The ESX-5 Associated eccB5-eccC5 Locus Is Essential for Mycobacterium tuberculosis Viability

The recently described ESX-5 secretion system of Mycobacterium tuberculosis is one of the most important modulators of host-pathogen interactions due to its crucial impact on PPE protein secretion, cell wall stability and virulence. Although various components of the ESX-5 secretion machinery have been defined, other ESX-5 core components still remain to be characterized. In this study, we focused on EccB5 and EccC5, a transmembrane protein (EccB5) and a membrane-bound ATPase (EccC5), both predicted to be building blocks of the M. tuberculosis ESX-5 membrane-associated complex. In vitro expression studies demonstrated that EccB5 and EccC5 encoding genes constitute an operon. The expression of this operon is essential for M. tuberculosis, since the deletion of the eccB5-eccC5 genomic segment at the ESX-5 locus is possible only after the integration of a second functional copy of eccB5-eccC5 genes into the M. tuberculosis chromosome. The characterization of two M. tuberculosis conditional mutant strains (MtbPptreccB5 and MtbPptreccC5), in which the eccB5-eccC5 operon or the eccC5 gene, respectively, were expressed under the control of an anhydrotetracycline-repressible promoter, confirmed that the repression of eccB5-eccC5 genes is detrimental for growth of M. tuberculosis both in vitro and in THP-1 human macrophage cell line. Moreover, analysis of the secretome of MtbPptreccB5-eccC5 and MtbPptreccC5 strains revealed that both EccB5 and EccC5 are required for secretion of ESX-5 specific substrates, thus confirming that they are indeed components of the ESX-5 secretion machinery. Taken together these findings demonstrate the importance of an intact and functional ESX-5 system for viability of M. tuberculosis, thus opening new interesting options for alternative antimycobacterial control strategies.


Introduction
Throughout evolution, numerous bacterial pathogens have acquired specialized protein secretion pathways to deliver effector proteins to host cells. These pathways, which are distinct from the ubiquitous Sec pathway, are critical for mediating interactions during infection and for allowing pathogen survival in the hostile environment of the host. The recently described mycobacterial type VII secretion systems are specialized secretion systems of small, highly immunogenic proteins lacking a classical N-terminal signal sequence, belonging to the Esx or WXG-100 family [1,2], which are distantly related to protein transport systems of Gram-positive bacteria [3]. The genome of Mycobacterium tuberculosis, the causative agent of human tuberculosis, encodes five type VII secretion systems (ESX-1 -ESX-5), the genes of which are arranged in highly conserved clusters [4][5][6]. Each ESX cluster typically carries a pair of esx genes flanked by esx conserved components (ecc) genes coding for predicted core components of the ESX secretion machineries responsible for the ATP-dependent transport of the corresponding ESX substrates outside the cell [5,7]. ESX-5 is the most recently evolved ESX cluster and is only present in the group of slow-growing mycobacteria that includes all major pathogenic species [8]. In the fish pathogen M. marinum, ESX-5 modulates host-pathogen interactions [9,10] and is responsible for secretion of several PPE and PE proteins, two of the most important classes of mycobacterial proteins involved in virulence/pathogenicity and implicated in immune evasion strategies used by pathogenic mycobacteria to survive in host tissues [11][12][13][14][15]. To date, PPE and PE proteins identified as being transported to the cell surface/secreted by the M. marinum ESX-5 are PE25-PPE41 [16,17], LipY [18], and members of the PPE_MPTR and PE_PGRS subgroups [19], the most recent subclasses of PE and PPE proteins, genes of which were suggested to have evolved by duplication/insertion events from ancestral genes encoded at the ESX-5 locus [8]. By the characterization of several M. tuberculosis knock-out mutants for ESX-5 components, we recently demon-strated that ESX-5 plays a crucial role in host pathogen interaction also in M. tuberculosis. The M. tuberculosis ESX-5 system mediates PPE protein transport and secretion [20] with impact on T-cell immunogenicity of ESX-5-and non-ESX-5encoded PPE and PE proteins. As such, ESX-5 is a major modulator of the host immune response [21] and a key virulence determinant of M. tuberculosis: inactivation of ESX-5 core components results in a strong attenuation of the corresponding mutant strains, which are unable to replicate both in immunodeficient and in immunocompetent mice [20,21]. Various components of the ESX-5 secretion machinery, such as EccD 5 (the predicted transmembrane channel) or EccA 5 (a cytosolic ATPase belonging to the AAA+ family) have been characterized, and their impact on secretion of ESX-5 specific substrates has been investigated [20]. However, other putative building blocks of the ESX-5 secretion apparatus in M. tuberculosis still remain to be characterized. In this study, we focused on EccB 5 and EccC 5 , a transmembrane protein and an ATP-binding protein belonging to the FtsK/SpoIIIE-like protein family, respectively [7,22]. Encoded at the ESX-5 locus upstream of the ppe25-pe19 cluster, EccB 5 and EccC 5 are both predicted to be components of the M. tuberculosis ESX-5-membrane-associated complex [7,20,23]. We demonstrated that eccB 5 -eccC 5 genes constitute an operon, expression of which is required for an efficient secretion of ESX-5 specific substrates. Moreover, by constructing/characterizing multi-copy gene variants and conditional mutants, in which the eccB 5 -eccC 5 genes were deleted or expressed under the control of an anhydrotetracycline-repressible promoter, we demonstrate that an intact eccB 5 -eccC 5 locus is essential for M. tuberculosis and that disruption/repression of single core components of the ESX-5 secretion machinery strongly impacts the M. tuberculosis in vitro growth properties. Taken together, the results obtained demonstrated the importance of an intact and functional ESX-5 for M. tuberculosis viability, emphasizing the key role of this secretion system in the biology of this human pathogen.
