GENE-40000; No. of pages: 4; 4C: Gene xxx (2014) xxx–xxx
Contents lists available at ScienceDirect
Gene journal homepage: www.elsevier.com/locate/gene
Original Research Article
Compositional features are potentially involved in the regulation of gene expression of tumor suppressor genes in human tissues Mohammadreza Hajjari a,⁎, Atefeh Khoshnevisan a, Mehrdad Behmanesh b a b
Department of Genetics, Faculty of Science, Shahid Chamran University of Ahvaz, Ahvaz, Iran Department of Genetics, Faculty of Biological Sciences, Tarbiat Modares University, Tehran, Iran
a r t i c l e
i n f o
Article history: Received 30 April 2014 Received in revised form 2 October 2014 Accepted 6 October 2014 Available online xxxx Keywords: Tumor suppressor gene Codon usage Expression level In silico analysis
a b s t r a c t Different mechanisms regulate the expression level of tissue specific genes in human. Here we report some compositional features such as codon usage bias, amino acid usage bias, codon frequency, and base composition which may be potentially related to mRNA amount of tissue specific tumor suppressor genes. Our findings support the possibility that structural elements in gene and protein may play an important role in the regulation of tumor suppressor genes, development, and tumorigenesis. The data presented here can open broad vistas in the understanding and treatment of a variety of human malignancies. © 2014 Elsevier B.V. All rights reserved.
1. Introduction Much progress has been made toward the identification of molecular pathogenesis of cancers. Several studies have been done to understand cancer related genes including tumor suppressor genes. This class of genes includes many genes which are involved in hereditary and nonhereditary cancers (Haber and Harlow, 1997). Tumor suppressor genes govern a wide range of normal cellular activities such as cell cycle checkpoint and DNA repair which might be deregulated in cancer cells. Inactivation of these genes in the germ line may lead one to be predisposed to cancer. Tumor suppressor genes are also inactivated by somatic mutations arising during tumor development (Haber and Harlow, 1997; Hajjari et al., 2014). Understanding a detailed characterization of tumor suppressor genes and their cellular function and regulation has the potential to open broad vistas in the treatment of a variety of human malignancies. A diverse array of mechanisms regulates gene expression level in human. Transcriptional control has been the primary focus of gene regulatory research. Also, deciphering the relationship between transcriptional, posttranscriptional, translational, and transport processes provides new insights into gene regulation. Some important features, which are
Abbreviations: TSG, tumor suppressor genes; CAI, codon adaption index; CDS, coding domain sequence. ⁎ Corresponding author. E-mail addresses: [email protected], [email protected] (M. Hajjari).
challenging subjects in human gene regulation, are codon usage, amino acid usage, and base composition of the human genome. In recent years, numerous studies have attempted to understand the relationship between gene expression levels and compositional features such as biased codon and amino acid usage (Goetz and Fuglsang, 2005; Ingvarsson, 2007; Misawa and Kikuno, 2011; Prabha et al., 2012; Sharp et al., 1986), whole genome regulatory networks (Gao et al., 2004), base composition (Arhondakis et al., 2004), and intron length (Castillo-Davis et al., 2002). Besides many studies, the correlation between these characteristics and expression level of human genes has been controversial. In order to get more precise conclusions, some studies have focused on tissue specific genes (Plotkin et al., 2004; Sémon et al., 2006). These studies are interesting because they can better elucidate the processes involved in differentiation. It has been reported that in most cases the tissue-specific codon usage has been selectively preserved throughout the evolution of human and mice from their common ancestor, yet the biological mechanism and impact of this phenomenon certainly require further study. Regarding the importance of tumor suppressor genes in the development and progression of cancer, we prompted to understand the potential features which may impact the regulation of these genes in human. In this study, we have correlated the compositional features and expression level of tumor suppressor genes in human tissues. Elucidating any correlation between these factors can lead to a better biological understanding of tumor suppressor genes. Also, the results may help to find some clues about the regulation of these genes in cancer initiation and progression.
http://dx.doi.org/10.1016/j.gene.2014.10.011 0378-1119/© 2014 Elsevier B.V. All rights reserved.
