Cold temperature improves mobility and survival in Drosophila models of autosomal-dominant hereditary spastic paraplegia (AD-HSP)

Hereditary spastic paraplegias (HSPs) are a group of neurodegenerative disorders marked by lower-limb spasticity (stiffness) and weakness, leading to progressive difficulty walking (Fink, 2013; http://www.sp-foundation.org/). Supportive treatments exist, but are of mixed efficacy and do not restore mobility. The most common form, pure autosomal dominant HSP (AD-HSP), accounts for 70–80% of HSP-afflicted families. AD-HSP pathology is characterized primarily by degeneration of the longest descending axons of the central nervous system (CNS). These originate from the upper motor neurons in the cortex and terminate in the lumbar spine, innervating the α1 motor neurons that control leg movement. The relative specificity of the affected neuronal population is striking, but still not understood.

Over 50% of cases of pure AD-HSP are caused by mutations in spastin (Hazan et al., 1999), which encodes one of a small family of microtubule-severing proteins – hexameric ATPases that disassemble microtubules along their length (Roll-Mecak and McNally, 2010; Sharp and Ross, 2012). Although it is not yet clear why Spastin is important in neurons, one thought is that its severing activity shortens microtubules for efficient transport into axons (Errico et al., 2002; Yu et al., 2008). Other models suggest that Spastin is required for net microtubule loss (Trotta et al., 2004) or growth (Sherwood et al., 2004) at motor neuron synapses, axon guidance (Wood et al., 2006; Butler et al., 2010) and axon transport (Kasher et al., 2009; Fassier et al., 2013). Most recently, data has emerged supporting roles for Spastin in membrane regulation, promoting tubular endoplasmic reticulum (Park et al., 2010) and endosomal tubule formation (Allison et al., 2013). Much progress has been made, but Spastin’s relevant functions, regulatory pathways and the specific mechanisms by which its mutations lead to axonal degeneration remain unclear.

Drosophila melanogaster is an effective model system for study of a wide variety of neurodegenerative diseases because of its high conservation of neuronal gene function with humans, short generation time, well-characterized features and the availability of a wide range of genetic and experimental tools (Bilen and Bonini, 2005). In regards to providing a model for AD-HSP, Drosophila Spastin, like its vertebrate orthologs, severs purified microtubules and those in Drosophila S2 cells (Roll-Mecak and Vale, 2005). Knocking down fly Spastin using a RNA-interference (RNAi) transgene (Trotta et al., 2004) or deletion of the endogenous gene (Sherwood et al., 2004) both cause synaptic defects at the Drosophila larval neuromuscular junction (NMJ), supporting a role for spastin in regulating synaptic morphology and function. Orthologs of several other HSP causative genes studied in Drosophila also exhibit progressive neurodegeneration, supporting the relevance of flies in providing insights into mechanisms underlying this disease (Wang and O’Kane, 2008; Ozdowski et al., 2014).

The spastin gene is completely deleted in the spastin^5.75 Drosophila model of AD-HSP (Sherwood et al., 2004). Spastin^5.75 larvae are homozygous-viable and have no obvious behavioral defects, but electrophysiological analysis at the NMJ reveals that synaptic strength is reduced. Morphological analysis reveals smaller, more numerous synaptic boutons that are often arranged in bunches and contain only sparse microtubules. Adult spastin^5.75 flies rarely eclose (emerge from their pupal cases) and those that do have severely reduced lifespans, neither fly nor jump, walk and climb only poorly and, even when still, hold their legs unsteadily. These phenotypes are equivalently rescued by low level, neuron-specific expression of Drosophila or human wild-type spastin transgenes, indicating that Spastin is predominantly required in neurons, and functions similarly in both organisms (Sherwood et al., 2004; Du et al., 2010).

Surprisingly, spastin^5.75 eclosion improves dramatically when animals are raised at a reduced temperature of 18°C, compared with the typical rearing temperature of 24–25°C. Maintaining flies at 18°C is a standard fly husbandry technique that slows development roughly twofold, but is not associated with systematic loss of phenotype or enhancement of viability and can in fact adversely affect stock health. Furthermore, although temperature-sensitive mutations are common, these typically arise from differences in mutant versus wild-type protein folding, which is impossible in spastin^5.75 animals given that the gene is completely deleted. This suggests that the cold-rescuing effect is specific to Spastin dysfunction, and could therefore provide a novel therapeutic approach to AD-HSP.

