Molecular and electrophysiological features of spinocerebellar ataxia type seven in induced pluripotent stem cells

Spinocerebellar ataxia type 7 (SCA7) is an inherited neurodegenerative disease caused by a polyglutamine repeat expansion in the ATXN7 gene. Patients with this disease suffer from a degeneration of their cerebellar Purkinje neurons and retinal photoreceptors that result in a progressive ataxia and loss of vision. As with many neurodegenerative diseases, studies of pathogenesis have been hindered by a lack of disease-relevant models. To this end, we have generated induced pluripotent stem cells (iPSCs) from a cohort of SCA7 patients in South Africa. First, we differentiated the SCA7 affected iPSCs into neurons which showed evidence of a transcriptional phenotype affecting components of STAGA (ATXN7 and KAT2A) and the heat shock protein pathway (DNAJA1 and HSP70). We then performed electrophysiology on the SCA7 iPSC-derived neurons and found that these cells show features of functional aberrations. Lastly, we were able to differentiate the SCA7 iPSCs into retinal photoreceptors that also showed similar transcriptional aberrations to the SCA7 neurons. Our findings demonstrate that iPSC-derived neurons and photoreceptors from SCA7 patients express molecular and electrophysiological differences that are indicative of impaired neuronal health. We hope that these findings will contribute towards the ongoing efforts to establish the cell-derived models of neurodegenerative diseases that are needed to develop patient-specific treatments.


23
Spinocerebellar ataxia type 7 (SCA7) is an inherited neurodegenerative disease 24 caused by a polyglutamine repeat expansion in the ATXN7 gene. Patients with this disease 25 suffer from a degeneration of their cerebellar Purkinje neurons and retinal photoreceptors 26 that result in a progressive ataxia and loss of vision. As with many neurodegenerative 27 diseases, studies of pathogenesis have been hindered by a lack of disease-relevant 28 models. To this end, we have generated induced pluripotent stem cells (iPSCs) from a 29 cohort of SCA7 patients in South Africa. First, we differentiated the SCA7 affected iPSCs 30 into neurons which showed evidence of a transcriptional phenotype affecting components of 31 STAGA (ATXN7 and KAT2A) and the heat shock protein pathway (DNAJA1 and HSP70). 32 We then performed electrophysiology on the SCA7 iPSC-derived neurons and found that 33 these cells show features of functional aberrations. Lastly, we were able to differentiate the 34 SCA7 iPSCs into retinal photoreceptors that also showed similar transcriptional aberrations 35 to the SCA7 neurons.
Our findings demonstrate that iPSC-derived neurons and 36 photoreceptors from SCA7 patients express molecular and electrophysiological differences 37 that are indicative of impaired neuronal health. We hope that these findings will contribute 38 Introduction 41 Spinocerebellar ataxia type 7 (SCA7) is an inherited neurodegenerative disease 42 caused by a CAG repeat expansion in the ATXN7 gene. Since the translation of this CAG 43 repeat leads to an expanded polyglutamine (polyQ) tract within the resultant protein, SCA7 44 is classified as a polyQ repeat disorder. Other diseases with a similar pathophysiology 45 include five different SCAs (SCA 1, 2, 3, 6 and 17), as well as Huntington disease, 46 dentatorubral-pallidoluysian atrophy and spinal bulbar muscular atrophy [1]. Clinically, 47 SCA7 patients present with ataxia, dysarthria and visual loss. This is caused by a selective 48 degeneration of cerebellar Purkinje neurons and retinal photoreceptors [2]. Symptoms 49 progressively worsen over a period of 10 to 30 years, leading ultimately to brainstem 50 dysfunction, blindness, physical disability and death. 