Top Ten Tips for Producing ¹³C ¹⁵N Protein in Abundance

Deborah A. Berthold, Victoria J. Jeisy, Terry L. Sasser, John J. Shea, Heather L. Frericks, Gautam Shah, and Chad M. Rienstra
Departments of Chemistry and Biochemistry
University of Illinois at Urbana-Champaign

What could be easier than overexpressing an E.coli protein in E.coli? You don’t have to be an old hand at protein expression to know that this can often be more difficult than it sounds. We tested our skills recently with DsbA, a 20 kDa protein that catalyzes disulfide bond formation in the E.coli periplasm. The wildtype DsbA expressed well in LB medium, and also in a Bio-Express-supplemented ¹³C ¹⁵N labeling medium. Likewise, the DsbA C33S mutant expressed well in LB. But when we first tried to label C33S, our luck ran out-- we saw no expression at all. Today we are producing ¹³C ¹⁵N DsbA C33S at a yield of 100 mg per liter. Here’s our top ten tips for expressing recalcitrant proteins.

10. Stop the leaks. Leaky expression (i.e. expression in the absence of inducer) of a “toxic” protein or even a “less-than-healthful” mutant protein can slow cell growth, resulting in a suboptimal level of expression. In addition, because the half-life of ampicillin in a dense culture is less than thirty minutes, at later stages of growth there is a loss of selection for cells with ampicillin-resistant expression plasmids. So, any selection pressure stemming from leaky expression would give advantage to cells that have lost their expression plasmids. At best, the portion of plasmidfree cells in the culture at the time of induction are taking up ¹³C-labeled glucose without contributing to expression. At worst, leaky expression slows cell growth to the point that there is no growth at all in the labeling medium.

Promoters show variation in their leakiness, so use of a highly-regulated promoter, such as the T7 promoter of the pET vector system, may be the solution. In our case, DsbA is on a high-copy plasmid (pUC119) and is expressed from a lac promoter (Kisigami, et al.,1995). In addition to this promoter being well-known for its leakiness, in optimizing the expression of the wildtype DsbA, we had changed E. coli strains. The original strain, E.coli M15/pREP4 contained additional copies of the lac repressor on pREP4; our preferred strain, E. coli C43(DE3), had no additional repressor other than the copy in the host genome. It was likely that the repressor binding sites on our expression plasmid titrated out all the copies of the repressor in the cell, leaving some promoters un-repressed. Our response: we put the compatible pREP4 plasmid into our E.coli C43(DE3) so that the cell would produce enough repressor for all the plasmid binding sites. Another solution might be to clone lacI^q (a constitutive lac repressor) directly onto the expression plasmid; this is found, for example, in Qiagen’s QE80 series of His-tag expression vectors.

9. Slow down the train. For high levels of protein expression, the rate of transcription needs to be coupled to that of translation, which in turn needs to be coupled to any essential co- or post-translational events, such as folding, cofactor binding or membrane insertion. When transcription outstrips translation, loss of cell viability can occur (along with the destruction of ribosomal RNA and induction of proteases; see Dong, et al., 1995). There are several methods to tweak the rates of cellular metabolism to try to bring transcription, translation and post-translational processing in line. One can change promoters (Makrides, 1996; Baneyx, 1999). Strong promoters, such as T7 can be replaced with weaker promoters (arabinose, T5, tac). One can change cell lines. Two E. coli strains particularly suited for expression optimization, C41(DE3) and C43(DE3), were originally obtained from a selection for mutations that overcame lethality associated with overexpression from a T7 promoter (Miroux and Walker, 1996). These two strains also show an increase in plasmid stability relative to their parent BL21(DE3) strain (Dumon-Seignovert et al., 2004). And, perhaps most easily, one can change the temperature during expression. For DsbA, the interplay between growth temperature and E.coli strain can be seen in Figure 1. Using BL21(DE3), slowing cell processes by growing at 25°C gives a large increase in expression over growth at 37°C (Lanes 9-10 vs. Lanes 7-8). In contrast, the strain C43(DE3), which is thought to have a mutation slowing the rate of transcription, gives a greater yield of expressed protein at 37°C (compare lanes 3-4 vs. 5-6). For DsbA C33S, we can choose between expressing at 37°C in C43, or expressing at 25°C in BL21(DE3). For some other proteins we have seen both C43(DE3) and a lower temperature are required for the highest expression level.

