© 2016. Published by The Company of Biologists Ltd | Disease Models & Mechanisms (2016) 9, 553-562 doi:10.1242/dmm.022848

RESEARCH ARTICLE Spontaneous shaker rat mutant – a new model for X-linked tremor/ ataxia Karla P. Figueroa1, Sharan Paul1, Tito Calì2, Raffaele Lopreiato2, Sukanya Karan1, Martina Frizzarin2, Darren Ames3, Ginevra Zanni4, Marisa Brini5, Warunee Dansithong1, Brett Milash3, Daniel R. Scoles1, Ernesto Carafoli6 and Stefan M. Pulst1,*

ABSTRACT mode of inheritance. Here, we describe the genetic analysis of the The shaker rat is an X-linked recessive spontaneous model of shaker rat, a model of Purkinje cell (PC) degeneration. This mutant progressive Purkinje cell (PC) degeneration exhibiting a shaking arose spontaneously and was observed in Sprague Dawley (SD) ataxia and wide stance. Generation of Wistar Furth (WF)/Brown outbred stock in 1991 at Saint Louis University, first described by Norwegian (BN) F1 hybrids and genetic mapping of F2 sib-sib La Regina et al. (1992), and the phenotype of whole-body tremor, ‘ ’ offspring using polymorphic markers narrowed the candidate ataxia and wide stance designated as shaker . The shaker trait was region to 26 Mbp denoted by the last recombinant genetic marker reported as an X-linked recessive trait. DXRat21 at 133 Mbp to qter (the end of the long arm). In the WF Various animal models of spontaneously occurring mutants that background, the shaker mutation has complete penetrance, results in parallel some aspects of human hereditary ataxia have been a stereotypic phenotype and there is a narrow window for age of reported; for example, weaver, lurcher, stumbler, tottering and disease onset; by contrast, the F2 hybrid phenotype was more varied, teetering mice (Chou et al., 1991; Frankel et al., 1994; Green and with a later age of onset and likely non-penetrance of the mutation. By Sidman, 1962; Guyer et al., 2002; Meier, 1967; Soha and Herrup, deep RNA-sequencing, five variants were found in the candidate 1995). Although the identification of candidate can arise from region; four were novel without known annotation. One of the variants several methodologies, including whole-transcriptome shotgun caused an arginine (R) to cysteine (C) change at codon 35 of the sequencing (WTSS, RNA-seq), whole-genome sequencing ATPase, Ca2+ transporting, plasma membrane 3 (Atp2b3) gene (WGS) and whole-exome sequencing (WES), knowledge of encoding PMCA3 that has high expression in the cerebellum. genetic variation between strains is imperative to the mapping of The variant was well supported by hundreds of overlapping reads, candidate genes. For rat laboratory strains, information on genetic and was found in 100% of all affected replicas and 0% of the wild-type variation is currently limited to microsatellite markers and a small (WT) replicas. The mutation segregated with disease in all affected number of single-nucleotide polymorphisms (SNPs) (Canzian, animals and the amino acid change was found in an evolutionarily 1997; Thomas et al., 2003). conserved region of PMCA3. Despite strong genetic evidence for Our objectives were to fine-map the shaker locus, to identify pathogenicity, in vitro analyses of PMCA3R35C function did not show candidate genes responsible for the shaker phenotype and to any differences to WT PMCA3. Because Atp2b3 mutation leads to establish a model system for the investigation of ataxia. Here, we congenital ataxia in humans, the identified Atp2b3 missense change report the finding of a candidate gene, Atp2b3, which encodes 2+ in the shaker rat presents a good candidate for the shaker rat PMCA3, a plasma membrane Ca ATPase transporter that is 2+ phenotype based on genetic criteria, but cannot yet be considered a abundant in human cerebellum. Ca signaling regulates some of definite pathogenic variant owing to lack of functional changes. the most important neuronal functions; thus, the concentration of Ca2+ within neurons must be precisely regulated. The plasma KEY WORDS: Ataxia, X-linked, shaker, PMCA3, Atp2b3 membrane Ca2+- (PMCAs or calcium pumps) are essential for this regulation, with two of the four basic PMCA pump INTRODUCTION isoforms, PMCA2 and PMCA3, being highly expressed in the brain Hereditary ataxias in humans comprise a diverse group of diseases and the cerebellum (Brini and Carafoli, 2009; Eakin et al., 1995; that affect the cerebellum and various other regions of the nervous Filoteo et al., 1997; Jensen et al., 2007). system. The classification of these disorders is complex owing to heterogeneity in clinical signs, age of onset, disease progression and RESULTS Behavior The shaker rat exhibits a wide-based hind-limb stance, ataxic 1Department of Neurology, University of Utah, Salt Lake City, UT 84112, USA. 2Department of Biomedical Sciences, University of Padova, Padova, Italy. movements and whole-body tremor with onset as early as 9 weeks 3Bioinformatics Shared Resource, Huntsman Cancer Institute, University of Utah, of age in the Wistar Furth (WF) background. Early in the disease Salt Lake City, UT 84112, USA. 4Unit of Molecular Medicine for Neuromuscular and course, a high-frequency tremor was felt upon the handling of the Neurodegenerative Disorders, Department of Neurosciences, Bambino Gesù Children’s Hospital, IRCCS, Rome, Italy. 5Department of Biology, University of animal but not visible by cage behavior. This high-frequency tremor Padova, Padova, Italy. 6Venetian Institute of Molecular Medicine (VIMM), Padova, Italy. is best described as a whole-body vibration, similar to that exhibited by an animal in fear. This high-frequency tremor was more difficult *Author for correspondence ([email protected]) to discern once the disease had fully developed and animals were This is an Open Access article distributed under the terms of the Creative Commons Attribution visibly shaking. The generalized tremor involves the entire trunk License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution and reproduction in any medium provided that the original work is properly attributed. and body (Movie 1). Although 90% of affected animals exhibited a marked shaking

Received 17 August 2015; Accepted 29 February 2016 tremor phenotype by 12 weeks of age, we did not observe any Disease Models & Mechanisms

553 RESEARCH ARTICLE Disease Models & Mechanisms (2016) 9, 553-562 doi:10.1242/dmm.022848 feeding or grooming abnormalities. The shaker rat appeared to be Fmr1 repeat expansion more docile than wild-type (WT) rats, exhibited more exploratory Because Fmr1 mutation causes a tremor/ataxia phenotype and the behavior and did not have ledge fear. We did not encounter Fmr1 gene is located in the candidate region, we excluded the Fmr1 infertility, or gestational or litter-size complications, and lifespan repeat expansion as a causative mutation for the shaker rat by was not impacted. amplification and sequencing of the 5′-UTR region that contains the CGG repeat. We amplified genomic DNA from both WT males and Morphology females, carrier females and affected males. The gel-purified Fmr1 Significant PC loss in shaker cerebella was reported previously in PCR amplicon was sequenced and no expansions were detected. detail (La Regina et al., 1992). We therefore did not conduct a repeat morphometric study. Illustrative samples of midline RNA-seq cerebellar sections at 2 and 6 months of age are shown in To identify the potential disease-causing variant, we performed Fig. 1. Progressive loss of calbindin staining and loss of PC deep RNA-seq on whole cerebella of WT and affected 5-week-old bodies was evident. WF males. We chose an age of 5 weeks to ensure that PCs were still present in normal numbers. Based on prior observations (La Regina Genetic mapping et al., 1992; Tolbert et al., 1995), PC counts and molecular layer By observation of segregation of the phenotype, X-linked thicknesses of 6-week-old shaker rats and age-matched controls recessive transmission was confirmed. In order to estimate the were not significantly different. We have previously demonstrated, rate of transmission, we crossed WT males and obligatory carrier in a different rodent model of cerebellar ataxia, that the expressions females for both WF and F2s. An obligatory carrier female was of Calb1, Pcp2 and Grid2, as well as that of some other genes highly defined as the female offspring of an affected male crossed with expressed in PCs, significantly reduce as the disease phenotype a WT female. We expected a 50/50 ratio of affected and WT progresses (Dansithong et al., 2015; Hansen et al., 2013). Analyses males, assuming 100% penetrance by 36 weeks of age. We of Calb1, Pcp2 and Grid2 transcript abundance by quantitative PCR observed 43/105 (41%) affected males in the WF background (qPCR) demonstrated that these PC transcripts were not and 32/92 (35%) affected males in the F2 cross. Lack of male-to- significantly reduced in cerebella of WF shaker rats at 5 weeks of male transmission on either background further verified X-linked age (Fig. S1). transmission. RNA quality was high, with 89 million and 102 million raw reads Although segregation analysis was consistent with X-linked for WT and affected rats, respectively, and with 96% alignment of segregation, we wanted to confirm X-linked inheritance using DNA sequences. The RNA sequence revealed five variants in our 26-Mbp markers and refine the shaker locus using an F2 hybrid mapping region of interest from 133 Mbp to qter (the end of the long arm of strategy. Several recombination events were seen for markers the ) (Table 1): two intergenic SNPs, two intergenic DXRat17 and only one for DXRat21, located at 112,204,377 and deletions, and one DNA variant leading to an amino acid (aa) 132,940,994, respectively (Table S1). This hybrid mapping strategy change. The non-synonymous substitution c.33C>T was found in mapped the shaker locus to a 26-Mbp region of distal rat Atb2b3, which encodes PMCA3, a plasma membrane Ca2+ ATPase chromosome X (rat genome version RGSC3.4). transporter that is abundant in rat brain (Ensembl transcript

