All ESX protein secretion systems that have been studied to date have been shown to secrete at least one WXG100 family protein homologous to the prototypic ESX-1 substrate EsxA [13], [16], [20], [21]. In B. subtilis, this protein is encoded by yukE. Therefore, our first experimental objective was to determine whether YukE is secreted from the B. subtilis cell. To address this question, we grew cultures of the wild-type domesticated strain of B. subtilis (PY79) in nutrient-rich LB medium to mid-exponential phase, harvested whole cell pellets, and filtered the culture supernatants. Proteins in the culture supernatant were concentrated by TCA precipitation and analyzed by SDS-PAGE. Presence of YukE was assessed using a primary antibody raised against recombinant full-length YukE. As a lysis control, we tested for the presence of the cytosolic protein RNA polymerase sigma factor SigmaA by immunoblotting with α-SigmaA antibodies [35]. In these experiments, we detected YukE in both the pellet and supernatant fractions (Figure 1B). These data confirm the prediction and recent demonstration that YukE is secreted from the cell [34]. In contrast to the previous work, we were able to detect YukE secretion in a domesticated laboratory strain. We found that YukE was secreted in all conditions tested, ranging from growth in nutrient-rich media to the nutrient-limiting conditions that promote competence and biofilm formation (Figure S1).

Next, we asked whether YukE secretion depends upon the other gene products in the yuk/yue locus. To address this question, we created a series of yuk/yue knockout strains. Each yuk/yue gene was individually replaced with an antibiotic resistance cassette and the yuk promoter (Pyuk) was reinserted after the resistance cassette to drive expression of the downstream operon genes. We used the intergenic region between yukE and adeR as the yuk promoter, and confirmed that Pyuk was transcriptionally active by inserting a Pyuk-lacZ construct at an ectopic integration site (amyE::Pyuk-lacZ) and assessing transcriptional activity. The β-galactosidase activity in this strain was approximately three-fold lower than the β-galactosidase activity in a strain with lacZ integrated at the endogenous yuk operon start site (ΩPyuk-lacZ) (Figure S2). This was ultimately useful, because genome-wide expression studies indicate that yukE expression is at least twice as high as the expression of other yuk operon genes [36]. Therefore, we reasoned that using our weaker Pyuk should result in approximately wild-type levels of transcription of the downstream genes. We confirmed that the reinserted Pyuk drove expression of downstream yuk genes, although resulting protein levels were approximately two-fold higher than native levels, as assessed by semi-quantitative immunoblotting (Figure S2).

To determine whether the genes of the yuk/yue locus are required for YukE secretion, we tested whether YukE is produced and secreted in each of the yuk/yue knockout strains. Currently, there are five annotated genes in the yuk operon: yukE, yukD, yukC, yukBA, and yueB [31], [32]. Knocking out each gene in the annotated yuk operon (yukE-yueB) individually abolished YukE secretion in all five of these strains (Figure 1B). Recently, transcriptomic profiling has implicated yueC and/or yueD as potential members of the yuk/yue operon as well [33]. Therefore, we also tested whether YukE is secreted in ΔyueC and ΔyueD strains. YukE was not secreted in the ΔyueC strain, demonstrating that YueC is required for YukE export, but it was secreted in the ΔyueD strain, suggesting that YueD is not required for YukE export (Figure 1B).

To demonstrate the specificity of these results, we constructed complementation strains by inserting the corresponding yuk/yue gene at an ectopic integration site under the control of an inducible promoter. We attached a C-terminal Myc or HA tag to each of the complementation constructs (except for the untagged YukE complementation construct), thereby allowing us to verify presence of the complementing protein by immunoblot (Figure S3). YukE secretion was restored to wild-type levels in the ΔyukD, ΔyukBA, and ΔyueC strains upon expression of yukD-myc, yukBA-myc, and yueC-myc respectively (Figure 1B). Densitometric analysis of secretion levels in each strain is presented in Table 1; values indicate the percentage of total YukE in each strain that is localized to the pellet versus culture supernatant. Complementation of ΔyukC with yukC-myc did not restore YukE secretion to wild-type levels, but partial restoration of YukE secretion can be seen in an overexposed blot (Figure 1B). We were unable to complement YukE secretion in the ΔyueB strain, despite attempts with untagged and several tagged versions of YueB. Nonetheless, YukE secretion appears dependent upon the yueB gene product and a recent study produced a complementing construct which confirms the specificity of a yueB deletion [34]. Thus we conclude that YukE secretion requires the full yuk operon as well as yueC, but not yueD.

The divergently transcribed gene adeR (formerly annotated as yukF) is a predicted transcription factor. Since regulatory proteins are often coded in the general vicinity of the genes they regulate, we also tested for YukE secretion in an adeR knockout strain, and found that YukE was still secreted in this background (Figure S4). This result is consistent with the idea that yuk/yue activity is perhaps principally regulated through stress response pathways including those governed by DegS/U and Spo0A [33], [34], [37]–[40], although inputs from other regulatory pathways may remain to be discovered.

