We used L7-tau-gfp mice (Sekirnjak et al., 2003; Watt et al., 2009) (generous gifts from both S. du Lac and M. Häusser), and C57BL6/J mice (purchased from Jackson laboratories, Bar Harbor, ME, USA) to characterize developmental torpedoes. For the SCA6^84Q/84Q disease model we used transgenic mice harboring a humanized 84-CAG repeat tract in the mouse Cacna1a locus (Jackson laboratories; strain B6.129S7-Cacna1a^tm3Hzo/J; stock number: 008683) (Watase et al., 2008). Mice [postnatal day (P)5-P15] were anesthetized with isoflurane, sacrificed, and the brain was rapidly removed into ice-cold ACSF (in mM) 125 NaCl, 2.5 KCl, 2 CaCl2, 1 MgCl2, 1.25 NaH2PO4, 26 NaHCO3, and 20 glucose. Mice (P15 and older) were transcardially perfused with 4% PFA (EMS, Hatfield, PA, USA) and the brains were quickly removed. In some cases, P30 mice were anesthetized with isoflurane, sacrificed, and the brain was rapidly removed into an ice-cold partial sucrose replacement solution (in mM) 111 Sucrose, 50 NaCl, 2.5 KCl, 0.65 CaCl2, 10 MgCl2, 1.25 NaH2PO4, 25 NaHCO3, and 25 glucose. For mice of all ages, the brains extracted (as above) were then quickly transferred into 4% PFA (EMS, Hartfield, PA, USA), and stored at 4°C on a rotary shaker at 70 RPM for at least 48 h, then transferred to PBS with 0.05% sodium azide. Results were compared for tissue prepared with and without perfusion, but no significant changes were observed and this data was pooled.

The vermis of the cerebellum was then submerged in 0.05% sodium azide in 0.01 M PBS and sliced into 100 μm parasagittal slices on a Leica Vibratome 3000 plus (Concord, ON, Canada). The slices were stored at 4°C in 0.05% sodium azide in 0.01 M PBS. Animal procedures were approved by the McGill Animal Care Committee and conform to the guidelines set in place by the Canadian Council on Animal Care.

All staining was performed in a blocking solution with 5% BSA, 0.05% sodium azide, and 0.4% Triton X in 0.01 M PBS. The primary antibodies used were mouse anti-Neurofilament 200 kD (1:500, Millipore, Temecula, CA, USA; Cat. No.: MAB5266), mouse anti-Myelin Basic Protein (MBP; 1:500, Biolegend, San Diego, CA, USA; Cat. No.: 836501), rabbit anti-Green Fluorescent Protein (1:500, Millipore, Temecula, CA, USA; Cat. No.: AB3080P) (in P5 and P7 animals to amplify the GFP signal), guinea pig anti- (1:500, Cedarlane, Burlington, Ontario, Canada; Cat. No.: 131004(SY)) rabbit anti-Iba1 (1:500, Wako Chemicals, Osaka, Japan, Cat. No.: 019-19741), and rabbit anti-Calbindin D-28k (1:1000, Swant, Switzerland; Cat. No.: CB38). The slices were incubated in primary antibody at 4°C for 72 h on a rotary shaker at 70 RPM. Slices were then rinsed three times in a solution containing 0.4% Triton X in 0.01 M PBS.

The secondary antibodies were used at dilutions corresponding to the primary antibody. The secondary antibodies used were Alexa Fluor 594 donkey anti-rabbit (Jackson ImmunoResearch, West Grove, PA, USA; Cat. No.: 711-585-152), Alexa Fluor 488 donkey anti-mouse (Jackson ImmunoResearch; Cat. No.: 715-545-150), Alexa Fluor 594 donkey anti-guinea pig (Jackson ImmunoResearch; Cat. No.: 706-585-1480) and Alexa Fluor 594 donkey anti-mouse (Jackson ImmunoResearch; Cat. No.: 715-585-10). AffiniPure Fab Fragment donkey anti-mouse (Jackson ImmunoResearch; Cat. No.: 715-007-003) was used during secondary staining with antibodies made in mouse to avoid cross reactivity. The secondary staining was performed at 4°C for 90 min on a rotary shaker at 70 RPM. Slices were rinsed with 0.4% Triton X in 0.01 M PBS, mounted onto slides with Prolong Gold Antifade mounting media (Invitrogen), and stored in the dark at 4°C. Chemicals were purchased from Sigma unless otherwise indicated.

