Growth differentiation factor 11 modulates metabolism, mitigating the pro-tumoral behavior provided by M2-like macrophages in hepatocellular carcinoma-derived cells

Abstract

BACKGROUND

Hepatocellular carcinoma (HCC) is one of the most aggressive tumors worldwide. Chronic inflammation contributes to tumor evolution, and the infiltration of tumor-associated macrophages (TAMs), also known as M2-like macrophages, is associated with the most aggressive behavior. Therefore, these macrophages provide the primary growth and migratory factors to the tumor cells, including those of HCC. Current therapies are not well optimized for eliminating transformed cells or neutralizing the tumor immune microenvironment leukocytes, such as TAMs. Growth differentiation factor 11 (GDF11) may represent a promising dual therapeutic target due to its reported anti-tumorigenic and immunomodulatory properties.

AIM

To characterize the effects of GDF11 in M2-like macrophages and the HCC cell interaction using a functional in vitro model.

METHODS

This research used THP-1 and Huh7 cell lines. We applied recombinant GDF11 (50 ng/mL) every 24 hours on THP-1 differentiated macrophages with M2-like polarization using interleukin-4 and interleukin-13. Firstly, the GDF11 effects on signaling, viability, proliferation, metabolism, and redox state in macrophages were characterized. Subsequently, we extracted conditioned media (CM) from macrophages and performed indirect co-cultures with Huh7 cells. The functional parameters were proliferation and migration assays. Finally, we characterized secretion in the CM using the cytokine array membrane assay.

RESULTS

The present study demonstrated that GDF11 activates the canonical pathway Smad2/3 without cytotoxic or proliferative effects. We provide evidence that GDF11 also diminishes the pro-tumoral properties of M2-like macrophages. GDF11 promoted the reduction of the M2-like macrophage marker, cluster of differentiation 206, indicating a loss of pro-tumoral properties in these leukocytes. Furthermore, this molecule induced changes in metabolism and an increase in reactive oxygen species. Using CM derived from GDF11-treated M2-like macrophages, we observed a reduction in the proliferation and migratory capacity of liver cancer cells. Moreover, the cytokine profile was affected by GDF11 stimulus, demonstrating that this molecule alters the pro-tumoral properties of TAMs, which in turn impact the behavior of HCC-derived cells.

CONCLUSION

This in vitro study suggests that mitigating tumor-promoting or M2-like macrophages with GDF11 may be an effective strategy for controlling the aggressiveness of HCC.

Key Words: Hepatocellular carcinoma; Growth differentiation factor 11; Tumor-associated macrophages; M2-like macrophages; Tumor immune microenvironment

Core Tip: Hepatocellular carcinoma (HCC) is an aggressive tumor, and the infiltration of tumor-associated macrophages or M2-like macrophages increases aggressiveness. We need therapies to mitigate this pathology in the immune context. In the present study, our in vitro model demonstrates that growth differentiation factor 11 (GDF11) counteracts the pro-tumorigenic properties of M2-like macrophages. This was shown in the reduction of cluster of differentiation 206; furthermore, it induces changes in metabolism and the redox state. Conditioned media from GDF11-treated M2-like macrophages possessed changes in their cytokine profile that neutralized the proliferation and migration of HCC cells, demonstrating that GDF11 alters the pro-tumoral properties of macrophages. This study suggests that mitigating tumor-associated macrophages may be an effective strategy for controlling HCC.

INTRODUCTION

Hepatocellular carcinoma (HCC) represents a severe health problem worldwide, considered one of the most aggressive liver cancer subtypes and ranked as the sixth and third leading cause of incidence and mortality, respectively[1,2]. HCC arises from hepatotropic virus infection, alcohol consumption, xenobiotic toxicity, or the consumption of enriched-lipid or fructose diets[3-6]. Despite tumor etiology, this led to chronic inflammation, which exacerbates tumor development[7]. It has been widely described that the most aggressive liver tumors exhibit an increase in vascularization and high leukocyte infiltration, such as tumor-associated macrophages (TAMs), which is associated with the worst prognosis, as observed in cholesterol-enriched diet models[6,8]. This provides evidence that TAMs are a significant component of the tumor microenvironment, specifically the tumor immune microenvironment (TIME), enhancing the aggressiveness of many tumors, including HCC[9-11].

Macrophages, which belong to the innate immune system, execute specific responses, such as eliminating pathogens or transformed cells through phagocytosis, via a mechanism that depends on the production of reactive oxygen species (ROS)[12,13]. In cancer, TAMs, also known as M2-like macrophages [cluster of differentiation 206 (CD206)], tend to be activated in an alternative manner, exhibiting pro-tumor and immunosuppressive responses, in contrast to M1 or classically activated macrophages, which induce anti-tumor effects. High interleukin (IL) levels in the tumor, such as IL-4/IL-13 or transforming growth factor (TGF)-β, lead to alternative activation and tumor promotion[14-16]. Additionally, TGF-β confers aggressive features to the tumor core and various components of TIME[17,18].

We previously reported that growth differentiation factor 11 (GDF11), a member of the TGF-β superfamily, exhibits tumor suppressor properties, unlike TGF-β, particularly in HCC-derived cells by decreasing aggressiveness through a mechanism dependent on canonical signaling mediated by Smad 2/3[19]. Furthermore, GDF11 reduces proliferation, invasion, and migration, promoting a mesenchymal-to-epithelial transition and decreasing lipid and bioenergetic metabolism[20-22]. The therapeutic approaches for HCC must target cancer cells and the TIME components. We hypothesize that a narrow relationship exists between GDF11 and the mitigation or neutralization of pro-tumoral effects in leukocytes from TIME, such as TAMs, as reported in cancer cells. This new two-axis proposal could represent a groundbreaking strategy to reduce HCC aggressiveness in conjunction with conventional immunotherapy and other targeted therapies. Furthermore, this current research details the neutralization of the pro-tumoral response that GDF11 executes in vitro against M2-like differentiated macrophages and their relationship with HCC-derived cells in the context of aggressive behavior.

