Codon usage bias as a function of generation time and life expectancy

Transcription

1 open acce Hypothe Volume 8(3) odon uage ba a a functon of generaton tme and lfe expectancy Ram N Mahd 1 & Erc Rouchka 2 * 1Well ornell Medcal ollege, Department of Genetc Medcne, New York, NY USA 10065; 2 Unverty of Louvlle, Speed School of Engneerng, Department of omputer Engneerng and omputer Scence, Duthe enter Room 208, Louvlle, KY USA 40292; Erc Rouchka - Emal: Erc.rouchka@louvlle.edu; *orrepondng author Receved September 21, 2011; Accepted January 27, 2012; Publhed February 03, 2012 Abtract: It ha recently been demontrated that human natural codon uage ba optmzed toward a hgher bufferng capacty to mutaton (meaured a the tendency of ngle pont mutaton n a DNA equence to yeld the ame or mlar amno acd) compared to random equence. In th work, we nvetgate th phenomenon further by analyzng the natural DNA of four dfferent pece (human, moue, zebrafh and frut fly) to determne whether uch a tolerance to mutaton correlated wth the lfe pan and age of exual maturaton for the correpondng organm. We alo propoe a new meaure to quantfy the bufferng capacty of a DNA equence to mutaton that take nto account the oberved mutaton rate wthn every genome and the effect of the correpondng mutaton. Our reult ugget there a propenty for tolerance to mutaton that potvely correlated wth the lfe expectancy of the condered organm. Moreover, random equence that are contraned to produce the ame proten a the naturally occurrng equence are found to be more buffered than completely random equence whle beng le buffered than the natural equence. Thee reult ugget that optmzaton toward protectve mechanm tolerant to mutaton correlated wth both lfe expectancy and age to exual maturty at both the level of codon uage ba and the ba of the natural equence of codon telf. Keyword: Bufferng capacty, odon ba, Sequence evoluton Background: In gene codng regon, DNA put together n trplet to form 64 dtnctve codon. 61 of thoe codon yntheze for the 20 amno acd, each of whch can be ynthezed by a few a one or a many a x ynonymou codon. An ntutve aumpton that codon and amno acd are evenly ued throughout dfferent genome. However, t ha been dcovered that amno acd ncorporaton a well a codon uage are baed n all organm from bactera to mammal [1-3]. It antcpated then that electve force are actng to mantan the current balance between mutaton and electon, reultng n optmal codng regon. Some of the propoed factor nclude degree and tmng of gene expreon, codon-antcodon nteracton, trancrpton and tranlaton rate and accuracy, codon context, and global and local G+ content. In human, ba ha been reported on nucleotde abundance n the three codon poton, dnucleotde uage, and d-codon uage wthn codng regon [4]. Th ba lead to a hgh uage of acdc amno acd a well a reflect the avodance of top codon. Ba n codon uage ha been attrbuted to electon toward a more effcent tranlaton model. Th ugget codon uage n hghly expreed gene baed toward optmal codon correpondng to more abundant trna. Such model take nto conderaton both the elongaton rate and the tranlaton accuracy [5-6]. In more recent tude [7-8], t ha been hown that the codon uage ba found to relate to gene expreon wth hghly expreed gene favorng codon correpondng to the more abundant trna (thu ncreang tranlaton effcency) and lower expreed gene favorng le ISSN (onlne) (prnt) Bonformaton 8(3): (2012) Bomedcal Informatc

2 open acce abundant trna. In addton, hghly expreed gene n human are found to be horter n general and contan le ntronc content [9]. Th tudy alo report a ba n the amno acd uage where hghly expreed gene avod complex amno acd, where complexty baed on the weght and hape of the amno acd. Other electon factor propoed nclude the effect of DNA polymerae and repar mechanm, methylaton, pg land [10-11], tue or organelle pecfcty [12], ncreang mrna tablty, trancrpton rate [13-14] and evolutonary age [15]. None of the prevouly mentoned factor manage to provde a complete explanaton for codon uage ba. It lkely that all of thee force work together compettvely wth electon, organelle and organm pecfcty to produce the current ba found n every gene and every genome. In a prevou paper [16], we demontrated that n human, ba n codon uage make the codng DNA gnfcantly more tolerant to pont mutaton than random equence. If only ubttuton event are condered n a codon, the degeneracy of the genetc code allow for a ngle nucleotde converon to reult n a codon repreentng the ame amno acd, a dfferent amno acd (mene mutaton), or a top codon (nonene mutaton). Analy baed on ngle nucleotde mutaton rate and mlarty among amno acd howed that pont mutaton n the natural codng equence of human are gnfcantly