Please cite this article as: Hajjari, M., et al., Compositional features are potentially involved in the regulation of gene expression of tumor suppressor genes in human tissues, Gene (2014), http://dx.doi.org/10.1016/j.gene.2014.10.011
2
M. Hajjari et al. / Gene xxx (2014) xxx–xxx
2. Materials and methods 2.1. Sequence retrieval, alignments, and expression data acquisition In total, 69 tumor suppressor genes (the list are in Supplementary File 1) which code proteins with different molecular functions and localizations in the cell, were extracted along with their normalized mRNA expression directly from SOURCE, 2009 (Stanford Microarray) database (source.stanford.edu). This database links to microarray experiments that included the queried gene. Their CDS and protein sequences were obtained from NCBI database. Because of minimizing the statistical errors, multiple alignments were performed for obtained sequences by ClustalW program (http://www.ebi.ac.uk). Twenty five tissues were selected and in each tissue the average expression level, the number of expressed genes, and the highest expressed gene were recognized. 2.2. Sequence compositional features Calculating the base content (totally and in each position for all codons) in percent, GC content (totally and GC1, 2, 3) in percent, number of codons and their frequencies and synonymous codon usage features – the percentage of each synonymous codon in each codon family that codes for the same amino acid – was done with FREQSQ program (http://www.bioinfo.hku.hk). Also, for estimating the codon usage bias for each gene, codon adaptation index (CAI) was assessed for each gene (http://genomes.urv.es/CAIcal/). 2.3. Amino acid sequence characteristics For each protein, “Protéines: courbe de titrage (ABIM)” program (http://sites.univ-provence.fr) was used in calculating the amino acid composition in number and in percent, aliphatic amino acids percentage, theoretical Pi, and molecular mass. 2.4. Statistical analyses For each tissue, the correlation between gene expression level of tumor suppressor genes and compositional features of CDSs and proteins was analyzed with MINITAB and GraphPad. P-Values below 0.01 were considered significant. However, in some tissues, because of the lack of p-value less than 0.01, we considered the characteristics with p-values below 0.05 as the most significant features. Finally, to pinpoint which features are truly significant, Bonferroni correction was done to decrease the statistical errors. Bonferroni correction was taken into account for each parameter separately by SISA program (http://www.quantitativeskills. com/sisa/calculations/bonhlp.htm) with the α-value = 0.05 in the whole correlation analyses. 3. Results 3.1. Correlation between gene expression and compositional features Besides analyzing the correlation between tumor suppressor gene expression level and their compositional features in 25 tissues, 6475 correlations were obtained, in which 117 are with p-value b 0.01 (Supplementary Table 1). The most significant features that have the lowest p-value for each tissue are listed in Table 1 and presented in Fig. 1. It should be pointed out that the adopted threshold for a significant level (p b 0.01) is not extremely low. Therefore, approximately 64 (1% of the total number 6475) features may be significant by chance that probably lead to some artifacts of the present results. Nevertheless, the majority of these significant features are informative. After the Bonferroni correction, 22 significant features were achieved (Supplementary Table 2). It should be noticed that the most significant features in each tissue listed in Table 1 also appear in supplementary Table 2 if that they have passed the Bonferroni correction.
Table 1 Most significant features of tumor suppressor genes which have relation with their expression levels in human tissues. Tissue
Number of genes
Feature
p-Value
Correlation coefficient
Bladder Kidney Embryonic tissue Liver skin Uterus Mammary-gland Testis Prostate Ovary Nerve-tissue Salivary-glands Stomach Lung Brain Bone Pancreas Muscle Heart Placenta Spleen Thyroid Lymph node Eye Bone marrow
40 58 56 52 58 61 55 64 60 54 32 28 51 66 67 53 62 59 54 58 37 18 48 64 35
ATGa AAGa AGGa AAGa ATCa ATCa AAGa TTGc (L) Rb Tb CGGa CTAa Rb CGTa GCTa ATCa ATCa TGTa ATCa Tc GCTa CGTa AGGa Rb Rb
b0.001 0.019 b0.001 0.002 b0.001 0.001 0.001 0.001 0.002 0.001 b0.001 0.001 b0.001 0.003 0.003 b0.001 0.001 0.003 0.001 0.023 0.007 0.001 b0.001 b0.001 b0.001
0.407 0.278 0.520 0.357 0.477 0.302 0.381 0.383 0.363 −0.376 0.466 −0.371 0.456 0.352 0.351 0.411 0.379 0.345 0.379 −0.269 0.317 0.401 0.427 0.435 0.404
CAI: codon adaptation index. a Codon frequency. b Amino acid frequency. c Codon Usage feature.
3.2. Common features in different tissues Among the characteristics in our analyses (given in Supplementary Table 1), the significant features in various tissues include codon content and amino acid content in percent (65 and 25 features respectively), Pi (9 features), and synonymous codon usage (16 features). It is important to note that most of the significant features are common between different tissues. This result supports the conclusion that the common mechanisms may be responsible for regulating gene expression in different tissues. The results show that the significant codon contents include some codons in which the number of codons coding arginine has more proportions to other amino acids (26/65) (Supplementary Table 1). On the other hand, the amino acid contents, which correlate significantly with the expression level of tumor suppressor genes, are arginine (in 13 tissues), tryptophan (in one tissue), threonine (in five tissues), asparagine (in three tissues), isoleucine (in one tissue), methionine (in one tissue), and lysine (in one tissue). The results indicate that the only significant feature for nucleotide compositional features is G2 (frequency of guanine in the second base of codons) in the stomach. Furthermore, among synonymous codon usage features, which correlate with the expression level of tumor suppressor genes, three codons are common between some tissues (AGC, CTA and CTC). To find the level of codon bias, codon adaptation index (CAI) for each gene was measured. We found some correlations between CAI and expression level of tumor suppressor genes in some tissues (p-value less than 0.05). These correlations were in skin, ovary, nerve tissues, salivary glands, lung, and pancreas. 3.3. Bonferroni correction Since multiple correlations were done in the present study, the Bonferroni correction was done to decrease the statistical errors (Supplementary Table 2). The most common features in these results are codon content for AUC and amino acid content for arginine.
Please cite this article as: Hajjari, M., et al., Compositional features are potentially involved in the regulation of gene expression of tumor suppressor genes in human tissues, Gene (2014), http://dx.doi.org/10.1016/j.gene.2014.10.011
M. Hajjari et al. / Gene xxx (2014) xxx–xxx
3
Fig. 1. The coefficient of the correlation between the expression level of tumor suppressor genes and the most significant compositional feature.
