A Progressive Loss of phosphoSer138-Profilin Aligns with Symptomatic Course in the R6/2 Mouse Model of Huntington’s Disease: Possible Sex-Dependent Signaling

The R6/2 transgenic mouse model of Huntington’s disease (HD) carries several copies of exon1 of the huntingtin gene that contains a highly pathogenic 120 CAG-repeat expansion. We used kinome analysis to screen for kinase activity patterns in neural tissues from wildtype (WT) and R6/2 mice at a pre-symptomatic (e.g., embryonic) and symptomatic (e.g., between 3 and 10 weeks postnatal) time points. We identified changes in several signaling cascades, for example, the Akt/FoxO3/CDK2, mTOR/ULK1, and RAF/MEK/CREB pathways. We also identified the Rho-Rac GTPase cascade that contributes to cytoskeleton organization through modulation of the actin-binding proteins, cofilin and profilin. Immunoblotting revealed higher levels of phosphoSer138-profilin in embryonic R6/2 mouse samples (cf. WT mice) that diminish progressively and significantly over the postnatal, symptomatic course of the disease. We detected sex- and genotype-dependent patterns in the phosphorylation of actin-regulators such a ROCK2, PAK, LIMK1, cofilin, and SSH1L, yet none of these aligned consistently with the changing levels of phosphoSer138-profilin. This could be reflecting an imbalance in the sequential influences these regulators are known to exert on actin signaling. The translational potential of these observations was inferred from preliminary observations of changes in LIMK-cofilin signaling and loss of neurite integrity in neural stem cells derived from an HD patient (versus a healthy control). Our observations suggest that a pre-symptomatic, neurodevelopmental onset of change in the phosphorylation of Ser138-profilin, potentially downstream of distinct signaling changes in male and female mice, could be contributing to cytoskeletal phenotypes in the R6/2 mouse model of HD pathology.

The genetic defect involves a CAG (trinucleotide coding for glutamine/Gln/Q) repeat expansion within exon 1 of the huntingtin gene (HTT). Above the pathological threshold (≥ CAG39; Q39), there is a strong correlation between the number of repeats and the age of onset and/or severity of clinical manifestation (Ross and Tabrizi 2011).
The wildtype (WT) HTT protein functions in a number of processes including energy metabolism, synaptic function, protein transport, transcription, survival, autophagy, and cytoskeletal dynamics [reviewed in (Saudou and Humbert 2016)]. A reduction in WT HTT has been implicated in axonal trafficking defects (Trushina et al. 2004), while the deletion of WT HTT significantly attenuates regeneration, hence implicating it in neural plasticity after injury (Poplawski et al. 2020;Belin et al. 2015). Consistent with this, the mutant huntingtin protein (mHTT) can impact proteins involved in a diverse range of biological processes (Culver et al. 2012;Hosp et al. 2017). The mHTT targets 1 3 primarily medium spiny neurons in the striatum, a structure enriched in dopamine (DA) neurotransmission (Roze et al. 2011), which increases in early stage of the clinical course of HD (leading to hyperkinetic movements) and decreases as pathology progresses, including in mouse models of HD (Cepeda et al. 2014). The progressive disruption in striatal DA transmission and any synaptic plasticity defect might rely, in part, on interference of the critical interaction between DA receptors, components of the actin cytoskeleton (especially Filamin-A, which participates in the anchoring of membrane proteins to the actin cytoskeleton), and downstream DA signalling molecules (Lin et al. 2001); this might occur during development and affect synaptogenesis (McCarthy et al. 2011;Zhang et al. 2010).
The potential for sex-dependent differences in HD progression and severity of phenotype is unclear. For example, women can present with a slightly more severe phenotype and a faster progression of HD than men (Zielonka et al. 2013), although contrasting reports also suggest that symptom onset is later (Roos et al. 1991) and that disease progression is milder (Roos et al. 1991;Chen et al. 2009) in women. Part of the substantial loss of striatal DA receptors and severity of phenotype in symptomatic male (versus female) transgenic HD rats (Q51) has been attributed to lower levels of the neuroprotective sex hormone 17β-estradiol in the male rat (Bode et al. 2008). A role for estrogen in clinical HD is supported by the demonstration that the phytoestrogen, genistein, promotes the breakdown of mHTT in HD fibroblasts (Pierzynowska et al. 2019) and by a much earlier demonstration that premarin could improve motor symptoms, but in less than 30% of patients (Koller et al. 1982). Clearly, examination of a role for estrogen in the context HD is warranted.
