Although unicellular eukaryotes such as yeast cannot serve as models for the bacterial infection of host cells, the utility of S. cerevisiae
for investigating the molecular activities of effector proteins is well established. Until now, little has been done to exploit the genetic tractability of yeast to determine cellular targets of effector proteins. In this study, we present the first use of yeast systems biology to identify a function for a poorly understood bacterial effector protein. By integrating complementary systems-level snapshots of the cellular perturbation caused by OspF, we were able to determine that OspF targets a well-characterized yeast-signaling pathway. These observations led us to hypothesize and subsequently demonstrate that OspF inhibits highly conserved MAPK signaling pathways including those that regulate cell wall biogenesis in yeast and the host innate immune response in mammals. These results are in agreement with the recent demonstration that OspF is a MAPK phosphatase specific for ERK and p38 [16
]. Although, there is also evidence that OspF can activate ERK phosphorylation in polarized epithelial cells [36
One of the hallmarks of Shigella
infections is the dramatic inflammatory response characterized by the recruitment of polymorphic neutrophils to sites of infection. This response facilitates access of Shigella
to the basolateral surface of epithelial cells where this intracellular pathogen mediates its own uptake into cells via a type III secretion system. Shigella
trigger activation of the innate immune response, at least in part, through the action of pathogen-associated molecular patterns like LPS [37
] and peptidoglycan [28
also down regulate the innate immune response by the action of several proteins including OspF. The Shigella
effector proteins, OspG and IpaH9.8, inhibit NF-κB activity resulting in decreased production of proinflammatory cytokines [33
]. In addition, the ShiA protein down regulates the innate T-cell response [38
]. OspF-mediated downregulation of the innate immune response by inhibition of MAPK signaling complements the actions of these other Shigella
proteins. These proteins presumably allow Shigella
to fine-tune the host innate immune response over the course of infection.
OspF homologs are found in pathogens of both plants and animals including Salmonella typhimurium
(SpvC) and Pseudomonas syringae
(HopAI1) (Figure S6
). HopAI1 was recently demonstrated to inhibit activation of the plant innate immune response by an unknown mechanism [39
]. These observations suggest the existence of a new class of bacterial proteins, common to pathogens of both plants and animals that modulate the host innate immune response by inhibiting MAPK phosphorylation.
Our study demonstrates how genome-wide yeast screens can help generate testable hypotheses about the roles of bacterial effector proteins in pathogenesis. We applied a standard technique of statistical data mining, enrichment analysis, to integrate our screen results with two of the many types of systems data available for S. cerevisiae, genetic interactions and gene ontology annotations. This implicated CWI in OspF toxicity. The juxtaposition of this observation with the apparent reverse regulation of the CWI pathway in microarray experiments provided the crucial insight. Our approach was necessarily heuristic. For example, the set of genes exhibiting differential expression in response to OspF was of a size that one might dismiss as uninformative if it were the only set of data available, but the implication of CWI involvement by the hypersensitivity screens made the presence of any CWI pathway–regulated genes among those differentially regulated conspicuous.
Considerable evidence also motivated our focus on the cell wall. For example, at least five OspF hypersensitive deletion strains not accounted for in the enrichment analyses are impaired in protein mannosylation and glycosylphosphatidyl-inositol anchor biosynthesis. Both of these post-translational modifications are abundant among yeast cell wall proteins [40
]. The corresponding ontologies were not statistically significant in our analyses, but their biological significance to cell wall biogenesis is well established. Similarly, six of the eight genes annotated to “ubiquitin-dependent protein catabolism” (italicized in ) are involved in regulation of an alternative cell wall biogenesis pathway which only becomes essential when the normal cell wall biogenesis pathway is perturbed [41
]. These and other observations highlight the limitations of statistical methods and the importance of complementing formal analysis with an in-depth literature review to accurately interpret screen results.
The demonstrated inhibition of the CWI pathway by OspF accounts for much, but not all, of the observed hypersensitivity in deletion strains. The CWI pathway is constitutively active in at least three of the OspF hypersensitive cell wall biogenesis strains (Δgas1, Δsmi1,
]. Presumably this activation is essential for their survival since these deletion strains are synthetically lethal in combination with mutations that impair the CWI pathway (Δslt2
]. However, other pathways and processes were also implicated to which no clear CWI connection exists.
