Heterogeneous tissue-specific macrophages orchestrate metastatic organotropism of breast cancer: implications for promising therapeutics

Compelling evidences have manifested that breast cancer cells prefer to metastasize to certain distant organs, including brain, lung, bone and liver. According to the canonical “seed and soil” theory, this prominent biological behavior, termed as metastatic organotropism, involves intricate interactions between breast cancer cells (the “seeds”) and specific residents in the tumor microenvironment (the “soil”), initiating from pre-metastatic niche formation to metastatic outgrowth. Recently, multifaceted heterogeneity of tissue-specific macrophages (TSMs) and their roles played in organotropic metastases of breast cancer are incrementally unveiled. Herein, we decipher multiple diversities of TSMs, including evolvement, profiles, functions and metabolic characteristics under different polarization states. Further, we elaborate on bidirectional effects of TSMs on metastatic organotropism of breast cancer (both to the “seeds” and “soil”), and unearth underlying signaling pathways based on updated mechanistic researches. Lastly, we compile a series of clinical trials, hoping to illuminate promising TSM-targeting therapies against breast cancer organotropic metastases.

Graphical Abstract

Breast cancer (BC) is an intricate aggregation of primary and disseminated malignancies with prevailing disease burden and remarkable biological heterogeneity [1]. According to the expressing status of hormone receptors (HR, including estrogen receptor [ER] and progesterone receptor [PR]) and human epidermal growth factor receptor 2 (HER2), BC can be further stratified into three major molecular subtypes, namely luminal-like, HER2-positive and triple-negative breast cancer (TNBC) [2]. Of note, metastatic breast cancer (MBC) cells are prone to spread preferentially to distant organs, including brain, lung, bone and liver. To be specific, the luminal-like subtype shows a metastatic propensity for bone, HER2-positive tends to metastasize to lung, liver and bone, whereas brain is especially favored for TNBC [3]. This intriguing metastatic inclination towards certain destined organs is termed as metastatic organotropism, a distinctive subtype-related feature of MBC with decisive prognostic and therapeutic values [4]. Therefore, it is imperative to dissect the underlying mechanisms and dig out effective targeted modalities.

The lately proposed concept of tissue-specific macrophages (TSMs), also termed as tissue-resident macrophages or resident tissue macrophages, refers to heterogeneous clusters of macrophages that persistently reside in specific organs or tissues and exert specified biological functions, such as microglia in the central nervous system (CNS), alveolar macrophages (AMs) and interstitial macrophages (IMs) in lung, osteoclasts in bone, and Kupffer cells (KCs) in liver. As an integral component of the innate immune system, the pivotal pathophysiological roles of TSMs are appealing to worldwide attention. Accumulating evidences have demonstrated that TSMs not only eliminate aberrant cells and invading pathogens, regulate cell growth and tissue remodeling, but also exert profound impacts on nutrition metabolism and energy utilization, collectively highlighting their essential roles in homeostasis and pathogenesis [5, 6].

Currently, the complicated crosstalk between MBC cells and the tumor microenvironment (TME) stands as a cutting-edge frontier, especially for TNBC with the utmost aggressiveness and worst prognosis [7]. Thanks to contemporary techniques like single-cell RNA sequencing (scRNA-seq), the cellular and molecular heterogeneity of MBC is getting incrementally untangled [8]. To date, several characters propelling MBC organotropism within the TME have been elegantly reviewed. Among them, Chin et al., Wong et al. and Huang et al. focused on exosomes secreted by MBC cells [9,10,11], whereas Wang et al. [12] and Yu et al. [13] discussed about the roles of breast cancer stem cells and circulating tumor cells respectively. As for macrophages, abnormally activated tumor-associated macrophages (TAMs) have been associated with immunosuppression and therapeutic resistance in MBC [14]. How TSMs affect metastatic organotropism of BC, however, has not yet been systematically delineated by now as far as we know. Considering the overwhelming intricacy of MBC, the theoretical landscape demands to be constantly replenished.

In this review, we enumerate multifaceted heterogeneity of TSMs from various angles. Further, based on the prestigious “seed and soil” hypothesis [15], we strive to figure out respective implications of TSMs on brain, lung, bone and liver metastases of BC. Finally, we compile an array of discerned signaling pathways and feasible regimens based on completed and ongoing clinical trials (graphical abstract). It is believed that the enriching knowledge of TSMs could provide enlightening insights into their interactions with both MBC cells (the “seeds”) and milieus of the four destined organs (the “soil”), thus shedding light on the development of promising therapeutic strategies against MBC in the coming future.

TSMs corporately constitute an enormous family with substantial molecular heterogeneity. Deciphering these diversities from various angles will pave the way for us to better understand their roles played in metastatic organotropism of BC. In this part, we will frame out the multifaceted heterogeneity of TSMs residing in CNS, lung, bone and liver, discussing about their ontogeny and development, polarization settings, molecular profiles, specified functions and metabolic features grounded on acquired evidences (Fig. 1).

An overview of heterogeneous TSMs residing in the CNS, lung, bone and liver in a patient with MBC. TSMs inside metastatic organs of an MBC patient are represented by alveolar and intestinal macrophages in lung, Kupffer cells, liver capsular macrophages, central vein macrophages and lipid-associated macrophages in liver, microglia and CNS-associated macrophages in the CNS, osteoclasts, osteomacs and EBI macrophages in bone. They are characterized by multifaceted heterogeneity in terms of ontogeny and development, polarization settings, molecular profiles, specified functions and metabolic features. CNS: central nervous system; EBI: erythroblastic island; MBC: metastatic breast cancer; TSMs: tissue-specific macrophages

Full size image

Ontogeny and development

Amid all identified TSMs, the best-researched ones should be entitled to macrophages in the CNS, including the major microglia population in the brain parenchyma, and CNS-associated macrophages (CAMs) in the interfaces. Microglia originate from yolk sac-derived erythro-myeloid progenitors (EMPs) and pre-macrophages with the ability to renew themselves independently [16], and were found to mature sequentially from early, preliminary to adult states in a stepwise order under the regulation of immune response pathways [17]. CAMs refer to macrophages situated in the dura mater, subdural meninges, choroid plexus and perivascular sites, harboring remarkable distinctions in origination, genetic characteristics and many other aspects as compared to microglia [18]. For instance, meningeal macrophages share an identical prenatal precursor with microglia, whereas perivascular macrophages were discovered to arise from perinatal meningeal macrophages dependent on integrin merely in the postnatal stage [19]. Besides, it was shown that TGF-β is indispensable to the development of microglia, but dispensable to the generation of CAMs [20]. Under pathological circumstances such as traumatic spinal cord injury and neuroinflammation, meningeal-resided macrophages can get replenished by surrounding cranial and vertebral bone marrow instead of circulating monocytes [21].

In pulmonary niches, AMs and IMs function as local guardians of the airway and vasculature. It was manifested that AMs of both mice and humans are stemmed from circulating fetal monocytes (CD115/116 + , CD64–) in the embryonic period, and get self-renewed during lifetime by locally proliferating at a low level [22]. In the absence of inflammatory signals, IMs are derived from the yolk sac as well. They either replenish themselves without the aid of circulating monocytes, or be displaced by monocytes in diverse proportions and speeds depending on the concrete subtype [23, 24]. Of note, the defending troop can be augmented, when CD115 + macrophages from CCR2 + Ly6C^hi monocyte precursors are recruited into the battle field in response to infection and inflammation [23].

