NK cell dysfunction and interferon-γ production underlie autoinflammation in mevalonate kinase deficiency.

Abstract
In brief The mechanisms linking defective protein prenylation to systemic inflammation in mevalonate kinase deficiency remain unclear. Munoz et al. reveal that mice and humans with MKD have defective NK cells, with impaired granule trafficking, reduced cytotoxicity, and increased IFNγ production. Their findings help explain the cause of inflammatory flares in MKD and provide a rationale for testing JAK inhibitor therapy.


Article


NK cell dysfunction and interferon-γ production
underlie autoinflammation in mevalonate kinase deficiency


Graphical abstract
Highlights • Mice and humans with MKD have decreased frequencies of mature NK cells • NK cells in MKD have defective protein prenylation and reduced cytotoxicity • Mice with MKD have an impaired anti-viral response and elevated serum IFN-γ • JAK inhibitor therapy may be beneficial in the treatment of MKD Authors Marcia A. Munoz, Iona S. Schuster, James Cremasco, ..., Joost Frenkel, Mariapia A. Degli-Esposti, Michael J. Rogers Correspondence m.rogers@garvan.org.au In brief The mechanisms linking defective protein prenylation to systemic inflammation in mevalonate kinase deficiency remain unclear. Munoz et al. reveal that mice and humans with MKD have defective NK cells, with impaired granule trafficking, reduced cytotoxicity, and increased IFNγ production. Their findings help explain the cause of inflammatory flares in MKD and provide a rationale for testing JAK inhibitor therapy. Munoz et al., 2026, Immunity 59, 1–17 June 9, 2026 © 2026 The Author(s). Published by Elsevier Inc. https://doi.org/10.1016/j.immuni.2026.03.027 ll


Article


NK cell dysfunction and interferon-γ
production underlie autoinflammation in mevalonate kinase deficiency Marcia A. Munoz,1,2 Iona S. Schuster,3,4,18 James Cremasco,1,18 Etienne N. Masle-Farquhar,1,2,18 Oliver P. Skinner,1 Zoe J. Vandeleur,1,2 Maté Biro,1,5 Daryan Kempe,5 William D. Renton,6,7,8 Sam Mehr,6 Charlotte Abell-King,1,2 Szun Szun Tay,1 Ryan C. Chai,1,2 Samar Ojaimi,7,8,9 John J. Zaunders,10 Geetha Rao,1 Ariel Castro-Martinez,1 Lisette van de Corput,11 Andrew N. McCorkindale,1 Leonard D. Goldstein,1,2 Xiaohong Li,12 Flore Wouters,13 Daniel L. Kastner,14 Ignatius Chua,9,15 Nicole L. Fewings,16 Fiona C. McKay,16 Catharina M. Mulders-Manders,17 Robert A. Brink,1,2 Stuart G. Tangye,1,2,9 Ivona Aksentijevich,14 Cindy S. Ma,1,2,9 Jeroen van der Hilst,13 Joost Frenkel,11 Mariapia A. Degli-Esposti,3,4 and Michael J. Rogers1,2,9,19,* 1Garvan Institute of Medical Research, Darlinghurst, NSW, Australia 2School of Clinical Medicine, UNSW Sydney, Sydney, NSW, Australia 3Biomedicine Discovery Institute, Monash University, Clayton, VIC, Australia 4Centre for Experimental Immunology, Lions Eye Institute, Perth, WA, Australia 5EMBL Australia, Single Molecule Science node, School of Biomedical Sciences, UNSW Sydney, Sydney, NSW, Australia 6Royal Children’s Hospital, Melbourne, VIC, Australia 7Monash Health, Melbourne, VIC, Australia 8Monash University, Clayton, VIC, Australia 9Clinical Immunogenomics Research Consortium Australasia (CIRCA) 10Centre for Applied Medical Research, St Vincent’s Hospital, Sydney, NSW, Australia 11Utrecht University Medical Center, Utrecht, the Netherlands 12UT Southwestern Medical Center, Dallas, TX, USA 13Hasselt University, Hasselt, Belgium 14National Human Genome Research Institute, Bethesda, MD, USA 15Christchurch Hospital, Christchurch, Canterbury, New Zealand 16Centre for Immunology and Allergy Research, Westmead Institute for Medical Research, The University of Sydney, Sydney, NSW, Australia 17Radboud University Medical Center, Nijmegen, the Netherlands 18These authors contributed equally 19Lead contact *Correspondence: m.rogers@garvan.org.au https://doi.org/10.1016/j.immuni.2026.03.027


SUMMARY
Prenylopathies such as mevalonate kinase deficiency (MKD) are an emerging family of monogenic autoinflammatory diseases with an underlying defect in isoprenoid lipid synthesis and protein prenylation. The mechanisms linking defective protein prenylation to systemic inflammation remain unclear. We revealed that mice and humans with MKD had significant decreases in the frequency of mature natural killer (NK) cells, impaired trafficking of cytolytic granules, reduced cytotoxic activity, and increased production of the cytokine interferon γ (IFN-γ). Mice with MKD failed to clear murine cytomegalovirus (MCMV) infections and had elevated serum IFN-γ and inflammatory pathology, likely the result of decreased and dysregulated cytotoxic cells. Finally, we describe the beneficial effect of cytokine signaling blockade with a Janus kinase (JAK) inhibitor in an infant with severe MKD. Together, these findings reveal a fundamental role for dysregulated cytotoxic cells and IFN-γ production in MKD and likely other prenylopathies. Importantly, this work provides a rationale for the use of JAK inhibitors in the treatment of MKD.


