The HDAC2-SP1 Axis Orchestrates Protumor Macrophage Polarization

Cancer onset and progression depend not only on the tumor cell phenotype and its functional characteristics but also on the tumor microenvironment (TME), including innate and adaptive immune cells (1). Macrophages are the most abundant innate immune cells in the TME, which may either hinder or support tumor growth depending on their phenotype. Although the macrophage phenotype is typically discussed at a single point in time, it is important to understand that environmental changes such as cytokine and growth factor secretion, inflammation, infection, injury, hypoxia, and other conditions can cause phenotype switching. Moreover, macrophage phenotype profiling in TME showed that many TAM subgroups display a mixture of antitumorigenic (M1) and protumorigenic (M2) cell markers, which have yet to be fully identified (2). Using IHC and multiplex immunofluorescence staining, we previously demonstrated that TAM density and spatial distribution as well as gene-expression profiles and associated phenotypic features are prognostic factors for overall survival in lung cancer (3). We also demonstrated that tumor cell–TAM cross-talk alters the transcriptional programming of macrophages and switches antitumor M1 TAMs to protumor M2 TAMs, thereby supporting tumor development (4).

The transcriptionally dynamic nature of macrophages is a precondition for adapting fast to changing microenvironmental conditions, e.g., by transitioning rapidly from a resting to a stimulated state by polarizing into dedicated and specialized subsets, or by interconverting between subsets. This versatility affects abundance, spatial distribution, and differentiation potential of the macrophages (5). Accordingly, there are some commonalities in overall gene and transcription factor (TF) expression but also distinct differences in gene regulatory and phenotypic features between these subsets (6, 7). Such genomic and phenotypic versatility of macrophages demands coordinated remodeling of epigenome and chromatin as underlying mechanisms, which are not yet well understood.

Histone deacetylases (HDAC) catalyze the removal of acetyl groups from lysine residues on histones and nonhistone proteins. HDACs are encoded by a family of 18 genes that fall into four classes. The classic HDACs are class I (HDAC1, 2, 3, 8), class IIa (HDAC4, 5, 7, 9), class IIb (HDAC6, 10), and class IV (HDAC11), and seven members of sirtuins are referred to as class III HDACs (8). The role of HDAC-mediated epigenetic modulation of macrophages has been addressed in several studies, but most of these studies have been executed using classic M1 macrophages in the context of infectious diseases (9). So far, only a few studies have directly or indirectly demonstrated that the functions of macrophages in the TME could be fine-tuned via histone acetylation/deacetylation profiles (10). For example, miRNA-145–mediated downregulation of HDAC11 induces an M2-like macrophage phenotype in colorectal carcinoma (11). Whereas tumor-infiltrating leukocyte-specific USP24 induces NFKB1 and IL6 in TME by stabilizing P300 and β-TrCP, which increases the expression of these genes in M2 macrophages to promote tumor growth (12). Further, the ACKR3/STAT1/HDAC axis plays a major role in M2 macrophage migration and immune escape in hepatocellular carcinoma (13).

Many individual or class-specific HDAC inhibitors are being tested in anticancer clinical trials to determine their safety and efficacy (14). Whereas these studies largely focus on direct pharmacologic effects on cancer cells, the impact of such an approach on tumor-associated innate and adaptive immune cells has largely been neglected. Exploration of the pharmacologic effects of HDAC inhibitors on tumor-associated immune cells in general, and TAMs in particular, might well reveal their full potential as antitumor agents. Exploiting the differences between protumorigenic and antitumorigenic macrophages using these modulators might provide a means for specifically targeting M2-like TAMs in the TME, thereby blocking these key tumor-supporting cells. Considering the complex, context-dependent, temporally and spatially restricted, and cell- or tissue-specific nature of HDAC expression, precise ex vivo and in vivo testing of individual or class-specific HDAC manipulation in specific TME is, however, crucial for exploitation of their therapeutic immunomodulatory potential. In the present study, we took up this challenge by focusing on the impact of a single-HDAC, i.e., HDAC2, on lung cancer TAMs in mouse and human coculture systems, primary cells, and lung tumor models.