Fig. 1. Significant cerebellar pathology in shaker rats. Three (40 μm) midline cerebellar sections of each group – wild-type (WT) and shaker F1 (50/50 WF/BN) hybrid rats – were analyzed at 2 and 6 months of age (n=3/group). Sections were stained with an antibody to mouse monoclonal anti- calbindin-D-28K (Calb1; 1:500) and counterstained with DAPI. Calb1 staining in Purkinje cells (PCs) was reduced in shaker 2-month-old animals, and was virtually absent in somata and dendrites in PCs of shaker 6-month-old animals. PCs are denoted by arrows. PF, primary fissure. Disease Models & Mechanisms

554 RESEARCH ARTICLE Disease Models & Mechanisms (2016) 9, 553-562 doi:10.1242/dmm.022848

Table 1. Variants in region of interest Genetic evaluation of Atp2b3 Location and We performed genomic PCR of DNAs from six rat strains [Brown Variant sequence Comments Norwegian (BN), SD, F344, Long-Evans, WF and Wistar Kyoto] and found that the arginine encoded by codon 35 was conserved SNP1 150049620 Intergenic, found in two other strains: aaaaA/Tttac SR/Jr and SS/Jr in all. This was also true by in silico analysis of the genomes of SNP2 150050694 Intergenic, found in two other strains: 16 additional rat strains. In addition, R35 was evolutionarily agaacC/ SR/Jr and SS/Jr conserved in all of the 48 sequenced taxa, including humans Tgttcttt (Fig. S3). Del 1 150051300 Deletion is intergenic, not found in any The human and rat Atp2b3 genes are 89% identical at the cDNA tttaTA(del) other strain level and 98% identical at the level. Human and rat PMCA3 aaaaa Del 2 150052790 Intergenic deletion causes this is almost identical, with the exception of an additional exon in rat gagaCA(del) dinucleotide repeat to no longer be composed of 38 aa at the C-terminal end, at c.3343, p.1115 gaga interrupted, not found in any other (Ensembl transcript ID ENSRNOT00000078100, A0A0G2K9Q6). strain Because protein-variant scoring algorithms are optimized for Atp2b3 159422011 c.33C>T base-pair change causing human , we utilized human PMCA3 for in silico analysis of actaC/Tgcag amino acid change R35C on the the aa variant. This analysis was performed using a suite of Atp2b3 gene, variant not found in 22 other strains bioinformatics tools: PolyPhen-2, SIFT, PROVEAN, Mutation Taster and CADD. PolyPhen-2 predicted the variant as benign, in SR/Jr, Dahl salt-hypertension resistant; SS/Jr, Dahl salt-hypertension direct contrast to SIFT, PROVEAN, Mutation Taster and CADD, sensitive. which all predicted it as a deleterious or disease-causing mutation. This variant is not known in humans (Table S2). ENSRNOT00000024025, Rat RGSC3.4). The resulting aa change was found at codon 35, substituting an arginine for a cysteine (p. In vitro characterization of PMCA3 R35C). The variant was well supported by 286 reads with a 99.3% SH-SY5Y cells were transfected with expression plasmids encoding variant allele frequency based on read counts with a P-value of the WT PMCA3 or PMCA3R35C protein, with an independent − 3.1791E 166 found in all affected and none of the WT transcripts. CMV-GFP cassette (Fig. 3A). Construct expression was analyzed RNA expression levels across the candidate gene interval, including 48 h post-transfection by western blot analysis; blots were those of Atp2b3, were not different for WT and mutation carriers at then sequentially stained with antibodies against PMCA3, GFP 5 weeks of age (Fig. S2). and β-actin. GFP and β-actin staining was utilized to verify uniform Once the Atp2b3 variant was identified, all animals were loading of wells. Western blot results confirmed the expression of genotyped. 100% of all mutation-positive rats in the WF WT PMCA3 or PMCA3R35C recombinant PMCA3 at equal levels background exhibited an affected phenotype by 20 weeks of age. (Fig. 3B). A total of 105 WF and 92 F2 males were produced. All male progeny was aged to 36 weeks of age before euthanization. In the F2 Immunocytochemistry for PMCA3 background, 95% of all mutation carriers exhibited an affected Immunocytochemical staining of PMCA3 in frozen cerebellar phenotype by 36 weeks of age. The phenotype of these affected tissue sections revealed no differences between WT and affected hybrid progeny was the same as that of affected WF rats. We male rats at 9 weeks of age (Fig. 3C). concluded that the 5% (5/92) of mutation carriers that did not exhibit a phenotype by 36 weeks of age were likely non-penetrant (Fig. 2). Ca2+ handling by exogenous WT PMCA3 or PMCA3R35C Reverse-transcriptase (RT)-PCR of the entire coding region for To selectively explore the ability of the mutated PMCA3 pump to Atp2b3 was performed to rule out other mutations that might affect handle cellular Ca2+ fluxes in living cells, the constructs encoding transcription or exon splicing. No alterations were detected (data not the splicing variants a and b of the PMCA3 pump were employed. shown). Both the full-length b and the C-terminally truncated isoform a of

Fig. 2. Age-of-onset graph for Wistar Furth (WF) and F2 male offspring. Graph represents 105 WF and 92 F2 males. All male progeny were aged to 36 weeks of age before euthanization. 100% of WF and 95% of F2 mutation carriers exhibited an affected phenotype by 36 weeks of age. ‘Non- penetrant’ was defined as mutation-positive but not exhibiting a phenotype by 36 weeks of age. Reduced penetrance was observed only in the F2 cohort. Disease Models & Mechanisms

555 RESEARCH ARTICLE Disease Models & Mechanisms (2016) 9, 553-562 doi:10.1242/dmm.022848

Fig. 3. Characterization of PMCA3. (A) Schematic representation of GFP- tagged PMCA3 clones. GFP-tagged wild-type (WT) PMCA3 or PMCA3R35C (i, ii) or non-tagged (pAdTrack) WT PMCA3 or PMCA3R35C (iii, iv) with an independent GFP cassette under the transcriptional control of CMV promoters are shown. The bottom images show a plasmid map delineating restriction used to clone the insert, and the selecting antibiotic used. (B) Western blot of SH-SY5Y cell lysates following transfection by GFP-tagged PMCA3 plasmids. Control lanes represent untransfected cell extract. Western blot results confirmed the expression of WT PMCA3 or PMCA3R35C recombinant PMCA3. The experiment was replicated three times. (C) Twelve (10 μm) midline cerebellar sections of each group – WT and shaker WF rats – were analyzed at 9 weeks age (n=2). Sections were stained with an antibody to rabbit polyclonal anti-PMCA3 ATPase (1:2000) and counterstained with DAPI. PMCA3 staining shows no difference between WT and shaker rats. ML, molecular layer; PC, Purkinje cells; GC, granular cell layer.

