Nicotine promotes brain metastasis by polarizing microglia and suppressing innate immune function

Approximately 170,000 patients in the United States are diagnosed with brain metastasis every year (Ellis et al., 2012). Lung, breast, and skin cancers are the most common tumor types that metastasize to the brain (Bacha et al., 2018; Huang et al., 2018; Redmer, 2018; Xing et al., 2018). Patients with metastatic lung cancer have a very poor prognosis (Mujoomdar et al., 2007), and those with brain metastasis have a 1-yr survival rate of 4–20% (Gaspar et al., 1997; Sperduto et al., 2008; Verger et al., 2005). Despite significant improvements in the treatment of metastatic cancer, the therapeutic options for brain metastasis are still limited mainly due to inefficient drug delivery to brain as well as limited surgical approaches to multiple metastases. Radiation therapies, such as stereotactic radio surgery and fractionated whole brain radiation, are standard therapeutic techniques employed for brain metastasis; however, they provide only limited local tumor control and, more often than not, intracranial failure leading to neurological death. The development of lung cancer brain metastasis is a complex process, requiring the invasion of primary tumor cells into surrounding tissue and vessels, trafficking through the circulatory system, and colonizing into the brain parenchyma (Fidler, 2015). Therefore, understanding the underlying mechanisms that drive brain metastasis and developing more effective therapies are urgently needed.

Cigarette smoking is one of the major risk factors for lung cancer (Lee et al., 2012). Our previous study have shown that smoking is correlated to rapid progression of lung cancer brain metastasis (Shenker et al., 2017). Tobacco smoke contains a toxic mix of >7,000 chemicals. Among them, nicotine is a major cigarette component associated with human tobacco consumption and addiction (Braverman, 1999; Heishman, 1999). Nicotine is not a carcinogen; however, several studies have reported the tumor-promoting effects of nicotine via activation of nicotinic acetylcholine (nAch) receptor in the tumor cells. The nAch receptor is comprised of combination of multiple subunits, and they are expressed in a variety of brain cells including neuron, astrocytes, endothelial cells, and microglia (Mazzaferro et al., 2017). The α4β2 is the most abundant class of nAch receptor in the brain and is the principal mediator of nicotine dependence (Hawkins et al., 2004; Mazzaferro et al., 2017). This suggests that nicotine may be capable of reprogramming the brain tumor microenvironment (TME) via activation of its receptor to promote tumor progression.

Several studies have shown that the interaction of neoplastic cells with microglial cells, brain-residing macrophages, profoundly affects tumor progression in glioblastoma (da Fonseca and Badie, 2013; Roesch et al., 2018; Wei et al., 2013). In brain metastatic lesions of lung cancer patients, abundant activated microglia were found to be infiltrated (He et al., 2006). Importantly, nAch receptor has been shown to be expressed on microglia (Morioka et al., 2015; Sadigh-Eteghad et al., 2016), raising the possibility that nicotine may stimulate microglia’s immunomodulatory pathways, which affects their immune function. Microglia are known to polarize into classical (M1) or alternative (M2) phenotypes by various microenvironment stimuli (Wei et al., 2013; Wu and Watabe, 2017). M1 cells suppress tumors by releasing cytotoxic factors and also by phagocytosis. In contrast, M2 cells have a pro-tumor response by activating immunosuppressive factors. It is virtually unknown how metastatic cells escape immune attack of microglia and then colonize in the brain. In this study, we showed that nicotine promoted lung cancer brain metastasis by enhancing both tumor progression and stemness by skewing microglial polarization and inhibiting its anti-tumor phagocytic ability. We also found that a natural compound, parthenolide (PTL), effectively suppressed tumor progression and stemness by blocking nicotine-induced polarization of the microglia in the brain.

To examine the effect of tobacco smoking on brain metastasis, we analyzed the incidence and mortality rate of 281 patients with lung cancer brain metastasis (Fig. S1) with or without a history of smoking. All patients were diagnosed with stage IV disease. Among them, 79 patients were diagnosed with brain metastasis. 59 patients were current smokers at the time of diagnosis and treatment, while 222 of them had either a previous or no smoking history. We found that the incidence of brain metastasis was significantly higher in current smokers compared with the never or the former smokers group (Fig. 1 A). The results of Kaplan–Meier analyses indicate that the current smokers were associated with worse brain metastasis progression free (Fig. 1 B) and overall survival (Fig. 1 C) compared with those who were not currently smoking. Since activation of microglia was previously reported to be associated with brain tumor progression (da Fonseca and Badie, 2013; He et al., 2006; Roesch et al., 2018; Wei et al., 2013), we examined the status of microglia activation in brain metastatic lesions and found that current smokers exhibited a significantly increased number of Iba1⁺ (pan-microglial marker) cells in the brain metastatic lesions (Fig. 1 D) compared with noncurrent smokers. Importantly, the majority of infiltrated microglia in current smokers was identified to be pro-tumor M2 microglia (CD206⁺ cells; Fig. 1 D). These results suggest a role of smoking-mediated M2 microglial polarization in brain metastatic tumor growth.