Human lung adenocarcinoma cells (A549) and mouse Lewis lung carcinoma cells (LLC1) were obtained from ATCC. Human umbilical vein endothelial cells (HUVEC) were obtained from PELOBiotech. Cells were cultured as follows: A549 in DMEM/F12 supplemented with 10% FCS, and 1% penicillin/streptomycin (1% P/S; Invitrogen); HUVECs in EGM-2 Bulletkit (Lonza) supplemented with 2% FCS and VEGF until passage seven, LLC1 in RPMI-1640 supplemented with medium supplemented with L-glutamine, 10% FCS, and 1% penicillin/streptomycin. Human primary lung cancer cells were generated using an explant-cell culture method (15). NSCLC tumors were collected from consenting patients with no prior chemotherapy. The explants were observed to discard fibroblast outgrowth and preserve epithelial monolayer outgrowth, which was transferred into a new dish for subculturing to generate a primary cancer cell line. Coculture of LLC1 and macrophages was performed using a transwell system with a 6-well layout and maintained for 24 hours before further analysis. All cells were cultured at 37°C and 5% CO₂ according to the manufacturer's recommendations. The cell line was authenticated by the manufacturer and checked for Mycoplasma, using LookOut Mycoplasma PCR Detection Kit to guarantee all cells were mycoplasma-free.

For the generation of human macrophages, peripheral blood monocytes (PBMC) were isolated from human buffy coats using Ficoll (GE Healthcare; ref. 4). For mouse macrophage generation, the bone marrow from the murine femur and tibia was isolated, depleted of erythrocytes, and cultured in RPMI medium in the presence of 20 ng/mL M-CSF (R&D Systems). M1 macrophages were polarized using 100 ng/mL LPS (Sigma-Aldrich) and 100 U/mL IFNγ (R&D Systems), whereas M2 macrophages were polarized using 20 ng/mL IL4 for 24 hours (R&D Systems). For TAMs derived from lung tumor tissue, specimens were collected at the time of surgery from patients at the University Hospital Giessen in Germany before the start of chemotherapy, after obtaining informed consent. Human lung cancer tissues were dissociated using the Tumor Dissociation Kit (Miltenyi Biotec). 10⁷ cells were incubated with 10 μL CD68-PE antibody (Miltenyi Biotec) in 90 μL magnetic-activated cell sorting (MACS) buffer at 4°C for 15 minutes, washed, and incubated with 10 μL anti-PE microbeads (Miltenyi Biotec). For mouse TAM isolation, mouse tumor tissues were digested with 1 mg/mL collagenase and 10 μg/μL DNase at 37°C for 30 minutes. Cells were incubated with anti-F4/80-PE antibody (Miltenyi Biotec), washed, and incubated with anti-PE microbeads. Labeled mononuclear cells were separated from unlabeled cells in the presence of a magnetic field.

Animal studies were approved by the Regierungspräsidium Darmstadt, the local regulatory authorities for animal research in Hessen, Germany (Animal proposals B2/1154 and B2/2038). Wild-type C57Bl/6J, transgenic Kras^LA2, Hdac2^f/f, LysM^Cre, and Rag1 mice were purchased from The Jackson Laboratory. Hdac2^f/fLysM^Cre mice were generated through crossing Hdac2^f/f mice with LysM^Cre knockin mice and Kras^LA2Hdac2^f/fLysM^Cre mice were generated through crossbreeding Kras^LA2 with Hdac2^f/fLysM^Cre. The following models were used in the current study: (i) Intratracheal lung tumor model: mice were intratracheally inoculated with 1×10⁶ LLC1 cells. Mice were sacrificed on day 15 after tumor cell implantation and lungs were collected to isolate macrophages for further analysis. (ii) Tumor relapse and metastatic model: primary tumor developed following the subcutaneous injection of LLC1 cell to mice. After 10 days, mice were intubated, anesthetized, and the subcutaneous tumor was extracted, followed by wound sewing. All mice were intensively observed over a period of 20 to 40 days (4, 16); (iii) Kras^LA2 model: a murine model of lung adenocarcinoma that was driven by Kras^LA2 was genotyped according to The Jackson Laboratory recommended protocol (17). (iv) Intravenous lung tumor model: mice were intravenously injected with 1×10⁶ LLC1 cells. Mice were sacrificed on day 20 after tumor cell implantation and lungs were collected for further analysis. Lung tumor burden was quantified macroscopically and intracavitary malignancies were monitored based on microcomputerized tomography (micro-CT) scanning as previously reported (4, 16).

For primary lung cancer cells and TAMs, tumor tissue specimens were collected from patients with lung cancer at the time of surgery before chemotherapy after obtaining informed consent at the University Hospital Giessen in Germany. The study protocol for tissue donation was approved by the ethics committee (“Ethik Kommission am Fachbereich Humanmedizin der Justus Liebig Universität Giessen”) of the University Hospital Giessen (Giessen, Germany) in accordance with national law and with “Good Clinical Practice/International Conference on Harmonisation” guidelines. Written informed consent was obtained from each patient or the patient's next of kin (reference AZ 58/15).

