VGF is required for recovery after focal stroke

The high incidence of ischemic stroke worldwide and poor efficacy of neuroprotective drugs has increased the need for novel therapies in stroke recovery. Transcription of the neurosecretory protein VGF (non-acronym) is enhanced following ischemic stroke and proposed to be important for stroke recovery. To determine the requirement for VGF in recovery, we created Vgf fl/fl :Nestin-Cre conditional knockout ( Vgf cKO) mice and induced a photothrombotic focal ischemic stroke. Naïve Vgf cKO mice had significant less body weight in the absence of gross defects in brain size, cortical lamination, or deficits in locomotor activity compared to wildtype controls. Following a focal stroke, the Vgf cKO mice had greater deficits including impaired recovery of forepaw motor deficits at 2- and 4-weeks post stroke. The increase in deficits occurred in the absence of any difference in lesion size and was accompanied by a striking loss of stroke-induced migration of SVZ-derived immature neurons to the peri-infarct region. Importantly, exogenous adenoviral delivery of VGF (AdVGF) significantly improved recovery in the Vgf cKO mice and was able to rescue the immature neuron migration defect observed. Taken together, our results define a requirement for VGF in post stroke recovery and identify VGF peptides as a potential future therapeutic.


Introduction
Stroke is the third leading cause of death and disability in Canada, affecting 10% of those over the age of 65 (Canada, 2009). This devastating neural event evokes a cascade of biological processes such as cell death and neuroinflammation (Khaku and Tadi, 2020). The increase in survival rates after stroke has shifted the focus of research to enhance stroke recovery, by extending the critical period for recovery with the goal to reduce long-term deficits (Joy and Carmichael, 2021;Kannangara et al., 2018;Leker et al., 2007).
The post stroke neural environment is in a state of shock and dysregulation, with attempts at minimizing the damage through processes such as oligodendrogenesis, the production of adult generated neurons, and modulation of the inflammation response to stroke. Indeed, improving remyelination and limiting an inflammatory response is thought to be the formula for preventing many neurodegenerative disorders including multiple sclerosis (Kim and Kornberg, 2021). Similarly, harnessing the potential of adult stem and progenitor cells to facilitate brain repair has been a hot topic since the discovery of this cellular niche and its robust response to stroke. Indeed, stroke-induced increase in proliferation and ectopic migration of cells from the subventricular zone (SVZ) to the infarct region is well described but the functional role of these cells in recovery remains controversial (Ceanga et al., 2021;Williamson et al., 2019). Our work has shown that a minority of the cells that migrate from the SVZ integrate into the neural circuitry (Kannangara et al., 2018), while others recently have shown that increasing the activity of the peri-infarct cortex can increase the integration and synaptic connections of these newborn cells that are required for stroke recovery (Liang et al., 2019). Mediators of the post stroke inflammatory response attempt to rid the damaged region of pathogens and cellular debris to re-establish homeostasis through the propagation of an inflammatory cascade (Anrather and Iadecola, 2016). Although this systemic response plays a role in the preservation of neurons and overall recovery, it can also further exacerbate tissue injury (Mo et al., 2020).
Innate immune cells, astrocytes and microglia, are key mediators of this response, participating in pathogen clearance and facilitating improved functional recovery (Domingues et al., 2017;Fehlings and Nguyen, 2010;More et al., 2013). Complement signaling pathways are responsible for mediating inflammatory responses and their pathologies, influencing both beneficial and exacerbating effects (Alawieh et al., 2015;Stokowska et al., 2017). In this regard, nasal delivery of C3a peptides in a mouse photothrombotic model of stroke resulted in post reperfusion secondary injury when administered immediately post stroke, but increased pathogen clearance and enhanced recovery when given at later times (Ma et al., 2019). The dual role of damage and repair is most likely due to the anaphylatoxin properties of C3a, which elicit multiple downstream applications (Ma et al., 2019).
Another peptide that signals through the C3a receptor is TLQP-21, a C-terminal peptide of the precursor protein VGF (Cero et al., 2014). VGF was first identified as a rapidly upregulated gene in PC-12 cells following nerve growth factor stimulation but has since been shown to be upregulated by exercise and other neurotrophic factors including BDNF (Levi et al., 1985). The VGF gene encodes a secreted precursor to a wide range of neuroendocrine peptides that are associated with a diverse set of functions and are recognized as a biomarker for several neurodegenerative conditions (Ferri et al., 2011;Quinn et al., 2021). Most notable are their anti-depressant properties and their causal role in protecting against neurodegeneration in models of amyotrophic lateral sclerosis (ALS) and Alzheimer's disease (Beckmann et al., 2020;Jiang et al., 2018;Jiang et al., 2017b;Mizoguchi et al., 2019;Shimazawa et al., 2010;Thakker-Varia et al., 2014;Zhao et al., 2008). Previously, we have demonstrated the ability of VGF and its secreted peptides to influence the proliferation of oligodendrocyte progenitor cells in a model of developmental brain damage, where ectopic delivery of adenoviral VGF to mice with severe cerebellar ataxia showed significantly lengthened survival and an increase in the proliferation of oligodendrocyte progenitor cells (Alvarez-Saavedra et al., 2016). Moreover, removal of the C3a receptor enhanced gliosis and cell death that increased the severity of the phenotype and further supported a role for C3a and VGF signaling in limiting neuronal loss and/or by promoting a reparative environment (Young et al., 2019). Collectively, these studies have indicated the importance of VGF in proper neurological functioning and its role in a variety of neurological conditions. With respect to stroke, two separate studies identified an acute upregulation of VGF transcript levels in the post ischemic brain that was suggested to be an NGF-mediated response to promote brain repair processes (Kury et al., 2004;Sakamoto et al., 2015). Here we test the in vivo requirement of VGF in post stroke recovery and whether it alters the stroke-induced SVZ-derived neurogenic response.

