PDIMs and PGLs are both re quired for virulence of...

1

PDIMs and PGLs are both required for virulence of Mycobacterium marinum 1

2

Jia Yu1, 2 &, Vanessa Tran3, &, Ming Li3, Xinghua Huang1, Chen Niu1, Decheng Wang1, Jianghua 3

Zhu1, Jianping Wang1, Qian Gao1,* and Jun Liu3,* 4

5

1Key Laboratory of Medical Molecular Virology, Institute of Biomedical Sciences and Medical 6

Microbiology, 2Laboratory of Population and Quantitative Genetics, Institute of Genetics and 7

Biostatistics, School of Life Sciences, Fudan University, Shanghai, China, 3Department of 8

Molecular Genetics, University of Toronto, Toronto, ON M5S 1A8, Canada 9

10

&These authors contributed equally. 11

12

*Corresponding authors: 13

Jun Liu, 4382 Medical Science Building, University of Toronto, 1 King’s College Circle, Toronto, 14

Ontario, Canada. Phone: 416-946-5067; Fax: 416-978-6885. E-mail: [email protected] 15

Qian Gao, Shanghai Medical College, Fudan University, 138 Yi Xue Yuan Road, Shanghai 200032, 16

China. Phone: 86-21-54237195; Fax: 86-21-54237195. E-mail: [email protected] 17

18

Running title: PDIMs and PGLs are required for mycobacterial virulence 19

20

21

22

Copyright © 2012, American Society for Microbiology. All Rights Reserved.Infect. Immun. doi:10.1128/IAI.06370-11 IAI Accepts, published online ahead of print on 30 January 2012

on May 31, 2018 by guest

http://iai.asm.org/

Dow

nloaded from

http://iai.asm.org/

2

Abstract 23

Phthiocerol dimycocerosates (PDIMs) and structurally related phenolic glycolipids (PGLs) are 24

complex cell wall lipids unique to pathogenic mycobacteria. While these lipids have been 25

extensively studied in recent years, there are conflicting reports on some aspects of their 26

biosynthesis and on the role of PDIMs and especially PGLs in virulence of Mycobacterium 27

tuberculosis (M. tb). This has been complicated by the natural deficiency of PGLs in many 28

clinical strains of M. tb and the frequent loss of PDIMs in laboratory M. tb strains. In this study, 29

we isolated seven mutants of Mycobacterium marinum deficient in PDIMs and/or PGLs, in 30

which multiple genes of the PDIM/PGL biosynthetic locus were disrupted by transposon 31

insertion. Zebrafish infection experiments showed that M. marinum strains lacking one or both 32

of these lipids were avirulent, suggesting that both PDIMs and PGLs are required for virulence. 33

We also found that these strains were hypersensitive to antibiotics and exhibited increased cell 34

wall permeability. Our studies provide new insights into the biosynthesis of PDIMs/PGLs and 35

may help to understand the role of PDIMs and PGLs in M. tb virulence. 36


http://iai.asm.org/

Dow

nloaded from

http://iai.asm.org/

3

INTRODUCTION 37

Pathogenic mycobacteria produce two structurally related, methyl-branched fatty acid-38

containing lipids called phthiocerol dimycocerosates (PDIMs) and phenolic glycolipids (PGLs). 39

PDIMs and PGLs have long chain fatty acid backbones consisting of 3-methoxy (or 3-keto, 3-40

hydroxy), 4-methyl, 9,11-dihydroxy glycols (phthiocerols) and p-glycosylated phenylglycols 41

(glycosyl phenolphthiocerols), respectively, that are diesterified with di-, tri-, and tetra-42

methylbranched acyl chains (mycocerosates)[reviewed in (28)]. PDIMs have been identified in 43

Mycobacterium tuberculosis (M. tb), M. africanum, M. bovis, M. leprae, M. marinum, M. 44

ulcerans, M. kansasii, M. haemophilum, M. microti, and M. gastri, all of which are pathogenic 45

for humans or animals. PGLs are produced by the same set of pathogenic mycobacterial 46

species except that in M. tb only a subset of clinical isolates produces PGLs. 47

