Hi Artem and Stephen,

Hmmm....while I went away to do some more cloning and to look at the sequencing results of one of the clones I just made for fusion protein production in bacteria, something healthy has been brewing at the ccp4bb end.. Something I may have inadvertently help brew  :)

I completely agree with all that you have both said. I do not dispute the value of tags in improving expression or solubility or in facilitating crystallization. Fusion tags are one of the most popular and one of the most successful tricks of the protein-production trade. To clarify, fusion protein tags are one of my first few things to try for any new clone.  

BUT...I recently had a similar problem to what Mo outlines -- human protein in E. coli, 'zero induction', wondering if the protein is toxic based on colony morphology and effect on cell growth, and all the good stuff that Mo describes. In my case, for my human protein:
1. Disorder-prediction and secondary-structure-prediction programs indicate stretches of disordered protein
2. After the usual tricks of bacterial expression with a His-tagged version (temp, IPTG variation, autoinduction, cells), I got 'zero induction'. Then I switched to larger fusion tags like those both of you mention and then I could see expressed protein, though there was only a small fraction of full-length protein and copious amounts of proteolysis and/or truncated products (mass spec, Western blot etc. etc.). 
3. When one tag didn't work, I switched to another. After switching a bunch of tags, I consistently faced the same kinds of problematic issues like persistent proteolytic degradation, persistent aggregation even when fusion tag was left intact, and hard-to-remove chaperone contamination during purification. I was unable to recover purified full-length protein and these issues could not be mitigated by buffer, pH optimization etc. during purification etc.
4. Literature indicates that the protein might function in a complex

In my case, I used all of the above information collectively to hypothesize that my target protein, under the given conditions, was likely poorly folded on its own. In this specific case, even before I determined whether to move to a different expression system or not, I did still have at least a few choices that I can think of:
a. To attempt co-expression with a protein binding partner 
b. To attempt co-expression in a chaperone-aided expression system
c. To resort to a synthetic gene 
(Chopping up the protein and going after individual domains was not a real option in my case.)

Is my example the norm or the exception? I leave that open. 

Even if my case is extremely rare, my original intention was to caution that there may come a time with a beast of a protein-- after a LOT has been attempted and tried-- when one has to start to quickly distinguish between a promising approach and a futile time-sink.

To me, it seems we concur and there is not much scope for a healthy debate :)!
Raji




On Feb 26, 2009, at 10:07 PM, Artem Evdokimov wrote:

I personally think that debate is healthy.
 
Having said this – I do not advertise one fusion partner over another – and I did not intend to overweight the MBP success in my previous message, it was just a familiar example. As I mentioned, SUMO also worked for us – perhaps not as frequently as MBP but there is no statistical significance in this because the number of studied cases was way under a thousand (which intuitively sounds like a useful number, but in fact even a thousand is probably too small to be meaningful).
 
Unlike choosing one brand of OS over another, the choice of fusion partner (or indeed whether to use or not to use one) is not an expensive one since cloning is generally very cheap and expression in E. coli is also extremely affordable. Therefore it is perfectly easy to try as many expression tricks as one desires. Having said this I also should mention that enough trials in E. coli eventually out-price a few trials in insect cells or other ‘higher order’ system so everyone should apply their own judgement as to when the aggregate cost advantage disappears. I don’t think that anyone would disagree that chances of successful expression of a eukaryotic protein in e.g. insect cells are considerably higher compared to the same odds in bacteria, but unfortunately so is the price both in terms of money and of labor.
 
Removal of fusion tags is a separate kettle of fish and certain guidelines can be drawn from experience: viral proteases such as TEV, TVMV, etc. are precise and somewhat slow (typical use ratio is 1:10-1:50). Thrombin is very fast and furious (typical use ratio 1:1000 to 1:5000) but it can sometimes cause undesired cutting if you have either additional sites or dirty protease (it’s amazing how much difference does thrombin purity make!). SUMO hydrolase seems to be pretty fast but in my hands it stubbornly refused to process quite a few protein fusions – which is probably about as statistically significant as Stephen’s experiences to the opposite J The menu of options is quite diverse so that everyone can find something they like.
 
So to recap – my key and only argument is that fusion with host proteins (whatever they might be) is a valid and experimentally successful method for salvaging expression of *some* difficult proteins in bacterial hosts. The issues associated with this method are also quite real – indeed, solubility enhancement is not equal to folding enhancement but there is enough evidence in the field that fusion proteins can provide more than just solubility and brute yield enhancement. The experimenter has to be savvy enough to figure out what’s going on quickly, not waste time trying to optimize a hopeless case, and decide whether to proceed with more trials or to go in a different direction.
 
