Colon-26 adenocarcinoma cells (C26 cells) (Medical Science Experimentation Center of Sun Yat-Sen University, China) and Lewis lung carcinoma cells (LLC cells) (Shanghai Branch of Chinese Academy of Science, China) were cultured in Dulbeccos modified Eagles medium (DMEM) plus 10% fetal bovine serum (FBS) with 1% penicillin/streptomycin at 37°C and 5% CO₂. Before injection of C26 cells into CD2F1 mice (C26 model) or injection of LLC cells into C57BL/6 mice (LLC model), cells were counted and resuspended at 5×10⁷ cells/ml in sterilized PBS. The right flanks of the mice were shaved, and they were administered a subcutaneous (s.c.) injection of either 5×10⁶ C26 cells or LLC cells suspended in 100 µl sterilized PBS (tumor-bearing mice, TB mice) or 100 µl sterilized PBS without cells (control mice, CN mice). Eight-week-old male CD2F1 or C57BL/6 mice were allocated randomly into one of four experimental groups: i) tumor-bearing mice in early cachexia (ECC mice); ii) tumor-bearing mice in late cachexia (LCC mice); iii) ECC-matched control mice (ECC-CN mice); and iv) LCC-matched control mice (LCC-CN mice). The animals were monitored daily and were euthanized separately at 24 days (ECC and ECC-CN mice) and 36 days (LCC and LCC-CN mice) following injection (7,9). Tumors, quadriceps, tibialis anterior, soleus, and gastrocnemius muscles, hearts, spleens, and epididymal fat were immediately harvested and weighed. For subsequent studies, tibialis anterior muscles were fixed in 4% paraformaldehyde, and the other tissues were quickly frozen in liquid nitrogen and stored at −80°C. All experiments were approved by the Animal Care and Use Committee of Tongji Medical College of Huazhong University of Science and Technology.

To determine the myofiber cross-sectional area (CSA), hematoxylin and eosin (H&E) staining was performed on a middle cross-section of the tibialis anterior. Images were acquired using a digital camera and were quantified using ImageJ software (NIH, Bethesda, MD, USA). Within each section, five view fields with 100 myofibers per field were measured (10).

To visualize the outlines of myofibers, 10 µm sections were obtained from the middle of the tibialis anterior. The sections were then incubated with Alexa Fluor 350-conjugated wheat germ agglutinin (Invitrogen, Carlsbad, CA, USA) for 2 h and subsequently washed in PBS. Images were acquired using a digital camera (3,49). Representative view fields were elected and recorded.

RNA was extracted from quadriceps muscles using TRIzol reagent (Invitrogen) according to the manufacturers instructions. The concentration and purity of the RNA solution were determined by Epoch microplate spectrophotometer (BioTek Instruments, Inc., Winooski, VT, USA). RNA (1 µg) was used for reverse transcription. Reverse transcription of mRNA was performed using a RevertAid First-Strand cDNA Synthesis kit (Thermo Fisher Scientific, Inc., Rockford, IL, USA) in a total reaction volume of 10 µl. Dilution (1:10) of the RT product was used as template for the quantitative real-time PCR (qPCR). qPCR was performed with the 2X SYBR-Green Mix (Thermo Fisher Scientific) using a LightCycler^® 480 (Roche Diagnostics, Mannheim, Germany) in a total reaction volume of 10 µl with the primers from Sangon Biotech, Co., Ltd. (Shanghai, China). The amplification procedure was 95°C pre-denaturation for 10 min followed by 95°C for 15 sec, 60°C for 10 sec and 72°C for 30 sec for a total of 40 cycles. The data were normalized to glyceraldehyde-3-phosphate dehydrogenase (GAPDH) expression and the relative expression was calculated using the formula: 2^−ΔCt (ΔCt = Ct _gene - Ct _GAPDH). The primer sequences were as follows: Myostatin: F-AGTGGATCTAAATGAGGG CAGT and R-GTTTCCAGGCGCAGCTTAC; PGC1α: F-AA CCACACCCACAGGATCAGA and R-TCTTCGCTTTAT TGCTCCATGA; FoxO3a: F-GCAAGCCGTGTACTGTGGA and R-CGGGAGCGCGATGTTATCC; MuRF1: F-AGCAT CAAGATCCGTCTGACA and R-CCAGAGCCGTCCACA ACAAT; Atrogin1: F-ACACATCCTTATGCACACTGG and R-TCTCCATCCGATACACCCACA; GAPDH: F-GGTGAA GGTCGGAGTCAACGG and R-GAGGTCAATGAAGGGG TCATTG.

