Pathogenic Effect of TP73 Gene Variants in People With Amyotrophic Lateral Sclerosis

Abstract

Objective

To identify novel disease associated loci for amyotrophic lateral sclerosis (ALS), we used sequencing data and performed in vitro and in vivo experiments to demonstrate pathogenicity of mutations identified in TP73.

Methods

We analyzed exome sequences of 87 patients with sporadic ALS and 324 controls, with confirmatory sequencing in independent ALS cohorts of >2,800 patients. For the top hit, TP73, a regulator of apoptosis and differentiation and a binding partner and homolog of the tumor suppressor gene TP53, we assayed mutation effects using in vitro and in vivo experiments. C2C12 myoblast differentiation assays, characterization of myotube appearance, and immunoprecipitation of p53-p73 complexes were performed in vitro. In vivo, we used clustered regularly interspaced short palindromic repeats (CRISPR)/Cas9 targeting of zebrafish tp73 to assay motor neuron number and axon morphology.

Results

Four heterozygous rare, nonsynonymous mutations in TP73 were identified in our sporadic ALS cohort. In independent ALS cohorts, we identified an additional 19 rare, deleterious variants in TP73. Patient TP73 mutations caused abnormal differentiation and increased apoptosis in the myoblast differentiation assay, with abnormal myotube appearance. Immunoprecipitation of mutant ΔN-p73 demonstrated that patient mutations hinder the ability of ΔN-p73 to bind p53. CRISPR/Cas9 knockout of tp73 in zebrafish led to impaired motor neuron development and abnormal axonal morphology, concordant with ALS pathology.

Conclusion

Together, these results strongly suggest that variants in TP73 correlate with risk for ALS and indicate a role for apoptosis in ALS disease pathology.

Amyotrophic lateral sclerosis (ALS) is a fatal degenerative disease of motor neurons in the brain and spinal cord.¹ Much remains unknown about the genetics and pathophysiology underlying this disease. While known gene variants are critical determinants of 68% of familial ALS, they account for only 17% of the more common sporadic ALS (SALS) cases.^2,3 However, up to 61% of SALS risk has been attributed to genetic factors,^4,5 suggesting that unknown genetic loci contribute to the development of SALS. Furthermore, it is unclear what key shared pathophysiology leads to ALS.

Our goal was to identify additional genetic risk factors for SALS using next-generation sequencing of a cohort of 87 patients with SALS, together with validation in 2 additional cohorts and in vitro and in vivo experiments. We show that rare, nonsynonymous mutations identified in TP73 among patients with SALS have a deleterious effect on protein function and that protein 73 (p73) is necessary for motor neuron health.

TP73 is part of the p53 family of tumor suppressor transcription factors, which modulate the expression of target genes to affect cell cycle arrest, apoptosis, and cellular differentiation.^6,7 p73^−/− mice exhibit developmental brain defects without a spontaneous tumor phenotype,⁸ demonstrating that p73 is essential in CNS development.⁷ Our findings also illustrate that regulation of apoptosis may have an unexpected role in contributing to ALS pathology. Future studies could investigate whether the apoptotic pathway could become a suitable target for novel therapeutic intervention during motor neuron degeneration.

Methods

Standard Protocol Approvals, Registrations, and Patient Consents

Patients with SALS seen at the University of Utah School of Medicine provided written consent and were included in genetic studies. Experiments were approved by the University of Utah Institutional Review Board. All zebrafish experiments followed guidelines from the University of Utah Institutional Animal Care and Use Committee, regulated under federal law by the US Department of Agriculture and the Office of Laboratory Animal Welfare at the NIH and accredited by the Association for Assessment and Accreditation of Laboratory Care International.

Data Availability

Anonymized data will be shared by request from any qualified investigator.

Sequencing and Screening of Patients With ALS for TP73 Variants

Exome sequencing results of 87 European patients with SALS² and 324 Simons Simplex Collection control individuals⁹ were used. These 87 patients were considered the discovery cohort, and their sequencing data were analyzed by VAAST/Phevor, which identified TP73 as a top hit. We then tested for additional TP73 variants in a second ALS cohort that was part of a University of Utah sequencing effort, the Utah Heritage 1K (H1K) Project. There were 70 patients with ALS in the H1K cohort, which included 9 exome-sequenced non-European patients with SALS. However, 26 of the 70 patients with ALS were initially from the discovery cohort of 87 exome-sequenced patients who had been selected for whole-genome sequencing. Therefore, this allowed screening in a total of 53 patients with ALS. One additional TP73 coding variant was found. The next cohort (called ALSdb) screened for TP73 variants is publicly available (alsdb.org/) and consists of 2,800 whole-exome–sequenced patients with SALS. Combining the discovery cohort of 87 patients with SALS, the 53 patients with ALS from the H1K cohort, and the 2,800 patients with ALS from ALSdb yields a total of 2,940 patients with ALS who were screened for TP73 variants.

