It has been shown that in the mouse colon 26 tumor model, tumors grown in the subcutis (subcutis colon 26) caused early onset of cachectic syndromes, whereas those in the liver (liver colon 26) did not. Both interleukin (IL)-6 and parathyroid hormone-related protein (PTHrP) were involved in the development of cachectic syndromes in this tumor model. However, whether expression of PTHrP and IL-6 is differently regulated in the tumor microenvironment is unclear. In the present study, culturing the colon 26 cells under different conditions in vitro revealed that IL-6 production was increased by monolayer culture under a low-glucose condition but not by spheroid culture. In contrast, PTHrP production was increased by spheroid culture but not by monolayer culture, even under a low-glucose condition. Gene expression profiling revealed that the expression of cyclooxygenase (COX)-2 was up-regulated in both subcutis colon 26 and spheroid cultures, and that COX-2 inhibitor NS-398 suppressed PTHrP production in spheroid cultures. Furthermore, administration of NS-398 decreased the PTHrP level without affecting the tumor growth in mice bearing subcutis colon 26. These results demonstrate that production of PTHrP and IL-6 largely depends on the microenvironments in which tumors are developed or metastasized and that up-regulation of COX-2 in a necrobiotic environment leads to PTHrP production, thereby causing cachectic syndromes. (Cancer Sci 2007; 98: 1563–1569)
Cancer cachexia is a complex disease, and the development of cachectic syndromes does not correlate with tumor mass or metastases.(1) Several factors, such as tumor necrosis factor (TNF)-α, interleukin (IL)-1β, IL-6, leukemia inhibitory factor (LIF), interferon (IFN)-γ, and parathyroid hormone-related protein (PTHrP) are thought to cause cachectic symptoms in animal models.(2) Among these, PTHrP plays a key role in the development of cachexia. Nude rats and nude mice bearing the human lung cancer xenograft that expresses PTHrP and several inflammatory cytokines, including IL-6, develop cachectic syndromes such as body-weight loss, hypercalcemia, hypoglycemia, and hypolipidemia.(3) Administration of an anti-PTHrP antibody nearly completely cured cachectic symptoms as revealed by the restoration of body-weight gain and blood calcium, but neither affected the serum levels of IL-6 nor abolished acute-phase reactions.(3) Human cachectic cancer patients with elevated blood levels of PTHrP also showed increasing levels of multiple inflammatory cytokines, including IL-1β, IL-6, IL-8, IL-11, TNF-α, and IFN-γ.(4) Furthermore, gastroesophageal cancer patients with elevated blood PTHrP levels have been shown to have shorter survival than those with low blood PTHrP levels.(5) However, whether the expression of PTHrP and other inflammatory cytokines such as IL-6 is differently regulated in the microenvironments of tumors remains unexplored.
It was demonstrated that mice with subcutaneously transplanted colon 26 (subcutis colon 26) developed cachectic symptoms, whereas those with intrahepatically transplanted colon 26 (liver colon 26) did not.(6) In addition, blood levels of PTHrP and IL-6 were elevated in mice with subcutis colon 26(7) and both PTHrP and IL-6 have been shown to play important roles in the development of cachexia in the subcutis colon 26 model.(7,8)
In the present study, the authors show that in mice bearing subcutis colon 26, PTHrP and IL-6 are differently expressed under different conditions and play different roles in the cachectic syndrome. In addition, the production of PTHrP is induced by a necrobiotic environment, and the induction of PTHrP is at least in part mediated by cyclooxygenase (COX)-2. The results provide new insights into the mechanism underlying the induction of PTHrP, which may explain the tumor site-dependent development of cachectic syndromes.
Materials and Methods
Animal experiments. A subclone of colon 26, colon 26 c20,(8) was used in the present study. The cells were cultured in RPMI 1640 (Sigma-Aldrich, St Louis, MO, USA) containing 10% fetal bovine serum (FBS). Male 5-week-old CDF1 mice obtained from SLC (Hamamatsu, Japan) were used as the host animals. For intraliver inoculation, 104 cells of colon 26 were injected into the superior mesenteric vein. For subcutaneous inoculation, 5 × 105–1 × 106 cells of colon 26 cells were injected into the subcutis of the right flank. Body weights were measured twice per week. Weights of tumors in the liver were estimated by subtracting the average liver weight of four non-tumor-bearing (normal) mice from the liver weights of each tumor-bearing mouse. Tumors excised on day 17 were used for reverse transcription–polymerase chain reaction (RT-PCR) and DNA microarray analysis. A humanized anti-PTHrP antibody raised against PTHrP(1-34),(9) and an anti-IL-6 antibody(8) were administered intravenously at doses of 30 µg/mouse and 1 mg/mouse, respectively, on days 10 and 14. Human IgG was purchased from MP Biomedicals (Eschwege, Germany) and used as a control antibody. N-(2-cyclohexyloxy-4-nitrophenyl)-methane sulfonamide (NS-398); Wako, Osaka, Japan) was suspended in 5% gum arabic solution and administered orally twice daily from day 7 to day 13. Day 0 represents the day tumors were inoculated into the mice.