Statistical analyses were performed with Prism ver. 6.0 and 9.0 (GraphPad Software). t test was used to compare the two groups. Data are expressed as mean ± SEM. When more than two groups were compared, differences among the groups were determined by one-way ANOVA with Tukey posttest for unpaired nonparametric variables. The Kaplan–Meier method was used to estimate overall survival, and differences were assessed using the log-rank test.

Expression profile data analyzed in this study were deposited at Gene-Expression Omnibus at GSE195440 and GSE227560. All other raw data are available upon request from the corresponding author.

Functional assays upon cytokine neutralization in TAMs conditioned medium (CM; Supplementary Table S1), quantitative real-time PCR (Supplementary Table S2), RNA sequencing, Western blotting (Supplementary Table S3), chromatin immunoprecipitation (Supplementary Table S4), immunocytochemistry (Supplementary Table S5), multiplexed immunofluorescence (Supplementary Table S6 and Supplementary Table S7), fluorescence-activated cell sorting (Supplementary Table S8), functional assays, treatment of macrophages with inhibitors, transient knockdown of HDACs and transient overexpression of HDAC2, cytokine array, hematoxylin and eosin staining, enzyme-linked immunosorbent assay (ELISA), coimmunoprecipitation and HDAC activity assay are provided in an online data supplement file.

Protumor M2-like macrophages were found to have significantly higher HDAC activity than antitumor M1-like macrophages (Fig. 1A). RNA sequencing (Fig. 1B; Supplementary Fig. S1A) revealed that among the classic HDAC family members (class I, II, and IV), HDAC2 expression was significantly increased in M2-like macrophages, but decreased in M1-like macrophages. This was confirmed by protein expression analysis (Fig. 1C) and HDAC2 activity assay (Fig. 1D). Multiplex immunofluorescence staining of human lung tumor tissue from adenocarcinoma, squamous cell carcinoma, and large-cell carcinoma demonstrated increased infiltration of M2 TAMs (CD68⁺CD163^hiALOX15^hiIL12^lowCCR7^low) compared with M1 TAMs (CD68⁺IL12^hiCCR7^hiCD163^lowALOX15^low) in TME (Supplementary Fig. S1B). An inclusion of HDAC2 staining in the above-mentioned antibody panel revealed higher HDAC2 intensity/expression in M2 TAMs compared with M1 counterparts (Fig. 1E and F). Additionally, HDAC2 showed a stronger positive correlation with M2 macrophage markers (ALOX15, CD163) than with M1 macrophage markers (IL12, CCR7; Supplementary Fig. S1C and S1D). The relative spatial distribution of TAMs and tumor cells revealed that TAMs with high HDAC2 expression (HDAC2^high) were located more proximal to tumor cells as compared with HDAC2^low TAMs (Fig. 1G and H). Interestingly, overall survival was significantly lower in patients with HDAC2^high TAMs compared with those with HDAC2^low TAMs (Fig. 1I).

To evaluate the role of macrophage-specific HDAC2, transgenic mice with macrophage-specific HDAC2 depletion (Hdac2^f/fLysm^Cre) were developed. To understand the effect of HDAC2 inactivation in macrophages from tumor-bearing mice, LLC1 cells were intravenously injected in Hdac2^f/f and Hdac2^f/fLysm^Cre mice, followed by the harvesting of bone marrow at day 20 to generate bone marrow–derived macrophages (BMDM). mRNA expression profiling of M1 (Tnf, Ccr7, Il12b, and Nos2) and M2 (Il10, Alox15, Arg1, and Chit1) macrophage markers in BMDMs from these mice revealed the HDAC2-deficient macrophages were shifted toward an M1-like phenotype (Fig. 2A; Supplementary Fig. S2A and S2B), resulting in reduced proliferation and migration, and enhanced apoptosis of tumor cells, which were exposed to CM of HDAC2-deficient macrophages (Supplementary Fig. S2C). Further, RNA sequencing analysis revealed differential gene regulation in BMDM from tumor-bearing Hdac2^f/fLysM^Cre mice compared with Hdac2^f/f mice (Fig. 2B). Additionally, pathway analysis from differentially regulated genes showed that inactivation of HDAC2 alters, among others, the interleukin signaling pathway, the inflammation-induced chemokine and cytokine signaling pathway, the EGFR signaling pathway, T-cell activation, and angiogenesis in TAMs (Fig. 2C).

