Morphological regulation of wound repair astrocytes by leucine zipper-bearing kinase-AKT signaling after spinal cord injury

Abstract

Following focal CNS injury, a salient feature of astrocytes lining the lesion is their remarkable morphological transformation into an interwoven cellular border that serves their protective function in wound closure. Despite the importance of morphology in determining function of lesion border astrocytes and injury outcome, there is sparse knowledge of how cell shape is regulated temporally and mechanistically in border-forming astrocytes. We report a transcriptional program of actin and microtubule reorganization that is induced in lesion border astrocytes after spinal cord injury in mice. By genetic gain- and loss-of-function analyses in vivo, we show that leucine zipper-bearing kinase (LZK) is a positive regulator of injury-responsive transcription of cytoskeleton remodeling genes in lesion border astrocytes, with consequences on morphological adaptation of border-forming astrocytes. Functional validation of LZK-dependent cytoskeleton rearrangement in vitro demonstrates its ability to enhance astrocytic process extension, cell movement, and associated structural reorganization of actin and microtubules. We further identify LZK-dependent activation of AKT in astrocytes in vitro and in vivo, which is required for transcriptional regulation of the cytoskeleton by LZK, and to a similar extent as STAT3. Lastly, loss of astrocytic LZK impairs motor recovery after spinal cord injury. Our findings define temporal and molecular regulation of morphological transformation of lesion border astrocytes that may be targeted for CNS repair.

1. Introduction

Astrocytes respond to central nervous system (CNS) insults with a range of morphological and molecular alterations – broadly described as reactive astrogliosis – that shapes their functional adaptations under pathological conditions. Following focal CNS injury including traumatic spinal cord injury in humans and experimental rodent models, severe astrogliosis occurs at the injury site to form a dense cellular border that encloses the lesion core (Hemati-Gourabi et al., 2022; Silver and Miller, 2004; Sofroniew and Vinters, 2010). Border-forming astrogliosis facilitates wound repair and neural recovery (Okada et al., 2006; Herrmann et al., 2008; Faulkner et al., 2004; Bush et al., 1999; Hara et al., 2017; Ono et al., 2023). In contrast to other CNS pathologies in which astrocyte cell shape changes also occur but their causal contribution to pathogenesis and outcome is unclear (Baldwin et al., 2024), focal injury-induced morphological adaptation of lesion border astrocytes serves protective functions of restricting inflammation and compacting the wound (Wanner et al., 2013; Renault-Mihara et al., 2017). Border-forming astrocytes are predominantly derived from astrocyte cell division in situ (Bush et al., 1999; Wanner et al., 2013; Bush et al., 1998; Barnabe-Heider et al., 2010; O’Shea et al., 2024), and they undergo pronounced morphological transformations characterized by cytoskeletal hypertrophy, process elongation, and cellular reorganization into a compact astrocytic meshwork (Wanner et al., 2013), all of which require substantial cytoskeletal remodeling. Migration of astrocytes has also been described in border formation (Okada et al., 2006; Hara et al., 2017; Ono et al., 2023; Renault-Mihara et al., 2017; Renault-Mihara et al., 2011; Robel et al., 2011; Burda and Sofroniew, 2014), the occurrence of which may be injury type-dependent (Tsai et al., 2012; Bardehle et al., 2013). Experimental disruption of astrocytic cytoskeleton, process elongation, or cell movement impairs astrocytic border formation and neurorepair following CNS injury (Okada et al., 2006; Wanner et al., 2013; Renault-Mihara et al., 2017; Pekny et al., 1999). Despite such importance of cytoskeletal remodeling in generating cell form that is critical to the barrier function of lesion border astrocytes, little is known about its regulation. Defining temporal and molecular regulation of the astrocytic cytoskeleton under injury conditions will expand astrocyte-based therapeutic strategies for CNS repair.

Leucine zipper-bearing kinase (LZK) is a positive regulator of astrocyte reactivity, whose endogenous protein expression is upregulated in reactive astrocytes (Chen et al., 2018; Chen et al., 2022). Its gene deletion in adult astrocytes attenuates astrocytic border formation and wound closure following spinal cord injury, whereas its gene overexpression enhances astrocyte-mediated wound closure (Chen et al., 2018). LZK promotes astrogliosis by stimulating astrocyte cell division (Chen et al., 2018), in part by activating signal transducer and activator of transcription 3 (STAT3) (Chen et al., 2018) – a key regulator of border-forming astrogliosis following spinal cord injury (Okada et al., 2006; Herrmann et al., 2008; Wanner et al., 2013). Besides impacting astrocyte proliferation, aberrant morphology of lesion border astrocytes was observed in LZK mutant mice (Chen et al., 2018) that suggests an additional role of LZK in directly regulating cell shape of lesion border astrocytes. However, this has not been directly tested.

Cytoskeleton reorganization fundamentally drives cell shape changes, this study therefore investigates if and how LZK regulates the cytoskeleton of lesion border astrocytes. Transcriptomics analyses reveal injury-dependent transcriptional control of actin and microtubule dynamics in perilesional astrocytes. Using LZK mutant mice that allow for inducible astrocyte-targeted LZK deletion or overexpression, we found that LZK positively regulates injury-induced transcription of cytoskeleton remodeling genes in lesion border astrocytes and their morphological adaptation. In vitro assays functionally validate positive effects of LZK on astrocyte process extension and corresponding structural modifications of actin and microtubule components. We further identified LZK-AKT as an important signaling axis in astrocytic cytoskeleton regulation, and provide evidence for its beneficial effects on motor recovery after spinal cord injury. Our discovery provides temporal and mechanistic insights into the regulation of morphological plasticity of lesion border astrocytes with implications on strategies for CNS wound repair.

2. Results

2.1. LZK-dependent morphological adaptation of lesion border astrocytes after spinal cord injury

To assess effects of LZK specifically on morphology of lesion border astrocytes, we applied spinal cord injury model of complete crush at thoracic level T8 to astrocyte-targeted, tamoxifen-inducible LZK mutant mice. Genetic loss-of-function approach utilized GFAP-CreER^T2; LZK^f/f mice, abbreviated as LZK-knockout (LZK-KO) from hereon (Chen et al., 2018). Age-matched LZK^f/f littermates lacking GFAP-CreER^T2 were used as genotype controls. Prior to injury, tamoxifen was administered in genotype control and LZK-KO mice of both sexes at 8–10 weeks of age, which induced LZK gene deletion only in LZK-KO mice (Chen et al., 2018; Chen et al., 2022). At one month after injury, immunofluorescence staining for glial fibrillary acidic protein (GFAP) demarcated a GFAP⁺ astrocytic border surrounding a largely GFAP⁻ lesion core (Fig. 1). Morphologically, formation of an astrocytic border to enclose damaged tissue requires extension and reorganization of astrocytic processes from a perpendicular to parallel orientation with respect to the lesion. Orientation of astrocytic processes can be quantified by GFAP⁺ process-to-lesion angle, with an angle of 90 degrees indicating a perpendicular orientation (Wanner et al., 2013). At one month after injury, the average process-to-lesion angle was 16 degrees in control mice, whereas it was increased to 72 degrees (more perpendicular) in LZK-KO mice (Fig. 1A-B). This shows that lesion border astrocytes in LZK-KO mice are impaired in aligning their processes parallel to the lesion. Additionally, extension of astrocytic processes was also examined. Because dense overlap of astrocytic processes at the lesion border makes it difficult to determine process lengths of individual astrocytes, we looked 300 μm and 600 μm rostral or caudal to the lesion border where individual GFAP⁺ astrocytes can be more easily identified. At 300 μm from the lesion border, the average length of the longest cellular process per astrocyte was reduced by 23 % in LZK-KO mice compared to control (Fig. 1A, C). At 600 μm from the lesion border however, there was no difference in length of longest process between control and LZK-KO mice (Fig. 1A, D). These findings show reduced process extension and disrupted process alignment of lesion border astrocytes in LZK-KO mice.

Fig. 1.

LZK-dependent morphological adaptation of lesion border astrocytes after spinal cord injury. (A) Representative images of GFAP and DAPI staining in injured spinal cords from control (top) and LZK-KO (bottom) mice, 1 month (1 M) after injury. Boxed insets are enlarged to show astrocytes at the lesion border (A1), or located at 300 μm (A2) or 600 μm (A3) from the lesion border; lesion border is outlined in white; A1 panels show examples of astrocytic process-to-lesion border angle measurement (yellow outline); scale bar = 100 μm. Based on GFAP staining, quantification of (B) Left, astrocytic process-to-lesion angle at the lesion border; right, illustration of astrocytic process-to-lesion angle measurement: angle is measured between lesion border (outlined in white) and GFAP⁺ process (red overlaid with yellow dotted line). (C) longest cellular process per astrocyte located at 300 μm or (D) at 600 μm from the lesion border in control and LZK-KO mice. n = 5 mice per genotype; graphing group mean with SEM; parametric, unpaired t-test, *p < 0.05. (E) Representative images of GFAP and DAPI staining in injured spinal cords from control (top) and LZK-OE (bottom) mice, 7 days post injury (7dpi). Boxed insets are enlarged to show astrocytes at the lesion border (A1), or located at 300 μm (A2) or 600 μm (A3) from the lesion border; astrocytic border is outlined in white; A1 panels show examples of astrocytic process-to-lesion border angle measurement (yellow outline); scale bar = 100 μm. Based on GFAP staining, quantification of (F) astrocytic process-to-lesion angle at the lesion border, (G) longest cellular process per astrocyte located at 300 μm or (H) at 600 μm from the lesion border in control and LZK-OE mice. n = 3 mice per genotype; graphing group mean with SEM; parametric, unpaired t-test, *p < 0.05. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

In a gain-of-function approach, we used GFAP-CreER^T2; LZK^OE mice that overexpress LZK in astrocytes (Chen et al., 2018), abbreviated as LZK-overexpression (LZK-OE) from hereon. Age-matched LZK^OE littermates lacking GFAP-CreER^T2 served as genotype controls; all mice received tamoxifen prior to injury. We reasoned that if LZK promotes morphological maturation of lesion border astrocytes, then its gain-of-function phenotype could be better detected at an early post-injury timepoint. We therefore harvested injured spinal cords at 7 days post injury (dpi). At 7dpi, the average process-to-lesion angle of lesion border astrocytes was 85 degrees in control mice, compared to 70 degrees in LZK-OE mice (Fig. 1E-F), which indicates a more parallel orientation in LZK-OE mice (Fig. 1E-F). Furthermore, average length of the longest process per astrocyte was increased by 16 % and 15 %, measured at 300 μm or 600 μm from the lesion border, respectively (Fig. 1E, G, H). Together, these results demonstrate positive effects of astrocytic LZK on morphological transformation of lesion border astrocytes, assessed based on their process extension and alignment.

2.2. An injury-induced transcriptional program of cytoskeleton reorganization in lesion border astrocytes after spinal cord injury

Cytoskeleton rearrangement drives cell shape changes and cell movement. Next, to determine if LZK promotes morphological adaptation of lesion border astrocytes via cytoskeleton reorganization, we first identified injury-induced changes of cytoskeletal components in lesion border astrocytes in an unbiased manner, by transcriptomics analyses. Using an available RNA-sequencing dataset from astrocytes purified from the lesion epicenter of contused spinal cords of adult mice in sub-acute and chronic phases of injury (GEO: GSE153720) (Wei et al., 2021), gene ontology (GO) analysis of differentially expressed genes (DEGs) identified organization of actin and microtubule, but not intermediate filaments, as biological processes enriched in lesion border astrocytes after spinal cord injury (DEGs defined by log2FC > 1 in expression levels compared to sham, biological processes defined by inclusion of ≥5 DEGs each, FDR < 0.05) (Fig. 2A, Table 2–1). Next, two gene lists – one of genes associated with actin and the other with microtubule organization – were generated based on GO terms (Table 2–2) and used to identify DEGs involved in the regulation of these cytoskeletal components after injury (Fig. 2B-C, Table 2–3 and Table 2–4). We focused on actin and microtubules because: i) while they have been implicated in cell shape control of border-forming astrocytes (Renault-Mihara et al., 2017; Abd-El-Basset and Fedoroff, 1997; Boukhelifa et al., 2003; Schiweck et al., 2018), they are under-examined in astrogliosis compared to intermediate filaments (Hol and Pekny, 2015); ii) knowledge from cell biology points to their likely contribution to cellular process extension, orientation, and organization of border-forming astrocytes, in contrast to intermediate filaments that chiefly provide structural support for cell shape stabilization; and iii) regulation of actin and microtubules by LZK is unknown.

