A role for ErbB signaling in the induction of reactive astrogliosis

Abstract

Reactive astrogliosis is a hallmark of many neurological disorders, yet its functions and molecular mechanisms remain elusive. Particularly, the upstream signaling that regulates pathological responses of astrocytes is largely undetermined. We used a mouse traumatic brain injury model to induce astrogliosis and revealed activation of ErbB receptors in reactive astrocytes. Moreover, cell-autonomous inhibition of ErbB receptor activity in reactive astrocytes by a genetic approach suppressed hypertrophic remodeling possibly through the regulation of actin dynamics. However, inhibiting ErbB signaling in reactive astrocytes did not affect astrocyte proliferation after brain injury, although it aggravated local inflammation. In contrast, active ErbB signaling in mature astrocytes of various brain regions in mice was sufficient to initiate reactive responses, reproducing characterized molecular and cellular features of astrogliosis observed in injured or diseased brains. Further, prevalent astrogliosis in the brain induced by astrocytic ErbB activation caused anorexia in animals. Therefore, our findings defined an unrecognized role of ErbB signaling in inducing reactive astrogliosis. Mechanistically, inhibiting ErbB signaling in reactive astrocytes prominently reduced Src and focal adhesion kinase (FAK) activity that is important for actin remodeling, although ErbB signaling activated multiple downstream signaling proteins. The discrepancies between the results from loss- and gain-of-function studies indicated that ErbB signaling regulated hypertrophy and proliferation of reactive astrocytes by different downstream signaling pathways. Our work demonstrated an essential mechanism in the pathological regulation of astrocytes and provided novel insights into potential therapeutic targets for astrogliosis-implicated diseases.

Keywords: anorexia; brain injury; glia; gliosis; receptor tyrosine kinase.

Figure 1

Astrocyte-specific expression of dnEGFR blocked injury-induced ErbB activation in reactive astrocytes. (a) Representative images of TRE-YFP expression in astrocytes of Mlc1-tTA mice at 1-month old. AAV-TRE-YFP was stereotaxically injected into indicated brain regions. Fixed brains collected 1 day later were sectioned and immunostained for Acsbg1 or GFAP to label astrocytes in the cerebral cortex or hippocampal hilus, respectively. White arrows, double-positive cells. (b) Efficient inhibition on activity of each ErbB receptor by dominant-negative mutant dnEGFR. Complementary DNA sequence of dnEGFR was amplified from genomic DNA of TRE-dnEGFR mice by PCR and cloned into pcDNA3.1-His/myc vector. pcDNA3.1-dnEGFR-myc was transfected into HEK293 cells by polyethylenimine (PEI) together with one of ErbB1–4 plasmids or an empty vector. Cells were lysed 24 h later and processed into WB with indicated antibodies. Any ErbB receptor when overexpressed in HEK293 cells would autophosphorylate itself independent of ligand stimulation. Shown are representative WB results, demonstrating the inhibition of dnEGFR on phosphorylation of each ErbB receptor. (c) Quantitative analyses of experiments in (b). ***P<0.001; **P<0.01; n=3 for each ErbB receptor, paired t-test. (d) Schematic illustration of the Tet-off system in Mlc1-tTA;TRE-dnEGFR (Mlc1-dnEGFR) mice. (e, f) Active ErbB3 (pErbB3), but not total ErbB3 levels, was suppressed in the reactive astrocytes of Mlc1-dnEGFR cortex. Cortical slices from indicated mice 3 days post injury were immunostained with antibodies against pErbB3 (e) or total ErbB3 (f) together with GFAP. White arrows, double positive cells. Arrowheads, cells positive for GFAP alone.

