Roles of trpE2, entC and entD in salicylic acid biosynthesis in Mycobacterium smegmatis


  • Editor: Dieter Jahn

Correspondence: Present address: Nivedita Nagachar, Immunology Department, Lund University, BMC D14, Lund 22184, Sweden. Tel.: +46 4622 24081; fax: +46 4622 24218; e-mail:


Mycobacterium smegmatis acquires extracellular iron using exochelin, mycobactin and carboxymycobactin. The latter two siderophores are synthesized from salicylic acid, which, in turn, is derived from chorismic acid in the shikimic acid pathway. To understand the conversion mechanism of chorismic acid to salicylic acid in M. smegmatis, knockout mutants of the putative key genes, trpE2, entC and entD, were created by targeted mutagenesis. By enzymatic assays with the cell-free extracts of the various knockout mutants, we have shown that TrpE2 converts chorismic acid into isochorismic acid and is thus an isochorismate synthase. The gene products of both entC and entD are involved in the conversion of isochorismic acid into salicylic acid, and hence correspond to salicylate synthase.


Mycobacteria, when grown under low iron conditions, overproduce salicylic acid (Ratledge & Winder, 1962), which is the aromatic moiety of mycobactin and carboxymycobactin. Mycobactin is the major intracellular siderophore of most mycobacteria, including the major pathogens, Mycobacterium tuberculosis and Mycobacterium avium. However, due to its lipophilicity, mycobactin acts as a repository for holding iron within the cell envelope before its release into and through the cytoplasmic membrane. Iron acquisition from the external environment is then achieved either using carboxymycobactin (which occurs in both pathogenic and saprophytic mycobacteria) or using chemically unrelated siderophores, the exochelins, which occur only in the saprophytic species (Ratledge, 1999; Ratledge & Dover, 2000).

Salicylic acid is also a precursor of other bacterial siderophores, for example in Azospirillum, Burkholderia, Pseudomonas, Vibrio and Yersinia (Saxena et al., 1986; Ankenbauer & Cox, 1988; Sokol et al., 1992; Okujo et al., 1994; Serino et al., 1995; Gehring et al., 1998). The conversion of salicylic acid to mycobactin was demonstrated some years ago (Ratledge, 1969; Hudson & Bentley, 1970; Ratledge & Hall, 1970, 1972) as was the synthesis of salicylic acid via the shikimic acid pathway. This latter pathway was shown to proceed via the formation of chorismic and isochorismic acids (Marshall & Ratledge, 1971, 1972; Ratledge & Dover, 2000). In M. tuberculosis, although not all steps or enzymes for mycobactin biosynthesis have been identified, most of them are clustered in the putative mbt operon (extending from Rv2377c to Rv2386c) (Cole et al., 1998). trpE2, reannotated as mbtI, encodes isochorismate synthase (ICS) that catalyzes the first step in the formation of salicylate from chorismic acid (Quadri et al., 1998). Indeed, trpE2 shows greater homology to pchA, coding for ICS in Pseudomonas aeruginosa, than to trpE, coding for anthranilate synthase that is required for tryptophan biosynthesis and from which it derives its name (Gaille et al., 2003). In P. aeruginosa, pchA, the last gene of the pchDCBA operon, codes for ICS in the first step of pyochelin biosynthesis (Serino et al., 1995, 1997; Gaille et al., 2003).

Based on the homology studies, only a few genes (trpE2, entC and/or entD) have been considered in the overall conversion of chorismic acid to salicylic acid: entD is suggested to be involved in some aspect of iron utilization (Cole et al., 1998), but without assignment to a specific enzyme activity. entC and trpE2 of Mycobacterium smegmatis show sequence homology to ICS of Escherichia coli, Bacillus subtilis and P. aeruginosa both at DNA and protein levels (Cole et al., 1998; blastn and blastp searches); entC is also adjacent to entD. In addition, the TGA stop codon of entD overlaps with the putative GTG start codon of entC. This feature of overlapping of genes also occurs in P. aeruginosa in the pchB/pchA and pchE/pchF operon encoding the enzymes involved in the formation of salicylate from chorismate and pyochelin from salicylate (Sokol et al., 1992; Visca et al., 1993; Serino et al., 1995). The proteins, MbtI of M. tuberculosis, YbtS of Yersinia pestis and Irp-9 of Yersinia enterocolitica, are all members of the ICS family (Gaille et al., 2003). Unlike PchA, they may carry out the direct conversion of chorismate to salicylate as a single reaction because of the absence of the PchB homolog in these organisms in the vicinity of the ICS genes (Gehring et al., 1998; Quadri et al., 1998).

