Overexpression of Slit 2 decreases neuronal exitotoxicity, accelerates glymphatic clearance and improves the cognition in multiple microinfarcts

Background: Cerebral microinfarcts (MIs) lead to progressive cognitive impairment in elderly, but the mechanism has kept unknown. Dysfunction of GABAergic transmission induces excitotoxicity, which contributes to stroke pathology, but the mechanism has kept unknown. The secreted leucine-rich repeat (LRR) family protein slit homologue 2 (Slit 2) upregulates GABAergic activity and protects against global cerebral ischemia, but the neuroprotective ecacy of Slit 2 against MI has not been examined. Methods: Middle-aged Wild type and Slit 2-Tg mice were divided into sham and MIs groups. MIs in parietal cortex of was induced by laser-evoked arteriole occlusion. Spatial memory was examined by Morris water maze, neuronal activity and glymphatic clearance in peri-infarct areas were monitored by two-photon imaging. GABAergic transimission and neuroinammation were detected by immunouorescent staining or western blotting. Results: MIs increased the intracellular Ca 2+ amplitude and frequency, decreased the neuron survival and neuronal connectivity of parietal cortex, decreased the GABAergic transmission, induced the neuroinammation, impaired the glymphatic clearance and cognition in middle-aged mice. Slit 2 overexpression attenuated dysfunctional neuronal Ca 2+ signaling, protected against the neuronal death in the peri-infarct area as well as loss of parietal cortex connectivity, increased the GABAergic transmission and attenuated the neuroinammation, improved glymphatic clearance and eventually improved spatial learning and memory. Conclusion: Our results strongly supported overexpression of Slit 2 protected against the dysfunction in MIs, which is a potential therapeutic target for cognition impaiment in the elderly.


Introduction
Cerebral microinfarct (MI) is wedge-shaped ischemic lesion that result from occlusion of penetrating arteriole (Luo, et (Hartmann DA, et al. 2018). Accumulating evidence suggests that MI produce persistent brain in ammation (Sofroniew and Vinters, 2010) and disorganized axonal structure in both subcortical (Hinman et al., 2015) and cortical tissues (Coban et al., 2017), thereby expanding regional injury and dysfunction. However, MI is di cult to detect in living human brain and the extent of these lesions is only revealed by postmortem histological examination (Arvanitakis Z, et al. 2011). Population aging is currently increasing the global dementia burden, so it is critical to understand the etiology and pathophysiology of MI to aid in the development of effective and safe preventative treatments.
Excessive release of glutamate and concomitant overstimulation of glutamate receptors during and following ischemic stroke (termed excitotoxicity) induces both acute and delayed neuronal death due to excessive calcium in ux, oxidative stress, degradation of macromolecules, and activation of apoptotic pathways (Lai TW, et al. 2014). Release of gamma-aminobutyric acid (GABA) can counteract glutamate excitotoxicity by inhibiting glutamatergic transmission at presynaptic sites and counteracting glutamatemediated depolarization in postsynaptic neurons, thereby reducing intracellular calcium deregulation and downstream processes leading to neuronal death (Louzada PR, et al. 2004). Indeed, augmenting GABAergic transmission can protect against ischemic damage (Costa C, et al. 2004; Bi M, et al. 2017).
However, it is uncertain whether glutamate/GABA imbalance and ensuing excitotoxicity contributes to MI pathology during aging. In addition, research on excitotoxic neuronal damage in cerebral ischemia has focused mainly on the dynamics of excitatory mediators, and much less is known regarding the changes in GABAergic activity (Schwartz-Bloom and Sah, 2001).
The secreted leucine-rich repeat (LRR) protein slit homologue 2 (Slit 2) regulates the migration, development, and axonal path-nding of GABAergic interneurons by stimulating roundabout (Robo) receptors (Andrews W, et al. 2006). In addition to regulating GABAergic neuron development and circuit formation, recent studies have also implicated Slit 2 signaling in cellular senescence (Gupta KP, et al. 2015) and improved glymphatic clearance (Ge Li, et al. 2018) as well as inhibition of neuroin ammation and protection against global cerebral ischemia (Altay T, et al. 2007). In the present study, we examined the effects of Slit 2 on MI induced by two-photon irradiation in aging mouse brain.

