Coordinated regulation of hepatic and adipose tissue transcriptomes by the oral administration of an amino acid mixture simulating the larval saliva of Vespa species
© The Author(s) 2016
Received: 20 November 2015
Accepted: 15 June 2016
Published: 11 July 2016
VAAM is an amino acid mixture that simulates the composition of Vespa larval saliva. VAAM enhanced physical endurance of mice and have been used by athletes as a supplementary drink before exercise. However, there is no information on the effect of VAAM on the physiology of freely moving animals. The purpose of this study was to obtain information about the VAAM-dependent regulation of liver and adipose tissue transcriptomes.
Mice were orally fed a VAAM solution, an amino acid mixture mimicking casein hydrolysate (CAAM) or water under ad libitum feeding conditions for 5 days. Comparisons of the hepatic transcriptome between VAAM-, CAAM-, and water-treated groups revealed a VAAM-specific regulation of the metabolic pathway, i.e., the down-regulation of glycolysis and fatty acid oxidation and the up-regulation of polyunsaturated fatty acid synthesis and glucogenic amino acid utilization. Similar transcriptomic analyses of white and brown adipose tissues (WAT and BAT, respectively) indicated the up-regulation of phospholipid synthesis in WAT and the negative regulation of cellular processes in BAT. Because the coordinated regulation of tissue transcriptomes implied the presence of upstream signaling common to these tissues, we conducted an Ingenuity Pathways Analysis. This analysis showed that estrogenic and glucagon signals were activated in the liver and WAT and that beta-adrenergic signaling was activated in all three tissues.
We found that VAAM ingestion had an effect on multiple tissue transcriptomes of freely moving mice. Utilization of glycogenic amino acids may have been activated in the liver. Fatty acid conversion into phospholipid, not to triacylglycerol, may have been stimulated in adipocytes contrasting that a little effect was observed in BAT. Analysis of upstream factors revealed that multiple hormonal signals were activated in the liver, WAT, and BAT. Our data provide some clues to understanding the role of VAAM in metabolic regulation.
The amino acid composition of food has a large impact on the metabolic homeostasis of individuals, especially when they are supplied as a major nutrient. As clearly understandable by the polyribosome model of protein synthesis, a shortage of the first-limiting amino acids readily affects the supply of proteins as the source of enzymes and other tissue-constructing materials. In this regard, essential amino acids (EAAs) are critical for the growth of the animal owing to their inability to synthesize EAAs from an organic backbone and nitrogen donor. In addition, many amino acids possess organic acid moieties that can serve as substrates for energy production (glucogenic and ketogenic amino acids).
In addition to these nutritive functions of amino acids, some dietary amino acids play roles as extracellular signals. Branched chain amino acids (BCAAs) constitute approximately 50 % of dietary EAAs and are known to activate protein anabolism via the mammalian target of rapamycin (mTOR)-dependent pathway [1, 2]. Glycine, glutamate, phenylalanine, histidine, and tryptophan act as or are metabolized to neurotransmitters and thus are able to exert a modulatory effect on the nervous system . Portions of gastrointestinal epithelial cells express l-amino acid-sensing taste receptors and stimulate gastrointestinal hormone secretion in response to phenylalanine, leucine, glutamate, and tryptophan administration [4, 5]. These non-nutritive functions are attributable to a small number of amino acids but not to their whole constituent amino acids.
We have investigated the regulatory role of an amino acid mixture that simulates Vespa species (VAAM) on motor performance in animals at the dosing condition of sub-nutritional level (0.38 % of total food-derived energy/day). VAAM consists of 17 amino acids, among which glycine, tryptophan, and proline are present at an approximately twofold higher molar ratio than diet-derived amino acids (Additional file 1: Table S1). We first searched the optimum condition of VAAM administration to achieve significant enhancement of exercise performance . In one study, a preceding oral administration of VAAM at the dose of 0.67 g/kg body weight improved the swimming performance of mice; this improvement was not observed with an amino acid mixture simulating the amino acid composition of bovine casein (CAAM) . In addition, VAAM can lower serum lactate levels and activate fatty acid release into mice sera and into the media of cultured rat adipocytes [6, 8]. VAAM solution, as a sport supplement drink, has been in product line since 1995 and consumed by many athletes. In the human studies, the preceding administration of VAAM at the dose of one eighth of our animal studies significantly lowered the respiratory quotient during the 60 min of cycling exercise , and the 12 weeks of VAAM administration at the same dose as above increased maximal oxygen uptake in the older women groups . These results suggest that ingested VAAM may affect the metabolic state of multiple tissues, such as liver and adipose tissues, to facilitate energy supply to the locomotor system. However, there is no information on the effect of VAAM on the physiology of resting or freely moving animals. Here we present data indicating this type of regulation in multiple tissues at the gene transcriptional level.
