Dysmyelination by Oligodendrocyte‐Specific Ablation of Ninj2 Contributes to Depressive‐Like Behaviors

Abstract Depression is a mental disorder affecting more than 300 million people in the world. Abnormalities in white matter are associated with the development of depression. Here, the authors show that mice with oligodendrocyte‐specific deletion of Nerve injury‐induced protein 2 （Ninj2) exhibit depressive‐like behaviors. Loss of Ninj2 in oligodendrocytes inhibits oligodendrocyte development and myelination, and impairs neuronal structure and activities. Ninj2 competitively inhibits TNFα/TNFR1 signaling pathway by directly binding to TNFR1 in oligodendrocytes. Loss of Ninj2 activates TNFα‐induced necroptosis, and increases C‐C Motif Chemokine Ligand 2 (Ccl2) production, which might mediate the signal transduction from oligodendrocyte to neurons. Inhibition of necroptosis by Nec‐1s administration synchronously restores oligodendrocyte development, improves neuronal excitability, and alleviates depressive‐like behaviors. This study thus illustrates the role of Ninj2 in the development of depression and myelination, reveals the relationship between oligodendrocytes and neurons, and provides a potential therapeutic target for depression.


Introduction
Depression is a common mental disorder that affects more than 300 million people in the world. The World Health Organization has predicted that major depressive disorder will become the 1st ranking global burden of disease by 2030. [1] The causation of depression is multifactorial. [2] Abnormal neurogenesis is highly associated with the development of depression. [3] In central nervous system (CNS), oligodendrocytesformed myelin sheath is the major component of the white matter, and it is essential for normal neuronal functions. The ensheathment of myelin around the axons allows rapid saltatory impulse propagation along the nerve fibers. [4] Furthermore, oligodendrocytes metabolically support the neurons and facilitate neuron development. [5] Oligodendrocyte development is a highly regulated process controlled by intrinsic factors including chromatin remodeling, transcriptional regulation, and posttranslational modifications, [6] or extrinsic factors such as neurotransmitters and the metabolites released from the nearby neurons. [7] Recently, impaired white matter integrity has been connected to the development of depression. [8] However, further studies are in need to show the mechanism underpinning this connection.
Nerve injury-induced protein 2 (Ninj2), a membrane adhesion molecule, was first found in the myelin-forming Schwann cells in the peripheral nervous system. It induces neurite outgrowth by promoting homophilic cellular interaction. [9] In CNS, Ninj2 is predominantly expressed in the oligodendrocyte lineage cells, [10] however, its function remains elusive. Single nucleotide polymorphisms studies points out that mutations of Ninj2 could be risk factors of ischemic stroke and multiple sclerosis, [11] and even associated with suicide attempts. [12] Recent studies indicate that the expression of Ninj2 is downregulated in mice with stress-induced depression. [13] However, whether and how Ninj2 plays a role in the development of depression is largely unknown.
In our present study, to demonstrate the function of Ninj2 in depression, we generated oligodendrocyte-specific Ninj2knockout mice to evaluate the effect of Ninj2 in depression. We found that oligodendrocyte dysfunction was closely associated with depression. Loss of Ninj2 in oligodendrocytes stimulated TNF -induced necroptosis, hence simultaneously inhibited myelination and led to depressive-like behaviors. Mechanistically, the production of C-C Motif Chemokine Ligand 2 (Ccl2) was significantly induced during the activation of necroptosis in the Ninj2-deficient oligodendrocytes, and it might mediate the conversation between oligodendrocytes and neurons. These mechanistic insights into the modulation of depression by Ninj2 broadened the biological significance of oligodendrocytes in CNS, and provided a therapeutic target for depression in clinical studies.

Oligodendrocyte-Specific Knockout of Ninj2 Leads to Depressive-Like Behaviors
Accumulating evidences have indicated that myelin is associated with depression. Regenold et al. showed that unipolar major depression subjects have reduced white matter intensity compared with normal control subjects. [14] In chronic social defeat mice showing depressive-like behaviors, the expression levels of the myelin-related genes are largely downregulated. [15] Furthermore, a single-nucleus transcriptomic analysis of the prefrontal cortex in major depressive disorder implicate that oligodendrocyte precursor cells (OPCs) plays a role in depression. [16] To investigate the role of oligodendrocytes during the development of depression, we generated an Olig1 cre/+ ;Gi-DREADD mice strain, in which the activity of oligodendrocytes was inhibited upon clozapine-N-oxide (CNO) treatment ( Figure S1A, Supporting Information). Using this strain, we performed behavioral tests to examine whether inhibition of oligodendrocyte activity would lead to depressive-like behaviors. As shown, when the Olig1 cre/+ ;Gi-DREADD mice were treated with CNO, they exhibited longer immobility durations in tail suspension test (TST) and forced swimming test (FST) than control mice ( Figure  S1B,C, Supporting Information). Meanwhile, the CNO-treated Olig1 cre/+ ;Gi-DREADD mice showed reduced sucrose preference ( Figure S1D, Supporting Information). The depressive-like behaviors in the CNO-treated Olig1 cre/+ ;Gi-DREADD mice clearly reflected that dysfunction in oligodendrocytes was sufficient to trigger depression.
