Mycobacterial polyketide-associated proteins are acyltransferases: Proof of principle with Mycobacterium tuberculosis PapA5

Abstract

Mycobacterium tuberculosis (Mt) produces complex virulence-enhancing lipids with scaffolds consisting of phthiocerol and phthiodiolone dimycocerosate esters (PDIMs). Sequence analysis suggested that PapA5, a so-called polyketide-associated protein (Pap) encoded in the PDIM synthesis gene cluster, as well as PapA5 homologs found in Mt and other species, are a subfamily of acyltransferases. Studies with recombinant protein confirmed that PapA5 is an acetyltransferase. Deletion analysis in Mt demonstrated that papA5 is required for PDIM synthesis. We propose that PapA5 catalyzes diesterification of phthiocerol and phthiodiolone with mycocerosate. These studies present the functional characterization of a Pap and permit inferences regarding roles of other Paps in the synthesis of complex lipids, including the antibiotic rifamycin.

Polyketides (PKs) are a family of complex lipids (1). Several secreted PKs and PK-containing compounds act as offensive, defensive, or adaptation effectors that contribute to mycobacterial virulence. These compounds include (phenol)phthiocerol (POL) and phthiodiolone (PONE) dimycocerosate esters (PDIMs) (2, 3) (Scheme 1, which is published as supporting information on the PNAS web site). Mycobacterium tuberculosis (Mt) mutants deficient in PDIM production are attenuated in mice, and PDIMs of Mycobacterium leprae (phenolic glycolipids or mycoside B) promote Schwann cell tropism (2-4).

The global burden of tuberculosis underscores the need to understand the biosynthesis of PDIMs and other Mt virulence-enhancing PKs. The Mt H37Rv genome contains 24 PK synthase genes (pks genes) organized in several loci (5, 6), some of which are relevant for virulence (2, 3, 7-13). Various genes associated with these loci encode proteins of unknown functions (5, 6). Among these are five so-called PK-associated protein (Pap) genes presumed to be involved in PK synthesis. Genes encoding Mt Pap homologs occur in M. leprae and Mycobacterium bovis, where they are also linked to pks gene-containing loci (14, 15).

We selected PapA5, a suspected membrane-associated protein encoded in the Mt PDIM synthesis gene cluster (2, 3, 12, 13, 16-18), as a representative Pap for functional characterization. Current models for PDIM synthesis propose involvement of at least three pks gene systems: (i) ppsA-E, required for POL and PONE synthesis (12); (ii) mas, required for mycocerosic (MYC) acid synthesis (12), and (iii) pks15/1, required for incorporation of the phenolic group in mycoside PDIM variants (18). Recent evidence suggests that additional pks genes are involved in PDIM synthesis (8, 10, 19). Despite progress made since early studies on PDIM synthesis described 40 years ago (20, 21), the enzyme(s) catalyzing diesterification of POL and PONE with MYC acid remains unknown. In this article, we demonstrate that Mt PapA5 is an acyltransferase (AT) required for PDIM synthesis. We propose that PapA5 catalyzes diesterification during PDIM synthesis, and that Paps are a subfamily of ATs prevalent in mycobacteria and are primarily involved in PK synthesis.

Materials and Methods Strains, Growth Conditions, and Reagents. Escherichia coli was grown in LB media (22). Media were supplemented with kanamycin (30 μg/ml) for strains with plasmids pSMT3, p2NIL, pMV261, or their derivatives or ampicillin (100 μg/ml) for strains with pGOAL. Mt Erdman was grown in Middlebrook media supplemented with 10% oleate-acbumin-dextrose-catalase (Becton Dickinson) (23) and, when needed, with antibiotics and/or sucrose as reported (24). Radiolabeled compounds were from American Radiolabeled Chemicals (St. Louis) or Sigma. Chloramphenicol acetyl transferase (CAT) was from Sigma and BSA from NEB. Other reagents were from Invitrogen, NEB, Novagen, or Sigma.