Figure 2

Inhibition of endogenous ErbB signaling suppressed hypertrophic expansion of reactive astrocytes. (a) GFAP immunostaining of cortices 3 days post injury showed there were many GFAP⁺ cells in Mlc1-dnEGFR cortices smaller than those in littermate controls. ‘Δ’ represents the injury sites. (b) Quantitative analyses of GFAP⁺ cell densities near the injury sites in the cortices of Mlc1-dnEGFR and littermate control mice on different days post injury. n=3 for each time point, paired t-test. (c) Quantitative analyses of the percentage of GFAP⁺ cells induced by injury with GFAP-immunoreactive area bigger than 300 μm². n=3 for each time point, and over 600 cells were analyzed for each mouse. *P<0.05, paired t-test. (d) Less increase of GFAP protein in injured cortical tissues from Mlc1-dnEGFR mice (M) in comparison with that from littermate TRE-dnEGFR controls (T) 3 days post injury. Shown are representative WB result (left) and quantitative analysis result (right, **P=0.002, n=4 for each group, paired t-test). (e) Transcription of GFAP isoform 1 was specifically suppressed in injured Mlc1-dnEGFR cortices. Total RNA was extracted from injured cortical tissues of Mlc1-dnEGFR and littermate mice. mRNA levels of GFAP isoform 1 and isoform 2 were evaluated by real-time RT-PCR with specific primers. **P=0.001, n=3, paired t-test. (f) Hypertrophy labeled by diffused YFP in reactive astrocytes induced by stab injury. Injured or uninjured cortices of Mlc1-tTA mice were stereotaxically injected with AAV-TRE-YFP on 1.5 day post injury, and collected 1 day later for immunostaining of non-selective astrocyte marker Acsbg1. All YFP-labeled (YFP⁺) cells were positive for Acsbg1 (white arrows). Note YFP⁺ cells in the injured cortex were hypertrophic and easy to be observed, whereas that in the uninjured cortex were small and hardly observed in the area farther from the injection sites. The asterisk indicates the needle track for viral injection. (g) Hypertrophic remodeling was independent of GFAP expression in cortical astrocytes in response to injury. Similar samples from experiments in (f) were immunostained for GFAP. Note many reactive astrocytes with hypertrophy exhibited no GFAP signal (white arrowheads). (h) YFP-labeled sizes of reactive astrocytes were strikingly reduced in the injured cortices of Mlc1-dnEGFR mice. Brain-injured Mlc1-dnEGFR and Mlc1-tTA mice were stereotaxically injected with AAV-TRE-YFP near the injury sites on 3 or 7 days post injury (dpi) and brains were collected 1 day later for immunostaining of GFAP. Note all YFP⁺ cells were positive for GFAP at these time points (white arrows). Images were taken within 300-μm areas near the injury sites. (i) Quantitative analyses of YFP-labeled cell sizes induced by stab injury in Mlc1-dnEGFR mice and littermate Mlc1-tTA controls. n=5 for each time point. **P=0.008 for 4, and *P=0.028 for 8, days post injury (dpi), paired t-test.

Figure 3

Inhibition of endogenous ErbB signaling in reactive astrocytes did not affect their proliferation but reduced scar thickness. (a, b) Comparable proliferation of reactive astrocytes in Mlc1-dnEGFR mice and littermate controls. Cortical slices from indicated mice 3 days post injury were immunostained for GFAP with either Ki67 (a) or Olig2 (b), respectively. White arrows, representative double positive cells. (c) Quantitative analyses of experiments in (a, b). n=3 for each group, paired t-test. (d) Representative images of scars formed by reactive astrocytes at the injury sites as shown by increased immunoreactivities of AQP4 and GFAP. Mouse cortices were stabbed by a blade parallel to the longitudinal fissure. Brains on 3, 7 or 15 days post injury (dpi) were isolated for sections crossing the injured grooves. Note there was no glial scar formed on 3 dpi. Dotted lines outlined the scar surface and the scar boundary. (e) AQP4 immunoreactivity shaped scar-forming astrocytes well. Injured cortices on 7 dpi were immunostained for AQP4 and GFAP. Note scar-forming astrocytes had abundant AQP4 immunoreactivity that exhibited more cell bodies and processes than GFAP immunoreactivity did, whereas AQP4 immunostaining in non-scar area failed to shape astrocytes. Dotted lines outlined a scar-forming astrocyte in wound surface. (f) The thickness of glial scars formed in the cortices of Mlc1-dnEGFR mice was reduced as compared with that formed in littermate controls 7 or 15 days post injury (dpi). Scar width was measured from the wound surface to the scar boundary according to the immunoreactivities of AQP4 plus GFAP. *P<0.05, paired t-test, n=3 for each time point.

