1. Microbiology and Infectious Disease
Download icon

A broadly distributed toxin family mediates contact-dependent antagonism between gram-positive bacteria

  1. John C Whitney
  2. S Brook Peterson
  3. Jungyun Kim
  4. Manuel Pazos
  5. Adrian J Verster
  6. Matthew C Radey
  7. Hemantha D Kulasekara
  8. Mary Q Ching
  9. Nathan P Bullen
  10. Diane Bryant
  11. Young Ah Goo
  12. Michael G Surette
  13. Elhanan Borenstein
  14. Waldemar Vollmer
  15. Joseph D Mougous Is a corresponding author
  1. University of Washington School of Medicine, United States
  2. Newcastle University, United Kingdom
  3. University of Washington, United States
  4. McMaster University, Canada
  5. Advanced Light Source, United States
  6. Northwestern University, United States
  7. Santa Fe Institute, United States
  8. Howard Hughes Medical Institute, University of Washington School of Medicine, United States
Research Article
Cited
3
Views
2,233
Comments
0
Cite as: eLife 2017;6:e26938 doi: 10.7554/eLife.26938

Abstract

The Firmicutes are a phylum of bacteria that dominate numerous polymicrobial habitats of importance to human health and industry. Although these communities are often densely colonized, a broadly distributed contact-dependent mechanism of interbacterial antagonism utilized by Firmicutes has not been elucidated. Here we show that proteins belonging to the LXG polymorphic toxin family present in Streptococcus intermedius mediate cell contact- and Esx secretion pathway-dependent growth inhibition of diverse Firmicute species. The structure of one such toxin revealed a previously unobserved protein fold that we demonstrate directs the degradation of a uniquely bacterial molecule required for cell wall biosynthesis, lipid II. Consistent with our functional data linking LXG toxins to interbacterial interactions in S. intermedius, we show that LXG genes are prevalent in the human gut microbiome, a polymicrobial community dominated by Firmicutes. We speculate that interbacterial antagonism mediated by LXG toxins plays a critical role in shaping Firmicute-rich bacterial communities.

https://doi.org/10.7554/eLife.26938.001

eLife digest

Most bacteria live in densely colonized environments, such as the human gut, in which they must constantly compete with other microbes for space and nutrients. As a result, bacteria have evolved a wide array of strategies to directly fight their neighbors. For example, some bacteria release antimicrobial compounds into their surroundings, while others ‘inject’ protein toxins directly into adjacent cells.

Bacteria can be classified into two groups known as Gram-positive and Gram-negative. Previous studies found that Gram-negative bacteria inject toxins into neighboring cells, but no comparable toxins in Gram-positive bacteria had been identified. Before a bacterium can inject molecules into an adjacent cell, it needs to move the toxins from its interior to the cell surface. It had been suggested that a transport system in Gram-positive bacteria called the Esx pathway may export toxins known as LXG proteins. However, it was not clear whether these proteins help Gram-positive bacteria to compete against other bacteria.

Whitney et al. studied the LXG proteins in Gram-positive bacteria known as Firmicutes. The experiments reveal that Firmicutes found in the human gut possess LXG genes. A Firmicute known as Streptococcus intermedius produces three LXG proteins that are all toxic to bacteria. To avoid being harmed by its own LXG proteins, S. intermedius also produces matching antidote proteins. Further experiments show that LXG proteins are exported out of S. intermedius cells and into adjacent competitor bacteria by the Esx pathway. Examining one of these LXG proteins in more detail showed that it can degrade a molecule that bacteria need to make their cell wall.

Together, these findings suggest that LXG proteins may influence the species living in many important microbial communities, including the human gut. Changes in the communities of gut microbes have been linked with many diseases. Therefore, understanding more about how the LXG proteins work may help us to develop ways to manipulate these communities to improve human health.

https://doi.org/10.7554/eLife.26938.002

Introduction

Bacteria in polymicrobial environments must persist in the face of frequent physical encounters with competing organisms. Studies have revealed Gram-negative bacterial species contend with this threat by utilizing pathways that mediate antagonism toward contacting bacterial cells (Konovalova and Søgaard-Andersen, 2011). For instance, Proteobacteria widely employ contact-dependent inhibition (CDI) to intoxicate competitor cells that share a high degree of phylogenetic relatedness (Hayes et al., 2014). Additionally, both Proteobacteria and bacteria belonging to the divergent phylum Bacteroidetes deliver toxins to competitor Gram-negative cells in an indiscriminate fashion through the type VI secretion system (T6SS) (Russell et al., 2014a, 2014b). Although toxin delivery by CDI and the T6SS is mechanistically distinct, cells harboring either pathway share the feature of prohibiting self-intoxication with immunity proteins that selectively inactivate cognate toxins through direct binding.

Few mechanisms that mediate direct antagonism between Gram-positive bacteria have been identified. In Bacillus subtilis, Sec-exported proteins belonging to the YD-repeat family have been shown to potently inhibit the growth of contacting cells belonging to the same strain (Koskiniemi et al., 2013); however, to our knowledge, a pathway that mediates interspecies antagonism between Gram-positive bacteria has not been identified. Given that Gram-positive and Gram-negative bacteria inhabit many of the same densely populated polymicrobial environments (e.g. the human gut), it stands to reason that the former should also possess mechanisms for more indiscriminate targeting of competing cells.

Contact-dependent toxin translocation between bacteria is primarily achieved using specialized secretion systems. Gram-negative export machineries of secretion types IV, V, and VI have each been implicated in this process (Aoki et al., 2005; Hood et al., 2010; Souza et al., 2015). A specialized secretion system widely distributed among Gram-positive bacteria is the Esx pathway (also referred to as type VII secretion) (Abdallah et al., 2007). This pathway was first identified in Mycobacterium tuberculosis, where it plays a critical role in virulence (Stanley et al., 2003). Indeed, attenuation of the vaccine strain M. bovis BCG can be attributed to a deletion inactivating ESX-1 secretion system present in virulence strains (Lewis et al., 2003; Pym et al., 2003). Subsequent genomic studies revealed that the Esx pathway is widely distributed in Actinobacteria, and that a divergent form is present in Firmicutes (Gey Van Pittius et al., 2001; Pallen, 2002). Though they share little genetic similarity, all Esx pathways studied to-date utilize a characteristic FtsK-like AAA+ ATPase referred to as EssC (or EccC) to catalyze the export of one or more substrates belonging to the WXG100 protein family (Ates et al., 2016). Proteins in this family, including ESAT-6 (EsxA) and CFP10 (EsxB) from M. tuberculosis, heterodimerize in order to transit the secretion machinery.

The presence of the Esx secretion system in environmental bacteria as well as commensal and pathogenic bacteria that specialize in colonizing non-sterile sites of their hosts, suggests that the pathway may be functionally pliable. Supporting this notion, ESX-3 of M. tuberculosis is required for mycobactin siderophore-based iron acquisition and the ESX-1 and ESX-4 systems of M. smegmatis are linked to DNA transfer (Gray et al., 2016; Siegrist et al., 2009). In Firmicutes, a Staphylococcus aureus Esx-exported DNase toxin termed EssD (or EsaD) has been linked to virulence and contact-independent intraspecies antibacterial activity (Cao et al., 2016; Ohr et al., 2017).

Aravind and colleagues have noted that Esx secretion system genes are often linked to genes encoding polymorphic toxins belonging to the LXG protein family (Zhang et al., 2012). Analogous to characteristic antimicrobial polymorphic toxins of Gram-negative bacteria, the LXG proteins consist of a conserved N-terminal domain (LXG), a middle domain of variable length, and a C-terminal variable toxin domain. The LXG domain is predicted to adopt a structure resembling WXG100 proteins, thus leading to speculation that these proteins are Esx secretion system substrates (Zhang et al., 2011). Despite the association between LXG proteins and the Esx secretion system, to-date there are no experimental data linking them functionally. However, an intriguing study performed by Hayes and colleagues demonstrated antibacterial properties of B. subtilis LXG RNase toxins via heterologous expression in E. coli (Holberger et al., 2012). This growth inhibition was alleviated by co-expression of immunity determinants encoded adjacent to cognate LXG genes. We show here that LXG proteins transit the Esx secretion system of Streptococcus intermedius (Si) and function as antibacterial toxins that mediate contact-dependent interspecies antagonism.

Results

LXG proteins are Esx secretion system substrates

We initiated our investigation into the function of LXG proteins by characterizing the diversity and distribution of genes encoding these proteins across all sequenced genomes from Firmicutes. As noted previously, the C-terminal domains in the LXG family members we identified are highly divergent, exhibiting a wide range of predicted activities (Figure 1a) (Zhang et al., 2012). LXG protein-encoding genes are prevalent and broadly distributed in the classes Clostridiales, Bacillales and Lactobacillales (Figure 1A). Notably, a significant proportion of organisms in these taxa are specifically adapted to the mammalian gut environment. Indeed, we find that LXG genes derived from reference genomes of many of these gut-adapted bacteria are abundant in metagenomic datasets from human gut microbiome samples (Figure 1A and Figure 1—figure supplement 1). An LXG toxin that is predicted to possess ADP-ribosyltransferase activity – previously linked to interbacterial antagonism in Gram-negative organisms – was particularly abundant in a subset of human gut metagenomes (Zhang et al., 2012). Close homologs of this gene are found in Ruminococcus, a dominant taxa in the human gut microbiome, potentially explaining the frequency of this gene (Wu et al., 2011).

Figure 1 with 1 supplement see all
The LXG protein family contains diverse toxins that are broadly distributed in Firmicutes and found in the human gut microbiome.

(A) Dendogram depicts LXG-containing genera within Firmicutes, clustered by class and order. Circle size indicates the number of sequenced genomes searched within each genus and circle color represents percentage of those found to contain at least one LXG protein. For classes or orders in which no LXG domain-containing proteins were found, the number of genera evaluated is indicated in parentheses; those consisting of Gram-negative organisms are boxed with dashed lines. Grey boxes contain predicted domain structures for representative divergent LXG proteins. Depicted are LXG-domains (pink), spacer regions (light grey) and C-terminal polymorphic toxin domains (NADase, purple; non-specific nuclease, orange; AHH family nuclease, green; ADP-ribosyltransferase, blue; lipid II phosphatase based on orthology to TelC (defined biochemically herein), yellow; EndoU family nuclease, brown; unknown activity, dark grey). (B) Heatmap depicting the relative abundance (using logarithmic scale) of selected LXG genes detected in the Integrated Gene Catalog (IGC). A complete heatmap is provided in Figure 1—figure supplement 1. Columns represent individual human gut metagenomes from the IGC database and rows correspond to LXG genes. Grey lines link representative LXG toxins in (A) to their corresponding (≥95% identity) IGC group in (B).

https://doi.org/10.7554/eLife.26938.003

We next sought to determine whether LXG proteins are secreted via the Esx pathway. The toxin domain of several of the LXG proteins we identified shares homology and predicted catalytic residues with M. tuberculosis TNT, an NAD+-degrading (NADase) enzyme (Figure 2—figure supplement 1A) (Sun et al., 2015). Si, a genetically tractable human commensal and opportunistic pathogen, is among the bacteria we identified that harbor a gene predicted to encode an NADase LXG protein (Claridge et al., 2001); we named this protein TelB (Toxin exported by Esx with LXG domain B). Attempts to clone the C-terminal toxin domain of TelB (TelBtox) were initially unsuccessful, suggesting the protein exhibits a high degree of toxicity. Guided by the TNT structure, we circumvented this by assembling an attenuated variant (H661A) that was tolerated under non-induced conditions (TelBtox*) (Figure 2—figure supplement 1A) (Sun et al., 2015). Induced expression of TelBtox* inhibited E. coli growth and reduced cellular NAD+ levels (Figure 2A, Figure 2—figure supplement 1B). The extent of NAD+ depletion mirrored that catalyzed by expression of a previously characterized interbacterial NADase toxin, Tse6, and importantly, intracellular NAD+ levels were unaffected by an unrelated bacteriostatic toxin, Tse2 (Hood et al., 2010; Whitney et al., 2015). Furthermore, substitution of a second predicted catalytic residue of TelB (R626A), abrogated toxicity of TelBtox* and significantly restored NAD+ levels (Figure 2—figure supplement 1B–C).

Figure 2 with 1 supplement see all
LXG-domain proteins of S. intermedius are secreted by the Esx-pathway.

(A) NAD+ levels in E. coli cells expressing a non-NAD+ -degrading toxin (Tse2), the toxin domain of a known NADase (Tse6tox), an inducibly toxic variant of the C-terminal toxin domain of TelB (TelBtox*), a variant of TelBtox* with significantly reduced toxicity (TelBtox*R626A) and TelBtox* co-expressed with its cognate immunity protein TipB. Cellular NAD+ levels were assayed 60 min after induction of protein expression and were normalized to untreated cells. Mean values (n = 3) ± SD are plotted. Asterisks indicate statistically significant differences in NAD+ levels compared to vector control (p<0.05). (B) NAD+ consumption by culture supernatants from the indicated Si strains. Fluorescent images of supernatant droplets supplemented with 2 mM NAD+ for 3 hr; brightness is proportional to NAD+ concentration and was quantified using densitometry. Mean values ± SD (n = 3) are plotted. Asterisks indicate statistically significant differences in NAD+ turnover compared to wild-type SiB196 (p<0.05). (C) Regions of the SiB196 genome encoding Esx-exported substrates. Genes are colored according to functions encoded (secreted Esx structural components, orange; secreted LXG toxins, dark purple; immunity determinants, light purple; WXG100-like proteins, green; other, grey). (D) Western blot analysis of TelC secretion in supernatant (Sup) and cell fractions of wild-type or essC-inactivated SiB196.

https://doi.org/10.7554/eLife.26938.005

Determination of the biochemical activity of TelB provided a means to test our hypothesis that LXG proteins are substrates of the ESX secretion pathway. Using an assay that exploits fluorescent derivatives of NAD+ that form under strongly alkaline conditions, we found that concentrated cell-free supernatant of an Si strain containing telB (SiB196) possesses elevated levels of NADase activity relative to that of a strain lacking telB (Si27335) (Figure 2B) (Johnson and Morrison, 1970; Olson et al., 2013; Whiley and Beighton, 1991). Furthermore, the NADase activity present in the supernatant of SiB196 was abolished by telB inactivation. Export of Esx substrates relies on EssC, a translocase with ATPase activity (Burts et al., 2005; Rosenberg et al., 2015). Inactivation of essC also abolished NADase activity in the supernatant of SiB196, suggesting that TelB utilizes the Esx pathway for export.