We performed multigenerational analysis on all samples from Utah and relatedness software on our discovery cohort of 87 patients with SALS. Patients were separated by at least 3 degrees of relatedness. In the H1K cohort, patients were separated by at least 6 degrees of relatedness. In the ALSdb cohort, the majority are White sporadic cases, having no first- or second-degree relatives with ALS, and this cohort is similar in composition to our ALS cohorts. The patients in all cohorts do not display any close relatedness or family history of ALS. The aligned genomic reads underwent joint variant calling with 95 long-lived individuals (longevity cohort) and 291 European individuals from the 1000 Genomes Project. All patients in all cohorts were diagnosed by El Escorial revised criteria by neuromuscular physicians.

VAAST/Phevor Analysis

Exome variants were used in VAAST and Phevor^10,11 analysis. Genomic regions covered by <5 reads on average were omitted. Variants with an Exome Aggregation Consortium¹² European minor allele frequency (MAF) >0.001 were removed. VAAST was run assuming a dominant model of disease inheritance. VAAST output was then analyzed by Phevor to identify burdened genes that share functional characteristics similar to ALS-associated genes. Human Phenotype Ontology terms used included HP:0007354, HP:0002450, HP:0007373, and HP:0002145. SIFT, FATHMM, PolyPhen-2, LRT, MutationTaster, MutationAssessor, PROVEAN, and CADD assessed nonsynonymous variants.

Cloning of Wild-Type and Mutant ΔN-p73

ΔN-p73α coding DNA sequence was amplified from HA-p73α-pcDNA (gift from William Kaelin; Addgene No. 22102). Patient mutations were introduced using site-directed mutagenesis (NEB) and cloned into pDONOR221. Next, LR reactions between pDONOR221 vectors and pCLX-pTF-R1-DEST-R2-EBR65 (gift from Patrick Salmon, Addgene No. 45952) were performed. HEK293T cells were transfected using calcium phosphate with psPAX2 (gift from Didier Trono, Addgene No. 12260), pCAG-Eco (gift from Arthur Nienhuis and Patrick Salmon, Addgene No. 35617), and wild-type/mutant pCLX–ΔN-p73 vectors. pCLX-pTF-eGFP was used to produce control lentivirus. Supernatant was harvested 72 hours later.

C2C12 In Vitro Testing of ALS TP73 Variants

C2C12 myoblasts were cultured in high-glucose Dulbecco modified Eagle medium plus 20% heat-inactivated fetal bovine serum and 1% penicillin/streptomycin. At 48 hours after infection, cells were incubated in growth medium plus 10 μg/mL blasticidin. Twenty-four–well plates were prepared for C2C12 differentiation by adding 0.1% gelatin in dH20, rinsing, and then seeding 50,000 cells in technical duplicate. Cells were exposed to growth medium containing 1 μg/mL of doxycycline or no doxycycline. After 24 hours, cells were incubated in differentiation medium (10 μg/mL recombinant human insulin [Gibco, Waltham, MA], 2% heat-inactivated horse serum, and 1% penicillin/streptomycin) with or without doxycycline. Fixation with 70% ethanol was performed 4 days later. Differentiation medium was changed every 48 hours. After fixation and peroxidase inactivation, cells were blocked for 1 hour with blocking solution (5% donkey serum in phosphate-buffered saline [PBS], 0.3% TX-100, 0.2% Na Azide). Cells were incubated with anti–myosin heavy chain (MHC) antibody (MF-20, DSHB, 1:100) at 4°C overnight. Cells were washed with PBS-Tween (PBST) and incubated with donkey anti-mouse conjugated to biotin for 45 minutes at room temperature (RT) (Jackson ImmunoResearch, West Grove, PA). Cells were washed; Vector ABC reagent was added; and cells were incubated for 35 minutes. Cells were washed and Vector DAB substrate was added. Cells were imaged with an EVOS Plate Reader (Fluorescence Microscopy Core Facility, HSC cores, University of Utah). Control plates were analyzed in biological duplicate and doxycycline-exposed plates in biological triplicate. Four images were taken of each technical replicate well per genotype. Three distinct measurements of myotube width were performed per image with Fiji. Myotube width and intensity values were normalized to the mean value of myotube width and mean intensity of pCLX-pTF-EGFP cells.

