A rare genetic variant confers resistance to neurodegeneration across multiple neurological disorders by augmenting selective autophagy

Summary

The study of disease-modifiers is a powerful way to identify patho-mechanisms associated with disease. Using the strong genetic traits of Huntington’s disease (HD), we identified a rare, single nucleotide polymorphism (SNP) in WDFY3 associated with a delayed age-of-onset of up to 23-years. Remarkably, the introduction of the orthologous SNP into mice recapitulates this neuroprotection, significantly delaying neuropathological and behavioral dysfunction in two models of HD. The SNP increases expression of the protein Alfy, an autophagy adaptor protein for the clearance of aggregated proteins, whose ectopic overexpression is sufficient to capture the neuroprotective effects of the variant. Increasing Alfy expression protects not only against HD, but also against the toxicity due to phospho-αSynuclein and AT8-positive accumulates. Taken together, by combining human and mouse genetics, we have uncovered a pathway that protects against multiple proteinopathies, revealing a much sought after, shared therapeutic target across a broad range of neurodegenerative diseases.

eTOC blurb for Croce et al

In Croce et al, the discovery of a rare SNP in the Venezuelan HD kindreds that is associated with a delayed age of onset by up to 23 years reveals how preventing aberrant protein accumulation is protective not only against polyglutamine toxicity, but α-synuclein- and tau-mediated toxicity as well.

Introduction

A powerful means to identify the cellular pathways that play a pivotal role in disease is through the study of disease-modifying genes. These modifiers are often genetic variants that can tip the balance to drive or diminish pathogenesis. The difficulty of using this approach stems from requiring a quantifiable and continuous outcome measure that can be correlated to pathogenic events central to the disease under study. For most adult-onset neurodegenerative diseases, given their sporadic nature, this is not a trivial task. In contrast, the genetic features of the autosomal dominant, heritable disorder Huntington’s disease (HD) offer a powerful platform from which disease modifiers can be identified. HD is caused by a expansion of CAG within the coding region of the HD gene, giving rise to an abnormally expanded polyglutamine (polyQ) stretch within the encoded protein Huntingtin (HTT)^1,2. Given its typical midlife onset and other neuropathological features^3–5, HD serves as a valuable paradigm disorder for sporadic, adult-onset neurodegenerative diseases. Moreover, due to the well-known inverse correlation between the length of the CAG expansion mutation and the predicted age of onset⁶, this provides the continuous outcome measure that is lacking in the most prevalent neurodegenerative diseases. Several studies have identified potential genetic modifiers in both the European and Venezuelan HD kindreds, most notably those genes implicated in DNA repair that may influence the CAG length itself⁷.

A well-known cellular feature across neurodegenerative diseases is the aberrant intracellular accumulation of proteins, but its relevance as a therapeutic target has been frustratingly unclear. Across model systems, the abnormal accumulation of aggregated proteins can drive neuropathology, but little success has been met by attempts to target the event therapeutically. This reflects the complex biology of protein accumulation, which intersects with the most fundamental cellular pathways from protein translation, chaperone-mediated protein folding, protein degradation and stress response pathways⁸. Impacting these homeostatic pathways have proven limited, largely due to the resulting pleiotropic effects stemming from an inability to impact the aggregated form of the protein directly.

Given the intricacy of protein aggregation and accumulation in disease, we hypothesize that their role in neurotoxicity can be revealed through a thorough understanding of the cellular pathways responsible for eliminating complex aggregated structures. Aggregated proteins are eliminated by the lysosome-mediated degradation pathway autophagy^9–11, which uses adaptor proteins to permit their selective turnover¹¹. Several adaptor proteins have been identified to play a role in the turnover of aggregated proteins, such as Sequestosome-1 (SQSTM1/p62), Calcoco3/TAX1BP1 and Autophagy linked FYVE protein (Alfy) (reviewed in¹¹). Despite the potential importance of autophagy on protein accumulation, it is notable that autophagy genes have yet to be overwhelmingly identified to modify disease onset or susceptibility. Mutations in adaptor proteins have been implicated in very rare forms of familial amyotrophic lateral sclerosis-frontotemporal dementia (ALS-FTD)^12–16, as well as other diseases of the periphery^17,18. Nonetheless, given how those implicated are also responsible for the turnover of different cargoes, including mitochondria^9,11, whether diminishing aggregate turnover is central to pathogenesis of these diseases is difficult to establish. Further, when autophagy genes are implicated in disease (reviewed in¹¹), the diseases are developmental, reflecting the overarching importance of autophagy to cellular and organellar homeostasis.

In this study, we identify a rare genetic variant of the gene WDFY3 that is associated with a delayed age of onset of up to 23 years within the Venezuelan HD kindred. The variant leads to increased levels of expression of its encoded mRNA and protein, WDFY3, also known as Alfy. Either introduction of the orthologous polymorphism in Wdfy3 or ectopic overexpression of a cDNA encoding human ALFY (hAlfy) enhances aggregate-turnover by autophagy and prevents the neuropathological and behavioral changes in both a transgenic and knock-in mouse model of HD. Moreover, although identified in a cohort of HD patients, we report that the beneficial effects of this modifier can be extended to different proteinopathies; increasing Alfy expression protects against the neurotoxicity associated with the α-synuclein (αSyn) preformed fibril (PFF) injection model of Parkinson’s disease (PD), as well as the PS19 P301S tauopathy model (PS19). Taken together, through characterization of a rare, protective single nucleotide polymorphism (SNP) within an HD cohort, we have identified a common pathway that can prevent protein accumulation and thereby confer neuroprotection across different neurodegenerative diseases.

Results

rs17368018 delays age of onset in HD patients

A genetic linkage study in the Venezuelan HD kindreds¹⁹ reported several loci modifying age of onset, including one at chr4q21²⁰. To resolve potential modifier genes of interest within this locus, we genotyped HD individuals with corresponding clinical data and performed a genome-wide association analysis (see Methods, Table S1). This approach recapitulated a significant signal within chr4q21 specifically mapping to variants proximal and within the gene WDFY3 (Fig. S1A), a large gene expressed highly in the brain^21,22. The most significantly associated SNP, rs17368018, encoded a missense mutation (A>G) within the WDFY3 coding region (WDFY3^rs17368018 (WDFY3^rs), Fig. 1). We examined the Venezuelan pedigree for patients who carried this signal and identified an extended family with delayed ages of onset by 6–23 years (Fig.1A).

Fig. 1: Introduction of rs17368018 into the mouse genome recapitulates the delayed age of onset of the HD-like phenotype (See also Fig. S1. Complete statistics can be found in Table S2).

A. Graphic representation of the pedigree identified in the HD Venezuelan cohort. Only a partial pedigree is shown that indicates patients positive for HD. Square (male) or circle (female) is indicated for HD positivity (left) and the presence of the SNP (right), the CAG (Q) repeat length, and the residual age of onset (AO, +/− years) below. The reported CAG repeat length is the value obtained upon sequencing. Blue, later AO; red, earlier AO; hatched, rs17368018+; empty, rs17368018-.

B. Sequence alignment of exon 60 of human WDFY3 and exon 57 murine Wdfy3 which share 83.23% sequence identity. Grey indicates sequence differences in mouse. rs17368018 indicated by *. Variant encodes a change of A to G (green).

C. Box plots of total distance traveled by 6 m/o Q140 mice with Wdfy3^Var. RM-ANOVA with Fisher PLSD. Sex considered independent a priori. Female: Ctrl n=17, Q140 n=16, Q140::Wdfy3^Var n=18; Male: Ctrl n=16, Q140 n=15, Q140::Wdfy3^Var n=21. *, p < 0.05.

D-G. Immunostaining and quantification for D. astrocytosis (GFAP), E. microgliosis (Iba1), F. MSN (FoxP1) and G. mHtt aggregates (MAB5492). Activated microglia indicated by black arrows and scored as described in Fig. S1E. Q140 and Q140 with one or two copies of Wdfy3^Var. ANOVA with Fisher PLSD. Bar graphs of Mean±St.Dev with individual data points. n=3/genotype. *, p < 0.05.

Recreation of SNP rs17368018 in mice recapitulates disease protection

WDFY3^rs is a rare SNP with a frequency of 0.8 to 1% in the broader population (dbSNP Build 157). Coupled with the rarity of HD, we were unable to obtain sufficient patient numbers to improve the study’s power. We therefore tested the hypothesis that should this SNP affect the biology of HD as significantly as the pedigree suggests, creation of the orthologous SNP on an HD background in mice could phenocopy the outcome. Given that WDFY3 shares significant sequence identity, over 96% conservation, with murine Wdfy3 (NCBI), we identified and introduced the corresponding A>G change into Wdfy3 (Wdfy3^Var) (Fig. 1B, Fig. S1B). Wdfy3^Var mice demonstrated no change in breeding, home cage behavior, or spontaneous locomotor activity compared to littermate wild-type (WT) mice (Fig. S1C, data not shown).