MATERIALS AND METHODS

Human recombinant GDF11 was purchased from Peprotech (120-11, Rocky Hill, NJ, United States). For monocyte-macrophage differentiation, we used phorbol-12-myristate-13-acetate (PMA; Sigma-Aldrich, P1585, Saint Louis, MO, United States); for macrophage activation, we applied lipopolysaccharide (LPS; Sigma-Aldrich, L2630, MO, United States), interferon-γ, IL-4, and IL-13 (Peprotech 300-02, 200-04, and 200-13, Rocky Hill, NJ, United States).

Cell culture

THP-1 monocytes and Huh7 were obtained from the American Type Culture Collection (TIB-202 and CRL-10317, Manassas, VA, United States). Both cells were cultured in RPMI 1640 (Gibco, 31800-014, New York, United States) and Williams’ Media E (Sigma-Aldrich, 4125-1, MO, United States), respectively. All cultures used 10% fetal bovine serum (Hy-Clone, SH30088.03, Logan, UT, United States), 100 U/mL ampicillin, and 100 μg/mL streptomycin (Gibco, 15240-062, New York, NY, United States). Cells were maintained at 37 ℃ in a 5% CO₂ and 90% humidity atmosphere. All cell lines were mycoplasma-free.

Cell viability and proliferation

We seeded 5 × 10³ or 1 × 10⁴ for Huh7 and THP-1 cell lines, respectively, in 96-well plates (Corning Inc., 3599, MA, United States). Cell Counting Kit-8 (Dojindo Lab, CK04, Kumamoto, Japan) or crystal violet (Sigma-Aldrich, 42555, MO, United States) addressed cell viability and proliferation, following the manufacturer’s instructions. We quantified the final absorbance at 450 nm every 24 hours. Cytarabine (Sigma-Aldrich, C1768, MO, United States) 10 μmol/L was used as a negative control for proliferation, and dimethyl sulfoxide (Sigma-Aldrich, 472301, MO, United States) for viability.

Macrophage polarization protocol

For macrophage polarization, we followed the reported protocol with some modifications[23]. We used 1 × 10⁶ THP-1 monocytes, which were differentiated into inactivated macrophages (M0) using PMA (200 ng/mL) for 24 hours. To induce M1-like macrophage polarity acquisition, interferon-γ (20 ng/mL) or LPS (100 ng/mL) were applied for 24 hours; additionally, for M2-like activation, we used the anti-inflammatory cytokines, IL-4 (20 ng/mL) and IL-13 (20 ng/mL) for 72 hours. As mentioned, the last treatments were co-treated using GDF11 (50 ng/mL), three doses every 24 hours, without exchange media to simulate tumor acidification. Finally, cells were used, conditioned media (CM) were collected, and stored at -20 ℃. Graphical experimental design can be found in Supplementary Figure 1. Positive controls of M2-like macrophages used CM derived from 1 × 10⁶ Huh7 cells and applied on M0 (1:1 dilution) for 24 hours and 72 hours.

GDF11 post-treatment protocol

M2-like macrophages were treated for an additional 72 hours in the presence of IL-4/IL-13 (post M2-like) in combination with GDF11, using three doses, each added every 24 hours (post M2-like + GDF11). We used the cells for the subsequent experiments and recovered CM as mentioned above, also described in Supplementary Figure 1.

Flow cytometry analysis

Flow cytometry (FCM) analysis was performed by analyzing 1 × 10⁴ events in duplicate. The CD206 antibody (Santa Cruz Biotechnology, sc-58986, TX, United States) was used to identify M2-like macrophage subsets. The specifications of primary and secondary antibodies used in this study are listed in Supplementary Table 1. The study utilized a BD FACSCalibur flow cytometer (Becton Dickinson, Franklin Lakes, NJ, United States), and the data obtained were analyzed using FlowJo version 10.8 software (BD Biosciences, CA, United States).

Protein isolation and quantification

Protein extraction was performed using M-PER lysis buffer (Thermo Fisher Scientific, 78501, MA, United States), which contained protease inhibitors (complete, Sigma-Aldrich, 11836153001, MO, United States) and phosphatase inhibitors (Sigma-Aldrich, 4906845001, MO, United States). Following the manufacturer’s instructions, protein quantification was performed using the bicinchoninic acid kit (Pierce, BCA Protein Assay Kit, Thermo Fisher Scientific, 23225, MA, United States). We used 30-100 micrograms of protein for THP-1 macrophages and Huh7 cells in the western blot assays.

Western blotting

Western blot was performed according to the previously reported protocol (Gerardo-Ramírez et al[20]) using primary and secondary antibodies and specific dilutions listed in Supplementary Table 1. Blot images were visualized using a ChemiDoc Imaging System (Bio-Rad, Hercules, CA, United States), and blot densitometry was analyzed using ImageJ software (National Institutes of Health, Bethesda, MD, United States). All data were normalized using β-actin (Sigma-Aldrich, A3854, MO, United States) as a loading control.

