From striking out to striking gold: discovering type VI secretion targets bacteria

Summary

Specialized secretion systems are infamous for their contribution to host–pathogen interactions. Our discovery that the type VI secretion system delivers toxins between bacterial cells has broadened our understanding of how both pathogens and non-pathogens interact with one another, whether within or outside of the host.

We microbiologists take pride in our sterile technique, but sometimes it may be in our best interest to do a bit less autoclaving. For instance, if we as a field hadn’t been so obsessed with sterility, the discovery of CRISPR might have happened at least a few years earlier. This focus on pure culture experiments likely also postponed the recognition of the predominant function of the bacterial type VI secretion system (T6SS) in interbacterial interactions. In 2010, our group reported in Cell Host and Microbe that the activity of this pathway can be a key determinant of survival when two Gram-negative bacterial cells make contact, opening a new direction for the burgeoning field of interbacterial antagonism (Hood et al., 2010).

From striking out to striking gold

When the Mougous lab started at the University of Washington in late 2007, it was with a mix of excitement and anxiety. Excitement, because we sensed that the newly discovered secretion system we were charged with characterizing did something cool. Anxiety, because we had no idea what that was. The prevailing thought was that it would fit the mold of other specialized secretion systems and contribute to pathogenesis via effector delivery. We hoped that identifying substrates of the secretion system, which were virtually unknown at that time, could direct us toward its physiological function.

A lot of things fell into place in our search. As a postdoc in John Mekalanos’ lab, Joseph had studied regulation of a T6SS in the opportunistic pathogen Pseudomonas aeruginosa (Mougous et al., 2007). These studies provided a means for tricking the bacterium into turning the system “on” under conditions in which it would otherwise be firmly switched “off.” In the experiment that yielded the first three Pseudomonas T6S substrates, we compared the extracellular proteome of our “on” strain to a strain in the “off’ state via an inactivating mutation of the secretion system. We collaborated with proteomics guru Dave Goodlett, who was kind enough to let Pragya Singh, a senior graduate student in his lab, conduct what amounted to a mini-sabbatical with our group. Pragya’s expertise allowed us to surmount the analytical hurdles we encountered, while Tuzun Guvener, an exceptionally careful senior researcher in the Mougous lab, ironed out methods for protein preparation. Together, their efforts allowed us to identify what turned out to be very low abundance proteins.

Combing through the data with all standard filtering criteria thrown to the wind, we fished out three candidate T6S substrates. We called these proteins type six exported 1-3 (Tsel-3), and quickly realized that the hard work lay ahead since their sequences offered no clues to their function. Luckily, the gene encoding one of our substrates, tse2, had been included in a screen testing ~500 P. aeruginosa open reading frames for yeast toxicity. Remarkably, Tse2 (then known as the hypothetical product of PA2702), was second only to a well established toxin in its lethality (Arnoldo et al., 2008). With help from Alex Merz’s lab, we confirmed the high degree of toxicity elicited by Tse2 and found, initially to our dismay, that Tse1 and Tse3 did not share this property of Tse2.

Around the same time the yeast experiments were taking place, Rachel Hood, a newly minted lab technician, began to systematically inactivate each of the tse genes and those immediately adjacent to them. This work was going smoothly until she attempted to knock out the gene downstream of tse2. Repeated attempts to generate this mutation failed, but simultaneously mutating tse2 and its neighbor was straightforward. Together with its toxic effects in yeast, this gave us the first hint that Tse2 was part of a toxin–immunity pair. FoSheng Hsu, a talented technician in the lab, went on to show that Tse2 and its adjacently-encoded immunity protein directly interact.

At this point, we were thrilled to have identified T6S substrates and to learn that one of these was part of a toxin–immunity system. We submitted the first draft of our manuscript to Cell Host & Microbe in August 2009. While the paper was out for review, we were busy sorting out whether Tse2 toxicity extended to other organisms. Was Tse2 toxic to bacteria? Yes; expressed ectopically, it inhibited the growth of every Gram-negative bacterial species we tested, including P. aeruginosa, but only in the absence of the immunity protein. Was Tse2 toxic to other types of eukaryotic cells? Again, yes; transfected HeLa cells succumbed to the toxin.

With this promising preliminary host cell toxicity data, Joseph rushed off to his colleague Arne Rietsch’s lab in Cleveland to determine whether Tse proteins could be delivered to mammalian cells by the T6SS. Dozens of experiments later, he returned to Seattle empty handed.

