Extracellular PKM2 modulates cancer immunity by regulating macrophage polarity

Recombinant protein expression and purification

The cDNAs that encode human PKM1, PKM2 and G415R were purchased from Adgenes. The cDNAs were subcloned into bacterial expression vectors. The recombinant proteins were expressed and purified from bacterial lysates by a two column procedure as in previous reports [13, 20].

Cell Line and mouse bone marrow derived macrophages

Cell lines: Cell lines Raw264.7, 4T1, and B16.F10 were purchased from ATCC. Cells were cultured according to the vendor’s instruction.

Mouse bone marrow derived macrophage (BMMφ) adopted from [23]: Briefly, mice were sacrificed and using aseptic technique. Bone marrow was harvested from the tibia and femur bones after the surrounding muscles were removed. To extract the bone marrow, the joints were cut with a scalpel, and the marrow was flushed out of the bones using a 25-gauge needle attached to a 10 ml syringe filled with PBS. All the clumps were gently dissociated and passed through a 70 μm cell strainer. The cell suspension was then centrifuged at 250 g for 5 min at room temperature to pellet the cells. To differentiate into BMMφ, isolated cells were cultured in DMEM with 10% heat inactivated FBS and 10 ng/mL M-CSF1 for 5 days. The differentiated BMMφ is ready for subsequent experiments.

Patient tissue analyses were carried out under the guidelines of NIH and adhered to the Declaration of Helsinki. The study was approved by GSU Institutional Review Board (IRB). All tissue samples are de-identified. It falls under IRB exemption 4. Samples were sectioned and analyzed by different staining. Tumor and normal tissues were purchased from US Biomax Inc.

Animal model and treatments

All animal experiments conformed to the guidelines of the NIH Guide for the Care and Use of Laboratory Animals and were approved by Georgia State University IACUC. At the end of each experiment, the mice were euthanized by CO₂ inhalation. Animal euthanasia conformed to the guidelines of the NIH Guide for the Care and Use of Laboratory Animals. Animal models: The same procedures similar to our previous reports were followed to generate 4T1 and GEM-NSCLC models [24, 25]. For B16 tumor model, 7–8-week-old C57BL/6 mice were restrained by hand. B16 cells (4 × 10⁵, suspension in 100 μl HBSS) were injected underneath the skin (s.c.). The mice returned to their cages. The tumors typically grow to ~ 200 mm³ in volume in 7–8 days in our hands. Treatments: the designated treatments began; (1) for 4T1 model, when tumors grew to average of 200 mm³ in volume (~ 8 days after 4T1 cell inoculation), (2) for B16, 8 days after B16 cell inoculation, and (3) for GEM-NSCLC, 8 weeks after AdV-Cre delivery. Mice were randomly assigned to the treatment groups. The agents as indicated in the figure panels and legends were administered via i.p. injections. Animals were euthanized 2 days after the last dose treatments. The treatment for assessment of survival of B16 tumor mice, treatments (i.p. injection) start 8 days after B16 cell inoculation, PKAb (4 mg/kg, b.i.w), αPD-1 (4 mg/kg, b.i.w) and the combination for 4 weeks in total 9 doses. The mice were maintained in cages for survival assessment.

Macrophage polarization inductions

Raw264.7 or the isolated BMMφ were cultured under standard conditions. To induce M1 polarization, LPS was added into medium to final concentration of 10 ng/ml. To induce M2 polarization, IL-4 was added into medium to final concentration of 20 ng/ml. The cells were maintained for designated. The treated cells were then harvested for subsequent analyses.

FACS analyses strategies

Tissue process for FACS analysis Adopted from [26]: Briefly, tumors were sliced out from the mice and minced (1 × 1 mm) and digested with collagenase D. After digestion and filtration, the recovered cells were resuspended and washed with FACS staining buffer. The resultant cells were subjected to FACS analysis.

