Keywords

1 Introduction

The establishment of “memory” is the cardinal and classical features of adaptive immunity and has served as the guiding principle of empirical vaccine generation for millennium. Adaptive immunity develops lasting memory responses toward highly specific antigens through somatic recombination-mediated generation of T cell receptors (TCR) and/or B cell receptors (BCR), followed by clonal expansion via interaction with selective antigen presenting cells. In contrast, innate immune cells can only respond to general molecular patterns associated with pathogens through innate receptors (Kawai and Akira 2007). Given limited repertoire of innate receptors, innate immune cells were not historically considered to be memory generating entities. However, emerging data from the last decade reveal fascinating, complex, and dynamic “memory”-like behaviors of innate immune cells that transcend beyond the classical adaptive immune memory phenotypes. The distinct features of innate memory are reflected in signal-strength and history-dependent behaviors such as priming, tolerance, and exhaustion (Li et al. 2020). The generation of innate memory may have profound consequence related to pathophysiology of both acute and chronic inflammatory diseases (Morris et al. 2014).

2 Mechanisms for the Generation of Innate Immune Memory

In the classical sense of immune memory, adaptive immune cells such as T cells and B cells gain the capability to uniquely recognize and memorize highly distinct antigens through somatic VDJ recombination. In sharp contrast, innate immune cells do not have the machinery for VDJ recombination and thus rely upon limited innately encoded receptors to recognize general molecular patterns (e.g., PAMPs – pathogen-associated molecular patterns; DAMPs – damage associated molecular patterns). Despite its limited specificity, innate immune cells can differentiate the signal strength and history of challenges, exhibiting “memory-like” behavior of priming, tolerance, and exhaustion (Geng et al. 2016; Yuan et al. 2016a; Xiong and Medvedev 2011; Foster et al. 2007; Lin et al. 2020). The establishment of such memory-like behavior is clearly distinct from the acquisition of adaptive memory and does not require genetic recombination. Instead, closely intertwined intra-cellular circuitries involving redox signaling, sub-cellular trafficking, metabolic and epigenetic processes are likely involved to establish transient memory states with limited stability and plasticity (Yuan et al. 2016b; Baker et al. 2014, 2015; Maitra et al. 2012; Chan et al. 2005; Netea et al. 2016) (Table 1).

Table 1 Key features of innate and adaptive immune memories
Full size table

A cardinal example of innate memory can be seen with monocyte/macrophage responses to rising dosages of bacterial endotoxin (Yuan et al. 2016b; Lu et al. 2015). While a prolonged challenge with higher dosages of lipopolysaccharide (LPS) can lead to reduced expression of pro-inflammatory cytokines, commonly known as endotoxin tolerance (Morris et al. 2014), prolonged stimulation with a subclinical super-low dose LPS can polarize monocyte/macrophage into a “primed” low-grade inflammatory state with sustained expression of inflammatory mediators (Yuan et al. 2016a, b). The mechanisms of endotoxin tolerance likely involve the activation and induction of molecular suppressors at multiple levels such as cytoplasmic signaling suppressors interleukin-1R-associated-kinase (IRAK)-M, and phosphatidylinositol-3-kinase and protein kinase B (PI3K/AKT) (Xiong and Medvedev 2011; Piao et al. 2009), as well as nuclear transcriptional suppressor RelB (Maitra et al. 2012; Chan et al. 2005). On the other hand, the generation of primed low-grade inflammatory monocyte/macrophage requires the clearance of suppressors such as IRAK-M and PI3K/AKT (Geng et al. 2016; Maitra et al. 2012). At the sub-cellular level, subclinical super-low dose LPS preferentially disrupts the homeostatic processes of autophagic flux as well as pexophagy, leading to the accumulation of reactive oxygen species involved in the establishment of low-grade inflammation (Yuan et al. 2016a; Geng et al. 2019). Innate leukocytes may sense the signal strength and duration of LPS via distinct usage and assembly of intra-cellular adaptor molecules such as myeloid differentiating factor 88 (MyD88) and TRIF-related adaptor molecule (TRAM), with TRAM preferentially directing the cellular response to sustained stimulation of super-low dose LPS (Yuan et al. 2016b; Rahtes and Li 2020). On the other hand, MyD88 is preferentially involved in response to higher dose LPS during both the acute response phase and the compensatory phase of tolerance (Cheng et al. 2015; Laird et al. 2009). The intra-cellular processes responsible for priming and tolerance may likely compete with each other forming multi-tiered competitive circuitries, assisting the decision-making processes of innate leukocytes in adopting dynamic activation behaviors (Morris et al. 2014; Fu et al. 2012; Kadelka et al. 2019) (Fig. 1). The generation of mutually competitive circuitries is also a fundamental principle for the clear differentiation and activation of other immune cells such as T helper cells (Hong et al. 2011, 2012).