Because M2 microglia are abundantly infiltrated in brain metastatic lesions of current smokers, we examined the possibility that nicotine promotes brain metastasis by polarizing microglia to M2 microglia. We first examined the expression of nicotine receptors in microglia cells and in tumor cells and found that microglia expressed high levels of α4β2, which is the most sensitive receptor of nicotine in the brain (Fig. 2 A). We also found that microglia expressed a significantly higher level of α4β2 receptor than macrophage (Fig. S2 A). Moreover, the expression of α4β2 receptor in microglia dramatically increased upon nicotine treatment (Fig. S2 A). We then examined the effect of nicotine on microglia polarization and tumor progression using the LL/2 mouse Lewis lung carcinoma model. The LL/2 cell is originated from the C57BL/6 mouse; however, this cell line can grow tumors in both C57BL/6 and BALB/c mice as reported previously (Janker et al., 2018; Kamei et al., 2009; Scholar et al., 1989). Therefore, we tested the effect of nicotine in both models. We intracardially inoculated luciferase-labeled LL/2 cells into both wild-type mice (C57BL/6 or BALB/c) followed by injecting nicotine (1 mg/kg) every 3 d until the end point. We found that nicotine significantly increased the incidence and growth of brain metastasis in both models (Fig. S2, B–D, for C57BL/6 model; Fig. 2 B and Fig. S2, E and F, for BALB/c model). On the other hand, nicotine did not promote bone metastasis (Fig. 2 B). The result of our immunohistochemical analysis for brain metastatic lesions revealed that high levels of activated microglial cells accumulated in the tumor mass in the nicotine-treated group (Fig. 2 C). Importantly, the majority of these microglia in the nicotine-treated mice were found to be M2 microglia (CD206⁺/Iba1⁺), and fewer M1 microglia (Iba1⁺/F4/80⁺) were identified in the brain lesions (Fig. 2 D and Fig. S2, G and H), suggesting that nicotine promoted microglia polarization to the M2 phenotype in the brain metastatic lesions. Infiltrated peripheral macrophages/microglia are known to modulate brain tumor progression (Bacha et al., 2018; Ellis et al., 2012). To define the exact role of macrophage and microglia on brain metastasis, we stained brain metastatic tumors with antibody against the macrophage marker CD45. We also examined the ratio of the number of microglia (CD45⁻/Iba1⁺) and peripheral macrophages (CD45⁺/Iba1⁺) in the brain metastatic lesions and found that macrophages comprised only 12–14% of CD45⁺/Iba1⁺ cells, and the number of macrophages was not altered between nicotine-treated and -untreated groups (Fig. S2 I). These results strongly suggest that nicotine promotes brain metastasis by skewing microglia to the M2 phenotype. To further validate this hypothesis, we examined the effect of depletion of “original” microglia by treating animals with the inhibitor for CSF 1 receptor (CSF1R), PLX3397. CSF1 signaling is necessary for microglia viability, and treating mice with PLX3397 has been previously shown to eliminate ∼99% of all microglia in the brain (Elmore et al., 2014). We therefore treated immune-competent wild-type mice with PLX3397 and transplanted them with LL/2 mouse lung cancer cells followed by nicotine treatment. To avoid a whole-body effect of PLX3397, the PLX3397 was locally injected into brain through intracranial injection 1 wk after the transplantation of LL/2 cells followed by nicotine treatment (Fig. S2 J). Strikingly, blocking microglia by PLX3397 significantly suppressed nicotine-related brain metastasis of lung cancer (Fig. 2, E–G), prolonged brain metastasis-free survival (Fig. 2 H), and repressed the nicotine-related M2 microglial polarization (Fig. 2, I and J) in the BALB/c mouse model. Similarly, PLX3397 suppressed nicotine-related brain metastasis and inhibited M2 microglial polarization in the C57BL/6 mouse model (Fig. S2, K–N). We also treated nude mice with PLX3397 and transplanted them with H2030BrM cells via an intracardiac injection followed by nicotine treatment (Fig. S2, O–R). Nicotine-induced brain metastasis was significantly suppressed after PLX3397 treatment (Fig. S2, O–Q). The brain metastasis–free survival was also significantly prolonged by PLX3397 treatment (Fig. S2 R). PLX3397 treatment showed no detectable liver toxicity (Fig. S2 S). These results suggest that the nicotine-induced brain metastasis is indeed mediated by the effect of nicotine on microglia.

As shown in Fig. 2 D, the number of M2 microglia was increased after nicotine treatment, suggesting that nicotine promotes microglial polarization. To further investigate the effect of nicotine on microglia polarization, the microglial cells were treated with nicotine followed by measurement of expression of M1 and M2 markers. Our results showed that nicotine did not affect tumor cell proliferation (Fig. S2 T) at 1 µM concentration, but the same dose of nicotine significantly increased the mRNA expression of M2-related genes, arginase-1 (Arg1), arginase-2(Arg2), and CD204, in both human and mouse microglial cells lines (Fig. 3, A and B) and mouse primary microglial cells (Fig. 3 C). Since STAT3 pathway is involved in M2 microglia (Qin et al., 2017), we examined the status of phospho-STAT3 and JAK expression in microglia with or without nicotine treatment, and found that nicotine treatment indeed enhanced the expressions of JAK/STAT3 (Fig. 3 D). We then treated the microglia with nicotine in the presence of STAT3 inhibitor (STATTIC) and found that inhibiting the STAT3 by STATTIC significantly suppressed the nicotine-induced STAT3 activation (Fig. 3 E). The Arg1 gene is an M2 marker and is known to be regulated by STAT3 (Yi and Kim, 2017). We found that nicotine-induced up-regulation of the Arg1 gene promoter was significantly suppressed by STATTIC treatment (Fig. 3 F). Moreover, inhibition of STAT3 reversed the nicotine-mediated suppression of Iba1⁺/F4/80⁺ (M1) and promotion of Iba1⁺/CD206⁺ cells (M2; Fig. 3, G and H). These results suggest that the nicotine promotes M2 microglial polarization via activation of the STAT3 pathway.

To examine the effect of nicotine-induced M2 microglia on tumor progression, we treated lung cancer brain metastatic cells, H2030BrM and PC9BrM, with conditioned medium (CM) generated from human microglia that were treated with or without nicotine. We found that CM derived from nicotine-treated microglia significantly increased sphere formation (Fig. 4 A) and population of ESA⁺/CD44⁺ or aldehyde dehydrogenase (ALDH⁺) cancer stem cells (CSCs; Fig. 4 B). The expression of stemness-inducing genes, SOX2 and NANOG, was also significantly up-regulated when the brain metastatic cells were treated with the CM of nicotine-treated microglia (Fig. 4, C and D). Furthermore, nicotine-treated microglial CM also promoted the colony-forming ability of both H2030BrM and PC9BrM cells (Fig. 4 E). We then sought to identify the factor(s) in CM that are responsible for the increase in stemness. We performed the cytokine array analysis (consisting of 99 factors) and found that nicotine significantly augmented the secretion of CCL20 and IGF-1 in the microglial CM (Fig. S3 A). The results of screening were verified using human microglial cells (Fig. S3 B). Furthermore, blocking STAT3 by STATTIC significantly suppressed nicotine-mediated up-regulation of CCL20 and IGF-1 (Fig. S3 B). We observed that microglia produced higher levels of IGF-1 and CCL20 than macrophage (Fig. S3 C). The expression of receptors of IGF-1 and CCL20 were significantly increased in H2030BrM cells after the treatment with CM of microglia that were exposed to nicotine (Fig. 4 F). We then treated lung cancer cells with recombinant CCL20 and IGF-1 and examined their effect on CSC properties. We found that both cytokines significantly enhanced the sphere formation (Fig. 4 G), ESA⁺/CD44⁺ CSC population (Fig. 4 H), and mRNA expression of CSC markers SOX2 and NANOG (Fig. 4, I and J). In addition, IGF-1 and CCL20 significantly increased the colony-forming ability of brain metastatic lung cancer cells (Fig. 4 K). To understand the clinical relevance of these results, we used The Cancer Genome Atlas (TCGA) dataset and found that high expression of CCL20 was associated with poor overall survival in lung cancer patients (Fig. S3 D); however, no such correlation was observed for the IGF-1 gene (Fig. S3 E). Furthermore, expressions of both CCL20 and its receptor, CCR6, were up-regulated in the primary tumor site of lung cancer patients compared with normal lung tissue in the same dataset (Fig. S3, F and G). Moreover, microglia expressed a higher level of CCL20 mRNA than lung cancer cells, indicating that microglia are potentially major sources of CCL20 in brain TME (Fig. S3, H and I). To examine whether CCL20 is involved in the growth-promoting effect of CM derived from nicotine-treated microglia, CCL20 was depleted in the CM using CCL20 neutralizing antibody. Our result indicates that depletion of CCL20 significantly suppressed the generation of CSC by nicotine-treated CM (Fig. 4 L). To further validate our results, we examined the levels of CCL20 in brain and serum of the nicotine-treated mice represented in Fig. 2 E and found that CCL20 indeed increased in the brain and serum of nicotine-treated animals (Fig. 4, M and N). In contrast, depletion of microglia by PLX3397 significantly suppressed the levels of nicotine-induced CCL20 in brain and serum of nicotine-treated animals (Fig. 4, M and N). Furthermore, consistent with our in vivo results, we found that the serum level of CCL20 was significantly higher in lung cancer patients who were current smokers compared with noncurrent smokers (Fig. 4 O).