Artem
 
---
When the Weasel comes to give New Year's greetings to the Chickens no good intentions are in his mind.
From: CCP4 bulletin board [mailto:[log in to unmask]] On Behalf Of Stephen Weeks
Sent: Thursday, February 26, 2009 9:31 PM
To: [log in to unmask]
Subject: Re: [ccp4bb] Off topic: Mammalian gene expression in E. coli
 
Hi all,
    Once again I seem to have managed  to kick up a minor debate on the bulletin board (Note to self no more posts on SUMO or Apple  :-[ ). With quite a few years of experience working with SUMO I feel I can safely state that it is a good enhancer of fusion protein production in E. coli. I am personally NOT convinced that it is a "solubility" enhancer like MBP or NusA but the fusions main benefit is it's easy  and specific removal. By default I do 24oC inductions in Bill Studiers fantastic auto-inducing media so I haven't really fought with solubility issues for a while.  We make and use our own hydrolase in the absence of any detergent (unless of course it the target protein requires it) , and I find it better than TEV and PreScission (admittedly I have not tried the new more soluble clones of the former). Typically we get 100 mg/L using autoinduction media which we dilute to 0.5 mg/ml in 50% glycerol buffer, salt and DTT, of which I'll use 100-200 ul for a fusion protein prep of 100 mg plus. 
  Addressing Mo's original question I shall restate my answer as: that it would be cheaper to stick his construct into any fusion vector he can lay his hands on before handing money over to the gene synthesizers to see if he can get detectable expression. Thinking downstream, if it works, you need to consider the expense of the removal of the fusion partner. Clones are available for TEV, PreScission (Rhinovirus 3c protease) and of course SUMO hydrolase ;-)   (plus there are few systems out there for removal of tags without a protease). No single fusion system is a  panacea for all our protein expression woes and stating a position on one is equivalent to choosing sides in the Mac vs. PC debate. (Actually I have an idea for an advert featuring SUMO,  the small and hip fusion partner and MBP, the dull old  and overweight workhorse).

Stephen 


[log in to unmask] wrote:
Hi,
 
I respectfully disagree with the doom&gloom feelings regarding fusion
proteins. In my not very limited experience, fusion proteins *can* fix
expression issues. Do they always work - of course not :) But there are
very few things in this field that work most of the time. Is it better to
try a fusion protein or to go into a higher-order expression system? If
you can afford it, usually higher order systems tend to work better. But
what if you cannot afford it?
 
Regarding precipitation upon cleavage - consider the example of PTPbeta
catalytic domain: this protein expresses very poorly on its own, however
it expresses extremely well with a His-MBP N-terminal fusion, and the
activity of the fusion protein is very high. If you cleave the protein in
'just buffer' then PTPbeta rapidly precipitates. Bad news, right? However
if you cleave the fusion in the presence of 0.1% BOG the protein stays
perfectly soluble and monomeric, concentrates to 15 mg/ml and produces
marvellous crystals (about six structures in the PDB). So - do not be too
quick to dismiss fusion proteins as a way to try and salvage your
desperate cases, especially if going to a different expression system is
hard for some reason.
 
Regarding SUMO - I have personally tested it on about 30-35 proteins. It
only worked for *one* - but it made the protein nice and soluble, and it
stayed soluble after cleavage (note - we do not use the SUMO-protease,
just regular protease sites).
 
Is ratio like that worht the trouble? You decide :)
 
Artem
  
Some thoughts about SUMO tags and fusion tags in general.
 
Fusion tags also follow the "Garbage In, Garbage Out" philosophy.
Yes, if for many of the reasons already hashed out extensively on
CCP4BB, one is dealing with lack of expression or miniscule
expression, often tagging the protein with a fusion/cleavable tag
does indeed bump up the expression and lead to 'improved solubility'.
Sometimes, it's very important to ask: improved solubility of what
though?
 
Everything that Phoebe describes, namely the chaperone contamination,
precipitation after cutting off tag etc., reeks of an intrinsically
misfolded/unstable/unhappy protein. My experience-- and those of many
others-- is that the fusion tag and fusion tag alone can only fix
little in cases: 1) when one observes lots of degradation of the
untagged protein, 2) where the untagged protein is made as an
intrinsically misfolded/unstable protein. In these cases, the carrier
protein then notoriously comes along for the ride in the soluble
fraction with the fusion/cleavable tag, initially giving the
impression of improved expression and improved solubility. Even then,
one might even see multiple degradation products with the tagged
expression product. Next, cleave the tag off in such a case and lo
and behold! all protein precipitates and you are back to square one.
 
I am not trying to discourage anyone from using fusion tags -- to
improve expression, solubility, crystallization etc. We all know of
many examples where fusion tags have worked wonders. I only caution
that if your favourite protein is intrinsically misfolded in a
particular expression system and then you have tried tagging a fusion/
cleavable tag onto the protein in the same expression system and you
observe all that Phoebe describes, perhaps it is time to bang your
head against a different wall now. In many difficult cases, I am
unaware that a fusion tag actually aids in the proper folding of a
carrier protein. I will not rule out this possibility but I do not
know that this is the general rule.
 
