Case presentation and functional studies
Our index patient (M107) was born to consanguineous healthy parents of Moroccan ancestry (Fig. 1A) and hospitalized soon after birth for failure to thrive with clinical and laboratory evidence of severe, sterile systemic inflammation (Fig. 1B, Suppl. Material). Her disease progressed despite high-dose steroids and broad-spectrum antimicrobials (Suppl. Material). A recessive disorder was suspected and WES identified a homozygous OTULIN variant [ENST00000284274.5: c.595T > A; p.(Trp199Arg), henceforth referred to as W199R] consistent with the patient’s presentation and segregating appropriately in her parents (Fig. 1A,C). This variant is absent from large population databases such as gnomAD v.4.0, has not been previously reported in the literature or in large variant databases such as ClinVar, is predicted to result in the non-conservative substitution of a highly conserved OTU domain residue, and is considered deleterious by multiple in silico algorithms.
Unfortunately, our index patient passed away before any functional studies could be performed. Therefore, we performed in silico comparative 3D structural modeling of WT and W199R OTULIN and observed that the latter alters a number of intra-protein interactions (Fig. 1D, S2). Specifically, hydrogen bonds involving Leu195, Ala203, and Leu202 are lost, while those with Gly144 and His300 are gained, while hydrophobic and/or aromatic interactions involving Ala138, Met139, Ala142, Pro146, Trp148 and Leu149 are reduced. W199R also results in greater protein instability with a ∆∆G (change in Gibbs free energy) of − 2.061 Kcal/mol [29] and a change in electrostatic potential via increased positive surface area (Figure S2A), which may affect interactions with other proteins. Additionally, MD modeling studies found increased dynamic fluctuation (RMSF) of catalytic core-proximal loop residues 281–285, which are involved in polyUb interaction (Figure S2B-C), so OTULIN’s affinity for polyUb may also be affected.
To directly address the functional consequences of W199R, we cloned and transfected this variant into HEK293 cells and observed severely reduced (~ 75%) protein levels of OTULIN-W199R relative to WT or the known pathogenic OTULIN-L272P variant (Fig. 1E,G, S2D). This may be attributable to a splicing defect leading to mRNA instability (c.595T is the first nucleotide of exon 6) and/or to the protein instability mentioned above. Since mRNA levels of this variant expressed in HEK293T cells remained stable (Figure S2E), we concluded that the latter is more likely. The increased Ub chain accumulation seen after NEMO immunoprecipitation shows that both OTULIN-W199R and -L272P mutants failed to deubiquitinate NEMO to the same extent as the WT protein in vitro (Fig. 1F). As additional support for its pathogenicity, OTULIN-W199R led to increased NF-kB activity by luciferase reporter assay to the same extent as OTULIN-L272P (Fig. 1H).
Systematic literature review and genetics
Our comprehensive literature search identified 13 relevant research/review articles (12 from PubMed and one from MedRxiv) reporting three OTULIN-related conditions [9, 13, 19, 32–41]. In total (including our Moroccan family), we identified 16 families and 116 individuals (Table S1, Figure S3), 56 of whom carry monoallelic or biallelic variants (Table 1). Among these 56 individuals, 9 from 7 families were reported to have ORAS - with 8 homozygotes and 1 compound heterozygote. An additional 15 individuals from 9 families were diagnosed with IMD107, including subjects H077 and H079, who were obligate/presumed heterozygotes for the familial variant. Finally, we assigned the diagnosis of DN-ORAS to 2 individuals from 2 families, who shared the same dominant-negative catalytic domain variant p.(Cys129Ser), and to the recently reported patient heterozygous for p.(A240V) variant because of her autoinflammatory phenotype without evidence for infection [41]. The remaining 29 individuals harbored variants but were unaffected (Table 1).
As previously noted [20], the same patients and family members may be reported in more than one independent study, so we examined the existing literature for such redundancies and found that 1 patient had been included in 4 articles (A023), 6 patients in 3, and 31 individuals in 2 (Fig. 2A). Of the 28 reported patients, 27 were genetically confirmed, 26 had detailed clinical data available, 16 had functional data performed on their primary cells, and 10 had available immunophenotyping data. However, all four datasets were present for only 7 patients (Fig. 2B).
For this study, we evaluated all 38 reported variants (including our patient’s novel W199R and the 2 variants reported as Likely Pathogenic in ClinVar) (Table 2). Unfortunately, we could not confirm the appropriate cDNA information for the variant “TQK100-102AAA” reported by Keusekotten et al. [9] or for the variant “c.395_396ins, p.Leu131_Arg132insLeuCysThrGlu” reported by Gezgin et al. [38], so these cannot be included until we receive further details from the authors. Of note, 10 of the 16 disease-associated variants reported to date could not be found in OMIM or ClinVar (Fig. 2A).