Fig. 2.

An injury-induced transcriptional program of cytoskeleton reorganization in lesion border astrocytes in vivo. (A) Biological processes related to actin and microtubule reorganization are enriched in lesion border astrocytes isolated from the lesion epicenter of mouse spinal cords at 7 days post-injury (dpi) and 1 month (M) after injury (FDR < 0.05, p < 0.05), based on RNA-seq dataset GSE153720. (B) Heatmap of actin-related genes differentially expressed (Log2FC > 1, p < 0.05) in lesion border astrocytes isolated from injured mouse spinal cord at the indicated post-injury timepoints. Color key in Log2FC scale. (C) Heatmap of microtubule-related genes differentially expressed (Log2FC > 1, p < 0.05) in lesion border astrocytes isolated from injured mouse spinal cord at the indicated post-injury timepoints. Color key in Log2FC scale. (D) Number of differentially expressed genes in astrocytes (DEGs with cutoff Log2FC > 1 and p < 0.05, upregulated genes in orange, downregulated genes in purple) isolated from sham versus the spinal cord lesion at the indicated post-injury timepoints. (E) Venn diagrams show the number of actin-related genes or (F) microtubule-related genes that are upregulated (Log2FC > 1, p < 0.05) in lesion border astrocytes at the indicated timepoints after spinal cord injury. (G) Actin-based biological processes and associated FDRs at the indicated timepoints after injury, identified based on injury-induced actin-related genes with peak expression at each timepoint. FDR < 0.05 in red. (H) Microtubule-based biological processes and associated FDRs at the indicated timepoints after injury, identified based on injury-induced microtubule-related genes with peak expression at each timepoint. FDR < 0.05 in red; FDR > 0.05 in blue. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

In lesion border astrocytes, actin-related structural and regulatory genes make up 7.7 % of injury-induced DEGs on 7dpi (total of 3141 DEGs), 8.0 % at 1 M (total of 2526 DEGs), and 7.8 % at 3 M (total of 925 DEGs) after injury, whereas microtubule-related genes make up 10.6 % of DEGs on 7dpi, 10.2 % at 1 M, and 4.1 % at 3 M (Table 2–3 and Table 2–4). Spinal cord injury has a main effect of transcriptional upregulation in lesion border astrocytes, that is also evident specifically in actin- and microtubule-related genes from sub-acute to chronic phases of injury (Fig. 2D). Of the upregulated actin- and microtubule-related genes across the three post-injury timepoints, 81 % and 83 % were induced on 7dpi, respectively (Fig. 2E-F). By 3 M after injury, 55 % of actin- and 85 % microtubule-related genes that were upregulated on 7dpi and/or at 1 M returned to at or below baseline level as in sham control (Table 2–3 and Table 2–4).

The appearance of timepoint-specific peaks of expression in DEG heatmaps prompted us to examine whether there exist injury phase-specific processes involving cytoskeletal remodeling in astrocytes (Fig. 2B-C). GO analyses of injury-induced actin-related genes with peak expression at 7dpi (consisting of 87 genes), 1 M (consisting of 41 genes), or 3 M (consisting of 19 genes) identified actin filament organization as common actin-based processes shared among all timepoints (Fig. 2G, FDR < 0.05 at 7dpi, 1 M, 3 M). In contrast, analyses of injury-induced microtubule-related genes with peak expression at 7dpi (consisting of 106 genes) or 1 M (consisting of 39 genes) revealed cell division as the overrepresented biological process unique to 7dpi (Fig. 2H, FDR < 0.05 on 7dpi), whereas post-mitotic regulation of microtubules persisted up to at least 1 M after injury (Fig. 2H, FDR < 0.05 at 1 M). Analyses of microtubule-related genes with peak expression at 3 M (consisting of 8 genes) returned no result based on cutoff criteria (inclusion of ≥5 genes per biological process, FDR < 0.05). Of the actin- and microtubule-related genes that remained upregulated at 1 M–3 M after injury, 74 % and 76 % were induced at 7dpi respectively, indicating that a majority of persistently upregulated cytoskeleton components were induced early after injury (Fig. 2E-F).

Together, these findings demonstrate that spinal cord injury activates a transcriptional program of actin and microtubule organization in lesion border astrocytes. Analyses of microtubule genes show primary contribution of cytoskeleton remodeling to cell division at 7dpi, consistent with current knowledge of the timeline of injury-induced astrocyte proliferation (Wanner et al., 2013; O’Shea et al., 2024). In contrast, continuous transcriptional upregulation of cytoskeleton components at 1–3 months after injury is post-mitotic, and reveals cytoskeleton remodeling that persists past establishment of the astrocytic border traditionally characterized to be complete by 2 weeks after injury based on histology (Okada et al., 2006; Herrmann et al., 2008; Chen et al., 2018). This suggests a window of morphological plasticity of lesion border astrocytes that is longer than previously recognized. Importantly, from this transcriptomics analyses, we identified injury-responsive cytoskeleton genes associated with persistent actin and microtubule reorganization that can be used to test morphological regulation by LZK in lesion border astrocytes.

2.3. LZK upregulates injury-induced transcription of cytoskeleton remodeling genes in lesion border astrocytes in vivo, and can be recapitulated in vitro

Of the cytoskeleton organization genes that are persistently upregulated in border-forming astrocytes following spinal cord injury (Fig. 2B-C, 2E-F, Table 2–3, Table 2–4), six representative actin and/or microtubule organization genes – Cotl1, Coro1a, Iqgap3, Tpx2, Rgs14, Rac1 – were selected to further test for LZK-dependent regulation based on: i) their defined roles in cellular process extension, cell movement, and epithelial-mesenchymal transition (Lechuga et al., 2024; Castro-Castro et al., 2011; Jinawath et al., 2020; Brunet et al., 2004; Wittmann et al., 2003; Chan et al., 2011; Schatz et al., 2003; Yang et al., 2015; Alfaro-Aco et al., 2017; Waterman-Storer et al., 1999; Palazzo et al., 2001; Watanabe et al., 2015; Wang et al., 2007; Provost et al., 2001; Kim et al., 2014; Hartwig et al., 2014; Vellano et al., 2013; Martin-Mccaffrey et al., 2005) (Fig. 3A), all of which are known to occur in border-forming astrocytes (Okada et al., 2006; Wanner et al., 2013; O’Shea et al., 2024) and can be functionally assessed in vitro; and ii) their persistent or delayed transcriptional upregulation at 1–3 M after injury (Fig. 3B), indicative of their contribution to post-mitotic cytoskeletal remodeling.

Fig. 3.

LZK positively regulates transcription of actin- and microtubule-remodeling genes in lesion border astrocytes and primary astrocytes. (A) Six representative actin- and microtubule-related genes tested for LZK-dependent transcription and their published functions in cytoskeleton remodeling. (B) Heatmap shows injury-induced, persistent or delayed upregulation of representative actin- and microtubule-remodeling genes in lesion border astrocytes at the indicated timepoints after spinal cord injury (Log2FC > 1, *p < 0.05). Color key in Log2FC scale. (C) Experimental timeline of five doses of tamoxifen administration, spinal cord injury (SCI, complete crush at T8), and astrocyte isolation from primary lesion of LZK^f/f (control) and GFAP-CreER^T2; LZK^f/f (LZK-KO) mice for gene analyses shown in D. (D) Fold change (FC) of gene expression of Lzk and indicated cytoskeleton remodeling genes in lesion border astrocytes isolated from LZK-KO mice compared to control mice, experiment outlined in C. Transcript levels quantified by qRT-PCR. n = 3 mice per genotype; pairwise comparison between control and LZK-KO by parametric, unpaired t-test for each gene. (E) Experimental timeline of two doses of tamoxifen administration, SCI, and astrocyte isolation from primary lesion of LZK^OE (control) and GFAP-CreER^T2; LZK^OE (LZK-OE) mice for gene analyses shown in F. (F) Fold change (FC) of gene expression of Lzk and indicated cytoskeleton remodeling genes in lesion border astrocytes isolated from LZK-OE mice compared to control, experiment outlined in E. Transcript levels quantified by qRT-PCR. n = 3–4 mice per genotype; pairwise comparison between control and LZK-OE by parametric, unpaired t-test for each gene at each timepoint. (G) Experimental timeline of 4-hydroxytamoxifen (4-OHT) treatment and use of primary cortical astrocytes isolated from control, LZK-KO, or LZK-OE mice for qRT-PCR or other in vitro assays. (H) Quantification of percentage of GFAP⁺ cells out of all DAPI⁺ cells to assess purity of astrocyte culture (white bar), and percentage of GFAP⁺tdTomato⁺ cells of all GFAP⁺ cells to assess efficiency of 4-OHT-induced Cre-mediated recombination in primary astrocytes (grey bar). n = 3 mice; graphing mean with SEM. (I) Immunofluorescence staining for GFAP and DAPI in 4-OHT-treated primary cortical astrocytes purified from GFAP-CreER^T2; LZK^OE mice that express tdTomato in a tamoxifen-inducible, Cre-dependent manner. (J) Fold change (FC) of gene expression of Lzk and indicated genes in primary LZK-KO astrocytes compared to control astrocytes, or LZK-OE astrocytes compared to control, experiment outlined in G. Transcript levels quantified by qRT-PCR. n = 3 mice per genotype; pairwise comparison between control and LZK-KO, or between control and LZK-OE, by parametric, unpaired t-test for each gene. (K) Additional actin- and microtubule-related genes tested in vitro for LZK-dependent transcription and their published functions in cytoskeleton remodeling. (L) Fold change (FC) of gene expression of indicated cytoskeleton-related genes in primary LZK-KO astrocytes compared to control astrocytes, or LZK-OE astrocytes compared to control, experiment outlined in Fig. 2G. Transcript levels quantified by qRT-PCR. n = 3 mice per genotype; pairwise comparison between control and LZK-KO, or between control and LZK-OE, by parametric, unpaired t-test for each gene.

To determine if LZK regulates these cytoskeleton-related genes in border-forming astrocytes in vivo, weperformed complete crush of the spinal cord at T8 in LZK-KO mice (Fig. 3C) and LZK-OE mice (Fig. 3E) with respective controls. Astrocytes isolated from the lesion of LZK-KO mice on 7dpi showed greater than 90 % reduction in LZK mRNA levels compared to genotype control, validating efficient astrocyte-specific LZK gene deletion (Fig. 3D, Table 3). Correspondingly, transcription of the six actin and/or microtubule remodeling cytoskeletons was reduced by 50–70 % in border-forming astrocytes of LZK-KO mice, compared to control (Fig. 3D, Table 3). These results indicate a requirement for endogenous astrocytic LZK in injury-dependent, transcriptional upregulation of cytoskeleton remodeling genes in lesion border astrocytes. In LZK-OE mice, astrocytes isolated from the injury site showed a 4-fold increase in LZK transcription compared to control, validating astrocyte-specific LZK overexpression (Fig. 3F, Table 3). Transcription of representative cytoskeleton-associated genes in border-forming astrocytes of LZK-OE mice was increased by a range of 2–100 times compared to control on 7dpi (Fig. 3F, Table 3). Even at the chronic timepoint of two months after injury, transcription of LZK and all representative cytoskeleton genes remained 3–50 times higher in LZK-OE astrocytes (Fig. 3F, Table 5). In astrocytes isolated from uninjured cerebral cortices of LZK-KO mice, expression of representative cytoskeleton genes was also dampened (Fig. S1A), supporting a requirement for LZK in maintaining steady state cytoskeleton dynamics in addition to injury-induced cytoskeletal remodeling in astrocytes. Correspondingly, LZK-OE alone was sufficient to weakly upregulate these genes by up to 4-fold without injury in astrocytes in vivo (Fig. S1B, Table 3), though to a much lesser extent compared to up to 183-fold upregulation observed in injured LZK-OE mice (Fig. 3F, Table 3). These results show compounded positive effects of LZK and injury on transcription of cytoskeleton genes. Taken together, LZK gene manipulations in vivo identified LZK as a positive regulator of injury-induced transcription of cytoskeleton organization genes in border-forming astrocytes after spinal cord injury.

