Research ArticleIMMUNOLOGY

IGF2R-initiated proton rechanneling dictates an anti-inflammatory property in macrophages

See allHide authors and affiliations

Science Advances  25 Nov 2020:
Vol. 6, no. 48, eabb7389
DOI: 10.1126/sciadv.abb7389


Metabolic traits of macrophages can be rewired by insulin-like growth factor 2 (IGF2); however, how IGF2 modulates macrophage cellular dynamics and functionality remains unclear. We demonstrate that IGF2 exhibits dual and opposing roles in controlling inflammatory phenotypes in macrophages by regulating glucose metabolism, relying on the dominant activation of the IGF2 receptor (IGF2R) by low-dose IGF2 (L-IGF2) and IGF1R by high-dose IGF2. IGF2R activation leads to proton rechanneling to the mitochondrial intermembrane space and enables sustained oxidative phosphorylation. Mechanistically, L-IGF2 induces nucleus translocation of IGF2R that promotes Dnmt3a-mediated DNA methylation by activating GSK3α/β and subsequently impairs expression of vacuolar-type H+-ATPase (v-ATPase). This sequestrated assembly of v-ATPase inhibits the channeling of protons to lysosomes and leads to their rechanneling to mitochondria. An IGF2R-specific IGF2 mutant induces only the anti-inflammatory response and inhibits colitis progression. Together, our findings highlight a previously unidentified role of IGF2R activation in dictating anti-inflammatory macrophages.


Monocytes patrolling the bloodstream emigrate into adjacent tissues where they mature into macrophages and acquire either a pro-inflammatory or anti-inflammatory phenotype (1, 2). Emerging evidence has demonstrated that these divergent immune phenotypes in macrophages can persist for a prolonged period, a phenomenon termed “innate immune memory” (35). Epigenetic modifications and metabolic rewiring are critical for building this innate immune memory (6, 7). It has been established that pro-inflammatory macrophages use mainly aerobic glycolysis for their energy needs (4), while mitochondria-driven oxidative phosphorylation (OXPHOS) predominates in anti-inflammatory macrophages (8). Although mitochondrial metabolism and dynamics are known to be closely linked to monocyte homeostasis and macrophage maturation (9), little is known about how external stimulation affects the outcome of macrophage maturation by modulating mitochondria activities.

Recently, it was demonstrated that stimulation through insulin-like growth factor 1 receptor (IGF1R) endows macrophages with a persistent pro-inflammatory phenotype (6, 10). This pro-inflammatory phenotype can be induced by many IGF1R ligands and factors, such as insulin-like growth factor 1 (IGF1), mevalonate, and apoptotic bodies (6, 11, 12). In contrast, IGF2, a mitogenic polypeptide, can bind to either IGF1R or IGF2R (11). While ligation of IGF1R is believed to mediate many of effects of IGF2 in regulating somatic growth, development, and tissue repair (13, 14), the mechanism of IGF2R function is largely unknown, beyond its role as a decoy receptor or carrier protein (11).

Here, we show that IGF2 ligation of IGF2R during macrophage maturation induces the rechanneling of protons from cytosol to mitochondria initiating the reprogramming of cellular metabolism toward OXPHOS, and thus resulting in an anti-inflammatory phenotype. At the same time, however, very robust activation of IGF1R by IGF2 imposes a strong bias toward aerobic glycolysis, thus countering any anti-inflammatory–inducing effect of IGF2R. These findings reveal an unappreciated biological function of the IGF2-IGF2R axis in regulating the fate of maturing macrophages to acquire either a pro- or anti-inflammatory phenotype. Therefore, targeting IGF2R activation can potentially be used to reduce inflammation in the treatment of inflammatory disease.


IGF2 exhibits dual opposing roles in determining macrophage phenotype

Our recent study demonstrated that IGF2 alleviates both experimental autoimmune encephalomyelitis (EAE) and dextran sulfate sodium (DSS)–induced colitis by programing macrophages to acquire an anti-inflammatory phenotype (15). When IGF2 was applied to treat colitis, we found that its anti-inflammatory effect occurred only in a specific dose range. Administration of L-IGF2 (≤50 ng per mouse; L-IGF2) reduced the severity of DSS-induced colitis, as revealed by significant improvement in body weight, survival rate, stool score, bleeding score, colon length, as well as inhibition of mononuclear cell infiltration into the colon (Fig. 1, A to D, and fig. S1A). Histopathological analysis showed obvious inhibition of DSS-induced colon damage by L-IGF2 (Fig. 1C). Consistent with our previous results, L-IGF2 administration inhibited immune cell infiltration into colonic lamina propria, with a significant reduction in interleukin-1β (IL-1β)–expressing CD11b+F4/80+ macrophages (Fig. 1E). In contrast, H-IGF2 (1000 ng per mouse; H-IGF2) failed to ameliorate colitis and instead exacerbated its progression and promoted inflammation and infiltration by IL-1β–expressing CD11b+F4/80+ macrophages (Fig. 1, A to E, and fig. S1A).

Fig. 1 IGF2 exhibits dual role in shaping macrophage activation phenotype.

(A and B) Changes in body weight and survival rate of mice with DSS-induced colitis treated with IGF2 at low doses (≤50 ng per mouse; L-IGF2) or high dose (1000 ng per mouse; H-IGF2), compared to phosphate-buffered saline (PBS)–treated controls. (C) Representative hematoxylin and eosin (H&E) staining showing tissue damage and immune cell infiltration into the colons of DSS-induced colitis mice after treatment with low- or high-dose IGF2 (L-IGF2 or H-IGF2, respectively) for three times. Scale bars, 100 μm. (D) Extent of colonic infiltration of mononuclear cells was quantified as the total number of mononuclear cells per colon in DSS-induced colitis mice after treatment with low- or high-dose IGF2. (E) IL-1β–positive cell percentage in macrophages isolated from the colons of DSS-induced colitis mice treated with low- or H-IGF2, as revealed by immunofluorescence staining and flow cytometric analysis. PE, phycoerythrin; FITC, fluorescein isothiocyanate. (F and G) Changes in body weight and survival rate of DSS-induced colitis mice treated with adoptive transfer of IGF2-induced PMs. PMs were isolated from mice with Ftg-induced peritonitis after two times of low- or H-IGF2 administration. (H) IL-1β–positive cell percentage and (I) PD-L1 expression in PMs isolated from Ftg-induced peritonitis mice treated with low- or H-IGF2, as revealed by immunofluorescence staining and flow cytometric analysis. MFI, mean fluorescence intensity. (J) Relative mRNA expression of iNOS and Retnla in PMs isolated from peritonitis mice treated with low- or H-IGF2. PMs were stimulated with or without LPS or IL-4/IL-13 for 6 or 24 hours in vitro. mRNA was assayed by real-time polymerase chain reaction (PCR). (K) Changes in body weight of DSS-induced colitis mice treated with IGF1 at low doses (≤50 ng per mouse; L-IGF1) or high dose (1000 ng per mouse; H-IGF1). All data are given as mean ± SEM, n ≥ 3. All results shown are representative of three independent experiments. *P < 0.05, **P < 0.01, ***P < 0.001.

