Insulin-Like Growth Factor-I Regulates the Liver Microenvironment in Obese Mice and Promotes Liver Metastasis

Among the mechanisms implicated in the tumor-promoting effects of obesity, signaling by insulin-like growth factor-I (IGF-I) and insulin has received considerable attention. However, the emerging realization that obesity is associated with chronic inflammation has prompted other consideration of how the IGF-I axis may participate in cancer progression. In the present study, we used two mouse models of chronic (LID) and inducible (iLID) igf-1 gene deficiency in the liver to investigate the role of IGF-I in regulating the host microenvironment and colorectal carcinoma growth and metastasis in obese mice. Obese mice had a heightened inflammatory response in the liver, which was abolished in mice with chronic IGF-I deficiency (LID). In control animals changes to the hepatic microenvironment associated with obesity sustained the presence of tumor cells in the liver and increased the incidence of hepatic metastases after intrasplenic/portal inoculation of colon carcinoma cells. These changes did not occur in LID mice with chlonic IGF-1 deficiency. In contrast, these changes occured in iLID mice with acute IGF-1 deficiency, in the same manner as the control animals, revealing a fundamental difference in the nature of the requirement for IGF-1 on tumor growth and metastasis. In the setting of obesity, our findings imply that IGF-1 is critical to activate and sustain an inflammatory response in the liver that is needed for hepatic metastasis, not only through direct, paracrine effect on tumor cell growth, but also through indirect effects involving the tumor microenvironment. Cancer Res; 70(1); 57–67

The tissue microenvironment plays an important role in tumor development and progression both at the primary site and at sites of metastases (1). The microenvironment consists of the tumor-associated stroma that includes resident fibroblasts, adipocytes, blood and lymph vessel–lining endothelial cells and their respective extracellular matrix proteins, as well as resident and infiltrating host inflammatory and immune cells. The malignant cells can communicate with the stroma through the release of, and response to, soluble mediators, such as growth factors and cytokines, through cell-cell and cell-matrix adhesion and through the release of proteolytic enzymes that remodel tissue barriers around the tumor cells and contribute to neovascularization and tumor migration and invasion (2, 3).

The host inflammatory response, macrophages in particular, plays an important role in tumor progression (4, 5). Chronic inflammation has been linked to tumor initiation and progression in different cancer types, including gastric (6), hepatocellular (7–10), and colorectal (11) carcinoma, and treatment with anti-inflammatory drugs such as cimetidine or aspirin was shown to reduce cancer risk (12). Moreover, a high density of tumor-associated macrophages was found to correlate with poor prognosis in more than 80% of studies published (13, 14). Animal studies have also shown that the host inflammatory response can be acutely activated during the early stages of metastasis and thereby promote tumor cell arrest and extravasation in the invaded organ (15, 16).

Obesity is a common metabolic disorder, frequently associated with chronic low-grade inflammation (17), increased oxidative stress (18), and lipotoxicity, all of which can alter the tissue microenvironment. Obesity often leads to lipid accumulation in the liver (hepatosteatosis) and chronic liver inflammation. Although it has been shown that obesity (and obesity-associated inflammation) can alter the host antitumor immune response and contribute to tumor cell escape from immune surveillance (19), it is unknown how this state of chronic inflammation affects the ability of metastatic tumor cells to induce the more acute and transient inflammatory reactions that are associated with the early stages of organ colonization and known to promote metastasis (15, 20, 21).

The liver is the main site of insulin-like growth factor-I (IGF-I) production. IGF-I has been identified as a critical regulator of tumor growth and metastasis, but its role in inflammation-associated tumor development or the obesity-associated increase in inflammation, and tumor growth, are presently unknown. Likewise, the role of IGF-I in regulating changes to the tumor-associated stroma that affect tumor cell growth in the liver in normal or obese individuals have not been studied. IGF-I is a known activator of the Janus-activated kinase/signal transducer and activator of transcription (STAT) pathway (22) and has been implicated in NF-κB–mediated transcriptional regulation of inflammatory cytokines and vascular endothelial cell adhesion receptors such as intercellular adhesion molecule-1 (ICAM-1; ref. 23). It may therefore play a role both in the maintenance of a chronic inflammatory state and in the induction of the acute inflammatory response that is triggered by tumor cells during the early stages of liver metastasis (24, 25).

