Iterative genetic testing identifies SAMHD1 deficiency caused by a homozygous balanced translocation

Excessive and sustained production of the antiviral type-I IFN (IFN-I) family of cytokines can lead to inflammation and cell death, which, if left unchecked, may result in an autoinflammatory feedback loop and neurological abnormalities. In the absence of an infectious or traumatic trigger, these IFN-I–driven disorders are categorized as interferonopathies, a subset of autoinflammatory diseases. Familial chilblain lupus (FCL) is a rare interferonopathy that typically presents from early life, in contrast to spontaneous chilblain lupus, which predominantly manifests in middle-aged women. FCL may present with cutaneous violaceous patches and papules or plaques on the nose, ears, hands, and feet, which are exacerbated by exposure to cold temperatures. These erythematous lesions can progress to nail loss and tissue disfigurement in the affected areas (1). Autosomal dominant inheritance of FCL has been observed in patients carrying heterozygous variants in STING1 (2) and TREX1 (3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14). SAMHD1 has previously been associated with autosomal dominant FCL (15, 16); however, population data are no longer supportive of the association described by Ravenscroft and colleagues, while the case described by Linggonegoro and colleagues lacks supporting genetic data. Monoallelic or biallelic variants in TREX1, biallelic variants in SAMHD1, or variants in a number of other genes have also been identified as monogenic causes of Aicardi–Goutières syndrome (AGS) (17). More complex AGS patients may additionally present with symmetric intracranial calcification, systemic inflammation, arthritis, and complex neurological symptoms.

SAMHD1 encodes the sterile alpha motif (SAM) domain and histidine-aspartic acid (HD) domain-containing protein 1 (SAMHD1), which possesses deoxynucleotide triphosphate (dNTP) triphosphatase activity. SAMHD1 limits the cellular pool of free dNTPs, thereby restricting available material for unregulated activation of nucleic acid–sensing pattern recognition receptors (18, 19). This has been reported across many SAMHD1-deficient individuals to result in systemic overproduction of IFN-I and elevated and chronic transcription of IFN-stimulated genes (ISGs) (20, 21, 22). Recently, Rabinowitz and colleagues have shown in SAMHD1-deficient THP-1 monocytes that this elevated ISG transcription is dependent on functional detection of double-stranded DNA (dsDNA) by cyclic GMP-AMP synthase and stimulator of IFN genes (STING). Correspondingly, knocking out the cytoplasmic RNA sensors melanoma differentiation-associated protein 5 (MDA5) and retinoic acid–inducible gene I (RIG-I) had no effect on ISG levels in the THP-1 cells, suggesting that, at least in monocytes, the lack of SAMHD1 activity is primarily detected as cytoplasmic accumulation of dsDNA (19).

We present here the unique case of a translocation event between chromosomes 17 and 20 resulting in disruption of SAMHD1. The proband presents with symptoms consistent with FCL, but without the neurological complications typical of AGS. The patient also carries a de novo likely pathogenic variant in MN1, adding further complexity to the clinical presentation.

A 13-year-old male presented with early-onset severe perniosis with autoamputation of several toes, mild erosion of other extremities including fingers and ears, and widespread small joint arthritis (active joint count 14). He had been affected by severe perniosis from late infancy (Fig. 1 a). He had mild facial dysmorphism, microcephaly (Z-score −4), severe sensorineural hearing loss and unilateral myopia in addition to growth restriction (height Z-score −3.2) despite treatment with recombinant human growth hormone (Fig. 1 b). Magnetic resonance imaging of the brain showed no signs of cerebral calcifications or other lesions, and the patient had no other remarkable neurological symptoms and an age-appropriate intellect. High-resolution computerized tomography scanning revealed no signs of interstitial lung disease. Clinical records reported late eruption of permanent teeth, laryngeal inflammation, mild mitral regurgitation, and resolved eczema, as well as immune abnormalities characterized by accelerated erythrocyte sedimentation rate (ESR), but normal C-reactive protein, persistent mild anemia but normal ferritin levels, mild neutropenia, natural killer (NK) cell lymphopenia, hypergammaglobulinemia, and elevated CD19⁺ B cells, although his switched memory B cell counts (CD19⁺, CD27⁺, IgM⁻, and IgD⁻) were very low (Table 1). Despite the observed hypergammaglobulinemia, autoantibody titers for antinuclear antibodies, anti-extractable nuclear antibodies, anti-dsDNA, anti-proteinase-3, anti-myeloperoxidase, and antiphospholipid antibodies were normal. Both parents were phenotypically normal at consultation and reported an uncertain degree of consanguinity through their respective paternal lines (Fig. 1 c). Microarray analysis of the proband revealed a male molecular karyotype with no clinically significant copy number variations. However, extensive long continuous stretches of homozygosity were detected, representing ∼6.6% of the genome, indicative of such consanguinity.

