Thrombotic Microangiopathy in Inverted Formin 2–Mediated Renal Disease

Abstract

The demonstration of impaired C regulation in the thrombotic microangiopathy (TMA) atypical hemolytic uremic syndrome (aHUS) resulted in the successful introduction of the C inhibitor eculizumab into clinical practice. C abnormalities account for approximately 50% of aHUS cases; however, mutations in the non-C gene diacylglycerol kinase-ε have been described recently in individuals not responsive to eculizumab. We report here a family in which the proposita presented with aHUS but did not respond to eculizumab. Her mother had previously presented with a post–renal transplant TMA. Both the proposita and her mother also had Charcot–Marie–Tooth disease. Using whole-exome sequencing, we identified a mutation in the inverted formin 2 gene (INF2) in the mutational hotspot for FSGS. Subsequent analysis of the Newcastle aHUS cohort identified another family with a functionally-significant mutation in INF2. In this family, renal transplantation was associated with post-transplant TMA. All individuals with INF2 mutations presenting with a TMA also had aHUS risk haplotypes, potentially accounting for the genetic pleiotropy. Identifying individuals with TMAs who may not respond to eculizumab will avoid prolonged exposure of such individuals to the infectious complications of terminal pathway C blockade.

Genetic abnormalities in the alternative pathway of C have been demonstrated to account for many cases of the thrombotic microangiopathy (TMA) atypical hemolytic uremic syndrome (aHUS) (MIM 235400).¹ Understanding the role of C in the pathogenesis of aHUS has resulted in the successful introduction of the C inhibitor eculizumab into clinical practice.²

Recently, mutations in the non-C gene DGKE have been demonstrated to be associated with aHUS (MIM 615008).³ Individuals with DGKE mutations display phenotypic variability with some patients presenting with membranoproliferative GN.⁴ As might be expected with a genetic cause which does not appear to be C-mediated, individuals have not responded to treatment with eculizumab.³

Despite recent advances, the genetic basis of many cases of familial aHUS remain unsolved. In this study, we describe the finding of inverted formin 2 gene (INF2) mutations in two families with TMA.

Our index case, patient III:2, presented aged seven with pex cavus and difficulty walking and was diagnosed with Charcot–Marie–Tooth (CMT) (Figure 1). Aged 15 she presented 5 days after a sore throat with microangiopathic hemolytic anemia on blood film (Hb, 9.4 g/dl), thrombocytopenia (platelets 119 × 10⁹/L), and renal failure (creatinine 879 µmol/L). Haptoglobins were undetectable, lactate dehydrogenase was 844 U/L, and there was proteinuria (4.8 g/L). BP on admission was 185/110 mmHg. Hemodialysis was commenced at presentation and plasma exchange was undertaken before commencement of eculizumab. Initially, there was an improvement in the platelet count to 173 × 10⁹/L, but subsequently this fell to 100 × 10⁹/L. A bone marrow biopsy was unremarkable and trough eculizumab concentration was adequate with a completely suppressed CH50. Three months after presentation a renal biopsy was undertaken demonstrating characteristic changes of a thrombotic microangiopathy (Figure 2A). There were also features of a distinct glomerulosclerosis with small glomeruli and arteriosclerosis (Figure 2B). After 9 months with no renal recovery eculizumab was withdrawn.

Figure 1.

Pedigrees of families with INF2 genetic variants. The pedigrees demonstrate the segregation of the renal/neurologic phenotype with the rare genetic variant, c.305T>A (p.V102D) in family 1 (A), and c.530G>A (p.R177H) in family 2 (D). Individuals tested but not carrying the mutation are shown (nmd, no mutation detected). The number of alleles carrying the aHUS risk haplotype CFH-H3 (CFH_risk) and CD46_GGAAC (CD46_risk) are shown on the pedigree. Sanger sequencing trace of wild type (WT) and mutant (Mut) for c.305T>A (p.V102D) (B) and c.530G>A (p.R177H) (E). Alignment of human, chimpanzee, orangutan, mouse, rat, dog, opossum, platypus, and zebrafish INF2 demonstrating amino acid conservation (C and F) (performed using http://genome.ucsc.edu/cgibin/hgTrackUi?hgsid=309786867&c=chr21&g=cons46way#a_cfg_phyloP).

Figure 2.

