Bidirectional interactions between the peripheral immune system and microglia are facilitated by cytokines crossing the BBB, primarily composed of endothelial cells, sealed together by tight junctions, covered by pericytes, astrocytic endfeet, and microglial processes (Bisht et al., 2016; Joost et al., 2019). A disrupted BBB can increase the influx of inflammatory molecules and cells to the CNS (Figure 1). Studies looking at CSF or serum concentrations of proteins typically found in the CNS, for instance, CSF hyaluronic acid (Ventorp et al., 2016) and albumin CSF/serum ratio (Bayard-Burfield et al., 1996), revealed that the BBB is compromised in individuals with SB and suicide ideation (Falcone et al., 2010; Torres-Platas et al., 2014b). Suggestive of decreased BBB integrity, mRNA downregulation of the tight junction protein Cldn5 is present in the NA of individuals with MDD who died by suicide (Menard et al., 2017; Dudek et al., 2020). Compared to resistant and healthy controls, male mice susceptible to CSDS exhibited increased cortical leakage of intravenously injected dyes (Lehmann et al., 2018). Microglia isolated from the susceptible mice selectively expressed transcripts involved with extracellular matrix breakdown, inflammatory response, cytokine production, ROS and metabolic processes, which highly correlated with the transcriptomic changes associated with aging (Lehmann et al., 2018). Microglial stress susceptibility could be connected to the BBB disruption, requiring future investigation, particularly in older individuals, where this effect might be exacerbated and contribute to increasing SB risk. Indeed, robust data from animal models supports that aging is linked to microglial pro-inflammatory and oxidative stress activities, reduction in CNS surveillance, as well as appearance of microglial states, such as primed microglia (see section “Microglial Priming Is a Hallmark of Early-Life Adversity and May Contribute to Stress Susceptibility”), connected to synaptic dysfunction (Tay et al., 2018; Stratoulias et al., 2019). Correspondingly, although SB, in general, are more common at younger ages, suicide deaths rates are higher in older individuals in almost all countries (Conejero et al., 2018).

Alternatively, microglial resistance to stress could be associated with the protection of the BBB. Microglia were suggested to infiltrate the BBB toward the periphery and rescue its permeability via CLDN5 protein expression early on during systemic inflammation caused by daily LPS injections in adult male mice (Haruwaka et al., 2019). This protective role, however, changed at later time points, whereby microglia expressed higher levels of CD68 protein, a phagolysosomal activity marker present in human microglia/macrophages and associated with inflammation (Walker and Lue, 2015; Haruwaka et al., 2019). Elevated CD68 occurred along with an increased number of microglial inclusions of aquaporin 4, notably expressed in blood vessels and astrocytic endfeet, suggesting their phagocytosis (Haruwaka et al., 2019). Accordingly, CSDS in adult male mice increases the number of brain CD68-high cells, as revealed by flow cytometry (Lehmann et al., 2016). No significant differences in mRNA Cd68 expression and CD68 immunostaining, however, were observed in the dACC and DPFWM or VPFWM, respectively, of individuals who died by suicide compared to PMI- (Torres-Platas et al., 2014b), sex-, psychiatric diagnosis- (Schnieder et al., 2014), and age-matched controls (Supplementary Table 1). Contrastingly, increased mRNA Cd68 expression was detected in the DLPFC but not ACC of individuals with MDD who died by suicide, compared to age and PMI-matched controls (Zhang L. et al., 2021). Despite strong associations between stress, inflammation and phagocytosis, prospective studies are needed to understand if microglia play a detrimental or beneficial role in BBB integrity specifically in the context of SB.