INTRODUCTION
Protein prenylation is a form of post-translational modification that is necessary for the membrane tethering and function of more than 300 proteins, particularly the superfamily of small GTPases such as Rab and Rho GTPases.1 These proteins act as molecular switches, regulating signaling pathways that control diverse cellular functions, including cytoskeletal arrangement, cell polarization, mitophagy, and vesicular trafficking.2 The long-chain isoprenoid lipids required for protein prenylation are synthesized by the ubiquitous mevalonate pathway. In humans, a rare inborn error of metabolism, caused by combinations of complete or partial Immunity 59, 1–17, June 9, 2026 © 2026 The Author(s). Published by Elsevier Inc. 1 This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/). ll OPEN ACCESS (legend continued on next page) 2 Immunity 59, 1–17, June 9, 2026 loss-of-function mutations in MVK, encoding mevalonate kinase (MK), disrupts this pathway and prevents downstream protein prenylation,3,4 leading to an autoinflammatory condition that appears in early childhood.5 Individuals with a milder form of MK deficiency (MKD), also known as hyper-immunoglobulin D (IgD) syndrome with periodic fever (HIDS, OMIM 260920, with 1%–20% residual enzyme activity), have recurrent episodes of high fever, aphthous ulcers, hepatosplenomegaly, lymphadenopathy, skin rashes, joint pain, and abdominal inflammation, while more severe loss of MK (<1% residual activity) leads to mevalonic aciduria (OMIM 610377) with additional features, including facial dysmorphia, psychomotor symptoms, and intellectual disability.6 The mechanisms linking impaired protein prenylation in MKD and inflammation remain unclear. It has long been thought that the underlying mechanism is enhanced production of pro-inflammatory interleukin (IL)-1β from monocytes/macrophages. This was based on early observations that peripheral blood mononuclear cells (PBMCs) from MKD patients produce excessive quantities of IL-1β (reviewed by Politiek and Waterham7) and that acute inhibition of protein prenylation with pharmacological agents enhances inflammasome activation in macrophages.8–10 However, in contrast to other inflammasome-driven autoinflammatory diseases, MKD remains particularly challenging to treat,11,12 with a high proportion of patients partially or completely unresponsive to treatment with IL-1β-inhibiting drugs such as anakinra.11–13 The symptoms of severe inflammatory flares in some MKD patients resemble secondary hemophagocytic lymphohistiocytosis (HLH)/macrophage activation syndrome (MAS),14–18 a hyperinflammatory state with excessive activation of T lymphocytes and macrophages19 that can occur in response to infection in patients with autoinflammatory disorders, including MKD.20 Recurrent infections, particularly of the respiratory tract,21–23 are a common feature in MKD and often trigger inflammatory flares and potentially life-threatening complications akin to HLH/MAS.15,24,25 MKD therefore appears to be a multi-cytokine disorder,26,27 driven by a complex and dysregulated immune response.28 To better understand the mechanisms underpinning inflammatory disease in MKD, we recently engineered Mvk mutant mice by introducing bi-allelic variants into the mouse genome.29 These Mvk mutant mice recapitulate the underlying metabolic defect in protein prenylation, the severity of which is related to the residual activity of the MK enzyme.29 Studying these and other Mvk mutant mouse lines, together with patient samples, led us to discover a deficiency in natural killer (NK) cells in mice and humans with MKD. By linking insufficient protein prenylation to dysregulation of cytotoxic cells, defective degranulation, and IFN-γ-associated hyperinflammation, our findings revealed an alternative paradigm for the underlying mechanism of disease pathogenesis, with important treatment implications for individuals with this debilitating and sometimes fatal condition. RESULTS Mice and humans with MKD are deficient in mature NK cells To investigate how disruption of the mevalonate pathway (Figure 1A) alters immune cell populations in vivo, we performed single-cell RNA sequencing (scRNA-seq) on PBMCs from MvkV377I/Δ91 mice that have defective protein prenylation that is similar to MKD patients but do not show signs of inflammation in the steady state under normal housing conditions.29 Apart from a small increase in the proportion of B cells, the major difference between MvkV377I/Δ91 and wild-type (WT) mice was a ∼60% decrease in the proportion of NK cells in mutant mice (Figures S1A–S1C), confirmed by flow cytometry as a profound reduction in blood NK1.1+ NK cells as a proportion of total PBMC (Figure 1B). We then extended the analysis to two lines of Mvk mutant mice that were either homozygous for a p.T322M mutation or compound heterozygous for p.T322M and a 91 bp deletion in exon 11 (Δ91).29 The p.T322M mutation, similar to a likely pathogenic variant p.T322S described in an MKD patient,6 was predicted to be damaging (PolyPhen2 score = 1.0) since it affects a highly conserved residue in an α-helical region of the dimerization interface of the MK enzyme.30,31 Major loss of function was confirmed by measurement of MK enzyme activity in the liver (approximately 4% and 2% residual enzyme activity in MvkT322M/T322M and MvkT322M/Δ91 mice, respectively; see Figure S1D). MvkT322M/T322M mice resembled MvkV377I/Δ91 (B) Representative plots and frequencies of murine blood NK cells (gated as B220− TCRb− NK1.1+ live singlets) as a percentage of total live cells. (C) Detection of unprenylated Rab GTPases (uRabs) in splenocyte lysates (data are representative of >5 independent experiments). (D) Plasma mevalonic acid concentration, measured by mass spectrometry (data are from a single analysis, and each symbol is an individual mouse). (E) Flow cytometric analysis of NK cells as a percentage of total live splenocytes from WT (white symbols) and Mvk+/T322M (blue) controls, compared with 3 lines of bi-allelic Mvk mutant mice (red symbols). (F) Representative fluorescence-activated cell sorting (FACS) plots of NK cell subsets in spleen in Mvk+/T322M and MvkT322M/Δ91 mice. Frequencies correspond to the percentage of total NK cells. (G) NK cell subsets in bone marrow, blood, and spleen from MvkT322M/Δ91 and control Mvk+/T322M mice, expressed as a percentage of total NK cells (mean ± SD, n = 5 mice per genotype). (H and I) Splenic NK cell subsets expressed as frequencies of total live cells and total numbers, respectively. (J) Representative FACS plots (left) and frequencies (right) of CD57+ NK cells (live CD56+ lymphocyte singlets) in PBMC from healthy controls and MKD patients (single experiment). (K) Representative FACS plots (left) and frequencies (right) of MitoSOX+ NK1.1+ splenic NK cells from Mvk+/T322M and MvkT322M/Δ91 mice. (L) Representative FACS plots (left) and frequencies (right) of MitoSOX+ CD56+ human NK cells from MKD patients and healthy controls (data pooled from 2 independent experiments). All graphs show mean ± SD, and each symbol is an individual mouse or human subject. Statistical analysis: (B, D, E, H, and I) Brown-Forsythe and Welch ANOVA/ Dunnett’s multiple comparisons test and (J and K) Welch t test (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001). Data shown in (B), (C), (E)–(I), and (K) are representative of at least 3 independent experiments. See also Figures S1–S3. Immunity 59, 1–17, June 9, 2026 3 mice in the extent of defective Rab GTPase prenylation and buildup of plasma mevalonic acid, whereas MvkT322M/Δ91 mice had the most severe defect in prenylation and most elevated plasma mevalonic acid (Figures 1C and 1D). Heterozygous mice, with about 50% residual MK activity29 (see Figure S1D), had no defect in prenylation or any buildup of mevalonic acid (Figures 1C and 1D). Importantly, all three genotypes of Mvk mutant mice (MvkV377I/Δ91, MvkT322M/T322M, and MvkT322M/Δ91) had significant decreases in the frequencies of splenic NK cells (Figure 1E). There was little difference in gene expression in the monocyte population in WT and MvkV377I/Δ91 mice (Figure S1E). As well as a reduction in NK cell frequency, scRNA-seq analysis revealed increased expression of genes associated with immature NK cells,32 whereas genes associated with NK cell maturity were decreased in MvkV377I/Δ91 mice (Figure S1F). Ly6c2 and Klra9 (encoding Ly49I), associated with NK cell maturation,32 were the two most markedly reduced genes expressed in circulating MvkV377I/Δ91 NK cells (Figure S1G). These transcriptomic findings were validated by flow cytometry, which showed reduced frequencies of Ly6C+ and Ly49I+ NK cells in the peripheral blood of MvkV377I/Δ91 mice compared with controls (Figure S1H). We then used differential expression of CD27 and CD11b to distinguish functionally immature (CD27+CD11b− ), intermediate (CD27+CD11b+), and mature (CD27− CD11b+) NK cells.33,34 When the NK cell pool was examined, Mvk mutant mice had a higher frequency of immature CD27+CD11b− and a lower frequency of mature CD27− CD11b+ NK cells (Figures 1F and 1G). These differences were most evident in the blood and spleen, where CD27− CD11b+ mature NK cells are the predominant subset, compared with bone marrow (Figure 1G). The frequency and total number of mature, CD27− CD11b+ NK cells were significantly lower in the spleens of all three Mvk mutant genotypes compared with WT or heterozygous controls, indicating impaired NK cell maturation (Figures 1H and 1I). Consistent with this, intermediate CD27+CD11b+ Mvk mutant NK cells were also reduced while the immature, CD27+CD11b− NK cell population remained unchanged (Figures 1H and 1I). Reconstitution of irradiated MvkV377I/Δ91 mice with WT bone marrow resulted in normal proportions of splenic NK cells and other immune cells (Figures S2A and S2B). However, WT recipients of Mvk mutant bone marrow showed significant reductions in the frequencies of total and mature (CD27− CD11b+) NK cells in the spleen (Figure S2B), accompanied by increases in immature (CD27+CD11b− ) and, to a lesser extent, intermediate (CD27+ CD11b+) NK cells (Figure S2B). Similar changes were observed in the spleens of mixed bone marrow chimeras generated by reconstitution of WT recipient mice with a mixture of WT and Mvk mutant bone marrow (Figures S2C–S2E). Together, these findings indicate a cell-intrinsic defect in the maturation and/or survival of Mvk mutant NK cells. We next determined whether MKD patients also have altered NK cells. In a cohort of MKD patients (genotypes shown in Table S1) and healthy controls, although the frequencies of CD56dim or CD56hi NK cells (and other immune cell populations) did not differ (Figures S3A–S3F), we found a significant reduction in the proportion of the most mature subset of NK cells, i.e., CD56dimCD57+ cells33,34 (Figure 1J), in patients compared with controls. To extend these findings, we assessed the frequencies of mature NK cells based on differential expression of CD56, CD16, and CD57 in a separate cohort of MKD patients and again found a clear reduction in the proportion of CD56+CD16+CD57+ NK cells in MKD patients compared with healthy controls (Figure S3G). Since terminal maturation of NK cells is affected by oxidative stress,35 we measured mitochondrial superoxide using MitoSOX and found a significant increase in the percentage of MitoSOX+ NK cells in the spleen of MvkT322M/Δ91 mice compared with controls (Figure 1K). Similarly, MitoSOX staining was elevated in NK cells from 2 MKD patients (MVKV377I/H380R and MVKV377I/C275R) compared with healthy donors (Figure 1L). Hence, there is a cell-intrinsic defect in the maturation and/or fitness of NK cells in mice and humans with MKD. Cytolytic granules are abnormal in NK cells in mice and humans with MKD Given that prenylated small GTPases, including Rab proteins, are required for the formation and trafficking of cytotoxic granules,36–39 we next examined the features of granules in NK cells from mice and humans with MKD. NK cells isolated from Mvk mutant mice had a profound accumulation of unprenylated Rab GTPases as well as unprenylated Rap1a, which was absent in WT or Mvk+/T322M NK cells (Figure 2A). Functional studies were then conducted using Mvk+/T322M and MvkT322M/Δ91 NK cells expanded ex vivo. Although the composition of NK cell subsets (based on CD11b and CD27 expression) differed in vivo between genotypes, NK cells expanded ex vivo with IL-2/IL-15 had similar expression of CD11b, CD27, or NK1.1 after 4 days in culture (Figure S4A). Unstimulated MvkT322M/Δ91 NK cells were similar in size (forward scatter) to control Mvk+/T322M NK cells but had significantly greater granularity (side scatter) (Figure S4B). Fluorescence confocal imaging revealed that there were no differences in average granule volume and total cell volume between genotypes. However, the number of granzyme B (GzmB)-positive granules per cell (and hence total granule volume per cell) was significantly increased in MvkT322M/Δ91 NK cells compared with Mvk+/T322M NK cells (Figures 2B and 2C). Furthermore, the granules were significantly more dispersed in MvkT322M/Δ91 NK cells than in controls (Figure 2C). We used dual pipette aspiration and concomitant live imaging of single NK/YAC-1 target cell interactions40 to follow granule dynamics during a 10 min period (Figure 2D; see also Figure S4C; Video S1). Approximately 75% of control Mvk+/T322M NK cells contracted upon target engagement (Figure S4D) and showed a corresponding decrease in tail volume (Figure 2E) that reflects an increase in cell tension and hydrostatic pressure.41 As the NK cell engaged with the target cell and contracted, the granules converged to the microtubule-organizing center (MTOC) and mobilized toward the cytotoxic immunological synapse (CIS) (Figure 2D), as evidenced by the progressive reduction in the distances to the interaction interface (Figure 2E). However, only 25% of MvkT322M/Δ91 NK cells showed evidence of cell contraction or granule polarization to the CIS, even though MTOC formation was not impaired (see Figures S4D and S4E). Consistent with these findings, MvkT322M/Δ91 NK cells displayed minimal, if any, change in mean tail volume or in the mean distance between the MTOC 4 Immunity 59, 1–17, June 9, 2026 (A) Detection of unprenylated Rab GTPases (uRabs) and unprenylated Rap1a (uRap1a) in purified, splenic NK cells (single experiment, cells pooled from 3 mice per genotype). (legend continued on next page) Immunity 59, 1–17, June 9, 2026 5 or granules to the CIS during the imaging period, in clear contrast to their Mvk+/T322M counterparts (Figure 2E). We next used the same live cell imaging approach to examine NK cells purified from two adult MKD patients (MVKV377I/C275R and MVKV377I/H380R) with defective protein prenylation in PBMCs (Figure 2F). NK cells from healthy donors and both MKD patients consistently engaged with K562 target cells (Figure 2G; see also Figures S4F and S4G). However, despite no difference in the ability to form an MTOC (Figure S4E), <20% of the patient NK cells showed evidence of contraction and granule polarization, compared with ∼75% of NK cells from healthy donors (Figure 2H). Moreover, during interaction with the target, healthy donor NK cells displayed a consistent reduction in tail volume, and the cytotoxic granules polarized toward the CIS, evidenced by a decrease in the mean distance of granules from the interaction interface (Figure 2I). This was not the case for NK cells from both MKD patients, where the mean tail volume remained unchanged, and the granules remained scattered and failed to polarize toward the interface (Figure 2I; see also Video S2). Together, these findings demonstrate that the trafficking of cytotoxic granules is defective in NK cells from mice and humans with MKD. Defective prenylation in MKD reduces NK cell cytotoxicity and increases IFN-γ production from NK and CD8+ T cells Consistent with abnormal trafficking of cytotoxic granules (Figure 2), NK cells from MvkT322M/Δ91 mice exhibited dysregulated targeted degranulation, with a higher frequency of cells expressing the degranulation marker CD107a in the absence or presence of YAC-1 target cells (Figure 3A). MvkT322M/Δ91 NK cells also had a ∼40% reduction in directed cytotoxicity toward YAC-1 target cells (Figure 3B). Importantly, the capacity of NK cells from two MKD patients (8-year-old male compound heterozygous for p.V377I and c.371+2 T>C; 16-year-old female homozygous for p.V377I) to lyse K562 target cells was also reduced (∼30%) compared with NK cells from healthy donors (Figure 3C). Murine MvkT322M/Δ91 NK cells stimulated by coculture with YAC-1 target cells produced higher concentrations of IFN-γ, as evidenced by higher proportions of IFN-γ+ NK cells compared with controls (Figure 3D). Secretion of tumor necrosis factor α (TNF-α) and granulocyte-macrophage colony-stimulating factor (GM-CSF), as well as IFN-γ, into the culture medium was also significantly higher from stimulated MvkT322M/Δ91 NK cells compared with controls (Figure 3E). Similar results were obtained with human NK-92 cells pre-treated with inhibitors that prevent the prenylation of either Rho/Rac/Cdc42/Rap-family GTPases (GGTI-298), Rab-family GTPases (NE10790), or both families of small GTPases (lovastatin).42 Pre-treatment of NK92 cells with each of the three inhibitors prevented protein prenylation according to their target specificity (Figure 3F). Importantly, each of the prenylation inhibitors significantly decreased the cytotoxicity of NK-92 cells toward K562 target cells (Figure 3G). To examine the effect on NK cells of immune activation other than by target cell interaction, we treated mice with C–phosphate–G (CpG) or the Infanrix vaccine. Treatment with a single dose of CpG resulted in a significantly higher frequency of IFNγ+ NK cells in MvkT322M/Δ91 mice compared with controls (Figure 3H). Furthermore, serum IFN-γ was undetectable in untreated mice but was elevated in MvkT322M/Δ91 mice compared with controls following CpG treatment (p = 0.052) (Figure 3I). Similarly, the frequency of IFN-γ+ NK cells in vivo was significantly higher in MvkT322M/Δ91 mice compared with Mvk+/T322M mice 24 h after treatment with a single dose of Infanrix (diphtheria, tetanus, and pertussis vaccine), with increased concentrations of serum IFN-γ (p = 0.071) (Figure 3J). Similar to NK cells (Figure 2A), freshly isolated CD8+ T cells from MvkT322M/Δ91 mice had a clear accumulation of unprenylated Rab GTPases as well as unprenylated Rap1a (Figure 4A). CD44+CD8+ effector T cells from MvkT322M/Δ91 mice were also significantly more granular than controls (Figure 4B), showed a significantly higher proportion with increased cell surface CD107a (Figure 4C), and also had significantly higher MFI for IFN-γ after PMA/ionomycin stimulation, compared with Mvk+/T322M CD8+ T cells (Figure 4D). Hence, the production of IFN-γ from activated NK and CD8+ T cells is increased in mice with MKD. Mice with MKD exhibit systemic inflammation and impaired virus control Since NK cells are key mediators of anti-viral immunity, we compared the ability of MvkT322M/Δ91 and control Mvk+/T322M mice to respond to murine cytomegalovirus (MCMV) infection. (B) Left: fluorescence confocal microscopy images of fixed Mvk+/T322M (top) and MvkT322M/Δ91 (bottom) NK cells, showing plasma membrane (cyan) and GzmBpositive cytotoxic granules (red). Right: rendered images show individual granules within each cell. (C) Violin plots showing measurements of granules from individual NK cells. Data (and representative images in B) were compiled from a total of 45–60 NK cells from 3 mice per genotype (*p < 0.05, **p < 0.01, ****p < 0.0001, Mann-Whitney t test). (D) Representative time-lapse rendered images (Imaris) of live NK cell/target cell interactions using a dual pipette aspiration assay. Data were collected during 10 min from the time of initial contact. NK cell (cyan), YAC-1 target cell (magenta), cytotoxic granules (yellow), and microtubule-organizing center (red). Scale bars, 5 μm. (E) Graphs show changes in NK cells during target cell interaction. Values (and representative images in D) were compiled from single NK/target cell interaction events from a total of 19 Mvk+/T322M (blue) and 29 MvkT322M/Δ91 (red) cells from 3 mice per genotype. Each interaction was normalized to time 0. (F) Defective Rab prenylation in PBMC isolated from an MKD patient (genotype MVKV377I/C275R). Representative of 3 independent analyses. (G) Representative time-lapse rendered images (Imaris) of a live, human NK cell from the same patient as in (F) and a healthy control. Images were captured during a 10-min NK/target cell interaction. NK cell (cyan), K562 target cell (magenta), cytotoxic granules (yellow). Scale bars, 5 μm. (H) Percentage of human NK cells that showed tail contraction or granule polarization to the cell interface. (I) Changes in NK cell tail volume and mean distance of cytotoxic granules to the cell interface. Graphs show measurements from single NK cells relative to the initial time of contact for each interaction (time 0). Data in (H) and (I) (and representative images in G) were compiled from 15 single cells from two MKD patients (MVKV377I/C275R or MVKV377I/H380R) and 15 single cells from two healthy controls. See also Figure S4 and Videos S1 and S2. 