51 The mechanism by which a polyQ expansion within the ubiquitously expressed ATXN7 52 protein leads to the selective degeneration of Purkinje neurons and photoreceptors remains 53 to be fully elucidated. ATXN7 is known to be a component of the mammalian transcription 54 co-activator complex, STAGA (SPT3-TAF9-ADA-GCN5 acetyltransferase) [3]. This protein 55 has been shown to facilitate the interaction between STAGA and the cone-rod homeobox 56 (CRX) transactivator of photoreceptor genes, linking the function of ATXN7 with the retinal 57 phenotype observed in SCA7 patients [4]. In addition, several studies have highlighted the 58 role of transcriptional aberrations in the neuronal cell dysfunction that precedes the onset of 59 disease symptoms [3][4][5][6][7]. These gene expression changes may arise either directly from 60 alterations in transcriptional regulation by mutant ATXN7, or indirectly, as a consequence of 61 a generalised cellular response to the presence of mutant ATXN7. More recently, 62 9 41Q and CON B showed lower levels of OCT4 expression compared to the remaining three 151 lines (although still significantly higher than differentiated fibroblasts and retinal cells). 152

Neural differentiation and characterisation
153 Differentiation of iPSCs into neural progenitors was performed by treatment of iPSCs with 154 3µM glycogen synthase kinase 3 (GSK3) inhibitor (CHIR99021) and 2µM TGFβ inhibitor 155 (SB431542) as previously described, with neural progenitors appearing in culture after 156 seven days. For neuronal differentiation, neural progenitors were seeded at a density of 157 150 000 cells/well onto a Matrigel-coated six-well plate in neural induction medium [25]. 158 After two days, medium was changed to neuronal differentiation medium (DMEM/FBS 159 supplemented with N2, B27, 300ng/ml cyclic AMP (Sigma), 0.2mM ascorbic acid (Sigma), 160 10ng/ml BDNF (Peprotech) and 10ng/ml GDNF (Peprotech)) and the cells maintained in 161 culture for 14 to 21 days, with neurite outgrowth typically observed after one week of 162 culture. Medium was changed every 2-3 days. Characterisation was performed by 163 immunocytochemistry and qRT-PCR (for antibodies and primers see S1-3 Tables). 164

165
Neurons cultured between 2-3 weeks on glass cover slips (uncoated) were removed from 166 the incubator and rapidly transported to the recording chamber of a Zeiss Axioskop Upright 167 Microscope (Zeiss). Electrophysiological recordings were made in neuronal differentiation 168 medium at room temperature and were restricted to the first 5 hours following cell removal 169 from the incubator environment. Patch pipettes of 13-20 MOhm tip resistance were pulled 170 from filamental borosilicate glass capillaries (2.00 mm outer diameter, 1.58 mm inner 171 diameter, Hilgenberg), using a horizontal puller (Model P-1000, Sutter). The pipettes were 172 filled with an internal solution containing (in mM): K-gluconate (126); KCl (4); Na 2 ATP (4); 173 NaGTP (0.3); Na 2 -phosphocreatinine (10) and HEPES (10). Osmolarity was adjusted to 174 between 290 and 300 mOsM and the pH was adjusted to between 7. 38 and 7.42 with KOH. 175 Cells were visualised using a 40x water-immersion objective (Zeiss). Digital images were 176 obtained using a CCD camera (VX55, TILL Photonics). Individual cells were selected for 177 recordings based on a small round or ovoid cell body (diameters, 5-10 μ m) and typically 178 two or more extended processes.