Determination of levels of calcium, glucose, IL-6, PTHrP and PDGF-BB. Blood levels of ionized calcium (iCa) were determined using the electrode method with a 634 Ca2+/pH autoanalyzer (Bayer Medical, Tokyo, Japan). Levels of glucose, calcium, IL-6, PTHrP, and platelet-derived growth factor (PDGF)-BB in plasma and conditioned media were determined using a glucose CII test (Wako), calcium C-test (Wako), a mouse IL-6 ELISA kit (Pierce, Rockford, IL, USA), PTHrP IRMA (Mitsubishi Chemical, Tokyo, Japan), and a mouse/rat PDGF-BB ELISA kit (R&D systems, Minneapolis, MN, USA), respectively. The DNA contents of the cells were determined with a FluoReporter blue fluorometric dsDNA quantitation kit (Molecular Probes, Eugene, OR, USA), and the amounts of IL-6 and PTHrP in conditioned media were normalized to the cellular DNA contents.
RNA extraction. Total RNA was extracted from tissues using Sepazol RNA I Super (Nacalai Tesque, Tokyo, Japan) and purified with an RNeasy Mini kit (Qiagen, Hilden, Germany) according to the manufacturer's instructions. cDNA was synthesized with 20 µg of RNA as the template using the SuperScript system (Invitrogen, Carlsbad, CA, USA).
RT-PCR. RT-PCR was carried out using an RNA LA PCR kit (Takara, Otsu, Japan) according to the manufacturer's instruction. Primers used for the mouse COX-2 were 5′-ACTCACTCAGTTTGTTGAGTCATTC-3′ and 5′-TTTGATTAGTACTGTAGGGTTAATG-3′, and those for the mouse GAPDH were 5′-TTCACCACCATGGAGAAGGC-3′ and 5′-GGCATGGACTGTGGTCATGA-3′. PCR was carried out with 27 cycles of incubation at 95°C for 30 s (denaturation), at 60°C for 30 s (annealing), and at 72°C for 30 s (extension). The amplified DNA was separated on a 2% agarose gel and visualized by staining with ethidium bromide.
For real-time RT-PCR, 1 µL of cDNA solution (equivalent to the cDNA from 0.2 µg of initial RNA) was used as the template. Real-time PCR was performed with a LightCycler (Roche Diagnostics, Penzberg, Germany) and Light Cycler-FastStart DNA Master SYBR-Green I (Roche Diagnostics) with 40 cycles of consecutive reactions of denature (95°C for 15 s), annealing (58°C for 2 s), and extension (72°C for 15 s).
Sequences of the primers used for PCR were 5′-AAGCAGTGCCCGAACCCCCATT-3′ and 5′-CCCGTTGATTTCCACGTGGAGT-3′ for mouse transforming growth factor (TGF)-β1 and 5′-CTCCTGCACCACCAACTG-3′ and 5′-GAGGGGCCATCCACAG-3′ for mouse GAPDH. Levels of mRNA for TGF-β1 were normalized with those of GAPDH.
Cell cultures. Unless otherwise specified, cells of colon 26 were cultured in RPMI 1640 supplemented with 10% FBS at 37°C under 5% CO2. When cells were cultured under a low glucose condition, they were cultured in glucose-free RPMI 1640 medium (Nikken Bio Medical Laboratory, Tokyo, Japan) supplemented with 10% FBS. For the hypoxic condition, cells were cultured under 1% O2 and 5% CO2. To make spheroids, 5 × 104 cells per well of colon 26 cells were plated onto a round bottom 96-well plate that had been coated with poly 2-hydroxyethyl methacrylate and cultured in the presence or absence of NS-398 for 2 days. To examine the effects of PDGF on the production of PTHrP and IL-6, colon 26 cells were cultured as a monolayer in the presence or absence of PDGF-BB, PDGF-AB, PDGF-AA, TGF-β1 (R&D systems), and NS-398.