The aim of the present study was 2-fold: (a) to assess host inflammation parameters and liver metastasis in mice with diet-induced obesity and (b) to determine the role of IGF-I in the development of liver metastases in the obese state. We used diet-induced obesity and an experimental liver metastasis assay to investigate the effect of a high-calorie diet on the hepatic microenvironment and the development of liver metastasis. To elucidate the role of serum IGF-I in tumor growth under conditions of obesity, we used mice with a liver-specific IGF-I deficiency (the LID model). These mice have no detectable IGF-I mRNA in the liver and a 75% reduction in serum IGF-I levels (26). Additionally, we generated a second mouse model of inducible liver IGF-I deficiency (iLID). The iLID mice express the Cre recombinase under the control of the anti-trypsin 1α promoter, specifically in liver, and a single tamoxifen injection results in gene recombination to produce a liver-specific IGF-I deficiency (iLID). The LID and iLID models were used here to dissect the effects of acute versus chronic IGF-I depletion on tumor cell growth and metastasis.

Male mice were genotyped as described previously (26). Both LID and iLID mouse models were bred on a C57B6/J background. All chows were purchased from Research Diets⁶

(D12450B: 10 kcal % fat, control diet; D12492: 60 kcal % fat, high-calorie diet). The diets did not contain phytoestrogens.⁷ Animal care and maintenance were provided through the Mount Sinai School of Medicine AAALAC-Accredited Animal Facility. All procedures were approved by the Animal Care and Use Committee of the Mount Sinai School of Medicine.

Body fat mass was measured in nonanesthetized mice using a Bruker minispec nuclear magnetic resonance analyzer mq 10 (Bruker Optics).

The colon adenocarcinoma MC-38 cells were maintained in DMEM (BioSource) supplemented with 10% FCS (Invitrogen) and glutamine (BioSource; ref. 27).

Experimental hepatic metastases were generated using the intrasplenic/portal route of inoculation as described previously (15). Mice were anesthetized using isoflurane. Spleens were exposed through a small abdominal incision; 0.5 × 10⁵ tumor cells in 100 μL of saline or saline alone were inoculated and the mice were splenectomized 1 min later. The livers were removed at the indicated time intervals, snap frozen in liquid N₂, and stored at −80°C until further analysis.

Total RNA was extracted from snap-frozen livers using a combined Trizol/Qiagen RNeasy Midi kit. One microgram of total RNA was reverse transcribed and real-time PCR was performed using TaqMan reagents and specific primers. Real-time PCR was performed with the QuantiTect SYBR green PCR kit (Qiagen) according to the manufacturer's instructions in ABI PRISM 7900HT sequence detection systems (Applied Biosystems). Each transcript in each sample was assayed three times and the fold-change ratios between experimental and control samples for each gene used in the analysis were calculated using β-actin or 18S levels as a reference.

Serum IGF-I and leptin were determined as described in detail previously (27).

Cells were fluorescently labeled using 5 mol/L 5-chloromethylfluorescein diacetate (CMFDA; Molecular Probes), according to the manufacturer's recommendation (28). At 1, 6, 18, and 24 h after the intrasplenic injection of 3 × 10⁵ cells, mice were euthanized by CO₂ inhalation. Livers were dissected and then imaged by inverted fluorescent videomicroscopy (Leica DM IRB) at ×100 magnification. Ten random frames were captured and the OpenLab software was used to define and count fluorescent (green) CMFDA tumor events larger than 10 pixels. The means of the total number of events per liver were calculated based on five mice per time point.

