Sustained ErbB Activation Causes Demyelination and Hypomyelination by Driving Necroptosis of Mature Oligodendrocytes and Apoptosis of Oligodendrocyte Precursor Cells

Oligodendrocytes are vulnerable to genetic and environmental insults and its injury leads to demyelinating diseases. The roles of ErbB receptors in maintaining the CNS myelin integrity are largely unknown. Here, we overactivate ErbB receptors that mediate signaling of either neuregulin (NRG) or epidermal growth factor (EGF) family growth factors and found their synergistic activation caused deleterious outcomes in white matter. Sustained ErbB activation induced by the tetracycline-dependent mouse tool Plp-tTA resulted in demyelination, axonal degeneration, oligodendrocyte precursor cell (OPC) proliferation, astrogliosis, and microgliosis in white matter. Moreover, there was hypermyelination before these inflammatory pathologic events. In contrast, sustained ErbB activation induced by another tetracycline-dependent mouse tool Sox10+/rtTA caused hypomyelination in the corpus callosum and optic nerve, which appeared to be a developmental deficit and did not associate with OPC regeneration, astrogliosis, or microgliosis. By tracing the differentiation states of cells expressing tetracycline-controlled transcriptional activator (tTA)/reverse tTA (rtTA)-dependent transgene or pulse-labeled reporter proteins in vitro and in vivo, we found that Plp-tTA targeted mainly mature oligodendrocytes (MOs), whereas Sox10+/rtTA targeted OPCs and newly-formed oligodendrocytes (NFOs). The distinct phenotypes of mice with ErbB overactivation induced by Plp-tTA and Sox10+/rtTA consolidated their nonoverlapping targeting preferences in the oligodendrocyte lineage, and enabled us to demonstrate that ErbB overactivation in MOs induced necroptosis that caused inflammatory demyelination, whereas in OPCs induced apoptosis that caused noninflammatory hypomyelination. Early interference with aberrant ErbB activation ceased oligodendrocyte deaths and restored myelin development in both mice. This study suggests that aberrant ErbB activation is an upstream pathogenetic mechanism of demyelinating diseases, providing a potential therapeutic target. SIGNIFICANCE STATEMENT Primary oligodendropathy is one of the etiologic mechanisms for multiple sclerosis, and oligodendrocyte necroptosis is a pathologic hallmark in the disease. Moreover, the demyelinating disease is now a broad concept that embraces schizophrenia, in which white matter lesions are an emerging feature. ErbB overactivation has been implicated in schizophrenia by genetic analysis and postmortem studies. This study suggests the etiologic implications of ErbB overactivation in myelin pathogenesis and elucidates the pathogenetic mechanisms.


Introduction
Oligodendrocytes form myelin to speed up the axonal impulse conduction while supporting neuronal function and integrity in the CNS. White matter is expanded, with a growth peak during juvenile to adolescence, by oligodendrocyte precursor cell (OPC) proliferation. Terminally differentiated OPCs differentiate into newly-formed oligodendrocytes (NFOs) under stringent regulation by extrinsic and intrinsic signals (Bergles and Richardson, 2015;Emery and Lu, 2015). NFOs progress from premyelinating oligodendrocytes to newly myelinating oligodendrocytes. Myelinating oligodendrocytes effectively generate myelin sheaths in a short time window. After myelin formation, mature oligodendrocytes (MOs) are extremely stable in maintaining myelin sheaths (Watkins et al., 2008;Czopka et al., 2013;Xiao et al., 2016;Tripathi et al., 2017;Hughes et al., 2018). However, a plethora of factors, including genetic alteration, inflammatory mediators, ischemia, trauma, viruses, and toxins, can damage oligodendrocytes and result in demyelinating diseases (Bradl and Lassmann, 2010). Understanding the pathogenetic mechanisms is essential for developing strategies to combat myelin loss and promote myelin restoration.
Although it is well established that ErbB2 and ErbB3 are crucial to peripheral myelin development (Nave and Salzer, 2006), the roles of ErbB receptors in oligodendrocytes have long been debated. For example, ErbB3 has been reported to be dispensable for oligodendrocyte development (Schmucker et al., 2003), and ErbB3/ErbB4 double knock-out from embryonic age does not result in CNS myelin alteration (Brinkmann et al., 2008). However, there is another report that ErbB3 depletion in oligodendrocytes from postnatal day (P)19 results in adult hypomyelination (Makinodan et al., 2012). Studies on oligodendrocytespecific knock-out mice have validated the expression of ErbB3 and ErbB4 in oligodendrocytes (Brinkmann et al., 2008), while phosphorylated EGFR is detected in oligodendrocytes by immunostaining (Palazuelos et al., 2014). Despites the controversial results from loss-of-function studies, overexpressing NRG1 Type I or Type III in neurons, or hEGFR in oligodendrocytes (CNP-hEGFR), promotes the CNS myelination (Aguirre et al., 2007;Brinkmann et al., 2008). Interestingly, transgenic mice CNP-hEGFR increases the numbers of myelinated axons but not myelin thickness (Aguirre et al., 2007), whereas NRG1-overexpressing mice exhibit hypermyelination of individual axons (Brinkmann et al., 2008). These results suggest that EGF-ErbB and NRG-ErbB receptors play different roles in oligodendrocytes. However, the possible synergistic effects of overactivating NRG-ErbB and EGF-ErbB receptors in oligodendrocytes have not been investigated.
To understand the consequences of co-activating ErbB receptors in oligodendrocytes, we employed a pan-ErbB strategy and activated endogenous ErbB receptors mediating either NRG or EGF signaling in oligodendrocytes. The results demonstrated that overactivating ErbB receptors in oligodendrocytes caused deleterious outcomes in white matter. Further, with the finding that the two inducible mouse tools, Plp-tTA and Sox10 1/rtTA , differentially targeted MO and OPC-NFO stages, we demonstrated that ErbB overactivation induced primary MO necroptosis and OPC apoptosis, causing inflammatory demyelination and noninflammatory hypomyelination, respectively. Early interference with the aberrant activation of ErbB signaling was effective to cease these pathologic progressions and restore oligodendrocyte numbers and myelin development.
Tet-Off or Tet-On treatment of mice Mice with Tet-system contain genes of tetracycline-controlled transcriptional activator (tTA) or reverse tTA (rtTA) driven by cell-specific promoters. When fed with Dox, these mice are able to switch on or off expression of a gene under the control of tetracycline-responsive element (TRE), specifically in rtTA-expressing or tTA-expressing cells, which are called "Tet-on" or "Tet-off," respectively. The offspring of Sox10 1/rtTA during the indicated periods were fed with Dox (2 mg/ml Â 10 ml/d from P21 to indicated test day) in drinking water to induce the expression of ErbB2 V664E in Sox10-ErbB2 V664E mice (Tet-On). For the offspring of Plp-tTA, Dox was given (Tet-off) from the embryonic day (through pregnant mothers) to their weaning day at P21 to inhibit the expression of ErbB2 V664E during this period in Plp-ErbB2 V664E mice (0.5 mg/ml Â 10 ml/d of Dox before P21). Water bottles were wrapped with foil to protect Dox from light. All used littermate control mice were treated the same.