We next assessed whether LZK-dependent transcriptional regulation of cytoskeleton genes can be recapitulated in vitro, with the goal to utilize primary astrocytes to examine effects of LZK on actin and microtubule structural reorganization and to determine underlying mechanism. To modulate LZK gene expression in vitro, primary astrocytes were isolated from cerebral cortices of early postnatal LZK-KO, LZK-OE, or littermate control pups lacking Cre-ERT2, and treated with 4-hydroxytamoxifen (4-OHT) (Fig. 3G). Purity of primary astrocyte culture was confirmed to be 90 % based on GFAP expression (Fig. 3H-I). 4-OHT treatment resulted in successful gene recombination in 78 % of astrocytes based on expression of fluorescent reporter tdTomato in astrocytes purified from LZK-OE mice that express tdTomato in a Cre-dependent manner (Chen et al., 2018) (Fig. 3G-I). Following 4-OHT treatment, Lzk transcription was reduced by 60 % in LZK-KO astrocytes, but increased over 200-fold in LZK-OE astrocytes compared to control astrocytes (Fig. 3J, Table 3). As observed in vivo, representative cytoskeleton remodeling genes (Colt1, Coro1a, Iqgap3, Tpx2, Rgs14, Rac1) exhibited LZK-dependent expression, with 50–70 % reduction in LZK-KO astrocytes and 2–5-fold increase in LZK-OE astrocytes (Fig. 3J, Table 3). Additional actin and microtubule organization genes induced in lesion border astrocytes (Nuak, Dock2, Met, Tubb4a and Phldb2) (Xiao et al., 2022; Curiel et al., 2017; Lansbergen et al., 2006; Xiang et al., 2017; Chmielowiec et al., 2007; Menard et al., 2014; Cote et al., 2005; Yang et al., 2009; Vallenius et al., 2011; Yuan et al., 2018; Sellers and Pato, 1984; Tan et al., 2001) also showed LZK-dependent expression in primary astrocytes (Fig. 3K-L, Table 3). Together, these in vivo and in vitro findings demonstrate that: i) LZK is necessary and sufficient for transcription of selective injury-responsive cytoskeleton remodeling genes (Fig. 3D, F, J, L); ii) while LZK overexpression alone is sufficient to enhance transcription of these genes in vitro, its effect is less pronounced than that of injury and LZK overexpression combined (Fig. 3F, J); and iii) in vitro recapitulation of LZK-dependent transcriptional regulation of cytoskeleton genes as observed in vivo supports the use of primary astrocytes to functionally isolate effects of LZK on the cytoskeleton and to identify underlying mechanisms.

2.4. LZK-dependent cytoskeleton reorganization in primary astrocytes

Based on established functions of the representative injury-responsive, LZK-regulated cytoskeleton genes in cell movement and protrusion formation (Fig. 3A, K), we sought to determine LZK’s effects on these cellular outcomes of cytoskeleton reorganization, starting with the classic scratch wound assay (Etienne, 2006). Primary astrocytes isolated from genotype control, LZK-KO, or LZK-OE mice (the last two of which also express the fluorescent reporter tdTomato in a Cre-dependent manner) were grown to a confluent monolayer and treated with 4-OHT prior to scratch induction. Primary astrocytes, when grown to confluency, do not undergo cell division. Within 24 h after making a scratch, which is shorter than the doubling time of primary astrocytes (Ikeshima-Kataoka et al., 2007), scratch or “wound” closure is attributed to astrocyte cell movement and quantifiable by change in cell-free gap area over time. Astrocyte cell count at 12 h and 24 h after scratch showed minimal increase in cell number (Fig. S1C); and quantification of Ki-67⁺GFAP⁺ astrocytes showed no significant difference in the number of proliferating astrocytes within 24 h of seeding (Fig. S1D-E). Following scratch, compared to control that had 41 % and 10 % of the original gap area remaining at 12 h and 24 h after “wounding” respectively, gap area was increased in LZK-KO astrocytes (58 % at 12 h and 48 % at 24 h), indicative of slower cell movement in filling the gap (Fig. 4A-B). In contrast, gap area was reduced at the same timepoints in LZK-OE astrocytes (18 % at 12 h and 3 % at 24 h), indicative of faster gap closure by LZK-OE astrocytes (Fig. 4A-B). These results show that LZK can functionally effect cytoskeleton reorganization to produce cell movement in scratch wound closure.

Fig. 4.

LZK effects cytoskeleton reorganization that enhances scratch wound closure, actin-based process extension, and actin polymerization in vitro. (A) Phase contrast images taken at 12 h and 24 h after scratch wounding a confluent monolayer of primary astrocytes isolated from control, LZK-KO, and LZK-OE mice, the latter two of which also express fluorescent reporter tdTomato. Cell-free area resulting from scratch is outlined in black in the center of image. Scale bar = 300 μm. (B) Comparison of cell-free area at 12 h and 24 h normalized to 0 h (immediately after scratch) as a percentage. n = 3–4 mice per genotype; graphing group mean with SEM. Two-way ANOVA with Tukey’s post hoc multiple comparison test *p < 0.05 (See ANOVA Statistics Table in Appendix). (C) Actin-based process extension at the leading edge visualized by phalloidin staining in control, LZK-KO, and LZK-OE primary astrocyte monolayer at 12 h and 24 h following scratch wound. Measurement of length and width of actin-based protrusion is shown in first image panel. Scale bar = 100 μm. (D) Comparison of length-to-width ratio of actin-based extension into cell-free space among astrocytes of the indicated genotypes. n = 25–45 astrocytes from 4 mice per genotype; graphing group mean with SEM; two-way ANOVA with Tukey’s post hoc multiple comparison test, *p < 0.05 (See ANOVA Statistics Table in Appendix). (E) Immunoblot of F-actin and G-actin purified from primary astrocytes treated with or without phalloidin to validate F- and G-actin fractionation protocol. (F) Immunoblot of F-actin and G-actin purified from control and LZK-KO astrocytes. (G) Comparison of F/G-actin ratio between control and LZK-KO astrocytes. n = 4 mice per genotype; graphing group mean with SEM; parametric, unpaired t-test, *p < 0.05. (H) Immunoblot of F-actin and G-actin purified from control and LZK-OE astrocytes. (I) Comparison of F/G-actin ratio between control and LZK-KO astrocytes. n = 4 mice per genotype; graphing group mean with SEM; parametric, unpaired t-test, *p < 0.05.

Transcriptomics analyses predict injury-induced actin filament organization in lesion border astrocytes (Fig. 2A-B, 2G) and LZK-regulated cytoskeleton genes promote actin-based process extension (Fig. 3), we therefore next examined effects of LZK on actin-driven process elongation. Scratch wound assay allows for direct assessment of formation of actin-based lamellipodia protrusions at the leading edge of astrocytes, in which actin polymerization provides the primary force for directional cell movement. To quantify lamellipodia extensions of astrocytes lining the “wound” edges, actin filaments were first visualized with phalloidin staining. Then the length (L) of phallodin⁺ lamellipodia of each astrocyte at the cell front – measured from the back of the nucleus to the tip of the longest extension in the direction of cell-free gap, and the width (W) – measured at the widest part of the lamellipodia perpendicular to the length, were used to calculate the length to width (L/W) ratio (Liao et al., 2015; Devitt et al., 2021) (Fig. 4C). L/W > 1 indicates lamellipodia extension into the cell-free gap; and the greater the L/W ratio, the longer the protrusion. At 12 h and 24 h after scratch, control astrocytes exhibited an average L/W ratio of 2 indicative of lamellipodia formation, whereas LZK-KO astrocytes showed a reduced L/W ratio of 1 indicative of lack of directional process elongation (Fig. 4C-D). In contrast, LZK-OE astrocytes doubled the L/W ratio compared to the control, indicative of longer lamellipodia extension (Fig. 4C-D). These findings demonstrate the ability of LZK to enhance actin-based process extension in astrocytes.

Polymerization of monomeric globular actin (G-actin) into filamentous actin (F-actin) creates the protrusive force required for cellular elongation (Kiuchi et al., 2011; Skruber et al., 2018; Ramaekers et al., 1981), we therefore tested if LZK alters relative abundance of G- and F-actin in astrocytes by fractionation (Lee et al., 2021; Song et al., 2013). As a quality control of G- and F-actin isolation, astrocytes were treated with the F-actin stabilizer phalloidin. F-actin enrichment resulting from phalloidin treatment validated our isolation technique (Fig. 4E). Fractionation and quantification of F- and G-actin following genetic manipulation of LZK showed a 63 % reduction in F/G-actin ratio in LZK-KO astrocytes (Fig. 4F-G), and a two-fold increase in LZK-OE astrocytes (Fig. 4H-I), compared to control. These results together demonstrate that LZK promotes actin polymerization to facilitate astrocytic process elongation.

In addition to actin, microtubule organization is also transcriptionally regulated in an injury- and LZK-dependent manner in astrocytes in vivo and in vitro (Fig. 2–3). Microtubules are a key structural contributor to process extension in astrocytes (Etienne-Manneville, 2004; Etienne-Manneville and Hall, 2001). Reported functions of injury-responsive and LZK-regulated microtubule genes point to microtubule stabilization in lesion border astrocytes (Fig. 3). Supporting this is the observation that αTAT1 – a tubulin acetyltransferase responsible for α-tubulin (αK40) acetylation (Akella et al., 2010; Shida et al., 2010) associated with microtubule stabilization (LeDizet and Piperno, 1986; Piperno et al., 1987; De Brabander et al., 1976) – is transcriptionally upregulated by two-fold in border-forming astrocytes on 7dpi and 1 M after spinal cord injury (GEO: GSE153720). To test effects of LZK on microtubule acetylation, we quantified the abundance of αK40-tubulin against total α-tubulin in primary astrocytes following LZK genetic manipulation. Both immunofluorescence staining and immunoblotting showed a minor but statistically significant 20–30 % decrease of αK40 acetylation in LZK-KO astrocytes compared to the control (Fig. 5A-D), and a corresponding increase of αK40 acetylation in LZK-OE astrocytes (Fig. 5A-B, 5E-F). These results indicate positive effects of LZK on microtubule acetylation that may act synergistically with the actin network to effect process extension of astrocytes. Collectively, these in vitro experiments functionally validate in vivo transcriptional indications of injury- and LZK-dependent regulation of actin and microtubule dynamics in astrocytes that are consistent with the morphological phenotype of LZK mutant astrocytes previously observed in vivo (Chen et al., 2018).

Fig. 5.

