Apolipoprotein E promotes white matter remodeling via the Dab1‐dependent pathway after traumatic brain injury

Abstract Introduction Axonal injury results in long‐term neurological deficits in traumatic brain injury (TBI) patients. Apolipoprotein E (ApoE) has been reported to activate intracellular adaptor protein Disabled‐1 (Dab1) phosphorylation via its interaction with ApoE receptors. The Dab1 pathway acts as a regulator of axonal outgrowth and growth cone formation in the brain. Aims We hypothesized that ApoE may alleviate axonal injury and regulate axonal regeneration via the Dab1 pathway after TBI. Results In this study, we established a model of controlled cortical impact (CCI) to mimic TBI in vivo. Using diffusion tensor imaging to detect white matter integrity, we demonstrated that APOE‐deficient mice exhibited lower fractional anisotropy (FA) values than APOE+/+ mice at 28 days after injury. The expression levels of axonal regeneration and synapse plasticity biomarkers, including growth‐associated protein 43 (GAP43), postsynaptic density protein 95 (PSD‐95), and synaptophysin, were also lower in APOE‐deficient mice. In contrast, APOE deficiency exerted no effects on the levels of myelin basic protein (MBP) expression, oligodendrocyte number, or oligodendrocyte precursor cell number. Neurological severity score (NSS) and behavioral measurements in the rotarod, Morris water maze, and Y maze tests revealed that APOE deficiency caused worse neurological deficits in CCI mice. Furthermore, Dab1 activation downregulation by the ApoE receptor inhibitor receptor‐associated protein (RAP) or Dab1 shRNA lentivirus attenuated the beneficial effects of ApoE on FA values, GAP43, PSD‐95, and synaptophysin expression, and neurological function tests. Additionally, the effects of ApoE on axonal regeneration were further validated in vitro. In a mechanical scratch injury model of primary cultured neurons, recombinant ApoE protein treatment enhanced axonal outgrowth and growth cone formation in injured neurons; however, these effects were attenuated by Dab1 shRNA, consistent with the in vivo results. Conclusion Collectively, these data suggest that ApoE promotes axonal regeneration partially through the Dab1 pathway, thereby contributing to functional recovery following TBI.


| INTRODUC TI ON
Traumatic axonal injury (TAI) is involved in almost all forms of brain trauma, especially traumatic brain injury (TBI), which results in longterm disability. TAI tends to disrupt the normal axoplasmic transport and electric signal transmission of axons, resulting in progressive axonal degeneration. However, axonal regeneration begins at some point after injury. Guided by the newly regenerated growth cone, residual axons spontaneously sprout from residual injured axonal tips and, therefore, potentiate the extension of axons to the destined area. However, the sprouting ability of injured neurons in the adult central nervous system (CNS) is very limited, accounting for the poor functional recovery after TBI. 1 Axonal regeneration is essential for synapse formation and synaptic maintenance, both of which are essential for neuronal circuit formation and function and ultimately contribute to functional recovery. Hence, strategies to promote axonal regeneration may provide promising benefits for TBI therapies.
Apolipoprotein E (ApoE) is a plasma protein responsible for transporting lipids and cholesterol. Moreover, ApoE is pivotal in modulating CNS neuronal responses to injury. 2 As the dominant apolipoprotein in the CNS, ApoE has been suggested to support a protective role of ApoE in TBI in many studies. For example, ApoE regulates neurogenesis, blood-brain barrier (BBB) integrity, inflammatory responses, and cell apoptosis after TBI. [3][4][5] Furthermore, neurite outgrowth in peripheral nerves is impaired in APOE-deficient mice, 6 while ApoE-mimetic peptide promotes axonal regeneration and myelination after peripheral axonal injury. 7 However, whether ApoE also contributes to axonal regeneration in CNS neural injury remains largely unknown.
ApoE contains two major structural domains: The N-terminal domain contains the low-density lipoprotein receptor (LDLR) binding region, and the C-terminal domain contains the lipid-binding region.
By binding to its receptors in the neural cell membrane, ApoE activates several intracellular kinases, resulting in the phosphorylation of intracellular adaptor protein Disabled-1 (Dab1). 8,9 Extensive evidence has revealed that ApoE receptors, especially the very low-density lipoprotein receptor (VLDLR) and the apolipoprotein E receptor 2 (ApoER2), cluster upon ligand binding and recruit Dab1 to an NPxY motif in their cytoplasmic domains, which leads to the tyrosine phosphorylation of Dab1 by nonreceptor tyrosine kinases of the Src family. 10 Various downstream signaling pathways of Dab1 connect ApoE to the actin and microtubule cytoskeleton. A central step in Dab1-driven dendrite outgrowth and growth cone formation is facilitated by the activation of the Rho-GTPase protein cell division control 42 (Cdc42). 11 Cdc42 is a monomeric G protein that is switched on when bound to GTP and switched off when bound to GDP. Activated Cdc42 is pivotal in regulating the cytoskeleton dynamics of actin filaments and microtubules. Cdc42 promotes actin accumulation in the growth cone to potentiate axon elongation. 12 The abovementioned evidence supports the activation of the Dab1 pathway as a therapeutic strategy in axonal regeneration. However, the effect of the Dab1-Cdc42 pathway in TBI needs to be further clarified. In the current study, we hypothesized that ApoE may promote axonal regeneration following TBI via a Dab1-dependent pathway.