Fig. 1
figure 1

Illustration of innate memory dynamics based on signal strength and duration. Innate immune leukocytes such as monocytes and macrophages can finely sense the strength and duration of external danger signals (e.g., lipopolysaccharide, LPS) and undergo distinct adaptations to generate dynamic memory states. In the case of LPS, a prolonged challenge with subclinical super-low dose LPS (<1 ng/mL) will induce a sustained low-grade inflammatory states due to the positive-feedback signals involving mutually activating TRAM adaptor, SRC kinases, IRF1/5/7, and STAT1. In contrast, while higher dose LPS acutely induces a transient and robust inflammatory response through the activation of NFkB, prolonged stimulation with higher LPS signals will trigger the expression of inhibitory kBs (IkBs) and RelB, leading to a tolerant state with reduced expression of inflammatory mediators such as CD86, CD40, and CCR5. Tolerant leukocytes still maintain a skewed expression of profile of selected immune suppression genes such as PD-L1, and eventually adopt an exhausted state characterized by pathogenic inflammation and immune suppression

Full size image

Sustained challenges with higher dose endotoxin lead to not only endotoxin tolerance, but also an exhausted state characterized by pathogenic inflammation and immunosuppression often seen during the progression of sepsis (Efron et al. 2018; Horiguchi et al. 2018). “Endotoxin tolerant” cells are not inert and can still robustly respond to endotoxin stimulation, with a significantly altered landscape of gene expression potentially contributing to pathogenic inflammation and immune exhaustion (Foster et al. 2007; Lu et al. 2015). For example, monocyte/macrophage with prolonged LPS stimulation exhibits robust induction of iNOS and PD-L1 (Foster et al. 2007; Lu et al. 2015). Persistent iNOS expression may contribute to pathogenic inflammation, and PD-L1 is a major contributor mediating immune suppression. Recently, we demonstrate that endotoxin exhaustion is not limited to monocyte/macrophage and can also be seen in neutrophils with prolonged challenge of higher dose LPS (Lin et al. 2020). Exhausted neutrophils with prolonged LPS treatment manifest enhanced expression of pathogenic inflammatory mediators such as LTB4 and ICAM1, contributing to altered migratory and swarming behaviors reminiscing septic neutrophils (Lin et al. 2020). Exhausted neutrophils similarly express elevated PD-L1, potentially contributing to immune suppression (Lin et al. 2020).

3 Innate Immune Memory During the Pathogenesis of Acute and Chronic Diseases

3.1 Low-Grade Inflammatory Memory Monocyte in Atherosclerosis

Atherosclerosis and related cardiovascular complications are among the leading causes of morbidity and mortality in the world (Libby et al. 2019). Previously considered as a lipid storage disease, atherosclerosis is nowadays well recognized as a chronic low-grade inflammatory disease that occurs within the arterial wall (Back et al. 2019). The programming of low-grade inflammatory monocytes is crucially involved in the pathogenesis of atherosclerosis. Non-resolving low-grade inflammatory monocytes and monocyte-derived macrophages are the key mediators for the formation and progression of atherosclerotic plaques (Jongstra-Bilen et al. 2006; Libby and Hansson 2015). Monocytes can be primed by risk factors present in the circulation and in the vessel wall, such as pathogen-associated molecular patterns, oxidized lipoproteins, shear stress, and oxidative stress. Excessive inflammatory signals tend to trigger compensatory anti-inflammatory tolerance and therefore the expression of pro-inflammatory mediators in monocytes is transient and subsequently suppressed due to the induction of homeostatic negative regulators (Nathan and Ding 2010; Biswas and Lopez-Collazo 2009; Adib-Conquy and Cavaillon 2009). In contrast, under non-resolving low-grade inflammatory conditions, monocytes may fail to develop tolerance and are programmed into a sustained inflammatory state that favors the development of atherosclerosis (Baker et al. 2014; Maitra et al. 2012; Deng et al. 2013).

