The proSAAS chaperone provides neuroprotection and attenuates transsynaptic α–synuclein spread in rodent models of Parkinson’s disease

Parkinson’s disease is a devastating motor disorder involving the aberrant aggregation of the synaptic protein synuclein (aSyn) and degeneration of the nigrostriatal dopaminergic tract. We previously showed that proSAAS, a small secreted chaperone protein widely expressed in neurons within the brain, is able to block aSyn-induced dopaminergic cytotoxicity in primary nigral neuron cultures. We show here that coinjection of proSAAS-encoding lentivirus profoundly reduced the motor asymmetry caused by unilateral nigral AAV-mediated human aSyn overexpression. This positive functional outcome was accompanied by significant amelioration of the human aSyn-induced loss of both nigral tyrosine hydroxylase-positive cells and striatal tyrosine hydroxylase-positive terminals, demonstrating clear proSAAS-mediated protection of the nigro-striatal tract. ProSAAS overexpression also reduced the content of human aSyn protein in both the nigra and striatum and reduced the loss of tyrosine hydroxylase protein in both regions. Since proSAAS is a secreted protein, we tested the possibility that proSAAS is able to block the transsynaptic spread of aSyn from the periphery to the central nervous system, increasingly recognized as a potentially significant pathological mechanism. The number of human aSyn-positive neurites in the pons and caudal midbrain of mice following administration of human aSyn-encoding AAV into the vagus nerve was considerably reduced in mice coinjected with proSAAS-encoding AAV, supporting proSAAS-mediated blockade of transsynaptic aSyn transmission. We suggest that proSAAS may represent a promising target for therapeutic development in Parkinson’s disease. Significance This paper describes two independent avenues of research that both provide support for the in vivo neuroprotective function of this small chaperone protein. In the first approach, we show that proSAAS overexpression provides remarkably effective protection against dopaminergic neurotoxicity in a rat model of Parkinson’s disease. This conclusion is supported both by three independent assays of motor function as well as by quantitative analysis of surviving dopaminergic neurons in brain areas involved in the control of motor function. In the second line of research, we show that in mice, the spread of human synuclein across synapses can be blunted by proSAAS overexpression.


Significance
This paper describes two independent avenues of research that both provide support for the in vivo neuroprotective function of this small chaperone protein. In the first approach, we show that proSAAS overexpression provides remarkably effective protection against dopaminergic neurotoxicity in a rat model of Parkinson's disease. This conclusion is supported both by three independent assays of motor function as well as by quantitative analysis of surviving dopaminergic neurons in brain areas involved in the control of motor function. In the second line of research, we show that in mice, the spread of human synuclein across synapses can be blunted by proSAAS overexpression.