Three populations of macrophages have been found to specifically reside in bone, including osteoclasts, osteomacs and erythroblastic island (EBI) macrophages [6]. Although their unique functions have already been clearly illustrated (to be discussed later), their origins remain relatively elusive. For osteoclasts colonizing embryonic ossification centers, it has been revealed that they belong to the lineage of EMP-derived macrophage populations, and get developed after fusing with haematopoietic stem cell-derived monocytic progenitors mediated by CSF1 and RANKL. During the postnatal phase, the combination between circulating monocytes and persisting osteoclast syncytia takes place repetitively in order to sustain the osteoclast population [25]. The ontogeny of osteomacs and EBI macrophages, nevertheless, is still poorly understood, awaiting more elaborately designed fate-mapping models and systematic histological analyses.

In addition to the well-characterized KCs, liver capsular macrophages (LCMs), central vein macrophages (CVMs) and lipid-associated macrophages (LAMs) are also involved in the entity of liver-specific macrophages [6]. KCs are widely accepted to arise from fetal EMPs rather than circulating monocytes in the stable status with a lifelong self-sustainable capability [26]. Accounting for no more than 10% amid the whole entity, the ontogeny of LCMs remains controversial and perplexing. It was once deemed that the majority of LCMs are replenished by monocytes in an adult murine model [27], but later it was concluded that only approximately 30% of LCMs originate from monocytes in another fate-mapping research [28]. LAMs are situated in the bile ducts, and are associated with regional lipid exposure as was revealed by spatial proteogenomic datasets [29], but the concrete developmental trajectory of LAMs as well as that of CVMs needs to be further explored.

Profiles and functions in different polarization states

Once induced by exogenous stimuli, macrophages tend to exhibit a spectrum of polarization states with disparate phenotypes and functions, in which the M1/M2 dichotomy is the most frequently adopted nomenclature despite emerging limitations [30]. Generally speaking, the classically activated or M1-polarized macrophages are stimulated by IFN-γ, TLR agonists and/or others with pro-inflammatory and anti-tumor properties. In contrast, the alternatively activated or M2-polarized macrophages are stimulated by IL-4 and/or others with anti-inflammatory and pro-tumor properties, which can be further divided into subsets including M2a, M2b, M2c and M2d [31]. Provoked by numerous activators, typical TSMs under different polarization states display different biomarkers, release multiple cytokines and exert distinctive biological effects (Table 1).

Metabolic characteristics in different polarization states

Macrophages are apt to exhibit distinctive metabolic characteristics in different polarization settings. To be concrete, there are both similarities and differences in terms of cellular metabolism among TSM populations. It is mostly applicable that oxidative phosphorylation (OXPHOS) is notably elevated in the steady state, whereas glycolysis turns predominant upon stimulation. Taking a step further, microglia are known for their impressive metabolic plasticity, which means they are capable of adapting to the environment by flexibly altering the major energy supplier between glucose and glutamine in a mTOR-mediated manner [32]. As to pulmonary macrophages, since AMs are mainly in charge of clearing surfactant, devouring inhaled particles and executing immune surveillance, they are endowed with intensified OXPHOS, lipid degradation and cholesterol disposal with weakened glycolysis, while IMs tend to demonstrate a totally opposite metabolic pattern when infected by mycobacterium tuberculosis [33]. Osteoclasts possess more mitochondria in quantity, size and intricacy than their counterparts, which is in line with their outstanding speciality in dissolving collagen and driving bone mineralization. In addition to a higher level of OXPHOS, enhanced fatty acid oxidation and glutamine catabolism are observed in tranquil osteoclasts as well. It has also been verified that boosted glycolysis, lactate accumulation and HIF-1α are associated with the osteolytic activity after osteoclasts are exposed to bone powders [34]. In KCs, iron metabolism (driven by Spi-C, NRF-2 and HO-1) and lipid metabolism (driven by PPARγ and LXRα) are remarkably upregulated, with more metabolism-related genes expressed than LCMs [33].

Setting out from the primary site, MBC cells are about to go through a series of events before constructing sizable lesions in destined organs, including pre-metastatic niche (PMN) formation [35], epithelial-to-mesenchymal transition, invasion, intravasation, circulating dissemination, adhesion, extravasation, dormancy, reactivation, proliferation and colonization [36]. In concordance with the prestigious “seed and soil” hypothesis that metastasis is manipulated by coordination between cancer cells and their destinations, enriching contemporary studies have verified that this multistep process is indeed delicately orchestrated by various parenchymal and stromal cells along the journey to create a favorable niche for tumor growth and immune escape [15]. In this section, we will introduce epidemiological and clinical highlights of BC organotropic metastases in turn, and then summarize reciprocal interactions between typical TSMs and both MBC cells (the “seeds”) and niches of involved organs (the “soil”) from brain and lung to bone and liver (Fig. 2), with discerned signaling pathways that regulate each metastatic scenario (Fig. 3 & Fig. 4).

Implications of diverse TSMs on BC metastatic organotropism. Incremental studies have uncovered intricate interactions between TSMs and residents in the tumor microenvironment, which play pivotal tuning roles in organotropic metastases of breast cancer. (a) interactions between microglia and BCBrM; (b) interactions between AMs and BCLuM; (c) interactions between osteoclasts and BCBoM; (d) interactions between KCs and BCLiM. Note: The “dysfunction” arrow directing from T cells to MBC cells in this figure means the loss of T cells’ normal tumoricidal effect on MBC cells. AB: apoptotic body; AM: alveolar macrophage; AP: antigen presentation; BBB: blood–brain barrier; BC: breast cancer; BoM: bone metastasis; BrM: brain metastasis; C5aR: complement C5a receptor; Cav-1: Caveolin-1; CCL: C–C chemokine ligand; CTC: circulating tumor cell; EVP: extracellular vesicle and particle; GF: growth factor; IFN: interferon; IL: interleukin; KC: Kupffer cell; LDC: lung dentritic cell; LiM: liver metastasis; LuM: lung metastasis; MBC: metastatic breast cancer; MHC: major histocompatibility complex; M-MDSC: monocytic myeloid-derived suppressor cell; MMP: matrix metalloproteinase; Mo: monocyte; Mreg: regulatory macrophage; Neu: neutrophil; NK: natural killer; OPG: osteoprotegerin; PGE2: prostaglandin E2; PMN: pre-metastatic niche; PPAR: peroxisome proliferator activated receptor; PTHrP: parathyroid hormone-related peptide; RANK: receptor activator of NF-κB; RANKL: RANK ligand; ROS: reactive oxygen species; TGF: transforming growth factor; TNF: tumor necrosis factor; TREM2: triggering receptor expressed on myeloid cells 2; TSM: tissue-specific macrophage; XIST: X-inactive-specific transcript