Real-time metabolism determination

In real-time, the oxygen consumption rate (OCR) and the extracellular acidification rate (ECAR) were determined using 5 × 10⁴ THP-1 differentiated macrophages previously characterized using the Seahorse XFe24 Flux Analyzer and XF Cell Energy Phenotype Test (Agilent Technologies, Santa Clara, CA, United States) according to the manufacturer’s instructions (data not shown). The XF Cell Mito Stress Test Kit (Agilent Technologies, 103015-100, Santa Clara, CA, United States) was used on activated macrophages, and XF RPMI (Agilent Technologies, 103576-503, Santa Clara, CA, United States) was used to quantify the parameters. The XF media was supplemented with 1 mmol/L pyruvate (Sigma-Aldrich, S8636, MO, United States), 2 mmol/L glutamine (Sigma-Aldrich, G8540, MO, United States), and 10 mmol/L glucose (Sigma-Aldrich, G7021, MO, United States), adjusted to a pH of 7.4. We used 500 μL as the final volume. The assay followed the sequential injection of 1.5 μmol/L oligomycin A (Sigma-Aldrich, 75351, MO, United States), 1 μmol/L carbonyl cyanide-4-(trifluoromethoxy)phenylhydrazone (Sigma-Aldrich, C29201, MO, United States), and 0.5 μmol/L antimycin A (Sigma-Aldrich, A86741, MO, United States) plus Rotenone (Sigma-Aldrich, R88751, MO, United States). OCR and ECAR measurements were normalized with total protein according to a standardized protocol.

ROS determination

ROS detection was performed using 5 μmol/L of dihydroethidium (DHE; Thermo Fisher Scientific, D11347, MA, United States). DHE diluted in phosphate buffered saline 1 × (Sigma-Aldrich, D5773, MO, United States) was applied on THP-1 differentiated macrophages for 15 minutes at 37 ℃ in the dark; images were captured using the cyanine 3 filter of Axio Vert.A1 microscope (Carl Zeiss, Jena, Germany); finally, images were analyzed (200 ×) every 24 hours, using ImageJ software (National Institutes of Health, Bethesda, MD, United States). We used rotenone (2 μmol/L) and antimycin (5 μmol/L) as positive controls. For the phagocytosis assay, 24 hours previously, macrophages were treated using GDF11, and the latex beads (Sigma-Aldrich, LB6, MO, United States) were used at a 1:100 cell-bead ratio for 90 minutes.

Cholesterol quantification

Total cholesterol was quantified, as previously reported[6]. Briefly, approximately 1 × 10⁶ cells were saponified with alcoholic potassium hydroxide solution. Hexane (Sigma-Aldrich, 34859-2, MO, United States) and distilled water were added and shaken to ensure complete mixing. The hexane or superior layer was evaporated with a vacuum (Speed Vac, Savant, Cranbury, NJ, United States). Cholesterol was measured using O-phthalaldehyde (Sigma-Aldrich, P1378, MO, United States) dissolved in acetic acid (0.5 mg/mL, Merck, 9508-05, Germany) at an absorbance of 550 nm. We used liver tissue from Western diet-fed mice as a positive control.

Wound-healing assay

We seeded 3 × 10⁴ cells using wound-healing inserts (Ibidi, 81176, Gräfelfing, Germany). Dishes were washed gently with phosphate-buffered saline 1 × to remove any detached cells. Subsequently, Williams’ Medium E media was added in the presence or absence of CM (1:10 dilution, CM-RMPI and Fresh Williams’ media, respectively), previously centrifuged to discard cellular debris. The healing closure was monitored, and images (100 ×) were taken every 24 hours up to 72 hours. The wound area (μm²) and migration capacity (%) were analyzed using ImageJ software. The anti-GDF11 (Santa Cruz Biotechnology, Inc., sc-81952, United States) antibody (1.62621 × 10¹² molecules) was mixed in CM derived from treated macrophages, previously filtered, and stirred overnight at 4 ℃ before use.

Cytokine detection

Cytokine detection was performed using the Human Cytokine Antibody Array Membrane (Abcam, ab133997, United Kingdom) according to the manufacturer’s instructions. We used undiluted CM derived from treated and untreated macrophages. Blot images were revealed using the ChemiDoc^TM imaging system (Bio-Rad, United States), and the dot blot analysis was performed using ImageJ software. The array data normalization of mean pixel density was performed using the formula X(Ny) = X(y) × P1/P(y).

Statistical analysis

GraphPad Prism 8 (GraphPad Software, San Diego, CA, United States) was used for data analysis on a Windows platform. All data are presented as the mean ± SEM. The analysis of variance test was used to compare mean values among groups. A P-value < 0.05 was considered statistically significant.

RESULTS

GDF11 activates the Smad signaling pathway in macrophages without affecting cell viability and proliferation

To evaluate GDF11 canonical signaling in THP-1 differentiated macrophages (M0), we applied GDF11 at early times (0-60 minutes) and assessed the phosphorylation of Smad 2/3 by western blot (Figure 1). We observed the activation of both proteins from 5 minutes, being stronger in Smad2 than in Smad3 (Figure 1A-C). At these times, we evaluated the content of the primary canonical inhibitor of the signaling pathway, Smad ubiquitination regulatory factor 1 (Figure 1A and D), observing no significant changes in its content. We previously reported that GDF11 exerts its biological function over a long period of treatment[20,21]. We decided to analyze whether GDF11 affects cell viability and proliferation, as reflected in the number of cells at 24 hours, 48 hours, and 72 hours, with GDF11 treatment administered every 24 hours, as depicted in the experimental design (Supplementary Figure 1). We were unable to detect any changes in cell viability (Figure 1E) or proliferation (Figure 1F). This result agrees with no significant changes in cellular morphology (Figure 1G). Our data suggest that GDF11 does not induce any cytotoxic response in this differentiated macrophage cell line.