INTRODUCTION
Neurodegenerative diseases such as Alzheimer's disease (AD) and Parkinson's disease (PD) are increasingly considered to arise, at least in part, from the cumulative effects of abnormal aggregating proteins. Environmental toxins and other stressors accumulated during aging disrupt normal proteostatic mechanisms, resulting in the accumulation of misfolded and aggregated proteins. Cellular chaperones represent an important component of the neuronal proteostasis network that works both to combat initial protein misfolding and to direct misfolded proteins towards degradative pathways. Consistent with this idea, overexpression of chaperone proteins has been found to be cytoprotective against toxic protein aggregates in animal models of a variety of neurodegenerative diseases (reviewed in (1-3)). Specific examples include cytoprotection by overexpression of heat shock proteins in fly and rodent models of PD (4)(5)(6)(7); reviewed in (8). In addition, disaggregase-type chaperones, such as the yeast disaggregase Hsp104 (9) and the mammalian disaggregase Hsp110 (10) have been shown to block alpha-synuclein (aSyn) pathology and cytotoxicity. The beneficial effects of chaperone overexpression on brain proteostasis are associated with improvements in both motor and cognitive function (reviewed in (2,11,12)).
While cytoplasmic chaperones such as heat shock proteins have been frequently studied in the context of neurodegenerative disease, relatively little attention has been paid to the possible contribution of secreted brain chaperone proteins to brain proteostasis. Secreted chaperones, such as the ubiquitouslyexpressed glycoprotein clusterin (also known as ApoJ), are uniquely positioned to intercept and/or to blunt the toxicity of extracellular aggregates both prior to and following endocytic cell entry (reviewed in (13)). However, as yet there are few in vivo studies showing rescue of neurodegenerative defects by secreted chaperones other than clusterin (reviewed in (14)). Thus the question of whether other secreted chaperones play a role in blocking the cytotoxicity and/or the pathogenic transmission of aggregating proteins has not been adequately addressed.
Given the devastating consequence of proteostatic dysregulation in the CNS, and the apparent ability of pathogenic aggregating proteins to travel throughout the brain (reviewed in (15)) it would not be surprising to find that neurons express specific secreted chaperones to combat the formation and propagation of pathogenic aggregates. Of the few known secreted chaperones preferentially expressed in brain, the small protein known as proSAAS (encoded by the gene PCSK1N) exhibits many of the features expected of such a chaperone. ProSAAS is expressed in neurons rather than glia (16), where it is stored within secretory vesicles (17,18) and presumably released into the synaptic cleft, well positioned to function as an extracellular chaperone at the synapse. Our past work has shown that recombinant 5 proSAAS blocks the aggregation of Abeta 1-42 and aSyn at highly sub-stoichiometric ratios (19). More importantly, both proSAAS overexpression as well as exogenous application of recombinant proSAAS are able to reduce Abeta 1-42 andaSyn-mediated neurotoxicity, both in neuronal cell lines and in primary cultured neurons (19,20). However, whether proSAAS overexpression is able to functionally protect neurons in vivo has yet to be established. Here we demonstrate the profound behavioral and neuropathological protective effect of proSAAS overexpression in a virus-mediated nigral human aSyn overexpression rat model of Parkinson's disease. Additionally we show that proSAAS overexpression decreases the pathogenic transmission of human aSyn in a mouse model of aSyn brain spreading following injection of viral vectors expressing human aSyn into the vagus nerve.

ProSAAS overexpression protects rats from human aSyn-induced motor asymmetry
In order to test the protective potential of proSAAS overexpression on human aSyn-induced motor dysfunction, rats received unilateral nigral injections of either proSAAS-encoding or GFP-encoding lentivirus together with AAV-encoding human aSyn, a treatment known to produce degeneration of nigrostriatal neurons. Additional control groups of rats received either GFP-encoding lentivirus plus GFPencoding AAV, or PBS. Rats were subjected to a weekly battery of tests 2 weeks prior to, and 8 weeks following injections to assess motor asymmetry. Of note, this fully controlled experiment was preceded by an independent pilot experiment -using a different batch of proSAAS-expressing lentivirus -that produced similar evidence of both the behavioral-and neuroprotective effect of proSAAS overexpression. Figure 1 shows data from each of the 3 motor asymmetry tests conducted -cylinder test (Figure 1A), bracing test (Figure 1B), and forelimb placement test ( Figure 1C). Profound motor asymmetry in the human aSyn + GFP group is apparent in all 3 tests, the progressive nature of which is reflected in a significant group by time interaction using a 2-way repeated measures ANOVA: cylinder test F (27,  For the cylinder test, the preference for use of the forepaw ipsilateral to the injection evident in the aSyn + GFP group was significantly curtailed in the aSyn + proSAAS-treated animals (main effect of group F (3,34) = 13.52, p < 0.0001; post-hoc Tukey comparison aSyn + GFP vs. aSyn + proSAAS, p < 0.0001).
Similarly, for the bracing test, the reduced number of adjustment steps contralateral to the injected side 6 relative to the ipsilateral forepaw, apparent in the aSyn + GFP group, was largely absent in the aSyn + proSAAS-treated animals (main effect of group F (3,34) = 74.23, p < 0.0001; post-hoc Tukey comparison aSyn + GFP vs. aSyn + proSAAS, p < 0.0001). Finally, in the forelimb placement test, the reduction in number of forelimb placements contralateral to the injection relative to ipsilateral in aSyn + GFP-treated animals was significantly curtailed by co-expression of proSAAS (main effect of group F (3,34) = 40.98, p < 0.0001; Tukey comparison aSyn + GFP vs. aSyn + proSAAS, p < 0.0001). Taken together, these data provide clear evidence that overexpression of proSAAS provides substantial protection from aSyn-induced motor asymmetry.