6 Immunity 59, 1–17, June 9, 2026 A D +/T 32 2M T3 22 M/ ∆9 1 J unstimulated 0 10 20 30 40 % IF N γ+ N K ce lls 0 10 20 30 40 50 Se ru m IF N γ ( pg /m L) p=0.071 +/T 32 2M T3 22 M/ ∆9 1 +/T 32 2M T3 22 M/ ∆9 1 ** +Infanrix +Infanrix 1:1 2:1 4:1 0 20 40 60 Control MKD 1 MKD 2 NK:Target cell ratio C % ta rg et c el l k illi ng b y hu m an N K ce lls 1:1 2:1 5:1 NK:Target cell ratio % ta rg et c el l k illi ng b y m ur in e N K ce lls *** 0 20 40 60 80 100 +/T322M T322M/∆91 B % C D 10 7a + N K ce lls +/T 32 2M T3 22 M/ ∆9 1 +/T 32 2M T3 22 M/ ∆9 1 +YAC-1unstimulated 0 20 40 60 80 100 * * +/T322M T322M/∆91 CD107a N or m al is ed to m od e E 3000 0 1000 2000 IF N γ (p g/ m L) 0 50 100 150 200 TN Fα (p g/ m L) 0 50 100 150 200 250 G M -C SF (p g/ m L) +/T 32 2M T3 22 M/ ∆9 1 +YAC-1 +/T 32 2M T3 22 M/ ∆9 1 unstimulated +/T 32 2M T3 22 M/ ∆9 1 +YAC-1 +/T 32 2M T3 22 M/ ∆9 1 unstimulated +/T 32 2M T3 22 M/ ∆9 1 +YAC-1 25 +/T 32 2M T3 22 M/ ∆9 1 +/T 32 2M T3 22 M/ ∆9 1 +YAC-1unstimulated 0 5 10 15 20 % IF N γ+ N K ce lls * N K92 c yt ot ox ic ity (r el at iv e to c on tro l) con tro l +lo va sta tin +G GT I-2 98 +N E1 07 90 0.0 0.5 1.0 1.5 **** *** *F G ** ** *** p=0.052H % IF N γ+ N K ce lls 0 10 20 30 40 50 ** +/T 32 2M T3 22 M/ ∆9 1 +CpG 0 103 104 105 0 20 40 60 80 100 N or m al iz ed to m od e IFNγ T322M/∆91 +/T322M Se ru m IF N γ (p g/ m L) +/T 32 2M T3 22 M/ ∆9 1 baseline 0 5000 10000 15000 20000 +/T 32 2M T3 22 M/ ∆9 1 +CpG I 21kDa 75kDa 25kDa uRabs loading control } uRap1a con trol +lo vas tati n +G GT I-29 8 +NE 107 90 Figure 3. NK cells with defective prenylation have decreased cytotoxicity and increased IFN-γ production (A) Left: representative flow cytometry histogram showing CD107a fluorescence intensity in unstimulated murine NK cells (blue = Mvk+/T322M, red = MvkT322M/Δ91). Right: percentage CD107a+ NK cells in the absence or presence of stimulation with YAC-1 cells. (B) Percentage YAC-1 target cell lysis by mouse NK cells, with increasing NK/target cell ratio. Data in (A) and (B) are representative of 3 independent experiments. (legend continued on next page) Immunity 59, 1–17, June 9, 2026 7 MCMV infection is characterized by the maturation and expansion of splenic NK cells,43 the predominant subset expressing the activating receptor Ly49H, which functions as a receptor for the MCMV m157 glycoprotein.43,44 The total number of splenic NK cells, in particular mature CD27− CD11b+ NK cells, remained significantly lower in MvkT322M/Δ91 mice compared with controls even after infection (Figures 5A and S5A). Consistent with reduced NK cell numbers in MvkT322M/Δ91 mice, the number of Ly49H+ NK cells was also significantly lower at steady state, and there was no expansion of Ly49H+ NK cells after MCMV infection in MvkT322M/Δ91 mice compared with controls (Figure 5B). The expansion of CD8+ T cells, including M45+ anti-viral CD8+ T cells, was also drastically reduced in MCMV-infected MvkT322M/Δ91 mice (Figure 5C). When viral loads were examined, MvkT322M/Δ91 mice showed significantly higher viral titers in spleen, liver, and lung than in control Mvk+/T322M mice (Figure 5D), consistent with an impaired capacity to control MCMV infection due to defects in NK and CD8+ T cell responses. Higher frequencies of GzmB+ NK cells were observed in infected MvkT322M/Δ91 mice compared with controls (Figure 5E), suggesting NK cell hyperactivation as well as dysfunction. Similarly, effector (CD44+) CD8+ T cells in infected MvkT322M/Δ91 mice, although reduced in number, also showed significantly higher frequencies of GzmB+ cells compared with infected controls (Figure 5F). After infection, significantly higher concentrations of serum IFN-γ and other pro-inflammatory cytokines and chemokines, including IL-6, CCL4, and IL-18 (Figures 5G and S5B), were also observed in MvkT322M/Δ91 mice compared with controls. Although pharmacological inhibition of protein prenylation with statins can enhance IFN-γ production in cultured monocytes,45 in our studies the proportion of IFN-γ+ monocytes was low after MCMV infection and did not differ between genotypes (Figure S5C). Thus, monocytes were unlikely to be the source of elevated serum IFN-γ in Mvk mutant mice. Histopathology analysis revealed significant liver disease, reflected in higher liver pathology scores, in infected MvkT322M/Δ91 mice compared with infected control mice (Figures 5H and 5I). In addition to increased numbers of virally infected cells with cytoplasmic and intranuclear inclusions, infected MvkT322M/Δ91 mice had more periportal and parenchymal infiltration of inflammatory cells and notably displayed significant areas of hepatocellular necrosis, cytologic degeneration, and non-focal hemorrhage (Figure 5I). Flow cytometric analysis of liver leukocytes demonstrated significantly increased neutrophil and monocyte infiltration in infected MvkT322M/Δ91 mice (Figure 5J) despite reduced numbers of NK cells, CD8+ T cells, and innate lymphoid cells type 1 (ILC1) cells (Figure S5D). Therefore, the pathology observed in the livers of infected MvkT322M/Δ91 mice was consistent with immune-mediated damage caused by unresolved viral infection accompanied by a sustained inflammatory response. JAK inhibition reduced clinical and biochemical flares in a patient with severe MKD Our studies in mice suggested that altered maturation and dysfunction of NK cells in MKD lead to dysregulated IFN-γ production and a pathogenic systemic inflammatory response following viral infection. Measurements of serum from 5 MKD patients also showed significantly elevated concentrations of IFN-γ and CXCL10 (an IFN-γ-induced chemokine) compared with healthy controls (Figure 6A). These findings suggest that reducing the activity of IFN-γ with a Janus kinase inhibitor (JAKi) could provide therapeutic benefit. JAKi are becoming established as a treatment to reduce IFN signaling in certain autoinflammatory disorders,46–49 but there are currently no data on the use of these agents in MKD.50 In light of this, we treated a male infant with severe MKD (mevalonic aciduria) with a JAKi. Born at 30+3 weeks of gestation with a birth weight of 1,600 g, the antenatal period was complicated by fetal hepatosplenomegaly and transfusion-dependent anemia and thrombocytopenia. He soon became critically unwell with fever and cardiorespiratory failure. First-line investigations showed a hyperinflammatory response with signs of HLH. Urine metabolic testing was consistent with mevalonic aciduria, and exome sequencing confirmed a diagnosis of MKD (MVK c.709A>T; p.T237S/c.1139A>G; p.H380R). Analysis of the patient’s PBMC revealed a clear buildup of unprenylated Rab proteins compared with PBMC from either heterozygous parent or from a healthy volunteer (Figure 6B). Despite improvement with corticosteroids and high-dose anakinra (10 mg/kg/day; Figure 6C), the child experienced persistent irritability, abdominal pain, loose stools, and feed intolerance. Colonoscopy demonstrated moderately friable (C) Percentage K562 target cell lysis by human NK cells from 1 healthy volunteer (blue) and 2 MKD patients (red and pink), with increasing NK/target cell ratio (single experiment). (D) Percentage of IFN-γ+ murine NK cells following coculture for 3 h with YAC-1 target cells (representative of 3 independent experiments). (E) Concentration of IFN-γ, TNF-α, and GM-CSF in culture supernatant, following coculture of NK cells for 3 h with YAC-1 target cells. Data show mean ± SD of technical replicates from 1 mouse and are representative of 2 independent experiments. (F) Detection of unprenylated Rab GTPases (uRabs) and unprenylated Rap1a (uRap1a) in NK-92 cells following treatment for 24 h with lovastatin, GGTI-298, or NE10790 (representative of 2 independent experiments). (G) Cytotoxicity of NK-92 cells toward K562 target cells after pre-treatment of the NK-92 cells with the protein prenylation inhibitors lovastatin, GGTI-298, or NE10790. Cytotoxicity is expressed relative to vehicle-treated control NK-92 cells. Values show the mean ± SD of quadruplicate replicates from 1 experiment, representative of 3 independent experiments. (H) Histogram (left) showing fluorescence intensity of intracellular IFN-γ in NK cells (blue = Mvk+/T322M, red = MvkT322M/Δ91), and percentage of IFN-γ+ splenic NK cells (right), 24 h after intraperitoneal (i.p.) injection of CpG. (I) Concentration of serum IFN-γ in MvkT322M/Δ91 and control mice at baseline and 24 h after CpG treatment. Data in (H) and (I) are from a single experiment. (J) Percentage of IFN-γ+ NK cells in spleen (left) and serum IFN-γ (right) from Mvk+/T322M and MvkT322M/Δ91 mice, 24 h after i.p. injection of Infanrix vaccine (single experiment). All graphs show mean ± SD, and each symbol represents an individual mouse unless stated otherwise. Statistical analysis: (A and D) Brown-Forsythe and Welch ANOVA/Dunnett’s multiple comparisons test, (B) two-way ANOVA/Sidak’s multiple comparisons test, (E and H) Welch t test, (G) one-way ANOVA, and (I and J) Mann-Whitney t test (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001). 8 Immunity 59, 1–17, June 9, 2026 mucosa with areas of spontaneous bleeding consistent with very early onset inflammatory bowel disease. At 11 months of age, treatment with the JAKi baricitinib was commenced. Dosing was extrapolated from reports in other autoinflammatory diseases,51 titrated to a dose of 0.42 mg/kg/day split across three doses. Anakinra was continued in conjunction with baricitinib as a strategy used in other IFN-γ-dependent autoinflammatory diseases.52 After starting baricitinib, steroid dosing was cautiously reduced without exacerbating symptoms, and fewer clinical and biochemical flares were observed (Figure 6C). Follow-up colonoscopy was much improved from prior assessment, and biopsy showed no active inflammation. Treatment was complicated by influenza A pneumonitis for which non-invasive ventilation support was briefly used. No other adverse effects were noted. The child underwent a matched unrelated donor hematopoietic stem cell transplant (HSCT) at age 14 months. Baricitinib was discontinued shortly after transplant conditioning was completed. The early post-transplant period was largely uncomplicated, with engraftment noted on day 21. On day 42, he developed fevers and increased respiratory effort. Respiratory failure progressed, and the child passed away 52 days post- transplant, aged 16 months, due to presumed severe thrombotic microangiopathy. Even though this patient succumbed postHSCT, there were sufficient clinical indications to suggest that JAKi treatment ameliorated disease severity, at least during the time he was monitored. DISCUSSION Defective post-translational prenylation of small GTPase proteins is the most likely cause of systemic autoinflammatory disease in MKD.7 However, the molecular mechanisms involved remain poorly understood. Using bi-allelic Mvk mutant mouse models of MKD that recapitulate the underlying defect in protein prenylation in immune cells,29 we describe a cell-intrinsic deficiency in NK cell number, specifically the most mature subset of CD27− CD11b+ NK cells, possibly the result of increased cell death during NK cell proliferation and/or maturation. Similarly, we found a decrease in the frequency of the most terminally differentiated, CD57+ NK cells in MKD patients. This phenomenon has likely been overlooked until now because routine measurements of NK cells in the clinic are often based on singular cell counts of total NK cells,53 i.e., CD3− CD56+CD16+ cells. CD8+ T cellsA B C Immunity 59, 1–17, June 9, 2026 9 (legend continued on next page) 10 Immunity 59, 1–17, June 9, 2026 Lack of protein prenylation in MKD has been suggested to compromise the removal of damaged mitochondria by mitophagy,54,55 which, like autophagy, requires prenylated Rab and other small GTPases.56,57 Mitochondrial superoxide was elevated in NK cells from Mvk mutant mice and MKD patients. The terminal maturation of human NK cells is associated with a proliferative burst,58 during which they are particularly susceptible to oxidative stress.35 Consistent with this, a reduction in CD56dimCD57+ NK cells in lupus patients was associated with increased mitochondrial reactive oxygen species (ROS) production.59 The formation of memory-like NK cells after MCMV infection in mice is also dependent on mitophagy to promote cell survival.60,61 Defective protein prenylation in mice and humans with MKD may therefore hinder mitophagy, resulting in excessive production of mitochondrial ROS and increased NK cell death during maturation. However, we cannot exclude defects in other pathways involved in NK cell differentiation. In humans and mice, deficiency of myocyte enhancer factor 2C (MEF2C) also results in lower frequencies of mature NK cells.62 This effect was ascribed primarily to disruption of sterol regulatory elementbinding protein (SREBP)-dependent lipid homeostasis and metabolic reprogramming required for terminal NK cell maturation. However, the effect of decreased SREBP-dependent isoprenoid lipid biosynthesis on protein prenylation was not examined.62 NK cells from mice and humans with MKD displayed defective directed delivery of cytolytic granules and reduced cytotoxic activity. The formation, maturation, and trafficking of cytolytic granules in NK cells is highly dependent on numerous prenylated small GTPases,36 including Rab27a, Cdc42, RhoG, and the guanine nucleotide exchange factor RASGRP1.37,39,63,64 We confirmed that Mvk mutant NK cells had deficient prenylation of Rab GTPases, as well as Rap1a, and found that cytolytic granules in unstimulated NK cells were more numerous and highly dispersed and underwent increased nondirected degranulation. Despite MTOC formation, granules in Mvk mutant murine NK cells, as well as from two MKD patients, failed to polarize to the NK/target cell interface following cell conjugation. Thus, Mvk mutant NK cells shared common features with cytotoxic cells from pearl and gunmetal mice. Pearl and gunmetal are models of Hermansky-Pudlak syndrome, carrying genetic defects in other components of the vesicular trafficking pathway.65,66 Like Mvk mutant NK cells, pearl cytotoxic T cells have impaired movement of enlarged granules to the CIS.65 Furthermore, similar to Mvk mutant mice, gunmetal mice have defective prenylation of Rab GTPases, including Rab27a,67–69 and gunmetal cytotoxic cells have more dispersed granules that fail to polarize to the CIS.37 It is likely that dysregulation of cytolytic granules in MKD is caused by a lack of prenylation of a variety of Rab proteins, as well as other small GTPases such as Rap1a, Ras, and Cdc42. Consistent with this, pharmacologic inhibitors that selectively prevent the prenylation of either Rho/Rac/Cdc42/Rap1 GTPases (GGTI-298), Rab GTPases (NE10790), or both (lovastatin)42 reduced the cytotoxic activity of human NK-92 cells, similar to Mvk mutant NK cells. A p.R186C variant in CDC42 that causes autoinflammatory disease and HLH was also associated with a partial loss of NK cell cytotoxicity.70 Mutations in cytolytic granule components or the secretory machinery are among the genetic causes of familial HLH,19,53,71 leading to fulminant inflammation driven by excessive IFN-γ production from cytotoxic cells in response to viral infection.72,73 In C57BL/6 mice, effective control of the early stage of viral infection with MCMV is highly dependent on NK cell function.74 Consistent with a reduction in the number and cytotoxic capacity of NK cells, Mvk mutant mice were more sensitive to MCMV infection than control mice, maintained high viral loads, and exhibited exacerbated inflammatory pathology, including higher concentrations of serum IFN-γ. Inflammatory flares in MKD patients are often associated with respiratory infection21–23 and may occasionally progress to more serious HLH/MAS-like complications.15,24,25 Viral infection may therefore play a major, but hitherto overlooked, role in MKD pathology. The role of NK cells in autoinflammatory disorders, including MKD, is poorly understood.75 A previous case report described abnormal NK cell degranulation and MAS-like symptoms in an infant with MKD, although this patient was also heterozygous for a PRF1 variant.18 Jeyaratnam et al. described a decrease in cytotoxic activity of 30% in an infant with MKD that presented with HLH,17 very similar to the decrease in NK cell cytotoxicity that we observed in two MKD patients. Akin to familial HLH, poorly cytotoxic NK cells in Mvk mutant mice produced elevated amounts of the potent macrophage-activating factor IFN-γ, most likely as a secondary consequence of impaired cytotoxic function. Hypersecretion of IFN-γ is a characteristic feature of impaired cytotoxic cells,76 caused by mechanisms that include prolonged NK cell activation due to failed disengagement of the CIS77 and lack of termination of the immune response to a pathogen.78 Our studies with Mvk mutant NK cells provide another example of this phenomenon of increased IFN-γ production associated with decreased cytolytic activity. Flares in MKD are known to be associated with increased concentrations of serum IFN-γ as well as the IFN-γ-induced metabolite neopterin.79 Consistent with this, we found significantly increased IFN-γ and CXCL10 (an IFN-γ-induced chemokine) in MKD patient sera, and results from a clinical trial showed higher expression of type II and I interferon response genes in MKD patients.27 (F) Frequencies (left) and representative FACS plots from infected mice (right) of GzmB+ splenic effector CD44+CD8+ T cells. (G) Concentration of serum cytokines, determined by multiplex ELISA. (H) Liver pathology of infected MvkT322M/Δ91 and control Mvk+/T322M mice, scored as grade 0 (unremarkable), grade 1 (mild, 1–2 foci), grade 2 (mild to moderate, 3–6 foci), grade 3 (moderate, 7–12 foci), grade 4 (moderate-severe, multifocal), or grade 5 (severe, diffuse). (I) Liver histology (arrows indicate foci of mononuclear cell infiltration, and boxes indicate interstitial hemorrhage; scale bar, 100 μm). (J) Flow cytometry analysis of total numbers of liver monocytes (CD45+CD3− TCRb− CD19− NK1.1− F4/80− NK1.1− CD11b+CD115+) and neutrophils (CD45+ CD3− TCRb− CD19− NK1.1− F4/80− NK1.1− Ly6C+Ly6G+). Statistical analysis: (A–C, F, and J) Brown-Forsythe and Welch ANOVA/Dunnett’s multiple comparisons test, (E and G) Kruskal-Wallis ANOVA/Dunn’s multiple comparisons test, and (D and H) Mann-Whitney t test (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001). See also Figure S5. Immunity 59, 1–17, June 9, 2026 11 Together, these observations suggest that elevated IFN-γ is a consistent underlying feature of MKD. Antigens or adjuvants in vaccines can also promote NK cell activation via cytokines such as IL-12 and IL-18.80 Treatment of Mvk mutant mice with a single dose of Infanrix (diphtheria, tetanus, and pertussis toxin vaccine) or CpG oligodeoxynucleotide 1018 (a TLR9 agonist, which induces strong Th1 responses and is used as an adjuvant in several human vaccines)81 led to an increased proportion of IFN-γ+ NK cells in Mvk mutant mice compared with controls. Thus, a dysregulated NK cell response provides one explanation why inflammatory flares in individuals with MKD can be triggered by vaccination.82–85 Further studies are now required to understand the mechanisms of NK cell hyperactivation and IFN-γ production in vaccination-induced flares in patients. Although our studies focused on NK cells, Mvk mutant CD8+ T cells also had a clear defect in the prenylation of Rab GTPases 12 Immunity 59, 1–17, June 9, 2026 as well as Rap1a, a small GTPase that regulates T lymphocyte adhesion and homing86 and modulates the response of T cells to antigen presentation.87 Like NK cells, unstimulated CD8+ T cells from Mvk mutant mice were more granular and had increased cell surface CD107a and produced more IFN-γ upon ex vivo stimulation. Their expansion in vivo was also profoundly reduced following MCMV infection. This lack of proliferation could also be due to insufficient endogenous cholesterol biosynthesis via the mevalonate pathway during blastogenesis.88,89 Together, our observations suggest that both NK and CD8+ T cells are dysregulated in MKD, and this could potentially extend to other cytotoxic cells such as ILC1 and γ,δ T cells. Additional cell types could also play a role in NK/CD8+ T cell dysregulation, for example, by enhancing IFN-γ expression and release from NK cells via increased IL-12 and IL-18 production from prenylation-deficient monocytes.45,90 30 years ago, it was proposed that MKD monocytes and macrophages may be primed in vivo to produce a variety of pro-inflammatory cytokines.91 Our findings suggest that the classic macrophage-activating factor IFN-γ, a mediator of systemic autoinflammatory disease,92 together with enhanced production of other cytokines such as TNF-α and GM-CSF by prenylationdeficient NK cells, may be responsible for such priming. This may explain the difficulty in managing inflammatory symptoms using biologic agents that neutralize solely IL-1β activity, TNF-α, or IL-611–13 in some patients, particularly those with severe forms of MKD. A role of IFN-γ in MKD provides a rationale for testing the use of IFN-γ-neutralizing antibodies, such as emapalumab or JAKi (to block signaling by IFN-γ as well as other proinflammatory cytokines such as IL-6), in patients with severe, life-threatening forms of MKD (mevalonic aciduria) or in MKD patients that fail to respond to single biologic agents such as anakinra. We report that the JAKi baricitinib was well tolerated and reduced clinical and biochemical flares in an infant with severe MKD. Importantly, our observations may be extended to the emerging family of autoinflammatory disorders that we term ‘‘prenylopathies,’’ caused by loss-of-function variants in genes encoding other enzymes of the mevalonate pathway, such as phosphomevalonate kinase (PMVK).93–95 The extent to which protein prenylation is disrupted in PMVK deficiency compared with MKD remains to be determined. However, we predict that variants in PMVK or other mevalonate pathway genes that result in a similar defect in small GTPase prenylation will have NK cell dysfunction and elevated IFN-γ as an over-arching cause of disease in prenylopathies in general.