To gain insight into possible function(s) of the yuk/yue system, we next sought to determine whether there are additional secreted proteins dependent upon the yuk/yue locus for secretion. Besides YukE, there is one other predicted WXG100 protein encoded in the B. subtilis genome, YfjA, and therefore this protein was a candidate yuk/yue substrate. [11]. In addition, secretion of LXG-motif proteins and non-WXG100 proteins has been reported in other ESX secretion systems, and these proteins are often encoded away from the primary ESX/Ess locus [20], [41]. Therefore, we decided to use an unbiased, quantitative proteomics approach to analyze the full profile of yuk/yue-dependent proteins in the culture supernatant.

In addition to the virulence factor polypeptides, the FtsK/SpoIIIE family ATPases are a signature of ESX loci. Thus, using quantitative mass spectrometry, we compared the proteins in culture supernatants of the wild-type domesticated strain and the ATPase deletion strain ΔyukBA grown in defined media. Consistent with our immunoblot assay, we detected YukE in the supernatant of the wild-type strain in a manner that was dependent upon yukBA (Figure 2A, 2B). YukE secretion was restored in the YukBA complementation strain (Figure 2B). Ninety-five YukE-specific peptide spectra were detected in the supernatant from the wild-type strain, no peptides were detected in the ΔyukBA strain and 116 YukE-specific peptide spectra were detected in the ΔyukBA; yukBA-myc complementation strain. We detected high levels of YueB peptides in the culture supernatant of the ΔyukBA and complement strains (Figure 2A, 2B), which is an expected consequence of the strain design. Briefly, the yuk promoter was reinserted after the yukBA deletion to drive expression of the downstream genes, as otherwise this would be a polar mutation. Most surprisingly, we did not detect any other proteins with the same secretion profile as YukE in these conditions. Therefore, by this method and under these growth conditions, we found YukE to be the only protein that requires the ATPase YukBA for secretion.

The biological function of the yuk/yue locus remains unknown but it is highly unusual for a secretion system to have only a single substrate. Further, since all conditions we tested yielded secreted YukE, we speculated that the yuk/yue knockout strains might display a growth or competition phenotype. We first tested whether various yuk/yue knockout strains have a growth defect compared to the wild-type domesticated strain by conducting growth assays. The growth curves of the yuk/yue knockout strains were statistically indistinguishable from the growth curve of the wild-type domesticated strain, indicating that the yuk/yue knockout strains do not have a growth defect under standard, nutrient-rich laboratory conditions (Figure 3A). Next, we performed competition assays between the wild-type domesticated strain and yuk/yue knockout strains. We found that the yuk/yue knockout strains did not have a statistically significant competitive advantage or disadvantage compared to the wild-type domesticated strain in nutrient-rich or nutrient-limiting media (Figure 3B and Figure S5).

General methods for molecular cloning and strain construction were performed according to published protocols [50]. Chromosomal DNA isolated from the prototrophic domesticated strain PY79 was used as a template for all PCR amplification. Introduction of DNA into PY79 derivatives was conducted by transformation [51]. The bacterial strains used in this study are listed in Table 2. Complete strain construction information including oligonucleotide primers is included in Supporting Information.

For general propagation, B. subtilis strains were grown at 37°C in LB (lysogeny broth) [52] (10 g tryptone per liter, 5 g yeast extract per liter, 5 g NaCl per liter) or on LB plates containing 1.5% Bacto agar. Where indicated, B. subtilis strains were grown in the nutrient-limiting medium B. subtilis Medium for Competence (1XMC) [53]. When appropriate, antibiotics were included in the growth medium as follows: 100 µg mL⁻¹ spectinomycin, 5 µg mL⁻¹ chloramphenicol, 5 µg mL⁻¹ kanamycin, 10 µg mL⁻¹ tetracycline, and 1 µg mL⁻¹ erythromycin plus 25 µg mL⁻¹ lincomycin (mls). When required, 100 µM IPTG (isopropyl-β-D-thiogalactopyranoside) was added to cultures or solid media to induce protein expression.