Full size image

TSM-related signaling pathways in BCBrM (upper), BCBoM (right) and BCLiM (left). In general, metastasis-promoting factors are put in red boxes, intermediate factors are put in yellow boxes, while metastasis-inhibitory factors are put in blue boxes. Arrow represents promoting effects, while vertical line represents inhibitory effects. The circled P represents phosphorylation of the substrate. ANXA1: annexin-A1; BC: breast cancer; BMP: bone morphogenetic protein; BoM: bone metastasis; BrM: brain metastasis; CCL/R: CC-motif chemokine ligand/receptor; CSF1R/2Rb: colony-stimulating factor 1 receptor/2 receptor b; CX3CL/R1: C-X3-C motif chemokine ligand/receptor 1; CXCR4: CXC chemokine receptor 4; ERMAP: erythroid membrane-associated protein; HA: meso-Hannokinol; JNK: c-jun N-terminal kinase; KC: Kupffer cell; LiM: liver metastasis; MCL: macrophage C-type lectin: also known as dectin-3; MMP: matrix metalloproteinases; MSP: macrophage-stimulating protein; MST-1(R): macrophage-stimulating 1 (receptor); OPN: osteopontin; PARP: poly(ADP-ribose) polymerase; PI3K: phosphoinositide 3-kinase; ROS: reactive oxygen species; RUNX2: Runt-related transcription factor 2; SDF1: Stromal cell-derived factor 1; sGRP78: secretory glucose-regulated protein 78; SIRPA: signal regulatory protein α; STAT5: signal transducer and activator of transcription 5; TLRs: Toll-like receptors; TNBC: triple-negative breast cancer; TSM: tissue-specific macrophage; VCAM1: vascular cell adhesion molecule 1; YTHDF2: YT521-B homology domain family member 2; ZEB: Zinc finger E-box binding homeobox 1

Full size image

AM-related signaling pathways in BCLuM. They are categorized as M2 polarization (upper), signal transduction (left), migration & infiltration (lower) and secretion (right). In general, metastasis-promoting factors are put in red boxes, intermediate factors are put in yellow boxes, while metastasis-inhibitory factors are put in blue boxes. Arrow represents promoting effects, while vertical line represents inhibitory effects. A3: Anemoside A3; A-FABP: adipocyte/macrophage fatty acid-binding protein; AM: alveolar macrophage; ANAX1: annexin-A1; BC: breast cancer; BPIFB1: bactericidal/permeability-increasing-fold-containing family B member 1; C5aR: complement C5a receptor; CCL/R: CC-motif chemokine ligand/receptor; CECR2: cat eye syndrome chromosome region candidate 2; CHI3L1: chitinase 3-like protein 1; E-Cig: e-cigarette; ECM: extracellular matrix; ERK: extracellular signal-regulated kinase; EV: extracellular vesicle; FPR2: formyl peptide receptor 2; FSTL1: follistatin-like protein 1; GA: glycyrrhetinic acid; G-CSF: granulocyte colony-stimulating factor; GGT1: γ-glutamyltransferase 1; GM-CSF: granulocyte–macrophage colony-stimulating factor; Gpr132: G protein-coupled receptor 132; HLF: hepatic leukemia factor; ICJ: immunological activity of Chamaejasmin B; IL: interleukin; IRF-8: interferon regulatory factor-8; ISGs: interferon-stimulated genes; JNK1/2: c-jun N-terminal kinase 1/2; KLHL21: Kelch-like protein 21; LuM: lung metastasis; MAPK: mitogen-activated protein kinase; MCP: modified citrus pectin; MIF: macrophage migration inhibitory factor; MMP: matrix metalloproteinase; mTOR: mammalian target of rapamycin; NF-κB: nuclear factor κB; Orp: glucan extract of Oudemansiella raphanipes; RKIP: Raf kinase inhibitor protein; RYP: Ruyiping; SIPA1: signal-induced proliferation-associated 1; STAT: signal transducer and activator of transcription; TGF-β: transforming growth factor β; TNF-α: tumor necrosis factor α

Full size image

CNS-specific macrophages and breast cancer brain metastasis (BCBrM)

BCBrM takes place in around 30% of the whole MBC population (particularly frequent in HER2-positive and TNBC subtypes), causing significant impairment of life span and quality of life with various neurological sequelae [37]. Recently, a German multi-centered registry analysis concluded that several factors may predict better prognosis and eligibility for strengthened treatment in BCBrM population, including younger diagnostic age, HER2-positive tumors, better physical condition, fewer cranial and visceral metastatic lesions [38]. BCBrM poses a remarkable therapeutic challenge for oncologists due to the presence of blood–brain barrier (BBB), the natural obstacle that restricts most drugs and immune cells from entry, thus providing a permissive sanctuary for MBC cells to develop rampantly in the CNS [39]. Although combinations of locoregional and/or systemic agents could reverse the dilemma to some extent in several exploratory clinical trials (e.g. tucatinib plus trastuzumab and capecitabine for HER2-positive BCBrM patients in the phase 2 HER2CLIMB trial [40], radiotherapy plus pyrotinib and capecitabine for HER2-positive BCBrM patients in a Chinese phase 2 trial [41], trastuzumab deruxtecan for HER2-positive BCBrM patients in the phase 2 TUXEDO-1 trial [42]), the long-term survival is still limited, calling for more effective tactics to treat BCBrM.

The intercellular interactions between MBC cells, microglia and other local cells like T cells are demonstrated in Fig. 2a. For circulating MBC cells, the first and foremost step to form an intracranial metastatic site is to traverse the separative BBB by undermining its integrity, which is normally comprised of vascular endothelia, pericytes and astrocyte endfeet [43]. By upregulating the expression of COX-2, HBEGF, ST6GALNAC, GM-CSF, CX3CL1, CXCL12, CXCR4 and other markers, MBC cells can fire up neuroinflammation, resulting in higher permeability of the BBB and recruitment of immune cells including microglia [39, 44]. In response, after sensing these proliferative and migratory signals, microglia would answer back by releasing cytokines like TNF-α to stimulate endothelial cells, further improving BBB penetrability and infiltration of surrounding immune cells [39].

After MBC cells survive and colonize the CNS, it is about time that angiogenesis be put on the agenda, when M1-polarized microglia are converted into the M2 phenotype via PPARγ [44]. The M1-to-M2 repolarization of microglia is also modulated by lowered expression of XIST, a long non-coding RNA (lncRNA) whose loss boosts secretion of exosomes with miRNA-503 to restrain proliferation of T cells [45]. By secreting VEGF, IGF-1, BDNF as well as expressing VEGF and CXCL2, M2-polarized microglia gather around BCBrM sites, promoting vascularization and supporting further tumor growth [44].

In general, microglia are commonly considered as the major cell population driving shortened survival, behavioral disorders and neuroinflammation in the setting of BCBrM. For instance, intratumoral infiltration of CD163 + M2-polarized microglia in BrM lesions is correlated with poor prognosis for luminal-like and TNBC patients [46], while peritumoral accumulation of activated microglia was observed to cause cortical dysfunction in heterozygous Cx3cr1^GFP/+ mice by means of living imaging and extracorporal electrophysiology [47]. A recent breast tumor-bearing BALB/c mice model-based research also favored the hypothesis that microglia are responsible for BC-induced neuroinflammation [48]. Collectively, these results suggest the critical roles of microglia in fostering BCBrM and concomitant cognitive complications.

Conventionally, microglia were regarded as a group of tumor supporters that propel the establishment of an immune-suppressive and tumor-permissive niche in prior studies [39, 44, 49]. In fact, microglia can also suppress BCBrM in a pro-inflammatory pattern, participating in phagocytosis, apoptosis, antigen presentation (AP), recruitment of CD8 + T cells, and TME remodeling by inhibiting TLRs, CCRs, CSF1R, CX3CLR and myeloid checkpoints [44]. This impressive anti-tumor effect of microglia in BCBrM was recently validated by scRNA-seq in a fms-intronic regulatory element-knockout (FIRE-KO) mice model born without microglia. It was analyzed that microglia in the humanized mouse model could enhance three pivotal pro-inflammatory events against BCBrM, including IFN signaling, AP and secretion. The paucity of microglia would exhaust anti-tumor activities of NK and T cells, thus tumor regression was largely limited. Evans and colleagues partly ascribed the difference to more complete and restricted depletion of microglia, as well as born without microglia for the FIRE-KO mice model [50]. Further mechanistic studies are demanded to figure out how microglia promote anti-tumor effects in the CNS.