ProSAAS overexpression attenuates human aSyn-induced nigrostriatal DA neuron degeneration
To determine if the functional protection provided by proSAAS overexpression was accompanied by sparing of the nigrostriatal tract from aSyn-induced damage we examined brain slices through the SNc and striatum immunostained for tyrosine hydroxylase (TH). Representative sections from each of the 4 groups for which behavioral data are presented above are shown in Figure 2A Quantitation of striatal terminal TH density was conducted at 3 rostro-caudal levels, designated anterior, middle and posterior, and is shown in Figure 2C. Statistical analysis of optical density by 2-way ANOVA revealed a main effect of group (F (3,34) = 57.81, p < 0.0001), a main effect of rostro-caudal segment (F (2,68) = 14.42, p < 0.0001), but no group x segment interaction (F (6,68) = 1.243, p = 0.296). Similar to nigral cell counts, striatal terminal density was significantly lower in human aSyn + GFP-treated animals compared to animals in the human aSyn + proSAAS group (post-hoc Tukey, p < 0.0001). Moreover, the protection offered by proSAAS was such that TH terminal density did not differ significantly from that of PBS controls (post-hoc Tukey, p > 0.05). Notably, however, there remained a significant difference from the GFP + GFP group, which itself was significantly higher than the other three groups (post-hoc Tukey, p < 0.0001 in each comparison). Further, this apparent increase in TH density was bilateral (Supplementary Figure 2).
We conclude that lentiviral proSAAS overexpression provides significant protection from human aSyninduced neuronal damage to the nigro-striatal DA tract.

aSyn phosphorylation in the SNc is reduced in proSAAS-treated animals
Colocalization of human aSyn and TH in surviving neurons in the SNc of animals treated with proSAAS was readily apparent ( Figure 2D). Importantly, however, aggregation-prone phosphorylated aSyn staining (31)(32)(33)(34)(35) was considerably reduced by proSAAS treatment. Figure 2E shows representative examples of extensive p129 aSyn staining in the SNc of animals treated with AAV-aSyn + lenti-GFP 8 weeks postinjection, and minimal staining in animals treated with AAV-aSyn + lenti-proSAAS. No such staining was apparent in sections from GFP/GFP-treated animals.

ProSAAS overexpression reduces virus-produced human aSyn protein levels in SNc and striatum and attenuates the reduction in TH protein 3 weeks post-treatment
Since one function of chaperones is to promote the degradation of toxic aggregating proteins, we sought evidence for reduced protein levels of human aSyn in the SNc and striatum of animals injected with proSAAS-and human aSyn-encoding viruses relative to those receiving aSyn-plus GFP-expressing viruses.
This was assessed by Western blotting 3 weeks after unilateral injection, a time of predicted maximal viral expression but before significant cell loss was expected. A GFP-GFP control group was included to demonstrate specificity, and examples of gels from each of the 3 groups are shown in Figure 3A.
Quantitation of blot intensities, normalized to -actin loading controls, are shown in Figure 3B. Interestingly, however, TH protein levels were lower in aSyn + GFP treated animals relative to GFP + GFP animals on both sides of the brain, perhaps reflecting a compensatory mechanism, or a significant contribution of down-regulation in crossed nigro-striatal neurons. This effect of aSyn contralateral to the site of injection was also attenuated by unilateral proSAAS co-administration such that considerable asymmetry in TH levels persisted.

ProSAAS expression attenuates the caudo-rostral transmission of human aSyn in mice
In order to determine whether proSAAS expression can block the transmission ofaSyn, we injected human aSyn-encoding AAV together with either proSAAS-or GFP-encoding AAV unilaterally into the vagus nerve of mice (Figure 4). This experimental strategy was designed to specifically overexpress human aSyn in the medulla oblongata (MO) and then to follow its temporal spreading into rostral regions of the brain. As Similar to the observations in the rat SNc and striatum (Figure 3), the intensity of aSyn immunoreactivity in the mouse MO was reduced in the proSAAS-co-expressing group (scores: +, n = 1; ++, n = 3; +++, n = 1) compared to the GFP-co-expressing group (scores: +++, n = 5). This reduction in the MO was accompanied by decreased numbers of human aSyn-positive-axons in the pons and caudal midbrain. Thus, transsynaptic spread from the medulla oblongata was significantly reduced in proSAAS-treated mice relative to their GFP-treated counterparts ( Figure 4C and D). No such diminishment was apparent in the rostral midbrain (rMB), but it should be noted that aSyn transmission had only minimally spread this far in controls. These data replicate a prior experiment and support the idea that proSAAS overexpression attenuates the trans-synaptic spread of aSyn through the brain.