Slices were imaged with a custom-built two-photon microscope with a Ti:Sapphire laser (MaiTai; Spectra Physics, Santa Clara, CA, USA) tuned to either 890 nm (GFP) or 775 nm (non-GFP). Image acquisition was done using ScanImage running in Matlab (Mathworks, Natick, MA, USA) (Pologruto et al., 2003). All imaging was done in lobule III of the cerebellar vermis. Whole lobule images (Figure 1) were obtained with a Leica upright fluorescent microscope using GFP filter.

Image analysis was performed in ImageJ. Torpedo density was measured as number of torpedoes/number of Purkinje cells per acquisition when signal to noise was high. However, for very young slices and very old slices, we found that Calbindin labeling gave high background, and we thus counted the number of torpedoes/section instead. Torpedo length was measured as the distance from the beginning to the end of the swelling parallel to the axon. Torpedo width was measured perpendicular to the length, and close to the widest point of the swelling. Axial ratio was defined as the length/width of each torpedo. To compare the density of microglia around torpedoes and axons, a 25 μm × 25 μm box was centered around either Purkinje cell axonal torpedoes or Purkinje cell axons without neighboring torpedoes while viewing only the green (Purkinje cell label) channel. Switching to the red channel, the red-labeled microglia that were in the boxed region were counted. To strengthen our analytical power, we performed analysis in tandem for key findings, including developmental torpedo count. Two individuals analyzed data images independently, and results were found to be consistent across experimenter. Data was collected from equal number of acquisitions from at least three animals from a minimum of two litters (typically three) for each comparison.

Data are reported as Mean ± SEM or as percentage of totals for distributions. Comparisons were made using one-tailed ANOVA tests followed by Tukey’s HSD test for normal distributions in JMP software (SAS, Cary, NC, USA), or Student’s t-tests in Igor Pro (Wavemetrics, Portland, OR, USA). For non-normal distributions, we used Wilcoxon/Kruskal–Wallis multiple comparison tests followed by the Benjamini–Hochberg procedure using a false discovery rate (FDR) of 0.05 in JMP software (SAS, Cary, NC, USA) or a Mann–Whitney U test in Igor Pro. Multiple-comparison corrections were performed with Excel scripts. P-values of <0.05 were considered significant.

Focal axonal swellings, or torpedoes, have been observed on the axons of developing Purkinje cell from postnatal rats (Gravel et al., 1986), and have been reported in developing mice (Baurle and Grusser-Cornehls, 1994). To characterize developmental axonal torpedoes found in the developing mouse cerebellum on Purkinje cell axons, we imaged Purkinje cell axons from several postnatal ages (Figure 1A) from L7-tau-GFP mice that express a tau-GFP fusion protein in Purkinje cells, resulting in strong axonal labeling, as previously described (Sekirnjak et al., 2003; Watt et al., 2009). To determine how prevalent developmental torpedoes were, we counted their numbers normalized to Purkinje cell number at several ages from both the apex and bank of anterior Lobule III of the vermis. We observed a rapid and transient increase in large focal axonal swellings, or developmental torpedoes, which were largely absent before P9, peaked at P11, and had decreased significantly by P30 (Figures 1A,B; P5: 7 torpedoes counted/702 Purkinje cells total, or 1.1 ± 0.44% normalized to acquisition, see “Materials and Methods” for details; P7: 16 torpedoes/690 Purkinje cells or 2.4 ± 1.1%; P9: 140 torpedoes/736 Purkinje cells, or 20.0 ± 4.5%; P11: 273 torpedoes/674 Purkinje cells, or 39.7 ± 2.5%; P13: 140 torpedoes/633 Purkinje cells, or 22.4 ± 4.3%; P15: 145 torpedoes/766 Purkinje cells or 19.5 ± 3.2%; P30: 141 torpedoes/774 Purkinje cells, or 18.6 ± 2.6%). Developmental torpedoes could be observed along Purkinje cell axons in both the granule cell layer and white matter (Figures 1A,C); however, because of the bundling of the Purkinje cell axons in white matter, which makes individual axons and torpedoes difficult to resolve, we limited our analysis to Purkinje cell axons in the granule cell layer. Interestingly, although torpedoes were not observed in younger ages, Purkinje cell axons in the first postnatal week displayed multiple small varicosities, as previously reported (Baurle and Grusser-Cornehls, 1994), which were distinguishable from torpedoes based on their smaller size (compare Figure 1A, left, middle). We thus restricted our analysis to developmental torpedoes, which showed a similar time course to those reported in rat (Gravel et al., 1986).