the pump are significantly expressed in cerebellar neurons (Carlson et al., 2009; Eakin et al., 1995). They only differ by the presence of a premature stop codon in the a variant, which is caused by a shift in the reading frame during the splicing process that truncates the C-terminal tail of the pump in the middle of the calmodulin-binding domain (Brini and Carafoli, 2009). The loss of about half of the calmodulin-binding domain compromises its ability to auto-inhibit the pump, which becomes considerably more active in the resting state than the original full-length variant. The plasmids encoding isoform a of a GFP-tagged (GFP) and a non-tagged (AdTrack) WT PMCA3 or PMCA3R35C were introduced into HeLa cells along with the plasmid encoding the recombinant Ca2+-sensitive photoprotein aequorin (cytAEQ). The same procedures were also applied to the b variant of the pump and were expressed in HeLa cells in the WT and the mutant form. To study the effect of the R35C mutation on the activity of the overexpressed pump, we monitored cytosolic Ca2+ transients generated upon stimulation of HeLa cells with histamine, an inositol 1,4,5-trisphosphate-linked agonist that induces Ca2+ release from intracellular stores, and the consequent influx of the Ca2+ from the extracellular ambient through channels activated by their emptying (the SOC channels). The statistical analysis and the average traces shown in Fig. 4 indicate that clearance of the Ca2+ transient, which was greatly accelerated in both cases with respect to control cells only transfected with cytAEQ, did not differ in cells overexpressing WT or PMCA3R35C. Average peak values of the transients were essentially the same in control cells and in cells in which GFP-tagged or un-tagged PMCA3 isoform a was overexpressed. Similar results were obtained for the full-length b variant of the pump (Fig. 5). Thus, the R35C mutation failed to significantly affect the ability of the pump to counteract cytosolic Ca2+ transients.

Effect of the R35C variant on the basal activity of the PMCA3 pump To study the effect of the R35C mutation on the basal activity of the PMCA3 pump, a yeast-based functional assay was employed (Ton and Rao, 2004). In this assay, the lethal phenotype of triple- mutant K616 yeast cells in Ca2+-free conditions is rescued by the ectopic expression of active Ca2+ pumps. The coding sequences of the WT or the mutant full-length PMCA3 pump were inserted in the galactose-inducible plasmid pYES, and transformed in yeast K616 cells. As a positive control, we used a C-terminally truncated Atp2b3 mutant (PMCA3-ΔCter), lacking the last 158 C-terminal residues, that is constitutively active because it is not auto-inhibited by the C-terminal tail of the pump. The constitutively active pump variant

is a suitable positive control in the assay. Two other mutants were Disease Models & Mechanisms

556 RESEARCH ARTICLE Disease Models & Mechanisms (2016) 9, 553-562 doi:10.1242/dmm.022848

Fig. 4. Cytosolic Ca2+ measurements in HeLa cells overexpressing the GFP-tagged and untagged wild-type (WT) and R35C mutant PMCA3a variant. HeLa cells were co-transfected with cytAEQ and the expression plasmid for the WT and the R35C mutant GFP-tagged and untagged PMCA3a variant. Average peak values (A,C) and cytosolic Ca2+ transients (B,D) were recorded following 100 μM histamine (His) stimulation. Bars in panels A and C represent mean [Ca2+] values upon stimulation (μM; ±s.e.m.). Peak values in µM are: 4.20±0.15, n=27 for cytAEQ, 3.97±0.21, n=29 for GFP-tagged WT PMCA3a, and 3.82±0.18, n=28 for GFP-tagged R35C PMCA3a; 3.95±0.23, n=30 for un-tagged WT PMCA3a, and 4.09±0.26, n=29 for un-tagged R35C PMCA3a. The experiment was replicated three times. tested. An ATPase-dead mutant, produced by replacing the catalytic Our assays included growth controls of yeast transformed with aspartate with an alanine (PMCA3 D465A), was used to verify the the different Atp2b3 constructs in a Ca2+-containing permissive detrimental effects of a known PMCA-inactivating mutation. A medium (Fig. 6A) and in an inducible Ca2+-free medium (EGTA W1104A mutant was generated by site-directed mutagenesis and 7.5 or 10 mM) (Fig. 6B and C, respectively). As expected, the used as another positive control because it promotes the basal constitutively active PMCA3-ΔCter mutant rescued the full viability activity of the protein (Penheiter et al., 2002), probably by slowing of K616 yeast cells in Ca2+-depleted medium (Fig. 6B,C, third row), its inactivation rate. whereas the auto-inhibited WT Atp2b3b (PMCA3b) restored it only

Fig. 5. Cytosolic Ca2+ measurements in HeLa cells overexpressing the untagged wild-type (WT) and R35C mutant PMCA3b variant. HeLa cells were co-transfected with cytAEQ and the expression plasmid for the WT and the R35C mutant untagged PMCA3b variant. Average peak values (A) and cytosolic Ca2+ transients (B) were recorded following 100 μM histamine (His) stimulation. Bars in panel A represent mean [Ca2+] values upon stimulation (μM; ±s.e.m.). Peak values in µM are: 4.50±0.21, n=10 for cytAEQ, 4.16±0.14, n=11 for un-tagged WT PMCA3b, and 4.30±0.20, n=7 for un- tagged R35C PMCA3b. The experiment was replicated three times. Disease Models & Mechanisms

557 RESEARCH ARTICLE Disease Models & Mechanisms (2016) 9, 553-562 doi:10.1242/dmm.022848

Fig. 6. Functional complementation assay in K616 yeast cells. Yeast K616 cells were transformed with pYES2-derived vectors carrying one of the following PMCA3b variants: wild type Δ (WT), C-terminally truncated (PMCA3 Cter), R35C mutant, D465A mutant and W1104A mutant. Exponentially growing cells have been serially

diluted and spotted on glucose+CaCl2 medium (A), galactose+7.5 mM EGTA (B) and galactose +10 mM EGTA (C). Cells carrying empty plasmid Δ or expressing constitutively active PMCA3 Cter are negative and positive internal controls of the functional assay, respectively. The experiment was replicated three times.