Given the important role of IGF2 in macrophage regulation, we wondered whether the divergent effects of IGF2 on colitis severity can be attributed to its action on macrophages. To address this question, we used adoptive transfer of the purified peritoneal macrophages (PMs) and confirmed the localization of these transferred cells in the inflamed colon (fig. S1, B and C). Donor mice were injected with thioglycollate broth media (Ftg) to induce peritonitis and then treated with IGF2 at either low doses (L-IGF2) or high doses (H-IGF2). Macrophages were then isolated from the peritoneal cavity of these mice and injected into mice with DSS-induced colitis. We found that adoptively transferred PMs from the L-IGF2–treated animals ameliorated disease severity and prolonged mouse survival (Fig. 1, F and G). However, such therapeutic effects were not achieved by adoptive transfer of PMs from H-IGF2–treated mice and worsened disease severity (Fig. 1, F and G). Thus, L-IGF2–induced macrophages were anti-inflammatory, while H-IGF2–induced macrophages were pro-inflammatory. Accordingly, we examined the PMs used for adoptive transfer. Consistent with our previous study (15) and with observations from a mouse colitis model, macrophages induced by L-IGF2 exhibited low levels of IL-1β and high levels of PD-L1 expression (Fig. 1, H and I). Such effects were reversed by H-IGF2 induction (Fig. 1, H and I). To further confirm their anti- and pro-inflammatory functions, PMs induced with L-IGF2 or H-IGF2 were stimulated with lipopolysaccharide (LPS) or IL-4/IL-13 in vitro. The L-IGF2–induced PMs maintained their anti-inflammatory status, as demonstrated by reduced mRNA expression of iNOS, Cxcl10, Cd86, Il18, Tnfa, and Cd40 markers that correlate with pro-inflammatory macrophage activation and enhanced mRNA expression of Retnla, Ym1, Ccl22, and Clec7a, which are associated with anti-inflammatory macrophage activation (Fig. 1J and fig. S1D). H-IGF2 promoted the expression of Cd40 but had little influence on the expression of iNOS, Cxcl10, Retnla, Ym1, Ccl22, and Clec7a. On the other hand, H-IGF2 weakly inhibited the expression of Cd86, Il18, and Tnfa in PMs (Fig. 1J and fig. S1D). Therefore, IGF2 has dual opposing roles in determining the anti- or pro-inflammatory phenotype of activated macrophages and subsequently their effects on colitis progression.

IGF2 is considered to be an analog of IGF1 (16). Therefore, we next examined the effect of IGF1 on DSS-induced colitis. Similar to H-IGF2, administration of IGF1 exacerbated the progression of colitis in a dose-dependent manner: In comparison to a control group, mice treated with IGF1 showed more severe signs of colitis, such as weight loss, decreased survival, and reduced colon length (Fig. 1K and fig. S1, E to G). IGF1 was also found to increase IL-1β expression and inhibit PD-L1 expression in PMs (fig. S1, H and I). In addition, upon stimulation with LPS, IGF1-treated PMs generated more nitric oxide and lactic acid and consumed more glucose, compared to controls (fig. S1, J to L). Together, these data reveal that IGF2, unlike IGF1, can promote the development of either anti- or pro-inflammatory macrophages, depending on the dose. Thus, IGF2 has dual opposing roles in programming macrophages to acquire an anti- or pro-inflammatory phenotype.

IGF2 modulates the phenotype of macrophages via IGF2R and IGF1R in context of dosage

To parse out the distinct effects of IGF2 and IGF1 on the development of the inflammation-regulating macrophages, we investigated downstream signaling from their receptors. IGF1R is known to bind both IGF1 and IGF2. In contrast, IGF2R can only be ligated by IGF2, and this occurs with much higher affinity than to IGF1R (11). To delineate the role of IGF1R and IGF2R in regulating inflammation, these receptors were genetically deleted using a myeloid cell–specific conditional knockout system. We established mice with myeloid cell–specific ablation of Igf1r (IGF1Rfl/flLyz2Cre, IGF1RCKO) or Igf2r (IGF2Rfl/flLyz2Cre, IGF2RCKO) (fig. S2, A and B), using the Lyz2-Cre conditional knockout cassette, which resulted in specific deletion of IGF1R or IGF2R, respectively, in macrophages (17). We found that IGF1Rfl/flLyz2Cre mice had less severe signs after induction of DSS-induced colitis, including better body weight and prolonged survival (Fig. 2, A to F, and fig. S2C), demonstrating that IGF1R deficiency in macrophages confers resistance to colitis. H-IGF2–driven exacerbation of colitis progression was not observed in IGF1Rfl/flLyz2Cre mice, compared to controls (Fig. 2, A to F, and fig. S2C), indicating that IGF1R mediates the H-IGF2–induced macrophage biasing toward a pro-inflammatory phenotype.

Fig. 2 IGF2 modulates macrophage phenotype via IGF2R and IGF1R in context of dosage.

(A to D) Changes in body weight, survival rate, stool score, and bleeding score for mice with conditional knockout of IGF1R (IGF1Rfl/flLyz2Cre; IGF1RCKO), compared to wild-type (WT) mice (IGF1Rfl/fl). Mice were treated with DSS to induce colitis and then with or without H-IGF2. (E) Representative H&E stains showing tissue damage and immune cell infiltration into colons of mice treated as in (A) to (D). Scale bars, 100 μm. (F) Total numbers of mononuclear cells counted per colon isolated from mice treated as in (A) to (D). (G to J) Changes in body weight, survival rate, stool score, and bleeding score for mice with conditional knockout of IGF2R (IGF2Rfl/flLyz2Cre; IGF2RCKO) compared to WT mice (IGF2Rfl/fl). Mice were treated with DSS to induce colitis and then treated with or without low-dose IGF2 (L-IGF2). (K) Representative H&E stains showing tissue damage and immune cell infiltration in the colons of mice treated as in (G) to (J). Scale bars, 100 μm. (L) Total numbers of mononuclear cells counted per colon isolated from mice treated as in (G) to (J). (M and N) PD-L1 expression and IL-1β–positive cell percentage in PMs isolated from IGF1RCKO and WT mice with Ftg-induced peritonitis, treated with or without H-IGF2. (O and P) PD-L1 expression and IL-1β–positive cell percentage in PMs isolated from IGF2RCKO and WT mice with Ftg-induced peritonitis, treated with or without L-IGF2. All data are given as mean ± SEM, n ≥ 3. All results shown are representative of three independent experiments. *P < 0.05, **P < 0.01, ***P < 0.001.

After induction of colitis, although no difference in disease severity was observed between IGF2Rfl/flLyz2Cre mice and their littermate controls, the therapeutic effect of L-IGF2 in treating colitis was abolished in mice with macrophage-specific IGF2R ablation. When L-IGF2 was administered to IGF2Rfl/flLyz2Cre mice, they showed more severe disease, including weight loss, poorer survival, and increased stool scores and bleeding scores, as well as more robust mononuclear cell infiltration into colonic lamina propria (Fig. 2, G to L, and fig. S2D), in comparison to controls. Also, H-IGF2 aggravated colitis in the mice with IGF2R deficiency in myeloid cells (fig. S2, E and F). These data suggest that the L-IGF2–driven programming of macrophages toward an anti-inflammatory phenotype is mediated by IGF2R. Further analysis showed that IGF1R deletion resulted in PMs expressing greater levels of PD-L1 and diminished levels of IL-1β in response to H-IGF2 (Fig. 2, M and N). In marked contrast, deletion of IGF2R reverted the L-IGF2–enhanced expression of PD-L1 in PMs and dramatically reversed the inhibition of L-IGF2 on IL-1β expression (Fig. 2, O and P). Together, these results show that IGF2R activation by L1 IGF2 can induce macrophages to acquire an anti-inflammatory phenotype, whereas H-IGF2 acts on IGF1R to drive cells toward a pro-inflammatory phenotype, further supporting the idea that IGF1R activation can counterbalance signaling from IGF2R.