DISCUSSION
The data presented here showing that proSAAS exerts strong protection from aSyn-induced nigro-striatal DA tract degeneration provide the first in vivo evidence of the potential utility of this chaperone as a 9 therapeutic target for PD. AAV-mediated overexpression of human wildtype aSyn unilaterally in the rat SNc produced a robust progressive motor asymmetry over an 8-week period accompanied by significant loss of TH-positive nigral cells and TH-positive striatal terminals measured post-mortem, consistent with previous reports describing similar viral-mediated aSyn models (reviewed in (36)). All three of these measures of aSyn toxicity were profoundly blunted by nigral co-injection of proSAAS-expressing lentivirus.
Further, the rostral spread of aSyn in the brain of mice following vagal injection of AAV-expressing human aSyn was similarly attenuated by co-vagal injection of AAV-proSAAS suggesting that therapeutic strategies targeting proSAAS may have potential for slowing progression of PD.
ProSAAS, a small neuronal and endocrine protein first identified in an unbiased peptidomics screen from brain extracts over 20 years ago (37), has now been identified by ten independent proteomics groups as a consistently lowered CSF biomarker in AD as well as in frontotemporal dementia (see meta-analysis in (38); see also (39,40)); one recent report also indicates lower proSAAS levels in PD CSF (39).
Immunoreactive proSAAS is associated with Lewy bodies in the brains of PD patients (19), with amyloid plaques in AD patients (20), and with tau tangles in various forms of dementia (41); reviewed in (14), suggesting brain sequestration of proSAAS during disease progression. In agreement with this idea, transcriptomics studies of AD patient tissues indicate increased brain proSAAS RNA expression during AD progression (42). Increased brain proSAAS is also seen in patients with cerebral amyloid angiopathy (43) as well as in proteomics studies of two animal models of neurodegenerative disease, equine serum sickness (44) and a rat model of Usher's syndrome (45). In rodents, hypothalamic levels of proSAAS increase following stressors including dehydration, hypoxia, and cold temperature (46)(47)(48), and recent work from our laboratory has shown that proSAAS levels are upregulated in Neuro2A cells during tunicamycin-induced endoplasmic reticulum stress (49). Collectively, these data, obtained from a variety of human disease and cell and animal models, support the idea that proSAAS is a stress-responsive protein that contributes to brain proteostasis during the progression of neurodegenerative disease.
aSyn, an abundant brain protein, is thought to participate in synaptic vesicle recycling and may have additional functions at the synapse (reviewed in (50)). Aggregation of aSyn into Lewy bodies, a key signature of Parkinson's disease, is believed to result in the loss of normal aSyn function as well as a gain of toxic function, leading to general synaptic dysfunction and ultimately neuronal cytotoxicity (51); reviewed in (52). The relative susceptibility of DA neurons to the toxic effect of aSyn overexpression and aggregation remains a subject of debate but may involve a role for DA or its metabolites in the aggregative process itself and/or a general elevated level of oxidative stress imparted by DA metabolism, which may be heightened when synaptic cycling and therefore vesicle packaging of DA is disrupted (53); see (54,55) for review. The precise cellular mechanism underlying the strong cytoprotective action of proSAAS observed in the current in vivo study and in our previous cellular investigation (19) is not yet clear, but most likely depends upon its demonstrated chaperone activity (reviewed in (14)). If proSAAS were acting simply as a secreted trophic factor to promote neuronal health (56), we might expect transsynaptic transmission of aSyn to be increased in the presence of increased proSAAS; instead, we found diminished transsynaptic spread of this protein.
As noted in the Introduction, proSAAS is one of only a few identified brain-specific secretory chaperones.
The protective effect of cytoplasmic heat shock protein overexpression in animal models of PD has been confirmed in multiple studies (reviewed in (14)). Effective reversal of cytotoxic, and in some cases, motor deficits in PD rodent models by cytoplasmic chaperone overexpression has been accomplished using Hsp70 (5, 6); Hsc70 (7); the Hsp70-interacting protein BAG1 (57); and the disaggregases Hsp104 and Hsp110 (9,10). Most recently, the ubiquitin ligase TRIM11 has been shown to possess aSyn chaperone activity and can rescue aSyn-mediated cell death in cell and animal models of PD (58). These chaperones are thought to exert their cytoprotective actions by directly blocking the intracellular formation of toxic aSyn aggregates, providing strong support for the idea that chaperone overexpression can restore normal proteostasis.
While it is clear that proSAAS is a highly potent anti-aggregant attenuating aSyn fibrillation (19), its sequestration to the secretory pathway limits its presence in the cytoplasm to endosomal uptake. While the cytoplasm clearly represents the predominant cellular location of aSyn aggregates, a small portion of cytoplasmic aSyn is known to be secreted (reviewed in (15)), indicating a possible extracellular location of interaction with proSAAS. Thus proSAAS may exert its effects on aSyn both within the synapse and following endosomal reuptake. These actions could take several forms. For example, secreted proSAAS, upon binding to extracellular aSyn, could block the uptake of aSyn into neurons (and/or other cells such as astrocytes). Indeed, our observation that proSAAS overexpression attenuates the spread of aSyn through the brain would be consistent with such a mechanism. ProSAAS may also exert cytoplasmic chaperone action following reuptake of proSAAS-bound toxic oligomers, promoting the direction of toxic aSyn aggregates to degradative pathways. In this case, functional interaction of proSAAS with aSyn could take place both in the extracellular space and in the cytoplasm. While the Western blot data presented above appear to support a facilitating effect of proSAAS on aSyn degradation, it is also possible that they instead reflect the blockade by proSAAS of the assembly of toxic oligomeric species, which again, could 11 be initiated either extracellularly or intracellularly. Finally, given the association of aSyn with synaptic vesicle cycling and the important protective role that vesicular packaging of DA is believed to play in dopaminergic neurons, proSAAS' neuroprotective efficacy may lie not only in its general ability to prevent the buildup of toxic aSyn assemblies, but also in its ability to maintain normal aSyn function at its primary site of action in the synaptic vesicle cycling machinery.
In conclusion, the data presented here indicate that local manipulation of brain proSAAS levels exerts a strong protective effect on nigrostriatal DA neurons against direct aSyn-mediated toxicity; and further, that increasing proSAAS levels within the vagus nerve limits the spread throughout the brain of similarly targeted aSyn overexpression. We suggest that these findings offer promise for the utility of PD therapies targeting proSAAS in halting the progression of PD, perhaps even following peripheral routes of intervention. A separate cohort of rats was injected with combinations of viruses as described above (4-5 per group) and euthanized 3 weeks later for Western analysis of nigral and striatal aSyn and tyrosine hydroxylase (TH) protein content as described below.