Doxycycline-Induced ΔN-p73 Expression in C2C12 Cells

Mutant and wild-type C2C12 cells were incubated in growth medium plus or minus doxycycline for 48 hours. Cells were then lysed with radioimmunoprecipitation assay buffer containing Pierce protease inhibitor tablets (Thermo Fisher, Waltham, MA). Protein (100 μg) was loaded per well in 4% to 12% Bis-Tris gels (Thermo Fisher). Proteins were transferred to membranes and blocked in 5% milk/PBST. Blots were incubated in primary antibody/5% milk/PBST overnight at 4°C (anti–β-actin No. MA5-15739,1:10,000 dilution, anti–ΔN-p73 No. MA5-16183 1.5 μg/mL). Cells were washed and incubated in secondary antibody for 1 hour at RT (goat anti-mouse, Thermo Fisher No. 31430, 1:20,000 for anti–β-actin blots, 1:5,000 for anti–ΔN-p73). Blots were developed with SuperSignal West Pico PLUS (Thermo Fisher No. 34578) before visualization with film.

Coimmunoprecipitation Experiments in Neuro-2a Cells

Neuro-2a (ATCC CCL-131) cells were infected with lentiviral vectors used in C2C12 cells and grown in Eagle minimal essential medium with 10% heat-inactivated fetal bovine serum and blasticidin. Cells were kept in doxycycline medium for 72 hours (media changed every 48 hours, 100 ng/mL doxycycline). Cells were lysed using Thermo Fisher Co-IP lysis buffer with protease inhibitor tablet. Magnetic beads were cross-linked to anti–ΔN-p73 or PCNA antibody with the Thermo Fisher cross-linking kit. Cross-linked beads were then incubated with 1,000 µg lysate overnight at 4°C. Protein bound to ΔN-p73 was eluted from beads with acidic elution buffer. Equal amounts of protein were loaded in 10% Bis-Tris gels and blotted for ΔN-p73, p53 (CST No. 2524), or PCNA (CST No. 13110).

Zebrafish Modeling of tp73 Loss of Function

An sgRNA was generated to target exon 4 in zebrafish: tp73, 5′-CGGCCATCCCTTCCAATACA-3′. After injection, embryos were collected at 24 hours post-fertilization (hpf) for analysis of mutagenesis using high-resolution melt analysis and Sanger sequencing.¹³ One-cell-stage wild-type embryos were injected with either mnx:GFP (18 ng/µL) alone or in combination (mnx:GFP, 50 ng/µL) with tp73 sgRNA (420 ng/μL). Tg(Hb9:Gal4; UAS:GFP) embryos were injected with 450 ng/μL tp73 sgRNA. All clustered regularly interspaced short palindromic repeats (CRISPR)–injected embryos were also injected with 800 ng/μL Cas9 protein (IDT). At 72 hpf, embryos were fixed and dehydrated for immunofluorescence.¹⁴ Antibodies used were chicken anti-GFP (1:1,000, Aves [Davis, CA], GFP-1020) and goat anti-chicken Alexa 488 (1:400, Invitrogen [Carlsbad, CA], A11039). In addition, nuclei were stained with DAPI (1:1,000). Terminal deoxynucleotidyl transferase dUTP nick-end labeling (TUNEL) was performed on whole-mount larvae (ApopTag Rhodamine In situ Apoptosis Detection Kit; Millipore [Burlington, MA]). Embryos were fixed, dehydrated, and permeabilized.¹⁴ Embryos were incubated in equilibration buffer for 1 hour, followed by overnight incubation in terminal deoxynucleotidyl transferase enzyme at 37°C. The following day, end labeling was terminated with incubations in stop/wash buffer, followed by several washes in 1× PBST. Next, embryos were incubated in working-strength sheep anti-digoxigenin rhodamine for 1 hour at RT. Labeling was stopped by several washes, and embryos were cleared and mounted.

Spinal Motor Neuron and Axon Quantification

To assess the number of spinal motor neurons (SMNs), 72 hpf Tg(Hb9:Gal4; UAS:GFP) transgenic embryos were stained for green fluorescent protein (GFP) and DAPI and mounted. Slides were imaged on a Nikon (Tokyo, Japan) A1 instrument (University of Utah HSC Imaging Core). A set of 21 slices (step size 5 μm) of the upper third of zebrafish body (starting point below the head) was obtained with identical confocal settings (10× objective, 2.5× zoom, 1,024 pixels, 6.2 milliseconds per pixel dwell time). Confocal stacks were projected in Fiji and composed with Adobe Photoshop for quantification of number of SMNs per hemi-segment. To assess axonal outgrowth, transient transgenic 72 hpf immunostained embryos were mounted. Sets of 21 slices of 5-μm step size were imaged. Confocal stacks of maximal intensity were projected in Fiji, and primary and secondary axonal branching lengths were measured with Neuron J.

Apoptosis Quantification

Seventy-two–hpf embryos of the above-mentioned transient transgenic line were immunostained, TUNEL labeled, mounted, and imaged. Statistical analyses used the Student t test.

