Analyse statistique des séquences génomiques

DEA en bioinformatique

Lausanne, 16 mai 2000Laurent Duret

duret@biomserv.univ-lyon1.fr

Plan

• Taille des génomes, paradoxe de la valeur C

• Contenu informationnel

• Séquences répétées

• Organisation en isochore des génomes de vertébrés

• Régions régulatrices non-codantes

• Usage des codons synonymes

What is a genome ?1911 - gene:

Elementary unit, responsible for the transmission of hereditary characters

1920 - genome:

Set of genes of an organism

1944 - Avery et al.

DNA is the molecule of heredity

1950-70 :

Double helix

Genetic code

Mycoplasma genitalium0,6 Mb

Genomes sizes E. coli 4,7 Mb

YeastS. cerevisiae

13,5 Mb

NematodeC. elegans

100 Mb

ProtozoaAmoeba dubia

700 000 Mb

PufferfishFugu

rubripes400 Mb

Man3400 Mb

Xenopuslaevis

3100 Mb

AmphibiaNewt

100 000 Mb

DipnoiLungfish

150 000 Mb

ProkaryotesEukaryotes

13%2%85%Drosophila85%2%13%E. coli70%2%28% YeastS. cerevisiae

5%2%93%Man27%2%71%NematodeC. elegansProportion of functional elements within genomes

2%98%0,01%Lunfish (dipnoi)Coding (protein)RNANon-coding

How many genes in the human genome ?Fields et al. 1994

Technique Gene estimate Comments/assumptions

Estimation 100,000 if average size = 30 kb

Estimation 300,000 if average size = 10 kb

Transcribed genome 20,000

RNA reassociation 20,000 - 92,000

Genomic sequencing (1994) 71,000 biased toward gene-rich region?

CpG islands 67,000 assumes 66% human genes have CpG islands

EST analysis (1994) 64,000 matching with GenBank; 50% EST redundancy

Chromosome 22 (1999) 45,000 correction for high gene density on chrom. 22

Chromosome 21 (2000) + 22 40,000

Functional elements in the human genome

3.4 109 nt 50,000-100,000 protein-coding genes

81% no known function43%38%introns4%12%protein-coding regions

centromeres, telomeres,

RNA2%intergenic

Untranslated RNAs: Xist, H19, His-1, bic, etc.

Regulatory elements: promoters, enhancers, etc.

2% ??

Structure of human protein genes

• 1396 complete human genes (exons + introns) from GenBank (1999)

• Average size (25%, 75%)

– Gene 15 kb ± 23 kb (4, 16) (10% > 35 kb)

– CDS 1300 nt ± 1200 (600, 1500)

– Exon (coding) 200 nt ± 180 (110, 200)

– Intron 1800 nt ± 3000 (500, 2000)

– 5'UTR 210 nt (Pesole et al. 1999)

– 3'UTR 740 nt (Pesole et al. 1999)

• Intron/exon

– Number of introns: 6 ±3 introns / kb CDS

– Introns / (introns + CDS): 80%

– 5' introns in 15% of genes (more ?), 3 ’introns very rare

• Alternative splicing in more than 30% of human genes (Hanke et al. 1999)

Structure of human protein genes

• GenBank: bias towards short genes

• 1396 complete human genes (exons + introns)

≤949596979899Publication date48121620Gene size (coding exons+introns) kb

5101520253035≤949596979899Publication dateGene size (coding exons+introns) kb

Structure of human protein genes

• GenBank: bias towards short genes

• 1396 complete human genes (exons + introns)

• 9268 complete human mRNA

Sequence:cDNA

complete gene (exons+introns)

400800120016002000889092949698Average CDS size (nt)Publication date

Overlapping genesN-myc geneN-cym geneIA IB II IIIIII II IPOL II

promoter

polyadenylationsite

transcriptionmaturation5'pm7GpppAAAAAAAAAAsmall nucleolar

RNA

mRNAoverlapping protein genessmall nucleolar RNA genes within intronsof protein genes

Tandem repeatsRepeated sequencesInterspersed repeatsDNA transposons retroelementsRecombination- DNA polymerase slippage:

CACACACACACA

- unequal crossing-over

motif bloc size % humangenome

satellite: 2-2000 nt up to 10 Mb 10%minisatellite: 2-64 nt 100-20,000 bp ?microsatellite: 1-6 nt 10-100 bp 2%

ADN satellite: centromères

CTTCGTTGGAAACGGGAsatellite α (171 )pb

(17 )site CenpB pb répétitions de satelliteα

répétitions d'ordre, supérieur spécifiques

de chaque chromosome( 10 )jusqu'à Mb

centromèrechromosome

Reverse transcriptase:RetroelementsNucleusCellRNADNAtranscriptionreverse transcriptionintegrationLINEs (long interspersed elements): 6-8 kbretroposons

SINEs (short interspersed elements):80-300 bpsmall-RNA-derived retrosequences (tRNA), pol III

Endogenous Retroviruses: 5-10 kb

LTR gag pol env LTRRetrovirusRetrotransposonRetroposonRetroséquenceRetrovirus

NucleusCellLINE reverse transcriptaseRetrosequences:opportunist retroelements

reverse transcriptionDNARNALINERNART protein

Pseudogenes1 - gene duplication generepeated elementinequal crossing-over2 - gene inactivation mutationA - pseudogenes derived from recombinationB - retropseudogenesgenepromoterAAAAAAtranscription + maturationmRNADNAretrotranscription + integrationAAAAAADNA

Retropseudogènes

• 23,000 à 33,000 retropseudogènes dans le génome humain

• Les gènes qui génèrent des retropseudogènes sont généralement de type housekeeping

• Gonçalves et al. 2000

Fréquence des éléments transposables dans le génome humain

• Total = 42% (Smit 1999)

0%4%8%12%AluLINE1MIRLINE2LTR elementsDNAtranposon

Fréquence des éléments transposables dans le génome humain (Smit 1999)

0%5%10%15%20%25%30%AluLINE1AutosomesChromosome X

Isochore organization of vertebrate genomes

Organisation en isochore des génomes de vertébrés: mise en évidence expérimentale

densité à l'équilibre, g/cm31,6901,7001,71030%41%51%satellitetaux C+Gcomposantmajeur

Fractionnement du génome de la souris par centrifugation en gradient de densité (Bernardi et al. 1976)

Analyse statistique des séquences publiées dans les banques de données.

Corrélation entre la composition en base en position 3 des codons et celle de l'envirronement génomique dans lequel se trouve le gène

Analyse statistique des séquences publiées dans les banques de données.

Distribution en fréquence des gènes dans les différentes classes d'isochores

CDS GC3%

0

1

2

3

4

5

6

7

0 20 40 60 80 100

Moy = .612Ecart-t = .1585447 séq

0

1

2

3

4

5

6

7

0 20 40 60 80 100

Moy = .639Ecart-t = .171818 séq

Homme Poulet

Nb

de

gèn

es (

%)

0

2

4

6

8

10

12

0 20 40 60 80 100

Moy = .509Ecart-t = .106703 séq

Xénope0 20 40 60 80 100

Danio

Moy = .580Ecart-t = .103173 séq

0

2

4

6

8

10

12

14

Evolution de la structure en isochore chez les vertébrés

Isochore organization of vertebrate genomes

• Insertion of repeated sequences (A. Smit 1996)

• Recombination frequency (Eyre-Walker 1993)

• Chromosome banding (Saccone, 1993)

• Replication timing (Bernardi, 1998)

• Gene density (Mouchiroud, 1991)

• Gene expression ?? -> No

• Gene structure (Duret, 1995)

isochore %C+G % total genomic DNA

L1+L2 : 33%-44% 62 %

H1+H2 : 44%-51% 31%

H3 : 51%-60% 3-5%

H1+H2L1+L2H3H1+H2L1+L2L1+L2>300 kbBernardi et al. 1985

Isochores and insertion of repeat sequences (Smit 1999)

4%8%12%16%20%AluLINE-1LTR-

elements

Density in repeat sequencesG+C content of genomic sequence:G+C < 39%G+C > 47%G+C 39%-47%