To evaluate the functional role of macrophage-specific HDAC2 expression in experimental lung tumorigenesis, various approaches were used to induce lung tumors in Hdac2^f/fLysm^Cre mice, including (i) intratracheal lung tumors (Fig. 2D), (ii) metastatic lung tumors (Fig. 2E), and (iii) Kras^LA2 oncogene-driven lung tumors (Fig. 2F). Mice with HDAC2-deficient TAMs showed a significant reduction in lung tumor burden (Fig. 2G,–I). Additionally, von Willebrand factor (vWF) staining of lung tumor sections showed a reduced number of vessels in tumors that developed in HDAC2-deficient macrophage tissue (Supplementary Fig. S3A–S3C). Coimmunostaining of HDAC2 and macrophage marker (F4/80) in lung tumor sections confirmed a significant reduction in HDAC2 expression in TAMs from Hdac2^f/fLysM^Cre mice in all models (Supplementary Fig. S3D–S3F). mRNA expression profiling of TAMs sorted by MACS revealed increased expression of M1 markers (Tnf, Ccr7, Nos2, and Il12b) and reduced expression of M2 markers (Il10, Arg1, Chit1, and Alox15) in TAMs from tumors with HDAC2-deficient macrophages (Fig. 2J,–L; Supplementary Fig. S3G–S3I). Notably, tumors developing in HDAC2-deficient macrophage tissue showed reduced CD206⁺ macrophages and CD4⁺ T cells, and increased CD80⁺ macrophages and CD8⁺ T-cell infiltration in the tumor microenvironment (Fig. 2M–O; Supplementary Fig. S3J–S3L). Collectively, these results confirmed that macrophage-specific HDAC2 plays a pivotal role in in vivo lung tumor growth and progression, linked to macrophage-mediated, and macrophage T-cell–mediated protumor immunity.

To dissect the role of class I HDACs in macrophage activation, we performed siRNA-mediated knockdown of the genes HDAC 1, 2, 3, and 8 in M2 macrophages, followed by mRNA and protein expression profiling of macrophage markers and treatment of tumor cells with CM. siRNAs targeting of HDAC1 and HDAC2 caused a phenotypic switch from M2 to M1 macrophages [increased M1 (TNF, CCR7, IL1B, IL8, IL12B) and decreased M2 (IL10, ALOX15, CCL18, CSF1R, IL1RA, MRC1) marker expression; Fig. 3A–E; Supplementary Fig. S4A–S4C; Supplementary Fig. S5A–S5C] and reduced the in vitro tumorigenicity of CM (reduced proliferation and migration and increased apoptosis of A549 cancer cells; augmented HUVEC tube formation; Fig. 3F–I). Although the degree of HDAC1 and HDAC2 knockdown in M2 macrophages was similar, HDAC2 knockdown led to more substantial phenotypic and functional repolarization effects than HDAC1 knockdown (Supplementary Fig. S5). Furthermore, concomitant knockdown of HDAC1 and HDAC2 in M2 macrophages did not lead to significant synergistic repolarization in terms of macrophage marker alteration and functional effects on A549 cells (Supplementary Fig. S5). In primary TAMs, HDAC2 knockdown also induced phenotypic M2 to M1 repolarization and reduced in vitro tumorigenicity in primary lung cancer cells (Fig. 3J–M). Knockdown of HDAC3 (Supplementary Fig. S6A-S6D) and HDAC8 (Supplementary Fig. S6E–S6H) did not significantly affect the macrophage phenotype nor its impact on the proliferation of tumor cells. Interestingly, the knockdown of HDAC2 in M1 macrophages further strengthened their antitumor properties (Supplementary Fig. S7A–S7D). Vice versa, a gain of function of HDAC2 in M1 macrophages phenotypically and functionally skewed these cells to the M2 phenotype (Supplementary Fig. S7E and S7F). Additionally, except for the reduction in migration, HDAC2_KD_M2 macrophages did not show alteration in other cellular functions such as proliferation, apoptosis, and phagocytosis (Supplementary Fig. S7G), confirming the pivotal role of HDAC2 in macrophage activation to M2 phenotype.