Clinical trio exome sequencing revealed a de novo, heterozygous, likely pathogenic, N-terminal variant in the MN1 gene (c.1306c>t, p.G436*). Premature truncation of MN1 has previously been associated with the gain-of-function MN1 C-terminal truncation syndrome (23). N-terminal truncating variants are less reported and are expected to lead to nonsense-mediated decay resulting in loss-of-function and haploinsufficiency associated with mild learning problems, conductive hearing loss, growth delay, and dysmorphism, although individuals reported to date remain limited (24, 25, 26). While MN1 haploinsufficiency likely explains our patient’s growth restriction and facial dysmorphism, this diagnosis is not consistent with the observed erosion of extremities, small joint arthritis, or immune abnormalities, such as elevated ESR and immunoglobulins (Table 2). Given these unexplained autoinflammatory features, the patient was referred to the Australian Autoinflammatory Diseases Registry for research investigation.

Bulk RNA sequencing (RNA-seq) was performed from the patient’s peripheral blood mononuclear cells (PBMCs) and compared to averaged parental samples. Gene Ontology (GO) enrichment analysis on the significant differentially expressed genes (DEGs) revealed altered expression of pathways including “regulation of innate immune response,” “regulation of immune effector process,” and “leukocyte-mediated immunity,” which are consistent with elevated innate immune activation and IFN-I responses. Similar analysis of the proband’s transcripts compared to those of healthy donor controls returned a nearly identical pattern with the addition of GO terms associated with responses to infectious agents such as viruses and bacteria, again suggesting immune activation and an IFN-I response (Fig. 2 a). Several ISGs were among the most upregulated significant DEGs when comparing the proband’s sample to the parental or healthy donor transcripts, including IFI44L, ISG15, EIF2AK2, and USP18; however, another known ISG, SAMHD1, was instead identified as one of the most significantly downregulated DEGs with P values of 1.6 × 10⁻⁴ and 2.3 × 10⁻⁵ when compared to parents and healthy donor datasets, respectively (Fig. 2 b). Following ISG filtering, we found that SAMHD1 was the most significantly downregulated ISG (by >50-fold) in the patient vs. parents dataset and the only downregulated ISG when comparing the patient to healthy donor controls.

Alignment of the RNA-seq reads within the SAMHD1 locus revealed only a subset of reads to be downregulated in the patient, with reads through to exon 4, but not throughout the 12 exons 3′ of that locus. This was in contrast to complete reads throughout SAMHD1 from the maternal sample (Fig. 3 a). This observation was confirmed by direct quantitative PCR (qPCR) using patient, parental, and healthy donor PBMC RNA and primers specific to exons 2–3, exons 4–5, or exons 6–7 of SAMHD1. This showed reduced expression of all the tested exons in both parental samples compared to healthy donors but further reduction of exons 4–5 and complete absence of exons 6–7 in the patient’s sample (Fig. 3 b), indicating a homozygous deletion of exons 4–7 and a heterozygous deletion in the parents.