We wondered whether the transient changes in Purkinje cell torpedoes we observed were due to the presence of the tau-GFP protein in our transgenic mice, since tau is important for axonal integrity, and overexpressing tau can be pathological in Alzheimer Disease (Duan et al., 2012; Krstic and Knuesel, 2013). To determine whether our results were influenced by overexpression of tau, we examined Purkinje cell axons from C57BL6/J mice and counted the number of developmental torpedoes at three different ages: P7, P11, and P30 (Figure 1E; P7: 0.84 ± 0.26 torpedoes/section; P11: 5.6 ± 1.2 torpedoes/section; P30: 0.29 ± 0.08 torpedoes/section). Broadly speaking, we observed a similar shape in the developmental profile of torpedoes from C57BL6/J mice as in L7-tau-GFP mice, with a peak of torpedoes at P11 (significantly different from P7 and P30; P < 0.0001). These values are lower than corresponding torpedo counts from L7-tau-GFP mice (P7: 1.1 ± 0.42 torpedoes/section; P11: 18.4 ± 1.7 torpedoes/section; P30: 8.8 ± 1.2 torpedoes/section, data not shown). However, we cannot directly compare these values since image quality varies for calbindin antibody labeling and endogenous GFP expression, and it is possible that calbindin does not label all torpedoes. To address this, we processed a subset of L7-tau-GFP sections for calbindin and counted the number of torpedoes that we observed. Remarkably, we detected only 46.4% of torpedoes with calbindin compared to the same sections examined in the GFP channel (N = 15 sections). This suggests that either only a subset of torpedoes are labeled with calbindin, and/or we detect fewer torpedoes with immunocytochemistry due to poorer image quality. Thus, our data indicates that L7-tau-GFP is a good model to study developmental torpedoes because more torpedoes can be detected than with standard immunocytochemistry. Although it is possible that the presence of tau increases torpedo numbers in developing axons, this likely accounts for a relatively small proportion of torpedoes at these ages.

Purkinje cell axon torpedoes have been observed in diseased (Louis et al., 2006, 2009, 2014) and aged cerebellum (Baurle and Grusser-Cornehls, 1994), where they have often been associated with Purkinje cell death (Louis et al., 2014). We wondered whether developmental torpedoes might also be related to Purkinje cell death, since developmental apoptosis of Purkinje cells is a normal part of cerebellar development (Light et al., 2002; Jankowski et al., 2009). To address this question, we measured the number of Purkinje cells in a sagittal slice from an entire lobule III at several ages (Figure 1C), and found that Purkinje cell numbers were elevated at both P5 and P7, but remained constant after P9 (P5: 59.8 ± 1.9 Purkinje cells/mm; P7: 57.2 ± 2.8 Purkinje cells/mm; P9: 45.0 ± 1.8 × 10^-2 Purkinje cells/mm; P11: 46.7 ± 2.1 Purkinje cells/mm; P13: 47.2 ± 1.8 Purkinje cells/mm; P15: 46.4 ± 1.7 Purkinje cells/mm; P30: 45.1 ± 2.7 Purkinje cells/mm; at least four whole-lobule images from four mice included at each age point). Our findings are in agreement with previous reports demonstrating that developmental Purkinje cell apoptosis peaks at P3 and is complete by P9 (Light et al., 2002; Jankowski et al., 2009). Our results argue that the majority of developmental torpedoes occur on axons of Purkinje cells that will persist into adulthood, since they appear after the peak of Purkinje cell apoptosis. However, the small fraction of torpedoes that appear at or before P9 might arise on axons of Purkinje cells that will undergo developmental apoptosis.

Purkinje cell axons extend collaterals that are extensively pruned during the second week of postnatal development (Gianola et al., 2003), and we wondered whether developmental torpedoes are associated with axonal collateral branch points. Collateral-associated torpedoes are structurally distinct, with a triangular rather than oval shape (Figure 2A). To address this, we counted how often Purkinje cell torpedoes were associated with an axon collateral (Figures 2A,B), and found that only a small fraction of torpedoes are located at collateral branch points (Figures 2A,B). Disease-related torpedoes have been characterized morphologically using the axial ratio measurement: length/width of torpedoes (Baurle and Grusser-Cornehls, 1994). We measured the length and width of torpedoes throughout postnatal mouse development (Figure 2C), and calculated their axial ratio (Figure 2D). We found that both the length and width of torpedoes increased with age (Figure 2C) while the axial ratio decreases from P15 onward (Figure 2D), suggesting that the torpedoes that persist after P11 increase in width more rapidly than they increase in length. Consistent with their different shape, the axial ratio of the collateral-branch-point-associated triangular torpedoes is significantly lower than non-collateral associated oval torpedoes (Figure 2E). Thus, developmental axonal torpedoes are typically not associated with axonal collaterals, and are structurally dynamic across development.

