M1 microglia secrete a large number of inflammatory factors, proteases, chemokines, prostaglandins, and other toxic substances. Since multiple damage-inducing factors overlap, brain cells die in various forms such as apoptosis, necrosis, pyroptosis, ferroptosis, which leads to the irreversible destruction of brain structure (Xi et al., 2014; Zhou et al., 2014).

In brain parenchyma, M1 microglia are the major source of inflammatory mediators, such as IL-1β, IL-6, IL-12, IL-23, and TNF-α (Jiang et al., 2020). Although inflammation is essential for innate immunity, it is the chief culprit to the sustained neurological deterioration in a sterile environment (Zhu et al., 2019). While inflammatory cytokines diffuse, functional neurons and neuroglia quickly die under the stress condition (Shen et al., 2017). The diffused inflammatory cytokines also promote polarization of surrounding microglia towards the M1 phenotype, cause the inflammatory region to expand, which forms a vicious circle. In patients with ICH, the levels of IL-1β, TNF-α, and IL-6 in plasma and brain tissues are significantly increased within 1–3 days, and the increasing degree is related to 90-days poor prognosis (Jiang et al., 2020). During pathological processes, oxidative stress and inflammation mutually reinforcing, which is no exception in ICH (Hu et al., 2016; Yao et al., 2021). M1 microglia express large amounts of peroxidases, iNOS, and reduced form of nicotinamide-adenine dinucleotide phosphate (NADPH) oxidase, which produce excessive free radicals and damage surrounding cells by attacking cellular membranes and DNA (Yang et al., 2013; Duan et al., 2016; Hu et al., 2016; Xiong et al., 2016). Moreover, M1 microglia contribute to the activation of matrix metalloproteinases (MMPs), including MMP2 and MMP9, which markedly destruct the blood-brain barrier (BBB) and cause severe vasogenic brain edema by degrading extracellular matrix constituents and attacking endothelial claudin-family tight junction proteins (Montaner et al., 2019). In ICH patients, increased MMP2/9 levels were independently associated with perihematomal edema volume (Li et al., 2013). In addition, M1 microglia also release chemokines including CXCL8, CCL2, and CCL5, which diffuse into peripheral blood through the ruptured blood vessel and attract peripheral leukocytes such as neutrophils, monocytes, and lymphocytes into brain parenchyma through disrupted BBB (Trettel et al., 2020). It was reported that chemokines concentrations in plasma were proportional to the infiltration degree of peripheral immunocytes in ICH patients (Guo et al., 2020). The infiltrated immunocytes not only express and secrete inflammatory factors and aggravate inflammatory response but also release toxic substances after their apoptosis (Lambertsen et al., 2019). In ICH patients, CCL2 concentrations in plasma within 24 h were associated with poor functional outcomes at day 7 after ICH (Hammond et al., 2014). Also, inhibiting CCL2 in animal models reduced brain edema and improved neural function (Yan et al., 2020).

Noticeably, there is an evident cooperativity effect on tissue damage induced by inflammatory cytokines, protease MMPs, and chemokines. Inflammatory cytokines not only attack vascular endothelial cells and tight junction proteins but also induce endothelial cells to secrete intercellular cell adhesion molecule-1 (ICAM-1), which promotes the adhesion and infiltration of peripheral leukocytes (Aslam et al., 2012). The direct damage on neurons induced by MMPs exacerbates inflammatory response, disrupts BBB to facilitate peripheral leukocytes infiltration (Kim et al., 2005). The infiltrated peripheral leukocytes secrete inflammatory factors and MMPs, which aggravates inflammatory response and BBB destruction in turn (Tschoe et al., 2020).