Livers were isolated and fixed overnight in 4% paraformaldehyde in PBS. The tissues were then embedded in paraffin and 5-μm sections were stained by H&E. Immunostaining was performed with an anti-F4/80 antibody (Abcam) at a dilution of 1:150 and detected with the ImmunoCruz Staining System (Santa Cruz Biotechnology).

Values are presented as mean ± SEM. Statistical significance was determined by one- or two-way ANOVA or by a t test using the SigmaStat software (SPSS, Inc.). Means ± SEM are indicated. P < 0.05 was considered statistically significant.

In our previous study, we explored the effect of diet-induced obesity on local subcutaneous growth of colon cancer using the murine colorectal carcinoma MC-38 cells (27). Our results revealed that diet-induced obesity increased the tumor incidence and growth rate in both male and female syngeneic mice, suggesting that obesity contributed to enhanced tumor development. To elucidate the role of serum IGF-I in the obesity-associated enhancement of colon carcinoma growth, we therefore used liver IGF-I–deficient (LID) male mice (26). These mice exhibit 75% to 80% reductions in serum IGF-I levels without changes in extrahepatic igf-1 gene expression, and their overall growth is not significantly different from that of control mice (26). When male LID and control mice were maintained on high-calorie diet for 10 to 14 weeks, they exhibited increased body adiposity and serum leptin levels, compared with age- and sex-matched mice that were fed regular chow (lean mice; Fig. 1A and B) while still maintaining low serum IGF-I levels (i.e., 25% of control levels; Fig. 1C). As expected, local tumor growth in obese mice injected s.c. with MC-38 cells was significantly increased relative to lean mice. However, this increase was abolished in obese LID mice where mean tumor weight at 21 days postinjection was essentially identical to that in lean LID mice (Fig. 1D). This identified IGF-I as a causative factor in enhanced tumorigenicity in these mice. It is also important to note that the decreased tumor growth in LID mice was seen despite increased circulating levels of insulin (secondary to elevations in serum GH; ref. 29).

LID mice are IGF-I deficient throughout their entire life span. Our results indicated that the chronic reduction in IGF-I levels attenuated tumor growth. To distinguish the potential effects that this long-term IGF-I deprivation can have on the host and thereby on tumor growth from the direct paracrine effect that IGF-I has on tumor cell growth in the liver, we developed an inducible, liver IGF-I deficiency model (iLID) whereby the Cre recombinase is expressed under the trypsin-1α promoter and can be induced by a single tamoxifen injection (Fig. 2A). This mouse model allowed us to deplete IGF-I from the liver at different time intervals relative to tumor inoculation, thereby providing the means for a more precise control of IGF-I bioavailability, while also avoiding the long-term effects that serum IGF-I deficiency can have on the host. RNase protection assay (Fig. 2B) was performed on livers dissected from control and iLID mice injected or not with tamoxifen. Dose-response (Fig. 2C) and kinetic (Fig. 2D) analyses revealed that a single i.p. injection of tamoxifen (0.3–0.4 mg) significantly decreased liver IGF-I gene expression 24 hours postinjection (Fig. 2D) and was sufficient to deplete serum IGF-I levels by 60% to 70% for at least 4 weeks postinjection.

To assess the effects that an acute IGF-I depletion had on the metabolic parameters characteristic of obesity, mice were fed a high-calorie (or regular) diet for 8 to 10 weeks, injected with 0.4 mg tamoxifen, and the serum samples were collected and analyzed 2 weeks later. Similarly to LID mice, we found that serum IGF-I levels in iLID lean or obese animals were decreased relative to tamoxifen-treated controls (Supplementary Fig. S1). Serum leptin and insulin levels increased in obese compared with lean iLID mice. Blood glucose and serum adiponectin levels were comparable with those in lean mice. These data suggested that both LID and iLID mice had increased body adiposity in response to a high-calorie diet and developed a state of insulin resistance.