partially (Fig. 6B,C, second row). Notably, the recovery of the yeast Essential tremor with the R35C mutation in the absence of Ca2+ was not unlike the The phenotype of the shaker rat was reminiscent of severe essential WT, suggesting that the aa substitution did not affect the ability of tremor (ET) rather than cerebellar ataxia in early disease stages, but the full-length PMCA3 pump to eject Ca2+ in the resting, not progressed to significant gait ataxia later. The associations between activated, condition (Fig. 6B,C, compare the second with the fourth ET and ataxia of gait and balance are well known. In many patients, rows). As expected, the ATPase-dead D465A control mutant failed tremor is not the only feature of ET, and can be accompanied by to rescue the viability of the K616 yeast cells (Fig. 6B,C, fifth row). neuropsychiatric changes and deficits of gait and balance (Arkadir Also as expected, the W1104A mutant, which removed the auto- and Louis, 2013; Louis and Okun, 2011). ET is one of the most inhibition of the pump, increasing its activity, produced a yeast common genetic neurological disorders and an inheritance pattern viability phenotype that was only slightly less pronounced than that consistent with autosomal-dominant inheritance is seen in many observed for the ΔCter mutant. Thus, the R35C mutant of the families (Gonzalez-Alegre et al., 2014; Tanner et al., 2001). PMCA3 was active and properly auto-inhibited in its basal, The main clinical manifestation of ET is postural tremor, but unstimulated state. involvement of the extremities, head and trunk might be seen as well. Tremors can be aggravated by emotions, physical exhaustion DISCUSSION and hunger, as well as temperature variations. Pathologic reports General have identified two distinct types. The first shows cerebellar The shaker mutation arose spontaneously in an outbred SD strain degeneration with significant PC loss; the second, Lewy bodies and was subsequently transferred to an inbred WF background confined to the locus ceruleus (Louis, 2009). (Clark et al., 2000) owing to fertility problems. It was inferred that the mutation was X-linked based on segregation studies. Wider X-chromosomal location and human disease hind-limb stance for carrier females versus WT rats was attributed to We mapped the shaker locus using an F2 intercross of male WF and mild loss of PCs, as determined by immunofluorescent staining for BN rats with the assumption that presence of WF alleles would calbindin (Clark et al., 2000; Tolbert et al., 1995). This phenotype denote proximity to the shaker locus. We used 44 DNA markers and was originally described as the ‘mild’ trait of the shaker rat mapped shaker distal to DXRat21 located at 133 Mbp (rat genome (La Regina et al., 1992; Tolbert et al., 1995). Utilizing DNA version RGSC3.4). The synteny region between human and rat markers on the in an F2 intercross, we confirmed includes the Fmr1 gene. Fragile X-associated tremor/ataxia X-linked inheritance. syndrome is a human adult-onset neurodegenerative disorder We found evidence for linkage of shaker to chromosome Xq36 to caused by an expanded trinucleotide repeat in the FMR1 gene qter. The combination of genetic mapping in an F2 cross and analysis (Hagerman et al., 2001). We excluded a repeat expansion in this of pre-symptomatic cerebellar transcriptomes led to identification of a candidate gene because an expansion would not have been easily likely pathogenic DNAvariant. Later disease onset in F2s and the rare detected by RNA sequencing. presence of non-penetrance highlight the effects of genetic X-linked recessive spinocerebellar ataxias are a heterogeneous background on PC dysfunction and degeneration. The candidate group and, to date, five have been reported. SCAX1 mutation segregated with disease in all affected animals and the aa (OMIM#302500, http://omim.org/entry/302500) was initially change was found in an evolutionarily conserved region. The arginine described by Bertini et al. as a neurological disorder characterized at codon 35 is conserved in 22 different rat strains. by hypotonia at birth, delayed motor development, difficulty standing, cerebellar fits (reminiscent of ET observed in the shaker Phenotype rat), dysarthria, and slow eye movements followed by ataxia in the As the name implies, the shaker phenotype is characterized by gait first years of life (Bertini et al., 2000). It is now known to be caused ataxia and total-body tremors. On previous examination by Clark by a mutation in the ATP2B3 gene, encoding PMCA3 (Zanni et al., et al., gait evaluation showed abnormalities in gait symmetry, stride 2012). Precise genetic locations for SCAX2 (Malamud and Cohen, length and stride width (Clark et al., 2000). The walking tracks of 1958) and SCAX3 (Schmidley et al., 1987) are unknown. SCAX4 carrier females were reported as unchanged (Clark et al., 2000). (Farlow et al., 1987) was excluded from Xq26-qter by genetic Disease Models & Mechanisms

558 RESEARCH ARTICLE Disease Models & Mechanisms (2016) 9, 553-562 doi:10.1242/dmm.022848 linkage analysis and SCAX5 was excluded from the SCAX1 region and Carafoli, 2009). PMCAs are also likely involved in local Ca2+ in Xq25-q27.1 (Zanni et al., 2008). signaling via interaction with caveolin-1 (Daniel et al., 2006). Although we did not detect major changes in subcellular Variant detection by transcriptome analysis localization, it is possible that more subtle changes exist in PCs. We hypothesized that transcriptome analysis at a stage prior to PC A single-residue substitution could affect the transfer of the PMCA loss would likely identify a loss-of-function allele by deep RNA-seq pump to the plasma membrane, because a change in one aa in in cerebellar RNAs, and that the number of variants would be transmembrane domain 5 caused PMCA to remain in the greatly constrained by our refined linkage mapping. Indeed, only endoplasmic reticulum (Guerini et al., 2000). Second, it is well few variants were found (Table 1). The most likely variant was a known that a number of proteins interact with the pump and affect its missense change in the Atp2b3 gene, a gene that is mutated in function (Linde et al., 2008), such as the 14-3-3 (epsilon) protein at SCAX1. the N-terminal region. Most of these proteins interact with the The changed aa is located in the N-terminal portion that protrudes C-terminal tail of the pump (DeMarco and Strehler, 2001), such into the cytosol. The PMCA3 protein belongs to the family of P-type as CaM (calmodulin), NOS-1 (nitric oxide synthase 1), CASK ion-transport ATPases characterized by the formation of an aspartyl (Ca2+/calmodulin-dependent serine kinase), NHERF2 (Na+/H+ phosphate intermediate during the reaction cycle. Mammalian exchanger regulatory factor 2) and several members of the MAGUK plasma-membrane calcium ATPases are encoded by at least four (membrane-associated guanylate kinase) family; others interact separate genes. Expression of these four genes is regulated during with the central body of the PMCA pump, such as RASSF1 (a Ras development and in a tissue- and cell-type-specific manner. effector protein) (Armesilla et al., 2004). Several bioinformatic tools (PolyPhen, SIFT, PROVEAN, PMCAs are also part of a postsynaptic complex formed by Homer Mutation Taster and CADD) were used to assess the impact of as well as metabotropic glutamate receptors (mGluRs) and inositol PMCA3 aa change on protein function. These analyses included triphosphate receptors (InsP3Rs; ITPRs) (Sgambato-Faure et al., change in protein structure and function (PolyPhen, Mutation 2006). The levels of mGluR1 and InsP3R1 are reduced in PCs Taster), conservation of the changed aa residue across species in several rodent models of ataxia. Similarly, late-onset human (SIFT, Mutation Taster), and potential effects on mRNA expression degenerative ataxias with pure cerebellar phenotypes are associated or splicing (Mutation Taster) (Schwarz et al., 2010), as well as with disrupted Ca2+- and ITPR-dependent pathways (Schorge et al., estimation of the relative pathogenicity of the variant (CADD) 2010). The importance of maintaining calcium homeostasis and (Kircher et al., 2014). PolyPhen predicted this mutation as benign. signaling with precision might explain why a subtle and currently In contrast, the newer programs SIFT, PROVEAN, Mutation Taster undetectable defect of the PMCA3 pump could generate a severe and CADD predicted the variant to be damaging. cerebellar phenotype. Further studies are required to explore the Additional evidence came from the fact that we did not detect the possibility that the N-terminal mutation of the shaker rat PMCA3 variant in an additional 22 rat strains and that this variant was not pump affects its interaction with still-unidentified partners that found in the 60K human exome project (http://exac.broadinstitute. could have a functional role in PC degeneration. org/gene/ENSG00000067842, accessed January 18, 2016). On the other hand, the differentiation between disease-causing Codon 35 was well covered, with a total allele count of >36,000 variants and rare benign variants might be complex, as shown by the for neighboring codons. difficulties identifying the SCA31 mutation. A heterozygous single- nucleotide change (-16C>T) in the PLEKHG4 (puratrophin-1) gene Functional characterization had been identified within the candidate linkage region and had Because the possibility existed that PMCA3R35C was just a rare been proposed as the candidate SCA31 mutation, especially because private mutation that had arisen in the WF strain maintained in Saint this variant in the UTR of the PLEKHG4 gene seemed to Louis, we embarked on a series of in vitro and functional experiments show in vitro functional effects as well (Ishikawa et al., 2005). to evaluate the biological consequence of the variant. Presence of Subsequently, Ohata et al. reported one patient with ataxia who did R35C did not affect Atp2b3 transcript abundance (Fig. S2) or not carry the -16C>T transition in a family showing tight linkage localization in PCs (Fig. 1B). We then investigated whether it altered with the UTR variant, suggesting that this variant was only closely functioning of the PMCA3 calcium pump. The disease-linked linked to the causative repeat mutation (Ohata et al., 2006). Further PMCA3G1107D mutation had shown dramatic alteration of Ca2+ analysis of the now refined genetic candidate region led to the transients (Zanni et al., 2012). The pump was studied in model cells identification of an expanded pentanucleotide repeat within an by exogenous protein expression of WT PMCA3 and PMCA3R35C. intron of the BEAN and TK2 genes as the cause of SCA31 (Sato The function of PMCAR35C was virtually identical to the WT protein et al., 2009). In the absence of functional in vitro effects of the (Fig. 5). An additional yeast complementation assay likewise did not PMCA3R35C variant, the possibility remains that this genetic change show any difference between the WT and R35C mutant (Fig. 6). is not causative, but very closely linked to the causative shaker Thus, PMCA3R35C was functionally normal, both under conditions mutation. in which it was maximally activated and in conditions with a low resting concentration of Ca2+ in the cytosol. Therefore, the MATERIALS AND METHODS substitution at codon 35 did not drastically impair the Ca2+ Animals extrusion ability of the pump. All animal work was done under an approved IACUC protocol at the Nonetheless, it is still possible that the pump variant identified in University of Utah. Animals were housed 4-5 animals per cage under a shaker is causative of the neurological phenotype. In addition to 12:12 h light cycle with food and water provided ad libitum. Animals were 2+ fed Teklad Global Rodent Diet (Denver, CO). their function in controlling basal Ca levels, PMCAs can serve as The shaker rat colony used in these studies was obtained from a breeding signaling molecules (Brini, 2009). These functions might include colony at Saint Louis University. This animal model occurred as a the regulation of the cytoskeleton through interaction with 14-3-3 spontaneous mutant in the SD strain and the mutation was later moved into proteins via the cytosolic N-terminal tail of the pump (Linde et al., the WF strain via cross-intercross methods, owing to fertility problems in the