We undertook a preliminary comparison of neural stem cell cultures and observed dysregulation of the cytoarchitecture (e.g. a loss of neurite outgrowth) in HD (Q45) cultures (cf. healthy control). Western blotting confirmed changes in molecular signatures implicated in actin signalling, such as changes in LIMK1, cofilin, and SSH1L (all components of the Rho-Rac signalling pathway). In order to identify changes in signaling pathways during the symptomatic course of HD, we chose to use the R6/2 mouse model. This transgenic mouse carries copies of a fragment of exon 1 of the human huntingtin gene containing a Q120 repeat expansion that is sufficient to trigger a progressive behavioral and neurological HD-like phenotype that manifests by 4-6 weeks of age (Mangiarini et al. 1996). We used kinome analysis (Berard et al. 2018) to screen for potential phosphorylation events (Jalal et al. 2009) and identified several affected signaling cascades, including the Rho-Rac GTPase cascade that has been associated with cytoskeletal phenotypes in various models of HD (Puigdellivol et al. 2015;Tourette et al. 2014;Tousley et al. 2019). A role for this cascade was corroborated by evaluating the phosphorylation status of key signaling proteins such as ROCK, LIMK1, SSH1L, cofilin, and profilin in R6/2 mouse tissues. Our observations strongly suggest distinct signaling changes in male and female mice, and as importantly, indicate an onset of signalling defect centered on the HTT-and actin-binding regulator profilin (Shao et al. 2008;Angeli et al. 2010) at embryonic stages, thus corroborating the suggestion that HD progression might have a neurodevelopmental origin (Wiatr et al. 2018).

Neural Stem Cell Cultures
Neural stem cells (NSCs) were derived from iPSCs (induced pluripotent stem cells) obtained from a female HD (Q45) donor (ax0021) and from an age-matched female healthy control (HC) donor (ax0016) (Axol Bioscience, Cambridge, UK). Culture dishes were coated with Axol Sure Bond coating solution (ax0041) prepared in PBS (without calcium or magnesium; D-PBS) overnight at 37 °C. iPSCs were seeded at a density of 10,000-50,000 cells/cm 2 in Axol Neural Maintenance Media (ax0031) supplemented with the Axol Sure Boost serum (ax0045) for 2 h and then cultured for 48 h in Neural Maintenance Medium supplemented with the Axol Sure Growth serum (ax0047). Thereafter, cells were cultured in the Neural Maintenance Medium alone. For passaging and harvesting of NSCs, the cultures were rinsed with PBS and detached using the Axol Neural Unlock solution (ax0044). An Olympus CKX41 light microscope was used for assessing neurite outgrowth, cell number, and soma size (quantitation was performed using Neurolucida360 software: MBF Bioscience, Williston, VT).

Animal Tissue Harvest
All animal procedures were performed in accordance with Canadian Council on Animal Care guidelines and were approved by the University of Saskatchewan's Animal Research Ethics Board. Animals had access to food and water ad libitum, and were housed under constant temperature (± 22 °C) and humidity (± 45%) with a 12:12 h light/ dark cycle. Age-matched breeding pairs of R6/2 transgenic mice (#6494) were purchased from the Jackson Laboratory (Farmington, CT). Tissues were harvested at embryonic day 14 (E14), when the striatum begins to develop (Voorn et al. 1988), as well as at a postnatal pre-symptomatic stage (at 3 weeks of age), a stage when striatal mHTT immunoreactivity is first detectable (at 5 weeks), and a stage when overt symptoms-such as brain and body weight loss, and a visible motor phenotype-are evident (at 10 weeks) (Davies et al. 1997) (Supplementary Fig. 1). Given the lack of a 1 3 defined striatum at E14, whole brain were used for analyses at this time-point. The striatum was used for analyses at postnatal stages, e.g. 3-week (3w), 5w, and 10w. In all cases, mice were euthanized by cervical dislocation.

Peptide Arrays and Kinome Analyses
DAPPLE 2 (https ://saphi re.usask .ca/saphi re/dappl e/) was used to design the peptide arrays (Trost et al. 2013a). The customized peptide microarray (JPT Peptide Technologies GmbH, Germany) contained 1268 peptides (corresponding phosphosites are listed in Supplementary Table 1) designed to cover key signaling pathways (Jalal et al. 2009). Only those murine proteins (corresponding peptides) that have a human homolog were selected to populate the microarray. The selection was accomplished using web-based online databases such as Phosphosite plus (Hornbeck et al. 2012). There were five biological replicates (5 separate arrays) performed, with nine technical replicates per array. The resulting 45 intensity values for every peptide per biological sample were normalized using 'variance stabilization normalization' transformation and the difference in the fold-change and P-values were calculated using PIIKA-2 (Platform for Intelligent, Integrated Kinome Analysis), as described in detail elsewhere (Jalal et al. 2009;Trost et al. 2013b) (Supplementary Table 2). The software (InnateDB) considers both fold-changes and P-values to define significantly dysregulated pathways (Breuer et al. 2013) and draws upon information from several major academic databases, including KEGG, REACTOME, and INOH, to generate a list of candidate pathways. The pathways analysis revealed upregulated as well as downregulated pathways [provided in Supplementary Tables 3 and 4, respectively].

Western Blot Analysis
Tissues were sonicated in RIPA buffer on ice with five 40 mA pulses (3 s each, separated by a 10 s pause), centrifuged at 12,000×g (4 °C, 30 min), and supernatants were heat-denatured. Samples (20 µg protein) were resolved by SDS-PAGE and transferred to nitrocellulose membranes, which were blocked and probed with primary antibodies. Detection relied on Image Studio™ Lite software (LI-COR) and densitometry was normalized to α-/β-tubulin levels.