The genome of SiB196 encodes two additional LXG proteins, which we named TelA and TelC (Figure 2C). To determine if these proteins are also secreted in an Esx-dependent fashion, we collected cell-free supernatants from stationary phase cultures of wild-type and essC-deficient SiB196. Extensive dialysis was used to reduce contamination from medium-derived peptides and the remaining extracellular proteins were precipitated and identified using semi-quantitative mass spectrometry (Liu et al., 2004). This technique revealed that each of the LXG proteins predicted by the Si genome is exported in an Esx-dependent manner (Table 1). Western blot analysis of TelC secretion by wild-type and the essC-lacking mutant further validated Esx-dependent export (Figure 2D). Together, these data indicate that LXG proteins are substrates of the Esx secretion system.

Table 1

The Esx-dependent extracellular proteome of S. intermedius B196.

https://doi.org/10.7554/eLife.26938.007
Locus tagWild-typeΔessCRelative abundance
(Wild-type/ΔessC)
Esx functionName
SIR_0169*19.670Not detected in ΔessCLXG proteinTelA
SIR_017614.670Not detected in ΔessCStructural componentEsaA
SIR_148912.000Not detected in ΔessCLXG proteinTelC
SIR_15169.330Not detected in ΔessC-Trigger Factor
SIR_01795.330Not detected in ΔessCLXG proteinTelB
SIR_0166140.0017.488.01Structural componentEsxA
SIR_027315.332.286.73--
SIR_162615.002.286.58-GroEL
SIR_083212.338.361.48-Enolase
SIR_190449.0037.241.32-Putative serine protease
SIR_138226.0019.761.32-Fructose-bisphosphate aldolase
SIR_064821.6717.481.24-50S ribosomal protein L7/L12
SIR_021247.0039.521.19-Elongation Factor G
SIR_00818.677.601.14-Putative outer membrane protein
SIR_167616.3314.441.13-phosphoglycerate kinase
SIR_152312.6712.920.98-DnaK
SIR_115410.3310.640.97-Putative bacteriocin accessory protein
SIR_102763.0067.640.93-Elongation Factor Tu
SIR_145514.0015.960.88--
SIR_075813.0015.200.86--
SIR_13879.3311.400.82-Putative extracellular solute-binding protein
SIR_049212.3315.200.81-Putative adhesion protein
SIR_103317.6724.320.73--
SIR_135914.0019.760.71-Penicillin-binding protein 3
SIR_001112.3317.480.71-Beta-lactamase class A
SIR_15468.3312.160.69--
SIR_0040101.67160.360.63-Putative stress protein
SIR_160811.0018.240.60-Putative endopeptidase O
SIR_15497.3312.160.60--
SIR_167579.00132.240.60-Putative cell-surface antigen I/II
SIR_141811.3321.280.53-Putative transcriptional regulator LytR
SIR_008011.0021.280.52--
SIR_102528.3363.840.44-Lysozyme
SIR_011310.6724.320.44--
SIR_02978.3324.320.34--
  1. *Rows highlighted in green correspond to proteins linked to the Esx pathway.

  2. Values correspond to average SC (spectral counts) of triplicate biological replicates for each strain.

  3. Functional link of LXG proteins to Esx secretion pathway defined in the study.

Contact-dependent interspecies antagonism is mediated by LXG toxins

The export of LXG proteins by the Esx pathway motivated us to investigate their capacity for mediating interbacterial antagonism. The C-terminal domains of TelA (TelAtox) and TelC (TelCtox) bear no homology to characterized proteins, so we first examined the ability of these domains to exhibit toxicity in bacteria. TelAtox and TelBtox* inhibited growth when expressed in the cytoplasm of E. coli, whereas TelCtox did not exhibit toxicity in this cellular compartment (Figure 3A). Given the capacity of some interbacterial toxins to act on extracellular structures, we assessed the viability of Si cells expressing TelCtox targeted to the sec translocon. In contrast to TelCtox production, overexpression of a derivative bearing a signal peptide directing extracellular expression (ss-TelCtox) exhibited significant toxicity (Figure 3B).

Figure 3 with 2 supplements see all
S. intermedius LXG proteins inhibit bacterial growth and mediate contact-dependent interbacterial antagonism.

(A) Viability of E. coli cells grown on solid media harboring inducible plasmids expressing the C-terminal toxin domains of the three identified SiB196 LXG proteins or an empty vector control. (B) SiB196 colonies recovered after transformation with equal concentrations of constitutive expression plasmids carrying genes encoding the indicated proteins. ss-TelCtox is targeted to the sec translocon through the addition of the secretion signal sequence from S. pneumoniae LysM (SP_0107). Error bars represent ± SD (n = 3). Asterisk indicates a statistically significant difference in Si transformation efficiency relative to TelCtox (p<0.05). (C) Viability of E. coli cells grown on solid media harboring inducible plasmids co-expressing the indicated proteins. Empty vector controls are indicated by a dash. Mean c.f.u. values ± SD (n = 3) are plotted. Asterisks indicate statistically significant differences in E. coli viability relative to vector control (p<0.05) (D) Intra-species growth competition experiments between the indicated bacterial strains. Competing strains were mixed and incubated in liquid medium or on solid medium for 30 hr and both initial and final populations of each strain were enumerated by plating on selective media. The competitive index was determined by comparing final and initial ratios of the two strains. Asterisks indicate outcomes statistically different between liquid and solid medium (n = 3, p<0.05). (E) Intra-species growth competition experiments performed as in (D) except for the presence of a filter that inhibits cell-cell contact. No contact, filter placed between indicated donor and susceptible recipient (∆telBtipB) strains; Contact, donor and susceptible recipient strains mixed on same side of filter. Asterisks indicate statistically different outcomes (n = 3, p<0.05). Note that recipient cell populations have an Esx-independent fitness advantage in these experiments by virtue of their relative proximity to the growth substrate. (F) Inter-species growth competition experiments performed on solid or in liquid (E. faecalis) medium between Si wild-type and ∆essC donor strains and the indicated recipient organisms. Si23775 lacks tipA and tipB and is therefore potentially susceptible to TelA and TelB delivered by SiB196. Asterisks indicate outcomes where the competitive index of wild-type was significantly higher than an ∆essC donor strain (n = 3, p<0.05). Genetic complementation of the mutant phenotypes presented in this figure was confounded by inherent plasmid fitness costs irrespective of the inserted sequence. As an alternative, we performed whole genome sequencing on strains ∆essC, ∆telB, ∆telC, ∆telBtipB, and ∆telCtipC, which confirmed the respective desired mutation as the only genetic difference between these strains. Sequences of these strains have been deposited to the NCBI Sequence Read Archive (BioProject ID: PRJNA388094).

https://doi.org/10.7554/eLife.26938.008

We next evaluated whether the Tel proteins, like the substrates of interbacterial toxin delivery systems in Gram-negative bacteria, are inactivated by genetically linked specialized cognate immunity determinants. By co-expressing candidate open reading frames located downstream of each tel gene, we identified a cognate tip (tel immunity protein) for each toxin (Figure 3B–C and Figure 3—figure supplement 1). We then sought to inactivate each of these factors to generate SiB196 strains sensitive to each of the Tel proteins. In SiB196, telA tipA loci are located immediately upstream of conserved esx genes (Figure 2C). We were unable to generate non-polar telA tipA-inactivated strains, and thus focused our efforts on the other two tel tip loci.

We reasoned that if LXG toxins target non-self cells, this process would occur either through diffusion or by facilitated transfer, the latter of which would likely require cell contact. Since we detect TelA-C secretion in liquid medium, we began our attempts to observe intercellular intoxication with wild-type and toxin-sensitive target cell co-culture. These efforts yielded no evidence of target cell killing or growth inhibition, including when co-incubations were performed at cell densities higher than that achievable through growth (Figure 3D, Figure 3—figure supplement 2A). The application of concentrated supernatants or purified TelC (to a final concentration of 0.1 mg/mL) to sensitive strains also did not produce evidence of toxicity (Figure 3—figure supplement 2B–C). This result is perhaps not surprising given the barrier presented by the Gram-positive cell wall (Forster and Marquis, 2012).

Next, we tested conditions that enforce cell contact. In each of these experiments, donor and recipient strains were grown in pure culture before they were mixed at defined ratios and cultured on a solid surface for 30 hr to promote cell-cell interactions. We observed significant growth inhibition of TelB- or TelC-susceptible strains co-cultured with wild-type, but not when co-cultured with strains lacking telB or telC, respectively (Figure 3D). A strain bearing inactivated essC was also unable to intoxicate a sensitive recipient. In competition experiments performed in parallel wherein the bacterial mixtures were grown in liquid culture, TelB and TelC-susceptible strains competed equally with wild type, suggesting that Esx-mediated intoxication requires prolonged cell contact. To further probe this requirement, we conducted related experiments in which wild-type donor cells were segregated from sensitive recipients by a semi-permeable (0.2 μm pore size) membrane (Figure 3E). This physical separation blocked intoxication, which taken together with the results of our liquid co-culture experiments and our finding that purified TelC is not bactericidal, strongly suggests that the mechanism of Esx-dependent intercellular LXG protein delivery requires immediate cell-cell contact.

In Gram-negative bacteria, some antagonistic cell contact-dependent pathways display narrow target range, whereas others act between species, or even between phyla (Hayes et al., 2014; Russell et al., 2014a). To begin to determine the target range of Esx-based LXG protein delivery, we measured its contribution to SiB196 fitness in interbacterial competition experiments with a panel of Gram-positive and -negative bacteria. The Esx pathway conferred fitness to SiB196 in competition with Si23775, S. pyogenes, and Enterococcus faecalis, an organism from a closely related genus (Figure 3F). On the contrary, the pathway did not measurably affect the competitiveness of SiB196 against Gram-negative species belonging to the phyla Proteobacteria (E. coli, Burkholderia thailandensis, Pseudomonas aeruginosa) or Bacteroidetes (Bacteriodes fragilis). These results demonstrate that the Esx pathway can act between species and suggest that its target range may be limited to Gram-positive bacteria.

TelC targets the bacterial cell wall biosynthetic precursor lipid II

The Esx pathway is best known for its role in mediating pathogen-host cell interactions (Abdallah et al., 2007). Given this precedence, we considered the possibility that the antibacterial activity we observed may not be relevant physiologically. TelB degrades NAD+, a molecule essential for all cellular life, and therefore this toxin is not definitive in this regard. We next turned our attention to TelC, which elicits toxicity from outside of the bacterial cell (Figure 3B). This protein contains a conserved aspartate-rich motif that we hypothesized constitutes its enzymatic active site (Figure 4—figure supplement 1A). To gain further insight into TelC function, we determined the crystal structure of TelCtox to 2.0 Å resolution (Table 2). The structure of TelCtox represents a new fold; it is comprised of distinct and largely α-helical N- and C-terminal lobes (Figure 4A). The single β element of TelCtox is a hairpin that protrudes from the N-terminal lobe. Although TelCtox does not share significant similarity to previously determined structures, we located its putative active site within a shallow groove that separates the N- and C-terminal lobes. This region contains a calcium ion bound to several residues that comprise the conserved aspartate-rich motif. Site-specific mutagenesis of these residues abrogated TelC-based toxicity (Figure 4B,C, Figure 4—figure supplement 1B).

Figure 4 with 3 supplements see all
TelC is a calcium-dependent lipid II phosphatase.

(A) Space-filling representation of the 2.0 Å resolution TelCtox X-ray crystal structure. Protein lobes (red and blue), active site cleft (white) and Ca2+ (green) are indicated. (B) TelCtox structure rotated as indicated relative to (A) with transparent surface revealing secondary structure. (C) Magnification of the TelC active site showing Ca2+ coordination by conserved aspartate residues and water molecules. (D) Viability of S. aureus cells harboring inducible plasmids expressing the indicated proteins or a vector control. ss-TelCtox is targeted for secretion through the addition of the signal sequence encoded by the 5’ end of the hla gene from S. aureus. Mean c.f.u. values ± SD (n = 3) are plotted. Asterisk indicates a statistically significant difference in S. aureus viability relative to vector control (p<0.05) (E) Representative micrographs of S. aureus expressing ss-TelCtox or a vector control. Frames were acquired eight and 12 hr after spotting cells on inducing growth media. (F) Thin-layer chromotography (TLC) analysis of reaction products from incubation of synthetic Lys-type lipid II with buffer (Ctrl), TelCtox, or TelCtox and its cognate immunity protein TipC. (G) Partial HPLC chromatograms of radiolabeled peptidoglycan (PG) fragments released upon incubation of Lys-type lipid II with the indicated purified proteins. Schematics depict PG fragment structures (pentapeptide, orange; N-acetylmuramic acid, dark green; N-acetylglucosamine, light green; phosphate, black). Known fragment patterns generated by PBP1B + LpoB and colicin M serve as controls. (H) TLC analysis of reaction products generated from incubation of buffer (Ctrl), TelCtox or TelCtox and TipC with undecaprenyl phosphate (C55–P) (left) or undecaprenyl pyrophosphate (C55–PP) (right).

https://doi.org/10.7554/eLife.26938.011
Table 2

X-ray data collection and refinement statistics.

https://doi.org/10.7554/eLife.26938.015
TelC202-CT (Semet)
Data Collection
Wavelength (Å)0.979
Space groupC2221
Cell dimensions
a, b, c (Å)127.4, 132.7, 58.3
α, β, γ (°)90.0, 90.0, 90.0
Resolution (Å)49.20–1.98 (2.03–1.98)*
Total observations891817
Unique observations34824
Rpim (%)6.6 (138.5)
II11.4 (0.8)
Completeness (%)100.0 (99.9)
Redundancy25.6 (23.4)
Refinement
Rwork / Rfree (%)22.4/24.6
Average B-factors (Å2)53.8
No. atoms
 Protein2539
 Ligands3
 Water145
Rms deviations
 Bond lengths (Å)0.008
 Bond angles (°)0.884
Ramachandran plot (%)
 Total favored96.9
 Total allowed99.7
Coordinate error (Å)0.28
PDB code5UKH
  1. *Values in parentheses correspond to the highest resolution shell.