Production of Recombinant Proteins. PapA5 and four mutated variants were produced as isopropylthiogalactoside (IPTG)-inducible N-terminal His-6-Smt3 fusions (25) in E. coli BL21(DE3) CodonPlus (Stratagene) by using TOPO-adapted pET-based Smt3 expression vector (Invitrogen). The PapA5 coding region was PCR amplified with primers 5xpf (ATGTTTCCCGGATCTGTGATCCGAAA) and 5xpr (CACCTCACTCCATGATCCAGCCATACTCCGA) from Mt genomic DNA. PapA5 mutant alleles were engineered by using QuickChange Mutagenesis Kit (Stratagene). Primer pairs 123Af (CTGACGCTATACCTCGCCCACTGCATGGCCGATGGTCATCA) and 123Ar (ATCGGCCATGCAGTGGGCGAGGTATAGCGTCAGCTCGGCT), 124Af (CTGACGCTATACCTCCATGCCTGCATGGCCGATGGTCATCA) and 124Ar (ATCGGCCATGCAGGCATGGAGGTATAGCGTCAGCTCGGCT), 128Af (CATCACTGCATGGCCGCTGGTCATCACGGGGCCGTTCTCGT) and 128Ar (GGCCCCGTGATGACCAGCGGCCATGCAGTGATGGAGGTATAGCGT), and 143Ff (GAGCTGTTCTCCCGCTTCACCGACGCGGTCACTACCGGTGA) and 143Fr (AGTGACCGCGTCGGTGAAGCGGGAGAACAGCTCGTCGA) were used to generate plasmids expressing PapA5 variants PapH123A, PapH124A, PapD128A, and PapY143F, respectively. DNA manipulations followed standard methods (22). Cultures (5 liters) were grown by fermentation at 37°C to an OD₆₀₀ of 2, adjusted to 1 mM IPTG, and incubated for 4 h at 30°C. Cells were harvested and resuspended in 20 mM Tris·HCl (pH 8)/350 mM NaCl/10 mM imidazole/20% sucrose/1 mM β-mercaptoethanol/20 μg/ml lysozyme. Suspensions were sonicated, and insoluble material was removed by centrifugation. His-6-tagged proteins were purified by metal-affinity chromatography and gel filtration (Pharmacia; Superdex75). His-6-Smt3 tags were removed by digestion with the Smt3-specific protease Ulp1, leaving a nonnative Ser at the N terminus of each protein. Tag-free proteins were purified by gel filtration (Superdex75) with yields of ≈10 mg/liter of culture, concentrated to 8 mg/ml in 10 mM Tris·HCl, pH 8/100 mM NaCl/1 mM DTT, and stored at -80°C. His-6-tagged Pseudomonas aeruginosa arylation enzyme PchD and His-6-tagged Mt synthetase MbtB ArCP domain were purified as reported (26, 27).

Mt ΔpapA5 Construction. Deletion of papA5 in Mt Erdman was made with the p2NIL/pGOAL-based method (24). The papA5 deletion cassette, including a 500-bp segment upstream of papA5, papA5 start and stop codons, and a 495-bp segment downstream of papA5, was generated with gene splicing by overlap extension PCR (28). Primer pairs 5ufx (TCGACAGGTACCCTCTCGAATCCTGGCAAACGCGATTCGGAC) and 5urx (TGGTTCGTTAGGTCACATCTCAACACTCCCAACTCGTCACGGCAT) and 5dfx (GGGACTGTTGAGATGTGACCTAACGAACCAGCCCGCCGATCGGGC) and 5drx (TGTCGATTAATTAATGTTCGCATCCACTGGGCAGCGGTCAAGGG) were used to amplify the 5′ and 3′ ≈500-bp papA5 flanking regions, respectively, from genomic DNA. The resulting PCR products were used as template in a PCR with primers 5ufx and 5drx to generate the cassette.

Mt ΔpapA5 Complementation. Complementation was done with pMV261-based plasmids (29) expressing papA5 or its mutant alleles from the hsp60 promoter. A fragment with papA5 flanked by 18 and 9 bp at the 5′- and 3′-ends, respectively, was PCR-amplified with primers 5cf (GAATTCCGAGTTGGGAGTGTTGAGATGTTT) and 5cr (GCTAGCTTCGTTAGGTCACTCCATGATCCA) from genomic DNA. Same-size fragments for mutant alleles papH₁₂₄A and papY₁₄₃F were generated by PCR and splicing by overlap extension (28). Fragment papH₁₂₄A was generated with primer pairs 5cf and 124Ar and 124Af and 5cr. Primer pairs 5cf and 143Fr and 143Ff and 5cr were used to generate fragment papY₁₄₃F. Fragments were cloned into pCR2.1TOPO (Invitrogen), recovered as EcoRI-NheI fragments, and subcloned into EcoRI-NheI digested pMV261. The resulting plasmids, pCPapA5 (expressing papA5), pCPH124 (expressing papH₁₂₄A), pCPY143 (expressing papY₁₄₃F), and pMV261 were introduced in Mt by electroporation (23).