Detection of early, but not late, exons of the SAMHD1 transcript suggested that transcription of the 3′ end of the gene may have become uncoupled from the promoter in cells of the proband. Whole-genome sequencing was performed to investigate a suspected intronic variant following SAMHD1 exon 4. Short-read sequencing revealed a suspected homozygous translocation, as zero reads from either direction mapped across the region. Investigation of soft-clipped bases showed mapping to poorly defined decoy sequences within the reference build. This indicated the other side of the potential translocation was possibly in a difficult-to-map region, such as repetitive sequence.

Long-read genomic sequencing was performed; however, initial attempts to align the distal end of SAMHD1 to the GRC38/hg38 reference genome were unsuccessful. Subsequent mapping to the telomere-to-telomere (T2T) reference genome (T2T-CHM13v2.0) (40) provided sufficient resolution within repetitive regions to reveal that the reciprocal breakpoint mapped proximal to the centromere in the p-arm of chromosome 17 (p11.2), a poorly annotated region in earlier human genome builds. Therefore, the proband was found to be homozygous for a balanced t(17;20)(p11.2;q11.23) translocation, with the chromosome 20q11.23 breakpoint occurring within intron 4 of the SAMHD1 gene (Fig. 4 a). Chromosomal analysis was performed on a peripheral blood sample taken from the proband, and G-banding revealed a male karyotype with two identical, apparently balanced reciprocal translocations involving both copies of chromosomes 17 and 20, with breakpoints at 17p11.2 and 20q11.23, consistent with the long-read sequencing data (Fig. 4 b). To validate that this translocation event resulted in loss of SAMHD1 expression, western immunoblotting was performed on whole-cell lysate samples prepared from proband, paternal, and healthy donor PBMCs. While bands were detected at ∼72 kDa and 60–70 kDa in the paternal and healthy donor samples, corresponding to full-length SAMHD1 (UniProt entry Q9Y3Z3-1) and shorter isoforms (possibly UniProt entries Q9Y3Z3-3 or Q9Y3Z3-4), no bands were detected from the proband sample, suggesting complete loss of SAMHD1 protein expression (Fig. 4 c).

The patient was treated with tofacitinib, a JAK inhibitor preferentially inhibiting JAK1 and JAK3, at doses extrapolated from other childhood inflammatory syndromes (41). qPCR analysis of a standardized subset of ISGs including IFI27, IFI44L, IFIT1, ISG15, SIGLEC1, and RSAD2 confirmed the interferonopathy from which the patient suffered pre-treatment (Fig. 5 a). Both parental samples showed similar expression of ISGs to healthy donor controls. No increase was detected in IFNB1 itself in the patient, which may reflect the transient and tightly regulated nature of IFN cytokine expression or elevation of IFN-I family members other than IFNβ. Little to no elevation of transcripts from other inflammatory pathways was observed, including TNF, IL6, and IL1B. Similar samples taken from the patient following tofacitinib treatment illustrated the efficacy of JAK inhibition in suppression of this IFN signature (Fig. 5 a). This correlated with marked improvement of symptoms following tofacitinib treatment, including improved weight and accelerated growth trajectory (Fig. 5 b), decreased perniosis (Fig. 5 c), and improved joint range of motion with a complete resolution of active joint count from 14 pre-tofacitinib to 0 after treatment. NK cell numbers were completely normalized, and switched memory B cells increased by more than threefold (Table 1). Treatment with a more targeted therapy, anifrolumab, an IFN α receptor 1 antagonist, was considered but declined owing to the degree of improvement while on tofacitinib.

Exon-specific transcriptional analysis indicated complete loss of expression of the 3′ end of SAMHD1 in the proband. Due to the loss of the SAMHD1 3′ untranslated region, stability of the truncated transcript is expected to be severely reduced, and western blotting showed complete loss of protein product in patient PBMCs. While it is possible that truncated SAMHD1 protein does not contain the epitope detected by the antibody used in this study, loss of SAMHD1 expression beyond exon 4 is predicted to result in ablation of the HD domain, implying any undetected protein product would be deficient in dNTP triphosphatase activity. Similar analysis of the parental samples showed only a partial loss of expression of the 3′ end of the transcript, suggesting that both parents likely possess single copies of the chromosomal derivatives arising from the t(17;20)(p11.2;q11.23) translocation and subsequently are heterozygous carriers of the disrupted SAMHD1 allele. Further interrogation of the parental samples, however, showed no defect in SAMHD1 protein expression, nor regulation of ISGs, implying that the autoinflammatory phenotype in the proband is recessive.