We next assessed the morphology of cells undergoing intoxication by TelCtox. Due to the potent toxicity of TelCtox in Si, we employed an inducible expression system in S. aureus as an alternative. S. aureus cells expressing extracellularly-targeted TelCtox exhibited significantly reduced viability (Figure 4D), and when examined microscopically, displayed a cessation of cell growth followed by lysis that was not observed in control cells (Figure 4E, Videos 12). Despite eliciting effects consistent with cell wall peptidoglycan disruption, isolated cell walls treated with TelCtox and peptidoglycan recovered from cells undergoing TelC-based intoxication showed no evidence of enzymatic digestion (Figure 4—figure supplement 2A–D). These data prompted us to consider that TelC corrupts peptidoglycan biosynthesis, which could also lead to the lytic phenotype observed (Harkness and Braun, 1989).

Video 1
Time-lapse series of S. aureus USA300 pEPSA5 growth.

Cells were imaged every 10 min.

https://doi.org/10.7554/eLife.26938.016
Video 2
Time-lapse series of S. aureus USA300 pEPSA5::ss-telCtox growth.

Cells were imaged every 10 min.

https://doi.org/10.7554/eLife.26938.017

The immediate precursor of peptidoglycan is lipid II, which consists of the oligopeptide disaccharide repeat unit linked via pyrophosphate to a lipid carrier (Vollmer and Bertsche, 2008). Likely due to its distinctive and conserved structure, lipid II is the target of diverse antibacterial molecules (Breukink and de Kruijff, 2006; Oppedijk et al., 2016). To test activity against lipid II, we incubated the molecule with purified TelCtox. Analysis of the reaction products showed that TelCtox cleaves lipid II – severing the molecule at the phosphoester linkage to undecaprenyl (Figure 4F–G, Figure 4—figure supplement 3A). Reaction products were confirmed by mass spectrometry and inclusion of TipC inhibited their formation. Consumption of lipid II for peptidoglycan assembly generates undecaprenyl pyrophosphate (UPP), which is converted to undecaprenyl phosphate (UP), and transported inside the cell. The UP molecule then reenters peptidoglycan biosynthesis or is utilized as a carrier for another essential cell wall constituent, wall teichoic acid (WTA). Our experiments showed that TelCtox is capable of hydrolyzing cleaved undecaprenyl derivatives but displays a strict requirement for the pyrophosphate group (Figure 4H), indicating the potential for TelC to simultaneously disrupt two critical Gram-positive cell wall polymers. Consistent with its ability to inhibit a conserved step in peptidoglycan biosynthesis, TelC exhibited toxicity towards diverse Gram-positive species including Si (Figure 2B), S. aureus (Figure 4D) and E. faecalis (Figure 4—figure supplement 3B). These data do not explain our observation that cytoplasmic TelC is non-toxic, as the substrates we defined are present in this compartment. The substrates may be inaccessible or TelC could be inactive in the cytoplasm. It is worth noting that TelC contains a calcium ion bound at the interface of its N- and C-terminal lobes. Many secreted proteins that bind calcium utilize the abundance of the free ion in the milieu to catalyze folding. Taken together, our biochemical and phenotypic data strongly suggest that TelC is a toxin directed specifically against bacteria. While we cannot rule out that TelC may have other targets, we find that its expression in the cytoplasm or secretory pathway of yeast does not impact the viability of this model eukaryotic cell (Figure 4—figure supplement 3C–D).

WXG100-like proteins bind cognate LXG proteins and promote toxin export

The majority of Esx substrates identified to-date belong to the WXG100 protein family. These proteins typically display secretion co-dependency and are essential for apparatus function. M. tuberculosis ESX-1 exports two WXG100 proteins, ESAT-6 and CFP10, and the removal of either inhibits the export of other substrates (Ates et al., 2016; Renshaw et al., 2002). LXG proteins do not belong to the WXG100 family; thus, we sought to determine how the Tel proteins influence Esx function in Si. Using Western blot analysis to measure TelC secretion and extracellular NADase activity as a proxy for TelB secretion, we found that telB- and telC-inactivated strains of Si retain the capacity to secrete TelC and TelB, respectively (Figure 5A–B). These data indicate that TelB and TelC are not required for core apparatus function and do not display secretion co-dependency.

Figure 5 with 1 supplement see all
LXG domain proteins are independently secreted and require interaction with cognate WXG100-like partners for export.

(A) NAD+ consumption assay of culture supernatants of the indicated SiB196 strains. Mean densitometry values ± SD (n = 3) are plotted. Asterisk indicates statistically significant difference in NAD+ turnover compared to wild-type SiB196 (p<0.05). (B) Western blot analysis of TelC secretion in supernatant (Sup) and cell fractions. (C) Western blot and coomassie stain analysis of CoIP assays of TelC-his6 co-expressed with either WxgB-V or WxgC-V proteins. (D) Bacterial two-hybrid assay for interaction between Tel and WXG100-like proteins. Adenylate cyclase subunit T25 fusions (WXG100-like proteins) and T18 fusions (Tel proteins and fragments thereof) were co-expressed in the indicated combinations. Bait-prey interaction results in blue color production. (E) Model depicting Esx-dependent cell-cell delivery of LXG toxins between bacteria. The schematic shows an Si donor cell containing cognate TelA-C (light shades) and WxgA-C (dark shades) pairs intoxicating a susceptible recipient cell. Molecular targets of LXG toxins identified in this study are depicted in the recipient cell.

https://doi.org/10.7554/eLife.26938.018

Interestingly, we noted genes encoding WXG100-like proteins upstream of telA-C (wxgA-C) (Figure 2C); however, these proteins were not identified in the extracellular proteome of Si (Table 1). Given the propensity for Esx substrates to function as heterodimers, we hypothesized that the Tel proteins specifically interact with cognate Wxg partners. In support of this, we found that WxgC, but not WxgB co-purified with TelC (Figure 5C). Moreover, using bacterial two-hybrid assays, we determined that this interaction is mediated by the LXG domain of TelC (Figure 5D). To investigate the generality of these findings, we next examined all pairwise interactions between the three Wxg proteins and the LXG domains of the three Tel proteins (TelA-CLXG) (Figure 5—figure supplement 1). We found that WxgA-C interact specifically with the LXG domain of their cognate toxins (Figure 5D). The functional relevance of the LXG–WXG100 interaction was tested by examining substrate secretion in a strain lacking wxgC. We found that wxgC inactivation abrogates TelC secretion, but not that of TelB (Figure 5A–B). In summary, these data suggest that cognate Tel–Wxg interaction facilitates secretion through the Esx pathway of Si (Figure 5E).

Discussion

We present multiple lines of evidence that Esx-mediated delivery of LXG toxins serves as a physiological mechanism for interbacterial antagonism between Gram-positive bacteria. Our results suggest that like the T6S pathway of Gram-negative bacteria, the Esx system may mediate antagonism against diverse targets, ranging from related strains to species belonging to other genera (Schwarz et al., 2010). This feature of Esx secretion, in conjunction with the frequency by which we detect LXG genes in human gut metagenomes, suggests that the system could have significant ramifications for the composition of human-associated polymicrobial communities. Bacteria harboring LXG toxin genes are also components or pathogenic invaders of polymicrobial communities important in agriculture and food processing. For instance, LXG toxins may assist Listeria in colonizing fermented food communities dominated by Lactobacillus and Lactococcus (Farber and Peterkin, 1991). Of note, the latter genera also possess LXG toxins, which may augment their known antimicrobial properties. Our findings thus provide insights into the forces influencing the formation of diverse communities relevant to human health and industry.

Palmer and colleagues recently reported that the Esx system of Staphylococcus aureus exports EssD, a nuclease capable of inhibiting the growth of target bacteria in co-culture (Cao et al., 2016). The relationship between these findings and those we report herein is currently unclear. S. aureus EssD does not possess an LXG domain and was reported to be active against susceptible bacteria during co-incubation in liquid media, a condition we found not conducive to LXG toxin delivery (Figure 3D). It is evident that the Esx pathway is functionally pliable (Burts et al., 2005; Conrad et al., 2017; Gray et al., 2016; Gröschel et al., 2016; Manzanillo et al., 2012; Siegrist et al., 2009); therefore, it is conceivable that it targets toxins to bacteria through multiple mechanisms. The capacity of EssD to act against bacteria in liquid media could be the result of its over-expression from a plasmid, although we found that the exogenous administration of quantities of TelC far exceeding those likely achievable physiologically had no impact on sensitive recipient cells (Figure 3—figure supplement 2C). A later study of EssD function found no evidence of interbacterial targeting and instead reported that its nuclease activity affects IL-12 accumulation in infected mice (Ohr et al., 2017).

Our data suggest that, like a subset of substrates of the Esx systems of M. tuberculosis, LXG family members require hetero-dimerization with specific WXG100-like partners to be secreted (Ates et al., 2016). Hetero-dimerization is thought to facilitate secretion of these substrates due to the requirement for a bipartite secretion signal consisting of a YxxxD/E motif in the C-terminus of one partner in proximity to the WXG motif present in the turn between helices in the second protein (Champion et al., 2006; Daleke et al., 2012a; Poulsen et al., 2014; Sysoeva et al., 2014). While the canonical secretion signals found in other Esx substrates appear to be lacking in the LXG proteins and their interaction partners, structure prediction algorithms suggest they adopt similar helical hairpin structures, which could facilitate formation of an alternative form of the bipartite signal. Unlike previously characterized Esx substrates, we found that the LXG proteins are not co-dependent for secretion, and we failed to detect secretion of their WXG100-like interaction partners. This suggests that WxgA-C could function analogously to the EspG proteins of M. tuberculosis, which serve as intracellular chaperones facilitating delivery of specific substrates to the secretion machinery (Daleke et al., 2012b; Ekiert and Cox, 2014). Alternatively, Wxg–Lxg complexes could be secreted as heterodimers, but for technical reasons the Wxg member was undetected in our experiments. The paradigm of Lxg-Wxg interaction likely extends beyond S. intermedius, as we observe that LXG proteins from other species are commonly encoded within the same operon as Wxg homologs.

Our study leaves open the question of how Esx-exported LXG proteins reach their targets. In the case of TelC, the target resides on the extracellular face of the plasma membrane, and in the case of TelA and TelB, they are cytoplasmic. Crossing the thick Gram-positive cell wall is the first hurdle that must be overcome to deliver of each of these toxins. The size of LXG toxins exceeds that of molecules capable of free diffusion across the peptidoglycan sacculus (Forster and Marquis, 2012). Donor cell-derived cell wall hydrolytic enzymes may facilitate entry or the LXG proteins could exploit cell surface proteins present on recipient cells. Whether the entry of LXG toxins is directly coordinated by the Esx pathway is not known; our experiments do not rule-out that the requirement for donor-recipient cell contact reflects a step subsequent to secretion by the Esx pathway. Once beyond the sacculus, TelA and TelB must translocate across the plasma membrane. Our study has identified roles for the N- and C-terminal domains of LXG proteins; however, the function of the region between these two domains remains undefined and may participate in entry. Intriguingly, the central domains of TelA and TelB are each over 150 residues, whereas the LXG and toxin domains of TelC, which does not require access to the cytoplasm, appear to directly fuse (Figure 5—figure supplement 1). Based on the entry mechanisms employed by other interbacterial toxins, this central domain – or another part of the protein – could facilitate direct translocation, proteolytic release of the toxin domain, interaction with a recipient membrane protein, or a combination of these activities (Kleanthous, 2010; Willett et al., 2015).

We discovered that TelC, a protein lacking characterized homologs, adopts a previously unobserved fold and catalyzes degradation of the cell wall precursor molecule lipid II. This molecule is the target of the food preservative nisin, as well as the last-line antibiotic vancomycin, which is used to treat a variety of Gram-positive infections (Ng and Chan, 2016). Lipid II is also the target of the recently discovered antibiotic teixobactin, synthesized by the soil bacterium Eleftheria terrae (Ling et al., 2015). A particularly interesting property of this potential therapeutic is the low rate at which resistance is evolved. The apparent challenge of structurally modifying lipid II in order to subvert antimicrobials may explain why interbacterial toxins targeting this molecule have evolved independently in Gram-negative (colicin M) and -positive (TelC) bacteria (El Ghachi et al., 2006). We anticipate that biochemical characterization of additional LXG toxins of unknown function will reveal further Gram-positive cell vulnerabilities that could likewise be exploited in the design of new antibiotics.