Beyond Wnt/β-catenin, CCL2/CCR2, SDF1/CXCR4, TLRs and PI3K that were previously summarized by Caffarel and Braza [39], microglia in the setting of BCBrM are mediated by some other aberrant signaling pathways (Fig. 3 upper). To begin with, ANXA1 and its downstream FPR were reported to regulate the activation, recruitment and migration of microglia [51]. Galectin-3, whose depletion would induce pro-inflammatory responses in the TME to curb MBC proliferation, also plays a vital role in the context by boosting the activation of an immunosuppressive microglia phenotype [52]. In addition, the polarization status of microglia in BCBrM is also regulated by estrogen, for dense enrichment of M2-polarized microglia were found in BrM sites of premenopausal BC patients. Tamoxifen, a widely used selective estrogen receptor modulator that competitively blocks estrogen signaling, could suppress BCBrM in ER-negative specimens by deviating microglia polarity from M2 to M1, thus cutting down secretion of CCL5 and restraining stemness of TNBC cells [53]. The combining application of AC4-130 (a CSF2Rb-STAT5 inhibitor) with BLZ945 (a CSF1R inhibitor) could effectively reverse adaptive resistance to BLZ945 by blocking the activity of pro-inflammatory tumor-associated microglia, suggesting the driving role of CSF1R and compensatory role of CSF2Rb/STAT5 in microglia-mediated BCBrM [54].

Lung-specific macrophages and breast cancer lung metastasis (BCLuM)

BCLuM is a common event for MBC with noteworthy incidence and lethality. According to the statistics disclosed from the China National Cancer Center (CNCC), around 35.7% of MBC patients were synchronously detected with LuM at initial diagnosis in the Chinese Han population, whose median overall survival (mOS) was 41.7 months in average. Among all subtypes, TNBC occupied the leading BCLuM proportion (up to 42.3%) and was responsible for the shortest OS (only 26.8 months). It was also summarized that aged over 50, Eastern Cooperative Oncology Group (ECOG) grade 2, M1, HR–/HER2 + or TNBC tumors, and disease-free survival (DFS) over 2 years are associated with a high incidence of BCLuM, while ECOG grade 2, TNBC, liver metastases, multiple metastatic sites and DFS no longer than 2 years were inverse prognostic factors [55]. Based on the Surveillance, Epidemiology, and End Results (SEER) database, similar subtype-related incidence rates, survival spans and risk factors of BCLuM were concluded by Xiao et al. and Liu et al. respectively [56, 57].

The intercellular interactions between MBC cells, AMs and other local cells like lung dentritic cells (LDCs) and Th cells are demonstrated in Fig. 2b. Affected remotely by primary BC cells via VEGF, PGF [58], exosomal Cav-1 [59] and other potential communicative factors, AMs get self-proliferated and polarized in the pulmonary PMN before the arrival of MBC cells, which is mediated by C5aR. These accumulated AMs promote BCLuM by directly curbing the generation of IFN-γ in Th1 cells, thus suppressing their potent tumoricidal responses while supporting Th2 cells with poorer antitumor effects instead. The expression of MHC-II on tumor cells is also downregulated by AMs to avoid immune attacks from cytotoxic Th1 cells. In the meanwhile, LDCs are blunted by AMs as well with reduced expression of CD80, CD86 and MHC-II due to increased TGF-β in the TME. Depletion of AMs in synergy with C5aR deprivation could remarkably alleviate the burden of BCLuM with restored functions of LDCs and Th1 cells [60].

As specific macrophages residing in lung, AMs play a critical role in reprogramming the pulmonary niche during the process of BCLuM. AMs could trigger the recruitment of T cell-inhibitory monocytic myeloid-derived suppressor cells (M-MDSCs) and sequential formation of an immunosuppressive PMN by overexpressing CCL12, which was found to be induced by CXCL10-activated CXCR3 and TLR4. Correspondingly, deficient or inhibited CXCR3 and/or TLR4 could result in decreased expression of CCL12 in AMs, which further ameliorates BCLuM with less enrichment of M-MDSCs in the TME. Thus, the CXCL10-CXCR3/TLR4-CCL12 axis in AMs is unveiled to contribute to BCLuM and may function as a promising therapeutic target [61]. TLR4 can also be stimulated by tenascin C to promote the formation of an endothelia-mediated pro-metastatic niche induced by NO and TNF from activated perivascular macrophages in a VEGF-independent manner [62].

In addition to AMs, in fact, the pulmonary PMN is also filled up with other immune cells, chronologically including inflammatory monocytes and neutrophils (prior to BCLuM) and suppressive macrophages (concordant with BCLuM), as was revealed from the single-cell perspective. It is interesting to note that the majority of immune cells are centrally gathered in the TME, except for TREM2-positive regulatory macrophages (Mregs) that are situated in the periphery of the metastatic site in both mice models and human specimens. These marginally distributed Mregs facilitate the establishment of an immunosuppressive pulmonary milieu, thus lung-metastasized BC cells can evade immune surveillance for further development [63].

The underlying signaling network for AMs in LuM has been thoroughly studied over the past decade. To start with broadly, the Wnt/β-catenin/TNF-α axis in AMs is involved in the regulation of LuM. Kramer and colleagues found that the raised activity of β-catenin in AMs could lead to inflammatory dysfunction and drive LuM in a TNF-dependent manner, because TNF-α blockade could significantly intercept the progress of LuM [64]. Focusing specifically on BCLuM, numerous signaling pathways have been identified (Fig. 4), whose effects on BCLuM via AMs can be categorized into four main aspects, namely M2 polarization, transmembrane signal transduction, migration & infiltration, and secretion.

Bone-specific macrophages and breast cancer bone metastasis (BCBoM)

Bone stands as the most predisposed destination of MBC distant metastases. BCBoM at initial BC diagnosis accounted for 65.1% among the whole MBC population by analyzing the SEER database and Fudan University Shanghai Cancer Center (FUSCC) cohort. In terms of survival, mOS for BCBoM patients from the SEER dateset and FUSCC cohort was 30.0 and 68.2 months respectively [65]. Another similar SEER-based study reported that the mOS for stage IV BCBoM patients reached at 38.0 months, and 5 year survival rate was 33.9% [66]. Specifically, the HR + /HER2– MBC patients were burdened with the highest BoM incidence (73.9% to be accurate), the HR + /HER2 + subtype held the longest survival period, while TNBC was presented with the worst prognosis [65]. MBC patients with BoM are usually tortured by pain, frailty and various skeletal-related events (SREs) including pathological fracture, spinal compression, hypercalcemia and forced requests for additional bone-targeted therapies (e.g. surgery, radiotherapy, and bone-modifying agents [BMAs] like zoledronic acid [ZA] and denosumab), all of which considerably undermine their living quality and consume social resources [67].