The quadriceps muscles were homogenized, and total protein was extracted using RIPA protein lysis buffer (P1003; Beyotime Institute of Biotechnology, Nantong, China) with freshly added protease inhibitor cocktail and phenylmethylsulphonyl fluoride (PMSF). The protein concentration of the samples was measured using BCA method. A total of 80 µg of protein was subjected to a 10% SDS-PAGE gel to separate the proteins by gel electrophoresis, and they were then transferred onto polyvinylidene fluoride (PVDF) (0.45 µm; Millipore, Boston, MA, USA) membranes. The membranes were blocked for 1 h at 37°C in 5% (w/v) non-fat dried skim milk (blocking buffer) and incubated with primary antibodies in blocking buffer overnight at 4°C. The primary antibodies were as follows: anti-atrogin1 antibody (#AP2041), purchased from ECM Biosciences, Versailles, KY, USA; anti-PGC1α antibody (ab54481), purchased from Abcam, Cambridge, MA, USA; anti-Phospho-FoxO3a (#9466) and anti-FoxO3a (#2497) antibodies obtained from Cell Signaling Technology, Danvers, MA, USA; anti-C/EBPβ (sc-7962), anti-HDAC1 (sc-7872), anti-HDAC2 (sc-7899), and anti-HDAC3 (sc-11417) antibodies acquired from Santa Cruz Biotechnology, Santa Cruz, CA, USA. The membranes were washed and incubated with the appropriate horseradish peroxidase-conjugated secondary antibody (Invitrogen) in blocking buffer for 2 h at room temperature. Finally, the membranes were washed before detection. Quantitative analyses of protein expression were performed using ImageJ software (25).

All values were represented as the mean ± standard error (SEM) unless stated otherwise. Differences between group means were determined using the Students t-test with Graphpad Prism 5 unless otherwise specified. A two-sided P-value of <0.05 was considered to indicate statistically significant result.

For the C26 model mice, at 24 days following C26 tumor implantation, the body weights of the ECC-CN and ECC mice were both increased (Fig. 1A). From the 30th day, the body weights of the LCC mice started to decrease, and this trend was maintained until the 36th day, when the mice in this group were sacrificed (Fig. 1B). For the LLC model mice, in contrast with the C26 model mice, the body weights of both the ECC and LCC mice consistently increased until day 36 (Fig. 1C and D). Nevertheless, the ECC mice had already developed cancer cachexia, as the tumor-free body masses of these mice were significantly decreased for both the C26 and LLC models. Interestingly, the tumor masses of the LCC mice were higher than that of ECC mice in LLC model. But no significant differences existed between the ECC mice and LCC mice in C26 model (Fig. 1E and F).

The tumor-free masses of the ECC mice were reduced by ~18 and 13% compared with those of ECC-CN mice for the C26 and LLC models, respectively. A similar finding was observed for the LCC mice, but with higher rates of reduction (~28 and 29% compared with the C26 and LLC model LCC-CN mice, respectively). The tumor-free masses of the LCC mice were obviously less than those of the ECC mice for both models (Figs. 2A and 3A).

For the above reasons, we defined the TB mice sacrificed on the 24th day as ECC mice (their tumor-free body masses were decreased by <20%) and the TB mice sacrificed on the 36th day as LCC mice (their tumor-free body masses were decreased by >20%).

As previously reported, C26 cachexia results in skeletal muscle, epididymal adipose and heart mass losses (7). The masses of the organs and muscles harvested from the C26 and LLC model mice are listed in Tables I and II. We obtained similar findings for these two models (Figs. 2 and 3), except that the heart mass did not exhibit a substantial change (Fig. 3F) in the LLC model. We found that the LCC mice had much greater losses of muscle and epididymal adipose mass than the ECC mice for both models (Figs. 2 and 3), except for the soleus muscle mass (Figs. 2D and 3D). These data further demonstrated that the LCC mice suffered from more severe cancer cachexia than the ECC mice.

C26 cancer cachexia has been reported to result in a large increase in the mass of the spleen (3). This conclusion is consistent with our results (Fig. 2G), and we obtained the same results with the LLC model (Fig. 3G). However, the spleen masses did not significantly differ between the ECC and LCC mice for either model (Figs. 2G and 3G). Additionally, the heart masses did not significantly differ between the ECC or LCC mice and their matched CN mice for the LLC model (Fig. 3F). These results differed from those for the C26 model mice but are consistent with those of a previous study (14).

