https://www.abbs.info e-mail:[email protected] ISSN 0582-9879                             ACTA BIOCHIMICA et BIOPHYSICA SINICA 2003, 35(6): 580-586                             CN 31-1300/Q

Short Communication

Factors
Affecting Codon Usage in Yersinia pestis

HOU Zhuo-Cheng, YANG Ning*

( College of Animal Science and
Technology, China Agricultural University, Beijing 100094, China )

Abstract        The
complete genome of Yersinia pestis which was the causative agent of the
systemic invasive infectious disease classically referred as plague, had been
recently sequenced. In order to have a further insight into the synonymous
codon usage evolution, factors shaping synonymous codon usage pattern of Yersinia
pestis were analyzed in this paper. The coding sequences larger than or
equal to 300 bp were used in codon usage analysis. Though “G”+“C” content in Y.
pestis genome was slightly lower (47.64%), the highly expressed genes
tended to use “C” or “G” at synonymous sites compared with lowly expressed
genes. Conversely, lowly expressed genes tended to prefer “A” or “T” at
synonymous positions. Gene expression level was strongly correlated with the
first axis of the correspondence analysis (COA) (R=0.63, P<0.0001). By the analyses of the codon usage pattern of highly and lowly expressed genes, it was confirmed that gene expression level was partially responsible for the codon usage bias. GC-skew analysis showed that codon usage suffered replication-transcriptional selection. Codon adaptation index (CAI), frequency of “C”+“G” at the synonymous third position of codon (GC3s) and the effective number of codons (Nc) values showed some differences among different gene length groups. “G”+“C” content of genes was strongly correlated with the first axis of the COA (R=0.72, P<0.0001). It could be concluded that gene expressivity, replication-transcriptional selection, gene length and gene composition constraints were the main affecting factors of codon usage variation in Y. pestis.

Key words     codon usage;
correspondence analysis; gene expression level; coding sequence length; Yersinia
pestis

The fast-growing
data of genomes give us new opportunities to study genome evolution on the
molecular level. It is well known that codon usage pattern is nonrandom and
species-specific, and the inter-genomic variation of the codon usage pattern is
a widespread phenomenon. There were also some reports that different genes have
different codon usage patterns in a same organism[1]. Biased codon usage of
codons might be influenced by various factors, such as translational
selection[2], mutation[3], compositional constraints[4], physical location of
the gene on chromosome[5], replication-translational selection[6],
hydrophobicity of each gene[7], etc. In Y. pestis,  the analysis of the codon usage pattern
intrigued researchers greatly, because it was essential for studying major
codon evolution, predicting ORF, and designing primers for PCR.

Yersinia pestis, a Gram-negative bacterium, had been considered as the causative
agent of the systemic invasive infectious disease classically referred as
plague, and had been responsible for three human pandemics: the Justinian
plague, the Black Death, and modern plague. The complete genome of Y. pestis
had been recently published[8]. Many genes in the Y. pestis genome seem
to have been acquired from other bacteria and viruses. There are also evidences
that Y. pestis has undergone large-scale genetic flux. Y. pestis
provides a unique insight into the ways in which new and highly virulent
pathogens evolve.

In this paper,
the Y. pestis codon usage pattern and the main factors that influence
the codon usage of Y. pestis were analysed by using the whole genome
datasets. The aim of this study was to facilitate the further study on
codon  evolution, ORF prediction,
and primers designing.

1    Materials
and Methods

The complete DNA
sequences of the Y. pestis genome were downloaded from the Sanger Center
(ftp://ftp.ebi.ac.uk/pub/databases/embl/genomes/Bact-eria/ypestis―CO92). The
length of all used coding region sequences is equal or greater than 300 bp. The
149 pseudogenes and 3 plasmids found in the Y. pestis genome were
excluded from our datasets. 3444 genes (coding region sequences) were totally
analyzed in this study. The coding sequences from the complete genome were
retrieved with a program developed in our lab
(ftp://202.205.81.236/download/soft/applying software/CDsRead).

Relative
synonymous codon usage(RSCU)[9], the effective number of codons(Nc)[10],
frequency of “C” + “G” at the synonymous third position of codon (GC3s) and
correspondence analysis (COA) were calculated by using the program CodonW1.3
[written and provided by Dr. John Peden (Oxford University), see
http://molbiol.ox.ac.uk/cu]. A3s, T3s, G3s, C3s were the distributions of “A”,
“T”, “G” and “C” at the synonymous third position of codons, respectively.
Codon adaptation index(CAI)[11] was calculated by using genes encoding the
ribosomal proteins and elongation factors as the referenced dataset (totally,
71 genes). CAI value had been proved to be the best gene expression theory
value and had been extensively used as a measure of gene expression
level[4,6,7,12]. In this study, CAI value was used as a presumed expression
level. Higher CAI value meant higher codon usage bias and higher gene
expression level[11].