Animal husbandry
Mice were singly housed in preparation for subsequent experiments. Wildtype C57BL/6 J mice (7-weeks) were purchased from Charles River Labs for use at 10 weeks. C57BL/6 J Vgf flox/flox mice were generously provided by Dr. Lei Cao (Ohio State University) and originally generated by Dr. Stephen Salton (Icahn School of Medicine, Mount Sinai, NY, USA). Vgf flox/flox mice were bred to Nestin-cre mice (Berube et al., 2005) to generate conditional knockout animals (Vgf cKO) for analysis. Animals were housed in a facility under SPF (specific pathogen-free) conditions on a 12/12 light:dark cycle with water and food ad libitum. All animal experiments were approved by the University of Ottawa's Animal Care ethics committee, with the guidelines set out by the Canadian Council on Animal Care.

Photothrombotic stroke and adenoviral treatment
Ischemic stroke was induced in the right sensorimotor cortex using the photothrombosis (PT) model (Watson et al., 1985). Briefly, mice were anesthetized (5% isofluorane; 1% oxygen) and body temperature was maintained at 37 • C with the use of a heating pad and rectal temperature monitoring. To induce focal cerebral ischemia the mice were intraperitoneally injected with Rose Bengal dye (10 mg/ml, 198,250-5G; Sigma Aldrich) and transferred to a stereotaxic frame. A small incision was made to expose the skull to a green laser (10 min; 532 nm, 20 mW, MGM-20; Beta Electronics) with the following brain coordinates (+0.7 anterior-posterior, +1.5 medial-lateral, 3 cm laser height, relative to Bregma). The scalp was then closed with tissue glue and 2% transdermal bupivacaine was administered as an analgesic immediately after surgery and four hours post-surgery. Sham surgeries were treated identically but were injected with saline rather than Rose Bengal dye.
The generation of the VGF and control adenovirus have been described previously (Alvarez-Saavedra et al., 2016). An adenovirus expressing VGF (AdVGF) or an empty control virus (CtrlAd) were diluted to 1 × 10 12 viral particles per kilogram in saline to a final volume of 100 μL and introduced intravenously by tail-vein injection into the mice 48 h after PT surgery.

Infarct volume quantification
Magnetic Resonance Imaging (MRI) was performed 24 h post PT to determine infarct volume. Mice were anesthetized (2% isofluorane) for 10-15 min while scans were generated using a small-animal scanner (7 T General Electric/Agilent MR901). Cardiovascular and respiratory function was monitored throughout the procedure. After the infarct region was located through a preliminary scan, T2-weighted structural images were obtained (15 transverse sections; thickness = 800 μm). MRI images were analyzed using Fiji (Schindelin et al., 2012), through manual tracing and measuring the cortical damage in each slice. Infarct volume was then calculated by multiplying the sum of traced areas in each slice by the slice thickness.

Behavioural analysis
General locomotor function was measured in Vgf cKO and wild type (WT) control mice by placing each mouse in a novel mouse shoebox sized home cage during their light cycle surrounded by a metal frame equipped with infrared detectors (VersaMax Legacy Open Field, Omnitech Electronics Inc., Columbus, OH). The number of beam breaks were recorded over 48 h to assess movement in two dark (active) cycles and one light (inactive) cycle.
Forelimb function was assessed using a test battery including both cylinder and horizontal ladder tests before and and up to 28 days after PT-induced stroke, in alignment with the guidelines from the International Stroke Recovery and Rehabilitation group to enhance clinical translation (Corbett et al., 2017). The mice were singly housed under normal conditions with a 12-h light cycle (7 am-7 pm) two weeks prior to beginning behavioural testing to measure stroke-induced deficits. Baseline tests were performed one week before inducing stroke. A minimum of 9 mice were analyzed per genotype (n = 9-16) for each behavioural test and experimental condition.
The cylinder test was used to assess spontaneous forelimb function using previously described protocols (Balkaya et al., 2013;Baskin et al., 2003). Briefly, the mouse was placed into a 10 × 15 cm glass beaker and was observed under red light for a minimum of 20 rears. The test was recorded using Ethovision XT 11 for the amount of time spent on each forelimb and the videos were scored blind to genotype. Percentage of time spent on the right forelimb was calculated using the following formula: [(Time on right forelimb)/(Time on right forelimb + Time on left forelimb)]*100.
The horizontal ladder test was performed using a protocol previously established by Farr et al., 2006(Farr et al., 2006. Briefly, training occurred over two consecutive days, performing 4-5 training trials per day. The test was performed using an elevated plexiglass ladder (69.5 cm × 15 cm) with irregularly spaced metal rungs spanning a clean cage on one side and their home cage on the other side. For each timepoint, the mice were video recorded as they crossed the ladder unassisted 3 consecutive times with the last two trials used for scoring. The recorded videos were analyzed blinded to genotype at 0.12× speed to examine forelimb placement, noting the number of 'successful steps', 'missteps' (foot faults) where the mouse did not grasp the ladder rung, as well as 'cheats' where the mouse used the walls of the apparatus for support to cross the ladder. Percentage of error was calculated for each animal using the following formula: [(Missed steps)/(Successful steps + missed steps + cheats)]*100.

Tissue collection
Brains collected for immunofluorescence were intracardially perfused with 4% paraformaldehyde (PFA, pH 7.4) in 0.1 M phosphatebuffered saline (PBS) and brains quickly dissected and fixed in 4% PFA overnight at 4 • C. The following day brains were washed with 0.1 M PBS then cryoprotected in 30% sucrose solution containing 1% sodium azide in 0.1 M PBS for 48 h, then placed in a 1:1 solution of 30% sucrose:OCT (VWR) overnight at 4 • C. For storage, the brains were flash frozen in an embedding mold at − 80 • C until required for sectioning. Brains collected for RNA or protein extracts, were quickly dissected after euthanasia, flash frozen in liquid nitrogen and stored at − 80 • C.