225
The length of the disease-causing CAG repeat in ATXN7 was determined from DNA by 226 means of PCR and automated fluorescent genotyping. The PCR reaction mix consisted of 227 0.4µM each, forward and reverse primer (S3 Table) To determine whether any transcriptional differences could be detected between SCA7 261 patient-and unaffected control-derived cell types, a panel of candidate genes was selected 262 which had previously been shown to be dysregulated in the retinal and cerebellar tissue of 263 SCA7 mouse models and patient lymphoblasts [5,6,30,31]. The panel included the 264 following genes: ATXN7, brain expressed, X-linked 1 (BEX1), DnaJ (Hsp40)  GABAergic neurons did not appear to vary between SCA7 patients and controls (Fig 1E-F). 294 No obvious differences in morphology were observed when comparing neurons derived 295 from SCA7 patient iPSCs with those derived from controls. In order to test functional 296 differences between SCA7 neurons and controls, electrophysiological measurements were 297 performed on a subset of these cells. 298 SCA7 iPSC-derived neurons demonstrate functional differences 299 that may affect cell excitability 300 Cells were assayed for physiological properties between 14-and 23-days post induction of 301 neuronal differentiation. Cells were targeted for whole-cell recordings based on their 302 morphological properties. This included a small round or ovoid cell body with diameters 303 between 5-10 μ m and typically two or more extended processes. Following the attainment 304 of a whole-cell patch, current pulses of between 0 and 10 pA were applied. Individual cells 305 displayed four general types of spiking responses: a purely passive (Fig 2A), abortive spike 306 ( Fig 2B), single spike ( Fig 2C) and recurrent spiking response (Fig 2D). These spiking 307 properties are similar to those observed in acute human fetal brain slices [34], hESC-308 derived neurons [35][36][37] and iPSC-derived neurons [38]. A postmitotic neuron matures by 309 inserting voltage-gated channels into its plasma membrane [39]. Therefore, the spiking 310 response of a cell to current injection can be used to determine the maturation stage of a 311 differentiating neuron: passive (least mature) Spiking responses were collected in current-clamp mode from cells derived from four 325 separate iPSC lines: two control lines CON A (n = 70) and CON B (n = 42), and two patient 326 lines 47Q (n = 44) and 41Q (n = 72). Although the fraction of cells which fell into each 327 spiking response category was significantly dependent on the iPSC line from which the cells 328 were derived (p < 0.0001, Chi-squared test), no trend could be discerned between control 329 and patient lines (Fig 2E). For example, the control line CON A had the most mature 330 phenotype with the highest fraction of cells in the single spike and recurrent spiking 331 categories whilst 47Q demonstrated a relatively immature phenotype, with the majority of 332 cells displaying spiking responses falling into the passive category. However, the other 333 lines did not corroborate this difference as the control line CON B displayed a less mature 334 phenotype than patient line 41Q. 335 Next, we compared the resting membrane potential (Vm) of cells derived from each cell line. 336 This parameter was significantly dependent on the iPSC line from which the cells were 337 derived ( Fig 2F, p = 0.002, ANOVA). The mean resting membrane potential +/-SEM for the 338 CON A , CON B , 47Q and 41Q cell lines were -57.0 +/-1.7, -54.7 +/-3.2, -67.3 +/-2.7 and -339 62.5 +/-2.5 mV respectively. 340 Next, we performed voltage-clamp recordings of the neurons in order to measure the input 341 resistance as well as the voltage-gated sodium and potassium currents (Fig 3A-B). A lower 342 input resistance is associated with neurite outgrowth and increased numbers of ion 343 channels inserted into the plasma membrane during the process of neuronal maturation. 344 Once again, a cell's input resistance was significantly dependent on the cell line to which it 345 belonged (Fig 3C, p = 0.005, ANOVA). Input resistance +/-SEM was 4987 +/-421, 5484 +/-346 243, 5979 +/-513 and 7094 +/-455 mΩ for the CON A , CON B , 47Q and 41Q cell lines 347 respectively. 348 markers. The cells were stained for either the disease-causing protein ATXN7 (Fig 4A) or 376 the retinal cell markers CRX and RCVRN (Fig 4B-C). No obvious differences in morphology 377 were observed between SCA7 patient and control iPSC-derived cells. The differentiated 378 cells displayed varying expression levels of the retinal genes CRX, PAX6, RCVRN and 379 OTX2 (Fig 4D-G). 380 NPCs) may be indicative of a possible differential response by the different cell types to the 408 presence of mutant ATXN7. As with the NPCs, ATXN7 CAG repeat alleles were confirmed 409 using an RT-PCR (S Fig 5B). 410  (25,26,33), which found no 424 difference in differentiation potential between patient and control iPSCs. In addition, no 425 obvious difference in the ability to generate GABA-positive processes could be detected 426 between SCA7 and control iPSC-derived neurons. 427

Discussion