The intercellular interactions between MBC cells, osteoclasts and other local cells like osteoblasts and CD8 + T cells are demonstrated in Fig. 2c. In most cases, BCBoM scenarios are dominated by the detrimental osteolytic pattern, in which osteoclast-driven bone resorption prevails osteoblast-driven bone formation [68]. After awakening from the dormant state in the bone microenvironment (BME), MBC cells can act on osteoblasts by secreting a series of cytokines, including IL-6/8/11, TNF, PGE2 and PTHrP, which (in)directly stimulate osteoblasts to produce RANKL. The binding between RANKL and RANK expressing on precursory osteoclasts facilitates the survival, recruitment and differentiation of these osteoclast precursors, maturing them to destroy bones, which can be antagonized by OPG derived from osteoblasts. Besides, MBC cells also promote PMN formation by transferring miR-21-containing exosomes to osteoclasts [69]. Reciprocally, the damaged bone matrix would release calcium, collagen and growth factors including TGF-β, IGF-1, FGF, PDGF and bone morphogenetic proteins (BMPs) to boost tumor growth by stimulating PTHrP, Jagged-1 (the ligand of Notch) and other markers on tumor cells, thus shaping the vicious tumor proliferation-bone destruction cycle in the BME [67, 70].

The interactions between osteoclasts and BME components, especially T cells, were partly unveiled by recent researches. On the one hand, osteoclasts undergoing apoptosis give rise to a myriad of apoptotic bodies (ABs) throughout the intermittent process of bone remodeling. Rather than being negligible wastes remaining in the BME, these osteoclast-derived ABs were proved as executors of immunosuppressive effects instead. They can curb the activation of naive CD8 + T cells by combining Siglec15 to TLR2, which results in insufficient infiltration of immune cells, thereby promoting not only BCBoM but also secondary metastasis to other vital organs [71]. On the other hand, inactivated T cells could augment osteoclastogenesis and fuel the tumor-bone vicious cycle, while activated T cells could inhibit the formation of osteoclasts by releasing IL-4 and IFN-γ [72].

Actually, the vicious cycle of BCBoM is under the regulation of a complicated network comprising numerous osteoclast-related signal pathways beyond the classical RANK/RANKL and Jagged/Notch axis (Fig. 3 right). For example, the kinase Fam20C from different origins could exert completely opposite effects via different substrates. Myeloid-derived Fam20C promotes phosphorylation of OPN to suppress osteoclastogenesis and BCBoM, whereas BC-derived Fam20C promotes BMP4 phosphorylation to enhance osteoclastogenesis and BCBoM [73]. In addition to TGF-β, IGF-1/IGF-1R, PI3K/AKT/mTOR, Wnt and Hippo that have been reviewed by Song et al. [74], other typical signaling pathways include VCAM-1 with its receptor integrin α4β1 [75] and MSP (a.k.a. MST-1) with RON kinase (a.k.a. MST1R) [76] that stimulate osteoclast activity, as well as the ROS/JNK/ZEB1 axis that inhibits matrix metalloproteinases (MMPs) and curbs osteoclast activity [77]. Besides, the binding between CD137 and Fra1 could promote osteoclast differentiation [78], whereas YTHDF2-mediated m⁶A degradation of the lncRNA FGF14-AS2 boosts RUNX2 translation and RANKL transcription, thus promotes osteoclast differentiation as well [79].

Targeting cancers with DNA homologous recombination repair deficiency, poly (ADP-ribose) polymerase (PARP) inhibitors represented by olaparib have been officially approved to treat BC patients harboring germline BRCA1/2 mutation [80]. However, olaparib was found to promote BCBoM and secondary bone loss by enhancing differentiation of osteoclasts via PARP2 (not PARP1) in the myeloid lineage (not in BC cells) [81]. In the meanwhile, osteoclasts featured by upregulated glutamine production could increase the resistance of MBC cells to olaparib and cisplatin [82]. As a result, more tailored researches are warranted for BCBoM patients receiving the treatment of PARP inhibitors.

Liver-specific macrophages and breast cancer liver metastasis (BCLiM)

Inferior to bone and lung, liver is the third-ranking preferred destination of MBC, although LiM is mostly of gastrointestinal origins, especially from colorectal cancer (CRC) [83]. Among all BC subtypes, the HR–/HER2 + subgroup is associated with the highest incidence of BCLiM [84, 85]. As for survival, only LiM (especially as the primary metastasis) represents a poor prognosis among all metastatic settings for HER2-positive MBC patients [86]. Besides, based on the SEER database and FUSCC cohort, Ji et al. concluded that the total mOS for BCLiM patients was 20.0 months (SEER) and 27.3 months (FUSCC) respectively. The HR + /HER2 + subtype represented the best prognosis, while TNBC remained the opposite [87]. This conclusion was supported by another SEER-based analysis by Zhao et al. [84], but Lin et al. found that the HR + /HER2– subtype is associated with the longest survival (38.2 months for HR + /HER2– BCLiM patients versus 31.4 months for overall BCLiM patients) by analyzing the data from CNCC [85]. In order to provide optimal cares for BCLiM patients, comprehensive implementation of systemic and locoregional modalities are recommended to be personally formulated in multidisciplinary team (MDT) patterns [83].

By and large, the process of LiM can be separated into several sequential periods: the formation of PMN before tumor invasion, and the microvascular, pre-angiogenic, angiogenic and growth phase after tumor invasion. At each phase, metastatic cancer cells are about to interact, either remotely or closely, with residing cellular populations such as KCs in the hepatic microenvironment [88]. For instance, integrin αvβ5-expressing exosomes from tumor cells incorporate specifically with KCs to mediate liver-tropic metastasis [89]. In addition, tumor-derived extracellular vesicles and particles (EVPs) containing fatty acid (especially palmitic acid) could activate KCs to produce TNF, thus boosting the formation of fatty liver and establishment of pro-inflammatory niches [90].

The intercellular interactions between MBC cells, KCs and other local cells like neutrophils and NK cells are demonstrated in Fig. 2d. Overall, KCs exhibit bidirectional effects towards invasive MBC cells in different phases. Initially, KCs are able to destruct MBC cells by secreting cytotoxic molecules like ROS and protease, as well as recruiting immune cells like NK cells via GM-CSF and IFN-γ [91]. However at the terminal stage of BCLiM, these previously potent phagocytes could not contain the deteriorating condition anymore owing to their sharp loss in the TME and inaccessibility to MBC cells [92]. Worse than that, KCs also actively conspire to facilitate BCLiM by producing various growth factors (e.g. VEGF and HGF), cytokines (e.g. IL-6) and MMPs (e.g. MMP-9 and MMP-14), which has already been clearly verified in CRC models [91].

Currently, immunotherapy represented by immune checkpoint inhibitors (ICIs) that target at programmed cell death protein 1 (PD-1), its ligand PD-L1, and cytotoxic T lymphocyte-associated antigen-4 (CTLA-4) is emerging as a revolutionizing regimen for the treatment of MBC, especially promising for TNBC patients [93,94,95]. Nevertheless, its clinical application is largely restricted by diverse immune-related adverse events (irAEs), which are difficult to be distinguished from desired antitumor reactions. By inducing Th1-promoting immunotherapy in mice models, Siwicki et al. demonstrated that KCs contributed to immune-related hepatotoxicity by releasing IL-12 after detecting boosted IFN-γ from Th1 cells. Actually, it is the ensuing neutrophil response, neither IFN-γ nor IL-12, that is responsible for liver damage induced by immunotherapy, but the mediating role of KCs should not be neglected [96].