AT Assay. Unless otherwise indicated, AT reactions (25 μl) contained 75 mM Mes, pH 6.5, 100 mM NaCl, enzyme, and substrates (acyl-CoA thioester/alcohol pairs, one ¹⁴C-labeled per pair) at indicated concentrations. After incubation at 37°C for the indicated periods, reactions were stopped with CHCl₃ (50 μl) and centrifuged to separate aqueous and CHCl₃ layers. CHCl₃ layers were analyzed (5 μl) by TLC on aluminum-backed 250-μm-thick silica gel plates (EM Science) with 3:1 hexane/ethyl acetate. Plates were exposed (48 h) to a phosphor screen, and products were quantified with a Storm 860 Imaging System (Molecular Dynamics).

Lipid Analysis. Mt apolar lipids were extracted as reported (30). Cells from Mt cultures (100 ml, OD₆₀₀ = 1) were harvested and resuspended in 20 ml of 10:1 MeOH/0.3% NaCl. Suspensions were extracted twice with 10 ml of petroleum ether (PE). Upper layers were recovered, mixed with 20 ml of CHCl₃, and dried. Dried extracts were suspended in 1:1 PE/CHCl₃ and subjected to TLC for PDIM analysis (31). Extracts were resolved by 1D TLC with PE/ethyl ether (9:1) or 2D TLC with PE/ethyl acetate (98:2) in one direction and PE/acetone (98:2) in a second direction, and stained with 5% ethanolic phosphomolybdic acid. ¹⁴C-labeled PDIMs were obtained as reported (3, 13). Mt cultures (100 ml, OD₆₀₀ = 0.8) were pulsed with ¹⁴C-propionate [10 μCi, 50 μCi/mmol (1 Ci = 37 GBq)] for 24 h before lipid extraction and TLC analysis of labeled lipids as above.

Mass Spectrometry. Compounds from CHCl₃ extracts of AT reactions were recovered by preparative TLC with CHCl₃/MeOH (2:1) and analyzed by electrospray ionization MS (ESI-MS) at the Sloan-Kettering Institute Organic Chemistry Facility (New York). PDIMs were recovered by preparative 1D TLC of Mt apolar lipid extracts with CHCl₃/MeOH (2:1) and analyzed by atmospheric pressure photoionization MS at the Hunter College MS Facility (New York).

Results

Sequence Analysis of Paps. Mt Paps A1-A5 are conserved proteins of unknown function (5). Homologs of each of these Paps were found in M. bovis (15), and Pap A3 and A5 homologs were identified in M. leprae (14). We performed protein-protein blast (blastp) and translated database (tblastn) similarity searches with default parameters via www.ncbi.nlm.nih.gov/blast by using Mt Paps A1, A2, A3, and A5 as queries (excluding N-terminally truncated A4). Analysis of blastp hits indicated that the only consistent similarity of Paps to proteins of known function was to condensation domains (CDs) of peptide synthetases (data not shown). Searches for conserved domains via www.ncbi.nlm.nih.gov/blast showed a possible CD motif (conserved domain database: pfam00668.8) in each Pap (scores 49-60, E values 10^-7 to 10^-10). CDs (≈450 amino acids) catalyze peptide bond formation during nonribosomal peptide synthesis (32-34) and contain a Hx₃Dx₁₄Y motif, which is also found in ATs (35).

Sequence alignment revealed a Hx₃Dx₁₄Y motif in the Mt, M. bovis, and M. leprae Paps noted above and in 15 Pap homologs identified with tblastn searches (Fig. 1). Pap homologs were identified in every mycobacterial genome examined, and only two homologs were found in nonmycobacterial genomes, one in Amycolatopsis mediterraeni S699 and another in Streptomyces coelicolor A3 (2) (Fig. 1). A possible CD motif was found in each mycobacterial Pap and in S. coelicolor Pap (scores 45-72, E values 10^-5 to 10^-13). Analysis of genome regions encoding Pap homologs indicated that, except in S. coelicolor, Pap genes were in close proximity to pks or lipid metabolism genes (data not shown). These observations suggest that Paps may be a subfamily of ATs involved in lipid synthesis.

Fig. 1.

Conservation of Hx₃Dx₁₄Y motif. Protein names are based on similarity to Mt Paps, except for Rif20 (with highest similarity to PapA5). Numbers in front of each protein segment indicate position of the conserved His. Total number of residues of each protein is indicated. Percent similarity of full-length Paps with their closest Mt Pap homolog is shown. Sequence alignment and similarity values were obtained with Clustal method (49). Am, A. mediterranei S699; Ma, Mycobacterium avium; Mp, M. avium subsp. paratuberculosis; Mb, M. bovis; Ml, M. leprae; Mm, Mycobacterium marinum; Ms, Mycobacterium smegmatis; Mt, M. tuberculosis; and Sc, S. coelicolor A3 (2). Two additional Paps from Mm are not shown.