4. Discussion In this study, the correlation between tumor suppressor gene compositional features and its expression level has been revealed. Since dysregulation of tumor suppressor genes in normal tissues might lead to tumor formation and progression, any knowledge about the mechanism of the regulation of these genes can help us to understand cancer better. The significant correlations between sequence features and mRNA expression level of tumor suppressor genes have important implications for understanding the molecular mechanisms of tissue-specific oncogenesis. Also, these results could confirm some evidence obtained from other studies. In the current study, the positive correlation between arginine content and the expression level of tumor suppressor genes has been shown in 13 tissues. This may support the criticality of this amino acid in translational efficiency and accuracy. Arginine is critical for the growth of human cancers, particularly those marked by de novo chemoresistance and a poor clinical outcome (Delage et al., 2010). Arginine has a noticeable role in nitric oxide (NO) metabolism which is involved in tumor formation (Lind, 2004; Shen et al., 2006). The correlation between arginine content and the expression level of tumor suppressor genes may indicate the role of this amino acid in the expression regulation of tumor suppressor genes in different tissues. Regarding the role of arginine in tumorigenesis, our hypothesis is that the increasing level of arginine may lead to the up-regulation of tumor suppressor proteins in normal tissues. Furthermore, correlations between some other amino acids and the expression level in specific tissues are interesting. For example, according to our results, amino acid asparagine has a positive correlation with the expression level of tumor suppressor gene in ovary, nerve tissues, and embryonic tissues. There are some evidences about the role of this amino acid in the control of the cell growth. It has been reported that asparagine synthetase is a potential biomarker in ovarian cancer (Lorenzi and Weinstein, 2009). Also, it is known that asparagine can be the inductor of apoptosis in nervous tissue (Chalisova et al., 2002). These data may support the importance of this amino acid in the regulation of tumor suppressor gene expression. We found that threonine has negative correlation with the expression level of tumor suppressor genes in the liver, ovary, prostate, placenta, and bone. In a study by Tschudy et al., it was found that the activity of
the enzyme threonine dehydrogenase, which catalyzes the conversion of L-threonine to aminoacetone, was decreased in the livers of animals bearing advanced sarcoma (Tschudy et al., 1964). Meanwhile, in our study, the amino acid lysine has a significant correlation with the gene expression level in bladder. In other studies, it has been shown that the activity of metalloproteinases (MMPs) can be inhibited by a nutrient mixture including amino-acids L-lysine and L-proline that work synergistically to stop the spread of cancer cells through a connective tissue (Roomi et al., 2006; Roomi et al., 2010). Altogether, it may be inferred that the content of amino acids in tissues is very important for the regulation of the expression level of tumor suppressor genes. We hypothesize that correlations are crucial to balance the growth of the normal cells. Furthermore, the enzymes which are involved in the metabolism of specific amino acids in tissues could be potential biomarkers in cancer. In summary, the correlation between composition features of proteins and gene expression level reveals some potential regulatory mechanisms in gene regulation. The correlation between frequency of codons and the expression level in some human tissues may be attributed to the frequency of specific tRNAs in specific tissues for the translation of tumor suppressor mRNAs. Some studies demonstrate the existence of tissue-specific expression of tRNA species and a role for tRNA heterogeneity in regulating translation and possibly the additional processes in human such as the development of cancer (Dittmar et al., 2006; Pavon-Eternod et al., 2009). Also, some studies indicate an association between codon usage of cancer related genes and relative tRNA expression in cancer (Pavon-Eternod et al., 2009). This may regulate the rate and the accuracy of translation. We assume that with the high level of certain tRNAs, using their related codons may prevent ribosome pausing on mRNA. So, this may stabilize the ribosome and elevate the efficiency of translation. The association of phase-specific base compositional features with gene expression levels may reflect the corresponding codon usage frequencies. G2 (frequency of guanine in the second base of codons) significantly correlates with gene expression level in the stomach. It is important to mention that over half of the codons which correlate significantly with gene expression level have guanine in second position. Translational selection may be responsible for the correlation between synonymous codon usage features and gene expression level in some tissues. Isoaccepting tRNA abundance in different human tissues
Please cite this article as: Hajjari, M., et al., Compositional features are potentially involved in the regulation of gene expression of tumor suppressor genes in human tissues, Gene (2014), http://dx.doi.org/10.1016/j.gene.2014.10.011
4
M. Hajjari et al. / Gene xxx (2014) xxx–xxx
may account for this correlation (Lavner and Kotlar, 2005; Rocha, 2004). We found some correlations between gene expression and CAI (as an indicator of codon usage bias). Codon bias in tissue-specifically expressed genes has been reported previously. This may prompt the insight that such biases can be related to potential tissue-dependent differences in tRNA expression (Plotkin et al., 2004). In summary, because of the great value of assessing human tumor suppressor gene expression in cancer prevention and therapy, this study was done to provide us with a good opportunity to analyze the relationship between human gene expression and sequence compositional features. Zhou et al. identified some similar significant correlations between tissues specific oncogenes and sequence compositional features in human tissues (Zhou et al., 2007). Our results are consistent with their observation and support the idea that some translational selections exist on synonymous codon usage pattern in genes involved in tumor progression. The understanding of different mechanisms involved in the regulation of tumor suppressor genes, in normal or tumor tissues, might lead to better images about cancer initiation and progression. It appears that these results may provide some new potential clues to understanding the translational selection on sequence features of human tumor suppressor genes. Besides, tissue-specific codon usage also has implications for the optimal design of gene therapies targeted at specific tissues. Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.gene.2014.10.011. Conflicts of interest No conflicts of interest exist. Acknowledgments The research was supported by Shahid Chamran University of Ahvaz in Iran. References Arhondakis, S., Auletta, F., Torelli, G., D'Onofrio, G., 2004. Base composition and expression level of human genes. Gene 325, 165–169. Castillo-Davis, C.I., Mekhedov, S.L., Hartl, D.L., Koonin, E.V., Kondrashov, F.A., 2002. Selection for short introns in highly expressed genes. Nat. Genet. 31, 415–418. Chalisova, N., Penniyanen, V., Khaaze, G., 2002. Regulating role of some amino acids in the development of apoptosis in organotypic culture of nervous and lymphoid tissues. Russ. Physiol. J. 5, 627–633.