Although the treatments aiming at inflammatory cytokines are currently limited in animal experiments, TNF-α antibody has shown huge therapeutic potential by significantly reducing the number of perihematomal activated microglia and improving neurological outcomes in mouse stroke models (Mayne et al., 2001; Lei B. et al., 2013; Chen A.-Q. et al., 2019). Inhibition of TNF-α not only reduces the microglial activation/macrophage recruitment via decreasing cleaved caspase-3 level (Mayne et al., 2001; Lei B. et al., 2013; Chen A.-Q. et al., 2019) but also reduces the activation of TNF receptor 1 (TNFR1) on endothelium therefore reducing endothelium necroptosis and ameliorating disruption of BBB (Mayne et al., 2001; Lei B. et al., 2013; Chen A.-Q. et al., 2019). Predictably, inhibition of specific inflammatory factors is becoming the central theme of ICH therapeutic researches.

M2 microglia primarily express anti-s and facilitate tissue regeneration (Lan et al., 2017b). Thereby, the injured brain acquires comprehensive and effective recovery. Due to the large amounts of anti-inflammatory cytokines and antioxidants, the inflammatory response and oxidative become diminished tardily (Zhu et al., 2019). More importantly, the anti-inflammatory factors promote surrounding microglia and other immune cells to transform into anti-inflammatory phenotype. It’s found that patients with higher TGF-β levels in plasma had a better prognosis at 90 days after ICH (Jiang et al., 2020).

At the same time, M2 microglia engulf the hematoma and cells debris, remove harmful substances and provide space for tissue regeneration. With the increase of the number of M2 microglia, the volume of the hematoma is eliminated promptly in 7–21 days after ICH. B-scavenger receptor CD36, one of the M2 microglial makers, is the main executive of microglial phagocytosis activity, which is obviously induced to upregulate by IL-10 (Fang et al., 2014; Yang et al., 2015; Li et al., 2021). In the mouse ICH model, CD36 knockout significantly inhibits hematoma absorption, and leads to the aggravation of neurological disorders (Fang et al., 2014). Instead, adoptive transferring CD36-positive microglia to CD36 knockout mice showed a significant improvement of neurological function after ICH (Yang et al., 2015). In fact, M2 microglia express CD163 and CD91 to absorb hemoglobin and heme, respectively (Dang et al., 2017; Garton et al., 2017). It should be noted that CD163 levels expressed by microglia may not be the only limiting factor in hematoma clearance. As a protective mechanism against severe hemolysis, the Haptoglobin (Hp) secreted by oligodendrocytes can capture free hemoglobin (Hb) to form a stable Hp-Hb complex, which is then englobed through CD163, thus reducing the toxicity of Hb. Similarly, hemopexin (Hx), secreted by neurons, binds with heme and is devoured via CD91 (Ma et al., 2016).

Particularly, M2 microglia are the drivers of brain tissue regeneration and remodeling. M2 microglia express various growth factors and trophic factors, such as insulin-like growth factors-1 (IGF-1), Brain-derived neurotrophic factor (BDNF), glial cell line-derived neurotrophic factor (GDNF), neurotrophin 3 (NT-3), NT-4/5, which could promote neurogenesis and neural circuit reframing (Xi et al., 2014; Ma et al., 2017). IGF-1 promotes the proliferation, migration, and differentiation of the neuro precursor cells in the subventricular zone, and facilitates the regenerated neurons’ functional integration into a new neural circuit (Thored et al., 2009). In a mouse ICH model, IGF-1 antibody promotes microglial M1 polarization, leading to more residual behavioral defects (Sun et al., 2020). BDNF and GDNF stimulate axon regeneration, which takes part in new neural connections (Madinier et al., 2009). The neurotrophic factors, including NT3 and NT4/5, are not only beneficial to the survival of residual neurons but also essential for the improvement and stability of the newborn neuron (Ma et al., 2017). During the remodeling of brain tissue, M2 microglia secrete clotting substance chitinase 3-like 3 (Ym-1) to prevent the degradation of extracellular matrix components (Girard et al., 2013). M2 marker Arg-1 not only converts arginine into polyamine which contributes to extracellular matrix subsidence but also competes with iNOS for reaction substrates to inhibit the excessive oxidative stress (Munder, 2009).

