The Effect of Exercise on Gene Expression and Signaling in Mouse

BASIC SCIENCES , 2 an, CHINA; ’ To screen for can- least moderate physical activ- Purpose: , QUANSHENG SU 1 software was utilized to identify EXERCISE, MELANOMA, GEO protein interaction network revealed that Department of Kinesiology, Coastal 1494, 2020. – 3 – Key Words: cytokine receptor interaction, chemokine signaling path- , XIAOMIN DING – 1 d, including 294 upregulated and 21 downregulated genes. Melanoma is the deadliest and most metastatic form of skin cancer survivors should perform at ity (6). However, there isactivity limited and data on exercise theand on effects posttreatment. There of prognosis are physical clear differences andof between the exercise quality effects on of cancer prevention, lifeand versus exercise during would treatment never and be survival, prescribedtraditional for treatment. most Moreover, there cancers is without no evidencein that itself exercise can directly eradicate aan tumor. Thus, adjunct exercise therapy is along more with like a traditional better clinical understanding treatments, of and thelinking underlying physical biological activity mechanisms andto exercise outcomes to in cancer cancer surviors prevention is and warranted. cancer which affects tens ofannually, and thousands the of prevalence people is worldwide growingtype of faster solid than cancer (7). any The other risk factorsknown, for melanoma including are both well intrinsic andfor extrinsic variances factors. in Except genotype andextrinsic phenotype, the cause most is important sunthe risk exposure. of skin Exercise cancer, however, mayture the current is help state of ambiguous. reduce the For litera- creased example, Moore levels et al. ofassociated found with that physical in- an increased activitycase-control study risk found in of that higher levels melanoma of leisure (8), physical activity while time a were gg may be the critical genes in the complement and coagulation cas- , Vol. 52, No. 7, pp. 1485 exercise and nonexercise mice. The R immune response, inflammatory response, and positive regulation of the on and pathway analysis were performed. Results were visualized using Exercise may ameliorate the immune response and inflammatory response 1485 infection, cytokine ABSTRACT , QIANJIN WANG 3 Conclusions: st significant module identified from the protein The GSE62628 expression profile was downloaded from Gene Expression Omnibus data- Med. Sci. Sports Exerc. ® 4,5 Staphylococcus aureus Methods: Department of Physical Education, Federal University of Maranhão (UFMA), São Luís-MA, 4 in the infection pathway. , JASON CHOLEWA [email protected]. In total, 315 differentially expressed genes were obtaine 2 Results: S. aureus s Web site (www.acsm-msse.org). ’ 5). According to the continuous update – Laboratory of Cellular and Molecular Biology of Skeletal Muscle (LABCEMME), BRAZIL 5 , and NELO EIDY ZANCHI 1 an, China. E-mail: zhix , HUAYU SHANG ’ Copyright © 2020 by the American College of Sports Medicine. Unauthorized reproduction of this article is prohibited. Copyright © 2020 by the American College of Sports Medicine. Unauthorized reproduction 1,2 didate hub genes associated with the effects of exercise onthis melanoma tumor process tissues and using to bioinformatics review the analysis. potential signaling pathwaysbase. involved in This data set contains 10 melanoma tumor tissues from two groups of XIA, Z., H. SHANG, J. CHOLEWA,sion Q. and WANG, Signaling X. in DING, Mouse Q. Melanoma SU, Tumors. Y. ZHAO, and N. E. ZANCHI. The Effect of Exercise on Gene Expres- differentially expressed genes between samples, andCytoscape functional software. annotati The functional analysis showed that these genesERK1/2 were mainly cascade enriched in in biological processAldh3b1, functional and groups. Apob. The The top pathway 10 analysis candidate of hub the genes mo were C3, Kng1, C3ar1, Ptafr, Fgg, Alb, Pf4, Orm1, the complement and coagulation cascades, way and phagosome were mainly involved.cades C3, pathway, C3ar1, and Kng1, Ptafr, and F DATABASE, BIOINFORMATICS ANALYSIS, DIFFERENTIALLY EXPRESSED GENES, BIOLOGICAL PATHWAYS in melanoma tissue, and further studies exploring their relationships are warranted. ancer is ranked as oneworldwide, of resulting the in leading a causes heavy ofburden social on death and the economic public health sector. A high volume of