The role of PDIMs in virulence was first suggested by two independent studies using 48

signature-tagged transposon mutagenesis, which identified mutants of M. tb that were unable 49

to either produce or properly localize PDIMs to the cell wall, and demonstrated that these 50

mutants were attenuated in animal models of infection (8, 12, 33). Since then, circumstantial 51

evidence supporting a role for PDIMs in M. tb virulence accumulated. The role of PGLs in M. tb 52

virulence is less clear and is confounded by the fact that laboratory strains (H37Rv, Erdman) and 53

many clinical isolates, including CDC1551 and MT103, are naturally deficient in PGL production 54

due to a 7-base-pair deletion in pks15/1, while some clinical isolates of the East-Asian lineage 55

have an intact pks15/1 gene and produce PGLs (31). Mutations of pks15/1 in M. tb HN878, a 56

strain that produces both PDIMs and PGLs and exhibits a hypervirulent phenotype in infected 57

animals, abolished PGL synthesis and decreased the virulence of the mutant to the level of non-58


http://iai.asm.org/

Dow

nloaded from

http://iai.asm.org/

4

PGL producing strains H37Rv and CDC1551 (31, 40). As such, it was suggested that the 59

production of PGLs in some M. tb strains of the Beijing family is associated with their increased 60

virulence. However, this correlation has been less than perfect (35). Moreover, an H37Rv 61

strain engineered to produce PGLs was not more virulent than the parental strain in the mouse 62

or rabbit model of infection (35). Therefore, the role of PGLs in M. tb virulence remains 63

controversial. In M. bovis (20) or M. marinum (2, 9), simultaneous loss of PDIMs and PGLs has 64

been associated with decreases in virulence. However, the contribution of individual PDIMs 65

and PGLs to virulence in these organisms has not been evaluated. 66

Significant progress on the biosynthesis of PDIMs and PGLs has been made in recent 67

years. Genes involved in PDIM/PGL synthesis were first identified by Kolattukudy and co-68

workers (5, 6). They showed that disruption of polyketide synthase (PKS) genes ppsB and ppsC 69

in M. bovis BCG abolished the production of PDIMs and PGLs (5). Subsequent genetic studies in 70

M. tb identified other genes involved in PDIM and/or PGL biosynthesis, including fadD26, 71

fadD28, mmpL7, ddrC, and pks15/1, which are in the same genetic region encompassing the 72

ppsA-ppsE genes (8, 12, 31). Together, these studies have uncovered a genetic locus which 73

consists of more than 30 genes believed to be involved in PDIM/PGL biosynthesis and major 74

biochemical steps leading to PDIM/PGL production have been proposed [Fig. 1 and (28)]. 75

Accordingly, phthiocerols and phenolphthiocerols are generated by five PKSs, PpsA-E, which are 76

arranged as a modular PKS system consisting of predicted domains required for the co-linear 77

assembly of the 3-methoxy, 4-methyl, 9,11-dihydroxy segment. A different PKS, Mas, 78

synthesizes the methyl-branched lipids, mycocerosates. Diesterification of phthiocerols with 79

mycocerosates mediated by an acyltransferase, PapA5, generates PDIMs (27). Similarly, PapA5 80


http://iai.asm.org/

Dow

nloaded from

http://iai.asm.org/

5

catalyzes the diesterification of phenolphthiocerols with mycocerosates and additional 81

glycosyltransferases are required for the glycosylation to produce the final PGL products (30). 82

While this general scheme is likely to be true, some aspects of this pathway remain unclear. 83

For example, while fadD28 and mas are thought to be responsible for the synthesis of 84

mycocerosates, a common precursor of both PDIMs and PGLs, disruption of mas or fadD28 in 85

M. bovis BCG abolished PGL but not PDIM production (6, 16). Genetic disruption of mas in M. 86

tb has not been reported, but inactivation of fadD28 in M. tb Erdman or MT103 strain 87

abrogated the production of PDIMs (7, 12). It is unclear whether this discrepancy reflects 88

differences in the strains studied. For PpsA-E, biochemical evidence is consistent with their role 89

in PDIM/PGL synthesis (39), but specific deletion studies have only been described for ppsB and 90

ppsC in M. bovis BCG (5), and more recently, ppsD in M. tb H37Rv (24). 91

In this study, we isolated seven mutants of M. marinum in which ppsA, ppsB, ppsD, ppsE, 92

mas, fadD28 and fadD26 genes were individually disrupted by transposon insertions. We found 93

that ppsA::Tn, ppsB::Tn, ppsD::Tn, and ppsE::Tn mutants were unable to produce PDIMs and 94