Thrombotic microangiopathy in renal biopsy from patients with INF2 mutations. Renal biopsies. Native renal biopsy from family 1, patient III:2. (A) Two arterioles (right) showing features of active thrombosis and a small artery (left) with relatively slight intimal edema with fibrosis (hematoxylin and eosin stain [H&E]). (B) Glomerulus (left) showing global sclerosis and occluded arteriole (right) (periodic acid–Schiff [PAS]). Native renal biopsy from family 1, patient II:2 demonstrating end-stage changes with diffuse global sclerosis (C). Renal transplant biopsy from family 1, patient II:2 demonstrating (D) an occluded arteriole (PAS) and (E) a capillary loop with abundant subendothelial fluffy material (electron microscopy). Native renal biopsy from family 2, patient III:2 demonstrating (F) a sclerosed glomeruli and (G) a segmental sclerosing lesion. Renal transplant biopsy from family 2, patient III:1. Mucoid intimal thickening is seen in an interlobular artery with red cell fragmentation in the wall and luminal thrombus (H). Mesangiolysis is also seen 3–6 o’clock (I) (silver stain). Native renal biopsy from family 2, patient III:2 showing a subacute/chronic arterial TMA with fibroproliferative obliteration of small arteries and arterioles (J) (H&E) and (K) (trichrome). Renal transplant biopsy from family 2, patient III:2 demonstrating end stage change with fibrous obliteration of arteries. (L) (H&E).

Screening for known inherited and acquired causes of aHUS did not reveal any abnormality.¹ The C5 variant c.2654G>A (p.R885H), which impairs eculizumab efficacy, was not present.⁵

Family history revealed that the proposita’s mother, patient II:2 (Figure 1), also had CMT, had presented with ESRD aged 17, and had a post-transplant TMA (Figure 2, D and E) (case history, Supplemental Material).

Because of the absence of an abnormality in a known aHUS-associated gene we sequenced the exomes of the two affected individuals in this family.

This revealed a rare variant in INF2, c.305T>A (p.V102D) (Figure 1). Mutations in INF2 are the commonest cause of familial autosomal dominant nephrotic syndrome.^6–8 In a minority of these cases the mutations cause a syndromic form of FSGS associated with the demyelinating peripheral neuropathy, CMT^9,10 (Figure 3).

Figure 3.

Hotspots for inverted formin 2 (INF2) variants in FSGS, CMT, and aHUS. A representation of the domain structure of INF2 showing the DID domain, formin homology domains (FH1, FH2), and the DAD domain. Genetic variants associated with isolated FSGS are shown below the domain structure. Genetic variants with a combined FSGS/CMT phenotype are shown above the domain structure. Variants from Mademan et al.⁵⁷ The locations of the aHUS-associated mutations demonstrated in the study are shown in red.

To assess whether mutations in INF2 account for other cases of familial or sporadic TMA previously referred to the Newcastle aHUS center,¹¹ we analyzed 28 familial cases by exome sequencing and undertook Sanger sequencing on an additional 161 sporadic aHUS cases. In one family, in which no known genetic risk factors had been found, we identified a functionally-significant mutation in INF2, c.530G>A (p.R177H), which segregated with the disease (Figures 1 and 2, Supplemental Material). No INF2 variants were identified in sporadic aHUS cases.

INF2 is a ubiquitously-expressed formin protein¹² which accelerates actin polymerization and depolymerization, thus regulating a range of cytoskeleton-dependent cellular functions including the secretory pathway.^13,14 INF2 comprises formin homology 1 and 2 domains, an N-terminal diaphanous inhibitory domain (DID), and a C-terminal diaphanous autoregulatory domain (DAD).¹⁵ Mutations in INF2 predominate in the DID domain.^6,7,10 Functional analysis of INF2 mutations in disease has demonstrated disorganized cytoskeletal functions^7,9 although the precise mechanism of disease remains elusive.

The two INF2 variants we describe here reside in the mutational hotspot for disease.^6–10 The c.305T>A (p.V102D) variant resides in exon 2 whereas c.530G>A (p.R177H) resides in exon 4 (Figures 3 and 4). Structural modeling reveals that the p.V102D variant is in close proximity to the DAD binding region. Modeling does not predict a surface-exposed residue but instead the variant may be expected to disrupt the architecture of the eighth α-helix of the DID domain (Figure 4). The p.R177H variant resides before the 13th α-helix of the DID domain and is surface-exposed (Figure 4). Amino acids at both these positions are conserved across species with GERP++ scores of 4.76 (p.V102D) and 4.48 (p.R177H) (Figure 1).