Figure 1 Growth differentiation factor 11 induces Smad signaling pathway activation in THP-1-derived M0 macrophages. A: Representative images of western blots; B-D: Densitometry analysis (β-actin was used as a loading control); E: Cell viability (H₂O₂ was used as a negative control); F: Cell proliferation (cytarabine was used as a negative control); G: Cellular morphology, each image is representative of at least three independent experiments. Scale bars: 150 μm (200 ×, original magnification). Each column or point represents the mean ± SEM of at least three independent experiments carried out in triplicate. ^aP ≤ 0.05 vs not treated cells. GDF11: Growth differentiation factor 11; NT: Not treated cells; AraC: Cytarabine.

M2-like macrophage generation

We differentiated the THP-1 cells (monocytes) into M0 macrophages using PMA for 24 hours. After that, we treated M0 cells with IL-4 (20 ng/mL) in combination with IL-13 (20 ng/mL) for 24 hours and 72 hours to obtain an M2-like polarity; LPS (20 ng/mL) or interferon-γ (20 ng/mL) were used to induce M1-like subsets. Figure 2A-C shows M2-like macrophages using an anti-inflammatory stimulus according to cytometry analysis of CD206-positive cells (CD206⁺). Furthermore, we observed an extended morphology (bi-polar morphology) in IL-4/IL-13-treated macrophages at 72 hours, features of M2-like macrophages in vitro. Unlike the M2-like phenotype, M1-like cells possess a single pole (Figure 2D). Moreover, it was relevant to compare in vitro macrophages using a pure anti-inflammatory stimulus vs CM derived from Huh7 (1:1) cells because CM from cancer cells contains a rich cocktail of mediators that could simulate the TME signals. In both cases, we observed an increase in CD206⁺ cells; however, the first protocol undoubtedly represents a purer model, as it excludes other factors that could bias our study (Supplementary Figure 2).

Figure 2 The anti-inflammatory stimulus acquires M2-like macrophage polarization. A: Fluorescence histogram of the expression of the cluster of differentiation 206 (CD206) marker obtained by flow cytometry; B: Percentage of CD206⁺ cells in IL-4/IL-13-treated macrophages (M2-like); C: Represents the mean fluorescent intensity of the CD206⁺ subset; D: Macrophage morphology under the different treatments, orange arrows show specific morphology. Images are representative of at least three independent experiments. Scale bars: 150 μm (200 ×, original magnification). Each column represents the mean ± SEM of at least three independent experiments. ^aP ≤ 0.05 vs M0; ^bP ≤ 0.05 vs M2-like (24 hours). CD206: Cluster of differentiation 206; MFI: Mean fluorescent intensity; LPS: Lipopolysaccharide; IFN: Interferon; IL: Interleukin.

GDF11 reduces the CD206 marker in M2-like macrophages

Once macrophage activation was standardized, we evaluated the CD206⁺ cells by FCM in M2-like macrophages at 72 hours (early response) of treatment with GDF11. We observed that GDF11 reduced the percentage of CD206⁺ subsets from approximately 40% to 25%, suggesting the possible tampering or re-polarization of these cells (Figure 3A-C). CD206, a mannose receptor, is a marker of aggressiveness in cancer cells and macrophages; hence, this result supports the notion that GDF11 affects M2 polarization and suggests a decrease in pro-tumoral activity due to the loss of this specific surface protein marker.

Figure 3 Growth differentiation factor 11 decreases cluster of differentiation 206 in M2-like macrophages. FCM analysis was performed to evaluate cluster of differentiation 206 (CD206) expression. Cells were treated with growth differentiation factor 11 for 72 hours, as stated in the materials and methods section. A: Fluorescence histogram; B: Percentage of CD206⁺ cells; C: Represents the median fluorescence intensity of the subset. Cells were treated with growth differentiation factor 11 for 144 hours, as stated in the materials and methods section; D: Fluorescence histogram; E: Percentage of CD206⁺ events; F: Mean fluorescent intensity in subset. Each column represents the mean ± SEM of at least three independent experiments. ^aP ≤ 0.01 vs M0; ^bP ≤ 0.05 vs post M2-like + growth differentiation factor 11; ^cP ≤ 0.001 vs post M2-like. CD206: Cluster of differentiation 206; GDF11: Growth differentiation factor 11; MFI: Mean fluorescent intensity.

In our protocol with an additional 72 hours of treatment (post-72 hours; extended response), we found that GDF11 reduced the percentage of CD206⁺ macrophage subsets from 90% to 30%, indicating a reduction in the phenotype of pro-tumoral macrophages (Figure 3D-F). We could hypothesize that GDF11 displays a re-educating, or neutralizing effect in M2-like macrophage subsets. To standardize this protocol, we first recognized that the absence of an anti-inflammatory effect after 72 hours can reduce CD206 content. Therefore, we add IL-4/IL-13 treatment and GDF11 as a co-treatment to simulate competence (data not shown).

M2-like macrophages present differences in mitochondrial metabolism

M2-like macrophages exhibit decreased mitochondrial function, as reflected in the OCR depicted in the metabolic profile (Figure 4A). Notably, GDF11 treatment increased and restored OCR levels, suggesting a role for GDF11 in mitochondrial metabolism, which aligns with our previous work in HCC cell lines[21]. Interestingly, the GDF11 treatment in M2-like macrophages (dark red; M2-like + GDF11) appears to be resetting the parameters observed in M0 (black line). M0 macrophages have a higher OCR value than M2-like macrophages in these experimental conditions. ECAR was also quantified by Seahorse, showing that LPS effects in media acidification suggest an alternative metabolic route for anti-inflammatory macrophages, which are characterized by low ECAR levels (Figure 4B).