IGF2R activation sets macrophages in a metabolic preference to OXPHOS

The preferred glucose metabolic pathway used by macrophages is determined during their activation and polarization into pro-inflammatory or anti-inflammatory macrophages. Pro-inflammatory macrophages preferentially use aerobic glycolysis to generate adenosine triphosphate (ATP) (4), whereas anti-inflammatory macrophages depend mostly on OXPHOS for their energy needs (8). To determine which glycolytic pathways are active in PMs isolated from peritonitis mice treated with low- or high-dose IGF2 administration, their oxygen consumption rates (OCRs) and extracellular acidification rates (ECARs) were analyzed using the Seahorse platform, with M1 and M2 macrophages as references (fig. S3, A and B) (18). No differences were observed between H-IGF2–treated macrophages and control macrophages (Fig. 3, A to C). Notably, however, L-IGF2–treated macrophages exhibited a greatly enhanced OCR, a lower ECAR, and a higher OCR to ECAR ratio (Fig. 3, A to C). These L-IGF2–treated macrophages produced less lactic acid, even in response to LPS stimulation, which normally induces glycolysis (Fig. 3D). Thus, macrophages induced by L-IGF2 show a preference for OXPHOS, while H-IGF2–induced macrophages use mainly glycolysis to generate energy.

Fig. 3 IGF2R activation biases macrophages toward the OXPHOS metabolic pathway.

(A) OCR, (B) ECAR, and (C) ratio of OCR to ECAR for PMs isolated from peritonitis mice treated with IGF2 at a low dose (L-IGF2) or high dose (H-IGF2). (D) Lactic acid production by PMs from peritonitis mice treated with low- or high-dose-IGF2, with or without LPS stimulation for 24 hours in vitro. (E and F) LPS-induced lactic acid production in vitro by PMs treated with IGF2 in the presence or absence of IGF1R inhibitor (NVP, NVP-AEW541) or IGF2R-blocking antibody (anti-IGF2R). (G to J) Lactic acid production and glucose consumption by PMs from mice with conditional knockout of IGF1R (IGF1Rfl/flLyz2Cre; IGF1RCKO) or IGF2R (IGF2Rfl/flLyz2Cre; IGF2RCKO), with treatment of low- or high-dose-IGF2. These cells were stimulated with or without LPS for 24 hours in vitro. All data are given as mean ± SEM, n ≥ 3. All results shown are representative of three independent experiments. *P < 0.05, **P < 0.01, ***P < 0.001.

Using an IGF1R inhibitor (NVP, NVP-AEW541) or IGF2R-blocking antibody (anti-IGF2R), we further verified that specific ligation of IGF2R confers a preference for OXPHOS metabolism and an anti-inflammatory phenotype in macrophages, while H-IGF2–driven IGF1R signaling leads to utilization of glycolysis (Fig. 3, E and F). Similar results were obtained when using macrophages with IGF1R- or IGF2R-specific deletion (Fig. 3, G to J). Also, the metabolic regulation in IGF1-treated PMs was similar to H-IGF2–treated PMs, exhibiting high levels of nitric oxide secretion, lactic acid production, and glucose consumption (fig. S1, J to L). Thus, distinct from the activation of IGF1R by IGF1 or H-IGF2, activation of IGF2R by L-IGF2 confers on macrophages a metabolic preference for OXPHOS.

IGF2R activation induces OXPHOS via proton rechanneling

Our previous studies demonstrated that IGF2 induces the development of an anti-inflammatory phenotype in macrophages during their maturation process (15). We therefore classified PMs from peritonitis mice into two groups based on their expression levels of macrophage markers and flow cytometric laser light scatter characteristics: maturing macrophages (include monocytes), which are FSClowCD11blowF4/80mid, and mature macrophages, which are FSChighCD11bhighF4/80high (fig. S3C). Unlike the mitochondrial depolarization observed upon LPS stimulation (fig. S3, D and E) (19), L-IGF2 was found to significantly increase mitochondrial membrane potential (MMP), a characteristic of energy generation through OXPHOS, in both maturing and mature PMs from IGF2-treated peritonitis mice (Fig. 4A and fig. S3F). In addition, the mean fluorescence intensity (MFI) ratio of TMRE (tetramethylrhodamine, ethyl ester) and mito-tracker also revealed that L-IGF2 increased the MMP and the quantity of mitochondria (fig. S3, G to J). Thus, MMP up-regulation in macrophages treated with L-IGF2 may occur during the process of macrophage maturation.

Fig. 4 Activation of IGF2R induces OXPHOS via proton rechanneling in maturing macrophages.

(A) MMP (measured as the ratio of red:green MFI emissions after JC1 staining) in maturing and mature PMs isolated from peritonitis mice treated with low- or high-dose IGF2. Mφ, macrophages. (B) Effect of IGF2 on levels of cytosolic pH in PMs, as indicated by pHrodo Green staining; the white bar charts are pH standard. (C) Effects of the fine-tuned cytosolic pH on MMP in PMs. (D) Lysosomal pH in PMs from peritonitis mice treated with IGF2 doses. (E) V-ATPase activity in lysosomes of PMs from peritonitis mice treated with IGF2 doses. (F to H) Levels of cytosolic pH (F), lysosomal pH (G), and MMP (H) in PMs after BafA1 treatment for 0.5 hours in vitro. (I to K) Glucose consumption (I), lactic acid production (J), and nitric oxide secretion (K) by PMs treated with or without LPS for 3 hours after BafA1 treatment in vitro. (L) Heat map showing changes of cytosolic pH in THP1 cells treated with or without IGF2 during their maturation. Among the THP1 cells, FSClow cells are maturing cells and FSChigh cells are mature cells. (M) Cytosolic pH and (N) MMP in BMDMs from Lyz2Cre (WT) mice, IGF1Rfl/flLyz2Cre (IGF1RCKO) mice, and IGF2Rfl/flLyz2Cre (IGF2RCKO) mice, with or without IGF2 addition during their maturation. (O) Cytosolic pH (indicated by pHrodo Green staining) in BMDMs treated with IGF2 doses with or without IGF2R-blocking antibody during their maturation. All data are given as mean ± SEM, n ≥ 3. All results shown are representative of three independent experiments. *P < 0.05, **P < 0.01, ***P < 0.001.

Mitochondria generate ATP by coupling the tricarboxylic acid cycle (TCAC), electron transport chain (ETC), and OXPHOS, a process regulated by the NAD+ (nicotinamide adenine dinucleotide)/NADH (reduced form of NAD+) ratio (20). We then evaluated the amounts of NAD+ and NADH in phosphate-buffered saline (PBS)– and IGF2-treated maturing macrophages and calculated the ratio of NADH/NAD+. When the amount of NADH was compared among PBS-, H-IGF2–, and L-IGF2–treated maturing macrophages, there is little difference. However, L-IGF2–treated maturing macrophages had lower amount of NAD+. This resulted in an enhanced ratio of NADH/NAD+ (fig. S3K). Also, there was no difference in the activity of mitochondrial complex I among maturing macrophages treated with PBS or different amounts of IGF2 (fig. S3L). Therefore, we excluded the possibility that the substrate-driven fueling for OXPHOS induced by L-IGF2 is dependent on the TCAC and ETC activities. Because the outer mitochondrial membrane is highly permeable to cytosolic content (21), we further investigated whether L-IGF2–boosted MMP results from utilization of cytosolic protons. Using pHrodo Green staining, a lower pH value (as low as 5.5) was observed in the cytosol of L-IGF2–treated maturing and mature macrophages (Fig. 4B). To verify this finding, we fine-tuned the pH value in the macrophage cytosolic compartment and found that cytosolic acidification could elevate the MMP (Fig. 4C). However, we found that L-IGF2 did not acidify the cytosol or lysosomes, nor did it alter MMP in mature macrophages (fig. S4, A to C). These results suggest that L-IGF2 can preprogram a metabolic preference of OXPHOS in maturing macrophages via cytosolic acidification.