LZK increases α-tubulin acetylation associated with microtubule stabilization in vitro. (A) Immunofluorescence staining for total α-tubulin and acetylated α-tubulin at K40 (Ac. α-tubulin) in primary astrocytes isolated from control, LZK-KO, and LZK-OE mice. Scale bar = 100 μm. Square insets in LZK-OE panel show increased α-tubulin acetylation in a tdTomato⁺ LZK-OE astrocyte (outlined in white) compared to a neighboring tdTomato⁻ non-LZK-OE cell (outlined in yellow). (B) Comparison of signal intensity of acetylated α-tubulin normalized to total α-tubulin among the three genotypes. 35–75 astrocytes from 4 mice per genotype; graphing group mean with SEM; one-way ANOVA and Bonferroni post hoc multiple comparison test, *p < 0.05. (C) Total lysates from control and LZK-KO astrocytes were immunoblotted for acetylated and total α-tubulin. (D) Quantification of acetylated α-tubulin normalized to total α-tubulin. n = 5 mice per genotype; graphing group mean with SEM; parametric, unpaired t-test, *p < 0.05. (E) Total lysates from control and LZK-OE astrocytes were immunoblotted for acetylated and total α-tubulin. (F) Quantification of acetylated α-tubulin normalized to total α-tubulin. n = 5 mice per genotype; graphing group mean with SEM; parametric, unpaired t-test, *p < 0.05. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

2.5. AKT is a downstream effector of LZK in astrocytic cytoskeleton regulation

The phosphoinositide 3 kinase (PI3K)-AKT pathway regulates the cytoskeleton in cell shape and movement (Liliental et al., 2000; Higuchi et al., 2001; Jacinto et al., 2004; Sarbassov et al., 2004; Enomoto et al., 2005; Sarbassov et al., 2005; Liao et al., 2022; Andrews et al., 2020). We therefore tested AKT as a downstream effector of LZK in cytoskeleton organization of astrocytes. We first assessed the ability of LZK to activate endogenous AKT in primary astrocytes. In LZK-KO astrocytes, basal AKT activation – detectable by AKT phosphorylation at serine 473 (pAKT) – was reduced by almost half compared to control by immunoblotting (Fig. 6A-B). As an alternative approach to assess AKT activation, we also immunoblotted for phosphorylation of protein S6 at serine 235 and 236 (pS6), an indicator of activation of the AKT-mTOR-S6K pathway. Consistent with results based on pAKT detection, basal pS6 was reduced by 80 % in LZK-KO astrocytes (Fig. 6C-D). In contrast, both pAKT and pS6 levels were increased by three-fold in LZK-OE astrocytes compared to control (Fig. 6E-H). These findings demonstrate the ability of LZK to activate AKT signaling in astrocytes.

Fig. 6.

LZK-AKT signaling promotes transcription of cytoskeleton-remodeling genes and cell movement in vitro. (A) Immunoblot of pAKT (S473), total AKT, LZK, and β-actin in total cell lysates of control and LZK-KO astrocytes. (B) Quantification of signal intensity of pAKT normalized to total AKT based on immunoblots; graphing mean with SEM; parametric, unpaired t-test, *p < 0.05. (C) Immunoblot of pS6 and β-actin in total cell lysates of control and LZK-KO astrocytes. (D) Quantification of signal intensity of pS6 normalized to β-actin based on immunoblots; graphing mean with SEM; parametric, unpaired t-test, *p < 0.05. (E) Immunoblot of indicated proteins in total cell lysates of control and LZK-OE astrocytes. (F) Quantification of signal intensity of pAKT normalized to total AKT based on immunoblots 6E and 6 K; graphing mean with SEM; parametric, unpaired t-test, *p < 0.05. (G) Immunoblot of pS6 and β-actin in total cell lysates of control and LZK-OE astrocytes. (H) Quantification of signal intensity of pS6 normalized to β-actin based on immunoblots; graphing mean with SEM; parametric, unpaired t-test, *p < 0.05. (I) Primary wildtype astrocytes were treated with vehicle or AKT inhibitor. Total cell lysates were immunoblotted for endogenous pS6 and β-actin. (J) Based on immunoblots, quantification of signal intensity of pS6 normalized to that of β-actin. n = 3; graphing mean with SEM; parametric, unpaired t-test, *p < 0.05. (K) Immunoblotting of indicated proteins in total cell lysates of control, LZK-OE, and LZK-OE astrocytes treated with or without AKT inhibitor. (L) Immunoblot of indicated proteins in total cell lysates of primary astrocytes, non-transfected, transfected with FLAG-LZK plasmid (LZK), treated with or without AKT inhibitor. (M) Fold change (FC) of gene expression of Lzk and indicated cytoskeleton remodeling genes in primary LZK-OE astrocytes treated with or without AKT inhibitor compared to vehicle-treated control astrocytes. Transcript levels quantified by qRT-PCR. n = 3 mice per genotype; p-values determined by one-way ANOVA and Tukey’s post hoc pairwise comparison to vehicle-treated genotype control. (N) Phase contrast images taken at 12 h and 24 h after scratch wounding a monolayer of primary astrocytes transfected with pBI empty vector (EV), pBI-LZK plasmid (LZK), or pBI-LZK-K195A (LZK-K195A) plasmid, with or without AKT inhibitor. Cell free area resulting from scratch is outlined in black in the center of image. Scale bar = 50 μm. (O) Comparison of cell-free area at 12 h and (P) 24 h normalized to 0 h (immediately after scratch) as a percentage. n = 3 independent biological replicates per condition; graphing group mean with SEM; one-way ANOVA with Tukey’s post hoc pairwise comparison, *p < 0.05 (See ANOVA Statistics Table in Appendix). (Q) Primary astrocytes were transfected with plasmid expressing FLAG-LZK only, or co-transfected with plasmids expressing FLAG-LZK and HA-AKT. Astrocyte total cell lysates (input) or protein immunoprecipitation with HA antibody (IP) was immunoblotted for the indicated proteins.

Next, to determine whether AKT activity is required for LZK-dependent transcription of injury-responsive cytoskeleton genes, we tested if AKT inhibition can reverse transcriptional upregulation of cytoskeleton remodeling genes in LZK-OE astrocytes. The allosteric AKT inhibitor VIII, applied at 2.5 μM, effectively reduced basal AKT activation in control astrocytes by 80 % as assessed by pS6 immunoblotting (Fig. 6I-J). When applied to LZK-OE astrocytes, AKT inhibitor abrogated LZK-dependent AKT activation based on pAKT (Fig. 6K). Interestingly, protein expression but not transcription of LZK itself was also reduced upon AKT inhibition (Fig. 6K, M). Even when switched to a plasmid-based approach to overexpress LZK, AKT inhibition was still observed to downregulate LZK protein levels that likely resulted from protein destabilization (Fig. 6L). These results suggest positive, post-translational feedback regulation between LZK and AKT in astrocytes that is reminiscent of characteristics of LZK-JNK signaling in neurons (Chen et al., 2016). Inhibition of AKT completely abolished transcriptional upregulation of actin and microtubule remodeling genes in LZK-OE astrocytes (Fig. 6M, Table 6). These results demonstrate that AKT activation is required for LZK-dependent transcription of cytoskeleton remodeling genes in astrocytes. Correspondingly, AKT inhibition blocked LZK-dependent acceleration of astrocyte cell movement (Fig. 6N-P). Expression of a catalytically inactive LZK mutant (K195A) (Chen et al., 2016; Dickson et al., 2010) enlarged gap area compared to control (Fig. 6N-P), indicating that LZK-K195A acts as a dominant negative that interferes with endogenous LZK function. Furthermore, LZK-K195A alone impaired gap closure to a similar extent as AKT inhibition, with no additive effects observed in LZK-K195A-expressing astrocytes treated with AKT inhibitor (Fig. 6N-P). These results show that LZK kinase activity and AKT activation are required for LZK to promote cytoskeleton reorganization in scratch wound closure.

To further examine how LZK may activate AKT, we tested for their ability to interact by protein co-immunoprecipitation. Primary astrocytes were co-transfected with plasmids expressing FLAG-tagged LZK or HA-tagged AKT. FLAG-LZK-only transfection served as negative control. Immunoprecipitation of HA-AKT pulled down LZK, showing their interaction in one protein complex in astrocytes (Fig. 6Q). Whether LZK can directly phosphorylate AKT, however, remains to be determined.

Lastly, given previous identification of STAT3 (Chen et al., 2018; Chen et al., 2022) as a downstream effector of LZK, we next compared the relative contribution of AKT versus STAT3 to LZK-dependent transcription of cytoskeleton genes. In primary astrocytes overexpressing LZK, inhibition of either STAT3 (by the high affinity inhibitor C188–9) or AKT was sufficient to completely abrogate transcriptional upregulation of cytoskeleton remodeling genes (Fig. S1F, Table 6). These findings indicate a requirement for AKT and/or STAT3 activity in LZK-dependent transcription of cytoskeleton genes, and further suggest cross-regulation between AKT and STAT3. Because STAT3 plays a dual role in regulating cell division and morphology of lesion border astrocytes (Okada et al., 2006; Wanner et al., 2013; Renault-Mihara et al., 2017), we further examined the contribution of AKT and STAT3 to LZK’s pro-proliferative activity (Chen et al., 2018). We found that inhibition of either AKT or STAT3 reduced expression of proliferation genes Mkif67 and Topo2a³⁰ to a similar extent in control and LZK-OE astrocytes (Fig. S1F, Table 6). These results show that LZK-AKT signaling positively regulates transcription of injury-responsive cytoskeleton remodeling genes with an impact on scratch wound closure, and that AKT and STAT3 cooperatively contribute to LZK-dependent transcriptional control of cytoskeleton reorganization and cell proliferation of astrocytes, two critical cellular responses that define lesion border astrocytes (Wanner et al., 2013; Renault-Mihara et al., 2017; O’Shea et al., 2024).

2.6. LZK activates AKT in lesion border astrocytes and is beneficial to motor recovery after spinal cord injury

Having identified AKT as a novel target of LZK signaling in vitro, we then assessed the ability of LZK to activate ATK in lesion border astrocytes following spinal cord injury. Astrocytes with activated AKT were quantified based on GFAP⁺pS6⁺ cell count within 250 μm of the lesion border. At 7 days after injury, GFAP⁺pS6⁺ astrocytes were evident surrounding the lesion (Fig. 7A, top). In LZK-OE mice, the number of GFAP⁺pS6⁺ astrocytes at the lesion border was more than double that of the control (Fig. 7A-B). Furthermore, LZK overexpression is sufficient to activate AKT in astrocytes, based on appearance of pAKT⁺GFAP⁺ cells, in the uninjured spinal cord (Fig. 7C). These observations demonstrate pS6 in lesion border astrocytes after spinal cord injury that can be further enhanced by overexpression of astrocytic LZK, and sufficiency of LZK in AKT activation in astrocytes of the naïve spinal cord.

Fig. 7.