Representative images of H&E-stained tibialis anterior middle cross sections from the mice in each group are shown in Fig. 4A. To better visualize the outlines of muscle fibers, skeletal muscle cross sections taken from tibialis anterior muscles were incubated with fluorescently labeled wheat germ agglutinin (Fig. 4B). The average muscle fiber CSAs declined by 15 and 45% in the ECC and LCC mice, respectively, compared with their matched CN mice for the C26 model (Fig. 5A), and these values declined by 13 and 43%, respectively, for the LLC model (Fig. 5B). Additionally, the changes in muscle mass were confirmed by analyses of the size distributions of myofibers in each group. These results indicated that the muscle fiber CSA of the LCC mice, but not the ECC mice, was obviously less than that of the LCC-CN mice for both models (Fig. 5C and D).

To determine the mRNA levels of some molecules involved in muscle wasting, five prominent molecules were selected for analysis in each group. The mRNA levels of these molecules did not obviously change in the ECC mice of both models, except for that of atrogin1, which was increased in the ECC mice compared with the ECC-CN mice for the C26 model, but not the LLC model (Fig. 6A and C). In contrast with the ECC mice, the levels of several mRNAs were increased in the muscles from the LCC mice of the two models. The mRNA levels of atrogin1 and FoxO3a were increased in the LCC mice of both models (Fig. 6B and D). Additionally, the mRNA expression of myostatin was increased in the muscles from the LLC model LCC mice (Fig. 6D). A previous study revealed that the mRNA level of PGC1α is consistently decreased (25). However, we found no significant difference in this mRNA level between the CN and TB mice of either model (Fig. 6).

To explore the underlying mechanism of the increased severity of cancer cachexia in the LCC mice compared with the ECC mice, the protein levels of some crucial molecules involved in muscle wasting, such as atrogin1, FoxO3a, PGC1α, C/EBPβ and class I HDACs, were determined. The protein level of atrogin1 was increased in the TB mice compared with their matched CN mice for both models (Figs. 7 and 8). We subsequently detected the expression of the molecules that may regulate the expression of atrogin1. The FoxO3a (not phospho-FoxO3a) protein level was also found to be increased in the muscles from the TB mice compared with their matched CN mice for both models (Figs. 7 and 8). The PGC1α protein level was obviously decreased in the LCC mice compared with the LCC-CN mice, but no significant difference was observed between the ECC and ECC-CN mice (Figs. 7 and 8). In contrast, the C/EBPβ protein level was obviously increased in the LCC mice compared with the LCC-CN mice, but no significant difference was detected between the ECC and ECC-CN mice (Figs. 7 and 8).

Furthermore, the protein levels of three class I HDACs were determined, and those of HDAC1 and HDAC3 were found to be slightly increased in the LCC mice compared with the LCC-CN mice for both models, while only HDAC2 was increased in the LLC model LCC mice (Figs. 7 and 8).

Cancer cachexia has been widely studied. A previous report demonstrated that lipid metabolism in adipose tissue differs between C26 model ECC and LCC mice. ECC was defined by the author as occurring no more than 12 days following C26 tumor implantation, when the white adipose tissue mass in cachectic mice is moderately reduced (34–42%) and weight loss is <10% of the initial body weight (54). Normally, loss of fat always occurs before muscle wasting in cancer cachexia. Therefore, in the present study, we prolonged the period defined as ECC for the optimal assessment of muscle wasting. We found that this definition was suitable for research of muscle wasting in the C26 and LLC models.

Prior to this study, many research groups focused on muscle wasting only in ECC. Thus, we questioned whether the molecules regulating muscle wasting in LCC are similar to those in ECC. The aim of the present study was to determine the differences between muscle wasting in ECC and LCC.

The alterations in the tumor-free body masses, the masses of various tissues and the cross-sectional areas (CSAs) of muscle fibers differed between the ECC and LCC mice and their matched CN mice. These results demonstrated that obvious differences existed between ECC and LCC. From this point of view, the definitions of ECC and LCC in the C26 and LLC models were also feasible.

Myostatin plays an important role in many types of muscle atrophy (19). However, its mRNA level was only altered in the muscles from the LLC model LCC mice. This result might indicate that the mRNA expression of myostatin is not a sensitive indicator of muscle wasting in our models, especially in the C26 model. Although the mRNA level of myostatin did not obviously change, the expression of the downstream molecule FoxO3a was altered. The protein level of FoxO3a was increased in the TB mice of both models, and its mRNA level was only increased in the LCC mice, but not in the ECC mice. Atrogin1 and MuRF1 are both important E3 ubiquitin ligases involved in muscle wasting (28), but only the mRNA level of atrogin1, and not that of MuRF1, was increased in our models. In addition, the protein level of atrogin1 was increased in the TB mice of both models. These results suggested that atrogin1 might be the more crucial gene involved in muscle wasting in our models. The altered atrogin1 expression directly induced muscle wasting in the TB mice, and no significant difference in its expression was detected between the ECC and LCC mice. Collectively, the myostatin-FoxO3a-atrogin1 axis indeed played an important role in muscle wasting in our models.