For analysis of in vitro growth of conditional mutants, mycobacterial strains in exponential growth phase were diluted to an optical density at 600 nm (OD 600 ) of 0.5, and 5 ml of 10-fold serial dilutions were spotted on Middlebrook 7H11 medium, added or not with different concentrations of anhydrotetracycline (ATc) (100, 200 and 400 ng/ml). Alternatively, bacterial strains were grown in Middlebrook 7H9 medium containing or not 400 ng/ml ATc. After 3 days, cultures were diluted to OD 600 = 0.04 in fresh medium containing or not 400 ng/ml ATc, and the bacterial growth was monitored by daily OD 600 measurements.

RNA Extraction and Reverse Transcription RT-PCR/59 RACE Assays
RNA extraction was performed from M. tuberculosis H37Rv cultures in exponential growth phase as previously described [24]. Briefly, bacteria were recovered by centrifugation and broken in 1 ml of TRIzol (Applied Biosystems) in presence of zirconia beads (0.1 mm diameter) in a MM300 apparatus (Qiagen) (30 sec at maximum speed). RNA was obtained by extraction with 0.2 vol of chloroform and precipitation for 1 h at -80uC with 0.1 vol of 3 M sodium acetate and 0.45 vol of isopropanol. Removal of contaminating DNA was performed using DNAfree kit (Applied Biosystems/Ambion), according to the manufacturer's instructions. RT-PCR reactions were performed using Superscript one-step RT-PCR kit (Applied Biosystems) as recommended by the producer. Sequences of primers used in amplification reactions are listed in Table S1. 59 Rapid Amplification of cDNA Ends (RACE) was performed using the 59/39 RACE kit (Roche Molecular Biochemicals). One microgram of RNA and eccB 5 -specific primer (Table S1) were incubated at 70uC for 5 min. Denatured RNA and primer were then incubated at 55uC for 1 h in the presence of 1X cDNA synthesis buffer, 1 mM dNTPs, 40 U Protector RNase Inhibitor and 25 U Transcriptor Reverse Transcriptase. The cDNA obtained was purified by using the High Pure PCR Product Purification Kit (Roche Molecular Biochemicals), and used in poly(A) tailing reaction (30 min at 37uC in the presence of 0.2 mM dATP and 80 U Terminal Transferase). Nested PCR amplification on poly(A)-tailed cDNA was performed using an oligo dTanchor primer and an eccB 5 specific primer (Table S1). The single amplification product obtained was directly sequenced.

Construction of eccB 5 -eccC 5 Deletion and Conditional Mutants in M. tuberculosis
Deletion of the eccB 5 -eccC 5 segment in M. tuberculosis H37Rv was performed by allelic exchange using the ts-sacB technology [25]. Briefly, a 2445 bp and a 1227 bp fragment encompassing the eccB 5 -upstream region and the eccC 5 -downstream region, respectively, were amplified by PCR from M. tuberculosis H37Rv genomic DNA (See Table S1 for primer details). The amplicons were digested with SpeI/XbaI and XbaI/NotI, respectively, and cloned into the SpeI/NotI-digested pPR27 vector to obtain pMDL92. The aph cassette, conferring resistance to kanamycin, was amplified by PCR from the pUC4K plasmid, digested by XbaI and inserted into the XbaI-digested pMDL92 plasmid. The resulting plasmid, pMDL92-aph, was used in allelic exchange experiments. Kanamycin-resistant/sucrose-resistant (Kana r /Sac r ) transformants were screened by PCR using primers specific for the aph cassette and for the rv1780 or rv1786 genes (see Table S1 for primer sequences).