| Controlled cortical impact (CCI) model
The CCI model was produced by the TBI-0310 TBI Model system (Precision Systems and Instrumentation, USA), as previously described. 13 Each mouse was anesthetized with an induction gas mixture (3% isoflurane with 1 L/min 100% oxygen). Once the mouse was fully sedated (steady breath rate coupled with the absence of toepinch reflex), the mouse was removed from the induction chamber, placed abdomen-down on a heating pad (37℃), and administered a maintenance gas mixture (1.5% isoflurane (range: 0.9%-1.8%) with 0.5 L/min 100% oxygen) for the remainder of the surgery. Briefly, a 5-mm left lateral craniotomy centered at 1.0 mm lateral to the midline and 3.0 mm anterior to lambda was performed. The CCI injury was produced using a pneumatic cylinder with a 3-mm diameter flattip impounder at an impact velocity of 6.0 m/s, a dwell time of 40 ms, and a cortical contusion depth of 0.6 mm. The body temperature of each mouse was maintained at 36-37°C throughout the duration of the surgery.

| Diffusion tensor imaging (DTI)
DTI was used to detect axonal regeneration in the corpus callosum (CC) after CCI. All magnetic resonance imaging (MRI) scans were performed on a Bruker 7T (70/20) system (Bruker Biospin). Mice were imaged before injury using DTI so that each mouse could serve as its own control. Each mouse was anesthetized and mounted on a Bruker animal bed, and body temperature was maintained at 37°C with respiratory rate continuously monitored. DTI images were acquired with a single-shot spin-echo echo-planar imaging (EPI) sequence

| Behavioral tests
For all behavioral testing, experimenters were blinded to injury, genotype status, and experimental conditions. N = 10 for each group.
Neurological severity score (NSS) test: Before and after the CCI model (one day preinjury and day 1, day 3, day 7, day 14, and  No difference was detected between the two groups in total axon length. Scale bar = 100 µm. (F) Representative bands and (G) quantitative analysis of GAP43 before injury and at 3 days or 28 days after CCI. Compared with the level in the APOE +/+ group, the GAP43 level was significantly decreased in the APOE −/− group at both 3 and 28 days postinjury (P < .05). N = 6/group. *, P < .05 A goal platform 8 cm in diameter was positioned 1 cm below the surface of the water (hidden platform). After five sets of hidden platform trials, two sets of visible platform trials were performed. The maximum time allotted to reach the platform was 90 s. Performance in the hidden and visible platform trials was quantified as latency to reach the platform in seconds.
Y maze: Spontaneous alternation behavior, as an index of attention, was evaluated by the Y maze test. The test was performed using the reported procedure and was conducted between 9 AM and 4 PM at 28 days postinjury. The mice were moved into the behavioral room for at least 1 hour before testing. The experiment was performed at 35 lux. Before the behavioral test, the mice were placed in one of the compartments and allowed to move freely in one of the arms for 10 min. Each mouse performed one trial. An arm entry was defined as three legs entering one of the arms.

| Western blot analysis
For the in vivo study, mice from each group were sacrificed, brain tissues were obtained, and the ipsilateral hemisphere was col- for each group.