4419 human genomic sequences > 50 kb4419 human genomic sequences > 50 kb

Isochores and gene density

MHC locus (3.6 Mb) MHC locus (3.6 Mb) (The MHC sequencing consortium 1999)(The MHC sequencing consortium 1999)

Class I, class II (H1-H2 isochores): 20 genes/Mb, many pseudogenesClass I, class II (H1-H2 isochores): 20 genes/Mb, many pseudogenesClass III (H3 isochore): 84 genes/Mb, no pseudogeneClass III (H3 isochore): 84 genes/Mb, no pseudogene

Class II boundaries correlate with switching of replication timingClass II boundaries correlate with switching of replication timing

isochore % total genomic DNA %total genes

L1+L2 : 62 % 31%

H1+H2 : 31% 39%

H3 : 3-5% 30%

2060100140Number of genes / MbL1+L2H1+H2H3Mouchiroud et al. 1991

Isochores and introns length

• 760 complete human genes• L1L2: intron G+C content < 46%• H1H2: intron G+C content 46-54%• H3: intron G+C content >54%

Average intron length (bp)Gene compaction (intron length/coding region length)40080012001600200024681012L1L2H1H2H3L1L2H1H2H3

Duret, Mouchiroud and Gautier, 1995

Prediction of functional elements (1) Ab initio methods

Ruled-based or statistical methods e.g.: protein genes prediction, promoter prediction, … Very useful but ...

Limits in sensibility/specificity No method available for many functional elements (non-coding RNA genes, regulatory elements, …)

Large scale transcriptome projects: ESTs, full-length cDNA Identification of transcribed genes (protein or non-coding RNA) Information on alternative splicing, polyadenylation (Hanke et al. 1999, Gautheret et al. 1998),

expression pattern Very useful but ...

Problems with genes expressed at low level, narrow tissue distribution, stage-specific expression, … Limited tissue sampling Artifacts in ESTs (introns, partially matured RNA, …) Limited to polyadenylated RNA

Prediction of functional elements (2) Comparative sequence analysis (phylogenetic footprinting)

Function => selective pressure

Corollary Sequence conservation = selective pressure = function

provided the number of aligned homologous sequences represents enough evolutionary time for the accumulation of mutations at the less constrained (presumably selectively neutral)

base positions.

• Evolutionary rate in non-functional DNA: ~ 0.3% / My (± 0.069)

Man/Mouse: ~ 80 Myrs 46-58% identity

Mammals/Birds: ~ 300 Myr 26-28% identity

Random sequences 25% identity

Analyse comparative des gènes de -actine de l'homme et de la carpe

CarpeHomme5’UTR 3’UTR site polyA échelle de similarité: pas de similarité significative70 - 80% identité80 - 90% identitérégions codantes: éléments régulateurs:introns:ATGcodon stop

Phylogenetic footprinting Advantages

Works for all kinds of functional elements (transcribed or not, coding or not) as far as the information is in the primary sequence

Does not require any a priori knowledge of the functional elements

Limits Absence of evolutionary conservation does not mean absence of function No efficient method to detect unknown conserved secondary structure in RNA Function, but what function ? Depends on the sequencing status of other genomes

Human, mouse, fugu, C. elegans, drosophila, yeast, A. thaliana

Number of sequences to compare : > 200 Myrs of evolution Mammals/birds: 310 Myrs Human + mouse + bovine : 240 Myrs

Prédiction de régions régulatrices

• Méthodes ab initio

– Prédiction de promoteurs

– Îlots CpG

• Approche comparative

Prédiction de promoteurs eucaryotes

• Combinaison de sites de fixation de facteur de transcription (ordre, orientation, distance)

• Motifs courts, dégénérés– Difficile de distinguer les vrais sites des faux positifs:– Motif à 4 bases: ≈1/256 pb (1/128 pb sur les deux brins)

• Boîtes TATA, CAAT , GC: absents dans beaucoup de promoteurs

• Banques de données de sites de fixation de facteurs de transcription (TRANSFAC), de promoteurs caractérisés expérimentalement (EPD)

• PromoterScan (Prestridge 1995): Mesure de la densité en sites potentiels de fixation de facteurs de transcription de long de la séquence (pondération en fonction de la fréquence des sites dans ou en dehors des vrais promoteurs)