2    Results

2.1   Genome
and gene composition constraint analysis

The “G”+“C”
content could be one of the most important factors in the evolution of genomic
structures[13]. The genome of Y. pestis was slightly compositionally
biased, since its “G”+“C” content was 47.64%. The GC3s values of genes ranged
from 16.5% to 69.8%, with a mean of 47.08% and standard deviation of 7.6%. The
Nc values of different genes in Y. pestis ranged from 28.07 to 61, with
a mean of 50.82 and standard deviation of 4.47. Wright[10] suggested that a plot
of Nc against GC3s could be effectively used in explaining the codon usage
variations among the genes. This method had been used to investigate the
evolution of many genomes[6,14]. If the codon usage of a gene had not suffered
from “G”+“C” composition constraints and natural selection, the Nc value of the
gene would fall on the continuous Nc-plot curve. In Y.pestis, it was found that
although there were a small number of genes lied on the Nc-Plot curve, the Nc
values of most genes fell below the expected Nc-plot curve (Fig.1), which
indicating that compositional constraints had some effects on the codon usage
among the most genes.

Fig.1       Nc-plot of Y.
pestis genes

The
continuous curve represented the expected curve between GC3s and Nc under
random codon usage.

2.2   The
relationship among gene length, gene expression and codon bias

It had been
considered that energetically costly longer genes had higher codon usage bias
to maximize translation efficiently. Selection might also be acting to reduce
the size of highly expressed genes, and the effect was particularly pronounced
in eukaryotes[15]. In this paper, we classified genes according to the gene
length into 5 groups (length≤500 bp, 500 bp－999 bp, 1000 bp－1999 bp, 2000 bp－2999 bp, >3000 bp), and then examined the effects of gene length
on gene expression level, Nc and GC3s for each group in Y. pestis.

Results of gene
length study showed that the shorter the gene length the higher the Nc value
(Table 1). However, the analyses of the GC3s and CAI presented different
results (Table 1). The shorter gene length (length≤500 bp) had the lower GC3s compared
with the longer gene lengths (length≥1000 bp). As to the CAI, it expressed a continuous variation in some
degree, but appeared increasing with length totally. The longer length resulted
in the higher CAI which meant the larger codon usage bias and higher expression
level. In total, genes with length longer than 3000 bp had lower Nc, larger
GC3s, larger codon usage bias and higher expression level (Table 1).

The positions of
the genes along the first axis were also significantly correlated with the gene
length (R=0.197, P<0.0001), and highly expressed genes were longer (Table 1). Similar results were also found in P.aeruginosa[12]. Longer genes were thought to impose constraints on the codon usage bias[15]. Powell & Moriyama[16] argued that the selective advantage for the speed of translation of an optimal codon would depend on the length of the translated message. If the gene length was longer, the effect of each individual mutation from non-optimal to optimal codons would be less effective to reduce the total time needed to translate the whole protein[17]. In Drosophila, longer coding regions had both a lower codon bias and higher synonymous substitution rates, and were affected less efficiently by selection[18]. Though the real reason was not quite clearly understood yet, the gene length played an important role in shaping the codon usage bias in Y. pestis.

Table 1   Comparision of Nc, GC3s and CAI
among different length groups

Group	Gene length (bp)	Number of observations	Means±Standard deviation
			Nc	GC3s	CAI
1	>3000	60	49.614±3.486b	0.533±0.079a	0.544±0.054b
2	2000�D2999	193	49.854±3.726b	0.513±0.070b	0.558±0.060a
3	1000�D1999	1201	50.574±3.783ab	0.506±0.067b	0.552±0.061ab
4	500�D999	1424	51.335±4.246a	0.484±0.072c	0.540±0.060b
5	<500	566	50.463±6.388ab	0.456±0.084d	0.539±0.082b

*There
existed significant difference between two groups in a same column if there was
no same superscripts letter (a, b, c, or d) between them (P<0.05).

2.3   Correspondence
analysis (COA)

Correspondence
analysis has been widely used to investigate codon usage patterns in different
species[6,7,12,14]. In the present study, we applied COA to RSCU values of each
gene to minimize effects of amino acid composition. The first axis generated by
the analysis represents 12.99% of the total variability, while the second axis
explains 8.67% and the third axis 5.56%. As the first axis explained only a
partial amount of variation of codon usage among the genes in this bacterial
genome, it was postulated that there were several major factors in shaping Y.
pestis gene codon usage. The first axis of the COA explained values far
less than other genomes studied previously[6,12], which suggests that in Y.
pestis the major trend in codon selection is not as strong as in other
species.