more lkely to yeld the ame or mlar amno acd than would be the cae n random DNA equence. Moreover, the codng equence baed on the natural trancrpton readng frame wa hown to be more buffered to mutaton than the other two poble hfted readng frame. Here, we ay a codon buffered to or tolerate pont mutaton when t tend to tranlate to the ame or mlar amno acd a the orgnal one more often than would be expected by random when expoed to pont mutaton. Methodology: Bufferng capacty Bufferng capacty refer to the tendency of a codng equence to allow for pont mutaton event that reult n a equence of the ame or mlar amno acd, a oppoed to mene or nonene mutaton. In order to quantfy th property, we have developed a mathematcal model that approxmate the bufferng capacty for an ndvdual equence by takng nto account the codon compoton, ndvdual nucleotde mutaton rate, and a dtance matrx between amno acd that reflect the reultng damage caued by ndvdual amno acd ubttuton(see upplementary materal). Selecton toward error mnmzaton or ncreaed bufferng not a new concept. For example, the natural codng or the natural mappng between the 20 amno acd and the 61 codng codon beleved to have come to t current tate by evoluton and electon and t ha been hown to provde hgh tolerance to mutaton or tranlaton error [17-19]. In [19], the author have hown that only two n a bllon randomly generated mappng would provde better error mnmzaton than the natural genetc code. In th paper, we tet for a poble correlaton between the expected lfe pan of an organm and the bufferng capacty n t codng DNA. Th hypothe motvated by the fact that an organm that lve long wll requre gene buffered agant mutaton to protect agant harmful change to proten equence whle a pece wth a hort lfe pan doe not need uch a property nce t wll not be ubected to nearly a many pont mutaton throughout t lfe. Furthermore, organm that have a long lfe pan uually reach exual maturty at a later age at whch tme ther DNA get paed to the next generaton. Therefore, t expected that the longer the lfe pan or age to exual maturty of an organm, the hgher the bufferng. To calculate the bufferng capacty of a gven codng DNA equence, we ued a mlar meaure to the one ued n [16]. The computed bufferng capacty take nto conderaton both the organm-pecfc rate of mutaton and the conequence of the correpondng mutaton. For mutaton rate, we have ued the neghbor-dependent ubttuton rate reported n a recent tudy baed on oberved neutral mutaton wthn nerted retrotranpoable element [20]. To etmate the conequence of a mutaton, an amno acd mlarty matrx ued to meaure the dtance between the orgnal amno acd and the amno acd reultng from the mutaton [19]. Th mlarty matrx baed on computaton of the change n the tructure and foldng free energy of a proten when a ngle amno acd mutated to another one at all poton n a et of 141 dfferent proten. On the other hand, a nonene mutaton thought to be the wort type of mutaton; nonethele there no natural way to wegh a nonene mutaton compared to other type of mene mutaton. A an approxmaton, n th paper we conder t to be three tme a detrmental a the wort mene mutaton. Dataet The pece elected nclude Droophla melanogater (frut fly), Dano rero (zebrafh), Mu muculu (moue), and Homo apen (human). Thee were elected baed upon varyng rate of average lfe expectancy and tme to exual maturty a well a the avalablty of equence data and pece pecfc mutaton rate. The rate for lfe expectancy and tme to exual maturty were derved from the Human Agng Genome Reource [21]. Thee rate are provded n Table 1 (See upplementary materal). For the purpoe of th tudy, human gene equence from GenBank buld 35.1 were downloaded from the human Exon- Intron Databae (EID) [22] from the webte 16,800 human gene n th dataet were condered. Thee equence were further fltered to exclude gene whoe nucleotde equence not a multple of three, or who do not begn wth the tart codon ATG. In addton, gene equence that do not end n one of the three top codon (TAA, TAG, or TGA) were removed. For zebrafh, moue, and frut fly, mrna equence were downloaded from the RefSeq databae. The total number of gene ued for each organm provded n Table 2 (See upplementary materal). DS categorzaton The data downloaded repreent known mrna equence from the codng equence (DS) of genc regon. Thee dataet were categorzed and labeled a natural. For each known DS, a randomly generated mrna equence wa ISSN (onlne) (prnt) Bonformaton 8(3): (2012) Bomedcal Informatc