Materials and methods

Bacterial strains and growth conditions

S. intermedius strains used in this study were derived from the sequenced strains ATCC 27335 and B196 (Supplementary file 1). S. intermedius strains were grown at 37°C in the presence of 5% CO2 in Todd Hewitt broth (THYB) or agar (THYA) supplemented with 0.5% yeast extract. When needed, media contained spectinomycin (75 μg/mL) or kanamycin (250 μg/mL). S. aureus USA300 derived strains were grown at 37°C in tryptic soy broth (TSB) or agar (TSA) supplemented with chloramphenicol (10 μg/mL) and xylose (2% w/v) when needed. E. faecalis OG1RF and S. pyogenes 5005 were grown at 37°C on Brain Heart Infusion (BHI) media. P. aeruginosa PAO1 and B. thailandensis E264 were grown at 37°C on THYA. B. fragilis NCTC9343 was grown anaerobically at 37°C on Brain Heart Infusion-supplemented (BHIS) media. E. coli strains used in this study included DH5α for plasmid maintenance, BL21 for protein expression and toxicity assays and MG1655 for competition experiments. E. coli strains were grown on LB medium supplemented with 150 μg/mL carbenicillin, 50 μg/mL kanamycin, 200 μg/mL trimethoprim, 75 μg/mL spectinomycin, 200 μM IPTG or 0.1% (w/v) rhamnose as needed. For co-culture experiments with S. intermedius strains, E. coli, B. thailandensis, P. aeruginosa, S. aureus, E. faecalis, S. pyogenes were grown on THYA. BHIS agar supplemented with sheep’s blood was used when B. fragilis was grown in co-culture with S. intermedius. S. cerevisiae BY4742 was grown on Synthetic Complete -uracil (SC-ura) medium at 30°C.

S. intermedius mutants were generated by replacing the gene to be deleted with a cassette conferring resistance to spectinomycin (derived from pDL277) or kanamycin (derived from pBAV1K-T5), as previously described (Tomoyasu et al., 2010). Briefly, the antibiotic resistance cassette was cloned between ~800 bp of sequence homologous to the regions flanking the gene to be deleted. The DNA fragment containing the cassette and flanking sequences was then linearized by restriction digest, gel purified, and ~250 ng of the purified fragment was added to 2 mL of log-phase culture pre-treated for two hours with competence peptide (200 ng/ml) to stimulate natural transformation. Cultures were further grown for four hours before plating on the appropriate antibiotic. All deletions were confirmed by PCR.

DNA manipulation and plasmid construction

All DNA manipulation procedures followed standard molecular biology protocols. Primers were synthesized and purified by Integrated DNA Technologies (IDT). Phusion polymerase, restriction enzymes and T4 DNA ligase were obtained from New England Biolabs (NEB). DNA sequencing was performed by Genewiz Incorporated.

Informatic analysis of LXG protein distribution

A comprehensive list of all clade names in the Firmicutes phylum was obtained from the List of Prokaryotic names with Standing in Nomenclature (http://www.bacterio.net/; updated 2017-02-02), a database that compiles comprehensive journal citations for every characterized prokaryotic species (Euzéby, 1997). This list was then compared with results obtained from a manually curated Jackhmmer search and LXG-containing Firmicutes were tabulated at the order, family, and genus levels (Finn et al., 2015; Mitchell et al., 2015). These results were binned into three categories based on the number of sequenced species and then further differentiated by the number of LXG-positive species within each genus. For species belonging to orders containing no predicted LXG encoding genes, the number of genera examined was tabulated and included in the dendogram.

Identification of LXG genes in human gut metagenomes

The 240 nucleotide tags from the toxin domains were mapped using blastn to the Integrated Gene Catalog (Li et al., 2014) – a large dataset of previously identified microbiome genes and their abundances in several extensive microbiome studies (including HMP [Human Microbiome Project Consortium, 2012], MetaHiT [Qin et al., 2010], and a T2D Chinese cohort [Qin et al., 2012]). Genes to which at least one tag was mapped with >95% identity and >50% overlap were labeled as LXG genes. This set of LXG genes was further manually curated to filter out genes that lack the LXG targeting domain. In analyzing the relative abundance of the LXG genes across samples, relative abundances < 10−7 were assumed to represent noise and were set to 0. LXG genes that were not present above this threshold in any sample and samples with no LXG genes were excluded from the analysis.

Determination of cellular NAD+ levels

Measurement of cellular NAD+ levels was performed as reported previously (Whitney et al., 2015). Briefly, E. coli strains harboring expression plasmids for Tse2, Tse6tox, TelBtox*, TelBtoxR626A, TelBtox*–TipB and a vector control were grown in LB media at 37°C to mid-log phase prior to induction of protein expression with 0.1% (w/v) rhamnose. 1 hr post-induction, cultures were diluted to OD600 = 0.5 and 500 μL of cells were harvested by microcentrifugation. Cells were then lysed in 0.2 M NaOH, 1% (w/v) cetyltrimethylammonium bromide (CTAB) followed by treatment with 0.4 M HCl at 60°C for 15 min. After neutralization with 0.5 M Tris base, samples were then mixed with an equal volume of NAD/NADH-Glo Detection Reagent (Promega) prepared immediately before use as per the instructions of the manufacturer. Luciferin bioluminescence was measured continuously using a Synergy H1 plate reader. The slope of the luciferin signal from the linear range of the assay was used to determine relative NAD+ concentration compared to a vector control strain.

NADase assay

S. intermedius strains were grown to late-log phase before cells were removed by centrifugation at 3000 g for 15 min. Residual particulates were removed by vacuum filtration through a 0.2 um membrane and the resulting supernatants were concentrated 100-fold by spin filtration (30 kDa MWCO). NADase assays were carried out by mixing 50 μL of concentrated supernatant with 50 μL of PBS containing 2 mM NAD+ followed by incubation at room temperature for 2 hr. Reactions were terminated by the addition of 50 μL of 6M NaOH and incubated in the dark at room temperature for 15 min. Samples were analyzed by UV light at a wavelength of 254 nm and imaged using a FluorChemQ (ProteinSimple). Relative NAD+ consumption was determined using densitometry analysis of each of the indicated strain supernatants using the ImageJ software program (https://imagej.nih.gov/ij/).

Bacterial toxicity experiments

To assess TelA and TelB toxicity in bacteria, stationary phase cultures of E. coli BL21 pLysS harboring the appropriate plasmids were diluted 106 and each 10-fold dilution was spotted onto 3% LB agar plates containing the appropriate antibiotics. 0.1% (w/v) L-rhamnose and 100 μM IPTG were added to the media to induce expression of toxin and immunity genes, respectively. For TelB, plasmids containing the wild-type toxin domain (under non-inducing conditions) were not tolerated. To circumvent this, SOE pcr was used to assemble a variant (H661A) that was tolerated under non-induced conditions. Based on the similarity of TelBtox to M. tuberculosis TNT toxin, this mutation likely reduces the binding affinity of TelB to NAD+ (Sun et al., 2015). To generate a TelB variant that exhibited significantly reduced toxicity under inducing conditions, a second mutation (R626A) was introduced in the toxin domain of TelB. For examination of TelC toxicity in S. intermedius, the gene fragment encoding TelCtox was fused to the constitutive P96 promoter followed by a start codon and cloned into pDL277 (Lo Sapio et al., 2012). For extracellular targeting of TelCtox in S. intermedius, the gene fragment encoding the sec-secretion signal (residues 1–30) of S. pneumoniae LysM (SP_0107) was fused to the 5’ end of telCtox, each of the telCtox site-specific variants and the telCtox–tipC bicistron. 500 ng of each plasmid was transformed in S. intermedius B196 and toxicity was assessed by counting the number of transformants. For examination of TelC toxicity S. aureus, the gene fragment encoding TelCtox was cloned into the xylose-inducible expression vector pEPSA5. For extracellular targeting, the gene fragment encoding the sec-secretion signal for hla was fused to the 5’ end of telCtox and telCtoxD401A. TelC-based toxicity was assessed in the same manner as was done for the above E. coli toxicity experiments except that xylose (2% w/v) was included in the media to induce protein expression. Detailed plasmid information can be found in Supplementary file 2.

Time-lapse microscopy

S. aureus USA300 pEPSA5::ss-telC202-CT and S. aureus USA300 pEPSA5 were resuspended in TSB and 1–2 μL of each suspension was spotted onto an 1% (w/v) agarose pad containing typtic soy medium supplemented with 2% (w/v) xylose and sealed.

Microscopy data were acquired using NIS Elements (Nikon) acquisition software on a Nikon Ti-E inverted microscope with a 60× oil objective, automated focusing (Perfect Focus System, Nikon), a xenon light source (Sutter Instruments), and a CCD camera (Clara series, Andor). Time-lapse sequences were acquired at 10 min intervals over 12 hr at room temperature. Movie files included are representative of three biological replicates for each experiment.

Extracellular proteome

200 mL cultures of S. intermedius B196 wild-type and ΔessC strains were grown to stationary phase in THYB before being pelleted by centrifugation at 2500 × g for 20 min at 4°C. Supernatant fractions containing secreted proteins were collected and spun at 2500 × g for an additional 20 min at 4°C and subsequently filtered through a 0.2 μm pore size membrane to remove residual cells and cell debris. Protease inhibitors (1 mM AEBSF, 10 mM leupeptin, and 1 mM pepstatin) were added to the filtered supernatants prior to dialysis in 4L of PBS using 10 kDa molecular weight cut off tubing at 4°C. After four dialysis buffer changes, the retained proteins were TCA precipitated, pelleted, washed in acetone, dried and resuspended in 1 mL of 100 mM ammonium bicarbonate containing 8 M urea. The denatured protein mixture was then desalted over a PD10 column prior to reduction, alkylation and trypsin digestion as described previously (Eshraghi et al., 2016). The resulting tryptic peptides were desalted and purified using C18 spin columns (Pierce) following the protocol of the manufacturer before being vacuum dried and resuspended in 10 µL of acetonitrile/H2O/formic acid (5/94.9/0.1, v/v/v) for LC-MS/MS analysis.

Peptides were analyzed by LC-MS/MS using a Dionex UltiMate 3000 Rapid Separation nanoLC and a linear ion trap – Orbitrap hybrid mass spectrometer (ThermoFisher Scientific). Peptide samples were loaded onto the trap column, which was 150 µm x 3 cm in-house packed with 3 µm C18 beads, at flow rate of 5 µL/min for 5 min using a loading buffer of acetonitrile/H2O/formic acid (5/94.9/0.1, v/v/v). The analytical column was a 75 µm x 10.5 cm PicoChip column packed with 1.9 µm C18 beads (New Objectives). The flow rate was kept at 300 nL/min. Solvent A was 0.1% formic acid in water and Solvent B was 0.1% formic acid in acetonitrile. The peptide was separated on a 90 min analytical gradient from 5% acetonitrile/0.1% formic acid to 40% acetonitrile/0.1% formic acid.

The mass spectrometer was operated in data-dependent mode. The source voltage was 2.10 kV and the capillary temperature was 275°C. MS1 scans were acquired from 400 to 2000 m/z at 60,000 resolving power and automatic gain control (AGC) set to 1 × 106. The top ten most abundant precursor ions in each MS1 scan were selected for fragmentation. Precursors were selected with an isolation width of 1 Da and fragmented by collision-induced dissociation (CID) at 35% normalized collision energy in the ion trap. Previously selected ions were dynamically excluded from re-selection for 60 s. The MS2 AGC was set to 3 × 105.

Proteins were identified from the MS raw files using Mascot search engine (Matrix Science). MS/MS spectra were searched against the UniprotKB database of S. intermedius B196 (UniProt and UniProt Consortium, 2015). All searches included carbamidomethyl cysteine as a fixed modification and oxidized Met, deamidated Asn and Gln, acetylated N-terminus as variable modifications. Three missed tryptic cleavages were allowed. The MS1 precursor mass tolerance was set to 10 ppm and the MS2 tolerance was set to 0.6 Da. A 1% false discovery rate cutoff was applied at the peptide level. Only proteins with a minimum of two unique peptides above the cutoff were considered for further study. MS/MS spectral counts were extracted by Scaffold 4 (Proteome Software Inc.) and used for statistical analysis of differential expression. Three biological replicates were performed and proteins identified in all three wild-type replicates were included in further analysis. After replicate averaging, low abundance proteins (less than five spectral counts in wild-type) were excluded from the final dataset.

Secretion assay

Overnight cultures of S. intermedius strains were used to inoculate 2 ml of THYB at a ratio of 1:200. Cultures were grown statically at 37°C, 5% CO2 to mid-log phase, and cell and supernatant fractions were prepared as described previously (Hood et al., 2010).

Antibody generation and western blot analyses

Full-length TelC protein was expressed and purified as described below (see protein expression and purification) except that PBS buffer was used instead of Tris-HCl for all stages of purification. Ten milligrams of purified TelC protein was sent to GenScript for polyclonal antisera production.

Western blot analyses of protein samples were performed using rabbit α-TelC (diluted 1:2000) or rabbit α-VSV-G (diluted 1:5000, Sigma) and detected with α-rabbit horseradish peroxidase-conjugated secondary antibodies (diluted 1:5000, Sigma). Western blots were developed using chemiluminescent substrate (SuperSignal West Pico Substrate, Thermo Scientific) and imaged with a FluorChemQ (ProteinSimple).

Bacterial competition experiments

For intraspecific competition experiments donor and recipient strains were diluted in THYB to a starting OD600 of 0.5 and 0.05, respectively. Cell suspensions were then mixed together in a 1:1 ratio and 10 μL of the mixture was spotted on THYA and grown at 37°C, 5% CO2 for 30 hr. The starting ratio of each competition was determined by enumerating donor and recipient c.f.u. Competitions were harvested by excising the agar surrounding the spot of cell growth followed by resuspension of cells in 0.5 mL of THYB. The final donor and recipient ratio was determined by enumerating c.f.u. For all intraspecific experiments, counts of donor and recipient c.f.u. were obtained by dilution plating on THYA containing appropriate antibiotics. To facilitate c.f.u. enumeration of wild-type S. intermedius B196, a spectinomycin resistance cassette was inserted into the intergenic region between SIR_0114 and SIR_0115.