2008), a domain that differs significantly among isoforms (Brini SD strain. We maintain the shaker colony in the WF background. Disease Models & Mechanisms

559 RESEARCH ARTICLE Disease Models & Mechanisms (2016) 9, 553-562 doi:10.1242/dmm.022848

Genetic mapping Exclusion of Fmr1 repeat expansion WF shaker rats were crossed with BN rats. BN rats were selected for the Amplification of the 5′ UTR region containing the CGG repeat of the Fmr1 hybrid cross because they are genetically most distant from all other strains gene was performed using HotStar Taq Plus (Qiagen Inc., Valencia, (Canzian, 1997). F1s were generated by mating of affected WF males CA), with standard genotyping conditions with the following primers: with BN WT females and in order to reduce cost and animal numbers. F1s Fmr1F (5′-TCGGCCTCAGTCAGTCTTGCG-3′) and Fmr1R (5′-CTGT- were also produced by mating WF affected females with BN WT males, and CCGGTAGCCCGTTACCT-3′). PCR conditions were as follows: 95°C carrier WF females with BN WT males. BN WT rats were purchased from 5 min, then 35 cycles of 95°C 1 min, 60°C 30 s, 72°C 1 min, followed by a Charles River (http://www.criver.com/). F2s were generated by breeding final extension at 72°C for 5 min. affected F1 males with carrier females. Ninety-two F2 males were retained for phenotype observation and genetic mapping. RNA expression analyses by quantitative RT-PCR Rat cerebella were removed and flash frozen in liquid nitrogen. Tissues were DNA isolation and microsatellite marker genotyping kept at −80°C until the time of processing. Total RNA was extracted from rat DNA was extracted using the Puregene Genomic DNA Purification Kit cerebella using the miRNeasy mini-kit (Qiagen Inc., Valencia, CA) (Gentra Systems; Minneapolis, MN) from whole brain tissue or 3-5 mm of according to the manufacturer’s protocol. DNAse-I-treated RNAs were tail tip (soft tissue only). We utilized SNPlotyper on the RGD (http:// used to synthesize cDNA using the ProtoScript cDNA synthesis kit (New snplotyper.mcw.edu/analyses) to select markers that would differentiate BN England Biolabs Inc., Ipswich, MA). Quantitative RT-PCR was performed from SD and WF. SNPlotyper is an analysis and visualization tool built on using the QuantStudio 12K Flex Real-Time PCR System (Thermo Fisher the rat SNP genotype data developed by the STAR consortium. We Scientific, Grand Island, NY) with the Power SYBR Green PCR master mix identified 44 markers that differentiated WF and BN alleles. Genotyping of (Applied Biosystems Inc., Grand Island, NY). PCR reaction mixtures the F2 progeny was performed and a haplotype table generated (Table S1). contained SYBR Green PCR master mix and 2.5 µmol primers, and PCR The total genetic map length was 137 cM. The average distance between amplifications were carried out for 45 cycles: denaturation at 95°C for 10 s, markers was 3.2 cM. annealing at 60°C for 10 s and extension at 72°C for 40 s. The threshold We PCR-amplified tail DNA of all F2s and assigned a calling code for cycle for each sample was chosen from the linear range and converted to a each sample based on standards produced from the donor strains. DNAs starting quantity by interpolation from a standard curve run on the same from WF and BN WT rats were amplified and sequenced for all markers. plate for each set of primers. All gene expression levels were normalized to The amplicons were then utilized as the standard for each marker as actin mRNA levels. Primer pairs designed for qPCR are given as forward illustrated in Table S1. and reverse, respectively, for all genes: Calb1, Pcp2, actin, Grid2 and Atp2b3 (Table S3). The sample size was n=3 for all assays and only male rats RNA-seq were utilized. RNA Illumina HiSeq Sequencing was performed on whole cerebella of WT and affected 5-week-old F2 males. Total RNA was isolated using an Genotyping of Atp2b3 exon 3 on genomic DNA miRNeasy mini-kit (Qiagen Inc., Valencia, CA) according to the Amplification of Atp2b3 exon 3 was performed with standard PCR manufacturer’s protocol. RNA quality was obtained using the Bioanalyzer conditions with the following PCR primers: RS-2A (5′-CCACCCGTCC- 2100 Pico Chip (Agilent, Santa Clara, CA). Samples with an RNA integrity CTAGCTATT-3′) and RS-2B (5′-TCTGTGTCCTTCATGGGTCA-3′) number (RIN) >8 were used for library preparation using Illumina TruSeq followed by Sanger sequencing utilizing the same primers. PCR RNA Sample Prep v2 (Illumina, San Diego, CA) with oligo-dT selection. conditions were as follows: 95°C 5 min, then 5 cycles 95°C 1 min, 57°C Three biological replicates from each group were analyzed. Samples of 30 s, 72°C 1 min, followed by 30 cycles of 95°C 1 min 30 s, 55°C 30 s, WT and affected cerebella were barcoded and run on a single lane to reduce 72°C 1 min, and final extension at 72°C for 5 min. batch effects. Single-end 50-bp reads were generated on a Hiseq 2000 machine at the University of Utah Microarray and Genomic Analysis Shared Genotyping of Atp2b3 via RT-PCR Resource using Illumina Version 4 flowcells. Reads were then aligned to the Amplification of the Atp2b3 coding region was performed using cDNA rat RGSC3.4 reference genome using Novoalign. We used the SAMTools with eight individually paired PCR primers (RT1 through RT8) listed mpileup on our alignment files, called variants using VarScan2, focusing on in Table S3. PCR conditions were as follows: 95°C 5 min, then 5 cycles variants found on chrX:112,204,377 to qter. Owing to limited annotation of 95°C 1 min, 60°C 30 s, 72°C 1 min 30 s, followed by 30 cycles of 95°C the rat genome, mouse and databases were used to curate 1 min 30 s, 65°C 30 s, 72°C 1 min 30 s, and final extension at 72°C for missing areas of coverage. 5 min. We utilized VarScan2 (Koboldt et al., 2012) (http://varscan.sourceforge. net/) and DESeq2 (Love et al., 2014) (https://bioconductor.org/packages/ Cell culture and transfection release/bioc/html/DESeq2.htm) to examine the level of gene expression and SH-SY5Y cells were cultured and maintained in DMEM media containing mutation analysis. VarScan2 is a platform-independent mutation caller for 10% fetal bovine serum (FBS). For the overexpression of proteins, the cells targeted, exome and whole-genome resequencing data. In contrast to other were transiently transfected with DNA constructs using Lipofectamine 2000 software packages that utilize probabilistic frameworks, to detect variants reagent (11668-027, Invitrogen, Grand Island, NY) according to the and assess confidence in them, VarScan2 utilizes a robust heuristic/statistic manufacturer’s protocol. 48 h post-transfection, the cells were harvested for approach to call variants that take into account read depth, base quality, western blot analyses. We chose to use pAdTrack-CMV (Addgene, variant allele frequency and statistical significance. DESeq2 is a software Cambridge, MA) because it is a vector used for the expression of package that uses a method of estimate variance-mean dependence in count transgenes when a GFP tracer is desired. data from high-throughput sequencing assays and tests for differential expression based on a model using the negative binomial distribution. Preparation of protein lysates and western blot analyses Cellular extracts were prepared by the single-step lysis method (Paul et al., In silico structural analysis 2011). SH-SY5Y cells [CRL-2266, American Type Culture Collection In silico structural analysis of the variants was done using PolyPhen, SIFT, (ATCC), Manassas, VA] were harvested and suspended in SDS-PAGE PROVEAN, Mutation Taster and CADD (http://genetics.bwh.harvard.edu/ sample buffer (Bio-Rad, Hercules, CA) and then boiled for 5 min. Equal pph/, http://sift.jcvi.org/, http://provean.jcvi.org/, http://www.mutationtaster. amounts of the extracts were subjected to western blot analysis to determine org/, http://cadd.gs.washington.edu/, respectively). For evolutionary protein levels using the antibodies listed below. Protein extracts were conservation, protein sequences were downloaded from NCBI (http:// resolved by SDS-PAGE and transferred to Hybond P membranes www.ncbi.nlm.nih.gov/protein), and the alignments were calculated using (Amersham Bioscience, Piscataway, NJ). After blocking with 5% skim ClustalW (http://www.ebi.ac.uk/clustalw2). milk in 0.1% Tween 20/phosphate buffered saline (PBS), the membranes Disease Models & Mechanisms