Enzymes involved in mycobactin biosynthesis are now important targets for the design of specific inhibitors that could then be useful for the treatment of the diseases caused by mycobacteria. The conversion of chorismic acid to salicylic acid in M. smegmatis by targeted mutagenesis has not been reported in comparison with other organisms producing siderophores. Hence, as a step further to this aspect, we have studied the functions of three key genes, trpE2, entC and entD, in salicylate biosynthesis by carrying out targeted mutagenesis of each one in M. smegmatis and then assessing their efficiency in converting chorismic acid to salicylic acid.

Materials and methods

Organisms and growth

The wild-type strain M. smegmatis mc2155 was used throughout.

Initial cloning experiments were performed in E. coli DH5α as a host, where all the genes of interest were internally deleted and the final suicide delivery vector was constructed for homologous recombination with the M. smegmatis genome.

Mycobacterium smegmatis was grown in a chemically defined (glycerol/asparagine) minimal medium (Ratledge & Hall, 1971). The medium (100 mL in 250 mL conical flasks with shaking) was supplemented with Fe2+ at 0.01 μg mL−1 (for iron-deficient growth) or at 2 μg mL−1 (for iron-sufficient growth).

Genomic DNA was isolated from both wild type and mutants grown in Lab Lemco medium (Belisle & Sonnenberg, 1998) as the growth of mutants was better in the enrichment medium compared with the minimal medium, whereas the production of siderophores was studied by growing them in the minimal medium as the iron concentration in the medium could be controlled as required.

Gene isolation and disruption

Primers were designed using the primer 3 analysis program ( to amplify trpE2, entC and entD from M. smegmatis genomic DNA and genes were flanked by 0.5–1 kb on both the ends. Primers were modified with EcoRI at the 5′-end of the primers to facilitate the subsequent ligation reaction. The genes were disrupted either by selecting appropriate restriction sites within the gene, which were not present in the vector and thereby deleting the internal gene fragment by restriction enzyme digestion, or by designing the primers in such a way that 5′- and 3′-ends of the gene were amplified so as to exclude the middle sequence of the gene. Using the two halves of the gene as a template, PCR was performed again, yielding a deleted version of the wild-type gene. The positive recombinants were selected based on kanamycin resistance and the deletion was confirmed by sequencing.

Construction of a suicide delivery vector

The two series of plasmids were used to develop a simple cloning strategy (Gordhan & Parish, 2001). The first series pNIL (p2NIL) was used for cloning and manipulating the genes. The second series pGOAL (pGOAL19) was used for generating and storing a number of marker gene cassettes (p2NIL and pGOAL19 plasmids were a kind gift from Prof. N. Stoker).

The target gene was amplified by PCR, cloned into the p2NIL vector, the required deletion was made in the gene and the construct was sequenced for confirmation. The marker cassette from plasmid pGOAL19 was cloned into p2NIL vector containing the disrupted gene. The final suicide delivery vector carrying the appropriate deleted gene was electroporated into M. smegmatis (Parish & Stoker, 1998), allowing for homologous recombination to occur.

Strategy used to generate gene knockouts in mycobacterium

To generate gene knockouts, the wild-type cells were electroporated with linearized DNA having a deleted version of the gene using an alkali lysis method (Gordhan & Parish, 2001). The transformants were then plated on Lemco agar with kanamycin (20 μg mL−1), hygromycin (50 μg mL−1) and X-gal (80 μg mL−1) and incubated at 37 °C for 3–7 days, until blue colonies appeared. These colonies, which were single crossovers (SCOs), were then streaked on Lemco agar with no antibiotics and incubated at 37 °C for 3–7 days, allowing the second crossover to occur.