The OTULIN protein consists of an N-terminal PUB-interacting motif (PIM) domain, an ovarian tumor (OTU) domain and a C-terminal PDZ binding motif (Fig. 2C). All 16 mutations associated with OTULIN-related diseases (Fig. 2C), including W199R, are located within the large OTU domain (aa79-348), which is required for M1-Ub chain binding and hydrolysis. In particular, conformational regulation of the catalytic triad Cys129-His339-Asp341 within this domain is important for determining OTULIN function and specificity [42]. Twelve missense mutations are associated with either DN-ORAS or ORAS and/or IMD107; 2 frameshift mutations are associated with ORAS and/or IMD107; 1 stop codon is associated with IMD107; and 1 splice-altering variant [c.864 + 2T > C; (EX6 + 2T > C); p.(Trp199_Gln288del)] associated with both ORAS and IMD107 results in the production of smaller transcripts corresponding to skipping of exons 5 and/or 6 or retention of 17 nucleotides between exons 4 and 5 followed by exon 6 skipping [34]. Thus, these mutations may affect OTULIN catalytic activity, its binding to linear chains and/or protein stability.
Clinical features and treatment outcomes of OTULIN-related diseases
We compared clinical phenotypes across the 9 ORAS, 15 IMD107 and 3 DN-ORAS patients (Fig. 3A). Although the number of reported patients for each condition was limited, we can see that all ORAS or DN-ORAS patients, regardless of mutational mechanism, were affected by various manifestations of systemic autoinflammation. These included but were not limited to recurrent fevers, arthritis/arthralgias, diarrhea, lipodystrophy and erythematous rashes with painful subcutaneous nodules (panniculitis) (Fig. 3A). Laboratory studies were notable for elevated inflammatory markers, leukocytosis and neutrophilia in the absence of known infection, as well as evidence of neutrophilic skin infiltration on histopathology. No evidence of immunodeficiency in these patients was reported/described. Of note, compound heterozygosity for 2 hypomorphic mutations has been associated with later-onset disease characterized by life-threatening, multi-organ sterile abscesses involving the skin, lung, and spleen [37]. When all ORAS and DN-ORAS patients are considered (12 total), we noted a mean diagnostic delay of ~ 7 years between symptom onset and achievement of molecular diagnosis (Fig. 3B).
Significantly more phenotypic variability is seen for IMD107. Most patients with IMD107 experience their first infection episode during adolescence [39], with some more severely affected than others. Clinical involvement for many parents of ORAS patients may be subtle and may not be revealed without dedicated clinical re-evaluation [13, 32, 34, 35]. Levels and functions of immune cells when measured appear normal, supporting the hypothesis that the molecularly relevant defect may reside in non-hematopoietic cells [39]. Transmission of IMD107 in the families reported was consistent with autosomal dominant inheritance with variable expressivity and incomplete penetrance. We calculated that 38.2% of confirmed heterozygotes are clinically affected, but this penetrance estimate decreases to 34.9% with the inclusion of presumed/obligate heterozygous carriers for pathogenic variants (Fig. 3C).
Regarding management outcomes, data were available for 12 individuals from 10 families, including 3 DN-ORAS cases (Fig. 3D, Table S2). All reported patients received steroids and all except our ORAS patient (M107) also received at least one form of immunomodulation (up to a maximum of 3 different classes). The most common agent used was TNF inhibitor (n = 7), followed by IL-1 inhibitor/anakinra (n = 5) and methotrexate (n = 4), azathioprine (n = 2), and colchicine or JAK inhibitor/ruxolitinib (n = 1). Almost all individuals responded positively to steroids to some extent but many went on to be trialed with other agents. All 7 patients who received TNF inhibition (four ORAS and three DN-ORAS) showed at least moderate, if not robustly, positive responses. By comparison, anakinra, colchicine, ruxolitinib and methotrexate elicited only partial clinical responses in one individual at most. One patient ultimately received a curative hematopoietic stem cell transplant (HSCT) after failing several kinds of immunomodulation (Fig. 3D, Table S2). By contrast, the use of antimicrobials in the non-IMD107 patients was much less frequently reported (n = 4), with known or positive outcomes in only one patient treated for pneumonia (not shown).
Molecular and cellular consequences of OTULIN LOF
Patient cell data
For 7 ORAS patients, 8 IMD107 patients, and 2 unaffected heterozygotes (F066, F067), more detailed cellular characterization at baseline or in response to specific stimuli was available (Figure S4, Table S3). Levels of OTULIN mRNA were comparable to controls for 8/11 tested patients and decreased for 1 ORAS (C039) and 2 IMD107 (H080 and H083) patients [Row 1]. Expectedly, cDNA sequence was abnormal for 2 individuals (E062 and E065) sharing a splice-altering variant [Row 2]. All patients except one with DN-ORAS (N110) showed decreased or nearly absent OTULIN protein levels [Row 3]. Protein (and mRNA) levels of LUBAC components (HOIP, HOIL-1 and SHARPIN) were comparable to controls in some tested patients but mildly to clearly reduced in others, with low expression of one, two or all three components [Rows 4–9].