Electron microscopy (EM)
Mice were anesthetized and transcardially perfused with 4% sucrose, 4% paraformaldehyde (PFA) and 2% glutaraldehyde in 0.1 M phosphate buffer (PB; pH 7.4). The brains and optic nerves were isolated carefully. The corpora callosa and prefrontal cortices were further dissected carefully under stereoscope. Tissues were postfixed overnight at 4°C in 1% glutaraldehyde in 0.1 M PB. 24 h later, samples were washed by 0.1 M PB and osmicated with 2% osmium tetroxide 30-60 min at 4°C, washed by 0.1 M PB and by deionized H 2 O at 4°C, and dehydrated in graded (50-100%) ethanol. Samples were incubated with propylene oxide and embedded with embedding resins. Ultrathin sections were stained with 2% uranyl acetate at 4°C for 30 min, and then photographed with Tecnai 10 (FEI). EM images were analyzed by ImageJ (NIH). To eliminate the bias on circularity, g-ratio of each axon was calculated by the perimeter of axons (inner) divided by the perimeter of corresponding fibers (outer). Axonal diameters were normalized by perimeters through equation: diameter = perimeter/p . This procedure allows for inclusion of irregularly shaped axons and fibers and helps to eliminate biased measurement of diameters based on circularity. For quantitative analysis, cross sections of each neural tissue were divided into five areas, and more than two images, randomly selected from each area, were examined. The g-ratio of detectable myelinated axons in these images was measured. The numbers of detectable axons from each mouse were not the same. g-ratio from all detectable axons of one mouse was averaged and averaged g-ratio of three to five mice were used for statistical analysis.
Luxol fast blue (LFB) staining Brain slices were obtained as described in Immunofluorescence staining. After sufficient washing with 0.01 M PBS, the slices were transferred into a mixture of trichloromethane and ethanol (volume ratio 1:1) for 10 min and then 95% ethanol for 10 min. They were next incubated in 0.2% LFB staining solution (0.2 g solvent blue 38, 0.5 ml acetic acid, 95% ethanol to 100 ml) at 60°C overnight. The next day, tissues were incubated for 5 min each in turn in 95% ethanol, 70% ethanol and ddH 2 O for rehydration, followed by incubation alternatively in 0.05% Li 2 CO 3 , 70% ethanol and ddH 2 O for differentiation until the contrast between the gray matter and white matter became obvious. After that, tissues were incubated for 10 min each in 95% and 100% ethanol to dehydrate, and then 5 min in dimethylbenzene to clear, before quickly mounting with neutral balsam mounting medium (CWBIO). All steps were operated in a ventilation cabinet. The LFB intensity in the corpus callosum was measured by Image J with background subtraction, and normalized to that of controls.
Terminal deoxynucleotidyl transferase (TdT)-mediated deoxyuridine triphosphate (dUTP) nick-end labeling (TUNEL) assay Apoptotic cells were examined with terminal TUNEL assay according to the manufacturer's instructions (Vazyme; Yeasen). In brief, brain slices which were obtained as that described in Immunofluorescence staining were digested for 10 min by proteinase K (20 mg/ml) at RT. After washing twice with PBS, brain slices were incubated with Equilibration buffer for 30 min at RT, and subsequently with Alexa Fluor 488-12-dUTP Labeling Mix for 60 min at 37°C. After washing with PBS three times, brain slices were stained with DAPI before being mounted under coverslips. For colabeling of apoptotic nuclei in slices with immunofluorescence staining, TUNEL assay was performed after washing of the secondary antibody.

OPC cultures and differentiation in vitro
The cortices of mice at P2-P5 were isolated and sheared into pieces, and digested by 0.25% Trypsin in Ca2 1 Mg2 1 -free D-Hanks' for 30 min at 37°C. Trypsinized tissues were triturated and passed through 40 mm mesh. OPCs were obtained from the dissociated cells using MS Columns (Miltenyi Biotec 130-042-201) with magnetic beads (Miltenyi Biotec 130-101-547) to positively select PDGFRa 1 cells according to the manufacturer's recommendations. Isolated OPCs were plated into a poly-L-lysine-coated glass coverslips at a density of 20,000 cells per cm 2 with DMEM/F12 supplemented with 10% fetal bovine serum. Four hours later, the medium was replaced with DMEM/F12 medium with 2% B27, 20 ng/ml bFGF and 10 ng/ml PDGFAA. For differentiation induction, the medium was replaced with DMEM/F12 medium with 2% B27 and 40 ng/ml triiodothyronine. For cells from Sox-ErbB2 V664E mice, 3 mg/ml Dox was added into the medium 24 h after plating.

Statistical analysis
All data were analyzed using Prism (GraphPad) and presented as mean 6 SEM. Unpaired two-tailed Student's t test was used for analysis between two groups with one variable. Statistical significance was set at *p , 0.05, **p , 0.01, ***p , 0.001. For Western blotting and staining results, statistical analyses were performed after subtraction of the background intensity and normalization with controls in each batch of experiments to minimize the influences of batch-to-batch variations. Detailed statistical information for each experiment was included in the figure legends.

Results
ErbB overactivation caused inflammatory demyelination in Plp-ErbB2 V664E mice To manipulate ErbB receptor activities specifically in oligodendrocytes in vivo, we employed tetracycline-controlled systems whose induction or blockade depends on the presence of doxycycline (Dox). We generated Plp-tTA;TRE-ErbB2 V664E (Plp-ErbB2 V664E ) bi-transgenic mice by crossing Plp-tTA with TRE-ErbB2 V664E mice. Breeding pairs, as well as their offspring, were fed with Dox to Tet-off the TRE-controlled transcription. Dox was withdrawn when the offspring were weaned on P21 (Fig. 1A). Among ErbB1-4 receptors, ErbB2 that does not bind to any known ligand is the preferred partner to other ligand-bound ErbB members. ErbB2 V664E contains an amino acid mutation (Vla 664 /Glu 664 ) within the transmembrane domain facilitating its dimerization with other ErbB receptors and potentiating their downstream signaling (Chen et al., 2017). ErbB2 V664E expression was significantly induced in the white matter of Plp-ErbB2 V664E mice within 1 d post-Dox withdrawal (dpd; Fig. 1B). Notably, Plp-ErbB2 V664E mice around P35 exhibited severe ataxia while walking on a grid panel (Fig. 1C). Moreover, Plp-ErbB2 V664E mice showed difficulty in rolling over, indicating severely impaired motor coordination. This was unexpected as this phenotype has not been reported for NRG1-overexpressing or hEGFR-overexpressing mice (Aguirre et al., 2007;Brinkmann et al., 2008).