LPS, also known as endotoxin, is the major stimulant to prime monocytes, which are the primary immune cells responding to LPS given their relatively high expression of TLR4. Trace amount of gut microbiota-derived LPS may leak into circulation via increased gut permeability, leading to subclinical endotoxemia (Frazier et al. 2011; Lassenius et al. 2011). According to epidemiological studies endotoxemia levels as low as 50 pg/mL may serve as a strong risk factor for the development of atherosclerosis (Stoll et al. 2004). Indeed, atherosclerosis patients have low but significantly elevated serum LPS level as compared with healthy individuals (79.0 ± 10.7 vs. 43.5 ± 11.9 pg/mL, p < 0.001). This concentration of LPS is sufficient to up-regulate Nox2 expression and elevate oxidative stress in human monocytes (Carnevale et al. 2018). In the murine model of atherosclerosis, ApoE−/− mice fed with high-fat diet exhibit significantly higher level of serum LPS as compared to the counterparts fed with regular diet. Oral administration of Akkermansia muciniphila decreases the circulating LPS level, alleviates atherosclerosis progression, as well as reduces monocyte/macrophage accumulation in the plaques (Li et al. 2016). These findings indicate that low-grade inflammatory monocytes primed by low-dose LPS are critically involved in the pathogenesis of atherosclerosis.

Chronic injection of subclinical dose LPS to high-fat diet-fed ApoE−/− mice (a murine model of atherosclerosis) significantly exacerbates the pathogenesis of atherosclerosis accompanied by higher levels of circulating Ly6CPositive low-grade inflammatory monocytes as well as increased number of macrophages within the plaque areas. The surface level of inflammatory chemotaxis receptor CCR5 is significantly elevated while the surface expression of SR-B1, a modulator for anti-inflammation and lipid metabolism, is reduced on circulating monocytes from the high-fat diet-fed ApoE−/− mice conditioned with super-low dose LPS. The monocytes that are primed with subclinical dose LPS for a long-term exhibit similar phenotype, as characterized by enhanced levels of CCR5 and reduced levels of SR-B1. Adoptive transfer of these LPS primed monocytes to high-fat diet-fed ApoE−/− mice results in significant elevation of plaque size and lipid deposition, suggesting that these low-grade inflammatory monocytes programmed by subclinical dose LPS can directly contribute to atherosclerosis progression. Mechanistically, super-low dose LPS treatment induces increased level of miR-24, which mediates the suppression of SR-B1, and reduction of IRAK-M, which is a critical negative-feedback regulator. IRAK-M deficiency in turn leads to elevated miR-24 levels, forming a positive-feedback loop sustaining the low-grade inflammatory state conducive to atherosclerosis (Geng et al. 2016). There are two competitive pathways transducing signals following LPS stimulation, namely the MyD88-dependent pathway and the MyD88-independent pathway mediated by TRIF and TRAM (Palsson-McDermott and O’Neill 2004). Intriguingly, the low-grade inflammatory monocyte primed by super-low dose LPS is dependent upon TRAM/TRIF but not MyD88 (Yuan et al. 2016b). By employing a bone marrow transplantation strategy, Lundberg et al. have shown that hematopoietic deficiency of TRAM and TRIF but not MyD88 adaptor-like (MAL) significantly reduces atherosclerosis in Ldlr−/− mice (another murine model of atherosclerosis). TRAM deficiency also leads to down-regulated level of pro-inflammatory mediators, such as TNF-α, IL-6, IL-12, CCL2, CCL5, and CXCL10, in the aorta of atherosclerotic mice (Lundberg et al. 2013). These data suggest that the priming of low-grade inflammatory monocytes by subclinical dose LPS during atherosclerosis is mainly mediated by TRAM, and targeting TRAM may promote effective generation of resolving monocytes for the prevention and treatment of atherosclerosis.