Other cases of SAMHD1 deficiency have been reported with a broad spectrum of symptoms (29). At a single center, three cases with identical homozygous SAMHD1 c.625G>A, p.G209S variants, presented with a highly divergent clinical spectrum, including both FCL and AGS (42). Therefore, it is likely that there are environmental or genetic phenotypic modifiers of the symptomatic spectrum observed in SAMHD1-deficient individuals. Among interferonopathies, there is precedent for phenotypic modifier alleles, such as the HAQ allele of STING1, one of a number of factors that may suppress the clinical presentation of COPA syndrome (43). Further interrogation of whole genome sequencing from our patient shows that he is also heterozygous for the STING1 HAQ allele, so the lack of typical AGS5 symptoms (e.g., encephalopathy, intellectual developmental delay) despite homozygous loss of SAMHD1 expression may be partially attributable to a similar suppressive effect of this haplotype of STING1. However, in the absence of a larger cohort of SAMHD1-deficiency patients to confirm such a correlation, this remains hypothetical. Furthermore, analysis of a broader cohort of COPA syndrome patients has uncovered individuals without the STING1 HAQ allele who are asymptomatic and one HAQ allele carrier who suffers from COPA-driven kidney disease, indicating that the HAQ allele is not the only factor modulating penetrance or severity of STING-related interferonopathies (44). Indeed, the clinical presentation of the current proband may be influenced by a number of other genetic or environmental factors (45), especially considering that he also carries a de novo variant in MN1. Of the symptoms reported from the proband in this study, most are consistent with previously reported cases of recessive loss-of-function variants in SAMHD1, with only growth restriction, microcephaly, and mild facial dysmorphism being consistent with previous reports of MN1 haploinsufficiency (Table 2).

Balanced reciprocal translocations occur in 0.09% of unselected newborns (46). Each individual translocation within this group is exceedingly rare, with most individual non-Robertsonian translocations being represented at most once with any given cohort (47). The autozygous balanced nature of the current proband’s genotype is also remarkable, considering the probability of two parents carrying an identical balanced t(17;20)(p11.2;q11.23) translocation is incredibly low, making the genomic architecture of the proband strikingly unique. In this family, there is potential fourth-degree consanguinity, meaning that the balanced translocation has been successfully transmitted at least three times on each side. It remains technically possible that the balanced translocation could be present in the ancestral population at a low frequency; however, the long regions of homozygosity and potential consanguinity within this family mean the translocated allele is much more likely to have arisen from a recent shared ancestor.

We are unaware of any heterozygous germline t(17;20)(p11.2;q11.23) translocations in the literature. One other report of a homozygous t(17;20) translocation has been published; however, as the breakpoints were between distinct areas of both affected chromosomes (t(17;20)(q21.1;p11.21)), the severe developmental abnormalities in that homozygous neonate are unrelated to those in the proband in this study (48). The most recent meta-analysis of homozygosity in balanced reciprocal translocations was published in 2010, with only five families recorded at the time and very few additional case reports since (49). The likelihood that reports of homozygosity only arise upon investigation of pathology, as was the case for the family reported here, makes determining the true rate of homozygous reciprocal translocation, including those who are phenotypically normal, a challenging exercise.

This appears to be only the third pathogenic autosomal recessive gene disruption event due to a homozygous balanced translocation described (50, 51). Identification of this balanced translocation causing an inborn error of immunity highlights the utility in certain cases of iterative genetic testing, including transcriptome analysis and long-read sequencing (Fig. S1). While the approach we have described was valuable for identification of the structural variation carried by this patient, the particular suite of genetic tests and decisions required to identify other rare inborn errors should be assessed on a case-by-case basis. Nonetheless, because immune-related genes are often rapidly evolving and typically nonessential for survival, it is plausible that, in some cases, disease may be driven by exceptionally rare or otherwise improbable genetic alterations that require nonstandard genomic diagnostics such as those employed here.