DNA microarray. cDNA and cRNA were synthesized using a reverse SuperScript Choice system (Invitrogen) and a MEGAscript T7 kit (Ambion, Austin, TX, USA), respectively, according to the manufacturers’ instructions. After removing the mononucleotides and short oligonucleotides with a CHROMA SPIN STE-100 column (Clontech, Palo Alto, CA, USA), gene expression was examined with GeneChip Mouse Expression Array 430 A (Affymetrix, Santa Clara, CA, USA) according to the manufacturers’ instructions. Expression of mRNA was calculated with Affymetrix GeneChip version 3.3 and Affymetrix Microarray Suite version 5.0 at a target intensity of 500.
Statistical analysis. Results were analyzed with an SAS statistical package (version 6.12) using the Dunnett's multiple comparison test, and P-values between the two groups of <0.05 for in vivo animal experiments and <0.0001 for in vitro cell culture experiments were considered significant.
Production of PTHrP and IL-6 in mice bearing subcutis colon 26 and liver colon 26. When subcutaneously transplanted with colon 26, the mice started losing body weight on day 12 and lost on average more than 5 g of body weight by day 17 (18.9% of initial body weight). In addition, they developed humoral hypercalcemia of malignancy symptoms by day 17 as revealed by elevated levels of blood calcium and decreased levels of plasma glucose. In contrast, mice intrahepatically transplanted with colon 26 did not show significant body-weight loss until day 17. Body-weight loss in mice bearing subcutis colon 26 was independent of tumor burden; the average tumor weight in liver was 1.16 g on day 17, whereas in the subcutis the average tumor weight was only 0.27 g (see Supplementary Fig. S1). These results are consistent with previously reported data.(6) Furthermore, earlier onset of cachectic syndromes in mice bearing subcutis colon 26 was not the consequence of genetic changes in the tumor cells because the subcutis colon 26 recovered from the mice did not induce early onset of cachectic symptoms when re-inoculated into the liver (not shown).
Previously, it was reported that blood levels of IL-6 were higher in mice bearing subcutis colon 26 than those with liver colon 26.(6) Because PTHrP is also involved in the development of wasting and other cachectic symptoms,(3) the plasma PTHrP levels in mice with subcutis colon 26 and those with liver colon 26 were compared. Significant levels of PTHrP were detected in mice with subcutis colon 26 but not in the mice with liver colon 26. In mice with subcutis colon 26, plasma levels of PTHrP and IL-6 were 12 pmol/L and 1 ng/mL, respectively, on day 12 and continued to increase, reaching 18 pmol/L and 1.7 ng/mL, respectively, on day 17 (Fig. 1a,b). In contrast, in mice with liver colon 26, plasma levels of PTHrP and IL-6 were more or less the same as those of non-tumor-bearing mice on day 12 and remained at low levels (4.7 pmol/L of PTHrP and 0.4 ng/mL of IL-6), as observed even on day 17. Although TGF-β has been implicated to augment the PTHrP expression of breast cancer cells, particularly at the sites of bone metastasis,(10) the level of TGF-β mRNA on day 17 was higher in liver colon 26 than in subcutis colon 26 (Fig. 1c).
Induction of PTHrP production from colon 26 in spheroid cultures. The above and other results(7,8) suggest that early onset of wasting and cachectic syndromes in mice bearing subcutis colon 26 is largely attributable to elevated levels of IL-6 and PTHrP. To clarify conditions representing the microenvironments that induce production of IL-6 and PTHrP from colon 26 cells, the culture conditions were explored. Only low levels of IL-6 and PTHrP were detected in the culture media of conventional monolayer cultures of colon 26. Because microvessel densities are considered to be higher in the liver than in the subcutis, tumor cells in the subcutis might grow under the more hypoxic and poorer nutrient conditions of the subcutis rather than in the liver. Therefore, whether production of PTHrP and IL-6 is augmented by a hypoxic or a low-glucose condition in monolayer cultures was examined. A hypoxic condition (1% O2) strongly induced HIF1-α expression (not shown) but did not increase the production of IL-6 and PTHrP from colon 26 cells (Fig. 2a,b). The amount of IL-6 in the culture medium, however, was significantly elevated when monolayers of colon 26 cells were cultured in a low-glucose medium, whereas the amount of PTHrP was not (Fig. 2a,b).