Animals
Transgenic mice overexpressing human Slit 2 (slit 2-Tg) (Li G, et al. 2018) were generated and supplied by the Institute of Biochemistry and Cell Biology (CAS, Shanghai, China) (Han HX, et al. 2011). These slit 2-Tg mice were bred at Guangdong Animal Centre (Guangzhou, China). Wild type (WT) C57BL/6J mice were supplied by Guangdong Animal Centre (Guangzhou, China). Mice at 14 months of age were randomly divided into sham and microinfarct (MI) groups. Animals in sham groups received the same surgery except for irradiation by a femtosecond laser to induce microinfarct formation.

Induction of microinfarcts
Anesthesia was induced with 5% iso urane and maintained by 2.5% iso urane in oxygen, a 2 × 2 mm 2 cranial window was created using a microdrill over the right parietal cortex (Nishimura N, et al. 2010). Fluorescein isothiocyanate-dextran (FITC-d2000, 1.5% in saline) was injected into the tail vein to image the neurovasculature through a 25 × water immersion objective lens. Five penetrating arterioles (PAOs) of 20-25 µm diameter were selected as targets for photobleaching-induced clotting as described (Luo CM, et al. 2018, Nishimura N, et al. 2006). Successful microinfarct induction was con rmed at 24 h postocclusion by two-photon imaging of target PAOs.

Morris water maze
Morris water maze testing was performed 1 week after MI modeling as described (He XF, et al. 2016).
Brie y, mice were rst examined for spatial learning during ve consecutive days of hidden platform training with four trials per day. On day six, the platform was removed and each mouse was tested for spatial memory on a single 60-s probe trial. Swim paths were recorded, the latency to reach the platform during water maze training as well as the number of crossings over the former platform location (target area) and time spent in the target quadrant during the probe trial were analyzed. Two-photon Ca 2+ imaging Two weeks after MI induction, the incision was re-opened and the agarose and coverslip over the cranial window were removed, intracellular Ca 2+ imaging was performed on the region surrounding the infarct area ( Fig. 2A & B) using a two-photon microscope (Leica, Wetzlar, Germany) as described previously (He XF, et al. 2015). Brie y, peri-ischemic target cells were stained with Oregon Green 488 BAPTA-1 AM (OGB-1 AM) at a concentration of 10 mM by multicell bolus loading (MCBL) (Garaschuk, et al. 2006

Assessment of the glymphatic e ciency
Glymphatic clearance were evaluated as described (Ge Li, et al. 2018). Brie y, the animals were anesthetized, the incision was re-opened, FITCconjugated dextran (40 kDa) dissolved in ACSF was injected into the subarachnoid space via cisterna magna puncture with a microsyringe pump controller, and 200 µL of rhodamine B (70 kDa) was injected intravenously immediately prior to imaging. Images were acquired 5, 15, 30, 45, and 60 min following intra-cisternal FITCconjugated dextran injection. Mean pixel intensity of the FITC-tracer in the paravascular space was quanti ed to evaluate clearance by the glymphatic system and BBB permeability, respectively.

Biotinylated dextran amine (BDA) injection and measurements of axon density
To investigate the effect of microinfarcts on axonal connectivity, 0.5 µL BDA solution (5% in 0.1 M PBS) the neuroanatomical tracer biotin dextran amine (BDA, MW 10000) was injected into the ipsilateral (exposed/injured) right parietal cortex and imaged in three cortical target regions. Animals were perfused through the heart 2 weeks later, and brains were xed, frozen, and coronally sectioned at 10 µm. Six sections spaced 100 µm apart were stained with Alexa Fluor® 488 Streptavidin, embedded in Fluoroshield™ containing DAPI for nuclear counterstaining and enclosed under a coverslip.

Histology
Sections were treated with 0.3% Triton and 10% goat serum for 1 h at room temperature, then incubation overnight at 4 °C with the indicated primary antibody. Sections were then incubated with the indicated secondary antibodies at room temperature in PBS containing 10% normal goat serum for 1 h. Slices were mounted onto slides, embedded in Fluoroshield™ with DAPI, and enclosed under a coverslip. Images were acquired using a Nikon uorescence microscope or a confocal microscope equipped with a 63× (N.A.

1.25) glycerol immersion objective.
Western blot analysis 20 µg total protein per lane separated by SDS-PAGE using 12% precast polyacrylamide gels at 120 V for 90 min. Separated proteins were then transferred to polyvinylidene uoride membranes at 100 V for 2 h. Membranes were blocked with 5% BSA at room temperature for 1 h and incubated with the indicated primary antibodies overnight at 4 °C, followed by incubation with anti-rabbit or anti-mouse immunoglobulin G secondary antibody for 1 h.

