Clinical, neuropathological, and biochemical characterization of ALS in a large CHCHD10 R15L family

Abstract

Familial forms of ALS are potential candidates for gene-directed therapies, but many recently identified genes remain poorly characterized. Here, we provide a comprehensive clinical, neuropathological, and biochemical description of fALS caused by the heterozygous p.R15L missense mutation in the gene CHCHD10. Using a cross-sectional study design, we evaluate five affected and nine unaffected individuals from a large seven-generation pedigree with at least 68 affected members. The pedigree suggests a high (68 - 81%) but incomplete disease penetrance. Through cloning of the disease-allele from distant members of the family, we establish the disease haplotype in the family. Notably, the haplotype was distinct from that of a previously reported p.R15L mutation carrier with ALS, demonstrating that the variant is in a mutational hotspot. The clinical presentation was notable for being highly stereotyped; all affected individuals presented with the rare ALS variant Flail Arm Syndrome (FAS; also known as, brachial amyotrophic diplegia or Vulpian-Bernhardt Syndrome), suggesting greater involvement of the cervical spinal cord. Consistently, neuropathology from one family member demonstrated substantially increased CHCHD10 protein aggregation and neuronal loss (though absent TDP-43 pathology) in the cervical vs. lumbar spinal cord. This FAS phenotype could be captured by a simple timed finger tapping task, suggesting potential utility for this task as a clinical biomarker. Additionally, through analysis of fibroblast lines from 12 mutation carriers, isogenic iPSC cells, and a knockin mouse model, we determined that CHCHD10 with the R15L variant is stably expressed and retains substantial function both in cultured cells and in vivo, in contrast to prior reports. Conversely, we find loss of function (LoF) variants are more common in the population but are not associated with a highly penetrant form of ALS in the UK Biobank (31 in controls; 0 in cases). Together, this argues against LoF and in favor of toxic gain-of-function as the mechanism of disease pathogenesis, similar to the myopathy-causing variants in CHCHD10 (p.G58R and p.S59L). Finally, through proteomic analysis of CSF of variant carriers, we identify that CHCHD10 protein levels are elevated approximately 2-fold in mutation carriers, and that affected and unaffected individuals are differentiated by elevation of two neurofilaments: neurofilament light chain (NfL) and Peripherin (PRPH). Collectively, our findings help set the stage for gene-directed therapy for a devasting form of fALS, by establishing the likely disease mechanism and identifying clinical and fluid biomarkers for target engagement and treatment response.

Keywords: CHCHD2; Lou Gehrig’s Disease; coiled-coil-helix-coiled-coil-helix domain containing 10; mitochondrial disorders; motor neuron disease; motor neurone disease.

Figure 1.. Pathogenic CHCHD10 R15L variant in large family with familial ALS.

(A) Simplified pedigree shows segregation of the heterozygous CHCHD10 p.R15L variant in a large family. Solid black symbols indicate those affected with ALS. Red dots indicate those testing positive for the heterozygous CHCHD10 p.R15L variant. (B, top) Common polymorphisms surrounding the pathogenic CHCHD10 p.R15L variant (red font) are shown for the indicated individuals from branches of the USALS#5 family and a sporadic ALS patient that was previously reported (Canadian Individual 1). Haplotype was established by topo cloning of each allele following long-range PCR of genomic CHCHD10 and Sanger sequencing. Common single nucleotide polymorphisms (SNPs) not matching the haplotype of the USALS#5 have an orange background. The haplotype surrounding p.R15L for Canadian Individual 1 is distinct from that surrounding distantly related members of the USALS#5 family. (B, bottom) the location of two common SNPs surrounding p.R15L are indicated (orange).

Figure 2.. CHCHD10 R15L ALS presents clinically as Flail Arm Syndrome.