We next asked whether this acute IGF-I depletion could affect tumor cell growth in the subcutaneous site. IGF-I depletion was induced in iLID mice by a tamoxifen injection 4 days before or after s.c. inoculation of MC-38 cells. We found that an acute IGF-I depletion before or after tumor inoculation failed to significantly reduce the growth of local subcutaneous tumors in lean mice and did not suppress the increase in tumor growth seen in obese mice (Fig. 3). This indicated that the effects of chronic and acute IGF-I depletion on tumor growth were fundamentally different and suggested that the tumor-promoting effects of IGF-I in vivo were not restricted to a direct paracrine, growth-stimulating effect on the tumor cells and entailed other IGF-I–mediated mechanisms, likely involving the host.

The liver is the primary site for colorectal carcinoma metastasis, and the IGF-I axis was identified as a positive regulator of liver metastasis. It was therefore of interest to analyze the effect of obesity on colon carcinoma liver metastasis. Mice were inoculated with 5 × 10⁴ MC-38 cells through the intrasplenic/portal route to generate experimental hepatic metastases. Livers were dissected 2 weeks later, the incidence of hepatic metastases was recorded (Fig. 4A), and the increase in liver weight was measured to determine the metastatic burden (Fig. 4B). We found that the incidence of metastases and the liver weights dramatically increased in obese compared with lean mice (Supplementary Figs. S2 and S3; Fig. 4). The incidence of liver metastases was markedly reduced and the difference between lean and obese mice was abolished in the LID mice, where only 1 of 10 lean and 2 of 10 obese mice had visible metastases 2 weeks after tumor cell injection.

To assess the effects of acute serum IGF-I depletion on hepatic metastasis, tamoxifen was injected into lean and obese control and iLID animals 4 days before or after intrasplenic/portal tumor cell inoculation, as depicted in Fig. 4A. As expected, the incidence of hepatic metastases increased in obese compared with lean mice (Fig. 4A). This increase was not diminished by a reduction in serum IGF-I levels that was induced 4 days before tumor inoculation (Fig. 4A and B). Tamoxifen-induced reduction in serum IGF-I 4 days after tumor cell inoculation reduced the incidence of metastases in obese iLID mice to the level seen in lean iLID mice. In contrast to the LID mice, the incidence of metastases in the obese iLID mice remained high and this was reflected in increased relative liver weights (Fig. 4B). Histologic examination of the livers 2 weeks after tumor cell inoculation revealed no morphologic differences between tumors that developed in LID, iLID, or control obese mice (Supplementary Figs. S2 and S3).

To begin to understand how obesity affects the hepatic microenvironment and what role, if any, IGF-I played in regulating these effects, we first measured tumor cell arrest following entry into the hepatic microcirculation, using in vitro vital microscopy. MC-38 cells were labeled with a fluorescent dye (CMFDA), injected through the intrasplenic/portal route, and their presence in the liver was monitored at different time intervals thereafter. When analyzed 1 hour after inoculation, no significant difference could be observed between the number of tumor cells per field detectable in lean and obese mice (Fig. 5), suggesting that similar number of cells reached the livers in both groups. However, although the number of tumor cells detectable in the livers declined steadily in lean mice reaching 10% of the initial number at 24 hours postinjection, the number of tumor cells present in livers of obese mice decreased more slowly with up to 70% of the initial number still detectable 24 hours postinjection (Fig. 5). These results suggested that the liver microenvironment in obese mice was more permissive for the sustained presence of metastatic cells during the first 24 hours after inoculation.