In Plp-ErbB2 V664E mice at P21 before Dox withdrawal, activities of endogenous ErbB receptors and the downstream Akt and Erk (MAPK) in white matter were not altered (Extended Data Fig. 1-1A). In Plp-ErbB2 V664E mice at P35 with 14 dpd, with the expression and phosphorylation of ectopic ErbB2 V664E , endogenous ErbB receptors in the white matter, including both EGFR that mediates EGF signaling and ErbB3/ErbB4 that mediate NRG signaling, were strikingly phosphorylated and activated (Fig. 1D,E). Activating ErbB receptors stimulated downstream signaling including both Erk and Akt, as their phosphorylation significantly increased in the white matter of Plp-ErbB2 V664E mice (Fig. 1D,E).
Overactivation of ErbB receptors caused lower myelin staining intensity as exhibited in the corpus callosum of Plp-ErbB2 V664E mice at 9 dpd after LFB staining (Fig. 1F). White matter tracts in the lateral part of corpus callosum of Plp-ErbB2 V664E mice at 9 dpd were obviously fragmented (black arrows). Notably, at 14 dpd, myelin loss became more evident throughout the corpus callosum, as that LFB staining intensities dropped dramatically in both the middle and lateral parts (Fig.  1F). Immunostaining of MBP also revealed dramatic myelin loss in the corpus callosum at 14 dpd ( Fig. 1F), suggesting that Plp-ErbB2 V664E mice were undergoing CNS demyelination after Dox withdrawal. Western blotting revealed that the induced ErbB2 V664E expression was associated with MBP loss in multiple brain regions of Plp-ErbB2 V664E mice (Fig. 1G), indicating a global CNS demyelination. Indeed, the EM examination revealed that myelin sheaths of some axons in Plp-ErbB2 V664E mice ruptured or underwent breakdown ( Fig. 1H-J). Because of demyelination, only a few intact axons were detected in the midline of the corpus callosum of Plp-ErbB2 V664E mice at 14 dpd ( Fig. 1I,J). When axonal tracts in the corpus callosum of Plp-ErbB2 V664E mice at 14 dpd were immunostained by TuJ1, the antibody recognizing neuronal specific b -tubulin III, the immunoreactivity dramatically reduced (Fig. 1K). Inflammatory demyelination is usually complicated and aggravated by the pathologic responses from nearby astrocytes and microglia (Bradl and Lassmann, 2010;Burda and Sofroniew, 2014). Indeed, in the white matter of Plp-ErbB2 V664E mice, astrogliosis and microgliosis were revealed after Dox withdrawal (Extended Data Fig. 1-1B,C; Fig. 1L,M). It was likely that inflammation exacerbated myelin degradation and axon degeneration in the corpus callosum, given the fact that most axons were preserved in the optic nerves where inflammatory status, as indicated by transcription of GFAP and proinflammatory cytokine IL-1b , was significantly milder than that in the corpus callosum of the same mice at 14 dpd ( Fig. 1N).
Plp-ErbB2 V664E white matter exhibited hypermyelination before demyelination Interestingly, despite the demyelination, the detectable axons, throughout the corpus callosum, optic nerve, and prefrontal cortex in Plp-ErbB2 V664E mice, were hypermyelinated (Fig. 1I). Myelinated axons detected in the brain of Plp-ErbB2 V664E mice had significantly smaller g-ratio (axon diameter/fiber diameter), a quantitative indication of myelin thickness for individual axons with different diameters (Fig. 1I). Myelin thickness showed no difference between TRE-ErbB2 V664E and littermate Plp-tTA, two single transgenic control mice, after Dox withdrawal (Extended Data Fig. 1-2A). Therefore, hypermyelination of detectable axons in Plp-ErbB2 V664E mice was a result of ErbB signaling overactivation.
Hypermyelination of individual axons in Plp-ErbB2 V664E mice phenocopied that observed in NRG1-overexpressing mice (Brinkmann et al., 2008), as well as that in the transgenic mice with Akt or Erk signaling overactivation in oligodendrocytes (Harrington et al., 2010;Ishii et al., 2013). We immunostained MBP in Plp-ErbB2 V664E mice, and found that myelin structures and distributions remained mostly normal in the middle part of corpus callosum at 9 dpd ( Fig.  2A, inset 2' vs inset 2), despite that MBP loss and myelin fragmentation were prominent at the cortical projection tips in the same mice ( Fig. 2A,C). An examination of the ultrastructure in these mice by EM revealed similar results that most axons were intact in the midline of the corpus callosum, although they have been significantly hypermyelinated (Fig. 2B, D). These results confirmed that hypermyelination occurred early in Plp-ErbB2 V664E mice, consistent with the reported function of ErbB signaling overactivation (Brinkmann et al., 2008;Harrington et al., 2010;Ishii et al., 2013). Overactivation of ErbB receptors activates multiple downstream signaling, and transduce more active intracellular signaling than transduction through the increase of ligands. Rapid demyelination and axonal degeneration after 9 dpd may be attributed to inflammation that exacerbated the pathologic progression in Plp-ErbB2 V664E mice (Burda and Sofroniew, 2014).
ErbB overactivation caused oligodendrocyte number reduction and noninflammatory hypomyelination in Sox10-ErbB2 V664E mice Next, we examined the effects of ErbB overactivation on CNS myelin by a Tet-on system generated in Sox10 1/rtTA ;TRE-Figure 2. Plp-ErbB2 V664E white matter exhibited hypermyelination before demyelination. A, MBP immunostaining of brain slices from indicated mice. Note the myelin structure and distribution exhibited relatively more normal in the middle part of corpus callosum (CC; inset 2') than that in the Layer II/III of cortex (inset 1'), in Plp-ErbB2 V664E mice at 9 dpd. B, EM examination of axons in the midline of the CC and the prefrontal cortex (PFC) in Plp-ErbB2 V664E mice at 9 dpd. g-ratio was calculated for myelinated axons and averaged g-ratio were analyzed by unpaired t test (inset). For CC, t (4) = 3.226, p = 0.032; for PFC, t (4) = 2.529, p = 0.065. Red arrows indicate the axons with myelin breakdown. C, Quantitative data of MBP signal distribution in immunostaining results. Data were presented as mean 6 SEM and analyzed by unpaired t test. In middle CC, t (6) = 13.11, p , 0.0001; in cortex Layer II/III, t (6) = 26.31, p , 0.0001. D, The densities of myelinated axons as well as the percentages of axons with myelin breakdown in EM analysis. Data were presented as mean 6 SEM and analyzed by unpaired t test. For myelinated-axon density in CC, t (4) = 0.805, p = 0.466. For axons with myelin breakdown at 9 dpd examined by EM: in CC, t (4) = 3.567, p = 0.023; in PFC, t (4) = 4.146, p = 0.014.