Delage, B., Fennell, D.A., Nicholson, L., McNeish, I., Lemoine, N.R., Crook, T., Szlosarek, P.W., 2010. Arginine deprivation and argininosuccinate synthetase expression in the treatment of cancer. Int. J. Cancer 126, 2762–2772. Dittmar, K.A., Goodenbour, J.M., Pan, T., 2006. Tissue-specific differences in human transfer RNA expression. PLoS Genet. 2 (12), e221. Gao, F., Foat, B.C., Bussemaker, H.J., 2004. Defining transcriptional networks through integrative modeling of mRNA expression and transcription factor binding data. BMC Bioinforma. 5, 31. Goetz, R.M., Fuglsang, A., 2005. Correlation of codon bias measures with mRNA levels: analysis of transcriptome data from Escherichia coli. Biochem. Biophys. Res. Commun. 327, 4–7. Haber, D., Harlow, E., 1997. Tumour-suppressor genes: evolving definitions in the genomic age. Nat. Genet. 16, 320–322. Hajjari, M., Khoshnevisan, A., Lemos, B., 2014. Characterizing theRetinoblastoma 1 locus: putative elements for Rb1 regulation by in silico analysis, Front. Genet. 5 (2). http:// dx.doi.org/10.3389/fgene.2014.00002. Ingvarsson, P.K., 2007. Gene expression and protein length influence codon usage and rates of sequence evolution in populus tremula. Mol. Biol. Evol. 24. Lavner, Y., Kotlar, D., 2005. Codon bias as a factor in regulating expression via translation rate in the human genome. Gene 345, 127–138. Lind, D.S., 2004. Arginine and cancer. J. Nutr. 134, 2837S–2841S. Lorenzi, P.L., Weinstein, J.N., 2009. Asparagine synthetase: a new potential biomarker in ovarian cancer. Drug News Perspect. 22, 61–64. Misawa, K., Kikuno, R.F., 2011. Relationship between amino acid composition and gene expression in the mouse genome. BMC Res. Notes 27, 4–20. Pavon-Eternod, M., Gomes, S., Geslain, R., Dai, Q., Rosner, M.R., 2009. tRNA over-expression in breast cancer and functional consequences. Nucleic Acids Res. 37 (21), 7268–7280. Plotkin, J.B., Robins, H., Levine, A.J., 2004. Tissue-specific codon usage and the expression of human genes. Proc. Natl. Acad. Sci. U. S. A. 101, 12588–12591. Prabha, R., Singh, D.P., Gupta, S.K., Farooqi, S., Rai, A., 2012. Synonymous codon usage in Thermosynechococcus elongatus (cyanobacteria) identifies the factors shaping codon usage variation. Bioinformation 8, 622–628. Rocha, E.P., 2004. Codon usage bias from tRNA's point of view: redundancy, specialization, and efficient decoding for translation optimization. Genome Res. 14, 2279–2286. Roomi, M.W., Ivanov, V., Kalinovsky, T., Niedzwiecki, A., Rath, M., 2006. Inhibition of matrix metalloproteinase-2 secretion and invasion by human ovarian cancer cell line SK-OV-3 with lysine, proline, arginine, ascorbic acid and green tea extract. J. Obstet. Gynaecol. Res. 32, 148–154. Roomi, M.W., Monterrey, J.C., Kalinovsky, T., Rath, M., Niedzwiecki, A., 2010. Comparative effects of EGCG, green tea and a nutrient mixture on the patterns of MMP-2 and MMP-9 expression in cancer cell lines. Oncol. Rep. 24, 747–757. Sémon, M., Lobry, J.R., Duret, L., 2006. No evidence for tissue-specific adaptation of synonymous codon usage in humans. Mol. Biol. Evol. 23, 523–529. Sharp, P.M., Tuohy, T.M., Mosurski, K.R., 1986. Codon usage in yeast: cluster analysis clearly differentiates highly and lowly expressed genes. Nucleic Acids Res. 14, 5125–5143. Shen, L.J., Beloussow, K., Shen, W.C., 2006. Modulation of arginine metabolic pathways as the potential anti-tumor mechanism of recombinant arginine deiminase. Cancer Lett. 231, 30–35. Tschudy, D.P., Welland, F.H., Collins, A.R., Rechcigl, M.J.R., 1964. The effect of tumor growth on hepatic threonine dehydrogenase. Cancer Res. 24, 2033–2037. Zhou, Y., Ma, B., Zhang, H., 2007. Human oncogene tissue-specific expression level significantly correlates with sequence compositional features. FEBS Lett. 51, 4361–4365.