Figure 4.

Predicted structure of DID domain of inverted formin 2 (INF2). A structural model of the DID domain (amino acids 1–234) was generated using Phyre2 (A).⁵⁸ Blue spheres represent the amino acids involved in binding to DAD domain: (R106, N110, A149, I152). The position of the p.V102D and R177H mutations are highlighted in red. The p.V102D lies in the eighth α-helix of the DID (amino acids) domain. (B) Surface representation of the modeled structure highlighting the surface-exposed amino acids responsible for DAD binding (R106, N110, A149, I152). The mutant p.V102D is not buried but may be expected to disrupt the architecture of the eighth α-helix. The p.R177H variant resides before the 13th α-helix of the DID domain and is surface-exposed. (C) Semi transparent representation of the surface of the INF2 model highlighting the variants.

Disease-causing mutations in FSGS have mainly been found to occur in the DID domain, in exons 2–4, with only one report of a mutation in exon 6 (Figure 3).^6–8,16 Within this hot-spot there is a cluster between nucleotides 300 and 500 which accounts for those with FSGS and CMT.^9,10 The p.V102D mutation resides in this region and this family have CMT, whereas the p.R177H mutation resides downstream of this region and this family has no neurologic phenotype. The p.R177H mutation has previously been reported in three unrelated pedigrees and in all cases had the nonsyndromic form of FSGS.^6,8 Functional analysis of this mutation demonstrated altered INF2 localization and disruption of the actin cytoskeleton.⁹

It is well reported that even in individual families with the same INF2 mutation there is phenotypic variability. Most commonly, individuals present with disease in adolescence with mild proteinuria, developing ESRD in the third or fourth decade, although individuals have been reported to be unaffected into their sixth or seventh decade.¹⁶ Variable intrafamilial penetrance has also been reported for the neurologic phenotype.¹⁶ The clinical and pathologic disease pleiotropism we describe with INF2 mutations is also seen in individuals with recessive DGKE mutations, where some individuals present with proteinuria and progressive renal failure whereas others present with aHUS. Likewise, the biopsy findings in DGKE-associated disease are also heterogenous ranging from a membranoproliferative pattern to a TMA. These findings can vary according to the time of presentation. It is only with genetic analysis that the underlying pathologic process can be identified and a therapeutic intervention sought. It has been suggested that genetic background or environmental factors modify the penetrance and phenotype of disease.^16,17 It is interesting to note that in family 1 both affected members, II:2 and III:2, were homozygous for the aHUS at-risk CFH-H3 haplotype,¹⁸ and both carried one risk CD46 allele.¹⁹ In family 2, patient III:1 was homozygous for the CD46_GGAAC risk haplotype whereas III:2 was heterozygous. III:1 also carried one copy of the CFH-H3 haplotype (Figure 1).

Although INF2 mutations have not previously been associated with a renal thrombotic microangiopathy, aHUS has been reported in patients with primary FSGS,^20–23 and FSGS is a frequent pathologic sequalae of sporadic Stx-HUS.²⁴ A thrombotic microangiopathy has also been associated with other causes of nephrotic syndrome^25,26 and primary GN, including IgA nephropathy,^27–29 Henoch Schonlein Purpura,²³ ANCA-associated vasculitis,^22,23 and anti-GBM nephropathy.²³

It has been hypothesized that either direct or indirect (via impaired VEGF secretion from podocytes) endothelial injury leads to a constricted microvasculature with perturbed hemodynamic flow, leading to the formation of platelet microthrombi and a thrombotic microangiopathy. Loss of coagulation regulators with upregulation of procoagulation factors has also been suggested as a contributory factor in those individuals with coexistent nephrotic syndrome.^30–33

It is intriguing that all three patients who had renal transplants had biopsy-proven evidence of a thrombotic microangiopathy in their renal allografts. The risk of FSGS recurrence post-transplant in those with genetic defects of the glomerular filtration barrier is low due to correction of the underlying defect.^34–36 An exception to this is in those individuals with complete deficiency of NPHS1 due to the presumed generation of antibodies against this immunologically novel protein in the allograft. Such a scenario would not be expected in a dominantly-inherited condition.