Results

VAAST and Phevor Identify TP73 as an ALS Risk Gene

To identify novel ALS risk loci, we performed whole-exome sequencing on a previously described cohort of 87 patients with SALS from the University of Utah and 324 healthy control individuals from the Simons Simplex Collection.^2,9 VAAST¹⁰ was used to identify genes with excess deleterious nonsynonymous single nucleotide variants (SNVs) in patients with ALS compared to unaffected controls. We excluded SNVs with an Exome Aggregation Consortium¹² non-Finnish European minor MAF >0.001. Phevor¹¹ was then used to rerank the VAAST candidate genes on the basis of biochemical, cellular, and pathologic functions similar to known ALS risk genes (supplement figure 1, A and B, doi:10.5061/dryad.4qrfj6q94). Two known ALS risk genes, MAPT (rank 3) and SOD1 (rank 5), were among the top 10 genes ranked by combined VAAST and Phevor analysis.

The only gene to rank among the top 5 genes in both VAAST and Phevor was TP73, which encodes tumor protein p73. Five different heterozygous, rare, missense SNVs in TP73 were found among 4 patients with SALS (figure 1B). An additional rare 7-amino-acid in-frame deletion variant (chr1:3646605 CCATGAACAAGGTGCACGGGGG>C; TP73:p.PMNKVHGG413-420P), which was not assessed by VAAST, was found in a patient with SALS in a search for indels (table 1). Five of the 6 total TP73 protein-coding variants were confirmed by Sanger sequencing.

Figure 1. Twenty-Four Rare (ExAC NFE MAF <0.001) TP73 Nonsynonymous Variants Were Found Across Multiple Cohorts.

(A) TA-p73 and ΔN-p73, which have opposing tumor suppressive/oncogenic functions, are expressed from 2 separate promoters. Alternative splicing events also give rise to a multitude of different isoforms (such as ΔN-p73α and ΔN-p73γ). (B) Primary structure of TA-p73α and ΔN-p73α. Number within the circles refers to the variant number in table 1. ExAC = Exome Aggregation Consortium; MAF = minor allele frequency; NFE = non-Finnish European.

Table 1.

Information For All 24 TP73 Coding Variants

TP73 Is Affected by Deleterious Variation at a Similar Rate to Known Risk Genes in Patients With ALS

We next tested whether rare and deleterious TP73 protein-coding variants could be found in 2 replication cohorts. A TP73 missense SNV (chr1:3647559 G>A; TP73:p.A472T) was identified in an independent cohort of 53 sequenced University of Utah patients with ALS, part of the Utah H1K project (table 1, Methods). An additional 18 rare, nonsynonymous variants in TP73 were found on additional screening in the ALSdb cohort. ALSdb consists of 2,800 White whole-exome–sequenced patients with ALS.¹⁵ When all analyzed cohorts were combined, 24 different rare TP73 coding sequence variant sites (22 SNVs and 2 in-frame indels) were found among 2,940 patients with ALS (table 1 and supplement table 1, doi:10.5061/dryad.4qrfj6q94), similar to the relative contribution of many other ALS risk genes.³ All 22 SNVs were predicted to be deleterious and had an MAF <0.0005 in the control non-Finnish European cohort. Four TP73 variants identified in University of Utah patients with ALS that produced coding changes in isoform ΔN-p73α were chosen for subsequent in vitro testing. Supplemental figure 2 shows the protein sequence alignment of ΔN-p73α for multiple species and illustrates that 3 of the 4 selected mutations are widely conserved across species.

TP73 Protein-Coding Variants Specific to Patients With ALS Impair or Alter the Function of ΔN-p73 During C2C12 Differentiation

TP73 gives rise to 2 different main protein isoforms, TA-p73 and ΔN-p73, which have opposing functions (figure 1A).⁷ TA-p73, which possesses an N-terminal transactivation domain, induces expression of gene targets. Conversely, N-terminally truncated p73 (ΔN-p73) has oncogenic and transformative qualities because it inhibits the function of p53 and TA-p73 through direct binding.^7,16,17 Of note, ΔN-p73 (supplement figure 3, doi:10.5061/dryad.4qrfj6q94) is the primary p73 isoform found in the brain and promotes neuronal survival by providing resistance to apoptotic insults.^18,19 ΔN-p73 function can be assayed by overexpression in serum-deprived C2C12 myoblasts because it inhibits cell cycle withdrawal, impairs myoblast differentiation into elongated, multinucleated myotubes, and decreases MHC staining.²⁰ Two known ALS-causing proteins, SOD1 and VAPB, have also been shown to function in C2C12 differentiation similar to ΔN-p73.^21,22 Multiple cell types in addition to motor neurons have been shown to be involved in ALS pathology, and the mouse myoblast C2C12 cell line has been used previously to study ALS.^23-26 This evidence, in addition to the known role of ΔN-p73 in C2C12 differentiation, led us to investigate whether the identified TP73 protein-coding variants impair the function of p73.