Please cite this article as: Hajjari, M., et al., Compositional features are potentially involved in the regulation of gene expression of tumor suppressor genes in human tissues, Gene (2014), http://dx.doi.org/10.1016/j.gene.2014.10.011
Contents lists available at ScienceDirect
Gene journal homepage: www.elsevier.com/locate/gene
Original Research Article
Compositional features are potentially involved in the regulation of gene expression of tumor suppressor genes in human tissues Mohammadreza Hajjari a,⁎, Atefeh Khoshnevisan a, Mehrdad Behmanesh b a b
Department of Genetics, Faculty of Science, Shahid Chamran University of Ahvaz, Ahvaz, Iran Department of Genetics, Faculty of Biological Sciences, Tarbiat Modares University, Tehran, Iran
a r t i c l e
i n f o
Article history: Received 30 April 2014 Received in revised form 2 October 2014 Accepted 6 October 2014 Available online xxxx Keywords: Tumor suppressor gene Codon usage Expression level In silico analysis
a b s t r a c t Different mechanisms regulate the expression level of tissue specific genes in human. Here we report some compositional features such as codon usage bias, amino acid usage bias, codon frequency, and base composition which may be potentially related to mRNA amount of tissue specific tumor suppressor genes. Our findings support the possibility that structural elements in gene and protein may play an important role in the regulation of tumor suppressor genes, development, and tumorigenesis. The data presented here can open broad vistas in the understanding and treatment of a variety of human malignancies. © 2014 Elsevier B.V. All rights reserved.
1. Introduction Much progress has been made toward the identification of molecular pathogenesis of cancers. Several studies have been done to understand cancer related genes including tumor suppressor genes. This class of genes includes many genes which are involved in hereditary and nonhereditary cancers (Haber and Harlow, 1997). Tumor suppressor genes govern a wide range of normal cellular activities such as cell cycle checkpoint and DNA repair which might be deregulated in cancer cells. Inactivation of these genes in the germ line may lead one to be predisposed to cancer. Tumor suppressor genes are also inactivated by somatic mutations arising during tumor development (Haber and Harlow, 1997; Hajjari et al., 2014). Understanding a detailed characterization of tumor suppressor genes and their cellular function and regulation has the potential to open broad vistas in the treatment of a variety of human malignancies. A diverse array of mechanisms regulates gene expression level in human. Transcriptional control has been the primary focus of gene regulatory research. Also, deciphering the relationship between transcriptional, posttranscriptional, translational, and transport processes provides new insights into gene regulation. Some important features, which are
Abbreviations: TSG, tumor suppressor genes; CAI, codon adaption index; CDS, coding domain sequence. ⁎ Corresponding author. E-mail addresses: [email protected], [email protected] (M. Hajjari).
challenging subjects in human gene regulation, are codon usage, amino acid usage, and base composition of the human genome. In recent years, numerous studies have attempted to understand the relationship between gene expression levels and compositional features such as biased codon and amino acid usage (Goetz and Fuglsang, 2005; Ingvarsson, 2007; Misawa and Kikuno, 2011; Prabha et al., 2012; Sharp et al., 1986), whole genome regulatory networks (Gao et al., 2004), base composition (Arhondakis et al., 2004), and intron length (Castillo-Davis et al., 2002). Besides many studies, the correlation between these characteristics and expression level of human genes has been controversial. In order to get more precise conclusions, some studies have focused on tissue specific genes (Plotkin et al., 2004; Sémon et al., 2006). These studies are interesting because they can better elucidate the processes involved in differentiation. It has been reported that in most cases the tissue-specific codon usage has been selectively preserved throughout the evolution of human and mice from their common ancestor, yet the biological mechanism and impact of this phenomenon certainly require further study. Regarding the importance of tumor suppressor genes in the development and progression of cancer, we prompted to understand the potential features which may impact the regulation of these genes in human. In this study, we have correlated the compositional features and expression level of tumor suppressor genes in human tissues. Elucidating any correlation between these factors can lead to a better biological understanding of tumor suppressor genes. Also, the results may help to find some clues about the regulation of these genes in cancer initiation and progression.
http://dx.doi.org/10.1016/j.gene.2014.10.011 0378-1119/© 2014 Elsevier B.V. All rights reserved.