To quantify the effects of cooling, we determined eclosion rate, climb rate and lifespan for spastin null animals reared at 18°C, and compared them with values for untreated (non-temperature-shifted) mutant animals raised at 24°C, as well as with control white-CantonS (WCS) animals raised in parallel conditions. We looked at whether the temporal parameters for the cold effect made sense in the context of when spastin is required. We repeated these experiments using transgenic strains that express human spastin genes, to address whether hypothermic rescue held true for flies with partial human spastin function, which are more representative of the human disease. Finally, we tested the hypothesis that cold functionally substitutes for Spastin’s severing activity through the promotion of microtubule disassembly, and discovered instead that the alleviatory effects of cooling also extend to other mutations causing synaptic dysfunction. Mild hypothermia is commonly employed for neuroprotective purposes during treatment of stroke, cardiac arrest and other ischemic trauma; our data indicate that cooling might also be beneficial in AD-HSP and other neurodegenerative contexts.

Most Drosophila lacking spastin survive into metamorphosis but fail to emerge from the pupal case. However, spastin^5.75 nulls reared at 18°C rather than 24°C were considerably more successful at reaching adulthood. We quantified this cold-induced alleviation of pupal lethality and tested for a developmental period(s) during which it is effective. Cooling could be required throughout development, or alternatively, be necessary only at a particular stage. The former would support a broad effect, such as general metabolism or developmental rate. The latter, if coinciding with temporal requirements for Spastin, would support a mechanism relevant to the loss of Spastin function in the nervous system.

White-CantonS (WCS) controls and spastin^5.75 mutant Drosophila were raised in parallel at 24°C (‘untreated’) or moved to 18°C during either their larval or pupal stage of development, and differences in eclosion rates determined for each condition (Fig. 1A,B). Compared with the 86% eclosion rate of WCS flies, only 18% of homozygous spastin^5.75 flies maintained at 24°C eclosed (Table 1). Cold treatment specifically during the larval stage resulted in slightly, but not significantly, higher mutant eclosion (Table 1; Fig. 1B). Only pupal stage cold treatment produced a significant increase to levels that, although still well below wild type, were 70% greater than those for untreated mutants (Fig. 1B). This effect was specific to the spastin mutation, as cold did not affect WCS eclosion.

We next looked at whether cold also improves the dramatically weakened mobility of spastin mutant flies, by measuring adult climbing rates (Table 2; Fig. 2A). All mutants that eclosed following larval stage cold treatment did not climb, dying shortly after eclosion. However, spastin^5.75 flies cold-treated as pupae climbed 46% faster than mutants maintained at 24°C, and when cold-treated as adults, climbed 85% faster. By contrast, WCS climbing was unaffected by pupal and adult cold treatment, and also impaired by larval-stage treatment (Fig. 2A).

Besides compromised mobility, spastin^5.75 flies are short-lived compared with wild type (Sherwood et al., 2004). Whereas WCS flies at 24°C lived 3–4 weeks on average, mutants survived only 1 week (Fig. 2B,C, Table 3). As in the climb rate experiments, larval cold treatment was deleterious to mutant fly survival. Adult-stage cold treatment extended the average lifespan 2.5-fold but this was not different from WCS flies, which lived 2.3-fold longer when cold-treated as adults (Fig. 2B,C). However, whereas pupal cold treatment left WCS flies unaffected, spastin^5.75 flies lived twice as long as untreated mutants (Fig. 2B). The survival curve for pupal cold-treated spastin^5.75 flies (dark blue solid line) showed a clear right shift of the entire population relative to untreated mutant flies (red long dashed line; Fig. 2C). Notably, about 10% of treated mutants had considerably extended lifetimes, which were threefold longer than untreated mutants and comparable to the maximal lifespan of control WCS flies.

These measurements indicate that pupal development is an effective period for cold to enhance eclosion, mobility and lifespan of adult spastin^5.75 mutants, although it has little effect on wild-type flies. Additionally, cooling significantly improves mobility even when applied in spastin^5.75 adults, when the nervous system has already matured. Larval cooling, by contrast, exacerbates spastin mutant phenotypes in adults. Taken together, these data indicate that cooling-induced rescue of adult phenotypes arises from the effects of mild hypothermia on the adult nervous system, which is assembled de novo during pupal stages as part of metamorphosis.