In general, M2 microglia resist inflammation and engulf hematomas to create a calm and stable microenvironment, which contributes to the neuro-angiogenesis and matrix deposition, and allows brain tissue to regain structure and function. Nevertheless, because of M1 microglia domination, only a third of new neurons survive inflammation in the acute phase. Therefore, promoting a beneficial microglial phenotypic transformation is a promising way in ICH treatment.

After ICH, blood carrying red blood cells (RBCs) and plasma proteins including thrombin and fibrinogen infiltrate into the brain parenchyma, and trigger the initiation of early cellular and molecular pathological processes. Hematoma not only contains the agents that directly activate microglia but also promote microglial M1-polarization indirectly through tissue damage. Figure 3 provides an overview of M1-polarization.

Because of energy exhaustion and cytotoxicity, RBCs in the hematoma begin to lyse within 1 day and continue for weeks after ICH (Righy et al., 2016). The damaged RBCs release Hb, peroxiredoxins (Prxs), and Carbonic Anhydrase-1 (CA-1), which induce microglia differentiating into M1-phenotype (Guo et al., 2012; Liu et al., 2016; Bian et al., 2020). Hb and the decomposed product hemin can directly promote microglial M1-polarization through Toll-like receptors (TLRs; Lin et al., 2012; Wang et al., 2014). Therefore, clearing hematoma is of importance in reducing brain damage. Since no reliable clinical benefits are provided from surgical hematoma removal at present, promoting hematoma devouring by microglia is of great significance.

During the formation of hematoma, thrombin and complements are produced in the brain, which are also important factors for M1-polarization. Thrombin, a serine protease that promotes blood clotting, is detected in the brain within 1 h after ICH (Zhu et al., 2019). Thrombin directly activates M1 microglia by binding to the proteinase-activated receptor-1 (PAR-1; Wan et al., 2016). In mouse models, delayed administration of thrombin inhibitor hirudin in 7–28 days after ICH significantly reduced the number of pro-inflammatory microglia (Li et al., 2019). However, thrombin regulation is difficult to apply to clinical therapy because of its two-sided effects. Though the inhibition of thrombin shows a beneficial effect in inflammation reduction, a suitable thrombin concentration is necessary for helping stop hemorrhage and protect neurons. Complements, anaphylatoxins, are activated within 24 h through various proximal cascaded pathways (Ducruet et al., 2009; Yuan et al., 2017). Complement composition C3a activates microglia cells by binding to the specific receptor C3aR. Membrane attack complex (MAC), the end product of complement cascade, attacks cell membrane, and leads to erythrocyte lysis and neuronal death, which indirectly exacerbates microglial M1-polarization. In animal models, complement inhibitor N-acetyl heparin inhibits microglia activation and ameliorates neurological deficits (Wang M. et al., 2019).

Besides, brain tissue primary damage also contributes to microglial polarization (Zhang et al., 2017). Neurons and astroglia around the hematoma express inflammatory factors such as IL-15 and IL-17, playing a vital role in M1 polarization (Yu et al., 2016; Shi et al., 2018, 2020). Likewise, damaged neurons and glia release damage-associated molecular patterns (DAMPs), including high mobility group protein-1 (HMGB1), heat shock proteins, and extracellular matrix fragments (Mracsko and Veltkamp, 2014; Bobinger et al., 2018). HMGB1 is a non-chromosome-related protein widely expressed in the nucleus of all eukaryotic cells (Mu et al., 2018). Under physiological conditions, HMGB1 helps stabilize chromosomes and regulate the transcription of many survival-based genes, but once it is dissociated from the nucleus and released outside the cell, HMGB1 becomes a powerful inflammatory mediator that promotes microglial M1 polarization by binding to TLRs on microglia (Ohnishi et al., 2011; Wang D. et al., 2017). In rodent ICH models, glycyrrhizin attenuates intracerebral hemorrhage-induced injury in a concentration-dependent manner via inhibiting HMGB1 (Ohnishi et al., 2011; Mu et al., 2018). HMGB1 inhibitor Ethyl-pyruvate significantly reduced microglia activation and inflammatory factors levels via inhibiting nuclear factor kappa B (NF-κB) DNA binding activity (Su et al., 2013).