Characterization of genes regulated by VAAM administration in murine liver
Regulation of liver metabolism by VAAM
Response of adipose tissue transcriptomes to VAAM
Search for upstream regulators common among liver and adipose tissues
In this study, we examined the non-nutritive effect of a specific amino acid mixture, VAAM, on the liver, WAT, and BAT transcriptomes in mice under freely moving conditions. From the gene ontology (GO) analysis of VAAM-regulated genes, it was predicted that VAAM may up-regulate sugar and lipid anabolism in these tissues. In addition, a comparative analysis of IPA upstream factors among these tissues suggested that estrogenic and glucagon signals were activated in the liver and WAT and that beta-adrenergic signaling was activated in the three tissues.
Our previous studies [6, 7] on VAAM were conducted using a mouse model under forced exercise conditions, where the glycogen storage might have been exhausted and the muscle might have continued to move with the help of a fatty acid supply from the adipose tissues. The major difference between the previous studies and the present study resides in whether or not moving conditions were present (forced swimming or free motion) and in the dosing condition (one shot or five shots). However, there are some correlations in the metabolic regulation among these studies. We observed a down-regulation of beta-oxidation and an up-regulation of glucogenic amino acid utilization in the liver (Fig. 3). The former may save the fatty acid pool in total and the latter makes it possible to supply amino acid-derived substrates to gluconeogenesis in response to exercise stimulus [11–13]. In the case of WAT, four enzyme genes for acylglycerol synthesis were regulated in a distinct manner: Mogat1 and Dgat1 were up-regulated, and Mogat2 and Dgat2 were down-regulated. Especially, the opposite regulation of Dgat1 (up) and 2 (down) should be mentioned because Dgat1 facilitates fatty acid/acylglycerol shuttling, and Dgat2 is responsible for triacylglycerol accumulation . This regulation may support fatty acid release to the serum. It is also possible that the up-regulation of Slc37a4 may decrease lactate production by facilitating glucose regeneration from glucose-6-phosphate (Fig. 3e). The role of BAT in exercise performance is not clear because our experiments were performed at a moderate temperature, where no additional thermogenesis is required. However, it should be noted that the two catecholamine responsive genes, deiodinase, iodothyronine, type II (Dio2) and peroxisome proliferator-activated receptor gamma, coactivator 1 alpha (Ppargc1a, also known as Pgc1a), are up-regulated. These genes are reported to be induced in mice and rats exposed to cold temperature [15, 16]. In the moderate temperature condition as those we adopted, this kind of regulation may result in an increased basal metabolic rate under resting conditions [12, 17, 18]. In conclusion, the VAAM-dependent regulation of multiple tissue transcriptomes seems to be a tuning mechanism that makes the metabolic state resistant to more intense exercise. It would be valuable to examine the effect of middle term (approximately 1 week) pre-administration of VAAM on exercise performance in a human interactive study.
Whether VAAM acts on these tissues directly or indirectly is an open question. From the IPA upstream factor analysis, it was predicted that the three endocrine factors adrenaline, estrogen, and glucagon seemed to stimulate these tissues. In particular, adrenaline is the most plausible factor involved in VAAM signaling because several IPA terms related to adrenergic signals were detected in the three tissues (Table 1) and its increase in the mice sera was detected in a previous study . Considering that a major source of adrenaline is the hypothalamus-sympathetic nerve-adrenal medullary (HSA) axis, we think it is highly possible that VAAM controls terminal tissues via this system. This assumption raises another question as to what type of mechanism transduces VAAM signals in the HSA axis. There are two supporting information. First is that VAAM’s effect depends on the overall ratio of amino acids (Additional file 1: Table S1), and second, the maximal effect of VAAM was obtained with the pretreatment of animals 30 min before the exercise. These considerations suggest that the sensing mechanism for VAAM might have a broad spectrum to amino acids and that the mechanism may reside in the gastrointestinal tract that faces the VAAM solution but not in metabolizing tissues nor locomotor systems. Accordingly, we assume that gastrointestinal chemosensory cells are one of the candidate targets of VAAM stimulus. Further investigation is needed to elucidate the VAAM-sensing and VAAM-transducing machinery.