Four nonsynonymous TP73 variants in ΔN-p73α (figure 1B, table 1, and supplement table 1, doi:10.5061/dryad.4qrfj6q94), the canonical isoform of ΔN-p73, were selected for functional testing. We tested the ability of wild-type ΔN-p73α and each selected variant to inhibit myoblast differentiation and MHC staining to determine whether these mutations rendered the protein dysfunctional. During myoblast differentiation, myoblasts fuse together to form myotubes while other myoblasts undergo apoptosis with dysfunctional myoblast apoptosis being observed in muscular degeneration.^27-30 It is estimated that ≈30% of myoblasts go through apoptosis after differentiation is induced, and the remaining cells proceed to exit the cell cycle.^27,30 Impeding apoptosis severely hinders myoblast fusion, leading to decreased myotube formation.²⁸ When myoblast fusion is increased through the addition of apoptotic myoblasts, myotubes enlarge and contain more nuclei.

To measure the extent of myoblast differentiation, we quantified MHC intensity and myotube width (figure 2 and supplement figure 4, doi:10.5061/dryad.4qrfj6q94). These parameters were used because increased MHC expression is a marker of differentiated myotubes and myotube width/size increases with increased number of incorporated nuclei in the myotubes.^20,28 On overexpression of wild-type ΔN-p73α, normalized white light intensity (inversely related to MHC staining intensity) increased. As expected, wild-type ΔN-p73α expression decreased MHC staining and inhibited differentiation (figure 2A and supplement figure 4, C and E, p = 2.06E-05, 95% confidence interval [CI] −0.188 to −0.088). Strikingly, expression of ΔN-p73αp.V187I (mut 1) did not inhibit differentiation. This is demonstrated by no difference in the MHC staining or myotube width between expression of ΔN-p73αp.V187I and enhanced GFP (EGFP) (p = 0.296, 95% CI −0.083 to 0.022; p = 0.800, 95% CI −0.072 to 0.102, figure 2, A and B and supplement figure 4, F and G). Expression of mutant ΔN-p73αp.PMNKVHGG364-371P (mut 2) did cause decreased MHC staining (p = 9.82E-07, 95% CI −0.149 to −0.067, figure 2A). However, myotubes were significantly wider (p = 8.80E-16, 95% CI −0.259 to −0.854, supplement figure 4, H and I). We hypothesize that this is due to increased myoblast fusion resulting from aberrant apoptosis caused by mutant ΔN-p73α, leading to larger myotubes. Expression of ΔN-p73αp.A423T (mut 3) did not inhibit differentiation correctly, as shown by no difference in MHC staining compared to EGFP (p = 2.96E-01, 95% CI −0.010 to 0.031). In addition, myotube width was substantially increased (p = 8.80E-16, 95% CI −1.051 to 0.771, supplement figure 4, J and K). Expression of ΔN-p73αp.V537M (mut 4) did inhibit differentiation (p = 1.95E-06, 95% CI −0.298 to −0.157). However, myotube width was significantly increased, as seen with mutants 2 and 3 (p = 1.94E-07, 95% CI −0.783 to −0.365, supplement figure 4, L and N). Mutant 1 does not inhibit differentiation, while mutants 2 through 4 have significantly larger myotubes. We hypothesize that this myotube enlargement is due to dysfunctional apoptosis.

Figure 2. ALS Variants in TP73 Render p73 Dysfunctional in C2C12 Differentiation Demonstrated by Quantification of MHC Staining Intensity and Myotube Width Compared to EGFP.

(A) Quantification of normalized myosin heavy chain (MHC) staining in differentiated C2C12s, with or without exposure to doxycycline, expressing enhanced green fluorescent protein (EGFP), wild-type (WT) ΔN-p73α, or mutant ΔN-p73. WT ΔN-p73α expression results in inhibited C2C12 differentiation shown by reduced MHC immunohistochemistry staining compared to EGFP. ΔN-p73α^Mut1+3 expression fails to inhibit C1C12 differentiation shown by no difference in MHC staining compared to EGFP (**p < 0.001, Wilcoxon rank-sum test with Bonferroni correction). (B) Normalized myotube width quantification. ΔN-p73α^Mut2-4 expression results in myotubes with significantly larger width compared to EGFP. ALS = amyotrophic lateral sclerosis.