Figure 4

Ectopically activated ErbB signaling in astrocytes throughout the brain induced reactive responses comprehensively. (a) ErbB2^V664E induction in Mlc1-ErbB2^V664E mice by Dox withdrawal according to the illustrated timetable. Mice were fed with Dox from embryonic stages, and killed to isolate different brain regions on P40 (C, no Dox withdrawal; 2, Dox withdrawal from P21 to P40), or P30 (1, Dox withdrawal from P21 to P30), respectively, for WB. Note GFAP increased concurrently with ErbB2^V664E. (b) GFAP immunoreactivity increased prevalently in the brain of Mlc1-ErbB2^V664E mice 20 days after Dox withdrawal. (c) GFAP and Acsbg1 colocalized in cortical astrocytes of Mlc1-ErbB2^V664E mice. Cortical slices from Mlc1-ErbB2^V664E and littermate mice after Dox withdrawal were immunostained for GFAP and Acsbg1. Arrowheads, cells positive for Acsbg1 alone. White arrows, double positive and hypertrophic cells. (d) Many GFAP⁺ cells were positive for Ki67 in the brain of Mlc1-ErbB2^V664E mice. White arrows, GFAP⁺Ki67⁺ cells. A nucleus (DAPI⁺) for a GFAP⁺ cell was identified by the association with its main cell body. (e) ErbB2^V664E was well localized in GFAP⁺ cells in the brain of Mlc1-ErbB2^V664E mice. White arrows, GFAP⁺ErbB2⁺ cells. (f) Expression of nestin in GFAP⁺ cells in the cortices of Mlc1-ErbB2^V664E mice. White arrows, GFAP⁺nestin⁺ cells. (g) Many GFAP⁺ cells had Olig2⁺ nuclei in the brain of Mlc1-ErbB2^V664E mice. White arrows, GFAP⁺Olig2⁺ cells. A nucleus (DAPI⁺) for a GFAP⁺ cell was identified by the association with its main cell body. (h) Quantitative analyses of GFAP⁺ or Olig2⁺ cell densities in the Cx or CC in Mlc1-ErbB2^V664E and littermate control mice. *P<0.05; **P<0.01; n=3 for each brain region, paired t-test. (i) Quantitative analyses of the percentage of GFAP⁺Olig2⁺ or Ki67⁺Olig2⁺ cells in Olig2⁺ cells in different brain regions of Mlc1-ErbB2^V664E mice. n=3 for each brain region. Cx, cortex; Th, thalamus; CC, corpus callosum; Hp, hippocampus.

Figure 5

Prevalent astrogliosis and associated inflammation in the brain induced anorexia in Mlc1-ErbB2^V664E mice. (a) Representative Mlc1-ErbB2^V664E and littermate control mice at P40 with Dox withdrawal from P21. Note Mlc1-ErbB2^V664E mice were much smaller than their littermate controls. (b) Quantitative analysis of the weight of Mlc1- ErbB2^V664E and littermate control mice at P40 with Dox withdrawal from P21. **P=0.0018, n=4 for each group, paired t-test. (c) Isolated digestive systems from Mlc1-ErbB2^V664E and littermate control mice at P40 with Dox withdrawal from P21. Note the atrophic gastrointestinal tract and accessory organs of Mlc1-ErbB2^V664E mice because of malnourishment during adolescent development. (d) No pathological changes in digestive system of Mlc1-ErbB2^V664E mice at P40 with Dox withdrawal from P21. H & E staining results showed that stomach walls of Mlc1-ErbB2^V664E mice folded in on itself because of no content in gastrointestinal tract. (e) Reactive astrogliosis and associated inflammation in the hypothalami of Mlc1-ErbB2^V664E mice as indicated by increased immunoreactivities of GFAP and Iba1, respectively. (f) Brain appearance and gross structure showed no apparent difference except enlarged ventricles in Mlc1-ErbB2^V664E mice. Shown are the whole brain and H & E stained brain slices of Mlc1-ErbB2^V664E and littermate control mice at P40 with Dox withdrawal from P21.

Figure 6

Molecular signaling activated in reactive astrocytes induced by cell-autonomous ErbB activation. (a) Various intracellular signaling proteins were activated in the cerebral cortices of Mlc1-ErbB2^V664E mice, as increases of their phosphorylation levels were indicated by WB results. (b) Quantitative analyses of the protein phosphorylation levels revealed by WB. Phosphorylation levels of indicated proteins were normalized by its total protein levels. ***P<0.001; **P<0.01; *P<0.05; n=3 for each protein, paired t-test. (c) Specific elevation of both total protein levels and activity of STAT3 in the reactive astrocytes of Mlc1-ErbB2^V664E mice. Cortical slices of Mlc1-ErbB2^V664E and littermate control mice were co-immunostained for STAT3 or phosphorylated STAT3 (pSTAT3) together with GFAP. White arrows, representative double positive cells. (d) Specific activation of FAK and Src in the reactive astrocytes of Mlc1-ErbB2^V664E mice. Cortical slices of Mlc1-ErbB2^V664E and littermate control mice were co-immunostained by mouse antibody against GFAP and rabbit/goat antibodies against the active forms of FAK or Src, respectively. White arrows, representative double positive cells. Note that one of the active form of FAK (pY397) exhibited similar subcellular distribution pattern to the active form of Src (pY418).