PBMCs were isolated from the patient and family members at Monash Health and from healthy donors at the Walter and Eliza Hall Institute or Red Cross Lifeblood. PBMCs were isolated from whole blood using Ficoll (GE Healthcare) before being frozen in fetal bovine serum (Bovogen) with 10% DMSO (Sigma-Aldrich) and either used directly or stored in liquid nitrogen until use (cell handling between groups was consistent within each experiment).

Buccal swab samples were taken from the proband and both parents, and exome sequencing was performed by Blueprint Genetics Whole Exome Family Plus Test (v3, 2023) utilizing paired-end 150 bp reads on the Illumina platform. Reads were mapped to the Homo sapiens genome (GRCh37/hg19) using Burrows-Wheeler Aligner (BWA-MEM) software (52).

1 × 10⁷–4 × 10⁷ PBMCs were lysed using TRIzol reagent (Invitrogen) for 5 min at room temperature. Samples were either stored at −80°C or immediately processed. RNA extraction was performed as per the manufacturer’s instructions. cDNA was generated from 1 to 2 μg of total RNA using the Invitrogen SuperScript III First-Strand Synthesis System for reverse transcription PCR (Invitrogen) utilizing oligo(dT)s. qPCR was performed using SYBR Green/ROX qPCR Master Mix (Thermo Fisher Scientific) on ViiA 7 or QuantStudio 6 Real-Time PCR systems (Thermo Fisher Scientific). Primers for genes of interest and housekeeping genes are listed in Table 3. Each sample was run in duplicate, and samples were normalized using the housekeeping gene ACTB. Results were analyzed using the ΔΔCt method and represented as fold change (FC) of the average of healthy donor samples from the same experiment.

ISG Score was calculated from SYBR green qPCR results for the following ISGs: IFI27, IFI44L, SIGLEC1, ISG15, IFIT1, and RSAD2 (20, 53). Expression of each ISG was calculated using the formula 2^−ΔΔCt, then normalized to the average of ACTB expression. For each ISG, individual samples were then normalized to the average of all healthy donors run in the experiment. Finally, the geometric mean of the relative expression of these six ISGs was used to calculate the final IFN score.

RNA was extracted from PBMCs isolated from the patient, parents, and healthy donors, as described above, and sent for sequencing by the Australian Genome Research Facility. mRNA was purified using oligo(dT) beads and, following fragmentation, the cDNA library was prepared using the Illumina TruSeq stranded RNA protocol. Sequencing using the Illumina NovaSeq X Plus platform was performed to generate paired-end 150 bp reads. Base calling was performed with Illumina DRAGEN BCL Convert (07.021.645.4.0.3 pipeline). Following demultiplexing and quality control, sequences were mapped to the H. sapiens genome (GRCh38/hg38) using the STAR aligner (v2.3.5a) (54). A gene counts matrix was assembled using the featureCounts v1.5.3 utility of the Subread package (55). StringTie (v2.1.4) was employed to obtain estimated transcript structure and abundance from aligned BAM files (56). Differential expression analysis was performed using a generalized linear model in the edgeR package (v4.0.9) (57) using R (v4.3.1) (58). GO analysis was completed with the org.Hs.eg.db package (59) to compare the significant genes (P value <0.05) to the genome-wide annotation for the Human database; clustering genes based on Biological Processes (Ont. = BP). Overrepresented sets of DEGs are shown by the circle size and colored by the degree of statistical significance. Differential gene expression analysis compared a predetermined significance threshold and expression FC. To capture all potential genes of interest, a standard P value <0.05 was used to highlight statistical significance, and absolute log₂FC >0.6 was used to take a sensitive approach to identifying the FC expression of genes. Volcano plots were created using base R (v4.5.1) plotting functions (60), and those above the assigned significance threshold were colored in blue and red to represent downregulated and upregulated genes, respectively. ISGs were identified within the lists of significant DEGs by first filtering against the Interferome database (61) with the search conditions “type-I IFN,” “in vivo,” and “H. sapiens.” To further filter for genes directly induced by IFN signalling, the Interferome-filtered DEGs were then cross-checked against genes with STAT1-binding sites within 1 kb of the transcriptional start site, generated from the ChIP-Atlas Target Genes tool (62), selecting dataset SRX150550 (K562 cells treated for 6 h with IFNα).

