Bacterial species exhibit diversity in their mechanisms and capacity for protein disulfide bond formation

Abstract

Protein disulfide bond formation contributes to the folding and activity of many exported proteins in bacteria. However, information about disulfide bond formation is limited to only a few bacterial species. We used a multifaceted bioinformatic approach to assess the capacity for disulfide bond formation across this biologically diverse group of organisms. We combined data from a cysteine counting method, in which a significant bias for even numbers of cysteine in proteins is taken as an indicator of disulfide bond formation, with data on the presence of homologs of known disulfide bond formation enzymes. These combined data enabled us to make predictions about disulfide bond formation in the cell envelope across bacterial species. Our bioinformatic and experimental results suggest that many bacteria may not generally oxidatively fold proteins, and implicate the bacterial homolog of the enzyme vitamin K epoxide reductase, a protein required for blood clotting in humans, as part of a disulfide bond formation pathway present in several major bacterial phyla.

Fig. 1.

Disulfide bond formation pathway of E. coli. (arrows indicate flow of electrons)

Fig. 2.

Exported proteins show a unique bias for even numbers of cysteines. Cysteine distribution in E. coli K12 proteins - cytoplasmic and exported (classes 1 and 5).

Fig. 3.

Cysteine counting in exported proteins of all bacterial genomes. The z score for the fraction of exported proteins with even numbers of cysteine (Efrac), is plotted against the Cpref (frequency of cysteine in exported proteins/frequency of cysteine in all proteins) for each genome (a Cpref of <1.0 indicates a bias against incorporation of cysteine into exported proteins). Region A: Disulfide bond formation is predicted to occur in exported proteins in these organisms. These genomes (z > 2.57) have significantly more exported proteins with even numbers of cysteine than is predicted by the random model. (i.e., there is a probability of 0.01% that the genome would have a z score of >2.57 if the cysteines were distributed at random.) Region B: These organisms are predicted to have no, or limited, disulfide bond formation in exported proteins. These genomes (2.57 > z > −2.57) do not have a significant fraction of exported proteins with even numbers of cysteine.

Fig. 4.

Combined results of disulfide predictions based on cysteine counting and homology searches. Genomes with significant numbers of exported proteins with even numbers of cysteines (z > 2.57) are indicated in red, and the distribution of DsbA (dark blue) and DsbB (light blue) homologs are shown in a representative subset of all organisms analyzed. The genomes containing a homolog of VKOR are indicated in green.

Fig. 5.

A bacterial VKOR homolog restores disulfide bond formation to E. coli deleted for dsbB. Disulfide bond formation was assayed by using motility plates, because motility requires active disulfide bond formation. Expression of the M. tuberculosis VKOR homolog restores motility to an E. coli ΔdsbB strain, but not a ΔdsbAΔdsbB strain.

Fig. 6.

The E. coli protein LivK does not become disulfide bonded when expressed in B. fragilis. Determination of the redox state of the E. coli protein LivK-myc expressed in B. fragilis. Samples are TCA precipitated then treated as follows, Lane 1: DTT was added to fully reduce the sample to provide a control for unlabeled protein (position of band is indicated with an open arrow), Lane 2: Mal-PEG alkylation. If the cysteines in the protein are not disulfide bonded, they will react with the 2kD alkylating agent, resulting in an increase in molecular weight. The filled arrow indicates the position of the expected shift as a result of alkylation of both cysteines, Lane 3: Control for full alkylation of the protein, samples were first reduced with DTT, then alkylated with Mal-PEG. The asterisk indicates a cross-reacting band, which also shifts upon alkylation (data not shown).