The impact of variants found in family 8 was determined by using a Pigc-deficient cell model to assess if they lead to a loss-of-function. cDNA of PIGC with variants p.(C265Ffs*11) and p.(G96R) were cloned into a pME expression vector. This vector is driven by a strong SRα promoter. By electroporation, plasmids were transfected into T1M1-Thy1(-)c (TIMI) cells [23]; these are Pigc-defective murine lymphoma cells. Transfection efficiency was monitored by co-transfecting luciferase-expression vector. Pathogenicity of the two variants was studied by assessing the restoration of the surface expression of GPI-APs by flow cytometry using mouse CD90 (PharMingen, San Diego, CA, USA), FLAER (Cedarlane, Burlington, NC, USA) and mouse CD48 (BioLegend, San Diego, CA, USA).
We ascertained 18 probands from 10 unrelated families, each with biallelic pathogenic variants in PIGC (Fig. 1A). The phenotypic details are summarized in Table 1, while genotypic details, including variant interpretation, are summarized in Table 2.
Exome sequencing identified eight novel variants in PIGC, illustrated in Fig. 2: homozygous c.437_445del, p.(T146_T148del) in P1-11, c.859 G > T, p.(E287*) in P12, c.77 A > G, p.(D26G) in P13, compound heterozygous variants c.422 C > T, p.(T141I) and c.138 C > A, p.(Y46*) in P14, compound heterozygous variants c.794_796delinsT, p.(C265Ffs*11), and c.286 G > A, p.(G96R) in P15-16, compound heterozygous variants c.77 A > T, p.(D26V) and c.286 G > A, p(G96R) in P17, and homozygous, potential start-loss variant c.3 G > A, p.(M1?) in P18. Sanger sequencing confirmed these variants in the probands, and parental testing revealed heterozygosity.
All variants are rare in frequency databases (allele frequency ranging from 0 to 1.922 × 10 in gnomAD v4.1.0); none have been reported in homozygous state in healthy probands. In silico analysis consistently predicted these variants as pathogenic (Table 2).
Due to the absence of experimental structural data, we used a predicted AlphaFold2 model to analyze disease-associated mutations in PIGC. The predicted structure reveals eight transmembrane helices (TMHs), with both N- and C-termini located in the cytosol. Apart from unstructured N- and C-terminal regions, PIGC is largely devoid of loop domains and is thus embedded in the membrane (Fig. 3A).
Nonsense mutations, including those at Tyr46 and Arg21, as well as a frameshift mutation at Cys265, are expected to result in truncated, non-functional proteins. However, the nonsense mutation at Glu287 produces a protein with a relatively short truncation of only 10 residues, which is unlikely to disrupt overall folding. Thus, this residue might play a role in mediating PIGC's interactions with other subunits of the GPI-GnT complex. Similarly, the D26G/D26V mutations could also impact subunit binding.
The deletion mutation Δ146-148 likely disrupts protein folding, as these residues are positioned at the start of TMH4. The short preceding loop connecting TMH3 and TMH4 is likely insufficient to compensate for this deletion, impairing structural integrity.
Mutations within transmembrane regions likely induce steric clashes, leading to destabilization of PIGC. Gly96 on TMH2, which packs against TMH6, if replaced by bulky, positively charged arginine, could cause steric clashes with adjacent residues, such as Leu216 and Trp213 (Fig. 3B). Thr141 on TMH3 packs closely with TMH5 and TMH6 at the membrane periphery; replacement with isoleucine may introduce clashes with nearby residues including Cys188, Val187, and Ser191 (Fig. 3C). Leu189 on TMH5 is in proximity to aromatic residues Trp72 on TMH1 and Tyr150 on TMH4. Replacement with tryptophan at this position could cause significant steric clashes (Fig. 1D). Finally, Leu212, located in the middle of TMH6, when mutated to proline, is expected to disrupt α-helical structure and cause misfolding.
The majority (16/18, 89%) of probands had profound ID/DD, being nonverbal and unable to walk or sit unsupported. Proper head control, social smiling, and fixing-and-following were frequently absent. For proband P2, clinical assessment was limited due to her passing at 1 month of age, possibly secondary to pneumonia. However, as the vast majority of our cohort, she was severely underweight (12/18, 67%) and hypotonic (15/18, 83%). Only two probands, P14 and P17, had ID/DD described as mild or moderate. Proband P14 had mild ID and the ability to communicate using few sentences by the age 26 years; she was able to sit by 6 months and walk by 17 months. P14 and P17 represented 2/4 of probands with compound heterozygous variants, while all probands with homozygous variants had severe to profound ID/DD. These findings suggest that the PIGC-related encephalopathy is characterized by severe neurodevelopmental impairment, but moderate or even mild phenotypes are also possible.
Developmental anomalies were seen in all proband but one proband (P17), who underwent cerebral MRI (14/15). These included cerebellar atrophy (11/14), prominent cortical and/or subcortical volume loss (8/14), abnormal corpus callosum (7/14), hypomyelination (4/14), and enlarged ventricles (4/14). An abnormal signal in the central tegmental tracts was detected in P13 and P15. Additionally, proband P18, exhibited widening of temporal horns and narrowing of superior peduncles. These findings suggest that the PIGC-related encephalopathy is tightly linked to abnormal findings in cerebral MRI.