Immunofluorescence
Brains were sectioned using a microtome (SM 2010R; Leica) at a thickness of 40 μm. Serial sections were placed into nine wells of a 12well plate containing 1× PBS with 0.1% NaN 3 and were stored until staining. Free-floating sections were washed three times in 1× PBS before use. Antigen retrieval was performed by incubating the floating section at 80 • C in sodium citrate (pH 6) solution for 30 min and then cooled slowly to room temperature. After antigen retrieval, sections were washed three times in 1× PBS and placed into a blocking solution (10% horse serum, 0.1% Triton X-100, 0.1% Tween-20 in 1× PBS) for one hour at room temperature. Sections were incubated in primary antibody in a PBS solution (0.04% Triton X-100, 3 mg/ml bovine serum albumin in 1× PBS) overnight at 4 • C before washing and incubating for one hour at room temperature with the appropriate secondary antibody. All antibodies and the concentrations used are listed in Supplemental Table 1. Sections were then washed in 1× PBS three times, incubated in Hoechst stain (10 mg/ml; 62,249, ThermoFisher) diluted 1:2000 in 1× PBS for 10 min at room temperature and then washed twice in 1× PBS. Free floating sections were then mounted onto SuperFrost Plus slides (Fisher, 12-550-15) using immunomounting solution (S3023, Agilent Tech.) and coverslips for imaging. Images of immunofluorescent sections were acquired using a Zeiss LSM800 AxioObserverZ1 confocal microscope with both 10× and 20× objectives. Emission wavelengths of 488, 555 and 647 were used and images were acquired using optical zstack sectioning through ZEN Canada, 2009 acquisition software (Zeiss). Images were then exported into Adobe Photoshop for organization into final figures.

Protein isolation and immunoblots
Harvested brain tissue was suspended in protein lysis buffer (150 mM NaCl, 50 mM Tris, 5 mM EDTA, 1% NP40, 0.1% SDS in ddH 2 O), supplemented with protease inhibitor cocktail (Calbiochem, 539,134-1SET) and sheared in a TissueTearer to generate protein lysates. Lysates were then placed on ice for 15 min then centrifuged for 10 min at 4 o C at 13,000 RPM before collecting the protein supernatant. Proteins were separated on 4-12% Bis-Tris pre-cast gels (Invitrogen, NP0321BOX) by gel electrophoresis at 170 V for 80 min (XCell, Sure-Lock Mini-Cell, ThermoFisher) in 1× MOPS running buffer (Fisher, BP308-500). Proteins were then transferred onto PVDF membranes (BioRad, cat # 162-0177) for 70 min at 100 V in 1× transfer buffer (20% MeOH, 50 mM Tris, 40 mM Glycine). Membranes were washed 3 times in 1× PBS for 10 min each and blocked with 5% milk in TBS-T for one hour at room temperature. The membrane was then incubated in primary antibody (see Suppl. Table 1) in a 5% milk and TBS-T solution overnight at 4 • C and then washed in 1× TBS-T three times for 10 min each. Subsequently, the membranes were placed in HRP-conjugated species-specific secondary antibody in a 5% milk solution for one hour at room temperature and then washed in 1× TBS-T three times before being incubated in ClarityTM Western ECL Blotting Substrate (BioRad, cat # 170-5061) for 5 min. Protein signals were then detected by exposing membranes to film (Harvard Apparatus Canada, DV-E3012).
In addition to the commercial antibodies listed in Suppl. Table 1 we also generated a custom rabbit antibody raised against a KLHconjugated peptide corresponding to the C-terminal 62 amino acids of VGF (Cedarlane).

RNA isolation and RT-qPCR
Peri-infarct regions were dissected from adult mice using a Leica MZ95 stereomicroscope (Meyer). The cortical plug was removed and cortical tissue in adjacent areas between the infarct and lateral ventricle was dissected. The tissue was placed into 1 ml TRIzol (Life Technologies, cat # 15596018) and RNA isolation was performed as per the manufacturer's instructions and quantified using a NanoDrop™ (Thermo-Fisher, ND-1000) spectrophotometer. Total RNA (5 μg) was reverse transcribed using RevertAid Reverse Transcriptase (Thermo Fisher Scientific), and synthesized cDNA was further diluted 1:10 prior to use. qPCR analysis was carried out using the SensiFAST SYBR Lo-ROX Master Mix (FroggaBio Inc., cat # BIO-94020) under the following conditions: one cycle at 95 • C for 2 min, and then 40 consecutive cycles at 95 • C for 5 s, 60 • C for 30 s, and 72 • C for 20 s with an Agilent Stratagene Mx3000P machine. Primers used are listed in Supplemental Table 2. All primers were analyzed by melt curve analysis after qPCR amplification. The ΔΔCt method was used to compare fold-change. GAPDH and 18S mRNAs were used as normalizers in separate experiments.

Statistics
Statistical analyses were performed using Prism 6 (GraphPad). All data were reported as the mean ± SD. For cell counts, data was determined from a minimum of 5 field of views and 3 mice per genotype or treatment group. When two independent variables were analyzed, an unpaired t-test was used; dependent groups were analyzed using a paired t-test. Data with two variables were analyzed using a two-way analysis of variance (ANOVA), using a repeated measure design for behaviour measured over time. P < 0.05 was accepted as statistically significant and p-value reporting was as follows: *, p < 0.05; **, p < 0.01; ***, p < 0.001; ****, p < 0.0001.