Data and statistical analyses
Data were analyzed by an experimenter blinded to treatment history. All data are expressed as mean ± standard deviation. Immunohistochemical staining and western blotting were analyzed using ImageJ. Mean Slit 2 expression levels on western blots were compared by independent-samples t test while other group means were compared by two-way repeated measures ANOVA with Tukey's post hoc tests for multiple comparisons. All statistical analyses were conducted using SPSS 19.0. A P < 0.05 (two tailed) was considered statistically signi cant for all tests.

Results
Slit 2 was overexpressed in neurons and astrocytes but not in microglia of transgenic mice Western blotting was performed to con rm expression of the human Slit 2 transgene protein in transgenic (Tg) mouse brain. Expression was signi cantly elevated in slit 2-Tg mice compared to WT mice (P < 0.001) ( Supplementary Fig. 1A & B). Co-immuno uorescence staining using anti-Flag for detection of Slit 2 and cell type-speci c antibodies revealed overexpression in neurons ( Supplementary Fig. 1C) and astrocytes ( Supplementary Fig. 1D) but not in microglia ( Supplementary Fig. 1E).
Overexpression of Slit 2 improved Morris water maze performance in mice with parietal microinfarcts The posterior parietal cortex (PPC) is a multimodal association area involved in spatial navigation as evidenced by performance de cits in the MWM following PPC lesions (Olsen GM, et al. 2017). To explore the protective e cacy of Slit 2 against cognitive dysfunction due to PPC microinfarcts in aged (14month-old) mice, we compared MWM performance between WT and slit 2-Tg mice following sham treatment or MI induction (Fig. 1A). As shown in Fig. 2B, on days 4 and 5 during training, there were no signi cant differences in escape latencies between WT sham and slit 2-Tg sham groups (Both P > 0.05), but the escape latencies in slit 2-Tg MIs mice exhibited shorter than WT MI mice (P < 0.05 and P < 0.001, respectively). These ndings suggest that Slit 2 overexpression protects against spatial learning impairment following MIs induction.
During the probe trial (Fig. 1C & D), the number of crossings was signi cantly reduced in MIs group compared with sham group, for WT mice (P < 0.05) but not for slit 2-Tg mice (P > 0.05), which was signi cantly reduced in the WT MIs group compared with slit 2-Tg MIs group (P < 0.05). Similarly, the target quadrant time was signi cantly decreased in MIs group compared with sham group, for WT mice (P < 0.05) but not for slit 2-Tg mice (P > 0.05), which was shorter in WT MIs mice than WT MIs mice (P < 0.05). These results suggest that MIs cause spatial memory de cits that can be improved by Slit 2 overexpression.
Overexpression of Slit 2 inhibited neuronal hyperactivation in the peri-infarct area Two-photon Ca 2+ imaging in the peri-infarct area was performed two weeks after MI induction to assess excitatory-inhibitory balance ( Fig. 2A & B). Amplitude was signi cantly greater in MIs mice than sham group, for WT mice (P < 0.001) (Fig. 2C, D, G), but not for slit 2 Tg mice (P > 0.05) (Fig. 2E, F, G). Frequency was signi cantly greater in MIs group compared to sham group, for WT mice (P < 0.05) (Fig. 2C, D, G), but not for slit-2 Tg mice (P > 0.05) (Fig. 2E, F, G). Collectively, these ndings indicate that Slit 2 overexpression suppress neuronal Ca 2+ transients in the peri-infarct area.
Overexpression of Slit 2 protected against axonal damage after multiple cortical microinfarct induction The anatomical tracer BDA was injected into right parietal cortex and assessing transport to ipsilateral hippocampus, ipsilateral entorhinal cortex, and contralateral parietal cortex (Fig. 3A). The BDA-positive cell number in ipsilateral hippocampus was signi cantly higher in slit 2-Tg MIs mice compared to WT MIs mice (P < 0.05) (Fig. 3b1 & C). BDA-positive cell number in ipsilateral entorhinal cortex was signi cantly lower in WT MIs mice compared to slit 2-Tg MIs mice (P < 0.05) (Fig. 3b2 & C). Finally, BDA-positive cell number in contralateral parietal cortex was signi cantly higher in slit 2-Tg MIs mice compared to WT MIs mice (P < 0.05) (Fig. 3b3 & C). These results indicating disruption of cortical projections. However, Slit 2 overexpression preserved these projections following MIs induction.
Overexpression of Slit 2 increased GABAergic transmission in the peri-infarct area Effect of Slit 2 overexpression on excitatory-inhibitory balance in peri-infarct areas were measured the immuno-expression of the GABAergic interneuron markers GAD67 (Fig. 4A) and vesicular GABA transporter (VGAT) (Fig. 4B). The number of GAD67-positive neurons surrounding the infarcts area was signi cantly lower in MIs group compared to sham group, for WT mice (P < 0.05), but not for slit 2-Tg sham mice (P > 0.05) (Fig. 4C). VGAT intensity was signi cantly lower in MIs group than sham group, for WT mice (P < 0.05), but not for slit 2-Tg mice (P > 0.05) (Fig. 4C). Finally, we performed western blotting to verify the expressions of GAD67 and VGAT in peri-infarct areas (Fig. 4D & E). GAD67 expression was signi cantly lower in WT MIs mice compared to WT sham mice (P < 0.01) but did not differ between slit 2-Tg MIs and slit 2-Tg sham mice (P > 0.05). Furthermore, GAD67 expression was signi cantly lower in WT MIs mice than slit 2-Tg MIs mice (P < 0.01). VGAT expression was signi cantly lower in WT MIs mice than WT sham mice (P < 0.01) but did not differ between slit 2-Tg MIs and slit 2-Tg sham mice (P > 0.05). Moreover, VGAT expression was signi cantly lower in WT MIs mice than slit 2-Tg MIs mice (P < 0.01).

