If the question is "why" this is happening, the answer may be complicated.

You can (and should) check your recombinant protein for predicted regions of disorder or long, unstructured loops. Without knowing anything about your target, my guess is that a protein that size likely has many discrete folded domains. The host may be cleaving the regions between these domains, which are commonly unstructured coils that would be susceptible to protease cleavage. I can see at least three discrete bands on your second gel that appear to be specific cleavage sites - do the sizes of any of these bands correspond approximately to the loss of a single domain, or to a free domain cleaved from the others?

I assume with the MBP/His10 tag this is a soluble protein expressed in the cytoplasm, so this would probably be happening in the host during protein expression. You can try reducing induction temp. and/or time to see if you can limit target degradation by the host. The inclusion of protease inhibitors will only be helpful if the degradation occurs after lysis.

The faint, smear-like bands are a bit more difficult to diagnose. If your protein is membrane-associated or participates in complex protein-protein interactions, it may be pulling down other E. coli proteins with it that remain reasonably tightly bound through purification. Alternatively, they may be nonspecific proteolytic degradation. If possible, you might want to attempt to digest a few micrograms of your purified samples in a MS-sequencing grade protease (e.g. trypsin) and run an LC-MS based proteomic analysis of the contents. This might help illuminate if you are pulling down E. coli proteins or if you're seeing your target degraded.

You may also want to Western blot these samples - if any of the degradation products contain the MBP/His10 tag, they will also blot, which can quickly tell you what bands belong to the protein and if those fragments contain the tagged termini.

Thank you Alexander for your feedback.

I ran one of the sequences through IUPred3 and got this graph back. I am going to read the paper and learn more about how this program works. I was wondering what you used in order to detect regions of disorder and unstructured loops.

I believe the 2 of the 3 bands located around the 55 kDa marker are MBP (45.5 kDa) and naturally occurring amylase (63 kDa) since the image you are referring to is purification of the cell lysate through an amylose column and that band disappears after a second step of purification but the other two persist. Other than those proteins, no other protein in my fusion can account for the 3rd band. I will check the domains and correlate it to the IUPred3 prediction and see if that could give me an answer.

We currently induce in 20C, but I will look into potentially lowering it to 18, 16 and maybe 14. I remember trying to induce at lower temperatures but the expression wasn't the best when compared to 20C, but perhaps there is a trade off between yield and amount of degraded of the POI.

I will ask about sending a sample for LC-MS. I will see if we have Anti-His in storage. I will answer back once I have responses and data.

As for observing if proteins are interacting with my POI and are being carried into the purified product, would it be possible to run a native and denaturing gel side by side to observe if complexes are forming?

Hi Daniel Arroyo

There are a variety of different servers that can predict disorder, IUPred is among the most popular. They really all do the same thing. Looks like you don't have any regions that look like entire disordered domains, BUT it looks pretty convincingly like there are a few regions that might represent the boundaries between discrete domains.

You can try structure prediction servers and domain/motif searching to see if you can identify those domains and their boundaries. InterPro, AlphaFold2, Swissmodel, Phyre2. Each of these can give you an idea about what discrete folded domains are present and how they map to your sequence.

It would be very typical to see the MBP tag cleaved and the amylase on your column, so that's a reasonable assignment of those two bands. Where I'm going with this is that if you can identify the major 3rd cleavage band, you might be able to attempt mutagenesis of your target at that linker to prevent the cleavage, and the secondary degradation products may be related to limited cleavage of the separated domains. It may do nothing, or you may be able to boost your yield/purity of the intact protein.

I often use rich media (super broth: 32g tryptone, 20g yeast extract, 10g NaCl per L) and induce at 15°C overnight, which both eases the metabolic stress of expressing a large/toxic/problematic protein and also helps to stretch the yield per L. There are also strains of E. coli engineered for lower temp induction, I believe one is called ArcticExpress and there are other commercial alternatives from the major brands.

An anti-His/anti-MBP blot is a really quick and easy way to get lots of info about your purifications, would really recommend it!

Yes, denaturing would help - but not in the gel, since at that point your proteins have already co-purified. You'd need to repeat a purification from scratch with urea (or guanidine, but this is PAGE-incompatible) and check to see what changes on the gel following.

Hi Alexander C. Anderson

I've compared the disordered domains against the protein structures. The intense peaks from the graph line up with the areas that the linkers are present.

I have found that we are using sequences that have been codon optimized for human expression, not E coli. Both proteins are similar and either have 2,261 or 2,211 amino acids. The AGG codon (makes up ~4% of Arg anticodons) occurs either 1-2% in the whole sequence making up 22-44 a.a.

I was curious as to the affect this would have on the purification process. I read that mRNA may dislodge from a ribosome if it has to wait too long for tRNA to arrive. Since this would make all the proteins have to compete for resources I wonder if this is what is occurring. I used this webservice (Rare Codon Analysis - Online Calculation and Plot Tool - BiologicsCorp) to get the codon representation frequency of both sequences (attached).

I ran a western against one of the proteins apart of the fusion and those results are attached. There is a spread and I can't decide if it's degradation or truncated version of the protein caused by the codons used. I performed the purification process with my supervisor and our results haven't changed.

Hi Daniel Arroyo

Great - looks like you've built some evidence for the case that your large, multi-domain protein is experiencing some degradation. It seems you've used a target-specific antibody which is even better and validates your assignment of one band to the loss of MBP-His and the other to amylase. It's very unlikely that your target is co-purifying with other host proteins, so you can focus your efforts on screening conditions to maximize your intact yield.

The first thing I'd suggest is screening expression conditions. You can try different media, temperatures, induction times, induction strengths, additives etc. One hugely beneficial tip I learned for fragile recombinants is to add 1-3% v/v ethanol to the media which will induce the expression of stress-related protein chaperones without sacrificing growth rate or viability. When screening conditions, I like to use a small (5mL) scale and then after expression harvest the cells, being sure to normalize their loadings. I directly lyse whole cells in SDS loading buffer and directly blot whole cell lysates to compare yields quickly without the need for purification and work-up. In less than a week you can quickly screen dozens of conditions and may find that a certain expression protocol will improve your intact yield.

Beyond that, you can go for gene/protein engineering as you mentioned. Codon optimization for E. coli is really hit or miss, in some cases it makes a world of difference and in other cases I've found it did nothing. You can try E. coli Rosetta or CodonPlus, which are strains of E. coli that carry an accessory plasmid that encodes rare tRNAs. This would let you test the idea without the need to entirely re-synthesize the gene and reclone it. One can also attempt protein engineering and swap the codons inside linker regions to see if a more stable protein can be generated - but this is a much longer workflow and can take weeks to screen through even a modest sized set of protein variants.

Best of luck!

ACA