Because maintenance of the cytosolic proton balance relies on acidic organelles, especially lysosomes, the effect of IGF2 on lysosomal pH was examined. We found that lysosomes in L-IGF2–treated PMs were enlarged (fig. S4D) and exhibited a higher pH (around 6.5) than control macrophages (pH 4.5 to 5.5), indicating an impairment in the channeling of protons into lysosomes (Fig. 4D). Consistently, compared to control and H-IGF2–treated PMs, L-IGF2–treated PMs showed a much lower activity of vacuolar H+-dependent adenosine triphosphatase (v-ATPase) (Fig. 4E), an enzyme responsible for pumping protons from the cytosol into lysosomes. Inhibition of v-ATPase with bafilomycin A1 (BafA1) effectively allowed proton accumulation in the cytosol and subsequent rechanneling of protons into mitochondria in mature macrophages (Fig. 4, F and G). Like L-IGF2–treated PMs, these BafA1-treated mature macrophages had a higher MMP (Fig. 4H). In response to a 3-hour exposure to LPS, these BafA1-treated macrophages consumed less glucose, produced less lactic acid and nitric oxide, and maintained a metabolic inclination for OCR (Fig. 4, I to K, and fig. S4, E and F). These data strongly suggest that inhibition of v-ATPase in macrophages promoted OXPHOS inclination in a proton rechanneling–dependent manner. Therefore, enhanced MMP in L-IGF2–treated maturing PMs is associated with rechanneling of protons into the mitochondria during macrophage maturation. Similar results were obtained with mouse bone marrow–derived macrophages (BMDMs) and THP1 cells, a human monocytic cell line, with L-IGF2 added during their maturation (fig. S4, G to I). IGF2-induced proton rechanneling could be detected even after 7 days in both THP1 cells and PMs (Fig. 4L and fig. S4, J to L).

Given that macrophage activation through IGF1R and IGF2R results in differential rewiring of their metabolic status, we further analyzed this effect in IGF1R-knockout and IGF2R-knockout BMDMs (derived from IGF1Rfl/flLyz2Cre mice and IGF2Rfl/flLyz2Cre mice, respectively) by comparing the level of cytosolic protons and MMPs, with or without IGF2 treatment. Compared to controls, greater cytosolic proton accumulation was observed in IGF1R-knockout macrophages (Fig. 4, M and N). Controversially, decreased cytosolic proton accumulation occurred in IGF2R-knockout macrophages (Fig. 4, M and N). Also, consistently, antibody blockade of IGF2R abolished cytosolic proton accumulation in maturing L-IGF2–treated BMDMs (Fig. 4O). However, no effect of IGF2R blockade was observed in H-IGF2–treated macrophages (Fig. 4O). Therefore, activation of macrophages via IGF2R, but not IGF1R, during their maturation induces proton rechanneling and presets their metabolic preference for OXPHOS.

Nuclear translocation of IGF2R induces proton rechanneling via GSK3 signaling

We next investigated how IGF2R activation initiates proton rechanneling. Adding IGF2 to THP1 cells substantially decreased the cell surface expression of IGF2R in a dose-dependent manner (Fig. 5, A and B). Using the subcellular protein fractionation assay (22), we then analyzed the distribution of IGF2R among different cellular components. We detected IGF2R in the nucleus after treatment with either low-dose or high-dose-IGF2 (Fig. 5C), indicating that ligation of IGF2R by IGF2 can induce its nuclear localization. This was further supported by immunofluorescence staining (Fig. 5D). However, we cannot observe notable nuclear translocation of IGF1R in PMs treated with IGF2 (fig. S5A). Moreover, treatment with ivermectin, an inhibitor of the nuclear transport receptor importin-α/β (23), blocked nuclear translocation of IGF2R and reduced IGF2R-induced proton rechanneling in maturing BMDMs (Fig. 5, E and F).

Fig. 5 Nuclear translocation of IGF2R induces proton rechanneling via GSK3 signaling.

(A and B) Cell surface expression of IGF2R by THP1 cells after treatment with graded doses of IGF2 (two low and two high doses) for 24 hours. (C) Immunoblotting analysis of IGF2R from THP1 cells treated with graded doses of IGF2 reveals that its distribution between total cellular and nuclear-associated is dose dependent. (D) Representative immunofluorescent images of IGF2R staining in PMs isolated from peritonitis mice treated with low- or high-dose IGF2. IGF2R is stained red, and nucleus is stained blue. Scale bars, 5 μm. (E) Immunoblotting analysis of nuclear IGF2R expression in BMDMs treated with low- or high-dose IGF2 during maturation, in the presence or absence of ivermectin (5 μM), an inhibitor of protein nuclear translocation. (F) Cytosolic pH in BMDMs treated with low- or high-dose IGF2 during maturation, with or without ivermectin. (G) The distribution of GSK3α/β among subcellular components of THP1 cells treated with graded doses of IGF2 as revealed by immunoblotting analysis. (H) Immunoblotting analysis of chromatin-bound GSK3α/β in L-IGF2-treated PMs, with or without GSK3α/β inhibitor (GSK3i). (I) Effect of GSK3α/β inhibition on cytosolic pH in IGF2-treated PMs. (J) Immunoblotting analysis of total and phosphorylated IGF1R, AKT, GSK3α, GSK3β, and mTOR in THP1 cells treated with graded doses of IGF2. All results shown are representative of three independent experiments. All data are given as mean ± SEM, n ≥ 3. ***P < 0.001.

Murine and human IGF2R share a similar motif for classical nuclear localization (fig. S5, B to D). According to database annotations and automatic text mining of the biomedical literature, the ability of IGF2R to localize to the nucleus was predicted (fig. S5, E and F). However, no consensus DNA or RNA binding sequences were predicted (fig. S5, G and H). Previous studies have demonstrated that IGF2 is necessary for memory consolidation and enhancement through the IGF2R–glycogen synthase kinase 3β (GSK3β) signaling process (14), and, ironically, GSK3 can be negatively regulated by IGF1R-phosphorylated AKT (24). We therefore checked GSK3α/β expression in different subcellular components and found much more chromatin-bound GSK3α/β in L-IGF2–treated macrophages (Fig. 5G). GSK3 inhibitor (SB216763) abolished L-IGF2–induced proton rechanneling and suppression of pro-inflammatory gene by decreasing chromatin-bound GSK3α/β in PMs (Fig. 5, H and I, fig. S6, A to D). Together with the inhibition of ivermectin in chromatin-bound GSK3α/β, we conclude that L-IGF2 induces GSK3α/β binding with chromatin via IGF2R (fig. S6, E to G). These results suggest that GSK3α/β is critical for maintenance of the high levels of cytoplasmic protons in L-IGF2–treated macrophages.

We next asked why chromatin-bound GSK3α/β is decreased in the presence of high doses of IGF2. We found that H-IGF2 triggers IGF1R and activates its related AKT-mTOR (mammalian target of rapamycin) kinase signaling, which is responsible for the phosphorylation and inactivation of GSK3 (Fig. 5J and fig. S6H), leading to diminution of IGF2R function, even though nuclear translocation of IGF2R still occurred (Fig. 5, A to D). Together, this indicates that L-IGF2–induced proton rechanneling in maturing macrophages relies on the nuclear translocation of IGF2R and its subsequent promotion of GSK3 binding to chromatin. This mechanism regulates the metabolic bias toward OXPHOS in macrophages.