In terms of the BCLiM modulating signaling pathways related to liver-specific macrophages (Fig. 3 left), it is notable that the initially robust phagocytotic activity of KCs against MBC cells is motivated by a protein complex consisting of ERMAP (a pro-phagocytosis transmembrane protein in charge of conveying the “eat me” signal to KCs) and dectin-2 (an innate immune receptor belonging to the CLR family) via the bridging molecule galectin-9, as was unraveled by a recent in vivo genome-wide CRISPR-Cas9 knockout screening-based study [97]. This dectin-2-dependent tumor phagocytosis of KCs was actually controlled and selectively endowed by the nuclear factor ID3, which also masters the activation state of liver NK cells and CD8 + T effector cells against tumors [98]. In fact, a preliminary conclusion has come out dating back to 2016, when Kimura et al. reported the critical functions of not only dectin-2, but also its ligation partner MCL (a.k.a. dectin-3) to suppress BCLiM in KCs [99], thus jointly composing the sequential signaling link from ID3, ERMAP, galectin-9 and dectin-2 to MCL or dectin-3. In addition, sGRP78 from tumor cells [100] and ANXA1 with its receptor FPR2 [101] propel BCLiM jointly by promoting macrophage M2 polarization. The activation of CX3CL1 and CX3CR1 in the liver PMN, which further upregulates MMP-9, also enhance the migration of macrophages and invasion of MBC cells [102].

In the contemporary era of cancer research and treatment, targeting macrophages, especially TAMs in the TME, has emerged as a promising strategy [103,104,105]. These modalities are usually designed based on distinctive characteristics of macrophages (e.g. functional [106] and metabolic [107] plasticity), and exert versatile antitumor effects by intervening various macrophage-involved processes (e.g. proliferation, recruitment, infiltration and M2 polarization) [108], or engineering macrophages into chimeric antigen receptor macrophages (CAR-M) with reinforced capabilities to phagocytize cancer cells, present antigens, secrete pro-inflammatory cytokines and trigger immune responses [109, 110]. Drawing lessons from these available experience, we plan to delve specifically into TSMs to explore their potential as therapeutic targets against MBC in this section. We will list published clinical trials with positive results (Table 2) and ongoing (including completed but not published) clinical trials (Table 3) about small-molecule inhibitors and monoclonal antibodies (mAb) of TSM-related signaling pathways in MBC, as well as assistance from microbes and microwave ablation (MWA).

Signaling pathway blockade

The activation of the JAK-STAT axis has been found to contribute to BCBrM. In preclinical experiments, a subcluster of reactive astrocytes with activated STAT3 was associated with expansion of CD74-expressing microglia to accumulate and exert pro-growth effects in the BrM site [111]. Indeed, the combination of a STAT3/5 inhibitor (TTI-101) and a PARP inhibitor (olaparib) could effectively overcome acquired resistance of palbociclib, a cyclin-dependent kinase 4/6 inhibitor (CDK4/6i) in ER-positive BC cells [112], while synergistic inhibition of TrkA by entrectinib and JAK2 by pacritinib (also inhibits IRAK1) could suppress BC growth and metastasis [113]. Therefore, STAT3 inhibitors (e.g. TTI-101 and silibinin), JAK1/2 inhibitors (e.g. ruxolitinib and pacritinib) and JAK-targeting multiple kinase inhibitors (MKI) like tinengotinib (a.k.a. TT-00420) have been widely tested, either as monotherapy or in combination with other drugs, to treat MBC, especially for BCBrM patients.

In HR-positive MBC, activated PI3K/AKT/mTOR (PAM) pathway was reported to drive tumor growth and metastasis, with several targeted inhibitors having been approved, including the PI3K inhibitor alpelisib, the AKT inhibitor capivasertib and the mTOR inhibitor everolimus [114]. Since PI3K was regarded as a principal regulator of not only MBC cells but also pro-metastasis microglia and macrophages [115], it is no wonder to apply PAM inhibitors in BCBrM mice and populations. In 2016, for example, Ni et al. reported that rather than monotherapy, only joint inhibition of PI3K by buparlisib (a.k.a. BKM120, a pan-PI3K inhibitor with the ability to penetrate the BBB) and mTOR by everolimus (a.k.a. RAD001) could elicit potent remissions in BCBrM patient-derived xenograft (PDX)-bearing mice models [116], providing novel insights for future translational practices. The phase 3 BELLE-2 trial (NCT01610284) examining buparlisib and fulvestrant in postmenopausal endocrine-resistant HR + /HER2– MBC attained a significantly improved PFS in the total, known PI3K status and PI3K-activated population, suggesting the effectiveness of combining PI3K inhibition and endocrine therapy in the indicated patients [117].

In accordance with the above-mentioned viewpoints, the CSF signaling is at least responsible for the progression of BCBrM and BCLuM. Taking a step further, depletion of microglia and infiltrating macrophages by PLX3397, a CSF-1R inhibitor, remarkably lessened BCBrM burden especially in young murine models [118]. Similarly, local pulmonary administration of PLX3397 in synergy with intravenous paclitaxel showed an inspiring efficacy and acceptable toxicity in the treatment of BCLuM mice, which was accompanied by gaining superiority of M1-polarized macrophages than their pro-tumor counterparts [119]. In addition to PLX3397, other small-molecule inhibitors like LY3022855 and anti-CSF-1R mAb like axatilimab have been adopted to treat MBC in several clinical trials.

Wnt/β-catenin is another pivotal signaling pathway that involves in various developing processes of MBC, including cell proliferation and metastasis, TME regulation, stemness maintenance, therapeutic resistance and so on [120]. As were previously reported, the hexapeptide foxy-5 derived from Wnt-5a protein could undermine motility (migration and invasion) of MBC cells, thus significantly inhibited BCLuM and BCLiM concluded from in vitro analyses and in vivo experiments [121]. Further, the Wnt/β-catenin signaling in AMs is in positive correlation with BCBrM in a TNF-dependent pattern [64]. Thus, Wnt mAb and inhibitors such as vantictumab (a.k.a. OMP-18R5) and foxy-5 may well hold promise for treating BCLuM by blocking the Wnt/β-catenin pathway.

Represented by ZA and denosumab, BMAs are widely used to counteract SREs in BCBoM patients partly by targeting the RANKL/RANK interaction to interfere osteoclast maturation and function, blocking the formation of PMN and modulating immune response in the TME [122]. In addition to efficacy and toxicity profiles that have been verified by numerous clinical trials, the de-escalated administration strategy (every 12 weeks or even every 24 weeks) instead of the traditional 4-week scheme is getting incrementally favored to function as standard of care. A systematic review and meta-analysis pointed out that there seemed to be no difference in terms of pain or SREs between the de-escalated and traditional dosing [123]. Besides, ZA every 3 months is more cost-effective with non-inferior effectiveness to reduce the risk of SREs in comparison to monthly denosumab extended from the CALGB 70604 (Alliance) trial (NCT00869206) [124].