For interspecies competition experiments, donor and recipient strains were diluted in THYB to a starting OD600 of 0.75 and 0.00075, respectively. Cell suspensions were then mixed together in a 1:1 ratio and 10 μL of the mixture was spotted on THYA and grown at 37°C, 5% CO2 for 30 hr. The starting ratio of each competition was determined by enumerating donor and recipient c.f.u. Competitions were harvested by excising the agar surrounding the spot of cell growth followed by resuspension of cells in 0.5 mL of THYB. The final donor and recipient ratio was determined by enumerating c.f.u. Counts of donor and recipient c.f.u. were obtained by dilution plating on THYA containing appropriate antibiotics (S. intermedius), BHI under standard atmospheric conditions (E. coli, E. faecalis and S. pyogenes), LB under standard atmospheric conditions (P. aeruginosa and B. thailandensis) or BHIS supplemented with 60 μg/mL gentamicin under anaerobic conditions (B. fragilis).

Statistically significance was assessed for bacterial competition experiments through pairwise t-tests of competitive index values (n = 3 for each condition).

Protein expression and purification

Stationary phase overnight cultures of E. coli BL21 pETDuet-1::telC, E. coli BL21 pETDuet-1::telC202-CT (encoding TelCtox) and E. coli BL21 pETDuet-1::tipCΔss were used to inoculate 4L of 2 x YT broth and cultures were grown to mid-log phase in a shaking incubator at 37°C. Upon reaching an OD600 of approximately 0.6, protein expression was induced by the addition of 1 mM IPTG followed by incubation at 18°C for 16 hr. Cells were harvested by centrifugation at 6000 g for 15 min, followed by resuspension in 35 mL of buffer A (50 mM Tris-HCl pH 8.0, 300 mM NaCl, 10 mM imidazole). Resuspended cells were then ruptured by sonication (3 pulses, 50 s each) and cellular debris was removed by centrifugation at 30,000 g for 45 min. Cleared cell lysates were then purified by nickel affinity chromatography using 2 mL of Ni-NTA agarose resin loaded onto a gravity flow column. Lysate was loaded onto the column and unbound proteins were removed using 50 mL of buffer A. Bound proteins were then eluted using 50 mM Tris-HCl pH 8.0, 300 mM NaCl, 400 mM imidazole. The purity of each protein sample was assessed by SDS-PAGE followed by Coomassie Brilliant Blue staining. All protein samples were dialyzed into 20 mM Tris-HCl, 150 mM NaCl.

Selenomethionine-incorporated TelC202-CT was obtained by growing E. coli BL21 pETDuet-1::telC202-CT in SelenoMethionine Medium Complete (Molecular Dimensions) using the expression conditions described above. Cell lysis and nickel affinity purification were also performed as described above except that all buffers contained 1 mM tris(2-carboxyethyl)phosphine.

Crystallization and structure determination

Purified selenomethionine-incorporated TelC202-CT was concentrated to 12 mg/mL by spin filtration (10 kDa cutoff, Millipore) and screened against commercially available crystallization screens (MCSG screens 1–4, Microlytic). Diffraction quality crystals appeared after 4 days in a solution containing 0.1 M Sodium Acetate pH 4.6, 0.1 M CaCl2, 30% PEG400. X-ray diffraction data were collected using beamline 5.0.2 at the Advanced Light Source (ALS). A single dataset (720 images, 1.0° Δφ oscillation, 1.0 s exposure) was collected on an ADSC Q315r CCD detector with a 200 mm crystal-to-detector distance. Data were indexed and integrated using XDS (Kabsch, 2010) and scaled using AIMLESS (Evans and Murshudov, 2013) (table S2).

The structure of TelC202-CT was solved by Se-SAD using the AutoSol wizard in the Phenix GUI (Adams et al., 2010). Model building was performed using the AutoBuild wizard in the Phenix GUI. The electron density allowed for near-complete building of the model except for N-terminal residues 202–211, two C-terminal residues and an internal segment spanning residues 417–434. Minor model adjustments were made manually in COOT between iterative rounds of refinement, which was carried out using Phenix.refine (Afonine et al., 2012; Emsley et al., 2010). The progress of the refinement was monitored by the reduction of Rwork and Rfree (Table 2).

Peptidoglycan hydrolase assay

Purified TelCtox was dialyzed against 20 mM sodium acetate pH 4.6, 150 mM NaCl, 10 mM CaCl2. Cross-linked peptidoglycan sacculi and lysostaphin endopeptidase pre-treated (non-cross-linked) sacculi from S. aureus were then incubated with 5 μM TelCtox, 2.5 μg of cellosyl muramidase or buffer at 37°C for 18 hr. Digests were then boiled for 5 min at 100°C and precipitated protein was removed by centrifugation. The resulting muropeptides were reduced by the addition of sodium borohydride and analyzed by HPLC as described previously (de Jonge et al., 1992).

For the analysis of cell walls isolated from TelC-intoxicated cells, 1L of S. aureus USA300 pEPSA5::ss-telCtox and S. aureus USA300 pEPSA5::ss-telCtoxD401A cells were grown to mid-log phase prior to induction of protein expression by the addition of 2% (w/v) xylose. 90 min post-induction, cultures were rapidly cooled in an ice-water bath and cells were harvested by centrifugation. After removal of supernatants, cell pellets were resuspended in 40 mL of ice-cold 50 mM Tris-HCl pH 7.0 and subsequently added dropwise to 120 mL boiling solutions of 5% SDS. PG was isolated as described (de Jonge et al., 1992) and digested with either cellosyl muramidase or lysostaphin endopeptidase and cellosyl, reduced with sodium borohydride and analyzed by HPLC as described above.

Lipid II phosphatase assay

Purified TelCtox and TelCtox–TipCΔss complex were dialyzed against 20 mM sodium acetate pH 4.6, 150 mM NaCl, 10 mM CaCl2. C14-labelled Lys-type lipid II was solubilized in 5 μL of Triton X-100 before being added to 95 μL of reaction buffer containing 15 mM HEPES pH 7.5, 0.4 mM CaCl2 (excluded from the PBP1B-LpoB reaction), 150 mM NaCl, 0.023% Triton X-100 and either PBP1B–LpoB complex, TelCtox, TelCtox–TipCΔss complex or Colicin M followed by incubation for 1 hr at 37°C. The reaction with PBP1B-LpoB was boiled and reduced with sodium borohydride. All the reactions were quenched by the addition of 1% (v/v) phosphoric acid and analyzed by HPLC as described (Bertsche et al., 2005). Three biological replicates were performed for each reaction. The lipid II degradation products of TelCtox digestion were confirmed by mass spectrometry. Lipid II was kindly provided by Ute Bertsche and was generated as described previously (Bertsche et al., 2005).

For thin-layer chromatography (TLC) analysis of Lys-type lipid II degradation by TelC, TelCtox or TelCtox–TipCΔss, 40 μM lipid II was solubilized in 30 mM HEPES/KOH pH 7.5, 150 mM KCl and 0.1% Triton X-100 before adding either 2 μM TelCtox, 2 μM TelCtox–TipCΔss complex or protein buffer, followed by incubation for 90 min at 37°C. Samples were extracted with n-butanol/pyridine acetate (2:1) pH 4.2 and resolved on silica gel (HPTLC silica gel 60, Millipore) in chloroform/methanol/ammonia/water (88:48:1:10). For the undecaprenyl phosphate reactions 100 μM undecaprenyl phosphate (Larodan) was solubilized in 20 mM HEPES/KOH pH 7.5, 150 mM KCl, 1 mM CaCl2 and 0.1% Triton X-100 before adding 2 μM TelCtox (final concentration), 2 μM TelCtox–TipCΔss or protein buffer, followed by incubation for 5 hr at 25°C and 90 min at 37°C. Samples were extracted and separated by TLC as indicated above. For the undecaprenyl pyrophosphate synthesis reactions coupled to the degradation by TelCtox0.04 mM Farnesyl pyrophosphate and 0.4 mM isopentenyl pyrophosphate were solubilized in 20 mM HEPES/KOH pH 7.5, 50 mM KCl, 0.5 mM MgCl2, 1 mM CaCl2, 0.1% Triton X-100 and incubated with 10 μM UppS and 2 μM TelCtox (final concentrations) or protein buffer for 5 hr at 25°C and 90 min at 37°C. Samples were extracted and separated by TLC as indicated above.

Yeast toxicity assay

To target TelC to the yeast secretory pathway, telCtox was fused to the gene fragment encoding the leader peptide of Kluyveromyces lactis killer toxin (Baldari et al., 1987), generating ss-telCtox. S. cerevisiae was transformed with pCM190 containing telCtox, ss-telCtox, a known toxin of yeast or empty vector and grown o/n SC-ura +1 ug/mL doxycycline. Cultures were resuspended to OD600 = 1.5 with water and serially diluted 5-fold onto SC-ura agar. Plates were incubated at 30°C for 2 days before being imaged using a Pentax WG-3 digital camera. Images presented are representative of three independent replicate experiments. Proteolytic processing of the leader peptide of ss-TelCtox was confirmed by western blot.

Isothermal titration calorimetry

Solutions of 25 μM TelC202-CT and 250 μM TipCΔss were degassed prior to experimentation. ITC measurements were performed with a VP-ITC microcalorimeter (MicroCal Inc., Northampton, MA). Titrations consisted of 25 10 μL injections with 180 s intervals between each injection. The ITC data were analyzed using the Origin software package (version 5.0, MicroCal, Inc.) and fit using a single-site binding model.

Bacterial two-hybrid analyses

E. coli BTH101 cells were co-transformed with plasmids encoding the T18 and T25 fragments of Bordetella pertussis adenylate cyclase fused to the proteins of interest. Stationary phase cells were then plated on LB agar containing 40 mg/mL X-gal, 0.5 mM IPTG, 50 mg/mL kanamycin and 150 mg/mL carbenicillin and grown for 24 hr at 30°C. Plates were imaged using a Pentax WG-3 digital camera. Images representative of at least three independent replicate experiments are presented.

Immunoprecipitation assay

E. coli BL21 (DE3) pLysS cells were co-transformed with plasmids encoding TelC-his6 and WxgB-V or TelC-his6 and WxgC-V. Cells were grown to an OD600 of 0.6 prior to induction of protein expression with 0.5 mM IPTG for 6 hr at 30°C. Cultures were harvested by centrifugation and cell pellets were resuspended in Buffer A prior to lysis by sonication. Clarified lysates were then incubated with Ni-NTA resin and incubated at 4°C with rotation for 90 min. Ni-NTA resin was then washed four times with Buffer A followed by elution of bound proteins with Buffer B. After the addition of Laemmli sample buffer, proteins were separated by SDS-PAGE using an 8–16% gradient TGX Stain-Free gel (Bio-Rad). TelC-his6 was visualized by UV activation the trihalo compound present in Stain-Free gels whereas WxgB-V and WxgC-V were detected by western blotting.

References

  1. 1
  2. 2
  3. 3
  4. 4
  5. 5
    Type VII secretion: a highly versatile secretion system
    1. LS Ates
    2. EN Houben
    3. W Bitter
    (2016)
    Microbiology Spectrum, 4, 10.1128/microbiolspec.VMBF-0011-2015, 26999398.
  6. 6
    A novel leader peptide which allows efficient secretion of a fragment of human interleukin 1 beta in Saccharomyces cerevisiae
    1. C Baldari
    2. JA Murray
    3. P Ghiara
    4. G Cesareni
    5. CL Galeotti
    (1987)
    The EMBO Journal 6:229–234.
  7. 7
  8. 8
  9. 9
  10. 10
  11. 11
  12. 12
  13. 13
  14. 14
  15. 15
  16. 16
    Peptidoglycan composition of a highly methicillin-resistant Staphylococcus aureus strain. the role of penicillin binding protein 2A
    1. BL de Jonge
    2. YS Chang
    3. D Gage
    4. A Tomasz
    (1992)
    The Journal of Biological Chemistry 267:11248–11254.
  17. 17
  18. 18
  19. 19
    Features and development of coot
    1. P Emsley
    2. B Lohkamp
    3. WG Scott
    4. K Cowtan
    (2010)
    Acta Crystallographica Section D Biological Crystallography 66:486–501.
    https://doi.org/10.1107/S0907444910007493
  20. 20
  21. 21
  22. 22
    How good are my data and what is the resolution?
    1. PR Evans
    2. GN Murshudov
    (2013)
    Acta Crystallographica Section D Biological Crystallography 69:1204–1214.
    https://doi.org/10.1107/S0907444913000061
  23. 23
    Listeria monocytogenes, a food-borne pathogen
    1. JM Farber
    2. PI Peterkin
    (1991)
    Microbiological Reviews 55:476–511.
  24. 24
  25. 25
  26. 26
  27. 27
  28. 28
  29. 29
    Colicin M inhibits peptidoglycan biosynthesis by interfering with lipid carrier recycling
    1. RE Harkness
    2. V Braun
    (1989)
    The Journal of Biological Chemistry 264:6177–6182.
  30. 30
  31. 31
  32. 32
  33. 33
  34. 34
    The alkaline reaction of nicotinamide adenine dinucleotide, a new transient intermediate
    1. SL Johnson
    2. DL Morrison
    (1970)
    The Journal of Biological Chemistry 245:4519–4524.
  35. 35
    XDS
    1. W Kabsch
    (2010)
    Acta Crystallographica. Section D, Biological Crystallography 66:125–132.
    https://doi.org/10.1107/S0907444909047337
  36. 36
  37. 37
  38. 38
  39. 39
  40. 40
  41. 41
  42. 42
  43. 43
  44. 44
  45. 45
  46. 46
  47. 47
  48. 48
  49. 49
  50. 50
  51. 51
  52. 52
  53. 53
  54. 54
  55. 55
  56. 56
  57. 57
  58. 58
  59. 59
  60. 60
  61. 61
  62. 62
  63. 63
  64. 64
  65. 65
  66. 66
  67. 67
  68. 68
  69. 69
  70. 70
  71. 71
  72. 72
  73. 73

Decision letter

  1. Michael T Laub
    Reviewing Editor; Massachusetts Institute of Technology, United States

In the interests of transparency, eLife includes the editorial decision letter and accompanying author responses. A lightly edited version of the letter sent to the authors after peer review is shown, indicating the most substantive concerns; minor comments are not usually included.