Colonies from nonselective (without antibiotics) agar plates were streaked on Lemco agar plates containing 2% (w/v) sucrose and X-gal (80 μg mL−1) and incubated at 37 °C. The resulting colonies on the sucrose plates were either spontaneous sucrose-resistant (sucR) mutants (but still SCOs) or double crossovers (DCOs). The rate of spontaneous sucR colonies ranged from 10−4 to 10−5 (Gordhan & Parish, 2001). Spontaneous sucR colonies are blue because they still carry the lacZ gene, whereas any DCOs are white, having lost the lacZ marker gene along with hygromycin and sacB genes. The potential DCOs from the Lemco/sucrose/X-gal agar plate were streaked on the plates with and without kanamycin to confirm the loss of the marker gene cassette after homologous recombination.

Preparation of cell-free extracts (CFE)

Cells (both the wild type and the mutants) were grown with shaking in 100 mL minimal medium held in 250 mL conical flasks and the contents of 15 flasks were then combined and centrifuged at 10 000 g for 10 min at 4 °C. Cells were washed three times with phosphate-buffered saline (PBS) and once with 0.1 M Tris/HCl buffer (pH 8), centrifuging each time between washes at 10 000 g for 10 min. One milliliter of 0.1 M Tris/HCl buffer (pH 8) was added to the pellet to make a thick paste of cells and the cell suspension was disrupted using a One Shot Cell Disruptor. The cell debris was removed by centrifugation at 10 000 g for 10 min and the cell-free extract was recovered. The concentration of protein was estimated immediately using the biuret method with bovine serum albumin as a standard.

Synthesis of salicylic acid from chorismic acid

One milliliter CFE (8–10 mg protein) of M. smegmatis (wild type and mutants) was incubated at 37 °C for 2 h with 10 μM Mg2+, 1.5 μM NAD+, 250 μM Tris/HCl buffer at pH 8 both with and without 2 μM chorismate (Sigma) as a substrate, in a final volume of 2.3 mL (Marshall & Ratledge, 1971). The reaction was terminated by adding 0.1 mL 5 M HCl and the mixture was extracted twice with ethyl acetate (2 × 5 mL). The ethyl acetate extract was evaporated under vacuum and the residue was dissolved in 5 mL 0.1 M KH2PO4/KOH buffer, pH 7. Salicylic acid was estimated spectrofluorimetrically by its fluorescence at 410 nm following excitation at 305 nm.

Synthesis of salicylic acid from chorismic acid via isochorismic acid

One milliliter of each CFE prepared from mutants (trpE2, entC and entD), each containing approximately 10 mg protein, was incubated at 37oC for 1 h with 10 μM chorismate, 10 μM Mg2+, 1.5 μM NAD+ and 250 μM Tris/HCl buffer at pH 8 in a final volume of 2.3 mL. At the end of the reaction, the pH of each mixture was carefully adjusted to 2 using 6 M HCl and extracted twice with ethyl acetate (2 × 5 mL) to remove any isochorismic acid that had been formed. Each ethyl acetate extract was evaporated under vacuum and the residue was taken up in 3 mL 0.1 M Tris/HCl buffer, pH 8. Each suspension was then divided into three aliquots of 1 mL (yielding nine samples in total) and each aliquot was incubated with 1 mL fresh CFE (containing approximately 10 mg of protein), prepared from the other two mutants with 10 μM Mg2+, 1.5 μM NAD+ in a final volume of 2.3 mL (Table 1). The third aliquot served as a control and was incubated without CFE. After 1 h, the reaction was terminated using 0.1 mL 5 M HCl, the mixture was extracted and salicylic acid was estimated as described above.