Despite the link between stress and CNS immune infiltration, previous post-mortem studies provide conflicting data on the presence of IBA1-positive cells associated with/within vessels, or on CD163-positive cell infiltration in the ACC, DPFWM and VPFWM of individuals who died by suicide compared to PMI- (Torres-Platas et al., 2014b), sex-, psychiatric diagnosis- (Schnieder et al., 2014, 2019), and age-matched controls. CD163 is a marker related to phagocytosis that is expressed by both human microglia and perivascular macrophages, although mainly in the latter (Swanson et al., 2020). CD163 protein expression was decreased in CD11b-positive microglia isolated from individuals with MDD compared to healthy controls, although without a specific effect on the one suicide death included in the study (Snijders et al., 2020). Moreover, elevated mRNA expression of Ccl2, alias Mcp-1, a chemokine involved in the recruitment of peripheral cells, and Cd45 mRNA, usually enriched in peripheral macrophage populations, was found in the dACC of individuals with MDD who died by suicide compared to age- and PMI-matched healthy controls (Torres-Platas et al., 2014b). Contrastingly, it is important to consider that microglial populations can express high levels of CD45 protein in male mouse models of stroke and aging (Honarpisheh et al., 2020). Furthermore, decreased Ccl2 transcripts were instead detected in the VLPFC of individuals with MDD compared to healthy controls, although lacking a suicide subgroup effect (Clark et al., 2016). Numerous factors could underlie the variable peripheral immune cell recruitment in SB, such as CNS regions, sexes, age, as well as inter-individual differences in stress susceptibility and resistance of the cohorts studied (Lehmann et al., 2018). Additionally, elevated CD163 immunostaining in the DPFWM was associated with a suicide-specific effect on microglial migration to abluminal surfaces of vessels, instead of peripheral immune cell infiltration (Schnieder et al., 2014, 2019). Previously reported elevations in IBA1-positive cells adjacent to vessels in the DPFWM of individuals who died by suicide compared to age-, sex- and psychiatric diagnoses-matched controls were not paralleled by changes in CD163-positive densities, indicating they likely arose from resident microglial staining (Schnieder et al., 2014, 2019). Increased association of microglia with vessels was shown upon systemic inflammation caused by daily LPS injections in male mice (Haruwaka et al., 2019) and a similar phenomenon might be present in SB. The dynamic interaction between microglia and blood vessels should be accounted for in future research looking at peripheral immune CNS invasion in SB.

Microglia, perivascular macrophages and subdural leptomeningeal macrophages derive from yolk-sac erythromyeloid progenitors matured into myeloid precursor cells that seed the brain around E9.5 in mouse (Ginhoux et al., 2010; Kierdorf et al., 2013). Peripheral infiltrating monocytes or macrophages, by contrast, arise from erythromyeloid progenitors that travel to the fetal liver and later transition to the bone marrow starting around birth (Hoeffel et al., 2015; Stremmel et al., 2018). Similar ontogeny contributes to the large overlap in markers such as CX3CR1, CSF1R, CD11b, IBA1, HLA-DR, CD163, CD68, CD45 between microglia, BAM, and peripheral monocytes (Walker and Lue, 2015; Snijders et al., 2020; Swanson et al., 2020; Prinz et al., 2021). Region-specific analyses using largely microglia-specific markers such as TMEM119, P2RY12, SALL1, and HEXB (Prinz et al., 2021) are warranted to clarify the contribution of microglia versus non-microglial immune cells possibly acting together during SB.

MIA induced using a maternal high-fat diet results in exacerbated responses of plasma cytokines, such as IL6, to LPS, and decreased microglial interactions with the extracellular space in the HIP of male and female adolescent mouse offspring (Bordeleau et al., 2020), which might be relevant for structural remodeling of synapses or myelination (Nguyen et al., 2020). Indeed, in a follow up study, maternal high-fat diet was associated with decreased postnatal presence of structures that support the axon-myelin unit known as cytosolic myelin channels, in the corpus callosum of adolescent males, contrasted by increased microglial synaptic contacts in both sexes (Bordeleau et al., 2021). An altered microglial support of myelination could be connected to the myelin impairment detected in the ACC epigenome and the transcriptome of post-mortem brain samples from individuals with MDD and a history of ELA who died by suicide (Lutz et al., 2017). Considering the marked sexual dimorphism observed in ELA and MIA, it will be crucial that studies addressing this question consider sex differences in microglial behavior.