Demonstration of PapA5 AT Activity. We selected Mt PapA5 as a representative Pap to assess the suspected AT activity. As noted, PapA5 is encoded in the PDIM gene cluster. PDIMs are synthesized via diesterification of POL and PONE with MYC acid (donated as an unknown intermediate, possibly a CoA- or Mas synthase-bound thioester) by an as-yet-unidentified AT. Sequence analysis suggests that PapA5 could be such an AT. We developed a TLC-based assay to investigate PapA5 AT activity using ¹⁴C-labeled surrogate substrates (i.e., aliphatic alcohols and acyl-CoA thioesters) and recombinant PapA5 (Fig. 2). We reasoned that in this reaction setup, alcohols would replace the POL and PONE nucleophiles, whereas CoA thioesters would substitute for the mycocerosyl thioester.

Fig. 2.

Purified PapA5 and PapA5 variants. Coomassie blue-stained SDS/PAGE (12%) of purified PapA5 (lane 2), PapH₁₂₃A (lane 3), PapH₁₂₄A (lane 4), PapD₁₂₈A (lane 5), and PapY₁₄₃F (lane 6) and molecular marker (lane 1).

TLC analysis of reactions with PapA5, 1-octanol (1-OCL), and ¹⁴C-palmitoyl-CoA (PCoA) showed two products (Fig. 3A). The lower R_f product migrated like palmitate (data not shown), and ESI-MS confirmed its identity (m/z 256, Fig. 3B). Palmitate was also formed, although in variable quantities, in reactions without alcohol or protein, in reactions with control proteins (Fig. 3A), and in reactions (data not shown) with mutated PapA5 variants described below. However, PCoA presence was needed for palmitate formation (Fig. 3A). These results suggest that palmitate is formed by nonspecific hydrolysis of PCoA. In contrast, the higher R_f product was formed only in reactions with PapA5 and both substrates (Fig. 3A). ESI-MS revealed a m/z [M+Na]⁺ of 391.2, in accordance with the mass of octyl-palmitate sodium adduct positive ion (Fig. 3C). PapA5 also catalyzed ester formation with [¹⁴C]hexadecanol (Fig. 3A). In this case, the higher R_f product comigrated with hexadecyl-palmitate (data not shown).

Fig. 3.

AT assays revealing PapA5 activity. (A) TLC analysis of AT reactions. Reactions with 18 μM[¹⁴C]PCoA, 180 μM 1-OCL, 75 mM Mes (pH 6.5) 100 mM NaCl, and PapA5 (1), no protein (2), BSA (3), His-6-tagged P. aeruginosa arylation enzyme PchD (4), His-6-tagged Mt synthetase MbtB ArCP domain (5), UlpI (6), or CAT (7) were incubated for 8 h. Analysis of reactions with PapA5 and [¹⁴C]PCoA but lacking 1-OCL (8), PapA5 and [¹⁴C]hexadecanol but lacking PCoA (9) and PapA5 [¹⁴C]PCoA and hexadecanol (10) is shown. Products octyl-palmitate (OP) in section 1, hexadecyl-palmitate (HP) in section 10, palmitate (PA) in all sections except 9, where the spot is unreacted [¹⁴C]hexadecanol, and TLC origin (Or) are marked. Assays were done in triplicate. Only duplicates are shown in sections 8-10. (B) ESI-MS of low R_f product from section 1 showing PA mass (m/z 256-H). (C) ESI-MS of high R_f products from section 1 showing OP mass (m/z 368 + Na).

Ester formation was pH-dependent (Fig. 4A), with an optimum at pH 6.5. As expected, palmitate accumulation resulting from PCoA hydrolysis increased at basic pH. Ester formation was also time-dependent, reaching a plateau with 40% conversion stoichiometry relative to PCoA after 8 h. No significant palmitate accumulation was observed during a 46-h reaction course (Fig. 4A).

Fig. 4.

PapA5-catalyzed octyl-palmitate formation is time- and pH-dependent. (A) Time course of product formation. (B) pH effect on product formation. Formation of octyl-palmitate (solid circles) and palmitate (open circles) is expressed as percentage of [¹⁴C]PCoA initial concentration. Reactions contained 2 μM PapA5, 18 μM[¹⁴C]PCoA, 180 μM 1-OCL, 100 mM NaCl, and 75 mM buffer (Mes, pH 6.5, for the time course and Mes, pH 6.0 or 6.5, Hepes, pH 7.0, and Tris·HCl, pH 7.5, 8.0, or 8.5, for pH-dependence analysis). Reactions were incubated for 12 h for pH-dependence analysis. Average of duplicates ± SE is shown.