School of Sports Medicine and Health, Chengdu Sport University, Chengdu, CHINA; Exercise Physiology and Biochemistry Laboratory, College of Physical Education, Jinggangshan University, Ji Carolina University, Conway, SC; YAN ZHAO BRAZIL; and Copyright © 2020 by the American College of SportsDOI: Medicine 10.1249/MSS.0000000000002291 0195-9131/20/5207-1485/0 MEDICINE & SCIENCE IN SPORTS & EXERCISE Address for correspondence: ZhiBiochemistry Xia, Laboratory, College Ph.D., of Exercise PhysicalUniversity, Physiology Ji Education, and Jinggangshan Huayu Shang, Jason Cholewa, andthis Nelo work. Eidy Zanchi contributed equally to Submitted for publication August 2019. Accepted for publication January 2020. Supplemental digital content is available for thisappear article. in Direct URL the citations printed textof and this are article on provided the in journal the HTML and PDF versions C 2 ZHI XIA and Signaling in Mouse Melanoma Tumors The Effect of Exercise on Gene Expression research has been conducted to exploreprevent potential and/or strategies to treat cancerence and (1), its with secondary an harmful increasingfects emphasis influ- on of the exercise preventative (2 ef- project expert report of 2018search released by Fund/American World Institute Cancer Re- for Cancer Research, all

Downloaded from http://journals.lww.com/acsm-msse by BhDMf5ePHKbH4TTImqenVBLNPkpHwi7kiznfBx+f7nKqKvM/3IIQKQjS9RocO1Rj on 07/04/2020 Downloaded from http://journals.lww.com/acsm-msse by BhDMf5ePHKbH4TTImqenVBLNPkpHwi7kiznfBx+f7nKqKvM/3IIQKQjS9RocO1Rj on 07/04/2020 are associated with a 30% reduction in melanoma risk (9). 4hinadigestionenzymemixtureofRoswellParkMemorial Physical activity and exercise are often done in thin clothing Institute 1640 with 0.4 mg·mL−1 collagenase I and 10 U·mL−1 and outdoors, which poses an increased risk of sun exposure pulmozyme. Digested tumor fragments were resuspended and and sunburns. In addition, Moore et al. found that exercise was filtered (70 μm cell strainer cap) and washed twice in phosphate more associated with melanoma in areas with high ultraviolet ra- buffer saline + 2% fetal bovine serum. Total RNA was extracted diation, and suggested that solar irradiance was an important trig- using the TRIzol reagent. In regard to the exercise intervention, a ger for this relationship. It is currently unclear whether exercise 12-cm-diameter running wheel was placed in the cage as a model protects against melanoma, and, if so, via which mechanisms. of voluntary exercise. Access to running wheels was given 4 wk During the last decade, microarray technology and bioinfor- before tumor cell inoculation and during the 2-wk tumor chal- matics analysis have been widely used to screen genetic alter- lenge. Total running distance was evaluated by placement of bi- ations at the genome level, which have helped to identify cycle computers on the running wheels, and the average wheel differentially expressed genes (DEGs) and signaling pathways running distance was 4.1 km per mouse per day. The results involved in the carcinogenesis and progression in individual showed that exercise before and during B16F10 tumor chal- cancers (10). The bioinformatics strategy is an uncommon lenge reduced tumor growth and volume by 61% (P <0.01) analysis in the field of sports medicine, and original data sets and 67% (P < 0.05), respectively (14). on this topic in the public databases are scarce. Thus, this is Microarray data. Themicroarraygeneprofile(ID: the first research to our knowledge to investigate the potential GSE62628, published by Pedersen et al. [14]) of melanoma targets and signalings of voluntary exercise on melanoma tu- tumor tissue from two groups of exercise (n = 5) and mor tissues via the bioinformatics strategy. Recently, Wang nonexercise C57BL/6 mice (n = 5) was downloaded from et al. used gene profile analysis to compare 2 sets of subcu- the Gene Expression Omnibus (http://www.ncbi.nlm.nih. taneous white adipose tissue samples in C57BL/6 mice to gov/geo/). Gene expression levels were measured using discover DEGs of sedentary and exercise interventions. GPL6246 [MoGene-1_0-st] Affymetrix Mouse Gene 1.0 ST SLC27A1, COX7A1, PPARGC1A, FABP3, and UCP1 were Array (Affymetrix Inc. Santa Clara, CA). identified as top 5 hub genes in white adipose tissue after 11 d Microarray data preprocessing. The platform and raw

