It's not. Unless your virus is a new pathogen to the host cells.

I'm sure you've reviewed lots of papers, check this one: http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2933640/ they study codon bias in influenza virus. It may help or enlight your way searching for an explanation.

Interesting question. My first answer would also be "no" because it would lead to a lower translation efficiency and hence a hindrance for viral replication.

But: There is a concept called "survival of the flattest". Some viruses gain an "evolutionary" advantage by replicating on a low level because that allows them to stay under the radar of the immune system (especially in non-immunosupressed hosts). So it might be possible that your virus achieves just that by having a non-favorable codon usage?!

Great Q. It really depends. Papillomaviruses do prefer to use rare codons. The reason might be the lower level of viral protein expression leading to no/lower detection by host immune system. In fact, in our field we have to codon optimize the viral protein for expression in vitro.

It not normal, but I think that hepatitis A virus is an example of that case. Use of rare codons is thought to be used to slow protein synthesis to favor the correct folding of viral proteins.

HIV also differently uses codons than human cells.

For more information, you can read:

Codon and amino acid usage in retroviral genomes is consistent with virus-specific nucleotide pressure. AIDS Res Hum Retroviruses. 2002 Jan 20;18(2):133-41.

http://www.ncbi.nlm.nih.gov/pubmed/11839146

Some viruses can infect to several different kinds of hosts. For example, influenza can infect to birds and various mammals. Dengue virus infect to insects and mammals. Therefore, I am not surprised at differences of codon usages.

In addition, there is room to discuss how strictly codon usage regulates translation. You could fine rare codons on host major proteins.

It depends on the virus. For example, lentiviruses are well-known for preferring to use rare codons (that's why most lentiviral genes are not well-expressed in transgenic system unless they are codon-optimized). Indeed, our innate immune system has apparently evolved an anti-viral ISG, Schlafen11 (see Nature. 2012 Nov 1;491(7422):125-8. doi: 10.1038/nature11433. Epub 2012 Sep 23), that specifically inhibits viral protein synthesis (for HIV) by means of codon-bias discrimination. SLFN11 binds to tRNAs of the bias codons but the mechanism remains unclear at present. Many other viruses, especially in specific genes, also prefer to use rare codons . . . depends on what is the fittest strategy for the reproductive cycle of the virus in question. We have had experience in codon-optimizing many viral genes which enhances their protein production, but many codon-optmization procedures also take into account removal of cryptic splice sites, which are sometimes required for the viruses themselves to maintain different ORFs. So, the question is complicated beyond "What is normal?" It is a matter of their conjoined evolutionary histories.

Thank you for you answers. It makes sense @Tobias and @Hang that it may serve as a "stealth" tactics of the virus to avoid detection.Good point. @Jordi thank you for your comment. I never expected that correct protein folding has something to do with translation efficiency. @fushimi great! that could also be the reason why DENV infects monkeys, a fact that remained unanswered for me for quite sometime. @Lee wow! I think that is an interesting area for research. That was the explanation I should have added in my research in DENV. Thank you for your answers! your explanations were all good and I really appreciate your efforts to share. cheers.

Another consequence of the codon bias may result in a change of the protein folding with all its implications. The preferences of certain codons, linked to tRNA availability, would affect the translation speed with pauses being indicated by low-usage codon series. Pauses and full speed translation may produce conformational changes in the newly synthetized protein. This was suspected while expressing some antigens from malaria parasites in prokaryotic expression systems with dismal results. We successfully used a “codon harmonization” algorithm developed at WRAIR to match the codon usage frequencies from the parasite (native host). The harmonization algorithm method is described here: http://www.ncbi.nlm.nih.gov/pubmed/21125377

Thank you @grawe. That was very helpful.

@Bastian: Do the effects you are mentioning also affect "standard" heterologous expression of non viral proteins (e.g. mammalian proteins in E. coli)? Will a standard codon usage optimization (i.e. avoiding rare codons), as offered for gene synthesis by various companies, affect protein stability and/or folding?

@all: are you by chance aware of any tool / algorithm to detect any of such pitfalls when designing sequences for heterologous expression?

Viruses are known for their high variability at the nucleotide level. For a same virus it exists multiple strains, and at each replication, new mutations are succeptible to appear. This is a strategy for the virus to escape host immune system and to survive. Influenza virus is the most famous example.

On the other way, your virus can have different host, and could be better adaptated to one compared to the others.

So I think it is not surprising to find such differences in codon usage.

Codon usage affects translation rate and folding. Producing a protein too rapidly may lead to incorrect folding and aggregation of viral protein products. It is still likely that host codon usage is important. If your virus is very different from the host it may indicate your virus is regulating some of its protein synthesis through translational control.

@Bastian "Sorry for this long answer!" :) Thanks! I just wonder how many of these nice options and concerns are applied when you hit the "optimize" button at any of these synthetic gene companies' websites, as they do not disclose their algorithms, of course. They also do not give you a warning if that special restriction site that you insist to have will be some sort of signal in the expression host that you have selected.