Male mice receiving an intraperitoneal injection of LPS during postnatal development display microglial upregulation of CX3CR1 which, in a TLR4-dependent manner, causes heightened synaptic engulfment in the ACC and results in MDD-like symptoms upon acute exposure to unpredictable stressors during adolescence (Cao et al., 2021). Notably, elevated TLR4 and TLR3 expression was found in the post-mortem DLPFC of individuals who died by suicide, irrespective of MDD diagnoses (Pandey et al., 2014). Moreover, the TLR2 was among the top ten genes involved in the DLPFC of individuals with MDD who died by suicide, compared to individuals with MDD who died of other causes and healthy controls (Zeng et al., 2020; Supplementary Table 1). According to the results obtained by Cao et al. (2021), TLR expression changes in post-mortem brain tissue could be indicative of increased postnatal peripheral inflammation leading to microglial synaptic pruning and elevated susceptibility to SB, warranting further investigation.

Postnatal exposure to alcohol in adult male rats increased HYP protein expression of CD11b, a component of complement receptor 3 mediating microglial pruning of axon terminals (Schafer et al., 2012; Chastain et al., 2019). Upregulation of CD11b was accompanied by heightened mRNA levels of Tnf, Il6, Tl4, and Csf1r both in vivo and in vitro, as well as protein levels of epigenetic modulators in microglia from the HYP paraventricular nucleus (Chastain et al., 2019). Particularly, Chastain et al. (2019) revealed a decreased global DNA methylation but increased acetylation of factors that upregulate inflammatory gene transcription, such as histone H3 lysine 9, which prevailed throughout adulthood. Less transcriptional repression in microglia was paralleled by an elevated GC response to a LPS challenge during adulthood and reversed by minocycline (Chastain et al., 2019), a tetracycline normalizer of microglial cytokine release and phagocytosis (Šimončičová et al., 2022). GC are produced by the HPA axis following exposure to physiological or biological threats and have numerous functions, largely affecting inflammatory and metabolic processes (de Kloet et al., 2005; Lucassen et al., 2014; Picard et al., 2021), as well as synaptic plasticity, behavior, and memory (McEwen and Magarinos, 2001). Polymorphisms and differential epigenetic regulation of numerous factors along the HPA axis can underlie impaired stress responses, together with increased psychiatric risk, SB and suicide ideation (Roy et al., 2017; Ke et al., 2018; Menke et al., 2018; Berardelli et al., 2020; Nold et al., 2021). It is possible that inflammatory and epigenetic responses mediated by HYP primed microglia and impacting the HPA support the corresponding increase of SB risk in adolescents exposed to alcohol during development (O’Connor et al., 2019), although warranting more substantial evidence. Additionally, this HPA effect could be mediated via CD11b-complement 3 synaptic remodeling. Correspondingly, decreased mRNA Cd11b expression was detected in the ACC of individuals with MDD who died by natural causes but not suicide, compared to age and PMI-matched controls (Zhang L. et al., 2021). Moreover, increased complement protein 3 was detected by microarray in the AMY, HIP, TLM of individuals who died by suicide compared to healthy controls (Glavan et al., 2021). Future studies are needed to explore the role of complement synaptic pruning in SB.