Thank you for submitting your article "A broadly distributed toxin family mediates contact-dependent antagonism between gram-positive bacteria" for consideration by eLife. Your article has been reviewed by three peer reviewers, one of whom is a member of our Board of Reviewing Editors, and the evaluation has been overseen by Gisela Storz as the Senior Editor. The following individual involved in review of your submission has agreed to reveal his identity: Jeffery S Cox (Reviewer #3).

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

Summary:

In this paper, the authors investigate the LXG protein family to test the overarching hypothesis that LXG proteins are toxins used for inter-species bacterial antagonism. Specifically, they focus on two LXG proteins produced by Streptococcus intermedius (TelB and TelC). They show that TelB and TelC are independently secreted by the Esx pathway in S. intermedius. The authors claim that TelB has NADase activity, and demonstrate that TelC is a calcium phosphatase acting on the bacterial peptidoglycan precursor Lipid II. Their data show that Esx-mediated secretion of TelC and TelB is likely contact-dependent, and that TelB mediates inter-species bacterial antagonism. The authors also present intriguing but preliminary data that WXG100-like proteins interact with their cognate LXG domain proteins, and that WXG100-like proteins promote LXG toxin secretion. Although the reviewers were enthusiastic about some aspects of the work, a number of shortcomings were identified that would need to be addressed if the authors wish to resubmit.

Essential revisions:

1) Although the paper was submitted as a 'Short Report', this length restriction led to some results being poorly explained or, in some cases like the WXG-100 data, almost neglected. There was also an extremely limited introduction, which is not ideal for a broad interest journal like eLife. The authors should flesh out the Introduction, Results, and Discussion to make the paper more accessible to a wider range of readers.

2) There was some confusion about a fundamental aspect of this system: On the one hand, the authors have argued that LXG proteins can only inhibit via a contact-dependent mechanism, but the proteins are easily detected in the supernatant suggesting they're constitutively secreted. This seems wasteful – why would a contact-dependent inhibition system spew toxin into the environment when it can only be active via direct delivery? Maybe more importantly, can the authors rule out the possibility that toxins are secreted and can kill in a contact-independent way, but are somehow getting stuck on the membranes used for the filter-well experiments? On a related note, the authors claim at one point that "The discovery that LXG toxin family members are delivered between bacteria…" But there's no direct evidence that TelC is transferred to recipient cells. The co-culturing experiments are consistent with such a model, but not definitive. For instance, are the fitness effects observed in co-culture on solid media really due to "intoxication" of the recipient strain? Is it possible to over-express the antitoxins in the recipient cell, in the experiments presented in both Figure 2E and G? Likewise, it would be nice to show specificity of the two systems (TelB/TipB and TelC/TipC) by performing competitive experiments with all combinations of mutants. In short, stronger, more direct evidence is needed that the LXG toxins are transferred from donors to recipients.

3) It would appear that Video 2 only shows what happens when ToxCtox is expressed by cells, i.e. self-intoxication after overexpression. But ToxC is normally is secreted into the extracellular medium. So the authors need to show that it can be toxic 'in trans', presumably by co-culturing cells on a solid surface given that co-culturing in liquid doesn't work. This is important to concluding that, if/when transferred as a full-length protein (not just the tox domain used in Video 2), ToxC targets lipid II.

4) The authors tested to see if production of the toxin TelC, which cleaves lipid II, is toxic in yeast cells. They were curious about this because these eukaryotes produce essential polyisoprenoid-linked glycoconjugates that are required for N-glycosylation. The authors do not see any toxicity. It is possible that TelC is not toxic because it cannot hydrolyze those essential lipid-linked saccharides. However, there are a lot of trivial reasons that could explain the lack of toxicity. For example, most of the protein might not be properly targeted since most of it still has a signal sequence (Figure 3H). A better test to determine substrate specificity would be to determine if TelC can cleave other bacterial undecaprenyl-linked saccharides. Although ideal, such work is not necessary and if not done, the authors should remove the text and Figure 3 panels G and H corresponding to the aforementioned test.

5) The somewhat grander claims regarding the implied role these LXG proteins play in polymicrobial communities is perhaps a bit over-speculative. In general, it seem that there is a disconnect between the data the authors present and their emphasis in the Abstract (and throughout the manuscript) on human gut metagenomes and polymicrobial dynamics. Throughout the manuscript, the authors emphasize the abundance/prevalence of LXG proteins in human microbiomes, but those data are neither very convincing nor novel. Thus, the authors have overstated the implications of the inter-species antagonism mediated by Esx secretion, as they do not present data illustrating the functional outcome of these interactions in an actual polymicrobial community.

6) The authors conclude that LXG toxins are specific to Gram + inter-species antagonism, but have only tested a single Gram negative species (E. coli). Testing a broader range of Gram negatives would be an easy way to bolster this claim.

7) The notion that TelB is a NAD+ degrading toxin is less than compelling. Two important controls are missing in Figure 2A – mutation of the conserved residues that comprise the active site as predicted in Figure 2—figure supplement 2, and rescue of the toxic effect by co-expressing TipB. Complementation of the Si telB mutant using wild-type and mutant forms of TelB would also be important in this regard as well. Finally, the assay in Figure 2B is somewhat unconvincing. An in vitro NADase activity assay with purified TelB would be much more convincing and quantitative.

8) Given that the main emphasis in the paper is on TelC, it is disappointing that the authors do not demonstrate inter-species antagonism dependent on TelC. Perhaps we missed something, but it appears that the data in Figure 2E and F are all experiments done with S.i. strains. In Figure 2G, they show that a strain WT for TelB elicits the crucial inter-species antagonism, but it is not clear whether this strain is also making TelC?

9) The idea that all three LXG proteins are secreted from Si via Esx secretion is not well supported. Table 1 is confusing – the "locus tag" is not intuitive, making it difficult to understand which proteins are represented in the MS results. Importantly, it is critical to show that secreted proteins that are not associated with Esx are present in similar levels in wild-type and essC mutant supernatants.

10) Figure 4 is compelling but underdeveloped. The exclusive use of bacterial two-hybrid seems insufficient to show convincingly that the LSG and WXG proteins interact specifically, and thus any conclusions at this point seems speculative.

11) In Figure 2F, why does the CI decrease to well below 1 when the filter is present?

12) Roughly how much TelC is secreted in an exponential or overnight culture? It is not clear how to interpret Figure 2—figure supplement 3 where bacteria are incubated with purified "active toxin" without an estimate of the amount of toxin produced endogenously. Also, the authors don't show data to indicate that their purified recombinant TelC protein is active.

13) In Figure 1A, the authors highlight another Streptococcus species (parasanguinus) that has an LXG protein which they annotate as a Lipid II phosphatase. That raises the question of how novel their finding about TelC in Si is? This deserves clarification.

[Editors' note: further revisions were requested prior to acceptance, as described below.]

Thank you for resubmitting your work entitled "A broadly distributed toxin family mediates contact-dependent antagonism between gram-positive bacteria" for further consideration at eLife. Your revised article has been favorably evaluated by Gisela Storz as Senior editor, a Reviewing editor, and two reviewers.

The manuscript has been improved and the consensus is that the paper is suitable for eLife provided you can address the additional issues raised by one of the reviewers:

Reviewer #2:

The manuscript has improved significantly because of the re-write and the inclusion of new data. I appreciate the authors' detailed responses to the reviews and their efforts to address the issues. Overall, I think they addressed the main concerns satisfactorily. I would only like to request a few modifications to the text.

The main comment I have is about the cellular target of TelC. The authors show that TelC: a) needs to be exported to be toxic in both E. coli and Staph, and b) degrades both lipid II and undecaprenyl-PP (C55-PP) in vitro. They conclude that, in cells, TelC targets lipid II that is located in the outer leaflet of the cytoplasmic membrane. This could lead readers to mistakenly think that lipid II is only present at that location and that C55-PP is not. In reality, C55-PP is also present in the outer leaflet of the cytoplasmic membrane, and both lipid II and C55-PP are present in both leaflets of the cytoplasmic membrane. Therefore, I suggest that the authors state that their in vivo data cannot distinguish whether TelC targets lipid II and/or undecaprenyl-PP (C55-PP) and that it is unknown why toxicity is limited to an extracytoplasmic location given that these essential peptidoglycan intermediates transit through both leaflets of the cytoplasmic membrane. [A simple explanation for their data could be that the intermediate(s) is not accessible in the cytoplasmic side of the membrane because proteins that normally handle it during peptidoglycan synthesis protect it from TelC or the intermediate has a very short residence time at that location].

https://doi.org/10.7554/eLife.26938.026

Author response

Essential revisions:

1) Although the paper was submitted as a 'Short Report', this length restriction led to some results being poorly explained or, in some cases like the WXG-100 data, almost neglected. There was also an extremely limited introduction, which is not ideal for a broad interest journal like eLife. The authors should flesh out the Introduction, Results, and Discussion to make the paper more accessible to a wider range of readers.

We agree wholeheartedly and have performed a complete overhaul of the text. Among other improvements described below, at the suggestion of the reviewers, we have extended the Introduction, Results, and Discussion sections of the revised manuscript. We regret the lack of clarity brought about by the brevity of our original submission.

2) There was some confusion about a fundamental aspect of this system: On the one hand, the authors have argued that LXG proteins can only inhibit via a contact-dependent mechanism, but the proteins are easily detected in the supernatant suggesting they're constitutively secreted. This seems wasteful – why would a contact-dependent inhibition system spew toxin into the environment when it can only be active via direct delivery? Maybe more importantly, can the authors rule out the possibility that toxins are secreted and can kill in a contact-independent way, but are somehow getting stuck on the membranes used for the filter-well experiments? On a related note, the authors claim at one point that "The discovery that LXG toxin family members are delivered between bacteria…" But there's no direct evidence that TelC is transferred to recipient cells. The co-culturing experiments are consistent with such a model, but not definitive. For instance, are the fitness effects observed in co-culture on solid media really due to "intoxication" of the recipient strain? Is it possible to over-express the antitoxins in the recipient cell, in the experiments presented in both Figure 2E and G? Likewise, it would be nice to show specificity of the two systems (TelB/TipB and TelC/TipC) by performing competitive experiments with all combinations of mutants. In short, stronger, more direct evidence is needed that the LXG toxins are transferred from donors to recipients.

The reviewers pose an interesting question, but one that is difficulty to answer. Prior to speculating why a bacterium would secrete toxins that require cell-cell contact for delivery into the milieu during propagation in vivo, it is worth clarifying a few points. First, by spectral counts deriving from mass spectrometry data (Table 1), the abundance of the LXG proteins is below the average for other proteins we detect in culture supernatants (TelA-C, μ=12.3; All others, μ=26.50). Whereas proteins that are abundantly secreted by bacteria can be detected in unconcentrated culture supernatants using SDS-PAGE followed by coomassie stain (e.g. see Jacob-Dubuisson et al., Microbiol. 2000), we detect TelC by Western blot only after TCA precipitation of culture supernatants (e.g. see Figures 2C and Figure 5B). Second, it is worth noting that the detection of LXG proteins in the supernatant may not be an indication that they are constitutively secreted. The production and/or export of the protein may still be elevated or repressed considerably depending on conditions.

Though counterintuitive, there exists overwhelming precedent for cell contact-dependent secretion systems that release their substrates to detectable levels in the absence of recipient cells. This has been documented for (note – not an exhaustive list): type III secretion (Nikolaus et al., J. Bacteriol. 2001/Rietsch et al., PNAS, 2005/Anderson and Schneewind, Mol. Microbiol.,1999) type IV secretion (Souza et al., Nat. Commun. 2015) type VI secretion (English et al., Mol. Microbiol. 2012/Eshraghi et al., 2016/Schwarz et al. Infect. Immun. 2014)

Indeed, our own group has used this property of the type VI secretion systems of Francisella novicida and Burkholderia thailandensis to define the effector proteins released by these systems (Eshraghi et al., Cell Host Microbe. 2016/Schwarz et al. Infect. Immun. 2014). In some cases, a specific cue (e.g. calcium removal, potassium addition) is required for activation of the system. Since we do not know if such a cue affects the Esx pathway of S. intermedius, we cannot determine if we have fortuitously overridden strict negative regulation by inadvertently providing an activation cue in the in vitro growth conditions we utilize.

Despite the considerations mentioned above, we have contemplated evolutionary mechanisms that could provide an explanation for apparent constitutive export of proteins that are dependent on cell contact for efficacy. We speculate that bacteria that routinely face contacting competitor cells may evolve to relieve repression of cell contact-mediated toxin delivery systems. There is evidence for this in the type VI secretion systems of Pseudomonas aeruginosa and Vibrio cholera. Although many strains of these organisms display strongly repressed type VI secretion systems when grown in vitro, other isolates possess highly active systems under the same conditions (P. aeruginosa clinical isolates studied in Mougous et al., Nat. Cell Biol. and the V52 strain of V. cholerae identified in Pukatzki et al.,Proc. Natl. Acad. Sci. 2006). This may be the case for the S. intermedius strain we employ in the current study. Anecdotally, we can share that prior to working with S. intermedius, we failed to observe Esx-dependent LXG protein export in several organisms. We suspect that many of the challenges we encountered are a result of secretion system inactivity under the conditions we utilized.