The Mtb::eccB 5 -eccC 5 merodiploid strain was constructed by using the integrative vector pRBexint [26]. The genomic segment eccB 5 -eccC 5 was amplified by PCR (see Table S1 for primer details), digested with SpeI/HpaI and cloned into the SpeI/HpaI digested-pRBexint. The resulting plasmid (pExinteccB 5 -eccC 5 ) was used to transform M. tuberculosis, and hygromycin resistant clones were selected. To construct eccB 5 and eccC 5 conditional mutants the recently developed TetR/Pip OFF mycobacterial repressible system was used [27,28]. Briefly, a recombinant M. tuberculosis strain (Mtb::tetR-pip) was constructed, in which the genes encoding the tetracycline-sensitive repressor TetR and the Streptomyces pristinaespiralis Pip repressor, as well as the reporter gene lacZ were integrated into the genome at the attB site. Such a genomic organization allows the transcriptional repression of genes expressed under the control of the P ptr promoter as a consequence of the addition of ATc to the medium [27,28]. The functionality of the PiP ON/Tet OFF regulatory circuit was confirmed by ßgalactosidase activity assays performed on the Mtb::tetR-pip strain, demonstrating that the addition of ATc to the culture medium results in a dose-dependent reduction of the expression of the lacZ gene reporter ( Figure S1).
A 950-bp and 876-bp fragment encompassing the 59-portions of eccB 5 and eccC 5 genes, respectively, were amplified by PCR on M. tuberculosis genomic DNA (see Table S1 for primer details), digested by NsiI and cloned in frame with the S. pristinaespiralis promoter P ptr , into the NsiI site of suicide vector pFRA50 [27]. The two obtained plasmids, pMDL-eccB 5 and pMDL-eccC 5 were then electroporated in the Mtb::tetR-pip strain, and recombinant clones were selected for resistance to hygromycin. Genomic DNAs from hygromycin resistant clones were analyzed by PCR for the correct integration of the P ptr promoter immediately upstream the eccB 5 or eccC 5 coding regions, respectively. Sequences of primers used in screening PCR reactions are reported in Table S1.

Infection of THP-1 Human Macrophage Cell Line
THP-1 cells were cultured in RPMI medium supplemented with 10% heat-inactivated fetal calf serum (Euroclone) and 2 mM L-glutamine (Euroclone). Before infection, cells were seeded in 96well plates at a density of 7.5610 4 and differentiated into macrophages by incubation with 50 nM PMA for 1 day. The Mtb Pptr eccB 5 -eccC 5 and Mtb Pptr eccC 5 conditional mutants as well as the M. tuberculosis control strain were pre-grown for 48 h in ATccontaining medium, and used in cell infection assays at a multiplicity of infection (m.o.i.) of 1:20 (bacteria:cells). After phagocytosis (90 min), infected cells were washed twice with PBS to remove extracellular bacteria, and incubated for 6 days in culture media supplemented or not with 400 ng/ml ATc. Culture media, containing or not ATc, were replaced every 48 h as described elsewhere [27]. At different time points (immediately after phagocytosis, and 2, 4 and 6 days post infection), cells were lysed in PBS 0.01% Triton X-100 and the number intracellular bacteria was determined by plating 10-fold serial dilutions of cell lysates on solid medium.

Preparation of Culture Supernatants and Total Lysates, SDS-PAGE and Immunoblotting
Conditional mutants and M. tuberculosis control strain were grown in Middlebrook 7H9 medium as described above. Precultures in exponential growth phase were diluted to OD 600 0.04 in Middlebrook 7H9 fresh medium added with 0.1% (v/v) ADC, and containing or not 400 ng/ml ATc. After 3 and 6 days, culture supernatants were recovered and proteins were precipitated with 10% (w/v) TCA as previously described [19]. To obtain total lysates, mycobacterial pellets were washed twice and resuspended in 20 mM TrisCl. Bacterial cells were broken by shaking with acid washed-glass beads (106 mm diameter) for 8 min, in a Tissue Lyser apparatus (Qiagen). Suspensions were centrifuged at 5,000 rpm at 4uC for 30 min and the supernatant fraction obtained represented the total-cell lysate. Immunoblot analyses were performed with rabbit anti-EsxN [20] and anti-PPE41 [16] polyclonal sera, or anti-GroEL2 mouse monoclonal antibody (Colorado State University, NIH, NIAID contract NO AI75320) as previously reported [20].

Results
Characterisation of the eccB 5 -eccC 5 Locus in M. tuberculosis H37Rv EccB 5 and EccC 5 are encoded by genes eccB 5 (rv1782) and eccC 5 (rv1783-rv1784) located in the 59 region of the ESX-5 locus of M. tuberculosis ( Figure 1A). As for the ESX-1-associated EccC 1 , which is encoded by the two adjacent genes eccCa 1 (rv3870) and eccCb 1 (rv3871), EccC 5 was thought to be encoded by two separate genes, eccCa 5 (rv1783) and eccCb 5 (rv1784) in the M. tuberculosis H37Rv reference strain [7,22]. Conversely, it was reported to be encoded by a single non-divided eccC 5 gene in several mycobacterial species such as Mycobacterium leprae, Mycobacterium bovis and Mycobacterium marinum, as well as in other M. tuberculosis strains ( Figure 1B) [29][30][31][32]. Sequence verification of the M. tuberculosis H37Rv strain used in our study revealed that the in frame stop codon separating eccC 5 gene into two genes, previously reported in the H37Rv reference genome sequence [7,22], seems to be due to a sequencing error (A instead of T at position 2020563) ( Figure  S2). Thus, in agreement with sequence data reported for other M. tuberculosis H37Rv variants [32], a single eccC 5 is also present in the M. tuberculosis H37Rv strain used in this study ( Figure 1A).