The increased density of primed cells observed by Torres-Platas et al. (2014b) constitutes the only available microglial quantitative morphological analysis in the context of SB. Microglia use cues from the microenvironment to tune their branching processes, as well as the shape and size of their soma, and modulate how motile, mobile and interactive they are with their surroundings (Savage et al., 2019). Enlarged microglial soma size, present in primed cells, was linked to enhanced gene transcription in adult mice (Dungrawala et al., 2010; Kozlowski and Weimer, 2012; Marguerat and Bähler, 2012) and to microglial mobility in aged male and female mice (Hefendehl et al., 2014). Moreover, elevations in IBA1-positive cell area, also detected after priming, were associated with the uptake of neuronal inclusions which coincided with a decreased spine density in the PFC of adult male mice exposed to CUS (Milior et al., 2016; Wohleb et al., 2018). Microglial morphological adaptations related to phagocytosis of neuronal elements were also linked to MDD-like behaviors in CUS-exposed male mice (Milior et al., 2016; Wohleb et al., 2018). Future research is needed to evaluate whether priming-driven changes in microglial morphology are tied to synaptic remodeling and if they could help understand, for example, the impaired reward and decision making, as well as mental pain associated with SB in the DLS and NA (Schmaal et al., 2020). Although microglial morphology is intimately associated to their activity, analyses of microglial morphological states can be highly contradictory, as shown in the stress literature, probably due to experimental variability in the time course and types of stress paradigms used, as well as regions examined and various coping strategies employed (Picard et al., 2021). Qualitative data has argued for a lack of difference in morphology of HLA-DR-positive cells in the DRN of individuals with MDD who died by suicide compared to individuals with MDD who died by other causes and healthy controls (Brisch et al., 2017). In particular, this study compared cells with enlarged cell body and few ramifications, also known as ameboid, and microglia with numerous radial, thin ramifications, defined as ramified (Brisch et al., 2017). Similar lack of SB-specific effects was suggested after comparing the density of HLA-DR ameboid and ramified microglia within the ACC, DLPFC, HIP, and MTN of two individuals with SCZ who died by suicide compared to age-, sex-matched healthy controls (Steiner et al., 2006). Other qualitative assessments, for example, in the ACC and VLPFC, resulted in similar findings (Steiner et al., 2011; Schnieder et al., 2014; Clark et al., 2016; Petrasch-Parwez et al., 2020). Quantitative assessments of microglial morphology are notably required to verify more nuanced differences along the spectrum of morphologies microglia can assume, which goes beyond ramified and ameboid states (Savage et al., 2019). Prospective studies should consider the variety of factors that can influence the shape and size of microglia and contextualize in space and time their morphological and stereological investigations.

Alike stress, specific CNS regions were more susceptible to changes in IBA1 or HLA-DR upregulation in post-mortem samples. Increased density of cells immunostained for IBA1 or HLA-DR was found in the dACC of male and female individuals with MDD or SCZ who died by suicide (Supplementary Table 1) compared to sex- (Steiner et al., 2006, 2008) and age-matched healthy controls (Torres-Platas et al., 2014b). Elevated HLA-DR was also observed in the DLPFC and MTN of female and male individuals who died by suicide, regardless of their psychiatric diagnosis (Steiner et al., 2006, 2008). Conversely, decreased HLA-DR immunostaining was detected in the DRN of individuals with MDD who did not die by suicide, compared to those who did and to healthy controls (Brisch et al., 2017). By contrast, no significant differences in HIP HLA-DR staining were, however, detected in two individuals with SCZ who died by suicide, compared to healthy matched controls (Gos et al., 2014). Moreover, IBA1-positive cell density in the aMCC did not differ between individuals with SCZ who died by suicide compared to age-matched individuals with SCZ who died by other causes or healthy controls (Petrasch-Parwez et al., 2020).