Short-read whole-genome sequencing was performed on the proband and both parents by the Garvan Genomics Platform, using an Illumina NovaSeq 6000 and Illumina DNA PCR-Free Prep Kit. Genomes were sequenced to a mean coverage of ≥30×, and paired-end reads were aligned to the human genome reference sequence (GRCh37).

Library preparation and Oxford Nanopore Technologies sequencing were performed by the Garvan Genomics Platform. DNA was first extracted from patient PBMCs using the PacBio PanDNA kit. High molecular weight genomic DNA was then sheared to ∼20–30 kb fragment size using the Diagenode Megaruptor 3 DNA shearing system (speed 29) and visualized on an Agilent Femto Pulse using the Genomic DNA 165 kb Kit prior to library preparation. The sequencing library was generated using the SQK-LSK114 ligation sequencing kit with 3 µg of sheared DNA as input. Sequencing was performed on a PromethION run with a FLO-PRO114M flow cell with MinKNOW (v25.03.7, MinKNOW Core v6.4.8). High-accuracy basecalling was performed using Dorado (v7.8.3, basecalling module version [email protected]) and Bream (v8.4.4) to generate raw reads in FASTQ format. Quality filtering was performed to retain only those reads with a minimum Q score of 9.

Raw reads were quality trimmed using filtlong (v0.3.1, Python v3.10.14) (63) with parameters --min_length 1000, --keep_percent 90, and --target_bases 40000000000 to preserve approximately the best-quality half of all bases. Alignment of reads to the human T2T reference genome (T2T-CHM13v2.0) (40), and structural variant calling were performed using NanoVar (v1.8.3, with Python v3.11.13) (64) in Oxford Nanopore mode (-x ont, otherwise default parameters), which calls Minimap2 (v2.30-r1287) (65) for alignment steps. The aligned reads file was sorted using samtools (v1.22) sort, and an index file was generated using samtools index with default parameters for viewing within Interactive Genomics Viewer (66).

G-banded karyotype analysis was performed on phytohemagglutinin-stimulated T cells from this patient using standard methods.

DNA was extracted from a peripheral blood sample using the EZ1 DSP Blood kit (QIAGEN), and microarray was performed using the Illumina Infinium GSA-24 v3.0. Analysis was performed on NxClinical (Bionano) using data normalized by Genome Studio (Illumina).

PBMCs were isolated using Ficoll (GE Healthcare) as described above and underwent RBC lysis in RBC lysis buffer (156 mM ammonium chloride, 11.9 mM sodium bicarbonate, and 0.1 mM EDTA) for 5 min at room temperature before dilution to 30 ml with 1× PBS, followed by washing and resuspension in PBS for counting. 2 × 10⁶ cells were lysed in 1× radioimmunoprecipitation assay buffer (1% Triton X-100, 20 mM tris-HCl [pH 7.4], 150 mM sodium chloride, 1 mM EDTA, 0.5% sodium deoxycholate, 3.5 mM SDS, and 10% glycerol) supplemented with 10 mM sodium pyrophosphate, 5 mM sodium fluoride, 1 mM sodium orthovanadate, 1 mM phenylmethylsulfonyl fluoride and cOmplete protease inhibitors (Roche Biochemicals) for 30 min at 4°C. Samples were processed through Pierce centrifuge columns (Thermo Fisher Scientific) to shred genomic DNA. After addition of reducing SDS sample loading buffer (1.25% SDS, 12.5% glycerol, 62.5 mM tris-HCl [pH 6.8], 0.005% bromophenol blue, and 50 mM dithiothreitol) and denaturation at 95°C for 5–10 min, samples were separated on Bolt 4–12% precast SDS-PAGE gels (Thermo Fisher Scientific) with MES running buffer (Thermo Fisher Scientific) and subsequently transferred onto nitrocellulose membrane (Millipore). Membranes were blocked in 5% skim milk in Tris-buffered saline (TBS) containing 0.1% Tween 20 (Sigma-Aldrich) (TBS-T) before overnight incubation with specific primary antibody in 5% skim milk in TBS-T at 4°C: Rabbit anti-SAMHD1 (1:1,000 #12586-1-AP; Proteintech). Membranes were washed five times with TBS-T and incubated with appropriate HRP-conjugated secondary antibodies: Goat anti-Rabbit (1:10,000, #P044801-2; Agilent) or anti-actin–HRP (1:10,000, #sc-47778; Santa Cruz Biotechnology) and washed five times again. Finally, membranes were developed using Immobilon Forte Chemiluminescent HRP Substrate (Millipore) and imaged using the ChemiDoc Touch Imaging System (Bio-Rad).

