Analysis and modeling of mycolyl-transferases in the CMN group

Abstract

Mycolyl-transferases are a family of proteins that are specifically present in the CMN (Corynebacterium, Mycobacterium and Nocardia) genera and are responsible for the synthesis of cell wall components. We modeled the three-dimensional structures of mycolyl-transfersases from Corynebacterium and Nocardia using homology modeling methods based on the crystal structures of mycolyl-transferases from M. tuberculosis. Comparison of the models revealed significant differences in their substrate binding site. Some mycolyl-transferases identified by the following Gene Ids: Nfa25110, Nfa45560, Nfa7210, Nfa38260, Nfa32420, Nfa23770, Nfa43800, Nfa30260, Dip0365, Ncgl0987, Ce1488, Ncgl0885, Ce0984, Ncgl2101, Ncgl0336, Ce0356 are associated with a relatively larger substrate binding site and amino acid residue mutations (D40N, R43D/G, S236N/A) are likely to affect binding to trehalose.

Background

The CMN group constitutes the organisms of the genera Corynebacterium, Mycobacterium and Nocardia, which are grouped together on the basis of factors that include complex cell wall components, presence/type of mycolic acids, adjuvant activity, presence of cord factor, sulfo-lipids, iron-chelating compounds, polyphosphate, and serological cross-reactivity. The cell walls of the organisms that belong to the CMN group consists of interconnected peptidoglycan and polysaccharide-mycolate complex and are characterized by the presence of mycolic acid on their surface. [1] Mycolic acids are long chain fatty acids that form a part of the unique cell envelope, responsible for the pathogenesis and survival of the organism inside the host. The mycolic acids are named according to the individual genus from which they are isolated; i.e., corynomycolic acids from Corynebacterium comprising ~22-36 carbons, mycolic/eumycolic acids from Mycobacterium comprising ~60-90 carbons and nocardiomycolic acids from Nocardia comprising ~40-60 carbons. [2 –4]

In M. tuberculosis, the mycolyl-transferases are also termed antigen 85 or Ag85 complex enzymes. [5] These correspond to three secreted proteins; Ag85A (Gene Id: Rv3804), Ag85B (Gene Id: Rv1886) and Ag85C (Gene Id: Rv0129). These proteins comprise a signal peptide at the N-terminus followed by a carboxylesterase domain. It has been demonstrated that Ag85 enzymes catalyze the transfer of mycolyl residue from one molecule of α, α' ­ TMM (trehalose monomycolate) to another leading to the formation of α, α' ­ TDM (trehalose dimycolate) and hence these enzymes are termed mycolyl-transferases. [6] Also, in Corynebacterium and Nocardia, orthologous proteins synthesize TDCM (trehalose dicorynomycolate) and TDNM (trehalose dinocardiomycolate), respectively. Further, this family of enzymes is specific only to the CMN group of organisms because of their unique cell envelope. Mycolyl-transferases are also termed fibronectin-binding proteins, since they are involved in binding to fibronectin and entry of the organism into host cells. [7,8] Hence, it is important to understand the structure and function of the proteins responsible for the synthesis of cell wall components in CMN.

The structures of Ag85A (PDB Ids: 1SFR) [9], Ag85B (PDB Ids: 1F0N, 1F0P) [10] and Ag85C (PDB Ids: 1DQZ, 1DQY, 1VA5) [11] were determined for both native and substrate bound forms. The structure corresponds to a α/β hydrolase fold and the catalytic triad responsible for the mycolyl-transferase activity comprise the amino acid residues S126, E230 and H262 (numbering is according to PDB Id: 1F0P). The structural comparison of the three mycolyl-transferases (PDB Ids: 1SFR, 1F0P, 1DQZ) revealed that the active sites are virtually identical indicating that these share a common function. [9] However, in contrast to the high level of similarity within the substrate-binding site and the active site, it was observed that the surface residues disparate from the active site are quite variable indicating that all three Ag85 enzymes in M. tuberculosis are needed to evade the host immune system. The genome sequencing of M. tuberculosis [12], C. glutamicum [13], C. efficiens [14], C. diphtheria [15] and Nocardia farcinica [16] is completed. The M. tuberculosis comprising 3,986 genes is the causative agent of tuberculosis that causes 3 million deaths worldwide. The C. glutamicum comprising 3,002 genes is a soil bacterium and widely used by the industry in the production of amino acids. The C. efficiens comprising 3,069 genes is a non-pathogenic bacterium. The C. diphtheria comprising 2,320 genes is the causative agent of diphtheria. The genome of N. farcinica comprising 5,674 genes is the causative agent of nocardiosis, affecting the lung, central nervous system and cutaneous tissues of humans and animals.