Limitations of the study
Although we revealed a decrease in the number and frequency of mature NK cells in mice with MKD, further work is required to determine whether this was due to altered proliferation and maturation of NK cells and/or increased cell death, and whether a similar mechanism could account for the decreased frequency of CD57+ NK cells in MKD patients. While the dual pipette aspiration approach to live cell imaging revealed abnormal cytotoxic granule trafficking in NK cells, the forced conjugation with target cells did not enable the analysis of potential differences in the formation of the CIS. Although we show that protein prenylation was impaired and IFN-γ production was increased in CD8+ T cells from Mvk mutant mice, further studies are necessary to define the cause of the profound lack of CD8+ T cell proliferation following MCMV infection and whether humans with MKD also exhibit dysregulated CD8+ T cells. Finally, further clinical studies are needed to assess the efficacy of JAKi, or other agents that block IFN-γ activity, as treatments to prevent systemic inflammation in MKD. RESOURCE AVAILABILITY Lead contact Requests for further information and resources should be directed to and will be fulfilled by the lead contact, Michael Rogers (m.rogers@garvan.org.au). Materials availability Mvk+/T322M mice may be obtained from Prof. Bruce Beutler via a materials transfer agreement. Data and code availability scRNA-seq data have been deposited at GEO and are publicly available as of the date of publication. The accession number is listed in the key resources table. Microscopy data reported in this paper will be shared by the lead contact upon request. This paper does not report original code. Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request. ACKNOWLEDGMENTS We sincerely thank the patients and volunteers who participated in this study. We also thank the staff of the Garvan Institute’s Biological Testing Facility, Clinical Research Facility, Garvan Molecular Genetics, Garvan Genomics Platform, and Dr. Eric Lam for technical support; Dr. Weng Hua Khoo for bioinformatics advice; Esther Kristianto and the VCCRI Innovation Centre’s Metabolomics Facility for mass spectrometry analysis; Pat O’Young for logistics and coordinating patient sample collection and transport; A/Prof. Neil Romberg for critical comments on the manuscript; and Prof. Jordan Orange for helpful discussion. We are grateful to Prof. Bruce Beutler and Prof. Christopher Goodnow for generously providing Mvk+/T322M mice. This work was funded in part by NHMRC grants 1139644 and 2036891 to M.J.R. and M.A.M. and 1113904 to E.N.M.-F. M.J.R. and M.A.M. were also supported by NHMRC Centres of Research Excellence grant 2035298 and by grants from the St Vincent’s Clinic Foundation, the Allergy and Immunology Foundation of Australasia, the David and Dulcie Henshall Foundation, the Marian & E.H. Flack Trust, the Mrs. Janice Gibson and Ernest Heine Family Foundation, and by family members and friends of Maddison Dupond. S.G.T., C.S.M., and M.A.D.-E. were supported by NHMRC Investigator grants 1176665/2034593, 2017463, and 2026377, respectively. O.P.S., C.A.-K., and Z.J.V. were supported through an Australian Government Research Training Program Scholarship. A.N.M. and L.D.G. were supported by the Kinghorn Foundation. N.L.F. and F.C.M. were supported by grants from MS Research Australia, the Trish Foundation, and MND Research Australia. CIRCA investigators (S.O., I.C., S.G.T., C.S.M., and M.J.R.) were supported by the John Brown Cook Foundation, CORIO Foundation, and Jeffrey Modell Foundation. I.A. and D.L.K. were supported by the Intramural Research Program of the National Institutes of Health (NIH). The contributions of the NIH authors are considered works of the United States Government. The findings and conclusions presented in this paper are those of the authors and do not necessarily reflect the views of the NIH or the U.S. Department of Health and Human Services. AUTHOR CONTRIBUTIONS Conceptualization, M.J.R., M.A.M., M.A.D.-E., and J.F.; funding acquisition, M.J.R., M.A.M., and M.A.D.-E.; investigation, M.A.M., I.S.S., J.C., E.N.M.-F., O.P.S., Z.J.V., C.A.-K., S.S.T., G.R., A.C.-M., L.v.d.C., A.N.M., X.L., N.L.F., W.D.R., and S.M.; methodology, M.B., D.K., R.C.C., L.D.G., F.W., F.C.M., R.A.B., S.G.T., C.S.M., J.v.d.H., and J.J.Z.; project administration, M.J.R.; resources, S.O., I.C., C.M.M.-M., I.A., D.L.K., S.G.T., J.v.d.H., and J.F.; supervision, M.J.R. and M.A.M.; writing – original draft, M.J.R., M.A.M., W.D.R., and Immunity 59, 1–17, June 9, 2026 13 S.M.; writing – review & editing, I.S.S., J.C., E.N.M.-F., I.A., S.G.T., C.S.M., J.v.d.H., J.F., and M.A.D.-E.