FACS gating and analysis strategies for macrophages: (1) Analysis of macrophages in tissue samples, the isolated viable single cells were first gated for CD45⁺ cells to identify leukocytes. Macrophages were then selected based on F4/80⁺ expression (the FACS gating for macrophage selection please see supplementary Fig. 1A). The selected macrophages were then FACS sorted via CD206⁺ for M2 macrophages and iNOS⁺ for M1 macrophages. (2) His-tag for rPKM1 or G415R binding analysis, the isolated viable single cells were first gated by CD45⁺ F4/80⁺ macrophage identification. The selected macrophages were then FACS sorted via His⁺ for binding to his-tagged proteins. (3) Analysis with cultured cells, viable cultured cells were gated for F4/80⁺ cells to identify macrophages, followed by analysis for CD206⁺ cells to identify M2 macrophages, by CD86⁺ for M1 macrophages.

FACS gating and analysis strategies for T cells from tissue samples: The isolated viable single cells were first gated for CD3⁺ cells to identify T-cells. The isolated T cells were analyzed by CD4⁺ and FoxP3⁺ as T_reg cells CD8⁺ as CD8 + CTL.

The procedures for His-tag Co-precipitation, Western blotting, cell attachment assay, and Flow cytometry were adopted from our previous reports [20, 21, 27, 28].

Tissue section and cell staining

IHC and IF: IHC and IF staining procedures were similar to those of previous reports [21, 27].

PI3K and Arg1 activity Assays were performed by the procedures similar to our previous reports [21, 22, 27]

Statistical calculations

Statistical analyses were carried out using the GraphPad Prism 6.0 software. Kaplan–Meier survival curves were calculated using survival time for each mouse from all treatment groups. Statistical analyses in the survival experiments were performed by log-rank (Mantel-Cox) test. All in vitro experiments were carried out at least 5 times. For image quantifications, FACS, and other analyses, statistical significance was assayed by unpaired Student’s t-test. In all figures, *P < 0.05; **P < 0.01, ***P < 0.001; n.s. denotes not significant. All data are presented as mean ± s.e.m. or as box plots.

EcPKM2 in patient tumors and PKM2 is secreted by cultured cancer cells

High levels of PKM2 are presented in blood circulation of cancer patients, and the levels of circulating PKM2 correlated with tumor progression [19, 29]. ELISA analyses of culture medium of cancer cells demonstrated that PKM2 is released from cultured cancer cells (Fig. 1A). To further confirm presence of EcPKM2 in tumor, we carried out IHC staining of tissue samples from cancer patients using an in-house developed rabbit monoclonal antibody against PKM2, IgGPK [21]. High levels of PKM2 were detected in cancer patient tissues. The EcPKM2 was apparent in the staining. As control, almost no PKM2 staining was observed in adjacent normal tissues (Fig. 1B). Our results indicate that PKM2 is expressed intracellularly and released to extracellular space of tumor of cancer patients, and PKM2 is secreted by cancer cells.

EcPKM2 interacts with integrin α_vβ₃ on macrophage and promotes M2 polarity

We previously reported that EcPKM2 interacts with integrin α_vβ₃ on angiogenic endothelial cells and myofibroblasts and activates the integrin signaling [20, 22]. It was demonstrated that integrin α_vβ₃ is highly expressed in macrophages [30–32]. We confirmed the integrin expression on macrophages by probing integrin α_vβ₃ expression in cultured Raw264.7 (Raw) cells and bone marrow-derived macrophages (BMMφ) from healthy mice (Fig. 1C). We reasoned whether EcPKM2 interact with the integrin α_vβ₃ on macrophages. To this end, we first carried out cell attachment assay with Raw and BMMφ cells using culture plate coated with recombinant PKM2 (rPKM2) or PKM1 (rPKM1) that expressed and purified from bacterial E. Coli. or BSA. Clearly, more cells attached to rPKM2 compared to rPKM1 and BSA coated plates (Fig. 1D). The cell attachment is integrin α_vβ₃ dependent, as antibody against integrin α_vβ₃ LM609 blocked the Raw cells attaching to the rPKM2 coated plate, while IgG did not (Fig. 1E). Co-precipitation (by His-tag pull-down) of His-rPKM2 with integrin α_vβ₃ from extracts made from BMMφ, confirmed the interaction of PKM2 with the integrin (Fig. 1F). To verify that rPKM2 interacts with macrophage in tumor, macrophages were FACS sorted from the 4T1 tumors from the tumor-bearing mice that were treated with His-rPKM1 or His-rPKM2. The macrophages were then analyzed via FACS using anti-his-tag antibody. The his-rPKM2 interacted with macrophage much stronger compared to his-PKM1 in the tumor (Fig. 1G). We conclude from our results that EcPKM2 interacts with integrin α_vβ₃ on macrophages in tumors.