PGLs, providing the first genetic evidence for the role of ppsA and ppsE in PDIM/PGL synthesis. 95

Disruption of mas or fadD28 abolished the production of both PDIMs and PGLs, reinforcing 96

their role in mycocerosate synthesis. The fadD26::Tn mutant, in which the transposon inserted 97

in the promoter region, produced a trace amount of PDIMs and a low level of PGLs. Zebrafish 98

infection experiments showed that all mutants were severely attenuated and were essentially 99

avirulent under the experimental conditions. Complementation of the fadD28::Tn mutant 100

partially restored PDIM/PGL production and virulence in zebrafish, confirming the role of 101

PDIMs/PGLs in M. marinum virulence. Complementation of the fadD26::Tn mutant resulted in 102


http://iai.asm.org/

Dow

nloaded from

http://iai.asm.org/

6

an intermediate level of PDIMs but further reduction of PGLs, and consequently, the 103

recombinant strain remained avirulent in zebrafish. Taken together, our results suggest that 104

PDIMs and PGLs are both required for virulence of M. marinum. In addition, we have provided 105

new insights into the biosynthesis of PDIMs and PGLs and their role in mycobacterial virulence. 106

107

MATERIALS AND METHODS 108

Bacterial strains, media, and growth conditions 109

Mycobacterium marinum M strain (ATCC BAA-535) was used as the parental and wild type 110

strain for the transposon mutagenesis and subsequent experiments. M. marinum cells were 111

grown at 32oC in Middlebrook 7H9 broth (Difco) supplemented with 0.2% glycerol and 10% 112

oleic acid-albumin-dextrose-catalase (OADC) (Difco) or on Middlebrook 7H11 agar (Difco) 113

supplemented with 0.5% glycerol and 10% OADC. Escherichia coli strain DH5α was used for 114

routine manipulation and propagation of plasmid DNA. E. coli strain DH5α λ pir116 (1) was 115

used for isolation of transposon-containing plasmid. Antibiotics were added as required: 116

kanamycin, 50 µg/ml for E. coli and 25 µg/ml for M. marinum; hygromycin, 150 µg/ml for E. coli 117

and 50 µg/ml for M. marinum. 118

119

Generation and screening of M. marinum �MycoMar insertion library 120

Propagation of the �MycoMar transposon phage and preparation of phage lysates have been 121

described previously (1). For phage infection, M. marinum cells were washed and resuspended 122

in 50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 10 mM MgSO4, and 2 mM CaCl2. Phage were added at 123

a multiplicity of infection (MOI) of 10:1 and incubated at 37oC for 3 hrs to allow infection to 124


http://iai.asm.org/

Dow

nloaded from

http://iai.asm.org/

7

occur. Bacteria were then plated on Middlebrook 7H11 agar supplemented with kanamycin 125

and incubated at 32oC. Kanamycin-resistant (i.e., transposon-containing) M. marinum colonies 126

were patched onto Middlebrook 7H11 agar and colonies with unusual morphology were 127

identified by visual inspection. 128

129

Localization of the �MycoMar insertion 130

Localization of the transposon insertions was done as previously described (1). Briefly, total 131

chromosomal DNA of the transposon insertion mutant was cleaved with BamHI, then self-132

ligated with T4 DNA ligase and transformed into competent E. coli DH5α λ pir116 cells. The 133

�MycoMar element contains an R6K origin and an aph gene such that recircularized fragments 134

containing the transposon are able to replicate as kanamycin resistant plasmids. Plasmid DNA 135

was isolated from KmR E. coli transformants and �MycoMar-specific primers were used to 136

determine the DNA sequence at the transposon/chromosomal junction. These DNA sequences 137

were compared to the genome sequences of M. marinum 138

http://genolist.pasteur.fr/MarinoList/. 139

140

Molecular cloning and complementation 141

A 4494-bp fragment containing fadD28 (MMAR_1765) gene and mmpL7 (MMAR_1764) was 142

amplified by PCR using chromosomal DNA of M. marinum M strain as the template and the 143

forward primer 5’-GAAGATCTAGGAGTGATGCCCATGAGTGTGCGTTCCCTT -3’ and reverse primer 144