Neuroprotection study in rats
Motor asymmetry tests A battery of three tests known to be sensitive to dopamine (DA) depletion, without the need for extensive training, drug administration, or food or water deprivation, was employed, as previously described (21).
All animals were handled and habituated to the behavioral testing procedures for 4 days and baseline scores on each test were subsequently obtained once weekly for two weeks before virus injection. Weekly testing resumed one week after surgery and continued for eight weeks. Tests were scored manually without knowledge of group assignment. The cylinder test examines spontaneous forelimb utilization during vertical movements. Rats were placed in a clear Plexiglas cylinder (diameter = 20 cm, height = 30 cm) and video-recorded over a 5-min period. During rearing movements, placement of a single paw onto the cylinder wall was recorded as an ipsilateral or contralateral placement relative to the injected side.
When both paws were placed on the cylinder simultaneously or in rapid succession, both paws remaining on the cylinder, a score of 'both' was recorded. An asymmetry score was calculated as the difference between ipsilateral (right) and contralateral (left) forepaw placements as a percentage of total placements. The bracing test probes the ability of the rat to adjust postural balance in response to examiner-imposed lateral movement. The rat was held with only one limb (a forelimb) unrestrained to 13 support its weight and moved laterally at a constant speed across a flat surface for 1 meter in 5 seconds.
The number of adjusted steps was recorded twice for both leftward movements with the left forepaw and rightward movements with the right forepaw. Test scores were calculated by dividing the number of steps contralateral to the injected side (leftward movements) by the number of ipsilateral (rightward) steps.
The forelimb placement test examines sensorimotor/proprioceptive capacity. During the test, the rat's torso was held with both forelimbs hanging freely and moved slowly sideways toward a vertical flat surface until the vibrissae of one side touched the surface, evoking a characteristic placement of the adjacent forepaw on the surface. Tests of both ipsilateral and contralateral sides were repeated 10 times.
The score was calculated as the ratio between the number of successful placements of the contralateral and ipsilateral forelimbs.