The reviews from Cell Host & Microbe came back and the news could have been better. We were given what many call a “soft rejection.” Two of the three reviewers expressed concerns about the suitability of the study for Cell Host & Microbe. As one reviewer pointed out, “the T6SS is not associated with cytotoxicity or virulence in an animal model, suggesting that its function is not related to killing host cells.” Another wrote, “Their analysis, though valuable and interesting, does not really increase our understanding of Tse2’s role in T6SS interactions with host cells.” The reviewers were justified in pointing this out, but we were starting to suspect that it would be impossible to prove the substrates targeted host cells. Evidence was mounting that P. aeruginosa delivered Tse2 to other bacteria through its T6SS. Indeed, to rationalize the toxicity of Tse2 toward bacteria, we had entertained this possibility in the wildly speculative discussion of our first submission: “The ability of Tse2 to impact growth of a bacterial cell raises the intriguing possibility that its role may involve assisting P. aeruginosa in competition with other bacteria.” We supported this point with three pieces of evidence from the literature: “1) the secretion system is present and conserved in many non-pathogenic, solitary bacteria […], 2) the cumulative results of genome-wide screens for virulence factors suggests that loss-of-function mutations in most T6SSs do not greatly impact virulence, and 3) there is an experimentally demonstrated structure-function relationship of extracellular components of the secretion apparatus to bacteriophage [(Bonemann et al., 2009; Kanamaru, 2009; Mougous et al, 2006; Pukatzki et al., 2007)].”

Had we been more perceptive, we might have noticed that other findings – some dating back almost a decade – hinted that the T6SS mediates interbacterial interactions. In 2002, Das and colleagues observed that a T6SS mutant of Vibrio cholerae undergoes conjugation with E. coli at a higher frequency than wild-type (Das et al., 2002). (We now recognize that the T6SS of V. cholerae kills E. coli, effectively preventing conjugation.) Another hint came out of the Greenberg lab, who were practically our neighbors at the time. Their genome-wide screen for Proteus mirabilis mutants with defects in kin recognition hit a T6S gene cluster (Gibbs et al., 2008).

These inklings, together with Joseph’s defeat in Cleveland, led us to try our hand at interbacterial competition experiments. We knew that the T6SS of P. aeruginosa is quiescent under standard cultivation conditions, so we took advantage of a mutant strain with an activated system. Since we had defined a specific immunity protein that prevents P. aeruginosa from killing by Tse2, we used a strain lacking this protective factor as our recipient. Even with the deck stacked in our favor by these measures, our initial competition experiments failed. We had not accounted for one important variable. One day we decided that rather than mixing the cells together in liquid media, we should force P. aeruginosa cells into close, long-term contact. DNA transfer by conjugation works most efficiently when cells are placed onto an agar plate, so maybe this was also the case for delivery of substrates by the T6SS.

And voila! Just hours after mixing an equal number of donor and recipient cells on a plate, the donor dominated. Further experiments left no doubt that this result was a consequence of T6S-catalyzed transfer of Tse2. The final set of bacterial competition data that made it into the paper came just two days before we resubmitted, and dramatically changed not only the thesis of the study but the future of the Mougous lab.

Full steam ahead

The publication of our finding that P. aeruginosa can deliver a toxin into another bacterial cell via a T6SS led to a period of refocusing. On the heels of this work, we highlighted a large body of experimental findings inconsistent with the widespread involvement of T6S in mediating host cell interactions, and issued a call for others to consider its role in bacterial antagonism (Schwarz et al., 2010a). Following our own advice, we investigated the function of the five T6SSs of Burkholderia thailandensis. Out of these experiments came the first demonstration that a T6SS can target toxins between bacterial species (Schwarz et al., 2010b). By deciphering the activities of the substrates neglected in our first paper, Tse1 and Tse3, we garnered key biochemical support for our hypothesis that the T6SS evolved to target bacteria. These proteins turned out to be lytic enzymes that degrade peptidoglycan, a molecule not synthesized outside of the bacterial kingdom (Russell et al., 2011). In this same study, we found that specific cognate immunity genes accompany each substrate. The generality of this pattern across diverse organisms would be one feature that helped identify the first effector superfamily, a broadly distributed group of T6S substrates with related activity (Russell et al., 2012).

With the contributions of many groups – many of which we regrettably cannot highlight here – the T6S field has progressed immensely since those early days. The diversity of organisms shown to employ the T6SS for interbacterial targeting extends well beyond what we could have envisioned when we first described Tse2 (see timeline). Not only are these pathways found throughout Proteobacteria, a related secretion system that mediates interbacterial antagonism is prevalent in the Bacteroidetes phylum (Russell et al., 2014b). The demonstrated molecular targets of T6S toxins to date encompass several essential structures, including cellular membranes, electron carriers, the nucleoid, and the cell wall (Russell et al., 2014a).

Aided by several landmark structural studies, research into the mechanism by which the T6SS recognizes and translocates substrates has kept pace with advances in understanding its physiological function. Three distinct T6S apparatus protein assemblages have been identified: phage tail- and baseplate-related complexes, and a membrane complex containing homologs of type IV secretion system proteins. The visualization of an extended tubule formed by two conserved T6S proteins was an important advance in defining the phage tail-like complex (Bonemann et al., 2009). It is now clear that this structure is dynamic and behaves in a manner similar to contractile phage tails (Basler et al., 2012). More recently, the ultrastructure of the membrane complex formed by three T6SS proteins was determined. The unique, pentameric, closed, membrane-spanning complex reported suggested that exiting T6S substrates may promote channel opening through an iris-like mechanism (Durand et al., 2015). Ongoing studies will no doubt resolve the structure of the remaining protein assemblage (the baseplate), and help to complete the picture of how these complexes interact to facilitate protein delivery.