In addition to low-dose LPS, low concentrations of oxidized low-density lipoprotein (oxLDL) can also induce epigenetic reprogramming of monocytes into a pro-inflammatory state. Primary human monocytes trained with low doses of oxLDL (below 10 μg/mL) for 24 h exhibit an enhanced response to secondary stimulation 6 days later by expressing a series of pro-inflammatory mediators, including IL-6, TNFα, IL-8, MCP-1, MMP-2, and MMP-9. These trained monocytes have enhanced capacity to generate foam cells, elevated expressions of scavenger receptors (CD36 and SR-A), and reduced expression of cholesterol efflux transporters (ABCA1 and ABCG1). Therefore, these pro-inflammatory monocytes may contribute to the pathogenesis of atherosclerosis. The oxLDL-induced long-lasting proatherogenic profile can be significantly attenuated if the monocytes are pre-treated with histone methyltransferase inhibitor, suggesting that epigenetic histone modification is crucial for this innate immune memory of monocytes (Bekkering et al. 2014). It has been found that oxLDL treatment can cooperatively boost the activation of macrophages induced by low-dose LPS. Costimulation with oxLDL and low-dose LPS significantly up-regulates the genes transcribed by promoters containing an AP-1 binding site as well as induces the activation of ERK1/2. The combined effects of subclinical endotoxemia and oxLDL result in the establishment of pro-inflammatory state of macrophages and production of a series of inflammatory cytokines within atherosclerotic lesions (Wiesner et al. 2010).

3.2 Exhausted Memory Innate Leukocytes During the Pathogenesis of Sepsis

Sepsis is a systemic inflammatory response to severe infection and injury leading to multi-organ failure and remains one of the primary causes of death in hospitalized patients (Rhee et al. 2019; Perner et al. 2016). In 2017, global incidence of sepsis was around 48.9 million cases and sepsis-related deaths were estimated at 11.0 million cases (Rudd et al. 2020). The new coronavirus (SARS-CoV-2) in the ongoing outbreak and its associated disease COVID-19 pose tremendous threats to public health and drastically affect worldwide economies and societies (Kumar 2021a, b). Particularly, sepsis is the leading cause of death by COVID-19, which has been observed in nearly all deceased patients in numerous cohorts (Lopez-Collazo et al. 2020; Kumar 2020). The immune response of sepsis patients consists of a hyperinflammatory phase featured by “cytokine storm” and an immunosuppressive phase exemplified by immune cell exhaustion and dysfunction (Hotchkiss et al. 2016). Many clinical trials have been conducted to attenuate the hyperinflammatory effects by using anti-cytokine or anti-inflammatory agents, such as anti-IL-1β, anti-TNF-α, anti-LPS, and TLR inhibitors. Unfortunately, none of these approaches produces robust curative outcomes, and in some cases, the survival rate was even reduced (Brady et al. 2020; Abraham et al. 1997; Opal et al. 2013). A hallmark of sepsis is diminished clearance of primary pathogens and increased risk of secondary infection due to pathogenic inflammation and immune suppression (Efron et al. 2018). Over 70% of deaths occur after the first 3 days of sepsis, many of which occur weeks after sepsis onset (Otto et al. 2011). Thus, immunosuppression caused by leukocyte exhaustion has been increasingly recognized as a major factor for sepsis-induced mortality. A recent single cell study revealed that moribund COVID patients tend to have higher numbers of exhausted classical monocytes (Schulte-Schrepping et al. 2020).