Figure 7

Molecular signaling in the regulation of reactive astrocyte hypertrophy. (a) Activities of various signaling proteins in injured cortices of Mlc1-dnEGFR (M) and littermate control mice (T) 3 days post injury were examined by WB. Note the specific reduction of phosphorylated FAK (pFAK) and Src (pSrc), as well as profilin (pProfilin), in the injured cortical tissues from Mlc1-dnEGFR mice in comparison with that from littermate controls. (b) Quantitative analyses of the protein phosphorylation levels revealed by WB. Phosphorylation levels of indicated proteins were normalized by its total protein levels. **P<0.01; *P<0.05; n=3 for each protein, paired t-test. (c) Specific reduction of FAK and Src activities in the reactive astrocytes of Mlc1-dnEGFR mice in comparison with that of littermate controls. Cortical slices from injured Mlc1-dnEGFR and littermate control mice were co-immunostained by mouse antibody against GFAP and rabbit/goat antibodies against the active forms of FAK or Src, respectively. White arrows, representative double positive cells. ‘Δ’ represents the injury sites. (d) Percentage of reactive astrocytes with active FAK or Src according to immunostaining results. Only cells positive for GFAP were analyzed. **P<0.01; *P<0.05; n=4 for each group, paired t-test. (e) Average levels of FAK or Src phosphorylation in individual reactive astrocytes according to immunostaining results. Only cells positive for GFAP were analyzed. *P<0.05; n=4 for each group, paired t-test.

Figure 8

Inhibition of ErbB signaling in reactive astrocytes did not block inflammatory signaling induced by injury. (a) Similar inflammation induced by stab injury in the cortices of Mlc1-dnEGFR and littermate control mice. Shown are representative images of Iba1 immunostaining in injured cortical regions 3 days post injury. ‘Δ’ represents the injury sites. (b) Quantitative analysis of Iba1⁺ cell densities in injured cortical regions of Mlc1-dnEGFR and littermate control mice. n=3 for each group, paired t-test. (c) Real-time RT-PCR results of indicated cytokines in injured cortical tissues from Mlc1-dnEGFR and littermate control mice 7 days post injury. ***P<0.001; **P<0.01; n=3 for each group. (d) Activities of various signaling proteins induced by ErbB receptor ligands were reduced in primary Mlc1-dnEGFR astrocytes. Primary astrocytes from Mlc1-dnEGFR and control mice were treated with saline (Ctrl), rhEGF (1 μg/ml), or rhNRG1 (100 ng/ml) for 15 min, respectively, and indicated proteins and their phosphorylation levels in cell lysates were examined by WB. (e) Quantitative analyses of the protein phosphorylation levels revealed by WB in astrocytes stimulated by ErbB ligands. Phosphorylation levels of indicated proteins were normalized by its total protein levels. ^#P<0.05; ^##P<0.01; as compared with the treatment with saline. *P<0.05, as compared with control astrocytes with the same treatment. n=3 for each protein, paired t-test. (f) STAT3 activity induced by cytokine CNTF was not reduced in primary Mlc1-dnEGFR astrocytes. Primary astrocytes from Mlc1-dnEGFR or control mice were treated with saline (Ctrl) or CNTF (200 ng/ml) for 30 min and indicated proteins and their phosphorylation levels in cell lysates were examined by WB. (g) Quantitative analyses of the protein phosphorylation levels revealed by WB in astrocytes stimulated by CNTF. Phosphorylation levels of indicated proteins were normalized by its total protein levels. ^##P<0.01; ^###P<0.001; as compared with the treatment with saline. **P<0.01, as compared with control astrocytes with the same treatment. n=3 for each protein, paired t-test. (h) Schematic illustration of a working model for the role of ErbB signaling in the induction of reactive astrogliosis. ErbB activation in quiescent astrocytes initiates reactive astrogliosis via diverse downstream signaling pathways. Src and FAK, the non-receptor tyrosine kinases regulating actin polymerization, are activated by ErbB signaling to prompt hypertrophic remodeling in astrocytes. Other signaling proteins downstream of ErbB receptors, such as STAT3, could be activated through different pathways stimulated by multiple factors in the inflammatory environment. Inhibiting ErbB signaling in reactive astrocytes blocks hypertrophy through a direct inhibition on Src/FAK activities, whereas other signaling proteins such as STAT3 remain active to promote proliferation. Note that microglia could react to factors released from reactive astrocytes or other sources.