Statistical Analysis
A peptide was selected from PIIKA2 output for further analysis if its P-value was < 0.2 along with a fold-change Table 1 The list of hyper-and hypo-phosphorylated peptides in both sexes across four developmental time points The developmental time points at which the R6/2 murine tissue were harvested are listed in the first column. The next three columns list the number of peptides that were significantly (P < 0.2) hyper-and hypo-phosphorylated based on the kinome analysis in female and male mice. The last column lists the number of hyper-or hypophosphorylated peptides that were in common in both sexes 1 3 (FC) > ± 1 (Goel et al. 2018;Maattanen et al. 2013). The P-value of 0.2 was chosen as it is known that if the threshold were to be too conservative, then the likelihood of false negatives would increase, and if too relaxed, then the analysis might provide false positives. Further, given that a cellular phenotype is often the reflection of changes in the expression patterns of groups of signaling molecules with common biological functions, identifying a change in a group of these molecules is more biologically meaningful than a change in a single molecule. As importantly, it has been noted that 50-70% of the information from peptide arrays can be lost due to technical reasons during data normalization (Scholma et al. 2016). The cut-off threshold used for InnateDB pathway analysis was more stringent (P < 0.05 and FC > ± 1.5), with P-values being generated using the hypergeometric distribution test that confirms-prior to correction for multiple testing-whether a pathway is statistically more overrepresented in the uploaded dataset than expected by chance. P-values are automatically corrected using the Benjamini and Hochberg or by a conservative Bonferroni correction (Breuer et al. 2013). Our kinome analyses relied on 5 males and 5 females per genotype per test time-point. The priority of kinome analysis is to identify targets that can be validated by an independent approach, for example, Western blotting. Western blot bands were quantified using Image Studio Lite (LI-COR Biosciences) and the intensities were normalized using housekeeping control (α/β-tubulin). The phospho-proteins were expressed relative to total protein and the corresponding ratios were used for statistical analyses based on two-way analysis of variance (ANOVA) and post hoc Tukey's multiple comparison test (GraphPad v7, PRISM). Morphological features of NSCs were estimated using six separate fields from several HC and HD cultures, and averages were compared using the Student t-test. Statistical significance was set at P < 0.05. All data are expressed as mean ± standard error of the mean. Our Western blotting relied on 3 males and 3 females per genotype per test time-point.

Neurite Retraction in Patient-Derived HD Neuronal Cells
Comparison of NSC cultures derived from a healthy control (HC) (Fig. 1a, c, e) and an HD patient and from (Fig. 1b, d, f) did not reveal any significant loss of cell number (Fig. 1g) or change in shape of the cell (Fig. 1h). However, there was a 25% decrease in neurite length in the HD NSCs (P < 0.05) (Fig. 1i). Western blotting for selected proteins implicated in actin organization and cytoskeletal integrity revealed less phosphorylation of LIMK1 in the HD NSC lysate, but more phosphorylation of the LIMK1 substrate, cofilin; the latter might be reflecting the lower levels of phosphorylation (and, hence, inactivation) of the cofilin phosphatase, SSH1L (Fig. 1j). Levels of actin were higher in the HD NSC lysate, whereas those of β-tubulin remained unaltered. Preliminary Western blotting of the NSC extracts also revealed changes in the phosphorylation of the pro-survival kinase Akt (− 70%), the cell cycle regulator CDK2 (+ 169%), and the stress-activated kinase involved in cytoskeletal organization, cell cycle, and chromatin remodeling, MAPKAPK2 (+ 119-183%) ( Fig. 1k) (discussed below).
These data suggest that the loss of communication between cells in HD might rely primarily on a loss of axonal integrity and synaptic connectivity, and implicate a potential influence by the LIMK/SSH1L/cofilin pathway. However, the interpretation of these data is hindered by the fact that the iPSCs available from the commercial source at the time were limited to a single HC female donor and a single sex-/ age-matched HD donor (leaving us with a biological replicate of '1'). This precluded any possibility of concluding whether the observed changes were due to the sex of the donor, the diagnosis of HD, an interaction between sex and diagnosis, or even variation within the cultures (given that they were non-isogenic). Yet we feel that the observations, even if based on a single biological replicate, clearly indicated a cytoskeletal defect and implicated the LIMK/SSH1L/ cofilin pathway, and thus provided justification for in vivo studies. To this end and to explore whether sex might exert influence, we screened protein kinase activities in the R6/2 mouse model of HD (and WT mice) using kinome analysis. Our experiments included both male and female mice.

The Phospho-protein Profiles in R6/2 Mice Suggest Distinct Sex-Dependent Influences on Signaling
Kinome analysis identified peptides that were significantly hyper-and hypo-phosphorylated at a pre-symptomatic stage (E14) (Fig. 2a) as well as across all three postnatal, symptomatic time points (Fig. 2b-d). The analysis revealed peptides that were similarly hyper/hypo-phosphorylated in both sexes and others that were preferentially hyper/hypophosphorylated by sex (Table 1).