Please cite this article as: Hajjari, M., et al., Compositional features are potentially involved in the regulation of gene expression of tumor suppressor genes in human tissues, Gene (2014), http://dx.doi.org/10.1016/j.gene.2014.10.011
2
M. Hajjari et al. / Gene xxx (2014) xxx–xxx
2. Materials and methods 2.1. Sequence retrieval, alignments, and expression data acquisition In total, 69 tumor suppressor genes (the list are in Supplementary File 1) which code proteins with different molecular functions and localizations in the cell, were extracted along with their normalized mRNA expression directly from SOURCE, 2009 (Stanford Microarray) database (source.stanford.edu). This database links to microarray experiments that included the queried gene. Their CDS and protein sequences were obtained from NCBI database. Because of minimizing the statistical errors, multiple alignments were performed for obtained sequences by ClustalW program (http://www.ebi.ac.uk). Twenty five tissues were selected and in each tissue the average expression level, the number of expressed genes, and the highest expressed gene were recognized. 2.2. Sequence compositional features Calculating the base content (totally and in each position for all codons) in percent, GC content (totally and GC1, 2, 3) in percent, number of codons and their frequencies and synonymous codon usage features – the percentage of each synonymous codon in each codon family that codes for the same amino acid – was done with FREQSQ program (http://www.bioinfo.hku.hk). Also, for estimating the codon usage bias for each gene, codon adaptation index (CAI) was assessed for each gene (http://genomes.urv.es/CAIcal/). 2.3. Amino acid sequence characteristics For each protein, “Protéines: courbe de titrage (ABIM)” program (http://sites.univ-provence.fr) was used in calculating the amino acid composition in number and in percent, aliphatic amino acids percentage, theoretical Pi, and molecular mass. 2.4. Statistical analyses For each tissue, the correlation between gene expression level of tumor suppressor genes and compositional features of CDSs and proteins was analyzed with MINITAB and GraphPad. P-Values below 0.01 were considered significant. However, in some tissues, because of the lack of p-value less than 0.01, we considered the characteristics with p-values below 0.05 as the most significant features. Finally, to pinpoint which features are truly significant, Bonferroni correction was done to decrease the statistical errors. Bonferroni correction was taken into account for each parameter separately by SISA program (http://www.quantitativeskills. com/sisa/calculations/bonhlp.htm) with the α-value = 0.05 in the whole correlation analyses. 3. Results 3.1. Correlation between gene expression and compositional features Besides analyzing the correlation between tumor suppressor gene expression level and their compositional features in 25 tissues, 6475 correlations were obtained, in which 117 are with p-value b 0.01 (Supplementary Table 1). The most significant features that have the lowest p-value for each tissue are listed in Table 1 and presented in Fig. 1. It should be pointed out that the adopted threshold for a significant level (p b 0.01) is not extremely low. Therefore, approximately 64 (1% of the total number 6475) features may be significant by chance that probably lead to some artifacts of the present results. Nevertheless, the majority of these significant features are informative. After the Bonferroni correction, 22 significant features were achieved (Supplementary Table 2). It should be noticed that the most significant features in each tissue listed in Table 1 also appear in supplementary Table 2 if that they have passed the Bonferroni correction.
Table 1 Most significant features of tumor suppressor genes which have relation with their expression levels in human tissues. Tissue
Number of genes
Feature
p-Value
Correlation coefficient
Bladder Kidney Embryonic tissue Liver skin Uterus Mammary-gland Testis Prostate Ovary Nerve-tissue Salivary-glands Stomach Lung Brain Bone Pancreas Muscle Heart Placenta Spleen Thyroid Lymph node Eye Bone marrow
40 58 56 52 58 61 55 64 60 54 32 28 51 66 67 53 62 59 54 58 37 18 48 64 35
ATGa AAGa AGGa AAGa ATCa ATCa AAGa TTGc (L) Rb Tb CGGa CTAa Rb CGTa GCTa ATCa ATCa TGTa ATCa Tc GCTa CGTa AGGa Rb Rb
b0.001 0.019 b0.001 0.002 b0.001 0.001 0.001 0.001 0.002 0.001 b0.001 0.001 b0.001 0.003 0.003 b0.001 0.001 0.003 0.001 0.023 0.007 0.001 b0.001 b0.001 b0.001
0.407 0.278 0.520 0.357 0.477 0.302 0.381 0.383 0.363 −0.376 0.466 −0.371 0.456 0.352 0.351 0.411 0.379 0.345 0.379 −0.269 0.317 0.401 0.427 0.435 0.404
CAI: codon adaptation index. a Codon frequency. b Amino acid frequency. c Codon Usage feature.
3.2. Common features in different tissues Among the characteristics in our analyses (given in Supplementary Table 1), the significant features in various tissues include codon content and amino acid content in percent (65 and 25 features respectively), Pi (9 features), and synonymous codon usage (16 features). It is important to note that most of the significant features are common between different tissues. This result supports the conclusion that the common mechanisms may be responsible for regulating gene expression in different tissues. The results show that the significant codon contents include some codons in which the number of codons coding arginine has more proportions to other amino acids (26/65) (Supplementary Table 1). On the other hand, the amino acid contents, which correlate significantly with the expression level of tumor suppressor genes, are arginine (in 13 tissues), tryptophan (in one tissue), threonine (in five tissues), asparagine (in three tissues), isoleucine (in one tissue), methionine (in one tissue), and lysine (in one tissue). The results indicate that the only significant feature for nucleotide compositional features is G2 (frequency of guanine in the second base of codons) in the stomach. Furthermore, among synonymous codon usage features, which correlate with the expression level of tumor suppressor genes, three codons are common between some tissues (AGC, CTA and CTC). To find the level of codon bias, codon adaptation index (CAI) for each gene was measured. We found some correlations between CAI and expression level of tumor suppressor genes in some tissues (p-value less than 0.05). These correlations were in skin, ovary, nerve tissues, salivary glands, lung, and pancreas. 3.3. Bonferroni correction Since multiple correlations were done in the present study, the Bonferroni correction was done to decrease the statistical errors (Supplementary Table 2). The most common features in these results are codon content for AUC and amino acid content for arginine.
Please cite this article as: Hajjari, M., et al., Compositional features are potentially involved in the regulation of gene expression of tumor suppressor genes in human tissues, Gene (2014), http://dx.doi.org/10.1016/j.gene.2014.10.011
M. Hajjari et al. / Gene xxx (2014) xxx–xxx
3
Fig. 1. The coefficient of the correlation between the expression level of tumor suppressor genes and the most significant compositional feature.