Figure 4 Growth differentiation factor 11 promotes bioenergetics rewiring in M2-like cells. Mitochondrial stress test was performed to determine. A: Oxygen consumption rate; B: Extracellular acidification rates using Seahorse XF24e flux analyzer; C: Maximal respiration; D: Basal respiration; E: Adenosine triphosphate production; F: Spare respiratory capacity; G: Non-mitochondrial oxygen consumption; H: Defective metabolic profile based on relative values of oxygen consumption rate and extracellular acidification rate before and after metabolic inducing stress. Each column represents the mean ± SEM of at least three independent experiments. ^aP ≤ 0.05 vs M0; ^bP ≤ 0.05 vs M2-like. OCR: Oxygen consumption rate; ECAR: Extracellular acidification rate; GDF11: Growth differentiation factor 11; ATP: Adenosine triphosphate.

Maximal respiration is the most prominent parameter affected due to the exacerbated respiration of GDF11-treated macrophages (Figure 4C). Basal respiration (Figure 4D), adenosine triphosphate production (Figure 4E), and spare respiratory capacity (Figure 4F) exhibited similar behavior, with no changes observed in non-mitochondrial oxygen consumption (Figure 4G), indicating specific alterations within this organelle. Figure 4H shows the phenotypic status of treated macrophages in the OCR vs ECAR axis; the M2-like macrophage acquires a more energetic phenotype after GDF11 treatment (M2-like + GDF11), like GDF11 alone treatment or the M1-like macrophage, suggesting a rewiring in energy production and probably immune responses.

GDF11 induces ROS production

Data support the hypothesis that GDF11 could change the phenotype from an M2-like to an anti-tumoral phenotype. To gather more evidence, we investigated whether GDF11 regulates ROS production, a critical feature of macrophage activity. First, we assessed whether THP-1 macrophages respond to ROS production by measuring the ROS content as a response to the phagocytic engulfment of latex beads. We treated M0 macrophages with latex beads for 90 minutes and evaluated the ROS produced by measuring the DHE-derived fluorescence, indicating superoxide anion generation. Interestingly, GDF11 alone (previously treated for 24 hours) increased ROS production to similar values observed in bead-treated control cells; combining both treatments did not induce a significant change compared with the treated groups (Figure 5A and B). The result could suggest a GDF11-induced ROS-mediated anti-tumoral activity or microbicide properties due to activation of the phagocytosis mechanism. We continued to explore this effect and analyzed the ROS production after the differentiation process of THP-1 cells treated with PMA for 24 hours. We determined ROS content at 0 hours, 24 hours, and 72 hours in M0 cells. Figure 5C and D show the results; the ROS content remains unchanged, but at 72 hours, ROS levels decrease significantly. We moved to figure out the effect of GDF11 on M0 cells; we could not find any difference at 24 hours of ROS, even in cells treated with TGFβ-1 (Figure 5E and F), but at 72 hours we observed an increment of ROS only in cells treated with GDF11, with no effect in TGF-β1 (Figure 5G and H). Finally, we compared the ROS production between M2a (IL-4/IL-13) and M2c (TGF-β1) macrophage subsets. GDF11-treated macrophages increased ROS in response to anti-inflammatory and immune suppressor stimuli at 72 hours, indicating a possible microbicide and anti-tumoral capacity training (Figure 5I-K).

Figure 5 Growth differentiation factor 11 increases reactive oxygen species production in macrophages. Superoxide anion detection by dihydroethidium (DHE)-derived fluorescence. A: Representative images of DHE-derived fluorescence in the phagocytosis beads assay. Scale bars: 150 μm (200 ×, original magnification); B: Representative densitometry analysis of phagocytosis beads assay; C: Representative images of DHE-derived fluorescence in M0 macrophages. Scale bars: 150 μm (200 ×, original magnification); D: Representative densitometry analysis in M0 macrophages; E: Representative images of DHE-derived fluorescence at 24 hours. Scale bars: 150 µm (200×, original magnification); F: Representative densitometry analysis at 24 hours; G: Representative images of DHE-derived fluorescence at 72 hours. Scale bars: 150 μm (200 ×, original magnification); H-J: Represent the densitometry analysis at 72 hours; K: Representative images of DHE-derived fluorescence at 72 hours. Scale bars: 150 μm (200 ×, original magnification). Rotenone (2 μmol/L) and antimycin A (5 μmol/L) were positive controls. Each column represents the mean ± SEM of at least three independent experiments in triplicate. ^aP ≤ 0.05 vs not treated cells or M0; ^bP ≤ 0.05 vs M0; ^cP ≤ 0.01 vs M0. NT: Not treated cells; GDF11: Growth differentiation factor 11; TGF: Transforming growth factor.

GDF11 decreases cholesterol content in THP-1-derived macrophages

GDF11 decreases cellular cholesterol, as previously reported[21]. We aimed to investigate whether GDF11 has the same metabolic effects on lowering total cholesterol, specifically in macrophages. Supplementary Figure 3A shows a time-dependent decrement of total cholesterol induced by GDF11 at similar levels of such effect mediated by atorvastatin, an inhibitor of the 3-hydroxy-3-methyl-glutaryl-CoA reductase, the rate-limiting enzyme of the mevalonate pathway (Supplementary Figure 3B). These data provide evidence that GDF11 affects macrophages in a lipid-rich context.