Cx3cr1 transcripts are increased in the ACC but not in the DLPFC of individuals with SCZ who died by suicide compared to individuals with SCZ who died by other causes (Zhang L. et al., 2021; Supplementary Table 1). Further, CD11b-positive cells from MFG, STG, TLM, and SZ tissue presented increased mRNA expression of Cx3cr1 when isolated from individuals with MDD compared to age-matched healthy controls (Snijders et al., 2020). According to a microarray study, increased Cx3cr1 was also detected in the AMY, HIP, and TLM of individuals who died by suicide compared to healthy controls (Glavan et al., 2021). By contrast, no difference in Cx3cr1 mRNA expression was detected in the ACC or DLPFC of individuals with MDD who died by natural causes and suicide, compared to age and PMI-matched controls (Zhang L. et al., 2021). Microglia and neuronal cells use a wide variety of signaling molecules, including fractalkine, to maintain their balanced communication in the CNS, required for optimal function of both cellular types (Paolicelli et al., 2014). Fractalkine signaling relies on binding of CX3CR1, largely expressed by microglia in the CNS, to the neuronal chemokine fractalkine, or CX3CL1, which generally inhibits microglial activity (Harrison et al., 1998; Biber et al., 2007). CX3CR1-CX3CL1 activity is needed for neuronal maturation and survival, plasticity of synapses and behavior (Paolicelli et al., 2014). During postnatal development, CX3CR1 is required for microglial recruitment to the cortex and maturation of postsynaptic glutamate receptors in female and male mice (Hoshiko et al., 2012). Furthermore, CX3CR1 deficiency impairs HIP neuronal long-term potentiation, a common paradigm to study synaptic plasticity, and results in cognitive and neurogenesis deficits in adult male mice (Bachstetter et al., 2011; Rogers et al., 2011). Given that fractalkine signaling is needed for optimal function of both cellular types (Paolicelli et al., 2014), regional mechanisms involving CX3CR1 could participate in susceptibility to stressors and inflammation contributing to SB risk. Despite the robust role of fractalkine signaling in synaptic plasticity, following CSDS in adulthood, CX3CR1 knockout male mice did not develop anxiety-like behavior compared to wild-type mice, an effect attributed to their lack of peripheral immune cell recruitment to the CNS (Wohleb et al., 2013). Moreover, adult CX3CR1 knockout male mice exposed to CUS failed to modify their HIP microglial morphology, phagocytosis, neuronal plasticity and did not show MDD-like behaviors compared to wild-type mice (Milior et al., 2016; Rimmerman et al., 2017). Similar resistance to developing MDD-like behaviors was found in an adult male and female mouse model of chronic behavioral despair (Hellwig et al., 2016) and in adult male mice exposed to ELA and chronic variable stress in adulthood (Winkler et al., 2017). Additionally, in a human induced pluripotent stem cell microglial model, CX3CR1 knockout resulted in heightened phagocytosis, as assessed with a bead assay (Murai et al., 2020). Region-specific elevations in Cx3cr1 expression found in post-mortem samples could confer an increased stress susceptibility to microglia, disrupting phagocytosis and neuronal plasticity and contributing to SB (Figure 1), which warrants further research.

Disrupted mitochondria, dilation of the Golgi apparatus and endoplasmic reticulum, as well as cell shrinkage and abundant endosomes, are hallmarks of DM, a microglial state uncovered by our group (Bisht et al., 2016). DM often encircle synaptic structures and are occasionally surrounded by extracellular space containing debris, altogether suggesting ongoing synaptic pruning and extracellular digestion (Bisht et al., 2016). Albeit nearly absent during mature steady-state conditions in the HIP, PFC, HYP, and BLA, DM become abundant in male mice subjected to MIA and in mouse models of stress, including CSDS and CUMS (Bisht et al., 2016). Given the appearance of DM upon distal and proximal SB risk factors as well as their marked oxidative stress, it is tempting to speculate that DM are potentially found upon SB. Electron microscopy studies assessing post-mortem samples from suicide deaths could clarify this matter. Notably, “dystrophic” microglia that resemble DM are present in the post-mortem PFC of individuals with SCZ (Uranova et al., 2018). Correspondingly, DM could help explain the dendritic atrophy and reduced spine density found in the PFC and HIP of male and female rats and mice exposed to CRS and CUMS (Qiao et al., 2016), similar to what is observed in the same regions of individuals with MDD (Duman et al., 2016; Bollinger and Wohleb, 2019).

In addition to marked oxidative stress, DM are characterized by an electron-dense nucleus with altered nuclear heterochromatin (Bisht et al., 2016; St-Pierre et al., 2020). Nuclear heterochromatin remodeling is a process coupled to DNA methylation and transcriptome adaptations (Robertson, 2002). Notably, DNA methylation is one of the proponent mechanisms by which ELA translates into higher SB risk later in life (Lutz et al., 2017; Zeng et al., 2020). ELA-induced methylation could potentially drive the appearance of DM, albeit more direct evidence is needed to support this hypothesis. Initial studies indicate microglia can be the target of complex epigenetic modulation after exposure to psychological stress, such as ESI and ESS (Wang et al., 2017; Catale et al., 2020), or physiological challenges, for instance early-life alcohol exposure (Chastain et al., 2019). IBA1-positive cells located in the core and shell of the NA, dorsomedial striatum and DLS, CA1 and CA3 of the HIP, as well as BLA and central AMY showed a marked decrease of methylation in ESI mice compared to control and ESS groups (Catale et al., 2020). Additionally, the authors revealed a decreased global methylation in the CA1 of mice exposed to ESS compared to controls from both sexes (Catale et al., 2020). While similar mechanisms might participate in SB, microglial epigenetic studies in the context of SB neuropathology are currently lacking.