4. Discussion In this study, the correlation between tumor suppressor gene compositional features and its expression level has been revealed. Since dysregulation of tumor suppressor genes in normal tissues might lead to tumor formation and progression, any knowledge about the mechanism of the regulation of these genes can help us to understand cancer better. The significant correlations between sequence features and mRNA expression level of tumor suppressor genes have important implications for understanding the molecular mechanisms of tissue-specific oncogenesis. Also, these results could confirm some evidence obtained from other studies. In the current study, the positive correlation between arginine content and the expression level of tumor suppressor genes has been shown in 13 tissues. This may support the criticality of this amino acid in translational efficiency and accuracy. Arginine is critical for the growth of human cancers, particularly those marked by de novo chemoresistance and a poor clinical outcome (Delage et al., 2010). Arginine has a noticeable role in nitric oxide (NO) metabolism which is involved in tumor formation (Lind, 2004; Shen et al., 2006). The correlation between arginine content and the expression level of tumor suppressor genes may indicate the role of this amino acid in the expression regulation of tumor suppressor genes in different tissues. Regarding the role of arginine in tumorigenesis, our hypothesis is that the increasing level of arginine may lead to the up-regulation of tumor suppressor proteins in normal tissues. Furthermore, correlations between some other amino acids and the expression level in specific tissues are interesting. For example, according to our results, amino acid asparagine has a positive correlation with the expression level of tumor suppressor gene in ovary, nerve tissues, and embryonic tissues. There are some evidences about the role of this amino acid in the control of the cell growth. It has been reported that asparagine synthetase is a potential biomarker in ovarian cancer (Lorenzi and Weinstein, 2009). Also, it is known that asparagine can be the inductor of apoptosis in nervous tissue (Chalisova et al., 2002). These data may support the importance of this amino acid in the regulation of tumor suppressor gene expression. We found that threonine has negative correlation with the expression level of tumor suppressor genes in the liver, ovary, prostate, placenta, and bone. In a study by Tschudy et al., it was found that the activity of
the enzyme threonine dehydrogenase, which catalyzes the conversion of L-threonine to aminoacetone, was decreased in the livers of animals bearing advanced sarcoma (Tschudy et al., 1964). Meanwhile, in our study, the amino acid lysine has a significant correlation with the gene expression level in bladder. In other studies, it has been shown that the activity of metalloproteinases (MMPs) can be inhibited by a nutrient mixture including amino-acids L-lysine and L-proline that work synergistically to stop the spread of cancer cells through a connective tissue (Roomi et al., 2006; Roomi et al., 2010). Altogether, it may be inferred that the content of amino acids in tissues is very important for the regulation of the expression level of tumor suppressor genes. We hypothesize that correlations are crucial to balance the growth of the normal cells. Furthermore, the enzymes which are involved in the metabolism of specific amino acids in tissues could be potential biomarkers in cancer. In summary, the correlation between composition features of proteins and gene expression level reveals some potential regulatory mechanisms in gene regulation. The correlation between frequency of codons and the expression level in some human tissues may be attributed to the frequency of specific tRNAs in specific tissues for the translation of tumor suppressor mRNAs. Some studies demonstrate the existence of tissue-specific expression of tRNA species and a role for tRNA heterogeneity in regulating translation and possibly the additional processes in human such as the development of cancer (Dittmar et al., 2006; Pavon-Eternod et al., 2009). Also, some studies indicate an association between codon usage of cancer related genes and relative tRNA expression in cancer (Pavon-Eternod et al., 2009). This may regulate the rate and the accuracy of translation. We assume that with the high level of certain tRNAs, using their related codons may prevent ribosome pausing on mRNA. So, this may stabilize the ribosome and elevate the efficiency of translation. The association of phase-specific base compositional features with gene expression levels may reflect the corresponding codon usage frequencies. G2 (frequency of guanine in the second base of codons) significantly correlates with gene expression level in the stomach. It is important to mention that over half of the codons which correlate significantly with gene expression level have guanine in second position. Translational selection may be responsible for the correlation between synonymous codon usage features and gene expression level in some tissues. Isoaccepting tRNA abundance in different human tissues
Please cite this article as: Hajjari, M., et al., Compositional features are potentially involved in the regulation of gene expression of tumor suppressor genes in human tissues, Gene (2014), http://dx.doi.org/10.1016/j.gene.2014.10.011
4
M. Hajjari et al. / Gene xxx (2014) xxx–xxx
may account for this correlation (Lavner and Kotlar, 2005; Rocha, 2004). We found some correlations between gene expression and CAI (as an indicator of codon usage bias). Codon bias in tissue-specifically expressed genes has been reported previously. This may prompt the insight that such biases can be related to potential tissue-dependent differences in tRNA expression (Plotkin et al., 2004). In summary, because of the great value of assessing human tumor suppressor gene expression in cancer prevention and therapy, this study was done to provide us with a good opportunity to analyze the relationship between human gene expression and sequence compositional features. Zhou et al. identified some similar significant correlations between tissues specific oncogenes and sequence compositional features in human tissues (Zhou et al., 2007). Our results are consistent with their observation and support the idea that some translational selections exist on synonymous codon usage pattern in genes involved in tumor progression. The understanding of different mechanisms involved in the regulation of tumor suppressor genes, in normal or tumor tissues, might lead to better images about cancer initiation and progression. It appears that these results may provide some new potential clues to understanding the translational selection on sequence features of human tumor suppressor genes. Besides, tissue-specific codon usage also has implications for the optimal design of gene therapies targeted at specific tissues. Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.gene.2014.10.011. Conflicts of interest No conflicts of interest exist. Acknowledgments The research was supported by Shahid Chamran University of Ahvaz in Iran. References Arhondakis, S., Auletta, F., Torelli, G., D'Onofrio, G., 2004. Base composition and expression level of human genes. Gene 325, 165–169. Castillo-Davis, C.I., Mekhedov, S.L., Hartl, D.L., Koonin, E.V., Kondrashov, F.A., 2002. Selection for short introns in highly expressed genes. Nat. Genet. 31, 415–418. Chalisova, N., Penniyanen, V., Khaaze, G., 2002. Regulating role of some amino acids in the development of apoptosis in organotypic culture of nervous and lymphoid tissues. Russ. Physiol. J. 5, 627–633.