PapA5 Substrate Selectivity. To gain further insight into PapA5 catalytic competence, we explored its ability to use several acyl-CoA thioesters and nucleophiles and calculated preliminary kinetic parameters with the best substrate pair. Pairs of [¹⁴C]PCoA and one of 36 nucleophiles were tested. The nucleophiles included short-, medium-, and long-chain alcohols, diols, hydroxy esters, and acids, amines, and thiols. Conversely, each of 17 acyl-CoA thioesters from a panel with short, medium, and long acyl chains was assessed in pair with [¹⁴C]hexadecanol.

Ester formation was detected with 28 of 36 nucleophiles tested (Table 1). Preference for saturated medium chain alcohols was observed, with 1-OCL displaying the highest ester formation stoichiometry (50-60% relative to [¹⁴C]PCoA). PapA5 discriminated between enantiomers; 20- and 5-fold more ester formed with (R)-(-)2-octanol and (S)-(-)1,2-decanediol, respectively, than with their corresponding enantiomers. Preference for neutral nucleophiles over closely related negatively charged ones was also observed (e.g., 3-hydroxypalmitic acid methyl ester over 3-hydroxypalmitic acid and 2-octanol or 1,2-octanediol over 2-hydroxyoctanoic acid). Although 1-OCL was the preferred nucleophile, nonhydroxy nucleophiles of comparable chain length, such as octylamine and octanethiol, were not appreciably used.

Table 1. Substrate specificity of PapA5.

Compound tested*	Product formation, %†
Nucleophile
1-OCL	100.00 ± 13.08
1-Decanol	78.50 ± 6.70
1-Nonanol	43.39 ± 7.94
(±)2-Octanol (mix)	36.86 ± 3.61
(S)-(−)-1, 2-decanediol‡	me/de 25.28 ± 5.79/1.92 ± 0.55
(R)-(−)-2-octanol	20.48 ± 3.71
Myristyl alcohol	18.19 ± 3.56
(±)2-Undecanol	12.00 ± 2.65
1-Hexadecanol	9.10 ± 1.82
1-Eicosanol	6.06 ± 1.59
Stearyl alcohol	5.88 ± 0.58
(R)-1, 2-decanediol	me/de 5.61 ± 1.83/2.80 ± 0.50
Trans 2-octene-1-ol	4.93 ± 2.77
DL-3-OH-methyl palmitate	4.40 ± 1.66
(±)1,2-Octanediol	me/de 4.09 ± 3.21/3.93 ± 2.19
1-Pentanol	1.04 ± 0.21
(S)-(+)-2-octanol	0.99 ± 0.30
(±)1, 2-Hexanediol	me/de 0.51 ± 0.19/ND
Dihydroxyacetone	0.39 ± 0.16
Glycerol	0.35 ± 0.08
3,5-Dihydroxytoluene	0.32 ± 0.28
1-Hexacosanol	0.22 ± 0.18
1-Octylamine	0.21 ± 0.07
1,2,6-Hexanetriol	me/de/te 0.20 ± 0.18/ND/ND
Ethanol	0.19 ± 0.17
DL-3-OH-palmitic acid	0.19 ± 0.02
(±)2-OH-octanoic acid	0.13 ± 0.10
(R)-(−)-1,3-butanediol, (S)-(+)-1,3-butanediol, 2-ethyl hexane-1,3-diol (mix), 1,3-cyclohexanediol (mix), Isopropanol, Methanol, (R)-(−)-1,3-nonanediol, 1-octanethiol, (2R, 4R)-(−)-pentanediol, (2S,4S)-(+)-pentanediol, D-(+)-trehalose	ND
Acyl-CoA
Palmitoyl-CoA (C16)	100.00 ± 24.16
Elaidoyl-CoA (C18)	81.37 ± 17.10
Oleoyl-CoA (C18)	74.63 ± 19.04
Heptadecanoyl-CoA (C17)	52.11 ± 36.99
Lauroyl-CoA (C12)	38.94 ± 17.09
Stearoyl-CoA (C18)	38.30 ± 19.81
Acetyl-CoA (C2), Methylmalonyl-CoA (C3), Isovaleryl-CoA (C4), Hexanoyl-CoA (C6), Octanoyl-CoA (C8), Decanoyl-CoA (C10), Myristoyl-CoA (C14), Arachindoyl-CoA (C20), Docosanoyl-CoA (C22), Hexacosanoyl-CoA (C26), Benzoyl-CoA	ND

ND, not detected. Values are average of duplicates ± SE.