To determine if rs17368018 modified HD pathogenesis, we crossed Wdfy3^Var with HD mice (Fig. S1D). We selected the Q140^+/− (Q140) model which approximates the genetic environment of the HD gene mutation²³, as we were uncertain how the SNP might exert protection. Consistent with previous reports^24,25, 6-month-old (m/o) Q140 mice demonstrated a significant hypolocomotor phenotype when compared to littermate controls (Fig. 1C). Notably, the presence of Wdfy3^Var in these mice made them indistinguishable from littermate controls (Fig. 1C). Next, we measured additional, orthogonal HD-associated outcomes, by examining for indications of astrocytosis and microgliosis in the striatum, a well-established feature of HD pathology^5,26,27. Astrocytosis was scored by the number of hyper ramified astrocytes with higher expression of glial fibrillary acid protein (GFAP)²⁸, whereas reactive microglia were identified by changes in morphology revealed by staining against Ionized calcium binding adaptor molecule 1 (Iba1)(Fig. S1E)^29,30. Mixed strain 6 m/o Q140 striata showed profound astrocytosis and microgliosis, while co-expression of Wdfy3^Var significantly blunted these indications (Fig. 1D, E). Moreover, we found that expression of the medium spiny neuron (MSN) marker FoxP1,³¹ is significantly diminished in Q140 mice, but preserved in Q140::Wdfy3^Var, thereby indicating that different indicators of brain health are yet to demonstrate pathological changes (Fig. 1F). These data suggest that the impact of rs17368018 can be translated across species, and can capture the protective effect observed in HD patients.

The genomic location of rs17368018 is within a coding exon of WDFY3, which encodes for Alfy, an adaptor protein that specifies cargo for degradation by autophagy^32,33. Alfy is required for the clearance of pre-formed aggregates in cells^32,33 including in neurons of the adult brain^34,35. Given its previously described role, we performed stereological analyses to determine if Wdfy3^Var influences accumulation of mutant Htt protein in 6 m/o Q140 striata (Fig. 1G). Notably, stereological analyses of Q140::Wdfy3^Var mice reveal few to no aggregates at this age. These data indicate that the SNP is associated with diminished aggregate accumulation in the HD brain, and suggest that the variant may enhance Alfy function.

rs17368018 increases expression of the selectivity autophagy adaptor protein Alfy by enhancing mRNA stability

Alfy is a 390 kDa scaffold protein that is characterized by a series of protein-protein and protein-lipid interaction motives^32,36, that scaffolds aggregated proteins, potentially through an interaction with Sequestesome-1/protein 62 (p62)³⁷, and the growing autophagosome membrane through an interaction with Atg12–5:16L complex³², a LC3-interacting region (LIR)³⁸ and a FYVE domain³⁹ (Fig. 2A). rs17368018 leads to an amino acid change from Ile3032Var (I3032V). Examining the predicted structure of Alfy using AlphaFold3^40,41 shows Alfy is of two distinct regions, with the N-terminal two thirds comprising of helices and the C-terminal third being globular in nature (Fig. S2A–C). I3032V falls within this region of Alfy, within the region previous shown to interact with the autophagy receptor p62 (Fig. 2A). We have previously shown that C-terminal fragment (AlfyC) is sufficient to increase aggregate clearance in an autophagy-dependent manner^32,33,39,42. To determine if I3032V enhances the ability of AlfyC to increase aggregate clearance, we overexpressed AlfyC or AlfyC(I3032V) in a previously published stable cell line expressing mCFP tagged polyQ protein⁴³ (Fig. S2E). The presence of I3032V neither diminished or enhanced AlfyC activity, suggesting that the amino acid replacement is neutral and that the benefit of the variant is not due to changes in protein function.

Fig. 2: rs17368018 increases Alfy expression by stabilizing the WDFY3 mRNA transcript, which is recapitulated in mice (See also Fig. S2).

A. Schematic representation of Alfy. Interacting regions for p62/SQSTM1 and Atg5. Atg5 allows interaction with Atg12-5:16L complex. LIR domain binds to GABARAP and LC3C. The FYVE domain. The multi-color striped regions are indicative of predicted disordered domains. I3032 falls within the p62 binding region. Abbreviations: aa: amino acid; ARM: Armadillo; ConA: Concavalin A; FYVE: Fab1, YOTB, Vac1 and EEA1; LIR: LC3 interacting region; PH-BEACH: Plecstrin homology-beige and Chediak-Higashi syndrome protein; and WD: WD40 repeat.

B. WDFY3 transcript (top) and Alfy protein levels (btm) of patient brains from individuals genotyped with the rs17368018 variant. (Top) RT-qPCR. Two-tailed student t-test reveals significantly higher levels of WDFY3 transcript expression in the variant (p=0.025). (Btm) Immunoblots of cerebellum from the same individuals with Box plots of quantification normalized to vinculin. Student t-test reveals higher levels of WDFY3 expression in variant carriers (p=0.027); n=14 WDFY3^+/+, n=7 WDFY3^rs/+.

C. Wdfy3 transcript (top) and Alfy protein levels (btm) of brains from mice in which the point mutation orthologous to rs17368018 was introduced. (top) RT-qPCR (btm) Immunoblot of brain lysates normalized to vinculin. ANOVA with Fisher PLSD. Complete statistics can be found in Table S2. n=3/genotype. *, p<0.05.

D, E. WDFY3 transcript levels from HEK293 WT (white) and HEK293 WDFY3^Var cells (blue). RT-qPCR of RNA isolated from whole cell homogenate (n=4), cytosolic (n=3) and nuclear (n=4) fractions. Normalized to WT parental cells. E. Cells treated with or without ActD. Box plots with p-values generated from Student t-test.

We have previously shown in cells, fly and mouse brain that there is a direct relationship between Alfy expression levels and aggregate clearance^32–34, and thus determined if rs17368018 changed WDFY3 expression levels. To do so, we genotyped 597 samples of varying diagnoses from the New York Brain Bank and identified 8 independent rs17368018-positive samples, of which 6 were available for subsequent analyses. RT-qPCR on both WDFY3^WT and WDFY3^rs carriers revealed increased levels of WDFY3 transcript and Alfy protein expression in brain tissue from rs17368018 individuals versus controls (Fig. 2B). To confirm the sufficiency of rs17368018 to increase Alfy expression, we examined Alfy expression in Wdfy3^Var mice. Strikingly, the mice recapitulated the human data (Fig. 2C), demonstrating increased transcript and protein levels of Alfy in mouse brain.. These data indicate that like the neuroprotection, the mechanism elicited by the variant to increase Alfy expression levels is conserved across species. Given the conservation, we hypothesized that the mechanism was through mRNA stability, which growing studies indicate can be achieved through changes to the open reading frame. To test our hypothesis, we used CRISPR base pair editing to create HEK293 cells with biallelic introduction of rs17368018 (Fig. 2D,E). Introduction of the variant was sufficient to increase WDFY3 mRNA as indicated by RT-qPCR (Fig. 2D). Cell fractionation revealed that this increase could be observed in the cytosolic but not nuclear mRNA pool, suggesting that the SNP enhanced transcript stability opposed to transcription. To pursue this further, we next determined if the increase in WDFY3 mRNA was diminished by the inhibition of transcription by Actinomycin D (ActD)^44,45. A time course of ActD administration revealed that WDFY3 mRNA was nearly unchanged across the 8 hrs, indicating that WDFY3 encodes a long-lived transcript (Fig. S2F). Using a 4hr ActD treatment, we found that the variant maintained higher WDFY3 transcript levels, suggesting that transcription is not responsible for higher expression of WDFY3 (Fig. 2E, S2G,H). Taken together, these data suggest that rs17368018 exerts protection via higher Alfy expression achieved by stabilizing WDFY3 mRNA.

Ectopic upregulation of Alfy is sufficient to capture the protective effect of Wdfy3^Var

To test this hypothesis directly, we ectopically overexpressed Alfy and compared the outcome to orthologous introduction of rs17368018. To do so, we created a knock-in of full-length hAlfy cDNA preceded by a 3xFLAG-tag into the Rosa26 locus (Rosa^hAlfy, Fig. 3A,B, S3A, and data not shown). To confirm that hAlfy can complement murine Alfy and to ensure that the FLAG tag did not interfere with Alfy function, we crossed Rosa^hAlfy/+ on an Wdfy3 null (AlfyKO) background. AlfyKO are perinatal lethal and demonstrate neurodevelopmental deficits such as a loss of the midline crossing^22,46 (Fig. S3B–D). Ectopic overexpression of hAlfy completely rescued the AlfyKO phenotypes.

Fig. 3: Ectopically increasing Alfy expression levels is sufficient to recapitulate the protective effect of rs17368018 (See also Fig. S3).

A,B. Ectopic overexpression of hALFY (hA) in mice is created through a knockin of hAlfy cDNA into the Rosa26 locus (Bl6/129Sv). A. Immunoblot for Alfy normalized to vinculin. The levels of hAlfy as detected by its FLAG tag can be observed in Fig S3A. B. Box plot reveals a gene dose dependent increase in Alfy expression. ANOVA with Fisher PLSD. Complete statistics can be found in Table S2. n=3/genotype. *, p<0.05.

C. Breeding schema to introduce hAlfy overexpression (Bl6/129Sv) onto the Q140 background (strain C57BL/6J). Resulting littermates were used in all comparisons.

D. Hypolocomotion phenotype of 6 m/o Q140 mice. Box plots of total distance traveled. RM-ANOVA with Fisher PLSD of comparison to Ctrl. Ctrl mice have either one or two copies of hAlfy. Females: Ctrl n=15, Q140 n=16, Q140::Rosa^hAlfy/+ n=14, Q140::Rosa^hAlfy/hAlfy n=12; Males: Ctrl n=13, Q140 n=13, Q140::Rosa^hAfy/+ n=10, Q140^hAlfy/hAlfy n=15. *, p < 0.05 Complete statistics can be found in Table S2.