GSK3-Dnmt3a induced by L-IGF2 controls DNA hypermethylation

Epigenetic regulation plays a critical role in controlling immune phenotype development during monocyte-to-macrophage maturation, and the development of “trained immunity” (7). In IGF2-treated PMs and THP1 cells, we found that L-IGF2 enhanced the signal intensity of 5-methylcytosine (5-mc) and decreased the level of 5-hydroxymethylcytosine (5-hmc), while H-IGF2 had the opposite effect (Fig. 6A and fig. S7, A to D), suggesting that notable levels of hypermethylation can be induced by L-IGF2. Because a previous study has reported that DNA methylation can be regulated by MEK (mitogen-activated protein kinase kinase) and GSK3β (25), we next determined the requirement for GSK3α/β signaling in L-IGF2–induced hypermethylation in macrophages. We found that application of a GSK3α/β inhibitor abolished DNA hypermethylation in L-IGF2–treated PMs, accompanied by a reduction in Dnmt3a expression (Fig. 6, B to D). This is consistent with previous findings that deficiency in both GSK3α and GSK3β in mouse embryonic stem cells results in inhibition of Dnmt3a2, a de novo DNA methyltransferase (26).

Fig. 6 L-IGF2–induced DNA hypermethylation is mediated by GSK3 and correlates with impaired expression of ATPase subunits.

(A) Expression of 5-mc (red) and 5-hmc (green) in PMs isolated from peritonitis mice treated with or without IGF2, as revealed by immunofluorescence staining. Blue marks nucleus. Scale bars, 5 μm. (B to D) Effect of GSK3α/β inhibitor (GSK3i) on the expression level of 5-mc, 5-hmc, and Dnmt3a in PMs treated with low- and high-dose IGF2. (E) For low- and high-dose IGF2-treated cells, enrichment for overlapping DMP genes involved in the energy metabolic network was analyzed by clustering and mapping using OmicsNet. (F) Changes in DNA methylation in CpG, CHH, and CHG for genes related to v-ATPase in PMs treated low- and high-dose IGF2. Change rate < 0 indicates decreased methylation; change rate > 0 indicates increased methylation. (G) Immunoblotting analysis of the indicated v-ATPase–associated genes in low- or high-dose IGF2–treated PMs. Data in (A) to (D) and (G) are representative of three independent experiments. All data are given as mean ± SEM, n ≥ 3. ***P < 0.001.

It has been shown that the presence of multiple methylation sites at CpG islands of gene promoters leads to gene silencing at the transcriptional level (27). To examine the genome-wide methylation rewiring programmed by L-IGF2, PMs with or without IGF2 treatment were subjected to whole-genome bisulfite sequencing to determine the methylation level at gene promoters. Compared to PBS-treated PMs, we detected 8758 genes with differential methylation at promoters (DMP genes) in L-IGF2–treated PMs and 8674 DMP genes in H-IGF2–treated PMs (fig. S7E). A total of 4406 genes were overlapped among groups (fig. S7E). Although the numbers of up- or down-regulated DMP genes were similar between L-IGF2 and H-IGF2 groups (fig. S7E), only L-IGF2–treated macrophages exhibited greater methylation at CpG islands of DMP genes (fig. S7G). On the basis of a curated database of metabolite-protein interactions in OmicsNet (28), we constructed a network model that connected known proteins of a metabolite signaling pathway to the DMP genes whose functions were noticeably changed in L-IGF2–treated macrophages. We singled out ATP, H2O, and H+ as highly ranked points downstream of L-IGF2–initiated signaling (Fig. 6E and fig. S7H). Also, pathways associated with thermogenesis, OXPHOS, and protein processing in endoplasmic reticulum were enriched in L-IGF2– and H-IGF2–treated macrophages by Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis (fig. S8, A and B). In comparing the average methylation levels in those genes with up- or down-regulated methylation between L-IGF2– and H-IGF2–treated macrophages, we found that L-IGF2–treated macrophages showed more activity in genes associated with OXPHOS, but less activity in genes related to TCAC and lysosomes (fig. S7F).

Given that the events of transcriptional regulation of v-ATPase–associated genes occurred mainly during macrophage maturation (29), we analyzed methylation status at v-ATPase gene promoters in IGF2-treated PMs. We found that L-IGF2 increased methylation levels at v-ATPase gene promoters, regardless of the context of CpG, CHH, or CHG (Fig. 6F and fig. S8C). At the mRNA level, transcription of these v-ATPase genes was significantly decreased in L-IGF2–treated cells (fig. S8D). Furthermore, protein expression levels for several critical components of v-ATPase, including ATP6V0A1, ATP6V0D1, ATP6V0D2, ATP6V1A, ATP6V1B2, and ATP6V1E1, were sharply down-regulated in L-IGF2–treated macrophages (Fig. 6G). Thus, DNA methylation initiated by L-IGF2 is controlled by GSK3-Dnmt3a, which directly impedes the assembly of v-ATPase in maturing macrophages and subsequently leads to proton rechanneling in mitochondria.

Specific targeting of IGF2R ameliorates colitis

Several mutant IGF2 peptides have been found to bind selectively to IGF2R (30, 31). Among them, Leu27-IGF2 binds with high affinity to IGF2R and very lower affinity to IGF1R as compared to the natural IGF2 (32). We therefore used Leu27-IGF2 to treat peritonitis mice and examined the effects on PMs. Compared to treatment with the same dose of native IGF2, we found that Leu27-IGF2 induced lower cytosolic pH in PMs. This effect persisted even at doses as high as 1000 ng per mouse (Fig. 7A), a level at which the effects of native IGF2 were reversed due to its engagement of IGF1R. Upon LPS stimulation, these Leu27-IGF2–treated macrophages exhibited lower levels of nitric oxide generation and lactic acid production, suggesting an impairment of aerobic glycolysis (Fig. 7, B and C). Consistently, Leu27-IGF2–treated macrophages exhibited higher potentials in OCR than the control cells (fig. S9A). The metabolic alteration in lactic acid and glucose consumption induced by Leu27-IGF2 cannot be observed in macrophages with IGF2R deletion (fig. S9, B and C), suggesting that Leu27-IGF2 presets macrophages a metabolic preference to OXPHOS via IGF2R. We also verified that Leu27-IGF2 mutant inhibits the acidification of lysosomes and increases the proton accumulation in cytosol (fig. S9, D and E). These metabolic changes were not seen in macrophages without IGF2R (fig. S9E). Therefore, by specifically targeting IGF2R, Leu27-IGF2 mutant can only preset a preference for OXPHOS in macrophages by proton rechanneling. In addition, regardless of the dose of Leu27-IGF2 within 50 to 2000 ng used to treat peritonitis mice, the numbers of IL-1β–expressing PMs were significantly reduced, while the expression levels of PD-L1 were increased (Fig. 7, D and E). When used to treat DSS-induced colitis, administration of Leu27-IGF2 across the entire dosage range, from 50 to 2000 ng per mouse, resulted in significant amelioration of colitis, as shown by significant improvements in body weight, survival rate, stool score, bleeding score, and colon length (Fig. 7, F to J, and fig. S9F). In summary, compared to native IGF2, Leu27-IGF2 showed a greater than 40-fold increase in the upper limit at which its anti-inflammatory function persisted, both on peritonitis-derived PMs and in effective treatment of colitis (fig. S9G). Thus, very specific targeting of IGF2R may be a potentially useful strategy to regulate the inflammatory status of macrophages for the treatment of autoimmune and inflammatory diseases.

Fig. 7 Targeting IGF2R significantly reduces severity of colitis.

