on eFluor450+ cells with eFluor450 depicted on the Y-axis, and near-IR staining (dead cells) along the X-axis. One representative experiment is shown with percentages of lysed K562 cells enumerated. (B) A graph showing percent cytotoxicity at 3 different E:T ratios for lentivirus-transduced NK-92 cells [empty vector control (circles), Rab27A WT (squares), and Rab27A p.A87P mutation (triangle)]. Results are plotted as means ± SEM for 5 independent experiments. Two-way ANOVA analysis revealed statistically significant differences (p<0.05) among empty vector, the Rab27A WT-, and Rab27A p.A87P mutant-transduced NK-92 cells at all 3 E:T cell ratios. (C) K562 cytotoxicity (means ± SEM) from empty vector, WT, or Rab27A p.A87P mutant lentiviral transduced NK-92 cells in the presence or absence of exogenous IL-6 are shown for a 4:1 E:T ratio. IL-6 significantly decreased cytotoxicity of Rab27A WT (* p=0.0175) and p.A87P mutant (** p=0.0058) transduced cells (n=3).

Figure 3. Rab27A p.A87P mutation does not alter NK cell conjugate formation with target cells

NK-92 cells transduced with lentiviruses co-expressing GFP and empty vector control (left), Rab27A WT (middle), or Rab27A p.A87P (right) were mixed with eFluor450-labeled K562 cells and assayed for cell-to-cell conjugation at 30 minutes as assessed by FCM. (A) GFP expression (NK-92 cells) is shown along the Y-axis and eFluor450 fluoresence (K562 cells) along the X-axis. NK-92-K562 cell conjugates are noted in the upper right quadrants, and percentages of cells in each quadrant are noted. (B) The results shown are representative of 3 similar experiments, and the means ± SEM for percent conjugates are shown in the bar graph to the right.

Figure 4. The Rab27A p.A87P mutation is located in a putative Rab27A/Munc13-4 binding site

The crystal structure of Rab27A (green) is shown with Munc13-4 homologs, melanophilin (yellow) and Slp2-a (orange). The positions of Rab27A mutations p.A87P (purple) and p.I44T (gray), both of which disrupt binding to WT Munc13-4 in vitro, are shown.

Figure 5. Rab27A p.A87P mutation results in decreased interaction with Munc13-4 WT protein

(A) NK-92 cells were transduced with a lentivirus expressing FLAG-tagged Munc13-4 WT followed by superinfection with lentiviruses expressing nothing (top row), or HA-tagged Rab27A WT (middle) or Rab27A p.A87P mutant (bottom). The transduced NK-92 cells were stimulated with K562 cells for the indicated time periods prior to in situ PLA as described in the Methods. Particles representing protein-protein interactions specifically between transduced Munc13-4 and Rab27A are noted in orange. (B) Results from 3 independent experiments are summarized and plotted as in situ PLA particles per cell (means ± SEM) along the Y-axis and time of stimulation along the X-axis. Two-way ANOVA analysis showed statistically significant differences (p<0.05) between the means at the 1 and 2 hour time points for the Rab27A WT- and Rab27A p.A87P-transduced cells.

Figure 6. Rab27A p.A87P mutation decreases NK cell degranulation and alters cytolytic granule polarization to the immunologic synapse

(A) NK-92 cells transduced with lentiviruses expressing empty vector control (top row), Rab27A WT (middle), or Rab27A p.A87P (bottom) were mixed with K562 cells for the indicated times, and cell surface CD107a expression was detected by FCM. One representative set of results from 3 independent experiments is shown and gated on CD56+ NK-92 cells with CD56 expression depicted along the Y-axis and CD107a along the X-axis. Percentages (%) of CD107a+ cells and CD107a mean fluorescent intensities (MFI) are noted. (B) Percent decreased degranulation was calculated by the formula: 100% × (CD107a positive% × MFI from Rab27A WT transduced NK-92 minus CD107a positive% × MFI from Rab27A p.A87P transduced NK-92)/CD107a positive% × MFI from Rab27A WT transduced NK-92. The bar graph shows the means ± SEM from 3 independent experiments. (C) NK-92 cells transduced with lentiviruses expressing Rab27A WT (top panel) or Rab27A p.A87P (bottom panel) were incubated with CFSE-labeled K562 target cells. Confocal microscopy portrays granzyme B (present only in NK-92 cells) as pink to orange in color, F-actin as blue, and both NK-92 and K562 cells as green. The arrow in the top panel denotes NK-92 cell granzyme B polarization to the immunologic synapse formed with the K562 target cell, whereas the arrow in the lower panel points out a NK-92 cell bound to a K562 cell but lacking granzyme B polarization to the immunologic synapse.