Next, we were prompted to investigate the key genes, which were predominantly expressed in oligodendrocytes, and potentially involved in the development of depression. We analyzed two published datasets that contained the transcriptomic data of the mice showing depressive behaviors. In a chronic social defeat stress-induced depression mice model, RNA-seq analysis in their cortex samples indicated that the expressions of 5083 genes were significantly changed (fold change > 1.5, or < 0.67). [13b] In the other dataset obtained from the cortex samples of the mice with chronic variable stress-induced depression, the expression levels of 3857 genes were greatly changed, [13a] compared to control mice. After alignment, 171 common genes were found in both datasets. According to the gene expression pattern of the cortex, [10] 25 of those common genes were substantially expressed in the cell lineages in CNS, and Ninj2 was the only gene that is predominantly expressed in oligodendrocyte lineages. Therefore, we chose Ninj2 as the candidate, to explore its role in the development of depression.
We then generated two oligodendrocyte-specific Ninj2 deletion mice strains (Olig1 cre/+ ;Ninj2 fl/fl and Cnp cre/+ ;Ninj2 fl/fl ) by crossing the Olig1 cre/+ or Cnp cre/+ with Ninj2 fl/fl mice strains ( Figure S2, Supporting Information). Through a series of behavioral tests, we found that Ninj2-deficient mice showed longer immobility durations in TST and FST, and reduced sucrose preference, compared with their WT counterparts ( Figure 1A,B), suggesting that loss of Ninj2 in oligodendrocytes resulted in depressive-like behaviors, similar to what we observed in the CNO-treated Olig1 cre/+ ;Gi-DREADD mice. At the same time, the Ninj2-deficient mice did not show any difference in Morris Water Maze test, Y maze test, Novel object recognition test, elevated plus maze test, or the open field test, compared with the WT mice ( Figure S3, Supporting Information). These results suggested that loss of Ninj2 in oligodendrocytes had no significant effect on memory, recognition, or anxiety-like behaviors.
To show the physiological impact of Ninj2 deletion, we detected the pyramidal neurons functions in the prefrontal cortex and hippocampus CA1 area by whole-cell patch-clamp recording. As shown, Olig1 cre/+ ;Ninj2 fl/fl mice had reduced frequencies and amplitudes of miniature excitatory postsynaptic currents (mEP-SCs), but augmented frequencies of miniature inhibitory postsynaptic currents (mIPSCs) in both regions ( Figure 1C,D), suggesting the neuronal dysfunction in Ninj2-deficient mice. Dendritic complexity was closely related with neuronal function, and impaired synaptic plasticity was also associated with depression. [17] Furthermore, using Golgi staining, we found that the pyramidal neurons in the hippocampus CA1 area of Olig1 cre/+ ;Ninj2 fl/fl mice were less developed, as they showed significantly reduced dendritic complexity, as well as reduced density of dendritic spines, when compared with their WT counterparts ( Figure 1E-H and Figure S4, Supporting Information). These results supported our hypothesis that Ninj2 deficiency in oligodendrocytes changed neuronal structure and further caused neuronal dysfunctions, resulting in depressive-like behaviors.

Loss of Ninj2 Inhibits Myelinogenesis
Recently, the linkages between oligodendrocytes and mental disorders have been reported. [18] Since Ninj2 was predominantly expressed in oligodendrocytes, we hypothesized that loss of Ninj2 caused depressive-like behaviors through its modulation in oligodendrocytes. To test our hypothesis, we examined the oligodendrocyte development and myelination status in the Olig1 cre/+ ;Ninj2 fl/fl mice. Through immunofluorescent imaging, we found that loss of Ninj2 reduced MBP + area and Olig2 + oligodendrocytes in both prefrontal cortex and hippocampus CA1 area, but had no effect on the population of NeuN + neurons, GFAP + astrocytes, or Iba1 + microglia (Figure 2A,B), suggesting that loss of Ninj2 in oligodendrocytes primarily inhibited oligodendrocyte development.