The genomic organisation of eccB 5 /eccC 5 locus ( Figure 1A) suggested that the two genes might be cotranscribed and form a transcriptional unit independent from their flanking genes. Analysis of eccB 5 /eccC 5 transcripts performed by RT-PCR using different combinations of eccB 5 /eccC 5 gene-specific primers revealed a 670-bp amplicon when a combination of primers specific for eccB 5 and eccC 5 was used ( Figure 1C, lane 4), thus indicating that these two genes are indeed cotranscribed. Transcription data were confirmed by mapping the 59 end of the eccB 5 -eccC 5 transcript using 59 RACE, which identified only one specific transcript starting 139 bases upstream of the eccB 5 start codon ( Figure 1E).
The eccB 5 -eccC 5 Operon is Essential for in Vitro Growth of M. tuberculosis Insertional mutagenesis and transposon site hybridization (TraSH) studies performed in M. tuberculosis identified transposon insertion mutants in various genes coding for different components of the ESX-5 secretion apparatus, such as eccA 5 or eccD 5 , but no consistent data were available for eccB 5 -eccC 5 [33,34]. To assess the potential essentiality of the eccB 5 -eccC 5 operon, an attempt to delete both genes was performed by using a classical knockout strategy based approach, which employs the replicative thermosensitive vector pPR27, carrying the sacB counterselectable marker [25]. A recombinant pPR27 plasmid containing the aph cassette encoding resistance to kanamycin flanked by eccB 5 -upstream and eccC 5downstream regions was constructed and used in the allelic replacement experiments. According to this strategy, at the end of the procedure putative M. tuberculosis double cross over recombinants with a kan r /suc r phenotype were expected to carry the aph cassette replacing the eccB 5 -eccC 5 operon. However, PCR analysis of genomic DNAs from 150 clones, using primers specific for the aph cassette and rv1780 or rv1786 genes, revealed that none of them showed the expected pattern for an allelic exchange mutant ( Figure 2 A and B). While eccB 5 -and aph specific amplification products were detected in PCR control reactions ( Figure 2B, lanes 1-2 and 5-6, respectively), no aph-rv1786 specific amplification products were obtained ( Figure 2B, lanes 3-4), demonstrating the presence of an intact eccB 5 -eccC 5 operon at the ESX-5 locus, as well as the non homologous integration of the aph cassette into the genomes of the tested kan r /suc r clones. As in a previous work the same technique has allowed the deletion of genomic segments up to , 20 kb (in size) in M. tuberculosis [24], the inability to obtain eccB 5 -eccC 5 knock-out mutants in this study strongly suggests that the deletion of the eccB 5 -eccC 5 operon might be lethal for M. tuberculosis.
To confirm the requirement of an intact eccB 5 -eccC 5 operon for growth of M. tuberculosis an Mtb::eccB 5 -eccC 5 merodiploid strain, carrying an additional wild-type copy of eccB 5 -eccC 5 genes integrated into the attB site of the genome, was constructed. Allelic exchange experiments were then repeated in the Mtb::eccB 5 -eccC 5 merodiploid genetic background. Again, genomic DNAs from resultant kan r /suc r clones were tested by PCR, using combinations of primers specific for the aph cassette and rv1780 or rv1786 genes. As depicted in Figure 2D for a representative mutant (Mtb488), the presence of 2706-bp aph-rv1780specific and 1585-bp aph-rv1786specific amplification products confirmed the replacement of the wild-type eccB 5 -eccC 5 genomic segment at the ESX-5 locus with the aph cassette. PCR analysis revealed that the deletion of eccB 5 -eccC 5 genes at the ESX-5 locus had occurred in 5% of the recombinants, thus demonstrating that only the presence of an additional copy of eccB 5 -eccC 5 genes allowed the deletion of the chromosomal eccB 5 -eccC 5 segment to take place.