Our next question is what the functional role(s) of the interaction between EcPKM2 and integrin α_vβ₃ on macrophage is. Careful examination of morphology of Raw cells that were treated by different agents, including rPKM2, rPKM1, LPS, or IL-4, revealed that the cells treated with rPKM2 resembled IL-4 treated cells in morphology (Fig. 2A). It is well known that IL-4 activates M2 macrophage, while LPS stimulates M1 polarization. M2 macrophages secrete IL-10 and express cell surface marker CD206 [4, 33]. Thus, we examined secretion of IL-10 in culture medium of Raw and BMMφ cells that were treated with LPS, rPKM1, rPKM2, or IL-4. The cells treated with rPKM1, or LPS did not secrete IL-10, while the IL-4 and rPKM2 treated cells secreted high levels of IL-10 (Fig. 2B, C). FACS analysis of Raw and the BMMφ cells that treated with LPS, rPKM1, rPKM2, or IL-4 indicated that macrophages expressed high levels of CD206 upon IL-4 or rPKM2 treatment, while CD206 expression was almost undetectable in LPS or rPKM1 treated macrophages (Fig. 2D, E). It is well documented that expressions of iNOS and CD86 are important molecular signatures of M1 macrophage polarity. FACS analysis of M1 macrophage marker CD86 on Raw cells showed that PKM2 treatment decreased CD86⁺ cell population (Fig. 2F), while iNOS levels in Raw cells were reduced by rPKM2 but not rPKM1 (Fig. 2G). We further tested whether EcPKM2 could convert M1 macrophage to M2 subtype. To this end, Raw and BMMφ cells were incubated with LPS in the presence or absence of rPKM1 or rPKM2. Secretion of TNFα into culture medium was measured as an indication of M1 macrophage polarity [34]. Interestingly, the addition of rPKM2 reduced TNFα in culture medium of both Raw and BMMφ that treated with LPS (Fig. 2H, I), suggesting that EcPKM2 may be capable of converting M1 macrophages to M2 polarity.

EcPKM2 increased M2 macrophages and decreased M1 macrophages in tumors

To test whether EcPKM2 promotes M2 macrophage in vivo, we employed the breast cancer orthotopic 4T1 tumor model. The mice carrying 4T1 tumor were treated with G415R, a PKM2 mutant with higher solubility and dimer ratio [35], or rPKM1 when the tumor grew to an average size of ~ 200 mm³ in volume (Fig. 3A). Animals were euthanized 2 days after the last dose treatments. Tumor growth was measured, and tumors were weighted. Clearly, G415R but not rPKM1 increased tumor growth rate, size, and weight (Fig. 3B–D). Immune cells in tumor analyzed via Flow Cytometry and Fluorescence-Activated Cell Sorting (FACS). Figure 3E shows an example of FACS plots of macrophage polarity analysis. FACS analyses gated by F4/80⁺ and subsequent sorting via CD206⁺ iNOS⁺ indicated that G415R treatment increased percentage of M2 and decreased percentage of M1 in the total macrophage population in 4T1 tumor compared to the rPKM1 group (Fig. 3F). It is well documented that M2 macrophage induce CD4⁺ T_reg cells in tumor [36]. Evidently, CD4⁺ T_reg cells (CD4⁺/FoxP3⁺) increased in the tumors of G415R treated mice (Fig. 3G). On the other hand, G415R decreased cytotoxic CD8⁺ T-cells in the tumor (Fig. 3G). TAM polarity plays a vital role in cancer metastasis [2, 37]. Examination of metastatic nodules in lungs revealed that G415R treatment increased both metastatic nodular number and size compared to the groups treated with rPKM1 (Fig. 3H, I). We asked whether G415R treatment affected macrophage polarity in lung metastatic tumors. FACS analyses showed that G415R increased M2 macrophage percentage in lung metastatic tumors compared to the rPKM1 treated group (Fig. 3J). Similarly, G415R treatment also increased T_reg cells in the lung metastatic tumors (Fig. 3K).