ErbB2 V664E (Sox10-ErbB2 V664E ) mice (Fig. 4A). ErbB2 V664E expression was also significantly induced in the white matter within 1 d with Dox-feeding (dwd), suggesting similarly prompt responses of tTA/rtTA transcription in Sox10-ErbB2 V664E and Plp-ErbB2 V664E mice (Fig. 4B). Sox10-ErbB2 V664E mice with Dox feeding from P21 developed severe motor dysfunction, including ataxia and tremors, and died around P35. As a result, Sox10-ErbB2 V664E and littermate control mice were investigated at P30 with 9 dwd. These mice had smaller body sizes at P30 and walked with difficulty on a grid panel (Fig. 4C). Interestingly, Sox10-ErbB2 V664E mice performed normal rolling over. Despite walking slowly in the grid walking test, Sox10-ErbB2 V664E mice did not exhibit as many foot slips as Plp-ErbB2 V664E mice did. On the other hand, Plp-ErbB2 V664E mice did not exhibit tremors. The tremors in Sox10-ErbB2 V664E mice should be ascribed to the hypomyelination of peripheral nerves (Extended Data Fig. 4-1A-H).
On P21 before Dox treatment, no ErbB2 V664E was detected in the white matter of Sox10-ErbB2 V664E mice, indicating there was no leaky expression of the transgene. Consistently, there were similar protein and phosphorylation levels of endogenous ErbB receptors as well as downstream Akt and Erk (Extended Data Fig. 1-1D). After 9 dwd, Western blotting revealed the expression and phosphorylation of ectopic ErbB2 V664E accompanied with increases in phosphorylation of ErbB3 and ErbB4, but not that of EGFR, in the white matter of Sox10-ErbB2 V664E mice (Fig. 4D,E). Although only the activities of ErbB3 and ErbB4, the receptors to the NRG family ligands, were increased in the white matter of Sox10-ErbB2 V664E mice, both downstream Erk and Akt signaling were activated (Fig. 4D,E). With the ErbB signaling overactivation, brain slices stained by LFB or immunostained by MBP antibody exhibited lower staining intensity in the corpus callosum of Sox10-ErbB2 V664E mice (Fig. 4G), consistent with the lower MBP levels detected by Western blotting (Fig. 4D,F). Unexpectedly, the examination of the ultrastructure of Sox10-ErbB2 V664E white matter by EM did not find hypermyelination or myelin breakdown but revealed that the axons in the corpus callosum and optic nerve exhibited thinner myelin with significantly increased g-ratio (Fig. 4H). Sox10 1/rtTA is a knock-in mouse line, so that the allele with Sox10-rtTA would not transcribe Sox10 mRNA (Ludwig et al., 2004). We analyzed the ultrastructure of myelinated axons in two control mice, Sox10 1/rtTA and littermate TRE-ErbB2 V664E mice, at P30 with 9 dwd and did not observe any differences (Extended Data Fig. 1-2B). Therefore, we can exclude the possible effect of haploinsufficiency of Sox10 on late postnatal myelin development.
It is notable that in Sox10-ErbB2 V664E mice, the numbers of myelinated axons were not altered (Fig. 4I), and myelin sheaths exhibited normal morphology (Fig. 4H). Moreover, neither microgliosis nor astrogliosis was detected (Extended Data Fig. 1-1E,F; Fig. 4J,K). Consistently, there was no increases in proinflammatory cytokine (IL-1b and TNFa) expression (Fig. 4L). Because there was no indication of inflammatory pathogenesis, we can conclude that thinner myelin in Sox10-ErbB2 V664E white matter was caused by developmental deficits not pathologic conditions. Therefore, ErbB overactivation in Sox10-ErbB2 V664E mice induced CNS hypomyelination rather than inflammatory demyelination.
We further examined the states of oligodendrocytes and found the numbers of Olig2 1 , NG2 1 , and CC1 1 cells all significantly decreased in the corpus callosum of Sox10-ErbB2 V664E mice with six and more dwd (Fig. 5A-C). The number reduction of NG2 1 and CC1 1 cells was proportional to each other and to that of Olig2 1 cells, confirming that the deficits in Sox10-ErbB2 V664E mice were development dependent. OPC proliferation decreased in the corpus callosum (Fig. 5D), further supporting that the oligodendrocyte lineage states in Sox10-ErbB2 V664E white matter were different from that in Plp-ErbB2 V664E white matter. As neither OPC nor oligodendrocyte differences were observed between littermate Sox10 1/rtTA and TRE-ErbB2 V664E , two control mice, at P30 with 9 dwd (Extended Data Fig. 5-1A,  B), the developmental deficits in Sox10-ErbB2 V664E mice should be attributed to ErbB overactivation in oligodendrocytes.
Plp-tTA targeted mainly MOs, whereas Sox10 1/rtTA targeted OPC-NFOs To understand the cellular mechanisms of inflammatory demyelination (Plp-ErbB2 V664E mice) and noninflammatory hypomyelination (Sox10-ErbB2 V664E mice) induced by ErbB overactivation, we had to first determine the cellular targeting preferences of Dox-dependent transgenes in these two mouse models. The finding that Plp-ErbB2 V664E and Sox10-ErbB2 V664E mice had few overlapping phenotypes was unexpected since that Sox10 is reported to express throughout the oligodendrocyte lineage, and Sox10 1/rtTA knock-in mice have been used to investigate all oligodendrocyte lineage cells (Wegener et al., 2015). On the other hand, Plp-tTA transgenic mice have been reported to target MOs in the adult brain (Inamura et al., 2012); however, Plp gene transcribes a shorter splicing isoform DM20 in OPCs and NFOs during embryonic and early postnatal development (Trapp et al., 1997). The previous report for Sox10 1/rtTA -targeting specificity was concluded by using GFP reporter mice with Dox treatment for several weeks (Wegener et al., 2015). GFP is highly stable with a ;26-h half-life (Corish and Tyler-Smith, 1999). In contrast, transactivator tTA has a half-life ,6.5 h (Chassin et al., 2019). Therefore, GFP reporter generated in the original tTA/rtTA expressing cells can accumulate into the following cellular stages when tTA/rtTA has degraded, and thus the reporter-containing cells consist of tTA/rtTA-expressing cells and their progeny. In oligodendrocyte lineage, terminal differentiated OPCs can differentiate into NFOs in as quickly as 2.5 h and NFOs can mature in 1 d (Xiao et al., 2016). Because the induction of the Tet-on or Tet-off system by Dox feeding or Dox withdrawal has a delayed effect on gene expression, to examine the cellular targeting preferences under a condition with limited derivation from original tTA/rtTA-expressing cells, we acutely delivered a TRE-controlled fluorescence reporter carried by an adeno-associated virus (AAV) into the mouse brains at P14 or P35. In these AAV-TRE-YFP-infecting mice, rtTA-targeting or tTA-targeting cells were pulse-labeled, allowing us to trace their differentiation states within a short time window.