In spite of these enlightening theories and positive preclinical results, there are some frustrating research conclusions and clinical trials that cannot be neglected as well. To our surprise, it was reported that anti-CSF-1(R) mAb and small-molecule inhibitors of CSF-1R could instead promote spontaneous metastasis without affecting primary BC burden in a mice model. The paradoxical outcome was ascribed to upregulated serum G-CSF, neutrophils in the primary and metastatic sites, together with neutrophils and Ly6C(hi) monocytes in the peripheral circulation, which could be reversed by anti-G-CSFR mAb [140]. Ruxolitinib, the selective JAK1/2 inhibitor, failed to achieve expected objective response rate (ORR) in metastatic TNBC as monotherapy, despite observed on-target activity [141]. Neither did the addition of ruxolitinib to capecitabine showed significant survival benefit in OS and progress-free survival (PFS) in HER2-negative MBC, despite a numerically better ORR and health-related quality of life [142]. The phase 2 TBCRC 039 trial (NCT02876302) where preoperative ruxolitinib plus paclitaxel were used to treat inflammatory TNBC only achieved a pathological complete response (pCR) rate of 8.7% (n = 2 in 23) despite decreased levels of phosphorylated STAT3 [143]. Likewise, these failures could be partly imputed to increased expression of pro-tumor factors like COX-2 following JAK/STAT inhibition, since COX-2 depletion could sensitize MBC cells to ruxolitinib [144]. For the pan-PI3K inhibitor buparlisib, its single dose demonstrated negligible clinical benefit in metastatic TNBC despite downregulated PI3K activity in patients with stable disease [145]. Even buparlisib plus trastuzumab (capecitabine was also added in case of progressive BrM) in HER2-positive MBC [146], plus paclitaxel in HER2-positive MBC in the phase 2/3 BELLE-4 trial (NCT01572727) [147], or plus fulvestrant in postmenopausal HR + /HER2– MBC in the phase 3 BELLE-3 trial (NCT01633060) [148], these buparlisib-based combination therapies all ended up in failure due to limited efficacy or excessive toxicity. The MEK1/2 inhibitor selumetinib in combination with buparlisib or pazopanib (a PDGFR inhibitor) improved survival and suppressed relevant signaling in mice bearing some sensitive metastatic TNBC cell lines with intracranial metastases [149], but the addition of selumetinib to fulvestrant in aromatase inhibitor (AI)-progressed HR + /HER2– MBC patients even worsened the original efficacy, whose disease-free survival (DCR) dropped from 50 to 23%, mPFS from 5.6 to 3.7 months, whereas median time to treatment failure (mTTF) was shortened from 5.6 to 5.1 months according to the phase 2 SAKK 21/08 trial (NCT01160718) [150]. In parallel, the anti-frizzled mAb vantictumab (OMP-18R5) to inhibit the canonical Wnt signaling encountered a disappointing outcome as well. In a phase 1b trial (NCT01973309), vantictumab plus weekly paclitaxel led to an ORR of 31.3% and clinical benefit rate (CBR) of 68.8%, but 6 cases of unexpected fractures in 48 enrolled HER2-nagetive MBC hampered its further development [151].

In terms of BCBoM, as the first recognized proto-oncogene and a nonreceptor tyrosine kinase, aberrant activation of the SFK signaling was found in various malignant tumors including MBC, which plays a vital role in driving the development of BCBoM [152]. Targeting Src signaling by SFK inhibitors like dasatinib, saracatinib (a.k.a. AZD0530) and bosutinib (a.k.a. SKI-606) in MBC has undergone an array of attempts by preclinical and early phase clinical trials [153, 154]. Disappointingly, neither as single agent in HR-positive MBC [155] nor in combination with AI in HR-positive MBC [156] did saracatinib (AZD0530) exhibited expected beneficial effects. Similar unsatisfactory conclusions were reached as for bosutinib (SKI-606), whose partnership with either letrozole (an AI) in postmenopausal HR + /HER2– MBC [157] or capecitabine in MBC and other advanced solid tumors [158] gave rise to limited efficacy. Single dose of dasatinib in phase 2 trials enrolling metastatic TNBC, HER2-positive and/or HR-positive MBC and general MBC populations successively ended up in failure in 2011 [159,160,161], irrespective of dosing schedule (either 100 mg qd or 70 mg bid) [162]. When it comes to dasatinib-based combination therapy, four of pertinent clinical trials acquired positive results (listed in the Table 2), but a further phase 2 trial testing dasatinib and paclitaxel in MBC was terminated early without future study plan owing to slow accrual [163].

Other promising therapies

In addition to signaling pathway blockade, there are some other promising strategies to harness the potential of TSMs in MBC. To begin with, by manipulating the machinery of engineered E.coli-produced CRISPR-Cas9, the expression of MafB and c-Maf could be simultaneously inhibited, which reversed dysfunctional KCs at the terminal stage of BCLiM to regain their potency against MBC by acquiring numerical and functional improvement. These motivated KCs could also secrete pro-inflammatory cytokines to provoke T cell response, thus further boosts the antitumor efficacy [92]. Trained by influenza viruses and mediated by IFN-γ and NK cells, AMs could be armed with enhanced phagocytosis, cytotoxicity and enduring anti-BCLuM capability [164]. Last but not least, MWA is a kind of minimally invasive thermal therapy that has been exploited to treat MBC patients, whose underlying mechanism has been partially unveiled. After treating primary sites by MWA in MBC murine models, macrophages could be activated, generating IL-15 to activate NK cells. Therefore, BCLuM progression is contained and survival period is elongated [165]. Pertaining to the exploration of MWA in BC, the majority of available fundamental and clinical researches focused on early-stage BC [166,167,168,169], whereas its application in MBC is worth expecting in the future.

Macrophages, the ubiquitous “big eaters”, exhibit diverse heterogeneity and exert pivotal functions in various physiological and pathological conditions. In this review, we adopted the newly proposed concept of TSMs rather than TAMs that has been numerously dissected before, attempting to elucidate their multifaceted diversities and plasticity in terms of evolvement, polarization, profile, function and metabolism, to untangle intricate interactions between typical TSMs (microglia, AMs, osteoclasts and KCs) and MBC cells (the “seeds”) and corresponding TME (the “soil”) in four typical MBC scenarios (BrM, LuM, BoM and LiM), and to compile promising TSM-targeting therapies (signaling pathway blockade, macrophage engineering, MWA, etc.) against MBC in detail.

Despite these endeavors, we have to admit that there are several limitations remaining in our present work. Firstly, due to the paucity of relevant studies, we couldn’t collect enough data about other TSM populations demonstrated in Fig. 1 beside the four mentioned above, as well as BCLiM-specific mechanisms and therapeutics. Besides, mammary gland tissue-resident macrophages (MGTRMs) are not included in the main text. In fact, MGTRMs stand as the most enriched stromal cells in the early phase of TNBC [170], whose heterogeneous features also affect the pathogenesis and development of MBC [171, 172]. For instance, ductal macrophages play an essential role in guarding mammary epithelial conditions and regulating tissue remodeling [173]. Among MGTRMs, the Lyve-1-overexpressing population is responsible for ECM homeostasis [174], while the perivascular FOLR2 + subgroup interacts with CD8 + T effector cells and symbolizes better prognosis of BC patients [175]. With the advent of novel approaches (e.g. syngeneic BC models [176] and iPSC-employed large-scale production [177]) and brand-new perspectives (e.g. temporal phenotype dynamics [178]) to decipher the roles of TSMs, it is promising to expect more thorough understanding of the whole iceberg, and the development of more efficacious, safe and tailored therapeutic strategies against MBC in the approaching era.

Not applicable.