The correlation
coefficient between the position of genes along the first major axis against
GC3s, C3s, A3s and CAI were calculated (Fig.2). The first axis of the COA is
positively correlated with the GC3s (R=0.66, P<0.0001), C3s (R=0.58, P<0.0001), but has a negative correlation with the A3s(R=－0.70, P<0.0001). Furthermore, there is a high correlation between the position on the first axis of the COA and the gene expression level (CAI) (R=0.63, P<0.0001). In order to have a quantitative idea about the different codon usage between highly expressed genes and lowly expressed genes, a χ2 test was applied to compare the highest CAI value genes (151 genes) and the lowest CAI value genes (151 genes) (Table 2).

The presumed highly expressed sequences
tend to use C3s-rich and A3s-poor codes compared with lowly expressed genes.
The increment of C3s accompanies with a decrease in A3s, and the frequencies of
these two bases in third codon positions are negatively correlated (R=－0.45, P<0.001). We postulate that the first axis of the COA is determined by the gene expression level and the second axis of the COA might reflect the lowly expressed genes' effects. Although the “G”+“C” composition
of the genome is 47.64%, but the highly expressed genes tend to terminate with C-
or G- at synonymous position compared with lowly expressed genes. The CAI value
and GC3s also has a significantly correlation (R=0.18, P<0.001). These results support that the highly expressed genes tend to use C- or G- at synonymous positions compared with lowly expressed genes. It is also confirmed that gene expression level affects the codon usage. Codon usage of the highly expressed genes has suffered selective pressures at translation processing. On the contrary, most of the lowly expressed genes terminate with A- or T- at synonymous positions and this implies that the codon usage in lowly expressed genes has not suffered such severely selective pressures as in highly expressed ones.

The results showed
that the highly expressed genes displayed a pattern of codon usage that
differed from the lowly expressed genes in Y. pestis. This analysis
showed that 23 codons are used more frequently in highly expressed genes, while
other 30 codons are used more frequently in lowly expressed genes (Table 2).
There was a significant increment of “C” or “G” at the synonymous sites in
highly expressed genes compared with the lowly expressed genes, and only two
abnormal codons display “T” ending [Arg (CGT), Gly (GGT)]. The lowly expressed
genes use “A” or “T” at terminate sites more frequently than highly expressed
genes, except that two codons displayed “C” [Leu (CTC), Pro (CCC)] and two
other codons displayed “G”[Arg (AGG), Gly (GGG)] at ends. These results confirmed
that there was a bias towards “C” or “G” in the highly expressed genes, while
towards “A” or “T” in the lowly expressed genes at the synonymous positions. In
duets codon family, C-ending codon was preferred in the pyrimidine-ending codon
family; and G-ending codon was preferred in the purine-ending codon family
[exception Glu (GAG)]. In highly expressed genes, terminate codon (Ter) TAA
(134/151), TAG (9/151), TGA (7/151) were used in different frequencies.
Terminate codon TAA (74/151), TAG (27/151), TGA (50/151) were also used in
different frequencies in lowly expressed genes. Stop codon expressed the same
usage pattern. TAA was the most frequent stop codon among highly and lowly
expressed genes. In highly expressed genes stop codon used more biased usage pattern
than lowly expressed genes. TAA was the most popular stop codon in highly
expressed genes in B. burgdorferi[19], C. trachomatis[7], E. histolytica[17].
It was needed to investigate more genomes data so that to infer that TAA was
the most popular stop codon in highly expressed genes.

Fig.2       The
correlations between the first axis of the COA and GC3s, A3s, C3s, CAI

Table 2   Codon usage in highly and lowly
expressed genes in Y. pestis

AA, amino acid; N, number of codons;
RSCU, cumulative relative synonymous codon usage in 100 genes; Ter, terminate
codon. *P<0.01 or *P<0.05, in this codon was used significantly more often in the highly expressed genes. ##P<0.01 or #P<0.05, in this codon was used significantly more often in the lowly expressed genes. χ2 test was used. aHighly expressed gene group.bLowly expressed gene group.