We also examined the direct effect of nicotine on cell growth by treating the human lung cancer brain metastatic cells, H2030BrM, with nicotine alone or with nicotine-treated microglial CM. Although colony-forming ability increased in both groups (Fig. S3 J), the CSC population and expression of stem cell markers were significantly elevated only in cells treated with the nicotine-treated microglial CM (Fig. S3, K and L). Consistent with these results, treating the mouse lung cancer cell, LL/2, with CM derived from nicotine-treated mouse microglia significantly increased the levels of CSCs (Fig. S3 M) and promoted the colony-forming ability (Fig. S3 N) and tumor cell viability (Fig. S3 O). We also examined whether nicotine-induced brain metastasis is mediated via M2 microglial polarization by STAT3 activation. We blocked the M2 polarization of microglia by STATTIC and examined the expression of IGF-1 and CCL20 by quantitative real-time PCR (qRT-PCR). We found that the nicotine treatment significantly up-regulated the IGF-1 and CCL20, while STATTIC reversed the effect (Fig. S3, P and Q). The CM was then added to the culture of LL/2 cells. We found that CM derived from microglia that were treated with both nicotine and STATTIC significantly suppressed CSC population (Fig. S3 R) and tumor cell viability (Fig. S3 S).

Microglia are known to exhibit phagocytic activity against brain tumor cells, like macrophages (Hutter et al., 2019). However, to escape from microglia-mediated phagocytosis, tumor cells express CD47 on their cell surface, which activates so-called “do not eat me” signaling. To investigate if nicotine affects this immune function of microglia, we measured the expression of signal-regulatory protein α (SIRPα) on microglia and CD47 on tumor cells. We found that the mRNA and protein expression of SIRPα were up-regulated in human microglia after nicotine treatment (Fig. 5 A). We also found that nicotine significantly increased the mRNA and protein expression of CD47 in both H2030BrM and PC9BrM cells (Fig. 5 B). Importantly, the expression of CD47 is up-regulated in brain metastasis of lung cancer patients with smoking history (Fig. 5 C). We then examined the effect of nicotine on phagocytic ability of microglia. We prelabeled H2030BrM cells with the fluorochrome-PKH26, a red-fluorescent vital dye that binds tightly to cell membranes (Munn and Cheung, 1990). The labeled cells were incubated with microglia that were treated with or without nicotine. We found that the size of PKH26⁺ tumor cells and microglia cells was similar at 0 h in the co-culture (Fig. 5 D). However, after 24 h of incubation, the small and irregular shape of PKH26⁺ cells was accumulated in the microglia. This result suggests that tumor cells were rapidly degraded by nicotine-treated microglia through phagocytosis. As shown in Fig. 5, D and F, both human and mouse microglial cells showed strong phagocytic activities toward the H2030BrM and LL/2 cells. Importantly, this phagocytic ability was greatly compromised in the presence of nicotine (Fig. 5, E and G). Blocking STAT3 signaling by STATTIC treatment suppressed nicotine-mediated down-regulation of phagocytic ability of microglia (Fig. 5 E). We also examined the expression of SIRPα in microglia of the nicotine-treated mouse brain represented in Fig. 2 E and found that the CD47 and SIRPα expression were strongly elevated in the brain of mice treated with nicotine (Fig. 5, H–J). These results suggest that nicotine promotes M1/M2 conversion and suppresses the innate immune function of microglia, which promotes brain metastasis progression.

Our results strongly suggest that blocking M2 polarization of microglia may be an effective therapeutic approach for brain metastasis. Therefore, we screened a natural compound library to identify compounds that block M2 macrophage polarization using an Arg1 promoter reporter system. We screened a total of 103 compounds that are known to be blood–brain barrier (BBB) permeable. As shown in Fig. S4 and Fig. 6 A, we identified three compounds (>15-fold change, P < 0.001) that effectively suppressed the nicotine-induced activation of Arg1 promoter at a concentration of 1 µM. We selected the top compound, PTL, which suppressed the Arg1 promoter activity by >20-fold, for further study (Fig. 6 A). PTL is an abundant sesquiterpene lactone found in the medicinal herb Feverfew (Tanacetum parthenium). It has been used to treat inflammation and headache for centuries in Europe (Liu et al., 2018). To confirm the suppressive effect of PTL on nicotine-induced M2 microglial polarization, microglia were treated with nicotine in the presence or absence of PTL (Fig. S5 A), and expressions of M1 and M2 markers were examined. We found that PTL significantly decreased the nicotine-mediated M2 polarization of microglia (Fig. 6 B). Furthermore, the nicotine-mediated reduction of Iba1⁺/CD11b⁺ cells (M1) and elevation of Iba1⁺/CD206⁺ cells (M2) were significantly reversed by PTL (Fig. 6 C). Previous studies have shown that PTL modulated the JAK2/STAT3 signaling in response to cytokine stimulations (Liu et al., 2018; Sobota et al., 2000). We indeed found that PTL suppressed the nicotine-induced JAK2/STAT3 activation (Fig. 6 D), suggesting that PTL blocks the nicotine-induced M2 microglial polarization by targeting STAT3. Furthermore, the nicotine-induced suppression of phagocytic ability of microglia was restored when cells were treated with PTL (Fig. 6 E). The H2030BrM cells were treated with CM of microglia that were exposed to nicotine only or nicotine plus PTL. We found that the ESA⁺/CD44⁺ CSCs (Fig. 6 F) and colony-forming ability (Fig. 6 G) were significantly suppressed by CM of microglia treated with nicotine plus PTL. We also examined the mRNA expression of CCL20 and found that the nicotine-mediated secretion of CCL20 was down-regulated by the PTL treatment (Fig. 6 H). Next, to investigate the effect of PTL on nicotine-related brain metastasis in vivo, wild-type mice (C57BL/6 and BALB/c) were transplanted with LL/2 mouse lung cancer cells followed by administering nicotine (1 mg/kg) only or nicotine plus PTL (1 mg/kg) via an intraperitoneal injection. As shown in Fig. 6, I-K (BALB/c) and Fig. S5, B and C (C57BL/6), PTL treatment significantly suppressed nicotine-mediated brain metastasis and prolonged the brain metastasis–free survival (Fig. 6 L). To examine whether the PTL treatment also reverses microglia polarization in the brains of these mice, we performed flow cytometry (FACS) analysis of mouse brain and stained brain sections for M1 and M2 markers. A significant increase in Iba1⁺/F4/80⁺ (M1 marker) and decrease in Iba1⁺/CD206⁺ (M2 marker) were observed in brain microglia of PTL-treated mice (Fig. 6, M and N; and Fig. S5, D and E). To validate our results in human cell line, we used the xenograft model of H2030BrM cell and tested the efficacy of PTL. As shown in Fig. S5, F–K, PTL significantly suppressed the nicotine-induced brain metastasis and M2 microglia polarization. PTL treatment did not affect body weight of mice and showed no detectable liver toxicity (Fig. S5 L). We also examined the direct effect of PTL on cell viability and CSC population in vitro (Fig. S5 M). Our results showed that PTL suppressed the abundance of CD44⁺/ESA⁺ CSC and cell viability at the concentration of 10–20 µM, but not at 1 µM, which is the dose used throughout our in vitro experiments (Fig. S5, N and O), indicating that the anti-tumor effect of PTL is not mediated by direct cytotoxic effect, and it is rather mediated by the effect on microglia. Furthermore, PTL significantly suppressed stemness promoting effect of nicotine on microglia (Fig. S5 P) and the expression of stemness-related genes (Fig. S5, Q and R). These results suggest that PTL suppresses nicotine-related brain metastasis of lung cancer by skewing polarity of M1 microglia and up-regulating their phagocytic ability.