ABs:

Apoptotic bodies

AI:

Aromatase inhibitor

AMs:

Alveolar macrophages

ANAX1:

Annexin A1

AP:

Antigen presentation

BBB:

Blood–brain barrier

BC:

Breast cancer

BDNF:

Brain-derived neurotrophic factor

BMAs:

Bone-modifying agents

BME:

Bone microenvironment

BMPs:

Bone morphogenetic proteins

BoM:

Bone metastasis

BrM:

Brain metastasis

C5aR:

Complement C5a receptor

CAMs:

CNS-associated macrophages

CAR-M:

Chimeric antigen receptor macrophages

CBR:

Clinical benefit rate

CCL/CCR:

CC-motif chemokine ligand/receptor

CDK4/6i:

Cyclin-dependent kinase 4/6 inhibitor

CLR:

C-type lectin receptor

CNCC:

China National Cancer Center

CNS:

Central nervous system

COX-2:

Cyclooxygenase-2

CRC:

Colorectal cancer

CRISPR-Cas9:

Clustered regularly interspaced short palindromic repeats/CRISPR-associated nuclease 9

CSF-1(R):

Colony-stimulating factor 1 (receptor)

CTLA-4:

Cytotoxic T lymphocyte-associated antigen-4

CVMs:

Central vein macrophages

CXCL/CXCR:

CXC chemokine ligand/receptor

DCR:

Disease control rate

DFS:

Disease-free survival

EBI:

Erythroblastic island

ECOG:

Eastern Cooperative Oncology Group

EMPs:

Erythro-myeloid progenitors

ER:

Estrogen receptor

EVPs:

Extracellular vesicles and particles

FGF:

Fibroblast growth factor

FIRE-KO:

Fms-intronic regulatory element knockout

FPR:

Formyl peptide receptor

FUSCC:

Fudan University Shanghai Cancer Center

GM-CSF:

Granulocyte–macrophage colony-stimulating factor

HBEGF:

Heparin-binding EGF-like growth factor

HER2:

Human epidermal growth factor receptor 2

HGF:

Hepatocyte growth factor

HIF-1α:

Hypoxia-inducible factor 1α

HO-1:

Heme oxygenase 1

HR:

Hormone receptor

HSCs:

Haematopoietic stem cells

ICI:

Immune checkpoint inhibitor

IFN-γ:

Interferon-γ

IGF-1:

Insulin-like growth factor 1

IL:

Interleukin

IMs:

Interstitial macrophages

iPSC:

Induced pluripotent stem cell

irAE:

Immune-related adverse event

IRAK1:

Interleukin-1 receptor-associated kinase 1

JAK:

Janus kinase

KCs:

Kupffer cells

LAMs:

Lipid-associated macrophages

LCMs:

Liver capsular macrophages

LDCs:

Lung dendritic cells

LiM:

Liver metastasis

lncRNA:

Long non-coding RNA

LuM:

Lung metastasis

LXRα:

Liver X receptor α

mAb:

Monoclonal antibody

MBC:

Metastatic breast cancer

MCL:

Macrophage C-type lectin

MDT:

Multidisciplinary team

MEK:

Mitogen-activated protein kinase kinase

MGTRMs:

Mammary gland tissue-resident macrophages

MHC:

Major histocompatibility complex

MKI:

Multiple kinase inhibitors

M-MDSCs:

Monocytic myeloid-derived suppressor cells

MMPs:

Matrix metalloproteinases

mOS:

Median overall survival

Mregs:

Regulatory macrophages

MSP:

Macrophage-stimulating protein

MST-1:

Macrophage-stimulating 1

mTOR:

Mammalian target of rapamycin

mTTF:

Median time to treatment failure

MWA:

Microwave ablation

NK:

Natural killer

NO:

Nitric oxide

NRF-2:

Nuclear factor erythroid 2-related factor 2

OPG:

Osteoprotegerin

OPN:

Osteopontin

ORR:

Objective response rate

OS:

Overall survival

OXPHOS:

Oxidative phosphorylation

PAM:

PI3K/AKT/mTOR pathway

PARP:

Poly(ADP-ribose) polymerase

pCR:

Pathological complete response

PD-(L)1:

Programmed cell death protein (ligand) 1

PDGF(R):

Platelet-derived growth factor (receptor)

PDX:

Patient-derived xenograft

PFS:

Progress-free survival

PGE2:

Prostaglandin E2

PGF:

Placental growth factor

PI3K:

Phosphoinositide 3-kinase

PMN:

Pre-metastatic niche

PPARγ:

Peroxisome proliferator activated receptor γ

PR:

Progesterone receptor

PTHrP:

Parathyroid hormone-related peptide

RANK(L):

Receptor activator of NF-κB (ligand)

scRNA-seq:

Single-cell RNA sequencing

SDF1:

Stromal cell-derived factor 1

SEER:

Surveillance, Epidemiology, and End Results

SFK:

Src family kinase

sGRP78:

Secretory glucose-regulated protein 78

SREs:

Skeletal-related events

STAT:

Signal transducer and activator of transcription

TAMs:

Tumor-associated macrophages

TGF-β:

Transforming growth factor β

TLR:

Toll-like receptor

TME:

Tumor microenvironment

TNBC:

Triple-negative breast cancer

TNF-α:

Tumor necrosis factor α

TREM2:

Triggering receptor expressed on myeloid cells 2

TSMs:

Tissue-specific macrophages

VEGF:

Vascular epidermal growth factor

XIST:

X-inactive-specific transcript

ZA:

Zoledronic acid

Not applicable.

This work was supported by the National Natural Science Foundation of China (No.82403574 to K.X.), the China Postdoctoral Science Foundation (No.2023M741470 to K.X.), Jiangsu Association for Science & Technology Youth Science & Technology Talents Lifting Project (No.JSTJ-2024-386 to K.X.), the Research Project of Jiangsu Cancer Hospital (No.ZJ202301 and No.SZL202301 to K.X.), the General Program of Wuxi Medical Center of Nanjing Medical University (No.WMCG202354 to C.W. and No.WMCG202519 to Y.Z.), the Doctoral Talent Fund of the Affiliated Wuxi People's Hospital of Nanjing Medical University (No.BSRC202303 to C.W.), and the Research Project of Lianyungang Anti-Cancer Association (No.ZD202312 to Q.C.).

1. Cenzhu Wang, Pinchao Fan, Yu Zhou and Meican Ma have contributed equally to this work.

Authors and Affiliations

1. Department of Oncology, The Affiliated Wuxi People’s Hospital of Nanjing Medical University, Wuxi People’s Hospital, Wuxi Medical Center, Nanjing Medical University, Wuxi, 214000, China

   Cenzhu Wang

2. Department of Oncology, The First Affiliated Hospital With Nanjing Medical University, Nanjing, 210029, China

   Pinchao Fan

3. Clinical Research Center, The Affiliated Wuxi People’s Hospital of Nanjing Medical University, Wuxi People’s Hospital, Wuxi Medical Center, Nanjing Medical University, Wuxi, 214000, China

   Yu Zhou

4. Department of Rheumatology, The Affiliated Wuxi People’s Hospital of Nanjing Medical University, Wuxi People’s Hospital, Wuxi Medical Center, Nanjing Medical University, Wuxi, 214000, China

   Meican Ma

5. The Second School of Clinical Medicine, Nanjing Medical University, Nanjing, 211166, China

   Haowei Zhong

6. Department of Gastroenterology, The Affiliated Yixing Hospital of Jiangsu University, Yixing, 214200, China

   Lei Liu

7. Department of Oncology, Ganyu District People’s Hospital of Lianyungang City, Lianyungang, 222100, China

   Qin Chen

8. Department of Oncology, Jiangsu Cancer Hospital, Jiangsu Institute of Cancer Research, The Affiliated Cancer Hospital of Nanjing Medical University, 42 Baiziting, Nanjing, 210009, China

   Kun Xu

Contributions

K.X., Q.C. and C.W. provided direction and guidance for the manuscript. C.W. and P.F. wrote the whole manuscript. Y.Z., M.M., H.Z. and L.L. made significant revisions to the manuscript. All authors have read and approved the final manuscript.

Corresponding authors

Correspondence to Lei Liu, Qin Chen or Kun Xu.

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.

Reprints and permissions