We sought further evidence for the specificity of cold rescue to loss of Spastin function by performing the previous experiments in a Drosophila AD-HSP model that recapitulates spastin gene dosage and allelic severity of the human disease (Du et al., 2010). We used two fly strains designed to genocopy humans at the spastin locus. In the first, denoted H^WT,H^WT, flies lacking endogenous spastin (i.e. spastin^5.75 flies) instead express wild-type human Spastin, encoded by two copies of a human spastin transgene, in all neurons. The second genotype, H^L44,H^R388, mimics the most severe form of spastin-mediated AD-HSP in humans (Svenson et al., 2004). Instead of wild-type human spastin, these flies carry one copy each of transgenes encoding the S44L or K388R human Spastin mutations in the fly null background. Pan-neuronal H^WT,H^WT expression rescues spastin-null phenotypes equally as well as wild-type Drosophila spastin transgenes, demonstrating functional conservation between human and fly Spastin. Flies neuronally expressing H^L44,H^R388, however, are severely compromised in eclosion, mobility and survival. Thus, the H^WT,H^WT flies provided a genetic control group (because, like WCS, these express wild-type spastin) and the H^L44,H^R388 flies modeled severe AD-HSP.

As further evidence of the therapeutic effect of cold, H^L44,H^R388 flies cold-treated as pupae eclosed nearly twice as often as those kept at 24°C, whereas H^WT,H^WT eclosion was unaffected by cold (Fig. 3A; Table 4). H^L44,H^R388 flies also climbed 47% and 60% faster, respectively, if cold-treated as pupae or adults (Fig. 3B, Table 5), but cold treatment of H^WT,H^WT at these stages yielded adults that climbed only half as quickly. Although spastin transgene expression might have been reduced at 18°C, our previous studies showed that a single copy of H^WT rescues the null mutants as effectively as two copies, making it unlikely that temperature-based alterations in expression could account for the alleviatory effects seen here (Du et al., 2010). Overall, the parallels between these results and those for spastin^5.75 strongly support the specificity of cold alleviation to defects caused by Spastin dysfunction in the nervous system.

We next looked at whether cooling rescues the cell biology of neurons affected in spastin mutants. We were unable to address this issue in adults because the cellular loci of adult defects in spastin mutants remain unknown. We thus investigated spastin defects at the larval NMJ, which are well characterized (Sherwood et al., 2004). Although previous experiments showed that synaptic transmission in mutants is reduced relative to wild type at both normal and low temperatures, we examined whether other cellular defects at this stage could be alleviated by cold. At room temperature, spastin mutants have a distinctive NMJ morphology, with greater numbers of smaller synaptic boutons, which are sometimes arrayed in grape-like ‘bunches’ and contain only sparse microtubules (Fig. 4A) (Sherwood et al., 2004; Du et al., 2010; Ozdowski et al., 2011). These bunched terminal arrangements are not observed at wild-type NMJs, which consist of large, round and linearly arrayed boutons penetrated by a clear microtubule bundle. Comparison of synaptic terminals in 18°C-versus 24°C-reared larvae showed that cooling mitigated the spastin mutant morphology (Fig. 4A1,A2), both with respect to the total number of boutons per muscle and the number of terminal boutons (a measure of synaptic arbor branching). WCS synapse morphology was unaffected. However, we did not detect a change in stable microtubule penetration into terminal boutons, measured by the 22C10 antibody against the Drosophila MAP1b ortholog, Futsch (Fig. 4A3). Larval cooling thus partially rescues the cellular defects resulting from spastin loss, mitigating bouton morphology but not stable microtubule distribution or synaptic function.

We have shown that cooling elicits effects specifically on spastin mutant phenotypes in comparison with controls. To help in understanding whether cold exerts a general rescuing effect on defective synaptic growth or whether these effects are also specific to mutations in spastin, we examined animals with mutations in flower (fwe), a putative Ca²⁺ channel that regulates synaptic vesicle endocytosis and has no known link to microtubule regulation (Yao et al., 2009). Similar to spastin mutants, however, fwe^DB25/DB56 larval NMJs have smaller and more numerous boutons, often arrayed in bunches, and are compromised in synaptic transmission. Rearing fwe^DB25/DB56 larvae at 18°C significantly reduced synaptic bouton number and arrangement to resemble wild-type morphologies (Fig. 4B), suggesting that the effects of cooling are not limited to mutations in spastin function.

We also examined adult behavioral phenotypes in mutants of a gene closely related to spastin, kat-60L1. Like Spastin, Kat-60L1 severs microtubules and is required during larval and pupal neuronal development, although the precise roles of each protein are distinct (Lee et al., 2009; Stewart et al., 2012). We examined the effect of cold on adult fly mobility in kat-60L1^PBac, a partial loss of function line that acts as a strong hypomorph, and in kat-60L1^BE6, a transcriptional null (Stewart et al., 2012). Both alleles are homozygous-viable and survive more successfully than spastin^5.75 null mutants, but also climb more slowly than WCS flies. Pupal cold treatment increased the average climb rates of both mutants over 35% relative to untreated mutant controls (Fig. 4C), to rates comparable to WCS controls (compare with Table 2).