5’- GACTAGTTTGACCGGTGCCCAGTCGATTGC -3’, which contains a BglII and a SpeI site, 145

respectively (underlined). The PCR product was cloned into BamHI and SpeI sites of the 146


http://iai.asm.org/

Dow

nloaded from

http://iai.asm.org/

8

pSMT3LxEGFP shuttle vector (21), creating pFad28/MmpL7. Similarly, the intact fadD28 was 147

amplified using the forward primer 5’-GAAGATCTAGGAGTGATGCCCATGAGTGTGCGTTCCCTT -3’ 148

and reverse primer 5’-CCCAAGCTTCTAGACGTCCAGGCGGGCGAACTGC-3’. The amplicon was 149

digested with BglII and HindIII and cloned into pSMT3LxEGFP, creating pFadD28. A 1755-bp 150

fragment containing fadD26 (MMAR_1777) was PCR amplified using the forward primer 5’- 151

CGGGATCCAAGGATGTAGTGCGATGCCGGTGACCGACCG -3’ and reverse primer 5’- 152

GACTAGTTCATACCGTCACGTCCAGCCGATTG -3’. The fragment was cloned into BamHI and SpeI 153

sites of the pSMT3LxEGFP shuttle vector, generating pFadD26. The constructs were confirmed 154

by DNA sequencing. Constructs were electroporated into appropriate mutant strains of M. 155

marinum and transformants selected on Middlebrook 7H10 agar plates containing 50 μg/ml of 156

hygromycin. 157

158

Quantitative real-time PCR (qRT-PCR) 159

Determination of the expression level of fadD26, ppsA and ppsE was accomplished by qRT-PCR 160

using specific primers for each gene: fadD26, 5'-CGGATTTATCGGGGTCCCACTTTC-3'(forward) 161

and 5'-CCGTCTTGTGAGCTGGCGTATTTTG-3'(reverse); ppsA, 5'-162

GACAAGATGGACCCGCAGCAAC(F)-3' and 5'-TCAGACATGAGCCGGCGAAGAC-3'(R); ppsE, 5'-163

GAACCTCCTCTCCCAGTGGCTAT-3' (F) and 5'-GGCTGAACTGATCGAAGTTGAGC-3'(R). Primers 164

specific for sigA (5'-GAAAAACCACCTGCTGGAAG-3', and 5'-CGCGTAGGTGGAGAACTTGT-3') were 165

used as an endogenous control to normalize the amount of cDNA template added to each qRT-166

PCR sample. The cDNA used in these experiments was prepared from RNA samples obtained 167


http://iai.asm.org/

Dow

nloaded from

http://iai.asm.org/

9

from three biological replicates. qRT-PCRs were carried out in triplicate using a 7500 Real-Time 168

PCR System (Applied Biosystems) and iQ SYBR green supermixture (Bio-Rad). 169

170

Thin layer chromatography (TLC) analysis of PDIMs and PGLs 171

The apolar lipids were prepared from M. marinum cells (50 mg dry biomass) according to 172

published procedures (11, 32). These lipids were analyzed by two-dimensional thin layer 173

chromatography (2D-TLC) on silica gel 60 plates (EMD Chemicals Inc.). For detection of PDIMs, 174

apolar lipids were developed with petroleum ether/ethyl acetate (98:2, 3×) in the first 175

dimension and petroleum ether/acetone (98:2) in the second dimension. Lipids were visualized 176

by spraying plates with 5% phosphomolybdic acid followed by gentle charring of the plates. For 177

detection of phenolic glycolipids (PGL), the apolar lipid extract was developed with 178

chloroform/methanol (96:4, v/v) in the first direction and toluene/acetone (80:20, v/v) in the 179

second direction, followed by charring with α-naphthol. 180

For quantitative TLC analysis, the M. marinum strains were grown to mid-log phase in 181

7H9-OADC and incubated with 25 µCi of [14C]propionate for 24 hrs (specific activity: 54 182

mCi/mmol; American Radiolabeled Chemicals Inc.). The apolar lipids were prepared from M. 183

marinum cells (50 mg dry biomass) and analyzed by 2D-TLC as described above. Lipids were 184

visualized and quantified by phosphorimaging using a storage phosphor screen (GE Healthcare) 185

and Typhoon9400TM imager. 186

187

Zebrafish infection 188


http://iai.asm.org/

Dow

nloaded from

http://iai.asm.org/

10

Adult zebra-fish (Danio rerio) were infected as described previously (41). To assess the survival, 189