Circumstances surrounding cancer cells in 3-D spheroid cultures have been thought to be more relevant to in vivo conditions than to monolayer cultures. In fact, cancer cells proliferate actively in the periphery of spheroids, but necrotic cells are abundant in the middle of spheroids (Fig. 2c). In addition, the circumstances of colon 26 cells in spheroid cultures seems to be closer to that of subcutis colon 26 than of liver colon 26 because cells undergoing necrosis were also observed in subcutis colon 26 but not in liver colon 26 on day 17 (Fig. 2d,e). When colon 26 cells formed spheroids, the amount of PTHrP in the culture media markedly increased even under a normoxic condition with glucose, whereas the amount of IL-6 was not significantly different between spheroid and monolayer cultures (Fig. 2a,b).
Involvement of COX-2 in the production of PTHrP under necrobiotic circumstances. To further understand the mechanism underlying the tumor inoculation site-dependent induction of PTHrP expression, the gene expression profiles of subcutis colon 26, liver colon 26, and the spheroid and monolayer cultures of colon 26 cells were examined. A DNA microarray of 14 000 genes revealed that the expression of 69 genes was more than three-fold in subcutis colon 26 compared to liver colon 26. Similar analyses between the monolayer and spheroid cultures of colon 26 cells demonstrated that the expression of 311 genes was more than three-fold in spheroid cultures compared to monolayer cultures. The expression of 20 of the genes appeared to be commonly up-regulated (more than three-fold) in both subcutis colon 26 and spheroid cultures (Fig. 3).
The 20 genes that were up-regulated in both subcutis colon 26 and spheroid cultures include cytokines and growth factors such as BMP-2, epiregulin, and activin A, in addition to PTHrP. However, none of these increased PTHrP production when added to the monolayer cultures of colon 26 (not shown). In contrast, COX-2, which has been reported to mediate the induction of PTHrP production by IL-1α in synovial tissue(11) was also included in the above 20 genes. Therefore, the expression of COX-2 was confirmed using RT-PCR and it was found that levels of COX-2 mRNA were indeed higher in subcutis colon 26 than in liver colon 26 and were also higher in spheroid cultures than in monolayer cultures (Fig. 4a,b). Furthermore, although neither growth nor viability of colon 26 cells was affected by 10 µM or lower concentrations of COX-2 inhibitor NS-398,(12) production of PTHrP in the spheroid cultures was markedly suppressed by 1 µM of NS-398 (Fig. 4c), indicating that COX-2 is indeed involved in the production of PTHrP under necrobiotic circumstances. Induction of COX-2 expression in the spheroid cultures and inhibition of PTHrP production by NS-398 was not a non-specific event because COX-2 expression did not increase in the monolayer cultures of colon 26 cells under a low-glucose condition, and IL-6 production from colon 26 cells in a low-glucose condition was not inhibited even in the presence of 10 µM NS-398 (Fig. 4d). Whether COX-2 inhibition suppressed the production of PTHrP in vivo was also examined. Administration of 4 mg/kg of NS-398 to mice bearing subcutis colon 26 significantly and that of 40 mg/kg more profoundly decreased plasma PTHrP levels without inhibiting the growth of tumors (Fig. 5), confirming that COX-2 is involved in the production of PTHrP in tumor cells in vivo.
Induction of COX-2 expression by PDGF has been reported in several cell types, including fibroblasts, mesangial cells, and osteoblasts.(13–15) The authors also found that blood PDGF-BB levels were elevated in mice bearing subcutis colon 26 (Fig. 6a). Therefore, whether PDGF-BB induced the expression of COX-2 in colon 26 cells was examined. PDGF-BB significantly augmented the production of PTHrP from monolayer cultures of colon 26 in a dose-dependent manner, whereas it did not affect the production of IL-6 (Fig. 6b,c). Neither PTHrP nor IL-6 production was affected by PDGF-AA, PDGF-AB, or TGF-β (Fig. 6b,c, not shown). In addition, the COX-2 inhibitor NS-398 significantly inhibited PDGF-BB-induced PTHrP production from monolayer culture of colon 26 (Fig. 6d).