E-H. Neuropathological phenotypes of 6 m/o Q140 mice. Two tailed t-test comparing Q140 and Q140^hAlfy/hAlfy for E. aggregated mHtt, (p = 0.009) F. astrocytosis (GFAP, p=0.007), G. microgliosis (Iba1, p=0.007), and H. decreased expression of FoxP1 (p=0.034). Bar graphs of Mean±St.Dev with individual data points. n=3/genotype. *, p < 0.05.

I. Detergent soluble mutant Htt with respect to total protein (Vinculin (Vinc) as loading control) or to WT Htt. 6 m/o striatal lysates. Bar graphs of Mean±St.Dev with individual data points. n=3/genotype. ANOVA reveals no significant effect of genotype for mHTT/Vinc (F_(2,6)=0.987, p=0.426) or mHTT/HTT (F_(2,6)=0.142, p=0.870). Abbrev. hA/+: Rosa ^hAlfy/+; Var/+: Wdfy3^Var/+

Ectopic overexpression of Alfy in Q140 mice (Fig. 3C) was consistent with our Wdfy3^Var findings, preventing the hypolocomotion in Q140 mice (Fig. 3D, S3E), and the neuropathological changes including protein aggregation, neuroinflammation and FoxP1 expression (Fig. 3E–H). Given the established function of Alfy on aggregate clearance by autophagy, we explored whether Alfy overexpression led to changes in indicators of basal autophagy. Immunoblot analyses revealed no difference in p62 or LC3-II/LC3-I levels compared to WT animals (Fig S3F). This aligns with loss-of-function studies, which showed that changing expression of Alfy does not globally alter autophagy, consistent with its role as a selectivity adaptor protein^32,34. Moreover, despite the diminished amount of aggregated proteins in the presence of increased Alfy expression, immunoblot analysis revealed that detergent soluble levels of mutant Htt remain unchanged (Fig. 3I). This is consistent with Alfy degrading the aggregated form of the protein, rather than preventing aggregate-formation or diminishing levels of the soluble protein. Taken together, our data indicate that ectopic overexpression of Alfy is sufficient to recapitulate our observations in the Wdfy3^Var mice, and further supports the model that the SNP confers protection by augmenting levels of Alfy expression, and enhances the selective turnover of aggregated mutant Htt.

Alfy upregulation is protective by increasing the turnover of aggregating proteins

The CAG expansion mutation is postulated to drive toxicity via several mechanisms including the loss of HTT function^47,48. Therefore to further examine the impact of diminishing the proteinopathy, we next assessed how Wdfy3^Var or Rosa^hAlfy affected the N171-82Q (N171) transgenic model (Fig. 4A).⁴⁹ N171 mice demonstrate an aggressive phenotype thought to be driven by the expanded polyglutamine stretch itself⁵⁰. Crossing N171 mice with either Wdfy3^Var or Rosa^hAlfy led to a significant decrease of aggregates in the striatum and cortex (Fig. 4B,C). Subsequent neuropathological analyses also revealed that the loss of aggregation was accompanied by diminished neuroinflammation (Fig. 4D,E), and preservation of FoxP1 expression (Fig. 4F). Behavioral analyses revealed that Alfy overexpression also delayed the onset of the hypolocomotor phenotype (Fig. 5A, S4A) and latency to fall on the accelerated rotarod (Fig. 5B, S4B). Therefore, increasing levels of Alfy in the N171 model augments aggregate turnover by selective autophagy, and protects against the neuropathological outcomes of the accumulating expanded polyglutamine protein.

Fig. 4. The Alfy variant or ectopic overexpression of Alfy confers protection to 5 m/o N171 mice: Neuropathology.

A. Breeding schema for the creation of N171-82Q (N171)(B6C3F1/J) mice, expressing either the WDFY3 variant (Bl6/129Sv) or ectopically overexpressing hAlfy (Bl6/129Sv). Resulting littermates were used in all comparisons.

B. MAB5492⁺ mHtt aggregates in the cortex.

C-F. Immunostaining and quantification in striatum of C. Aggregate load. D. astrocytosis (GFAP), E. microgliosis (Iba1), and F. FoxP1 expression.

ANOVA with Fisher PLSD. *, p < 0.05. n=3/genotype. Complete statistics can be found in Table S2. Bar graphs of Mean±St.Dev with individual data points. n=3/genotype.

Fig. 5. The Alfy variant or ectopic overexpression of Alfy confers protection to N171 mice: Behavior and Survival (See also Fig. S4).

A. Hypolocomotor phenotype in 3 m/o N171 mice. Box plots of total distance traveled for male N171 in the presence or absence of the variant (Alfy Var, left) or ectopic overexpression (Alfy OE, right). Female N171 mice failed to become hypolocomotive in either group and were not analyzed further. Right: Ctrl n=11, N171 n=11, N171::Wdfy3^Var/Var n=13; Left: Ctrl n=20, N171 n=11, N171::Rosa^hAlfy/+ n=14. *, p < 0.05. Complete statistics can be found in Table S2.

B. Accelerated rotarod deficits in N171 mice. p-values generated from RM-ANOVA of the multiple trials per age, and posthoc analyses are listed below each line graph. Males: 2 and 3 m/o n=11 mice/genotype, 4 m/o n=11 Ctrl, n=10 N171, n=9 N171::Rosa^hAlfy/+, 5 m/o n=11 Ctrl, n=7 N171, n=6 N171::Rosa^hAlfy/+.

C. Survival curve. N171 (median age = 17wks) vs. N171::Wdfy3^Var/Var (median age = 26wks): Mann-Whitney U for time: Z=−6.4, p<0.001. N171 vs N171::Rosa^hAlfy/+ (median age = 21wks): Mann-Whitney U for time: Z = −4.524, p<0.001. Mixed sex cohorts. N171 n=66; N171:: Wdfy3^Var/Var n=29; N171::Rosa^hAlfy/+ n=61.

A unique feature of the N171 mice is that they suffer from premature lethality. The cause of the early demise is uncertain but monitoring fecal deposits and performing pathologic examinations suggest that gut motility defects might play a role (data not shown). Interestingly, increasing levels of Alfy prolongs normal fecal deposition and leads to a significant lifespan expansion of the N171 mice (not shown and Fig. 5C). Consistent with our cell based data, these data suggest that diminishing protein accumulation may protect across different tissues and not only the brain.

Upregulation of Alfy is protective in a mouse model of synucleinopathy

A shared feature across adult-onset neurodegenerative diseases is that a once functional, soluble protein becomes entangled in an insoluble complex, that ultimately consolidate into the inclusion bodies that are known neuropathological hallmarks. We previously found that autophagy degrades aggregated structures that are predominantly less than 1 micron³², and with others, found that Alfy can target different aggregated structures for autophagic degradation^{32,37,51–53}. This suggests that the biochemical and physical features of the structure, rather than the identity of the aggregating protein, are important for selection by Alfy. We therefore determined if augmenting Alfy levels is protective against two additional proteinopathies, synucleinopathies and tauopathies.

Synucleinopathies are a group of neurodegenerative diseases characterized by the presence of intracytoplasmic accumulates of aggregated αSyn and include disorders such as PD and Dementia with Lewy Bodies (DLB)⁵⁴. A prominent model for synucleinopathies is the α-Syn PFF model^55–57. Intracerebral introduction of α-Syn PFFs leads to a change of endogenous α-Syn into states that are reminiscent of pathological αSYN found in patients, such as the accumulation of phosphorylated αSyn at serine 129 (pS129-αSyn). WT and Rosa^hAlfy/hAlfy mice received a single striatal injection of mouse αSyn PFFs and were euthanized 120 days post-injection (Fig. 6A). Consistent with the literature^55,58–60, PFF-injections in WT mice led to the accumulation of pS129-αSyn) in the injected striatum, as well as regions of innervation including the cortex, amygdala, and substantia nigra pars compacta (SNpc) (Fig. 6C–G, S5). The aberrant accumulation was accompanied by a loss of tyrosine hydroxylase (TH)- or Nissl-positive neurons in the SNpc, attesting to their toxicity^55,56. In contrast, PFF-injections in Rosa^hAlfy/hAlfy mice revealed that despite a similar aggregation load at the injection site with WT (Fig. S5A), as well as similar staining for endogenous αSyn (Fig. S5D), pS129-αSyn accumulation appeared attenuated in projection areas (Fig. 6C, S5B–D). Subsequent quantification in SNpc revealed that there was significantly less aggregation in the Rosa^hAlfy/hAlfy mice (Fig. 6C,D), demonstrating that Alfy overexpression can also diminish pS129-αSyn accumulation. Moreover, quantification of neurons revealed a significantly greater number of TH-positive and Nissl-positive dopaminergic neurons present in PFF-injected Rosa^hAlfy/hAlfy mice. These data demonstrate that Alfy overexpression reduces aggregate burden and delays neurodegeneration in the αSyn PFF model.

Fig. 6. Increasing Alfy levels protects against αSYN accumulation and neurotoxicity in a PFF model of synucleinopathy (See also Fig. S5).

A. Schematic representation of experimental design. WT and Rosa^hAlfy/hAlfy littermates generated from the intercross of Rosa^hAlfy/+ mice were used.

B. Summarized representation of striatal injection site (red) per animal.

C. pS129-αSYN counterstained with Nissl in SNpc of mice injected with PFFs. D. Quantification of aggregates per field. Bar graphs of Mean±St.Dev with individual data points. n=4/condition. Two-tailed student t-test, *, p<0.001.