20 fish per group were infected by intraperitoneal injection with 20 µl of thawed bacterial 190

stocks that were diluted in PBS to reach a dosage of approximately 104 CFU per fish. A Kaplan-191

Meier curve was calculated for animals infected with each strain including mock-injected (PBS) 192

animals and statistical analysis comparing survival of different groups of animals was calculated 193

by log-rank test. 194

195

Drug sensitivity assay 196

The drug sensitivity of the M. marinum strains was tested using the agar dilution method. 197

Briefly, strains were grown in 7H9 supplemented with ADN (0.5% albumin, 0.2% dextrose, 0.085% 198

NaCl) to OD600 ≈1.0. Strains were then adjusted to a concentration 2.5x106 cfu/mL and 4 µL was 199

spotted onto 7H11 agar containing 2-fold dilutions of antibiotic. Plates were visually inspected 200

after 1 week incubation at 32°C to find the minimum inhibitory concentration (MIC). 201

202

Uptake assay 203

The accumulation of Nile Red (Sigma) and ethidium bromide (EtBr) (OmniPur) was measured. 204

Strains of M. marinum were grown in 7H9/ADN to an OD600 ≈ 1.0, washed twice with 50 mM 205

potassium phosphate buffer (pH 7), and resuspended in 2.5 mL of buffer. The OD600 of the 206

resuspended cells was determined, adjusted to OD600 = 0.4, and 100 µL of this cell suspension 207

was added in triplicate to a 96-well black fluoroplate (Greiner Bio-One). Nile Red and EtBr were 208

added to a final concentration of 2 µM and 6 µM, respectively. The accumulation of these dyes 209

was measured by fluorescence using a TECAN Infinite® M200 spectrofluorometer with an 210


http://iai.asm.org/

Dow

nloaded from

http://iai.asm.org/

11

excitation of 540 nm and emission of 630 nm for Nile Red and an excitation of 545 nm and 211

emission of 600 nm for EtBr. 212

213

RESULTS 214

Isolation and characterization of M. marinum mutants defective in cord formation 215

When growing in liquid medium, pathogenic mycobacteria such as M. tb and M. marinum form 216

serpentine cords, which are large bundles of bacterial cells aggregated in parallel along their 217

long axes (19, 25, 36). A link between cord formation and virulence was first suggested by the 218

observation that the virulent M. tb strain H37Rv formed cords whereas the avirulent strain 219

H37Ra did not (25). This was later supported by isolation of attenuated strains of M. tb and M. 220

marinum which were also defective in cording (17, 18). Therefore, as a general strategy to 221

uncover virulence genes, we first visually screened a transposon-insertion library of M. 222

marinum for altered colony morphology, followed by microscopic screening for colonies 223

defective in cord formation. The initial screen by colony morphology was done since factors 224

that contribute to cord formation are likely cell surface components. Thus, defects in cording 225

would likely give rise to colonies with altered morphology on agar plates, as demonstrated in 226

our earlier studies (10, 32). We identified 66 mutants after screening approximately 10,000 227

colonies, which exhibited altered colony morphology compared to the wild type strain. Of these, 228

11 mutants were defective in cord formation (see examples in Fig. 2A and 2B). 229

Genetic characterizations of these mutants revealed that 7 mutants contains 230

transposon insertion site individually within the open reading frames (orf) of ppsA, ppsB, ppsD, 231

ppsE, fadD28, mmaA3, and PPE38 genes. Three mutants had transposon insertion at different 232


http://iai.asm.org/

Dow

nloaded from

http://iai.asm.org/

12

sites within the orf of mas and one mutant had transposon inserted at the promoter region of 233

fadD26 (Fig. 1A). Remarkably, 7 of the 9 genes identified (except mmaA3 and PPE38) belong to 234

the presumed PDIM/PGL biosynthetic locus of M. marinum (Fig. 1A), which consists of more 235

than 30 genes spanning ∼70 kb in the genome (28). These mutants are the focus of our current 236

studies and the other two mutants (mmaA3::Tn and PPE38::Tn) will be dealt with elsewhere. 237