Could Purkinje cell torpedoes be associated with sites of axonal remodeling without being directly associated with a branch point? For example, could the swelling of a torpedo mark where a recent axonal collateral has been pruned? To address this, we examined the localization of activated microglia in the granule cell layer, which have been associated with axonal refinement and pruning during development (Pont-Lezica et al., 2011), and can be recruited to axonal swellings (di Penta et al., 2013; Kato et al., 2016). We looked at two ages: at P11, the peak of developmental torpedo density, and at P30 (Figure 3A), when torpedoes are larger but less numerous. We found that microglia density in the granule cell layer increased significantly over this period (Figure 3B; P11: 6.52 ± 0.37 cells/10⁶ μm³; P30: 7.71 ± 0.44 cells/10⁶ μm³; significantly different, P = 0.045), likely because microglia density throughout the cerebellum overall increases across development (Perez-Pouchoulen et al., 2015). To determine whether microglia were enriched around torpedoes, which might suggest that they are involved in axonal refinement, we measured the local density of microglia around torpedoes (Figure 3C) and compared this density to that around Purkinje cell axons that are not in the proximity of a torpedo (Figure 3C shows two examples for each). We found no significant enrichment of microglia around torpedoes compared to axons at either the peak age of torpedo enrichment, P11 (Figure 3D, left; P11 torpedo: 25/51 or 49.0% have 0 microglia in proximity, 20/51 or 39.2% are near 1 microglia, and 6/51 or 11.8% are near 2 microglia; P11 axon: 29/51, or 56.9% have 0 microglia in proximity, 21/51 or 41.2% are near 1 microglia, and 1/51 or 1.9% are near 2 microglia; not significantly different P = 0.84) or later when torpedoes are less prevalent but larger, P30 (Figure 3D, right; P30 torpedo: 11/45 or 24.4% have 0 microglia in proximity, 26/45 or 57.8% are near 1 microglia, and 8/45 or 17.8% are near 2 microglia; P30 axon: 14/48, or 56.9% have 0 microglia in proximity, 28/48 or 58.3% are near 1 microglia, and 6/48 or 12.5% are near 2 microglia; not significantly different P = 0.84). Taken together, our findings suggest that torpedoes are unlikely to be associated with axonal damage or refinement.

Some neurons, such as hippocampal dentate gyrus granule cells, have striking focal axonal swellings along their axons that are the morphological correlates of large presynaptic terminals (Chamberland et al., 2014). These large presynaptic terminals bear a resemblance to Purkinje cell axonal torpedoes. Although ultrastructure analysis of disease-related Purkinje cell axonal torpedoes reveals that they do not contain vesicles (Petito et al., 1973; Yagishita, 1978; Mann et al., 1980; Louis et al., 2009), suggesting that they are not presynaptic release sites, we wondered whether developmental Purkinje cell axonal torpedoes might be the presynaptic structure of a transient synapse that functions during postnatal development. Interestingly, Purkinje cells form transient synapses onto other Purkinje cells during postnatal development (Watt et al., 2009), and transient synapses are thought to be a common feature of developing brain circuits (van Welie et al., 2011), and can also be found in other cerebellar neurons (Trigo et al., 2010). To examine whether developmental torpedoes were axon terminals containing vesicles, we used immunolabeling for vesicular GABA transporter (VGAT; Figure 4A) (Watt et al., 2009), and quantified the number of torpedoes that colocalized with this presynaptic marker (Figures 4A,B). We found that the vast majority (>80%) of Purkinje cell torpedoes were negative for VGAT staining, both at P11 and P30 (Figure 4B; P11: 54/62, or 87.1% of torpedoes were VGAT negative; P30: 124/144, or 86.1% of torpedoes were VGAT negative, not significantly different, P = 0.62), which suggests that like disease-related torpedoes, developmental torpedoes do not appear to be inhibitory presynaptic terminals.