We inhibited the activities of class I and II HDACs in M2 macrophages by using class I (VPA and CAY10398) and class II (TMP269 and MC1658) HDAC inhibitors. mRNA expression profiling revealed that treatment of M2 macrophages with VPA and CAY10398 increased the expression of M1 markers (TNF, CCR7, IL8, IL1B, and IL12B) and decreased expression of M2 markers (IL10, MRC1, ALOX15, CCL18, and IL1RA; Fig. 4A; Supplementary Fig. S8A and S8B). These results were confirmed by protein expression analysis and ICC staining (Fig. 4B and C). In contrast, neither TMP269 nor MC1658 showed any ability to affect M2 to M1 macrophage repolarization (Supplementary Fig. S9A–S9D). Additionally, the treatment of A549 cells with CM from VPA-treated M2 macrophages resulted in reduced in vitro tumorigenicity (proliferation, migration, and apoptosis; HUVEC tube formation) compared with the control CM (Fig. 4D and E). Further, VPA treatment of ex vivo TAMs isolated from human lung tumors caused an M2-to-M1 phenotype switch (Fig. 4F) and reduced their in vitro tumorigenicity (proliferation, migration, and apoptosis) in primary lung cancer cells (Fig. 4F and G). Notably, intraperitoneal administration of class I HDAC inhibitor (VPA) to tumor-bearing mice showed compared with Hdac2^f/f, a 38% reduction of tumor growth in Hdac2^f/fLysM^Cre mice, whereas VPA treatment caused a further reduction of 26% (Fig. 5A and B). mRNA expression profiling of macrophage markers demonstrated a stronger M1-like macrophage phenotype in tumors from VPA-treated Hdac2^f/fLysM^Cre mice (Supplementary Fig. S10). Additionally, FACS analysis demonstrated reduced CD206⁺ macrophages and CD4⁺ T cells, and increased CD80⁺ macrophages and CD8⁺ T-cell infiltration in tumors treated with VPA (Fig. 5C and D), suggesting antitumor effects of VPA are dependent on macrophage T-cell cross-talk. To test this hypothesis, we treated wild-type (WT) and T-cell–depleted mice (Rag1) with VPA. Interestingly, tumor growth was significantly reduced in VPA-treated WT and Rag1 mice. However, VPA treatment of Rag1 mice did not show a further reduction in tumor growth in RAG1 mice compared with untreated Rag1 mice (Fig. 5E and F). FACS analysis confirmed the absence of T-cell populations in lung tumors from Rag1 mice compared with those from WT mice, and, VPA treatment of tumor-bearing WT mice reduced CD206⁺ macrophages, CD4⁺ T cells, and increased CD80⁺ macrophages, and CD8⁺ T cells infiltration (Fig. 5G and H), suggesting class I HDAC inhibitors reduce lung tumor growth via macrophage-mediated and macrophage T-cell–mediated antitumor immunity.

Because CM from HDAC2-deficient M2 TAMs consistently showed antitumor effects, we next examined their secretory profile. Using cytokine array, we found that primary HDAC2-deficient human TAMs showed reduced secretion of IL10, IL16, and macrophage migration inhibitory factor (MIF), and increased secretion of IL8 (Fig. 6A and B). ELISA of the CM demonstrated TAMs secreted approximately 10 ng/mL IL10, 50 ng/mL IL16, 50 ng/mL MIF, 5 ng/mL IL8 in CM, which then showed significant alteration in CM from HDAC2-deficient TAMs (Supplementary Fig. S11A). Treatment of IL10 (5 and 10 ng/mL), IL16 (50 and 100 ng/mL), and MIF (50 and 100 ng/mL) increased proliferation, migration, and reduced apoptosis (Supplementary Fig. S11B), the opposite effects were observed upon neutralizing IL10, IL16, and MIF in TAMs CM (Supplementary Fig. S11C). On the other hand, treatment of lung cancer cells with 10 ng/mL IL8 significantly increased the apoptosis of lung cancer cells; confirming the protumor role of IL10, IL16, MIF, and the antitumor role of IL8 (Supplementary Fig. S11D).

Notably, siRNA-mediated genetic ablation of HDAC2 in primary TAMs elevated the acetylation level of histone protein H3 (Fig. 6C). Similar effects were observed by ICC staining of M2 macrophages (Fig. 6D). Furthermore, the coimmunoprecipitation analysis demonstrated that suppression of HDAC2 in TAMs increased the acetylation of TF SP1 (Fig. 6E), indicating a loss in its DNA binding capacity at gene promoters (18). In silico analysis detected SP1 binding sites in the promotor regions of top downregulated genes (IL10, IL16, and MIF), but they were absent in the promotor region of the top upregulated gene (IL8; Supplementary Figs. S12 and S13), indicating HDAC2 may regulate gene transcription in macrophages by acting on the acetylation status of both histone and nonhistone proteins.