238

The ppsA::Tn, ppsB::Tn, ppsD::Tn, ppsE::Tn, fadD28::Tn and mas::Tn mutants of M. marinum 239

failed to synthesize PDIMs and PGLs 240

To confirm the role of the identified genes in PDIM/PGL synthesis, we performed two-241

dimensional thin layer chromatography (2D-TLC) analyses of PDIMs and PGLs using apolar lipid 242

fraction isolated from each mutant according to previously published procedures (11). 2D-TLC 243

analysis showed that the synthesis of PDIMs and PGLs were abolished in ppsA::Tn, ppsB::Tn, 244

ppsD::Tn, ppsE::Tn, fadD28::Tn and mas::Tn mutants (Fig. 3A and Supplementary Figure 1). 245

Unlike previous studies using M. bovis BCG, which found that fadD28::Tn and mas::Tn mutants 246

still produced shorter chain PDIMs (6, 16), the mas::Tn or fadD28::Tn mutant of M. marinum is 247

devoid of PDIMs. However, our result is consistent with previous studies of the fadD28::Tn 248

mutants of M. tb (7, 12). 249

To confirm the role of fadD28 in PDIM/PGL synthesis, we cloned the intact fadD28 gene 250

or fadD28 together with its downstream gene, mmpL7, and transformed the constructs into the 251

fadD28::Tn mutant. Complementation with each construct partially restored the production of 252

PDIMs and PGLs in the mutant (Fig. 3A). The identity of PDIMs and PGLs were confirmed by 253

mass spectrometric analysis and/or by radiolabeled 2D-TLC analysis. The MALDI-TOF spectra of 254


http://iai.asm.org/

Dow

nloaded from

http://iai.asm.org/

13

purified PDIMs revealed a series of major pseudomolecular ion [M + Na]+ peaks at m/z 1238, 255

1252, 1266, 1280 (data not shown), which corresponds to a C81–C84 composition of PDIMs and 256

agrees with PDIMs found in M. marinum (9). To quantify the relative abundance of PDIMs and 257

PGLs, we repeated the 2D-TLC analysis with bacterial cultures labeled with [14C]propionate, a 258

precursor of PDIMs and PGLs, followed by quantification by phosphor imaging analysis (Fig. 3C). 259

This analysis revealed that the levels of PDIMs and PGLs in the fadD28 complemented strain 260

were 25% and 35% of that in the wild type strain, respectively (Fig. 3D). The relatively low 261

levels of PDIMs and PGLs in the complemented strains compared to wild type may be due to 262

sub-optimal expression of fadD28 in the cloning vector. Previously, it was found that 263

complementation of the fadD28::Tn mutant of M. tb with intact fadD28 in a plasmid restored 264

only 15% PDIM production, presumably due to the same reason (7). 265

266

The fadD26 gene is involved in the synthesis of PDIMs but not PGLs in M. marinum 267

We found that unlike other mutants described above, the fadD26::Tn mutant still produced a 268

trace amount (2%) of PDIMs and a substantial amount (31%) of PGLs, although at a lower level 269

than wild type (Figs. 3B, 3C & 3D). In this mutant, the transposon inserted at 313 bp upstream 270

of the predicted start codon of fadD26, which reduces the fadD26 expression to about 9% 271

based on quantitative real time PCR (qRT-PCR) analysis (Fig. 4). This is consistent with previous 272

findings in M. tb, where transposon insertion at the upstream region (113 bp) of fadD26 273

resulted in significantly reduced level of PDIMs, and transposon insertion within the orf of 274

fadD26 abolished PDIM production (7). 275


http://iai.asm.org/

Dow

nloaded from

http://iai.asm.org/

14

The role of fadD26 in PGL synthesis has not been evaluated in M. tb because the strains 276

(Erdman and MT103) from which fadD26::Tn mutants were generated are naturally deficient in 277

PGL production due to the pks15/1 polymorphism (7, 8, 12). We found that in the fadD26::Tn 278

mutant of M. marinum, the amount of PGLs decreased, which would initially suggest that 279

fadD26 is also involved in PGL synthesis. However, complementation with the intact fadD26 280

gene of M. marinum did not increase the level of PGL production in the mutant. Instead, the 281

amount of PGLs was further reduced in the complemented strain (from 31% to 4%) (Figs. 3B, 3C 282