Prédiction de promoteurs: sensibilité, spécificité

• Sensibilité: fraction des promoteurs qui sont trouvés par le logiciel

– PromoterScan: sensibilité = 70% (promoteurs à boîte TATA)

• Spécificité: fraction des vrais promoteurs parmi ceux qui ont été prédits

– PromoterScan: spécificité = 20%– Un faux positif / 10 kb

• Génome humain: ≈100 000 gènes, ≈1 promoteur/30 kb

sensibilité=vrais_ positifs

vrais_ positifs+faux_ négatifs

spécificité=vrais_ positifs

vrais_ positifs+faux_ positifs

SpécificitéSensibilité1000Seuil

Prédiction de promoteurs eucaryotes: recherches en cours

• Prise en compte de l'orientation relative et des distances entre sites de fixation de facteurs de transcription– COMPEL (Kolchanov 1998): banque de données d'éléments composites– FastM : recherche dans une séquence génomique d'une combinaison de deux sites

de fixation de facteurs de transcription à une distance définie l'un de l'autre

• Recherche de corrélations entre sites– PromoterInspector (Werner 2000)

• Sensibilité: 40%• Spécificité: 45%

http://www.gsf.de/biodv/index.html

• Combinaison recherche ab initio / approche comparative: recherche de sites potentiels parmi les régions conservées

Îlots CpG• Génome de vertébrés :

– méthylation des C dans les dinucléotides 5 ’-CG-3 ’(CpG)

• Me-C fortement mutable -> T5 ’-CG- 3 ’ 5 ’-TG-3 ’ 5 ’-CA-3 ’

3 ’-GC- 5 ’ 3 ’-AC-5 ’ 3 ’-GT-5 ’

• Génome des vertébrés: globalement dépourvu en CpG (excès de TG, CA)

• Certaines régions (200 nt à plusieurs kb) échappent à la méthylation– Pas de déplétion en CpG: CpGo/e proche de 1

– Riche en G+C

– Îlot CpG:Longueur > 500 nt

CpGo/e > 0.6

G+C > 50%

ouou

CpGo /e =Nombre_ de_ CpG_ observéNombre_ de_CpG_ attendu

=0.25

01CpGo/e

La déamination des cytosines

Cytosine

N

C

CH

C

NH

NH2

O

CH

réparation

Cytosine

N

C

CH

C

NH

NH2

O

CH

Cytosine méthylée

N

C

CH

C

NH

NH2

O

CH

CH3

Uracile

HN

C

CH

C

NH

O

CH

O

déamination

Thymine

HN

C

CH

C

NH

O

CH

O

CH3

déamination TpGou

CpA

Îlots CpG: associé aux régions promotrices ?

• Bird (1986), Gardiner-Garden (1987) Larsen (1992) ref– 40% des gènes tissu-spécifiques possèdent un îlot CpG en 5 ’

– 100% des gènes ‘ housekeeping ’ possèdent un îlot CpG en 5 ’

• Rechercher des îlots CpG pour prédire des régions promotrices ?– Sensibilité: 40-100%

– Spécificité ?? (Quelle fraction des îlots CpG correspond effectivement à des régions promotrices ?)

• Ponger (1999): comparaison des îlot CpG qui recouvre ou non le site d ’initiation de la transcription

Fréquence des gènes humains avec un îlot CpG recouvrant le site d ’initiation de la transcription

– 800 gènes humains avec promoteur décrit

– Mesure de la distribution tissulaire à l ’aide d ’EST (20 tissus)

0%20%40%60%80%0-23-67-20Nombre de tissus où le gène est exprimé

Comparaison des îlots CpG recouvrant ou non le site d ’initiation de la transcription

– 272 îlots start CpG recouvrant le site d ’initiation de la transcription (start)

– 1078 îlots CpG en dehors d ’un promoteur connu (other) (en excluant les séquences répétées)

0 2460.50.70.91.11.350%60%70%80%otherstartG+C%CpGo/eLongueur(kb)

Recherche de régions régulatrices par analyse comparative (empreintes phylogénétiques)