To better understand the role of the HDAC2–SP1 gene regulatory axis in the transcription of M1 and M2 marker genes, we performed mRNA expression analysis of macrophage markers after siRNA-mediated knockdown of SP1 in M2 macrophages. SP1-deficient M2 macrophages demonstrated reduced expression and activity of HDAC2 (Fig. 6F), pointing to a role for SP1 in HDAC2 gene transcription. Like HDAC2-deficiency, SP1-deficient M2 macrophages also induced phenotypic (increased expression of M1 and decreased expression of M2 macrophage markers; Fig. 6G–I) and functional (CM from SP1-deficient M2 macrophages reduced proliferation, migration, and increased apoptosis of tumor cells; Fig. 6J) switching of M2 to M1 macrophages, suggesting that loss of SP1 binding either by acetylation or by transcriptional inhibition downregulates the M2 macrophage program, thereby upregulating the M1 macrophage program.

To test this hypothesis, we performed chromatin immunoprecipitation (ChIP) with M2 macrophages transfected with si_NS and si_HADC2 using antibodies for HDAC2, AcH3, RNA polymerase II, SP1, and a control IgG. The ChIP assay demonstrated that suppression of HDAC2 was associated with significantly impaired recruitment of RNA polymerase II, decreased level of acetylated histones H3, and reduced binding of SP1 to the promoter region of M2 marker genes (IL10, IL16, MIF, and ALOX15). The opposite effects were observed at the promotor regions of M1 marker genes (IL8, IL12B; Fig. 7A; Supplementary Fig. S14). Taken together, these results demonstrate that the transcriptional regulation of HDAC2 by SP1 and deacetylation of SP1 by HDAC2 upregulate the M2 macrophage program while downregulating the M1 macrophage program.

We generated four major findings with this study. First, increased infiltration of HDAC2-overexpressing M2-like macrophages and their close proximity to tumor cells significantly correlate with poor survival of lung cancer patients. Second, both macrophage-specific genetic ablation of HDAC2 and the use of pharmacologic inhibitors, specifically targeting class I HDACs, reinduce TAM-associated antitumorigenic activity in murine models of lung cancer, in primary human lung cancer-derived TAMs as well as in vitro–cultured M2 macrophages. Third, myeloid cell–specific suppression of Hdac2 (Hdac2^f/fLysm^Cre) and pharmacologic inhibition of class I HDACs in four different murine models (intratracheal, intravenous, tumor relapse, and Kras^LA2-driven) reduces lung tumor growth via macrophage-mediated and macrophage T-cell–mediated antitumor immunity. And fourth, HDAC2-mediated deacetylation of SP1 and SP1-mediated HDAC2 gene transcription drives the protumor M2 macrophage programming, thereby suppressing the antitumor M1 macrophage programming (Fig. 7B).

TAMs are the most prevalent innate immune cells in the TME. Based on their immunologic response, TAMs present with two distinct phenotypic states: antitumor M1 and protumor M2. These differentiation states are regulated by different mechanisms, including dynamic layers of epigenetic modifications. Histone acetylation and DNA methylation modulate gene transcription, either alone or cooperatively, by altering the chromatin status. Several studies have suggested that inhibition of HDACs not only targets histone acetylation but also DNA methylation, thereby inducing widespread alterations in the transcriptional program (19). Therefore, epigenetic modulators targeting HDACs may provide a promising approach for targeting TAMs and can be easily repurposed for cancers with high M2-like macrophage infiltration. In the current study, we demonstrated that both the expression and the activity of the class I HDAC family member HDAC2 are higher in M2-like macrophages than in M1-like macrophages and that the close proximity of high-HDAC2 M2-like macrophages to tumor cells is inversely associated with lung cancer patient survival. This is well in line with recent findings that poor patient survival in lung cancer correlates with M2-like macrophage predominance and their spatial proximity to tumor cells, and that tumor cell–TAMs cross-talk alters the cellular differentiation state of macrophages by switching M1 to M2 TAMs (20, 21).

A successful targeted (epi)genomic and phenotypic reprogramming of M2 to M1 status implies that altering macrophage plasticity might be used for therapeutic purposes. As a therapeutic strategy, it avoids the limitations of approaches that target all macrophages. Various techniques are now being examined at the preclinical level, including prior work from our lab that revealed the reprogramming potential of blocking the Wnt pathway in TAMs. Several mechanisms are responsible for the upregulation of the Wnt pathway in lung cancer, including decreased expression of the tumor suppressor gene, and adenomatous polyposis coli (APC; ref. 22). In corroboration, our previous work based on RNA sequencing demonstrated reduced expression of APC in primary human TAMs and in vitro–trained M2 TAMs (4). Interestingly, others have shown that on loss of APC the Wnt pathway induces aberrant expression of HDAC2, thereby regulating tumor growth and progression (23, 24). This lends further support to the potential of TAM-specific inhibition of HDAC2 in reprogramming M2 TAMs to M1 TAMs.