LZK activates AKT in lesion border astrocytes and is beneficial to motor recovery after spinal cord injury. (A) Immunofluorescence staining for GFAP and pS6 in injured spinal cords (7dpi) from control (top) and LZK-OE (bottom) mice. Boxed insets are enlarged to show individual and merged channels. Scale bar = 50 μm. (B) Quantification of GFAP⁺pS6⁺ cells per unit area within 250 μm of lesion border in control and LZK-OE mice. n = 3 mice per genotype; graphing group mean with SEM; parametric, unpaired t-test, *p < 0.05. (C) Immunofluorescence staining of GFAP and pAKT (Ser473) in naïve spinal cords from control or LZK-OE mice. Scale bar = 100 μm. (D) Immunofluorescence staining of GFAP and phalloidin (F-actin) in injured spinal cords collected at 1 month post-injury from control (top), LZK-KO (middle), and LZK-OE (bottom) mice. Scale bar = 100 μm. (E) Quantification of phalloidin⁺ signal intensity in GFAP⁺ area within 250 μm of lesion border in the indicated genotypes. n = 3 mice per genotype; graphing group mean with SEM; one-way ANOVA and Tukey’s post hoc pairwise comparison to control, *p < 0.05. (F) Quantification of lesion area in control, LZK-KO, and LZK-OE mice. n = 3 mice per genotype; graphing group mean with SEM; one-way ANOVA and Tukey’s post hoc pairwise comparison to control, *p < 0.05. (G) Quantification of gross hindlimb motor function by BMS in control and LZK-KO mice. n = 13 mice per genotype; graphing group mean with SEM; two-way ANOVA with mixed effects and Bonferroni post hoc multiple comparison test; main genotype effect: p = 0.0298 (See ANOVA Statistics Table in Appendix). (H) Quantification of percentage of mice within each genotype, control or LZK-KO mice from (G), able to obtain the indicated BMS score at each post-injury timepoint of assessment. n = 13 mice per genotype. (I) Quantification of gross hindlimb motor function by BMS in control and LZK-OE mice. n = 9–11 mice per genotype; graphing group mean with SEM; two-way ANOVA with mixed effects and Bonferroni post hoc multiple comparison test; main genotype effect: p = 0.0368 (See ANOVA Statistics Table in Appendix). (J) Quantification of percentage of mice within each genotype, control or LZK-OE mice from (I), able to obtain the indicated BMS score at each post-injury timepoint of assessment. n = 9–11 mice per genotype. (K) Quantification of skilled hindlimb stepping by regular horizontal ladder in control and LZK-KO mice. n = 13 mice per genotype; graphing group mean with SEM; two-way ANOVA with mixed effects; main genotype effect: p = 0.07 (See ANOVA Statistics Table in Appendix). (L) Quantification of skilled hindlimb stepping by regular horizontal ladder in control and LZK-OE mice. n = 9–11 mice per genotype; graphing group mean with SEM; two-way ANOVA with mixed effects; main genotype effect: p = 0.109 (See ANOVA Statistics Table in Appendix).

Next, LZK-dependent cytoskeleton changes in lesion border astrocytes were directly examined in vivo, based on the abundance of filamentous actin (F-actin) that is detectable by phalloidin staining (Tehrani et al., 2014; Garbuzova-Davis et al., 2021). Quantification of phalloidin signal intensity in GFAP⁺ cells lining the lesion edge showed 21 % decrease in phalloidin signal – indicative of reduced F-actin levels – in lesion border astrocytes of LZK-KO mice at 1 month after SCI (Fig. 7D-E). Conversely, a 55 % increase in F-actin levels was observed in lesion border astrocytes of LZK-OE mice (Fig. 7D-E). These findings in vivo and in vitro (Fig. 4) together demonstrate the ability of LZK to enhance actin polymerization in astrocytes. Abundance of F-actin positively correlates with the extent of wound closure that is impaired in LZK-KO mice, but enhanced in LZK-OE mice, as previously reported (Chen et al., 2018) (Fig. 7D-F).

Given that LZK positively regulates border-forming astrogliosis that is essential to CNS wound repair (Okada et al., 2006; Herrmann et al., 2008; Faulkner et al., 2004; Bush et al., 1999; Chen et al., 2018), we evaluated its effects on motor recovery after thoracic spinal cord injury. Without injury, neither LZK-KO or LZK-OE affected baseline gross hindlimb motor performance assessed by the Basso Mouse Scale (BMS) (Basso et al., 2006) (Fig. 7G-J) or skilled hindlimb stepping assessed by regular horizontal ladder (Farr et al., 2006) compared to genotype control mice (Fig. 7K-L). Severity of injury was comparable between control and LZK-KO mice (Fig. 7G-H, 2dpi), and separately between control and LZK-OE mice (Fig. 7I-J, 2dpi). Quantification of group average BMS score over 4 weeks after injury showed a statistically significant main genotype effect of LZK-KO on motor recovery, with LZK-KO mice reaching only half the point value compared to the control mice in the first two weeks after injury, indicating impaired recovery in LZK-KO mice (Fig. 7G). Further breakdown of percentages of mice able to achieve each BMS score showed that 30.8 % and 46.8 % of control mice were able to obtain score of ≥3 at two and four weeks after injury respectively, whereas only 0 % and 20 % of LZK-KO mice were able to obtain score of ≥3 at the same timepoints (Fig. 7H). In separate experiments comparing control and LZK-OE mice, quantification of average BMS scores revealed minor but statistically significant main genotype effect of LZK-OE mice performing better (Fig. 7I). Breakdown of percentages of mice able to obtain each BMS score showed that at four weeks after injury, 12 % of control mice scored above 3, whereas 54.4 % of LZK-OE mice scored above 3 (Fig. 7J). Regular horizontal ladder did not show any difference among the genotypes (Fig. 7K-L). These results show that LZK positively impacts gross locomotor recovery after spinal cord injury, with a more pronounced phenotype resulting from LZK-KO.

3. Discussion

While morphological transformation of lesion border astrocytes is critical to their protective barrier function in CNS wound repair (Okada et al., 2006; Wanner et al., 2013; Renault-Mihara et al., 2017; Robel et al., 2011), cytoskeleton regulation that drives cell shape changes is understudied in lesion border astrocytes. Aberrant morphology of border-forming astrocytes observed in LZK mutant mice (Chen et al., 2018) motivated us to investigate if and how LZK directly controls cell shape of lesion border astrocytes. We found that 1) focal CNS injury activates a transcriptional program of cytoskeleton reorganization that persists in the chronic phase of injury; 2) LZK positively regulates injury-induced transcription of cytoskeleton remodeling genes in lesion border astrocytes and their morphology after spinal cord injury; 3) in vitro assays functionally validate positive effects of LZK on process elongation and movement, with corresponding increase in actin polymerization and microtubule acetylation; 4) LZK activates AKT in astrocytes in vitro and in vivo, which is required for cytoskeleton regulation by LZK to a similar extent as STAT3; and 5) beneficial effects of LZK on motor recovery after spinal cord injury. These findings provide important temporal and molecular insights into the control of morphological adaptation of lesion border astrocytes.

3.1. LZK couples regulation of cell division and morphological transformation of lesion border astrocytes

Border-forming astrogliosis following focal CNS injury is typified by cell division and prominent cell shape changes, two astrocytic responses that are coupled both in function – as they participate in sequential and overlapping processes of tissue replacement and tissue remodeling respectively (Okada et al., 2006; Wanner et al., 2013; Renault-Mihara et al., 2017; Burda and Sofroniew, 2014), and in regulation – as exemplified by STAT3’s dual role in controlling these cellular behaviors (Okada et al., 2006; Wanner et al., 2013; Renault-Mihara et al., 2017). Together with previous report of LZK’s pro-proliferative activity in lesion border astrocytes (Chen et al., 2018), the present discovery of LZK’s ability to directly regulate actin and microtubule cytoskeletons identifies LZK as a signaling molecule that couples the regulation of astrocyte proliferation and morphological transformation.

Transcriptomics and histological analyses of lesion border astrocytes define the window of cell division to be within 7 days after injury (Wanner et al., 2013; O’Shea et al., 2024). In agreement, our analyses of cytoskeleton-related genes in lesion border astrocytes revealed cell cycle-specific microtubule functions at 7 days after injury, but not at later timepoints. Perhaps surprisingly, a majority of injury-induced actin- and microtubule-remodeling genes remains upregulated at 1 month, even until 3 months, in lesion border astrocytes after injury. Such persistent upregulation of cytoskeleton genes suggests their continuous contribution to morphological transformation of lesion border astrocytes that occurs after completion of cell division. We therefore selected representative cytoskeleton remodeling genes that exhibit persistent or delayed upregulation in lesion border astrocytes in order to test the role of LZK in post-mitotic regulation of cell shape. While we tested LZK-dependent regulation on only selective genes, transcriptional control of the cytoskeleton by LZK likely extends beyond these genes to coordinate process elongation, orientation, and local cellular organization of border-forming astrocytes. The prolonged transcription of cytoskeleton components closely reflects the extended process of tissue remodeling by cellular reorganization following tissue replacement by cell division after focal CNS injury (Burda and Sofroniew, 2014). It also discloses a potentially longer time window for modulating morphological plasticity than cell division, in considering approaches to modify the astrocytic border. Lastly, chronic upregulation of cytoskeleton genes at 3 months after injury, a majority of which are associated with actin organization, suggests their functions in maintaining the adapted cell shape of border-forming astrocytes.

3.2. LZK-AKT signaling in cytoskeleton regulation of astrocytes

This study identifies AKT as a novel effector of LZK signaling in the regulation of the astrocytic cytoskeleton. It remains to be determined whether LZK can activate AKT by direct phosphorylation, whether LZK-dependent activation of AKT requires canonical PI3K signaling, and what is the substrate of AKT that transduces LZK-AKT signaling to a transcriptional output. While our findings show that cytoskeleton regulation by LZK-AKT is transcriptional, AKT has been shown to act on the cytoskeleton post-translationally in other mammalian cell types. It is possible that these other AKT-mediated mechanisms are at play in astrocytes, including direct phosphorylation of actin-binding protein Girdin in lamellipodia formation (Enomoto et al., 2005), RhoA activation in F-actin organization (Liu et al., 2013; Deng et al., 2022), GSK-3 suppression in microtubule stabilization (Onishi et al., 2007) and vimentin stabilization in cell motility (Zhu et al., 2011).

LZK-dependent transcription of injury-induced cytoskeleton remodeling genes observed in lesion border astrocytes in vivo are notably recapitulated in vitro. The in vitro system therefore offered an opportunity to begin assessing relative contribution of AKT and STAT3, two effectors of LZK identified in astrocytes thus far by current and previous studies (Chen et al., 2018). The observation that either AKT or STAT3 inhibition was sufficient to completely abolish LZK-dependent effects on the cytoskeleton suggest parallel contribution and/or potential crosstalk between AKT and STAT3 downstream of LZK.

3.3. Cytoskeleton remodeling: a novel function of LZK in astrogliosis

Original cloning of LZK from the human cerebellum and its biochemical characterization as an upstream activator of the classic cellular stress pathway initially suggested its role in the control of stress response in the mammalian CNS⁹⁴. Previous work by us and others have demonstrated its functions in regulating the injury response of mammalian CNS neurons and astrocytes (Chen et al., 2018; Chen et al., 2022; Chen et al., 2016; Saikia et al., 2022; Welsbie et al., 2017). In focal CNS injury, astrocytes surrounding the lesion undergo cell proliferation and extensive cell shape changes to form an astrocytic border that is crucial to wound closure. It was in this context of wound repair astrogliosis following spinal cord injury that LZK was first discovered as an activator of astrocyte reactivity (Chen et al., 2018). Pro-proliferative activity of LZK, via activation of STAT3 (Okada et al., 2006; Herrmann et al., 2008), was previously identified as a mechanism underlying LZK’s ability to drive border-forming astrogliosis (Chen et al., 2018). The reported effects of LZK on astrocyte proliferation and hypertrophy, however, cannot fully explain defects in astrocytic process reorientation at the lesion border evident in LZK loss-of-function mutant, which involves dynamic rearrangement of the cytoskeleton.

The present study addresses this outstanding question by uncovering an additional layer of LZK-dependent control of wound repair astrogliosis, that is, direct transcriptional regulation of actin and microtubule components that contribute to post-mitotic cytoskeleton remodeling of border-forming astrocytes. Furthermore, we identified AKT as a novel downstream signaling target of LZK, and provided first insights into relative contributions of AKT and STAT3 to LZK-dependent cytoskeleton regulation in astrocytes. It is worth noting that while LZK is sufficient to activate its signaling mediators, as well as induce cell proliferation and cytoskeletal regulation of astrocytes in vivo without injury (Chen et al., 2018) (present study), formation of an astrocytic border requires the presence of injury and is potentiated by LZK. Lastly, of all genes differentially regulated by injury from sub-acute to chronic phases of injury in perilesional astrocytes, actin- or microtubule cytoskeleton-related genes comprise only 6–8 % or 4–11 % respectively (Wei et al., 2021), representing a small subset of injury-induced biological processes occurring in lesion border astrocytes (O’Shea et al., 2024). Based on our findings thus far that LZK regulates cytoskeleton remodeling, cell proliferation (Chen et al., 2018), and cytokine production (Chen et al., 2022) in reactive astrocytes in vivo, cytoskeleton regulation by LZK-AKT/STAT3 likely represents only one of many potential biological outcomes of these injury signal transducers acting in a concerted and/or independent manner in astrocytes following injury.