BASIC SCIENCES of exercise, and the PPAR signaling pathway, adipocytokine expression data in Series Matrix format were downloaded as signaling pathway, and the insulin resistance pathway were TXT files. The data were preprocessed by using the affy and screened out as the critical pathways involved in the enhanced preprocess Core package from Bioconductor (Bioconductor, relationship between the browning of white adipose tissue and Fred Hutchinson Cancer Research Center, Seattle, WA). insulin resistance by exercise (11). Similar analyses in the on- In this process, data were calibrated, standardized, and log2 cology research field can also be found in the works of Li et al transformed. Thereafter, probe identifier (ID) sets were con- (10), Trigos et al (12), and Vidotto et al (13) as examples. verted into gene symbol by using the Bioconductor annota- To our knowledge, the effects and potential mechanisms of tion package mogene10sttranscriptcluster.db in the microarray exercise on melanoma tumors require further study, and there platform (15). is no consensus among researchers regarding these issues. Re- Differential expressed genes screening. To identify cently, a preclinical study by Pedersen et al (14) investigated the DEGs between sedentary and exercised samples, limma the effects of 6 wk voluntary wheel running on the melanoma package (http://www.bioconductor.org/packages/release/ tumor tissues, using B16F10, a fast-growing transplantable tu- bioc/html/limma.html) of Bioconductor was used (16), which mor model. Exercise was found to decrease melanoma tumor provided unpaired t-test to calculate the P-value for expres- incidence and growth by over 60% through the regulation of sion differences. DEGs with a corrected P-value <0.05 and NK cell mobilization and trafficking in an epinephrine- and |Log Fold Change (FC)| ≥ 0.75 were considered as the signif- IL-6–dependent manner. In the present study, we performed icant DEGs. bioinformatics analysis based on Pedersen et al.’sdataset Cluster analysis of differential expressed genes. (GSE62628) to further compare those 2 sets of melanoma tu- Cluster analysis was performed using pheatmap package in mor samples to screen for DEGs of sedentary and exercise in- R. Euclidean distance was adopted to cluster the genes and terventions. We aimed to identify the potential hub genes and produce the dendrograms. signaling pathways using a protein–protein interaction (PPI) Functional enrichment analysis of differential network to predict a potential mechanic model of exercise. expressed genes. To investigate the biological functions of DEGs, the online tool, Database for Annotation, Visualiza- METHODS tion and Integration Discovery (DAVID, http://david.abcc. Animals and exercise intervention from the origi- ncifcrf.gov/) was utilized to perform the GO (Gene Ontology) nal experiment. In the original melanoma experiment con- (http://geneontology.org/) terms enrichment analysis (17). ducted by Pedersen et al., 3-month-old female mice were Pathway analysis involved by DEGs was performed based inoculated subcutaneously with 2 Â 105 B16F10 melanoma on the Kyoto Encyclopedia of Genes and Genomes (KEGG, cells. Two weeks after injections, the mice were sacrificed http://www.kegg.jp/) database using the KOBAS web server posteuthanization, and tissues of interest were excised. The (http://kobas.cbi.pku.edu.cn/) (18). P < 0.05 was selected as melanoma tumors were dissected into fragments and incubated the cut-off.

1486 Official Journal of the American College of Sports Medicine http://www.acsm-msse.org

Copyright © 2020 by the American College of Sports Medicine. Unauthorized reproduction of this article is prohibited. Associations between the enriched GO and KEGG (20), the criteria were node P value cut-off = 0.001, false discov- pathway terms. Cytoscape software 3.5.1 (http://www. ery rate q value cut-off = 0.1, edge cut-off (similarity) = 0.5. The cytoscape.org/) was used to investigate the associations between associations between two terms were calculated according to significantly enriched GO and KEGG pathway terms (19). To vi- the overlapped DEGs, and terms shown strong associations sualize the results using Enrichmentmap Plug-in of Cytoscape were used to construct the association networks. AI SCIENCES BASIC

FIGURE 1—Volcano boxplot of DEGs between two sets of samples (A) and hierarchical clustering heatmap of the top50 DEGs (B). (A) The red points represent upregulated genes screened on the basis of |LogFC| > 0.75 and P value <0.05. The green points represent downregulation of the expression of genes screened on the basis of |LogFC| < 0.75 and P value <0.05. The black points represent genes with no significant difference; (B) Red indicates that gene expression is relatively upregulated, green indicates that gene expression is relatively downregulated, and black indicates no significant changes in gene expression.

BIOINFORMATICS ANALYSIS FOR MELANOMA TUMOR Medicine & Science in Sports & Exercise® 1487

Copyright © 2020 by the American College of Sports Medicine. Unauthorized reproduction of this article is prohibited. Interactions between differential expressed genes. RESULTS The Search Tool for the Retrieval of Interacting Genes/Proteins (STRING, http://string-db.org/) online database was used to Screened differential expressed genes. When the ’ identify the PPI between proteins encoded by DEGs (21). data set GSE62628 from Pedersen et al. s previous report Cytoscape software was used to construct the PPI network (14) was screened using the limma package, 315 DEGs were (19). The network were shown with nodes and edges. The top obtained. Among them, 294 upregulated genes and 21 down- 10 nodes were screened out by degree filter and then visualized regulated genes were identified. The differential expression using plug-in Cytohubba (22) of Cytoscape. DEGs with high of multiple genes from the two sets of sample data is shown degree will be considered as hub genes that might play crucial in Figure 1A. The cluster heatmaps of the top 50 DEGs are roles in exercise mediated regulatory processes, as degree shown in Figure 1B. means the number of nodes linked to a certain node. Function of differential expressed genes. GO func- Module analysis of interaction network. Proteins in tional enrichment of all DEGs with P <0.05andDEGcount the same module of a PPI network usually have the same bio- ≥2 were obtained using the DAVID online tool. Results of logical processes and functions by co-expression. In this GO analysis were divided into three functional groups, i.e., study, we used plug-in MCODE of Cytoscape to mine the molecular function (MF), cellular component (CC) and biolog- modules from the PPI network with score >10 (23). Thereaf- ical process (BP) (see Table, Supplemental Digital Content 1, ter, plug-in Bingo was used to annotate the functional group Details of significant GO and KEGG pathway terms of DEGs, of those genes in the modules based on the hypergeometric http://links.lww.com/MSS/B924). The top 15 enrichments distribution (corr P < 0.01, DEG count ≥2) (24), and KOBAS in each functional group are shown in Figure 2. In the BP group, web server was used to identify the KEGG pathways. genes were mainly enriched in immune and inflammatory BASIC SCIENCES