8. Eat yer spinach. If Popeye were a microbiologist, he might well claim that his cultures were “strong to the finish” because he feeds them iron. Studier recommends that trace metals be added to defined media, and, remarkably, that if a trace metal mixture is not available, 100 µM FeCl₃ alone will give the nearly the same level of expression (Studier, 2005). It is recommended that trace metals be added to rich media, as a precaution against batch-to-batch variation of trace metal content. We have found that addition of trace metals improves the yield of expressed protein even in Bio-Express-supplemented (10 ml/L) media.

7. Be media-savvy. Media acidification is detrimental to E.coli growth and protein yield (Swartz, 2001), but can be prevented to some extent by using a well-buffered medium with good aeration. We are currently using Studier Medium P (a phosphate-buffered medium) for expression of DsbA and several other proteins (Table 1). In addition, it has been found that calcium, although often added in relatively high amounts in traditional E.coli media as M9, is actually not essential at such high levels (and we have noted, can sometimes cause precipitant to form in the medium). On the other hand, Studier (2005) reports that using 2 mM MgSO₄ , rather than 1 mM, can increase the cell density anywhere from 50% to 5-fold -- which would correspond to comparable increases in protein yield.

6. Exercise *aerobically*. The least expensive (and perhaps also least-appreciated) supplement for cultures is oxygen, in the form of increased aeration. Glucose utilized by the aerobic respiratory chain provides more than 10-times the energy of glucose fermented. For growth on glycerol, the situation is even more dramatic: E. coli cannot utilize glycerol as an anaerobic energy source. And, as noted above, culture acidification, as occurs under anaerobic and microaerophilic conditions, is often detrimental to protein expression. To enhance aeration, we routinely use a culture volume of only 250 ml in a baffled 2 L flask. Note, however, that all expressions cannot tolerate all volume/baffling geometries, as certain expressions may cause cells to become fragile and sensitive to shear forces. Therefore optimal culture volume and rate of shaking with a given baffle geometry should be tested, not presumed.

5. Timing is everything. (part I: Induction) A common protocol for protein expression requires that IPTG be added when a culture has reached a cell density (A₆₀₀) of 0.8. But, following induction the growth of the culture will often slow or stop entirely. Thus it could happen that while the labeling medium supports uninduced growth to cell density of A₆₀₀ = 4.0, the density of an induced culture ends up at harvest at only A₆₀₀ = 1.0, giving perhaps only one quarter the potential yield. A better “rule of thumb” for time of induction would be to induce expression at 50% the density of an uninduced culture at harvest, and to also test the effect of induction at 80% uninduced density (i.e. at A₆₀₀ = 3.2 for a culture that could grow to 4.0).

4. Timing is everything. (part II: Harvest) A typical protocol for a 37°C expression calls for harvest of E. coli cells 2-3 hours after protein expression is induced. However, it can be worthwhile to determine where expression is maximal, especially if one is using temperatures lower than 37°C. We now harvest our 25°C DsbA expression at 20 hours post-induction, and this more than doubles the protein yield of a 6-hour harvest.

3. Quadruple up on the cells. Marley et al. (2001) have reported increases in expression of labeled protein by growing an E. coli culture in LB medium to density of A₆₀₀ = 0.8, followed by harvest and resuspension at four-fold the density in labeling medium. Because cells containing the DsbA C33S expression plasmid grew well in rich medium, but poorly in labeling medium, this was an ideal solution for us. The four expression conditions of Figure 1 all used this method.

2. Just a spoonful of sugar (or four). Marley et al. (2001) also reported using 4 g/L glucose in their dense cell cultures. We tested 2 g/L glucose and 4 g/L glucose and found an increase of more than 5-fold in yield of DsbA C33S with a 2-fold increase in glucose (Fig. 2). Additional glucose (6 and 8 g/L) did further increase the yield, but not cost-effectively. At 4 g/L glucose, increasing the NH₄Cl from 2 g/L to 3 g/L gave a corresponding ~50% increase in expressed protein, but increase of NH₄Cl above 3 g/L had no further effect (data not shown).