^*Nucleophiles (180 μM) were tested in combination with [¹⁴C]PCoA (18 μM). Acyl-CoA thioesters (18 μM) were tested in combination with [¹⁴C]hexadecanol (180 μM).

^†Relative product formation of 100% in the nucleophile and acyl-CoA panels corresponds to 270 and 240 picomoles of product, respectively (maximum yields observed with compounds of the respective panels).

^‡Formation of monoester (me), diester (de), and triester (te) is indicated. Reactions were incubated for 12 h.

PapA5 used various acyl-CoA thioesters (Table 1). Ester formation was detected with 6 of 17 thioesters tested and was more efficient with long chain acyl thioesters. The highest yield (50-60%) was seen with PCoA. Monounsaturated C18 elaidoyl- and oleoyl-CoA were more efficiently used than saturated C18 stearoyl-CoA. Lauroyl-CoA (C12) was readily used, whereas no ester was detected with myrisotyl-CoA (C14). This is perhaps due to PapA5 inhibition by the latter compound.

Determination of kinetic parameters for ester formation with 1-OCL and PCoA, the preferred substrates among those tested, was complicated by substrate inhibition (Fig. 5). Inhibition was observed with both substrates but was particularly severe with PCoA. Inhibition was detected with other substrates as well (data not shown). Fitting of the equation v = V_max/[1 + K_m/S + S/K_i] permitted calculation of apparent kinetic parameters and K_i values. K_cat, K_m, and K_i values of 0.022 min^-1, 500 μM, and 3 mM, respectively, were obtained for 1-OCL. K_cat, K_m, and K_i values of 0.027 min^-1, 4 μM, and 9.6 μM, respectively, were derived for PCoA.

Fig. 5.

Effect of 1-OCL and PCoA concentration on the rate of PapA5-catalyzed octyl-palmitate formation. (A) Reaction velocity as a function of 1-OCL concentration with [¹⁴C]PCoA at 18 μM. (B) Reaction velocity as a function of [¹⁴C]PCoA concentration with 1-OCL at 1 mM. Reactions contained substrates, 0.2 μM PapA5, 100 mM NaCl, and 75 mM Mes, pH 6.5, and were incubated for 3 h. Means of triplicates ± SEM are shown.

PapA5 HHx₃DGx₁₃Y Motif Mutagenesis. Alignment of 25 Paps revealed a fully conserved Hx₃Dx₁₄Y motif and an extended HHx₃DGx₁₃Y motif in PapA5 and its close homologs (Fig. 1). Both motifs are present in ATs, where the fully conserved His and Asp residues are proposed to act as a catalytic base and an active site conformation stabilizer, respectively (36-38). We analyzed the activity of PapA5 variants PapH124A and PapD128A, with Ala substitutions in H124 (second His in the motif) and D128, respectively. We also examined the functional relevance of motif residues H123 and Y143 by investigating the activity of variants PapH₁₂₃A (Ala substituted) and PapY₁₄₃F (Phe substituted). Mutated variants were purified in the same way as PapA5 (Fig. 2). Representative results of AT assays with PapA5 variants are shown in Fig. 6. No ester formation was detected in reactions with 0.2 μM PapH₁₂₄A or PapD₁₂₈A after 3 h, whereas significant ester formed in reactions with PapA5 under the same conditions. Increasing variant concentration to 2 μM and extending incubation to 24 h resulted in only marginal ester accumulation, corresponding to 65- and 76-fold reduction for PapD₁₂₈A and PapH₁₂₄A, respectively, compared with PapA5. In contrast, the ester formed in reactions with PapH₁₂₃A and PapY₁₄₃F was only moderately reduced relative to those with PapA5 (1.7- to 2.3-fold for PapH₁₂₃A and ≤1.9-fold for PapY₁₄₃F), thus suggesting that H123 and Y143 have neither catalytic nor structural critical roles.

Fig. 6.

Activity of PapA5 mutant variants. TLC analysis showing ester formed in reactions with 18 μM[¹⁴C]PCoA/1 mM 1-OCL/75 mM Mes, pH 6.5/100 mM NaCl/0.2 μM enzyme incubated 3 h (Upper) or 2 μM enzyme incubated 24 h (Lower). Picomoles of ester are shown as mean of triplicates ± SEM. ND, not detected.

Effect of papA5 Deletion on PDIM Synthesis.To investigate the involvement of PapA5 in PDIM synthesis, we constructed a ΔpapA5 mutant by creating a deletion leaving only the start and stop papA5 codons. The deletion was confirmed by PCR (Fig. 7A). Gene papA5 (1.2 kb) was PCR amplified with gene-specific primers only from WT Mt, and PCR with priming outside papA5 gave the expected 2.2- and 1-kb products for WT and ΔpapA5 Mt, respectively. Southern analysis with two independent probes also confirmed the deletion (data not shown).