(A) PMs were isolated from peritonitis mice treated with low- or high-dose IGF2 or Leu27-IGF2 (IGF2R-specific peptide mutant), and effects on cytosolic pH (measured as pHrodo Green fluorescence) were determined. (B and C) Production of nitric oxide (B) and lactic acid (C) by PMs as in (A) after subsequent LPS stimulation for 24 hours in vitro. (D) IL-1β–positive cell percentage and (E) PD-L1 expression in PMs as in (A). (F to J) Changes in body weight (F), survival rate (G), stool score (H), bleeding score (I), and colon length (J) in DSS-induced colitis mice treated with various doses of Leu27-IGF2. All data are given as mean ± SEM, n ≥ 3. All results shown are representative of three independent experiments. *P < 0.05, **P < 0.01, ***P < 0.001.


Our findings reveal a fire-new mechanism in which IGF2R signaling induces proton rechanneling to the mitochondria, leads to a preferential utilization of OXPHOS for energy generation, and thus preprograms maturing macrophages to develop an anti-inflammatory phenotype. In IGF2-preprogrammed macrophages, the increase in cytosolic protons is derived from the impaired acidification of immature lysosomes, which results from impaired transcription of proton-pumping v-ATPase subunits. Chemical inhibition of v-ATPase activity in mature macrophages can mimic IGF2R-induced proton rechanneling and confer upon macrophages an inclination for OXPHOS. Notably, specific activation of IGF2R, not IGF1R, can effectively ameliorate inflammatory disease in mice.

IGF2 can bind to either of the receptors, IGF1R or IGF2R (11). Our findings highlight that activation of IGF2R during macrophage maturation endows these cells with an anti-inflammatory phenotype, while IGF1R activation leads to a pro-inflammatory phenotype (fig. S9H). Distinct from IGF1R and insulin receptor, IGF2R is a receptor with no intrinsic catalytic activity and thus has been regarded as a decoy receptor to internalize and degrade its ligands (11). Surprisingly, in the present study, we found that nuclear translocation of IGF2R is associated with GSK3 and preprograms macrophages to develop an anti-inflammatory phenotype, revealing a new role of IGF2R signaling in macrophages. Such regulation of GSK3 activity by IGF2R has been reported in IGF2-mediated memory consolidation and enhancement in the hippocampus (14). We found that IGF2R-associated GSK3α/β signals promote Dnmt3a expression and increase DNA methylation in the promoter regions of v-ATPase–related genes. The reduced levels of v-ATPase directly restrict lysosomal acidification and subsequently provide protons for mitochondrial respiration. Because de novo assembly of v-ATPase complex and lysosome maturation occur mainly during macrophage maturation, the IGF2-activated IGF2R signaling regulates development of an anti-inflammatory phenotype in macrophages during their maturation, but not directly in already mature macrophages. The v-ATPase inhibitor BafA1 can mimic lysosomal acidification restriction–enhanced OXPHOS inclination in mature macrophages. However, as the lysosomal acidification impairment caused by BafA1 is reversible (33), the ability and effective action of BafA1 on maintaining OXPHOS metabolic inclination were limited. A deeper understanding of the molecular mechanisms of IGF2R nuclear translocation and its regulation of the chromatin-binding potential of GSK3α/β may yield fundamental insights into how macrophage memory is dictated by various factors in the microenvironment.

Although activation through IGF2R can imprint on maturing macrophages an anti-inflammatory property, activation of IGF1R by IGF1 or IGF2 (which occurs with lower affinity than for IGF2R) can counteract the signaling by IGF2R. The data on IGF1R activation by H-IGF2 and its promotion of pro-inflammatory macrophages are strongly supported by complementary results from other studies on IGF1R and trained immunity. IGF1R can be activated by mevalonate, a metabolite in the cholesterol synthesis pathway, which induces epigenetic modification in monocytes and imparts on them a pro-inflammatory phenotype, an effect also called trained immunity (6). However, attempts at preventing IGF1R signaling by ablation of IGF1R in myeloid cells can exacerbate insulin resistance induced by a high-fat diet and worsen atherosclerosis (34, 35). These controversial results may be due to the diversity of macrophages in both phenotype and functionality among various tissues and disease conditions. Nevertheless, our study strongly supports the conclusion that IGF1R activation by IGF1 and H-IGF2 plays an important role in conferring a pro-inflammatory status on macrophages.

Emerging studies have demonstrated that epigenetic reprogramming and metabolic rewiring during macrophage maturation can drive cells into a pro-inflammatory phenotype or an anti-inflammatory one, a phenomenon termed innate immune memory. Our studies demonstrate that DNA hypermethylation induced by IGF2R activation during macrophage maturation suppresses the transcription of v-ATPase subunits, thereby impairing normal lysosome maturation. Such perpetuated changes induce the rechanneling of protons to the mitochondria and promote the utilization of OXPHOS for energy production, a key feature of anti-inflammatory macrophages. Following removal of IGF2R ligands, these preprogrammed macrophages can sustain their anti-inflammatory characteristics even in the presence of other external pro-inflammatory stimuli. IGF1R activation, however, induces DNA demethylation, which can diminish the effect of IGF2R on DNA hypermethylation. In the context of tissue homeostasis, a balance between IGF1R and IGF2R activation is necessary to maintain the proper pro- or anti-inflammatory status of macrophages to allow appropriate responses to infection or inflammation. This discovery could help in the development of novel approaches for the treatment of immunodeficient states and for the regulation of uncontrolled inflammation in autoimmune diseases.

Overall, we provide a new perspective on the role of IGF2R activation in macrophage phenotype and function and underscore its importance in epigenetic reprogramming and realignment of cellular energy production toward OXPHOS, which subsequently imprint on macrophages an anti-inflammatory role. On the basis of the substantially enhanced efficacy of Leu27-IGF2 in affecting peritonitis-derived PMs and treating colitis, efforts are underway to pharmacologically potentiate IGF2R. Thus, IGF2R is a promising target for the treatment of macrophage-regulated inflammatory diseases.



C57BL/6 mice were purchased from Shanghai Laboratory Animal Center of the Chinese Academy of Science (Shanghai, China). IGF1Rfl/fl mice were purchased from The Jackson Laboratory (Bar Harbor, ME). Lyz2Cre mice were provided by H. Wang (Shanghai Jiao Tong University School of Medicine). IGF2Rfl/fl mice were established by Shanghai Center of Model Organisms Center (Shanghai, China). All mice were on a C57BL/6 background. IGF2Rfl/fl mice carry the loxP sites flanking exon 2 of the insulin growth factor receptor 2 gene. IGF1Rfl/fl and IGF2Rfl/fl mice were crossed with Lyz2Cre mice to generate mice with IGF1R- or IGF2R-specific deletion in myeloid cells only. Mice used in this study were 9- to 12-week-old females. Mice were maintained in a specific pathogen–free facility of the Shanghai Institute of Nutrition and Health of the Chinese Academy of Sciences and used in accordance with guidelines of the Institutional Animal Care and Use Committee.

Peritonitis induction and treatment with IGF2

Peritonitis was induced by intraperitoneal injection of 2 ml of 4% (w/v) thioglycollate broth media (Ftg, BD). After 24 and 48 hours, mice were injected with IGF2 (R&D Systems) at either low doses (5 or 50 ng per mouse; L-IGF2) or a high dose (1000 ng per mouse, H-IGF2), or with LPS (100 ng per mouse, Sigma-Aldrich). Peritoneal cells were harvested on day 2 or day 3 and identified phenotypically by flow cytometry as monocytes (FSClow F4/80mid CD11bmid Ly6Chigh), maturing macrophages (FSClow F4/80mid CD11bmid Ly6Clow), or mature macrophages (FSChigh F4/80high CD11bhigh Ly6Clow). These peritoneal-derived monocytes or macrophages are collectively referred to as “IGF2-treated PMs” or “LPS-treated PMs,” respectively. In some peritonitis-induced mice, IGF1R was inhibited by intraperitoneal injection of NVP-AEW541 (NVP, Selleck) with or without concurrent IGF2 administration. Similarly, GSK3α/β activity was inhibited by intraperitoneal injection of SB216763 (GSK3i, 75 μg/kg).