FIGURE 2—Top 15 GO enrichment significance items in different functional groups.

1488 Official Journal of the American College of Sports Medicine http://www.acsm-msse.org

Copyright © 2020 by the American College of Sports Medicine. Unauthorized reproduction of this article is prohibited. response, and positive regulation of ERK1 and ERK2 cascade. terms (histidine, beta-alanine, arginine, and proline metabolism In the CC group, genes were mainly enriched in extracellular and metabolic pathways) (Fig. 3). space, extracellular region, and the external side of the plasma Associations between enriched terms. Among the membrane. In the MF group, genes were mainly enriched in 337 terms enriched by upregulated DEGs, 103 GO terms chemokine activity, cytokine activity, and binding function. and 27 KEGG pathway terms were strongly associated with As for the upregulated DEGs, they were significantly enriched each other, and these terms were mainly related with the mem- in 292 terms, including 209 BP terms, 41 CC terms, and brane or extracellular region. Among the 13 terms enriched by 42 MF terms. DEG downregulated were significantly enriched downregulated DEGs, no GO terms or KEGG pathway terms in 9 terms, including 5 BP terms, 1 CC terms (mitochondrion) were strongly associated with each other (see Figure, Supple- and 3 MF terms (oxidoreductase, NAD, and oxidoreductase mental Digital Content 2, Functional association networks of activity). Under the same criteria, DEGs upregulated were sig- DEGs, http://links.lww.com/MSS/B925). nificantly enriched in 45 KEGG pathway terms, and downreg- Protein–protein interaction network. The PPI network ulated DEGs were significantly enriched in 4 KEGG pathway was constructed using the STRING database (the minimum AI SCIENCES BASIC

FIGURE 3—Significant pathway enrichment of upregulated (A) and downregulated (B) DEGs. Blue represents the signaling pathway, red represents upregulated genes, and green represents downregulated genes.

BIOINFORMATICS ANALYSIS FOR MELANOMA TUMOR Medicine & Science in Sports & Exercise® 1489

Copyright © 2020 by the American College of Sports Medicine. Unauthorized reproduction of this article is prohibited. BASIC SCIENCES FIGURE 4—Protein–protein interaction network. Nodes represent genes, edges represent the interaction of proteins between genes. Size of node is positively correlated with degree, and edge between two nodes is positively correlated with the combined score.

required interaction score was set at the highest confidence of HBG2, and TTK. The top 10 nodes were C3, Kng1, C3ar1, Ptafr, 0.900) and Cytoscape software, with a total of 315 DEGs, in- Fgg, Alb, Pf4, Orm1, Aldh3b1, and Apob as shown in Figure 5. cluding 294 upregulated and 21 downregulated genes. After Module analysis of interaction network. MCODE removing the disconnected nodes in the network, a complex was used to mine the densely connected modules in protein– network of DEGs was constructed, as shown in Figure 4. The protein interaction network. With a score >10, one module 17 most significant genes showing significant interaction was screened out (Fig. 6). This module contains 41 nodes, were HBB, ZWINT, WNT2B, SPP1, HBA2, NUF2, ALDH1A1, and all hub genes were included. By using BINGO to perform FZD10, MMP7, MUC16, MUC1, OMD, OGN, AOX1, ADH1B, functional annotation, 192 terms, including 155 BP terms, 7 CC

FIGURE 5—Hub genes screened out by degree filter. Nodes with different colors mean the rank of degree.