Consistent with the suppressed oligodendrocyte development, we observed dysmyelination phenotypes in the Ninj2 deficient mice. Ultrastructural analyses on optic nerve and corpus callosum showed that Olig1 cre/+ ;Ninj2 fl/fl mice had significantly lower myelinated axon numbers, compared with WT mice (Figure 2C). Moreover, the MBP intensity and the protein levels of MBP and CNP in the corpus callosum were significantly reduced by Ninj2 ablation throughout the duration of myelin development ( Figure S5, Supporting Information). Consistently, Whole-cell patch-clamp recording was performed to show the representative traces and quantification of frequencies and amplitudes of mEPSCs or mIPSCs from prefrontal cortex pyramidal neurons (mEPSCs: n = 31 neurons for WT mice, n = 25 neurons for cKO mice; mIPSCs: n = 21 neurons for WT mice, n = 23 neurons for cKO mice; from six mice/genotype) (C), or hippocampal CA1 pyramidal neurons (mEPSCs: n = 23 neurons for WT mice, n = 24 neurons for cKO mice; mIPSCs: n = 21 neurons for WT mice, n = 22 neurons for cKO mice; from six mice/genotype) (D). E-H) WT and cKO mice were subjected to Golgi staining. Sholl analysis of dendritic complexity in hippocampal CA1 pyramidal neurons (n = 28 neurons from three mice for each group) was shown in (E,F), Apical distal spine numbers of the pyramidal neurons in hippocampal CA1 regions (n = 42 dendritic segments for WT mice, n = 49 dendritic segments for cKO mice, from three mice/genotype) were shown in (G,H), Scale bars = 20 μm. For all panels, mice were subjected to experiments at P60. All the quantification data are presented as mean ± SEM, p-values are calculated using two-tailed unpaired Student's t-test (A-D,H), or two-way ANOVA with Tukey's multiple comparisons test (F), F (interaction [F (19, 1080) = 2.014, p = 0.0061]),*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.
in Cnp cre/+ ;Ninj2 fl/fl mice, the myelinated axon number in the corpus callosum was also lower than that in WT mice (Figure 2D). However, loss of Ninj2 had no effect on G-ratio ( Figure  S6, Supporting Information). The hypomyelination phenotype in these Ninj2-deficient mice was further corroborated by their motor defects. Both strains showed significant reductions in their grip strength and motor coordination ( Figure S7, Supporting Information).

Ninj2 Ablation in Oligodendrocytes Promotes Canonical Necroptosis
To explore molecular mechanisms underpinning with the dysmyelination observed upon oligodendrocytic Ninj2 deletion, we used immunofluorescent imaging to evaluate the effect of Ninj2 on oligodendrocyte development, and found that ablation of Ninj2 substantially reduced the total number of Olig2 + oligodendrocytes at P14 and P30 in corpus callosum (Figure 3A). At the same time, upon Ninj2 deletion, an increased population of TUNEL + cells was observed, but the proportion of CC1 + , PDGFR + , Ki67 + and cleaved-caspase3 + cells in the total Olig2 + oligodendrocytes at the indicated time points were unaffected ( Figure 3A and Figure S8, Supporting Information), suggesting that loss of Ninj2 primarily induced nonapoptotic cell death in oligodendrocytes. Furthermore, western blot analysis of corpus callosum and spinal cord samples showed that loss of Ninj2 increased p-MLKL level, a surrogate marker for cell necroptosis, but had no effect on cleaved-caspase3 level ( Figure 3B). All these data suggested that loss of Ninj2 in oligodendrocytes reduced the population of Olig2 + oligodendrocytes by inducing necroptotic cell death. The quantification of NEUN + GFAP + , Iba1 + , and Olig2 + cell numbers, and the relative MBP intensity, were shown in the right panel. Scale bar = 100 μm (A), 250 μm (B). n = 3 mice/genotype. C) Electron microscopic examination of the optic nerve at P14 and P21 and corpus callosum at P21 and P60 from WT or cKO mice. Scale bar = 2 μm. n = 3 mice/genotype. D) Electron microscopic examination of the corpus callosum at P60 from WT or Cnp cre/+ ;Ninj2 fl/fl (CcKO) mice. The quantified myelinated axon numbers were shown on the right panel. Scale bar = 2 μm. n = 3 mice/genotype. All the quantification data are presented as mean ± SEM, p-values are calculated using two-tailed unpaired Student's t-test, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.
We then went on to confirm the results in cultured mouse oligodendrocytes. Consistently, ablation of Ninj2 greatly reduced the number of MBP + oligodendrocytes, and inhibited MBP, CNP, and MAG level ( Figure 3C,D). However, loss of Ninj2 had no effect on oligodendrocyte proliferation (Figure 3C). Furthermore, loss of Ninj2 induced necroptotic cell death, but had no effect on apoptosis ( Figure 3C,D), which further confirmed that Ninj2 ablation induced necroptosis in oligodendrocytes.
It has been previously reported that canonical necroptosis, mediated by RIPK1 and RIPK3, played an important role in cell death in oligodendrocytes. [19] We then wondered that whether blocking the canonical necroptosis pathway could rescue oligodendrocyte development. We treated primary cultured oligodendrocytes with Nec-1s and GSK-872, the inhibitors targeting RIPK1 and RIPK3, respectively. [20] We found that Nec-1s or GSK-872 treatment decreased intracellular p-MLKL level in the Ninj2-knockout oligodendrocytes ( Figure 3E). Moreover, it also reduced the population of TUNEL + cells ( Figure 3F), and increased the population of MBP + cells in the Ninj2-deficient oligodendrocytes ( Figure 3G). Collectively, our data indicate that loss of Ninj2 inhibits oligodendrocyte development and myelination due to its inductive effect on necroptosis.