Vgf cKO adult mice are smaller in size but have normal cortical development, brain mass, and patterns of locomotor activity
To examine the requirement for VGF in stroke recovery, we generated Vgf conditional knockout mice by obtaining Vgf fl/fl mice (Lin et al., 2015) and breeding them to the Nestin-Cre +/− driver line (Suppl. Fig. 1A). The Vgf fl/fl :Nestin-Cre +/− mice (herein referred to as Vgf cKO) Fig. 1. VGF cKO mice are smaller but lack motor deficits and display normal cortical development. A) Representative images of 8-week-old Vgf cKO mice (bottom) compared to control (WT) littermates (top). B) Plot of mean body weight (± SD) of P21 mice WT (grey) and Vgf cKO mice (blue). Each dot represents the weight of an individual mouse. ****, p < 0.0001. C) Representative brain images of WT and Vgf cKO P21 mice, and D) corresponding graph of mean brain weight (± SD) by genotype (WT, grey dot; Vgf cKO, blue dot) at the same age. NS = not significant. E) qRT-PCR expression of VGF and BDNF from 8-week old WT and Vgf cKO mice. Expression was normalized to Gapdh levels and plotted as Fold Change relative to the WT level. ****, p < 0.0001. F) Immunoblots of P0 cortical extracts showing VGF protein levels in WT and Vgf cKO samples. Actin is used as a loading control. G) Locomotor activity was plotted as the number of beam breaks each hour over 48 h (n = 6 WT, 7 cKO). Yellow regions highlight daytime hours while grey regions indicate evening hours. The X-axis shows the specific hour using the 24-h clock. H) P0 cortical sections from WT and Vgf cKO mice IF-stained with the cortical layer markers, Tbr1 (green, layer VI); Ctip2 (red, layer V) and Satb2 (green, layer II-IV). Slides were also stained with Hoeschst (blue) to counterstain cell nuclei (n = 3 WT, 3 KO).
Scale bar = 10 μm. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) were born at normal Mendelian ratios and were indistinguishable by size to their wildtype (WT) littermates at birth. By weaning age (postnatal day 21, P21) the Vgf cKO mice were significantly smaller in size (WT: 9.48 ± 0.37 g, n = 22; Vgf cKO: 5.76 ± 0.28 g, n = 12; t-test, two-tailed, p < 0.0001) with a 40% decrease in body weight (Fig. 1A, B). The Vgf cKO mice had a leaner body profile (Fig. 1A) that, upon dissection, was associated with no detectable body fat in their truncal region compared to WT littermates. The decreased size and leaner body profile of the Vgf cKO mice are consistent with findings from germline Vgf KO mice and in conditional Vgf mutants where a synapsin-Cre driver was used for neuronal-specific deletion (Hahm et al., 1999;Jiang et al., 2017a). At a gross level, the Vgf cKO mice showed no significant differences in brain mass or morphology compared to control littermates (Fig. 1C, D), which was also consistent with previous reports (Bozdagi et al., 2008;Hahm et al., 1999;Jiang et al., 2017a) and distinct from transgenic mice overexpressing Vgf that present with smaller brain weights and expanded lateral ventricles (Mizoguchi et al., 2017). Nonetheless, the smaller size of the Vgf cKO mice did result in a significant difference in brain weight:body weight ratio (Suppl. Fig. 1B).
To confirm that the Nestin-Cre driver resulted in effective Cre excision of the Vgf gene and subsequent loss of expression we performed RT-qPCR for Vgf and Bdnf transcripts using RNA isolated from cortical tissue of 8-week old mice. Vgf transcript levels were almost undetectable in the mutant mice while there was no significant difference in the level of Bdnf transcripts, despite a large variance between animals (Fig. 1E). Similarly, immunoblots for Vgf protein from Vgf cKO and WT P0 brain lysates confirmed that the protein was also ablated in the mutant animals (Fig. 1F).
Previous studies have indicated that Vgf KO mice have increased energy expenditure, some learning impairments, and succumb earlier in tests of forced depression or social defeat (Hunsberger et al., 2007;Jiang et al., 2018;Jiang et al., 2017a). Since this study was testing the effects of stroke in the Vgf cKO mice, we tested if the naïve Vgf cKO animals had any locomotor deficits that could confound the behavioural analysis of stroke recovery. Specifically, we measured the spontaneous activity of the mice for a 48-h period after being placed in a novel home cage environment (Tatem et al., 2014). Locomotor activity was plotted as the number of beam breaks each hour (Fig. 1G), or as movement time (Suppl. Fig. 1C). As expected, peak activity times occurred during the initial habituation period and during the evening hours, reflecting the nocturnal nature of the mice (Fig. 1G). Importantly, the Vgf cKO mice and WT animals showed no significant differences in locomotor activity (Fig. 1G, Suppl. Fig. 1C), supporting previous work that suggests Vgf does not alter normal rates of locomotor activity (Hunsberger et al., 2007;Jiang et al., 2017a).
Expression of the Nestin-Cre driver is first detectable at embryonic day 8.5 (E8.5) with high expression in the neural progenitors located in the ventricular zone (VZ) of the developing 6-layer laminar structure of the cortex (Berube et al., 2005). Vgf is widely expressed in the developing and adult rat brain including the cerebral cortex and hippocampus (van den Pol et al., 1994). To assess whether loss of Vgf in mice had any impact on corticogenesis, we performed immunohistochemical analysis to examine the expression of different layer-specific neuronal markers in the Vgf cKO and WT mice. Coronal sections from mutant and WT brains collected at E15 and P0 were stained using antibodies specific to the germinal layers, Pax6 (Ventricular zone, VZ) and Tbr2 (Subventricular zone, SVZ); or the cortical layers, Tbr1 (layer VI), Ctip2 (layer V), and Satb2 (layers II-V) to assess cortical lamination ( Fig. 1H; Suppl. Fig. 1D). Quantification of the proportion of marker-positive cells to the total number of cells within the cortex demonstrated no significant differences between Vgf cKO and WT littermates at either timepoint (Suppl. Fig. 1E, F). From these experiments we conclude that naïve Vgf cKO mice have no gross defects in brain size, cortical lamination, or deficits in locomotor activity. This finding suggested that the Vgf cKO mice would be a suitable model to examine the requirement for VGF in stroke recovery.