Table 1.   Involvement of the gene products of trpE2, entC and entD in the stepwise conversion of chorismic acid to salicylic acid
Substrate ICell-free extract (CFE)Substrate IICFESalicylic acid (in ng)
  1. In the first step of reaction, with chorismic acid as a substrate, CFEs, prepared from knockout mutants of trpE2, entC and entD, were incubated at 37°C with 10 μM Mg2+, 1.5 μM NAD+ and 250 μM Tris/HCl buffer at pH 8 in a total volume of 2.3 mL and extracted for isochorismic acid. In the second step, assuming isochorismic acid to act as a substrate, either no CFE (labelled ‘None’) or CFE from entCΔ and trpE2Δ were used where entDΔ was used in the first reaction. The same logic applies to the CFE from the entC and trpE2 knockouts, allowing all different combinations of three mutants in the reaction so that their involvement could be deduced at the end of the two reactions by estimating the salicylic acid concentration as explained in Materials and methods. Δ means that the gene mentioned has been silenced, for example entCΔ means entC is silenced.

Chorismic AcidentDΔIsochorismic AcidNone0.1

Estimation of mycobactin

Mycobacterium smegmatis, grown in minimal media, was harvested by centrifugation at 10 000 g for 20 min at 4 °C and the cells were freeze-dried and weighed. The dried cells were resuspended in ethanol and left for 0.5 h at room temperature (Snow & White, 1970). The cells were filtered through Whatman filter paper No. 1 and a saturated solution of FeCl3 in absolute ethanol was added dropwise to the filtrate until there was no further color change. The resultant red solution was filtered through Whatman filter paper No. 1, an equal volume of chloroform was added to the filtrate and water was then added to generate two phases. The chloroform layer, containing the mycobactin, was removed and evaporated under vacuum. The residue was stirred with 25 mL ethanol and any ethanol-insoluble material was carefully removed. The concentration of mycobactin was estimated from its 1% A450 nm value of 43 in ethanol.


Gene knockout mutants of trpE2, entC, entD and entDtrpE2 (a double mutant) in M. smegmatis were created by targeted mutagenesis (see Materials and methods). The growth of mutants was not as good as the wild type in iron-deficient minimal medium; hence, much larger volumes of culture (1.5 L) were used to obtain sufficient cells to yield cell-free extracts (CFE) with 10 mg protein mL−1.

Salicylic acid was identified by HPLC and quantified both by HPLC and by spectrofluorimetry using appropriate controls, with 6-fluorosalicylic acid as an internal standard, to assess its efficiency of extraction and, using appropriate standards of salicylate, to quantify its response in the spectrofluorimeter. Using the conditions described, salicylate was the sole metabolite recognized by HPLC when the eluate was monitored at 296 nm.

Conversion of chorismic acid to salicylic acid

To evaluate the ability of mutants to convert chorismic acid to salicylic acid in comparison with the wild-type strain, CFE (∼10 mg protein mL−1) of the mutants and the wild type were incubated with and without chorismic acid at 37 °C and salicylic acid was extracted. Using CFE prepared from wild-type M. smegmatis, grown iron deficiently on minimal medium, 550 ng salicylic acid was formed per 10 mg protein over 2 h. Salicylic acid, however, was not synthesized to any appreciable extent from chorismic acid by extracts prepared from any of the mutants grown similarly: ∼1 ng by CFEs from knockouts of trpE2, entC and entD as single mutants and 0.25 ng by entDtrpE2 as a double mutant (Fig. 1). These very low conversions suggest that a combination of the gene products from trpE2, entC and entD or, probably and more likely, that all three genes play a role in the synthesis of salicylic acid from chorismic acid.

Figure 1.

 Comparison of the direct conversion (single reaction) of chorismic acid to salicylic acid using CFE from wild-type Mycobacterium smegmatis and from its knockout mutants of salicylic acid pathway genes (trpE2, entC, entD and entDtrpE2). Salicylic acid (SA) concentration was estimated by spectrofluorimetry and the values represented in ng are the average of two independent experiments.

To evaluate which genes are involved in the conversion of chorismic acid to isochorismic acid and then in the conversion of isochorismic acid to salicylic acid, the above experiment was modified such that isochorismic acid was extracted before estimating salicylic acid and hence could confirm the involvement of trpE2, entC and entD in the stepwise conversion. Accordingly, the CFEs of each of the three single mutants were prepared. Each contained approximately 10 mg protein mL−1 and were incubated individually with chorismic acid as a substrate at 37 °C in a total volume of 2.3 mL (Marshall & Ratledge, 1971). After 1 h, the reaction was stopped with HCl and each mixture was extracted with ethyl acetate (see Materials and methods) to remove any isochorismic acid that had been formed. Each of these solvent extracts, now in an aqueous buffer, was then divided into three equal aliquots and each of these was placed in separate test tubes. For each batch of three solvent extracts, one was incubated without addition of CFE (control), and the other two were incubated with a CFE other than the one that had been used originally (Table 1). In other words, this was a cross-over biochemical reaction.