3 contructed that yeld the exact ame proten. Thee correpond to a group labeled a contraned random nce they are randomly generated equence contraned by the property that they mut yeld an dentcal amno acd equence upon tranlaton. A thrd et of equence labeled random wa contructed contng of mrna equence of the ame length a the natural group. Analy of bufferng capacty For each of the natural, contraned random and random equence, the bufferng capacty wa calculated a decrbed n the bufferng capacty ecton. Mutaton rate for frut fly, zebrafh, and human were downloaded from an extng dataet [20]. Mutaton rate for the moue were calculated ung the publcly avalable web erver ung the method decrbed n [20] wth the moue B1_Mur1 repettve element from RepBae a the ancetral equence and a et of 4,010 annotated B1_Mur1 occurrence n the moue genome a the decendant equence. A dtrbuton of the bufferng capacte for each of thee three group wa examned. A dtrbuton cloe to zero would ndcate a et of equence that are hghly buffered to pont mutaton at the mrna level that do not gnfcantly affect the correpondng tranlated proten equence. The man hypothe that f a pece ha a bult-n tolerance to mutaton, the dtrbuton of bufferng capacty core between the naturally occurrng mrna equence and the random equence et hould be gnfcantly dfferent. open acce gnfcance of the ncreaed bufferng n the natural codng equence a compared to the random equence. Second, for each of the four pece, we ue ohen d meaurement to quantfy the effect ze of codon uage ba n ncreang the tolerance to mutaton n the natural codng equence compared to the random equence and to quantfy how much more the contraned random equence are buffered than the completely random equence. Note that a drect comparon between the three organm codng equence would be nvald becaue the mutaton rate are tme ndependent and were computed wthn every genome eparately. The mutaton rate are relatve value wthn the ame genome only. However, the comparon between the natural codng and the random equence under the ame mutaton rate hould tell u how much ba there n the natural DNA tructure toward bufferng wthn every genome eparately. Therefore, the effect ze wthn every genome hould be a vald ndcator for comparon between the dfferent genome. Dcuon: Our reult how that both human and zebrafh DNA are gnfcantly more buffered than random equence. Bufferng more apparent n human whle beng abent n frut fly. Furthermore, random equence that are contraned to produce the ame proten a the natural one are found to be more buffered than completely random equence whle beng le buffered than the natural equence themelve n both human and zebrafh. Th property alo more apparent n human than zebrafh whle beng abent n frut fly. Thee reult ugget evdence that ba n the DNA tructure toward tolerance to mutaton correlated wth lfe pan and realzed on two level: proten electon and codon uage ba. Fgure 1: Bufferng capacty dtrbuton. Shown are the dtrbuton of bufferng capacte (larger x-value = more bufferng) for the naturally occurrng mrna equence (old lne), contraned random equence (dahed lne), and random equence (dotted lne). The pece repreented are frut fly (panel A), zebrafh (panel B), moue (panel ), and human (panel D). Stattcal analy For further analy of the reult two dfferent tattcal analye tool are ued. A t-tet ued to meaure the Fgure 2: Dtrbuton effect a a functon of lfe expectancy. Dm: Droophla melanogater (frut fly), Mm: Mu muculu (moue), Dr: Dano rero (zebrafh), H: Homo apen (human). Dtrbuton of bufferng capacte for frut fly, zebrafh, moue, and human are gven n Fgure 1. Note that the x-ax ISSN (onlne) (prnt) Bonformaton 8(3): (2012) Bomedcal Informatc

4 open acce for each of thee cannot be compared drectly due to the dfferent mutaton rate n each of thee organm. A Fgure 1 clearly ndcate, a dfference n the dtrbuton of bufferng capacty for the random, contraned random and natural mrna equence cannot be oberved n the frut fly whle the eparaton evdent n zebrafh and more pronounced n the moue and human. Analy of the effect ze Table 3 (ee upplementary materal) how that the eparaton between the naturally occurrng mrna and random equence greatet n the moue and n the human. When the effect ze compared to the lfe expectancy (Fgure 2) and tme to exual maturty (Fgure 3), a clear trend reult wth an excepton occurrng wth the moue bufferng dtrbuton. Th may be due to a dfference n the method for calculatng mutatonal rate wthn the moue. Fgure 3: Dtrbuton effect a a functon of age to exual maturty. Dm: Droophla melanogater (frut fly), Mm: Mu muculu (moue), Dr: Dano rero (zebrafh), H: Homo apen (human). In order to provde a more accurate meaure of the relatonhp between lfe expectancy and codon bufferng, a more thorough analy needed for organm wth a wder range of lfe expectance and age to exual maturty. Nonethele, the preented reult demontrate an overall gnfcant correlaton between bufferng capacty to mutaton a a reult of