2.4   Replication-transcription
and codon usage bias

The codon usage
might be selected at replication-transcriptional level[7,19]. As to several
prokaryotic genomes it had been shown that there was a change in the sign of
the skew near the origin of replication[20,21]. It had been reported that in
the leading strand there was an excess of “G” over “C”, and most of the genes
lied on the leading strand. So GC skew was generally used to locate the leading
and lagging strands of a prokaryotic organism. The GC skew of Y. pestis
was calculated along the genome sequences by taking a window of size 20 kb
(Fig.3). There is generally an excess of “C” over “G” between 800 kb and 1120
kb, while between 900 kb to 920 kb there is an excess of “G” over “C”. The
content of “G” is over that of “C” in the ranges between 2880 kb and 3120 kb,
and between 4000 kb and 4160 kb, which should be worth noting that the GC skew
displayed anomalies in this plot. The reason for this may be the genomic
rearrangement during growth of the organism[8]. We obtained 2014 (58.47%) genes
in the positive side of GC skew and 1430 (41.53%) genes in the negative side.
This excess of genes in the GC skew positive side was also noted previously in
several genomes[7] and probably implied the orientation of the genes. This
phenomena violated the neutral selection theory. Therefore, the transposition
of the lagging strand

Fig.3       GC skew
along the Y. pestis genome DNA

It was calculated using a sliding window of 20 kb. GC skew>0, meant
an excess of “G” over “C”; GC skew<0, meant an excess of “C” over “G”.genes to the leading strands would have a selective advantage. In highly expressed genes, the selective advantage of transposition to the leading strands was much more likely to overcome random genetic drift, and these genotypes became fixed more easily in the population[19]. So, it was postulated that the codon usage bias had suffered selection pressures on the replication-transcription level in Y. pestis.

3    Discussion

In prokaryote
genome, non-random codon usage is considered as the result of the mutation and
natural selection that acting on the translation level. Generally speaking, the
mutation is random and the probabilities of the third position to mutate to
other bases are the same. The major reason of affecting the ‘wobble’ position
is the natural selection acting on the translation level and then causes some
codon have higher frequencies in the genome and evolves into ‘optimal codons’.
More and more new factors are found to shape the codon usage when detailed
analysis on the different genome data.

The genome base
compositions affected the codon usage in E. histolytica genome[17]. Though the
C. reinhardtii[6] and Echinococcus spp.[22] genomes had high GC contents, there
were little evidences that the genome composition shaped the codon usages in
these two genomes. In D. melanogaster, the “G+C” content was uniformly higher
at silent sites in coding regions than in putatively neutrally evolving
introns[23] . C. elegans showed a weak, but statistically significant, negative
correlation between “G+C” content and gene expression levels[24]. In Y.
pestis genome, the GC content was 47.64%, but the ‘wobble’ position of the
codon tended to use “G” or “C” in the highly expressed genes. The highly
expressed genes also had the high GC content. The CAI value and GC3s also had a
significantly correlation(R=0.18, P<0.001). These results support that the highly expressed genes tend to use “C” or “G” at synonymous positions compared with lowly expressed genes. It was also confirmed that genes expression level affectd the codon usage in Y. pestis genome.

Apart from the
gene expression level and gene composition, the gene length also had played a
critical role in affecting Y. pestis codon usage and expresses some
differences compared with other genomes. In Drosophila[18] genome, longer genes
had lower codon usage bias. The natural selection had little effect on the
Drsosophila genome codon usage. But, the longer genes had higher expression
level and higher codon usage bias in P. aeruginosa[12] genome and S.
penumoniae[4] genome. In S.cerevisiae genome, gene length was positively
correlated with codon usage bias[25]. In this study, the moderate longer genes
(1000－2999 bp) had
higher expression level and higher codon usage bias. This indicated that
different genomes had different gene lengths which accommodated their own
genome best requirements. At least, there weren’t universal rules about gene
length and expression level in all genomes.

The analysis
results showed that gene expression level, gene composition, gene length and
replication-transcription selection were the main factors shaping the Y.
pestis codon usage. These factors couldn’t repel neutral theory or
mutation-selection-drift model, but neutral theory also couldn’t explain these
results perfectly. So, different genome codon usage should be analysed to
investigate the genome codon usage pattern. In theories, the genome evolution
forces could be studied, and in practices, these results could also be used to
redesign the gene or primers for PCR or transmit genes among different species.

