Background
During early life when the brain is exquisitely sensitive to signals, both exogenous factors such as environmental and maternal care, as well as endogenous mediators, including immunological molecules and neurotrophins, affect brain development. An emerging body of literature has demonstrated that immune activation during the early postnatal period may markedly induce cognitive impairment after a subsequent immune challenge later in life [
1]. Nevertheless, Bilbo et al. [
2] also reported that neonatal immune activation caused by
Escherichia coli protected against rather than aggravated stressor-induced depressive-like symptoms.
Another feasible scenario is that preconditioning with a low dose of lipopolysaccharide (LPS) can attenuate the pathological effects of a subsequent stimulus, such as a larger LPS challenge [
3], brain trauma [
4], or stress [
5,
6]. Although the precise mechanism is not fully understood, facilitating M2 activation of resident microglia induced by LPS preconditioning at least partially explains this favorable phenomenon [
7].
Interestingly, we recently reported that neonatal BCG vaccination modulates dendritic development, neurogenesis, and behavior in early life by inducing M2 microglial activation and a neurotrophic neuroimmune pattern in the brain [
8‐
10]. In another study, we showed that infiltrating T lymphocytes in the choroid plexus (CP) contribute to induce M2 microglial polarization and secret neurotrophic factors during a normal physiological state [
11]. In addition to T lymphocytes, monocyte-derived anti-inflammatory macrophages were also detected in the CP [
12] and the pathologic central nervous system (CNS) [
13]. This observation included CP activation after evoking the peripheral immune response without breakdown of the blood-brain barrier (BBB), where the myeloid cells exerts inflammation-resolving effects and thus fights off the CNS pathology. Importantly, mice trained in a spatial learning and memory test, the Morris water maze (MWM), led to recruitment of IL-4-producing T lymphocytes into the meninges, and depletion of T lymphocytes from the meningeal spaces skewed meningeal macrophages toward a pro-inflammatory phenotype [
14].
Exposure to an enriched environment (Enr) has been suggested to exert positive effects on plasticity in the hippocampus and other regions in the brain, including elevated learning and memory, neurogenesis, cell survival [
15], alterations in microglial phenotypes [
16], antigen expression [
17], and neuroimmune functions [
18]. Enr increases the expression of growth factors, such as BDNF and IGF-1, in the brain. Moreover, peripheral T lymphocytes are essential for maintaining normal neurogenesis and cognition. Ziv et al. [
17] also reported that an enriched environment did not enhance neurogenesis in immune deficient mice, whereas another form of cognitive training (MWM) indeed enhanced behavioral performance by an increasing meningeal T lymphocyte recruitment [
14].
The hippocampal plasticity of newborns is more susceptible to environmental exposure than that of adults. Therefore, early enriched experience by pre-weaning into an Enr may induce pronounced effects on neurogenesis and behavior [
19]. However, whether neonatal pretreatment with a BCG vaccination could further influence Enr-related elevated hippocampal neurogenesis and behavior and the precise mechanisms regulating this process remained elusive. Here, we show that BCG vaccination indeed further enhances hippocampal neurogenesis and spatial cognition caused by Enr by activating the CP, recruiting T lymphocytes to the CP, and increasing meningeal macrophage M2 polarization and neurotrophic factor expression.
Discussion
We previously reported that neonatal BCG vaccination elevated neurogenesis, synaptic plasticity, cognitive function, and the levels of neurotrophic factors in mice housed in a standard environment by affecting the neuroimmune cross talk between the periphery and the brain, which may be associated with a systemic Th1 bias [
8‐
10]. Here, we extended the neurobeneficial effects caused by BCG vaccination to the enriched housing condition. We further showed that peripheral immune activation induced by BCG vaccination resulted in some neurobeneficial effects, such as higher levels of BDNF and IGF-1, through skewing meningeal (pia mater and dura mater) macrophages toward an M2 phenotype, and these effects were prevented by minocycline and anti-IL-10 antibody treatment. Importantly, we verified that recruitment of T lymphocytes to the choroid plexus (CP) contributed to meningeal macrophage M2 activation, and these effects could be blocked after anti-TCR antibody treatment. The increased recruitment of T cells to the CP paralleled observed increase in macrophage M2 polarization in the meningeal space. Similarly, the expression of IL-10 (an anti-inflammatory cytokine) in the CP was decreased after anti-TCR antibody treatment, implying that depletion of T cells and change of cytokine IL-10 are coincidence in the CP. Collectively, the present work confirmed that T lymphocytes in the brain ventricle’s CP, meningeal macrophage M2 polarization, microglia M2 activation, and formation of pro-neurogenic niche may combine and contribute to cellular and behavioral differences observed between the BCG-Enr mice and PBS-Enr mice.