| Immunocytochemistry
Cultured cells were fixed with 4% paraformaldehyde, followed by incubation with 0.1% Triton X-100. Cell slides were incubated with anti-βIII-tubulin (1:500, Sigma) primary antibodies. The resulting slides were observed through a fluorescence microscope (DM6000, Leica), and images were captured with the Leica Application Suite.
For growth cone analysis, the same immunocytochemical protocol for ApoE (1:200, Santa Cruz) was followed, plus phalloidin (1:100, Sigma) incubation before laser confocal microscopy (Leica Microsystem) scanning. The images were captured with Nikon Elements Imaging software.
To analyze axonal regeneration, we assessed the total axon length of βIII-tubulin-positive axons crossing the scratched blank regions with ImageJ software (NIH). To assess growth cone formation, we determined the average area per growth cone and the number of growth cones per image area in the lesion gap. Ten images from five separate wells were randomly measured in each experimental group.
The growth cone area was determined by drawing a polygon around the palm of the growth cone, followed by area measurement using ImageJ. N = 6 for each group.

| Statistical analysis
The results are represented as the mean ± standard deviation (SD). Statistical analysis was performed using Prism version 5 (GraphPad). Motor and MWM test data from the APOE-deficient mice and Dab1 shRNA interference mice were compared to data from APOE +/+ controls using two-factor repeated-measures analysis of variance (ANOVA; for group and time). The remaining data were analyzed using Student's t test or randomized one-way ANOVA, followed by Tukey's post hoc test, to compare the differences between groups. Differences were considered statistically significant at P < .05.

| APOE deficiency inhibits axonal regeneration following CCI
To Subsequently, the total axon length in gray matter at the pericontusional cortex was compared between groups, as determined by SMI-32 immunoreactivity. Before injury, the intact axons were arranged in parallel, whereas CCI greatly disrupted the axon structure in the pericontusional cortex. No difference was detected between the two groups in total axon length before and after CCI ( Figure 1D,E).
To better quantify axonal regeneration in vivo, the level of GAP43 in the ipsilateral hemisphere was measured by Western blot.
As shown in Figure 1F,G, the GAP43 level was significantly lower in the APOE −/− group than in the APOE +/+ at both 3 and 28 days postinjury (P < .05).

| Alterations in related axonopathy by ApoE following CCI
In addition to axonal restoration, the potential effects of ApoE protein on synaptogenesis, myelination, and axonal injury were further assessed. For synaptogenesis assessment, we first examined the ex- In the Western blot analysis (Figure 2A), MBP expression did not differ between the APOE −/− and APOE +/+ groups (P > .05); however, APOE deficiency significantly suppressed synaptophysin and PSD-95 expression (P < .05). Accordingly, the immunofluorescence results ( Figure 2B) showed lower synaptophysin intensity in APOE-deficient mice than in APOE +/+ mice following CCI (P < .05) but no differences in APP intensity, PDGFR-α-positive cell (OPC) number or GST-πpositive (OC) cell number (P > .05).

| ApoE improves neurobehavioral function following CCI
To evaluate the potential benefits of ApoE-induced axonal regeneration, we conducted a series of behavioral tests in APOE −/− and APOE +/+ mice, that is, the NSS, rotarod, MWM, and Y maze tests.
As shown in Figure 3A

| The Dab1-dependent pathway is activated and has a beneficial effect in ApoE-induced axonal regeneration following CCI
To study whether the Dab1-Cdc42 pathway regulates the effects of ApoE on the reconstruction of the neural network after acute brain injury, we used Western blotting to detect changes in pathway regulators and the downstream target GAP43.
As shown in Figure 4A, Dab1 phosphorylation was significantly suppressed in the APOE −/− group compared to that in the APOE +/+ group (P < .05), whereas RAP, a competitive receptor blocker that strongly inhibits ligand binding to ApoE receptors, abolished this effect of ApoE (P < .05). Following injury, a higher level of GAP43 was observed in the APOE +/+ group than in the APOE −/− group (P < .05), but this difference in expression was inhibited by Dab1 shRNA ( Figure 4B) and ML141 (a Cdc42 activation inhibitor) treatment ( Figure 4D) (P < .05). Likewise, the GTP-Cdc42/Cdc42 ratio was significantly higher in the APOE +/+ group than in the APOE −/− group (P < .05); however, this ratio decreased when APOE +/+ animals were treated with Dab1 shRNA (P < .05), as shown in Figure 4C. Moreover, Dab1 inhibition abolished the beneficial effects of ApoE on white matter integrity ( Figure 5A) and synaptogenesis ( Figure 5B) (P < .05).