Fibroblasts or immune cells from all tested individuals showed increased accumulation of M1-Ub chains, consistent with loss of OTULIN-mediated deubiquitination [Rows 11–12] [9]. Fibroblasts and PBMCs from 3 ORAS patients (A023, B035, C039) also showed increased linear ubiquitination of specific targets such as NEMO, TNFR1, RIPK1 and ASC under stimulation by TNF or IL-1β [32]. Moreover, fibroblasts from 6 ORAS or IMD107 patients showed accumulation of high molecular weight caveolin-1 complexes, presumed or shown to be modified by K63-Ub [Rows 13–14].
Stimulated fibroblasts or T cells from ORAS patients (A023, B035, C039) showed increased phosphorylation of JNK, IKKα/β, I𝛋B-α, p38, p65-NF-𝛋B, suggesting activated NF-𝛋B and MAPK signaling [Rows 16–21], along with increased expression of pro-inflammatory cytokines (i.e. IL-18, TNF-α, IL-6, IL-12, IL-1β), and increased production of IFN-γ [Rows 22–34] (Fig. 4).
PBMCs and fibroblasts from 3 ORAS patients showed reduced caspase-like proteasome activity relative to controls, as well as downstream abnormalities in proteasome assembly and function [Rows 35–39], leading to accumulation of unfolded or K48-ubiquitinated proteins [Row 40]. Tryptic-like and chymotryptic-like proteasome activity was reduced in PBMCs but not fibroblasts from 2/3 patients, specifically implicating the immunoproteasome, a known OTULIN substrate [19]. This was associated with increased expression of IFN-stimulated genes in DN-ORAS (N110) and ORAS (C039) patients [Rows 41–54], along with increased levels of IFN-α and other inflammatory signals (i.e. IP10, MIG, RANTES, MCP1) in the PBMCs, monocytes, whole blood or serum of ORAS patients (A023, B035, C039) [Rows 55–60].
Finally, ORAS and DN-ORAS skin fibroblasts or hepatocytes (A027) showed increased apoptosis after stimulation by TNF + BV6, TNF + CHX, alpha-toxin, or S. aureus, but not TNF alone, while IMD107 skin fibroblasts showed increased levels of cell death only after stimulation with alpha-toxin or S. aureus [Rows 61–65] [39, 40].
Other cellular processes, such as proliferation, phagocytosis, or oxidative burst capacity, appeared largely unaffected in these patients [rows 66–71].
Functional studies
The subset of published variants were functionally evaluated in vitro predominantly in HEK293 cells, while V82I was modeled in silico. Assay data was available for 3 ORAS/IMD107 variants (G174Dfs*2, Y244C, EX6 + 2T > C), 4 ORAS-specific variants (M86I, W167S, L272P, G281R), 6 IMD107-specific variants (E95*, D246V, P254S, R263Q, D268Tfs*6, N341D), and 1 DN-ORAS variant (C129S) (Fig. 4, Table S4).
All tested missense variants resulted in normal OTULIN mRNA levels [Row 1], and nearly all in WT-like protein levels except for L272P and W199R [Row 2]. Protein levels for these latter two appeared to be reduced in some [13] but not other assays [32, 39], possibly as a consequence of reduced thermal stability. Stop-gain, frameshift and splice-altering variants all resulted in reduced or nearly absent protein levels.
In silico 3D structural modeling suggested that all variants examined led to aberrant protein conformations [Row 3], while 3/5 were also predicted to reduce protein thermal stability (W167S, L272P, G281R) [Row 4]. Predicted protein half-life was normal for 2 variants (M86I, W167S) [Row 5]. OTULIN interaction with LUBAC through HOIP was normal for 1 over-expressed frameshift variant (G174Dfs*2) and 2 missense (Y244C and L272P) variants [Row 6] [32]. By contrast, OTULIN binding affinity for M1-Ub chains was shown or predicted to be reduced for all but 2 DN-ORAS variants (C129A, C129S) [Row 7] [9].
A common consequence of almost all variants tested was the loss or reduction of OTULIN’s DUB activity with consequent accumulation of Met1-Ub chains and increased NF-𝛋B signaling, though there are some nuances in the data for Y244C (see Discussion). Two IMD107 variants (D246V and P254S) showed normal DUB activity and M1-Ub chain levels, but reduced inhibition of NF-𝛋B activity [Rows 8–9]. Presumed benign variant Q115H as well as the rare variant c.1033dup; p.(R345Kfs*4) showed WT-like levels of NF-𝛋B inhibition [Rows 10–12]. As opposed to the other frameshift variants (p.G174Dfs*2, p.D268Tfs*6), p.(R345Kfs*4) appears to lead to WT level production of OTULIN protein. Therefore, this variant may be an isomorph, though its function was only assessed in a single assay and other assays might show other uninterrogated functional abnormalities.
Caspase-like, but not chymotryptic-like or tryptic-like, proteasome activity was found to be decreased in cells expressing C129A or L272 [Rows 13–15], while C129S expression in THP-1 cells increased some inflammatory gene expression (i.e. IFNB1 and IL6, but not IRF7 or TNF) [Rows 16–19]. This supports observations in patient cells suggesting activation of multiple streams of inflammatory signaling, including type I IFN.