E. Images of TH+ and Nissl+ neurons in the SNpc. Quantification of F. TH+ and G. Nissl+ cells contralateral (C) and ipsilateral (I) to injection site. RM-ANOVA. *, p<0.05. Complete statistics can be found in Table S2. Bar graphs of Mean±St.Dev with individual data points. n=4/genotype.

Upregulation of Alfy is protective in a mouse model of tauopathy

Another commonly occurring proteinopathy is the abnormal accumulation of the neuronal microtubule-associated protein tau. Post-translational modifications of tau favor conformational changes in the protein, leading to fibril formation and insolubility⁶¹. The accumulation of phosphorylated, paired helical filaments (PHFs) of tau are found across numerous diseases from Alzheimer’s disease and Frontotemporal dementia (FTD), to rare disorders such as Progressive Supranuclear Palsy and Pick’s disease. The PS19 transgenic mouse model expresses the T34 isoform of tau with one N-terminal insert and four microtubule binding repeats encoding the human P301S mutation, which causes FTD with parkinsonism^62–64. These mice demonstrate an AT8-positive tauopathy with neurodegeneration of forebrain regions⁶². PS19 mice were crossed with Rosa^hAlfy/hAlfy mice, then characterized at 34 weeks-old for the accumulation of AT8-positive structures and neurodegeneration (Fig. 7). Consistent with the reduction of mutant Htt and pS129-αSyn, increased levels of Alfy prevented the accumulation of AT8-positive structures in the somatosensory (SS) cortex and hippocampus (Fig 7B). This difference was apparent despite showing no difference in soluble levels of total tau or AT8-positive phosphorylated tau (pTau), suggesting that diminished AT8 staining was neither due to diminished total tau levels, or increased turnover of soluble pTau (Fig 7D). In addition, the ratio between pTau and tau were also unchanged suggesting that increasing levels of Alfy does not change phosphorylation of the protein. Again, similarly to the models of HD and αSyn toxicity, diminished pTau accumulation was accompanied by diminished toxicity (Fig. 7C, E). PS19 mice have been reported to have approximately 15% cell loss in the perirhinal cortex by 12 weeks of age⁶⁵. At 34 weeks, we found that there was an even further loss of cells, but in the presence of Alfy overexpression, the loss was prevented (Fig. 7C). Moreover, like for N171 mice, ectopic expression of Alfy or the presence of the variant enhanced lifespan, again with a greater effect shown by the variant. Taken together, Alfy overexpression can confer resistance across a broad range of aggregating proteins from mutant Htt, pS129-αSyn, and PHF tau.

Fig. 7. Increasing Alfy levels protects against accumulation and neurotoxicity in the PS19 mouse model of tauopathy.

A. Breeding schema to introduce hAlfy (hA) (Bl6/129Sv) onto the PS19 model (B6C3F1/J).

B. AT8 staining from somatosensory (SS) cortex (top) and hippocampus CA1 (btm) from PS19 and PS19::Rosa^hAlfy/hAlfy. Counterstained with Nissl. Higher magnification below.

C. Quantification of Nissl+ neurons. ANOVA with Fisher PLSD. *, p < 0.05. Complete statistics can be found in Table S2. Bar graphs of Mean±St.Dev with individual data points. n=3/genotype.

D. Immunoblotting of detergent soluble HT7 (human) tau and AT8+ phosphorylated tau (pTau). Vinculin loading controls. Rosa^hA/hA littermate control (Ctrl) included to ensure specificity of HT7. Below, relative density of Tau and pTau and their ratio. Bar graphs of Mean±St.Dev with individual data points. n=3/genotype.

E. Survival curve of PS19 mice. PS19 (median age = 10.5 mos) vs. PS19::Rosa^hAlfy/+ (median age = 12.5mos): Mann-Whitney U for time: Z=−2.196, p=0.0281. PS19 vs PS19::Wdfy3^Var/+ (median age = 21mos): Mann-Whitney U for time: Z=−2.509, p=0.0122. Mixed sex cohorts. PS19 n=28; N171:: Wdfy3^Var/+ n=6; N171::Rosa^hAlfy/+ n=23

The impact of Alfy overexpression on gene expression is revealed on the background of a proteinopathy

By increasing levels of the autophagy adaptor protein Alfy and augmenting aggregate-turnover, we find that this delays neurotoxicity. To gain insight into how diminishing aggregate burden might be protective, we returned to the Q140 model which has a well-established transcriptional signature via total RNA sequencing²⁵.

We first asked if increased levels of Alfy leads to discrete changes in differentially expressed genes (DEGs) in the striata of WT versus Q140 mice (Fig. 8A–C, Table S3). The presence of one copy of hAlfy led to 133 up- and 67 down-regulated genes relative to WT mice, suggesting a modest impact of Alfy expression. In contrast, on the Q140 background, the number of changes increased in a dose-dependent manner from 173 up- and 132 down-regulated genes in Rosa^hAlfy/+::Q140 vs. Q140, to 921 and 773 in Rosa^hAlfy/hAlfy::Q140 vs. Q140 mice, respectively (Fig. 8A–C, Table S3). In addition to the change in DEG number, the −log10 False Discovery Rate (FDR) increased, suggesting increased certainty. As to the nature of the changes (Fig. 8D–G, Table S4–5), Gene Ontology (GO) analyses coupled with Search Tool for the Retrieval of Interacting Genomes (STRING) analyses revealed that Alfy over-expression, on the Q140 but not WT background, leads to changes that increased expression of inter-related genes, all involved in translation and ribosomal biogenesis; pathways shown to be dysfunctional broadly across neurodegenerative diseases^66–74 (Fig. 8D,E, Tables S3, S4). Decreased gene expression observed on both WT and Q140 backgrounds are involved in angiogenesis and cell migration. Taken together, these findings suggest that Alfy may negatively regulate genes involved in angiogenesis and cell migration, whereas its overexpression can correct the disruption in translation and ribosomal dyshomeostasis that occurs in the presence of a proteinopathy.

Fig. 8. Increasing Alfy levels leads to changes in ribosomal biogenesis, but not on the larger transcriptome-wide signature of HD (See also Fig. S6,S7).

A-C. Volcano plots of DEGs evoked by Alfy overexpression in A. WT vs. B,C. Q140 mice. See Table S3. Although several gene changes were noted by Alfy overexpression in a WT background, no discernable pattern were observed using independent approaches such as Gene Ontology or STRING analyses.

D-G. Gene ontology analyses and protein interaction network of DEG lists based on Table S3. D, G. GO analyses of D. increased and G. decreased gene expression changes from all comparisons. E, F. Protein-protein interactions using Cytoscape via the STRING database, v12.0. E. increased and F. decreased gene expression changes due to homozygous overexpression of hAlfy in Q140 mice to illustrate the inter-relationship of the DE increased and decreased genes with E. ribosomal biogenesis and F. angiogenesis.

The transcriptomic study we performed mirrored a previously published study that established an HD transcriptomic signature in the Q140 mice²⁵. In light of the positive impact Alfy has on the neuropathological and behavioral phenotypes we examined, we also asked whether increasing Alfy levels altered the Q140 transcriptional signature previously described²⁵ (Fig. S6). To do so, we first performed transcriptome-wide correlation analysis between the differential expression statistics of the striata of our Q140 vs. littermate WT mice and the previously published comparison of Q140 vs CAG20 (Q20)²⁵ (Fig. S6A). This revealed that our Q140 mice, despite being of mixed background (C57Bl6x129Sv), largely maintained the transcriptional signature as previously reported²⁵, with a positive correlation coefficient of 0.78. (Fig. S6A). Given the similarities, we determined the impact of Alfy overexpression on the HD transcriptional signature (Fig. S6B–D). Correlation analyses revealed that Alfy overexpression did not substantially alter the Q140 transcriptional signature (Fig. S6C,D). This was consistent with principal component analysis, which identified Q140 to account for the largest possible variance rather than Alfy levels (Fig. S6B). Although Alfy overexpression left the HD transcriptional signature largely unmodified, we identified a subset of Q140 dysregulated genes that are exacerbated (as indicated by a negative correlation) or rescued (as indicated by a positive correlation) (Fig. S6C,D, S7, Table S6,7). GO analyses revealed that transcripts involved in protein metabolism and focal adhesion protein were rescued, whereas transcripts involved in angiogenesis were exacerbated, but changes were minimal. Overall, these data suggest that the transcriptional changes observed in Q140 mice may reflect the disease-specific state evoked by the CAG expansion mutation in the Htt gene, whereas the modifying effect of Alfy overexpression is post-transcriptional and at the level of the proteinopathy.

Discussion

By leveraging the strengths of human and mouse genetics, we identified a potent modifier of neurodegenerative diseases that evokes protection by preventing protein accumulation, providing critical insight into the relevance of diminishing protein-mediated toxicity and positive outcomes in the adult brain.

The starting premise of this study aims to determine if a rare variant, rs17368018, underlies the delayed age of onset observed in an extended family of HD. Although modeling implicated this SNP as most highly correlated with the delayed age of onset (Fig. S1A), our inability to identify enough patients to reach power required an orthogonal approach to test our hypothesis. WDFY3 is exquisitely conserved across vertebrates, and thus mouse modeling permitted us to eventually capture the molecular consequence of this rare coding variant; it increases expression of the protein product by increasing transcript stability (Fig. 2). Although SNPs within a coding sequence are more associated to a change in protein function, SNPs can alter the transcript expression. For example, a synonymous SNP in corneodesmosin (CDSN) increases transcript stability by changing the sequence within a stability motif⁷⁵, leading to changes in RNA-protein interactions. Although the affected sequence in WDFY3 differs from CDSN, a similar mechanism might be at play; RNAs naturally interact with RNA-binding proteins (RBPs), RNA-protein complexes and other RNAs^76,77. Whether this is due to changes in direct binding or changes in RNA secondary structure will require a further and important line of study, since the mechanism may outline novel therapeutic approaches that can translate the findings we present here.