Full circle

The studies leading to the identification of T6S were motivated by an interest in bacterial-host interactions (Box 1). Although we subsequently learned that T6S primarily targets bacteria, questions about how the system influences the host, directly or indirectly, have persisted. A prescient reviewer of the Hood et al. study suggested in 2009 that we consider “a potential role of the T6SS effectors in mediating competition with other microbes in a host niche.” The first study demonstrating such a scenario was finally published in 2016. Sana et al. showed that antibacterial T6S proteins facilitate invasion of gut commensal communities by the enteric pathogen Salmonella typhimurium (Sana et al., 2016). It would be surprising if other examples of T6S-mediated interactions did not emerge soon, whether between pathogens and commensals, or between pathogenic species in polymicrobial infections. Similarly, the finding that genes encoding a T6-like pathway are widespread in Bacteriodes spp. suggests the pathway also plays an important role in establishing or maintaining the healthy human gut community (Russell et al., 2014b). Using the T6S as a handle for manipulating gut microbial communities is an exciting prospect to emerge from this possibility.

Box: “IAHP … who, what?”.

“Names and attributes must be accommodated to the essence of things, and not the essence to the names, since things come first and names afterwards.”

- Galileo Galilei, Discoveries and Opinions of Galileo

Finally, some of our own recent findings serve as a good reminder that “what goes around comes around.” While antibacterial targeting may be the ancestral function of the T6SS, in some cases the system clearly has evolved to specifically target host cells. This appears to be the case for the T6S-like pathway encoded on the Francisella pathogenicity island (FPI). This system is sufficiently distinct from the pathways found in other Proteobacteria and Bacteroidetes to merit classification as its own T6SS subtype, perhaps reflecting the extent of evolutionary divergence necessary to result in a specialization switch (Russell et al., 2014b). Just as defining the functions of the Tse proteins was paramount for advancing our understanding of bacterial-targeting T6SSs, we anticipate that a detailed comprehension of the FPI-encoded T6SS hinges on characterizing its substrates. Using a substrate identification approach not unlike that which led to the discovery of the Tse proteins, we have taken an important step toward this end (Eshraghi et al., 2016). Fittingly, the Francisella T6S substrates – just like the Tse proteins in their time – are a group of cryptic proteins with no known function. This is one of several frontiers that should leave researchers in the T6 field with much new ground to explore in upcoming years.

The T6S field began when Manning and colleagues identified a protein they called Hcp (hemolysin coregulated protein) secreted by Vibrio cholerae (Williams et al., 1996). They noted that Hcp lacks a recognizable signal sequence and proposed that it traverses the outer membrane via a “novel mechanism of secretion.” In a foretelling series of experiments, the authors found that hcp inactivation does not alter V. cholerae host colonization or virulence, and that the Hcp protein itself has no apparent cytotoxic activity. Fueled by the first sequencing revolution, a series of key bioinformatic observations followed. Wang et al. reported the linkage between rhs (recombination hot spot) genes and homologues of hcp from V. cholerae in 1998 (Wang et al., 1998). Five years later, Das and colleagues noticed that these rhs elements were often associated with a phylogenetically widespread cluster of genes they named IAHPs (IcmF associated homologous proteins) (Das and Chaudhuri, 2003). IcmF was of interest because of its role in intracellular growth of Legionella pneumophila. Based on the predicted sub-cellular localization of the conserved group of proteins encoded by IAHP clusters, Das and colleagues deduced that they likely encoded a secretion apparatus (Das and Chaudhuri, 2003).

The genetic foundation was laid – all that was missing was a functional link between Hcp secretion and the IAHP gene clusters. This came serendipitously out of work by Ka Yin Leung’s group that sought to associate attenuation of the fish pathogen Edwardsiella tarda with changes in its extracellular proteome (Rao et al., 2004). Among the proteins less abundantly secreted in attenuated E. tarda mutants was an Hcp homolog (EvpC). Without a genome sequence available, the authors employed genome walking to determine the genomic context of hcp. E. tarda hcp turns out to be embedded in an IAHP gene cluster, and the authors also showed that a non-polar deletion in evpB (now known as tssC) abrogates Hcp secretion. At this point, there was both genetic and biochemical evidence that IAHP gene clusters encode a novel secretion system. However, it would be another two years before Mekalanos and colleagues found that protein secretion by the IAHP cluster of V. cholerae influences predation by amoebae and named the pathway “type VI secretion” (Pukatzki et al., 2006).

Figure 1: Highlights from the history of T6S research.

Upper panel – timeline for the chronology of significant studies discussed in the text. The month of publication and first author of each study are indicated. Complete citations are included in the main body. Bottom panel – illustration of the diverse bacterial species with an experimentally demonstrated bacterial cell-targeting T6SS. Background colors correspond to those in the timeline and denote the year in which the antibacterial activity of a T6SS for a given organism was first demonstrated. The surname(s) of the corresponding author(s) for each study is provided in parenthesis. Due to space limitations, we are unable to provide complete references.