Currently, there is little information available as to the recurrence of FSGS post-transplantation in individuals carrying INF2 mutations. However, in one small study recurrence was seen in one of three individuals.⁸ INF2 is expressed ubiquitously¹² and recurrence of FSGS in an allograft suggests that a circulating factor or cell type is predisposing to recurrent disease. Such a factor may account for post-transplant TMA. It should be noted that INF2 has been demonstrated to complex with and alter the intracellular transport of the C regulators CD55 and CD59, which are present on all circulating cells including platelets. We cannot, however, rule out the possibility that the TMA was a consequence of the post-transplant milieu (e.g., viral diseases, ischemia reperfusion injury, donor-specific antibodies, immunosuppressive drugs).

In summary, we describe two families with mutations in INF2 in addition to common aHUS risk haplotypes who present with aHUS or a post-transplant TMA. Eculizumab was unsuccessful in preventing either ongoing TMA or ESRD as is seen with other non–C-mediated causes of aHUS. Identifying individuals who will not respond to eculizumab will avoid exposing these individuals to the infectious risks of terminal pathway C blockade. This study represents an initial application of whole-exome sequencing in personalized management of TMA.

Concise Methods

The study was approved by North East–York Research Ethics Committee, and informed consent was obtained in accordance with the Declaration of Helsinki.

C Assays

C3 and C4 levels were measured by rate nephelometry (Beckman Coulter Array 360). Factor H levels were measured by radial immunodiffusion (Binding Site). Screening for C autoantibodies was undertaken using ELISA as described previously.^37,38

Genetic Analysis and Multiplex Ligation–Dependent Probe Amplification

Mutation screening of CFH,³⁹ CFI,⁴⁰ CFB,⁴¹ MCP,⁴² C3,⁴³ and DGKE³ was undertaken using Sanger sequencing as previously described. Screening for genomic disorders affecting CFH, CFHR1, CFHR2, CFHR3, CFHR5, CFI, and CD46 was undertaken using multiplex ligation-dependent probe amplification.^44,45 Mutation screening of INF2 was undertaken using Sanger sequencing using the primer conditions in Supplemental Table 5.

Whole-Exome Sequencing

Enrichment from isolated DNA was performed using either Illumina Nextera Rapid Capture Exome by AROS AB (family 1) or Agilent SureSelect^XT Human All Exon V5 by GATC Biotech, Konstanz (family 2, III:1) as described previously.⁴⁶ Library preparation was performed postcapture, with adaptor sequences and indexing incorporated using proprietary methods of AROS AB and GATC Biotech, compatible for Illumina sequencing technology. Illumina sequencing was performed on the HiSeq2000 instrument (v3 chemistry) (Supplemental Table 2).

The quality of sequencing reads was firstly checked with FastQC.⁴⁷ Duplicated reads were removed with FastUniq.⁴⁸ The remaining reads were mapped to the human reference genome GRCh37 with BWA.⁴⁹ The alignments were refined with tools of the GATK suite.⁵⁰ Variants were called according to GATK Best Practice recommendations,^51,52 including recalibration. Freebayes was also used to call variants from the same set of samples.⁵³ The variants called by Freebayes with total coverage ≥5, minor allele coverage ≥5, and variants call quality ≥20 were added to those identified by GATK. Annovar was used for annotations and prediction of functional consequences.⁵⁴ Variants identified in family 1 were filtered as detailed (Supplemental Table 2). First, we selected for variants in high-effect regions and selected variants at a minor allele frequency <5% in 1000G and ESP6500. We then selected those variants segregating in a dominant fashion with disease. Variants predicted to be deleterious by Polyphen-2 HDIV and HVAR, Mutation Taster, Mutation Assessor, FATHMM, or RadialSVM were selected for further analysis. A more stringent minor allele frequency cut-off of <0.1% in 1000G and ESP6500 was applied and nonconserved variants (<2 by GERP++ and <0.5 by PhyloP) were discarded (Supplemental Table 3). Phenotypic data were then used to interrogate the remaining 34 genes providing only one candidate gene known to have both renal and neurologic conditions inherited in an autosomal dominant pattern (Supplemental Table 4).

Protein Modeling

Phyre2 was used to generate an approximate protein structure using the inputted amino acid sequence of INF2 (NP 071934.3, amino acids 1–250) using the intensive modeling mode. Protein domain boundaries for INF2 were taken from Pfam.⁵⁵ Three-dimensional protein structures were manipulated using PyMOL.⁵⁶

Disclosures

Newcastle University has received funding from Alexion Pharmaceuticals, Cheshire, UK, for consultancy work undertaken by T.H.J.G and D.K. D.K. is scientific advisor to Gyroscope Therapeutics, London.