DECLARATION OF INTERESTS
The authors declare no competing interests. STAR★METHODS Detailed methods are provided in the online version of this paper and include the following: • KEY RESOURCES TABLE • EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS ○ Mvk mutant mice ○ Case study ○ Human PBMC, NK cells and sera ○ Cell lines • METHOD DETAILS ○ Single cell RNA sequencing ○ Isolation and culture of murine cells ○ Detection of unprenylated proteins ○ Flow cytometry ○ Confocal microscopy ○ Dual pipette aspiration assay ○ Image analysis ○ Degranulation and cytotoxicity assays ○ MCMV infection and Infanrix/CpG treatment • QUANTIFICATION AND STATISTICAL ANALYSIS SUPPLEMENTAL INFORMATION Supplemental information can be found online at https://doi.org/10.1016/j. immuni.2026.03.027. Received: June 26, 2025 Revised: December 30, 2025 Accepted: March 27, 2026


KEY RESOURCES TABLE
REAGENT or RESOURCE SOURCE IDENTIFIER Antibodies Rat anti-CD11b BUV395 BD Bioscience BD Biosciences Cat# 563553 RRID:AB_2738276 Rat anti-CD11b APC-Cy7 BD Bioscience BD Biosciences Cat# 557657, RRID:AB_396772 Rat anti-mouse CD11b eF450 BD Bioscience BD Biosciences Cat# 557396, RRID:AB_396679 Rat anti-mouse CD11b AF700 BD Bioscience Cat#557960; RRID:AB_396960 Rat anti-mouse CD11b BUV805 BD Bioscience Cat#741934; RRID:AB_2941960 Mouse anti-CD19 eF450 eBioscience Cat# 48-0193-82; RRID:AB_2734905 Rat anti-mouse CD19 BV786 BD Bioscience Cat#563333; RRID:AB_2738141 Rat anti-mouse MHCII (i-A/I-E) FITC BD Bioscience Cat#562009; RRID:AB_394958 Rat anti-mouse MHCII (i-A/I-E) BV650 BD Bioscience Cat#563415 RRID:AB_2738192 Mouse anti-mouse CD64 PE Biolegend Cat#139303; RRID:AB_10613467 Rat anti-mouse CD44 PE BD Biosciences Cat#553134; RRID:AB_394649 Rat anti-mouse Ly6C BV605 BD Biosciences Cat#563011; RRID:AB_2737949 Rat anti-mouse Ly6C BV570 Biolegend Cat#128030; RRID:AB_2562617 Rat anti-mouse Ly6C PE-Cy7 BD Biosciences Cat#560593; RRID:AB_1727557 Mouse anti-mouse NK1.1 APC-Cy7 BD Biosciences Cat#560618; RRID:AB_1727569 Mouse anti-mouse NK1.1 PE eBioscience Cat#108738; RRID:AB_2562216 Mouse anti-mouse NK1.1 BV421 BD Bioscience Cat#562921; RRID:AB_2728688 Mouse anti-mouse NK1.1 PerCP-Cy5.5 BD Biosciences Cat#551114; RRID:AB_394052 Mouse anti-mouse NK1.1 PE-Cy7 BD Bioscience Cat#552878; RRID:AB_2743793 Hamster anti-mouse CD27 PE-Cy7 BD Bioscience Cat#563604; RRID:AB_2738309 Hamster anti-mouse CD27 APC BD Bioscience Cat#560691; RRID:AB_1727455 Hamster anti-mouse/rat/human CD27 BV650 Biolegend Cat#124233; RRID:AB_2687192 Rat anti-mouse CD45R/B220 BUV737 BD Bioscience Cat#612838; RRID:AB_2870160 Rat anti-mouse CD45 APC-Cy7 BD Biosciences Cat#557659; RRID:AB_396774 Rat anti-mouse CD107a FITC BD Bioscience Cat#553793; RRID:AB_395057 Hamster anti-mouse TCRb APC-Cy7 BD Bioscience Cat#560656; RRID:AB_1727574 Hamster anti-mouse TCRb BUV737 BD Bioscience Cat#612821; RRID:AB_2870145 Hamster anti-mouse TCRb PerCP-Cy5.5 Biolegend Cat#109228; RRID:AB_1575173 Rat anti-mouse CD8a BUV395 BD Bioscience Cat#563786; RRID:AB_2732919 Rat anti-mouse CD8a BUV805 BD Bioscience Cat#612898; RRID:AB_2870186 Rat anti-mouse CD4 BV786 BD Bioscience Cat#563331; RRID:AB_2738140 Rat anti-mouse CD3e APC eFluor-780 eBioscience Cat#47-0032-82; RRID:AB_1272181 Hamster anti-mouse CD3e PerCP-Cy5.5 BD Bioscience Cat#551163; RRID:AB_394082 Mouse anti-human Granzyme B AF-647 BD Bioscience Cat#560121; RRID:AB_11154033 Mouse anti-human Granzyme B AF700 BD Biosciences Cat#560213; RRID:AB_1645453 Rat anti-mouse CD62L PerCP-Cy5.5 BD Bioscience Cat#560513; RRID:AB_1061158 Hamster anti-mouse CD49b BUV563 BD Bioscience Cat#741280; RRID:AB _2870819 Hamster anti-mouse CD49a BUV737 BD Bioscience Cat# 741776; RRID:AB _2871130 Mouse anti-mouse Ly49H BV480 BD Bioscience Cat#746493; RRID:AB _2743793 Mouse anti-mouse Ly49I-Biotin (YLI-90) ThermoFisher Scientific Cat#MA5-28667; RRID AB_2745626 PE-Streptavidin BD Bioscience Cat#554060 RRID:AB _10053373 (Continued on next page) e1 Immunity 59, 1–17.e1–e8, June 9, 2026