We found that VAAM ingestion had an effect on multiple tissue transcriptomes of freely moving mice. Utilization of glucogenic amino acids may have been activated in the liver. Fatty acid conversion into phospholipid, not to triacylglycerol, may have been stimulated in adipocytes contrasting that a little effect was observed in BAT. Analysis of upstream factors revealed that multiple hormonal signals were activated in the liver, WAT, and BAT. Our data provide some clues to understanding the role of VAAM in metabolic regulation.
Five-week-old male ddY mice (SLC Japan, Shizuoka, Japan) were housed in plastic cages with food (MF, Oriental yeast, Tokyo, Japan) and deionized water ad libitum in a temperature- and humidity-controlled room with a 12-h light/dark cycle (light 08:00–20:00; dark 20:00–08:00). The mice were weighed every day. All animal experimental protocols were approved by the Animal Use Committee of the Faculty of Agriculture at the University of Tokyo with approval number: P12-669.
Glycine and all l-amino acids were purchased from Wako Pure Chemical Industries (Osaka, Japan). The amino acid mixtures VAAM and CAAM were prepared in the ratios described by Abe et al.  and were analyzed for amino acid composition using a Hitachi l-8500A amino acid analyzer (Hitachi High Technologies, Tokyo) (Additional file 1: Table S1). The concentrations of VAAM and CAAM were each adjusted to 1.8 %.
The experimental schedule is shown in Additional file 2: Figure S1. The mice (n = 21) were acclimated to a laboratory environment for 1 week. They were divided randomly into three treatment groups: VAAM, CAAM, or water. Using a feeding tube, 1.8 % VAAM, 1.8 % CAAM, or water was orally administered five times once a day at 10:00AM. The dosage of VAAM, CAAM, or water was adjusted to 37.5 μL per gram of body weight as described previously . On the day of last administration, the food was removed and the mice were moved to clean cages at 8:00. The last treatments began at 10:00. At 4 h after the last administration, the mice were euthanized by cervical dislocation and the blood, liver, WAT, and BAT were collected. The blood was separated to obtain the serum. Small hepatic pieces were immersed into RNAlater (Qiagen, Tokyo, Japan). The WAT and BAT were immediately frozen after extraction using liquid nitrogen. All samples were maintained at −80 °C until use. Two mice in CAAM group were excluded because of their morphological abnormality in the liver. Subsequently, the mice number of each group became n = 7 for the VAAM group, n = 5 for the CAAM group, and n = 7 for the water group.
Measurement of serum biochemical parameters
Serum glucose, non-esterified fatty acid (NEFA), and triacylglycerol were enzymatically assayed using the glucose C II test Wako, the NEFA C test Wako, and the triglyceride E test Wako (Wako Pure Chemical Industries, Osaka, Japan), respectively. Total lipids, total ketone bodies, total cholesterol, HDL cholesterol and LDL cholesterol were assayed on Nagahama Life Science (Shiga, Japan). Each of the values is represented as the mean ± standard error of the mean (SEM). Significant differences (p ≤ 0.05) between the experimental groups were assessed by Tukey-Kramer comparison.
DNA microarray assay
Total RNA was isolated from each hepatic, WAT, and BAT sample by TRIzol reagent (Invitrogen Japan, Tokyo, Japan) and purified using an RNeasy mini kit (Qiagen, Tokyo, Japan). The RNA Integrity Number (RIN) was estimated as an index of the quality of the total RNA using an Agilent 2100 Bioanalyzer (Agilent Technologies Japan, Tokyo, Japan). The values for RIN were over 8.3.