We have demonstrated that cold temperature alleviates reduced mobility and survival caused by loss of Spastin function in Drosophila. This is the case for flies lacking endogenous spastin, as well as those expressing pathogenic human Spastin. Cold treatment during the pupal stage of development was sufficient to enhance the eclosion rate, climbing ability and lifespan of spastin mutant adults. Furthermore, cold administered only after pupal development, to fully developed adults, also improved mutant mobility. The timing of these two effective periods is consistent with the idea that cold alleviates spastin mutant phenotypes by acting on the developing adult nervous system during pupal metamorphosis, but is also potent after the nervous system has matured. This is extremely promising from a clinical viewpoint, suggesting that the therapeutic window in AD-HSP includes both developing and mature nervous systems.

Although wild-type levels of mobility and survival were not often achieved, the temperature shift to 18°C conferred considerable improvement. Some cold-treated flies were able to jump and even fly briefly, behaviors not observed in untreated mutants. Cooling can match or exceed the efficacy of rescue by the microtubule destabilizing drug vinblastine, which has been proposed as a therapeutic approach for AD-HSP (Orso et al., 2005). Orso and colleagues showed that vinblastine doubled the ~12% eclosion rate of spastin^5.75 null mutants; in our hands the drug was ineffective for null and H^L44,H^R388 eclosion, but improved eclosion by 65% for H^WT,H^R388, which is a more common, representative AD-HSP genotype associated with milder pathogenesis (Du et al., 2010; Fang Du and N.T.S., unpublished results). In comparison, pupal cooling of spastin^5.75 null mutants increased eclosion by 70%.

Importantly, cooling during the pupal and adult stages did not affect eclosion or motor behavior in wild-type flies. This suggests that cooling not only compensates for defects in neuronal function caused by lack of Spastin (or other mutations), but is also innocuous to properly functioning neurons. Although cooling administered at the larval stage was ultimately deleterious to both control and spastin mutant adults, mutant larval synapses were effectively restored to wild-type morphologies. This suggests that cold was beneficial for some spastin-mediated defects at this stage, but also had nonspecific, toxic effects on a cell population required later, in adults.

What is the mechanism(s) underlying the rescuing effect of cold? Our demonstration that cold alleviates not just spastin mutant phenotypes, but also mutant phenotypes in fwe and kat-60L1, indicates that that rescuing effects of cold on nervous system function might be quite broad. All three genes are important in synapse formation, although kat-60L1 has been shown to act post-rather than pre-synaptically at larval and pupal stages. Reduced temperature could thus be generally beneficial to synaptic dysfunction, perhaps by reducing activity or metabolic load. Alternatively, fwe, spastin and kat-60L1 might share a common pathway component(s), as yet undiscovered, that is directly affected by cold. For example, cold itself is well known to destabilize microtubules, particularly at temperatures below 20°C (Delphin et al., 2012), and is often used in experiments to depolymerize microtubules (Baas et al., 1994; Cottam et al., 2006). Cold could thus substitute directly for the microtubule-severing function of Spastin by promoting microtubule destabilization. Cold-mediated rescue of Kat-60L1 mutants supported this idea; however, we did not observe obvious differences in stable microtubule distribution at cold-treated spastin^5.75 synapses or in Drosophila S2R+ cells (data not shown), and fwe mutants, which have not been implicated in microtubule dysregulation, were also rescued by cooling.

In humans, cooling has been shown to be generally neuroprotective, and mild or moderate therapeutic hypothermia (e.g. 33–35°C) has long had clinical applications, including reducing neurological injury in patients following cardiac arrest, traumatic brain injury, epilepsy and stroke (Hartemink et al., 2004; Yenari and Han, 2012). Furthermore, Yang and colleagues found that exposure to even near-freezing temperatures results in minimal neuropathology in rat and cat neocortex and hippocampus (Yang et al., 2006). Although commonly administered in situations involving acute brain injury, the mechanism by which cooling confers neuroprotection or therapeutic improvement is unknown, multifactorial and context-dependent (Choi et al., 2012; Yenari and Han, 2012).