In our earlier work [17], we identified mycolyl-transferases in C. glutamicum and C. efficiens genomes and modeled their three dimensional structures. We reported the relative binding of corynomycolyl-transferases towards trehalose. Our findings are in accordance with the experimental data [18, 19] that reported the gene deletion mutation studies and measured the concentration of TMCM / TDCM. The genomes of N. farcincia, a representative species from Nocardia and C. diphtheria were also subsequently sequenced and we now have complete data available in the public databases on all mycolyl-transferases from species that belong to the CMN group. Therefore we have carried out sequence analysis corresponding to all mycolyl-transferases and modeled the structures of Nocardia and C. diphtheria and compared their substrate binding sites. Such comparative analysis is relevant in situations when the structural information for proteins from only one organism is available and useful inferences can be made about the structure, function and nature of the substrate binding sites for related members from other organisms.

Methodology

Sequence data

The amino acid sequences corresponding to mycolyl-transferases from M. tuberculosis; Ag85A, Ag85B and Ag85C were obtained from the EBI (European Bioinformatics Institute) [20] and are represented by the following Ids; GI: 15610940, GI: 15609023, GI: 57116693, respectively as shown in Table 1.

Table 1. Mycolyl-transferases in CMN group.

Gene Id	GeneBank Id	Source	Protein Length	% similarity	BLASTP E-value
Rv1886c	GI:15609023	M. tuberculosis	325	100	9e-173
Rv3804c	GI:15610940	M. tuberculosis	338	90	1e-146
Rv0129c	GI:57116693	M. tuberculosis	340	81	3e-123
Rv3803c	GI:57117159	M. tuberculosis	299	52	2e-50
Nfa1830	GI:54022147	N. farcinica	345	53	5e-48
Nfa1810	GI:54022145	N. farcinica	347	51	2e-47
Nfa1820	GI:54022146	N. farcinica	353	48	1e-45
NCgl2777	GI:19554065	C. glutamicum	657	50	2e-44
Ce2709	GI:25029265	C. efficiens	669	52	5e-44
Nfa1840	GI:54022148	N. farcinica	624	50	1e-40
NCgl2779	GI:19554067	C. glutamicum	341	50	2e-38
Dip2193	GI:38234734	C. diphtheriae	638	49	3e-3
Ce2710	GI:25029266	C. efficiens	360	51	9e-3
Dip2194	GI:38234735	C. diphtheriae	338	49	7e-35
Nfa5610	GI:54022528	N. farcinica	319	48	2e-33
Nfa30260	GI:54024995	N. farcinica	341	45	8e-28
Nfa32420	GI:54025211	N. farcinica	351	44	9e-27
Nfa38260	GI:54025796	N. farcinica	353	42	2e-26
Nfa7210	GI:54022688	N. farcinica	340	42	4e-26
Ncgl0987	GI:19552252	C. glutamicum	411	45	8e-26
Nfa25110	GI:54024480	N. farcinica	311	45	5e-25
Ce1488	GI:25028044	C. efficiens	390	43	9e-24
Dip0365	GI:38232981	C. diphtheriae	355	43	1e-23
Nfa45560	GI:54026529	N. farcinica	324	44	4e-23
Ncgl0885	GI:19552148	C. glutamicum	483	43	5e-23
Ncgl2101	GI:19553383	C. glutamicum	483	43	8e-23
Nfa23770	GI:54024346	N. farcinica	339	42	4e-22
Nfa43800	GI:54026351	N. farcinica	337	43	9e-22
Dip2339	GI:38234873	C. diphtheriae	406	44	3e-20
Ce0356	GI:25026912	C. efficiens	381	41	5e-20
Ce0984	GI:25027540	C. efficiens	484	42	1e-19
Ncgl0336	GI:19551592	C. glutamicum	365	42	8e-18

Database searching

The homologous proteins were identified for the Mycobacterium, Corynebacterium, and N. farcinica using BLASTP [21] with the Ag85B as the query sequence against GenBank release 153 [22]. The BLOSUM62 matrices were used and the results were sorted using E-value (expected value) with the gap costs set to existence at 11 and extension at 1.