Overexpression of Slit 2 attenuated peri-infarct neuroin ammation and protected against local neuronal loss
Neuronal number was signi cantly reduced in MIs groups compared to corresponding sham groups, both for WT mice(P < 0.0001) and slit 2-Tg mice (P < 0.001) (Fig. 5A & B), but signi cantly greater in slit 2-Tg MIs mice compared to WT MIs mice (P < 0.001). Microglial number was signi cantly greater in both MIs groups compared to the corresponding sham controls, both for WTs mice (P < 0.0001) and slit 2-Tg mice (P < 0.01), but signi cantly lower in slit 2-Tg MIs mice compared to WT MIs mice (P < 0.001). Therefore, Slit 2 overexpression appears to protect neurons for MI-induced degeneration, possibly by quelling the ensuing neuroin ammatory response.
Overexpression of Slit 2 improved glymphatic clearance and increased the astrocytic AQP4 polarity As shown in Fig. 6A, following intra-cisternal injection, FITC-dextran tracer moved along the paravascular space and rapidly entered the interstitium of the parenchyma. The FITC intensity at 5 min showed no signi cant differences among the WT sham, WT MIs, slit 2-Tg sham and slit 2-Tg MIs groups (Fig. 6B &  C). In all groups, parenchymal/perivascular FITC-dextran uorescence intensity gradually increased over the rst 45 min after injection. Thereafter, intensity continued to increase in the WT MIs group, indicating dysfunction of glymphatic clearance, but decreased in the other three groups. At 60 min after FITCdextran injection (Fig. 6C), FITC intensity was signi cantly higher in MIs mice compared to sham mice, both for WT mice (P < 0.01) and slit 2-Tg mice (P < 0.05), Further, FITC intensity was signi cantly lower in slit 2-Tg MIs mice compared to WT MIs mice (P < 0.05) (Fig. 6C), suggesting that Slit 2 overexpression sustained glymphatic clearance following MIs induction.
Reactive astrocyte number was signi cantly greater in peri-infarct areas of MI groups compared to corresponding sham groups, both for WT (P < 0.001) and Slit 2-Tg mice (P < 0.05) (Fig. 6D & E). Astrocyte number was signi cantly lower in slit 2-Tg MIs mice compared to WT MIs mice (P < 0.01), suggesting that Slit 2 overexpression suppressed astrocyte reactivity in response to MIs induction. AQP4 polarity was signi cantly lower in MIs groups compared to corresponding sham groups, both for WT mice (P < 0.001) and slit 2-Tg mice (P < 0.01) and signi cantly greater in slit 2-Tg MIs mice compared to WT MIs mice (P < 0.05). These results indicate that MIs induction impaired glymphatic clearance and disrupted astrocytic AQP4 expression function, which was protected against by Slit 2 overexpression.