Continued
REAGENT or RESOURCE SOURCE IDENTIFIER Rat anti-mouse Ter119 PerCP-Cy5.5 Biolegend Cat#116228 RRID:AB _893636 PE-M45 tetramer (Phycoerythrin-conjugated tetramers for H-2Db-HGIRNASFI MCMV-M45) Brooks Laboratory, University of Melbourne N/A Rat anti-mouse Ly6G BV605 BD Bioscience Cat#563005; RRID:AB _2737946 Rat anti-mouse Ly6G APC BD Bioscience Cat#560599; RRID:AB _1727560 Hamster anti-mouse CD11c APC BD Bioscience Cat#550261; RRID:AB _398460 Rat anti-mouse IFNg FITC BD Bioscience Cat#554411; RRID:AB _395375 Rat anti-mouse IFNg PE-Cy7 eBioscience Cat#25-7311-82; RRID:AB _469680 Rat anti-mouse CD115 (CSF-1R) BUV395 BD Bioscience Cat#750886; RRID:AB _2874982 Rat anti-mouse F4/80 BV711 BD Bioscience Cat#565612; RRID:AB _2734769 Rat anti-mouse Siglec-H FITC eBioscience Cat#11-0333-82; RRID:AB _837163 Hamster anti-mouse CD49b BUV563 BD Bioscience Cat#741280; RRID:AB _2870819 Rat anti-mouse CD44 PE BD Bioscience Cat#553134; RRID:AB _394649 Purified rat anti-mouse CD16/CD32 (mouse BD Fc block) BD Bioscience Cat#553142; AB_394656 Mouse anti-human CD14 PerCP-Cy5.5 Biolegend Cat# 325621; RRID:AB _893252 Mouse anti-human CD14 BV711 BD Bioscience Cat# 563372; RRID:AB _2744290 Mouse anti-human CD14 APC-Cy7/APC-Fire750 Biolegend Cat# 302059; RRID:AB _2616706 Mouse anti-human CD19 BV750 BD Bioscience Cat# 747161; RRID:AB _2871897 Mouse anti-human CD3 Spark violet 500 Biolegend Cat#344869; RRID:AB _2936699 Mouse anti-human CD3 BV421 BD Bioscience Cat# 562426; RRID:AB _11152082 Mouse anti-human CD4 Spark blue574 Biolegend Cat# 344679; RRID:AB _2910397 Mouse anti-human CD4 BUV737 BD Bioscience Cat# 612748; RRID:AB _2870079 Mouse anti-human CD56 APC Biolegend Cat# 318309; RRID:AB _604098 Mouse anti-human CD56 APCvio770 Miltenyi Biotec Cat# 130-114-584; RRID:AB _2658740 Mouse anti-human CD57 APC BD Bioscience Cat# 560845; RRID:AB _10563760 Mouse anti-human CD57 BV605 Biolegend Cat# 393303; RRID:AB _2728425 Mouse anti-human CD8 BUV395 BD Bioscience Cat# 563795; RRID:AB _2722501 Mouse anti-human CD8 APC-Fire810 Biolegend Cat# 344763; RRID:AB _2860889 Mouse anti-human CD9 SparkNIR 685 Biolegend Cat# 302269; RRID:AB _2860769 Mouse anti-human HLA-DR PE/Fire 810 Biolegend Cat# 307683; RRID:AB _2904336 β-Actin (8H10D10) Mouse mAb Cell Signaling Cat#mAb #3700 RRID:AB _2242334 Goat anti-Rap1A Santa Cruz Biotechnologies Cat# sc-1482; RRID:AB_2177124 IRDye 680RD Streptavidin LICOR Cat#25-68079 Rabbit anti-gammaTubulin (EPR16793) Centrosome Marker Abcam Cat# ab179503; RRID:AB_2904198 Goat Anti-Rabbit IgG H&L (Alexa Fluor® 488) Abcam Cat# ab150077; RRID:AB_2630356 Biological samples PBMC samples from MKD patients Radboudumc Biobank of Autoinflammatory Disorders, Nijmegen, Netherlands https://www.radboudumc.nl/en/ research/technology-centers/ radboud-biobank PBMC samples from MKD patients Clinical Immunogenomics Research Consortium australasia (CIRCA) https://www.garvan.org.au/research/ collaboration/circa PBMC samples from MKD patients NIHGR, Bethesda, USA N/A Chemicals, peptides, and recombinant proteins Collagen Type I, rat tail Corning Cat#354236 Recombinant human IL-15 Miltenyi Cat#130-095-760 Recombinant human IL-2 Miltenyi Cat#170-076-146 Recombinant mouse IL-2 R&D Cat# 402-ML Recombinant Rab GGTase (GGTase II) Jena Bioscience Cat# PR-103 (Continued on next page) Immunity 59, 1–17.e1–e8, June 9, 2026 e2


Continued
REAGENT or RESOURCE SOURCE IDENTIFIER Recombinant REP-1-His Jena Bioscience Cat# PR-105 B-GPP Jena Bioscience LI-015 Lymphoprep Stemcell Catalog #18061 Ficoll-Paque Plus GE Healthcare Cat#17-1440-03 Cytiva Percoll Centrifugation Media Sigma-Aldrich/Cytiva Cat#GE17-0891-02 GolgiStop Protein Transport Inhibitor BD Bioscience Cat#554724 MitoSOX Red fluorescent stain Invitrogen Cat#M36008 Phalloidin Alexa Fluor Plus 555 ThermoFisher Scientific Cat#C10642 DAPI ThermoFisher Scientific Cat#62248 CellBrite Fix 488 Biotium Cat#30090 CellTracker™ Deep Red ThermoFisher Scientific Cat#C34565 CellTracker Orange (CMRA) Invitrogen Cat# C34551 Cell Trace Violet ThermoFisher Scientific Cat#C34571 Lysotracker Deep Red ThermoFisher Scientific Cat#L12492 Lysotracker Green DND-26 ThermoFisher Scientific Cat#L7526 CMFDA ThermoFisher Scientific Cat#C7025 CMTMR ThermoFisher Scientific Cat#C2927 propidium iodide ThermoFisher Scientific Cat# P3566 SPY555-tubulin Spirochrome SPY555-tubulin LEGENDplex Mouse Anti-Virus Response Panel BioLegend Cat#740621 Zombie Aqua Viability dye BioLegend Cat#423101 BD Horizon Fixable Viability Stain 440UV (FVS440UV) BD Bioscience Cat#566332; AB_2869748 Cytofix/Cytoperm™ Fixation/Permeabilization Kit BD Bioscience Cat# 554714; AB_2869008 Lovastatin Sapphire Bioscience Cat#TRC-L472225 GGTI-298 TFA salt Selleck Chemicals Cat#S7466 NE10790 Coxon et al.96 N/A Infanrix IPV GlaxoSmithKline AUST R 142370 Anakinra Swedish Orphan Biovitrum AUST R 082872 Baricitinib Eli Lilly & Co AUST R 277905 CpG ODN 1018 Vaccigrade Invivogen Cat#vac-1018-1 Bio-Plex Pro Mouse Cytokine 23-plex Bio-Rad Cat#M60009RDPD ProcartaPlex Human Inflammation Panel 20-plex ThermoFisher Cat#EPX200-12185-901 LEGENDplex Mouse Anti-Virus Response Panel BioLegend Cat#740621 Mouse CD8a+ T Cell isolation kit Miltenyi Cat#130-104-075 Mouse NK cell isolation kit Miltenyi Cat# 130-115-818 Deposited data Single-cell RNA-seq data This paper GEO: GSE298361 Experimental models: Cell lines Mouse: YAC-1 ATCC TIB-160; RRID:CVCL_2244 Human: NK-92 ATCC CRL-2407; RRID:CVCL_2142 Human: K562 ATCC CCL-243; RRID:CVCL_0004 Experimental models: Organisms/strains Mouse: C57BL/6JAusb Australian Bioresources N/A Mouse: C57BL/6JAusb.Mvk-T322M/Δ91 This paper N/A Mouse: C57BL/6J-MtgxR0206Btlr/Mmmh This paper R0206, MMRRC ID 38459 Mouse: C57BL/6JAusb.Mvk-T322M/T322M This paper N/A Mouse: C57BL/6JAusb.Mvk-V377I/Δ91 Australian Bioresources DOI: 10.1172/JCI160929 (Continued on next page) e3 Immunity 59, 1–17.e1–e8, June 9, 2026


EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS
Mvk mutant mice MvkV377I/Δ91 mice were generated as described previously.29 C57BL/6J mice heterozygous for an incidental missense mutation in Mvk (p.Thr322Met)98 were identified from an ENU mutagenesis screen99,100 (pedigree R0206, MMRRC ID 38459, strain C57BL/ 6J-MtgxR0206Btlr/Mmmh) and generously provided by Prof Bruce Beutler, UT Southwestern Medical Center, TX USA. Mvk+/T322M mice were rederived and backcrossed at least 10 times with C57BL/6J mice, and crossed with Mvk+/T322M or Mvk+/Δ91 mice29 to obtain homozygous MvkT322M/T322M and compound heterozygous MvkT322M/Δ91 mice, respectively. Unless specified, randomly allocated male and female mice were used for experiments, and sex-matched littermates or age- and sex-matched Mvk+/T322M and Mvk+/+ wildtype mice were used as controls, between 10-14 weeks old. Measurements of plasma mevalonic acid and mevalonate kinase activity in liver were performed as described in detail previously.29 To generate 100% bone marrow chimaeras, recipient female C57B6.SJL-PtprcaPepcb/BoyJArc mice, aged 8-9 weeks, were irradiated twice with 425 Rads (delivered from X-RAD 320 Biological Irradiator, PXI) administered 8 hours apart and injected intravenously with 5-10x106 bone marrow cells from a female MvkV377I/Δ91 donor. Conversely, recipient female MvkV377I/Δ91 mice were injected with bone marrow from a female C57B6.SJL-PtprcaPepcb/BoyJArc donor. To generate 50% bone marrow chimaeras, recipient female C57BL/6J mice were transplanted with bone marrow from donor Mvk+/+Ptprca mice in a 1:1 mixture with bone marrow from either a Mvk+/+Ptprcb or MvkV377I/Δ91Ptprcb female donor mouse. After 6 weeks, splenocytes were analysed by flow cytometry. Animals were bred and housed under standard SPF conditions with standard chow diet in Australian BioResources (ABR), the Garvan Institute Biological Testing Facility (BTF) and Monash Animal Research Platform (MARP). All animal experiments were performed with approval from the Garvan Institute/St Vincent’s Hospital (New South Wales) and the Monash (MARP4, Victoria) Animal Ethics Committees, in accordance with National Health & Medical Research Council guidelines.


Case study
The case study was a Caucasian male singleton born at 30+3 weeks gestation. He was the first child to supportive and well-resourced non-consanguineous parents. Consent to publish de-identified clinical details was obtained from both parents. Key medications included the following: 1) corticosteroids, initiated on day 10 of life - intravenous hydrocortisone, intravenous methylprednisolone and oral prednisolone used at different timepoints (doses used ranged from 0.5 to 2.5 mg/kg/day, equivalent prednisolone dose); corticosteroid dose was titrated to disease activity and limited by toxicity (growth failure, hypertension, hyperglycaemia, behavioural change, sleep disturbance, cataracts); 2) anakinra, initiated on day 10 of life – 100 mg / 0.67mL syringes used, starting dose of 0.8mg/kg twice daily by intermittent intravenous bolus dosing (method of administration extrapolated from other disease contexts101) and converted to subcutaneous dosing on day 157 of life, dose ranged from 1.6mg to 10 mg/kg/day, anakinra dose was titrated to disease activity and limited by toxicity (liver function test derangement); 3) baricitinib, initiated on day 343 of life - 2 mg film-coated tablets, split along the longest diameter if required, dispersed in 3 mL water and administered by nasogastric tube; the dose was increased every 4 days until the final dose was achieved as follows: 1 mg daily (∼0.14 mg/kg/day), then 1 mg twice daily (∼0.28 mg/kg/day), then 1 mg three times daily (∼0.42 mg/kg/day). Baricitinib doses were withheld during an influenza infection but no other toxicity was noted. Human PBMC, NK cells and sera Samples of cryopreserved human PBMC and serum were randomly selected from biobanks at RadboudUMC (Netherlands), NIH/ NIHGR (USA) or collected via Clinical Immunogenomics Research Consortium Australasia (CIRCA). Demographic information is shown in Table S1 (ancestry and ethnicity was not recorded). All MKD patients lacked fever or inflammatory symptoms at the time of blood collection. For live imaging of human NK cells, fresh blood samples were collected in lithium heparin tubes from healthy volunteers and from two adult females diagnosed with MKD and with bi-allelic variants in MVK. Patient 1: MVK c.1129G>A


Continued
REAGENT or RESOURCE SOURCE IDENTIFIER Software and algorithms FlowJo 10.6.2 or 10.8 BD Bioscience RRID:SCR_008520 GraphPad Prism Version 10.6.0 (statistical analysis software) GraphPad RRID:SCR_002798 Fiji (image processing and analysis software) Schindelin et al.97 RRID:SCR_002285; https://fiji.sc Image Studio v5.2.5 (western blot analysis software) LICOR RRID:SCR_013715 LEGENDplex (analyte analysis software) BioLegend https://www.biolegend.com/en-us/ immunoassays/legendplex/ support/software Imaris (image processing and analysis software) Bitplane AG RRID:SCR_007370 Immunity 59, 1–17.e1–e8, June 9, 2026 e4 (p.Val377Ile)/c.823T>C (p.Cys275Arg). Patient 2: MVK c.1129G>A (p.Val377Ile)/c.1139A>G (p.His380Arg). Buffy coat PBMC were prepared by centrifugation over Ficoll-Paque Plus (GE Healthcare) then NK cells were isolated using a human NK negative cell selection kit (Miltenyi) according to the manufacturer’s instructions. Purified NK cells were cultured overnight in a humidified incubator (5% CO2) at 37 oC in complete NK medium (NKM: Iscove’s Modified Dulbecco’s Medium, 0.58mg/ml L-glutamine, 25nM HEPES, 10% FBS, 100U/mL penicillin, 100μg/mL streptomycin, 50μM β-mercaptoethanol) supplemented with 50ng/mL recombinant human-IL-15 and 100ng/mL human-IL-2 (Miltenyi). IFNγ and CXCL10 were measured in serum samples using a ProcartaPlex 20-plex Human Inflammation Panel and Luminex platform. This study was approved by the Sydney Local Health District RPAH Zone Human Research Ethics Committee and Research Governance Office, Royal Prince Alfred Hospital, Camperdown, NSW, Australia (Protocols X16-0210/LNR/16/RPAH/257, X160210, 2019/ETH06359; and Protocol X20-0177 and 2020/ETH00998), by Sydney Children’s Hospital Network Human Research Ethics Committee (HREC/18/SCHN/403), and by the Committee on Human Research (CMO) Region Arnhem-Nijmegen (dossier number 2014-311) for PBMC samples obtained from Radboudumc Biobank of Autoinflammatory Disorders. All subjects and/or guardians gave written informed consent in accordance with the Declaration of Helsinki.