Alfy is highly expressed in the brain, and like other genes with similar profiles, it is essential for CNS development and maintenance^{22,46,53,78,79} and the loss of function has been implicated in neurodevelopmental disorders and developmental delay^46,79–81. Using heterozygous depletion of Wdfy3, we previously found that reducing Alfy was sufficient to accelerate aggregate accumulation and phenotypic onset in HD mice³⁴. Unlike the loss of function studies, expression of the SNP or ectopic overexpression of Alfy had no notable impact on peripheral tissue, as both the Wdfy3^Var and Rosa^hAlfy mice were indistinguishable from controls. Furthermore, transcriptome-wide analyses also revealed that there is little impact on basal transcription as well, consistent with the greater health-span observed within the Venezuelan patients carrying this variant. Notably, increasing the adaptor protein revealed that global upregulation of autophagy may not be necessary to harness its therapeutic potential. Given the broad homeostatic role of autophagy that may differ across different cell types and tissue, autophagy may be too complex to consider affecting protein accumulation alone. This has been highlighted by studies upregulating autophagy globally by decreasing levels of Rubicon, which can augment autophagic clearance of aggregated proteins in brain, but can have problematic effects in peripheral tissues, and lead to metabolic disease^82,83.

Although ectopic Alfy overexpression captured the general protective features of rs17368018, some of our findings indicate that the latter is more protective, suggesting that Alfy may be acting beyond autophagy alone. The difference is most evident in the survival of the N171 and PS19 models (Fig. 5C, 7E). Why the variant might be more protective is unclear. Ectopic overexpression limits genetic regulation of Wdfy3, but studies suggest that Alfy may also impact the turnover of NIPSNAP-positive mitochondria⁸⁴ and improve efferocytosis⁸⁵, suggesting potential anti-inflammatory outcomes. Recently, BEACH domain containing proteins have been shown to act as cargo sorting adaptors in secretory and endocytic pathways⁸⁶, consistent with the loss of the ability of AlfyKO neurons to respond to pathfinding cues during development²². How these other functions might be potentiated by increasing Alfy still needs to be explored. Our transcriptional study, however, suggests that Alfy’s ability to prevent protein accumulation is essential by maintaining ribosomal homeostasis. The direct disruption of protein translation due to the CAG repeat^87–90 has been reported and similar events due to α-Syn and tau are also suggested^73,74,91,92.

The need for better understanding of Alfy biology notwithstanding, we find that for three distinct, aggregation-prone and disease-causing proteins, slowing accumulation can be significantly protective against neurodegeneration. The ability to demonstrate the significance of protein aggregation and accumulation on disease pathogenesis, especially as a therapeutic target, has been impressively frustrating. The rationale underlying this study stems from a continuous line of cell- and animal-based studies showing that Alfy is required to clear aggregated proteins in the adult brain^32–34. Through the design of this study, we cannot determine if aggregate formation is slowed by increasing Alfy expression, but it clearly demonstrates that protein accumulation is detrimental, and its prevention is a potent way to help maintain function. How increasing Alfy levels might influence the different proteinopathies might vary, and for proteins such as oligomeric αSyn, it remains unclear if Alfy alters propagation kinetics or the end stage accumulation event. Given its role in macrophages⁸⁵, studies implicating the ability of microglia to uptake αSyn⁹³ may be at play. Future studies that overexpress Alfy in a cell type specific manner will permit greater mechanistic insight, and may indicate that augmenting its expression can complement existing therapeutics in development.

Taken together, by applying the complementary nature of human and mouse genetics, we address two major questions in adult neurodegeneration: the importance of the essential pathway autophagy and the significance of protein accumulation on pathogenesis. We believe these findings highlight the strength of mouse modeling and how it can serve to complement patient-related findings. As such, we believe they will serve as a strong foundation from which a potential therapeutic avenue for combatting these devastating diseases might be born.

Resource Availability

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Ai Yamamoto (ai.yamamoto@columbia.edu).

Material availability

All unique/stable reagents generated in this study will be made available by request to the Lead Contact with a completed Materials Transfer Agreement. We are glad to share mouse lines and other resources generated in this study, with reasonable compensation by requestor for shipping.

Data and code availability

The RNA sequencing datasets generated during the current study have been made immediately available in the Gene Expression Omnibus (GEO), GSE281823. No new code has been generated as part of this publication. Any additional information required to reanalyze the data reported in this work paper is available from the Lead Contact upon request.

STAR METHODS

Experimental model and study participant details

Human participants:

The participants are from the Venezuelan HD kindred, whose documentation began in 1979 and thanks to an international, interdisciplinary team that had traveled annually to Maracaibo, Venezuela. Genetic and clinical data from these kindreds have been used in this investigation and have been extensively described elsewhere.^20,94–96 All experiments with genetic material have been granted IRB exemption by the MIT Committee on the Use of Humans as Experimental Subjects. Patients were de-identified and only identified by code.

Human brain samples:

Carriers of rs17368018 were identified by genotyping 597 samples of varying diagnoses from the New York Brain Bank (NYBB) as described on Table S8. We identified 7 independent samples of varied patient histories positive for the SNP. Tissue from the Broadmann Area 9 (BA9) frontal cortex and cerebellum was used for RT-qPCR and immunoblotting, respectively. All experiments with patient material have been granted IRB exemption status by virtue of working with de-identified material from deceased individuals

Mice:

Animals were bred and housed in facilities at the William Black Medical Research Building. Same-sex animals of mixed genotypes, maximum 5 adult mice per cage, are housed in a humidity- and temperature-controlled room, with access to food and water ad libitum. Animals are maintained on a 12-hour light/dark cycle. For breeding and housing prior to experiments, animals were maintained on a cycle with lights on between 7:00 A.M and 7:00 P.M. For experimental and behavioral analysis, animals were maintained on a cycle with lights off between 9:00 A.M. and 9:00 P.M. All experiments were reviewed and approved by the Columbia University Medical Center’s Institutional Animal Care and Use Committee (IACUC).

Generation of conditional Alfy mice (Alfy^flox/flox, strain: Bl6/129Sv) were previously described^22,34 and Hprt^Cre/+ mice (strain: 129S1, stock no. 004302), were obtained from the Jackson Laboratory (JAX). By crossing Alfy^flox/+ males with Hprt^Cre/+ females we created the potential of Alfy heterozygous constitutive knockout (Alfy^Δ/+, strain: Bl6/129Sv)²². Conditional human Alfy knockin mice were generated by Ingenious Targeting Laboratory (Rosa^flox-hAlfy/+, strain: Bl6/129Sv). Targeted iTL BA1 (hybrid: C57BL/6 X 129Sv) embryonic stem cells were microinjected into CD-1 blastocysts. Resulting chimeras were mated to wild-type C57BL/6N mice to generate F1 heterozygous offspring. This knockin line expresses full-length human ALFY (hALFY) that is fused with a 3xFLAG tag. This sequence is preceded by a stop cassette flanked with LoxP sites, and was knocked into the Rosa locus and is driven by the Rosa promoter. Mice with a heterozygous hALFY allele (Rosa^flox-hAlfy/+) were crossed to obtain homozygous Rosa^{flox-hAlfy/flox-hAlfy} mice. Crossing with HPRT^Cre/+ females removed the floxed stop cassette, allowing for expression of hAlfy (Rosa^hAlfy/+). Wdfy3^Var/+ mice (strain Bl6/129Sv) were generated from Ingenious Targeting Laboratory. Targeted iTL HF4 (129/SvEv x C57BL/6 FLP) hybrid embryonic stem cells were microinjected into C57BL/6 blastocysts. Resulting chimeras were mated to C57BL/6 WT mice to generate Germline Neo Deleted mice. This knockin mouse model expresses the T>C point mutation identified in humans consisting of a T > C change at exon 59, codon 4090, at the Wdfy3 locus and driven by the Wdfy3 promotor. Q140 mice (strain C57BL/6J, stock no. JAX027409), N171-82Q mice (strain B6C3F1/J, stock no. JAX003627) and PS19 P301S mice (strain B6C3F1/J, stock no. JAX008169) were obtained from the Jackson Laboratory. Q140 mice contained a floxed neo cassette and were crossed with HPRT^Cre/+ females (C57BL/6J) resulting in deletion of the neo cassette. After crossing these models with either Rosa^hAlfy/+ or Wdfy3^Var/+ lines, the strains of all animals used in the study were as follows: Q140 (Bl6/129Sv), N171-82Q (Bl6/C3/129Sv), and PS19 (Bl6/C3/129Sv). PFF model was created in Alfy^flox/flox, and WT littermates created by intercrossing Alfy^flox/+ mice (strain: Bl6/129Sv)

Cellular models:

HEK293T WT and WDFY3^Var cells were maintained in high-glucose Dulbecco's modified eagle medium (DMEM; Life Technologies) supplemented with 20% fetal bovine serum (FBS; Life Technologies) at 37°C and in a 5% CO2-containing atmosphere. Tet-regulatable 103QmGFP cells were maintained as previously described⁴³.