Movement disorders were present in our cohort, although not previously described. Dystonia was noted in P4 and P9, while P13 exhibited dyskinetic movements in both upper and lower limbs. Spasticity was not noted in any of our cases, but some probands had brisk reflexes. Proband P15 had upper limb hypertonia and displayed dystonic movements of the arms. P4 also suffered from breath-holding spells, while P15 had recurrent respiratory tract infections and breathing difficulties. These findings suggest that the PIGC-related encephalopathy can present with additional neuromotor anomalies.
Seizures were reported in all probands who survived beyond the first month of age, except for P16. Seizures began at a mean age of 7 months (range 1 month to 18 months) (Table 1). Nine probands presented with focal epilepsy, two probands with generalized seizures, and the rest had either combined or unclassified epilepsy. Only 2/17 probands had isolated focal seizures, while 14/17 had focal to bilateral or generalized tonic-clonic seizures. Atonic and myoclonic seizures were noted in 3/17 and 6/17 probands, respectively. Proband P14 had atypical absences, and P5 had isolated epileptic spasms. Over the follow-up period, seven probands developed additional seizure types.
Focal tonic seizures could be provoked by fever, typically lasting 3-10 min and occurred several times weekly. Atonic seizures had an onset around 6-8 months of age and typically lasted a few seconds; they occurred several times daily without known triggers. The myoclonic jerks had an onset between 4 and 12 months of age, occurring several times daily, with frequency increasing due to fever. Convulsive status epilepticus occurred in at least 9/16 probands, often in the first 3 years of life.
EEG recordings were available for 13/16 probands with seizures, revealing interictal EEG patterns from an early age. They typically showed focal or diffuse background slowing with interictal multifocal epileptic discharges, particularly in the frontotemporal or posterior regions. While multifocal discharges were seen in 11/13 probands, P1 had unifocal epileptiform discharges. In those classified with generalized or combined focal and generalized epilepsy, generalized ictal spike-wave activity was recorded during seizures. In P15, EEG showed hypsarrhythmia, which improved after treatment but remained encephalopathic.
All probands with epilepsy experienced drug-resistant and ongoing seizures, contrasting with previously published cases where 2/5 (A-II-2 and A-II-4) were responsive to antiseizure medications (ASM) (Table 1). No single ASM proved consistently effective across multiple probands. While some ASMs appeared beneficial for certain probands, they were ineffective or exacerbated seizures in others. Levetiracetam, vigabatrin and topiramate were reported to reduce seizure frequency by at least 50% for a period of at least 6 months. None of the probands were placed on a ketogenic diet or given pyridoxine. Proband P15 developed infantile spasms at 2 months old and was treated with prednisolone and vigabatrin. Vigabatrin was stopped due to drowsiness and bradycardia. She was initially started on levetiracetam, which was eventually replaced with valproate and topiramate. Clobazam proved most helpful in controlling her seizures and she remained on a low-maintenance dose (5 mg twice a day). These findings suggest that epileptic activity in the PIGC-related encephalopathy tends to be severe and complex to treat.
Out of the 18 probands, 10 were deceased. The age at death ranged from 40 days to 7 years of life (median age of 40 months). Nine probands died within the first four years of life (Table 1). The causes of death were due either to respiratory failure (2/10) or possible sudden unexpected death in epilepsy (SUDEP) (8/10). These findings suggest that high and early mortality should be a concern among patients with PIGC-related encephalopathy.
Dysmorphic features were noted in nearly all probands (Fig. 1B-F). These included prominent/high forehead, deep-set eyes, large and low-set ears, wide mouth with full lips, full cheeks and pointed chin. Skeletal anomalies were also common, such as scoliosis (P5, P13 and P15), pectus excavatum (P5), joint laxity (P6, P9 and P12), clinodactyly (P4), arachnodactyly (P5 and P9), and pre-axial polydactyly (P15). Additionally, three probands presented with dysplastic fingernails (P4, P9 and P14). Anteriorly placed anus and hydronephrosis were noted in probands P15-16. These findings suggest that dysmorphisms are predominant in the PIGC-related encephalopathy.
Flow cytometry analysis revealed a 53% and 42% decrease in FLAER and CD16 levels in granulocytes of proband P12 (Fig. 4A). Although P12 had a nonsense variant, p.E287*, the premature stop codon was at the end of the last exon, likely allowing the transcript to escape NMD and produce a truncated protein, resulting in reduced surface expression of GPI-APs.
In proband P13 granulocytes, we showed a decrease of 63% in FLAER levels and 68% in both CD55 and CD16 (Fig. 4B). Proband P14 granulocytes showed reductions in FLAER, CD24 and CD16 to 56%, 42%, and 49% compared to controls, respectively (Fig. 4C). In fibroblasts from proband P11, levels of FLAER, CD73 and CD109 were decreased to 51%, 80%, and 45% compared to healthy probands (data not shown).
Functional analysis of variants in knockout cells is an alternative option to study variant consequences, offering insights when proband cells are unavailable. We performed functional analysis on two variants using Pigc-deficient TIMI cells, focusing on a newly reported missense variant, and a frameshift variant found in family 8. The experiments demonstrated that the mutants p.(C265Ffs*11) showed no activity, while the p.(G96R) mutant had decreased activity compared to the wildtype protein (Fig. 4D, E). Overall, these findings suggest that variants leading to PIGC-related encephalopathy are linked to decreased GPI cellular signal.