CM derived from M2-like macrophages increases cell aggressiveness in HCC-derived cells

Then, we focused on addressing the effects of GDF11 on restrained pro-tumoral properties of M2 macrophages concerning human HCC cell lines by using indirect co-culture with CM from M2-like macrophages. We assayed cell viability using CM dilutions (1:3, 1:4, 1:8, and 1:10) at 24 hours, 48 hours, and 72 hours. We found that optimal dilutions were 1:8 and 1:10, observing no cytotoxic effects in the Huh7 cell line (Figure 6A-C) and preserving the signaling of mediators released to the medium. The reduction in cell viability at 72 hours using 1:3 and 1:4 CM dilution may be due to a high concentration of different factors, such as interleukins, growth factors, and high acidification. To gain more confidence, we analyzed the migratory capacity using a wound healing assay up to 48 hours, employing ideal dilutions of 1:8 and 1:10. Figure 6D and E illustrate the effects on migratory capacity of CM. As expected, the media from M2 cells efficiently closed the wound in a time - and concentration-dependent manner, confirming that M2 macrophages display pro-tumoral signals that favor the aggressiveness of the HCC cells in our model.

Figure 6 Effects of the conditioned media from M2-like derived macrophages on the viability and migration of Huh7 cells. A-C: Cell viability was assessed using the Cell Counting Kit-8 kit (24 hours, 48 hours, and 72 hours, respectively), conditioned media from macrophages at different dilutions were used; D: Cell migration assay determined by a wound healing assay, representative images of at least three independent experiments. Scale bars: 300 μm (100 ×, original magnification); E: Plot of the percentage of migration determined by Image J software. Each column represents the mean ± SEM of at least three independent experiments in triplicate. ^aP ≤ 0.05 vs not treated cells. NT: Not treated cells; CM: Conditioned media; FBS: Fetal bovine serum.

CM from GDF11-treated M2-like macrophages lessens cell aggressiveness in HCC-derived cells

Based on our results that GDF11 induces a mitigation response in macrophages, we then aim to prove whether CM derived from M2-like macrophages treated with GDF11 can reduce the efficiency of wound closure. We observed that CM from M2-like macrophages treated with GDF11 (M2-like + GDF11 CM) hindered wound closure at 24 hours compared to M2-like macrophage CM (not treated with GDF11), indicating a reduction or mitigation in the aggressive phenotype of the HCC cell line (Figure 7A and B). All Huh7 cell wound groups repaired the wound before 48-72 hours. We did not find differences in cell proliferation and viability (data not shown), indicating that migration capacity is the primary process provided by macrophage secretion. Additional control groups and wound area quantification are presented in Supplementary Figure 4A and B.

Figure 7 Conditioned media from growth differentiation factor 11-treated macrophages mitigate migration capacity and proliferation in Huh7 cells. A: Cell migration assay determined by a wound healing assay, representative images of at least three independent experiments. Scale bars: 300 μm (100 ×, original magnification); B: Percentage of migration determined by Image J software; C: Cell migration assay determined by a wound healing assay, representative images of at least three independent experiments. Scale bars: 300 μm (100 ×, original magnification); D: Percentage of migration determined by Image J software; E: Cell proliferation; F: Cell viability at 24 hours, 48 hours, and 72 hours under different treatments; G: Cell migration assay determined by a wound healing assay using anti-GDF11. Scale bars: 300 μm (100 ×, original magnification); H: Percentage of migration determined by Image J software. Each column or point represents the mean ± SEM of at least three independent experiments carried out in triplicate. ^aP ≤ 0.05 vs M2-like conditioned media or post M2-like conditioned media; ^bP ≤ 0.05 vs post M2-like conditioned media; ^cP ≤ 0.05 vs not treated cells. Each column or point represents the mean ± SEM of at least three independent experiments carried out in triplicate. Anti-growth differentiation factor 11 was used to inhibit the remnants of the recombinant growth differentiation factor 11. NT: Not treated cells; CM: Conditioned media; GDF11: Growth differentiation factor 11.

Once we established the M2-like phenotype, we treated the cells for an additional 72 hours with GDF11 (Supplementary Figure 1), which we named post-M2-like macrophages. Figure 7C and D illustrate the effects of CM from GDF11-treated M2-macrophages (post M2-like + GDF11 CM). We found that the treatment delays wound closure compared with those without the GDF11 stimulus. Additional controls are presented in Supplementary Figure 4C and D. Consistent with these results, cell proliferation and viability decreased in treatments using post M2-like + GDF11 CM (Figure 7E and F), indicating again the reduction of aggressiveness provided by TAM. Extra controls for this experiment can be found in the Supplementary Figure 4E and F.

To rule out the possible effect of GDF11 remnants in the CM and cast doubt on the effect of macrophage secretion mediators, we supplied the CM with anti-GDF11 and explored migration at 72 hours. Figure 7G and H indeed show that the effect on migration is due to the mediators secreted by the macrophages present in the CM. Additional control treatments for this experiment are presented in the Supplementary Figure 5A and B. To test that the antibody effectively blocks GDF11, we performed a western blot of p-Smad3, shown in Supplementary Figure 6A and B. Anti-GDF11 successfully abolished the activation of the canonical signaling mediated by Smad3. The schematic representation of anti-GDF11 is shown in Supplementary Figure 6C. At this point, we demonstrated three relevant points: (1) M2-like macrophages induce tumor aggressiveness, such as proliferation and migration capacity on HCC cells; (2) GDF11 mitigates pro-tumoral properties on M2-like macrophages reflected on cancer cells; and (3) The possible GDF11 remnants in CM did not induce effects in our models (Supplementary Figure 7).