(A) Pattern of weakness on confrontational motor testing in affected participants is represented by MuscleViz visualization. Scoring follows the Medical Research Council (MRC) scale for muscle strength grading. (B) Upper extremity motor function for presymptomatic (PS) and symptomatic (S) CHCHD10 p.R15L mutation carriers, as measured by the number of finger taps in 10 seconds (top) and degree of arm swing during 25-foot instrumented walk over 4 trials (bottom). (C) Lower extremity motor function as measured by the number of foot taps in 10 seconds (top) and length of stride during 25-foot instrumented walk over 4 trials (bottom). In (B and C) data from left and right extremities appear as solid and open data points, respectively. (D) Gait speed on a 25-foot instrumented walk over 4 trials (top) and instrumented timed up and go (iTUG) (bottom). Gait speed is the average of 4 trials and iTUG is the average of two trials. (E) Articulatory coordination as measured using the PaTaKa speech test. Number of repetitions of the syllables Pa, Ta, and Ka in five seconds is represented. Two trials were averaged. (F) Cognition (top) and executive function (bottom) measured using the ALS-CBS and UDS3-EF, respectively. Dotted line (top) indicates the cutoff value for normal (≥ 17) in this assessment. Dotted line (bottom) indicated the average score in the NIH ALS cohort. (G) Axial T1-weighted MRI and FreeSurfer brain reconstruction highlighting the precentral gyrus (dark blue) and parietal lobe (light blue). Axial T2-weighted MRI from a presymptomatic participant (gray box) and four symptomatic participants. Precentral gyrus hypointensity (arrowhead) and adjacent white matter hyperintensity (arrow) is observed. (H) Precentral gyrus thickness (left) and gray-matter volume as a percentage of the total intracranial volume (TIV) (right). Data from right and left hemispheres appear as solid and open datapoints, respectively. (I) Parietal cortex gyral thickness (left) and parietal gray matter volume as a percentage of the TIV (right). Data from right and left hemispheres appear as solid and open datapoints, respectively. In all panels, ns, *, **, ***, ***” correspond to not-significant, p ≤ 0.05, p ≤ 0.01, p ≤ 0.001, and p ≤ 0.0001, respectively. Statistics were calculated using the Mann-Whitney test.

Figure 3.. CHCHD10 pathology in spinal cord and motor cortex.

(A) Spinal cord section from a non-neurological control showing weak diffuse cytoplasmic CHCHD10 immunohistochemistry. Scale bar 100 μm. (B) Cervical spinal cord showing scattered neuropil threads (arrows) and neuronal cytoplasmic inclusions in the soma of the residual motor neurons (arrowhead). Scale bar 100 μm. (C) Lumbar spinal cord showing rare threads (arrow) and neuronal cytoplasmic inclusions in the soma of the residual motor neurons (arrowhead). (D) High-power image of lumbar spinal cord highlighting fibrillar CHCHD10-positive aggregates in motor neurons (arrows). Scale bar 0 μm. (E) High-power image of lumbar spinal cord highlighting rare round and “glassy” CHCHD10-positive aggregates (arrowhead). Scale bar 100 μm. (F) High-power image of layer V of frontal cortex from a non-neurological control showing weak diffuse cytoplasmic staining in neurons. Scale bar 100 μm. (G) High-power image of layer V of motor cortex showing diffuse neuropil threads (arrow) and occasional neuronal cytoplasmic aggregates (arrow head). Scale bar 0 μm. (H) High-power image of layers II/III of motor cortex showing diffuse neuropil threads and neuronal cytoplasmic aggregates. Scale bar 0 μm. (I) Low-power image of motor cortex showing diffuse neuropil threads and neuronal cytoplasmic aggregates.

Figure 4.. CHCHD10 R15L is stably expressed and partially functionally in human cells.