The synthesis of salicylic acid occurred when CFEs from mutants of either entC or entD were used in the first reaction with chorismic acid as a substrate and followed by using the CFE of mutant trpE2 in the second reaction. The synthesis of salicylic acid was completely absent when a CFE of mutant trpE2 was used in the first reaction, irrespective of which CFE was used in the second reaction (Table 1).

Production of mycobactin

As salicylic acid is principally converted to mycobactin, with only about 5–10% being converted into carboxymycobactin (Ratledge & Ewing, 1996), we then studied the production of mycobactin in the knockout mutants. The wild type and the mutants of M. smegmatis were grown for 7 days in minimal medium under iron-deficient conditions (which are needed to maximize mycobactin formation) with and without salicylic acid added at 5 μg mL−1. The production of mycobactin by the mutants was drastically decreased in minimal medium compared with the wild-type strain (Fig. 2). However, when salicylic acid was included in the medium, the mutant cells had considerably more mycobactin than before, although the amounts were well below those in the wild-type strain (Table 2). Mycobactin reached up to 20% of the wild type by the entD and trpE2 knockout mutants, 14% by the entC mutant and 10% by the double knockout, entDtrpE2.

Figure 2.

 Comparison of mycobactin production in whole cells of wild-type Mycobacterium smegmatis with its knockout mutants (trpE2, entC, entD and entDtrpE2) grown for 7 dyas in minimal medium (MM) with and without salicylic acid (SA) (5 μg mL−1) under iron deficiency (0.01 μg mL−1). Mycobactin values represented in mg are the average of three independent experiments rounded off to the nearest number.

Table 2.   Production of mycobactin (mg g−1 cell weight) by wild type and mutants of Mycobacterium smegmatis grown in minimal medium (MM) & minimal medium with 5 μg salicylic acid (SA) mL−1
Strain/mutantMM (%)MM+SA (%)
  1. The values in parentheses are the percentages of mycobactin production by mutants in comparison with the wild-type Mycobacterium smegmatis. The concentration of mycobactin was estimated from its 1% A450 nm value of 43 in ethanol, and values are in mg g−1 cell dry weight.

mc215520 (100)21 (100)
trpE2Δ0.81 (4)3.8 (18)
entCΔ0.76 (3.8)2.9 (14)
entDΔ0.85 (4)4.5 (22)
(entDtrpE20.67 (3.5)2.3 (11)

Growth supplementation of mutants

Salicylic acid restored the growth of trpE2, entC, entD and (entDtrpE2) mutants, but only to a limited extent when added up to 5 μg mL−1 in the medium. Hence, the mutants are not strict auxotrophs of salicylic acid, but this may be because the deleted proteins also have an (unproven) involvement in the conversion of salicylic acid into both mycobactin and carboxymycobactin. Interestingly, although neither mycobactin nor carboxymycobactin individually restored the growth of the knockout mutants, they did so together (Fig. 3). This suggests that carboxymycobactin may be more important in iron metabolism than hitherto considered in spite of it being a minor siderophore in this organism (Ratledge & Ewing, 1996). The results also indicate that mycobactin is not converted to carboxymycobactin and vice versa as then there would have been no enhancement of growth when both siderophores were added together.

Figure 3.

 Growth of wild type and mutants of Mycobacterium smegmatis grown iron deficiently (0.01 μg mL−1) in minimal medium (MM) with salicylic acid (SA), carboxymycobactin (CMy), mycobactin (My), salicylic acid and carboxymycobactin, salicylic acid and mycobactin and carboxymycobactin & mycobactin (5 μg mL−1of each). Mycobactin and carboxymycobactin used were extracted and purified from M. smegmatis NCIMB 8548 (Ratledge & Ewing, 1996). The cell dry weights were calculated from the OD600 nm value (an OD value of 1=0.83 mg dry wt mL−1) after 7 days growth from 10 mL culture and are the average of three independent experiments.