The incidence of brain metastasis among lung cancer patients is 10–36% (Bacha et al., 2018; Barnholtz-Sloan et al., 2004; Schouten et al., 2002), although these data are considered to be underestimated (Bacha et al., 2018; Schouten et al., 2002). The majority of the lung cancer patients are diagnosed with advanced stage disease, most with stage IV cancer. More than 90% of lung cancer deaths are attributed directly to the metastatic disease and include those brain metastases (Ali et al., 2013; Kakusa et al., 2018). Cigarette smoking is a major risk factor for lung cancer (Lee et al., 2012), and our previous study showed a negative impact of smoking on lung cancer brain metastasis (Shenker et al., 2017). In the present study, we have found a significantly higher incidence of brain metastasis among current smokers with stage IV lung cancer.

Nicotine is not a carcinogen, but it causes addiction by activating nAch receptor, which is abundantly expressed by various brain cells including microglia. Direct tumor promoting effects of nicotine have been previously reported (Schaal and Chellappan, 2014; Sun et al., 2017; Wang et al., 2017; Yoneyama et al., 2016); however, how nicotine promotes tumor by activating TME is still poorly understood. The current report is the first to show the critical impact of nicotine on brain TME to promote brain metastasis by skewing microglia to M2 phenotype and suppressing their role in innate immunity. Our results indicate that the phenotypic conversion of microglia by nicotine is mediated through the nAch receptor–STAT3 pathway, and that inhibition of STAT3 blocked the M2 conversion, suppressed the secretion of pro-tumor M2 cytokines, and reactivated their phagocytic activity toward tumor cells (Fig. 6 O).

Previous studies showed that microglial cells and macrophages in the brain are derived from two different sources, parenchymal resident microglia and monocytes/macrophages that enter the brain from the bone marrow (Wu and Watabe, 2017). Although microglia and macrophage cells show similarity in their surface markers and physiological functions, recent studies have demonstrated that microglia and macrophages are two distinct myeloid populations with different developmental origins (Gomez Perdiguero et al., 2015; Prinz and Priller, 2014; Schulz et al., 2012). Microglial cells originate from erythromyeloid progenitors that begin on embryonic day (E) 7.5–E8.0. in the blood islands of the yolk sac. Until E9.5, erythromyeloid progenitors migrate to the developing central nervous system and mature into microglia (Gomez Perdiguero et al., 2015; Prinz and Priller, 2014; Schulz et al., 2012). These early microglial cells reside in the brain throughout life and are thought to sustain the local microglial population. In contrast, macrophages originate from the hematopoietic stem cells that start in the aorta–gonad–mesonephros region at E10.5. and then in the fetal liver at E12.5. After the postnatal stage, macrophages are produced from monocytes in the bone marrow (Gomez Perdiguero et al., 2015; Prinz and Priller, 2014; Schulz et al., 2012). Furthermore, recent studies demonstrated that microglial cells and macrophage have distinct and specific surface antigens (Feng et al., 2015). Studies on CX3CR1 (⁺/GFP)/CCR2 (⁺/RFP) knock-in fluorescent protein reporter mice revealed that microglial cells are CX3CR1⁺/CCR2⁻/CD45⁻, while the macrophages are CX3CR1⁻/CCR2⁺/CD45⁺ (Feng et al., 2015), strongly supporting the notion that microglial cells and macrophages are from different populations and they can be distinguished. Badie and Schartner performed flow cytometric analysis and characterized the distribution of microglial cells and macrophages in experimental gliomas and found that microglial cells (CD11b/c^high, CD45^low), mainly present at the site of tumor or tumor periphery, accounted for 13–34% of the tumor mass. In contrast, macrophages (CD11b/c^high, CD45^high) were less prominent within the tumors or the tumor periphery and accounted for just 4.2–12% of the tumor mass (Badie and Schartner, 2000). These results also suggest that microglial cells play a key role in mediating brain tumor progression.

We have also found that nicotine dramatically increased the mRNA expression of α4β2 receptor on microglia cell compared with macrophage. These results suggest that microglia is more responsive to nicotine compared with macrophage, resulting in increase in the incidence of brain metastasis. Furthermore, it was reported that the numbers of macrophage/microglia were significantly higher in the central nervous system than in the peripheral system after injury (Leskovar et al., 2000). In addition, the peak concentrations of three cytokines, TNFα, IL-1, and IL-6, appeared earlier and were significantly higher in the injured central nervous system than peripheral system (Leskovar et al., 2000), suggesting differences in the reparative responses of macrophage/microglia in the PNS and central nervous system to injury.