Notably, we observed that only class I HDAC inhibitors, and not class II inhibitors, induced a phenotype switch from M2 to M1-like macrophages. Among the four class I HDAC family members, the amino-acid sequences of HDAC1 and HDAC2 are highly identical (85%), suggesting an overlapping impact on the transcriptional machinery (25). HDAC2 was indeed significantly upregulated in M2-like macrophages, and genetic ablation of HDAC2 in M2 macrophages or primary human TAMs suppressed their protumor status by phenotypically and functionally shifting M2-like to M1-like macrophages. Knockdown of HDAC1 displayed a similar but less efficient effect on M2 to M1 macrophage switching, and knockdown of both HDAC1 and HDAC2 had no synergistic effect on M2 macrophage repolarization. In contrast to HDAC1 and HDAC2, suppression of HDAC3 or HDAC8 did not affect M2 macrophage repolarization. On the other hand, loss of HDAC3 was previously reported to promote an M2 phenotype induced by IL4. This phenotype is unable to activate approximately half of the inflammatory gene-expression program when stimulated with LPS (26, 27). In contrast, the central role of HDAC2 is further supported by the finding that overexpression of HDAC2 in M1 macrophages orchestrated their polarization toward the M2 phenotype.

Previously, it was shown that macrophages, IFNγ and CD8⁺ T cells are required for the antitumor microenvironment elicited by class IIa HDAC inhibitor, TMP195 treatment (28). Later, using syngeneic mouse models of melanoma, breast, and lung cancer, Li X and colleagues demonstrated that low doses of class I and II HDAC inhibitor trichostatin-A (TSA) modestly increases antitumor activity of T cells and TAMs, thereby synergizing with checkpoint-targeted therapy (PD-L1 antibody) to induce tumor regression (29). In this study, using four preclinical lung cancer models (Kras^LA2 oncogenic, intratracheal, subcutaneous, and metastatic lung tumor models), we demonstrated that myeloid cell–specific Hdac2 deletion and class I HDAC inhibitor, VPA, caused a shift of TAMs into an antitumoral M1-like phenotype. This led to altered pathways related to interleukin signaling, chemokine and cytokine signaling, T-cell activation, angiogenesis, and EGFR signaling, which reduced infiltration of CD4 T cells, increased recruitment and activation of CD8 T cells, and facilitated vascular normalization and tumor suppression. In the context of T cells, Guerriero JL and colleagues observed that TMP195 treatment increased granzyme-B⁺ CD8⁺ T cells in breast tumor models, whereas in our study, VPA treatment of mice bearing lung tumors reduced CD4⁺, CD4⁺CD25⁺FOXP3⁺, CD4⁺PD1⁺, CD4⁺TIM3⁺, and effector memory T-cell infiltration and increased CD8⁺T-cell infiltration (Supplementary Fig. S15), indicating that T-cell–mediated antitumor responses are highly dependent on T-cell subpopulations involved in the various tumor models and the applied HDAC inhibitors (28). Therefore, future studies need to focus on the comprehensive characterization and manipulation of T-cell subpopulations in respective tumor models to ensure successful clinical translation of HDAC inhibitors or the development of HDAC2-selective inhibitors.

In the present study, CM from HDAC2-deficient M2-like macrophages consistently showed antitumoral properties, suggesting that HDAC2-regulated secretory factors may drive a protumor communication network within the TME. We further found that CM from HDAC2-deficient primary TAMs had decreased levels of the protumor factors [e.g., IL10 (30), IL16 (31), MIF (32), and FGF19 (33)] and increased levels of antitumor factors [e.g., IL8 (21), IL12 (34), and IL31 (35)], indicating that HDAC2 regulates transcription of both M1- and M2-associated secretory products. It is increasingly clear that HDACs regulate cytokine and chemokine gene expression not only by directly changing the acetylation of their promoters but also by modifying nonhistone proteins such as SP1, STAT3, ZEB1, and GATA-1,2,3 (36). SP1 is highly regulated by posttranslational modifications and is known to be associated with class I HDACs (37). Both HDAC2 and SP1 knockdown switched M2 macrophages to M1. HDAC2 knockdown was, indeed, found to increase acetylation levels of histone H3 as well as of SP1. Notably, SP1 knockdown revealed its role in the transcription of HDAC2 in M2 macrophages. Furthermore, in silico and ChIP analysis demonstrated that the deacetylation of both histone H3 and SP1 is required to activate the M2 macrophage program, thereby suppressing the M1 macrophage program. Future studies are needed to understand the epigenetic events triggered by acetylated histone H3 at the promotor of M1 marker genes.