Electrophysiological studies were carried out to establish whether there were any 428 functional differences in the intrinsic properties of patient and control cells. The majority of 429 cells recorded were capable of generating spiking activity including single and multiple 430 action potentials, an indication of neuronal maturity [39]. Despite the presence of significant 431 differences in spiking responses between the four cell lines, we did not observe a reliable 432 trend between the control and patient derived neurons. We do acknowledge a major 433 limitation in these findings being significant variability across cell lines. This is likely due to 434 unappreciated differences in culturing conditions between cell lines, which may have 435 masked our ability to detect a reliable difference in spiking responses caused by the mutant 436 ATXN7. Alternatively, it may be that mutant ATXN7 does not affect the spiking properties of 437 neurons at early stages of development, suggesting that an extrinsic stressor of some kind 438 might be required to elicit a pathological phenotype. Indeed, similar functional analysis of 439 neurons derived from patients with SCA3, a related polyQ-repeat disorder, showed no 440 difference between control and patient cells until neurons were excited via bath application 441 of glutamate [33]. 442 22 Whilst we acknowledge that it is not possible to make conclusive claims due to the 443 differences between all the cell lines used, we did observe an apparent difference in resting 444 membrane potential and input resistance between control and SCA7 patient derived 445 neurons. Patient cells had more negative resting membrane potentials and increased input 446 resistance compared to control cells. A more hyperpolarised membrane potential is usually 447 associated with neuronal maturity whist the high input resistance is usually associated with 448 neuronal immaturity. These conflicting findings will need to be further explored in future 449 studies to provide a more conclusive understanding as to how SCA7 specific changes 450 cause changes in neuronal maturation or drive alternative functional alterations. We can, 451 however, postulate that the more hyperpolarised membrane potential in the patient cells  Fig 4A). This strongly suggests that the differentiated 480 cells represent a population of cells at an early stage of development, rather than 481 recapitulating the age or disease stage of the patient from which the primary cells were 482 derived. Previous studies employing similar models for the study of neurodegenerative 483 disease have raised concerns regarding the relevance of modelling adolescent-and adult-484 onset diseases over the short lifespan of cultured neurons [46,47]. These findings suggest 485 that pathological hallmarks of disease such as the formation of aggregates may take 486 decades to manifest, requiring the gradual accumulation of toxic proteins as a result of age-487 dependent deficiencies in protein homeostasis. Although some studies suggest that 488 24 aggregates may be detected at earlier stages, the major determinants of aggregate 489 formation remain the length of the polyQ expansion, and the levels of expression of the 490 polyQ-containing protein [48,49]. Thus, a cell line derived from an individual expressing 491 endogenous levels of a moderately expanded ATXN7 protein may be less likely to 492 demonstrate an observable cellular phenotype. Alternatively, the aggregation of mutant 493 protein may require prolonged periods in culture, or an exogenous trigger, such as 494 exposure to oxidative stress or neurotoxins, or excitation-induced calcium influx [33]. 495 The role of transcriptional dysregulation in the polyQ diseases has been extensively 496 documented, particularly in cell and animal models of SCA7 [5,6,30,31,50,51]. The 497 identification of gene expression changes, which precede the onset of symptoms, suggests 498 strongly that alterations in transcription may be among the earliest manifestations of 499 disease [52]. Thus, it follows that gene expression changes may be used as a tool to 500 identify a disease-associated phenotype in cells representing early stages of development 501 [53]. 502 In order to investigate gene expression changes in the SCA7 iPSCs and iPSC-derived 503 neurons generated here, a panel of candidate transcripts were selected, in which robust 504 changes had been previously demonstrated [4,6,31,[54][55][56]. The iPSC-derived NPCs and 505 retinal photoreceptors displayed changes in expression of these key transcripts, suggesting 506 that these cells may serve as useful models of neurodegenerative disease progression and 507 for the testing of potential therapies (Fig 1D and Fig 4H). It should be noted that the low cell 508 numbers and heterogeneity of mature neuronal cultures proved a challenge to obtaining 509 significant biological material to perform reliable expression analysis. Hence, the decision 510 was made to perform this analysis on the relatively homogeneous NPC cultures, with the 511 25 view that these cells may yield insights into the early cellular dysfunction underlying SCA7 512 pathology. 513 Of the three genes consistently downregulated across both differentiated cell types, two 514 (ATXN7 and KAT2A) encode components of the STAGA transcriptional co-activator 515 complex.