1. Save both the baby and the bathwater. A common protocol for extracting periplasmic proteins such as DsbA directs that the cells be suspended in a hyperosmotic medium (we use 20% sucrose), followed by resuspension in water to burst the outer membrane, releasing the periplasmic protein. We found our DsbA overexpressing cells to be so fragile that breakage occured even in the sucrose solution (Fig. 1, lanes A) and had we not saved this supernatant, we would have lost a substantial part of our yield (compare Fig. 1, lanes B). Another occasion where it can be useful to save and assay all fractions is when assessing growth conditions and expression strains. A low-speed centrifugation of a well-sonicated E.coli sample will pellet inclusion bodies. If one is attempting to optimize for a high yield of protein in a native conformation, a large amount of protein in an inclusion body fraction can indicate that expression needs to be slowed somehow (see Tip #9 above) to allow time for proper folding or post-translational events.

By applying these tips to our DsbA C33S expression, we were able to produce sufficient ¹³C ¹⁵N labeled protein for solid state NMR studies. Figure 3 shows a highly resolved 100 ms DARR spectrum of 5 mg microcrystalline DsbA taken on a 750 MHz spectrometer (Franks, et al. (2005)). We have found these tips to be generally useful: we have not only improved the expression of the E. coli periplasmic protein DsbA, but have been able to increase the expression yield of membrane proteins (e.g. E. coli cytochrome bo3 oxidase (Frericks, et al. (2006) and DsbB (Li, et al. (2007)), as well as heterologous (non- E. coli) proteins expressed in E. coli.

References:
Baneyx, F. (1999) Recombinant protein expression in Escherichia coli, Curr Opin Biotech 10, 411-421

Dong, H., Nilsson, L., and Kurland, C. G. (1995) Gratuitous overexpression of genes in Escherichia coli leads to growth inhibition and ribosome destruction, J Bacteriol 177, 1497-1504

Dumon-Seignovert, L., Cariot, G., and Vuillard, L. (2004) The toxicity of recombinant proteins in Escherichia coli: a comparison of overexpression in BL21(DE3), C41(DE3), and C43(DE3), Protein Expr Purif 37, 203-206

Frericks, H. L., Zhou, D. H., Yap, L. L., Gennis, R. B., and Rienstra, C. M. (2006) Magic-angle spinning solid-state NMR of a 144 kDa membrane protein complex: E. coli cytochrome bo3 oxidase, J Biomol NMR 36 , 55-71

Franks, W. T., Zhou, D. H., Wylie, B. J., Money, B. G., Graesser, D. T., Frericks, H. L., Sahota, G., and Rienstra, C. M. (2005) Magic-angle spinning solid-state NMR spectroscopy of the beta-1 immunoglobulin binding domain of protein G (GB1): ¹⁵N and ¹³C chemical shift assignments and conformational analysis, J Am Chem Soc 127, 12291-12305

Kishigami, S., Kanaya, E., Kikuchi, M., and Ito, K. (1995) DsbA-DsbB Interaction through their active site cysteines. Evidence from an odd cysteine mutant of DsbA, J Biol Chem 270, 17072-17074

Li, Y., Berthold, D. A., Frericks, H. L., Gennis, R. B., and Rienstra, C. (2007) Partial ¹³C and ¹⁵N chemical-shift assignments of the disulfide-bond-forming enzyme DsbB by 3D magic-angle spinning NMR spectroscopy, Chembiochem 8, 434-442

Makrides, S. C. (1996) Strategies for achieving high-level expression of genes in Escherichia coli, Microbiol Rev 60, 512-538

Marley, J., Lu, M., and Bracken, C. (2001) A method for efficient isotopic labeling of recombinant proteins, J Biomol NMR 20, 71-75

Miroux, B., and Walker, J. E. (1996) Over-production of proteins in Escherichia coli: Mutant hosts that allow synthesis of some membrane proteins and globular proteins at high levels, J Mol Biol 260, 289-298

Studier, F. W. (2005) Protein production by auto-induction in high-density shaking cultures, Protein Expr Purif 41, 207-234

Swartz, J. R. (2001) Advances in Escherichia coli production of therapeutic proteins, Curr Opin Biotech. 12, 195-201

BioExpress® is a registered trademark of Cambridge Isotope Laboratories

back to top