Vgf cKO mice have greater stroke-induced behavioural deficits and less recovery, accompanied by deficits in the ectopic migration of immature neurons
A PT-induced ischemic stroke was made in the left sensorimotor/ motor cortex of Vgf cKO mice and WT littermates (Watson et al., 1985). To exclude the possibility that the smaller size of the Vgf cKO mice might impact stroke size and location, we measured the infarct volume 24 h post ischemia in Vgf cKO mice compared to WT littermates. Representative MRI images demonstrated that the location of the stroke ( Fig. 2A) and volumetric calculations (Fig. 2B) were similar for the Vgf cKO and control mice (WT: 8.50 ± 0.73 mm 3 ; Vgf cKO: 8.32 ± 1.33 mm 3 ; n = 4; Mann Whitney test, 2-tailed, p-value = 0.886).
To assess functional recovery, 10-week-old Vgf cKO and control mice received baseline training and testing (BL-Be) on the cylinder and horizontal ladder test, one week prior to receiving a PT-induced stroke, as shown in Fig. 2C. As expected, the Vgf cKO and WT animals showed no differences in baseline testing for either the horizontal ladder (Fig. 2D) or cylinder tests (Fig. 2E), in alignment with no gross locomotor deficits in naïve Vgf cKO mice, as shown in Fig. 1G. Both of these behaviors were then re-analyzed at 1, 14, and 28 days post stroke (dps). In both tests, the affected limb is typically used less after the ischemic event resulting in asymmetry of limb use in the cylinder test and increased foot faults in the horizontal ladder (Balkaya et al., 2013;Farr et al., 2006). Both tests are also sensitive to examine recovery, with most animals showing marked improvement with increasing time after stroke (Balkaya et al., 2018).
One day post stroke, there was a significant increase in the number of foot misplacements in the Vgf cKO mice (34.1% error) compared to the WT (20.8% error) animals (Fig. 2D). A similar effect was observed in the cylinder test (Fig. 2E) with Vgf cKO mice utilizing their affected right paw significantly less (16.6% of the time) compared to control animals (28.7% of the time). Recovery, as assessed by change in improvement in performance over time after stroke induction, was observed in both the Vgf cKO mice and controls for both the ladder and cylinder task, yet in both tests the mice did not fully recover to pre-ischemic levels (Fig. 2D,  E). Thus, together these findings show the Vgf cKO mice had more acute stroke-induced behavioural deficits after stroke as well as deficits persisting throughout the 4-week recovery period.
Previous studies have indicated that VGF levels are increased within the tissue surrounding the infarct (Kury et al., 2004;Sakamoto et al., 2015). As such, we examined Vgf and Bdnf levels in the peri-infarct region 2 dps for increasing levels compared to control mice that did not receive a stroke (Fig. 2F). Consistent with previous results, we observed a massive 18.78-fold increase in Vgf levels in WT mice after stroke, while Vgf cKO mice showed essentially no change (1.14-fold increase) in Vgf transcript levels post stroke. Similarly, there was a 2.64-fold increase in Bdnf mRNA in the peri-infarct region of WT mice but no change in Bdnf levels in the Vgf cKO animals (1.13-fold increase). These findings confirm a striking induction of Vgf post PT-induced stroke that occurs concurrently with the significant behavioural deficits in the Vgf cKO mice.
To examine the generation and migration of SVZ-derived progenitor cells, we immunostained coronal sections of Vgf cKO and WT mice collected at 14 dps with doublecortin (DCX), a marker of immature neurons. In WT mice, we observed that 40% of DCX+ neurons were located in the SVZ while 60% were migrating through the ortex towards the infarct region (Fig. 2G, H; Suppl. Fig. 2). In contrast, in the Vgf cKO mice the vast majority of DCX+ cells were located in the SVZ with fewer (~20%) migrating towards the injury site. To determine why fewer DCX+ cells were found at the injury site, we examined sections for differences in proliferation and survival within the SVZ that could result in a secondary decrease in migration towards the infarct. Sections were co-stained with antibodies to DCX and Mcm2, a marker of the prereplicative complex that forms on origins of replication early in the G1 phase (Stoeber et al., 2001) to examine whether there was a (caption on next page) H.L. Gillis et al. proliferation deficit. While we observed a slight decrease in Mcm2+ proliferating cells in the SVZ of the Vgf cKO mice, there remained an increased number of DCX+ neurons, in particular, more cells that were DCX+/Mcm2-(Suppl. Fig. 3). Differences in neuronal cell death were assessed by staining with cleaved caspase 3 antibody. We observed very few labeled cells and no differences were detected in either genotype (data not shown). Collectively, these findings suggest that while there might be a slight defect in the proliferation of SVZ progenitor cells, VGF   Fig. 2. VGF cKO mice exhibit reduced migration of DCX+ newborn neurons after photothrombotic stroke. A) MRI images of WT and Vgf cKO mice 24 h after PT-induced stroke, highlighting the location of the infarct (bright region). B) Infarct volumes were calculated from MRI images and plotted as the mean infarct size ± SD (n = 4 WT, 4 KO). C) Schematic diagram showing the design of the PT-induced stroke experiment indicating timepoints and procedures. BL-Be, Baseline behaviour; PT, photothrombosis; Be, Behaviour (n = 9 WT, 9 KO). D) Foot misplacements of WT and Vgf cKO mice were scored as % error in the Horizontal Ladder behavioural test. E) Spontaneous use of the affected forelimb was scored as a percentage of total forelimb use in the Cylinder behaviour test. F) qRT-PCR expression of VGF and BDNF from RNA isolated from the peri-infarct and SVZ at 2 dps. Expression was normalized to Gapdh levels and plotted as fold change relative to the WT pre-stroke level. For all graphs, error bars represent standard deviation; *, p < 0.05; **, p < 0.01; ***, p < 0.001; ****, p < 0.0001; ns = not significant. G) Representative images of coronal brain sections from 8-week-old WT and Vgf cKO mice after stroke and immunostained for DCX (red) 14 dps. Nuclei were counterstained with Hoechst (blue). Dotted box in first and second panels indicate the location of the magnified images in the subsequent panels. Dotted line shows the infarct location. Scale bar = 1000 μm (left panel), 100 μm (other panels). CC, cerebral cortex; LV, lateral ventricle. H) The total number of DCX+ cells was quantified and plotted (mean ± SD) as a percentage located within the subventricular zone (SVZ) or the corpus callosum (CC) (n = 3 WT, 3 KO). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) Fig. 3. Exogenous VGF treatment improves functional recovery of forelimb use after stroke. A) Representative images of cortical sections from WT adult mice immunostained with VGF (red) and counterstained with Hoechst (blue) 12 days after control adenovirus (CtrlAd) or VGF adenovirus (AdVGF) injection and 14 dps (n = 3). White box insert showing schematic drawing of coronal adult brain section with red box indicates SVZ region imaged. Scale bar = 100 μm. B) Schematic of experimental design of PT-induced stroke with CtrlAd or AdVGF treatment on a cohort of WT mice (n = 14 CtrlAd, n = 16 AdVGF). C) Foot misplacements of CtrlAd and AdVGF treated WT mice were scored 14-and 28-dps for % error in the Horizontal Ladder test. D) Spontaneous use of the affected forelimb was scored as a percentage of total forelimb use in the Cylinder test. Both graphs show mean ± SD; *, p < 0.05, ***, p < 0.001). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) is critical for the efficient migration of adult-generated immature neurons from the SVZ towards the site of damage.