Mutations in TP73 Inhibit Binding of ΔN-p73α to p53

To investigate whether the mutations identified in patients with ALS affected the ability of ΔN-p73α to inhibit apoptosis, Neuro-2A (N2A) cells, a mouse neuroblastoma line, were infected with lentivirus expressing either wild-type or mutant ΔN-p73α. We tested the ability of mutant ΔN-p73α to bind p53 because this could help explain the mechanism underlying aberrant apoptosis caused by mutant ΔN-p73α. After N2A cells expressing mutant ΔN-p73α underwent doxycycline exposure and cell lysis, coimmunoprecipitation (pull-down) of ΔN-p73α with cross-linked magnetic beads was performed. PCNA encodes for proliferating cell nuclear antigen and was used as an immunoprecipitation control. Next, Western blotting was used to assess the amount of p53 bound to mutant and wild-type ΔN-p73α. Equal amounts of p53 and ΔN-p73α were present in the cell lysates before immunoprecipitation (figure 3A). Blotting for ΔN-p73α in the lysate that had undergone immunoprecipitation showed that all ΔN-p73α had been successfully immunoprecipitated by cross-linked beads. All 4 mutant ΔN-p73α proteins displayed impaired ability to bind p53. This is shown by the large difference in band intensity of p53 pulled down by mutant ΔN-p73α compared to N2A cells expressing wild-type ΔN-p73α (figure 3B). These results demonstrate that these mutations affect the ability of ΔN-p73α to respond to apoptosis, which could potentially render motor neurons more vulnerable to apoptosis upon cellular stress.

Figure 3. ΔN-p73α Containing ALS Patient Mutations Expressed in Neuro-2A Cells Displayed Impeded Binding to Proapoptotic p53.

(A) Western blots show equal input level of protein (proliferating cell nuclear antigen [PCNA], ΔN-p73α, and p53) before immunoprecipitation. Protein was immunoprecipitated with ΔN-p73α or PCNA antibody. Postimmunoprecipitation lysate demonstrates that all ΔN-p73α protein was immunoprecipitated with ΔN-p73α antibody. Blots were probed for ΔN-p73α, p53, and PCNA. PCNA was used as a loading control and was immunoprecipitated in tandem to ΔN-p73α. Red arrows denote less intense p53 protein bands immunoprecipitated by mutant ΔN-p73α compared to wild-type (WT) ΔN-p73α and enhanced green fluorescent protein (EGFP) (in EGFP lane, ΔN-p73α band is endogenous protein). (B) Quantification of p53 band intensity of Western blots performed after ΔN-p73α immunoprecipitation. Shown are mean intensity and SD for p53 bands from coimmunoprecipitations performed in biological duplicate for Neuro-2A cells expressing ST ΔN-p73α or mutant ΔN-p73α on doxycycline (dox) exposure. ALS = amyotrophic lateral sclerosis.

Loss of tp73 Results in Reduction of SMNs via Apoptosis

Next, we tested the effect of loss of tp73 function on motor neuron development and morphology. Previous studies have established zebrafish (Danio rerio) as an ALS model.³¹ A CRISPR/Cas9 system was used to determine whether loss of tp73 in zebrafish leads to an ALS-like phenotype. Using gRNA against exon 4 of zebrafish tp73 (tp73^CRISPR), which encodes a portion of the DNA binding domain, we found that CRISPR-injected embryos had >95% deleterious mutations at the target locus (supplement figure 5B, doi:10.5061/dryad.4qrfj6q94). To determine whether mutagenesis of tp73 disrupts motor neuron development, we quantified GFP+ SMNs in Tg(Hb9:Gal4; UAS:GFP) embryos (figure 4, A–C). As a control, we used a gRNA targeting tyrosinase (tyr^CRISPR)³² (supplement figure 5, C and D). tp73^CRISPR mutagenized embryos had a significant reduction in the number of GFP-labeled SMNs compared to uninjected or tyr^CRISPR-injected controls (figure 4C, uninjected 59.9 ± 2.1, n = 16; tp73^CRISPR 52.1 ± 1.5, n = 24, p < 0.01, 95% CI −12.94 to −2.75). These results demonstrate the effect loss of tp73 has on SMN development and motor neuron disease.

Figure 4. tp73 Mutation Decreases Motor Neuron Count, Increases Apoptosis, and Impairs Motor Neuron Axon Outgrowth.