Ninj2 Interrupts TNF /TNFR1 Signaling
In order to elucidate the molecular basis underlying the necroptosis induced by Ninj2 deletion, we performed RNA-Seq analysis of WT or Ninj2-deficient mouse oligodendrocytes. The analysis showed that Ninj2 deletion in oligodendrocytes induced the expression of 105 genes (fold change > 2), and reduced the expression of 195 genes (to at least 50%) ( Figure 4A). According to Gene Ontology analysis, the downregulated genes were Figure 3. Ablation of Ninj2 in oligodendrocytes induces necroptosis. A) Immunofluorescent staining against Olig2, TUNEL, and cleaved-caspase3 were performed in the corpus callosum sections from WT or Olig1 cre/+ ;Ninj2 fl/fl (cKO) mice at P7, P14, and P30, the number of the Olig2 + , TUNEL + cells, or the percentage of cleaved-caspase3 + cells in Olig2 + cells, were quantified and shown on the right panel. Scale bar = 50 μm. n = 3 mice/genotype. B) Western blotting analyses against p-MLKL, MLKL, and cleaved-caspase3 on the corpus callosum or spinal cord samples from WT or cKO mice at P7. Densitometric quantification of p-MLKL/MLKL or cleaved-caspase3/GAPDH ratio was shown at the bottom panel. n = 6 mice/genotype. C) Immunofluorescent staining against MBP, BrdU, TUNEL, and cleaved-caspase3 in cultured oligodendrocytes from WT or cKO mice, the percentages of MBP + , BrdU + , TUNEL + , and cleaved-caspase3 + cells were shown at the bottom panel. Scale bar = 50 μm. n = 3 independent experiments. D) Western blot assays with the indicated antibodies in oligodendrocytes from WT or cKO mice. Densitometric quantification of p-MLKL/MLKL or MBP, CNP, MAG, cleaved-caspase3/GAPDH ratio was shown at the bottom panel. n = 3 independent experiments. E-G) Oligodendrocytes from WT or cKO mice were treated with vehicle, Nec-1s (1 μm) or GSK-872 (0.25 μM) for 48 h respectively, and then subjected to western blot analyses against the indicated proteins (E) or immunofluorescent staining against TUNEL (F) and MBP (G). Densitometric quantification of p-MLKL/MLKL ratio was shown at the right panel of (E). The percentages of the TUNEL + or MBP + in total DAPI + cells were quantified and shown on the right panels of (F) and (G). Scale bar = 50 μm. majorly involved in axon ensheathment or myelination, and the upregulated genes were found in chemotaxis and cellular response to tumor necrosis factor ( Figure 4B). These results hinted that Ninj2 might play a role in TNF signaling, a well-known pathway to trigger necroptosis. [21] To study the potential role of Ninj2 in TNF signaling, we first examined the cellular response to TNF in WT or Ninj2deficient oligodendrocytes. As shown in Figure 4C, in the presence of TNF treatment, the level of p-MLKL was induced at a dose-dependent manner in both WT and Olig1 cre/+ ;Ninj2 fl/fl oligodendrocytes, but the level of p-MLKL increased more dramatically in the TNF -treated Ninj2-deficient cells, compared to the WT cells. Similarly, in cultured rat oligodendrocytes, TNF sufficiently induced intracellular p-MLKL, but it was greatly abolished by Ninj2 overexpression ( Figure 4D). We then speculated that Ninj2 could interrupt TNF /TNFR1 binding. To test the speculation, co-immunoprecipitation assays in HEK293T cells were performed and showed that Ninj2 directly interacted with TNFR1 ( Figure 4E). In rat oligodendrocytes, similar interaction between Ninj2 and TNFR1 was observed ( Figure 4F). In addition, Ninj2/TNFR1 interaction interrupted the recruitment of TNF to TNFR1. Overexpression of Ninj2 in oligodendrocytes attenuated the interaction between TNFR1 and the endogenous TNF ( Figure 4G). Collectively, our data illustrate that Ninj2 . Ninj2 inhibits TNF -induced necroptosis through interaction with TNFR1. A) Volcano diagram showing the genes whose expression level was significantly changed in the oligodendrocytes from Olig1 cre/+ ;Ninj2 fl/fl (cKO) mice. B) Gene Ontology (GO) analysis showing the biological processes in which the significantly changed genes involved. C,D) Western blotting analyses against p-MLKL, MLKL in the oligodendrocytes from WT or cKO mice that treated with vehicle, 10 or 50 ng mL −1 of TNF for 48 h, respectively (C), or in oligodendrocytes transfected with empty or Ninj2-overexpressing vectors, and then treated with 100 ng mL −1 of TNF for 48 h (D). Densitometric quantification of p-MLKL/MLKL ratio from at least three independent assays is shown at the right panel of (C) and (D). E) Co-immunoprecipitation analysis in HEK293T cells that transfected with Ninj2 and TNFR1-overexpressing vectors, alone or in combination as indicated. n = 3 independent experiments. F) Co-immunoprecipitation analysis in rat oligodendrocytes that transfected with empty or TNFR1-overexpressing vectors. n = 3 independent experiments. G) Co-immunoprecipitation analysis in rat oligodendrocytes that transfected with empty, Ninj2 or TNFR1-overexpressing vectors, alone or in combination as indicated. Densitometric quantification of IP/Input (TNF ) ratio from at least three independent assays was shown at the bottom panel. All the quantification data are presented as mean ± SEM, p-values are calculated using two-way ANOVA (C) or one-way ANOVA with Tukey's multiple comparisons test (D,G), C (interaction [F 2,12 = 26.56, p < 0.0001]), D (interaction [F 2,6 = 27.31, p = 0.001]), G (interaction [F 2,6 = 10.91, p = 0.01]), *p < 0.05, **p < 0.01, ****p < 0.0001. competitively inhibits TNF /TNFR1 signaling through interaction with TNFR1.