3.4. Comparison of cytoskeleton regulation by LZK and its closely related protein DLK

Together with previous finding that LZK increases protein expression of GFAP and vimentin (Chen et al., 2018), this study expands LZK’s ability to promote morphological transformation of astrocytes by acting on all three cytoskeletal systems – intermediate filaments, actin, and microtubules. Each of these components performs distinct functions, but they also interact and are coordinately regulated to effect astrocyte cell shape changes (Schiweck et al., 2018). While our experiments were guided by in vivo transcriptional outputs of LZK on actin and microtubules, we cannot exclude the possibility that LZK can regulate the cytoskeleton post-transcriptionally and/or post-translationally. Knowledge in these potential mechanisms may be drawn from studies on how dual leucine kinase (DLK), a protein closely related to LZK (Sakuma et al., 1997; Holzman et al., 1994), regulates cytoskeletal dynamics. In neurons, DLK has been shown to upregulate transcription of cytoskeletal genes in response to cytoskeletal stress (DeVault et al., 2024). In the context of axon regeneration that relies on microtubule stability for growth cone formation (Erturk et al., 2007), DLK promotes growth of axonal microtubules by phosphorylation-mediated inhibition of a microtubule depolymerizing kinesin (Ghosh-Roy et al., 2012), stabilizes microtubules (Hirai et al., 2011) via post-translational modification of tubulin (Ghosh-Roy et al., 2012), and facilitates retrograde transport of injury signal from axon to soma implicating a role in microtubule polarization (Shin et al., 2012). While neuronal LZK and DLK are functionally redundant in promoting CNS axon growth (Chen et al., 2016; Saikia et al., 2022), effects of LZK on cytoskeleton has not been examined in any cell type of the nervous system until this study.

Mechanistic insights from this study reveal parallels and divergence in cytoskeleton regulation by LZK in astrocytes and by DLK in neurons. Activation of either kinase elicits transcriptional upregulation of cytoskeletal genes to reorganize actin and microtubules, increases post-translational modifications of α-tubulin associated with microtubule stability, and restructures the cytoskeleton for process extension (current study and published work (DeVault et al., 2024; Ghosh-Roy et al., 2012)). In contrast to the primary outcome of axon growth in neurons, however, astrocytes remodel the cytoskeleton for cell division, process extension, and local cellular reorganization to form a cellular barrier. Additionally, these cell type-specific behaviors may be controlled by different signaling modules: whereas LZK predominantly signals through STAT3 (Chen et al., 2018; Chen et al., 2022) and the presently identified AKT in astrogliosis, LZK/DLK activates MKK4-JNK and is dispensable for PTEN-mTOR signaling in axon growth/regeneration (Chen et al., 2016; Saikia et al., 2022; Hammarlund et al., 2009; Xiong et al., 2010). This study thus reveals cell type-specific utilization of LZK signaling and consequent cytoskeleton-driven cellular behavior.

3.5. Targeting morphology of lesion border astrocytes for CNS repair

The ability of lesion border astrocytes to transform into an interwoven cellular meshwork is critical to their primary functions in damage control and wound compaction after focal CNS injury. Transcriptomics analyses of cytoskeleton structural and regulatory genes in lesion border astrocytes indicate a window of morphological plasticity that is longer than previously appreciated, lasting up to at least 3 months after injury in contrast to the previously characterized 14 days after injury based on histology (Okada et al., 2006; Herrmann et al., 2008; Chen et al., 2018). In our experiments, we manipulated LZK gene expression in astrocytes prior to injury. It remains to be determined whether LZK manipulation during the window of injury-induced morphological plasticity can alter cytoskeleton remodeling of lesion border astrocytes and wound repair, whether targeting downstream effectors of LZK would be more effective and beneficial than LZK itself given the detrimental effects of astrocytic LZK when highly expressed in a CNS-systemic manner (Chen et al., 2018), and whether the persistently upregulated cytoskeleton remodeling genes are required to maintain the adapted cell shape of lesion border astrocytes. One consideration for enhancing lesion border astrogliosis for neurorepair is its potentially negative effect on axon regeneration, as lesion border astrocytes have been traditionally associated with axon growth-inhibitory properties (Silver and Miller, 2004; Davies et al., 1999; Stern et al., 2021). In this regard, it would be of interest to test if manipulating LZK-AKT/STAT3 signaling in lesion border astrocytes following completion of wound closure can reverse compaction of astrocytic processes and impact axon regeneration. Lastly, while our current findings were made in spinal cord injury model that produces dramatic cell shape changes of lesion border astrocytes, they may translate to other CNS pathologies in which subtler alterations of astrocyte morphology are implicated in causation of disease (Endo et al., 2022; Soto et al., 2024). Our findings define temporal and molecular regulation of morphological transformation of lesion border astrocytes, with implications on targeting astrocyte cell shape as a strategy to enhance CNS repair.

4. Methods

4.1. Mice

All mouse husbandry and experimental procedures were performed in compliance with protocols approved by the Institutional Animal Care and Use Committee at the University of Kentucky. GFAP-CreER^T2;LZK^f/f (abbreviated as LZK-KO) mice and GFAP-CreER^T2;LZK^OE (abbreviated as LZK-OE) mice have been previously described (Chen et al., 2018). An additional line of LZK-KO mice expressing tdTomato reporter gene was obtained by breeding GFAP-CreER^T2;LZK^f/f to ROSA26-loxP-STOP-loxP-tdTomato^f/f reporter mice (Madisen et al., 2010). All mice are in C57BL/6 background. To delete LZK from adult astrocytes, LZK-KO mice at the age of 8–10 weeks were each given 75 mg/kg of tamoxifen by intraperitoneal injection once daily, consecutively for 5 days. Tamoxifen-treated LZK^f/f or LZK^f/f; tdT^f/f littermates served as genotype controls. To induce LZK overexpression in adult astrocytes, LZK-OE mice at the age of 8–10 weeks were each given 75 mg/kg of tamoxifen by intraperitoneal injection once daily, consecutively for 2 days. Tamoxifen dosing is reduced in LZK-OE mice to prevent lethality resulting from high expression of exogenous LZK (Chen et al., 2018). Tamoxifen-treated LZK^OE littermates served as genotype controls. 10 days after the last tamoxifen injection, mice were subjected to spinal cord surgery. All surgical procedures, animal dissection, tissue processing, and data analyses were performed by lab members blinded to mouse genotypes.

4.2. Surgical procedures

Mice underwent general anesthesia by intraperitoneal injection of ketamine and xylazine. Complete crush of the spinal cord at thoracic level T8-T9 was performed under a surgical microscope (Zeiss Stemi 2000-C). Similar to previously described (Wanner et al., 2013), following laminectomy of a single vertebra, complete crush of the spinal cord was made by compressing the cord at T8/T9 bilaterally for 5 s using Dumont #5 SuperFine forceps (Fine Science Tools).

4.3. Histology

At the indicated time points after complete spinal cord crush, terminal anesthesia was performed on mice by isoflurane overdose. Mice were perfused intracardially first with ice cold PBS containing heparin (10 units/mL), followed by 4 % paraformaldehyde (PFA). 0.8–1 cm of spinal cords centered at the injury site were then dissected and post-fixed at 4 °C overnight. Tissues were then cryoprotected in 30 % sucrose at 4 °C overnight before embedding. Spinal cords were embedded on dry ice for horizontal sections in the O.C.T compound (Tissue-Tek). Embedded spinal cord tissue blocks were sectioned at 15 μm thickness on a cryostat (Leica CM1860). For further histological examination, free-floating sections were collected in PBS with 0.01 % sodium azide.

4.4. Fluorescence immunohistochemistry

Free-floating spinal cord tissue sections were stained as previously described (Chen et al., 2018). Tissues were washed three times with wash buffer (0.2 % triton in PBS), blocked and permeabilized with blocking buffer (0.4 % triton, 5 % bovine serum albumin in PBS with 1× sodium azide) for 1 h at room temperature, and incubated with primary antibody (diluted in 0.2 % triton, 1 % BSA in PBS with 1× sodium azide) overnight at 4 °C with rotation. The next day, tissues were washed three times in wash buffer and incubated with secondary antibodies for 2 h at room temperature with rotation. After two washes in wash buffer, tissues were stained with DAPI (Sigma) (1 μg/mL in PBS) for 10 min, then mounted on glass slides using Fluoromount G media (Thermo Fisher).

4.5. Transcriptomics analyses of lesion border astrocytes after spinal cord injury

RNA-sequencing dataset GSE153720 (Wei et al., 2021) was used to determine transcription levels of cytoskeleton-related genes in lesion border astrocytes at 7dpi, 1 M, and 3 M after SCI. Differentially upregulated genes (Log2FC > 1, p < 0.05) and downregulated genes (Log2FC < −1, p < 0.05) at each post-injury timepoint compared to sham were used to identify gene ontology-biological process (GO-BP, p < 0.05) by Panther (http://www.pantherdb.org/). Biological processes with >2 fold enrichment and inclusion of ≥5 input genes were selected for further analysis (Table 2–1). Commonly upregulated GO-BPs related to actin and microtubule organization with FDR < 0.05 were compared. Actin- and microtubule-related gene lists were created based on GO-BP datasets using Mouse Genome Informatics (MGI) gene ontology database (Hayamizu et al., 2015) (Table 2–2). These two gene lists were compared to the DEGs identified from GSE153720 to generate heatmaps for actin- and microtubule-associated genes differentially expressed at each post-injury timepoint (Log2FC > 1, p-adj ≤ 0.05). ShinyGO v0.8 was used to identify enriched GO terms of interest from the GO-BP database.

4.6. Astrocyte isolation from adult mouse spinal cords

Mice were perfused intracardially with ice cold PBS containing heparin (10 units/mL) to clear blood. 0.5 cm of spinal cords centered at the injury site were then rapidly harvested and immediately used for cell dissociation by Neural Tissue Dissociation Kit (Miltenyi Biotec), debris and red blood cell removal by Debris Removal Solution (Miltenyi Biotec) and Red Blood Cell Lysis Solution (Miltenyi Biotec), according to manufacturer’s protocols. Lastly, astrocytes were magnetically labeled using a mouse anti-ACSA2 MicroBead Kit (Miltenyi Biotec) and separated from other neural cells using MACS columns following the manufacturer’s protocol.

4.7. mRNA quantification by qRT-PCR

Total RNA from astrocytes isolated from adult spinal cords or primary astrocytes was obtained using RNA Extraction Kit (Takara), and concentration was measured by NanoDrop spectrophotometer. According to the manufacturer’s protocol, 500 ng of RNA was converted into cDNA using Verso cDNA Synthesis Kit (Fisher Scientific). Quantitative reverse transcription polymerase chain reaction (qRT-PCR) was run on a QuantStudio 7 Flex System using iTaq Universal SYBR Green (Bio-Rad) at 95 °C for 10 min, followed by 45 cycles of 10 s at 95 °C and 30 s at 60 °C. Target specificity of all qRT-PCR primers used in this study was validated by appearance of a single peak on melting curve analysis, and primer sequences are listed at the end of Materials and Methods. Relative gene expression based on mRNA levels was quantified by 2^-ΔΔCt method using 18srRNA as a housekeeping gene.