Exogenous VGF improves cellular and behavioural metrics of post stroke recovery
We reasoned that if VGF was required to limit post stroke behavioural deficits and promote functional recovery afterward then exogenous Vgf delivery to the Vgf cKO mice should facilitate the recovery process. Moreover, it also provided the opportunity to test whether exogenous VGF treatment post stroke could enhance stroke recovery in WT animals. Previous work in our lab utilized an adenoviral vector containing a VGF expression cassette (AdVGF) and demonstrated that AdVGF injection into Smarca5 cKO mice resulted in increased plasma levels of VGF and prolonged the survival of the mutant animals (Alvarez-Saavedra et al., 2016). We tested new AdVGF virus stocks in HEK293T cells and observed VGF expression infected with AdVGF but not an adenoviral vector containing an empty expression cassette (CtrlAd; Suppl. Fig. 3A). We also performed tail vein injections of AdVGF into WT mice and observed VGF expression in liver cryosections at 2-and 7-days after injection (Suppl. Fig. 3B), as expected since the vast majority of injected Ad virus is targeted to the liver (Alvarez-Saavedra et al., 2016;Vrancken Peeters et al., 1996).
An initial cohort of WT mice was next used to test the effects of Adtreatment on animal survival, VGF expression in the brain, and behaviour responses following stroke. We observed no adverse effects after delivering the Ad virus by tail vein injections two days post stroke. Since VGF expression in the brain after injection of AdVGF is more likely a result of circulating VGF rather than viral infection within brain tissue (Alvarez-Saavedra et al., 2016;Vrancken Peeters et al., 1996), we examined the cells lining the lateral ventricle for uptake of circulating VGF protein or C-terminal peptides. WT control mice were randomized to be treated with CtrlAd or AdVGF 2 dps and brain sections were collected 14 dps and examined for VGF expression. Sections immunostained with a C-terminal VGF antibody clearly detected VGF protein/ peptides concentrated around the lateral ventricles in mice that received AdVGF treatment, but not in animals injected with the CtrlAd (Fig. 3A). This suggested that VGF is circulating in the brain and therefore we tested the effect of AdVGF injection compared to CtrlAd treatment following stroke.
Using a similar experimental design, WT mice underwent baseline cylinder and horizontal ladder testing, received CtrlAd or AdVGF virus injection 2 dps, followed by testing at 14 and 28 dps (Fig. 3B). Unexpectedly, WT mice treated with AdVGF had better recovery after stroke compared to WT mice treated with the CtrlAd (Fig. 3C, D). Specifically, the WT mice treated with AdVGF had a 6% improvement in the error rate on the horizontal ladder (14 dps: CtrlAd, 15.62% error; VGFAd, 9.42%; 28 dps: CtrlAd, 12.05% error; VGFAd, 6.05%) and approximately 8% increase in the use of the right forepaw in the cylinder test (14 dps: CtrlAd, 33.14% usage; VGFAd, 40.61%; 28 dps: CtrlAd, 39.38% usage; VGFAd, 47.73%; 2-way ANOVA, multiple comparisons) over both time points assessed. Moreover, the behaviour results for the CtrlAd treated mice were very similar to the stroke recovery observed in the previous experiment indicating that the virus treatment was not having any negative effects on recovery (compare CtrlAd in Figs. 3C, D to WT in Fig. 2D, E).
Encouraged by these results, we next tested whether the AdVGF treatment could rescue the deficits observed in the Vgf cKO animals. Vgf cKO mice were treated with CtrlAd or AdVGF and compared to WT mice treated with CtrlAd. Vgf cKO mice treated with AdVGF were sacrificed two days after adenovirus injection and confirmed VGF expression in liver sections (Suppl. Fig. 3C). Next, we assessed the behavioural recovery at 2-and 4-weeks post stroke. In the horizontal ladder test, the Vgf cKO mice injected with AdVGF showed a significant improvement in the number of foot misplacement errors with an average of 13.6% improvement over both timepoints analyzed (cKO: 26.78% and 19.07% error at 2-and 4-weeks; cKO + AdVGF: 12.04% and 6.56% error; n = 9; 2-way ANOVA, multiple comparisons; Fig. 4B). Indeed, the recovery was equivalent to that observed for WT mice that received CtrlAd (WT: 15.62% and 12.05% error at 2-and 4-weeks; n = 14; 2-way ANOVA, multiple comparisons; Fig. 4B). Similar results were obtained for the cylinder test. The VGF cKO mice that received AdVGF showed an average increase of 12.8% in the use of their affected forelimb at both timepoints (cKO: 25.91% and 29.79% usage at 2-and 4-weeks; cKO + AdVGF: 33.63% and 44.57% usage; n = 9; 2-way ANOVA, multiple comparisons; Fig. 4C). Moreover, the Vgf cKO mice that received AdVGF showed an equivalent recovery to the WT animals treated with CtrlAd (WT: 33.14% and 39.38% usage at 2-and 4-weeks; n = 14; 2-way ANOVA, multiple comparisons; Fig. 4C). These results indicate that AdVGF treatment significantly improves recovery of behavioural deficits from PT-induced stroke in Vgf cKO mice.
Following behaviour testing, the animals from each treatment group were sacrificed at 4 weeks for immunohistochemistry to assess the production and migration of immature neurons (Fig. 4D). VGF treatment increased the percentage of migrating DCX+ neurons in the Vgf cKO animals compared to CtrlAd treated Vgf cKO mice (Fig. 4D, E). Importantly, the AdVGF treatment restored the migration of DCX+ neurons towards the infarct region in the Vgf cKO mice, which reflected what was observed for the WT mice (Fig. 4D, bottom panels). Despite the increase in migrating DCX+ neurons in the AdVGF treated mice, the percentages remained 50% lower than the number of migrating DCX+ cells observed in WT mice (Fig. 4E). These experiments demonstrate that migration of immature neurons towards the infarct region is improved after delivery of AdVGF in the Vgf cKO mice. Collectively, AdVGF treatment was efficacious in reducing behavioural deficits and restoring the mobilization of immature neurons in the Vgf cKO mice.