560 RESEARCH ARTICLE Disease Models & Mechanisms (2016) 9, 553-562 doi:10.1242/dmm.022848 were incubated with primary antibodies in 5% skim milk in 0.1% Tween 20/ analyzed and calibrated with a custom-made macro-enabled Excel PBS for 2 h at room temperature or overnight at 4°C. After several washes workbook. Data are reported as means±s.e.m. Statistical differences were with 0.1% Tween 20/PBS, the membranes were incubated with the evaluated by Student’s two-tailed t-test for unpaired samples. A P-value of corresponding secondary antibodies conjugated with HRP in 5% skim milk ≤0.05 was considered statistically significant. The cell line was recently in 0.1% Tween 20/PBS for 2 h at room temperature. Following three authenticated and tested for mycoplasma contamination. additional washes with 0.1% Tween 20/PBS, signals were detected by using the Immobilon Western Chemiluminescent HRP Substrate (Millipore, Functional complementation assay in K616 yeast cells Temecula, CA) according to the manufacturer’s protocol. The following Saccharomyces cerevisiae strain K616 (Mat α; pmr1::HIS3; pmc1::TRP1; antibodies were used throughout the study: rabbit anti-calcium PMCA3 cnb1::LEU2; ade2; ura3) (Cunningham and Fink, 1994) has been used ATPase (ab3530, 1:3000, Abcam Inc., Cambridge, UK), mouse monoclonal previously to perform Ca2+-ATPase functional complementation assays, anti-calbindin-D-28K (C9848, 1:5000, Sigma Inc., St Louis, MO) and and was used in the present study (Baekgaard et al., 2006). Briefly, yeast mouse monoclonal anti-GFP (sc-9996, 1:5000, Santa Cruz Inc., Dallas, cells were transformed by standard protocols (Gietz and Woods, 2002) with TX). To control for protein quality and loading, the membranes were different pYES2-derived vectors, carrying either WT or mutagenized reprobed with monoclonal anti β-actin (A3854, 1:10,000, Sigma Inc., St rat Atp2b3b cDNAs. The c.33C>T Atp2b3 point mutation expressing Louis, MO). Secondary antibodies were: donkey anti-rabbit IgG-HRP (sc- PMCA3R35C was introduced by using the QuikChange II XL site-directed 2057, 1:5000, Santa Cruz Inc., Dallas, TX) and goat anti-mouse IgG-HRP mutagenesis kit (200521-5; Stratagene, Santa Clara, CA) using (A2304, 1:5000, Sigma Inc., St Louis, MO). The cell line was recently primers For (5′-CACGCTGGCAGAACTATGCAGCCTCATGG-3′) and authenticated and tested for mycoplasma contamination. Rev (5′-CCATGAGGCTGCATAGTTCTGCCAGCGTG-3′), and verified by sequencing. Positive clones were selected on minimal synthetic defined Immunohistochemistry (SD) medium (6.7% yeast nitrogen base with ammonium sulphate, 2% Rats were deeply anesthetized with isoflurane, then perfused with ice-cold glucose) lacking uracil and supplemented with 10 mM CaCl2. Functional PBS transcardially. Tissue was quickly removed and submerged into cold complementation assays (repeated for at least 3-5 independent clones) were 4% paraformaldehyde (PFA) (Electron Microscopy Sciences, Hatfield, PA) performed by growing yeast cells in SD+CaCl2 until mid-log phase, when a and kept at 4°C overnight. The following day, PFA was replaced with 10% similar number of yeast cells were serially diluted and spotted on minimal sucrose solution in 1× PBS, and kept at 4°C overnight. The following day, medium containing glucose (pYES2-PMCA3b repression) or galactose 10% sucrose was replaced with 20% sucrose solution at 4°C overnight. The (pYES2-PMCA3b induction) as the carbon source. Addition of 10 mM following day, 20% sucrose solution was replaced with 30% sucrose CaCl2, or 7.5 and 10 mM EGTA to glucose or galactose media reflects solution, stored at 4°C, and then mounted in Tissue-Tek O.C.T. Compound permissive and nonpermissive conditions for yeast K616 viability, (Sakura Finetek, Torrence, CA) and stored at −80°C until the time of respectively. sectioning. Tissue sections were cut into slices of 10 μmor40μm and placed on Acknowledgements Superfrost Plus microscope slides (Thermo Fisher Scientific, Grand Island, We thank Dr Daniel Tolbert for providing the shaker colony and Dr Mark Rinehart and NY). The slides were washed in a 1% SDS solution for 5 min, followed by a Christopher Nelson for their contributions and help with histology. wash in 1× PBS for 5 min to remove residual SDS. Slides were then washed four times for 15 min each in a washing buffer consisting of 0.01% Triton Competing interests X-100 and 0.005% Tween 20 in 1× PBS. Blocking solution was applied and The authors declare no competing or financial interests. incubated overnight at 4°C. Primary antibodies mouse monoclonal anti- calbindin-D-28K (C9848, 1:500, Sigma Inc., St Louis, MO) or rabbit Author contributions polyclonal anti-PMCA3 ATPase (ab3530, 1:2000, Abcam, Cambridge, K.P.F. – design, interpretation of data, execution of experiments, preparation and UK) were applied and incubated overnight at 4°C. Slides were washed four editing of article. S.P. – interpretation of data, execution of experiments, preparation times for 15 min each in aforementioned washing buffer to remove residual of article. T.C. – design, interpretation of data, execution of experiments, preparation primary antibodies. Secondary antibodies goat-anti-mouse IgG Dylight 550 of article. S.K. – interpretation of data, execution of experiments, editing of article. – – (84540, 1:500, Thermo Fisher Scientific, Grand Island, NY) or goat-anti- R.L. design, interpretation of data, execution of experiments. M.F. execution of – – rabbit IgG Dylight 550 (84541, 1:2000, Thermo Fisher Scientific, Grand experiments. D.A. interpretation of data, editing of article. G.Z. interpretation of data. M.B. – conception and design of experiments, interpretation of data, Island, NY) was applied and allowed to incubate for 2 h at room preparation of article. W.D. – interpretation of data, execution of experiments, temperature. Slides were washed four times for 15 min each with wash preparation of article. B.M. – interpretation of data, editing of article. D.R.S. – buffer to remove residual secondary antibodies. Slides were cover-slipped interpretation of data and manuscript editing. E.C. – conception and design of with Fluorogel II containing DAPI (17985-5051, Electron Microscopy experiments, interpretation of data, preparation of article. S.M.P. – conception, Sciences) and allowed to dry for 24 h. Slides were imaged using Nikon EZ- design, interpretation of data, preparation and editing of article. C1 3.80 Confocal software on a Nikon Eclipse Ti microscope. Funding Ca2+ measurements with recombinant aequorin National Institutes of Health (2R37NS033123, 5R21NS079852), National Ataxia Cytosolic Ca2+ concentration was monitored by using the recombinant Foundation. T.C. is supported by the University of Padova (Progetto giovani 2012 no Ca2+-sensitive photoprotein aequorin cytAEQ (Ottolini et al., 2014). GRIC128SP0) and the Ministry of University and Research (Bando SIR 2014 no RBSI14C65Z). Measurements were carried out in a Perkin-Elmer (Perkin-Elmer, Waltham, MA) Envision plate reader equipped with a two-injector unit. Data availability The day prior to Ca2+ measurements, transfected cells were plated onto 96-well plates. Recombinant cytAEQ was reconstituted by incubating HeLa Supplementary information cells (CCL-2, ATCC, Manassas, VA) for 1-3 h with 5 µM coelenterazine in Supplementary information available online at DMEM supplemented with 1% FBS at 37°C in a 5% CO2 atmosphere. After http://dmm.biologists.org/lookup/suppl/doi:10.1242/dmm.022848/-/DC1 reconstitution, cells were placed in 70 µl of Krebs Ringer Buffer (KRB; NaCl 135 mM, KCl 5 mM, KH2PO4 400 mM, MgSO4 1 mM, Hepes References 20 mM, pH 7.4) supplemented as indicated in the figures, and luminescence Arkadir, D. and Louis, E. D. (2013). The balance and gait disorder of essential from each well was measured for 1 min. Depending on the experiment, tremor: what does this mean for patients? Ther. Adv. Neurol. Disord. 6, 229-236. 100 µM histamine or 2 mM CaCl (final concentration) were injected to Armesilla, A. L., Williams, J. C., Buch, M. H., Pickard, A., Emerson, M., 2 Cartwright, E. J., Oceandy, D., Vos, M. D., Gillies, S., Clark, G. J. et al. (2004). generate Ca2+ transients. The experiments were terminated by lysing the 2+ Novel functional interaction between the plasma membrane Ca2+ pump 4b and cells with 100 µM digitonin in a hypotonic Ca -rich solution (10 mM the proapoptotic tumor suppressor Ras-associated factor 1 (RASSF1). J. Biol.