Multiple sequence analysis

Thirty-one mycolyl-transferase sequences were aligned using the CLUSTALW program [23] available at EBI. A penalty of 10 for gap opening, 0.05 for gap extension and 8 for gap separation (default parameters) was assigned for the alignment and shown in Figure 1.

Figure 1.

Multiple sequence alignment corresponding of CMN mycolyl-transferases. Conserved amino acid residues (*), sites of insertion (inverted triangle).

Homology modeling

The three-dimensional models were constructed using MODELER [24] available in InsightII (Accelrys Inc., USA). The structures of Ag85A (PDB Id: 1SFR), Ag85B (PDB Id: 1F0N) and Ag85C (PDB Ids: 1DQZ) were used as templates for modeling. MODELER is an automated comparative modeling program designed to find the most probable structure of a protein sequence, given its alignment with related structures. The model is obtained by the optimal satisfaction of spatial restraints derived from the alignment and is expressed as probability density function for the features restrained. The optimization procedure is a variable target function method that applies conjugate gradients algorithm to position all non-hydrogen atoms. [25] In all seventeen homology models were constructed for the mycolyl-transferases from N. farcincia and C. diphtheria species.

Model evaluation

The models were evaluated using PROCHECK. [26] The RMSD (root mean square deviation) values corresponding to topologically equivalent residues between the models and corresponding crystal structures obtained via structural superposition were derived using programs in InsightII (Accelrys Inc., USA) The method of Profiles-3D that measures the compatibility of an amino acid sequence to a protein of known three-dimensional structure was used to further assess the model. [27]

Substrate docking

The trehalose substrate was docked into the binding site of all protein models using QUANTA (Accelrys Inc., USA). The enzyme-substrate complex was refined using molecular mechanics (MM) and molecular dynamics (MD) calculations in order to understand their interactions. Hydrogen atoms were added to the structures at pH 7.00 using BIOPOLYMER in Insight II. The parameter ‘capping mode off’ was chosen so that the protein ends remain uncharged with the NH2 and COOH groups. The CVFF (Consistent Valence Force Field) force field was chosen and the ‘Fix’ option was used to select the potential atom types, partial charges and formal charges for the protein-substrate complex. The docked complex was subjected to energy minimization using 3000 steps steepest descent followed by conjugate gradient until an energy gradient < 0.01 kcal/mol/A⁰ was achieved. The energy minimized structures were further subjected for MD simulations which were performed in the canonical ensemble (NVT) at 298° K using CVFF force field implemented in Discover-3 and equilibrated for 3000 femtoseconds with step size of 1 femtosecond.

Results and Discussion

Sequence searches identified four mycolyl-transferases each in M. tuberculosis and C. diphtheria, six in C. glutamicum, five in C. efficiens, and thirteen in N. farcinica. The details of mycolyl-transferases analysed and modeled in this work are provided in Table 1. The mycolyl-transferases corresponding to the mycobacteria species; M. tuberculosis, M. leprae and M. bovis are highly similar. Therefore, the mycolyl-transferases from M. tuberculosis H37Rv strain are used in our analysis. Also, M. tuberculosis consists a mycolyl-transferase precursor protein MPT51 (Gene Id: Rv3803) that does not possess mycolyl-transferase activity [28] and was also therefore excluded from our analysis. The multiple sequence alignment of thirty-one mycolyl-transferases is shown in Figure 1. Despite low sequence similarity shared between these proteins, we observed 16 amino acid residues are conserved. These amino acid residues are; L39, W51, P71, D81, W82, W97, F100, G124, S126, S150, D192, G214, E230, G260, H262 and W264. The alignment also indicated some proteins have an insertion sequence of variable length (between 2 and 19 amino acid residues) that precedes the catalytic E230. Further, two N. farcinia proteins (Nfa1810 and Nfa1820) comprise a 27 amino acid residue insertion sequence rich in glycine and serine present between the conserved W82 and W97 (see Figure 1).