(A and B) Confocal images of spinal motor neurons (MNs; green) in uninjected and tp73^CRISPR injected embryos (confocal imaging of dorsal view of spinal cord). (C) tp73^CRISPR-injected embryos have a significantly reduced number of MNs compared to both uninjected and tyr^CRISPR-injected sibling controls (supplement figure 4, C and D, doi:10.5061/dryad.4qrfj6q94). (D and E) Confocal images of increased MN (green) apoptosis (red terminal deoxynucleotidyl transferase dUTP nick-end labeling [TUNEL]) in tp73^CRISPR embryos compared to uninjected. (F) Increased MN apoptosis in tp73^CRISPR embryos compared to uninjected, performed in mnx:GFP transgenic line. (G and H) Confocal images of MN primary and secondary axons (green) in uninjected and tp73^CRISPR-injected embryos. (I) MN primary axon length in tp73^CRISPR embryos is significantly shorter than that of uninjected sibling controls. (J) MN secondary axon length in tp73^CRISPR-injected embryos is significantly shorter compared to that of uninjected sibling controls. CRISPER = clustered regularly interspaced short palindromic repeats; hpf = high-power field.

Previous studies have shown that p73 is required to prevent neuronal apoptosis.^19,33 To determine whether the reduction in motor neuron number in tp73^CRISPR mutagenized animals was due to apoptosis, we quantified TUNEL+/GFP+ colabeled SMNs in Tg(Hb9:Gal4; UAS:GFP) embryos (figure 4, D–F). We found an ≈2-fold increase in TUNEL+ GFP+ SMN in tp73^CRISPR mutagenized animals compared to uninjected controls (uninjected 1.2 ± 0.3, n = 6; tp73^CRISPR 3.1 ± 0.6, n = 9, p < 0.05, 95% CI 0.166–3.723). This suggests that increased apoptosis in tp73^CRISPR mutants contributes to motor neuron number reduction.

Loss of p73 Results in Reduction of SMN Axonal Branching

Several ALS zebrafish models^34-36 have reported drastic reduction in SMN axon length early in development. To address SMN axon outgrowth, we injected mnx:GFP alone or in combination with tp73^CRISPR to quantify SMN axon length. We found that tp73^CRISPR-mutagenized embryos exhibited a significant reduction in primary and secondary axonal branching length (figure 4, G–I, primary axon lengths: uninjected 149.4 ± 4.4 μm, n = 47; tp73^CRISPR 132.6 ± 5.9 μm, n = 34, p < 0.05, 95% CI 2.129–31.567; figure 4, G, H, and J, secondary axon lengths: uninjected 34.7 ± 1.5 μm, n = 76; tp73^CRISPR 29.0 ± 2.0 μm, n = 64, p < 0.05 95% CI 0.622–10.711). Taken together, these data indicate that disruption of tp73 leads to a reduction in SMN number and axon branch length.

Discussion

Our results show that deleterious TP73 protein-coding variants occur in patients with ALS at a total frequency similar to that of other known ALS risk genes, that TP73 ALS variants impair normal function of p73 in both C2C12 and neuroblastoma cell lines, and that loss of p73 impedes motor neuron survival and axonal development in vivo. Several additional lines of evidence support a causal role of TP73 in ALS pathogenesis.

Aged mice carrying 1 p73 null allele (p73^+/−) have been described as a model for Alzheimer disease,³⁷ but aged p73^+/− heterozygous mice also display symptoms of motor neuron disease. While haploinsufficient mice develop neurodegeneration in a number of neuronal groups and brain regions, 16-month-old p73^+/− mice specifically display a 5% reduction in the volume of the motor cortex and a 25% reduction in the number of neurons in the motor cortex compared with wild-type mice. In addition, the number of argyrophilic neurons with degenerating processes is increased. Haploinsufficient mice also display increased clasping behavior, often seen in neurodegeneration of the motor system. Therefore, aged p73^+/− mice are an appropriate model of motor neuron disease, and their phenotype supports our argument that TP73 contributes to ALS pathology.

A hallmark of ALS pathology is protein aggregation.^38,39 Causal mutations in SOD1 cause the altered protein to form aggregates in motor neurons and glia.⁴⁰ However, it has also been shown that no matter the causal mutation, specific wild-type proteins will aggregate, as is the case with TAR DNA-binding protein 43.⁴¹ One of the proteins found in subsets of these aggregates is tau, which is encoded by MAPT.^42,43 Phosphorylated tau has been shown to be regulated by p73, with aged haploinsufficient p73^+/− mice containing high levels of phosphorylated tau filaments in the motor cortex.³⁷ In addition, accumulation of p73 occurred in the motor neurons of SOD1(G93A) mice during disease progression.⁴⁴ These attributes make p73 similar to other ALS-causing proteins because p73 accumulates during ALS disease progression and aberrant levels of p73 can lead to the accumulation of phosphorylated tau. We hypothesize that mutated p73 could lead to increased aggregation of tau in patients with ALS, a known feature of ALS pathology.