Then, amplified RNA (aRNA) was synthesized from 200 ng of purified total RNA, and biotinylated aRNA was obtained using a GeneChip® 3′IVT Express Kit (Affymetrix, Santa Clara, CA, USA). The aRNA was fragmented and hybridized to a GeneChip® Mouse Genome 430 2.0 Array (Affymetrix) for 16 h at 45 °C. The arrays were washed and stained with phycoerythrin using the GeneChip® Fluidics Station 450 (Affymetrix). The arrays were submitted to scanning on an Affymetrix GeneChip® Scanner 3000 7G (Affymetrix). The Affymetrix® GeneChip® Command Console® Software (Affymetrix) was used to make CEL files.
DNA microarray data analysis
The CEL files were quantified by the distribution free weighted method (DFW)  using the statistical language R (2.7.1) (http://www.r-project.org/)  and Bioconductor (2.2) (http://www.bioconductor.org/) . Hierarchical clustering was performed using the pvclust() function in R . The rank products (RP) method was used to identify differentially expressed gene probe sets of the DFW-quantified data . The probe sets with a false discovery rate (FDR) <0.05 were considered to be differentially expressed between two groups. RP was performed for a comparison of the water treatment group and the VAAM or CAAM treatment groups.
Gene ontology analysis
The up- and down-regulated probe sets picked out at FDR <0.05 were functionally classified by the biological process in gene ontology (GO) with the Functional Annotation Tool of the Database for Annotation, Visualization, and Integrated Discovery (DAVID) [24, 25] and Quick GO (http://www.ebi.ac.uk/QuickGO/) . EASE scores, which are modified Fisher’s exact test p values, were used to extract statistically overrepresented GO terms. GO terms with FDR-corrected p values <0.01 were regarded as significantly enriched.
Upstream regulator analysis
Predicted upstream regulators from VAAM- and CAAM-specific data sets were analyzed using QIAGEN’s Ingenuity® Pathway Analysis (IPA®, QIAGEN Redwood City, www.qiagen.com/ingenuity). The activation Z-score was calculated as a measure of upstream regulator analysis. An absolute Z-score ≥2 was judged as significantly activated or inhibited, while an absolute Z-score ≤1.0 could not judge a direction. Common upstream regulators that were predicted to be activated or inhibited in the liver, WAT, and BAT were selected from the all upstream regulators list.
Quantitative PCR analysis
Total RNA was prepared as described in “DNA microarray assay” section. Complementary DNA (cDNA) was synthesized with PrimeScript RT reagent kit (TaKaRa, Shiga, Japan) and subjected to PCR analysis using the following primer sets and SYBR Premix Ex Taq II kit (TaKaRa). The cDNA was amplified by 30 s at 95 °C followed by 40 cycles of incubation at 95 °C for 15 s and at 60 °C for 30 s in CFX96 real-time PCR detection system (Bio-Rad). The primers sets were as follows: Elovl3 forward; 5′-TTCTCACGCGGGTTAAAAATGG-3′, Elovl3 reverse; 5′- GAGCAACAGATAGACGACCAC-3′, G6pc forward; 5′-TTCAAGTGGATTCTGTTTGG-3′, G6pc reverse; 5′- AGATAGCAAGAGTAGAAGTGAC-3′, Pecr forward; 5′-GAAGGGATGGCATGCTGTGA-3′, Pecr reverse; 5′- TTGACAATCGACCCTCCGTG-3′, beta-actin forward; 5′-GATGTATGAAGGCTTTGGTC-3′, beta-actin reverse; 5′- TGTGCACTTTTATTGGTCTC-3′. Each cycle threshold (CT) value of an amplified fragment was compared with that of beta-actin and represented as a relative expression level. Values for VAAM or CAAM treatment group and water treatment group were examined for their significant difference using Dunnett’s test.
BAT, brown adipose tissue; CAAM, casein amino acid mixture; EAAs, essential amino acids; NEFA, non-esterified fatty acid; VAAM, Vespa amino acid mixture; WAT, white adipose tissue
We thank Hiromasa Morishita for technical assistance in amino acid analysis. A part of this study was supported by KYOWA HAKKO BIO CO., LTD. This work was supported by the Council for Science, Technology and Innovation (CSTI), Cross-ministerial Strategic Innovation Promotion Program (SIP), “Technologies for creating next-generation agriculture, forestry and fisheries”, and JSPS KAKENHI Grant number 26350177.