Figure 7. Polarization of granzyme B occurs at the plasma membrane of the immunologic synapse between NK-92 effector cells and K562 target cells but is decreased in those expressing mutant Rab27A p.A87P

NK-92 cells effector cells (E) transduced with lentiviruses expressing empty vector control (left), Rab27A WT (middle), or mutant Rab27A p.A87P (right) were mixed with K562 cells target cells (T) and assayed for granzyme B polarization to the plasma membrane at the immunological synapse by confocal microscopy. granzyme B is noted in green (top left panels), DNA (DAPI nuclear stain) in blue (upper right panels), and F-actin cytoskeleton (defining cell membranes) in red (lower left panels). Merged images of granzyme B, DNA, and F-actin are presented in the bottom right panels with yellow representing granzyme B and F-actin overlap, and light blue depicting granzyme B not polarized to the cell membrane immunological synapse.

Figure 8. Rab27A p.A87P mutation delays granzyme B polarization to the immunologic synapse

(A) Vector control (top row), Rab27A WT (middle), and Rab27A p.A87P mutant (bottom) transduced NK-92 cells were stimulated with K562 cells for the indicated time periods prior to intracellular analysis of granzyme B polarization as assessed by confocal microscopy. One representative experiment is shown, where green represents granzyme B staining and blue denotes cell nuclei (DAPI stain). (B) Three independent experiments are summarized and plotted as percentages (Y-axis) of granzyme B polarized NK-92 cells (means ± SEM) versus time (X-axis) for vector control (circles), Rab27A WT (squares), and Rab27A p.A87P (triangles) expressing lentivirus-transduced cells. Statistically significant differences (p<0.0001) in the means of granzyme B-polarized cells between the Rab27A WT- and Rab27A A87-expressing NK-92 cells was noted by two-way ANOVA analysis for the 0.5 and 1.0 hour time points.

Figure 9. Increased IFNγ expression by mutant Rab27A p.A87P, compared to WT Rab27A, expressing NK-92 cells following incubation K562 target cells

(A) Empty vector (CTL), mutant Rab27A p.A87P, or WT Rab27A lentiviral transduced NK-92 cells were incubated with K562 cells for 4 hours, and intracellular IFNγ levels of GFP⁺ NK-92 cells are plotted (X-axis) versus side scatter (Y-axis) for one representative experiment. (B) Average IFNγ levels (means ± SEM) are plotted for 3 similar experiments.

Figure 10. Heterozygous Rab27A p.A87P patient NK cells have decreased degranulation and cytotoxicity

(A) PBMC were obtained from the heterozygous Rab27A p.A87P sHLH Italian patient (black bar), his Rab27 p.A87P heterozygous father (checkered bar), his Rab27A WT mother (gray bar), and his Rab27A WT sister (white bar). PBMC were mixed with K562 cells at various E:T ratios as in the Methods. Percentages of NK cells undergoing degranulation (CD107⁺, CD56⁺, CD3⁻) were determined by FCM. (B) PBMC were obtained from the heterozygous Rab27A p.A87P sHLH Italian patient (at 2 time points over a 24-month period of disease inactivity; black bar), his Rab27A p.A87P heterozygous father (single time point; checkered bar), and from 3 normal healthy unrelated control individuals (white bar). Cell lysis of NK-sensitive K562 cells was analyzed at various E:T ratios as detailed in the Methods. Error bars represent means ± SEM for the patient (2 time points) and the average of the 3 controls.