T cell exhaustion driven by persistent exposure to infections during sepsis has been well documented in the literature. The exhausted T cells are defined by a progressive loss of T cell effector function, a state of vigilant transcription distinct from functional effector or memory T cells. A typical alteration of exhausted T cells is the overexpression of a series of inhibitory molecules, such as PD-1, CTLA-4, LAG-3, and TIM-3 (Wherry 2011). PD-1 is a critical negative regulator involved in suppressing lymphocyte responses. PD-1/PD-L1 pathway plays an important role in the initiation and promotion of immunosuppression (Liu and Li 2017). Multiple studies using mouse model of cecal ligation and puncture (CLP) have unveiled elevated PD-1 expression on splenic CD4+ and CD8+ T cells. There is a continuously increased PD-1 expression on CD4+ and CD8+ T cells with the progression of sepsis, associated with a drastic reduction of total T cell population. Similarly, sepsis patients also have significantly increased PD-1 expression on T cells in the peripheral blood, spleen as well as injured organs (Patil et al. 2017). These exhausted T cells from sepsis patients fail to efficiently produce inflammatory cytokines and their secretory profiles are potently compromised (O’Sullivan et al. 1995; Wick et al. 2000; Heidecke et al. 1999). The ploy-functionality of CD8+ cells is also significantly impaired in severe sepsis patients, and PD1 expression is inversely correlated with the number of poly-functional CD8+ T cells (Choi et al. 2017). PD-1 is considered as one of the most promising targets for immunomodulatory therapy to resume T cell function. However, anti-PD-1 treatment alone does not yield expected outcomes because multiple negative costimulatory molecules are expressed on the surface of exhausted T cells. For example, a recent study demonstrates that T cells co-expressing LAG3 and PD-1 are more significantly exhausted as compared to LAG3 or PD-1 single positive T cells in patients with acute sepsis. Furthermore, the frequency of co-expressing T cells is positively associated with the mortality and the length of hospital stay (Niu et al. 2019). Thus, therapies targeting these suppressor molecules may maximize the recovery of T cells.

Correspondingly, monocytes in sepsis patients tend to express higher levels of immune receptors including CD63, CD163, CD206, TLR2, and TLR4, presumably rending them with elevated responses to infections (Armstrong et al. 2004; Hirsh et al. 2001). However, studies reveal that monocytes from sepsis patients are less responsive than those from healthy individuals. They cannot efficiently produce TNF-α, IL-1β, and IL-8 when challenged with LPS ex vivo, and the reduced TNF-α production by monocytes is employed as an index to evaluate the immune suppression of patients with sepsis (Ryan et al. 2017). The diminished capacity to produce pro-inflammatory cytokines may be due to the elevated expression of IRAK-M, an inhibitory Toll receptor signaling molecule, in the monocytes from sepsis patients. The patients with higher IRAK-M levels on admission have a higher mortality rate (Wiersinga et al. 2009). Monocytes are specialized antigen presenting cells (APCs) that present surface MHC molecule-bound antigens to activate T cells. The exhaustion of these APCs potentially facilitates the immunosuppression during sepsis. Sepsis induces altered monocyte–T cell interactions because of reduced expressions of co-stimulatory molecules on monocytes. Indeed, monocytes in sepsis patients are found to express much lower levels of CD40, CD80, and CD86 (Sugimoto et al. 2003; Sinistro et al. 2008; Lissauer et al. 2009). On the contrary, PD-L1 surface expression is up-regulated on monocytes from septic mice models as well as sepsis patients, correlated with T cell exhaustion and immunosuppression via PD-1/PD-L1 signaling pathway (Patil et al. 2017). Monocyte PD-L1 expression can be used as an independent predictor of 28-day mortality in patients with septic shock (Shao et al. 2016).