Discussion
VGF is a multi-functional neurosecretory protein that is released from neuronal dense core vesicles and proteolytically processed into >10 active peptides. VGF levels serve as a biomarker for neurodegenerative diseases, including Alzheimer's disease (AD), and some psychiatric disorders such as bipolar and depressive disorders (Quinn et al., 2021). Recent studies have shown that VGF-derived peptide treatment has many positive effects on brain function. VGF can reduce neurodegeneration in a mouse model of AD, reduce anxiety-and depressionlike behaviors, improve cognitive function, neuronal maturation and neuritogenesis, and provide neuroprotection after optic nerve crush injury (Beckmann et al., 2020;Lin et al., 2021;Moutinho et al., 2020;Sato et al., 2022;Takeuchi et al., 2018). A role for VGF in promoting stroke recovery has also been suggested. Two independent studies screening for endogenous factors important for stroke recovery identified VGF as an upregulated gene in the cortical tissue surrounding the infarct following PT-induced stroke (Kury et al., 2004;Sakamoto et al., 2015). Moreover, VGF C-terminal peptides show sequence similarity with the complement protein C3a which has further increased interest, particularly since C3a treatment can promote stroke recovery when initiated 7 days after stroke (Stokowska et al., 2017). In this study, we confirmed that VGF is normally upregulated in the peri-infarct region after stroke and demonstrate the in vivo requirement of VGF during recovery following PT-induced stroke in Vgf cKO mice. Ablation of Vgf in the mice was associated with an increase in stroke-induced behavioural deficits, in the absence of size differences of the lesion. This finding correlated with a lack of Vgf and Bdnf upregulation and impaired migration of stroke-induced immature DCX+ neurons migrating towards the infarct. Exogenous delivery of VGF using an adenovirus vector was able to reduce the behavioural deficits and restore migration of the immature neurons towards the infarct suggesting that VGF presents as a promising therapeutic to promote stroke recovery.
Despite identifying a requirement for VGF, many questions remain including what cells secrete it, which peptides are key to the recovery Fig. 4. AdVGF post stroke treatment of Vgf cKO mice rescues newborn neuron migration and improves functional recovery. A) Schematic of experimental design of PT-induced stroke with CtrlAd or AdVGF treatment on cohorts of Vgf cKO and compared to WT mice treated with CtrlAd (n = 14 WT + CtrlAd; 9 Vgf cKO + CtrlAd; 10 Vgf cKO + VGF-Ad). B) Quantification of % error in Horizontal Ladder test at baseline, 2-and 4-weeks post-stroke. C) Quantification of mean time ± SD spent on affected forelimb at baseline, 2-and 4-weeks post stroke. E) Representative images of coronal brain sections from mice immunostained for DCX (red) 28-dps, with nuclei counterstained with DAPI (blue). Scale bar = 100 μm. All graphs show mean ± SD, *, p < 0.05; **, p < 0.01; ****, p < 0.0001. E) The percentage (mean ± SD) of migrating DCX+ cells within the CC (n = 3 WT, 3 KO). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) process, and what specific functions do these peptides have during recovery. Previous studies have suggested that Vgf is primarily expressed and secreted from neurons, although in the retina Vgf expression was prevalent in Muller glia cells and astrocytes after nerve crush injury (Quinn et al., 2021;Takeuchi et al., 2018). We observed increased levels of Vgf mRNA in tissue collected from the peri-infarct region, although it remains unclear whether the source is neuronal (either healthy or damaged) or, arises from the influx of microglial and astrocytes to the injury site.
Following stroke, the Vgf cKO mice showed a reduction in the number of immature DCX+ neurons migrating to the infarct. While we observed a decrease in the proportion of proliferating progenitors (Mcm2+) in the SVZ, there was a large increase in DCX+/Mcm2-cells suggestive of a defect in their mobilization from the SVZ to the infarct. The migration was dependent on VGF since exogenous delivery was able to rescue cell migration. These data suggest that VGF is a potential factor for improving stem cell or immature neuron cell mobilization and is the first data to show the requirement for VGF for the SVZ-derived neurogenic response to stroke. Delivery of VGF or the VGF-derived peptide TLQP-62 has been previously shown to enhance hippocampal neurogenesis in vitro and in vivo (Thakker-Varia et al., 2007). This finding raises the question of whether VGF and its peptides are required for hippocampal neurogenesis in the naïve and stroke condition. Additionally, it remains to be tested if the effect of VGF on migration is restricted to the peri-infarct region, or also occurs when the SVZ-derived cells migrate to the olfactory bulb. Lastly, our study utilized AdVGF to express the full-length VGF, thus it remains to be determined which VGF peptide promotes the immature neuron migration. The fact that both the WT and Vgf cKO mice had improved recovery after stroke accompanied by the increase in migration of immature neurons make it tempting to speculate that the immature neurons have a functional role, either directly through integration, or indirectly through the microenvironment, to improve stroke recovery. It should also be noted that a limitation of this study was that it was performed using young healthy animals. A similar study using older mice and/or confounding lifestyle stressors may further elucidate the importance of VGF peptides to stroke recovery.