References

1     Wada
K, Aota S, Tsuchiya R, Ishibashi F, Gojobori T, Ikemura T. Codon usage
tabulated from the GeneBank genetic sequence data. Nucleic Acids Res, 1990,
18(Suppl): 2367－2411

2     Bennetzen
JL, Hall BD. Codon selection in yeast. J Biol Chem, 1982, 257: 3026－3031

3     Sueoka
K, Kawanishi Y. DNA G+C content of the third codon position and codon usage
biases of human genes. Gene, 2000, 261: 53－62

4     Hou
ZC, Yang N. Analysis of factors shaping S. penumoniae codon usage. Acta
Genetica Sinica, 2002, 29(8): 747－752

5     Kerr
AR, Peden JF, Sharp PM. Systematic base composition variation around the genome
of Mycoplasma genitalium, but not Mycoplasma pneumoniae. Mol
Microbiol, 1997, 25: 1177－1179

6     Naya
H, Romero H, Carels N, Zavala A, Musto H. Translational selection shapes codon
usage in the GC-rich genomes of Chlamydomonas reinhardtii. FEBS Lett,
2001, 501: 127－130

7     Romero
H, Zavala A, Musto H. Codon usage in Chlamydia trachomatis is the result
of strand-specific mutational biases and a complex pattern of selective forces.
Nucleic Acids Res, 2000, 28: 2084－2090

8     Parkhill
J, Wern BW, Thomson NR, Titball RW, Holden MT, Prentice MB, Sebaihia M et al.
Genome sequence of Yersinia pestis, the causative agent of plague.
Nature, 2001, 413: 523－527

9     Sharp
PM, Tuohy TM, Mosurski KR. Codon usage in yeast: Cluster analysis clearly
differentiates highly and lowly expressed genes. Nucleic Acids Res, 1986, 14:
5125－5143

10    Wright
F. The ‘effective number of codons’ used in a gene. Gene, 1990, 87: 23－29

11    Sharp
PM, Li WH. The codon adaptation index─A measure of directional synonymous codon usage bias, and its
potential applications. Nucleic Acids Res, 1987, 15: 1281－1295

12    Gupta
SK, Ghosh TC. Gene expressivity is the main factor in dictating the codon usage
variation among the genes in Pseudomonas aeruginosa. Gene, 2001, 273: 63－70

13    Bellgard
M, Schibeci D, Trifonov E, Gojobori T. Early detection of G+C differences in
bacterial species inferred from the comparative analysis of the two completely
sequenced Helicobacter pylori strains. J Mol Evol, 2001, 53(4－5): 465－468

14    Jenkins
GM, Pagel M, Gould EA, de A Zanotto PM, Holmes EC. Evolution of base
composition and codon usage bias in the genus Flasvivirus. J Mol Evol, 2001,
52: 383－390

15    Moriyama
EN, Powell JR. Gene length and codon usage bias in Drosophila melanogaster,
Saccharomyces cerevisiae and Escherichia coli. Nucleic Acids Res,
1998, 26: 3188－3193

16    Powell
JR, Moriyama EN. Evolution of codon usage bias in Drosophila. Proc Natl
Acad Sci USA, 1997, 94: 7784－7790

17    Romero
H, Zavala A, Musto H. Compositional pressure and translational selection
determine codon usage in the extremely GC-poor unicellular eukaryote Entamoeba
histolytica. Gene, 2000, 242: 307－311

18    Comeron
JM, Kreitman M, Aguade M. Natural selection on synonymous sites correlates with
gene length and recombination in Drosophila. Genetics, 1999, 151: 239－249

19    McInerney
JO. Replicational and transcriptional selection on codon usage in Borrelia
burgdorferi. Proc Natl Acad Sci USA, 1998, 95: 10698－10703

20    Lobry
JR. Origin of replication of Mycoplasma genitalium. Science, 1996, 272:
745－746

21    Blattner
FR, Plunkett GM, Bloch CA, Perna NT, Burland V, Riley M, Collado-Vides J et al.
The complete genome sequence of Escherichia coli K-12. Science, 1997,
277: 1453－1474

22    Fernandez
V, Zavala A, Musto H. Evidence for translational selection in codon usage in Echinococcus
spp. Parasitology, 2001, 123(Pt2): 203－209

23    Kliman
RM, Hey J. The effects of mutation and natural selections on codon bias in the
genes of Drosophila. Genetics, 1994, 137(4): 1049－1056

24    Duret
L, Mouchiroud D. Expression pattern and, surprisingly, gene length shape codon
usage in Caenorhabditis, Drosophila, and Arabidopsis. Proc
Natl Acad Sci USA, 1999, 96: 4482－4487

25    Coghlan
A, Wolfe KH. Relationship of codon bias to mRNA concentration and protein
length in Saccharomyces cerevisiae. Yeast, 2000,16: 1131－1145

Received: December 26, 2002       Accepted:
April 2, 2003

This work was supported by the grants from
the National Science Fund for Distinguished Young Scholars (No. 30225032) and
the Teaching and Research Award Program for Outstanding Young Teachers in
Higher Education Institutions of Ministry of Education of China(No. 1999094)

*Corresponding author: Tel,
86-10-628927141; Fax, 86-10-62891351; e-mail, [email protected]