4.8. Primary astrocyte culture

Primary cortical astrocytes were cultured as previously described (Schildge et al., 2013). Briefly, prior to astrocyte isolation, T-25 flasks were coated with 50 μg/mL poly-D-lysine (Sigma) overnight, washed three times with Dulbecco’s Phosphate-Buffered Saline DPBS (Sigma), one time with sterilized water, then left to dry in biosafety cabinet. Postnatal day 2 (P2) mouse cerebral cortices were dissected, meninges removed, and minced with a razor blade. Dissociated tissue was then washed with DMEM (Thermo Fisher) supplemented with 1 % L-Glutamine (Thermo Fisher), and 1 % Penicillin-Streptomycin (Thermo Fisher), followed by incubation in 2.5 % trypsin (Thermo Fisher) and 100 μL/mL DNase I (StemCell Technologies) for 5 min at 37 °C. Tissue was then washed with culture media (DMEM supplemented with 10 % heat-inactivated FBS, 1 % L-Glutamine, and 1 % Penicillin-Streptomycin) to inactivate and remove trypsin and DNaseI and triturated with a 1 mL open ended pipette in culture media. Dissociated cells were passed through a 40 μM cell strainer to remove aggregates, pelleted by centrifugation, resuspended in culture media, and seeded at 1.5 × 10⁶ cells/25cm² flask. Cells were maintained in tissue culture incubator at 37 °C with 5 % CO₂. Culture media was replaced every two to three days. After 7 days in culture, astrocytes were enriched by flask shaking at 180 rpm for 48 h to remove non-adherent cells. Remaining adherent astrocytes were reseeded at 0.5 × 10⁶ cells/75cm² flask or 0.1 × 10⁶ cells/well in a 12-well plate for subsequent assays.

4.9. 4-OHT-induced gene recombination in vitro

To test effects of astrocytic LZK deletion in vitro, GFAP-CreER^T2; LZK^f/f pups, and LZK^f/f littermate controls were produced. To test effects of astrocytic LZK overexpression in vitro, GFAP-CreER^T2; LZK^OE pups, and LZK^OE littermate controls were produced. Each pup was individually genotyped using tail tissue, separately dissected for primary astrocyte isolation, and the resulting culture was separately maintained. See “Primary Astrocyte Culture” section for details of cortical astrocyte isolation. Primary cortical astrocytes grown on a 12-well plate for qRT-PCR and a 6-well plate for western blot to a monolayer of 70 % confluency were incubated with culture media containing 1 μM 4-hydroxy-tamoxifen (4-OHT) (Bio-techne) for three consecutive days, followed by replacement with complete media without 4-OHT. Astrocytes were maintained in culture media for three more days before subsequent analyses started or before cell collection.

4.10. Plasmid transfection

For subsequent plasmid transfection experiments, wildtype pups were produced for cortical astrocyte isolation, and cerebral cortices of wildtype littermates were combined to establish primary cortical astrocyte culture. Primary cortical astrocytes were seeded in PDL-coated, 12-well plates at a density of 0.1 × 10⁶ cells/well 24 h before transfection. Astrocyte monolayer was grown to 70 % to 80 % confluency for transfection. Culture media was replaced with Opti-MEM (Fisher Scientific) 30 min before transfection. Astrocytes were transfected with pBI empty vector, pBI-LZK, or pBI-LZK-K195A (Chen et al., 2016) as indicated, at 0.5 μg of DNA/well using Lipofectamine 3000 (Thermo Fisher) following the manufacturer’s protocol. For AKT or STAT3 inhibition experiments, AKT inhibitor VIII (Sigma) or STAT3 inhibitor C188–9 (Selleckchem) was added to astrocytes 48 h after transfection for cell collection or subsequent assays (see below).

4.11. AKT and STAT3 inhibitor treatments in vitro

AKT inhibitor VIII (Sigma) was used at 2.5 μM, STAT3 inhibitor C188–9 (Selleckchem) was used at 1 μM, and DMSO was used as vehicle control on primary cortical astrocytes. Primary astrocytes were treated with inhibitors for 16 h before cell collection for qRT-PCR or immuno-blotting. For experiments involving plasmid-transfected primary astrocytes, inhibitors were added to cells 48 h after completion of transfection, and cells were incubated with the inhibitors for 16 h before cell collection or the start of subsequent assays.

4.12. Scratch wound assay

Scratch wound assay was adapted from previously described (Etienne, 2006). Astrocytes were seeded on a PDL-coated 12-well plate at a density of 0.1 × 10⁶ cells/well. A day before performing the assay, culture media was replaced with low-serum media (DMEM supplemented with 1 % FBS, 1 % L-Glutamine, and 1 % Pen-strep). Scratch was made on astrocyte monolayer using a sterile 1 mL pipette tip. Each well was rinsed thoroughly with sterile PBS to remove detached cells and replaced with fresh low-serum media. Cells were maintained in tissue culture incubator at 37 °C with 5 % CO₂ throughout the assay. Images were taken at the indicated time points for analyses.

4.13. Protein co-immunoprecipitation

Primary astrocytes were transfected with pEF-FLAG-LZK and HA-AKT plasmids using Lipofectamine 3000 as described above. FLAG-LZK-only transfection served as negative control. 48 h after transfection, cells were lysed using RIPA lysis buffer (150 mM NaCl, 1 % NP40, 50 mM Tris-HCl, pH = 7.5, 5 mM EDTA, pH = 8, 5 mM EGTA, pH = 8) supplemented with cOmplete mini EDTA-free protease and phosphatase inhibitor cocktails (Sigma). 2.5 μg of HA antibody (Thermo Fisher, 26183) was first conjugated to protein A/G PLUS agarose beads (Thermo Fisher) by overnight incubation at 4 °C. Separately, total cell lysates were pre-cleared by incubating with the beads for 2 h at 4 °C. Per immunoprecipitation, lysates containing 0.5 μg of total protein were incubated with antibody-conjugated beads overnight at 4 °C. Beads were then washed three time in IP buffer on ice, followed by boiling in 2× sample buffer for 10 min at 95 °C for western blotting.

4.14. Immunoblotting

Primary astrocytes were lysed in RIPA buffer (150 mM NaCl, 1 % NP40, 50 mM Tris-HCl, pH = 7.5, 5 mM EDTA, pH = 8, 5 mM EGTA, pH = 8) supplemented with cOmplete mini EDTA-free protease and phosphatase inhibitor cocktails (Sigma) and boiled for 5 min in 4× SDS sample buffer. 40 μg of protein samples were separated by 10 % SDS-polyacrylamide gels and electro-transferred onto the nitrocellulose membrane. The membrane was blocked with 5 % non-fat milk or 5 % BSA (for phosphorylated proteins), followed by primary antibody incubation at 4 °C overnight. After repeated washes with PBST or TBST, the blot was incubated with secondary antibodies at room temperature for 1 h. Primary antibodies and secondary antibodies are listed at the end of Materials and Methods. Images were acquired using the Odyssey infrared imaging system (LI-COR Biosciences, USA) and Bio-Rad Chemidoc imaging system. Protein levels of each target were normalized to those of loading control β-actin based on signal intensity. Image Studio Lite version 5.2 or ImageJ software was used for quantification.

4.15. F-actin and G-actin isolation

Fractionation of filamentous (F-actin) and globular actin (G-actin) was performed following the manufacturer’s protocols (BK037 Cytoskeleton). Briefly, the monolayer of astrocytes was harvested by adding lysis buffer (100 μL of LAS01 buffer, 10 μL of 100 mM ATP, and 10 μL of 100× protease inhibitor cocktail stock) and homogenized by a 25G syringe. The cell lysate was incubated at 37 °C for 10 min. The lysate was centrifuged at 100,000 ×g at 37 °C for 1 h. After centrifugation, the supernatant containing G-actin was collected. The F-actin depolymerization buffer was added to the pellet and incubated on ice for 1 h. In the final step, 4× SDS sample buffer was added to each pellet and supernatant sample. Protein samples were separated by SDS-PAGE and transferred to a nitrocellulose membrane, with a reference amount of actin used for quantification.

4.16. Fluorescence immunocytochemistry

Primary astrocytes grown on PDL-coated glass coverslips were fixed with 4 % PFA for 5 min, permeabilized with wash buffer (0.1 % Triton ×−100 in PBS), and blocked with 5 % BSA in PBS with sodium azide for 30 min. Cells on coverslips were incubated with primary antibody buffer (1 % BSA in PBS) overnight at 4 °C and washed three times with wash buffer for 10 min each at room temperature. Cells were then incubated with secondary antibody buffer (1 % BSA in PBS) for 1 h at room temperature, washed three times with wash buffer, and stained with DPAI (0.5 μg/mL in PBS) for 10 min. After two washes in PBS, coverslips were mounted onto glass slides using Fluoromount-G mounting media (Thermo Fisher). Primary and secondary antibodies used in this study are listed at the end of Materials and Methods.

4.17. Phalloidin staining

Primary astrocytes were plated on PDL-coated glass coverslips and fixed with 4 % PFA for 5 min following wound scratch at the indicated time points. After fixation, cells were first permeabilized with 0.1 % Triton X-100 for 3 min and blocked with 1 % BSA in PBS-sodium azide for 30 min. Cells were then incubated with Alexa fluor 488 phalloidin (1:800, Invitrogen) for 45 min, washed three times in PBS, stained with DAPI (0.5 μg/mL) for 10 min, washed two times with PBS, and coverslips were mounted onto glass slides using Fluoromount-G mounting media (Thermo Fisher). For phalloidin staining of spinal cord tissue sections, slices were incubated with phalloidin for 45 min following secondary antibody incubation.

4.18. Microscopy and image analyses

Zeiss Axioscan Z1 Slide Scanner and Nikon Eclipse Ti 2 confocal microscope were used to take images. Software ImageJ was used for image analyses. For analyses of injured spinal cord, n = 3 mice per genotype, and 6 sections per mouse were used. For GFAP⁺ and pS6⁺ cell count, the number of cells double positive for GFAP and pS6 by placing sampling frames of 64 × 64 μm² within 250 μm of lesion border. 16 sampling frames were applied per section. To quantify phalloidin signal intensity in lesion border astrocytes, GFAP⁺ areas immediately surrounding the lesion (sampling frames of 200 × 200 μm² within 250 μm of GFAP⁺/GFAP⁻ interface) were selected as regions of interest. Integrated density of phalloidin signal within GFAP⁺ regions was measured. 8 sampling frames per section; 4 sections per mouse; n = 3–6 mice per genotype. For quantification of lesion size, area bound by GFAP⁺ lesion border was traced and measured in horizontal sections containing the injury site from 1 serie of adjacent sections per animal, then averaged per group (n = 3–6 mice per genotype).

For scratch wound assay, live cells were imaged immediately after scratch (0 h), 12 h, and 24 h after scratch. We performed two independent experiments, each including at least one biological replicate per genotype. In each experiment, we used three wells per condition per genotype. Imaging was performed by capturing 6 frames per well. Scratch wound closure was quantified as gap area at 12 h or 24 h normalized to cell-free area at 0 h within each image frame. Using the same image frames, the number of astrocytes was quantified at 0, 12, and 24 h after scratch. Cell counts at 12 and 24 h were normalized to those at 0 h.

To analyze the morphology of astrocytes at scratch border, length of the longest phalloidin⁺ protrusion extended into the cell free space in the direction of movement was measured from the back of the nucleus to the tip of the protrusion; the width of the protrusion was measured at the nucleus-proximal end of the protrusion perpendicular to the length, as previously described (Liao et al., 2015; Devitt et al., 2021). The ratio of length/width of each cell was quantified. n = 3–4 mice per genotype. In each experiment, we used three wells per condition per genotype. Imaging was performed by capturing 6 frames per well.