The three-dimensional models are useful to identify the positions of these highly conserved resides and regions of insertions. Further, we can also infer the nature of the substrate binding pockets defined by interactions with ‘trehalose’. Evaluation of the three-dimensional models corresponding to corynomycolyl-transferases and nocardiomycolyl-transferases according to PROCHECK indicated more than 85% amino acid residues are in the allowed regions of the Ramachandran plot [29] suggesting that the models are of good quality. Further, according to Profiles-3D, the ‘observed’ scores for the models lie between 124-134 as ‘expected’, suggesting the compatibility of structure and sequence. Also, the RMSD of the respective structures is ~0.68Å and residues that form the catalytic site S126, E230 and H262 can be highly superimposed. The conservation of catalytic residues and their positions in the three dimensional models indicated that all corynomycolyl transferases and nocardiomycolyl transferases must also retain catalytic activity. Examination of the models on computer graphics showed that, the conserved residues L39, P71, D81, W82, W97 and F100 constitute the ‘hydrophobic tunnel’. These are needed in order to accommodate the alkyl chain of mycolic acid, indicating a functional conservation in these proteins. The invariant S126 and G260 are close to the catalytic active site comprising E230. The indole side chains of W51 and W264 are perpendicular to each other and are in proximity to G124 associated with the β5 strand. The amino acid residue D192 is away from the active site indicating that the conservation extends beyond the catalytic site in CMN mycolyl-transferases. We observed that the disulphide connectivity patterns are different. The structures of 1SFR (Ag85A) and 1F0N (Ag85B) consist a disulphide bridge connecting half-cystine residues on β5 and β6 strands. In some proteins, half-cystine residue on the α10 helix and half-cystine residue on the loop connecting β6 strand and α6 helix are involved in the disulphide bridge. The information on the disulphide connectivity pattern is provided in Table 2. Based on the structural superposition, we observed that the differences between these structures are only in the loop regions. The 27 amino acid residue insertion in Nfa1810 and Nfa1820 is located between the β5 and β6 strands that is away from the active site and we therefore predict that it may not be involved in the activity of the protein. According to the structure of 1F0P (Ag85B bound to the substrate trehalose), two substrate binding pockets are present. We observed that the variable region preceding the E230 forms an “insertion loop” close to the trehalose 1151 binding site (Figure 1). The length and amino acid composition of this insertion loop is variable and is given in Table 2. The proteins with a long insertion loop formed a larger substrate binding pocket relative to the mycolyl-transferases. The corynomycolyl-transferases and nocardiomycolyl-transferases with larger substrate binding pocket are: Nfa7210, Nfa38260, Nfa32420, Nfa23720 Nfa43800, Nfa30260, Nfa45560, Nfa25110, Nfa5610, Ce0356, Ncgl0336, Dip0365, Ncgl2101, Ncgl0885 and Ce0984. In order to get an insight into the nature of interaction between the enzymes and substrate, trehalose was docked into the substrate binding site of all modeled structures and optimized using energy minimization. The specificity pockets defined by interaction with trehalose substrate were examined and the results are presented in Table 2. While some proteins retain the nature of residues lining the specificity pockets, mutations such as D40N, R43D/G, S236N/A are observed in Nfa25110, Nfa45560, Nfa7210, Nfa38260, Nfa32420, Nfa23770, Nfa43800, Nfa30260, Dip0365, Ncgl0987, Ce1488, Ncgl0885, Ce0984, Ncgl2101, Ncgl0336 and Ce0356. In these proteins specificity may be affected. Further, we observed that proteins with large substrate binding site were also associated with specific amino acid residue mutations. Therefore, in these proteins binding to trehalose is affected. Also, we observed that proteins comprising conserved amino acid residues in the substrate binding site are not associated with an insertion loop. Therefore, such proteins may bind trehalose.

Table 2. ‘Insertion loop’ amino acid sequence, disulphide bridges and substrate binding pockets in CMN mycolyl-transferases.