Tumor cell arrest in the liver microcirculation was previously linked to an upregulation of cytokine-induced expression of vascular endothelial cell adhesion molecules such as E-selectin and VCAM-1 (15, 21). A quantitative, real-time PCR analysis revealed that all indicators of an activated inflammatory response, such as the expression of the cytokines tumor necrosis factor α (TNF-α), interleukin (IL)-1β and IL-18, and the levels of MCP-1 (Fig. 6A), were elevated in obese relative to lean mice and that a chronic IGF-I depletion (as present in LID mice) caused a reduction in these indicators in both obese and lean animals. Analysis of the macrophage-specific marker F4/80 revealed a decrease in macrophage numbers and activity in the livers of these mice, and, in parallel, markers of cytokine signaling such as STAT-3 were also reduced (data not shown). In sharp contrast, an acute igf-1 gene deletion, as seen in iLID mice, did not significantly alter F4/80 or inflammatory cytokine mRNA expression in lean mice and actually increased the expression of some cytokines (e.g., TNF-α and IL-1β) in obese mice (Fig. 6B). These data were also confirmed by immunostaining with the macrophage-specific antibody (to F4/80), which showed an increase in macrophage numbers in livers of control and iLID obese mice and a decrease in their numbers in LID mice (Supplementary Fig. S4). Additionally, we observed microvesicular and macrovesicular steatosis (abnormal retention of lipids within hepatocytes) in livers of control and acutely depleted IGF-I (iLID) mice, whereas LID mice developed only a mild steatosis that was mainly microvesicular (Supplementary Fig. S4).

Vascular endothelial cell adhesion molecules such as the selectins, ICAM-1, and VCAM-1 are cytokine inducible, and their expression is upregulated in response to an activated inflammatory response (30). We found that P-selectin, VCAM-1, and ICAM-1 mRNA levels decreased in both lean and obese LID compared with control mice. In contrast, the expression levels of these CAMs in lean or obese iLID mice did not differ significantly from tamoxifen-treated controls (Fig. 6C). Taken together, these data indicate that chronic but not short-term (acute) liver IGF-I depletion alters the hepatic microenvironment.

The roles of IGF-I– and insulin-induced signaling in the tumor-promoting effects of obesity have received considerable attention. However, the emerging realization that obesity is associated with chronic inflammation has provided new insight into the possible role that obesity and the IGF-I axis in particular may play in cancer progression. Adhesion and growth of tumor cells often occur at sites of inflammation (31), tissue injury (32), and repair (33). Therefore, sites of inflammation are potentially “fertile ground” for the development of metastases. The present study was undertaken to determine whether enhanced tumor growth and progression in an obese state (where systemic inflammation occurs) is also regulated by IGF-I.

Our study shows the following: (a) the liver microenvironment in obese mice is more permissive for metastatic cells than in lean mice; (b) IGF-I plays a critical role in the induction of a chronic inflammatory state in the liver of obese animals and can thereby indirectly affect tumor growth in this site; and (c) there is a fundamental difference between the effects of chronic and acute liver IGF-I depletion on tumor growth and metastasis, implying that IGF-I promotes liver metastasis not only through a direct paracrine effect on tumor cell survival and proliferation but also through indirect effects, likely involving the host microenvironment and proinflammatory responses.

Hepatic tissue is a primary target for metastases by most disseminating malignant cancers (34, 35). Kupffer and perisinusoidal hepatic stellate cells release inflammatory cytokines, eicosanoids, plasminogen activator inhibitor, and reactive oxygen or nitrogen intermediates (36) that can lead to increased cell proliferation, expression of adhesion molecules, extracellular matrix production, cytotoxicity, changes in coagulation, and altered sinusoidal flow. Furthermore, inflammatory cytokines such as TNF-α and IL-1 (constitutively upregulated in a state of obesity) can stimulate the endothelium (37, 38) and lead to an increased expression of cell surface adhesion molecules. In line with previous studies, we show here that in mice with diet-induced obesity, the basal expression levels of inflammatory cytokines and cell adhesion molecules were upregulated. This increased expression was dependent on hepatic IGF-I, as it was abolished in LID mice. This effect was only seen in mice with a chronic reduction in hepatic IGF-I because acute depletion of liver IGF-I, as induced in obese iLID mice, was not associated with reduced expression of inflammatory cytokines and vascular adhesion molecules and in fact caused an increase in the expression of some cytokines, such as TNF-α and IL-1β, by a mechanism yet to be identified. Taken together, our data identify a new indirect role for IGF-I in promoting tumor growth, at least in the liver, namely, the regulation of the host innate immunity and tumor-promoting inflammation. Studies have shown that GH and IGF-I can each stimulate the expansion of primitive multilineage hematopoietic progenitor cells and increase several hematopoietic subpopulations, including lymphoid progenitor cells (39). In LID animals, we observed significant and consistent reductions (between 4 and 52 weeks of age) in the LeuCam surface protein, CD11b (macrophage marker) in the bone marrow (40), and a markedly reduced expression of F4/80 in the liver. F4/80 is a Kupffer cell marker that was accompanied by a reduction in mRNA levels for several macrophage-derived cytokines, including TNF-α, IL-1β, and IL-18. Together, these results suggest that cells committed to the myeloid/monocytic differentiation pathway may be particularly susceptible to a sustained (but not a short term) reduction in hepatic IGF-I levels and that the altered cytokine profile seen in the liver of LID mice was due, at least in part, to a depletion of hepatic Kupffer cells and/or their impaired function.