An understanding of Purkinje cell developmental torpedo function is currently unknown. Since the myelination of Purkinje cell axons is underway during the second postnatal week of development (Gianola et al., 2003), at the age when developmental torpedoes are prevalent, we wondered whether developmental torpedoes occur on myelinated or unmyelinated axons. To examine this, we used MBP to identify myelin, and examined whether torpedoes were surrounded by myelin (MBP+) or not (MBP-; Figures 5A,B), at P11, the peak of torpedo density, and at P30, when torpedo numbers are decreased and when Purkinje cell axons are expected to be fully myelinated (Gianola et al., 2003). We found that the vast majority of torpedoes at P11 (102/122, or 85%) and at P30 (78/81, or 96%) were myelinated (Figure 5B), with no significant difference in the percentage of non-myelinated axons across ages (P = 0.13). While we considered any local enrichment of MBP to be positive for myelin, we typically observed torpedoes to have good myelin coverage (Figure 5C shows three successive z-stack images). This suggests that action potential conductance may not be hindered in developing axons by the presence of a developmental axonal torpedo, although it is possible that we missed subtle changes in myelin that could affect propagation. We found no significant differences in the length (P = 0.41), width (P = 0.60), or axial ratio (P = 0.62) of myelinated or unmyelinated torpedoes at P11 (there were too few non-myelinated torpedoes to compare at P30), suggesting that these torpedoes are not distinct subpopulations. Rather, whether a developmental torpedo is myelinated or not is likely to depend on whether its parent axon is myelinated or not, and not on the torpedo. The high percentage of developmental torpedoes that are myelinated is in contrast to what is observed for disease-related torpedoes, which although occurring on both myelinated or non-myelinated axons, (Yagishita, 1978; Takeuchi et al., 1995; Louis et al., 2009), disease-related torpedoes are typically more common on non-myelinated axons (Louis et al., 2009).

One striking feature of disease-related torpedoes that has been observed in several diseases is that they are enriched in disorganized neurofilament (Petito et al., 1973; Yagishita, 1978; Mann et al., 1980; Louis et al., 2009), which may be associated with alterations in axonal transport. We wondered whether developmental torpedoes also contained neurofilament. To address this, we labeled neurofilament 200 kD (NF) and counted the torpedoes that were positive (NF+) and negative (NF-) at both P11 and P30 (Figure 6A). At the peak of Purkinje cell torpedo density at P11, although the majority of Purkinje cell axon torpedoes were NF+ (51/80, or 64% of torpedoes), there was a significant fraction that were not (29/80, or 36%; Figure 6B). Sometimes neurofilament positive and negative torpedoes were next to each other on the same axon (Figure 6A, inset), suggesting that these two types of torpedoes may have distinct functions that are locally determined. Consistent with this, NF+ torpedoes were significantly larger than NF- torpedoes (NF+ length: 7.6 ± 0.3 μm; NF- length: 6.1 ± 0.2 μm; significantly different, P = 0.0001; NF+ width: 3.4 ± 0.2 μm; NF- width 2.7 ± 0.1 μm; significantly different, P = 0.0007). Later in development at P30, the proportion of NF+ torpedoes had increased slightly although not significantly from P11 (NF+: 85/109, or 78% of torpedoes; NF-: 24/109, or 22% of torpedoes; not significantly different from P11, P = 0.09). Similar to what was observed at P11, NF+ torpedoes were also larger than NF- at P30 (NF+ length: 8.9 ± 0.3 μm; NF- length: 7.5 ± 0.6 μm; significantly different, P = 0.036; NF+ width: 4.7 ± 0.2 μm; NF- width: 3.5 ± 0.3 μm; significantly different, P = 0.003). These data raise the possibility that there may be functional subpopulations of developmental torpedoes, with the majority staining positive for neurofilament, and a significant minority that do not, in contrast to reports of disease-related torpedoes, where neurofilament accumulation in torpedoes is robust (Petito et al., 1973; Yagishita, 1978; Mann et al., 1980; Louis et al., 2009). An alternate explanation is that that the distinction between NF+ and NF- torpedoes is less distinct, with a continuum of neurofilament content in torpedoes ranging from low to high, rather than two separate populations.