In the later phase of ICH, an anti-inflammatory pathway, enlisting native microglia, occurs alongside neuroinflammation (Shtaya et al., 2021). Anti-inflammatory factors such as IL-4, IL-33, IL-10, TGF-β increase distinctly around the hematoma, which are mostly released by macrophages, mature lymphocytes, and mast cells (Taylor et al., 2017; Zhou et al., 2017; Chen Z. et al., 2019). The immune microenvironment changes shift microglial polarization from M1 to M2. Intraventricular injection of IL-4 in mice increases the proportion of M2 microglia and accelerates the recovery of neurological function after ICH (Yang J. et al., 2016). Some other molecular targets have also been recently identified up-regulated on microglia during M2-polarization after ICH, including Dopamine D1 receptor (DRD1), Cannabinoid receptor-2 (CB2R), Melanocortin receptor 4 (MC4R), and especially sphingosine-1-phosphate receptor (S1PR; Xu et al., 2013; Li L. et al., 2015; Zhang et al., 2015).

To recognize extracellular agents and transduce extracellular signals, microglia express various membrane receptors, nuclear receptors, and executive proteins to play roles in morphological and functional changes such as secretion, phagocytosis, and movement. Understanding microglial signal transduction is beneficial to the exploration of clinical targets. Here, we briefly introduce several important receptors and signaling molecules.

It is observed that single microglia express both M1/M2 phenotypic markers (Ransohoff, 2016; Tschoe et al., 2020). Neither M1 nor M2 should be considered as a microglial final differentiation form. The ability of microglia to switch between M1/M2 phenotypes is always a fascinating topic. However, the mechanism for this phenotypic transformation is really elusive. M1 and M2 microglia not only perform distinct cellular functions but also have incompatible polarization processes. For example, in vitro, PPAR-γ significantly inhibits the activation of NF-κB and STAT1/3 (Fang et al., 2014). In like manner, inflammatory cytokines and TLRs inhibit microglia in CD36 expression (Zhou et al., 2014; Yuan et al., 2015).

Recently, the relationship between microglia phenotype and metabolic status has attracted much attention. Microglia in different phenotypes show different oxidative metabolism (Eun Jung et al., 2020). Compared to M1 microglia, M2 microglia have significantly lower oxygen consumption (Orihuela et al., 2016). Therefore, it has been speculated that intracellular stress environment and energy crisis promote M2 polarization by influencing mitochondrial metabolism. The reactive oxygen species (ROS) released by M1 microglia has been found to activate Nrf2, which contributes to microglial M2 polarization (Duan et al., 2016; Hu et al., 2016; Qu et al., 2016). In addition, Adenosine 5‘-monophosphate activated protein kinase (AMPK), as a key molecule regulating bioenergy metabolism, is activated under cellular energy crisis and oxidative stress (Saikia and Joseph, 2021). Evidence indicates that AMPK contributes to Nrf2 activation as well (Zhao et al., 2018; Zheng et al., 2019). In other words, the initiative activation of M2 microglia may be a type of self-protection when M1 phenotype creates an immoderate oxidative stress (Barakat and Redzic, 2015). More in-depth research in the mechanism of microglial phenotypic transformation may provide insights into innovative therapeutic strategies for ICH.