(A and B) CHCHD10 protein levels were measured by immunoblotting from primary dermal fibroblast cell lines. Lines were established from 12 CHCHD10 p.R15L mutation carriers (5 affected and 7 unaffected), 5 healthy controls, and 7 unrelated disease controls (with idiopathic Parkinson’s disease). (C - E) CHCHD10 protein levels (C and D) and OPA1 cleavage by OMA1 (C and E) were measured by immunoblotting for isogenic WT, CHCHD10 KO, and CHCHD10 R15/R15L iPSC cell lines. OMA1 activation is reflected by the cleavage of L-OPA1 (a and b isoforms) to specific S-OPA1 isoforms (c and e isoforms). Statistical analysis was performed using the Brown-Forsythe ANOVA test followed by Dunnett’s T3 multiple comparisons test. (F) Schematic demonstrates the method used to generate induced Lower Motor Neuron like cells (iLMNs) from iPSCs through the doxycycline induced expression of transcription factors NGN2, ISL1, and LHX3. (G) Immunoblotting of isogenic WT, CHCHD10 indel, and CHCHD10 R15/R15L iLMNs after 9 days in vitro. Data is representative of two separate differentiations. (H) Representative confocal image of CHCHD10 in iLMNs from isogenic WT and R15L/R15L iLMNs. Mitochondria are marked by cytochrome c (CYCS) and neuronal processes by β3-Tubulin (TUBB). Scale bars = 10 μm. Data is representative of N = 2 wells from a single differentiation. (I and J) Volcano plot of affinity purification mass spectrometry data following immunoprecipitation of endogenous CHCHD10 from WT (left) and R15L/R15L (right) iPSC cells compared to CHCHD10 indel cells. Dark blue datapoints were significant interactors (FDR ≤ 0.05 and log2 fold change ≥ 1) of CHCHD10 in WT iPSC cells; light blue datapoints were significant interactors of CHCHD10 in R15L/R15L iPSC cells. Data is from N ≥ 3 wells of cells cultured and processed in parallel. In all panels, ns, *, **, ***, ***” correspond to not-significant, p ≤ 0.05, p ≤ 0.01, p ≤ 0.001, and p ≤ 0.0001, respectively. For (Fig. 3D an DE) an ordinary one-way ANOVA test followed by Dunnett’s multiple comparison’s test with a single pooled variance was performed, as the data was normally distributed with equal standard deviations.

Figure 5.. Expression of CHCHD10 R15L in brain and spinal cord of humanized knock-in mouse.

(A) Schematic demonstrating the generation of the Chchd10^hR15L/+ knock-in (KI) mouse line. The coding region of human CHCHD10 inclusive of introns was knocked into the endogenous mouse Chchd10 locus. The haplotype was identical to that of the USALS#5 family described in (Fig. 1). Loxp sites were added to the introns flanking exon 2. (B) Body weights of Chchd10^hR15L/+ knock-in (KI) and WT littermates. Chchd10^hR15L/+ mice were grossly normal up to 1 year. (C) Immunofluorescence of a coronal brain section at the level of the primary motor cortex from Chchd10^hR15L/+ mice. Within the cortex CHCHD10 immunostaining (red) was substantially higher in layer 5, identifiable by CTIP2 immunostaining (green). This layer contains the extratelecephalic upper motor neurons cells that project to the spinal cord through the corticospinal track. Within layer 5, CHCHD10 immunostaining was highest in CTIP2 bright neurons and lowest in non-neuronal cells (DAPI+ but NeuN-). Dotted rectangles in images indicate the approximate areas of the magnified images that appear immediately to the right. Neurons with high CTIP2 expression are indicated by arrows. Those with lower CTIP2 expression are indicated by closed arrow heads. DAPI positive nuclei that were negative for the neuronal nuclear marker NeuN are indicated by open arrowheads. Scale bars are 500 μm (left image), 200 μm (middle images), and 10 μm (right images), respectively. (D) Immunofluorescence of a coronal spinal cord section at the level of the cervical enlargement from Chchd10^hR15L/+ mice. CHCHD10 (red) immunostaining is strongest in the ChAT (green) positive lower motor neurons. Dotted rectangles in images indicate the approximate areas of the magnified images that appear immediately to the right. Scale bars are 500 μm (left and middle images) and 10 μm (right images), respectively.

Figure 6.. CHCHD10 R15L is functional in vivo.