Finally, to gather more evidence and determine whether wound closures were due to a migration process rather than proliferation, we described the morphology in the healing process. This observation shows the extended appearance in Huh7 cultures characteristic of the migratory process. We excluded the use of proliferation or migratory inhibitors, such as cytarabine, which could reduce cell viability, as we confirmed, and therefore affect the study (Supplementary Figure 8).

Secretome changes induced by GDF11 on post M2-like macrophages

We determined the cytokine profile in the CM derived from our cultures to demonstrate these changes in specific secretome-treated macrophages. The cytokine array showed that the content of interleukins, chemokines, and growth factors was differentially affected by GDF11 stimulus in post M2-like macrophages (Figure 8; Supplementary Figure 9). Notably, we observed that GDF11 reduced the expression of IL-6, epithelial-derived neutrophil-activating protein-78 (ENA-78), and angiogenin, which are key proteins that promote aggressiveness. However, we found an increase in IL-1β, IL-8, IL-13, tumor necrosis factor-α, monocyte chemoattractant protein (MCP)-1, MCP-2, MCP-3, regulated on activation normal T cell expressed, growth-regulated oncogene α, vascular endothelial growth factor, and platelet-derived growth factor-BB, which demonstrates a pro-inflammatory profile.

Figure 8 Cytokine profile changes induced by growth differentiation factor 11 on post M2-like macrophages. A: Interleukin and interferon family alteration by growth differentiation factor 11 (GDF11) on post M2-like macrophages, ^aP < 0.0001 vs post M2-like conditioned media (CM), ^bP = 0.0012 vs post M2-like CM; B: Chemokine family alteration by GDF11 on post M2-like macrophages, ^aP < 0.0001 vs post M2-like CM, ^bP = 0.0050 vs post M2-like CM; C: Growth factors alteration by GDF11 on post M2-like macrophages, ^aP ≤ 0.0001 vs post M2-like CM. The mean pixel density in each column on the heat map represents the mean ± SEM of at least three independent experiments. CM: Conditioned media; GDF11: Growth differentiation factor 11.

DISCUSSION

Some studies show evidence that GDF11 is downregulated in various tumors, including breast, pancreatic, and liver cancers[24-26]; this could make sense due to GDF11 displaying antitumoral properties. We have previously reported that adding GDF11 to HCC-derived cells exhibits prominent anti-tumoral properties; these effects have also been corroborated in other cancer types, such as breast cancer[20]. Furthermore, the anti-cancer properties are also associated with changes in aberrant metabolism, resulting in a decrease in lipogenic enzymes, such as fatty acid synthase, and those related to the mevalonate pathway. Bioenergetic metabolism is also affected, as evidenced by mitochondrial dysfunction, as indicated by a decrement in the metabolic profile, as measured by ECAR and OCR[20,21]. Recent evidence supports our findings, showing that GDF11 inhibits this malignant progression by regulating the mechanistic target of rapamycin complex 1-autophagy axis or even inducing apoptosis through the ROS-c-Jun N-terminal kinase pathway[26,27].

Recently, we also found that GDF11 has mitigating effects on T-cell acute lymphoblastic leukemia-derived cells, in a mechanism dependent on the decrease of the transcription factor forkhead box P3, a key factor involved in immune system responses[28]. We observed a decrement in the migratory capacity, demonstrating the additional impact of this molecule not only on carcinomas[29]. We wondered about the possible effects that could exist on other leukocytes present in TIME, such as TAMs. This suggests that GDF11 could provide a promising approach as a multitarget factor. Previously, we identified the relevant infiltration of macrophages (CD206⁺) in liver tumors raised under metabolic dysfunction-associated steatohepatitis derived etiology, particularly pro-tumoral with M2-like polarization[6]. In the present work, we focused on characterizing the roles of GDF11 on TIME leukocytes, specifically, TAMs with M2-like polarization, due to its pleiotropic properties. The central dilemma regarding the mechanism between TGF-β and GDF11, which are pro-tumoral and anti-tumoral factors, respectively, and the canonical pathways through Smad 2/3 that are activated in both contexts[19,30,31]. Notably, TAMs are the leading TGF-β producers and induce M2 polarization through the canonical Smad signaling[16,32,33]. Strikingly, our study demonstrates this mitigation response using GDF11 without cytotoxic effects, activating the same canonical pathway.

We utilize THP-1 cells to induce differentiation into macrophages, addressing the effects elicited by GDF11, particularly in the context of HCC. The use of this monocytic cell line provides a suitable model for macrophage research, simulating diverse polarities and immunological effects. Furthermore, THP-1 more closely mimics native monocyte-derived macrophages than other human myeloid cell lines[34,35]. Additionally, it has been reported that monocyte-derived macrophages are more abundant than resident macrophages in both the early and late stages of HCC tumor development, using diethylnitrosamine and non-alcoholic steatohepatitis models[36]. There is considerable evidence that many stimuli lead to the M2-like phenotype, which is secreted by cancer cells through a crosstalk mechanism. We can mimic this by adding CM derived from HCC cells. Instead, we employed IL4/IL-13 to obtain a simpler and purer model of M2-like macrophages[37].