In most cases from Sox10 1/rtTA mice, we noticed the total numbers of NG2 1 YFP 1 (or PDGFRa 1 YFP 1 ) and CC1 1 YFP 1 cells were fewer than that of Olig2 1 YFP 1 cells (Fig. 6C), which indicated the presence of reporter-containing cells that were neither positive for CC1 nor for NG2 (PDGFRa). These cells belong to the NFO stage, which includes both CC1 -NG2 -(or CC1 -PDGFRa -) and CC1 1 NG2 -(or CC1 1 PDGFRa -) oligodendrocyte lineage cells. The previous study for Sox10 1/rtTAtargeting specificity only examined the PDGFRa/NG2 and CC1 immunoreactivities in reporter-containing cells. It is known that most NFOs are also positive for CC1 immunostaining (Ye et al., 2009). The b -catenin effector TCF4 is specifically expressed in the NFO stage (Fancy et al., 2009;Fu et al., 2009;Ye et al., 2009), which was present in a subset of Olig2 1 cells but absent in PDGFRa 1 cells in mice at P30 (Fig. 6D). In Sox10 1/rtTA mice at P14, immunostaining revealed that ;49% of YFP 1 cells were NFOs (TCF4 1 ) 1 d after virus injection, but it reduced to 28% after one more day (Fig. 6E,G). TCF4 1 cells found in the corpus callosum at P35 were far fewer than P14, and these cells appeared as clusters. Interestingly, in Sox10 1/rtTA mice at P35, YFP 1 cells were mostly found in regions with TCF4 1 cell clusters (Extended Data Fig. 6-1E), where ;56% of YFP 1 cells were TCF4 1 1 d after virus injection and it reduced to 29% after one more day (Fig. 6F,G). The half reduction of NFO percentage in YFP 1 cells from day 1 to day 2 was consistent with the previous report that NFOs mature into MOs in 1 or 2 d (Xiao et al., 2016). There was another possibility that the transcriptional activity of Sox10-rtTA was low in MOs, and thus took more days to have detectable YFP levels. We analyzed the densities of TCF4 1 YFP 1 cells and found that they reduced to half from day 1 to day 2 after viral infection in Sox10 1/rtTA mice at either P14 or P35 (Fig. 6H). This result excluded the possibility that the reduction of NFO ratio in YFP 1 cells was because of the increase of targeted MOs, and confirmed the maturation rate of pulse-labeled NFOs from day 1 to day 2. A similar transition rate is applicable for targeted OPCs and NFOs from day 0 to day 1. Through analyses of the stage alteration rate of reporter-containing cells from day 1 to day 2 after viral infection, we can conclude that the Sox10 1/rtTA -targeting cells at day 0 should mostly be OPCs and NFOs.
YFP only have four amino-acid residue difference near the chromophore from GFP (Wachter et al., 1998). Therefore, the YFP-containing MOs (CC1 1 TCF4 -) in Sox10 1/rtTA mice can be ascribed to the high stability of YFP. We have measured and reported that the half-life of ErbB2 is ;8 h in HEK293 cells and ;9 h in breast cancer cells. Moreover, ligand binding stimulates the turnover of ErbB receptors (Liu et al., 2007;Tao et al., 2009Tao et al., , 2014. To clearly examine the oligodendrocyte stages with ErbB2 V664E expression, we purified OPCs and induced transgene expression in vitro. ErbB2 V664E was visualized by immunocytofluorescence in dissociated oligodendrocyte lineage cells from Sox10-ErbB2 V664E mice, but not those from control mice, with Dox treatment (Fig. 7A). Significant ErbB2 immunoreactivities were detected in OPCs (NG2 1 ) and NFOs (TCF4 1 ; Fig. 7B, open white arrows). Essentially, there were only a few differentiated oligodendrocytes with low MBP levels detected to have ErbB2 signals. Nevertheless, ErbB2 signals in these MBP 1 (low) cells, which just started maturation, became extremely weak (Fig.  7B, solid white arrows). Once oligodendrocytes matured to have high MBP levels, ErbB2 immunoreactivities disappeared (Fig.  7B, solid pink arrows), indicating that there were no new ErbB2 V664E synthesis and the existed ErbB2 V664E degraded during oligodendrocyte maturation. Despite a different oligodendrocyte stage proportion in the in vitro system from that in vivo, we detected 67.1 6 8.9% ErbB2 1 cells NG2 1 (OPCs) from the Sox10-ErbB2 V664E cultures during early differentiation. Four days in vitro after differentiation induction when many MBP 1 cells were newly developed, 58.7 6 4.0% ErbB2 1 cells were TCF4 1 (NFOs), and 18.1 6 2.4% were MBP 1 (low; Fig. 7C). There were ;2.92 6 1.8% ErbB2 1 were MBP 1 (high) where minimal ErbB2 intensities were detected (Fig. 7C,D). ErbB2 signals were highest in oligodendrocytes at OPC stage, and close to zero at MO stage (Fig. 7D), consolidating the concept that ErbB2 V664E expression induced by Sox10 1/rtTA is mainly confined at OPC-NFO stages.
AAV-TRE-YFP in Plp-tTA mice also labeled some TCF4 1 cells, which comprised only 7-12% of YFP 1 cells 1 or 2 d after virus injection at either age ( Fig. 6E-H). It was noticeable that Plp-tTA did not specifically target TCF4 1 cell clusters in the corpus callosum at P35 (Extended Data Fig. 6-1F). These results implied that Plp-tTA did not specifically target the OPC or NFO stage but randomly expressed in the OPC-NFO period at a low ratio. Conversely, 90% of the YFP 1 cells were TCF4and 92-97% were CC1 1 in Plp-tTA mice at either P14 or P35, suggesting that Plp-tTA steadily targeted MOs. Further, the in vitro analyses validated this idea by showing that 96.7 6 1.7% MBP 1 cells (Fig. 7E-G), including both cells with low (Fig. 7F, solid white arrows) and high MBP levels (Fig. 7F, solid pink arrows) from the Tet-off Plp-ErbB2 V664E mice, were ErbB2 1 . There were a few oligodendrocytes not MBP 1 , corresponding to 5.29 6 1.2% NFOs (TCF4 1 ) with the transgene expression (Fig. 7F, open white arrows). None of PDGFRa 1 cells (OPCs) was immunopositive for ErbB2 (Fig. 7G). Note the ErbB2 immunostaining intensities were comparable in NG2 1 cells from Sox10-ErbB2 V664E mice and MBP 1 cells from Plp-ErbB2 V664E mice (Fig. 7D), indicating comparable transactivation efficiencies of Sox10-rtTA and Plp-tTA at single cell levels. The in vivo and in vitro analyses corroborated that Sox10 1/rtTA and Plp-tTA target different oligodendrocyte differentiation stages (Fig. 7H), which explains why ErbB overactivation induced distinct cellular and histologic phenotypes in Plp-ErbB2 V664E and Sox10-ErbB2 V664E mice.