CaCl2 in H2O) to discharge the remaining aequorin pool. Output data were Chem. 279, 31318-31328. Disease Models & Mechanisms

561 RESEARCH ARTICLE Disease Models & Mechanisms (2016) 9, 553-562 doi:10.1242/dmm.022848

Baekgaard, L., Luoni, L., De Michelis, M. I. and Palmgren, M. G. (2006). The plant Kircher, M., Witten, D. M., Jain, P., O’Roak, B. J., Cooper, G. M. and Shendure, J. plasma membrane Ca2+ pump ACA8 contains overlapping as well as physically (2014). A general framework for estimating the relative pathogenicity of human separated autoinhibitory and calmodulin-binding domains. J. Biol. Chem. 281, genetic variants. Nat. Genet. 46, 310-315. 1058-1065. Koboldt, D. C., Zhang, Q., Larson, D. E., Shen, D., McLellan, M. D., Lin, L., Miller, Bertini, E., des Portes, V., Zanni, G., Santorelli, F., Dionisi-Vici, C., Vicari, S., C. A., Mardis, E. R., Ding, L. and Wilson, R. K. (2012). VarScan 2: somatic Fariello, G. and Chelly, J. (2000). X-linked congenital ataxia: a clinical and mutation and copy number alteration discovery in cancer by exome sequencing. genetic study. Am. J. Med. Genet. 92, 53-56. Genome Res. 22, 568-576. Brini, M. (2009). Plasma membrane Ca(2+)-ATPase: from a housekeeping function La Regina, M. C., Yates-Siilata, K., Woods, L. and Tolbert, D. (1992). to a versatile signaling role. Pflugers Arch. 457, 657-664. Preliminary characterization of hereditary cerebellar ataxia in rats. Lab. Anim. Brini, M. and Carafoli, E. (2009). Calcium pumps in health and disease. Physiol. Sci. 42, 19-26. Rev. 89, 1341-1378. Linde, C. I., Di Leva, F., Domi, T., Tosatto, S. C. E., Brini, M. and Carafoli, E. Canzian, F. (1997). Phylogenetics of the laboratory rat Rattus norvegicus. Genome (2008). Inhibitory interaction of the 14–3–3 proteins with ubiquitous (PMCA1) and Res. 7, 262-267. tissue-specific (PMCA3) isoforms of the plasma membrane Ca2+ pump. Cell Carlson, K. M., Andresen, J. M. and Orr, H. T. (2009). Emerging pathogenic Calcium 43, 550-561. pathways in the spinocerebellar ataxias. Curr. Opin. Genet. Dev. 19, 247-253. Louis, E. D. (2009). Essential tremors: a family of neurodegenerative disorders? Chou, H., Ogawa, N., Asanuma, M., Hirata, H. and Mori, A. (1991). Dopamine Arch. Neurol. 66, 1202-1208. deficiency in the weaver mutant mouse: an animal model of olivopontocerebellar Louis, E. D. and Okun, M. S. (2011). It is time to remove the ‘benign’ from the atrophy. Res. Commun. Chem. Pathol. Pharmacol. 74, 117-120. essential tremor label. Parkinson. Relat. Disord. 17, 516-520. Clark, B. R., LaRegina, M. and Tolbert, D. L. (2000). X-linked transmission of the Love, M. I., Huber, W. and Anders, S. (2014). Moderated estimation of fold change shaker mutation in rats with hereditary Purkinje cell degeneration and ataxia. Brain and dispersion for RNA-seq data with DESeq2. Genome Biol. 15, 550. Res. 858, 264-273. Malamud, N. and Cohen, P. (1958). Unusual form of cerebellar ataxia with sex- Cunningham, K. W. and Fink, G. R. (1994). Calcineurin-dependent growth control linked inheritance. Neurology 8, 261-266. in Saccharomyces cerevisiae mutants lacking PMC1, a homolog of plasma Meier, H. (1967). The neuropathy of teetering, a neurological mutation in the mouse. membrane Ca2+ ATPases. J. Cell Biol. 124, 351-363. Arch. Neurol. 16, 59-66. Daniel, E. E., El-Yazbi, A. and Cho, W. J. (2006). Caveolae and calcium handling, a Ohata, T., Yoshida, K., Sakai, H., Hamanoue, H., Mizuguchi, T., Shimizu, Y., review and a hypothesis. J. Cell. Mol. Med. 10, 529-544. Okano, T., Takada, F., Ishikawa, K., Mizusawa, H. et al. (2006). A -16C>T Dansithong, W., Paul, S., Figueroa, K. P., Rinehart, M. D., Wiest, S., Pflieger, substitution in the 5′ UTR of the puratrophin-1 gene is prevalent in autosomal L. T., Scoles, D. R. and Pulst, S. M. (2015). Ataxin-2 regulates RGS8 translation dominant cerebellar ataxia in Nagano. J. Hum. Genet. 51, 461-466. in a new BAC-SCA2 transgenic mouse model. PLoS Genet. 11, e1005182. Ottolini, D., Cali, T. and Brini, M. (2014). Methods to measure intracellular Ca(2+) DeMarco, S. J. and Strehler, E. E. (2001). Plasma membrane Ca2+- fluxes with organelle-targeted aequorin-based probes. Methods Enzymol. 543, isoforms 2b and 4b interact promiscuously and selectively with members of the 21-45. membrane-associated guanylate kinase family of PDZ (PSD95/Dlg/ZO-1) Paul, S., Dansithong, W., Jog, S. P., Holt, I., Mittal, S., Brook, J. D., Morris, G. E., domain-containing proteins. J. Biol. Chem. 276, 21594-21600. Comai, L. and Reddy, S. (2011). Expanded CUG repeats Dysregulate RNA Eakin, T. J., Antonelli, M. C., Malchiodi, E. L., Baskin, D. G. and Stahl, W. L. (1995). splicing by altering the stoichiometry of the muscleblind 1 complex. J. Biol. Chem. Localization of the plasma membrane Ca(2+)-ATPase isoform PMCA3 in rat 286, 38427-38438. cerebellum, choroid plexus and hippocampus. Brain Res. Mol. Brain Res. 29,71-80. Penheiter, A. R., Caride, A. J., Enyedi, A. and Penniston, J. T. (2002). Tryptophan Farlow, M. R., DeMyer, W., Dlouhy, S. R. and Hodes, M. E. (1987). X-linked 1093 is largely responsible for the slow off rate of calmodulin from plasma recessive inheritance of ataxia and adult-onset dementia: clinical features and membrane Ca2+ pump 4b. J. Biol. Chem. 277, 17728-17732. preliminary linkage analysis. Neurology 37, 602-607. Sato, N., Amino, T., Kobayashi, K., Asakawa, S., Ishiguro, T., Tsunemi, T., Filoteo, A. G., Elwess, N. L., Enyedi, A., Caride, A., Aung, H. H. and Penniston, Takahashi, M., Matsuura, T., Flanigan, K. M., Iwasaki, S. et al. (2009). J. T. (1997). Plasma membrane Ca2+ pump in rat brain: patterns of alternative type 31 is associated with “inserted” penta-nucleotide splices seen by isoform-specific antibodies. J. Biol. Chem. 272, 23741-23747. repeats containing (TGGAA)n. Am. J. Hum. Genet. 