METHOD DETAILS

Blinding, Replicates and Controls:

All human samples were de-identified. Identification codes of mice were assigned upon weaning prior to genotyping. All mice experiments were performed blind to genotype, with littermates and cagemates used as controls. Genotypes were revealed at time of analysis. Sex and age of mice, as well as experimental replicates are listed under each experiment and figure legends. For cell culture experimental replicates are from independent passages, whereas samples generated from the same passage number are considered a technical replicate.

Genome-wide genotyping and imputation of Venezuelan patients:

DNA was extracted from EBV-transformed lymphoblastoid cell lines established from blood samples of the Venezuelan individuals using the Qiagen Blood and Tissue DNeasy Kit with a fine-mapping approach that combined whole genome sequencing and an Illumina Core Exome single nucleotide polymorphism (SNP) array across 440 HD patients. Genotyping with CoreExome array was performed at the New York Genome Center on a HiScan Illumina machine. Population-based imputation was performed using Minimac4 to impute genotypes from the publicly available TopMed R2 reference panel and whole genome sequences from select Venezuelan patient samples.

Genome-wide association study of residual age of onset:

To perform quantitative genome-wide association testing using the SAIGE software package⁹⁷, a linear mixed model regression approach was implemented using residual age at motor onset as a phenotype and allowing for adjustment of empirical pairwise relatedness as well as other covariates, such as sex and population structure. We used age at onset of motor signs and CAG repeat length to derive the residual age at onset. In order to subtract the effects of CAG repeats from the age at onset of motor signs, we used a phenotype model previously developed⁹⁸ through a linear regression of natural log-transformed age at motor onset to CAG repeat length. In order to reduce skewness in the distribution of residual age at onset to more closely model a theoretical normal distribution, a Yeo-Johnson power transformation was applied to the distribution.

Reverse transcription (RT) and Quantitative PCR (qPCR) of ALFY: Human brain samples:

Total RNA was extracted from approximately 25mg of tissue using the total RNeasy kit (Qiagen). Residual genomic DNA was removed on the column using the RNase-Free DNase Set as described in the RNeasy kit. cDNA was reverse transcribed using the iScript cDNA synthesis kit (BioRad). Quantitative PCR was performed on a LightCycler 480II machine (Roche) using the Luna Universal qPCR master mix (NEB) and primer pairs specific to human WDFY3 and RPLP0 as a housekeeping gene for normalization. Ct values were calculated using the 2nd derivative method.

Mouse brain samples:

Total RNA was extracted using the RNeasy Plus Mini Kit (Qiagen), cDNA was reverse transcribed using random hexamers and SuperScriptIII reverse transcriptase (Invitrogen). qPCR was performed using KAPA SYBR FAST qPCR kit (Roche) primers against mouse ALFY (Mm_Wdfy3_1_SG, cat no QT00170583) or mouse TBP (Mm_Tbp_1_SG, cat no QT00198443)(QuantiTect) as a housekeeping gene for normalization. WDFY3^Var cells: Total RNA was extracted from WT and WDFY3^Var HEK293T cells using the total RNeasy kit (Qiagen). 1 ug or 2 ug of RNA was loaded into the reverse transcription reactions performed with Smartscribe Reverse Transcriptase (Takara). The program used is as follows; 90 min at 42 °C followed by 15 min at 72 °C. After RT, samples were diluted by 1:10 then used for qPCR with KAPA SYBR FAST or Fisher FAST SYBR mix. Primers were used to target human ALFY, c-Myc (short-lived transcript), and GAPDH (housekeeping gene/long-lived transcript). qPCR cycling conditions: 95 degrees for 3 min, 40 cycles of 95 degrees for 3s, 60 degrees for 30s.

Generation of PFF model:

Four-month-old WT and Rosa^hAlfy/hAlfy mice were anesthetized and stereotaxically injected in one hemisphere with 500 nL (2.5 ug) of fibrillized mouse αSyn (pre-formed fibrils, PFF), as previously described^99,100. A single needle insertion was used to target the left dorsolateral striatum (unilateral injection using the following coordinates relative to bregma; L, 2.3mm; A 0.9mm; and V 3.4mm). Animals were monitored regularly following recovery from surgery, and sacrificed at 120 days post-injection by transcardial perfusion for histological analysis.

Genotyping:

Genomic DNA was isolated from ear punch biopsies. Samples were lysed overnight at 55°C. DNA was precipitated by adding 350 μL isopropyl alcohol (Thermo Fisher) to the digested sample. Tubes were then inverted to mix thoroughly, followed by centrifugation at 4°C, 14,000 rpm for 30 min. Once the supernatant has been decanted, the pellet was washed with 800 μL 80% ethanol (Decon Laboratories) and spun at 4°C, 14,000 rpm for 10 min. The supernatant was carefully removed and the DNA pellet was air-dried for 5 min at room temperature. Samples are resuspended in 50 μL of nuclease free water. PCR was performed using DreamTaq Green Master Mix (Thermo Fisher). Cycling conditions for Alfy flox, Rosa^hAlfy, and Wdfy3^Var mice are as follows: 94 degrees for 2 minutes (min); 10 cycles of 94 degrees for 15 seconds (s), 65 degrees (decreasing by 1 degree/cycle) for 45s, and 72 degrees for 45s; 25 cycles of 94 degrees for 15s, 55 degrees for 30s, 72 degrees for 45s; 72 degrees for 7 min.

Mouse Behavioral Testing:

Animals were relocated and acclimated to the reverse-cycle Satellite Animal Facility one month prior to behavioral testing. Lights were off between 9 AM and 9 PM so that longitudinal assays could be conducted during the animals dark/active hours. All behavioral assessments were performed between hours 3 and 8 of the dark cycle. Littermates of the same sex were randomly assigned to experimental groups at weaning, and experimenters were blinded to genotype. Mice were maintained in group housing, and experiments were performed by cage.

Open field:

Mice were habituated to the behavior room for 30 minutes prior to testing. Locomotor activity (including total distance traveled) was recorded by an automated monitoring system using equipment and software from Med Associates. Each mouse was tested for a single exposure of 2 hours in the open field arena (43.2 cm × 43.2 cm × 30.5 cm) (Med Associates). Data was binned into 5-minute intervals. The Q140 cohort was assessed at 6 months old, and the N171-82Q^+/− cohort was assessed at 3 months old.

Accelerated rotarod test:

For the Accelerated Rotarod assay, the Mouse Rota-Rod apparatus (Ugo Basile) was used. The rod consists of textured drums to avoid slipping, and allowed 5 mice to be trained simultaneously. Training consisted of one day of initial exposure where each mouse performed three, 5-minute trials with the first trial at a constant speed of 5 rpm, a second trial with speed accelerating from 5 rpm to 15 rpm, and a third trial with speed accelerating from 5 to 25 rpm. Animals were given 10-minute rest intervals between trials. Training day was followed by three consecutive days of testing on the accelerated rotarod with speeds accelerating from 5 to 40 rpm over 5 minutes for a total of three trials each day. Mice were allowed 1 fall from the rotarod and on the second fall, the trial was terminated and latency to fall was recorded. Training was only conducted during the first exposure at 2 months, and testing was repeated every 4 weeks from 2 to 5 months old.

Fixed Tissue Preparation:

Mice were deeply anesthetized with isoflurane and transcardially perfused with 25 mL of ice-cold phosphate buffer saline (PBS) followed by 25 mL of 4% paraformaldehyde (PFA) in PBS. Brains were dissected and post-fixed in 4% PFA overnight at 4 °C, then washed in PBS and stored in 30% sucrose in 1X phosphate buffer (PB) for 72 hours minimum, until used for processing. Brains were snap-frozen in powdered dry ice then cryoprotected in OCT embedding medium (Thermo Fisher). Serial sections 30 μm thick were cut in sets of 8 through the striatum, and sets of 4 through the substantia nigra on a Leica CM 1950 cryostat and stored at 4°C in 1x PB containing 0.02% sodium azide as a preservative.

Tissue preparation for detergent soluble fractions:

Tissue (50–500 mg) was placed in a glass dounce homogenizer with an equal volume of 1X phosphate buffer saline (PBS) containing 1X Halt protease inhibitor cocktail (ThermoFisher) and disaggregated with 40 pumps of a pestle. The suspension was transferred to a tube and an equal volume of detergent (2% Triton X-100 [Tx-100] in PBS) was added to bring the final detergent concentration to 1%Tx-100 in PBS (PBS-Tx), and incubated for 30 minutes on ice, then spun in an Eppendorf 5417R centrifuge at 1000 rpm for 5 minutes. To segregate the Tx-100 soluble versus insoluble fraction, the supernatant (S1) was centrifuged at 14,000 rpm for 5 minutes. The resulting supernatant (S2) represents the detergent soluble fraction.