Construction of Mtb Pptr eccB 5 and Mtb Pptr eccC 5

Conditional Mutant Strains
To investigate more in depth the effect of inactivation of each gene of the eccB 5 -eccC 5 locus on growth of M. tuberculosis, two conditional mutant strains were constructed using the Pip ON/Tet OFF repressible system [27]. This system, which has been successfully used to construct and characterize an M. tuberculosis conditional mutant for the ESX-3 secretion system [28], is based on the expression of the gene of interest under the control of the anhydrotetracycline-repressible P ptr promoter [27]. As schematically depicted in Figure 3A, in the first mutant, the P ptr promoter is placed immediately upstream of the eccB 5 coding sequence, replacing the physiological eccB 5 promoter. As eccB 5 and eccC 5 genes are cotranscribed from the same promoter ( Figure 1 C and E), this mutant is actually a conditional mutant for the eccB 5 -eccC 5 operon and was thus referred as Mtb Pptr eccB 5 -eccC 5 . In the second mutant (Mtb Pptr eccC 5 ), the P ptr promoter is placed immediately upstream the eccC 5 coding region, so that only the eccC 5 is regulated by P ptr , while the eccB 5 gene is expressed from its own promoter. Both conditional mutants are derivatives of the Mtb::tetR-pip strain and carry the tetracycline-sensitive repressor TetR and the Pip repressor encoding genes integrated into the chromosome at the attB site. A 950-bp and 876-bp fragment encompassing the 59portions of eccB 5 and eccC 5 coding regions, respectively, were cloned in frame with the P ptr promoter into the suicide vector The eccB 5 -eccC 5 Locus of M. tuberculosis PLOS ONE | www.plosone.org pFRA50, and recombinant plasmids obtained were used to transform the Mtb::tetR-pip strain. Analysis by PCR on genomic DNA from selected hygromycin resistant clones confirmed the correct replacement of the eccB 5 -eccC 5 promoter with the P ptr promoter in the Mtb Pptr eccB 5 -eccC 5 strain, and the insertion of the P ptr promoter upstream the eccC 5 gene in the Mtb Pptr eccC 5 mutant ( Figure 3B). In these strains, the expression of eccB 5 -eccC 5 operon or the eccC 5 gene is exclusively regulated by the activity of the P ptr promoter, which in turn is strictly dependent on the absence/ presence of ATc in the culture medium ( Figure S3). In the absence of ATc, TetR binds to its operators, turning off the pip transcription, and thus allowing the expression of the eccB 5 -eccC 5 or the eccC 5 genes from the P ptr promoter (Figure S3 A). In the presence of ATc, the transcription of pip is allowed. The Pip production results in the block of the P ptr promoter activity, and finally in the repression of eccB 5 -eccC 5 or eccC 5 gene expression (Figure S3 B).
Repression of eccB 5 -eccC 5 Shows Stronger Impact on in vitro Growth of M. tuberculosis than Repression of eccC 5 The impact of EccB 5 and EccC 5 on M. tuberculosis viability was determined by evaluating the growth properties of Mtb Pptr eccB 5 -eccC 5 and Mtb Pptr eccC 5 conditional mutants both on solid and liquid media. Ten-fold dilutions of Mtb Pptr eccB 5 -eccC 5 and Mtb Pptr eccC 5 mutants or M. tuberculosis control strain were spotted onto Middlebrook 7H11 medium added with different concentrations of ATc, ranging from 100 to 400 ng/ml. As control, bacterial dilutions were also spotted on the same medium without the antibiotic. While the growth of the control strain was not affected by the presence of ATc even at the highest concentration of the antibiotic, the growth of the Mtb Pptr eccB 5 -eccC 5 mutant was strongly affected by the presence of the antibiotic in an ATc-concentration dependent manner (Figure 4). After 2 weeks of incubation, no growth was indeed observed for the Mtb Pptr eccB 5 -eccC 5 strain on plates containing the highest concentration of ATc, and some colonies were detectable for the mutant only on agar plates from two representative kan r / suc r clones, obtained in PCR reaction carried out by using primers specific for the aph cassette and eccB 5 -eccC 5 flanking genes (not included in the plasmid construct used in allelic exchange experiments), namely rv1780 (lanes 3-4) and rv1786 (data not shown). In these reactions, amplification products were expected only if the eccB 5 -eccC 5 operon at the ESX-5 locus is replaced by the aph cassette; in contrast, no amplification products were expected in case of non-homologous recombination. PCR control reactions were performed using primers specific for the aph cassette (lanes 1-2) or the eccB 5 gene (lanes 5-6). (C) Genomic organization of Mtb488, a representative mutant deleted for the eccB 5 -eccC 5 operon in the Mtb::eccB 5 -eccC 5 genetic background. The gray box represents the aph cassette replacing the eccB 5 -eccC 5 operon at the ESX-5 locus; dark gray arrows indicate the additional copies of eccB 5 and eccC 5 genes integrated at the attB site. (D) PCR profiles of genomic DNA from Mtb488 (lanes 1 and 3) amplified with primers specific for the aph cassette and rv1780 gene (lane 1) or the aph cassette and rv1786 gene (lane 3). Again, these PCR reactions were expected to give amplification products only if the native eccB 5 -eccC 5 operon was replaced by the aph gene. As expected, no amplification products were detected in PCR reactions performed/carried out on genomic DNA from wild-type H37Rv M. tuberculosis (lanes 2 and 4), which was used as negative control. doi:10.1371/journal.pone.0052059.g002 containing lower ATc concentrations. After 3 weeks of incubation, whereas bacterial growth was detected for the control strain at higher concentrations of antibiotic and less concentrated bacterial dilutions, some growth was detectable for the Mtb Pptr eccB 5 -eccC 5 mutant, but only at the most concentrated bacterial dilutions.