ErbB overactivation caused MO necroptosis whereas OPC apoptosis
With the advantages of the nonoverlapping phenotypes in Plp-ErbB2 V664E and Sox10-ErbB2 V664E mice, we were able to investigate the pathogenetic mechanisms that determine the different myelin responses in these two mouse models. MO number reduction occurred earlier than the time when demyelination was obviously observed in the corpus callosum of Plp-ErbB2 V664E mice, suggesting that primary oligodendropathy may be the cause of demyelination. We noticed there was an increase in the numbers of nuclei associated with CC1 1 cell debris, starting from 6 dpd, in the corpus callosum of Plp-ErbB2 V664E mice (Fig. 3A, white arrows). We examined the corpus callosum of Plp-ErbB2 V664E mice by TUNEL assay, and observed as little apoptotic signaling as that in controls ( Fig. 8A-C). Immunostaining for cleaved caspase-3, the key apoptotic signaling protein, also exhibited few signals there (Fig. 8B,C). These results revealed that the degenerating CC1 1 cells were necrotic rather than apoptotic. Consistently, the oligodendrocyte nuclei associated with the destroyed myelin sheaths in Plp-ErbB2 V664E mice were regular nuclei without apoptotic chromatin condensation (Fig. 1H, red asterisk). In support Figure 7. ErbB2 V664E mainly expressed in OPC-NFO stages from Sox10-ErbB2 V664E mice whereas in MOs from Plp-ErbB2 V664E mice. A, E, In vitro differentiation and Dox induction protocols for OPCs from Sox10-ErbB2 V664E (A) or Plp-ErbB2 V664E mice (E). DIV, day in vitro. B, F, Representative immunostaining images of ErbB2 with protein markers for different oligodendrocyte stages for cultured OPCs from indicated mice with illustrated differentiation and Dox treatment protocols in A or E. Open white arrows, ErbB2 1 cells without MBP signals; solid white arrows, oligodendrocytes with low MBP levels; solid pink arrows, oligodendrocytes with high MBP levels. C, The percentages of ErbB2 1 cells at different oligodendrocyte stages in total ErbB2 1 cells detected in each batch staining for cultures from Sox10-ErbB2 V664E (SB) mice. Note the proportion of oligodendrocytes at each differentiation stage in vitro is not the same as that in vivo, and Tet-on was not switched on for all oligodendrocytes in cultures with Dox supplementation 24 h after cell plating. D, The immunostaining intensities of ErbB2 in Sox10-ErbB2 V664E (SB) oligodendrocytes at each differentiation stage, as well as that in MBP 1 Plp-ErbB2 V664E (PB) oligodendrocytes. A.U., absolute unit. G, The percentages of ErbB2 1 cells at different oligodendrocyte stages in total ErbB2 1 cells detected in each batch staining for cultures from Plp-ErbB2 V664E (PB) mice, as well as the percentages of ErbB2 1 oligodendrocytes in total oligodendrocytes at each differentiation stage. Tet-off was switched on for all oligodendrocytes in cultures from PB mice, although the proportion of oligodendrocytes at each differentiation stage in vitro is not the same as that in vivo. H, Schematic summary of oligodendrocyte stage-targeting preferences of Plp-tTA and Sox10 1/rtTA . Figure 8. ErbB overactivation induced MO necroptosis in Plp-ErbB2 V664E mice but OPC apoptosis in Sox10-ErbB2 V664E mice. A, G, Dox treatment setting for indicated mice and littermate controls. B, H, Apoptotic cells (arrows) in the corpus callosum of Plp-ErbB2 V664E and control mice at 9 dpd (B), or Sox10-ErbB2 V664E and control mice at 6 or 9 dwd (H), were examined by TUNEL assays and immunostaining for cleaved caspase-3. Note the increased numbers of nuclei (DAPI 1 ) and nonspecifically stained hemorrhagic spots (asterisk) in the brain slice of Plp-ErbB2 V664E mice (B). C, Quantitative data of apoptotic cell densities in indicated mice. PB, Plp-ErbB2 V664E . SB, Sox10-ErbB2 V664E . Ctrl, control mice. Data were presented as mean 6 SEM and analyzed by unpaired t test. For TUNEL assay: PB, t (6) = 0.1128, p = 0.914; SB, t (5) = 8.344, p = 0.0004. For caspase-3: PB, t (7) = 0.3959, p = 0.704; SB, t (11) = 3.893, p = 0.0025. D, J, Co-immunostaining results of MLKL or RIP3 with CC1 in the corpus callosum of indicated mice with indicated Dox treatments. E, Western blotting results of MLKL and RIP3 in the white matter of Plp-ErbB2 V664E (PB) mice, or in that of Sox10-ErbB2 V664E (SB) mice, and in those of littermate control mice (Ctrl). ns, nonspecific bands. F, Quantitative data of immunostaining and Western blotting results of MLKL or RIP3 in indicated mice at 9 d with Dox treatments. Data were presented as mean 6 SEM and analyzed by unpaired t test. In Western blotting results, for RIP3 protein in PB, t (4) = 3.579, p = 0.023; for MLKL protein in PB, t (4) = 13.69, p = 0.00017. In immunostaining results, for percentage of RIP3 1 in CC1 1 cells in PB, t (4) = 6.002, p = 0.0039; for percentage of MLKL 1 in CC1 1 cells in PB, t (4) = 8.202, p = 0.0012. I, Apoptotic cells (TUNEL 1 or cleaved caspase-3 1 ) were OPCs (NG2 1 or PDGFRa 1 ) in Sox10-ErbB2 V664E mice at P30 with 9 dwd. Arrows, representative double positive cells. Note OPCs in Sox10-ErbB2 V664E mice were hypertrophic. The percentage of apoptotic OPCs in total OPCs were quantified and data were presented as mean 6 SEM and analyzed by unpaired t test. For TUNEL 1 NG2 1 cells, t (4) = 3.95, p = 0.0168. For Caspase3 1 PDGFRa 1 cells, t (4) = 26.96, p , 0.0001. of this theory, MLKL, the protein mediating oligodendrocyte necroptosis in multiple sclerosis (Ofengeim et al., 2015) as well as the peripheral myelin breakdown after nerve injury (Ying et al., 2018), demonstrated an increased expression in the callosal CC1 1 cells in Plp-ErbB2 V664E mice from 6 dpd ( Fig. 8D-F). Necroptosis is a programmed form of necrosis, and has been revealed to be a prominent pathologic hallmark of multiple sclerosis (Ofengeim et al., 2015). RIP3 is the kinase at the upstream of MLKL in this programmed death signaling pathway (Sun et al., 2012;Ofengeim et al., 2015). Notably, the expression of RIP3 was also elevated in the callosal CC1 1 cells in Plp-ErbB2 V664E mice from 6 dpd as revealed by both immunostaining and Western blotting (Fig. 8D-F). Significant MLKL signals also appeared in the corpus callosum of adult Plp-ErbB2 V664E mice at P70 with 10 dpd, indicating similar susceptibility of adult MOs to necroptosis induced by ErbB overactivation (Extended Data Fig. 3-1C,E). Cell necroptosis is a trigger to inflammation (Pasparakis and Vandenabeele, 2015). Based on the timeline, MO necroptosis was the primary defect induced in Plp-ErbB2 V664E mice, followed by inflammation, OPC regeneration, myelin breakdown, axonal degeneration, and other pathologic events as reported in multiple sclerosis (Bradl and Lassmann, 2010;Ofengeim et al., 2015).