1490 Official Journal of the American College of Sports Medicine http://www.acsm-msse.org

FIGURE 6—Module diagram of interaction network. Red represents upregulated genes, and green represents downregulated genes. terms, and 30 MF terms, were screened out with the criteria genes (e.g. C3, Kng1, C3ar1, Ptafr, Fgg, Alb, Pf4, Orm1, corr P < 0.01 and DEG count ≥2. Five KEGG pathways were Aldh3b1, Apob) were screened out, and analysis of the most enriched using the DAVID online tool with the criteria significant module identified from the PPI network showed P < 0.01and DEG count ≥2 (see Table, Supplemental Digital that the complement and coagulation cascade, S. aureus infec- Content 3, Top 5 significant enrichments for each functional tion, cytokine–cytokine receptor interaction, chemokine sig- group in module function, http://links.lww.com/MSS/B926). naling pathway and phagosomes were mainly involved. Biological functional enrichment analysis showed that genes Moreover, C3, C3ar1, Kng1, Ptafr, and Fgg may be the critical in this module were significantly enriched in response to stimuli, genes in the complement and coagulation cascades pathway response to wounding, immune system process, and inflammatory and S. aureus infection pathway. These results indicate that and immune response. Complement and coagulation cascades, voluntary exercise may exert positive effects on melanoma tis- Staphylococcus aureus infection, cytokine–cytokine receptor inter- sues by regulating inflammation and the immune response. action, chemokine signaling pathway and phagosome were Because the mechanistic effects of exercise in relation to clin- enriched in the KEGG pathway analysis (see Table, Supplemen- ical treatment and tumor-specific outcomes are largely un- tal Digital Content 4, Details of significant GO and KEGG path- known, further experimental research targeting these genes way terms in module analysis, http://links.lww.com/MSS/B927). and signaling pathways are warranted to enhance the etiology and treatment of melanoma, and to reveal the potential of exercise in the perioperative period. Moreover, these investigations DISCUSSION may also lay the theoretic foundation for the application of these In this study, 294 upregulated and 21 downregulated DEGs targets in clinical translation medicine. were identified in melanoma tumor tissue from two groups of Recent preclinical exercise-oncology studies have docu- exercise and sedentary mice based on the previous data set mented similar exercise-related protection against tumor inci- GSE62628. Membrane or extracellular space, extracellular re- dence and growth (3,27–29); however, only a few studies have gion, immune system process, immune response, and response investigated the anti-inflammatory and immunomodulation ef- to stimuli were the top terms in the functional association net- fects within tumor tissues induced by exercise. For example, after works which have been suggested as contributors to mela- implantation with four T1 cells, female BALB/c mice received noma tumor progression (25,26). The top 10 candidate hub 6 wk aerobic interval training on a treadmill (5 d·wk−1) consisting

BIOINFORMATICS ANALYSIS FOR MELANOMA TUMOR Medicine & Science in Sports & Exercise® 1491

Copyright © 2020 by the American College of Sports Medicine. Unauthorized reproduction of this article is prohibited. ̇ of ten 2-min intervals of running at 70% VO2max separated by C3ar1 expression on melonoma is not clear, but Nabizadeh’s ̇ 2 min of active recovery at 50% VO2max. The results showed that group (40) showed that development and growth of B16-F0 exercise significantly decreased tumor volume, and increased the melanomas is retarded in mice lacking C3aR (complement intratumoral oncostatin M and TNF-α levels (30). In MC4L2 C3a receptor), while growth of established melanomas can tumor-bearing mice model, 6 or 14 wk of progressive treadmill be arrested by the C3aR antagonist. There are no reports about running (5 d·wk−1) led to a significant reduction in the level of the relationship between melanoma and Kng1 to date. A previ- IL-6, tumor volume, and growth, which highlight the critical role ous study hypothesized that the high molecular weight kinino- of postimplantation exercise (31). Zhang et al. also found that gen may be considered as the angiogenic contributor in 42-d moderate swimming (8 min·d−1, 5 d·wk−1)posttransplan- melanoma progression (41), thus the inhibition of it may help tation significantly reduced the relative tumor weight and lung to decrease the tumor burden. Fgg encodes the gamma subunit metastases while increasing the mean median survival time by of the fibrinogen. Animal studies demonstrate that different 9 d in Hepa1–6 tumor-bearing mice, and the proportion of components of the coagulation system such as fibrinogen CD4+ CD25+ Foxp3+ Treg cells in tumor tissues were signif- can promote tumor progression (42). As for the effects of ex- icantly decreased in exercised mice (32). Similar results can ercise, ours is the first study we are aware of that suggests ex- alsobefoundinEhrlich,inEL-4,andinHCC-LM3tumor- ercise modulates these genes within tumor tissues. Finally, it bearing mice models (33–35). Therefore, our findings fit well has been established that oxidized Ptafr agonists augment mu- with these previous studies and the hypothesis that exercise rine B16F10 melanoma tumor growth in a platelet-activating ameliorates the immune and inflammatory responses in tumors. factor receptor-dependent manner because of its effects on Current evidence from preclinical studies indicates that host immunity (43), but the relationship between exercise moderate to high-intensity aerobic exercise is superior to light and Ptafr is still not clear. exercise, when aiming to target intrinsic tumor factors (36). Exercise-induced alterations in the systemic milieu influ- However, limited consensus can be drawn from animal studies ence key regulatory mechanisms in the tumor microenviron- regarding recommendations for the optimal mode of exercise, ment, such as metabolism and immune regulation, and thus as well as volume, duration, and intensity in humans. As was considered to have a cumulative antitumorigenic effect