Fig. 7.

Deletion of papA5 and lipid profiling. (A) Confirmation of papA5 deletion. Lanes 1 and 6, molecular marker. Lanes 2 and 3, PCR products with primers 5xpf and 5xpr from genomic DNA of WT and ΔpapA5 Mt, respectively. Lanes 4 and 5, PCR products with primers 5ufx and 5urx from genomic DNA of WT and ΔpapA5 Mt, respectively. (B) 1D TLC of ¹⁴C-labeled apolar lipids of WT Mt (lane 1), ΔpapA5 (lane 2), ΔpapA5 with pMV261 (lane 3), ΔpapA5 with pCPapA5 (lane 4), ΔpapA5 with pCPH124 (lane 5), and ΔpapA5 with pCPY143 (lane 6). Spots I and II are POL- and PONE-dimycocerosate, respectively. TLC origin (Or) is marked. (C) 2D TLC of ¹⁴C-labeled apolar lipids of WT and ΔpapA5 Mt. PDIM spots and origin are labeled as in B. (D) Atmospheric pressure photoionization MS of total PDIMs corresponding to TLC spots missing in ΔpapA5 Mt. POL monoester ion (C₆₃H₁₂₅O₃⁺) expected mass: 929.97 ± 14(n+m). PONE monoester ion (C₆₂H₁₂₁O₃⁺) expected mass: 913.94 ± 14(n+m). (Inset) Ionization of POL (R, -OCH₃) and PONE (R, =O) esters. l and m = 1-7, n = 0-2.

Apolar lipid analysis of Mt WT and several ΔpapA5 isolates by 1D and 2D TLC showed that the deletion abrogated PDIM synthesis (Fig. 7 B and C). Identity of PDIM spots corresponding to POL- and PONE-dimycocerosate was confirmed by atmospheric pressure photoionization (APPI) MS of compounds purified from TLC plates of WT Mt (Fig. 7D). APPI MS showed the characteristic PDIM mass pattern, with 14-atomic mass unit periodicity from chain length heterogeneity (31, 39). TLC analysis of polar lipid showed no apparent differences between WT and ΔpapA5 Mt (data not shown). To confirm that the deletion was responsible for PDIM deficiency, Mt ΔpapA5 was transformed with pCPapA5, expressing papA5. Lipid analysis showed that the transformant produced PDIMs at WT levels (Fig. 7B), thus ruling out the possibility that lack of PDIMs was due to polar effect. Transformation of Mt ΔpapA5 with vector pMV261 failed to restore PDIM synthesis (Fig. 7B). In agreement with the AT activity analysis, PDIM synthesis was also restored by transformation with pCPY143 (expressing PapY₁₄₃F) but not with pCPH124 (expressing PapH₁₂₄A) (Fig. 7B).

Discussion

PapA5 and four other Mt Paps are annotated as proteins of unknown function (5). Recently, some investigators speculated that PapA5 could be involved in PK transport (40), whereas others suggested that it might be an AT (35). Our sequence analysis suggests that PapA5, 3 other Mt Paps and 21 Pap homologs in other Actinomycetales are a subfamily of ATs characterized by a Hx₃Dx₁₄Y motif and sequence and size similarity to CDs of peptide synthetases. Notably, Paps are prevalent in Mycobacterium spp and are primarily associated with PK production pathways.

We used recombinant PapA5 and surrogate alcohol/acyl-CoA substrate pairs to provide proof of principle for PapA5 AT activity by demonstrating that the protein catalyzes hydroxy-ester formation. Various alcohols and acyl-CoA thioesters were investigated as substrates. The selection of these compounds was biased by the assumption that Pap5 catalyzes diesterification of POL and PONE with MYC acid donated from a thioester intermediate. PapA5 displays AT activity with a variety of alcohol/acyl-CoA pairs; however, the observed substrate specificity indicates that the active site has clear substrate recognition determinants. PapA5 utilizes acyl-CoA thioesters of 12-16 carbons and a variety of alkanols and alkanediols as nucleophiles in vitro. Interestingly, mycoside diesters with MYC and palmitic or stearic acids have been previously observed (21). Notably, PCoA is readily converted to ester; however, palmitic acid and palmitoyl chloride are not appreciably used. This may suggest that the neutrality and/or the structural features of the thioesterified substrate (e.g., thioester bond functionality, phosphopantetheinyl moiety, or CoA scaffold) are crucial for recognition. Overall, the hydrophobic character of the preferred surrogate substrates resembles the greasy characteristics of POL, PONE and MYC acid acyl chains.