In this study, we showed that blocking microglia CSF1R by PLX3397 significantly suppressed nicotine-mediated brain metastasis, suggesting that cancer cells lose growth support function from M2 microglia. Previous studies showed that PLX3397 reprogrammed tumor-infiltrating macrophages, enhanced antigen presentation (Zhu et al., 2014), and suppressed TAM infiltration in the tumor (Butowski et al., 2016; Cannarile et al., 2017; Cuccarese et al., 2017). We also found that PLX3397 skewed the population of microglia/macrophage to M1 phenotype. Therefore, while blocking the proliferation of microglia/macrophage, PLX3397 also promotes the residual microglia’s innate immune function to tumor cells. Our results indicate that nicotine increased the secretion of IGF-1 and CCL20 in the CM of microglial cells. IGF-1 is known to enhance tumor cell survival and proliferation via activation of the PI3K-Akt signaling (Park et al., 2016; Zhu et al., 2011). However, the role of CCL20 on tumor progression has not yet been well defined. CCL20 is a critical chemoattractant responsible for the recruitment of inflammatory immune cells (Nandi et al., 2014). The expression of CCL20 has been shown to be overexpressed in a variety of tumors including lung cancer (Liu et al., 2016; Wang et al., 2015, 2016). In pancreatic cancer, CCL20 has been found to be overexpressed in metastatic tumor tissues (Liu et al., 2016). It was reported to promote epithelial–mesenchymal transition and invasive ability of tumor cells by activating the MAPK and PI3K pathways and matrix metallopeptidase signaling (Marsigliante et al., 2016; Wang et al., 2016). Previous studies showed that CCL20 promotes cell proliferation through activation of MAPK pathway (Lu et al., 2017) and up-regulating transcription of c-Fos and c-Myc. Others have shown that CCL20 enhanced the chemotherapy resistance of triple-negative breast cancer by promoting ALDH⁺ population of breast cancer cells through PKCζ- and p38 MAPK-mediated NF-κB activation (Chen et al., 2018). We found that stem cell genes NANOG and SOX2 were up-regulated in CCL20-treated cells, suggesting that CCL20 promoted the stemness of lung cancer brain metastatic cells via NANOG/SOX2. One of the major sources of CCL20 is a tumor-associated macrophage, especially M2 macrophage (Liu et al., 2016). In this study, we found that nicotine strongly promoted phenotypic conversion of microglia to M2 and enhanced secretion of CCL20 to stimulate tumor cell growth and cancer cell stemness. The depletion of microglia in our animal model significantly reversed the effect of nicotine on brain metastasis. Furthermore, we showed that the effect of CCL20 on tumor stem cell was mediated by up-regulation of SOX2 and NANOG genes. These results suggest that CCL20 may serve as a novel therapeutic target and biomarker for brain metastasis of lung cancer.

SIRPα is expressed predominantly in myeloid and neuronal cells (Adams et al., 1998; Kharitonenkov et al., 1997). Normal cells express CD47 that interacts with SIRPα on macrophage/microglia to suppress their phagocytic activity. Several studies have shown that the expression of SIRPα is also associated with polarization of microglia/macrophage (Pan et al., 2013). Furthermore, knockdown of SIRPα in macrophages increased the production of M1 cytokines in the tumor environment (Pan et al., 2013). Blocking SIRPα also triggered the Akt survival signal in macrophages and enhanced macrophage migratory activities toward tumors (Pan et al., 2013), suggesting that SIRPα also affects microglia’s function. In this study, we found that the nicotine increased the expression of SIRPα on microglia, while nicotine also promoted the expression of CD47 on the tumor cells, resulting in suppression of phagocytic ability of microglia. There are two possible mechanisms that are involved in the anti-tumor effect of microglia. One is direct internalization of tumor cells into the lysosome of microglia by phagocytosis. The other possibility is that tumor cells are first killed by cytotoxic factors released by microglia followed by phagocytosis of the apoptotic debris by microglia. In addition, it is noteworthy that the CD47-SIRPα signal was previously found to modulate T cell differentiation and regulate T cell priming to the tumor site (Liu et al., 2015b; Zhang and Petro, 1996). Therefore, the CD47-SIRPα signal that was activated by nicotine plays a pivotal role in modulating anti-tumor innate as well as adaptive immune responses in TME.

By screening a BBB permeable natural compound library, we identified PTL as a potent inhibitor for nicotine-induced M2 conversion of microglia. PTL is an abundant sesquiterpene lactone found in the medicinal herb Feverfew (T. parthenium). It has been used for centuries as a traditional medicine to treat arthritis, fever, and headache (Liu et al., 2015a; Rummel et al., 2011). PTL was also previously shown to exhibit direct tumor suppressive activity (Talib and Al Kury, 2018) by modulating multiple pathways, including p53, vascular endothelial growth factor (Talib and Al Kury, 2018), B-Raf/MAPK/Erk (Lin et al., 2017), and STAT3 pathways (Carlisi et al., 2011; Skoumal et al., 2011; Song et al., 2014). However, these direct tumor cytotoxic effects mediated by PTL require a relatively high dose, typically >10 µM. In this study, we found that sub-micromolar dose (<1 µM) of PTL effectively suppressed nicotine-induced polarization of microglia to M2 phenotype and enhanced its M1 phenotype. On the other hand, the low concentration (1 µM) of PTL did not affect the population of CSC or tumor cell growth, which strongly suggests that the anti-tumor effect of PTL on nicotine-promoted brain metastasis shown in our study is mediated via modulation of microglial function and not by direct effect of PTL on tumor cells. Importantly, PTL was shown to be able to cross the BBB due to high membrane permeability (Könczöl et al., 2013). Considering the inhibitory effect of PTL on nicotine-induced microglial activation and their minimum toxicity at the effective dose, PTL could be used for the prevention and treatment of brain metastasis, particularly for patients with past and current smoking history.

Nicotine is highly addictive and considered as the major obstacle for smoking cessation (Henningfield et al., 1991). Unfortunately, many cancer patients have difficulty in quitting smoking even after their diagnosis due to nicotine addiction. E-cigarette, nicotine patch, and nicotine gum are nicotine replacement therapies for smoking cessation used by these patients (Hartmann-Boyce et al., 2018; Stead et al., 2012). Although nicotine, per se, is not a carcinogen, many reports document its tumor-promoting effects (Li et al., 2017; Schaal and Chellappan, 2014; Schaal et al., 2018) as well as its pro-inflammatory effect, which suppresses the phagocytosis ability of the alveolar macrophages (Scott et al., 2018). Considering our results that clearly demonstrated profound and long-term effects of nicotine on brain metastasis progression, extra caution should be paid in the use of nicotine for smoking cessation.