| Dab1 inhibition reverses ApoE-induced functional recovery following CCI
To validate the role of Dab1 in functional recovery after TBI, we repeated the behavioral tests in APOE −/− and APOE +/+ mice ( Figure 6).

| ApoE improves residual axon extension and growth cone formation following mechanical neuronal injury
To clarify the potential spatial association between ApoE and axons, we double-stained cultured primary cortical neurons with F-actin and ApoE ( Figure S1). ApoE colocalized with the F-actin-enriched To verify axonal outgrowth after TBI, we created a scratch injury in culture. As shown in Figure S1,

| D ISCUSS I ON
Following TBI, axonal damage is believed to be responsible for neurological deficits. However, spontaneous outgrowth from the axotomized axon tip has also been observed postinjury. To explore the potential pathway mediating axonal remodeling, we measured white matter tract integrity in CCI mice, as well as growth cone formation The MWM and Y maze tests were applied to evaluate the cognitive function of mice after injury. During the learning trials in the water maze, APOE +/+ mice showed an overall advantage in cognitive performance compared to APOE −/− mice following CCI (C). At the end of training, spatial memory was tested by target quadrant entries in each group, and APOE +/+ mice were more inclined to enter the target quadrant than the other groups (D). Likewise, APOE −/− mice showed a significant decrease in alternation behavior in the Y maze compared to the APOE +/+ group (E). However, there was no difference in total arm entries between the groups (F). N = 10/group. *, P < .05 and axonal outgrowth following mechanical injury in vitro. The In experimental animals and clinical patients, noninvasive DTI has recently been employed as a novel method for the detection of axonopathy in Alzheimer's disease, 15 neonatal hypoxia-ischemia, 16 multiple sclerosis, 17 traumatic spinal cord injury, 18 and TBI. 19 DTI is a quantitative measurement of water diffusion that provides information regarding the regional directional asymmetry (anisotropy) of white matter. The association between DTI parameters (relative anisotropy [RA], radial diffusivity [RD], FA) and histological axonal changes within pericontusion white matter following TBI has been studied, 20,21 and the FA value shows the strongest correlation with the regenerated axon number and diameter. 22 Moreover, the FA value is associated with neurological behavioral performance in hypoxia-ischemia stroke, 23 brain radiation injury 22 and TBI. 20 Hence, the FA parameter has been demonstrated to be a sensitive indicator of white matter integrity. We found that there was no difference in the FA value between the injury groups at 3 days postinjury, indicating that ApoE may exert no effects on acute axonal injury, which is supported by APP staining in white matter. In the regeneration phase, however, APOE-deficient mice exhibited significantly impaired white matter integrity, as indicated by lower FA values than those in APOE +/+ mice. In contrast, axon regeneration in gray matter was unchanged by APOE deficiency, which could be explained by the low baseline axon density in such an area. Due to the working principle of DTI, axonal regeneration is difficult to distinguish from remyelination with FA. To solve this puzzle, we measured markers of remyelination using multiple parameters. After measuring MBP expression, OPC number and OC number, we cautiously concluded that ApoE is efficient in promoting axonal regeneration, rather than remyelination, following TBI.
As a membrane-anchored intracellular growth-associated protein predominantly expressed in growth cones during development, GAP43 has been applied as a marker of axonal regeneration. 24,25 Western blot results indicated that ApoE significantly upregulated GAP43 expression at both 3 and 28 days post-CCI. This discrepancy in the MRI results implies that axonal regeneration was initiated starting in the early phase and eventually contributes to white matter reconstruction in the late phase. Moreover, the expression of synaptogenesis markers, including synaptophysin and PSD95, was also increased by ApoE. In accordance with the benefits of synaptogenesis, these data support a positive role of ApoE in axonal remodeling following TBI.