Why do we focus on mitigating the response provided by TAMs in HCC? The response is based on the nature of the liver. Hepatic macrophages (Kupffer cells), comprise 80%-90% of all macrophages in the body and cover up to 20% of non-parenchymal cells in this organ[38]. In the context of cancer, TAMs can comprise up to 50% of the tumor mass, depending on the type and grade of the tumor[9,39]. They provide signals that favor tumor progression, including growth, migration, and drug resistance factors[23]. Additionally, it has been reported that TAMs can also provide cholesterol to the tumor cells[40]. Cholesterol is a fundamental lipid for cell membranes, and others, such as hydroxycholesterol, act as a signal to regulate tumor aggressiveness[41]. Our data suggest that GDF11 reduces lipid synthesis in macrophages, potentially providing a mitigation strategy to prevent cholesterol donation to cancer cells.

Although the increase in ROS and the reduction of the CD206 marker are indicators of an M1-like phenotype, our data may only suggest a reduction in pro-tumoral behavior. It is well known that M1 macrophages generate more ROS than M2-like macrophages, which can engage tissue repair[42,43]. This fact may explain the mitigation of wound closure observed in GDF11-treated macrophages. On the other hand, the loss of CD206 or additional markers such as CD163 in cancer has been associated with a favorable prognosis in patients, as indicated by the high overall survival rates observed in intrahepatic cholangiocarcinoma, HCC, and other tumors[11,44-46].

As expected, we also observed changes in metabolic status with GDF11, which reset M2-like behavior. This finding could contradict the existing literature, as mitochondria and fatty acid oxidation are considered the primary energy sources in M2 macrophages, as reflected in OCR levels[47]. We hypothesized that this is due to cell line features, as neoplastic cells (acute monocytic leukemia) differ from bone marrow-derived macrophages in their characterization compared to primary non-neoplastic cells[48]. Furthermore, numerous protocols and conditions reported in the literature regarding the differentiation and polarization of macrophages may impact the response due to variations in stimulation time and concentration[49]. Therefore, this data suggests that mitochondria play an essential role in macrophage polarization, as mitochondria represent an immunological organelle[50].

We mentioned that M2-like macrophages are responsible for secreting pro-tumoral factors that induce aggressive behavior in HCC cells[23,51]. This led to the finding that this secretory response boosted the migratory capacity and proliferation of our in vitro models. We used Huh7 cells as a model for HCC cells due to their high proliferation and aggressive characteristics. Moreover, this cell line was used to demonstrate the effects observed in our previous publications, which were similar when using different cell lines[20,21]. Using CM derived from GDF11-treated M2-like macrophages was the key protocol for demonstrating the reduction in aggressive behavior in HCC cells, showing that the secretion could influence malignant processes. However, we noted neutralizing effects, rather than a reduction in the aggressive parameters (proliferation and migration capacity) vs NT cells. These outcomes suggest that GDF11-treated M2-like macrophages are resting their secretory capacity, or, as mentioned before, high ROS production by antitumor phagocytes hinders wound closure as a cytotoxic stimulus[42]. In summary, our data suggest that GDF11 neutralizes the pro-tumoral secretion derived from TAMs.

Later, we demonstrated that GDF11 affected the secretory factors of M2-like macrophages, specifically downregulating IL-6, ENA-78, and angiogenin. IL-6 is a pleiotropic cytokine that possesses pro-inflammatory effects. IL-6 upregulates IL6-R, which increases the aggressiveness of HCC[52]. Furthermore, IL-6 stimulates M2 polarization in other cancers[53]. GDF11 reduced IL-6, indicating a mitigation of this aggressiveness. ENA-78, also known as CXCL5, an indicator of tumor progression and facilitator of tumor growth and invasion[54,55], was reduced using GDF11. Similarly, angiogenin, which IL-6 could drive[56,57], was mitigated using GDF11. The increase in additional chemokines and growth factors induced by GDF11 may suggest that other family molecules could also induce anti-tumoral effects in the CM derived from GDF11-treated macrophages; hence, we can investigate this further through molecular sequencing or additional studies.

Due to the limited effectiveness of immunotherapy against certain types of liver cancer[58], such as metabolic dysfunction-associated steatohepatitis-HCC, the most aggressive form[59], we need to improve our treatment strategies. Our findings suggest that GDF11 could be a promising therapeutic option; however, more information is required, especially regarding the variety of leukocytes in an in vivo model. Nonetheless, we demonstrate that this factor reduces several parameters of tumor and inflammatory cell aggressiveness, including TAMs, one of the most critical tumor-infiltrating leukocytes, which could play a role in a potential treatment.

Finally, there is additional evidence regarding GDF11 in the context of other pathologies, particularly in relation to macrophage polarization, such as severe acute pancreatitis and ischemia-reperfusion-induced acute kidney injury[60,61], which affects M1/M2 polarization balance. The other mentioned GDF11 works as an anti-inflammatory molecule, reducing inflammatory arthritis[62], or antagonizes psoriasis-like skin inflammation[63]. Our current publication represents the first report in the context of GDF11 and macrophages in HCC. This could represent a novelty for the mechanism or effects; however, our data differ due to the experimental strategy employed. Our graphic abstract summarizes the study (Supplementary Figure 10).

CONCLUSION

This research provides evidence of GDF11’s dual effect on TAMs and HCC cells, as previously demonstrated in studies, and reveals the individual and interactive effects of its secretion. GDF11 decreases aggressiveness in HCC cells and reduces the influence of leukocyte components, such as TAMs, on M2-like polarization. This information could be important for developing new therapeutic targets to enhance current immunotherapy.

ACKNOWLEDGEMENTS

The authors would like to acknowledge the Confocal Microscopy Core Unit at Universidad Autónoma Metropolitana-Iztapalapa.