Recent studies have identified changes in basket cell innervation of Purkinje cells that contain axonal torpedoes in disease states (Erickson-Davis et al., 2010; Kuo et al., 2013). Since basket cell innervation of Purkinje cells is maturing during postnatal development (Ichikawa et al., 2011), we thought that such changes might be associated with developmental torpedoes as well. We identified Purkinje cell neurons that had axonal torpedoes on their axon and examined the extent of basket cell innervation of the parent Purkinje cell [Figure 7A shows examples of low (score = 1), medium (score = 3), and high (score = 5) degree of innervation], and compared this with similar neighboring neurons that did not have a torpedo on their proximal axon, using neurofilament labeling to measure basket cell innervation density, since basket cell processes contain neurofilament (Erickson-Davis et al., 2010). In contrast to what has been observed in essential tremor-related torpedoes (Erickson-Davis et al., 2010; Kuo et al., 2013), we found no differences between axonal-torpedo-containing Purkinje cells and those without axonal torpedoes, with both groups being similarly innervated by basket cell processes (Figures 7B,C; average basket cell innervation density, scored in arbitrary units from 1 to 5, of Purkinje cells with torpedoes: 2.23 ± 0.26; N = 13; average basket cell innervation of Purkinje cells without torpedoes: 2.23 ± 0.28; N = 13; not significantly different, P = 0.92). Aberrant basket cell innervation of Purkinje cells harboring disease-related axonal torpedoes suggests that circuit rewiring is associated with those neurons. In contrast, our results suggest that developmental torpedoes are not associated with differences in rewiring, but rather appear normally during normal development. Thus, although morphologically similar, there appears to be differences between developmental and disease-related torpedoes.

To understand whether a relationship exists between developmental and disease-related torpedoes, we examined Purkinje cell axons at several time points in a mouse model of SCA6, since torpedoes have been seen in postmortem tissue from SCA6 patients (Sasaki et al., 1998; Yang et al., 2000). We used SCA6^84Q/84Q mice harboring an expanded polyglutamine (poly-Q) repeat which have been shown to display ataxic symptoms at 7 months old (Watase et al., 2008; Jayabal et al., 2015). We have previously shown that there is no Purkinje cell loss at 7 months when disease symptoms are first observed, although Purkinje cell loss is detectable at 2 years (Jayabal et al., 2015). We first wondered whether disease-related torpedoes would be observed in these mice at 2 years. To address this, we labeled Purkinje cells with calbindin (Figure 8A) and counted the number of torpedoes we observed on Purkinje cell axons in both SCA6^84Q/84Q and litter-matched wild-type (WT) controls. We found that torpedo number was greatly increased at 2-years in SCA6^84Q/84Q compared to WT mice as predicted from human postmortem studies (2 year WT: 1.09 ± 0.27 torpedoes/section; 2 year SCA6^84Q/84Q: 4.48 ± 0.68 torpedoes/section; significantly different, P < 0.0001) (Sasaki et al., 1998; Yang et al., 2000). Interestingly, we have recently shown that transient functional and morphological changes occur during postnatal development in the developing SCA6^84Q/84Q cerebellum at the age when we observe developmental torpedoes (Jayabal et al., 2016b). This led us to wonder whether developmental torpedoes might be altered in these mice as well, since this might suggest that they are related to later pathophysiology. We found, however, that developmental torpedo density was normal in P11 SCA6^84Q/84Q mice compared to WT, suggesting that they are not directly related to later pathophysiology (P11 WT: 2.71 ± 0.90 torpedoes/section; P11 SCA6^84Q/84Q: 2.61 ± 0.54 torpedoes/section; not significantly different, P = 0.92; Figure 8B, left). We also measured the density of Purkinje cell torpedoes in SCA6^84Q/84Q and litter-matched mice at 7 months, when cerebellar-related motor deficits are observed without detectable Purkinje cell loss (Jayabal et al., 2015, 2016a), and found that torpedo density was low at 7 months and similar in both WT and SCA6^84Q/84Q mice (7 month WT: 0.57 ± 0.20 torpedoes/section; 7 month SCA6^84Q/84Q: 0.42 ± 0.13 torpedoes/section; not significantly different; P = 0.53; Figure 8B, middle). Note that we observe higher torpedo numbers in P11 mice than 7-month-old mice in both WT and SCA6^84Q/84Q mice, which is consistent with our observations of a transient developmental peak of torpedoes at P11 (Figure 1). These data suggest that developmental torpedoes are unlikely to be linked to later disease-related torpedoes and do not by themselves contribute to later pathogenesis. They furthermore suggest that disease-related torpedoes are not an early symptom of SCA6, but rather are observed later during disease progression, at the same time as Purkinje cell loss.