• Goodman et al. 1988: régulation de l’expression des gènes du cluster -globine au cours du développement

• Alignement de séquences orthologues de 6 mammifères (> 270 Ma d’évolution)

• 13 empreintes phylogénétiques: ≥ 6 nt, conservation 100%

• Analyse par retard de bande sur gel:

• 12/13 (92%) correspondent à des sites de fixation de protéines

• 1996: 35 empreintes phylogénétiques avec protéines fixatrices identifiées

• Enhancers de gènes HOX (Fugu/souris) (Aparicio et al. 1995)

• enhancer TCR α (homme/souris) (Luo, 1998)

• promoteur COX5B (11 primates) (Bachman, 1996)

• promoteur uPAR (homme/souris) (Soravia, 1995)

Large scale phylogenetic footprinting

Non-coding sequences : 325,247 sequences 145 Mb

everything except protein-coding regions and structural RNA genes (rRNA, tRNA, snRNA, scRNA)

Introns, 5' and 3' untranslated regions, intergenic sequences

Filtering of microsatellite repeats and cloning vectors: XBLAST

Similarity search: BLASTN + LFASTA

Vertebrates, insects, nematode

Metazoan Genome ProjectsMillion yearsPorifera (sponge)Nematodes (C. elegans)Arthropods (Drosophila)EchinodermsUrochordataCephalochordata (amphioxus)Jawless fisheschondrichthyes (ray, shark)actinopterygii (bony fishes)amphibians mammals birds reptiles600400200800VertebratesSequencing effort: 9 to 100 Mb 0.8 to 2.4 Mb less than 0.2 Mb

Sequence Similarities1- Identification of new genes

protein-genes, RNA-genes: intronic snoRNA genes

2- Retroviral elements, retrotransposons

3- Low complexity sequences:

GC-rich, AT-rich, cryptic microsatellites

4- Artefacts:

annotation errors, sample contamination (sponge insulin, ascidian RNA, chicken TGFB1)

5- 326 highly conserved regions (HCRs)

- do not code for proteins

- do not correspond to any known structural RNA

326 Highly Conserved Regions (HCRs)

• > 70% identity over 50 to 2000 nt after more than 300 Myrs

• Unique sequences

• Generally specific of only one gene

• Longest HCR:

84% identity over 1930 nt after 300 Myrs

3’UTR deltaEF1 transcription factor

• Oldest HCRs: 500 to 600 Myrs

• No HCR between vertebrates and insects or nematode

Oldest HCRsMillion yearsPorifera (sponge)Nematodes (C. elegans)Arthropods (Drosophila)EchinodermsUrochordataCephalochordata (amphioxus)Jawless fisheschondrichthyes (ray, shark)actinopterygii (bony fishes)amphibians mammals birds reptiles600400200800Sequencing effort: 9 to 100 Mb 0.8 to 2.4 Mb less than 0.2 MbHistone 3’UTRα- actin3’UTR

3 5’HOX UTRVertebrates

Conservation pattern in 3’UTRsposition relative to the stop codon (nt)10005000150020002400c-fosTransferrin receptorbirdmammalEndoplasmic-reticulum Ca2+ ATPase birdmammalbirdmammalsimilarity: <60% ≥60% ≥70% ≥80%

Distribution of HCRs within genes3'-non-coding5'-non-codingintrons0%10%20%30%40%mammals / birdsmammals / amphibiansmammals / bony fishes2841917296512563812 Frequency of orthologous

genes containing HCRs

HCRs and multigenic familiesHistone replacement variant H3.3A0400600100014001800AAAAAAAAUGStopAUGStopAAAAAAAHistone replacement variant H3.3BHistone replacement variant H3.3A and H3.3B, Calmodulinsnt• several genes coding for a same protein

• non-coding sequences are distinct, and conserved

Function of 3’HCRs: mRNA stability, translation

A+U-rich element: stability, translationposition relative to the stop codon (nt)10005000150020002400c-fosTransferrin receptorbirdmammalbirdmammalsimilarity: <60% ≥60% ≥70% ≥80%IRE : Iron Responsive Element