Immunoblotting:

20 μg of total protein lysate was resolved by SDS-PAGE (Thermo Fisher) at 180V. Proteins were transferred to PDVF membranes (Thermo Fisher) for 1.5 hours at 0.4A, and blocked for 1 hour in 3% BSA (Thermo Fisher) in 1x PBS with 0.1% Tween (Thermo Fisher) at room temperature. Membranes were probed overnight at 4 °C with primary antibodies. Primary and secondary antibody dilutions contained 0.1% PBS-Tween. Following three washes, membranes were incubated with HRP-conjugated secondary antibody (Thermo Fisher) for 1.5 hours at room temperature. Bands were visualized by the Clarity Western ECL Substrate (BioRad) and detected using the BioRad VersaDoc imaging system. Band intensities were analyzed using ImageJ and normalized to a loading control (vinculin) probed on the same blot

Immunohistochemistry:

Representative sections throughout the brain were stained. Sections were washed in PBS containing 0.02% Triton-X, followed by antigen retrieval in sodium citrate, pH 9.0 boiling at 70°C for 20 minutes. Sections were washed three times then endogenous peroxidases were blocked with 1% hydrogen peroxide in 0.5% Triton-X PBS for two 10-minute washes. Sections were washed three times before blocked in 0.4% Triton-X for 1 hour at room temperature, then incubated overnight in primary antibody dilutions containing 0.4% Triton-X PBS. For MAB5492 staining, sections were blocked using a Mouse on Mouse (M.O.M.) Immunodetection kit (Vector Labs) to reduce endogenous mouse Ig staining. These sections were briefly washed with contained M.O.M. Protein Concentrate (Vector Labs) in 0.4% Triton-X PBS before incubating overnight in primary antibody dilutions that also contained Mouse on Mouse Protein Concentrate in 0.4% Triton-X PBS. Sections were washed 3 times in PBS then were incubated in biotinylated secondary antibody at room temperature for 1 hour. Signal was amplified using the Vectastain ABC Kit (Vector labs) and detected using diaminobenzidine (DAB) (10 mg/25mL) in phosphate buffer containing 0.00001% H₂O₂. Following the DAB reaction, sections were washed in phosphate buffer then mounted on glass slides and air-dried. Once thoroughly dried, slides were dehydrated in ascending grades of ethanol, cleared in xylene (Thermo Fisher), and cover slipped with Permount mounting medium (Thermo Fisher).

Stereology:

Unbiased stereological counts were obtained from the striatum using Stereo Investigator Software (MBF Bioscience). The optical dissector method was used to count profiles in an unbiased random selection of serial sections in a defined volume of the striatum, cortex, substantia nigra pars compacta (TH/Nissl only), or perirhinal cortex (Nissl only). Total numbers were calculated using optical fractionator estimations and the variability within animals was assessed via the Gundersen Coefficient of Error (< 0.1). For all striatal stereological counts, the left hemisphere was traced at 2x magnification. Parameters for individual immunostains/structures are as follows:

mHtt inclusions (MAB5492):

Inclusions were classified as cytosolic or intranuclear and combined for total mHtt aggregate scoring. Sections were counterstained with Nissl following immunolabeling. The counting parameters involved a 300 × 300 grid with a 50 × 50 counting frame at 60x magnification, and a fixed tissue thickness of 30 μm.

FoxP1 positivity:

FoxP1 staining intensity was classified as weak, medium, and strong signal. Medium and strong nuclei counts were compared over total positive cells to assess variability across animals. The counting parameters involved a 250 × 250 grid with a 100 × 100 counting frame at 40x magnification, and a fixed tissue thickness of 30 μm.

Reactive astrocytes (GFAP):

The counting parameters involved a 500 × 500 grid with a 300 × 300 counting frame at 40x magnification, and a fixed tissue thickness of 30 μm²⁸.

Activated microglia (IBA1):

Concurrent with cell counting, Iba1-positive microglia were categorized as one of four types (representative images can be found in Fig. S1): Type 1: thin ramified processes, Type 2: thick long processes, Type 3: stout processes, or Type 4: round/ameboid shapes^29,30. Type 1/2 are considered ramified/quiescent and therefore grouped together, whereas Type 3/4 are considered activated. The counting parameters involved a 250 × 250 grid with a 100 × 100 counting frame at 40x magnification, and a fixed tissue thickness of 30 μm.

TH-positive neurons (SNpc):

Sections were counterstained with Nissl following TH immunolabeling. Glial nuclei were excluded from Nissl-positive neuronal counts based on size. Parameters involved a 50 μm × 50 μm counting frame systematically distributed across a grid 150 μm × 150 μm within the injected and uninjected SNpc^56,57. Neurons were counted at 60x magnification at a fixed tissue thickness of 30 μm.

Nissl-positive neurons (Prh):

Sections were counterstained with Nissl following AT8 immunolabeling. Glial nuclei were excluded from Nissl-positive neuronal counts based on size. Parameters involved a 40 μm × 40 μm counting frame systematically distributed across a grid 200 μm × 200 μm within the right hemisphere perirhinal cortex⁶⁵. Nissl-positive neurons were counted at 40x magnification at a fixed tissue thickness of 30 μm within 12–15 counting windows.

Counting phosphorylated α-synuclein inclusions (pSer129-positive aggregates):

Stereo Investigator Software (MBF Bioscience) was used to assess the total number of phosphorylated α-syn aggregates within the SNpc at 120 days post-injection. Sections were counterstained with Nissl following pSer129 immunolabeling. A contour was drawn around the SNpc of the injected hemisphere using a 4X objective and aggregates were systematically counted using the 20X objective. Based on the literature⁵⁶, the counting parameters selected for the SNpc involved a 250 × 250 grid with a 250 × 250 counting frame at 20x magnification, and a fixed tissue thickness of 30 μm. The total number of aggregates was recorded for every fourth section (1:4 series) throughout the rostro-caudal axis of the SNpc. The total number of aggregates for each section was compiled and reported as a total for each animal. Counts reflect actual numbers counted, not a population estimate derived from a sample within the SNpc.

Tissue selection and mRNA sequencing:

Striatum was selected for mRNA profiling at a 6-month time point. Samples consisted of four female and four male heterozygous Q140 mice with or without Alfy overexpression, in addition to control littermates with or without Alfy overexpression. Total genotypes included WT, Rosa^hAlfy/+, Q140, Q140::Rosa^hAlfy/+, Q140::Rosa^hAlfy/hAlfy. RNA extraction, sequencing, data processing and analysis was conducted as previously described by Langfelder, et al²⁵

Creation of HEK293 WDFY3^Var cells:

HEK293T cells were plated into 6 well plate at density 105 cells/well and transfected 24h later with prepared pGL3-U6-sgRNA-PGK-puromycin guide RNA vector and NG-ABE8e vector (gift from David Liu, Addgene plasmid # 138491) in 1:1 ratio using X-tremeGENE 9 DNA Transfection Reagent (Roche). Complete media was supplemented with 2 μg/ml of puromycin 24 hours after transfection and cells were cultured in this media for additional 48 hours. 72h after transfection (48h after addition of puromycin) cells were lysed in 150μl of DirectPCR Lysis Reagent (Cell) (Viagen Biotech), supplemented with 0.2 mg/ml of Proteinase K and incubated for 6h at 55 degrees followed by incubation at 85 degrees for 45 min. 0,5 μl of lysate was used as a template to PCR region of genomic DNA around I3032 of ALFY/WDFY3 using primers pair Fw1 : CCACCCAGCAGGTCTTGTAG Rev1: TGGCTAGGATCTCTCGGAGG or Fw2: CATTCCACCCAGCAGGTCTT Rev2: AGCCAGACCACAAAAGAGCA. Obtained PCR products were cloned into pCR-Blunt II-TOPO vector (Zero Blunt^™ TOPO^™ PCR Cloning Kit, Invitrogen) and transformed into Stbl3 chemically competent cells. Plasmid DNA was purified from 20 bacterial colonies and sequenced with T7 primer (TAATACGACTCACTATAGGG).

Preparation of pGL3-U6-sgRNA-PGK-puromycin guide RNA vector:

Oligonucleotides ccggaggtattcttgcggtggaac and aaacgttccaccgcaagaatacct were phosphorylated and 100μM of each annealed using T4 PNK (NEB). Annealed oligos were cloned into pGL3-U6-sgRNA-PGK-puromycin vector (gift from Xingxu Huang, Addgene plasmid # 51133) cut with BsaI-HFv2 restriction enzyme.

Aggregate clearance assay:

AlfyC constructs were created as previously described^32,33,39. I3032V modification was created by introducing the Ala to Gly mutation using QuikChange mutagenesis per manufacturer’s instructions. Constructs was transiently transfected into 103QmGFP cells using Lipofectamine 2000 per manufacturer’s instructions and assayed for aggregate clearance^43,101. Briefly 24 hrs post-transfection, cells are replated onto coverslips then 48 hrs later, fixed with 4% paraformaldehyde in PBS +/+ for 10 min at room temperature. Cells were washed two times with PBS, the incubated with 1 μg/ml Hoechst solution (Life Technologies) for two hours at room temperature. Images were acquired using a Leica TCS SP2 confocal microscope at 63x magnification and the accompanying software package. Aggregates and nuclei were counted using ImageJ.

Cell Fractionation: Nuclear isolation:

Nuclei EZ kit was used to isolate nuclei followed by immediate RNA isolation. Cytosol extraction: Either supernatant from Nuclei EZ prepped samples were used (after centrifugation step 1) or HLB buffer was added directly to cell pellet to extract cytosol¹⁰².

Actinomycin D treatment:

Cells were seeded at 3×10⁵ cells per well (6 well plate)⁴⁴. The following day, Act D (10μg) was added to the wells then samples were collected at designated time points (t=1, 2, 4, 8 hrs). The cell pellets were then used directly in RNA isolation with the RNeasy kit.