Neutrophils are the most abundant leukocytes in the circulation and play a crucial role in sepsis as the first line of defense in protecting the body from microbial invasion. The interaction between neutrophils and other immune cells is necessary for the resolution of excessive inflammation as well as effective host defense (Serhan and Savill 2005). Exhausted neutrophils with aberrant immune responses to infection have been observed in septic animal models and patients. Excessive bacterial products and pro-inflammation cytokines in sepsis induce the loss of CD62L expression but elevated integrin expression (e.g., CD11b) on the surface of neutrophils (Rosenbloom et al. 1999; Kovach and Standiford 2012). In addition, CXCR2 expression is reduced in the neutrophils of mice and patients with severe sepsis. Prolonged neutrophil stimulation results in up-regulation of iNOS and GRK2, which further promotes CXCR2 internalization (Paula-Neto et al. 2011). Therefore, neutrophils exhibit reduced rolling and migratory capacity and fail to be recruited to the primary infection foci. Instead, these exhausted neutrophils tend to form pathogenic aggregations in the vital organs, which can be mimicked through in vitro examination (Lin et al. 2020). The antimicrobial activities of neutrophils are also compromised during sepsis. Excessive bacterial load activates complement system, and high levels of C5a suppress the phagocytic function and ROS production of neutrophils. Various studies with septic mouse models and sepsis patients have revealed impaired phagocytosis, oxidant production as well as oxidative burst capacity of septic neutrophils (Shen et al. 2017; Bhan et al. 2016). Furthermore, neutrophils from septic mice and patients can induce immunosuppression and T cell apoptosis in a cell-contact dependent manner via the surface expression of PD-L1. Thus, the PD-L1 level on neutrophils, which is positively associated with sepsis severity, may serve as a biomarker for the prognosis of septic patients (Wang et al. 2015). Despite its clinical significance, the mechanisms of neutrophil exhaustion during sepsis are still poorly understood. We have recently reported that neutrophils treated with LPS in vitro for a prolonged period develop a phenotype of elevated ICAM1, CD11b, and PD-L1 expression as well as enhanced swarming and aggregation, which resembles the exhausted neutrophil phenotypes seen in sepsis. Importantly, the exhaustive profiles are significantly alleviated in TRAM deficient neutrophils after prolonged LPS challenge as compared with wild-type neutrophils. TRAM mediated neutrophil exhaustion may be dependent upon Src family kinases (SFK) and STAT1 activation, since SFK inhibitor can effectively block neutrophil exhaustion caused by prolonged LPS treatment. Furthermore, TRAM deficiency is protective against the development of severe systemic inflammation and multi-organ damage in mice (Lin et al. 2020). These data unveil a critical function of TRAM in promoting neutrophil exhaustion.

3.3 Innate Immune Memory During the Pathogenesis of Cancer

The phenotype and functionalities of myeloid cells (e.g., monocytes, macrophages, and neutrophils) are substantially changed by tumor-induced systemic environment and microenvironment, so that these cells usually acquire pro-tumor functions to promote cancer progression. On the other hand, the tumor-associated innate immune cells may provide ideal targets for fighting against cancer.

The number of circulating monocytes significantly increases in both humans and mice bearing tumors. Among several cancer types, patients with high blood monocyte counts have a poorer disease prognosis, and the ratio of lymphocytes to monocytes has become a prognostic factor for lung cancer, colorectal cancer, and ovarian cancer (Kiss et al. 2020; Olingy et al. 2019). The increased monocyte levels may be caused by two reasons: enhanced migration from bone marrow to circulation, and increased myelopoiesis. Patients with pancreatic cancer exhibit elevated circulating monocyte levels associated with decreased monocyte abundance in the bone marrow. CCL2, a critical chemokine for monocyte recruitment, is commonly present at higher levels in serum of both mice and humans with cancer, facilitating the egress of monocytes from the bone marrow (Kishimoto et al. 2019). Cancer is usually accompanied by elevated serum levels of cytokines that are involved in the myeloid cell differentiation and survival, such as macrophage-colony stimulating factor (M-CSF), granulocyte-colony stimulating factor (G-CSF), and granulocyte-macrophage-colony stimulating factor (GM-CSF) (Scholl et al. 1996; Ribechini et al. 2017; Katsumata et al. 1996). Excessive production of these cytokines and tumor-associated low-grade inflammation promote reprogramming of myelopoiesis. As a result, hematopoietic stem cell (HSC), common myeloid progenitor (CMP), and granulocyte-monocyte progenitor (GMP) populations are expanded, while common lymphoid progenitor (CLP) and megakaryocyte-erythrocyte progenitor (MEP) are not significantly altered. For example, increased frequency of HSC and GMP populations is observed in peripheral blood of patients with various types of solid tumors, indicating that tumor-associated environment favors myeloid hematopoiesis and expansion of circulating monocytes (Wu et al. 2014; Manz and Boettcher 2014; Casbon et al. 2015; Strauss et al. 2020). In addition to increased numbers, the phenotype of monocytes is also profoundly influenced by tumors. One of the well-documented features of cancer-educated monocytes is the acquisition of immunosuppressive properties. The monocytes from healthy individuals express high level of HLA-DR (the protein of MHC II) on the surface, while HLA-DR level is significantly down-regulated in the monocytes of cancer patients (Luczynski et al. 2004; Ugurel et al. 2004). The high level of CD14 HLA-DRlow monocytes is correlated with the lower levels of tumor-specific T-cells in the circulation of cancer patients. The patients with lower levels of CD14+ HLA-DRlow monocytes are more responsive to immune checkpoint blockade therapy (Weide et al. 2014; Weber et al. 2016). The surface expression of CD86, a co-stimulatory molecule for T cell activation, is also reduced on cancer-educated monocytes, inhibiting T cell function (Luczynski et al. 2004; Ugurel et al. 2004). Furthermore, monocytes in cancer patients have enhanced expression and activity of arginase-1, which limits the availability of L-arginine to T cells (Trovato et al. 2019; Hoechst et al. 2008). Up-regulation of PD-L1 and GPNMB may also contribute to the immunosuppressive activity of monocytes in cancer (Kobayashi et al. 2019). Intriguingly, monocytes from cancer patients exhibited increased level of phosphorylated STAT3, and STAT3 is potentially activated in the healthy monocytes after co-culture with cancer cells. Inhibition of STAT3 attenuates the immunosuppressive activity of cancer-educated monocytes (Trovato et al. 2019; Poschke et al. 2010). Therefore, STAT3 may be a crucial transcription factor for the reprogramming of monocytes by tumor-specific environment.