Respectfully, we believe that within reason we have ruled out the possibility that secreted TelC kills in the absence of a functional Esx system and cell contact. In addition to the membrane experiment referred to by the reviewer, our original submission included the demonstration that purified, full-length TelC, added to a final concentration of 0.1 mg/mL has no impact on the growth of TelC-susceptible cells (Figure 3—figure supplement 2C). We also showed that highly concentrated TelB and TelC donor cells (OD600nm = 20) are unable to inhibit the growth of susceptible recipient cells in liquid media (Figure 3—figure supplement 2A). Finally, as noted, we showed that a 0.2 μm pore diameter membrane separating donor and sensitive recipient cells abrogates the Esx-dependent growth advantage (Figure 3E).

We cannot rule out that LXG proteins are adhering to the 0.2 μm pore diameter membrane during our competition experiments. However, these are low protein binding track-etched polycarbonate membranes and we would expect that the minimal protein binding capacity they possess would be rapidly saturated by virtue of being coated with cells, rich media, and a variety of extracellular proteins. Nevertheless, to address this potential confounding factor, we experimentally measured the propensity of TelC to adsorb to the membranes used in our contact-dependency studies (Author response image 1). In summary, we failed to detect TelC adsorption to these membranes.

Author response image 1
TelC does not irreversibly associate with polycarbonate membranes to a detectable level.

The indicated quantity of TelC (Before) was incubated with 16 cm2 of filter material in PBS. After an extended incubation, the filter was removed and the protein was again quantified (After).

https://doi.org/10.7554/eLife.26938.022

We agree that we have not provided direct evidence of LXG transport into recipient cells. However, we believe that no other reasonably plausible mechanism is consistent with the totality of our data and we now strengthen this further with new data described in the subsequent paragraph. We show conclusively that two LXG proteins are toxins exported via the Esx secretion system (Figure 2), and that the export of these proteins is required for their capacity to inhibit the proliferation of recipient cell populations (Figure 3). We also show that this growth inhibitory activity of the toxins occurs only in strain populations rendered susceptible to the toxins by the removal of genes that encode proteins that bind to (in the case of TelC–TipC this is demonstrated in our paper in Figure 3—figure supplement 1) and inhibit the toxin-associated biochemical activities (both TelC–TipC and TelB–TipB are demonstrated in this study, see Figure 3B,C).

To add to the data in our original manuscript, we now provide evidence of toxin-mediated cell lysis occurring in co-cultures between wild-type Si and a mutant rendered susceptible to TelC (Author response image 2,described in detail in Reviewer Response to Essential Revision 3). We have additionally ruled out the possibility that secondary site mutations cause general growth defects in the telB tipB and telC tipC mutant strains by performing whole genome sequencing (new whole genome sequencing data described in the legend to Figure 3), and by demonstrating these strains compete equally with the wild-type parent in liquid culture (Figure 3D). Additionally, we find that the difference in competitive index between a wild-type strain and an essC-inactivated strain towards a susceptible recipient species can be attributed entirely to changes in the final growth yield of the recipient, consistent with growth inhibition mediated by toxin delivery (Author response image 3). Finally, it is worth noting that there is precedent for similar experiments as those we present here being used to first demonstrate contact-dependent intercellular toxin delivery by a specialized secretion systems (Souza et al., Nat. Comm. 2015 / Hood et al., Cell Host Microbe 2010 / Aoki et al. Science. 2005).

Author response image 2
Frequency of double-resistant colonies resulting from co-culture of a streptomycin resistant donor and a kanamycin resistant recipient is elevated under conditions conducive to intoxication via intracellular transfer of TelC.
https://doi.org/10.7554/eLife.26938.023
Author response image 3
The Esx-pathway of S. intermedius mediates inhibition of E. faecalis growth.

CFUs were determined by plating on selective media after 24 hours growth of co-cultures. Populations of E. faecalis were equal between cultures at the start of the experiment.

https://doi.org/10.7554/eLife.26938.024

3) It would appear that Video 2 only shows what happens when ToxCtox is expressed by cells, i.e. self-intoxication after overexpression. But ToxC is normally is secreted into the extracellular medium. So the authors need to show that it can be toxic 'in trans', presumably by co-culturing cells on a solid surface given that co-culturing in liquid doesn't work. This is important to concluding that, if/when transferred as a full-length protein (not just the tox domain used in Video 2), ToxC targets lipid II.

We assume that here and elsewhere, the reviewer is referring to TelC, as we did not study a protein designated ToxC. That aside, we believe there may be some confusion here, perhaps related to our oversight in not including legends for these videos. The bacteria in the video are S. aureus, and so we are not reporting native self-intoxication, but instead these cells are simply convenient Gram-positive heterologous hosts for observing potential phenotypic consequences of TelCtox over-expression. Video 2 is presented to illustrate that the toxin domain of TelC can promote cellular lysis when over-expressed and artificially targeted to the exterior of the cell. This observation, in conjunction with our finding that the overall structure of the peptidoglycan sacculus remains largely intact, is how we came upon the hypothesis that TelC acts upstream in cell wall synthesis to degrade lipid II. In the revised manuscript, we provide biochemical evidence that full length TelC is active on lipid II (Figure 4—figure supplement 3A). Nonetheless, we have now also obtained evidence that TelC-dependent cell lysis occurs during co-culture between Si bearing a streptomycin resistant cassette inserted at a neutral chromosomal location, and a mutant rendered susceptible to TelC through insertion of a kanamycin cassette in place of the telC tipC locus (Author response image 2). We find that after 30 hours of co-culture of these strains, a significant proportion of the resulting population acquires resistance to both antibiotics. The frequency of double-resistant colonies obtained from co-cultures in which the streptomycin-resistant strain contained inactivated telC is significantly lower. It is well established in Streptococcus that cell lysis provides DNA for uptake by natural transformation. Thus, the accumulation of cells containing both markers is a useful proxy for cell lysis. These results provide further evidence that TelC is a bacteriolytic toxin.

4) The authors tested to see if production of the toxin TelC, which cleaves lipid II, is toxic in yeast cells. They were curious about this because these eukaryotes produce essential polyisoprenoid-linked glycoconjugates that are required for N-glycosylation. The authors do not see any toxicity. It is possible that TelC is not toxic because it cannot hydrolyze those essential lipid-linked saccharides. However, there are a lot of trivial reasons that could explain the lack of toxicity. For example, most of the protein might not be properly targeted since most of it still has a signal sequence (Figure 3H). A better test to determine substrate specificity would be to determine if TelC can cleave other bacterial undecaprenyl-linked saccharides. Although ideal, such work is not necessary and if not done, the authors should remove the text and Figure 3 panels G and H corresponding to the aforementioned test.

We agree that there are trivial reasons for the apparent inability of TelC to inhibit yeast growth. However, we do believe that these negative data have value, as they show that contrary to the situation with bacteria, TelC is not toxic to a eukaryotic cell when expressed from a strong promoter. We show the toxin is produced to detectable levels in the cytoplasm and we provide evidence that our construct directing it to the secretory pathway is performing as expected. If this were a bona fide eukaryotic toxin, we would anticipate that levels detected by direct Western blot of lysed cells (immunoprecipitation not required) would likely be sufficient to induce detectable killing or growth inhibition, which we do not observe. This is an experiment that we feel others in the community would want to see – caveats notwithstanding – given that we are proposing TelC targets bacteria physiologically. We do appreciate the point raised by the reviewer, and so to meet in the middle on this matter, we have moved the yeast data to the supplement and we have simplified our interpretation (Figure 4—figure supplement 3C,D). We now avoid the dolichol discussion entirely and use the data simply to demonstrate a negative result consistent with our model.

5) The somewhat grander claims regarding the implied role these LXG proteins play in polymicrobial communities is perhaps a bit over-speculative. In general, it seem that there is a disconnect between the data the authors present and their emphasis in the Abstract (and throughout the manuscript) on human gut metagenomes and polymicrobial dynamics. Throughout the manuscript, the authors emphasize the abundance/prevalence of LXG proteins in human microbiomes, but those data are neither very convincing nor novel. Thus, the authors have overstated the implications of the inter-species antagonism mediated by Esx secretion, as they do not present data illustrating the functional outcome of these interactions in an actual polymicrobial community.

At the suggestion of the reviewers, we have toned down and limited references in the main body of the revised manuscript to the potential role of LXG toxins in human-associated polymicrobial communities. We have also edited the Abstract with the intent to minimize this potential implication of the prevalence of LXG toxins amongst human gut microbiome samples.

We feel that we must respectfully voice our disagreement with the reviewer comment pertaining to our metagenomic analyses that “those data are neither very convincing nor novel.” To our knowledge, LXG toxins have not previously been identified in the human gut microbiome (or another polymicrobial community) and we employed rigorous, accepted methods of showing that they are present in said communities in our study.

6) The authors conclude that LXG toxins are specific to Gram + inter-species antagonism, but have only tested a single Gram negative species (E. coli). Testing a broader range of Gram negatives would be an easy way to bolster this claim.

First, to clarify, we respectfully wish to correct the record by pointing out that we did not “conclude that LXG toxins are specific to Gram + inter-species antagonism.” The single statement we made addressing these data in our original submission, stated “These results demonstrate that the Esx pathway can act between species and suggest that its target range may be limited to Gram-positive bacteria.” We specifically chose the words “suggest” and “may” in this statement because we agree that stronger terms are difficult (or even impossible) to support.

At the suggestion of the reviewers, the revised manuscript now includes the results of experiments testing the capacity of the Esx secretion pathway to target additional Gram-negative bacteria. We report that under conditions similar to those used in the corresponding Gram-positive target cell experiments, the Esx system of S. intermedius did not detectably alter the outcome of co-cultivations with P. aeruginosa, B. thailandensis and Bacteroides fragilis (new Figure 3F). B. thailandensis and B. fragilis, a β-proteobacterium and a member of the phylum Bacteroidetes, respectively, significantly extend the phylogenetic range of Gram-negative organisms tested. While we now report more Gram-negative bacteria tested, we recognize that this does not bring us significantly closer to reaching the conclusion that Gram-negative bacteria are not or cannot be targeted by the Esx pathway. We thus have not modified our statement that “These results demonstrate that the Esx pathway can act between species and suggest that its target range may be limited to Gram-positive bacteria”.

7) The notion that TelB is a NAD+ degrading toxin is less than compelling. Two important controls are missing in Figure 2A – mutation of the conserved residues that comprise the active site as predicted in Figure 2—figure supplement 2, and rescue of the toxic effect by co-expressing TipB. Complementation of the Si telB mutant using wild-type and mutant forms of TelB would also be important in this regard as well. Finally, the assay in Figure 2B is somewhat unconvincing. An in vitro NADase activity assay with purified TelB would be much more convincing and quantitative.

In our original manuscript, we used two independent biochemical methods to demonstrate that TelB is an NAD+-degrading toxin. The first makes use of the fact that if NAD+ is not consumed by such a toxin, a product of its hydrolysis by base is fluorescent (Johnson et al. J. Biol. Chem. 1970). This assay showed that wild-type S. intermedius B196 possesses supernatant NAD+-degrading activity that is entirely dependent on TelB and EssC (Figure 2B). We also showed that another strain of S. intermedius that does not encode a TelB ortholog does not possess this activity in its supernatant (Figure 2B). The second biochemical assay we presented makes use of a very well established, specific, and sensitive cell-based assay for measuring NAD+ pools that is commercially available. We used this assay to show that induction of TelB and an established NADase, Tse6, lead to a dramatic drop in cellular [NAD+], whereas induction of an interbacterial toxin that does not possess this activity, Tse2, does not reduce cellular NAD+ levels (Figure 2A). In addition to these specific biochemical assays, we stated that “One such toxin domain shares homology and predicted catalytic residues with M. tuberculosis TNT, an NAD+-degrading enzyme” and presented a molecular model of TelB based on the TNT structure that showed conserved catalytic residues. Indeed, we noted in the methods section of the original submission that, due to its potent toxicity, we were unable to clone TelBtox, and “to circumvent this, SOE PCR was used to assemble a variant (H661A) that was tolerated under non-induced conditions. Based on the similarity of TelBtox to M. tuberculosis TNT toxin, this mutation likely reduces the binding affinity of TelB to NAD+ (Sun et al., 2015).”

For technical reasons relating to its stability, we are unable to purify TelB. These enzymes are well characterized and we do not believe that a detailed quantification of its activity is necessary to support the conclusions of this manuscript. However, at the request of the reviewer, we now include data demonstrating that the NAD+-degrading activity of TelB is abrogated by point mutations in the active site of the enzyme and by co-expression with TipB (new Figure 2A).

8) Given that the main emphasis in the paper is on TelC, it is disappointing that the authors do not demonstrate inter-species antagonism dependent on TelC. Perhaps we missed something, but it appears that the data in Figure 2E and F are all experiments done with S.i. strains. In Figure 2G, they show that a strain WT for TelB elicits the crucial inter-species antagonism, but it is not clear whether this strain is also making TelC?

We agree with the reviewers and the revised manuscript now includes the finding that TelC contributes significantly to the fitness of S. intermedius during interspecies competition (new Figure 4—figure supplement 3B). As a point of clarification, the “wild-type” strain used in Figure 2G of our original submission was indeed wild-type and thus possesses and presumably utilizes its full complement of LXG effectors.

9) The idea that all three LXG proteins are secreted from Si via Esx secretion is not well supported. Table 1 is confusing – the "locus tag" is not intuitive, making it difficult to understand which proteins are represented in the MS results. Importantly, it is critical to show that secreted proteins that are not associated with Esx are present in similar levels in wild-type and essC mutant supernatants.