The kinome analysis data (Supplementary Table 2) were uploaded onto InnateDB along with their respective P-values and fold-changes (FC); this generated a list of pathways that were up-/down-regulated across the time course in these mice (provided in Supplementary Tables 3 and 4). The top three most significantly upregulated pathways were 'Caspase mediated cleavage of cytoskeletal proteins' (P = 7.17E−04; REACTOME), 'Lysosomes' (P = 8.91E−04; KEGG), and 'Peptide ligand binding receptors' (P = 0.004190495; REACTOME) ( Supplementary Fig. 2). The top three most significantly downregulated pathways were 'Degradation of DVL' (P = 3.53E−04; REACTOME), 'Beta-catenin independent WNT signaling' (P = 0.001525682; REACTOME), and 'Degradation of GLI1 by proteasome' (P = 0.001658169; REACTOME) ( Supplementary Fig. 3). Other pathways identified were [upregulated] 'Depolymerisation of the nuclear lamina', 'disinhibition of SNARE formation', and 'metabolism of steroid hormones and vitamin D' and [downregulated] 'Trafficking of AMPA recycling', 'Recycling of L1′, 'Transmission across chemical synapses'. Although the RhoRac GTPase pathway was not specifically represented in the InnateDB analysis performed, we manually annotated members of the pathway ( Supplementary  Fig. 4) and identified several changes in phosphorylation of proteins in the kinome analysis. The major components of the ROCK (Rho-associated protein kinase) and PAK (p21-activated kinase) cascades ( Supplementary Fig. 4) involve serine/threonine protein kinases (and counterbalancing phosphatases), with primary regulatory effects on the actin cytoskeleton and ultimate phenotypic effects centered on neuronal growth and synaptic plasticity (Zhao and Manser 2012;Julian and Olson 2014). Other preliminary examinations of signaling pathways are included in Supplementary Figs. 5-8. These pathways are included for sake of information and confirm our preliminary Western blotting of the NSC extracts, which also revealed changes in the phosphorylation of the pro-survival kinase Akt, the cell cycle regulator CDK2, and the stress-activated kinase involved in cytoskeletal organization, cell cycle, and chromatin remodeling, MAPKAPK2 (Fig. 1k). These signaling pathways were not explored further in this report.

Validation of Kinome Analysis Data by Western Blot Analysis
We investigated the phosphorylation levels of selected proteins in the Rho-Rac pathway at time-points relevant to key stages of HD progression in the R6/2 mouse model, namely developmental (E14), pre-symptomatic (3w), early disease (5w), and late-stage disease (10w). The levels of β-actin tended to be more variable than those of α-tubulin between the sexes and across genotypes (Fig. 4a-i). Consequently, levels of α-tubulin were used to monitor protein loading in the Western blotting experiments.
At the E14 time-point, the phosphorylation of ROCK2 was significantly elevated in female R6/2 mice compared to female WT mice (P < 0.0001) and male R6/2 mice (P < 0.05)  Changes in phosphorylated peptides identified in the kinome analysis of the R6/2 mice: a Cluster analysis of kinome data sets of neural tissue samples of R6/2 mice (relative to WT mice). Kinome data sets were subjected to hierarchical clustering analysis using PIIKA-2. The age of the animal is indicated under the heatmap where number represents the time point (E14, 3w, 5w, and 10w) followed by sex (F/M) and the genotype, e.g. WT (W) and R6/2 (R). Each column depicts the kinome activity at that time point. Green represents hypo-phosphorylated peptides and red represents hyper-phosphorylated peptides. b Fold-change heatmap for the three indicate phosphopeptides across four time-points, e.g. E14, 3w, 5w, and 10w, in both sexes. The names of the peptides and the phosphosite are indicated at the top of each column. The color key represents positive values in red and negative values in green. c Scatterplot of fold-changes based on sex. Males are represented as squares and females as circles, with a different color assigned to each time-point, as indicated in the panel 1 3 (Fig. 5a, b). The phosphorylation of PAK was lower in both sexes in R6/2 mice compared to WT mice (Female: P < 0.05; Male: P < 0.001) (Fig. 5c, d), while the phosphorylation of LIMK1 and cofilin were not affected by sex or genotype ( Fig. 5e-h). The phosphorylation of SSH1L was lower in female (P < 0.05)-but not male-R6/2 mice (Fig. 5i, j). The phosphorylation of profilin was substantially elevated in both sexes in R6/2 mice compared to WT mice (female: P < 0.001; male: P < 0.05) (Fig. 5k, l).
There was significant variability in the phosphorylation of the proteins examined at the 3w time-point, when a mouse is considered juvenile and the brain is thought to be still maturing. For example, the phosphorylation of ROCK2 was higher in WT males than in WT females (P < 0.05), but was unaffected by genotype (Fig. 6a). The level of phosphorylation of PAK was higher (or lower) depending on the sex and genotype (Fig. 6). The phosphorylation of LIMK1 was lower in the female R6/2 mouse compared to the WT females (P < 0.05) (Fig. 6c), while the phosphorylation of cofilin was higher in the male R6/2 mouse (cf. male WT and female R6/2 mice) (P < 0.0001) (Fig. 6d). The phosphorylation of SSH1L was similar to that of PAK in that in that it was higher (or lower) depending on the sex and genotype (Fig. 6e).