IRP : Iron Regulatory Protein

CCAGUGN5'3'

Function of 3’HCRs:mRNA subcellular localization

Myosin heavy chain, c-myc, vimentin, -actin

chickencarp (bony fish)site poly(A)site poly(A)0200400600800position relative to the stop codon (nt)localization signalssimilarity: <60% ≥60% ≥70% ≥80%- 3’actin UTR

Comparaison des régions non-codantes de 77 gènes orthologues homme/souris (Jareborg et al. 1999)

0.20.40.60.81Upstream(1 kb)5’UTRIntrons3’UTRs

Upstream015’UTRcoding exon

intron3’UTR

Fraction des régions non-codantes conservées entre homme et souris

Synonymous codon usage61 codons, 20 amino-acids: genetic code degeneracy

Synonymous codon usage is not random: some codons are preferentially used

Synonymous codon usage bias

0.0%25.0%50.0%75.0%Example: Proline codon usage in Escherichia coli (4300 genes)

Mycoplasmacapricolum

Micrococcusluteus

Saccharomyces cerevisiae

Caenorhabditis elegans

Mus musculus0%25%50%75%100%CCUCCCCCACCGExample: proline codon usage in different speciesMST serine/threonine kinaseInositol 5-phosphatase0%25%50%75%CCU

CCC

CCA

CCG

Example: proline codon usage in different human genesSYNONYMOUS CODON USAGE BIAS VARIES ...... ACCORDING TO THE SPECIES... ACCORDING TO GENES WITHIN A SAME GENOME

HOW TO EXPLAIN BIASES IN SYNONYMOUS CODON USAGE ?SELECTIONIST AND NEUTRALIST HYPOTHESES

Selection of codons that are optimal for translation

(speed or accuracy)

Mutational biasCodon usage bias is correlated with gene expression level

Prefered codons correspond to the most abundant tRNAs

Selective pressure on synonymous sites => synonymous substitution rate is negatively corelated with codon usage bias

Examples: E. coli, B. subtilis, yeast

No relationship with gene expression level

Mutational biases affect all positions within a genome (not only synonymous codon position) => correlation between codon usage and whole genome base composition

Examples: - Borrelia burgdorferi:

whole genome : 29% G+C3rd codon position : 21% G+Cother positions: 32% G+C

- Mycobacterium tuberculosis: whole genome : 66% G+C3rd codon position : 79% G+Cother positions: 60% G+C

Selection-mutation-drift model

Mise en évidence d’une pression de sélection sur le choix des codons synonymes chez des organismes

pluricellulaires

• Corrélation entre biais d’usage des codons synoymes et niveau d’expression des gènes

• C. elegans • D. melanogaster • A. thaliana

• Corrélation entre l’abondance des ARNt et l’usage des codons synoymes

• D. melanogaster (White et al.1973, Moriyama & Powell 1997) • C. elegans

Expressed Sequence Tags (ESTs)Inventory of mRNAs expressed by a given organism in different tissues (or cell cultures), development stages, sexes, in response to stimuli, pathologies, etc.

Procedure• cDNA libraries

- polyA+ selection- normalization: reduce the abundance of

over-expressed mRNA

• automatic sequencing from 3' or 5' end, single read

Result• partial mRNA sequences (300-500 nt)• high error rate (3-6%)• high redundancy

In Silico Northern Blot• probe = mRNA sequence (or coding-region, when only the genomic sequence is known)

• filtering of repeated elements (microsatellites, Alu, LINEs, etc.)

• similarity search (BLASTN2) against a species-specific EST collection

• threshold for positive matches : ≥ 95% identity over 100 nt or more

- low enough to tolerate sequencing errors- high enough to distinguish different members of a gene family

• we only retained cDNA libraries that had been sufficiently sampled (> 10.000 ESTs)• cell culture, pathologies, unidentified cDNA libraries were not considered

number of ESTs number of different tissues/stages

Human 802,246 22/3C. elegans 67,987 2/2A. thaliana 36,200 1Drosophila 27,491 2/2

Expression pattern

• tissue distribution: tissue-specific vs. housekeeping

• expression level: number of EST matches / total number of ESTs in that cDNA library

No EST detected low moderate high

0.350.400.450.500.550.600.350.400.550.600.650.70Caenorhabditis elegansDrosophila melanogasterArabidopsis thalianaprotein length< 333 aa333-570 aa> 570 aaFavError bars indicate the 95% confidence interval.mRNA abundance:Frequency of favored codons (Fav),

protein length and expression level

No ESTLowInterme-diateHighExpressionIntron G+C contentD. melanogasterN = 439A. thalianaN = 2386C. elegansN = 7891

Mutational bias or selection ?In Drosophila, C. elegans and A. thaliana most of favored codons end in C or G.