Ninj2 Ablation Increases Ccl2 Production in Oligodendrocytes
As shown in Figure 4B, Ninj2-deficient oligodendrocytes showed increased expression levels of genes involved in chemotaxis and cellular response to TNF . As cytokines were tightly associated with the onset of depression, [22] we tested the expression levels of inflammatory cytokines and found that Ccl2 was the most significantly upregulated in the Ninj2-deficient oligodendrocytes, compared to the WT cells ( Figure 5A). Furthermore, Ccl2 content increased in conditioned media from Ninj2-deficient oligodendrocytes ( Figure 5B). Ccl2 production was increased during the process of necroptosis. [23] When we used Nec-1s to inhibit oligodendrocytic necroptosis, the mRNA level and content of Ccl2 in the conditioned media of Ninj2-deficient oligodendrocytes greatly reduced ( Figure 5C,D), providing additional evidence that the excessive Ccl2 production came from necroptosis triggered by TNF signaling overactivation in Ninj2-deficient oligodendrocytes. Previously, Ccl2 has been reported to affect the activities of different kinds of neurons. [24] Blockage of Ccl2/CCR2 signaling attenuated neuropathic pain and depression. [24c] Here, when the primary cultured pyramidal neurons were treated with Ccl2, whole-cell patch-clamp recording displayed that Ccl2 administration efficiently repressed the frequencies and amplitudes of mEP-SCs, and induced the frequencies and amplitudes of mIPSCs ( Figure 5E,F). These data provide a clue that the Ninj2-deficient oligodendrocytes impair neuronal activities by its excessive Ccl2 secretion.

Inhibition of Cell Necroptosis in Olig1 cre/+ ;Ninj2 fl/fl Mice Rescues Myelination and Alleviated Depressive-Like Behavior
To explore the physiological significance of protecting oligodendrocytes from necroptosis, we treated the mice with Nec-1s to show its effects in vivo. Consistent with the findings in vitro, after Nec-1s treatment, myelination in Olig1 cre/+ ;Ninj2 fl/fl mice was largely restored. In corpus callosum, their myelinated axon numbers significantly increased (Figure 6A). In both hippocampus CA1 area and prefrontal cortex, the MBP + area and Olig2 + numbers were also rescued, compared with the untreated mice ( Figure 6B,C). In keeping with the improved myelination, the motor defects in Olig1 cre/+ ;Ninj2 fl/fl mice were also alleviated, as they had improved grip strength and motor coordination ( Figure 6D,E).
Accompanied with the improved myelination, the neuronal structure and activities were recovered in Olig1 cre/+ ;Ninj2 fl/fl mice after Nec-1s treatment. In hippocampus CA1 area, the pyramidal neurons structure was restored, as indicated by the increased dendritic complexity, and density of spines (Figure 7A,B and Figure S9, Supporting Information). Most importantly, the neuronal activities were recovered in both hippocampus CA1 area and prefrontal cortex, after Nec-1s treatment. As shown, in both areas, the frequencies of mEPSCs were elevated, whereas the frequencies of mIPSCs were reduced, to similar ranges as that in WT mice ( Figure 7C,D). As a result of the Figure 7. Nec-1s treatment alleviates depressive-like behaviors. WT or Olig1 cre/+ ;Ninj2 fl/fl (cKO) mice received i.p. injection with vehicle or Nec-1s (10 mg kg −1 ) from P60 to P90, and then subjected to the following experiments. A,B) Mice were subjected to Golgi staining. Sholl analysis of dendritic complexity in hippocampal CA1 pyramidal neurons (n = 25 neurons for WT+Veh, n = 29 neurons for cKO+Veh, n = 32 neurons for WT+Nec-1s, n = 30 neurons for cKO+Nec-1s, from three mice/group) was shown in (A) (* indicated the significance of comparison between WT+Veh and cKO+Veh, # indicated the significance of comparison between cKO+Veh and cKO+Nec-1s). Apical distal spine numbers of the pyramidal neurons in hippocampal CA1 regions (n = 41 dendritic segments for WT+Veh, n = 45 dendritic segments for cKO+Veh, n = 40 dendritic segments for WT+Nec-1s, n = 41 dendritic segments for cKO+Nec-1s, from 3 mice/group) were shown in (B). C,D) Whole-cell patch-clamp recording was performed to show the representative traces and quantification of frequencies and amplitudes of mEPSCs or mIPSCs from hippocampal CA1 pyramidal neurons (mEPSC: n = 15 neurons for WT+Veh, n = 17 neurons for cKO+Veh, n = 18 neurons for WT+Nec-1s, n = 17 neurons for cKO+Nec-1s; mIPSC: n = 17 neurons for WT+Veh, n = 17 neurons for cKO+Veh, n = 18 neurons for WT+Nec-1s, n = 18 neurons for cKO+Nec-1s; from 5 mice/group) (C), or prefrontal cortex pyramidal neurons (mEPSC: n = 21 neurons for WT+Veh, n = 21 neurons for cKO+Veh, n = 21 neurons for WT+Nec-1s, n = 15 neurons for cKO+Nec-1s; mIPSC: n = 19 neurons for WT+Veh, n = 20 neurons for cKO+Veh, n = 17 neurons for WT+Nec-1s, n = 16 neurons for cKO+Nec-1s; from 5 mice/group) (D). E-G) Tail suspension test (TST), forced swimming test (FST), and sucrose preference test (SPT), were performed to evaluate their depressive-like behaviors (n = 6 mice/group). All the quantification data are presented as mean ± SEM, p-values are calculated using two-way ANOVA with Tukey's www.advancedsciencenews.com www.advancedscience.com recoveries on neuronal activities, the depressive-like behaviors in Olig1 cre/+ ;Ninj2 fl/fl mice significantly diminished. After Nec-1s treatment, Olig1 cre/+ ;Ninj2 fl/fl mice had significantly reduced immobility times in TST and FST, and increased sucrose preference, compared with the vehicle-treated counterparts ( Figure 7E-G). Combining all these data, we show that inhibition on the oligodendrocytic necroptosis in Ninj2-deficient mice is efficient to diminish their depressive-like behaviors.

Discussion
In our current study, we found that Ninj2 deletion in oligodendrocytes exhibited depressive-like behaviors. Loss of Ninj2 inhibited oligodendrocyte development by activating TNF -mediated cell necroptosis. The induced oligodendrocytic necroptosis impaired myelination, and simultaneously changed the neuronal structure and activities in the implicated depression-related brain regions, the prefrontal cortex and hippocampus CA1 area. Conversely, inhibition of necroptosis effectively counteracted the phenotypes in these mice. Our findings established a substantial connection between oligodendrocytes and depression.
After data mining with the online database in the depression mice models, [10,13] we chose Ninj2 as our research candidate, because of its predominant expression patterns in oligodendrocyte lineage cells. Until now, the function of Ninj2 in oligodendrocytes was still poorly studied. In our model, we found that Ninj2 protected oligodendrocytes from TNF -induced canonical necroptosis. Among the membrane-located cytokine receptors activating canonical necroptosis, TNFR1 was the most abundantly expressed in oligodendrocytes. [10] And our RNA-Seq analysis showed that a large set of genes that are involved in TNF signaling pathway was induced. Furthermore, a robust interaction between Ninj2 and TNFR1, and an inhibitory effect of Ninj2 on the recruitment of TNF to TNFR1 were observed. Combining those evidences, we summarize that Ninj2 functions as a protector for oligodendrocytes from canonical necroptosis. Notably, as previously reported, oligodendrocytes were susceptible to necroptotic cell death. [19,25] According to the database reported by Zhang et al., the expression levels of TNF and TNFR1 were both higher in oligodendrocytes than in neurons or astrocytes, which might partially contribute to the cell susceptibility to programmed cell death. [10] In this study, our findings implied that the oligodendrocyte-specific expressing Ninj2 could be a protective machinery for the cells during the course of development.
The importance of oligodendrocytes in mental health has been emerging. It has been reported that oligodendrocytes are involved in the pathologies of several mental illnesses, including depression, [18a,b] bipolar disorder, [18c] anxiety, [18d,e] and schizophrenia. [18f,g] The relationship between oligodendrocytes and depression has been drawing increasing attention. Abnormal myelin integrity appears at the early stage of depression, and worsen as the disease progressed. [8a] Neurotransmitters, [18a,b,26] microRNAs, [27] regulatory factors, [18b,28] and inflammatory cytokines [29] from oligodendrocytes have been implicated in the pathology of depression. In our study, according to the evidence obtained from different mouse models, we conclude that the oligodendrocytes dysfunction would be sufficient for the presence of depression. Using the chemogenetic mouse model, we illustrate that inhibition of oligodendrocytes activity leads to depressive-like behaviors. And with the Olig1 cre/+ ;Ninj2 fl/fl mice, we found that increased oligodendrocyte necroptosis could also be a cause for depression. Notably, when myelination in Olig1 cre/+ ;Ninj2 fl/fl mice was restored by Nec-1s treatment, the depressive-like behaviors in these mice synchronically ameliorated. Our results provided additional evidence to demonstrate the importance of oligodendrocytes in the pathology of depression.