It will be important to characterize the in vivo effects of cold in mouse models of AD-HSP (Kasher et al., 2009; Fassier et al., 2013). The specificity of the effect of cold on mutant and not wild-type animals in our experiments, together with the spatially localized neurodegeneration in AD-HSP, suggest that moderate hypothermia could be applied in a highly targeted manner in this disease context, with minimal negative effects. Future studies should furthermore elucidate the underlying cellular mechanisms and potentially broader applications of cold in alleviating neuronal dysfunction in neurodegeneration. Because Drosophila are ectothermic, with body temperatures that vary with their environment, they provide a straightforward system in which the cell biological effects of temperature change can be studied in vivo. Finally, our data highlight the potential sensitivity of neuronal phenotypes to variations in temperature and thus its importance as a consideration in studies of neuronal function, neurodegeneration and behavior.

Flies were reared on standard cornmeal/agar/molasses medium at either 24–25°C (untreated) or 18°C (cold treatment). Genetic controls were white-CantonS (WCS), the CantonS wild-type strain backcrossed to white ten times (gift of Anne Simon, Western University, Ontario, Canada) (Sherwood et al., 2004). Homozygous spastin null (spastin^5.75/spastin^5.75) animals were obtained by picking non-Tubby larvae from spastin^5.75/TM6B stocks. Transgenic line H^L44,H^R388 carried one copy each of transgenes encoding S44L and K388R human Spastin mutations, expressed via the inducible geneswitch elav-GAL4 driver (Osterwalder et al., 2001), in the spastin^5.75 null background. H^WT,H^WT flies had two copies of wild-type human Spastin in the same background (Du et al., 2010). Kat-60L1 mutants were generated as described previously (Stewart et al., 2011). Trans-heterozygous flower^DB25/DB56 larvae were maintained at 25°C or 18°C after collecting non-green progeny at ~stage L2, of fwe^DB25/TM3,Ser,twi-GAL4,UAS-GFP × fwe^DB56/TM3,Ser,twi-GAL4,UAS-GFP adults allowed to lay on yeasted grape juice plates.

Third-instar larvae of the appropriate genotype were picked with a cotton swab and placed in a clean vial. Eclosion rate was calculated as the number of adult flies from each vial divided by the number of larvae originally placed into that vial. After eclosing, each adult fly was moved to an individual food vial.

Adult mobility was assessed via a climb rate test: single flies were transferred to an empty vial, tapped to the bottom, and the rates at which they climbed up to an 8-cm mark measured. This was repeated three times per fly and the results averaged. Flies were then transferred back into their individual food vial and replaced at either 24°C or 18°C depending on experimental group. All tests were done at 24°C; adults maintained at 18°C were first equilibrated for 30 minutes at 24°C prior to being transferred to the empty vial for testing. Mobility was measured using flies aged 0–27 days. The average age of flies tested was not correlated with the average climb rates.

Adults were transferred to new food vials every 1–2 weeks to maintain healthy living conditions. Lifespan was the number of days from eclosion until death.

The Student’s t-test was used to measure statistical significance. Graphs are presented as percentage changes due to the large differences between wild-type and spastin mutant controls; raw values (mean ± s.e.m.) and number (n) of experiments are listed in the tables and figures.

Third-instar WCS and homozygous spastin^5.75 or trans-heterozygous flower^DB25/DB56 larvae were filleted, dissected and immunostained using standard methods (e.g. Ozdowski et al., 2011). Briefly, larvae were dissected in room temperature PBS and fixed for 30 minutes in 4% paraformaldehyde, immunostained at 4°C overnight using the neuronal membrane marker rabbit anti-HRP (1:100; Jackson ImmunoResearch, PA, 323-005-021) alone or with mAb 22C10 to label microtubules (mouse anti-Futsch, 1:50; Developmental Studies Hybridoma Bank, University of Iowa, Iowa City, IA). Secondary antibodies (Alexa Fluor 488 goat anti-rabbit A-11070 and Alexa Fluor 568 goat anti-mouse A-11031; 1:400; Life Technologies, Grand Island, NY) were incubated for 2–3 hours at room temperature. Fillets were mounted in H-1000 (Vector Laboratories, Burlingame, CA) and z-series images of muscle 4 synapses from larval segments 2–4 acquired on a Zeiss LSM 510 inverted confocal microscope using 63× 1.4 N.A. or 100× 1.2 N.A. PlanApo objectives (Oberkochen, Germany).

We are indebted to the Spastic Paraplegia Foundation and its donors, as well as Professors Doug Marchuk, Hunt Willard, Vann Bennett and Robin Wharton for their generous support of this work. We are also grateful to Drs Andrew Bellemer and Ken Honjo for experimental advice and assistance, Hugo Bellen and Chi-Kuang Yao for flower mutants and advice, and members of the Sherwood laboratory, particularly Fang Du and Emily Ozdowski, for their contributions and insights throughout this project.