PapA5-catalyzed ester formation is optimal with 1-OCL and PCoA, with apparent K_m values of 500 and 4 μM, respectively. With these substrates, acyltransfer proceeds with an apparent K_cat value of ≈0.025 min^-1, albeit under severe substrate inhibition. Although 1-OCL is a preferred nucleophile, neither octylamine nor octanethiol are significantly used. This may indicate O-specificity, a property consistent with PapA5 predicted function in vivo. Another property consistent with PapA5 expected function is its ability to catalyze diol diesterification. PapA5 also displays enantioselectivity. Altogether, these studies demonstrate that PapA5 is an AT and present the characterization of a Pap.

We identified a fully conserved Hx₃Dx₁₄Y motif in the Paps and an extended HHx₃DGx₁₃Y motif in PapA5 and its close homologs. Both motifs are present in ATs (32, 33). By analogy with CAT, a prototype AT, H124 of PapA5 would be expected to act as the catalytic base (41). By the same criterion, D128 in PapA5 should be required as an active site stabilizer (38, 42). Substitution of H124 or D128 of PapA5 by Ala decreased activity by 65- and 76-fold, respectively. These results are consistent with mutagenesis analysis of the Hx₃Dx₁₄Y motif of other ATs (38, 41).

Mutational analysis of the proposed PapA5 catalytic site motif is consistent with the catalytic mechanism proposed for CAT and other ATs. We have recently solved the PapA5 crystal structure (J.B. and C.D.L., unpublished work). Preliminary analysis reveals that PapA5 is a pseudodimer of two CAT domains, a property also observed in the stand-alone CD VibH (34). The N-terminal 150 residues of PapA5 share a high degree of structural alignment to CAT, dihydrolipoamide acetyltransferase (E2p) (43), and VibH, particularly with respect to the Hx₃D motif conserved residues. Consistent with Papa5 activities on model substrates, we also observed a deep hydrophobic channel that links the Papa5 surface to the proposed active site. Thus, the structural data support the PapA5 AT activity in vitro and the proposed roles of H124 and D128.

The location of papA5 suggests its involvement in PDIM synthesis. We demonstrated that the ΔpapA5 mutant is deficient in PDIM synthesis, a defect corrected by complementation with papA5. In agreement with the in vitro enzymology, episomal expression of papY₁₄₃F but not of papH₁₂₄A complements the mutant. These results establish PapA5 involvement in PDIM synthesis and provide a demonstration of an in vivo function for a Pap. Based on the in vivo and in vitro studies, we propose that PapA5 catalyzes diesterification of POL and PONE with MYC acid. Similarly, PapA5 homologs encoded in PDIM gene loci of other mycobacteria are expected to catalyze the diesterification required for synthesis of their PDIM variants. It had been suggested that FadD28, an acyl-CoA ligase encoded in the M. bovis PDIM gene locus and required for mycoside production, may catalyze phenol-POL diesterification or synthesize acyl-CoA precursors for MYC acid synthesis (17). In view of our results, and considering that FadD28 did not have AT activity in vitro (17), it is likely that FadD28 is required for the latter function.

Mt papA1, papA2, and papA3 are associated with loci encoding Mas-like PK synthases, which are thought to use methylmalonyl-CoA for methyl-branched PK synthesis (44). Mt papA1 and papA2 are linked to pks2, a gene believed to be required for synthesis of the methyl-branched PK found esterified to trehalose in sulfolipids (19, 45). Mt papA3 is linked to pks 3/4, which are proposed to be required for synthesis of two methyl-branched PK moieties of polyacyltrehaloses (46). Our analysis suggests that papA1, papA2, and papA3 may catalyze O-esterification of trehalose with the methyl-branched PKs. Notably, these PKs and MYC acid have not been found as free acids (12, 39), a finding consistent with the lack of thioesterases in their synthesis loci. The latter observation suggests that Mt Paps may use methyl-branched PKs thioesterified to carrier protein domains of PK synthases in a manner reminiscent of the ATs involve in lipid A synthesis (47).

Last, what we have learned during efforts to deconvolute the biosynthesis of virulence-enhancing PKs of Mt provides insight regarding a yet obscure step in the biosynthesis of the antituberculosis drug rifamycin. As recently noted (48), the AT required for C₂₅ O-acetylation of rifamycin has not been identified. It is likely that Rif20, a PapA5 homolog encoded in the rifamycin gene cluster, is such an AT.