Nowadays, the communication between oligodendrocytes and neurons has become a popular topic. Previously, we found that oligodendrocytes could regulate neuronal activities and controlled food intake by its lactate production. [30] For depression, as mentioned above, a few small molecules produced from oligodendrocytes have been implicated in the pathology of depression. [18] In our study, after Ninj2 ablation, among the TNF downstream cytokines, Ccl2 was the most significantly induced gene in oligodendrocytes. Treatment of Ccl2 significantly reduced the neuronal activities. Consistent with our results, Ccl2 has been reported to modulate neural activities in several kinds of neurons. [24] Therefore, we believed that Ccl2 mediated the signal delivery, at least partially, from oligodendrocytes to neurons that control the presence of depression. CCR2 was the receptor for Ccl2 in neurons, [31] and it is a G protein-coupled receptor that binds G i. [32] Recruitment of Ccl2 to CCR2 inhibited intracellular cAMP levels, which activates its downstream protein kinase A signaling pathway and plays a significant role in the development of depressive disorders. [33] Therefore, the Ccl2-CCR2-cAMP axis could be a potential machinery for the control of neuronal activities and the presence of depression. Besides Ccl2, we could not exclude the possibility that other molecules might also participate in this conversation, such as other cytokines, neurotransmitters, or even nutrients.
It was worth mentioning that we observed dysmyelination and depressive-like behaviors simultaneously in the Ninj2-deficient mice. Improper myelination was associated with impaired motor function. [34] Although we did not find any difference between the WT and mutant mice in the total walking distance in the open field test ( Figure S3, Supporting Information), we still could not completely rule out the possibility that the motor impairment caused by dysmyelination was at least partially associated with the poor performances in TST and FST. Interestingly, although the communication between oligodendrocytes and neurons should exist widely, the Ninj2-deficient mice only exhibited apparent www.advancedsciencenews.com www.advancedscience.com depressive-like behaviors, but not memory or recognition defects, or anxiety. The cause of the quite unique phenotype in the Ninj2deficient mice is still an open question. The oligodendrocytes in the depression-relevant regions, such as PFC and hippocampus, might have specified properties when communicating with their nearby neurons. And the neurons in these regions might also have specified susceptibilities to the molecules secreted by their nearby oligodendrocytes. These remaining questions are worthy for further studies to better show the specificity of oligodendrocytes/neurons communication.
Mouse Cortical Neuron Primary Culture: Primary neurons were dissected from E16.5 mouse embryos as described previously. [36] Briefly, brain cortices were dissected, the meninges were removed, and cortical tissue was dissociated by enzymatic digestion. Isolated primary cortical neurons were plated on poly-D-lysine coated dishes, and cultured in Neurobasal medium supplemented with 1% penicillin-streptomycin, 1% Lglutamine, 2% B27, and maintained in a 5% CO 2 incubator at 37°C.
Immunofluorescence: Mouse brains were perfused with 4% paraformaldehyde (PFA) followed by dehydration with 30% sucrose, then 16 μm frozen sections were prepared. Mouse brain sections or cultured cells were incubated in blocking solution for 1 h at room temperature. Primary antibodies were applied overnight at 4°C. The next day, sections were incubated with the secondary antibodies for 2 h at room temperature. Images were acquired using a confocal microscope (Leica SP8). The fluorescence intensity analysis and cell counting were performed using ImageJ software.
TUNEL: DeadEnd Fluorometric TUNEL System (Promega, Cat# G3250) was used by following manufacture's protocol to detect cell death.
Real-Time Quantitative PCR: Total RNA was extracted using TRIzol (Takara, Cat # 9109) according to the manufacturer's instructions. cDNAs were prepared from 1 μg total RNA using Hifair II 1st Strand cDNA Synthesis SuperMix for qPCR (Yeasen, Cat# 11123ES60) according to the manufacturer's instructions. Real-time quantitative PCR was performed using Hieff qPCR SYBR Green Master Mix (Yeasen, Cat# 11201ES08). Primers used are listed in Table S1, Supporting Information.
Co-Immunoprecipitation: Cells were transfected with plasmid for 48 h, then lysed and incubated with anti-HA antibodies overnight at 4°C. Then the complexes were precipitated with 20 μL of Protein A Magnetic Beads (MCE, Cat# HY-K0203) with gentle agitation at 4°C for 3 h. The beads were washed, and the immunoprecipitated protein complex was processed for western blotting.
Nucleofection of Rat Oligodendrocyte Precursor Cells: Nucleofection of rat OPCs was performed using the Amaxa Nucleofector protocol (Lonza). Briefly, the rat OPCs were harvested by trypsinization. The cell pellet was resuspended carefully in 100 μL Nucleofector Solution per sample. Plasmids (pcDNA3.3-HA-TNFR1/plvo3-Flag-Ninj2) were added to cell suspension. The mixture was subjected to nucleofection using Nucleofector Program O-017 in a certified cuvette. After nucleofection, 500 μL of the pre-equilibrated rat OPC growth medium were immediately added into the cuvette, and the samples were gently transferred into the prepared plate. After 48 h, the cells were lysed and incubated with anti-HA antibodies, followed with co-immunoprecipitation methods.
Enzyme Linked Immunosorbent Assay: Ccl2 levels in cell culture supernatant were measured using mouse Ccl2 enzyme linked immunosorbent assay (ELISA) Kits (Boster, Cat# EK0568) according to the manufacturer's instructions.