Remarkably, the levels of phosphoSer138-profilin, which were substantially higher in the R6/2 mice than in the WT mice at the E14 stage (see Fig. 5), remained high in the R6/2 mouse at the 3w stage (P < 0.05), were comparable to levels in the WT mice at 5w of age, and by 10w were significantly lower in the R6/2 mice than in the WT mice (female: P < 0.01; male: P < 0.05) (Fig. 6f). This pattern was not influenced by the sex of the mouse.

Discussion
Kinome analysis based on peptide arrays is a validated platform (Scholma et al. 2016 (Arsenault et al. 2013b(Arsenault et al. , 2013a. Systematic quantitative proteomics-supported by mass spectrometry-based on striatal tissues from the R6/2 mouse (Hosp et al. 2017) and post-mortem HD patient samples (Ratovitski et al. 2016) revealed a widespread loss of protein function that implicates Rho proteins, actin cytoskeleton signaling, and mitochondria (Ratovitski et al. 2016), as well as proteins related to energy metabolism and cellular transport/cytoskeleton (Wegrzynowicz et al. 2012). A recent quantitative proteomic study implicated HTT as a critical regulator of neural injury response in adult mice suggesting its importance in neuronal survival and axon regeneration (Belin et al. 2015). Several high-throughput studies of HD using mouse models, post-mortem brain, and patient-derived stem cells also implicated dysregulation of actin signaling, including a loss of profilin expression at early stages of the disease process (DiProspero et al. 2004;Goldberg 2003;Heng et al. 2010;Lorincz and Zawistowski 2009;McQuade et al. 2014;Niwa et al. 2002;Burnett et al. 2008).
Our kinome analysis of striatal tissues identified several candidate pathways (see Supplementary Figs. 5-8) including the Akt/FOXO3 pathway that is neuroprotective in HD (Farina et al. 2017) and CDK2 (which we also observed in our NSC extracts, Fig. 1), thus suggesting a dysregulation of cell cycle regulatory proteins (Sang et al. 2014). We identified a loss of phosphorylation of mTOR and a corresponding increase in the phosphorylation of ULK1, which would indicate an activation of autophagy, likely in response to cellular stress (Rui et al. 2015), and our observed loss of RAF/MEK/ CREB signalling over the symptomatic course in the R6/2 mice is consistent with a loss of ERK activation in cell death in models of HD (Bodai and Marsh 2012). ERK and Akt signalling deficits have been implicated in the loss of differentiation and neurite retraction in Q48-and Q89-expressing (but not Q16) PC12 cells (Song et al. 2002) and although these systems deserve to be characterized within the context of the R6/2 mouse, we focused on the Rho-Rac GTPase effector proteins, e.g. ROCK and PAK given our preliminary observations based on HC and HD NSC cultures. Although ROCK and PAK target unique substrates, there is abundant evidence that they both modulate LIMK-cofilin signaling and associated phenotypes. LIMK1 is highly expressed in the brain (Proschel et al. 1995) and the LIMK-cofilin association helps support the integrity and structure of dendritic spines (Linseman and Loucks 2008;Govek et al. 2005) as well as axonal growth (Heng et al. 2010;Koch et al. 2014). Part of cytoskeletal integrity might rely on the phosphorylation of cofilin-Ser3 by LIMK1, which inactivates cofilin and prevents its binding to actin (Yang et al. 1998). Loss of phosphorylation of cofilin-Ser3 (as we've seen in some of our samples) impairs cofilin function and monomeric actinturnover in the cytoplasm leading to motility and morphological deficits, such as cell shrinkage (Munsie et al. 2012;Bravo-Cordero et al. 2013). The dephosphorylation of cofilin is not necessarily negative; indeed, during cell stress dephosphorylated cofilin can be sequestered as cofilin-actin rods, thus freeing up a pool of ATP bound to cofilin for critical cellular processes (Bernstein et al. 2006). Wildtype HTT helps localize these cofilin-actin rods to the nucleus, but these rods then disappear with the relief of the cellular stress; in contrast, mHTT induces a dominant, persistent nuclear cofilin-actin rod phenotype that triggers, amongst other events, an increase in calcium levels and cell death (Munsie et al. 2011).
Profilin exerts the opposite action to cofilin on actin and its polymerization, and can affect neuronal growth cone and synaptic plasticity (Birbach 2008). Yet the roles of these pathways are not as straightforward as anticipated. Indeed, the Rho and Rac pathways have been shown to exert mutual antagonism in N-Cadherin-mediated contact mechanisms in myoblasts, but sequential roles for these kinases are essential for contact communication in these same cells (Comunale et al. 2007). Interestingly, inhibition of either Rho or Rac elicit opposite effects of actin-based repair mechanisms in gastric epithelium (Aihara et al. 2018) and while Rho does not exert much influence on the leading edge of lamellipodia in rat adenocarcinoma cells, its inhibition does unmask a Rac-mediated facilitation of edge growth (El-Sibai et al. 2008). The phosphatase SSH1L also targets cofilin-Ser3, thus promoting cofilin (re)activation (Romarowski et al. 2015). As with Rho and Rac, the role of SSH1L in LIMKcofilin-actin polymerization is viewed more as a contextdependent or sequence-dependent influence rather than simply as the phosphatase that targets cofilin.