We used a cohort of patients with a diagnosis of nonsmall cell lung cancer with or without brain metastases treated at Wake Forest Baptist Comprehensive Cancer Center between May 2015 and April 2018. We collected a total of 281 stage IV nonsmall cell lung cancer patients with or without a history of smoking. Among them, 79 patients were diagnosed with brain metastasis. 59 patients were current smokers at the time of diagnosis and treatment. We defined the “never smokers” (n = 44) as those who have no smoking history and those who had smoked <10 pack/year but stopped smoking. The “former smokers” (n = 181) are defined as those who smoked >10 pack/year but quit smoking in the past. Patient characteristics included histology status, smoking history, pack/years, progression status, and date of death. Patient information about age, gender, cancer type, and smoking status are described in Fig. S1. All samples used in this study were obtained under the approval of the institutional review boards of Wake Forest School of Medicine Institutional and Wake Forest Baptist Comprehensive Cancer Center. Written informed consent was obtained from all participants.

Human lung cancer cell lines, H2030BrM and PC9BrM, were kind gifts from J. Massagué (Memorial Sloan-Kettering Cancer Center, New York, NY). H2030BrM and PC9BrM cell lines were authenticated by using GenePrint 10 STR System (Promega, B9510). SIM-A9 was obtained from Kumi Nagamoto-Combs, through Kerafast.com. HMC3 and LL/2 cells were purchased from American Type Culture Collection (ATCC) and were authenticated by ATCC. The H2030BrM and PC9BrM were cultured in RPMI supplemented with 10% FBS, streptomycin (100 mg/ml), and penicillin (100 U/ml). LL/2 cells were cultured in DMEM medium supplemented with 10% FBS. SIM-A9 and HMC3 were cultured in DMEM/F12 medium supplemented with 5% FBS. All cells were grown at 37°C under 5% CO₂. All cell lines were obtained between 2010 and 2016, and they were ensured to be negative of mycoplasma. The HMC3 and SIM-A9 cells were seeded in a 10-cm dish. After reaching 70% confluence, cells were incubated with or without 1 µM nicotine in DMEM/F12 medium supplemented with 2% FBS or in medium containing nicotine plus the STAT3 inhibitor STATTIC (0.5 µM). After 24 h, cells were washed twice with PBS and then incubated in the fresh DMEM/F12 medium supplemented with 2% FBS for 24 h. The CM harvested from the cell culture were centrifuged at 300 g for 10 min to remove the cells and stored at −80°C. To investigate the effect of nicotine-derived CM on tumor progression, we harvested CM from microglial cells that were treated with or without nicotine, followed by treatment of tumor cells with the CM. All cell lines were tested by using universal mycoplasma detection kit (ATCC, 30-1012k, lot 70008746).

Animals were treated in accordance with the U.S. National Institutes of Health Animal Protection Guidelines, and the protocol was approved by the Wake Forest Baptist Health Institutional Animal Care and Use Committee. Female nude, BALB/c, and C57BL/6 mice of 5–6 wk of age were used. The mice were housed (four to five per cage) at a stable temperature (24 ± 1°C), in a 12-h light/dark cycle, and with unrestricted access to food and water. Several studies showed that human lung adenocarcinomas cell H2030BrM and mouse Lewis lung carcinoma cell LL/2 exhibited their highly metastatic abilities in the nude, C57BL/6, and BALB/c mice, respectively (Chen et al., 2016; Janker et al., 2018; Kamei et al., 2009; Malladi et al., 2016; Nguyen et al., 2009; Scholar et al., 1989). Actively growing luciferase-labeled H2030BrM and LL/2 cells were intracardially injected at a concentration of 2 × 10⁵ H2030BrM cells or 5 × 10⁴ LL/2 cells in 100 µl PBS into the left cardiac ventricle. To confirm a successful injection, the photon flux from whole body of the mice was immediately measured using IVIS Xenogen bioimager (Caliper). To examine the effect of nicotine, the mice received nicotine (1 mg/kg) by intraperitoneal injection every 3 d. To examine the role of microglia on nicotine-related brain metastasis, the mice were administered PLX3397 (20 mg in 7 µl, Selleck Chemicals USA), a selective inhibitor of CSF1R (also known as macrophage colony-stimulating factor receptor), which reduces viability of microglial cells. To avoid whole-body effect, the PLX3397 was directly injected into the right mouse brain after 1 wk of transplantation of mouse lung cancer cell LL/2. Mice were divided into four groups: (1) tumor transplant only, (2) tumor transplant plus nicotine, (3) tumor transplant plus PLX3397, and (4) tumor transplant plus PLX3397 and nicotine. To examine the effect of natural compound (PTL) on modulation of microglial cell polarization, the mice were administered PTL (1 mg/kg) plus nicotine by intraperitoneal injection. Mice were divided into three groups: (1) tumor transplant only, (2) tumor transplant plus nicotine, and (3) tumor transplant plus PTL and nicotine. Tumor growth was monitored by bioluminescence until day 60. For bioluminescent imaging, the mice were injected with D-luciferin intraperitoneally (100 mg/kg), followed by capturing images every week using IVIS Xenogen bioimager. The brain metastasis was monitored, and the luminescence was quantified once per week. Finally, whole brain was removed and incubated in PBS with 0.6 mg/ml luciferin for 5 min, and photon flux was measured by an in vivo image system (IVIS).

Mice were anesthetized with a ketamine/xylazine (90–120/7–10 mg/kg) mixture and fixed into a stereotactic head frame (Kopf stereotactic frame). A 1.5-mm bur hole was drilled 1 mm anterior to the coronal suture on the right hemisphere and 2 mm lateral from the midline. A Hamilton syringe fixed onto the head frame was used to inject LL/2 (2 × 10⁵ cells/5 µl) into the right frontal lobe of the brain. The skin incision was then closed with 4–0 silk thread. The ketoprofen (2 mg/kg) was given to the mice to reduce pain. After 1 wk of tumor transplantation, 36 mice were randomly divided into four groups: (1) tumor transplant only (control group), (2) tumor transplant plus nicotine, (3) tumor transplant plus PLX3397, and (4) tumor transplant plus PLX3397 and nicotine. The PLX3397 and PLX3397 plus nicotine groups received PLX3397 (20 mg in 7 µl) treatment by intracranial injection every 2 wk at the same area of tumor transplantation. The control group received the same volume of PBS.

To examine the nicotine-induced M2 microglial polarization and examine the effect of natural compounds on microglial polarization, SIM-A9 cells were first infected with lentivirus containing GFP gene, and GFP⁺ cells were sorted by flow cytometry (FACS). The GFP⁺ SIM-A9 cells were seeded in 96-well plates for 1 d and were transfected with the Arg1 promoter reporter plasmid (Addgene) using Lipofectamine 2000 (Invitrogen). After 24 h, the cells were treated with only nicotine or nicotine plus natural compound in presence of 2% FBS and cultured for another 24 h. Cells were washed twice and were treated with D-luciferin for 5 min. The expression of luciferase was detected by using an IVIS Xenogen bioimager. For Arg1 promoter luciferase normalization, the photon flux was divided by GFP signal, which was measured by a Multi-Mode Reader (Biocompare). Each experiment was conducted a minimum of three times.