Similar results were obtained when the growth kinetics of the Mtb Pptr eccB 5 -eccC 5 conditional mutant and control strain were analyzed in liquid medium ( Figure 5). Again, the growth of the M. tuberculosis control strain was not affected by the addition of ATc into the medium ( Figure 5C). In contrast, the growth of Mtb Pptr eccB 5 -eccC 5 mutant was impaired in the presence of 400 ng/ml ATc ( Figure 5A), a growth defect that was detectable after 3 days from the addition of the ATc. As EccB 5 and EccC 5 are predicted to be structural components of the M. tuberculosis ESX-5 membrane complex it is possible that EccB 5 and EccC 5 proteins already produced by bacterial strains before the exposure to ATc might compensate the effect of eccB 5 -eccC 5 gene repression immediately after the addition of the antibiotic. These results further confirm that the repression of the eccB 5 -eccC 5 operon strongly inhibits the M. tuberculosis in vitro growth.  The Mtb Pptr eccC 5 strain, in which only the eccC 5 gene is repressed in presence of ATc, showed a different phenotype. This mutant displayed impaired growth as compared to the control strain on agar plates containing ATc (Figure 4), indicating that the repression of eccC 5 affects the ability of M. tuberculosis to grow on solid medium. In contrast, no difference between the Mtb Pptr eccC 5 mutant and the control strain was detected when the growth was evaluated in liquid medium ( Figure 5). In this case, similarly to the M. tuberculosis control strain, the growth of EccC 5 conditional mutant was not affected by the addition of ATc, and comparable growth kinetics were observed for the Mtb Pptr eccC 5 strain in Middlebrook 7H9 medium added or not with the antibiotic marker ( Figure 5B). Taken together with the results from knockout experiments these data clearly emphasize that the eccB 5 -eccC 5 operon is essential for M. tuberculosis viability. eccB 5 -eccC 5 and eccC 5 Repression Impairs the Growth of M. tuberculosis in THP-1 Derived Macrophages Inactivation of single ESX-5-associated genes encoding building blocks of the ESX-5 secretion apparatus (e.g. eccD 5 ) results in the loss of the ability of M. tuberculosis to replicate in ex vivo models [20]. To assess the impact of eccB 5 -eccC 5 and eccC 5 repression on the M. tuberculosis intracellular growth properties the growth kinetics of Mtb Pptr eccB 5 -eccC 5 and Mtb Pptr eccC 5 conditional mutants were compared in THP-1-derived macrophages. Cells were infected with different strains at a m.o.i of 1:20 (bacteria:cells) and cultured in presence or not of 400 ng/ml ATc. At these concentrations, ATc is not toxic for THP-1, as evaluated by Alamar blu assay (data not shown), but can penetrate into the cells and down regulate the expression of P Ptr -controlled genes [27]. At different times post infection (after phagocytosis and at day 2, 4 and 6), the number of viable intracellular bacteria was determined. Results obtained are depicted in Figure 6. While the intracellular growth of the control strain was not affected by the presence of the antibiotic in the culture medium ( Figure 6 C and F), the growth of the eccB 5 -eccC 5 mutant was strongly inhibited in presence of ATc ( Figure 6 A and D). At day 6 post infection, the CFU ratio (CFU/ CFU at day 0) value for this mutant was 2.4 in THP-1 macrophages cultured in medium containing ATc, whereas it was 28.8 in cells incubated in the absence of the antibiotic (Figure 6  D). A similar attenuation was observed for the eccC 5 mutant (Figure 6 B and E), for which CFU ratio values were 3.5 and 28.4 in THP-1 macrophages cultured in presence or in absence of ATc, respectively (Figure 6 E). These data demonstrate that expression of eccB 5 and eccC 5 is required for optimal replication of M. tuberculosis in macrophages.

EccB 5 and EccC 5 are Required for Secretion of ESX-5 specific Substrates
ESX-5 is responsible for secretion/transport of EsxN, the Esx protein encoded by the ESX-5 locus, and PPE41, a representative member of the large PPE protein family [20]. To determine the impact of EccB 5 and EccC 5 on secretion of ESX-5 specific substrates, the presence of EsxN and PPE41 in culture supernatants from Mtb Pptr eccB 5 -eccC 5 and Mtb Pptr eccC 5 strains grown in the presence or not of ATc was investigated. As expected, similar amounts of EsxN were detected in the culture supernatants from EccB 5 and EccC 5 mutants grown in the absence of ATc, as well as in samples from M. tuberculosis control strain (Figure 7). In contrast, while the addition of ATc did not affect the EsxN secretion in the control strain, the presence of the antibiotic abolished the secretion of the protein in EccB 5 and EccC 5 conditional mutants. Three and six days after addition of ATc to the medium, no EsxN was indeed detectable in the culture supernatants from Mtb Pptr eccB 5 -eccC 5 and Mtb Pptr eccC 5 strains, despite the presence of an EsxN-specific band in the corresponding total lysate samples (Figure 7). Similar results were obtained when the secretion of PPE41 was analyzed. Comparable amounts of the protein were detected in the culture supernatants recovered at day 6 from all mycobacterial strains grown in the absence of ATc (Figure 7). Again, while the addition of the antibiotic did not impact the secretion of PPE41 in the control strain, the presence of ATc into the medium resulted in a significant reduction of the amount of PPE41 exported in culture supernatants form the Mtb Pptr eccB 5 -eccC 5 and Mtb Pptr eccC 5 mutants (Figure 7).