In the brain, L-kynurenine gets broken down into 3-hydroxykynurenine, QUIN and xanthurenic acid, all synthesized by microglia and peripheral macrophages according to human cells in vitro evidence (Heyes et al., 1992, 1996; Espey et al., 1997; Gos et al., 2014), while kynurenic acid and picolinic acid are produced by astrocytes (Raison et al., 2006; Fujigaki et al., 2017). QUIN is a marker of oxidative stress which was found to be significantly elevated in the CSF of individuals with a history of suicide attempts (Erhardt et al., 2013) and in the ACC of individuals with MDD who died by suicide compared to age- and sex-matched healthy controls (Steiner et al., 2011; Supplementary Table 1). In the same cohort, however, decreased QUIN immunoreactivity was found in the HIP CA1 of individuals with MDD or bipolar disorder who died by suicide compared to age- and sex-matched healthy controls, indicating that different brain regions likely present distinct microglial oxidative balances in SB (Busse et al., 2015). Similarly, individuals with SCZ presented reduced QUIN binding in the CA1 compared to controls without psychiatric diagnoses, despite a lack of specific changes in the two suicide deaths included in this study (Gos et al., 2014). Robust evidence links oxidative stress and SB; however, more studies are warranted to explore the extent through which microglia contribute to modulating region-specific tryptophan metabolism in particular mental health disorders and SB (Suzuki et al., 2019; Baharikhoob and Kolla, 2020).

Tryptophan is a precursor of serotonin (Bakunina et al., 2015; Fujigaki et al., 2017). Increased microglial degradation of tryptophan, supported by changes in IDO isoforms and QUIN levels, could help explain the decreased levels of serotonin metabolites measured in the CSF of individuals with MDD and SB (Brown et al., 1982; Mann et al., 1992; Steiner et al., 2011; Bakunina et al., 2015; Fujigaki et al., 2017; Figure 1). Notably, serotonin depletion is associated with increased impulsivity and aggressive behaviors (Stanley et al., 1982; Stanley and Mann, 1983), low mood, anxiety, as well as suicide risk (Achtyes et al., 2020; Baharikhoob and Kolla, 2020; Koweszko et al., 2020). More research could clarify the specific contribution of microglia to serotonin depletion in the context of SB. Simultaneously, serotonin affects the inflammatory activity, trophic support, phagocytosis and motility of microglia in a context-dependent manner (Turkin et al., 2021). For instance, serotonin treatment enhanced adult mouse cortical microglial process motility toward laser injury but reduced phagocytosis of beads by neonatal amoeboid microglia of the corpus callosum in situ (Krabbe et al., 2012). Adult male mice treated with the selective serotonin reuptake inhibitors fluoxetine exhibited different HIP inflammatory responses according to the quality of their living environment (Alboni et al., 2016). In mice exposed to CUMS, fluoxetine was associated with anti-inflammatory responses, such as reduced Tnf transcripts, increased microglial spacing index and cell body area, but decreased arborization in the HIP CA1 (Alboni et al., 2016). Conversely, in mice housed in an enriched environment, microglia had increased pro-inflammatory responses, involving IL1β protein and Tlr4 mRNA (Alboni et al., 2016). Stress appears to regulate serotonin levels along with microglial activity. Social isolation in adult male mice caused a decrease in DRN serotonin production accompanied by elevated IBA1 and C1q protein expression, as well as depressive-like behavior (Lee et al., 2021). Upregulation in complement expression was paralleled by a downregulation in synaptic proteins, i.e., postsynaptic density protein 95 and synaptophysin, which could indicate synaptic pruning (Lee et al., 2021). It is therefore possible that stress-induced serotonin depletion increases the susceptibility to psychiatric disorders or SB via adaptations in microglial pro-inflammatory activity, cell motility and microglial synaptic remodeling, an area that awaits future investigation.