Golgi Staining: Golgi staining was performed using the FD Rapid Golgi Stain kit according to the manufacturer's instructions (FD Neuro Technologies, INC). Briefly, mice brains were immersed in the impregnation solution, comprising equal volumes of Solutions A and B, and stored at room temperature for 2 weeks in the dark. Then the brains were transferred into Solution C and stored at 4°C in the dark for at least 48 h. The brains were sectioned using a vibratome (Leica 1200S) at a thickness of 150 μm. The sections were then stained with Solutions D and E. Golgistained hippocampal CA1 pyramidal neurons and the dendritic segments were imaged by microscopy (Olympus FV1000).
Analysis of Dendritic Complexity and Spine Density: As described previously, [38a,b] dendritic complexity and apical distal spine numbers were analyzed using ImageJ software. Dendritic intersection analysis was measured using the Sholl analysis plugin in ImageJ. For Sholl analysis, the dendritic trees were traced and reconstructed using the NeuronJ plugin in ImageJ, and then the traced dendritic trees were converted into a binary image. Sholl analysis parameters were as follows: starting radius, 10 μm; ending radius, 200 μm; radius step size, 10 μm. For spine density quantification, the apical distal dendrite was traced and its length was measured by ImageJ software, and then the number of spines was counted.
Behavior Tests: For all behavioral tests, adult male mice were used. Rotarod Test: Motor function and balance ability were evaluated by rotarod apparatus. Mice were placed on a rotarod (Ugo Basile). The rotarod was constantly accelerated from 5 to 40 rpm and the latency to fall was recorded. The procedure was repeated for five trials, which was averaged to give the latency to fall for each mouse.
Forelimb Grip Strength: The grip strength meter (Ugo Basile) was used to measure the forelimb grip strength. The mice were held by the base of the tail and allowed to freely grasp a grid that was connected to a force transducer with only their front paws, and were steadily pulled back horizontally. Each mouse was measured for five trials, and the average force was recorded.
Forced Swimming Test: Mice were placed in a clear cylinder filled with water (23 ± 1°C) for 6 min and recorded by a video tracking system (Smart 3.0). Immobility time in the total 6 min was measured.
Tail Suspension Test: Mice were suspended by their tails from a tail suspension frame for 6 min and monitored using a video tracking system (Smart 3.0). Escape-related behavior was assessed, and immobility time during the 6 min suspension period was recorded.
Sucrose Preference Test: Mice were trained to consume 1% sucrose and water from two different bottles. Sucrose and water bottles were placed on randomly assigned sides of the cage, and were switched to eliminate side preference. Then mice were allowed free access to 1% sucrose and water from two differing bottles, the total weight of liquid consumed from each bottle was measured after 24 h. Sucrose preference was calculated according to the following formula: Sucrose preferences (%) = sucrose consumption/(sucrose + water consumption) × 100%.
Y Maze Test: Y maze test was carried out in a three arms horizontal maze. Mice moved freely through the maze for 5 min, recorded by a video tracking system (Smart 3.0). Spontaneous alternation was defined as successive entries into the three arms.
Elevated Plus Maze Test: The elevated plus maze consisted of two open and two closed arms. Mice activities were tracked for 5 min with a video tracking system (Smart 3.0). The time spent in the open arms was measured.
Open Field Test: Mice were placed in the center of an open field box, and their activities were recorded for 10 min with a video tracking system (Smart 3.0). The time spent in the center zone, as well as the total distances traveled, were measured.
Novel Object Recognition Test: The test was carried out in an open box containing two identical objects. Mice were placed in the center of the apparatus and explored for 5 min. 2 h later, one of the objects was replaced by a novel object with different form and color, and the exploratory activity was recorded for 5 min by a video tracking system (Smart 3.0). The novel object recognition index was determined as the ratio of the time spent in exploring the novel object to the total time spent in exploring both objects.
Water Maze: The water maze was divided into four quadrants, mice were trained to escape from water by swimming to the platform, and recorded by a video tracking system (Smart 3.0). All mice were trained four times daily for 4 consecutive days. On day 5, the platform was removed and a probe test was conducted. The time spent in the platform-located quadrant, distance traveled in the platform quadrant, or times crossing the platform area was measured.
Statistical Analysis: The sample sizes (n) were indicated in the figure legends. For animal behavior tests, the data were automatically captured and analyzed by an automatic analysis program. All statistical analyses were performed using GraphPad Prism version 8.0. Data were present as Mean ± SEM. A two-tail test was chosen for the cases using an unpaired ttest. For cases using one-way ANOVA or two-way ANOVA, the Tukey's multiple comparisons tests were performed for posthoc tests. p values were indicated with single asterisk (*p < 0.05), double asterisks (**p < 0.01), triple asterisks (***p < 0.001), and quadruple asterisk (****p < 0.0001) on graphs. Data collected were normalized to the control group. No data were excluded from statistical analysis. www.advancedsciencenews.com www.advancedscience.com

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.