Does codon usage reflect a mutational bias towards G or C in highly expressed genes ?

0.300.400.500.60both in embryo and adultonly in adultonly in embryoNo ESTprotein length< 333 aa333-570 aa> 570 aaExpression during development :FavCaenorhabditis elegansFrequency of favored codons (Fav), protein length and expression during

development in C. elegans.

Favored codons are the same in embryo- and adult- specific genes

Mise en évidence d’une pression de sélection sur le choix des codons synonymes chez des organismes

pluricellulaires

• Corrélation entre biais d’usage des codons synoymes et niveau d’expression des gènes

• C. elegans • D. melanogaster • A. thaliana

• Corrélation entre l’abondance des ARNt et l’usage des codons synoymes

• D. melanogaster (White et al.1973, Moriyama & Powell 1997) • C. elegans

Isoaccepting tRNA gene copy numberFrequency of amino-acids (weighted according to protein

expression levels)

515253545R = 0.82p < 0.00010%2%4%6%8%ACDEFGHILMNPQRSTVWYK Relationship between the number of tRNA genes and amino-acid usage in

C. elegans genome

580 tRNA genes

Example: proline

tRNA Codon

CGG CCG

GGG CCC

AGG CCT

TGG CCA

Relationship between favored codons and tRNA abundance in C. elegans genome

Anticodon-codon recognition(Wooble):

Anticodon CodonA UC GG C, UU A, GI U, C, (A)

A’s at first anticodon position are predicted as I’s

In all cases, favored codons are decoded by the most frequent tRNA gene

Selection on codon usage =selection for higher translation efficiency

- elongation rate

- cost of proofreading

=> no relationship between codon usage and gene length

- translation accuracy

the energetic cost of translation and the probability of misincorporation increase with protein length:

=> stronger selective pressure on codon usage in long genes => positive correlation between codon usage and gene length

E. coli (Eyre-Walker, 1996; Moriyama & Powell, 1998)

Correlation between protein length and Fav (a) or Favmax (b) in C. elegans highly expressed genes

number of codons (log scale) 0.2 0.4 0.6 0.8 25 80 250 800 2500 8000 R = -0.60 p<0.0001 Fav number of codons (log scale) 0.2 0.4 0.6 0.8 25 80 250 800 2500 8000 R = -0.42 p<0.0001 Favmax (a) (b) Favmax = maximum Fav value along coding sequences, using a sliding window of 150 codons, moved by steps of 3 codons. Variation of codon usage along genes ?

Mutations and evolutionBase modificationReplication error

etc.

MutationDNA repairs < 0 : Negative selection|Ns| << 1 : Random genetic drift s > 0 :Positive selectionFixationselective advantage or disadvantage (s)Population (N)somatictransmission to the offspringIndividual|Ns| << 1 : rate of evolution = rate of mutationgermlineno transmission to the offspringallele lossPolymorphismRate of mutation:etc. Rate of molecular evolution

= f(mutation rate, probability of fixation)

environmentsensibility to mutagensfidelity of polymerasesefficiency of DNA repair

Mammals: small population size:(~104 i.e. 100 to 1000 times smaller than drosophila)

the small selective differences between alternative synonymous codons is not sufficient to overcome random genetic drift

C. elegans, A. thaliana , D. melanogaster: selection on synonymous codon usage (translation efficency)

Mammals: no evidence for selection on codon usage

Probability of fixation of a new mutation :

p = f(s, N)

s : selective advantage or disadvantage

N : population size

| Ns | << 1 : random genetic drift