Cell lines
All cell lines were obtained from ATCC without further authentication. The human male NK cell line NK-92 was maintained in T-Cell Medium (TCM: RPMI; 10% heat-inactivated fetal calf serum; 10mM HEPES, 2mM L-glutamine, 1mM sodium pyruvate, 100U/mL penicillin, 100μg/mL streptomycin, 55μM β-mercaptoethanol) complemented with 100 U/mL human-IL-2 and maintained between 2x105 - 1x106 cells/mL by subculturing every 2-3 days. The human female K562 cell line was cultured in TCM for cytotoxicity assays with NK-92 cells, or complete NKM when used in assays with primary human NK cells. K562 cells were maintained at 105-106 cells/mL by subculturing every 3 days. The murine male YAC-1 cell line was cultured in NKM and maintained between 2x105 - 2x106 cells/mL by subculturing every 2-3 days. All cell lines were maintained in a humidified incubator at 37oC with 5% CO2.


METHOD DETAILS


Single cell RNA sequencing
10-week-old female Mvk+/+ and MvkV377I/Δ91 mice (n=3 per genotype) were retro-orbitally bled and blood was pooled into EDTAcoated tubes (BD Vacutainer, BD Biosciences). PBMCs were isolated by centrifugation over Ficoll-Paque Plus (GE Healthcare) and erythrocytes were lysed using 0.83% NH4Cl, 0.1% KHCO3 for 5 minutes at room temperature. Cells were stained with DAPI and anti-Ter119-PerCP/Cy5.5 (BioLegend), and double negative (DAPI-Ter119-) cells were isolated by using a BD FACSAria III Cell Sorter. Single cells were encapsulated into emulsion droplets using the 10x Chromium platform (10x Genomics, California), with a targeted output of 2000 cells. scRNAseq libraries were constructed using Chromium Single Cell 3.0 v2 Reagent Kit according to the manufacturer’s protocol. Amplified cDNA and final libraries were evaluated on an Agilent BioAnalyzer using a High Sensitivity DNA Kit (Agilent Technologies Inc.). Individual libraries were diluted to 4nM and pooled for sequencing, with 150 cycle run kits (26bp Read1, 8bp Index1 and 98bp Read2) on a Nextseq500 Sequencing System (Illumina) to 80-90% saturation. Cell Ranger (version 2.1.1) was used to map the 10x Genomics FASTQ files to the mouse genome reference database. Sequencing data was analysed using Seurat version 3.0 with R studio (Boston, Mass). Cells with fewer than 500 genes or greater than 2,500 genes were removed from downstream analysis, in addition to cells expressing >10% mitochondrial gene content. In total, 5,173 wildtype and 4,065 MvkV377I/Δ91 cells were considered for downstream analysis. Haemoglobin genes from contaminating erythrocytes were removed from the gene expression matrix. The ‘‘SCTransform’’ function of Seurat package and ‘‘percent.mito’’ genes regression were used to normalise and scale the data. Isolation and culture of murine cells Whole bone marrow was flushed with PBS (Gibco) from the femur and tibia. Peripheral blood mononuclear cells (PBMC) were isolated by centrifugation over Lymphoprep (Stemcell). Splenocytes were obtained by mashing the spleen through a 70μm nylon filter (Falcon) and lysing erythrocytes using 0.83% NH4Cl, 0.1% KHCO3. Livers were perfused with 10 mL cold PBS before being processed through metal sieves. The single cell suspension was then centrifuged in 37.5 % isotonic Percoll (Cytiva) at 690g for 20 min at room temperature and pelleted cells underwent erythrocyte lysis to obtain liver leukocytes. Splenic NK cells and CD8+ T cells were isolated using a murine NK cell isolation kit (Miltenyi) or negative selection CD8a+ T Cell isolation kit (Miltenyi) respectively, according to the manufacturer’s instructions. Purified splenic NK cells from individual mice were cultured in 12-well plates containing 1.5mL/well complete NKM with 100ng/mL mouse-IL-2 (R&D Systems) and 50ng/mL human-IL-15 (Miltenyi), and maintained in a humidified incubator (5% CO2) at 37 ◦C. On day 3, cells were transferred to T25 cell culture flasks containing 10mL fresh NKM. The expanded NK cells were used on day 4 or 5 post-isolation.


Detection of unprenylated proteins
Unprenylated Rab GTPases were detected in lysates of freshly-purified mouse splenic NK cells (pooled from 3 mice per genotype), splenic CD8+ T cells, NK-92 cells and human buffy coat PMBC, using an in vitro prenylation assay with recombinant Rab GGTase, REP-1 and a biotinylated analogue of geranyl pyrophosphate (Jena Bioscience), as previously described.3,4,29 In vitro-prenylated (and hence biotinylated) Rab proteins were then detected on PVDF membranes using streptavidin-680RD (LiCOR). Blots were e5 Immunity 59, 1–17.e1–e8, June 9, 2026 scanned on a LI-COR Odyssey imager and analysed using Image Studio v5.2.5. An endogenous biotinylated protein, approximately 75kDa, was used as a loading control.4 Unprenylated Rap1a was detected by western blotting with 14μg of cell lysate from purified NK or CD8+ T cells, using a goat polyclonal antibody (sc-1482, Santa Cruz; 1/1000 dilution), as previously described.102 β-actin was used as a loading control (mouse monoclonal antibody 8H10D10, Cell Signaling Technology; 1/5000 dilution).


Flow cytometry
Murine: cells from mouse bone marrow, spleen, liver and peripheral blood were collected as described above and cell viability and total cell numbers were assessed by trypan-blue staining using a Corning CytoSMART cell counter or an ABX Micros 60 cell counter (HORIBA). Cells were pre-incubated with mouse-Fc block (1:200) and viability dye (Zombie Aqua, BioLegend or FVS440UV, BD Bioscience) in calcium/magnesium-free PBS (Gibco) for 15 minutes, before staining with fluorescently-conjugated antibodies or PE-M45 tetramer (Brooks Laboratory, University of Melbourne; 1/100 dilution), prepared in FACS washing buffer (FW: 2mM EDTA, 0.02% azide; 0.5% fetal calf serum, calcium/magnesium-free PBS) for 30-45 minutes on ice. Samples were washed 3 times in FW and fixed with 10% formalin neutral buffered solution (Sigma), washed and resuspended in FW. For intracellular proteins (including GzmB, perforin and IFNγ), after staining for extracellular markers, cells were fixed and permeabilised using a BD Cytoperm/Cytofix Fixation/Permeabilization kit (BD Bioscience) according to the manufacturer’s instructions, incubated on ice for 1 hour with antibody mix and rinsed 3 times in FW before analysis. For mitochondrial superoxide measurement, splenic NK cells were purified as described above, then 106 NK cells/mL were stained with 500nM MitoSOX Red (Invitrogen) in Hanks Buffered Salt Solution (HBSS, Gibco) for 30 minutes at 37oC. After washing 3x in warm HBSS, cells were stained for flow cytometry as described above without fixation, using DAPI for dead cell exclusion. Samples were run on a BD LSRII SORP or FACSymphony A3 flow cytometer/DIVA, and post-acquisition analysis performed using FlowJo 10.6.2 or 10.8 (BD Bioscience). Human: flow cytometric analysis was performed on cryopreserved PBMCs from a total of 12 MKD patients (Table S1) and 13 ageand sex-matched healthy donors. Broad human immunophenotyping was applied to 1x106 PBMC using a custom designed 28-parameter panel as previously described.103 For measurement of mitochondrial superoxide, cryopreserved human PBMC were thawed and cultured overnight in complete NKM. Cells were then stained with MitoSOX Red (Invitrogen) as described for murine NK cells, and with antibodies to human CD3, CD19, CD8, CD56, CD16, CD57.


Confocal microscopy
Expanded murine NK cells were seeded into 96-well glass-bottom imaging plates (Greiner). Before staining, media were removed and cells fixed (200μL Fixation/Permeabilization kit; BD Biosciences) for 20 minutes at room temperature, rinsed twice with PBS, and incubated for 1 hour with blocking solution (Perm/Wash™ buffer with 1% BSA). Cells were stained in 200μL antibody mix (Perm/Wash™ buffer, 0.1% BSA) by overnight incubation at 4oC. For MTOC staining, 2x105 murine or human NK cells/well were gently centrifuged and allowed to recover at 37◦C and 5% CO2 for 10 minutes, fixed as described above and stained with consecutive overnight incubations with primary antibody (6.45μg/mL anti-γ-tubulin; Abcam) then secondary antibody (10μg/mL anti-rabbit Alexa Fluor® 488), followed by staining for 20 minutes with phalloidin (0.165μM Alexa Fluor Plus 555; ThermoFisher) and DAPI (300nM, ThermoFisher) solution in PBS. Samples were washed twice with PBS between each step. For cytotoxic granule staining, expanded NK cells were embedded in a collagen matrix (Corning) as previously described.104 Briefly, 3x105 NK cells were labelled with CellBrite Fix 488 (Biotium) for 15 minutes at 37◦C/5% CO2, washed twice with NKM and resuspended in cold phenol red-free NKM (37.5μL) before the addition of collagen (10μL 10xPBS, 1.15μL NaOH, 50μL liquid-phase rat tail collagen I). Collagencell suspensions were transferred to imaging plates and placed in the dark at 37◦C/5% CO2 for 10 minutes before adding warm NKM for a further 10 minutes. Cells in matrix were fixed as described above and stained with anti-Granzyme B-AF647 antibody (0.1μg/mL, BD Bioscience) in Perm/Wash™ buffer kit (BD Bioscience) for 24 hours at 4oC. Cells were imaged using a Leica DMI 6000 SP8 confocal microscope with a 63x oil immersion objective (numerical aperture 1.2). Images were acquired in X, Y and Z, with a 0.5μm step size within Z-stacks. Dual pipette aspiration assay A dual-pipette aspiration (DPA) assay, comprised of two remote-controlled TransferMan 4r micromanipulators (Eppendorf) mounted onto a FV1000 confocal microscope (Olympus). was used to aspirate single cells onto glass pipettes to gently conjugate them for live cell imaging.40 Human NK cells were labelled with 100nM CMFDA (Thermofisher) and 200nM Lysotracker Deep Red (ThermoFisher), and K562 human tumour target cells were labelled with 1μM CellTracker™ Orange CMTMR (ThermoFisher). Alternatively, K562 cells were labelled with 100nM CMFDA, and human NK cells were labelled with 1μM CellTracker™ Orange CMTMR and 1μM Lysotracker Green DND-26 (ThermoFisher). Murine NK cells were labelled with 0.25μM CellTracker™ Deep Red (ThermoFisher) and 1μM Lysotracker Green DND-26, YAC-1 target cells were labelled with 100nM CMFDA, and imaging was performed with 1μM SPY555-tubulin (Spirochrome) in the culture medium. Approximately 104 total cells (NK cells and target cells) were resuspended in 100μL phenol-free NKM in 50mm glass bottom dishes (MatTek), with 3mL mineral oil (Merck, Millipore) covering the sample to mitigate sample evaporation. Dishes were placed inside the microscope incubation chamber at 37◦C, 5% CO2. All cells used in the DPA assay were aspirated at 2.5nN unless otherwise specified.40 Imaging was performed using a 40x oil objective with 0.9 NA, with Z-stacks comprised of 1μm optical slices acquired every 10-15 seconds for 10-15 minutes. Immunity 59, 1–17.e1–e8, June 9, 2026 e6