QUANTIFICATION AND STATISTICAL ANALYSIS

Statistical analyses:

Statistical analyses of survival, behavioral data, western blots, stereologically obtained counts and volumes were performed using Statview or SAS (SAS Institute) unless otherwise stated in the text. Statistical significance was accepted at p < 0.05. Normally distributed data were subject to student t-test or, for multiple comparisons, analysis of variance (ANOVA). Complete F-statistics and calculated p-values are found in Table S2, as indicated in the Figure Legends. Fisher PLSD was used as a post hoc test. Power analyses for behavioral and neuroanatomical studies were performed to achieve a power of 0.8 with a confidence of 0.95 using G*Power 3.1¹⁰³. Effect sizes were based upon pilot studies, previously published, or unpublished materials. n-values are indicative of biological replicates, and no data were excluded from analyses. Nonparametric tests such as Kruskal-Wallis and Whitney Mann U were used if normal distribution of data was not assumed. Data are represented as mean +/− SEM or SD as noted in figures. ImageJ (NIH) was used for western blot quantification of band intensity. RNAseq analyses were performed as previously described.²⁵

Image preparation:

All images were prepared using NIH ImageJ, Adobe Photoshop CS5, and Adobe Illustrator CS5

Supplementary Material

1

Table S1: Significant variants in linkage disequilibrium in the WDFY3 locus, related to Figure 1A. Excel spreadsheet indicating output from SNPEff, which was used to generate annotation for each SNP with added columns for GWA statistics and variant to protein prediction algorithms (SIFT, CADD, PolyPhen2)

2

Table S3: DEG lists (FDR < 0.10) are related to Figures 5A, B and C. Excel Spreadsheet with complete DEGs from respective comparisons from Q140 mice.

3

Table S4: GO Analysis Gene IDs, Increased Expression, as related to Figures. 5D and E. Excel spreadsheet with gene lists representing DEGs that showed increased expression

4

Table S5: GO Analysis Gene IDs, Decreased Expression, as related to Figures. 5F and G. Excel spreadsheet with gene lists representing DEGs that showed decreased expression

5

Table S6: Significant Rescue genes (FDR < 0.10) as related to Figure 6 and 7. Excel spreadsheet of genes showing a change in direction upon expression of Alfy (eg Gene whose expression is increased in a WT vs Q140 comparison becomes decreased in a Q140 vs Q140+Alfy comparison)

6

Table S7: Significant Exacerbate genes (FDR < 0.10) as related to Figure 6 and 7. Excel spreadsheet of genes showing a further enhancement of direction upon expression of Alfy

7

Table S8: NYBB cohort used for genotyping for presence of Alfy variant, as related to STAR METHODS as related to Figure 2. Excel spreadsheet indicating age, sex, PMI of samples used. Those samples used for subsequent analysis as indicated in Figure 2 are indicated.

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Rabbit monoclonal anti-Alfy	Gifted by Dr. Ben Hoffstrom, Fred Hutchinson Cancer Center	This publication
Mouse monoclonal anti-Huntingtin Antibody, a.a. 1-82	Millipore	Cat# MAB5492; RRID:AB_347723
Rabbit polyclonal anti- Glial Fibrillary Acidic Protein	Agilent	Cat# Z0334; RRID:AB_10013382
Rabbit monoclonal anti-Iba1/AIF-1 (E4O4W)	Cell Signal Technology	Cat# 17198; RRID:AB_2820254
Rabbit polyclonal anti-FOXP1	Abcam	Cat# ab16645; RRID:AB_732428
Rabbit monoclonal anti-Alpha-synuclein (phospho S129) [EP1536Y]	Abcam	Cat# ab51253; RRID:AB_869973
Rabbit polyclonal anti-Tyrosine Hydroxylase	Sigma-Aldrich	Cat# AB152;RRID:AB_390204
Rabbit recombinant monoclonal anti-Vinculin	Thermo Fisher Scientific	Cat# 700062, RRID:AB_2532280
Mouse monoclonal M2 anti-flag	Sigma-Aldrich	Cat# F1804; RRID:AB_262044
Mouse monoclonal anti-Huntingtin Antibody, clone mEM48	Millipore	Cat# MAB5374; RRID:AB_177645
Rabbit polyclonal anti-LC3B	Abcam	Cat# ab48394; RRID:AB_881433
Rabbit recombinant anti-SQSTM1 / p62 antibody [EPR4844]	Abcam	Cat# ab109012; RRID:AB_2810880
Mouse monoclonal Phospho-Tau (Ser202, Thr205) (AT8)	Thermo Fisher Scientific	Cat# MN1020; RRID:AB_223647
Mouse monoclonal Tau antibody (HT7)	Thermo Fisher Scientific	Cat# MN1000; RRID:AB_2314654
Rabbit monoclonal α-Synuclein (D37A6)	Cell Signal Technology	Cat# 4179
Rabbit monoclonal anti-Huntingtin antibody [EPR5526]	Abcam	Cat# ab109115; RRID:AB_10863082
Biological samples
Human Broadmann Area 9 (BA9) frontal cortex	New York Brain Bank	Table S8
Human cerebellum samples	New York Brain Bank	Table S8
Chemicals, peptides, and recombinant proteins
Bovine serum albumin (BSA)	Sigma	Cat# A4161
TRIzol Reagent	ThermoFisher	Cat# 15596026
Halt protease inhibitor cocktail	ThermoFisher	78430
3,3′-Diaminobenzidine tetrahydrochloride	Sigma-Aldrich	Cat# D5905-50TAB
Goat Anti-Rabbit HRP	ThermoFisher	Cat# 31460
Goat Anti-mouse HRP	ThermoFisher	Cat# 31430
Fisher FAST SYBR Mix	Applied Biosystems ThermoFisher	Cat# 4385612
KAPA FAST SYBR Mix	Kapa Biosystems Roche Diagnostics	Cat# 50-196-5206
Dulbecco's Modified Eagle Medium	ThermoFisher	Cat# 11965118
Heat inactivated Fetal Bovine Serum	ThermoFisher	Cat# 16140071
Actinomycin D	ThermoFisher	Cat# 11805017
Induro Reverse transcriptase	New England Biolabs	Cat# M0681S
Smartscribe Reverse Transcriptase	Takarabio	Cat# 639536
Fibrillized mouse α-Syn	This paper	Polinski, et al. Bieri et al.92,93
Critical commercial assays
DC protein assay	Biorad	Cat# 5000111
VECTASTAIN Elite ABC-HRP Kit	Vector Laboratories	Cat# PK-6100
M.O.M. Immunodetection Kit	Vector Laboratories	Cat# BMK-2202
Nuclei EZ Prep Kit	Sigma-Aldrich	Cat# Nuc101-1Kt
RNeasy Isolation Kit	Qiagen	Cat# 74104
DNeasy Blood and Tissue Kit	Qiagen	Cat# 69504
Deposited data
Transcriptional profiling of striatum from the Q140 mouse model of Huntington's disease.	GEO deposit	GSE281823
Experimental models: Cell lines
HEK293T cells	ATCC	Cat# CRL-3216
Experimental models: Organisms/strains
Mouse: Q140 (strain C57BL/6J)	Jackson Labs	JAX027409
Mouse: N171-82Q (strain B6C3F1/J)	Jackson Labs	JAX003627
Mouse: PS19 P301S (strain B6C3F1/J)	Jackson Labs	JAX008169
Mouse: HPRTCre/+ (strain 129S1/Sv)	Jackson Labs	JAX004302
Mouse: Wdfy3Var/+ (strain Bl6/129Sv)	This paper	N/A
Mouse: Rosaflox-hAlfy/+ (strain Bl6/129Sv)	This paper	N/A
Mouse: Alfyflox/flox (strain: Bl6/129Sv)	Dragich et al.22	N/A
Oligonucleotides (See Table for complete list)
Alfy Lox P1 Forward: GAA ACG AAG CTC GTT TAC GG	Dragich et al.22	N/A
Alfy Lox P1 Reverse: TGC AGT GAC ATT TCC TCT GG	Dragich et al.22	N/A
Rosa hAlfy Forward: AAA TCT CAT CCC CGG TGC G	This paper	N/A
Rosa hAlfy Reverse: CTG GCA ACT AGA AGG CAC AG	This paper	N/A
Rosa WT Forward: AGC ACT TGC TCT CCC AAA GTC	This paper	N/A
Rosa WT Reverse: TGA GCA TGT CTT TAA TCT ACC TCG ATG G	This paper	N/A
Alfy Var Forward: CAA GTG ACC AGA TTG TGA CTT GCC A	This paper	N/A
Alfy Var Reverse: GAG AAC CCA GAC CAG GCT TTG TCA	This paper	N/A
Software and algorithms
Stereo Investigator Software	MBF Bioscience	https://www.mbfbioscience.com/products/stereo-investigator
SAIGE software package	Zhou et al.94
Open field Activity Monitor	Med Associates	https://med-associates.com/product/activity-monitor-7-software/
ImageJ	Schneider et al.99	NIH
Statview	SAS Institute	N/A
Photoshop Creative Suite 5	Adobe	https://helpx.adobe.com/creative-suite.html
Illustrator Creative Suite 5	Adobe	https://helpx.adobe.com/creative-suite.html
Bio-Rad CFX Maestro	Bio-Rad	https://www.bio-rad.com/en-us/product/cfx-maestro-software-for-cfx-real-time-pcr-instruments?ID=OKZP7E15

Highlights.

1. A variant of WDFY3 delays the age of onset of HD in the Venezuelan kindred

2. Orthologous expression of the SNP in mice captures the protection seen in patients

3. The SNP augments expression of the autophagy adaptor Alfy by increasing mRNA stability

4. Increasing Alfy expression protects against different proteinopathies