Colitis induction and treatment

Colitis was induced in mice by feeding with 4% (w/v) DSS (MP Biomedicals) in drinking water for the indicated number of days. After colitis induction, mice were injected intraperitoneally with IGF2 at either low doses (5 or 50 ng per mouse; L-IGF2) or a high dose (1000 ng per mouse, H-IGF2), or with Leu27-IGF2 (GroPep Bioreagents), or IGF2-treated PMs (5 × 106 cells) once a day for 3 days starting 1 day after disease induction. Total body weight and survival rate were monitored. Bleeding scores and stool scores were evaluated as previously reported (36). On day 5, mice were euthanized, and colons were removed for gross examination.

In vitro treatments

Mature macrophage experiment. PMs were obtained from mice with Ftg-induced peritonitis by peritoneal lavage. Mature PMs were obtained by petri dish adherence of peritoneal lavage fluid during culture for 72 hours. For inhibition of v-ATPase, mature PMs were treated with BafA1 (Selleck, 50 nM) for 0.5 hours. For in vitro experiments, IGF2 was added at low doses (5 or 50 ng/ml, L-IGF2) or a high dose (1000 ng/ml, H-IGF2) at times 0 and 24 hours, and cell phenotype was analyzed on day 4.

BMDM experiment. Bone marrow–nucleated cells were isolated from femur and tibia and then cultured in RPMI 1640 medium supplemented with 10% (v/v) fetal bovine serum (FBS) (LONSERA) and macrophage colony-stimulating factor (20 ng/ml) (R&D Systems) for 4 days. IGF2 was added at low doses (5 or 50 ng/ml, L-IGF2) or a high dose (1000 ng/ml, H-IGF2) at times 0 and 48 hours, and cell phenotype was analyzed on day 4. For IGF2R blockade, maturing BMDMs were treated with IGF2R antibody (1 μg/ml) (Abcam) or isotype control (R&D Systems) for 30 min bes fore IGF2 addition. For nuclear transport receptor importin-α/β inhibition, ivermectin was added concurrently with IGF2. For M1 and M2 polarization, M1 polarization was induced by addition of interferon-γ (IFN-γ) (10 ng/ml, R&D Systems) and LPS (100 ng/ml, Sigma-Aldrich) for 24 hours, and M2 polarization was induced by IL-4 (10 ng/ml, R&D Systems) and IL-13 (10 ng/ml, R&D Systems) for 48 hours.

THP1 cell experiment. THP1 cells were stimulated with phorbol 12-myristate 13-acetate (PMA) (50 ng/ml) (Sigma-Aldrich) in RPMI 1640 medium without FBS for 12 hours, and then the adherent THP1 cells were cultured in the FBS-supplemented RPMI 1640 medium without PMA. The maturing THP1 cells were treated with IGF2 added at low doses (5 or 50 ng/ml, L-IGF2) or a high dose (1000 ng/ml, H-IGF2) at times 0 and 24 hours, and cell phenotype was analyzed on day 2 or day 3.

Isolation of mononuclear cells from colonic lamina propria

Mononuclear cells in colonic laminar propria were isolated according to our previously described protocol (15). Colons were cut in small pieces and digested in Hanks’ balanced salt solution (HBSS) (containing 1 mM dithiothreitol and 5 mM EDTA) with addition of collagenase VIII (Sigma-Aldrich) and deoxyribonuclease I (Sigma-Aldrich) at 37°C for 30 min. Cells were separated by discontinuous density gradient centrifugation, as follows. Cells were washed with PBS and centrifuged, and pellets were resuspended in 40% Percoll, then layered on top of an 80% Percoll cushion, and centrifuged for 20 min at 2000g.

Fine-tuning the cytosolic pH values

Cytosolic pH was fine-tuned using the Intracellular pH Calibration Buffer Kit (Life Technologies, catalog no. P35379). This kit contains ionophores, valinomycin, and nigericin, which can be used to balance intracellular and extracellular pH (37).

Immunofluorescence analysis

Mitochondrial membrane potential. MMP was measured by incubating cells with a membrane-permeant JC1 dye (Beyotime) at 37°C for 30 min, according to MMP assay instructions.

Cytosolic pH. Cytosolic pH was measured by incubating cells with pHrodo Green AM (Invitrogen) at 37°C for 30 min. The pH value was quantified using an intracellular pH calibration buffer kit (Invitrogen).

Lysosomal pH. Lysosomal pH was measured by incubating cells with LysoSensor Yellow/Blue DND-160 (Invitrogen) or LysoSensor Green DND-189 (Invitrogen) at 37°C for 30 min. The pH value was quantified using an intracellular pH calibration buffer kit (Invitrogen).

Lysosome. Lysosome was detected by staining cells with LysoTracker Red DND-99 (Invitrogen) for 30 min at 37°C.

Cell membrane surface proteins. Cell membrane surface proteins were detected after blocking nonspecific staining by incubation with anti-CD16/32 (eBioscience). Cells were then stained with antibodies for 30 min at room temperature. The antibodies specific for surface proteins F4/80, Ly6C, CD45, and CD11b were from eBioscience.

Intracellular and nuclear protein detection. Macrophages were subjected to the Fixation/Permeabilization Solution plus Perm/Wash Buffer (BD) and Foxp3 Transcription Factor Staining Buffer (eBioscience). For IL-1β staining, macrophages were stimulated with LPS (100 ng/ml) and brefeldin A1 (10 μg/ml) (Sigma-Aldrich) for 5 hours before cell analysis. The macrophages were counterstained with Hoechst 33342 to reveal nuclei (Invitrogen).

5-mc and 5-hmc staining. PMs and THP1 cells were fixed in methyl alcohol (precooled to −20°C) for 5 min, and then chromatin was unwound by treating with 2 M HCl for 30 min. After washing with PBS, cells were stained with antibodies against 5-mc or 5-hmc (Abcam). THP1 cells were treated with PMA (50 ng/ml) for 24 hours before 5-hmc staining. Staining was analyzed by fluorescence microscopy (Zeiss) or flow cytometry (FACSCalibur, BD Biosciences) and FlowJo software.

Extracellular flux analysis

PMs were cultured in XF medium (Agilent), supplemented with 10 mM glucose (Sigma-Aldrich), 2 mM glutamine (Gibco), and 2 mM pyruvate (Gibco). OCR and ECAR were tested with an extracellular flux analyzer (Seahorse, Agilent) in the presence or absence of the following reagents: 1 μM oligomycin (Oligo; Selleck), 0.75 mM FCCP (carbonyl cyanide p-trifluoromethoxyphenylhydrazone) (Sigma-Aldrich), 100 nM rotenone (Rot; Sigma-Aldrich), and 1 μM antimycin (Ant; BioVision). For evaluation of BafA1-induced metabolic inclination by Seahorse analysis, mature PMs were treated with LPS (100 ng/ml) for 3 hours following 0.5-hour BafA1 treatment.

Quantification of NAD+/NADH

NAD+ and NADH were quantified using the NAD+/NADH Quantification Colorimetric Kit (BioVision). Cells (2 × 105) were treated with 400 μl of NAD+/NADH extraction buffer using two freeze/thaw cycles. To detect total amounts of NAD+ and NADH, 50 μl of extracted samples was transferred into 96-well plates. To detect NADH, 50 μl of extracted samples was heated to 60°C for 30 min and transferred into 96-well plates. Then, the samples were mixed with NAD cycling buffer, enzyme mix, and NADH developer, and OD450 (optical density at 450 nm) was measured. The amounts of NADH and NAD+ were calculated using NADH standard provided in the kit.