Based on the pro-tumor characteristics of monocytes, a series of regimens have been developed to reverse monocyte reprogramming. Specific antibody against CCR2 or small molecules that block CCR2 signaling remarkably restrain the growth and metastasis of tumor cells in mouse models of lung, breast, prostate, pancreatic, and liver cancers (Olingy et al. 2019). A recent clinical study has revealed that oral administration of PF-04136309, a small molecule CCR2 antagonist, effectively reduces circulating monocytes and tumor-associated macrophages (TAMs) in pancreatic cancer (Nywening et al. 2016). Angiotensin II serves as a key regulator of cancer-induced myelopoiesis, and angiotensin-converting enzyme (ACE) inhibitors may suppress the excessive generation of pro-tumor monocytes. Enalapril, an ACE inhibitor, has been shown to reduce monopoiesis and TAMs as well as prolong the survival of mice with lung tumors (Cortez-Retamozo et al. 2013). Neutralization of tumor-derived GM-CSF reduces the emergence of CD11b+ Gr-1+ myeloid cells, leading to elevated anti-tumor activity of T cells and restrained tumor growth (Bayne et al. 2012). Arginase inhibition may also be an alternative approach to mitigate monocyte-mediated arginine depletion and consequent immunosuppression. Accordingly, a small molecule arginase inhibitor has been developed, which can increase plasma and tumor arginine levels, enhance anti-tumor T cell responses, prime immunity toward a pro-inflammatory state, and reduce tumor growth in mouse cancer models (Steggerda et al. 2017). Application of STAT3 inhibitors has also yielded promising anti-tumor outcomes by boosting anti-tumor immunity in mice. TAM receptor tyrosine kinases promote STAT3 phosphorylation, and administration of UNC4241, an inhibitor against TAM receptor tyrosine kinases, significantly alleviates the immunosuppressive activity of monocytes in a mouse model of melanoma (Holtzhausen et al. 2019).