We assume that the multiplicity of process ontologies enriched in the hypersensitive strains reflects pathway dependencies centered on the cellular perturbation caused by OspF. For example, if a given pathway is required to mitigate toxicity of OspF, then any cellular process on which that pathway depends, will probably also be intolerant of impairment by gene deletion, though perhaps to a lesser degree. From this perspective, it is possible that OspF inhibition of cell wall biogenesis via the CWI pathway accounts for the hypersensitivity of cell division-associated null alleles since successful cell division depends strongly on CWI. Because yeast are under high osmotic pressure, budding and cytokinesis must be tightly coordinated with cell wall growth and remodeling to maintain CWI throughout the process. Defects in the cell wall can result in failure of cell division due to incomplete cytokinesis or cell lysis [43
]. The congruence of chitin biosynthesis genes to OspF expression supports this speculation since chitin is a cell wall constituent involved in cell division [40
]. However, concrete evidence explaining hypersensitivity of the cell division-associated null alleles in terms of OspF inhibition of the CWI pathway is lacking. The hypersensitivity of these null alleles may also be related to OspF inhibition of multiple MAPK cascades. The unanimity of our analyses in implicating CWI is likely due to the fact that this is the only MAPK cascade with demonstrated basal activity in laboratory conditions.
Our study demonstrates how genome-wide yeast screens can help generate testable hypotheses about the roles of bacterial effector proteins in pathogenesis. The multiple perspectives gained from two genome-wide experiments (as well as from multiple analyses of individual experiments) allowed us to effectively “triangulate” the process being perturbed by OspF, as well as to identify the nature of the perturbation. Other choices of screens, e.g., genome-wide identification of OspF binding targets using protein arrays or coimmunoprecipitation assays or identification of toxicity-suppressing conditions in synthetic liquid media (in which OspF is more toxic to yeast), would likely have resulted in a different path to the same discovery. For example, a preliminary genome-wide screen for suppressors of OspF toxicity in liquid media suggested that Δsac7 suppresses OspF toxicity. SAC7 is a RhoGAP whose absence should lead to increased activation of Rho1, so it makes sense, given what we establish in this paper, that it would suppress OspF toxicity.
Genome-wide screens are potentially applicable to any bacterial effector that exhibits evidence of conserved activity in yeast such as toxicity or conserved localization. Hypersensitivity screens only make sense for effectors that exhibit mild or conditional toxicity, but other kinds of screens can be applied to extremely toxic effectors. For example, Alto and colleagues recently demonstrated how genome-wide suppressor screens of the yeast deletion strain collection can identify the cellular targets of effector proteins whose expression severely inhibits yeast growth [10
]. They confirmed that Shigella
IpgB2 mimics activated Rho1 in yeast when they observed that three deletion strains, all downstream components of the MAPK signaling cascade regulated by Rho1, were resistant to IpgB2 toxicity. In this case, since expression of IpgB2 activates a nonessential signaling pathway, they were able to isolate suppressors that blocked persistent activation of the pathway. However, if effector proteins are toxic because they directly target an essential cellular process, it may not be possible to isolate suppressors by screening yeast strains that contain null alleles of nonessential genes. Alternatively, suppressors might be found by screening for yeast genes whose overexpression suppresses toxicity. A genome-wide library of such constructs is now available [44
In summary, this study exemplifies how contemporary analytic tools and the simplest of eukaryotes, S. cerevisiae, can contribute to the study of bacterial effectors. Even when bacterial proteins target cellular processes like innate immunity, which are not conserved among all eukaryotes, if elements of these processes like MAPK signaling cascades are conserved, then screens in yeast can be helpful in elucidating a function for the effector proteins. This systems approach should be generally applicable to numerous microbial virulence proteins. Moreover, since this approach only requires bacterial DNA, it should be particularly valuable for pathogens difficult to genetically manipulate or dangerous to culture.