A correlation between profilin inactivation and altered cytoskeletal dynamics could affect neurite morphology, given that actin microfilaments tend to be concentrated at the synaptic terminals, dendritic spines, and growth cones (Matus et al. 1982;Gordon-Weeks 1987). Although we observed changes in phosphorylation of both ROCK2 and PAK in the R62 mice at E14, we did not observe any corresponding change in the phosphorylation of their purported targets LIMK1 or cofilin (although we did observe a tendency for an increase in phosphoSer3-cofilin in E14 female R6/2 mice: P = 0.055; Fig. 5g). The phosphorylation of profilin was significantly higher at E14 in R6/2 mice compared to WT mice, regardless of sex, which suggests that the phosphorylation of ROCK2 (cf. the loss of phosphorylation of PAK) might be driving the phosphorylation of Ser138profilin at this time point. This would presumably stabilize developmental dynamic actin structures and could help explain the overabundance of synaptic connectivity (likely due to a pruning defect) demonstrated elsewhere in a conditional knockout of Htt as well as a knock-in (Q175) mouse model of HD (McKinstry et al. 2014). Interestingly, these authors demonstrate that the loss of normal Htt function leads to an unanticipated early exaggerated increase in synapse formation; however, this phenotype is not sustainable and eventually leads to a loss of synaptic density by 5 weeks of age and a gliosis, but no neuronal death (McKinstry et al. 2014). Similarly, the expression of mHtt in the R6/1 mouse and in the Q7/Q111 knock-in mouse leads to a corticostriatal phenotype centered on deficits in cortical cell migration and electrophysiological properties, including a loss of longterm potentiation, a mechanism for strengthening synapses and critical for memory formation (Puigdellivol et al. 2015). This cortical phenotype precedes any loss of striatal synaptic integrity and associated motor deficits, and appears to be triggered by the loss (by two months of age) of Kalirin-7, a brain-specific Rho-guanine nucleotide exchange factor for Rac-like GTPases, that is expressed highly in dendritic spines of neuronal populations (Puigdellivol et al. 2015). The concurrent hyper-and hypo-phosphorylation of ROCK and PAK, respectively, observed at this time point (Fig. 5a, b) adds support to the notion of context-dependent and/or sequence-dependent influences of ROCK and PAK, as discussed in the previous paragraph.
Western blot analyses of components of the striatal ROCK2 and PAK pathways revealed dynamic changes with some level of similarity within sex/genotype across the postnatal time-points (Fig. 6). For example, the phosphorylation pattern of ROCK2 between the sexes and genotypes at 5w was more similar to the pattern observed at 3w than the pattern at 10w, while the pattern of phosphorylation of PAK and cofilin at 5w resembled more so that in the 10w sample set. Patterns of phospho-LIMK1 and phospho-SSH1L appeared to be more in flux at 5w. Recall, it is this age in the R6/2 mouse that striatal mHTT immunoreactivity is first detected (Davies et al. 1997) and it is this age in the Q175 mouse that is associated with a loss of synaptic density and increased gliosis (McKinstry et al. 2014). In our studies, this age also aligned with a change in the relative levels of phosphoSer138-profilin. Indeed, the levels of phosphoSer138-profilin were substantially higher in R6/2 mice at E14 and only slightly less so at 3w (again regardless of sex), but at 5w they were similar to levels in the WT mice and by 10w they were significantly lower than the levels in WT mice. While a screen of Rho pathway mRNA transcripts found a significant increase in profilin mRNA expression in autopsied HD patient cortical samples as well as in 13wold pooled (male + female) R6/2 mouse striatum (but not at 4w), there was no corresponding change in profilin protein expression (Narayanan et al. 2016) and these authors did not explore the phosphorylation status of profilin. It was shown elsewhere that Y-27632, a rho-kinase (ROCK) inhibitor, blocks the phosphorylation of profilin, which binds actin as well as Htt (Shao et al. 2008), reduces intracellular aggregation of Htt (Pollitt et al. 2003), and inhibits Htt toxicity in Drosophila and motor deficits in mice (treatment began at age 4 weeks) (Pollitt et al. 2003;Li et al. 2009). We are unclear as to why profilin shifts from a phosphorylated to an unphosphorylated state between 5 and 10 weeks of age and how this might align with pathology. Perhaps the hyperphosphorylated state of profilin observed at the embryonic stage releases mHTT to localize with, amongst other proteins, perinuclear α-actinin-1-enriched stress fibers (Tousley et al. 2019) and trigger disruption of the nuclear lamina (Gasset-Rosa et al. 2017) (as suggested by our pathways analysis; see Supplementary Fig. 2) and transport via the nuclear pore complex (Grima et al. 2017). In keeping with a nuclear phenotype, susceptibility to DNA damage or the induction of genes for cell cycle re-entry and transition from G1 to S phase (downstream of mitochondrial stress and normally leading to apoptosis in neurons) has also been shown to be proportional to the CAG repeat lengths (Q30, Q45, Q65, and Q81) in isogenic embryonic stem cell lines . Perhaps this hyperphosphorylated profilin destabilizes actin structures and interferes with DA signalling (Lin et al. 2001) as observed in the early