Measurement of mitochondrial complex I activity

Activity of mitochondrial complex I was evaluated using the Mitochondrial Complex I Activity Colorimetric Assay Kit (BioVision). Mitochondria were isolated from cells of interest, and a 5-μg sample was reacted with complex I assay buffer, decylubiquinone, and complex I dye, with or without rotenone addition. OD600 was measured in kinetic mode at 30-s intervals for up to 5 min at room temperature. Activity of mitochondrial complex I was calculated based on a standard curve.

Measurement of v-ATPase activity

Activity of v-ATPase was assayed using the ATPase Activity Assay Kit (BioVision). Lysosomes were isolated from cells and reacted with ATPase assay buffer at 4°C and then with ATPase substrate and ATPase assay developer at 25°C for 30 min. OD650 was measured in endpoint mode. Sample v-ATPase activity was interpolated from a standard curve.

Measurement of lactic acid, nitric oxide, and glucose

Cells were harvested and intracellular lactic acid was detected using the Lactate Colorimetric Assay Kit II (BioVision), nitric oxide was determined using Griess reagent (Sigma-Aldrich), and glucose was measured using Glucose Colorimetric Assay Kit II (BioVision), each according to the manufacturer’s instructions.

Real-time quantitative polymerase chain reaction

Total RNA was extracted with the RNAprep Pure Cell/Bacteria Kit (Tiangen Biotech) and reverse-transcribed using the PrimeScript RT Master Mix (Takara). The FastStart Universal SYBR Green Master (Roche) was added to each polymerase chain reaction (PCR) along with complementary DNA (cDNA) and specific primers to a total volume of 10 μl. Real-time quantitative PCR (qPCR) was performed on ABI Prism 7900HT (Applied Biosystems).

Immunoblotting analysis

Total cell lysates were prepared in radioimmunoprecipitation assay buffer (Millipore), with protease inhibitors (Roche) and phosphatase inhibitors (Sigma-Aldrich). Subcellular protein extraction was performed using the Subcellular Protein Fractionation Kit (Pierce). Protein concentration was measured using the BCA Protein Assay Kit (Pierce). Samples were boiled in loading buffer and then separated by SDS–polyacrylamide gel electrophoresis and transferred onto polyvinylidene difluoride membranes (Millipore). Subsequently, the membranes were incubated with primary antibodies at 4°C for overnight, followed by incubation with the appropriate secondary horseradish peroxidase–conjugated antibodies [Cell Signaling Technology (CST)] for 1 hour at room temperature, and developed using Clarity Western ECL Substrate (Bio-Rad). Antibodies specific for IGF2R, IGF1R, phospho-IGF1R (Tyr1131), phospho-IGF1R (Tyr1135/1136), GSK3α, phospho-GSK3α (Ser21), GSK3β, phospho-GSK3β (Ser9), mTOR, phospho-mTOR (Ser2448), AKT, phospho-AKT (Ser473), P70S6K, phospho-P70S6K (Thr389), 4EBP1, and phospho-4EBP1 (Thr37/46) were from CST.

Feature predictions of mouse and human IGF2R

Prediction of the nuclear localization signal of IGF2R was based on frequent pattern mining and linear motif scoring using the website (38). Subcellular location predication of mouse and human IGF2R was based on the database annotations and automatic text mining of the biomedical literature at the website (39). Protein, DNA, and RNA binding sites in mouse and human IGF2R were predicated by the website (40).

Whole-genome bisulfite sequencing

Total DNA was isolated using the QIAamp DNA Mini Kit (Qiagen). A bisulfite library was prepared according to the Illumina protocol, essentially as described (41). The 667 million read pairs were cleared by the NGS QC Toolkit (42), excluding reads with a mapping quality lower than 30 (probability to map at a correct position = 0.999) and ignoring bases with base quality lower than 20 (probability that the sequencing process produced the correct base = 0.99). Last, above 90-gigabyte data of each sample were executed using the bismark tool and a modified Mus musculus reference (GRCm38.p6) to output discriminates between cytosine in the CpG, CHG, and CHH context (43). Differentially methylated regions were detected by a methylKit (methylation level above 10% and q ≤ 0.05; Fisher’s exact test) (44). On the basis of the above analyses, 2 kb upstream of the transcription start site was used as the gene promoter region and DMPs were screened out for further KEGG analysis (methylation level above 10% and q ≤ 0.05; Fisher’s exact test). DNA methylation level (Methy level) was calculated by “Number of methylated cytosine site”/“Number of covered cytosine site” and then represented as heat map by Integrative Genomics Viewer (IGV) software. Average methylation level (Aver-methy level) was calculated by “total methylation level”/“gene number.” Change in DNA methylation level was calculated as “‘Methy level of IGF2-preprogrammed PMs’ minus ‘methy level of control PMs’”/“Methy level of control PMs.” DMP gene–associated metabolite-protein interactions were analyzed by OmicsNet ( (28), and the metabolic node filter degree was 10. DMP gene–associated cellular biological functions and metabolic pathways were enriched by KEGG analysis.

Histological analysis

Colons were harvested from mice, fixed in 4% paraformaldehyde (Sigma-Aldrich) in PBS, and embedded in paraffin (Sigma-Aldrich). Hematoxylin and eosin staining was performed according to standard protocols.

Statistical analysis

Data are provided as mean or mean ± SEM. Populations were compared using two-tailed unpaired Student’s t test, with a 95% confidence interval under the untested assumption of normality. For all tests, P values are indicated as *P < 0.05, **P < 0.01, ***P < 0.001, and ns means not significant. P < 0.05 was considered statistically significant.


Supplementary material for this article is available at

This is an open-access article distributed under the terms of the Creative Commons Attribution-NonCommercial license, which permits use, distribution, and reproduction in any medium, so long as the resultant use is not for commercial advantage and provided the original work is properly cited.


Acknowledgments: Funding: This work was supported by grants from the National Key R&D Program of China (2018YFA0107500), Scientific Innovation Project of the Chinese Academy of Sciences (XDA16020403), National Natural Science Foundation of China (81861138015, 81530043, 31961133024, 31771641, 81930085, and 81571612), Chinese Postdoctoral Science Foundation (2020 M671573), Key Laboratory of Tumor and Microenvironment of Chinese Academy of Sciences (no. 202004), Ministry of Health Italy-China cooperation grant and AIRC IG#20473 to G.M., Youth Innovation Promotion Association research fund from the Chinese Academy of Sciences, a start-up fund from Soochow University, and Department of Science and Technology of Jiangsu Province research fund (BE2016671). Author contributions: X.W., Ying Wang, and Y.S. designed the project. X.W. designed and performed most of the experiments. X.W., L.L., and B.L. provided the interpretation of data. L.L., B.L., Yu Wang, L.D., X.C., Q.L., K.L., M.H., and Y.X. provided technical assistance and material support. G.M. and C.S. made comments and suggestions. X.W., A.I.R., Ying Wang, and Y.S. wrote the manuscript with input from all authors. Competing interests: The authors declare that they have no competing interests. Data and materials availability: All data needed to evaluate the conclusions in the paper are present in the paper and/or the Supplementary Materials. Additional data related to this paper may be requested from the authors. The accession number of whole-genome bisulfite sequencing data of this study is NCBI GEO: GSE151404.

Stay Connected to Science Advances

Navigate This Article