Isochores, GC content, Codon Bias

varies among species: bacteria, plants, invertebrates

little variation among vertebrates

high veriation/ GC heterogenity within vertabrates:

• vertebrate genome can be divided into isochores = long stretches (100s of kb) of DNA with uniform %GC.

• long stretches of DNA with uniform %GC in vertabrates

• 100s of kb, sometimes >10 Mb

• High %GC = heavy isochores (H)

  • H1 (42-47%)

  • H2(47-52%)

  • H3(>52%)

• Low %GC = light isochores (L)

  • L1 (<37% GC)

  • L2 (37-42%)

    
    

    
    

Heavy isochores are found only in warm-blooded vertebrates (mammals, birds)

not in cold-blooded vertebrates (fish).

Selectionist hypothesis

• GC-pairing is stronger than AT-pairing (3 vs. 2 hydrogen bonds)

  -> may stabilize DNA at higher temperatures.

• Support: heavy isochores found in warm-blooded vertebrates

• Furthermore, heavy isochores are gene-rich.

Mutationist hypothesis

• The pool of available nucleotides changes over replication

  (which takes 8 or more hours for mammals).

  -> There is more GC available early in replication, so mutations will be biased towards G or C

• support: Over time, regions of the genome that replicate early

• non-random bio-chemical errors during replication and repair

Analysis 2005: By doing a sliding window analysis of the human genome DNA sequence and defining Isochores as segments >300 kb with distinct %GC and low heterogeneity:

• isochores covered 41% of the human genome

• most had low %GC

• four-family model with mean GC contents of

• 35%, 38%, 41%, and 48%

  
  

genomic regions are GC rich

Coding regions > introns > 5’ flanking regions > 3’ flanking regions

all of the synonymous codons for a particular amino acid are not used with equal frequency as would be expected at random

• The preferred codons correspond to the most abundant tRNA in each species, suggesting that selection favors the use of codons that increase the level of gene expression

Leucine: very high codon bias:

• Leu can be encoded by 6 different codons:

  • CTG, CTA, CTC, CTT, TTG, TTA.

• at radom: 1/6 (17%) of the time/ per codon

• In highly expressed E. coli genes, CTG is used ≈90% of the time.

• In yeast, TTG is used ≈90% of the time.

ENC = effective number of codons

• the average number of codons that are used to encode the

  20 amino acids

• lies btw 20 and 61

• Low ENC = high codon bias.

  
  

Fop = frequency of optimal codons

• the frequency with which the “optimal” codon is used for

  each amino acid

• Optimal codons def: the one used with the highest frequency in highly expressed genes.

• High Fop = high codon bias

Fop is species-specific

-> requires that optimal codons are known. For this, one must have many gene sequences and expression information.

ENC

can be applied to any species without prior knowledge of expression or codon usage.

Some observed patterns of codon bias:

• in highly-expressed genes

• in short genes compared to long genes

• in female-expressed compared to male-expressed genes

• laveal expressed genes

• Selection

• Mutation (neutral)

• use of optimal codons (those that correspond to the most abundant tRNA) make translation faster and more accurate.

• Codon bias could be used as a way to regulate gene expression post-transcriptionally

evidence:

• Highly expressed genes have higher codon bias

• Conserved protein motifs, such as DNA binding domains, have higher bias than other protein regions.

  (This suggests selection for accuracy of translation)

• Experimental replacement of optimal codons with non-optimal codons reduces the level of protein.

• Replacing sub-optimal leucine codons in the Adh gene with optimal codons increases ADH enzymatic activity in larvae, but decreases it in adults.

  -> This suggests that there may be a trade-off between optimal codon usage (or other factors) in different developmental stages. Codon bias is highest for larval-expressed genes.

Drosophila Alcohol dehydrogenase (ADH)

optimal leucine codons -> non-optimal codons

=> leads to a lower level of ADH protein in vivo and reduces ethanol tolerance in adult flies

Wa-F = wild-type, optimal leucine codons

1 leu = 1 luecine codon changed from optimal to non-optimal

6 leu = 6 leucine codons changed from optimal to non-optimal

10 leu = 10 luecine codons changed from optimal to non-optimal

In a comparison of ADH protein concentration, it was found that:

Wa-F > 1 leu > 6 leu > 10 leu

wild-type flies: more tolerant to ethanol that the mutant flies.

The LD50 = the ethanol concentration at which 50% of the flies were killed within 24 hours

The LD50 wild-type flies: 9%

for 10 leu flies: It was only 7.5%

Mutation bias specific bases?

eg:

• if mutations from A or T to G or C are more frequent and no selection on synonymous sites:

  -> should become GC-rich.

• The reverse bias in mutation would lead to synonymous sites being AT rich

= this would lead to non-random codon usage

• Most optimal codons end in G or C.

  ->hypothesis requires mutational bias towards G or C.

• But most observations: bias is towards A or T,

• But biased mismatch repair in favor of G or C.

  = strongest in regions of high recombination (where most of the highly codon-biased genes are located).

  could explain at least part of the observed codon bias

• The GC content at third codon positions is correlated with local genomic GC content, suggesting an effect of mutation of codon usage.

  
  

• The strength of selection acting on a particular codon is expected to be very weak (N*s=1 where N pop size, s selection coefficient )

  -> often very difficult to distinguish between selective and neutral explanations for codon bias.

• Typically, there is evidence for selection affecting codon usage in organisms with small genomes and large population sizes (bacteria, yeast, Drosophila).

• evidence is much weaker in humans and other vertebrates, where it appears that mutational/repair biases can explain the observed codon usage.