In LPS-treated adult male mice, minocycline and the IDO antagonist 1-methyltryptophan rescued the development of MDD-like behavior, independently of brain serotonin turnover (O’Connor et al., 2009). Therefore, at least partially, microglial IDO may influence SB independently of serotonin. As previously described, in addition to reducing tryptophan levels, IDO activity results in production of the metabolite QUIN in microglia. QUIN is an agonist of the NMDAR crucial for glutamatergic excitatory signaling in the CNS (Olloquequi et al., 2018). Elevated levels of QUIN, found in association with SB, as discussed above (see section “Microglial Tryptophan Metabolites Are Regionally Detected in Suicidal Behaviors”), can over-activate the NMDAR, generating abnormally high levels of intracellular Ca²⁺, toxic free radicals and disrupted ATP production that ultimately contribute to loss of neuronal cell function and cell death, a process formally known as glutamate excitotoxicity (Olloquequi et al., 2018). QUIN-mediated glutamate excitotoxicity could be another mechanism by which microglia participate in SB (Suzuki et al., 2019; Baharikhoob and Kolla, 2020; Figure 1). Correspondingly, several clinical trials with ketamine, a NMDAR antagonist, found a significant and rapid reduction in suicidal cognition, a reason why ketamine therapy is currently being evaluated for the treatment of suicidal ideation (Ballard et al., 2021). Ketamine was shown to rescue MDD-like behavior, decreasing brain IL6, TNF and QUIN protein production, as well as CX3CR1-positive cell area, whilst increasing CX3CR1-positive cell arborization in the PFC of adult male mice exposed to LPS (Verdonk et al., 2019). Partial depletion of microglia by the CSF1R inhibitor PLX3397 blocked the effects of R-ketamine, a more potent and longer-lasting enantiomer of (R,S)-ketamine in adult male mice susceptible to CSDS (Zhang K. et al., 2020). Hence, ketamine affects microglia in a manner that is partially required to achieve its anti-depressant effects. It is, however, unknown if similar mechanisms underlie the effects of ketamine in SB.

Despite positive results in the context of MDD, an inverse response to ketamine is found in SCZ, where it produces psychosis-like responses in healthy subjects and can temporarily worsen positive symptoms in individuals previously diagnosed with SCZ (Lahti et al., 2001). This indicates that NMDAR antagonism is to some degree involved in SCZ pathology (Stone et al., 2007; Straub et al., 2007; Nakazawa et al., 2012; Steiner et al., 2012) and that ketamine may possibly induce disorder-specific adaptations in microglia. It is postulated that hypofunctional NMDAR located on GABAergic inhibitory interneurons disinhibit excitatory pyramidal neurons, the principal neurons of the cerebral cortex, leading to a paradoxical increase in glutamatergic activity that contributes to SCZ (Stone et al., 2007; Nakazawa et al., 2012). Based on cortical post-mortem results of asphyxiation suicide deaths, cytokines such as IL6 affecting NOX2 expression in GABAergic neurons, can contribute to oxidative stress that causes the loss of inhibitory tone on glutamatergic neurons and increases glutamatergic excitotoxicity (Schiavone et al., 2016; Supplementary Table 1). Cortical glutamate excitotoxicity is hypothesized to elevate synaptic apoptosis, a sub-lethal form of apoptosis in terminal neurites and individual synapses that triggers synaptic elimination in the absence of cell death, in line with the reduced cortical volume in individuals with SCZ without associated pyramidal cell loss (Bennett, 2011; Schobel et al., 2013; Parellada and Gassó, 2021). While in the DLPFC and HIP apoptosis and neuronal death pathways were significantly correlated with suicide death (Zeng et al., 2020; Glavan et al., 2021), further studies are required to evaluate whether these changes reflect cellular or synaptic events. It is suggested that synaptic apoptosis causes exposure of internal phosphatidylserine and accumulation of complement on the synaptic membrane, such as C1q, which stimulate microglial chemotaxis and synaptic pruning (Nonaka and Nakanishi, 2019; Parellada and Gassó, 2021). Minocycline successfully dampened the aberrant synaptic engulfment of SCZ patient-derived microglial cells in vitro (Sellgren et al., 2019) and may be used for similar outcomes in SB.