What type of plasmid are you using? I had similar problem with some of my "bacteriotoxic" proteins. What I have done is I added 0.5-1% glucose on agar-plates and in the medium for on culture as well as for expression (in case of IPTG-inducible system). I also tried auto-induction medium. It does not work if your over produced protein makes a great impact on E. coli physiology. In my case, I was dealing with proteins hindering translation by hindering ribosomes during initiation stage.

Could you please tell us more about your plasmid, strain, as well as characteristics and possible toxicity of your over expressed gene?

What else has its influence on bacteria physiology is how you autoclaved your LB. According to my experience, you get different sugar composition in your LB upon autoclaving depending on the size of your autoclave, its occupancy with other bottles and the content of those bottles.

Yep tell us more about your plasmid!! I had inconsistent proteins expression using DHalpha cells but worked okay when i switched to BL21 cells. The former is bad for experiments involving expression!!

its a pET plasmid, we tried different strains, JM109(DE3), BL21(DE3) and others, BL21(DE3) seems to work better - the protein is part of a bacteriophage fibre

@Maggy. DH5alpha, TOP10, XL1-blue are RecA deficient and good for molecular cloning but not for protein over production. Whereas BL21, Tuner, Rosetta, etc., are E. coli B straing lacking proteases Lon and OmpT. Also, one should not forget about DE3-lysogen when dealing with T7-driven constructs.

A good way to see if you have a toxicity is to make a OD measurement through time once you have induced. If it turns out that overexpression of your protein is toxic for the cell, you can try other strains which are more stringent as pLysS or pLysE, in an extreme case C41 or C43. You can also optimize induction by decreasing the temperature (18-20 degrees) or use a lighter promoter (arabinose) or decrease concentration of IPTG. But one would need more details about the plasmids, the strains and the potential toxicity of the gene.

Check you competent cells, do the transormation without plasmid. It seems like you get some cantamination cell,

Try to grow the transformed culture from the very begining at 30Grad or even lower (for instance 27Grad)

I agree with Ziguo. Either cell or plasmid contamination.

Does this happen only with this protein? or you have this problem with other unrelated proteins? I would suspect that your protein is toxic, but there might be a technical problem somewhere...

@Mark. You mention that your protein comes from bacteriophage. Is it E. coli's pathogen? If yes, you should think about using other expression systems (in vitro translation or different bacteria species).

I also can recommend Tuner (DE3) strain where you can "tune" the over expression varying IPTG concentrations starting from as little as 10µM IPTG.

Also, keep in mind that bacteria will try its best to neutralize any toxic construct inside by mutating various parts of your construct, i.e., promoter, thus resulting in irreproducible results of over expression. You MUST add at least 0.8-1% Glucose on agar-plates and in the media in order to shut down leaky expression.

Sorry for unstructured answer. I just try to help. I ahve got a lot of experience in expression of proteins which are very toxic for E. coli but not with bacteriophages.

@Konstantin true that!!

hi there...

1. same question....do u see just the overexpression after induction using whole cell lysate?

2. what is the size of protein?

3. I hope u have checked the sequence of ur clone...no stop codons...right?

4. there may be stop codon in ur primer (fp)..it will not let ur protein to be expressed...for obvious reasons!!

5. rest of ur system looks fine

Some more details:

- the protein sizes is 50 kD, the construct was sequenced and found ok, we have gotten very good soluble expression a couple of times, using apparently exactly the same conditions.

- the protein expressed is the right protein because it behaves as expected (it has a very specific biophysical property) and crystallised in a spacegroup compatible with the size and symmetry of the protein, the protein is not a lysin or bacteria-killing enzyme

- will try the glucose in the plates and other suggestions

Dear Mark - you have my sympathies - and if anyone can provide a correct answer to your issue - I would like to know it!

We have experienced similar things in the past - i.e. huge variability of expression levels of the same protein in the same cells.

None of the suggestions above are the answer - to either your observations, or ours.

More thoughts please - this is not an uncommon problem.

On other thing that wasn't mentioned is what is the age of your culture cells? I had a similar problem with sf9 cells once they reached a certain passage. I am not sure if that applies to e. Coli cultures but I know their competency reduces overtime.

Hi, there is a paper dealing with high yield expression of proteins in E. coli: Sivashanmugam et al, DOI: 10.1002/pro.102. They describe a double-selection procedure, in which they select for stably expressing clones. Maybe this would work for you. Cheers, Magdalena

Hi Mark,

I second Lynne. We have similar issues with a plant protein that we express in E. coli. Expression levels are completely irreproducible. Similar to what you have described, we tried a lot of things to figure out what's going on, but to no avail. What we ended up doing was to check ~10 colonies after transformation in small scale for expression levels. Some colonies would generate no protein at all (although they carried the plasmid based on their antibiotic resistance), while others would express large amounts. Once we have identified the good guys, we move on to large scale expression. From a scientific point of view this may not be completely satisfying, but at the end of the day we cared more whether we have protein or not.

I am also a total fan of using glucose in my plates upon outgrowth after transformations as well as in my overnight pre-cultures the day before induction day. Have you waited to induce at a higher OD and cooler temps? Everything you can do to avoid selecting for the poor expressing guys helps.

Yes, different temperatures were also tried, best expression was actually observed at 37 degrees, but, as said, not reproducibly

I also had this issue and used Rosetta strains that seemed to solve the issue as it has the pRARE codon. I also used auto-induction medium and BugBuster to extract the protein without it getting damaged. Hope your problem gets solved.

See Meena Ali's response. Is the insert codon optimized?

+ this is a common feature in a lot of proteins. where some colonies produce more protein then others. One thing you could do is test a number of different colonies and make a glycerol stock from a colony you know expresses the protein well.