(A) Schematic shows the cross used in this set of experiments to test whether a single CHCHD10 R15L allele can suppress the mitochondrial integrated stress response (mt-ISR) activation observed in the CHCHD2/CHCHD10 double knockout mouse. This cross produces the four genotypes that are compared in (Fig. 5E). (B) Schematic shows activation of the OMA1-DELE1 mt-ISR. Activation of the mt-ISR is evident both in the cleavage of L-OPA1 by OMA1 and the upregulation of enzymes that are targets of the transcription factor ATF4. These targets include MTHFD2, PSAT1, P5CS, and PYCR1. (C) Solubility and localization of CHCHD10 was determined for Chchd10^+/+; Chchd2^+/+ (WT) and Chchd10^hR15L/−; Chchd2^−/− (RL) by immunoblotting following subcellular fractionation of heart tissues. Heart tissue from a Chchd10^S59L/+; Chchd2^+/+ (SL) mouse was additionally examined as a positive control. Heart tissues were fractionated by centrifugation into a cytosolic fraction and a mitochondria-enriched heavy membrane fraction. These fractions were further fractionated by differential detergent solubilization into soluble and insoluble fractions. CHCHD10 R15L was predominately in the soluble mitochondrial fraction. As expected, OXPHOS subunits for complex I - V were found predominately in the mitochondrial fraction. OXPHOS subunit expression was decreased in the SL mouse sample, as expected, but was not decreased in samples from RL mice relative to WT. These data are representative of 2 WT, 4 RL, and 2 SL mice. RL mice were approximately 8 weeks of age. WT and SL mice were approximately 40 weeks old. (D) Immunofluorescence of mouse heart tissue. Mitochondria are indicated by cytochrome c (CYCS) immunostaining. Stronger signal is observed for Chchd10^hR15L/+plausibly because of the higher affinity of the antibody for human CHCHD10. The distribution of CHCHD10, however is similar to Chchd10^+/+ mice and distinct from CHCHD10 aggregation observed in Chchd10^G58R/+ mice. (E - J) Immunoblotting of heart lysates from mice produced by the cross shown in (Fig. 5A). Suppression of OMA1 activation by a single Chchd10 hR15L allele is evident in the increased L-OPA1/S-OPA1 ratio (E and F) and restoration of OMA1 levels (which decrease on OMA1 activation) (E and G). Downstream of OMA1, enzymes that are transcriptionally upregulated by the mt-ISR are likewise normalized by expression of a single Chchd10 hR15L allele (E and H – J). Mice were approximately 8 weeks of age. In all panels, ns, *, **, ***, ***” correspond to not-significant, p ≤ 0.05, p ≤ 0.01, p ≤ 0.001, and p ≤ 0.0001, respectively. For (Fig. 5F, G, and J) an ordinary one-way ANOVA test followed by Dunnett’s multiple comparison’s test with a single pooled variance was performed, as the data was normally distributed with equal standard deviations. For (Fig. 5H and I) a Brown-Forsythe and Welch ANOVA test followed by a Dunnett’s T3 multiple comparisons test was performed as the standard deviations were not equal.

Figure 7.. CSF proteomics reveal CHCHD10 and NfL elevation in CHCHD10 R15L ALS.

(A - B) Volcano plots of CSF proteomics with the Olink Explore HT platform comparing symptomatic (A) and presymptomatic (B) CHCHD10 R15L mutation carriers to healthy controls. Uncorrected p-value is shown on the y-axis. Significant differentially expressed proteins (FDR < 5% in the one-way ANOVA across all groups and the t-test for the comparison of interest) are outlined in black. Mitochondrial proteins are blue with CHCH- domain containing proteins shown in dark blue and others in light blue. The neurofilaments NfL and PRPH are shown in red. (C) Dot plot of CHCHD10 protein levels for individuals measured by Olink proteomics in the same experiment depicted in (A – B). (D) Dot plot of CHCHD10 protein levels for individuals measured by Somascan proteomics in the same experiment depicted in (Fig. S5). (E) Dot plot of NfL protein levels for individuals measured by Olink proteomics in the same experiment depicted in (A – B). (E) Dot plot of NfL protein levels for individuals measured using a SIMOA assay. Dotted line indicates the cutoff value of 1431 pg/mL previously suggested for ALS. CHCHD10 presymptomatic (C10-PS) was used as the control for pairwise statistical comparisons. (G) Dot plot of PRPH protein levels for individuals measured by Olink proteomics in the same experiment depicted in (A – B). (H) Dot plot shows the ratio of PRPH to NfL protein levels for the indicated groups. Adjusted p-values are shown for (C, D, E, and G) to reflect multiple comparisons across proteins in the source dataset. P-values for (F and H) are adjusted for the multiple comparisons shown in graphs. In all panels, ns, *, **, ***, ***” correspond to not-significant, p ≤ 0.05, p ≤ 0.01, p ≤ 0.001, and p ≤ 0.0001, respectively.