Bacterial strains were grown in LB medium to an OD₆₀₀ of approximately 1.0–1.3. The cells were pelleted and the supernatant was collected. The pellet samples were processed to make whole cell lysates according to standard protocols [53]. Briefly, one milliliter of cells was harvested, lysed in the presence of lysozyme and then boiled for 15 minutes in 1× sample buffer (4% SDS, 250 mM Tris pH 6.8, 20% glycerol, 10 mM EDTA, 1% bromophenol blue, 10% β-mercaptoethanol (BME)). The culture supernatant samples were first filtered through a 0.2 micron filter and then incubated in 10% tricholoracetic acid (TCA) for 12–15 hours at 4°C. The following day, the samples were spun at 15,000xg for 20 minutes to pellet the precipitated proteins, the liquid was poured off, and the pellets were washed with ice-cold acetone. The pellets were resuspended in 100 µL of 1× sample buffer and the samples were boiled for 15 minutes. After processing the pellet and supernatant samples, the proteins were separated by SDS-polyacrylamide gel electrophoresis (SDS-PAGE) and analyzed by immunoblot analysis with appropriate antibodies. Pellet samples are equivalent to 0.1 OD units and twenty-fold more was loaded for supernatant samples. Precipitated supernatant samples were normalized based on Coomassie staining.

A hexahistidine-tagged version of YukE was utilized for antibody production. YukE was PCR-amplified with primers oLH067 and oLH068 using genomic DNA from the wild-type domesticated strain PY79 as a template. The sequence was inserted into an inducible E. coli expression vector to make pLH054, which was then transformed into E. coli BL21 cells. The cells were induced and YukE was purified from the E. coli extracts by nickel-affinity chromatography. Finally, a rabbit polyclonal serum was raised against this protein (Covance).

Proteins were separated by SDS-PAGE and transferred to nitrocellulose membrane. The membrane was probed with affinity-purified α-YukE (polyclonal), α-GFP (polyclonal), α-Myc (Novus Biologicals), and/or α-SigmaA (polyclonal) antibodies. Primary antibodies were diluted 1∶1000 (α-YukE), 1∶5,000 (α -GFP), 1∶10,000 (α-Myc) or 1∶1,000,000 (α-SigmaA) in 5% nonfat milk in TBS-0.05% Tween20. The primary antibody was detected using horseradish peroxidase-conjugated goat, α-rabbit immunoglobulin G (Bio-Rad or Jackson Laboratories). Supersignal West Femto chemiluminescent substrate (Thermo Scientific) was used to create a visible chemical reaction. The blots were imaged and densitometric quantitation of YukE secretion was performed using a FlourChem FC2 gel documentation system (Alpha Innotech) and provided software. The densitometry values in Table 1 indicate the proportion of total YukE in each strain that is localized to the pellet versus supernatant; values reflect normalization based on loading of an equivalent of 0.1 OD unit for pellet samples and twenty-fold more sample loaded for supernatant samples.

Bacterial strains were grown in MC media to an OD₆₀₀ of ∼2.0. The cells were pelleted and the supernatant was collected and filtered through a 0.2 micron filter. Total proteins in the supernatant were obtained by TCA precipitating 30 mL of sample as described above. The samples were prepared for mass spectrometry analysis as described previously [27]. Briefly, samples were separated by molecular weight on a 10–20% Tricine gel (Invitrogen), each lane of the gel was sectioned into 10 roughly equal sized segments, followed by in-gel reduction, alkylation and trypsin digestion. Samples were run on a Thermo Fisher Scientific LTQ Veloz Mass Spectrometer (Thermo Fisher Scientific, Cambridge, MA). Samples were injected onto a Proxeon Easy nLC system configured with a 5 cm×100 µm trap packed with 15–20 µm PS-DVB 300A media, and a 25 cm×100 µm ID resolving column packed with 200A C18AQ media. Buffer A was 96% water, 4% methanol, and 0.2% formic acid. Buffer B was 10% water, 10% isopropanol, 80% acetonitrile, and 0.2% formic acid; loading buffer (sample loading/rinsing buffer) was 96% water, 4% methanol, and 0.2% formic acid. Samples were loaded at 5 µL min⁻¹ for 9 min, and a gradient from 0–60% B at 375 nL min⁻¹ was run over 70 min, for a total run time of 115 min (including regeneration and sample loading). Injection standards (Michrom Medium Molecule test mix, 5 angios, and the TP4 peptides) were injected at 61 fmoles per sample. Velos was run in a data dependent 15 configuration, with a full scan run in the in enhance scan mode (3^e4 target), with up to 15MS2 events. Rejection of +1 ions was used in precursor ion selection.

Resulting spectra were searched against a composite database which contained the predicted open reading frames annotated in the genome of Bacillus subtilis 168 supplemented with common contaminates using SEQUEST (Thermo Scientific, San Jose, CA). Peptides were filtered at a 1% FDR with PeptideProphet and grouped into proteins with ProteinProphet [54] with a cutoff of 0.95. Spectral counts across the gel slices for three biological replicates were pooled, and then levels of protein abundance between strains were compared using an extended G-test [55]. Data was corrected for multiple testing (Benjamini and Hochberg) using a p value of ≤0.01; for a given protein, a criterion of having ≥5 peptides in at least one strain was set.