EcPKM2 promotes M2 macrophage by interacting with and activating integrin α_vβ₃. We reasoned that antibody disrupting the PKM2 and integrin α_vβ₃ interaction would exert opposite effects in tumor. We used an in-house developed monoclonal anti-PKM2 antibody (Previous IgGPK, changed the name to PKAb here) that disrupts PKM2 and integrin α_vβ₃ interaction [20, 21]. Tumor bearing mice were treated with PKAb or rabbit IgG as a control. PKAb treatment led to tumor growth inhibition and smaller tumors (supplementary Fig. 1B) and less and smaller metastatic nodules in lungs compared to the IgG treatment group (Fig. 3L–N). PKAb led to high M1 and low M2 macrophage populations (Fig. 3O) and decreased CD4⁺ T_reg cells (Fig. 3O, P) in primary tumors. FACS analyses of M2 macrophage and T_reg in lung metastatic tumors revealed that PKAb treatment decreased M2 macrophage percentage and T_reg in metastatic tumors (Fig. 3 Q&R). We conclude from our experiments that EcPKM2 promotes M2 macrophage, while antibody disrupt EcPKM2 and integrin α_vβ₃ interaction abrogates the effects of EcPKM2 on polarity of TAMs.

To test the commonality of the role of EcPKM2 in modulating TAM polarity in tumor, we analyzed the effects of rPKM2 with two additional murine cancer models: (1) Orthotopic xenograft of B16 cells, a murine melanoma cell line. (2) DuPage, M., and co-workers developed a genetic engineering mouse (GEM) lung cancer model (Ref to as GEM-NSCLC). The GEM-NSCLC mice spontaneously develop lung cancer 3 weeks after intratracheal intubation delivery of adenoviral coded Cre recombinase (AdV-Cre) due to specific expression of Kras^G12D and Trp53^R172H in the lung [25, 38]. The B16 tumor-bearing mice and the GEM-NSCLC mice were treated with rPKM1 or G415R; 8 days post tumor inoculation for B16, or 8 weeks after delivery of AdV-Cre for GEM-NSCLC respectively (Fig. 4A). Animals were euthanized 2 days after the last dose treatments. The B16 tumor size in G415R treated mice was larger than that in rPKM1 treated mice (Fig. 4B). The tumor nodule number and size in G415R treated GEM-NSCLC mice was higher and larger than those in rPKM1 treated mice (Fig. 4C). FACS analyses demonstrated that G415R treatment increased M2 macrophage percentage and T_reg cells in tumors of both B16 and GEM-NSCLC mice (Fig. 4D–G). Conversely, PKAb treatment led to smaller B16 tumor in weight (Fig. 4J). Importantly, PKAb decreased M2 macrophage percentage and T_reg cells (Fig. 4K–N) in tumors of both models. Examination of macrophage and T_reg in lung metastatic tumors of the treated B16 tumor bearing mice showed that G415R increased M2 macrophage percentage and T_reg cells in lung metastatic tumor (Fig. 4H, I), while PKAb reduced M2 macrophage percentage and T_reg cells in B16 lung metastatic tumor (Fig. 4O, P). These results further confirmed our conclusion that EcPKM2 promotes M2 macrophage polarity in both primary and metastasis tumors, while antibody that disrupt the interaction between EcPKM2 and integrin α_vβ₃ promotes M1 macrophage polarity.