In contrast to the marked reductions in cytokine expression in LID mice, we found that an acute depletion of IGF-I did not change basal cytokine production in lean mice, whereas in obese animals, the levels of these cytokines increased relative to wild-type mice. This could most likely reflect different biological effects of sustained (chronic) or acute IGF-I deletion on the number and function of hepatic Kupffer cells. Although the number of these cells in livers of LID mice was markedly reduced, the acute depletion of hepatic IGF-I did not alter the macrophage content in the liver and, as a consequence, basal cytokine production remained relatively normal. In obese mice, cytokine expression was actually augmented in response to acute IGF-I depletion, suggesting that the combined effect of “obesity” and an acute IGF-I depletion may cause a rapid activation of the hepatic macrophages resulting in increased cytokine production. Consistent with these findings, a phenotypic switch from the low cytokine–producing M2 to the “classically activated” M1 macrophages was observed in adipose tissue–associated macrophages of obese mice (41), and Kupffer cell dysfunction was reported in obese animals such as the leptin-deficient ob/ob mice (42). It is possible that in the present study, phenotypic/functional changes in the Kupffer cells in obese mice heightened their sensitivity to an acute drop in hepatic IGF-I levels and, as a consequence, cytokine expression was further activated.

In addition to the differences in the effects of chronic and acute IGF-I depletion on the macrophage population of the liver, differences in the metabolic effects caused by IGF-I depletion may have also contributed to the differential effects on hepatic cytokine production in these mice. For example, chronic depletion of IGF-I (as seen in the LID model) can lead to compensatory increases in serum GH levels, which persist even to an older age (up to 24 weeks).⁸

GH is one of the regulators of body adiposity and was shown to induce lipolysis (43). It is conceivable that the elevation in GH levels in LID mice improves their lipolytic response to the high-calorie diet in the liver or in the white adipose tissue, thereby leading to alterations in the liver microenvironment. On the other hand, acute depletion of IGF-I (as seen in the iLID model), although leading to a slight increase in GH secretion (∼4–7 days after induction), is unlikely to cause rapid and significant alterations in the liver microenvironment. Our finding that the incidence of hepatic metastases in obese iLID mice was not reduced and even somewhat increased relative to controls is consistent with this lack of effect on hepatic cytokine production and with the increase observed in TNF-α and IL-1β production. This finding also lends further support to the critical role that inflammatory cytokines play in the regulation of hepatic metastasis.

Taken together, our results identify hepatic IGF-I as a critical regulator of basal inflammatory responses in the liver of normal and obese mice. Furthermore, they implicate IGF-I in the obesity-induced increase in tumor cell growth and hepatic metastasis and show that its effect is mediated through a complex mechanism involving both direct effects on the tumor cells and the conditioning of the hepatic microenvironment.

No potential conflicts of interest were disclosed.

We thank Drs. Derek LeRoith and Scott Friedman (Mount Sinai School of Medicine) for their help and advice on this project.

Grant Support: Canadian Institute for Health Research grant MOP-81201.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.