stages of the disease progression (McCarthy et al. 2011;Zhang et al. 2010). Interestingly, our pathways analysis also reveals a 'disinhibition of SNARE formation' (Supplementary Fig. 2) as well as a loss of 'transmission across chemical synapses' (Supplementary Fig. 3). This could alter quantal DA release and trigger the hyperactivity observed in young R6/2 mice (which gradually disappears until the mice become hypoactive by 8w) (Carter et al. 1999). Whatever the mechanism, our data suggest an mHTT-induced developmental profilin phenotype. An additional a priori conclusion stemming from this study is that if the inconsistent phosphorylation profiles for LIMK1, cofilin, and/or SSH1L are contributing to the phosphorylation of Ser138-profilin and to the onset or development of the disease, then their roles are likely sex-dependent and either sequential or cascade-specific, as suggested elsewhere and discussed above. Yet, it is also possible that other regulators of the LIMK-cofilin pathways, such as the cofilin phosphatase PP2A (Pendleton et al. 2003) or the LIMK phosphatase PP1 (Vorster et al. 2011), could be exerting temporal or sequential influences. We also cannot discount the possibility of interference by the mHTT in the function of the WT HTT, e.g. transport and trafficking (Caviston et al. 2007;Gunawardena et al. 2003;Her and Fig. 5 Western blots of the proteins involved in Rho-Rac signaling at E14 in wild type (WT) and R6/2 mice. Densitometry was used to quantify the ratio of phosphorylated to total protein expression of a, b ROCK2, c, d PAK, e, f LIMK1, g, h Cofilin, i, j SSH1L, k, l Profilin. Each value was initially normalized to expression of α-tubulin in the corresponding lane. Note that the same α-Tubulin blot might appear in more than one panel because of re-probing for multiple targets with non-overlapping molecular weights on a given blot. The data are presented as mean ± sem (n = 3). *P < 0.05; **P < 0.01, between indicated groups ◂ Goldstein 2008; Orr et al. 2008) as trafficking and recycling pathways were also identified in our analysis (Supplementary Fig. 3). Any of these are potential mechanistic contributors to the phosphoSer138-profilin phenotype and symptomatic progression, and certainly warrant being investigated further in terms of a sex-dependent influence on HD.
Our initial observation of a reduction in neurite length in HD NSCs is consistent with reports of reduced neurite length or abnormal dendritic branching in HD (Ferrante et al. 1991;Rong et al. 2006;Liu et al. 2014). Recently, human HD patient cells and mouse (Q140/Q140) striatum and primary neurons were shown to be less sensitive to growth factor stimulation and this reflected a disruption of a Rac1:p85(PI3K):α-actinin-2 complex (e.g. the mHTT does not interact with p85), which is enriched in striatal neurons and spines (Tousley et al. 2019). The loss of a stable complex could impact any growth factor-mediated, PI3K-dependent neurite outgrowth. A role for Rac1 and Rho GTPase signalling in the context of HD was also demonstrated in a yeast two-hybrid screen designed to identify HTT binding partners (Tourette et al. 2014). That study identified a number of candidates found in Rho GTPase family signaling, including Ezrin, several PI3K family members, and BAIAP2 (brain-specific angiogenesis inhibitor 1-associated protein 2). Functional assays demonstrated that mHTT interferes with BAIAP2-mediated filopodia-like protrusions in mouse embryonic fibroblasts (Tourette et al. 2014). The effect of mHTT on neurite outgrowth seems quite specific given that iPSCs derived from HD (Q77; Q109; Q180) and control (Q18; Q21; Q33) can all be differentiated toward a healthy cortical neuronal fate; but the CAG repeat length in HD iPSCs correlate directly with decreased neurite lengths, without any overt effect on branching morphology (Mehta et al. 2018). Neurite outgrowth is essential for the proper wiring of the nervous system during development and regeneration (Miller and Suter 2018), and cytoskeletal proteinsalong with HTT (Burrus et al. 2020)-are critical in this process.
Finally, we address our observation that dysregulation of phosphoSer138-profilin in the R6/2 mouse model occurs long before the reported onset of phenotypic changes. Indeed, the phosphorylation of profilin at the embryonic stage suggests a developmental phenotype, potentially centered on a pruning defect, as discussed above. However, the progressive reduction in phosphoSer138-profilin at postnatal stages suggests a progressive defect in actin cytoarchitecture across the symptomatic course of the disease. This might involve exaggerated pruning, if one considers our observation of a loss of neurite length (e.g. enhanced retraction) observed in the HD NSCs as well as the massive pruning of glutamatergic terminals observed in the Q140 (Deng et al. 2013) and Q175 (Rothe et al. 2015) mouse models of HD, and the progressive loss of somatosensory cortical dendritic spine density over symptomatic stages of the R6/2 mouse (Murmu et al. 2013). This provides additional support for the suggestion that HD follows a developmental course centered on defects in cortical neurogenesis, axonal transport, and Golgi apparatus organization (Humbert 2010). Furthermore, our data support the suggestion that the HD genotype (e.g., 72 + CAG repeats) leads to defects through a lossof-function mechanism as early as the neurulation stage (Haremaki et al. 2019).