Protein	‘Insertion loop’ amino acid sequence	Disulphide bridge	Trehalose 1151 binding residues							Trehalose 1152 binding residues
1F0P		Cys 87-	40D	43R	126S	223N	262H	263S	264W	154D	157Q	159M	231N	232F	235S	236S	239K
	Cys 92
Rv0129			38D	41R	124S	221N	260H	261S	262W	152N	155E	157W	229G	230L	233R	234T	237T
Rv3804		Cys 87-	40D	43R	126S	223N	262H	263S	264W	154D	157Q	159M	231G	232F	235T	236S	239K
		Cys 92
Ncgl2777	AIGPA		40D	43R	121S	216G	261H	262S	263W	149D	152S	154G	231V	232I	235M	236T	239T
Ce2709	ATGPA		40D	43R	121S	215G	261H	262A	263W	149D	152S	154G	231L	232I	235M	236T	239T
Ncgl2779	DH		41D	44R	128S	223V	266H	267G	268W	156N	159A	161G	236F	237V	240T	241S	244I
Ce2710	DH		41D	44R	128S	223T	266H	267S	268W	156T	159A	161G	236A	237V	240A	241T	244A
Ncgl0987	SEKEPFYN		41D	44G	125S	219D	267H	268N	269W	153S	156D	158I	240S	241C	244A	245L	248S
Ce1488	YADEPFYN		41D	44G	125S	219E	267H	268N	269W	153S	156D	158I	240S	241C	244A	245L	248A
Ncgl0885	DNAPIDEDAFKNR		41G	44D	124S	-	272H	273A	274W	152E	155S	157M	241A	242M	245T	246C	249N
Ce0984	ENAPEDEKGLKNR		41G	44D	124S	-	272H	273A	274W	152E	155S	157M	241A	242L	245T	246C	249N
Ncgl2101	DNAPIDEDAFKNR		41G	44D	124S	-	272H	273A	274W	152E	155S	157M	241A	242M	245T	246C	249N
Ncgl0336	SPRFEGLNQQVQSIAMAET		41N	44D	124S	218D	276H	277S	278W	152A	155S	157L	246A	247A	250K	251C	254D
Ce0356	SPRFNGLDQAYLSLAMTET		41N	44D	124S	218N	276H	277S	278W	152S	155Q	157L	246A	247A	250K	251C	254D
Nfa1810	FG		40D	43R	153S	249N	291H	292N	293W	181N	184A	186G	260V	261L	264A	265N	268A
Nfa1820	FN		40D	43R	148S	244S	286H	287A	288W	176N	179A	181G	255A	256L	259A	260N	263A
Nfa1830	SPVGVFN		39D	42R	124S	218N	264H	265S	266W	152N	155A	157G	234A	235L	238V	239N	242A
Nfa1840	PGVST		41D	44R	122S	217S	263H	264S	265W	150T	153T	155G	233I	234L	237L	238T	241N
Nfa25110		Cys 146-	38A	41G	120S	-	252H	253T	254W	148W	151D	153P	222A	223I	226T	227C	230A
		Cys 227
Nfa45560	APGIDGNPLDLVER	Cys 146-	38N	41D	120S	241T	266H	267S	268W	148R	151D	153A	237T	238V	241A	242C	245P
		Cys 242
Nfa7210	GPYALPGSYGLANQ	Cys 149-	41N	44G	123S	218N	271H	272S	273W	151Q	154D	156V	241A	242G	245Y	246C	249N
		Cys 246
Nfa38260	GPHAMPGSDGLTNQ	Cys 150-	41A	44G	123S	217N	270H	271S	272W	151Q	154D	156V	240A	241G	244H	245C	248N
		Cys 246
Nfa32420	YLNAAPGPMGAVN-	Cys 150-	41N	44D	123S	218Y	270H	271Y	272W	151Q	154D	156T	240A	241A	244Q	245C	248N
		Cys 246
Nfa23770	NPRLHNDDRQLLNQ	Cys 157-	41N	44G	130S	224A	278H	279S	280W	158M	161D	163L	247S	248V	241L	252C	255R
		Cys 253
Nfa43800	AVGGDPMQLGYQ	Cys 149-	41N	44S	122S	-	267H	268A	269W	150R	153D	155Q	237A	238V	241M	242C	245Q
		Cys 243
Nfa30260	GPGIDADPLALADQ	Cys 149-	41N	44T	123S	217Q	270H	271S	272W	151P	154D	156R	240A	241V	244D	245C	248E
		Cys 245
Nfa5610	KPQLAEN	Cys 148-	41D	43D	122S	214L	260H	261S	262W	150D	153L	155T	230V	231G	234I	235C	238A
		Cys 235
Dip0365	SPRLAGKDPVTIFATNLIT		39N	42D	122S	216S	274H	275S	276W	150A	153S	155L	244A	245G	248M	249C	252D
Dip2339	PKEDGPFT		41D	44T	125S	219G	269H	270S	271W	153S	156N	158S	240R	241C	244E	245L	248S
Dip2193	ANKKG		40D	43R	121S	215G	261H	262D	263W	149D	152S	154G	231V	232I	235M	236T	239T
Dip2194	ND		41D	44R	125S	220Y	263H	264N	265W	153S	156V	158G	233I	234A	237V	238S	241I