The interaction between peripheral immune responses and the CNS has been addressed for decades. However, no plausible immune cellular mechanisms have yet been confirmed to explain the relationship between these two systems. Recently, the “brain border”-meninges including the outermost dura mater, the intermediate arachnoid mater, and the inner pia mater have received increased attention [
29]. Notably, the anatomical structures of meninges, such as the choroid membranes/plexus, connect peripheral blood system to the brain via CSF circulation [
31,
32]. Therefore, meningeal immunity could be a potential target for elaborating the periphery-to-brain communication pathway. Our findings of elevated recruitment of T lymphocytes into the CP and dura mater and meningeal macrophage M2 polarization in response to BCG vaccination provided evidence for the interior relationship between the periphery and the brain. However, how do the infiltrating T lymphocytes regulate macrophage activation in the meninges? We showed that the level of monocyte-derived anti-inflammatory IL-10, which supports beneficial effects on cognitive function, was elevated in the meninges including pia and dura mater, although IL-10 immunoreactivity was not only restricted to macrophages (Fig.
7h). Accordingly, Jonathan Kipnis’ group [
14] also demonstrated that accumulation of another inflammation-resolving cytokine IL-4 was detected in the meninges, in which the involved myeloid cells underwent an inflammatory phenotypic change. In the absence of IL-4, the meningeal myeloid cells were skewed toward a proinflammatory phenotype, which resulted in cognitive impairment in IL-4 knockout mice. Other mechanisms, such as IL-4 and Arg-1, may also play important roles in regulating of macrophage M2 polarization, since increased expression of Arg-1 was observed in the meninges, lateral and third ventricles, and hippocampal CA1 area of the brain (Additional file 3: Figure S4A-F). Choroidal astrocytes are in close proximity to the “brain borders” and the CSF and thus could be mutually regulated by cytokines from the meninges and the CSF. We also showed that astrocytes located in the choroid membrane of the brain ventricle expressed higher levels of BDNF in the BCG-Enr mice than in the PBS-Enr mice. Expectedly, the expression of neurotrophic factors (BDNF and IGF-1) was reversed in the hippocampus by inhibiting meningeal macrophage activation after minocycline or anti-IL-10 antibody treatment. In line with our findings, another anti-inflammatory cytokine IL-4 directly induced BDNF mRNA upregulation in astrocytes in previously published literature [
14]. Together with our findings that the levels of IL-10 in the pia and dura mater were also normalized when macrophage activation was inhibited via minocycline treatment, we concluded that meningeal macrophages-derived IL-10 not only directly regulated choroid astrocytes secretion of BDNF but also induced granular neurons expressing BDNF and microglia expressing IGF-1 in the hippocampus after IL-10 release into the CSF. Thus, the findings from us and others revealed that infiltrating T cells in the “brain border” indeed cross talk with resident macrophages secreting the anti-inflammatory cytokine IL-10 and choroidal astrocytes expressing BDNF in addition to cross talk with resident microglia secreting IGF-1 [
8,
11]. Notably, the benefits from Enr and even BCG vaccination on neurogenesis may not be exclusively caused by the M2 shift; other mechanisms, such as direct effects from circulatory endothelium-derived factors and MHC-II
+ cell-derived signals, may also mediate hippocampal neurogenesis [
33]. Vascular endothelial growth factor (VEGF) stimulates adult neuronal progenitor cell proliferation in the hippocampus and has neurogenic and angiogenic properties [
34,
35]. Since microglia located in the hippocampus are important for the regulation of adult neurogenesis, chemokine CX3CR1 deficient mice perform adult hippocampal neurogenesis deficit and cognitive impairment [
36]. Together, microglia-related signals, chemokines, and other growth factors that affect neural progenitor cell proliferation cannot be excluded.
