Mode of action of a novel lymphocyte inhibitory factor of attaching and effacing Escherichia coli
Bease, Andrew Graham
Attaching and effacing Escherichia coli are significant diarrhoeal pathogens that can spread between humans or via animal reservoirs. An important virulence factor produced by these bacteria is the large multifunctional protein lymphostatin (LifA), which has been reported to inhibit the mitogen- and antigen-stimulated proliferation of lymphocytes as well as mediate adherence to epithelial cells. Shiga toxin-producing E. coli lacking lifA are significantly impaired in their ability to colonise cattle. Little is known about the mode of action of LifA, however, in silico analysis has identified a putative glycosyltransferase domain homologous to that of large clostridial toxins and a putative cysteine protease domain homologous to that of C58 family proteases. LifA has recently been reported to bind uridine diphosphate-N-acetylglucosamine (UDP-GlcNAc) and mutation of a DXD motif within the predicted glycosyltransferase domain abolished this binding and lymphostatin activity. In this study, I sought to identify domains of LifA required for cell binding and lymphostatin activity, probe the role of the cysteine protease motif in intracellular processing and LifA activity, and to identify potential targets and interacting partners of the protein. Domains of LifA predicted by limited proteolysis were cloned, expressed and affinity purified. Robust assays for detecting interactions between LifA, or fragments thereof, and T lymphocytes were developed but none of the domains possessed lymphostatin activity alone or in combination. LifA was found to be cleaved within T cells, which by analogy with large clostridial toxins was hypothesised to be the result of autoproteolysis mediated by the cysteine protease domain. A C1480A substitution mutant of full-length LifA was constructed by site-directed mutagenesis to disrupt the predicted catalytic triad of the cysteine protease domain. The C1480A substitution resulted in a lack of intracellular processing of LifA and impaired the ability of the protein to inhibit mitogen-stimulated proliferation of bovine T cells, without obvious changes to the biophysical properties of the protein. LifA processing was also found to require endosomal acidification using the inhibitors bafilomycin A1 and chloroquine. Shotgun mass spectrometry and protein pull-downs were used to identify potential targets of LifA activity and interacting partners. Relatively few candidate proteins were identified and these were generally not consistently observed between repeated experiments. Based on analysis of signal transduction pathways perturbed by LifA, I explored if the cellular kinase Akt may be targeted by LifA directly or indirectly. Akt is known to control T cell proliferation and to be regulated by phosphorylation and GlcNAcylation. S473 phosphorylation of Akt in mitogen-stimulated T cells was inhibited by LifA in a manner dependent on the DXD and cysteine protease motifs, but O-GlcNAcylation of Akt was not detected. This inhibition only occurred in cells treated with LifA before mitogenic stimulation. Infection of T cells with an enteropathogenic E. coli strain inhibited Akt phosphorylation in a manner dependent on the Type III secretion system but not LifA or a homologous LifA-like protein. Taken together, this study advances our understanding of the mode of action of a key virulence factor of pathogenic E. coli and, in particular, identifies a key role for a cysteine protease motif in intracellular processing of the protein and lymphostatin activity.