Delage, B., Fennell, D.A., Nicholson, L., McNeish, I., Lemoine, N.R., Crook, T., Szlosarek, P.W., 2010. Arginine deprivation and argininosuccinate synthetase expression in the treatment of cancer. Int. J. Cancer 126, 2762–2772. Dittmar, K.A., Goodenbour, J.M., Pan, T., 2006. Tissue-specific differences in human transfer RNA expression. PLoS Genet. 2 (12), e221. Gao, F., Foat, B.C., Bussemaker, H.J., 2004. Defining transcriptional networks through integrative modeling of mRNA expression and transcription factor binding data. BMC Bioinforma. 5, 31. Goetz, R.M., Fuglsang, A., 2005. Correlation of codon bias measures with mRNA levels: analysis of transcriptome data from Escherichia coli. Biochem. Biophys. Res. Commun. 327, 4–7. Haber, D., Harlow, E., 1997. Tumour-suppressor genes: evolving definitions in the genomic age. Nat. Genet. 16, 320–322. Hajjari, M., Khoshnevisan, A., Lemos, B., 2014. Characterizing theRetinoblastoma 1 locus: putative elements for Rb1 regulation by in silico analysis, Front. Genet. 5 (2). http:// dx.doi.org/10.3389/fgene.2014.00002. Ingvarsson, P.K., 2007. Gene expression and protein length influence codon usage and rates of sequence evolution in populus tremula. Mol. Biol. Evol. 24. Lavner, Y., Kotlar, D., 2005. Codon bias as a factor in regulating expression via translation rate in the human genome. Gene 345, 127–138. Lind, D.S., 2004. Arginine and cancer. J. Nutr. 134, 2837S–2841S. Lorenzi, P.L., Weinstein, J.N., 2009. Asparagine synthetase: a new potential biomarker in ovarian cancer. Drug News Perspect. 22, 61–64. Misawa, K., Kikuno, R.F., 2011. Relationship between amino acid composition and gene expression in the mouse genome. BMC Res. Notes 27, 4–20. Pavon-Eternod, M., Gomes, S., Geslain, R., Dai, Q., Rosner, M.R., 2009. tRNA over-expression in breast cancer and functional consequences. Nucleic Acids Res. 37 (21), 7268–7280. Plotkin, J.B., Robins, H., Levine, A.J., 2004. Tissue-specific codon usage and the expression of human genes. Proc. Natl. Acad. Sci. U. S. A. 101, 12588–12591. Prabha, R., Singh, D.P., Gupta, S.K., Farooqi, S., Rai, A., 2012. Synonymous codon usage in Thermosynechococcus elongatus (cyanobacteria) identifies the factors shaping codon usage variation. Bioinformation 8, 622–628. Rocha, E.P., 2004. Codon usage bias from tRNA's point of view: redundancy, specialization, and efficient decoding for translation optimization. Genome Res. 14, 2279–2286. Roomi, M.W., Ivanov, V., Kalinovsky, T., Niedzwiecki, A., Rath, M., 2006. Inhibition of matrix metalloproteinase-2 secretion and invasion by human ovarian cancer cell line SK-OV-3 with lysine, proline, arginine, ascorbic acid and green tea extract. J. Obstet. Gynaecol. Res. 32, 148–154. Roomi, M.W., Monterrey, J.C., Kalinovsky, T., Rath, M., Niedzwiecki, A., 2010. Comparative effects of EGCG, green tea and a nutrient mixture on the patterns of MMP-2 and MMP-9 expression in cancer cell lines. Oncol. Rep. 24, 747–757. Sémon, M., Lobry, J.R., Duret, L., 2006. No evidence for tissue-specific adaptation of synonymous codon usage in humans. Mol. Biol. Evol. 23, 523–529. Sharp, P.M., Tuohy, T.M., Mosurski, K.R., 1986. Codon usage in yeast: cluster analysis clearly differentiates highly and lowly expressed genes. Nucleic Acids Res. 14, 5125–5143. Shen, L.J., Beloussow, K., Shen, W.C., 2006. Modulation of arginine metabolic pathways as the potential anti-tumor mechanism of recombinant arginine deiminase. Cancer Lett. 231, 30–35. Tschudy, D.P., Welland, F.H., Collins, A.R., Rechcigl, M.J.R., 1964. The effect of tumor growth on hepatic threonine dehydrogenase. Cancer Res. 24, 2033–2037. Zhou, Y., Ma, B., Zhang, H., 2007. Human oncogene tissue-specific expression level significantly correlates with sequence compositional features. FEBS Lett. 51, 4361–4365.
Please cite this article as: Hajjari, M., et al., Compositional features are potentially involved in the regulation of gene expression of tumor suppressor genes in human tissues, Gene (2014), http://dx.doi.org/10.1016/j.gene.2014.10.011