In contrast, for Sox10-ErbB2 V664E mice, a dramatic increase in cell apoptosis in the corpus callosum was observed (Fig. 8C,G, H). These apoptotic nuclei, labeled either by TUNEL assay or immunostaining of cleaved caspase-3, were localized in the NG2 1 and PDGFRa 1 cells (Fig. 8I), indicating apoptosis of OPCs. On the other hand, no increase of MLKL or RIP3 was detected (Fig.  8E,F,J), indicating there was no necroptosis. Developmental apoptosis does not arouse local inflammation, and hence no pathologic responses were observed in astrocytes and microglia in the white matter of Sox10-ErbB2 V664E mice ( Fig. 4J-L). It is also interesting to note that both the NG2 1 cells with and without TUNEL 1 nuclei were hypertrophic in Sox10-ErbB2 V664E mice (Fig. 8I). This phenomenon was not revealed for NG2 1 cells in Plp-ErbB2 V664E mice (Fig. 3B), further supporting that the OPC phenotypes in Sox10-ErbB2 V664E mice were attributed to the transgene expression in OPCs.
Early intervention ceased oligodendrocyte necroptosis/ apoptosis and rescued pathologic phenotypes in both mice Next, we explored whether the pathologic deficits induced by ErbB overactivation can resolve after correcting the ErbB activities in oligodendrocytes. Toward this end, we withdrew Dox for Sox10-ErbB2 V664E mice after 9 dwd or re-fed Plp-ErbB2 V664E mice with Dox after 9 dpd. However, these mice were not rescued. Their disease conditions progressed deleteriously as same as mice with continuous Tet-on or Tet-off. These failures indicated that the pathology in both mice had progressed into an irreversible phase. The delayed effects of Dox feeding or Dox withdrawal on Tet-on or Tet-off mice may worsen the situations because that the real time for fully suppressing ErbB overactivation was after 9 dwd or 9 dpd for Sox10-ErbB2 V664E and Plp-ErbB2 V664E mice. In contrast, Sox10-ErbB2 V664E mice with 6 dwd, and Plp-ErbB2 V664E mice with 6 dpd, could be rescued by the Dox-treatment switch. Although both mice at these conditions had exhibited oligodendrocyte necroptosis or apoptosis (Fig. 8D,H), decreases of myelinating oligodendrocytes (Figs. 3A, 5A), as well as dysmyelination (Fig. 9B,C,I,J), in the corpus callosum on the day before Dox-treatment switching, they became mostly normal after 10-d removal/addition of Dox. The MBP intensities and distributions returned to normal (Fig. 9A-C,H-J). There were no necroptotic or apoptotic signals in the corpus callosum (Fig. 9D,K), and numbers of oligodendrocytes (CC1 1 , Olig2 1 ) became as same as those in control mice , despite that proliferative OPCs in Plp-ErbB2 V664E mice had not completely returned to normal at this time ( Fig. 9E-G). In a word, MO necroptosis and OPC apoptosis can be ceased by suppressing the aberrant activation of ErbB receptors, and the dead oligodendrocytes, regardless of inflammatory status, can be cleaned to restore myelin development. These results suggest that the early interference with aberrant ErbB signaling is effective to prevent the white matter pathologic progression.

Discussion
The results from in vivo pulse-labeled reporter tracing and in vitro transgene-expressing cell examination were consistent with the fact that the death was occurred for OPCs in Sox10-ErbB2 V664E mice but for MOs in Plp-ErbB2 V664E mice. These results substantiate that Sox10 1/rtTA and Plp-tTA mice are valuable tools for in vivo exploration of molecular and cellular mechanisms that involve oligodendrocyte development and pathogenesis. In the present study, the differentiation-stage targeting preferences of Sox10 1/rtTA and Plp-tTA mice helped manipulate ErbB activities in nonoverlapping oligodendrocyte lineage cells, enabling us to distinguish the histologic and cellular responses induced by activating ErbB receptors in MOs and OPC-NFOs, respectively. The levels of ErbB2 immunostaining intensities in individual OPCs from Sox10-ErbB2 V664E mice and individual MOs from Plp-ErbB2 V664E mice were comparable, suggesting that the transactivation activities of tTA/rtTA at single cell levels induced in Sox10 1/rtTA and Plp-tTA mice are the same (Fig.  7D). Moreover, when using real-time RT-PCR to examine the initiation of the transgene in Sox10-ErbB2 V664E and Plp-ErbB2 V664E white matter, ErbB2 V664E transcript significantly increased in both mice at the first day with Dox addition or removal, suggesting that the Sox10-rtTA and Plp-tTA have comparable response time to the drug induction. Therefore, the nonoverlapping phenotypes observed in Sox10-ErbB2 V664E and Plp-ErbB2 V664E mice are attributed to the cellular targeting preferences of Sox10 1/rtTA and Plp-tTA.