In principle, treatment to promote axonal regeneration could be beneficial for functional recovery after brain injury. In survivors of TBI, the most impaired neurological functions are related to F I G U R E 4 ApoE induces the activation of Dab1 and Cdc42 after CCI. (A) Representative bands and quantitative analysis of the ratio of phosphorylated Dab1(pDab1) at 3 days after brain injury. Dab1 phosphorylation was significantly suppressed in the APOE −/− group compared to that in the APOE +/+ group, whereas RAP, a competitive receptor blocker known to strongly inhibit ligand binding to ApoE receptors, abolished this difference. (B) The level of Dab1 in APOE +/+ mice was suppressed by Dab1 shRNA lentivirus in the APOE +/+ +Dab1 shRNA group. Following injury, a higher level of GAP43 was observed in the APOE +/+ group than in the APOE −/− group. However, GAP43 levels were lower in the APOE +/+ +Dab1 shRNA group than in the APOE +/+ group. (C) The ratio of GTP-Cdc42/Cdc42 at 3 days following brain injury was measured. The level of activated Cdc42 was significantly higher in the APOE +/+ group than in the APOE −/− group. However, compared with the level in the APOE +/+ group, activated Cdc42 was lower in the APOE +/+ +Dab1 shRNA group, in which Dab1 was downregulated. (D) ML141 (a Cdc42 activation inhibitor) treatment showed a similar effect as that of Dab1 shRNA. GAP43 levels were lower in the ML141 treatment group than in the APOE +/+ group. The expression levels of each protein were normalized to those in the APOE −/− group. N = 6/group. *, P < .05  Along with the regenerated axonal extension, the number of growth cones was significantly higher after ApoE treatment than the number in the control group. Likewise, the average area per growth cone in the ApoE-treated group was significantly larger than that in the control group (P < .05). However, the effects of ApoE were abolished when Dab1 was downregulated by pretreatment with Dab1 shRNA. *P < .05. n = 6/group outgrowth. 31,32 Accordingly, Dab1 downregulation in vivo causes behavioral impairments during development. 33 These observations raise more questions about whether Dab1 exerts similar effects on the reconstruction of neural networks, especially after acute brain injury. We found that Dab1 phosphorylation was activated in response to ApoE following brain injury. Furthermore, the effects of ApoE on axonal regeneration were attenuated by Dab1 shRNA in vitro and in vivo. As the downstream effector of Dab1, Cdc42 has previously been documented to promote actin accumulation in the growth cone to further potentiate axon elongation. 12 More importantly, Cdc42 can be activated by pDab1 to promote axonal branching and growth cone motility in cultured primary neurons. 11 Here, we demonstrated that Cdc42 activity was enhanced in the presence of ApoE, but this en- Surviving axons must transform their damaged terminals into new growth cones to initiate regeneration. 34 With the enrichment of dynamic actin, the de novo assembly of the new growth cone is a prerequisite for axonal elongation from the axotomized tip. APOE gene expression was significantly increased in axotomized cortical neurons in a DNA microarray study in which a total of 305 genes were analyzed. 35 Our previous study demonstrated that mechanical neuronal injury promotes the synthesis of intracellular ApoE within neurons and the uptake of extracellular ApoE. 14 Here, we showed that ApoE was expressed in both existing and regenerated growth cones, suggesting that ApoE participates in growth cone formation.
There are several limitations in our study. A specific Dab1 activator would allow for a better assessment of the effect of Dab1 pathway on axonal regeneration, but it has yet to be discovered or created. It would be better to compare the effect of ApoE on Dab1 pathway by using a specific Dab1 activator in the future. Secondly, the human APOE gene has 3 polymorphic alleles, e2 (cys112, cys158), e3 (cys112, arg158), and e4 (arg112, arg158). The e4 isoform has been implicated in atherosclerosis, ischemic cerebrovascular disease, impaired cognitive function, and late-onset Alzheimer's disease. 36 Several studies have shown that patients with APOE e4 have a poorer outcome after TBI. 37 In this study, we explored the mechanism of wild-type murine ApoE on white matter remodeling after TBI; however, whether different ApoE isoform (e2, e3 or e4) has a similar function should be examined in future studies.

| CON CLUS ION
Given that a potential neuroprotective effect of ApoE has been previously recognized, the current study extends the benefits of ApoE in neural restoration following TBI via axonal regeneration. In vivo and in vitro results demonstrate that ApoE initiates the phosphorylation of Dab1, which leads to subsequent Cdc42 activation, ultimately contributing to axonal regeneration and functional recovery.