Low cell viability has been reported for GGD(/E)EF proteins previously expressed from various sources (Ryjenkov et al. 2005). The same was evident for our clones as well, and transformation efficiency was abysmally low. During expression and extraction, care had to be taken for obtaining active protein in the soluble fractions. Initially, recombinant wild type Sebox3 was obtained in small quantity in the supernatant fractions and had to be standardized against buffer conditions to improve purity. The need for larger culture volumes was necessitated by the low yield of the protein after cleavage of tag. Similarly, for the mutants as well, yield needed to be standardized to achieve workable quantities (Fig. 1). Again, the tagless proteins were found to be unstable in solution.
Mutation in the ‘GEE’ positions affects biofilm formation in V. cholerae
Site directed mutagenesis at the positions G, E and E of the GGEEF sequence in VC0395_0300 resulted in the generation of mutants with altered amino acids. While Sebox5 mutants had an arginine in place of the glycine, Sebox6 had the glutamate replaced by lysine, and Sebox7 also had a lysine in place of the glutamate. The generated mutants were checked by sequencing and also by blotting against anti-GST antibodies (GE Healthcare Life Sciences).
Growth of the mutants in LB medium and transfer to static conditions for observation of biofilm formation was undertaken for separate sets at 2, 4 and 7 days intervals. After both 4 and 7 days, it was observed that the biofilm formation was significantly lower for the mutant Sebox5, 6 and 7 cultures when compared to the Sebox3 wild type. While the formation of biofilms decreased almost by threefolds in Sebox5 and 7, a twofold decrease was evident in Sebox6 (Fig. 2a, b).
The complementary motility assay was also performed to visualize the loss of biofilm formation abilities in the mutants. The TTC assay essentially marks out the migration of the bacterium from the zone of inoculation by the needle. After inoculation in the stab culture, the test tubes were allowed to incubate at 37 °C for 2 days without shaking. For the wild type strain Sebox3, there was no migration from the initial stab line, and no regions of color were visible. However, for the mutant strains, there was visible spread of the red formazan zone away from the site of inoculation of the initial stab (Fig. 3). All the three strains showed more-or-less comparable zone of motility, confirming our initial observations about loss of biofilm formation ability in the three mutants.
The VC0395_0300 protein and the mutants do not have diguanylate cyclase activity but display in vitro phosphodiesterase activity
Since almost all GGD(/E)EF proteins display cyclic diguanylate activity, the proteins were initially assayed for the production of c-di-GMP, using the method employed previously (Ryjenkov et al. 2005). Briefly, a reaction mixture containing 5 μM protein, 10 mM MgCl2, and 50 mM NaCl was treated with 5 μM of GTP (Sigma Aldrich). The reaction mixture was prewarmed and the reaction was carried out for 0, 5, 15, 30, and 60 min. Aliquots were withdrawn at the given time intervals, heated in a boiling water bath for 3 min, 0.5 mM EDTA was added and centrifuged at 10,000g for 10 min.
The samples were injected into a X Terra RP18 column (250 mm × 4.6) from Waters and separated in an Agilent 1220 Infinity LC system for a total run time of 30 min at a flow rate of 0.6 ml/min. 20 mM triethyl amine in 9 % methanol and water was used as the solvent system for the process. NADP was added to the sample prior to injection as a standard. However, even after repeated runs with wild type and mutant proteins (at least four times for each), no peak corresponding to c-di-GMP was observed, indicating the absence of diguanylate cyclase activity in VC0395_0300 as well as its mutants.
The wild type and the mutants were both assayed for in vitro phosphodiesterase activity with bis(p-nitrophenyl) phosphate as the substrate. To our surprise, all the three mutants as well as the wild type demonstrated considerable activity with OD410 values ranging from an average of 0.41 (Sebox3) to 0.54 (Sebox7) (Fig. 4). Triplicate reactions confirmed the same for all four sets. Again, inhibition with CaCl2 reduced the OD410 value almost fourfolds indicating the inhibition of PDE activity by Ca2+, as had been the hallmark of other proteins having PDE activity. However, it is indeed notable that the VC0395_0300 protein does not have any EAL or HD-GYP domain in the correct orientation, ones that are usually associated with PDE activity. Since the mutants do not show any significant change in the PDE activity, it can also be safely concluded that the PDE activity is not associated with the GGEEF signature sequence.
Estimation of changes in structure of mutants through fluorescence
Intrinsic tryptophan fluorescence (λex at 295 nm) was observed to infer any changes in the proteins’ architecture due to the effect of the mutations. The fluorescence spectra after denaturation with guanidinium hydrochloride showed a shift in the emission maxima (λmax) for both the wild type and mutant proteins (Fig. 5). While this shift was more pronounced in the case of the wild type protein, it was considerably less for the mutants. One of the tryptophans is within the first four amino acids from the N-terminal, and possibly would not be folded into the interior. The other tryptophan is near about the middle of the sequence and contributes to the red shift. A bigger shift in λmax for an unfolded protein indicates a complete exposure on unfolding for a buried tryptophan. On the other side, a smaller shift signifies the tryptophan to be partially accessible for fluorescence in the folded state, which seems to be the case for the mutants. This was indicative of some structural rearrangement resulting in partial exposition of the interior tryptophan in the mutants vis-à-vis the wild type Sebox3.
Structural changes in the interface from homology models
The modeled structure was aligned against other crystallized GGDEF proteins structures (4h54, 3tvk, 3qyy, 3ign) using PyMOL (The PyMOL Molecular Graphics System 2002). All these proteins aligned almost perfectly at their GGD(/E)EF regions. The GGD(/E)EF domain in VC0395_0300 is made up of a five-stranded beta sheet surrounded by multiple alpha helices and the GGEEF sequence occurs in the turn between the first two antiparallel strands of the sheet. This is in agreement with the three other structures which have been cited before (Lim et al. 2007; Tamayo et al. 2007; Gao and Stock 2009).
Similarly, models were constructed for the mutant proteins Sebox5–7, and the model with the least free energy was chosen. All the generated models had more than 97 % of the residues in the favourable regions of the Ramachandran plot and the rest in the allowed regions. The mutant structures also show more or less similar architecture with regard to the domain, but, there is significant difference in the GGEEF turn between β1 and β2. This has been highlighted in Fig. 6.
While the structure of the PleD from P. aeruginosa and the XCC4471GGDEF protein from Xanthomonas campestris showed the presence of compact interaction between the two consecutive, highly conserved glycines in the first two sites, generation of a G to R mutation in Sebox5 has disrupted the compactness which is essential for interaction with the c-di-GMP. Comparison of the surface morphology of the wild type and mutant proteins reveal significant differences in the surface of the wild type and the mutant Sebox5. The introduction of the lysine in place of the first glutamate in Sebox6 also changes the surface of the turn. It is to be noted that both Sebox5 and Sebox6 show a significant loss of biofilm formation abilities compared to the wild type. However, the lack of diguanylate cyclase activity in any of the proteins does not allow us a clear hypothesis.