Together these results confirmed that EccB 5 and EccC 5 are building blocks of the ESX-5 secretion machinery, both required for transport of ESX-5 specific substrates outside the cell.

Discussion
In previous studies we demonstrated that selected ESX-5 encoded genes had an impact on secretion of ESX-5 specific substrates, cell wall integrity and virulence of M. tuberculosis [20,21]. However, for some other ESX-5 associated genes, i.e. eccB 5 /C 5 , initial gene-inactivation attempts were unsuccessful, which made us think that these genes might be essential for M. tuberculosis. As results from TraSH analysis were inconclusive for this genomic locus [34], the impact of the eccB 5 and eccC 5 on in vitro growth of M. tuberculosis was thus studied in more detail. The ability to construct an eccB 5 -eccC 5 deletion mutant only in the presence of a second functional copy of eccB 5 -eccC 5 genes, as well as the inability of Mtb Pptr eccB 5 -eccC 5 (in which the eccB 5 -eccC 5 operon is repressed) to grow on solid medium consistently demonstrate that an intact eccB 5 -eccC 5 locus is essential for viability of M. tuberculosis. In a recent study by Griffin and colleagues, in which M. tuberculosis essential genes were identified by using high density transposon mutagenesis and next generation sequencing techniques, no transposon insertions were found in rv1783 gene (encoding the N-terminal part of EccC 5 ), whereas only a few insertions were detected in rv1784 (encoding the C-terminal part of EccC 5 ) or in rv1782 (eccB 5 ) [35], further indicating that, in accordance with our data, an intact eccB 5 -eccC 5 (rv1782-rv1784) locus is indispensable for growth of M. tuberculosis. The finding that the only ESX-5 M. marinum mutants identified in transposon mutagenesis studies were inactivated for MMAR_2676 and MMAR_2680/eccA 5Mm genes (orthologous to the M. tuberculosis rv1794 and rv1798/eccA 5Mt genes, respectively) [16,19], suggests that the ESX-5 locus in M. marinum might also encode some ESX-5 components that are essential for its in vitro growth. It seems clear that the inactivation of ESX-5 via the simultaneous deletion/ repression of various genes encoding building blocks of the ESX-5 secretion machinery strongly affects the viability of M. tuberculosis, and might explain the reason why in this study it was not possible to obtain M. tuberculosis mutants deleted for the eccB 5 -eccC 5 operon. EccB 5 and EccC 5 are required for secretion of ESX-5 specific substrates ( Figure 7) and are both predicted to encode membranebound ESX-5 components. During the preparation of this manuscript, the characterization of the composition of the ESX-5 membrane-bound complex in M. marinum and M. bovis BCG was reported [23]. Although the impact of eccB 5 -eccC 5 inactivation on mycobacterial in vitro growth properties was not investigated, secretion and/or biochemical data confirmed the involvement of EccB 5 and EccC 5 as core components of the ESX-5 secretion apparatus.
ESX-5 has a strong impact in maintaining the mycobacterial cell wall integrity, and inactivation of a single core component of the ESX-5-transmembrane complex results in an increased sensitivity to detergents and hydrophilic antibiotics to which mycobacteria are naturally resistant [20]. It is possible that the disruption of large portions of the ESX-5 secretion apparatus, as is the case for the eccB 5 -eccC 5 deletion/repression, causes more extensive damage to the cell wall or more profound alterations to cell wall stability, resulting in the inability of the corresponding mutant strain to grow both on solid and liquid media. We also observed the intriguing phenomenon that inactivation of a single ESX-5 core component, such as EccC 5 , has an impact on the mutant's growth characteristics on solid medium, whereas it has not any discernable inhibitory effect on growth in liquid medium. Interestingly, eccC 5 repression strongly affects the intracellular growth properties of M. tuberculosis, further confirming that repression of single genes encoding ESX-5 structural components is sufficient to strongly impair the growth of M. tuberculosis in a restrictive environment such as the macrophage. A similar phenotype has indeed been previously observed for a different ESX-5 mutant, in which the eccD 5 gene encoding the predicted ESX-5 transmembrane channel was disrupted [20]. The EccD 5 mutant is unable to replicate in murine macrophages, displayed an impaired growth on solid medium, as revealed by a small colony morphotype on Middlebrook 7H11 agar plates, but showed no difference in growth as compared to the control strain in liquid medium, thus suggesting a potential role for the ESX-5 system in modeling the bacterial surface during colony formation on solid medium. A functional link between ESX secretion systems and mycobacterial cell wall has been demonstrated in M. marinum, where the ESX-1-encoded EccA 1 ATPase, involved in secretion of ESX-1 substrates, was also found to be required for optimal synthesis of mycolic acids [36]. The recent finding that a plethora of genes encoding enzymes involved in cell-wall synthesis or functioning as well as various ESX loci (ESX-1, ESX-2 and ESX-5) are regulated by the nucleoid-associated EspR regulator [37], provides further evidence of the link between two mycobacterial virulence hallmarks such as ESX-mediated protein secretion/ transport and cell envelop biogenesis.