Availability of data and materials
The data sets supporting the results of this article are available in the National Center of Biotechnology Information (NCBI) Gene Expression Omnibus (http://www.ncbi.nlm.nih.gov/geo/) and are accessible through accession number GSE 73266.
FS, TA, AK, and KA conceived and designed the experiment. KKu provided the technique and materials for VAAM preparation. FS, TA, AK, YW, and KS performed animal experiment. FS, AK, AY, KKo, SA, and TK analyzed the data. FS, AY, TK, and KA wrote the paper. All authors read and approved the final manuscript.
Author Kumagai K is an employee of KYOWA HAKKO BIO CO., LTD. He provided the technique and materials for VAAM preparation, but the company had no control over the interpretation, the writing, or the publication of this work. Other authors declare that they have no competing interests.
Ethics approval and consent to participate
All animal experimental protocols were approved by the Animal Use Committee of the Faculty of Agriculture at the University of Tokyo with approval number: P12-669.
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.
- Wu G. Functional amino acids in growth, reproduction, and health. Adv Nutr. 2010;1:31–7.View ArticlePubMedPubMed CentralGoogle Scholar
- Mattick JSA, Kamisoglu K, Ierapetritoua MG, Androulakisa IP, Berthiaume F. Branched chain amino acid supplementation impact on signaling and relevance to critical illness. Wiley Interdiscip Rev Syst Biol Med. 2013;5:449–60.View ArticlePubMedPubMed CentralGoogle Scholar
- Wu G. Amino acids metabolism, functions, and nutrition. Amino Acids. 2009;37:1–17.View ArticlePubMedGoogle Scholar
- Wang JH, Inoue T, Higashiyama M, Guth PH, Engel E, Kaunitz JD, Akiba Y. Umami receptor activation increases duodenal bicarbonate secretion via glucagon-like peptide-2 release in rats. J Pharmacol Exp Ther. 2011;339:464–73.View ArticlePubMedPubMed CentralGoogle Scholar
- Shirazi-Beechey SP, Daly K, Al-Rammahi M, Moran AW, Bravo D. Role of nutrient-sensing taste 1 receptor (T1R) family members in gastrointestinal chemosensing. Br J Nutr. 2014;111:S8–15.View ArticlePubMedGoogle Scholar
- Abe T, Takiguchi Y, Tamura M, Shimura J, Yamazaki K. Effects of Vespa amino acid mixture isolated from hornet larval saliva exercise in swimming mice of blood lactate and glucose and modified VAAM nutrients on endurance. Jpn J Phys Fitness Sports Med. 1995;44:225–38.View ArticleGoogle Scholar
- Abe T, Inamori M, Iida K, Tamura M, Takiguchi Y, Yasuda K. The activation of fatty acid metabolism by VAAM and related nutrients during endurance exercise in mice. Adv Exerc sports physiol. 1997;13:35–44.Google Scholar
- Shinozaki F, Abe T. Synergistic effect of Vespa amino acid mixture on lipolysis in rat adipocytes. Biosci Biotechnol Biochem. 2008;72:1860–8.View ArticlePubMedGoogle Scholar
- Demura S, Nagasawa Y, Kitabayashi T, Matsuzawa J. Effect of amino acid mixture intake on physiological response and rating of perceived exertion during cycling exercise. Percept Mot Skills. 2003;96:883–95.View ArticlePubMedGoogle Scholar
- Sasai H, Matsuo T, Fujita M, Saito M, Tanaka K. Effects of regular exercise combined with ingestion of vespa amino acid mixture on aerobic fitness and cardiovascular diseases risk factors in sedentary older woman: A preliminary study. Geriatr Gerontol Int. 2011;11:24–31.View ArticlePubMedGoogle Scholar
- Emhoff CA, Messonnier LA, Horning MA, Fattor JA, Carlson TJ, Brooks GA. Gluconeogenesis and hepatic glycogenolysis during exercise at the lactate threshold. J Appl Physiol (1985). 2013;114:297–306.View ArticleGoogle Scholar