BASIC SCIENCES mentioned above, nearly all studies utilized a protocol with (44). Although the majority of current studies support this moderate intensity and 5 d·wk−1 frequency for tumor- tumor-suppressing effect of exercise, we admit that a consider- bearing animals, with mixed results. Moreover, 80% of able publication bias may exist favoring positive findings. maximal workload swimming for 1 h·d−1,5d·wk−1,over And, little consensus can currently be drawn from these stud- 6 wk did not reduce the weight and volume of tumors in ies regarding recommendations for the optimal volume, inten- Ehrlich tumor-bearing mice, whereas 50% maximal work- sity, and duration of aerobic exercise. To further study the load reduced tumor volume (33). A recent study investigat- mechanism of exercise-dependent control of melanoma tumor ing the effects of 9 wk of moderate (8 min·d−1) and overload tissue, we systematically re-analyzed the gene expression pro- (16 and 32 min·d−1)swimmingonHepa1–6 tumor initiation file. However, there are still two major limitations in the pres- and progression reported similar results (35). Therefore, it ent study that could be addressed in future research. First, the is possible that voluminous high intensity exercise may sample size for gene profile analysis was small which might compromise the inflammatory and immunomodulation ef- lead to a certain rate of false positive results despite our at- fects of exercise that benefit cancer outcomes. As for the tempts to control for this with standardized analyzing pro- present study, voluntary wheel running was adopted in the cesses and strict screening thresholds. Second, it is difficult original experiment, intensity was not controlled, and only to draw a clear conclusion concerning the detailed effects of the average running distance was reported (14), thus, it is the intervention because of the design of the original experi- difficult to draw a definite conclusion regarding the exercise ment. Only running distance was reported in the melanoma intensity, dose, and duration. However, taking the afore- experiment, thus while we can discuss exercise in general, mentioned clues and the susceptibility of tumor-bearing an- we cannot further distinguish how exercise volume, intensity, imals into consideration, a programmed aerobic exercise and duration may affect these candidate genes and pathways protocol with moderate intensity and volumes, and higher in tumor tissues. exercise frequency (5 d·wk−1), seems effective at reducing In our humble opinion, further studies are warranted to in- tumorgrowthinanimals. vestigate the associations between exercise and melanoma Among the critical genes screened in the current study, C3, after detailed adjustment for sun exposure and ultraviolet C3ar1, Kng1, and Fgg were enriched in complement and co- radiation-related skin damage. The biological mechanisms agulation cascades, which have been identified in multiple by which exercise affects melanoma cancer processes require cancers, suggesting an essential role of this pathway in cancer further investigation, since the associations between exercise biology (37). Complement protein C3 had previously been and melanoma (particularly the intratumoral alterations) are suggested as a prognostic factor of lung cancer (38). Jandl et al critical to propose recommendations, and the current mecha- (39) found that radioimmunotherapy leads to increased C3 de- nisms through which this intervention operates (including the position in melanoma tumor, and activates the host immune mode of exercise, as well as the volume, intensity, and dura- system, leading to a suppression of the tumor. The effects of tion) are less clearly elucidated. Moreover, more efforts are

1492 Official Journal of the American College of Sports Medicine http://www.acsm-msse.org

Copyright © 2020 by the American College of Sports Medicine. Unauthorized reproduction of this article is prohibited. needed to explore the impact of exercise throughout the complement and coagulation cascades pathway and S. aureus course of life on melanoma risk. infection pathway. These findings indicate that exercise ame- liorated the immune response and inflammatory response in CONCLUSIONS melanoma tissue. The objective of the present study was to improve our understanding of the molecular mechanisms underlying the This study was supported by grants from the National Natural Sci- regulatory effects of exercise on melanoma tissue through ence Foundation of China (31960192, 31900842), the Natural Science Foundation of Jiangxi Province (20192BAB205081), the Foundation for bioinformatics analysis. We identified 294 upregulated genes Science and Technology in Jiangxi Provincial Department of Education and 21 downregulated genes in the exercise group and 10 can- (GJJ180560), and the Humanities and Social Sciences of Higher Insti- didate hub genes including C3, Kng1, C3ar1, Ptafr, Fgg, Alb, tutions in Jiangxi Province (TY17210). Results of the present study do not constitute endorsement by the American College of Sports Medicine. Pf4, Orm1, Aldh3b1, and Apob for exercise were predicted. The results of the study are presented clearly, honestly, and without fab- C3, C3ar1, Kng1, Ptafr, Fgg may be the critical genes in the rication, falsification, or inappropriate data manipulation.