Devastating effects of ErbB2 V664E in Plp-ErbB2 V664E mice may be because of its potent promotion of endogenous ErbB activation and multiple downstream signaling including both PI-3K/Akt and MAPK (Erk) pathways (Fig. 1D,E). There may be a dose-dependent effect for timing or severity of the pathologic responses; however, observations in Plp-ErbB2 V664E , Olig2-Cre; Pten flox/flox , and Plp-CreER;Mek/Mek mice corroborate the concept that continuously activating ErbB signaling in oligodendrocyte lineage cells is pathogenetic. Given the fact that Plp-CreER and Olig2-Cre mice target the full oligodendrocyte lineage including OPCs (Guo et al., 2009), the cellular mechanism for myelin pathogenesis in the previous reports was not determined. With the advantage of Sox10 1/rtTA mice that mainly target OPC-NFOs, we demonstrated that overactivation of ErbB receptors that mediate NRG signaling in OPCs induced apoptosis, leading to noninflammatory hypomyelination (Figs. 4,5,8). On the other hand, the study on Plp-ErbB2 V664E mice suggested that synergistic overactivation of ErbB receptors that mediate NRG (ErbB3/ErbB4) and EGF (EGFR) signaling in white matter leads to more deleterious pathologic changes stimulating inflammation (Figs. 1, 3, 8). The differences in Plp-ErbB2 V664E and Sox10-ErbB2 V664E mice demonstrated that ErbB overactivation in MOs, but not OPC-NFOs, triggers inflammatory pathogenesis.
The present study revealed that oligodendrocytes and OPCs are cell types susceptible to cell deaths induced by ErbB overactivation, different from astrocytes that become hypertrophic and proliferative with ErbB overactivation (Chen et al., 2017). Notably, it has been reported that overactivation of ErbB receptors can cause cell death, although in many cases ErbB signaling is demonstrated to promote cell proliferation and survival. A number of studies reveal the roles of ErbB4, EGFR, ErbB2 activation in necrosis or apoptosis of macrophages, cancer cells, and a variety of cell types (Högnason et al., 2001;Tikhomirov and Carpenter, 2004;Jackson and Ceresa, 2017;Schumacher et al., 2017). ErbB ligand-induced cell death mostly occurs to cells with high ErbB receptor expression, highlighting that the excessive activation of ErbB receptors is the key to inducing death pathways. Consistently, their downstream Akt and Erk signaling have also been causally linked to cell deaths in some cases (Maddika et al., 2007;Wu et al., 2009;Cagnol and Chambard, 2010). It is intriguing that ErbB receptor overactivation in MOs and OPCs induced distinct deleterious effects. Because that RIP3 and MLKL, the core components of necroptotic machinery, increased specifically in MOs (Fig. 8D), but not in astrocytes or microglia which were also pathologically activated in Plp-ErbB2 V664E white matter (Fig. 1L,M), the necroptosis signaling should be activated by intracellular signaling pathways downstream of ErbB receptors in MOs. Caspase-8 activation has been reported to be the key event to determine apoptotic fate of cells (Oberst et al., 2011), and defective activation of caspase-8 is critical for RIP1/RIP3/MLKL signaling to induce oligodendrocyte necroptosis in multiple sclerosis (Ofengeim et al., 2015). A cell type-specific RNA-sequencing transcriptome analysis suggests that caspase-8 is minimally expressed in postmitotic oligodendrocytes but is detectable in OPCs (Zhang et al., 2014), which may determine MO necroptosis but OPC apoptosis under continuous ErbB activation.
The demyelinating disease is now a broad concept that embraces more than multiple sclerosis, neuromyelitis optica, and leukodystrophy. Increasing evidence points to the correlation of the common demyelinating diseases with psychological symptoms (Diaz-Olavarrieta et al., 1999;Chiaravalloti and DeLuca, 2008). On the other hand, white matter lesion is an emerging feature that can be examined periodically by structural brain imaging in patients with psychiatric disorders (Fields, 2008;Mei and Nave, 2014). Remarkably, elevated ErbB activation has been repeatedly implicated in schizophrenia. For example, the mRNA and protein levels of NRG1 and ErbB4 are reported to increase in the prefrontal cortex and hippocampus of schizophrenic patients (Law et al., 2006;Chong et al., 2008;Joshi et al., 2014). These increases could be caused by genetic factors, such as schizophrenia-linked single nucleotidepolymorphisms (SNPs) SNP8NRG221132 and SNP8NRG243177, which are identified to increase the mRNA levels of NRG1 (Law et al., 2006). In addition to increased expression, it has been shown that the capacity for ErbB4 to be activated is markedly improved in the prefrontal cortex of schizophrenic patients (Hahn et al., 2006). Besides NRG1 and ErbB4 that have received extensive attention, EGFR is observed to increase in the brain of some schizophrenic patients (Futamura et al., 2002).
The white matter abnormalities observed in our mouse models are reminiscent of diverse myelin-related clinical characteristics in schizophrenic brains, including reduced white matter volume, decreased oligodendrocyte densities, reduced myelin gene products, apoptotic oligodendrocytes, and damaged myelin (Douaud et al., 2007;Uranova et al., 2007Uranova et al., , 2011Fields, 2008;Höistad et al., 2009). To our knowledge, we are the first to reveal that ErbB overactivation can primarily induce oligodendropathy and myelin pathogenesis in white matter, providing a possible predisposition of a genetic variability in the ErbB receptors or ligands to the white matter lesion. Notably, SNP8NRG243177 with T-allele, which increases NRG1 Type IV production (Law et al., 2006), is associated with the reduced white matter integrity in schizophrenic patients (McIntosh et al., 2008). The direct evidence provided by the present study that gain of function in ErbB receptors caused primary white matter lesions emphasizes that monitoring white matter in live patients and testing the genetic variability of ErbB signaling pathways are two necessary methods to help develop prognostic and therapeutic strategies for psychiatric disorders.
Intrinsic genetic alterations, apart from the pathogenetic immune responses, may also cause primary oligodendropathy in multiple sclerosis (Factor et al., 2020). Oligodendrocyte necroptosis has been revealed to be a hallmark in multiple sclerosis (Ofengeim et al., 2015), indicating that ErbB receptor activation is possible an upstream pathogenic mechanism for the most common demyelinating disease. It has been reported that inhibition of MAPK (Erk) activity by PD0325901, a MEK inhibitor, promotes OPC differentiation and myelin regeneration in two different animal models of the demyelinating disease (Suo et al., 2019). Moreover, an EGF antibody has been used to neutralize EGF in the animal model of multiple sclerosis and exhibits therapeutic effects by promoting oligodendrogenesis and ameliorating pathologic status (Amir-Levy et al., 2014). Therefore, our findings provide a novel molecular and cellular insight into the primary oligodendropathy in demyelinating diseases, and support that early intervention of ErbB overactivation is beneficial in the therapy of the diseases.