Image analysis
All NK cell and target cell components were segmented as spots and 3D surfaces using Imaris 10.2.0 (Bitplane AG). Fixed NK cells were segmented as surfaces by using membrane labelling to quantify cell volume. Cytotoxic granules were segmented as surfaces to quantify the mean and total granule volume per cell, and the number of granules per cell. The ‘‘split touching objects’’ function (seed points diameter setting of 0.722 μm) was used during surface generation to delineate granules that were in contact. Granules were additionally segmented as spots (XY diameter setting of 0.722 μm) to calculate the mean distance between granules per cell using the ‘‘spots to spots closest distance’’ Imaris plugin. NK cells and target cells imaged with the DPA assay shown in Figures 2D, 2G, and S4F were segmented as surfaces based on cytoplasm labelling. Generated NK cell surfaces were used to apply masks to the cytotoxic granule and target cell channel, allowing for their separation and segmentation. Granules and the MTOC (in murine NK cells) were segmented as spots. The cytotoxic immunological synapse (CIS) used for the quantification of granule and MTOC dynamics was generated from the overlapping voxels of the NK cell signal and the target cell signal. This area of contact was isolated by applying cell surface masks to both channels and setting the voxels of all signal outside the overlapping surfaces to 0. This newly generated synapse channel was then segmented as a surface. A distance transformation was applied to this surface to allow for the quantification of the mean granule distance and MTOC distance to the IS. To determine the contractility of NK cells in response to target cell conjugation, the tail volume of the aspirated NK cells was quantified as previously described for CTLs41 and defined as the volume within the aspirating pipette. To segment this, a second NK cell surface was generated within a defined x-axis region of interest occupied by the aspirating pipette containing the NK cell tail. Consequently, NK cell volume outside the pipette was omitted from surface generation. For analysis of the MTOC in murine NK cells, a median filter was applied to the tubulin channel to remove salt and pepper noise. Next, the MTOC was defined as the centre of highest tubulin intensity within each frame and was manually segmented using spots across all time frames. To quantify the mean granule distance to the MTOC, a distance transformation was applied to the MTOC spots. NK cells that formed a stable MTOC when conjugated with a target were defined as having a singular and obvious tubulin intensity centre during > 65% of total image acquisition. To determine the means and standard deviation of data sets with varying frame rates, all pooled DPA data were interpolated to achieve a uniform time increment of 1 second between data points. The proportion of resting NK cells displaying a clear MTOC was calculated from Z-projections of confocal images using ImageJ.


Degranulation and cytotoxicity assays
Murine NK cell degranulation was assessed by culturing 5x105 splenic NK cells (expanded ex-vivo for 4 days) with or without YAC-1 target cells (1:1 ratio). NK and target cells were mixed in complete NKM containing 5μg/mL anti-CD107a FITC (BD Bioscience), centrifuged for 30 seconds at 300g and incubated at 37◦C, 5% CO2 for 3 hours. After incubation, cells were fixed and stained for intracellular IFNγ (as described above) and analysed by flow cytometry. Supernatants were collected to measure secreted cytokines by flow cytometry on a BD FACs Symphony A5 using a cytometric bead array (LEGENDplex Mouse Anti-Virus Response Panel, BioLegend) according to the manufacturer’s instructions and analysed using LEGENDplex analysis software. Degranulation in freshly-isolated murine CD8+ T cells T cells was assessed by incubating 106 purified, unstimulated splenic CD8+ T cells/well in complete NKM with 5μg/mL anti-CD107a FITC (BD Bioscience) for 3 hours and analysed by flow cytometry. For INFγ production, 106 purified CD8+ T cells were treated with 100ng/mL PMA and 750ng/mL ionomycin (Sigma) in complete NKM for a total of 6 hours. GolgiStop (BD Bioscience) was added 2 hours after stimulation for the remaining 4 hours. Cells were fixed and stained for intracellular IFNγ and analysed by flow cytometry as described above. For cytotoxicity assays with primary murine NK cells, YAC-1 target cells were labelled with 100nM CMFDA (Thermofisher) and 1μM CellTracker™ Deep Red (ThermoFisher) for 20 min in PBS at 37◦C, 5% CO2, then quenched in complete NKM and washed twice with PBS. Cells were counted and assessed for viability before mixing 2.5x104 YAC-1 target cells and 2.5x104, 5x104 or 1.25x105 unlabelled mouse NK cells in 200μL NKM in duplicate wells. The co-cultures were incubated at 37◦C, 5% CO2 for 3 hours then 50μL of 105 counting beads (Spherotech) and 5μg/mL propidium iodide (ThermoFisher) were added to each well to determine total number of viable cells by flow cytometry. Percentage target cell lysis was calculated as the absolute number of target cells remaining after 3 hours of incubation with NK cells, divided by the absolute number of target control cells cultured in the absence of NK cells. Cytotoxicity assays with human NK cells were performed using fresh peripheral blood samples from a healthy volunteer and two patients with MKD (male 8 years old, MVK c.1129G>A, (p.Val377Ile); c.371+2 T>C; female 16 years old, homozygous MVK c.1129G>A, (p.Val377Ile). PBMC were isolated by Ficoll gradient separation and absolute numbers of NK cells were determined by flow cytometry (Trucount tube). Fixed amounts of NK cells (104, 2x104, 4x104) were incubated in duplicate wells for 4 hours with 104 K562 target cells pre-labelled with Cell Trace Violet (ThermoFisher) at effector:target cell ratios 1:1, 2:1 and 4:1. Co-cultures were performed in medium containing 5μg/mL propidium iodide, and the percentage of propidium iodide-positive K562 cells was measured by flow cytometry. In cytotoxicity assays using human NK-92 cells, the cells were pre-treated for 24 hours with either 5μM lovastatin (Sapphire Bioscience) 10μM GGTI-298 (Selleck Chemicals), 2mM NE10790 (a kind gift from Dr Hal Ebetino) or DMSO (vehicle control). 2x106 K562 target cells were incubated with 2μM CMTMR CellTracker Orange (Invitrogen) in 2mL of serum-free TCM for 15 minutes (mixing regularly). After labeling, cells were washed twice with complete TCM and resuspended in TCM supplemented with human IL-2. Pre-treated NK-92 cells were washed and cultured for 5 hours with labeled K562 cells in a 1:1 ratio (105 of each cell type per well/96-well plate, 4 replicate wells per treatment) in TCM+IL-2. Supernatants were collected for IFNγ measurements by cytometric e7 Immunity 59, 1–17.e1–e8, June 9, 2026 bead array (BioLegend) and cells were fixed with Cyto-Fast Fix/Perm Buffer Set (BioLegend) according to manufacturer’s instructions before analysis by flow cytometry. Cell viability was determined by trypan blue staining. MCMV infection and Infanrix/CpG treatment MvkT322M/Δ91 and control Mvk+/T322M mice were infected with 5x103 plaque forming units (PFU) of salivary gland-propagated murine cytomegalovirus (MCMV)-K181Perth intraperitoneally (i.p.). At day 7 post-infection (pi), mice were euthanised and spleen, liver and serum were collected for analyses. For flow cytometry, tissues were processed, cultured and stained as described above. For histological assessment of liver pathology, tissue was fixed in 10% normal buffered formalin for a minimum of 48 hours, then 5μm FFPE sections were stained with hematoxylin and eosin. Liver pathology was scored as mild (3-6 foci), moderate (7-12 foci), moderate-severe (multifocal) or severe (diffuse) changes. Viral loads in each tissue were assessed as previously described.105 In brief, tissues were homogenised in Minimal Essential Medium (MEM, Gibco) enriched with 2% newborn calf serum (NCS, ThermoFisher) before being centrifuged at 1,800g for 15 min at 4◦C to collect supernatants. Serial dilutions of the supernatants in MEM supplemented with 2% NCS were then transferred onto an M210B4 monolayer in 24-well tissue culture trays (ThermoFisher). After incubation for 1 hour at 37 ◦C, 5% CO2, the supernatant was removed and the monolayer overlayed with 1mL 0.01% carboxy-methylcellulose (Sigma) diluted 1:2 in MEM/2% NCS and incubated at 37◦C, 5% CO2 for 4 days. Monolayers were then fixed and stained for 24 hours before plaques were counted to calculate the number per organ. Serum samples were analysed for concentrations of cytokines and chemokines using a multiplex immunoassay (Bio-Plex Pro Mouse Cytokine 23-plex, Bio-Rad) and a MAGPIX platform (Luminex), according to the manufacturer’s instructions. MvkT322M/Δ91 and control Mvk+/T322M mice were treated by i.p. injection of a half-dose (225uL per mouse) of undiluted Infanrix IPV vaccine (GlaxoSmithKline), or 30μg/mouse CpG (ODN 1018 Vaccigrade, Invivogen). 24 hours later, mice were culled and splenocytes were isolated and analysed by flow cytometry as described above. Concentrations of serum IFNγ were measured using a Bio-Rad Bio-Plex assay as described above.


QUANTIFICATION AND STATISTICAL ANALYSIS
Numerical data were plotted and analysed using GraphPad Prism v10.3.0. The statistical analysis performed on each data set is stated in the figure legend. Normal distribution of the data was assessed before selecting statistical tests. Unpaired two-tailed t-tests were used when comparing means between two groups (Welch t-test: data displaying normal distribution not assuming equal SD; Mann-Whitney t-test: no assumptions of normality and equal SD). One-way ANOVA and multiple comparisons tests were used when comparing three or more independent groups (Brown-Forsythe and Welch’s ANOVA with multiple comparison Dunnett’s test for normally distributed data not assuming equal SD, or Kruskal-Wallis ANOVA with multiple comparisons Dunn’s test for no assumptions of normality and equal SD). Two-way ANOVA with Sidak’s multiple comparisons was used to assess the effect of genotype and effector/ target cell ratio on cytotoxicity. Unless otherwise stated, data are presented as the mean ± standard deviation (SD). Significance was defined as *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Alternatively, unless otherwise specified, data shown are representative of at least 3 independent experiments. Immunity 59, 1–17.e1–e8, June 9, 2026 e8


Metadata
Authors
Marcia A Munoz, Iona S Schuster, James Cremasco, Etienne N Masle-Farquhar, Oliver P Skinner, Zoe J Vandeleur, Maté Biro, Daryan Kempe, William D Renton, Sam Mehr, Charlotte Abell-King, Szun Szun Tay, Ryan C Chai, Samar Ojaimi, John J Zaunders, Geetha Rao, Ariel Castro-Martinez, Lisette van de Corput, Andrew N McCorkindale, Leonard D Goldstein, Xiaohong Li, Flore Wouters, Daniel L Kastner, Ignatius Chua, Nicole L Fewings, Fiona C McKay, Catharina M Mulders-Manders, Robert A Brink, Stuart G Tangye, Ivona Aksentijevich, Cindy S Ma, Jeroen van der Hilst, Joost Frenkel, Mariapia A Degli-Esposti, Michael J Rogers
Journal
Immunity
Publisher
Date
pm42066777
PM Id
42066777
PMC Id
Images