- Haase TN, Ringholm S, Leick L, Biensø RS, Kiilerich K, Johansen S, Nielsen MM, Wojtaszewski JF, Hidalgo J, Pedersen PA, Pilegaard H. Role of PGC-1a in exercise and fasting-induced adaptations in mouse liver. Am J Physiol Regul Integr Comp Physiol. 2011;301:R1501–9.View ArticlePubMedGoogle Scholar
- Suh SH, Paik IY, Jacobs K. Regulation of blood glucose homeostasis during prolonged exercise. Mol Cells. 2007;23:272–9.PubMedGoogle Scholar
- Yen CL, Stone SJ, Koliwad S, Harris C, Farese Jr RV. Thematic review series: glycerolipids. DGAT enzymes and triacylglycerol biosynthesis. J Lipid Res. 2008;49:2283–301.View ArticlePubMedPubMed CentralGoogle Scholar
- Watanabe M, Yamamoto T, Kakuhata R, Okada N, Kajimoto K, Yamazaki N, Kataoka M, Baba Y, Tamaki T, Shinohara Y. Synchronized changes in transcript levels of genes activating cold exposure-induced thermogenesis in brown adipose tissue of experimental animals. Biochim Biophys Acta. 2008;1777:104–12.View ArticlePubMedGoogle Scholar
- Hondares E, Rosell M, Díaz-Delfín J, Olmos Y, Monsalve M, Iglesias R, Villarroya F, Giralt M. Peroxisome proliferator-activated receptor (PPARα) induces PPARγ coactivator 1 (PGC-1α) gene expression and contributes to thermogenic activation of brown fat: involvement of PRDM16. J Biol Chem. 2011;286:43112–22.View ArticlePubMedPubMed CentralGoogle Scholar
- Hall JA, Ribich S, Christoffolete MA, Simovic G, Correa-Medina M, Patti ME, Bianco AC. Absence of thyroid hormone activation during development underlies a permanent defect in adaptive thermogenesis. Endocrinology. 2010;151:4573–82.View ArticlePubMedPubMed CentralGoogle Scholar
- Betz MJ, Enerbäck S. Therapeutic prospects of metabolically active brown adipose tissue in humans. Front Endocrinol (Lausanne). 2011;29:86.Google Scholar
- Chen Z, McGee M, Liu Q, Scheuermann RH. A distribution free summarization method for Affymetrix GeneChip arrays. Bioinformatics. 2007;23:321–7.View ArticlePubMedGoogle Scholar
- R Development Core Team. R: A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing; 2006. http://www.R-project.org.
- Gentleman RC, Carey VJ, Bates DM, Bolstad B, Dettling M, Dudoit S, Ellis B, Gautier L, Ge Y, Gentry J, Hornik K, Hothorn T, Huber W, Iacus S, Irizarry R, Leisch F, Li C, Maechler M, Rossini AJ, Sawitzki G, Smith C, Smyth G, Tierney L, Yang JY, Zhang J. Bioconductor: open software development for computational biology and bioinformatics. Genome Biol. 2004;5:R80.View ArticlePubMedPubMed CentralGoogle Scholar
- Suzuki R, Shimodaira H. Pvclust: an R package for assessing the uncertainty in hierarchical clustering. Bioinformatics. 2006;22:1540–2.View ArticlePubMedGoogle Scholar
- Breitling R, Armengaud P, Amtmann A, Herzyk P. Rank products: a simple, yet powerful, new method to detect differentially regulated genes in replicated microarray experiments. FEBS Lett. 2004;573:83–92.View ArticlePubMedGoogle Scholar
- Dennis Jr G, Sherman BT, Hosack DA, Yang J, Gao W, Lane HC, Lempicki RA. DAVID: database for annotation, visualization, and integrated discovery. Genome Biol. 2003;4:P3.View ArticlePubMedGoogle Scholar
- Huang DW, Sherman BT, Zheng X, Yang J, Imamichi T, Stephens R, Lempicki RA. Extracting biological meaning from large gene lists with DAVID. In: Current Protocols in Bioinformatics. New York: John Wiley & Sons, Inc; 2009. Chapter13. Unit 13.11.Google Scholar
- Hosack DA, Dennis Jr G, Sherman BT, Lane HC, Lempicki RA. Identifying biological themes within lists of genes with EASE. Genome Biol. 2003;4:R70.View ArticlePubMedPubMed CentralGoogle Scholar
- KEGG. Kyoto Encyclopedia of Genes and Genomes. [www.genome.ad.jp/kegg]. Accessed 2 Oct 2015.