Hi there,

Do you get nothing at all from your expression experiment in Rosetta? If yes it shows that overexpression of rare tRNAs in E. coli is not enough and therefore codon optimization could help. If you get something from your first expression experiment it might be worth trying optimizing culture condition in a first move in order to get more product.

Hi Matthew,

Can you expand on 'without much success'? We need to know what you do get e.g. low levels by western or coomassie, incomplete translation/truncations, something insoluble when you were anticipating a soluble protein etc?

I agree with Dominique that there could be many very simple optimization steps that should be investigated before resorting to codon optimisation (that can ironically in a small number of cases make matters worse!). So please let us know what fusion protein, (detailed) culture conditions etc. you have tried so far.

Do you if the problem is just rare codons ? it could be also mRNA folding which could influence the expression. Have you checked that? other factors include the first residue after methionine which  needs to be checked.  Rosettagami is a bit of a difficult strain and optimizing the expression of tRNAs might be difficult  as you don't have control over it, a better way of doing it is by  gene optimized sequence so it worth a shot.

In my experience with proteins that are difficult to express in E.coli codon optimization never helped... I agree that optimization of culture condition may be more useful.

In my experience, making generalisations never helps for protein expression issues. Codon optimisation rarely makes things worse in terms of expression level but it solves some issue often enough to make it worth testing. The only disadvantage might be the burden on your budget.

It would be important to clarify what your construct looks like. A "particular protein with a number of rare codons" and "without much success" is not very helpful. Does that mean you had no improvement in Rosetta compared to another strain or did you get no expression in Rosetta either after you tested another strain?

If you have the native start sequence at the 5'-end and no tag or fusion partner, low or lack of expression might be caused by non-optimal structure of the start region for the initiation of translation (=rate limiting step in protein synthesis). If that is the case, either N-terminal tagging or silent mutations in the region of the first 5-10 amino acids might help but multiple permutations would need to be tested (e.g. see PMID: 12034488).

If you have a N-terminal tag or fusion partner and you still don't see any expression of your target protein then it might be more a toxicity or degradation issue than a rare codon issue. Using a different expression system might be more promising in that case.

Optimizing culturing conditions (almost) never helps when you are working with the wrong construct.

Indeed, in my case it was more a toxicity problem, in fact the cells started to die after induction. The problem wasn't solved by changing rare codons. Instead, I was able to obtain a bit higher expression levels by inducing at high OD. But every protein is a story apart and generalzation never helps.

Thanks for all your commments. The protein is expressed from pet15b (with an N-terminal His-tag). The family of proteins of which the protein is a member, is notoriously difficult to work with in vitro. The main problem is insolubility but if we could increase expression levels it may "bypass" the issue. Culture conditions have been explored and there is apparently no difference with standard E.coli/Rosetta strains. Since the protein is so short, codon optimisation is in fact quite inexpensive so I am going to give this a go. My question really was a philosophical one and I know very well that there is no single rule for successful overexpression.

Hi,

Have you tried auto-induction? This can make the difference between an almost completely insoluble product and a workable soluble yield. Might be worth a try with your existing construct, after that I would be tempted to try different constructs with fusion proteins before I optimised the sequence with the same N-His tag. Our current favourites are Trx, Z-tag and SUMO-all available as N-his constructs so there is no need to change your primary purification step and, depending on the usage you may not need to cleave the fusion off before use.

Also it may be worth adding a little Triton or Tween 20 (0.2-1%) as occasionally, depending on both the protocol and the lysis buffer, apparently insoluble proteins can be easily extracted from your 'insoluble' fraction with a bit of detergent.

Hi,

I am about to do a generalisation, but sometimes the presence of very rare codons (that then should be in the 7 provided by rosetta) is not as much of a problem as the presence of rare codons clusters. In the latter case, you have a region of the gene that contains low occurrence codons that are not necessarily complemented by the rosetta plasmid. this then can lead to localised slow-downs/stops in the traduction, and thus immobilization of a part of the protein outside of the ribosome. this part could then either be degraded by proteases, or misfold (by itself or with other immobilized domains from neighbouring ribosomes), or even immobilize the ribosome for so long that it detaches and interrupts the synthesis, leading to a truncated protein.

In order to analyse this clustering effect, there is a nice tool online that makes a sliding window analysis of your sequence, instead of a simple codon-by-codon. This really helps to visualize the presence of that problem.

http://www.codons.org/

(by Clarke T.F. & Clark P.L. (2008) Rare Codons Cluster, PLoS ONE 3: e3412)

also, in "my" lab, my colleagues have had unexplainable lacks of production for a few proteins, even with codon-optimised plasmids. It eventually turned out to be a mRNA stability issue and thus using Star cells allowed a good expression. maybe worth considering.

There can also be a problem of protein toxicity, leading to early loss of plasmid due to leaky expression and thus death of plasmid-containing cells. Then a pLysS cell type (or adding an excess of glucose before induction, or the auto-induction medium mentioned by Nick Berrow) can help as well. Or different growth temperatures/times.

I have also read that among other things, making a clone double selection can help a lot, but I don't have a strong experience with that technique yet (Arun Sivashanmugam, Protein Sci. 2009 May; 18(5): 936–948. doi: 10.1002/pro.102)

good luck!