I have worked quite a bit with SUMO tags. As far as GST and SUMO tags
are concerned, I banged my head against the GST-tag and SUMO- tag
wall for my target protein for a frustrating while. I tried a His
tag, then a GST tag, then a SUMO tag. All had exactly the same
symptoms. In my case, clearly the problem lay with the carrier
problem but I was never allowed to conclude so.
 
Just my two cents, the worth of which will already have diminished by
the time you have read this email.
 
Raji
 
 
 
 
 
 
On Feb 26, 2009, at 11:30 AM, Phoebe Rice wrote:
 
    
We haven't tried SUMO, but had some frustrating results with
GST fusions.  They did improve expression and solubility - BUT
in one case the target protein precipitated immediately when
the tag was cleaved off, and resisted all attempts to bring it
back to life.  In another case, the fusion protein dragged
chaperones into the prep that were nearly impossible to get
rid of completely, thus ruining our ATPase assays.
 
Is SUMO, being smaller, less likely to drag such crud along
with it?
 
  Phoebe
 
 
---- Original message ----
      
Date: Wed, 25 Feb 2009 14:48:57 -0500
From: Mo Wong <[log in to unmask]>
Subject: Re: [ccp4bb] Off topic: Mammalian gene expression in
        
E. coli
      
To: [log in to unmask]
 
  Thanks to all who responded. Actually, this bulletin
  board is better for help with molecular biology than
  the molecular biology bulletin board I am subscribed
  to!
 
  On Tue, Feb 24, 2009 at 7:47 PM, Stephen Weeks
  <[log in to unmask]> wrote:
 
    Mo,
      Just to add my 50 cents, I didn't see any
    mention of the use of fusion proteins in your
    original post. GST, MBP or my personal, and
    completely biased, favourite SUMO (plus many more
    proteins) have been shown to enhance expression
    when fused to the amino terminus of a target
    protein. If you fear you have toxicity, simply
    tracking the OD600 pre and post induction normally
    tell you if this is happening. I've worked with
    proteins that basically baselined the cell growth
    upon induction and, as Artem stated, at least I
    knew my protein was being made albeit at very low
    levels.
 
    Stephen
 
     --
     Stephen Weeks, Ph. D.
     Drexel University College of Medicine
     Department of Biochemistry and Molecular Biology
     Room 10102 New College Building
     245 N. 15th St.
     Philadelphia, PA  19102
 
     Phone: (+) 215-762-7316
     Fax: (+) 215-762-4452
 
    Mo Wong wrote:
 
      I thought I'd post this to the CCP4bb, as
      judging by previous posts, it seems I could get
      some useful insight into my problem...
 
      This is question has probably been asked by
      people for a long as molecular biology has been
      around, but hopefully my question isn't a
      complete rehash of other peoples: I am trying to
      express a human protein in bacteria where the
      only modified amino acids are 3 phosphorylated
      serines. I’ve gone through the usual hoopla of
      trying to get it expressed in E. coli
      (Rosetta/Codon+ cells, varying IPTG, low
      temperature, etc). Sequencing confirms my insert
      is correct, but from coomassie gel inspection, I
      appear to get near zero induction (I need to do
      a Western to get a clearer assessment). I’ve
      heard about custom gene synthesis, and it
      appears Mr. Gene (https://www.mrgene.com/) would
      be a good avenue to look into as they optimize
      the ORF taking into account codon usage in E.
      coli (though I’m not sure they examine
      putative mRNA substructure formation like some
      companies do). It’s only 49c per base pair, so
      doesn’t seem too cost prohibitive. My only
      concern is that if this protein is toxic, I
      could be wasting money.
 
      So I was wondering, has anyone seen the
      expression for a particular protein change from
      zero in Rosetta/Codon+ cells using "native"
      sequeneces to being largely overexpressed in
      BL21(DE3) cells using codon optimized sequences?
      For folks who have had a similar problem to the
      one I've described, would you recommend that I
      first try using a codon optimized sequence in E.
      coli over testing protein expression in
      yeast/insect cells, or the other way round?
 
      Thanks!
        
Phoebe A. Rice
Assoc. Prof., Dept. of Biochemistry & Molecular Biology
The University of Chicago
phone 773 834 1723
http://bmb.bsd.uchicago.edu/Faculty_and_Research/01_Faculty/
01_Faculty_Alphabetically.php?faculty_id=123
 
RNA is really nifty
DNA is over fifty
We have put them
  both in one book
Please do take a
  really good look
http://www.rsc.org/shop/books/2008/9780854042722.asp