In the present study, similar tendencies were observed in the levels of cytokines in the serum and hippocampus, although each cytokine was expressed to a different extent. Our immunization approach led to an obvious elevation in cytokine IFN-γ level both in the serum and the hippocampus, which is known to induce the upregulation of specific trafficking molecules for recruitment of T lymphocytes into the meninges [
30,
37]. However, the levels of the pro-inflammatory cytokines TNF-a, IL-1β, and IL-6, which are associated with cognitive deficits, were decreased in the hippocampus at postnatal day (PD) 21. Indeed, the levels of pro-inflammatory TNF-a and IL-6 were increased during the initial 2 weeks after BCG vaccination and then decreased at PD 21 (Fig.
8c). The initial elevation of cytokines can be induced by peripheral adaptive immune responses. The meningeal macrophage M2 activation at PD 21 may ameliorate the pro-inflammatory immune response in the hippocampus and thereby decrease the levels of TNF-a and IL-6 mRNA. Compared with the PBS-Reg mice, we found significant elevation in hippocampal neurogenesis, LTP, and cognitive function in the BCG-Reg mice and PBS-Enr mice. Although both BCG vaccination and the enriched environment promoted the recruitment of peripheral T lymphocytes, which skewed macrophages toward an anti-inflammatory phenotype, to the CP and meninges, the levels of cytokines in BCG mice and Enr mice were different. This may contribute to the transience of the immune response and sustainability in an enriched environment. Of course, other molecular mechanisms may also explain the differences between these two groups.
In the present study, we have determined infiltrating T cells from periphery to the meninges were significantly increased 2 weeks after BCG vaccination and/or Enr exposure compared to the corresponding controls (Additional file 5: Figure S3A). However, we cannot rule out the possibility of the resident T cell proliferation in the meninges after BCG and/or Enr treatment. Moreover, quantitative analysis revealed no significant changes in CD11b
+ macrophages in the meninges among all groups, implying that macrophage M2 polarization rather than macrophage migration was affected in the meninges after BCG vaccination (Additional file 5: Figure S3B). Actually, BCG vaccination indeed had a long-lasting effect, since the increases of cytokine levels in the periphery and T cell count in the meninges were still detected 2 weeks after BCG vaccination (Additional file 5: Figure S3D). Together, we agree that resident macrophages were activated, at least in part, by infiltrating T cell-derived cytokines, in the meninges, but not migrate to the meninges from the periphery. Also, homing monocytes were not detected via their in situ labeling by intravenously injected FITC-conjugated anti-CD11b antibodies, further supported our conclusions (Additional file
6: Figure S5C and D).
Recent reports demonstrated that striking Th2 skewing of T lymphocytes potentially led to meningeal myeloid cell M2 polarization [
38,
39]. However, our findings regarding the T helper (Th) 1 cytokine-IFN-γ were continuously increased in the periphery and the brain and were specialized to recruit T lymphocytes to the CP and induce macrophage M2 polarization. It seems that our findings of Th1 biased-macrophage M2 activation are contradictory with those of previously reported studies. Notably, a recent study reported that depletion of T cells from meningeal space shewed meningeal myeloid cells toward a proinflammatory phenotype (M1 activation) [
14]. Therefore, we formulate a concept that infiltrating T cells and meningeal macrophages maintains a homeostasis under physiological conditions. Alteration of T cell count may modulate meningeal macrophage phenotype. The data revealed that treatment with IFN-γ induced the expression of CCL5, which is important to T cell recruitment. We have also demonstrated that infiltrating T cells to the CP is a requirement for meningeal macrophage M2 activation due to anti-TCR antibody blocking experimental. However, the underlying mechanism remains unclear. We are going to measure which immunomodulatory factors secreted by infiltrating T cells regulating macrophage M2 activation in the further research.
Recently, the brain’s CP activation for recruitment of inflammation-resolving monocytes and T lymphocytes attenuated Alzheimer’s disease (AD) pathology or disease progression in ALS model following glatiramer acetate [
40] or myelin-derived peptide [
13] immunization, with concomitant Th1-cytokine IFN-γ upregulation in the periphery. Our data now demonstrate that peripheral Th1-meningeal macrophage M2 activation may be an underlying mechanism for interpreting the beneficial effects observed in previous studies [
13,
40]. These findings may shed light on the migration of immunoregulatory cells to the CNS for fighting off the neuronal pathology by peripheral immunity.
Acknowledgements
We thank Prof. Zhibin Yao, Dr. Junhua Yang, Dr. Yaru Wen, and Dr. Xiao Wang for the discussions and suggestions for the experiments and the manuscript. We also thank Dr. Yingying Wu and Ms. Dan Song for the language modification.