Post-mortem histological assessment
Tissue preparation: Following the final motor asymmetry test, animals were anesthetized with pentobarbital (100 mg/kg) and transcardially perfused with PBS followed by 4% paraformaldehyde in PBS.
Brains were harvested, cryoprotected in 30% sucrose at 4˚C overnight, and stored at -80˚C. Brains were cut coronally on a cryotome (Leica CM1850) at 40 µm, and sections washed three times in 1x phosphate- Striatal TH optical density assessment: Immunostained, mounted, striatal sections were scanned using an Aperio ScanScope at 10 µm resolution and the striata ipsilateral and contralateral to the injection analyzed with ImageJ software (version 1.53). Boundaries of the striatum at three anterior-posterior levels relative to Bregma -anterior +2.16 mm, middle +1.08 mm, posterior -0.05 mm -were determined by comparing anatomical landmarks in sections with a rat brain atlas (27).
Immunocytochemical localization of virally expressed human aSyn in the SNc: Midbrain sections from perfused brains removed from rats injected unilaterally in the SNc with human aSyn-expressing AAV plus proSAAS-expressing lentivirus were processed as described above. Sections were first washed 3 times for 10 min each with PBS then incubated with 10mM citric acid and 0.05% Tween 20 for 10 min at 90˚C. Following cooling, sections were washed 3 times for 10 min/each with PBS before incubating in 5% donkey serum (Jackson ImmunoLab), 1% BSA and 0.5% Triton X-100 in PBS for 2 h at room temperature The comparison of aSyn and TH protein levels was made following normalization of bands to -actin.

Statistical analyses
Comparisons of motor asymmetry scores collected over multiple weeks were analyzed by ANOVA corrected for repeated measures. Group comparisons of post-mortem data were analyzed by ANOVA followed by a Tukey post hoc test when data were normally distributed. When the presumption of normal distribution was not justified, a Kruskal-Wallis test was performed followed by a Mann-Whitney U test with Bonferroni correction. Western blot data were analyzed using ANOVA with aligned rank transform followed by the least-squares means with Tukey corrections. All statistical analyses were performed with custom R code (R Project for Statistical Computing -http://www.R-project.org/).

Transsynaptic aSyn spread study in mice
Subjects Experiments were carried out on 12-week-old female C57BL/6JRj mice (Janvier). Experimental protocols were approved by the State Agency for Nature, Environment and Consumer Protection in North Rhine Westphalia. Mice were maintained on a 12-12 h light-dark cycle. Food and water were available ad libitum throughout the study.

Surgical procedure
Following anesthetization with 2% isoflurane mixed with O 2 and N 2 O, a 1-cm incision was made at the midline of the neck. The left vagus nerve was isolated from the carotid artery, and vector solution (800 nl) was injected at a flow rate of 350 nl/min using a 35-gauge blunt steel needle fitted onto a 10 ml NanoFil syringe as described previously (30). Mice received a mixture of either aSyn-AAV + proSAAS-AAV (n=5) or aSyn-AAV + GFP-AAV (n=5). After injection, the needle was kept in place for two additional minutes. Postsurgery analgesia was provided by subcutaneous injection with buprenorphine (Temgesic, 0.1 mg/kg). Six weeks after injection, mice were euthanized under pentobarbital anesthesia by perfusion through the ascending aorta first with saline containing heparin and then with ice-cold 4% (w/v) paraformaldehyde.

Post-mortem histological assessment
Tissue preparation: Brains were removed, immersion-fixed in 4% paraformaldehyde and cryopreserved in 30% (w/v) sucrose solution. Coronal sections (35 m) throughout the brain were cut using a cryostat and stored at -20°C in phosphate buffer (pH 7.4) containing 30% glycerol and 30% ethylene glycol.
Images were obtained using an Observer.Z1 microscope (Carl Zeiss) equipped with a motorized stage. Low