85, 544-557. Frankel, W. N., Sweet, H. O. and Davisson, M. T. (1994). The stumbler mutation Schmidley, J. W., Levinsohn, M. W. and Manetto, V. (1987). Infantile X-linked maps to proximal mouse chromosome 2. Mamm. Genome 5, 659-662. ataxia and deafness: a new clinicopathologic entity? Neurology 37, 1344-1349. Gietz, R. D. and Woods, R. A. (2002). Transformation of yeast by lithium acetate/ Schorge, S., van de Leemput, J., Singleton, A., Houlden, H. and Hardy, J. single-stranded carrier DNA/polyethylene glycol method. Methods Enzymol. 350, (2010). Human ataxias: a genetic dissection of inositol triphosphate receptor 87-96. (ITPR1)-dependent signaling. Trends Neurosci. 33, 211-219. Gonzalez-Alegre, P., Di Paola, J., Wang, K., Fabbro, S., Yu, H. C., Shaikh, T. H., Schwarz, J. M., Rodelsperger, C., Schuelke, M. and Seelow, D. (2010). Darbro, B. W. and Bassuk, A. G. (2014). Evaluating familial essential tremor with MutationTaster evaluates disease-causing potential of sequence alterations. novel genetic approaches: is it a genotyping or phenotyping issue? Tremor. Other Nat. Methods 7, 575-576. Hyperkinet. Mov. 4, 258. Sgambato-Faure, V., Xiong, Y., Berke, J. D., Hyman, S. E. and Strehler, E. E. Green, M. C. and Sidman, R. L. (1962). Tottering–a neuromusclar mutation in the (2006). The Homer-1 protein Ania-3 interacts with the plasma membrane calcium mouse. And its linkage with oligosyndacylism. J. Hered. 53, 233-237. pump. Biochem. Biophys. Res. Commun. 343, 630-637. Guerini, D., Zecca-Mazza, A. and Carafoli, E. (2000). Single amino acid mutations Soha, J. M. and Herrup, K. (1995). Stunted morphologies of cerebellar Purkinje in transmembrane domain 5 confer to the plasma membrane Ca2+ pump cells in lurcher and staggerer mice are cell-intrinsic effects of the mutant genes. properties typical of the Ca2+ pump of endo(sarco)plasmic reticulum. J. Biol. J. Comp. Neurol. 357, 65-75. Chem. 275, 31361-31368. Tanner, C. M., Goldman, S. M., Lyons, K. E., Aston, D. A., Tetrud, J. W., Welsh, Guyer, C. L., Stoica, G., Womack, J. E., Storts, R. W., Derr, J. N. and Abbott, L. C. M. D., Langston, J. W. and Koller, W. C. (2001). Essential tremor in twins: an (2002). A line of Berlin Druckrey IV rats proposed as a new model for human assessment of genetic vs environmental determinants of etiology. Neurology 57, hereditary ataxia. In Vivo 16, 255-263. 1389-1391. Hagerman, R. J., Leehey, M., Heinrichs, W., Tassone, F., Wilson, R., Hills, J., Thomas, M. A., Chen, C.-F., Jensen-Seaman, M. I., Tonellato, P. J. and Twigger, Grigsby, J., Gage, B. and Hagerman, P. J. (2001). Intention tremor, S. N. (2003). Phylogenetics of rat inbred strains. Mamm. Genome 14, 61-64. parkinsonism, and generalized brain atrophy in male carriers of fragile X. Tolbert, D. L., Ewald, M., Gutting, J. and La Regina, M. C. (1995). Spatial and Neurology 57, 127-130. temporal pattern of Purkinje cell degeneration in shaker mutant rats with Hansen, S. T., Meera, P., Otis, T. S. and Pulst, S. M. (2013). Changes in Purkinje hereditary cerebellar ataxia. J. Comp. Neurol. 355, 490-507. cell firing and gene expression precede behavioral pathology in a mouse model of Ton, V.-K. and Rao, R. (2004). Functional expression of heterologous proteins in SCA2. Hum. Mol. Genet. 22, 271-283. yeast: insights into Ca2+ signaling and Ca2+-transporting ATPases. Ishikawa, K., Toru, S., Tsunemi, T., Li, M., Kobayashi, K., Yokota, T., Amino, T., Am. J. Physiol. Cell Physiol. 287, C580-C589. Owada, K., Fujigasaki, H., Sakamoto, M. et al. (2005). An autosomal dominant Zanni, G., Bertini, E., Bellcross, C., Nedelec, B., Froyen, G., Neuhauser, G., cerebellar ataxia linked to chromosome 16q22.1 is associated with a single- Opitz, J. M. and Chelly, J. (2008). X-linked congenital ataxia: a new locus maps nucleotide substitution in the 5′ untranslated region of the gene encoding a protein to Xq25-q27.1. Am. J. Med. Genet. A 146A, 593-600. with spectrin repeat and Rho guanine-nucleotide exchange-factor domains. Zanni, G., Cali, T., Kalscheuer, V. M., Ottolini, D., Barresi, S., Lebrun, N., Am. J. Hum. Genet. 77, 280-296. Montecchi-Palazzi, L., Hu, H., Chelly, J., Bertini, E. et al. (2012). Mutation of Jensen, T. P., Filoteo, A. G., Knopfel, T. and Empson, R. M. (2007). Presynaptic plasma membrane Ca2+ ATPase isoform 3 in a family with X-linked congenital plasma membrane Ca2+ ATPase isoform 2a regulates excitatory synaptic cerebellar ataxia impairs Ca2+ homeostasis. Proc. Natl. Acad. Sci. USA 109, transmission in rat hippocampal CA3. J. Physiol. 579, 85-99. 14514-14519. Disease Models & Mechanisms

562