In M. smegmatis, salicylic acid is produced from the shikimic acid pathway via chorismic and isochorismic acids (Marshall & Ratledge, 1972). In P. aeruginosa, genetic and experimental evidences indicate that pchA and pchB genes encode ICS and isochorismate pyruvate-lyase, respectively, catalyzing in turn the conversion of chorismate to isochorismate and then isochorismate to pyruvate plus salicylate for the biosynthesis of pyochelin (Serino et al., 1995; Gaille et al., 2002). When the purified ICS from P. aeruginosa was examined for salicylate synthesis, there was no reaction in vitro (Gaille et al., 2003); additionally, in vivo, PchA did not display salicylate synthase activity. An entC mutant of E. coli carrying only the pchA gene also failed to produce salicylate, but when the same mutant had both pchA and pchB genes, salicylate synthesis took place (Serino et al., 1995). Hence, organisms that have no PchB protein homolog can carry out the direct conversion of chorismate to salicylate, for example MbtI of M. tuberculosis, Irp-9 of Y. enterocolitica and YbtS of Y. pestis (Gehring et al., 1998; Quadri et al., 1998). This proposition was supported by studies where native and purified protein MbtI from M. tuberculosis was shown, not to function as ICS like PchA, but instead acted as a salicylate synthase like Irp-9 (Harrison et al., 2006). In Yersinia spp., which again synthesizes salicylic acid for the production of yersiniabactin, the conversion of chorismic acid to isochorismic acid and then to salicylic acid is by a single gene product acting as a bifunctional salicylate synthase (Kerbarh et al., 2005) as was the case in M. tuberculosis (Harrison et al., 2006).

To elucidate genes for salicylate biosynthesis in M. smegmatis, we generated knockout mutants of the likely key genes trpE2, entC and entD by targeted mutagenesis. From the enzymatic analysis of salicylic acid biosynthesis by CFEs from the various mutants of M. smegmatis, the participation of entC and entD for the conversion of isochorismic acid to salicylic acid is now evident. As a corollary, these genes are not involved in the formation of isochorismic acid from chorismic acid. In addition, we have shown that trpE2 is involved in the conversion of chorismic acid to isochorismic acid (Table 1). The gene product of trpE2 thus corresponds to ICS and would be equivalent of PchA in P. aeruginosa (Gaille et al., 2002; Kunzler et al., 2005).

In this study, the targeted mutagenesis has elucidated the roles of trpE2, entC and entD genes in the conversion of salicylic acid from chorismic acid. Hence, salicylic acid seems to have only one function, although its involvement in the recognition of iron and its transfer cannot be ruled out completely. However, since we observed the salicylate nonauxotrophy of the mutants, the most viable explanation for this is that the gene products of salicylate biosynthesis interact with other proteins of the mycobactin pathway, making the conversion of salicylate to mycobactin and carboxymycobactin less efficient. The addition of salicylate (which cannot be converted to mycobactin and carboxymycobactin) at higher concentrations, over 5 μg mL−1, in the medium makes it toxic for the mutants, although the mechanism for this toxicity is not understood.

With these studies, we suggest that the organization of salicylate biosynthesis is different between M. smegmatis (current study) and M. tuberculosis (Harrison et al., 2006). Distinct from mbtI of M. tuberculosis, but in common with pchA of P. aeruginosa, trpE2 is coding for ICS in M. smegmatis. Hence, the conversion of chorismate to salicylic acid in M. smegmatis involves a multienzyme complex consisting of trpE2, entC and entD genes. Taken together, these data conclude that in M. smegmatis, the gene product of trpE2 corresponds to ICS; entC and entD code for salicylate synthase.


We thank Overseas Research Studentships (UK) for a research studentship to N.N. We are indebted to Prof. Neil Stoker (Royal Veterinary College, London) for his invaluable suggestions in creating knockout mutants and generously gifting p2NIL and pGOAL19 vectors.