Neutrophil is one of the major components of tumor-infiltrating innate immune cells. Cancer patients with poor prognosis often have an expanded pool of tumor-associated neutrophils (TANs), which exhibit complex and contradictory functions, promoting or limiting tumor growth (Powell and Huttenlocher 2016). The pro-tumor neutrophils produce high levels of VEGF, MMP9, and HGF, which facilitate tumor angiogenesis. MMP9 production and NET formation from neutrophils may provide ideal environment and niche for tumor cell intravasation and metastasis. Similar as cancer-educated monocytes, some TANs also possess immunosuppressive properties and suppress T cell proliferation via deprivation of L-arginine (Granot 2019). Compared with circulating and splenic neutrophils, TANs secrete higher amount of CCL7 (a Tregs chemoattractant) that promotes the recruitment of Tregs to tumor, thereby forming an immunosuppressive microenvironment (Fridlender and Albelda 2012). On the other hand, some neutrophils also exert anti-tumor activities by secreting cytotoxic mediators (ROS) to induce tumor cell apoptosis (Granot 2019). Intriguingly, accumulating data have suggested that the interaction between neutrophils and T cells is indispensable for the proper anti-tumor response of adaptive immunity (Eruslanov et al. 2014; Stoppacciaro et al. 1993). A subset of TANs has been found to exhibit both neutrophil and APC characteristics in the patients with lung cancer, and these cells can boost anti-tumor T cell responses (Singhal et al. 2016). The molecular mechanisms underlying the differential function of neutrophils are not well understood. A recent study has indicated that Tollip, an innate immunity signaling adaptor molecule, may contribute to the neutrophil reprogramming in a mouse model of colorectal cancer. Tollip-deficient neutrophils have STAT5-dependent elevation of CD80 and reduction of PD-L1 as compared to wild-type counterparts. Therefore, Tollip-deficient neutrophils may potently activate T cells to exert anti-tumor activity. Adoptive transfer of Tollip-deficient neutrophils but not Tollip-deficient monocytes promotes tumor immune surveillance and reduces colorectal cancer burden in vivo. Thus, Tollip may serve as a target to modulate the decision-making process of neutrophils for future cancer therapy (Zhang et al. 2019).

In recent years, the concept of trained immunity has drawn increasing attention due to its potential for the treatment of diseases such as cancer. Trained immunity represents an epigenetic and metabolic reprogramming process of innate immune cells to acquire long-term function and memory property (Netea et al. 2017). Instillation of Bacillus Calmette-Guérin (BCG) can prime monocytes from the patients with bladder cancer into a pro-inflammatory and anti-cancer profile, mediated by the autophagy genes ATG2B and ATG5 (Buffen et al. 2014). The patients who exhibit low responsiveness to trained immunity have a higher chance of recurrence and tumor progression (Netea et al. 2016). In LPS primed macrophages, 70% of the inducible genes are the targets of Jumonji domain containing 3 (JMJD3), also known as lysine demethylase 6B (KDM6B) (De Santa et al. 2009). KDM6B can stabilize tumor suppressor gene p53 (Ene et al. 2012), and higher KDM6B expression is a prognostic indicator for better survival in neuroblastoma patients (Yang et al. 2019). Thus, LPS-trained macrophages via KDM6B may have clinical potential for cancer therapy. Fungal-derived polysaccharide β-glucans have been used to treat various cancers for a long time, and they are also potential inducers to promote trained immunity. A series of studies have revealed that β-glucan treatment leads to marked phenotypical and functional alterations of monocytes/macrophages, such as elevated cytokine production, enhanced phagocytic capacity, and increased ROS generation (Netea et al. 2017; Lerias et al. 2019). A recent study has shown that training with β-glucan leads to a transcriptomic and epigenetic rewiring of granulopoiesis in mice with melanoma, which subsequently induces the anti-tumor phenotype of TANs. Importantly, the mice receiving a single injection of β-glucan still exhibit a significant inhibition of tumor growth after 28 days, indicating long-term anti-tumor effects of neutrophil-mediated trained Immunity (Kalafati et al. 2020).

4 Concluding Remarks

Despite these exciting advances, the field of innate immune memory has only been revealed as the very tip of an iceberg closely intertwined with almost all aspects of human pathophysiology. Innate leukocyte may adopt highly diverse disease and context-dependent memory states, requiring single cell approaches to further clarify their unique contribution to distinct pathogenesis of inflammatory diseases. Memory leukocytes may further propagate their phenotypes to neighboring cells, establishing unique memory niche within local environments (Ballesteros et al. 2020). Future efforts are needed to finely map the establishment as well as propagation of innate memory through genetic and chemical approaches in tracking the ontogeny and propagation of memory innate leukocytes, in response to varying degrees of signal strengths and intensities within complex host immune environments. Harnessing the potential of reprograming innate memory would hold enormous promise in the treatment of both acute and chronic human diseases.