Given the large number of peptides that can be generated from a single VGF protein, we suspect that the increase in migration of the immature neurons is likely only one aspect that VGF and its peptides effect in establishing a reparative micro-environment during recovery. Indeed, other studies have identified distinct brain reparative effects of VGF peptides. The VGF peptide TLQP-21 was shown to be chemotactic for microglia (Elmadany et al., 2020). This study also noted differential effects of TLQP-21 on microglial activation depending on whether it interacted with the C3aR1 or gC1q receptors, the former stimulating migration and phagocytic activity but the latter reducing such activities (Elmadany et al., 2020). Increased phagocytic activity of microglia is a well-established event necessary for repair. Similarly, we previously demonstrated that the C3aR1 receptor was critical to limit gliosis and granule neuron death in a developmental brain injury model in which survival was improved after VGF treatment (Alvarez-Saavedra et al., 2016;Young et al., 2019). Another study that explored the use of a C3aR antagonist (SB 290157) demonstrated a reduced number of inflammatory microglia following stroke suggesting that SB 290157 may be used to limit neuronal death (Surugiu et al., 2019). These studies raise the idea that VGF peptide -C3aR interactions may block phagocytic activity. In an unrelated study in which the Na/H exchanger (NHE1) gene was ablated in microglia to enhance phagocytic function, overall improvements in cognitive function were observed after stroke that were linked to improved synaptic remodeling, oligodendrogenesis, and remyelination (Song et al., 2022). These studies highlight the importance of microglia to recovery after stroke and the involvement of the known receptors for VGF peptides. Aside from microglial activation, other studies have shown that TLQP-21 and other C-terminal VGFderived peptides can limit neuronal death (Alvarez-Saavedra et al., Takeuchi et al., 2018), promote remyelination (Alvarez-Saavedra et al., 2016), and enhance dendritic arborization (Moutinho et al., 2020); each an important activity for post-stroke recovery. Finally, the synergistic trophic effects that VGF peptides invoke through a VGF-BDNF positive auto-feedback loop must also be considered a key factor in generating the repair environment for post stroke recovery (Lin et al., 2015). Collectively, our data coupled with other studies are beginning to forge an important role for VGF and its peptides in the recovery process following stroke.
The potential to exploit VGF as a future treatment for stroke remains a valid possibility. In this study, systemic delivery of AdVGF largely resulted in VGF expression from the liver. Other studies on AD mouse models introduced VGF peptides via intracerebral injection, or by genetic manipulation, both undesirable delivery methods for human trials. Further work is required to not only identify the proper timing, dose, and/or optimal peptide to use for treatment, but also to generate a less invasive delivery method. Recent advances include the development of VGF liposomal nanoparticles that have shown promise for VGF delivery to the brain (Arora and Singh, 2021). Similarly, intranasal administration of C3a peptides have reduced neurodegeneration in a hypoxicischemia injury model (Pozo-Rodrigalvarez et al., 2021), offering hope that a similar approach could be used with VGF peptides.
Overall, this study has defined a requirement for VGF in post stroke recovery and demonstrated the potential efficacy of exogenous delivery of VGF shortly after stroke. Despite a clear requirement for VGF, one of the limitations of this study was that no clear mechanistic insight was determined leaving many avenues open to further studies (eg. microglia activation, synaptic remodeling, or remyelination). Nonetheless, these findings should increase the enthusiasm for defining the mechanisms involved and for the further development of novel VGF therapeutic strategies.

Declaration
On behalf of all authors, I, David Picketts the corresponding author confirm that our submission to Experimental Neurology entitled "VGF is required for recovery after focal stroke" is not under consideration for publication elsewhere, that its publication is approved by all authors, and by the responsible authorities at the University of Ottawa and the Ottawa Hospital Research Institute where our research study was conducted.

Data availability
Data will be made available on request.