Previous studies in SCA7 patient fibroblasts and mouse models have 516 demonstrated a disease-associated increase in ATXN7 expression, mediated by non-517 coding RNAs [54,57]. The contradictory decrease in ATXN7 expression in SCA7 NPCs 518 and photoreceptors observed here could reflect the early developmental stage of the cells, 519 but further analysis of the regulatory pathways will be required in order to elucidate the 520 basis for this apparent decrease in the disease-causing protein in affected cell types. 521 shown to result in increased retinal degeneration in SCA7 mice [58]. 526 The interaction between ATXN7 and CRX has been hypothesised to be a key factor behind 527 the development of retinal degeneration in SCA7 patients [51]. Therefore, the expression of 528 multiple known CRX targets, which were previously shown to be down-regulated in SCA7 529 mice, were included in the gene expression experiments. None of these target genes 530 (including ARR3, GNAT1 or RHO) showed consistent changes in patient cells. However, 531 transcriptional changes in the expression of additional retinal genes, including OTX2 532 (involved in the determination of photoreceptor cell fate), RCVRN (expressed in 533 photoreceptors), and RPE65 (expressed in retinal pigment epithelial cells), were noted in 534 26 the patient derived cells. A significant degree of heterogeneity was observed in the 535 differentiated retinal cells, both in terms of morphology, and gene/protein expression (Fig 4), 536 therefore additional investigation will be required on additional iPSC cell lines to determine 537 whether these differences can be attributed to experimental differences or pathogenic 538 mechanisms. 539 Downregulation of the HSP genes HSP70 and DNAJA1 was observed in SCA7 patient 540 NPCs. A decrease in levels of these two HSPs has been previously reported in both SCA7 541 mice, and human patient lymphoblasts [6,31]. Although this decrease in expression was 542 hypothesised in mice to represent an advanced stage of disease progression, the early 543 developmental stage recapitulated by our model suggests that decreases in certain HSP 544 genes may instead be an inherent defect, which could predispose certain populations of 545 cells to degeneration. 546 The generation of patient-specific, disease-relevant cell types is particularly important in 547 neurodegenerative diseases; as such cells provide a unique model in which to evaluate 548 disease pathogenesis without the complications associated with transgene overexpression 549 in cell or animal models. In addition, the use of cells containing the patient's own genetic 550 background offers the opportunity to investigate potential modifiers of disease onset and 551 progression [60,61]. Perhaps most importantly to the South African context, iPSC-derived 552 neurons provide the first opportunity to evaluate the efficacy of the allele-specific RNAi-553 based therapy developed by Scholefield et al [16], in disease-affected cells. 554 One significant caveat of this study remains the small number of patients assessed -a 555 consequence of the rare nature of the condition -and the challenges associated with patient 556 recruitment in a developing world setting, in which many of those affected are unable to 557 27 access tertiary healthcare. Whilst, in this study, comparisons were not made between 558 isogenic gene-edited lines, they do retain significantly similar background genetics, as they 559 are all generated from the same immediate family, minimising (to a degree) the extensive 560 differences between unrelated individuals. Future studies will focus on recruitment, in order 561 to extend these investigations in a larger patient cohort. To control for the inherent genetic 562 variability associated with comparisons between unrelated patients, future work will also 563 focus on the generation of isogenic control lines, by means of CRISPR/Cas9-mediated 564 genome editing. Dastidar et al [62] have previously demonstrate this approach. It remains, 565 however, technically challenging to target the disease-causing repeat in ataxin-7 and retain 566 the endogenous regulatory landscape. Due to the sequence homology of the wild -type 567 allele, such a strategy would likely render the wild-type protein non-functional via a 568 frameshift. 569 Nevertheless, the SCA7 iPSCs generated here serve as a resource for differentiation into a 570 variety of disease-associated cell types, providing an ideal model in which to study 571 neurodegenerative diseases. The results of this study provide evidence of a disease 572 phenotype in iPSC-derived cells from the South African SCA7 patient cohort. We hope that 573 our data will contribute to the ongoing efforts to use iPSC cells to study the pathogenesis of 574 neurodegenerative disorders, which are needed in development of population-specific 575 therapies. 576 28 at the South African National Health Laboratory Service for karyotype analyses. We honour 580 the late Ms Ingrid Baumgarten for her invaluable assistance over many years, particularly in 581 establishing patient fibroblast cultures. 582