Discussion
The ensuing imbalance between excitatory and inhibitory transmission is the primary pathogenic mechanism for cell death in stroke (Costa C, et al. 2004), and inhibiting the excitotoxicity protects against the brain damage. Numerous inhibitors of glutamatergic transmission have been tested as neuroprotective agents following experimental stroke, but none has been shown to be of clinical value (Green AR, et al. 2000). GABA is the primary inhibitory neurotransmitter in the mammalian brain and has been demonstrated to counteract excessive glutamatergic excitation, thereby protecting neurons from excitotoxicity during stroke (Gale F, et al. 2000). However, many compounds shown to increase GABA function do not possess robust neuroprotective e cacy in focal stroke models (Green AR, et al. 2000).
Slit-Robo signaling regulates the development of interneuron populations in cerebral cortex (Andrews W, et al. 2008; Marín O and Rubenstein JL. 2003), we demonstrate that Slit 2 overexpression can increase peri-infarct GABAergic activity, which was protective in focal microstroke. It was reported that curcumin upregulate Slit 2 expression (Sirohi VK, et al 2017), which protect against stroke-mediated brain damage (Bhat A, et al. 2019). Curcumin is widely consumed as turmeric (Bavarsad K, et al. 2019), which may be a potential treatment to upregulate GABAergic activity around microinfarcts, a possibility that we will study in further study.
In addition, Slit 2 was reported to reduce in ammatory responses following ischemic insults (Altay T, et al. 2007). Consistent with these ndings, overexpression of Slit 2 inhibited local microglial and astrocytic activation and protected against peri-infarct neuronal loss. Slit 2 is expressed in multiple brain cell types, including neurons, astrocytes, endothelial cells (Wu JY, et al. 2001), and pericytes (Guijarro-Muñoz I, et al. 2012), and different ischemic models induce distinct cell type-speci c changes in Slit 2 expression. For instance, Park et al. (Park JH, et al. 2016) reported that Slit 2 was constitutively expressed in neurons of control rats and that transient forebrain ischemia upregulated Slit 2 ligand in reactive astrocytes but not neurons or activated microglia of the hippocampus. Fang et al. (Fang M, et al. 2010) reported that Slit 2 was expressed mainly by neurons of the temporal lobe during the acute and latent phases of temporal lobe epilepsy (TLE) in rats, but mainly in astrocytes during the chronic phase. In our study, Slit 2 was expressed in both cortical neurons and astrocytes but not in microglia of control mice, while MI increased astrocytic expression and reduced neuronal expression. Slit-Rob signaling is high in interneurons, so reduced in Slit 2 expression likely re ected loss of peri-infarct interneurons and ensuing local excitotoxicity. Indeed, Slit 2 overexpression protected against interneuronal loss and reduced excitotoxic sequela.
Slit 2 was reported to be overexpressed in these Tg mice throughout life (Han HX, et al. 2011, Li JC, et al. 2018Li G, et al. 2018), which promoted paravascular clearance (Li G, et al. 2018). Consistent with these ndings, we demonstrated that microstroke impaired AQP 4 function, paravascular clearance (Gaberel T, et al. 2014), and BBB permeability (Chen X, et al. 2018), all of which were protected by Slit 2 overexpression. Finally, we demonstrated that Slit 2 overexpression promoted neuronal plasticity after cortical microinfarct induction. Slit 2 has been reported to promote both axonal elongation and branching (Ma L and Tessier-Lavigne M, 2007) or dendritic growth and branching in developing cortical cells (Whitford KL, et al. 2002), as well as GABAergic function in mature brain. Collectively, these processes contribute to circuit recovery post-ischemia. Indeed, GABA-mediated inhibition is a critical modulator of cortical remapping, which is required for functional recovery after stroke (Hiu T, et al. 2016). It is worthy of note that parietal cortex contributes to route learning using proximal salient cues in the water maze task (Solari N, et al. 2018). Thigmotaxis (swimming alone the tub edge) is indicative of spatial learning failure (Vorhees CV, et al. 2006) and we observed thigmotaxis in WT MIs mice, but not Slit 2-Tg MI mice, suggesting that Slit 2 overexpression protected against the spatial learning impairment induced by parietal microinfarcts.

Figure 5
Overexpression of Slit 2 reduces peri-infarct neuronal loss and microglial activation. A.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.
Supplementary le.doc