Currently, increasing numbers of studies are focusing on the molecules that affect the myostatin-FoxO3a-atrogin1 axis. We found that the molecules involved in muscle wasting were not exactly the same in the ECC and LCC mice of each model. In addition, we focused on the molecules that were altered only in the muscles from the LCC mice. Although the mRNA level of PGC1α was not altered in the TB mice, its protein level was decreased in the LCC mice, but not in the ECC mice, of both models. These results indicated that C/EBPβ, HDAC1 and HDAC3 might play roles in promoting cancer cachexia, especially during the late stage. Correspondingly, PGC1α might play an opposite role. As previously reported, muscles from the TB mice had a higher level of phosphorylated C/EBPβ, along with a modest increase in total C/EBPβ, on day 14 for the LLC model (10,44). In our opinion, the LLC model TB mice sacrificed on day 14 were the ECC mice in this study. However, we measured the protein levels of total C/EBPβ in the muscles from the ECC mice of both models and found that they were not significantly different. This result is consistent with the previous report. In addition, we showed that the protein expression of HDAC1 was increased in muscles from the LCC mice of both models. The change in HDAC3 expression was similar to that in HDAC1 expression. In contrast, HDAC3 has been reported to be decreased in dexamethasone-induced muscle wasting (47). Although this finding is not consistent with our data, it suggests that HDAC3 is indeed involved in muscle wasting and might have different roles in different models. The role of PGC1α in protecting muscles from wasting has been proven (35,40). In our experiment, this role might be inhibited in both the C26 and LLC models.

Many studies have verified that the levels of microRNAs are altered in muscles from cancer cachectic mice. We used different miRNA target-predicting algorithms (for example, TargetScan and RegRNA) to identify potential miRNAs that could affect the aforementioned genes. We found conserved miR-30c sites in the 3UTRs of atrogin1, FoxO3a and HDAC3 (Fig. 9A). Moreover, we found a conserved miR-30c site in the 5UTR of PGC1α (Fig. 9A). Consequently, we observed that the miR-30c level was not altered in muscles from the ECC mice of the C26 model but that it was decreased in the LCC mice of both models (Fig. 9B and C). Our observations indicate that miR-30c might be involved in the process of cancer cachexia by interfering with the expression of PGC1α, atrogin1, FoxO3a and HDAC3. Further research needs to be performed to determine whether these genes are directly regulated by miR-30c.

By comparing the changes in the expression of crucial molecules involved in muscle wasting in both the ECC and LCC mice, we confirmed that some molecules exhibited varying degrees of change in our models. Although the expression levels of several other molecules did not obviously change in the ECC mice, they were significantly altered in the LCC mice, such as PGC1α, C/EBPβ and HDACs. However, it is still difficult to conclude that these unchanged molecules do not play roles in the ECC mice. For instance, the role of HDACs in muscle wasting has been realized in recent years, and pharmacological interventions with HDAC inhibitors have been shown to increase myofiber size and counter the functional decline of dystrophic muscles (55). In addition, class II HDACs promote neurogenic muscle atrophy by inducing E3 ubiquitin ligases (56). These findings suggest that HDACs might accelerate the process of muscle wasting induced by cancer. A previous report has shown that the total protein level of HDAC1 does not change in disused muscle but that the relative abundance of HDAC1 is decreased in the nuclear fraction and increased in the cytosol (49). These data suggest that HDAC1 may shuttle out of the nucleus to exert its function within the cytoplasm. In our models, the protein level of HDAC1 was increased in the LCC mice, but not in the ECC mice. This finding does not indicate that HDAC1 plays no role in muscle wasting in ECC mice. The function of this molecule might have been further enhanced when its level was increased in the LCC mice.

In conclusion, our results have revealed that the expression levels of several molecules are altered in muscles from LCC mice, but not in those from ECC mice. From our results we deduce that these changes may promote muscle wasting in late cancer cachexia. The data in this study may facilitate the further understanding of the underlying mechanism involved in the development of cancer cachexia. However, our present study on muscle wasting in late cancer cachexia merely sheds light on the underlying mechanism, which remains poorly understood. Thus, further investigation is warranted to delineate the foundation of late cancer cachexia to provide a solid basis for the clinical prediction and prevention of muscle wasting in cancer cachexia.