REFERENCES 1. Chen W, Zheng R, Baade PD, et al. Cancer statistics in China, 2015. 16. Ritchie ME, Phipson B, Wu D, et al. Limma powers differential

CA Cancer J Clin. 2016;66(2):115–32. expression analyses for RNA-sequencing and microarray studies. SCIENCES BASIC 2. Kushi LH, Doyle C, McCullough M, et al. American Cancer Society Nucleic Acids Res. 2015;43(7):e47. Guidelines on nutrition and physical activity for cancer prevention: 17. Huang da W, Sherman BT, Lempicki RA. Systematic and integrative reducing the risk of cancer with healthy food choices and physical ac- analysis of large gene lists using DAVID bioinformatics resources. tivity. CA Cancer J Clin. 2012;62(1):30–67. Nat Protoc. 2009;4(1):44–57. 3. Warner AB, McQuade JL. Modifiable host factors in melanoma: 18. Kanehisa M, Goto S. KEGG: Kyoto encyclopedia of genes and ge- emerging evidence for obesity, diet, exercise, and the microbiome. nomes. Nucleic Acids Res. 2000;28(1):27–30. Curr Oncol Rep. 2019;21(8):72. 19. Shannon P, Markiel A, Ozier O, et al. Cytoscape: a software environ- 4. Zhang X, Ashcraft KA, Betof Warner A, Nair SK, Dewhirst MW. ment for integrated models of biomolecular interaction networks. Can exercise-induced modulation of the tumor physiologic microen- Genome Res. 2003;13(11):2498–504. vironment improve antitumor immunity? Cancer Res. 2019;79(10): 20. Merico D, Isserlin R, Stueker O, Emili A, Bader GD. Enrichment 2447–56. map: a network-based method for gene-set enrichment visualization 5. Alejo LB, Pagola-Aldazabal I, Fiuza-Luces C, et al. Exercise and interpretation. PLoS One. 2010;5(11):e13984. prehabilitation program for patients under neoadjuvant treatment for 21. Szklarczyk D, Gable AL, Lyon D, et al. STRING v11: protein– rectal cancer: a pilot study. J Cancer Res Ther. 2019;15(1):20–5. protein association networks with increased coverage, supporting 6. Clinton SK, Giovannucci EL, Hursting SD. The World Cancer Re- functional discovery in genome-wide experimental datasets. Nucleic search Fund/American Institute for Cancer Research third expert report Acids Res. 2019;47(D1):D607–13. on diet, nutrition, physical activity, and cancer: impact and future direc- 22. Chin CH, Chen SH, Wu HH, Ho CW, Ko MT, Lin CY. cytoHubba: tions. JNutr. 2020;150(4):663–71. identifying hub objects and sub-networks from complex interactome. 7. Schadendorf D, van Akkooi ACJ, Berking C, et al. Melanoma. BMC Syst Biol. 2014;8(4 Suppl):S11. Lancet. 2018;392(10151):971–84. 23. Wu X, Guo S, Shen G, et al. Screening of osteoprotegerin-related fea- 8. Moore SC, Lee IM, Weiderpass E, et al. Association of leisure-time ture genes in osteoporosis and functional analysis with DNA micro- physical activity with risk of 26 types of cancer in 1.44 million adults. array. Eur J Med Res. 2013;18(1):15. JAMA Intern Med. 2016;176(6):816–25. 24. Maere S, Heymans K, Kuiper M. BiNGO: a Cytoscape plugin to as- 9. Shors AR, Solomon C, McTiernan A, White E. Melanoma risk in re- sess overrepresentation of gene ontology categories in biological net- lation to height, weight, and exercise (United States). Cancer Causes works. Bioinformatics. 2005;21(16):3448–9. Control. 2001;12(7):599–606. 25. Gajewski TF. Failure at the effector phase: immune barriers at the 10. Li L, Lei Q, Zhang S, Kong L, Qin B. Screening and identification of level of the melanoma tumor microenvironment. Clin Cancer Res. key biomarkers in hepatocellular carcinoma: evidence from bioinfor- 2007;13(18 Pt 1):5256–61. matic analysis. Oncol Rep. 2017;38(5):2607–18. 26. Prins RM, Vo DD, Khan-Farooqi H, et al. NK and CD4 cells col- 11. Wang H, Lee JH, Tian Y. Critical genes in white adipose tissue based laborate to protect against melanoma tumor formation in the brain. on gene expression profile following exercise. Int J Sports Med. J Immunol. 2006;177(12):8448–55. 2019;40(1):57–61. 27. Isanejad A, Alizadeh AM, Amani Shalamzari S, et al. MicroRNA-206, 12. Trigos AS, Pearson RB, Papenfuss AT, Goode DL. Altered interac- let-7a and microRNA-21 pathways involved in the anti-angiogenesis tions between unicellular and multicellular genes drive hallmarks of effects of the interval exercise training and hormone therapy in breast transformation in a diverse range of solid tumors. Proc Natl Acad cancer. Life Sci. 2016;151:30–40. Sci U S A. 2017;114(24):6406–11. 28. Higgins KA, Park D, Lee GY, Curran WJ, Deng X. Exercise-induced 13. Vidotto T, Nersesian S, Graham C, Siemens DR, Koti M. DNA dam- lung cancer regression: mechanistic findings from a mouse model. age repair gene mutations and their association with tumor immune Cancer. 2014;120(21):3302–10. regulatory gene expression in muscle invasive bladder cancer sub- 29. Ruiz-Casado A, Martin-Ruiz A, Perez LM, Provencio M, Fiuza-Luces types. J Immunother Cancer. 2019;7(1):148. C, Lucia A. Exercise and the hallmarks of cancer. Trends Cancer. 14. Pedersen L, Idorn M, Olofsson GH, et al. Voluntary running sup- 2017;3(6):423–41. presses tumor growth through epinephrine- and IL-6-dependent NK 30. Molanouri Shamsi M, Chekachak S, Soudi S, et al. Effects of cell mobilization and redistribution. Cell Metab. 2016;23(3):554–62. exercise training and supplementation with selenium nanopar- 15. Meisig J, Bluthgen N. The gene regulatory network of mESC dif- ticle on T-helper 1 and 2 and cytokine levels in tumor tissue ferentiation: a benchmark for reverse engineering methods. Philos of mice bearing the 4 T1 mammary carcinoma. Nutrition. 2019; Trans R Soc Lond B Biol Sci. 2018;373(1750):20170222. 57:141–7.