Fig. S1 shows the iterative testing workflow and reasoning behind each step used to identify the uncommon structural genetic variant carried by the proband in this study.

Informed consent was obtained from the patient and family, in accordance with local regulations, and a protocol for research on human subjects was approved by the institutional review boards of Monash Health (Human ethics number: HREC-15-MonH-31) and the Garvan Institute of Medical Research (Human ethics number: HREC X20-0177). Healthy donor samples were obtained from the Volunteer Blood Donor Registry at the Walter and Eliza Institute of Medical Research (Human ethics number: WEHI HREC 18/07) or the Australian Red Cross Lifeblood (Human ethics number: MonH-2025-483166). All participants, or their legal representatives, consented to take part in this study and to have the results of this research published, including patient images.

The raw genetic and transcriptional data underlying this study are not publicly available, given it is classified as protected health information. The data are available from the corresponding author upon reasonable request.

The authors would like to thank the patient and his family for their involvement in this study, Dr. Marianne Mégroz for administrative support, and Dr. Sophia Davidson for discussion and advice regarding data interpretation.

S.L. Masters is supported by Australian National Health and Medical Research Council Project Grants (2003159 and 2003756).

Author contributions: Paul J. Baker: conceptualization, data curation, formal analysis, investigation, methodology, project administration, visualization, and writing—original draft, review, and editing. Yaoyuan Zhang: data curation, formal analysis, investigation, validation, visualization, and writing—review and editing. Imogen Bishop: data curation, formal analysis, investigation, methodology, visualization, and writing—review and editing. Madeline L. Cleveland: data curation, formal analysis, investigation, methodology, visualization, and writing—review and editing. Alexandra L. McAllan: data curation, formal analysis, investigation, methodology, and writing—review and editing. Georgina E. Hollway: formal analysis, investigation, methodology, and writing—review and editing. Ravikiran Vedururu: formal analysis, investigation, project administration, and writing—review and editing. Amit Kumar: formal analysis and writing—review and editing. Raj Krishnaswamy: formal analysis, validation, and writing—review and editing. Ken L. Wan: data curation, formal analysis, investigation, visualization, and writing—review and editing. Peter A. Kaub: formal analysis, investigation, supervision, and writing—review and editing. Emma L. Brown: formal analysis, methodology, visualization, and writing—review and editing. Dhanya Lakshmi Narayanan: investigation and writing—review and editing. Ira W. Deveson: data curation, investigation, and writing—review and editing. Peter Gowdie: investigation and writing—review and editing. William D. Renton: conceptualization, investigation, validation, and writing—review and editing. Samar Ojaimi: conceptualization, investigation, resources, and writing—original draft, review, and editing. Andrew P. Fennell: conceptualization, investigation, and writing—review and editing. Seth L. Masters: conceptualization, data curation, formal analysis, funding acquisition, investigation, methodology, project administration, resources, supervision, validation, visualization, and writing—original draft, review, and editing.