Previously, the authors have reported that nude rats and nude mice bearing LC-6 human lung cancer xenografts secreted PTHrP and several inflammatory cytokines, including IL-6 and IL-8, and that they exhibited wasting, hypercalcemia, and other cachectic symptoms. Administration of an anti-PTHrP antibody markedly improved body-weight loss, hypercalcemia, and other cachectic symptoms without affecting the blood IL-6 levels or acute phase reactions caused by inflammatory cytokines(3) suggesting that expression of PTHrP and IL-6 is differently regulated. In the present study, the authors explored the mechanism underlying the elevated expression of IL-6 and PTHrP in subcutis colon 26 compared to liver colon 26. Culturing colon 26 cells under different conditions in vitro revealed that IL-6 production was increased by monolayer culture under a low-glucose condition but not by spheroid culture. In contrast, PTHrP production was increased by spheroid culture but not by monolayer culture even under a low-glucose condition, indicating that PTHrP and IL-6 are differently expressed under different conditions.
It is considered that TGF-β released from bone at the site of bone metastasis enhances the production of PTHrP in tumor cells, leading to further augmentation of bone resorption and cachectic symptoms including hypercalcemia. Indeed, TGF-β induced PTHrP expression in tumor cells in vitro.(10) Moreover, Gli2, a downstream transcriptional effector of the Hedgehog signaling pathway, augments PTHrP expression and osteolytic bone metastasis.(16) However, in the present study the levels of TGF-β and Gli2 mRNA were not elevated in either subcutis colon 26 or spheroid cultures. In contrast, COX-2 expression was elevated in subcutis colon 26 compared to in liver colon 26, and it was also higher in spheroid cultures than in monolayer cultures. The result that COX-2 inhibitor NS-398 not only suppressed PTHrP production in the spheroid cultures but also significantly decreased the serum PTHrP levels without inhibiting the tumor growth in vivo clearly demonstrate that COX-2 is at least in part involved in PTHrP production in some circumstances. Because cells in the central part of the spheroid undergo necrosis, production of PTHrP may be induced by necrobiotic processes in which COX-2 is involved rather than by hypoxia or by poor nutrient conditions. Induction of COX-2 expression in the necrotizing cells has also been reported in the spheroid cultures of human dermal fibroblasts.(17) In contrast, IL-6 is induced by poor nutrient conditions, such as low glucose, rather than by hypoxia or necrobiotic processes occurring in the spheroid cultures.
PDGF-BB induced the production of PTHrP from monolayer cultures of colon 26, and the PDGF-BB-induced production of PTHrP from the monolayer cultures of colon 26 was inhibited by NS-398. Thus, it is possible that elevated levels of PDGF-BB in tumor tissues increase the expression of COX-2 of colon 26 cells, leading to PTHrP production and cachectic symptoms. However, PDGF-B expression did not increase in the spheroid cultures of colon 26; PDGF-B may be secreted from stroma cells surrounding tumors in vivo. The factors that mimicked PDGF-BB and induced COX-2 in spheroid cultures remain to be elucidated. Interestingly, neither PDGF-AA nor PDGF-AB augmented PTHrP production from the monolayer culture of colon 26. One possibility is that colon 26 cells predominantly express PDGFR-β and a very small amount of PDGFR-α.
Previously, it was demonstrated that when administered into mice bearing a large burden of colon 26, indomethacin facilitated the growth of the tumor but alleviated cachectic symptoms.(18) Moreover, COX-2 inhibitors such as celecoxib and meloxicam have been shown to improve cachectic symptoms in mouse tumor models.(19–21) Although the mechanisms underlying the improvement of cachexia by COX-2 inhibitors have not been fully understood, reduction of IL-6 was observed after administration of COX-2 inhibitors.(20,21) In the present study, COX-2 inhibitor NS-398 significantly inhibited the production of PTHrP in spheroid cultures but did not affect IL-6 production induced by a low-glucose condition. In addition, NS-398 reduced the plasma PTHrP levels in vivo without inhibiting tumor growth. The results indicate an additional action of COX-2 inhibitors on cachexia, that is, impairment of PTHrP production from tumors.
In the mouse colon 26 tumor model, administration of the anti-PTHrP antibody restored the normal blood calcium level and significantly improved body-weight loss. In contrast, administration of an anti-IL-6 antibody significantly increased the blood glucose level but did not affect the blood calcium level, which led to partial improvement of body-weight loss.(9,10) Therefore, it seems likely that wasting is not directly associated with hypercalcemia and that IL-6 and PTHrP are differently involved in the development of cachectic syndromes. Interestingly, coadministration of the anti-IL-6 and the anti-PTHrP antibodies to mice bearing subcutis colon 26 did not give rise to an additive effect on body-weight loss compared to the anti-PTHrP antibody alone (Supplementary Fig. S2). The degree of involvement of PTHrP and IL-6 may be different at different stages during the development and progression of cachexia.