EcPKM2 activates integrin-FAK-PI3K signaling axis and upregulates Arg1 in macrophage

An open question was how the interaction between EcPKM2 and integrin α_vβ₃ regulated macrophage polarity. We previously showed that EcPKM2 interacted with integrin α_vβ₃ and activated the integrin signaling and downstream PI3K signaling axis in myofibroblasts and angiogenic endothelial cells [20, 22]. Thus, a reasonable speculation would be that EcPKM2 activates the same integrin-FAK-PI3K signaling axis in macrophage. To test the speculation, we first examined activation of FAK. Evidently, FAK was activated upon addition of rPKM2 to culture medium of Raw as shown by the FAK phosphorylation (Fig. 5A). The activation of FAK was diminished by the antibodies PKAb or 23C6, a mouse monoclonal antibody against integrin α_vβ₃ (Fig. 5B). It is well established that activation of integrin signaling FAK leads to activation of downstream PI3K [39]. Thus, we examined the activation of PI3K in the Raw cells upon rPKM2 treatments. Immunoblot demonstrated that PI3K was activated as shown by the PI3K phosphorylation (Fig. 5A). PI3K activity assay confirmed the activation of PI3K by PKM2 in the cells (Fig. 5C). The activation of PI3K by PKM2 was abrogated by a PI3K inhibitor and by a FAK inhibitor (Fig. 5A, C). It is well established that PI3K—PTEN pathway regulates Arg1 expression in macrophages and subsequently modulate macrophage polarity [40, 41]. We reasoned that EcPKM2 may modulate macrophage polarity by the same FAK-PI3K-PTEN signaling pathway and subsequently regulates the Arg1 expression/activity. We probed Arg1 levels and activity in the rPKM2 treated Raw and BMMφ cells. Clearly, addition of rPKM2 but not rPKM1 into culture medium upregulates Arg1 (Fig. 5A, D). Arg1 activity assay revealed that rPKM2 increased arginase activity (Fig. 5E). The effects of rPKM2 on Arg1 activity were abrogated by addition of the antibody 23C6 or the FAK inhibitor FI14 into culture medium (Fig. 5F). Knockdown of integrin β3 in Raw cells also abolished the effects of PKM2 in regulation of Arg1 (Fig. 6A, B), suggesting that upregulation of Arg1 and increase in arginase activity in macrophage by EcPKM2 were mediated through integrin α_vβ₃. We further analyzed the PTEN levels and activity in Raw and BMMφ. Evidently, the addition of rPKM2 but not rPKM1 into culture medium reduced PTEN in the cells (Fig. 6C), suggesting that EcPKM2 might play a role in downregulation of PTEN in macrophage to promote M2 subtype. To verify the role of EcPKM2 in regulating Arg1 expression and promoting M2 polarity via suppression of PTEN expression, PTEN was exogenously expressed in Raw via adenoviral vector carry either null or PTEN (AdV-Null/AdV-PTEN) (Fig. 6D). Exogenous expression of PTEN downregulated Arg1 levels (Fig. 6E) and decreased arginase activity in the rPKM2 treated cells (Fig. 6F).

PKAb enhances efficacy of checkpoint blockades

Addition of checkpoint inhibitors to various chemotherapies represents an important advancement in the treatment of cancers of different types [42, 43]. It is well documented that macrophage and polarity play a critical role in modulating cancer immunity by secreting pro- or anti-inflammatory soluble factors, such as cytokines/chemokines [4]. The anti-PKM2 antibody PKAb that disrupts the interaction between EcPKM2 and integrin α_vβ₃ promotes M1 macrophage polarity in tumor. It would be intriguing to test anti-cancer effects of PKAb in combination with checkpoint blockade. B16 is a murine melanoma model that is well tested with checkpoint blockade treatment. We therefore used this model to test efficacy of a combination of PKAb + anti-PD-1 antibody (αPD-1). Tumor bearing mice were treated with PKAb, αPD-1, and PKAb + αPD-1. PKAb and αPD-1 alone provided marginal or modest benefits on the animal overall survival and tumor growth. The combination provided strong anti-cancer effects (Fig. 6G).

Conflict interest

The authors declare no competing conflict of interests