I'm the one doing the experiments (Mark's student). We already tried Rosetta, but I couldn't see expression. I also tried Rossetta 2.

I thought the expression was better when I was using autoinduction media, but sometimes the expression improves with LB. Yesterday I tried the same colony in both and the best was LB.

The best one was the first time I expressed the protein. I was not in my lab, so maybe some of you are right and how we prepare the LB it's important.

First of all, I will try again with glucose. But all ideas are welcome!

Variability in protein expression from a colony to the other is very common. I strongly suggest doing the double selection protocol at least the first time you do the expression and then keep the glycerol stock of the colony that you select and to use it to do a single selection experiment. When you do selection from a freshly transformed plate, it is very common to find clones that don't show expression at all, as other said.

I will add something of "very low level" compared my predecessors:

to make a smear on LB+antibiotic from the broth where the DNA expressing the protein was isolated. This why recombinant colonies in the tranformation plate can be contaminated with empty E.coli, which overgrow during long time culture.

How about a colony blot? http://www.qiagen.com/literature/BenchGuide/pdf/1017778_BenchGuide_chap_4.pdf#71

A variation of this was published in Nature Methods (if I remember well they just use freeze-thaw cycles instead of SDS to break the cells so they can check for soluble expression instead of total). http://www.nature.com/nmeth/journal/v2/n7/full/nmeth767.html http://www.nature.com/nprot/journal/v1/n1/full/nprot.2006.39.html

The general idea being that not all colonies express at the same levels, so if you pick a high expressing one from the beginning you have better chances.

From personal experience, what you need to be careful with is not to have bacterial leftovers on the membrane as they may absorb the western part reagents. I have seen great variations (high expressing, low expressing and not-at-all expressing colonies). Oh, and of course you need to have a plate that is not very dense, but that's a good idea always, anyway. Good luck!

Maybe it's just a stupid question from a grad student but... if I have a pure plasmid and I transform competent cells with it... why not all the colonies are expressing the protein? I think it should be always the same, and it's why I always prepare a new transformation for each purification.

I can speculate that maybe it's a matter of the plasmid copy number, but really? Not a clue!

Have you checked to be certain that the media you are using is of the same composition? We've had issues with certain brands/compositions of Bactotryptone and yeast extract greatly affecting expressions, thanks to our "helpful" store providing the cheapest, but not necessarily the best media components.

I have also faced same problem where some colonies expressed more protein than others. Therefore we first used to check the expression at small level and take that colony which is expressing maximum protein for large scale induction and inocuation.

Hi Mark,

I would first test the toxicity of the expression system on plate using the inducer of the system. IPTG ? If cells do not grow in presence of the inducer, then it is likely that the plasmid is unstable even before induction. This you can check by counting the cells on plate with or without antibiotic just before to induce. If so then you have to down regulated you expression system. If you are in T7 then try C41, C43 or BL21AI from invitrogene. All the best, Bruno Miroux

Hi Mark and Carmela,

I'm not sure my post will make you go ahead... i agree with the fact that, in theory Carmela you're wright : transformation of competent bacteria (from the same colony) should give several identical clones, but... that's theory. After transformation, you have to check the level of expression in several colonies (small scale), then you choose one and make a stock of it ! In my opinion transform each time is a loss of time and too tricky.

But sometimes even with cells stocks, you can also have bad surprises : it happens with me few months ago. In that case, check everything :

Did you have some troubles with water system ? (think of everything you prepare with it)

Did you use different batch of LB ? IPTG ? antibiotics ?

Did you change the concentration of stocks (IPTG, antibiotics again) ?

Is someone new making these stocks ?

It can looks like paranoïa, but... In my case it worked (finally it was a combination of all the points i mentioned).

Hope you will find out what's going wrong !

Best regards

This is for Carmela

the theory says that ampicillin does not kill E.coli cells, just stop their growth . In a plate after a transformation with AMP- R plasmids you still have alive cells without plasmid, have you never seen satellite colonies? If this is not a problem when you need DNA plasmid because the low DNA yield can be compensated by a larger plasmid preparation, in protein expression the presence of E.coli cells without plasmid make the difference in deciding if the protein is expressed or not. During a long term culture of E. coli the AMP is inactivated consequently cells without plasmid overgrowth

I am having the same problem.. I have tried everything, but nothing has worked. My protein has a signal peptide that may be causing this. Does yours have signal peptide?

If you use ampicillin: I found that adding the antibiotics before induction and in the evening (if you do an over night expression) increases the yield a lot. This is probably because amp is degraded after few hours by resistant cells. I also found that using carbenicillin instead increases the expression level.

I agree with Ian

no dogma in science!

I agree with most of the suggestions above and suspect that the problem may simply be variations in the media composition...or variations in the water supply used to make up the media. However, I learned a trick many years ago regarding getting reproducible expression from a plasmid wherein the protein product is toxic to E. coli. In short, you take one colony from your fresh transformation and place it in 1 liter of medium to which glucose was added (to repress expression). Then after an overnight growth, you spin out the cells and resuspend in fresh medium (without glucose but with IPTG). This approach can work well for toxic proteins. The original suggestion came from someone working in industry, but we have published our version (Jaffe et al., JBC, 275:2619). We routinely do our first growth at 37 and our expression at 15...but this is protein specific.

It seems glucose helps somewhat - still not clear why we are not getting the great expression obtained twice before...

In any case thanks to all for the insightful responses

To Marina, Mark and Carmela, the common reason why independent transformants may yield different results is that there is a strong selection against expression. Most current expression systems are sufficiently physiologically stressful that mutants which abolish or reduce expression have a strong growth advantage. Hence when you grow up a culture, especially an overnight culture, faster growing variants will arise in the population. Adding glucose (assuming a lac based expression system) often reduces expression and hence reduces the selective pressure.