The expression of WT HTT tends to increase with brain development (Marques Sousa and Humbert 2013;Bhide et al. 1996). Previous studies have shown that WT HTT associates with microtubules and is involved with transport in both anterograde and retrograde directions, whereas mHTT interferes with these processes, ultimately affecting brain development and/or causing neuronal dysfunction and death (Caviston et al. 2007;Gunawardena et al. 2003;Her and Goldstein 2008;Orr et al. 2008). The appearance of striatal mHTT [at 5 weeks: (Davies et al. 1997)] and the reduction in levels of functional WT HTT [by week 7: (Zhang et al. 2003)] in the R6/2 mouse likely exacerbates the defects in neuronal connectivity and transport, and expedites the course of symptomatology. Transport defects initially led to the suggestion that the R6/2 mouse and its rapidly progressing phenotype is likely a better reflection of juvenile onset HD (more likely when the CAG repeat expansion is in excess of 70) (Mangiarini et al. 1996). The critical role for HTT in the brain has also led to the suggestion that HD is a neurodevelopmental disorder, rather than simply an adult neurodegenerative disorder (Wiatr et al. 2018) and has also led to the assumption that mHTT carriers experience normal brain development, but that an emerging degenerative phase ultimately leads to the appearance of clinical symptoms. A recent study based on human organoids suggests that the CAG/glutamine repeat length in HTT regulates neurogenesis during early development (Zhang et al. 2019), while a repeat length below the disease threshold benefits brain structure and general intelligence among children aged 6-18 years of age (Lee et al. 2017). These same authors also reported that a higher repeat length (as long as it is below disease threshold) gives females an advantage on cortical thickness and intelligence. Fig. 6 Densitometric analysis of the proteins involved in Rho-Rac signaling at postnatal time-points in wild type (WT) and R6/2 mice. The three time-points represent different stages of disease progression in the R6/2 mouse: e.g. 3 week old (3w) = preclinical; 5w = emergence of motor abnormalities; and 10w = overt pathology. Densitometry was used to quantify the ratio of phosphorylated to total protein a ROCK2, b PAK, c LIMK1, d Cofilin, e SSH1L, f Profilin. Each value was initially normalized to expression of α-Tubulin in the corresponding sample. The data are presented as mean ± sem (n = 3). *P < 0.05; **P < 0.01; ***P < 0.001, between indicated groups ◂ Conclusion Overall, our observations suggest a potential sex-dependent influence on cofilin/LIMK1/SSH1L signaling in HD. We are re-assured that several of the observations made using the R6/2 mouse tissues corroborated our observations in the HD NSCs (even if only a single biological replicate). This is even more re-assuring given that these HD NSCs were from a Q45 donor, while the R6/2 mice are Q120 and many other works that we cite herein are based on equally or more aggressive Q77, Q109, Q140, Q175, Q180 etc. genotypes. We were concerned that our analyses did not identify any consistent changes in metabolism, be it energy or substrate, which are often identified in HD-related screens. Part of this could be viewed as a limitation of the interpretation of kinome platform. For example, in males at E14, we observed a significant downregulation [P = 0.002189] as well as a significant upregulation [P = 0.03161] of 'metabolism of amino acids'. This apparent contradiction might simply reflect two distinct phosphopeptides being identified in the screen (one being upregulated, the other being downregulated) and the potentially different roles of the parent proteins in 'metabolism of amino acids'. We also acknowledge that a limitation of our study is that the data remain correlational; however, they do suggest a biological mechanism implicating a progressive phosphoSer138-profilin phenotype. Perhaps more importantly, our observations suggest that the phosphoSer138-profilin phenotype emerges in the earliest stages of brain development, well before any manifestation of symptoms, providing for a clinically targetable and modifiable event. This warrants investigating whether this phenotype is causative and, if so, is it specific to the R6/2 mouse model of HD or is it generalizable across models (e.g., Q77, Q140, Q175) and/or in clinical HD. As importantly, both males and females need to be included in future studies so as to better define potential sex-dependent mechanisms in HD onset and progression.
Author Contributions AB, WJR and DDM made substantial contributions to the conception and to the interpretation of data; AB and ZW were involved in the acquisition of data; AB, WJR and DDM drafted the work; all authors reviewed the manuscript critically and approved the version to be published; all authors agree to be accountable for all aspects of the work in ensuring accuracy or integrity of the data.
Funding AB was supported, in part, by Saskatchewan Agricultural Development Funds, by a Saskatchewan Innovation Opportunity Scholarship, and by departmental student assistantship funds. DDM is the Saskatchewan Research Chair in Alzheimer Disease and Related Dementias funded by the Alzheimer Society of Saskatchewan and the Saskatchewan Health Research Foundation. DDM also receives support from the College of Medicine, University of Saskatchewan.
Data Availability Available through public data repository.

Compliance with Ethical Standards
Conflict of interest The authors declare that they have no conflict of interest.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creat iveco mmons .org/licen ses/by/4.0/.