Orientation of astrocytic processes at the lesion border was determined by process-to-lesion angle. The angle of each GFAP⁺ process with respect to the closest lesion edge was measured using Directionality Plugin in ImageJ. Orientation of 30–60 astrocytic processes at the lesion border were quantified per genotype, based on 4 sampling frames per tissue section, 6 sections per animal, and 3–5 mice per genotype.

Length of the longest cellular process per astrocyte was measured at 300 μm or 600 μm rostral or caudal from the lesion border. Freehand line tool in ImageJ was used to trace the longest cellular process per astrocyte, measured from DAPI⁺ nucleus to the end of longest GFAP⁺ process. Total of 35–40 astrocytes at each distance were quantified per genotype, based on 4 sampling frames per tissue section, 6 sections per animal, and 3–5 mice per genotype.

4.19. Mouse behavioral analyses

Observers were blinded to genotype in all behavioral tests. Mice were acclimated to each task 3–7 days before injury. Basso Mouse Scale (BMS) open field test was used to evaluate gross hindlimb motor function (Basso et al., 2006). Each animal had one trial at each timepoint. Regularly spaced horizontal ladder (Farr et al., 2006) scoring was modified to assess post-SCI skilled hindlimb stepping. A weight-supported paw placement on a rung without slipping or misplacing the paw from step takeoff to landing was counted as a successful step. Number of successful steps was normalized to the total number of steps taken in each trial. Each animal had three trials at each timepoint that were averaged.

4.20. Statistical analysis

Statistical analysis was performed using GraphPad Prism software. All data are presented as mean value per group with SEM as error bars. “n” represents the number of animals or the number of cells per genotype or treatment. Unpaired parametric two-tailed t-test was used for single comparison between two groups. Depending on number of groups and variables, one- or two-way ANOVA was used for other comparisons (see details in Figure Legends, and ANOVA Statistics Table in Appendix). Post hoc pairwise comparisons were performed only if main effect was statistically significant, see method of correction applied in figure legends. In all tests, statistical significance was defined as p < 0.05.

Supplementary Material

Supplementary data to this article can be found online at https://doi.org/10.1016/j.expneurol.2025.115379.

Appendix

A.1. ANOVA Statistics Table

Number of Figure	Type of Test	p-values	F values
Fig. 4B	Two-way ANOVA with repeated measures	Time, p < 0.0001 Genotype, p < 0.0001 Interaction, p = 0.0122	Time, F (1, 14) = 56.24 Genotype, F (2, 14) = 92.99 Interaction, F (2, 14) = 6.141
Fig. 4D	Two-way ANOVA with repeated measures	Time, p = 0.0105 Genotype, p < 0.0001 Interaction, p < 0.0001	Time, F (1, 169) = 6.698 Genotype, F (2, 169) = 113.0 Interaction, F (2, 169) = 12.51
Fig. 5B	One-way ANOVA	Genotype, p < 0.0001	Genotype, F (2, 172) = 29.39
Fig. 6M-Lzk	One-way ANOVA	Treatment, p = 0.0106	Treatment, F (2, 6) = 10.64
Fig. 6M-Rac1	One-way ANOVA	Treatment, p = 0.0093	Treatment, F (2, 6) = 11.27
Fig. 6M-Rgs14	One-way ANOVA	Treatment, p <0.0001	Treatment, F (2, 6) = 253.3
Fig. 6M-Tpx2	One-way ANOVA	Treatment, p = 0.0049	Treatment, F (2, 6) = 14.67
Fig. 6M-Iqgap3	One-way ANOVA	Treatment, p = 0.0209	Treatment, F (2, 6) = 7.888
Fig. 6M-Coro1a	One-way ANOVA	Treatment, p = 0.0317	Treatment, F (2, 6) = 6.477
Fig. 6M-Cotl1	One-way ANOVA	Treatment, p = 0.0003	Treatment, F (2, 6) = 41.70
Fig. 6O	One-way ANOVA	Treatment, p < 0.0001	Treatment, F (5, 12) = 1598
Fig. 6P	One-way ANOVA	Treatment, p < 0.0001	Treatment, F (5, 12) = 976.9
Fig. S1F-Rac1	Two-way ANOVA	Treatment, p < 0.0001 Genotype, p = 0.0003 Interaction, p = 0.131	Treatment, F (2,12) = 32.84 Genotype, F (1,12) = 25.25 Interaction, F (2, 12) = 2.409
Fig. S1F-Rgs14	Two-way ANOVA	Treatment, p < 0.0001 Genotype, p < 0.0001 Interaction, p < 0.0001	Treatment, F (2, 12) = 584.8 Genotype, F (1, 12) = 418.5 Interaction, F (2, 12) = 349.6
Fig. S1F-Tpx2	Two-way ANOVA	Treatment, p < 0.0001 Genotype, p = 0.0008 Interaction, p = 0.0181	Treatment, F (2, 12) = 28.02 Genotype, F (1, 12) = 19.48 Interaction, F (2, 12) = 5.707
Fig. S1F-Iqgap3	Two-way ANOVA	Treatment, p < 0.0001 Genotype, p < 0.0001 Interaction, p < 0.0001	Treatment, F (2, 12) = 103.4 Genotype, F (1, 12) = 90.33 Interaction, F (2, 12) = 29.95
Fig. S1F-Coro1a	Two-way ANOVA	Treatment, p < 0.0001 Genotype, p < 0.0001 Interaction, p < 0.0001	Treatment, F (2, 12) = 462.8 Genotype, F (1, 12) = 219.7 Interaction, F (2, 12) = 223.0
Fig. S1F-Cotl1	Two-way ANOVA	Treatment, p < 0.0001 Genotype, p < 0.0001 Interaction, p < 0.0001	Treatment, F (2, 12) = 93.22 Genotype, F (1, 12) = 78.11 Interaction, F (2, 12) = 35.12
Fig. S1F-Topo2a	Two-way ANOVA	Treatment, p < 0.0001 Genotype, p < 0.0001 Interaction, p < 0.0001	Treatment, F (2, 12) = 136.7 Genotype, F (1, 12) = 66.69 Interaction, F (2, 12) = 51.45
Fig. S1F-Mkif67	Two-way ANOVA	Treatment, p < 0.0001 Genotype, p < 0.0001 Interaction, p < 0.0001	Treatment, F (2, 12) = 98.16 Genotype, F (1, 12) = 34.75 Interaction, F (2, 12) = 33.86
Fig. 7G	Two-way ANOVA with mixed effects 2 mice died on 14dpi.	Time, p < 0.0001 Genotype, p = 0.0298 Interaction, p = 0.08	Time, F (2.196, 50.50) = 300.8 Genotype, F (1, 24) = 5.337 Interaction, F (4, 92) = 2.134
Fig. 7I	Two-way ANOVA with mixed effects 1 mouse died on 14dpi and 1 mouse died on 21dpi.	Time, p < 0.0001 Genotype, p = 0.0368 Interaction, p = 0.26	Time, F (4, 71) = 320.1 Genotype, F (1, 18) = 5.087 Interaction, F (4, 71) = 0.8197
Fig. 7K	Two-way ANOVA with mixed effects 2 mice died on 14dpi.	Time, p < 0.0001 Genotype, p = 0.07 Interaction, p = 0.2	Time, F (1.512, 35.54) = 832.1 Genotype, F (1, 24) = 3.514 Interaction, F (2, 47) = 1.615
Fig. 7L	Two-way ANOVA with mixed effects 1 mouse died on 14dpi and 1 mouse died on 21dpi.	Time, p < 0.0001 Genotype, p = 0.109 Interaction, p = 0. 6	Time, F (2, 53) = 9386 Genotype, F (1, 53) = 2.493 Interaction, F (2, 53) = 0.4644

A.2. Antibodies

A.2.1. Primary antibodies used for immunostaining

Primary Antibodies	Catalog Number	Dilution
α-tubulin	Abcam, ab18251	1:1000
Acetylated α-tubulin (Lys40)	Sigma, T7451	1:15000
GFAP	Invitrogen, 13–0300	1:1000
Ki-67	BD Bioscience, 550609	1:1000
pS6	Cell signaling, 4858 T	1:2000

A.2.2. Primary antibodies used for immunoblotting

Primary Antibodies	Catalog Number	Dilution
β-actin	Thermo Fisher, MA5–11869	1:5000
AKT	Cell signaling, 9272	1:1000
phospho-AKT (Ser473)	Cell signaling, 9271	1:1000
MAP3K13 (LZK)	Sigma, HPA016497	1:300
α-tubulin	Abcam, ab18251	1:2000
Acetylated α-tubulin (Lys40)	Sigma, T7451	1:2000
pS6	Cell signaling, 4858 T	1:1000

A.2.3. Secondary antibodies for immunostaining

Secondary Antibodies	Catalog Number	Dilution
Alexa 647 conjugated anti-rabbit IgG	Life Technologies A21245	1:300
Alexa 488 conjugated anti-mouse IgG	Life Technologies A21202	1:300
Alexa 647 conjugated anti-Rat IgG	Life Technologies A21472	1:300

A.2.4. Secondary antibodies for immunoblotting

Secondary Antibodies	Catalog Number	Dilution
IRDye 680RD Goat anti Rabbit-926–68,071	D10603–15	1:10000
IRDye 800CW Goat anti-Mouse −926–32,210	D10825–15	1:10000
IRDye 680RD Goat anti Mouse −926–68,070	D10901–15	1:10000
IRDye 800CWGoat anti Rabbit −926–32,211	D10811–25	1:10000

A.3. Primers for qRT-PCR

Primer	Sequence
Rac1 Fw	AGGGGCAAAGACAAGCCGA
Rac1 Rv	ATCAAGCTTCGTCCCCACGAG
Iqgap3 Fw	CTGCTGGCCCTAGCCAAGAT
Iqgap3 RV	TGGACGTTGGCGAGGTTGAT
Met Fw	GACAAGACCACCGAGGATGGC
Met Rv	TCTGCCGTGAAGTTGGGGAG
Dock2 Fw	AGAGAACAGGAGGCTCTCAAGG
Dock2 Rv	GTACCCTCTGTACCAGTCTCCA
Cotl1 Fw	CTCGGCGGTCATCTGGGTGACTT
Cotl1 Rv	ACAAAGGCGAACAGCCGGAC
Coro1a Fw	ACTCAGGCACTTCACCTGCTT
Coro1a Rv	GTCCAAACACGTGGCGGAAT
Phdlb2 Fw	CCTGCTGATGCTGATGCTGC
Phdlb2 Rv	AAGTGCTGTTGCTCTTGCGG
Rgs14 Fw	AGACCAGGAAGTGCGACTGG
Rgs14 Rv	TCCACCAGGCCTTCAATGTCA
Nuak-2 Fw	TGAAGTGGGCAGATCACGCT
Nuak-2 Rv	CGGAAGAAATGCCTGGCGTC
Tubb4a Fw	AAGCCGGTCAATGCGGTAAC
Tubb4a Rv	AGATCTGGCCAAAAGGGCCG
Tpx2 Fw	GACTTCGGGGCGTTGGTATT
Tpx2 Rv	TCTCCTCAATAGACTGTCTTCATGT
LZK Fw	GGAGCTCAGGCATGCTCTGG
LZK Rv	CAGGTCGTCTGGCCGCTTAGTCTGC
18S Fw	GTTGGTTTTCGGAACTGAGGC
18S Rv	GTCGGCATCGTTTATGGTCG
Mkif67 Fw	GAGGAGAAACGCCAACCAAGAG
Mkif67 Rv	TTTGTCCTCGGTGGCGTTATCC
Topo2a Fw	CAAGCGAGAAGTGAAGGTTGCC
Topo2a Rv	GCTACCCACAAAATTCTGCGCC

Data availability

Data will be made available on request.