In view of the serious inflammatory brain injury, whole microglia population deletion by knocking out microglial survival signal receptor colony-stimulating factor 1 receptor (CSF1R) achieved an early therapeutic effect in rodent experiments (Li et al., 2017; Shi et al., 2019). Of course, increasing the M2/M1 phenotypic proportion of microglia usually brings more satisfactory results (Wang J. et al., 2018; Bai et al., 2020), and it has become the most frequently studied therapeutic method. Relying on the aforementioned targets, experimental therapeutic studies on the precise regulation on microglia phenotype are developing rapidly, and relevant drugs are summarized in Table 3 (Hu et al., 2011; Ohnishi et al., 2013, 2019; Yang et al., 2014a,b, 2018; Iniaghe et al., 2015; Zhao et al., 2015a,b, 2018; Flores et al., 2016; Shi et al., 2016; Sukumari-Ramesh and Alleyne, 2016; Zhang et al., 2016; Anan et al., 2017; Chen-Roetling and Regan, 2017; Lan et al., 2017a; Wang J. et al., 2017; Wei et al., 2017; Xu et al., 2017; Zeng et al., 2017; Chen C. et al., 2018; Chen S. et al., 2018; Fu et al., 2018; Han et al., 2018; Li X. et al., 2018; Qiao et al., 2018; Ren et al., 2018; Wang et al., 2018b; Liang et al., 2019; Song and Zhang, 2019; Xi et al., 2019; Zhou F. et al., 2019; Cheng et al., 2020; Ding et al., 2020).

Though how to regulate microglia to promote brain recovery remains worth pondering in some sense, there are latent misgivings that excessive inhibition of M1 microglia and promotion of M2 microglia may turn into adverse effects in ICH treatment, where we should keep watchful eyes.

On the one hand, the immunoreactive materials secreted from M1 microglia appear to have delayed beneficial effects on brain repair. Solid evidence indicates that MMPs are necessary for angiogenesis, myelin remodeling, and axonal regeneration in ICH later stage (Lei et al., 2015; Fields, 2019). As well, infiltrating neutrophils and monocytes have been found conducive to hematoma clearance and inflammation regression (Lambertsen et al., 2019). Besides, the over-suppressed inflammatory status may increase brain infection risk since the systemic immunity also decreases after ICH (Saand et al., 2019).

On the other hand, the early organizational disruption may build the basis of neogenesis. M1 microglia destroy dying and defunct neurons in pieces, which lends a convenience for M2 phagocytosis (Hu et al., 2015). In addition, the deconstruction of dense tissue matrix made by M1 microglia provides space for the migration of neural precursors and synaptic remodeling (Lei C. et al., 2013). Also, M1 microglia impair BBB integrity, which is in favor of the hematoma clearance by free diffusion, especially when microglial phagocytic receptors are of inefficiencies in the ICH early phase (Righy et al., 2016, 2018).

As for M2 microglia, superfluous and prolonged existing growth factors will predictably cause abnormal tissue repair. Overexpression of Arg1 has been found to cause tissue scarring and brain dysfunction (Hesse et al., 2001), and excessive polyamines extraordinarily promote inflammatory response (Dudvarski Stankovic et al., 2016). Resting microglia plays a special role in tissue repair and remodeling (Cherry et al., 2014), and M2-M0 may be a necessary functional transformation after ICH.

In summary, it is really improper to consider that M1 and M2 microglial phenotypes are thoroughly opposite. Instead, their interaction, cooperation, and even codependency are waiting to be explored in the future. A balance of M1 and M2 microglial, rather than extremely choosing M2 over M1, ought to be achieved for ICH individualized treatment, just like the balance of yin and yang.

Although drugs with the pleiotropic ability of immune regulation may bring more benefits, not a few medical experiments failed just because of uncontrolled side effects. It is a neglected consensus that many microglial receptors and signaling molecules are meanwhile expressed or activated in other brain cells, such as astrocytes, oligodendrocytes, endothelial cells, and neurons. It is unwise to judge the holistic functions of concerned targets in the brain by taking only microglia into account. For example, CD163 helps microglia engulf and break down hemoglobin, whereas, inhibition of CD163 in the ICH acute phase unexpectedly reduces brain damage, possibly because inhibition of CD163 expressed on neurons decreases the Hb neurotoxicity induced neuronal death (Righy et al., 2018).

Hence, we need a kind of drug that has a high targeting specificity to microglia. Preferably, it’s expected to have sufficient liposoluble ability to pass through BBB and concentrate on microglia. Furthermore, it’s recommended to conjunctive use advanced medical technology such as intranasal administration, nanomaterials, and genetic technologies to achieve better intervention results for ICH treatment.