Macrophages derived from embryonic precursor cells (resident-tissue macrophages) and bone marrow monocytes (monocyte-derived macrophages [MDMs]) may exhibit distinct functions in cancer (Schulz et al., 2012; Jacome-Galarza et al., 2019). For example, our recent studies showed that macrophages derived from Ly-6C+ inflammatory monocytes (Inflam-Monos), but not CD169+ bone-resident macrophages, are important for breast cancer bone metastasis growth (Ma et al., 2020). To characterize macrophage heterogeneity in our MycCaP-Bo model, we performed single-cell RNA-seq (scRNA-seq) of cells from healthy bone marrow, bone metastasis with vehicle treatment (naive), enzalutamide for 4 d (Enz-4d), 7 d (Enz-7d), and 18 d (resistant). Through Uniform Manifold Approximation and Projection (UMAP) clustering, we identified a total of 9,454 monocytes/macrophages that were further divided into five subsets based on the differentially expressed genes (Fig. 3 A). According to their signature gene expression, we defined these subsets as proliferating monocytes (Mki67), Inflam-Monos (Ccr2, Fos), resident tissue macrophages (RTMs; Hes1, Nr4a1), Isg15+ macrophages (Isg15, Stat1, Irf7), and Ftl1+ macrophages (Ftl1, Fabp5; Fig. 3, B and C). Among different samples, the abundance of RTMs and Inflam-Monos were higher in normal bone marrow (healthy) compared with bone metastasis (naive; Fig. 3 D), which agreed with the identification as RTM and Inflam-Mono from gene signature. In contrast, the other two macrophage populations increased in bone metastasis samples compared with normal, suggesting that they are MAMs (Fig. 3 D). Upon enzalutamide treatment, Isg15+ MAMs further increased in resistant tumors compared with naive tumors, while Ftl1+ MAMs abundance was not significantly different, despite some fluctuation at day 7 and large variation among samples at day 18 (Fig. 3 D). Pseudotime analysis indicated that Isg15+ MAMs were MDMs and potentially differentiated from Inflam-Monos (Fig. 3 E). Further pathway enrichment analysis identified major pathways enriched in different macrophage subsets (Fig. S3, A–E). Notably, Isg15+ MAMs and Inflam-Monos were predominantly enriched for inflammatory pathways, including positive regulation of response to cytokine, regulation of tissue remodeling, and cytokine production for Isg15+ MAMs (Fig. S3 A), and regulation of T cell cytokine production and regulation of α-β T cell activation for Inflam-Monos (Fig. S3 B). Together, these data suggested that Isg15+ MAMs differentiated from Inflam-Monos might be important for the promotion of anti-androgen resistance in MycCaP-Bo model.

To test this directly, we used a mouse model with genetic ablation of CC chemokine receptor 2 (Ccr2), the major chemokine receptor mediating the recruitment of Inflam-Monos (Palframan et al., 2001; Getts et al., 2008). Similar to breast cancer models, MycCaP-Bo bone metastasis growth was significantly inhibited in syngeneic FVB Ccr2−/− mice (Fig. 3, F and G; vehicle-treated groups [Veh]) deficient of Inflam-Monos as reported previously (Shi et al., 2011). In mice receiving enzalutamide treatment, the development of resistant tumors was also significantly inhibited (Fig. 3, F and G; Enz). Using TRAP staining, we determined that the density of osteoclasts located on bones surface adjacent to metastasis lesions was substantially reduced with pan-macrophage depletion by L-Clod, regardless of enzalutamide treatment (Fig. S3 I). In contrast, osteoclast abundance was not affected in monocyte-deficient Ccr2−/− mice (Fig. S3 K). These data indicated that CCR2-recruited MAMs were important for anti-androgen resistance of MycCaP-Bo bone lesions, in an osteoclast independent manner.

CD169 has been recognized as the marker for bone marrow RTMs (Hashimoto et al., 2013). CD169+ RTMs were recently illustrated to contribute to tumor initiation of lung cancer (Casanova-Acebes et al., 2021). In the MycCaP-Bo model, RTMs were enriched for pathways of chemokine signaling, inflammatory response, and phagocytosis (Fig. S3 C). To test the role of CD169+ RTMs, we used transgenic mice expressing DTR under the control of CD169 promoter (CD169-DTR), in which CD169+ bone-resident macrophages can be depleted upon DT treatment compared with the control treatment with Glu52-DT (Ma et al., 2020). Consistently, macrophages associated with MycCaP-Bo bone lesions can be significantly depleted in CD169-DTR mice regardless of enzalutamide treatment (Fig. 3 H and Fig. S3, F and G). We further confirmed that CD106, another resident macrophage marker (Kaur et al., 2018), was also expressed by CD169+ macrophages, and CD106+CD169+ macrophages were efficiently depleted in CD169-DTR mice with DT treatment (Fig. S3 H). Similar to breast cancer bone metastasis models (Ma et al., 2020), this depletion did not affect MycCaP-Bo bone metastasis growth in vehicle-treated mice. In contrast, depletion of CD169+ macrophages synergistically inhibited MycCaP-Bo bone metastasis growth in combination with enzalutamide (coefficient of drug interaction = 0.59; Fig. 3, I and J). This indicated that CD169+ RTMs contributed to the development of enzalutamide resistance in MycCaP-Bo model. Furthermore, in resistant tumors, the ablation of CD169+ macrophages significantly inhibited continuous growth of resistant bone lesions (Fig. 3, K and L) indicating the importance of their continuous presence. Depletion of CD169+ bone-resident macrophages did not affect bone surface osteoclast density measured by TRAP staining (Fig. S3 J). Together, these data indicated that although CD169+ RTMs contribute minimally to bone metastasis growth, they are critical for enzalutamide resistance, again in an osteoclast independent manner. Collectively, our data indicated that both CD169+ RTMs and CCR2 recruited MDMs are critical for enzalutamide resistance of MycCaP-Bo bone lesions.