In addition to the alterations of cell wall properties, it is plausible that such impairment in transport and secretion of ESX-5 specific substrates could contribute to the loss of viability observed after disruption of large portions of ESX-5. As a functional ESX-5 is required not only for secretion of PPE proteins [20,21] but also for their correct localization in the cell wall [20], it cannot be excluded that the massive intracellular accumulation of un-secreted or incorrectly localized ESX-5 substrates can exert a toxic, lethal effect for mycobacterial cells. ESX-5 systems of various mycobacterial species have been demonstrated or predicted to be involved in transport of PE_PGRS [19], and a number of PE/PPE proteins encoded in and outside the ESX-5 locus [20,21]. Consistent with previous data reported for another M. tuberculosis H37Rv ESX-5 knock-out strain [20], Western blot analyses of culture supernatants and cell lysates using a monoclonal antibody specific for the PGRS domain [19] did not reveal differences in the PE_PGRS profiles in samples from Mtb Pptr eccB 5 -eccC 5 or Mtb Pptr eccC 5 mutant strains, grown in the presence or not of ATc (data not shown). This finding thus argues against the possibility that the effects of eccB 5 -eccC 5 and eccC 5 repression on M. tuberculosis growth, as well as the phenotypic differences observed for eccB 5 -eccC 5 and eccC 5 mutants might be related to alterations/ differences in secretion or intracellular accumulation of PE_PGRS proteins. On the other hand, information on the presence of ESX-5 associated PE/PPE proteins and their homologs encoded outside the ESX-5 locus in samples from eccB 5 -eccC 5 and eccC 5 mutants are not available, due to the lack of antibodies specific for these proteins. Thus, at present it cannot be excluded that the defect in in vitro growth observed for eccB 5 -eccC 5 or eccC 5 mutants might also be caused in part by the lack of transport and/or by the intracellular accumulation of these substrates (or some of them).
Although an intact ESX-5 system is required for optimal growth of M. tuberculosis and M. marinum, the ESX-5 locus is absent in the genomes of fast growing, saprophytic mycobacterial species [8]. These findings suggest that during the evolution of the slowly growing, pathogenic mycobacteria, duplication-diversification events have led to the emergence of ESX-5 systems that were apparently linked to the expansion of the PE/PPE protein family. It might well be that fast-growing mycobacteria, which possess only a very limited set of PE/PPE proteins, have no need for an ESX-5 secretion system, while slow-growers that harbor a wide range of PE/PPE proteins require a fully functional ESX-5 system that exports these proteins to the cell envelop and beyond for their viability. The described situation resembles observations made for the ESX-3 system involved in mycobactin-mediated iron uptake, which is essential in M. tuberculosis but dispensable for growth in the saprophytic species M. smegmatis [27,38]. Furthermore, ESX-1 is a key virulence determinant in pathogenic mycobacteria but seems to regulate the DNA transfer in M. smegmatis [2,39]. Such differences represent an intriguing and unexplored aspect of ESX secretion systems, leading to the requirement to study the individual ESX systems in the context of the mycobacterial species concerned. Because of their crucial role in host-pathogen interactions as well as their involvement in basic biological processes of tubercle bacilli, ESX secretion systems of M. tuberculosis thus represent potential targets for new anti-tuberculosis drugs directed against key mycobacterial factors required for viability and virulence. Figure S1 Characterization of the TetR/Pip OFF system in the Mtb::tetR-pip strain. ß-galactosidase assay performed on Mtb::tetRpip strain grown on Middlebrook 7H11 plates containing X-gal (40 mg/ml) and different concentrations of ATc, ranging from 0 to 200 ng/ml. The ß-galactosidase activity was clearly detected when bacteria were grown in the absence of ATc, and decreased in the presence of increasing concentrations of the antibiotic. The ßgalactosidase activity was abolished when Mtb::tetR-pip was grown on Middlebrook 7H11 medium containing 200 ng/ml ATc.