codon uage ba and both the lfe expectancy and the tme to exual maturty of the correpondng organm. To our knowledge, th the frt tme uch a hypothe ha been teted. Our analy alo how that the optmzaton toward hgher tolerance to pont mutaton accomplhed at both level the codon uage and the equence of codon themelve. Thee reult add a new addtve explanaton to current theorem about codon uage ba n the natural trancrpt of thee organm. Fnally, a more organm get equenced and a more accurate and comprehenve tude of mutaton rate become avalable, we hope to contnue th work on a larger cale, wth addtonal phylogenetc nformaton ncluded a well. Acknowledgement: Support provded by NIH NRR grant P20RR16481 and P20RR16481S1. The content of th manucrpt are olely the reponblty of the author and do not repreent the offcal vew of NRR or NIH. Reference: [1] Eyre-Walker A, J Mol Evol : 442 [PMID: ]. [2] Ikemura T & Wada K, Nuclec Acd Re : 4333 [PMID: ]. [3] Grantham R et al. Nuclec Acd Re : r49 [PMID: ]. [4] Karln S & Mrazek J, J Mol Bol : 459 [PMID: ]. [5] Akah H. Genetc : 927 [PMID: ]. [6] Bulmer M. Nuclec Acd Re : 2869 [PMID: ]. [7] Lavner Y & Kotlar D, Gene : 127[PMID: ]. [8] Urruta AO & Hurt LD, Genetc : 1191 [PMID: ]. [9] Urruta AO & Hurt LD, Genome Re : 2260 [PMID: ]. [10] Hana R & Wada A, J Mol Evol : 321 [PMID: ]. [11] Taz J & Brd A, ell : 909 [PMID: ]. [12] Sharp PM & Mata G, urr Opn Genet Dev : 851[PMID: ]. [13] Anderon SG & Kurland G, Mcrobol Rev : 198 [PMID: ]. [14] hamary JV & Hurt LD, Genome Bol : R75 [PMID: ]. [15] Smth TF et al. Mol Bol Evol : 390 [PMID: ]. [16] Mahd RN & Rouchka E, Proc IEEE Int Symp Sgnal Proc Inf Tech : 29 [PMID: ]. [17] Hag D & Hurt LD, J Mol Evol : 412 [PMID: ]. [18] Freeland SJ & Hurt LD, J Mol Evol : 238 [PMID: ]. [19] Gl D et al..genome Bol : RESEARH0049 [PMID: ]. [20] Arndt PF et al.j omput Bol : 313 [PMID: ]. [21] de Magalhae JP et al.. Agng ell : 65 [PMID: ]. [22] Saxonov S et al. Nuclec Acd Re :185 [PMID: ]. Edted by P Kangueane taton: Mahd & Rouchka, Bonformaton 8(3): (2012) Lcene tatement: Th an open-acce artcle, whch permt unretrcted ue, dtrbuton, and reproducton n any medum, for non-commercal purpoe, provded the orgnal author and ource are credted. ISSN (onlne) (prnt) Bonformaton 8(3): (2012) Bomedcal Informatc

5 open acce Supplementry materal: Methodology: Bufferng capacty Bufferng capacty refer to the tendency of a codng equence to allow for pont mutaton event that reult n a equence of the ame or mlar amno acd, a oppoed to mene or nonene mutaton. In order to quantfy th property, we have developed a mathematcal model that approxmate the bufferng capacty for an ndvdual equence by takng nto account the codon compoton, ndvdual nucleotde mutaton rate, and a dtance matrx between amno acd that reflect the reultng damage caued by ndvdual amno acd ubttuton. For a gven DNA or mrna codng equence S = 1 23, n= 3 m, where each a nuclec acd redue and m the number of codon, the bufferng capacty etmated by: n 4 1 ( ) = (,,, ) (, ) n (1) m = 1 B S P L R D In (1), (,,, ) nucleotde of P L R the probablty that the nucleotde repectvely n the gven equence S. replaced by ( )., D the dtance between the two codon wll mutate to the codon n whch the nucleotde and gven that L and occur whle R are the left and the rght the ame codon where Here, we defne the dtance between two codon a a meaure of the dfference between the encoded amno acd and to approxmate th dtance, we ue the amno acd mlarty matrx propoed n [19] a follow: ( ) D (, ) ( ) ( ), ( ) max Sm Sm A A In (2), A( ) and ( ) = (2) A are the amno acd encoded by the codon and repectvely. max( Sm ) the maxmum mlarty value n the ame mlarty matrx. If the new codon a top codon (nonene mutaton), the dtance wll be condered three tme the hghet dtance gven by (2). Note that other value for the nonene mutaton dtance produce mlar reult (not hown). A an approxmaton of the neghbor dependent mutaton probablty (,,, ) functon of tme perod t and proportonal to the relatve neghbor dependent mutaton rate (,,, ) P L R, we aume the probablty dependent on a R L R. Snce we are comparng natural codng agant random DNA, replacng the mutaton probablte by the relatve mng perod mutaton rate hould have no effect on the gnfcance tet and the effect ze meaurement nce both tattcal meaure are cale ndependent Table 1: Lfe expectancy and exual maturty rate Spece Lfe expectancy (year) frut fly zebrafh moue 3 42 human Sexual maturty (day) Table 2: Number of mrna equence ued Spece mrna frut fly 2,479 zebrafh 1,294 moue 9,476 human 16,016 Table 3: Bufferng capacty dtrbuton effect ze Spece Natural v. Natural v. random ontraned random frut fly zebrafh moue human ISSN (onlne) (prnt) Bonformaton 8(3): (2012) Bomedcal Informatc