For qRT-PCR, the treated cells were washed with PBS and lysed using the TRIzol reagent. The isolated total RNA was quantified by a NanoDrop 2000 (Thermo Fisher Scientific). Then, RNA was reverse transcribed to cDNA using a Reverse Transcription Supermix kit (Bio-Rad). The cDNA was then amplified with a pair of forward and reverse primers for the following genes: hCD74-F 5′-GCG​ACC​TTA​TCT​CCA​ACA​ATG-3′, hCD74-R 5′-GTC​CTC​CAG​TTC​CAG​TGA​CT-3′, hCD80-F 5′-CTC​TTG​GTG​CTG​GCT​GGT​CTT​T-3′, hCD80-R 5′-GCC​AGT​AGA​TGC​GAG​TTT​GTG​C-3′, hCD86-F 5′-CCA​TCA​GCT​TGT​CTG​TTT​CAT​TCC-3′, hCD86-R 5′-GCT​GTA​ATC​CAA​GGA​ATG​TGG​TC-3′, hArginase-1-F 5′-GCG​GGC​GGA​CGC​TGG​CG-3′, hArginase-1-R 5′-GCA​GCG​GGA​CCA​TGC​TCC​ACT​C-3′, hArginase-2-F 5′-AGA​CGC​GAG​TGC​ATT​CCA​TC-3′, hArginase-2-R 5′-CAC​TAA​TGG​TAC​CGA​TTG​CCA-3′, hCD204-F 5′-GGT​TTC​AAT​TGT​AAA​GAG​AGA​GAA​G-3′, hCD204-R 5′-CTG​AGC​AAT​TCT​TCG​TTT​CCC-3′, mCD74-F 5′-AGA​GCC​AGA​AAG​GTG​CAG​C-3′, mCD74-R 5′-GAT​GCA​TCA​CAT​GGT​CCT​GG-3′, mCD80-F 5′-ACG​TAT​TGG​AAG​GAG​ATT​ACA​GCT-3′, mCD80-R 5′-TCT​GTC​AGC​GTT​ACT​ATC​CCG​C-3′, mCD86-F 5′-TTA​CGG​AAG​CAC​CCA​CGA​TG-3′, mCD86-R 5′-CCT​GTT​ACA​TTC​TGA​GCC​AGT-3′, hCHRNA7-F 5′-CGT​GGA​TGA​GAA​GAA​CCA​AG-3′, hCHRNA7-R 5′-CCA​TTG​GGG​ATA​TAG​CCA​CT-3′, hSOX2-F 5′-GGG​AAA​TGG​GAG​GGG​TGC​AAA​AGA​GG-3′, hSOX2-R 5′-TTG​CGT​GAG​TGT​GGA​TGG​GAT​TGG​TG-3′, hOCT4-F 5′-GCT​GGA​GAA​GGA​GAA​GCT​GGA​GC-3′, hOCT4-R 5′-GCT​AAG​CTG​CAG​AGC​CTC​AAA​GC-3′, hNanog-F 5′-CTT​GGA​AGC​TGC​TGG​GGA​AG-3′, hNanog-R 5′-GAT​GGG​AGG​AGG​GGA​GAG​GA-3′, hCD47-F 5′-GCT​AAA​ATA​TCG​TGT​TGT​TTC​ATG​GTT​TTC-3′, hCD47-R 5′-GCT​GTA​CTA​AAC​ACA​TAG​TAG​TGA​AGT-3′, hSIRP-F 5′-GCC​TTA​GTC​GTT​CGC​CCG​CG-3′, hSIRP-R 5′-GCC​GGC​TTG​GCC​CTA​CTC​CT-3′, hIGF-1-F 5′-GCT​GTG​CCT​GCT​CAC​CTT​CAC​CAG-3′, hIGF-1-R 5′-GCC​TCC​TTA​GAT​CAC​AGC​TCC​GGA​AG-3′, hCCL20-F 5′-TCT​CGC​CTC​CAG​CAT​GAA​AG-3′, and hCCL20-R 5′-CAG​ATC​TCC​TTG​GCC​ACA​ATG-3′.

PCR reactions were performed using iTaq Universal SYBR Green Supermix (Bio-Rad) and CFX Connect (Bio-Rad). The thermal cycling conditions composed of an initial denaturation step at 95°C for 5 min, followed by 35 cycles of PCR using the following profile: 94°C, 30 s; 60°C, 30 s; and 72°C, 30 s.

The human brain metastasis tissues of lung cancer were obtained from Wake Forest Baptist Comprehensive Cancer Center Tumor Tissue and Pathology Shared Resource. Tissue sections were stained using anti-CD206 (1:200, R&D Systems) for M2 microglia and anti-ionized calcium-binding adapter molecule-1 (Iba1; 1:1,000, Abcam) for microglia and CD45 (1:500, Abcam) for macrophage. Brain sections were then incubated with appropriate biotin-conjugated secondary antibodies and diaminobenzidine as the substrate. As a control, isotype antibodies were used. To examine the expression of SIRPα on microglia, 10-µm-thick frozen mouse brain sections were stained using rabbit anti-Iba1 (1:1,000; Abcam) for microglia and anti-SIRPα (1:200) (Cell Signaling). Brain sections were incubated for 1 h in a PBS solution containing 0.1% Triton X-100 and 3% bovine serum albumin at room temperature, and then transferred to a solution containing anti-SIRPα antibody overnight at 4°C. The secondary antibody conjugated with fluorescent dye Alexa Fluor 594 (1:1,000, Invitrogen) was used to detect the expression of SIRPα. Sections were then incubated overnight at 4°C in a solution containing anti-Iba1 antibodies. Antibodies against rabbit IgG conjugated with fluorescent dye Alexa Fluor 488 (1:1,000, Invitrogen) were used to detect the microglial cells. Fluorescence images were acquired using a microscope (BZ-X710, Keyence) connected to a computer equipped with imaging software (BZ-X Viewer). Appropriate isotype antibodies were used as controls. To determine the area of Iba1⁺ and CD206⁺ cells in brain metastasis, we chose three randomly selected fields in each tumor and measured staining intensity after taking images followed by using the Image-Pro software. We also chose three nontumor areas in the corresponding slide and measured intensity as background, and this background was subtracted from the intensity in the tumor area. The average intensity of three fields was used as the score of that patient. Therefore, each patient has a single score. We then divided the score range by setting the lowest score as 0 and the highest to be +3.