Not only do we postulate that mutant p73 contribute to disease progression, but additional evidence demonstrates that mutations in already known ALS risk genes disrupt pathways involving p73 such as the p53 signaling pathway. A meta-analysis of skeletal muscle gene expression microarray data from patients with ALS, controls, and the SOD1(G86R) and SOD1(G93A) ALS mouse models showed that only 3 transcription factors had abnormal activity in all datasets.²⁴ Among these 3, p53 had the largest number of target genes with dysfunctional expression. Gene expression of p53 target genes correlated with the severity of disease in muscle samples from patients with ALS. In addition, ΔN-p73 expression was decreased, unlike p73 expression levels, which increased substantially as disease progressed. This would suggest that aberrant apoptosis increases during disease progression due to the decreased expression of antiapoptotic ΔN-p73. Overexpression of SOD1(G86R) protein in C2C12 mouse myoblast cells affected the expression of downstream targets of the p53 family, with a pattern similar to that of muscle from patients with ALS.²⁴

In addition, it has recently been shown that the C9orf72 expansion alters access to DNA for only the p53 transcription factor family.⁴⁵ It was demonstrated that loss of p53 increased the lifespan of C9orf72 mice and prevented neurodegeneration. Our results also allude to overactivation of p53 functioning in ALS pathology. However, we came to this conclusion by investigating p73-p53 binding interactions and demonstrating that ALS patient mutations impede ΔN-p73 binding to p53 (figure 3). This illustrates that apoptotic pathways could have a more important function in ALS than previously thought with p73 as an important factor in disease progression.

The p73-related Src/c-Abl pathway is also disrupted in ALS disease progression. c-Abl is a tyrosine kinase that regulates apoptosis and is a direct activator of p73.⁴⁶ Gene expression studies of spinal cord motor neurons from patients with SALS showed >4-fold increased expression of c-Abl.⁴⁷ In addition, increased c-Abl activation is seen in SOD1(G93A) mouse spinal cords presymptomatically, and drug inhibition of c-Abl has been shown to prevent motor neuron toxicity.^48,49 Furthermore, in a large screen of induced pluripotent stem cell–derived motor neurons from patients with ALS, 14 of 27 compounds that promoted motor neuron survival affected the Src/c-Abl pathway, which was confirmed in an SOD1 mouse model.⁵⁰ Because c-Abl is a direct activator of p73, increased activation of c-Abl because of mutations in ALS-causing loci could increase p73 activity, leading to aberrant motor neuron apoptosis and ALS symptoms. These results indicate the potential of targeting apoptotic pathways in ALS treatment, and we hypothesize that p73 is an important part of these pathways during ALS progression.

Our data suggest that TP73 is a novel ALS risk gene and that apoptosis in motor neurons may play an important role in ALS pathology. Transcription factors that drive neuronal cell survival, differentiation, and tumor suppressor pathways have not been extensively studied in ALS. These findings reveal unexpected aspects of ALS genetic risk and pathology and may identify new approaches to treatment.

Glossary

ALS

amyotrophic lateral sclerosis

CI

confidence interval

CRISPR

clustered regularly interspaced short palindromic repeats

EGFP

enhanced GFP

GFP

green fluorescent protein

H1K

Heritage 1K

hpf

hours post-fertilization

MAF

minor allele frequency

MHC

myosin heavy chain

N2A

Neuro-2A

PBS

phosphate-buffered saline

PBST

PBS-Tween

p73

protein 73

RT

room temperature

SALS

sporadic ALS

SMN

spinal motor neuron

SNV

single nucleotide variant

TUNEL

terminal deoxynucleotidyl transferase dUTP nick-end labeling

Appendix. Authors

Study Funding

Funding provided by the NIH: TL1TR001066, R01GM059290, R01GM104390, R35GM118335, R37NS033123, and R21MH107039; Target; and ALS Bray Chair in Child Neurology Research. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH.

Disclosure

K.L. Russell, J.M. Downie, S. Tsetsou, M.D. Keefe, J.A. Duran, K.P. Figueroa, L. Charles Murtaugh, and M.B. Bromberg report no disclosures relevant to the manuscript. S.B. Gibson reports the following competing interests: Recursion Pharmaceuticals—shareholder and Cytokinetics—advisory board. J.L. Bonkowsky reports the following competing interests: consultant to Bluebird Bio, Inc; Calico, Inc; Denali Therapeutics; Neurogene, Inc; Enzyvant, Inc; and Passage Bio; owns stock in Orchard Therapeutics; is on the Executive Board of wFluidx; receives royalties from Manson Publishing; and spouse receive royalties from BioFire. S.M. Pulst reports the following competing interests: Progenitor Life Sciences—shareholder, Cedars-Sinai—royalties, University of Utah—royalties, and Ataxion Therapeutics—consultant. L.B. Jorde reports receiving royalties from Elsevier, Inc for his Medical Genetics textbook. Go to Neurology.org/N for full disclosures.