In summary, our results demonstrate that the proximity of high HDAC2-expressing M2-like macrophages to tumor cells is a prognostic indicator for reduced overall survival of patients with lung cancer. Suppression of HDAC2 in M2 TAMs upregulates M1 marker gene expression and downregulates M2 marker genes in support of an antitumor M1-like phenotype. By regulating the transcription of genes such as IL10, IL16, MIF, ALOX15, IL8, and IL12B via acetylation of both histone and nonhistone proteins, HDAC2 determines TAM phenotypic plasticity. We found the HDAC2–SP1 axis to play a key role in this process. Tumor regression via myeloid cell–specific suppression of HDAC2 and pharmacologic inhibition of class I HDACs was mediated by antitumor T cells linked to the M1-like phenotype. Therefore, targeting HDAC2 in lung cancer TAMs, in particular in the presence of immune-checkpoint inhibitors, may provide a novel therapeutic strategy.

A. Stenzinger reports personal fees from Aignostics, Amgen, AstraZeneca, Eli Lilly, Illumina, Janssen, MSD, Novartis, Pfizer, Qlucore, Roche, Seagen, Takeda, Bayer, BMS, and Incyte, and grants from Thermo Fisher Scientific outside the submitted work. T. Stiewe reports grants from Bundesministerium (Bildung und Forschung, Deutsches Zentrum) Lungenforschung (DZL), Hessisches Ministerium für Wissenschaft und Kunst, and LOEWE (iCANx) during the conduct of the study. W. Seeger reports personal fees from United Therapeutics, Tiakis Biotech AG, Liquidia, Pieris Pharmaceuticals, Abivax, Pfizer, and Medspray BV outside the submitted work. No disclosures were reported by the other authors.

X. Zheng: Conceptualization, data curation, software, formal analysis, validation, investigation, methodology, writing–original draft, writing–review and editing. P. Sarode: Data curation, formal analysis, validation, investigation, methodology, writing–original draft, writing–review and editing. A. Weigert: Conceptualization, data curation, formal analysis, validation, investigation, methodology, writing–review and editing. K. Turkowski: Data curation, formal analysis, methodology. P. Chelladurai: Investigation, methodology. S. Günther: Software, formal analysis, investigation, methodology, writing–review and editing. C. Kuenne: Software, formal analysis, investigation, methodology, writing–review and editing. H. Winter: Resources, data curation, writing–review and editing. A. Stenzinger: Resources, formal analysis, writing–review and editing. S. Reu: Resources, writing–review and editing. F. Grimminger: Resources, funding acquisition, writing–review and editing. T. Stiewe: Resources, writing–review and editing. W. Seeger: Conceptualization, funding acquisition, writing–original draft, writing–review and editing. S.S. Pullamsetti: Conceptualization, resources, formal analysis, supervision, funding acquisition, methodology, writing–original draft, writing–review and editing. R. Savai: Conceptualization, resources, data curation, formal analysis, supervision, funding acquisition, validation, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing.

The authors thank Yanina Knepper, Vanessa Golchert, Jeanette Knepper, and Praveen Mathoor for their excellent technical assistance. The authors thank Spyridoula Barboutsi for her support with the macrophage phagocytosis assay. Work in the laboratories of R. Savai was supported by Max Planck Society, the German Center for Lung Research (DZL). BMBF (KMU-innovative-22: miRTumorProst; 031B0768C), Deutsche Forschungsgemeinschaft (DFG, German Research Foundation): SA 1923/7-1, SFB1213 (Project A10N*); the Excellence Cluster Cardio-Pulmonary Institute (EXC 2026: Cardio-Pulmonary Institute (CPI), project 390649896), and the State of Hesse (LOEWE iCANx, Project A5 and Area C). Work in the laboratories of S.S. Pullamsetti was supported by Max Planck Society, the German Center for Lung Research (DZL). European Research Council (ERC) Consolidator Grant (#866051), Deutsche Forschungsgemeinschaft (DFG, German Research Foundation): SFB1213 (Project A01 and A05); the Excellence Cluster Cardio-Pulmonary Institute (EXC 2026: Cardio-Pulmonary Institute (CPI), project 390649896), and the State of Hesse (LOEWE iCANx, Project B4, B5 and Area C).

The publication costs of this article were defrayed in part by the payment of publication fees. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734.