Characterisation of putative glycan and drug binding proteins predicted using in silico screening methods

Abstract

This thesis describes the characterisation of six targets for a novel antimicrobial drug 3,4-methylenedioxy-beta-nitropropene (BDM-I) shown to inhibit bacterial, protozoal and fungal infections and the characterisation putative carbohydrate binding proteins (CBP) BT_411 and BT_3781. Furthermore, described here is the experimental validation of bioinformatic programs SPOT-Ligand, of which identified six putative drug targets for BDM-I, and SPOT-Struc, that predicted BT_411 and BT_3781 as CBPs. Specifically I expressed and purified the six BDM-I targets predicted by SPOT-Ligand and expressed the two SPOT-Struc predicted putative CBPs BT_411 and BT_3781. The drug targets were then used to identify a potential mechanism of action for BDM-I while the putative CBPs were likewise characterised to identify their carbohydrate binding affinity. Lastly the characterisation results were then utilised to evaluate the predictive capabilities of SPOT-Ligand and SPOT-Struc and in turn help clarify the role of bioinformatics in experimental research. The six BDM-I targets (Gene name: EF0414, ubiE, ftsZ5, Lebu_1328, acpD, and ubiH) were expressed in E. coli BL21 (DE3) cells with expressed recombinant soluble protein purified to homogeneity using HIS-select nickel affinity and size exclusion chromatography (SEC). Likewise, two putative CBPs BT_411 and BT_3781 as well as an additional novel mushroom lectin PSL-2 (used as a positive lectin control), were expressed in E. coli BL21 (DE3) cells and purified to homogeneity using affinity chromatography and SEC. All proteins were confirmed pure by SDS-PAGE and of those proteins containing a HIS-tag (all except PSL-2), western blot immunodetection. All purified proteins were then measured using circular dichroism (CD) spectroscopy as a quality control step to determine if they had denatured before proceeding with further characterisation. The six potential ii drug targets were analysed for BDM-I affinity using surface plasmon resonance (SPR) with those showing affinity for BDM-I further characterised by computational docking analysis. Similarly, putative CBPs BT_411 and BT_3781 in conjunction with PSL-2 were analysed for carbohydrate affinity with a wide variety of glycans using SPR followed by computational docking analysis of the CBPs with some of the SPR defined glycan matches. Results showed that BDM-I has high affinity for drug target AcpD from Salmonella enterica, serovar Typhimurium, a bacterial azoreductase protein responsible for the breakdown of azo dyes in S. typhimurium (S. typhimurium causes gastroenteritis in humans). SPR analysis of the binding of BDM-I to AcpD revealed an equilibrium dissociation constant (KD) of 0.58 μM with an association constant (ka) of 9.4 x 104 M-1s-1 and dissociation constant (kd) of 9.7 x 10-4 s-1, suggesting that BDM-I binds rapidly and easily to acpD followed by a slow dissociation. Docking analysis of BDM-I in AcpD showed hydrogen bonds between a BDM-I nitro group and residues Ala115 and Asn98, with a binding energy of – 6.7 kcal/mol. AcpD structural alignment with an azoreductase that also shows nitroreductase activity, PaAzoR from Pseudomonas aeruginosa, showed a high enough structural similarity with AcpD to computationally infer that function is likely shared between the two proteins. As such, it was computationally determined that AcpD may act on BDM-I as a nitroreductase via the binding of BDM-I’s nitro group in order to reduce it to chemical intermediates (similar to paAzoR nitroreductase activity). This reveals a possible mechanism of action for BDM-I; namely that bacterial nitroreductases may bind and reduce the nitro group on BDM-I resulting in the production of bacterially toxic intermediates. SPR analysis of BT_411 revealed carbohydrate affinity for β1,4 linked N-acetylglucosamines (GlcNAc) and 2,3 linked sialic acids (Neu5Ac) with KD’s between iii 0.10 μM and 0.23 μM. Computational analysis of BT_411 further indicated preferential affinity towards GlcNAc (-4.6 kcal/mol) over Neu5Ac and indicated that BT_411 is likely a carbohydrate binding module (CBM) showing possible functional attributes of a βGNase (hydrolase) that catalyses the reduction of GlcNAc glycosidic bonds. SPR analysis of BT_3781 indicated carbohydrate affinity for fucose (Fuc) and galactose (Gal) containing glycans; specifically blood group H, B and A showing high affinity with KD’s between 0.17 μM and 0.34 μM. Computational analysis of BT_3781 supported SPR analysis indicating affinity with Fuc, in particular blood group H disaccharide (Fucα1-2Gal) with a binding energy of -6.8 kcal/mol. Indicating that BT_3781 is also a CBM with functional similarities to that of a glycosidic hydrolase; namely a fucosidase that catalyses the reduction of Fucα1-2 linkages. SPR analysis of the positive control PSL-2 confirmed crystallography data of the protein that showed preferential binding to Gal and Fuc residues. Specifically, SPR indicated PSL-2 has high affinity towards blood group B (Galα1-3(Fucα1-2)Galβ1-4Glc), implying that PSL-2 may be a blood group lectin. Moreover, successful characterisation of PSL-2 validates the experimental procedures used in the characterisation of BT_411 and BT_3781. The characterisation results of BDM-I drug targets and CBPs BT_411 and BT_3781 validate the supportive role of bioinformatic programs SPOT-Ligand and SPOT-Struc in experimental research. Mainly in the structural predication of drug targets and identification of novel CBPs.