BIOINFORMATICS ANALYSIS FOR MELANOMA TUMOR Medicine & Science in Sports & Exercise® 1493

Copyright © 2020 by the American College of Sports Medicine. Unauthorized reproduction of this article is prohibited. 31. Shalamzari SA, Agha-Alinejad H, Alizadeh S, et al. The effect of ex- 38. Lin K, He S, He L, et al. Complement component 3 is a prognostic ercise training on the level of tissue IL-6 and vascular endothelial factor of nonsmall cell lung cancer. Mol Med Rep. 2014;10(2):811–7. growth factor in breast cancer bearing mice. Iran J Basic Med Sci. 39. Jandl T, Revskaya E, Jiang Z, Bryan RA, Casadevall A, Dadachova E. 2014;17(4):231–58. Complement-dependent cytotoxicity of an antibody to melanin in 32. Zhang QB, Meng XT, Jia QA, et al. Herbal compound Songyou yin and radioimmunotherapy of metastatic melanoma. Immunotherapy. 2013; Moderate swimming suppress growth and metastasis of liver cancer by 5(4):357–64. enhancing immune function. Integr Cancer Ther. 2016;15(3):368–75. 40. Nabizadeh JA, Manthey HD, Steyn FJ, et al. The complement C3a 33. Almeida PW, Gomes-Filho A, Ferreira AJ, et al. Swim training sup- receptor contributes to melanoma tumorigenesis by inhibiting presses tumor growth in mice. J Appl Physiol (1985). 2009;107(1): neutrophil and CD4+ T cell responses. J Immunol. 2016;196(11): 261–5. 4783–92. 34. Zielinski MR, Muenchow M, Wallig MA, Horn PL, Woods JA. Ex- 41. Khan ST, Pixley RA, Liu Y, et al. Inhibition of metastasis of syngeneic ercise delays allogeneic tumor growth and reduces intratumoral in- murine melanoma in vivo and vasculogenesis in vitro by monoclonal flammation and vascularization. J Appl Physiol (1985). 2004;96(6): antibody C11C1 targeted to domain 5 of high molecular weight kinin- 2249–56. ogen. Cancer Immunol Immunother. 2010;59(12):1885–93. 35. Zhang QB, Zhang BH, Zhang KZ, et al. Moderate swimming sup- 42. Palumbo JS, Kombrinck KW, Drew AF, et al. Fibrinogen is an im- pressed the growth and metastasis of the transplanted liver cancer in portant determinant of the metastatic potential of circulating tumor mice model: with reference to nervous system. Oncogene. 2016; cells. Blood. 2000;96(10):3302–9. 35(31):4122–31. 43. Hackler PC, Reuss S, Konger RL, Travers JB, Sahu RP. Systemic 36. Hojman P, Gehl J, Christensen JF, Pedersen BK. Molecular mech- platelet-activating factor receptor activation augments experimen- anisms linking exercise to cancer prevention and treatment. Cell tal lung tumor growth and metastasis. Cancer Growth Metastasis. Metab. 2018;27(1):10–21. 2014;7:27–32. 37. Krupp M, Maass T, Marquardt JU, et al. The functional cancer map: a 44. Koelwyn GJ, Quail DF, Zhang X, White RM, Jones LW. Exercise- systems-level synopsis of genetic deregulation in cancer. BMC Med dependent regulation of the tumour microenvironment. Nat Rev Genomics. 2011;4:53. Cancer. 2017;17(10):620–32. BASIC SCIENCES

1494 Official Journal of the American College of Sports Medicine http://www.acsm-msse.org