It is often observed that, during evolution, gene duplications, rearrangements and gene loss occur in genomes due to a complex, general purpose mechanism for rapid adaptation of the organism. As a result of gene duplication, extra copies of selected genes are evolved. Duplications are important because they effectively allow at least one of the gene copies to evolve while the function of the original gene can remain intact. Many new functions arise from duplication and subsequent change of old genes. In this way, duplication of pre-existing genetic information provides the raw material from which new gene functions can evolve thereby contributing to the genetic complexity during evolution. With reference to mycolyl-transferases in the CMN genera, the presence of varying number of proteins in each organism reflects gene duplication events during evolution of these organisms. Further, we identified that the overall structure, active site and hydrophobic tunnel are identical in all proteins, with significant differences in substrate specificity pockets which may be a result of selective pressure during evolution. From this work, we propose that trehalose is the original substrate and this binding is retained only in some corynomycolyl-transferases and nocardiomycolyl-transferases. During gene duplication, mutations in the substrate binding site have occurred such that the newly evolved proteins can bind to other sugars so as to synthesize organism specific polysaccharide-mycolate cell wall component.

Further, the mycolyl-transferases Nfa1840, Ncgl2777, Ce2709 and Dip2193 comprise a 300 amino acid residue C-terminal extension as a result of gene fusion events. Brand et al., 2003 reported that deletion of Ncgl2777 gene led to a 10-fold increase in the cell volume of the organism. We reported the identification of 55 amino acid residue tandem LGFP (conserved sequence motif; leucine, glycine, phenylalanine, proline) repeats in the C-terminal region of Ncgl2777 and Ce2709 [30] and suggested that the abnormal increase in the cell volume of C. glutamicum is due to the loss of C-terminal domain corresponding to the LGFP tandem repeats that may be responsible for maintaining the integrity of the cell wall. The presence of these LGFP repeats in C-terminal region of Nfa1840 and Dip2193 imply that these are also cell surface proteins and may be important in maintaining cell wall integrity in analogous manner.

Conclusion

This work describes the comparison of the three-dimensional models for mycolyl-transferases in CMN genera. Although the sequence identities in some cases is as low as 17%, yet the overall α/β fold characteristic of mycolyl-transferases is conserved. This conservation extends to the active site comprising amino acid residues; S126, E230 and H262. However, the amino acid residues comprising the substrate binding pockets defined by interactions with trehalose vary owing to certain mutations in some mycolyl-transferases. Also, significant differences are observed in the size of the substrate binding pocket owing to the close proximity of an insertion loop between the conserved W82 and W97. The size and nature of amino acid residues corresponding to the substrate binding pockets is likely to affect mycolyl-transferase substrate specificity. These observations lead us to believe that during the course of evolution, gene duplication events followed by mutagenesis at the substrate binding pockets, may have resulted in those mycolyl-transferases that are responsible for synthesis of polysaccharide-mycolate complex in an organism specific manner.

Figure 2.

The structural superposition of representative CMN mycolyl-transferases (PDB Id: 1F0P (brown), Ncgl0336 (yellow), Ncgl0987 (blue). The side chains of the active site residues S126, E230, H262 (red) and trehalose 1151 (green) are represented in ball and stick model.