We regret the lack of clarity in Table 1 of our original submission. At the suggestion of the reviewers, we have added a column to Table 1 that provides the common names of the proteins we detected. Importantly, SIR_0169, SIR_0179, and SIR_1489 are now identified as TelA, TelB, and TelC, respectively (note – this information was provided in Figure 2C of the original submission). There additionally may have been some confusion regarding the footnote that defined the values we provided for each entry in Table 1. These values represent the average spectral counts – a semi-quantitative correlate of protein abundance from mass spectrometry data (Liu et al.,Anal. Chem. 2004) – for each protein from triplicate biological replicates. The extracellular proteome of S. intermedius has not previously been defined; however, the wild-type:∆essC ratio of these spectral count values clearly indicates that known Esx-associated proteins and the Tel proteins are underrepresented in the extracellular proteome of ∆essC relative to 29 of the 30 other proteins we identified.

10) Figure 4 is compelling but underdeveloped. The exclusive use of bacterial two-hybrid seems insufficient to show convincingly that the LSG and WXG proteins interact specifically, and thus any conclusions at this point seems speculative.

We are unsure if there was disagreement amongst the reviewers as to whether the data in Figure 4 of our original submission was compelling or underdeveloped. Nevertheless, we agree that the exclusive use of the bacterial two-hybrid to demonstrate an interaction would be problematic and for this reason we included in vivo functional data in the original submission (Figure 4A and 4B) that supports the interaction between LXG proteins and cognate WXG-like proteins. In short, we showed that a deletion of wxgC abrogates TelC secretion while leaving TelB secretion intact. To further support this, we added to the revised manuscript an experiment demonstrating that in a heterologous host, TelC immunoprecipitates WxgC and not WxgB – additional evidence of their direct interaction (new Figure 5C). In summary, we now provide genetic, biochemical, and in vivo functional data supporting the specific interaction of Lxg proteins with cognate Wxg proteins.

11) In Figure 2F, why does the CI decrease to well below 1 when the filter is present?

This is an excellent question that we have clarified in the revised manuscript (see revised legend to Figure 3). In the experiments represented, recipient cell populations were placed below the filter, whereas the donor cells were above the filter. As the cells proliferate, those below the filter have an Esx-independent growth advantage by virtue of their immediate proximity to the growth substrate. It is worth noting that we conducted this experiment in both configurations and in neither case were non-contacting cells inhibited in an Esx-dependent manner.

12) Roughly how much TelC is secreted in an exponential or overnight culture? It is not clear how to interpret Figure 2—figure supplement 3 where bacteria are incubated with purified "active toxin" without an estimate of the amount of toxin produced endogenously. Also, the authors don't show data to indicate that their purified recombinant TelC protein is active.

As we noted above, TelC (and the other Tel proteins) are of lower than average abundance (as determined by spectral counts) among the 35 proteins identified in our secreted proteome analysis (Table 1). We are not aware of absolute measurements of secreted protein concentration during bacterial culture, but we estimate based on our own experience studying secreted proteins that those we can detect are present at 5-100 ng/mL. We choose a concentration of 0.1 mg/mL of TelC in order to ensure that the protein was present at levels exceeding those obtainable physiologically. To probe this directly and address the question raised, we present here a comparison of the relative levels of endogenous TelC to the levels of purified TelC we added exogenously to S. intermedius cells. As expected, endogenous TelC is below the detection limit of this assay, whereas 0.1 mg/mL purified TelC is readily detected (Author response image 4).

Author response image 4
TelC levels in cell free supernatants (CFS) of S. intermedius do not approach 0.1 mg/mL.

SDS-PAGE analysis of the supernatant fraction of S. intermedius cells in mid-log or stationary growth. To examine the relative levels of endogenous TelC versus the exogenous purified active TelC we applied, supernatants were doped with 0.1 mg/mL TelC (CFS + TelC) and the protein was visualized alongside CFS alone.

https://doi.org/10.7554/eLife.26938.025

With regard to the question pertaining to the activity of purified recombinant TelC, we first wish to clarify that in our original submission, all biochemical assays demonstrating the capacity of this protein to act on undecaprenylphosphate derivatives were performed using purified recombinant protein. Though it is not specified here, the reviewers may be referring to the fact that our initial biochemical assays were performed with the toxin domain of TelC (TelCtox), in contrast to the full-length protein used in our cellular inhibition assays. We use full-length TelC in lieu of TelCtox for growth inhibition assays to allow potential cell entry mechanisms that could utilize the N-terminal domains of the protein. At the suggestion of the reviewers, we have performed biochemical assays with purified full-length TelC. New Figure 4—figure supplement 3A shows that like the toxin domain alone, the full-length protein possesses the capacity to degrade lipid II.

13) In Figure 1A, the authors highlight another Streptococcus species (parasanguinus) that has an LXG protein which they annotate as a Lipid II phosphatase. That raises the question of how novel their finding about TelC in Si is? This deserves clarification.

The S. parasanguinus protein schematized in Figure 1A is a predicted ortholog of TelC and was assigned a predicted activity based solely on this relatedness. Were an LXG protein possessing lipid II phosphatase activity known, we certainly would have highlighted and cited such a highly relevant result. To prevent this potential confusion, we have modified the Figure 1 legend to read: “lipid II phosphatase based on orthology to TelC (defined biochemically herein)” in place of “TelC-like lipid II phosphatase” found in the original manuscript.

[Editors' note: further revisions were requested prior to acceptance, as described below.]

Reviewer #2:

The manuscript has improved significantly because of the re-write and the inclusion of new data. I appreciate the authors' detailed responses to the reviews and their efforts to address the issues. Overall, I think they addressed the main concerns satisfactorily. I would only like to request a few modifications to the text.

The main comment I have is about the cellular target of TelC. The authors show that TelC: a) needs to be exported to be toxic in both E. coli and Staph, and b) degrades both lipid II and undecaprenyl-PP (C55-PP) in vitro. They conclude that, in cells, TelC targets lipid II that is located in the outer leaflet of the cytoplasmic membrane. This could lead readers to mistakenly think that lipid II is only present at that location and that C55-PP is not. In reality, C55-PP is also present in the outer leaflet of the cytoplasmic membrane, and both lipid II and C55-PP are present in both leaflets of the cytoplasmic membrane. Therefore, I suggest that the authors state that their in vivo data cannot distinguish whether TelC targets lipid II and/or undecaprenyl-PP (C55-PP) and that it is unknown why toxicity is limited to an extracytoplasmic location given that these essential peptidoglycan intermediates transit through both leaflets of the cytoplasmic membrane. [A simple explanation for their data could be that the intermediate(s) is not accessible in the cytoplasmic side of the membrane because proteins that normally handle it during peptidoglycan synthesis protect it from TelC or the intermediate has a very short residence time at that location].

At the suggestion of the reviewer we have added text stating that C55-PP and lipid II are present both inside and outside of the cell, that thus that it is unknown why cytoplasmic TelC is non-toxic. In this text, we acknowledge the substrate accessibility explanation mentioned by the reviewer and, in addition, we note that the abundance of free calcium ions outside of the bacterial cell may be important for TelC folding.

https://doi.org/10.7554/eLife.26938.027

Article and author information

Author details

  1. John C Whitney

    Department of Microbiology, University of Washington School of Medicine, Seattle, United States
    Present address
    Department of Biochemistry and Biomedical Sciences, McMaster University, Hamilton, Canada
    Contribution
    JCW, Conceptualization, Investigation, Writing—original draft, Writing—review and editing
    Competing interests
    The authors declare that no competing interests exist.
    ORCID icon 0000-0002-4517-8836
  2. S Brook Peterson

    Department of Microbiology, University of Washington School of Medicine, Seattle, United States
    Contribution
    SBP, Conceptualization, Supervision, Funding acquisition, Investigation, Writing—original draft, Writing—review and editing
    Competing interests
    The authors declare that no competing interests exist.
    ORCID icon 0000-0003-2648-0965
  3. Jungyun Kim

    Department of Microbiology, University of Washington School of Medicine, Seattle, United States
    Contribution
    JK, Investigation, Writing—review and editing
    Competing interests
    The authors declare that no competing interests exist.
    ORCID icon 0000-0003-3793-4264
  4. Manuel Pazos

    Centre for Bacterial Cell Biology, Institute for Cell and Molecular Biosciences, Newcastle University, Newcastle, United Kingdom
    Contribution
    MP, Investigation, Writing—review and editing
    Competing interests
    The authors declare that no competing interests exist.
  5. Adrian J Verster

    Department of Genome Sciences, University of Washington, Seattle, United States
    Contribution
    AJV, Investigation, Writing—review and editing
    Competing interests
    The authors declare that no competing interests exist.
  6. Matthew C Radey

    Department of Microbiology, University of Washington School of Medicine, Seattle, United States
    Contribution
    MCR, Investigation, Writing—review and editing
    Competing interests
    The authors declare that no competing interests exist.
  7. Hemantha D Kulasekara

    Department of Microbiology, University of Washington School of Medicine, Seattle, United States
    Contribution
    HDK, Investigation
    Competing interests
    The authors declare that no competing interests exist.
  8. Mary Q Ching

    Department of Microbiology, University of Washington School of Medicine, Seattle, United States
    Contribution
    MQC, Investigation
    Competing interests
    The authors declare that no competing interests exist.
  9. Nathan P Bullen

    1. Michael DeGroote Institute for Infectious Disease Research, McMaster University, Hamilton, Canada
    2. Department of Biochemistry and Biomedical Sciences, McMaster University, Hamilton, Canada
    Contribution
    NPB, Investigation
    Competing interests
    The authors declare that no competing interests exist.
  10. Diane Bryant

    Experimental Systems Group, Advanced Light Source, Berkeley, United States
    Contribution
    DB, Investigation
    Competing interests
    The authors declare that no competing interests exist.
  11. Young Ah Goo

    Northwestern Proteomics Core Facility, Northwestern University, Chicago, United States
    Contribution
    YAG, Investigation, Writing—review and editing
    Competing interests
    The authors declare that no competing interests exist.
  12. Michael G Surette

    1. Michael DeGroote Institute for Infectious Disease Research, McMaster University, Hamilton, Canada
    2. Department of Biochemistry and Biomedical Sciences, McMaster University, Hamilton, Canada
    3. Department of Medicine, Farncombe Family Digestive Health Research Institute, McMaster University, Hamilton, Canada
    Contribution
    MGS, Investigation
    Competing interests
    The authors declare that no competing interests exist.
  13. Elhanan Borenstein

    1. Department of Genome Sciences, University of Washington, Seattle, United States
    2. Department of Computer Science and Engineering, University of Washington, Seattle, United States
    3. Santa Fe Institute, Santa Fe, United States
    Contribution
    EB, Funding acquisition, Investigation, Writing—review and editing
    Competing interests
    The authors declare that no competing interests exist.
  14. Waldemar Vollmer

    Centre for Bacterial Cell Biology, Institute for Cell and Molecular Biosciences, Newcastle University, Newcastle, United Kingdom
    Contribution
    WV, Funding acquisition, Investigation, Writing—review and editing
    Competing interests
    The authors declare that no competing interests exist.
  15. Joseph D Mougous

    1. Department of Microbiology, University of Washington School of Medicine, Seattle, United States
    2. Howard Hughes Medical Institute, University of Washington School of Medicine, Seattle, United States
    Contribution
    JDM, Conceptualization, Supervision, Funding acquisition, Investigation, Writing—original draft, Project administration, Writing—review and editing
    For correspondence
    mougous@uw.edu
    Competing interests
    The authors declare that no competing interests exist.
    ORCID icon 0000-0002-5417-4861

Funding

Canadian Institutes of Health Research

  • John C Whitney

Natural Sciences and Engineering Research Council of Canada

  • Adrian J Verster

National Cancer Institute (CCSG P30 CA060553)

  • Young Ah Goo

National Institutes of Health (AT007802-01)

  • Elhanan Borenstein

Medical Research Council (MR/N0026791/1)

  • Waldemar Vollmer

National Institutes of Health (AI080609)

  • Joseph D Mougous

Howard Hughes Medical Institute

  • Joseph D Mougous

Burroughs Wellcome Fund (1010010)

  • Joseph D Mougous

Defense Threat Reduction Agency (HDTRA1-13-1-0014)

  • Joseph D Mougous

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We thank Eefjan Breukink for providing lipid II, Joe Gray for mass spectrometry analyses, the Fang and Rajagopal laboratories for sharing reagents, Lynn Hancock and Gary Dunny for providing strains and plasmids, Corie Ralston for help with X-ray data collection, Ben Ross and Simon Dove for critical reading of the manuscript, and members of the Mougous laboratory for helpful discussions. This work was funded by the NIH (AI080609 to JDM and AT007802 to EB), the Defense Threat Reduction Agency (HDTRA1-13-1-0014 to JDM), and the Medical Research Council (MR/N0026791/1). Proteomics services were performed by the Northwestern Proteomics Core Facility, generously supported by NCI CCSG P30 CA060553 awarded to the Robert H Lurie Comprehensive Cancer Center and the National Resource for Translational and Developmental Proteomics supported by P41 GM108569. JCW was supported by a postdoctoral research fellowship from the Canadian Institutes of Health Research, AJV was supported by a postdoctoral fellowship from the Natural Science and Engineering Research Council of Canada, and JDM holds an Investigator in the Pathogenesis of Infectious Disease Award from the Burroughs Wellcome Fund and is an HHMI Investigator.

Reviewing Editor

  1. Michael T Laub, Reviewing Editor, Massachusetts Institute of Technology, United States

Publication history

  1. Received: March 20, 2017
  2. Accepted: July 10, 2017
  3. Accepted Manuscript published: July 11, 2017 (version 1)
  4. Accepted Manuscript updated: July 12, 2017 (version 2)
  5. Version of Record published: August 14, 2017 (version 3)
  6. Version of Record updated: August 24, 2017 (version 4)

Copyright

This is an open-access article, free of all copyright, and may be freely reproduced, distributed, transmitted, modified, built upon, or otherwise used by anyone for any lawful purpose. The work is made available under the Creative Commons CC0 public domain dedication.

Metrics

  • 2,233
    Page views
  • 505
    Downloads
  • 3
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Comments

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)