Cells were homogenized in the radioimmunoprecipitation assay buffer and then centrifuged at 17,000 g for 30 min at 4°C. The protein concentrations of the supernatants were determined by NanoDrop 2000. Supernatants (30 µg of total protein) were mixed with sample buffer containing 0.5 M dithiothreitol, heated to 80°C for 10 min, loaded into each well of 10% polyacrylamide gel, and resolved at 120 V for 2 h. The separated proteins bands were transferred to a polyvinylidene fluoride membrane, blocked with 5% milk, and probed with respective primary antibodies: total-STAT3 (1:1,000, Cell Signaling), phospho-STAT3 (1:1,000, Cell Signaling), GAPDH (1:10,000, Cell Signaling), SOX2 (1:500, Cell Signaling), OCT4 (1:1,000, Cell Signaling), NANOG (1:500, Cell Signaling), and SIRPα (1:1,000, Cell Signaling). The bound antibodies were detected using an enhanced chemiluminescence detection kit (GE Healthcare Lifescience). The band intensities were measured using an imaging system (Amersham Imager 600; UVP) and analyzed using ImageJ (1.43u; http://rsb.info.nih.gov/ij/). For gel loading control, membranes were incubated with monoclonal GAPDH antibody.

Mouse brain specimens were homogenized in radioimmunoprecipitation assay lysis buffer and centrifuged at 13,000 g for 30 min at 4°C. The protein concentrations of the supernatant were measured and adjusted to 1 mg/ml. The human serum of lung cancer patients was obtained from FOX CHASE Cancer Center. The serum was further diluted fourfold in assay buffer. The levels of CCL20 in the mouse brain supernatants and mouse and human serum were determined using mouse and human CCL20 ELISA kits (Sigma-Aldrich). Briefly, 96-well plates were first coated with anti-CCL20 monoclonal antibodies followed by addition of 100 µl of the supernatants and serum under investigation or the same volume of the CCL20 standards. The plate was incubated at 4°C overnight on a rocking platform. After washing the plates, the detection antibody and streptavidin solution were added to each well. After incubation, the 3,3′,5,5′-tetramethylbenzidine (TMB) and stop solution were added to each well. Finally, the plates were examined at an absorbance wavelength of 450 nm. Standard curves were obtained from values generated from known concentrations of mouse and human CCL20 provided by the kits.

Microglial cells (HMC3) were treated with or without nicotine (1 µM) plus STATTIC (0.5 µM) for 24 h. Cells were washed with PBS twice and incubated in the fresh DMEM/F12 medium supplemented with 2% FBS for an additional 24 h. The human monocyte/macrophage cells were treated with or without nicotine for 24 h. CM was collected and analyzed by the human IGF-1 ELISA kit (Sigma-Aldrich). Briefly, 96-well plates were first coated with anti–IGF-1 monoclonal antibodies followed by addition of 100 µl of the microglia conditional medium or the same volume of the IGF-1 standards. The plate was incubated at 4°C overnight on a rocking platform. After washing the plates, the detection antibody and streptavidin solution were added to each well. After incubation, the TMB and stop solution were added to each well. Finally, the plates were examined at an absorbance wavelength of 450 nm. Standard curves were obtained from values generated from known concentrations of mouse and human IGF-1 provided by the kit.

H2030BrM and LL/2 cells were labeled with PKH26 dye (Sigma-Aldrich). The labeled cells were then washed three times and cultured overnight to reduce nonspecific leaking of dye during the assay. Tumor cells were 95% viable after the incubation. Labeled tumor cells were mixed with microglia that were pretreated with or without nicotine. Co-cultured tumor and microglia cells in the culture slides were harvested after 24 h and fixed with 4% paraformaldehyde. Microglial cells were counterstained with anti-Iba1 antibody overnight. Secondary antibody conjugated with the fluorescent dye Alexa Fluor 488 (1:1,000, Invitrogen) was used to detect the microglia. The phagocytic activities were measured by immunofluorescence microscope and FACS. For quantification of phagocytosis, the phagocytic percentage was calculated as 100 × (percent Iba1⁺/PKH26⁺ cells/[percent Iba1⁺/PKH26⁻ cells + percent Iba1⁺/PKH26⁺ cells]).

Data are represented as mean ± SD. Significance was set at P < 0.05. The nicotine effect on M1/M2 gene expression, effect of nicotine-derived microglial CM on tumor growth, and effect of nicotine on microglia phagocytosis ability were analyzed using the unpaired Student’s t test. One-way ANOVA was used to analyze the effect of nicotine, PTL3397, or STATTIC treatment on M1/M2 gene polarization and tumor progression, the effect of different concentrations of IGF-1 or CCL20 on cancer progression, the effect of different natural compounds on M2 gene polarization, and the effect of nicotine or PTL treatment on M1/M2 gene polarization and cancer cell progression and nicotine effect in animal studies. Significance between groups was represented as *, P < 0.05; **, P < 0.01; and ***, P < 0.001.

Fig. S1 shows the effect of smoking and nicotine on brain metastasis. Fig. S2 demonstrates that the nicotine promotes brain metastasis in a mouse model. Fig. S3 shows that nicotine treatment of microglia increases secretion of CCL20 and IGF-1 followed by promotion of tumor cell stemness. Fig. S4 shows the effect of natural compounds on nicotine-mediated activity of Arg1 reporter. Fig. S5 shows PTL suppressed nicotine-promoted brain metastasis in syngeneic mouse model.

We thank Dr. Joan Massagué for providing the H2030-BrM cell line.

This work was supported by National Institutes of Health grants RO1CA173499, R01CA185650, and R01CA205067 (to K. Watabe). This research used Cancer Center Shared Resources, The Tumor Tissue and Pathology, Cell Engineering, Flow Cytometry, Biostatistics and Bioinformatics that are supported by the Comprehensive Cancer Center of the Wake Forest University National Cancer Institute, National Institutes of Health grant P30CA012197. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Cancer Institute.

Author contributions: S. Wu and K. Watabe designed the study and wrote the manuscript. S. Wu, F. Xing, and K. Wu conducted experiments and acquired, analyzed, and interpreted the data. S. Sharma, A. Tyagi, and K. Wu performed animal experiments. F. Xing and Y. Liu performed intracardiac injections. T. Lycan and A. Dothard performed clinical sample analysis. S. Wu, F. Xing, S. Sharma, A. Tyagi, K. Wu, Y. Liu, D. Zhao, R.P. Deshpande, Y. Shiozawa, T. Ahmed, W. Zhang, M. Chan, J. Ruiz, and K. Watabe reviewed and edited the manuscript and interpreted the data. K. Watabe supervised the study.