& 3D). As expected, complementation of the fadD26::Tn mutant increased the production of 283

PDIMs, although not to the wild type levels (57%). 284

In M. tb, fadD26 and its downstream genes ppsA to papA5 were shown to be 285

transcriptionally coupled (7). Consistently, we found by qRT-PCR analysis that the transcript 286

levels of ppsA and ppsE in the fadD26::Tn mutant were reduced to 7% and 52% of that in the 287

wild type strain, and that trans-complementation with fadD26 did not increase their 288

expressions (Fig. 4). This suggests that like M. tb, fadD26 and its downstream genes (ppsA-289

papA5) in M. marinum also form an operon. Accordingly, transposon insertion in the upstream 290

region of fadD26 causes a polar effect on the expression of ppsA-papA5 genes, which affects 291

the synthesis of both PDIMs and PGLs. As a consequence, fadD26 is overexpressed in the 292

complemented strain but the expression of ppsA-papA5 is still low. Because PpsA-E are PKSs 293

used for both PDIM and PGL synthesis, the elevated FadD26 may outcompete FadD29, which is 294

specific for the PGL pathway (Fig. 1), for the limited amount of PpsA-E, resulting in the further 295

reduction of PGL in the complemented strain. 296


http://iai.asm.org/

Dow

nloaded from

http://iai.asm.org/

15

The low expression level of ppsA-papA5 in the complemented strain may also explain 297

the partial restoration of PDIM production. Previously it was found that complementation of 298

the fadD26::Tn mutant of M. tb by intact fadD26 gene only restored 5% PDIM production (7). 299

Based on our results, we suggest that fadD26 is involved in the synthesis of PDIMs but not PGLs 300

in M. marinum. Our conclusion is in accordance with a recent study in M. bovis BCG, which 301

found that deletion of fadD26 abolished PDIM production and reduced the level of PGLs, and 302

that complementation only partially restored the PDIM production (34). Interestingly, in that 303

study, the amount of PGLs in the complemented strain remained unchanged, presumably 304

because fadD26 was expressed as a single copy gene (i.e., it was cloned in an integrative vector 305

for complementation experiments) and its expression level is not high enough to outcompete 306

FadD29 for the common enzymes (PpsA-E) involved in both pathways (Fig. 1). 307

308

PDIMs and PGLs are both required for M. marinum virulence in zebrafish 309

To assess the role of PDIMs and PGLs in M. marinum virulence, we performed zebrafish 310

infection experiments. Twenty fish per group were infected with the wild type M. marinum and 311

three PDIM/PGL deficient strains, mas::Tn, fadD28::Tn, ppsE::Tn, and were monitored for 312

survival. All three mutants exhibited attenuated phenotype and were essentially avirulent 313

under the experimental conditions (Fig. 5). All fish infected with the wild type strain died by 16 314

days post infection, whereas none of the fish infected with each of the mutants died at the time 315

the experiment was terminated (30 days post infection) (Fig. 5). Histological analysis revealed 316

no microscopic liver lesions in fish infected with PDIM/PGL deficient mutants. In contrast, 317


http://iai.asm.org/

Dow

nloaded from

http://iai.asm.org/

16

typical necrotizing granulomas were observed in fish infected with the wild type M. marinum 318

(data not shown). 319

To confirm that the attenuation of mutants was caused by the deficiency of PDIMs/PGLs, 320

we performed the zebrafish infection experiment with the fadD28::Tn strain complemented 321

with fadD28 or with fadD28 plus mmpL7. As expected, the fadD28::Tn strain complemented 322

with both fadD28 and mmpL7 partially restored virulence, whereas complementation with 323

fadD28 alone did not restore virulence (Fig. 5). MmpL7 is a membrane transporter required for 324

transporting PDIMs and possibly PGLs to the cell surface. Previous works in M. tb showed that 325

the mmpL7::Tn mutant, which still produced PDIMs, was attenuated in virulence because 326

PDIMs were not properly localized (7, 8, 12). The inability to restore virulence in the strain 327

complemented with fadD28 alone is likely due to the polar effect on mmpL7 expression caused 328

by the transposon insertion. Consistent with this hypothesis, fadD28 and mmpL7 are co-329

transcribed in M. tb (7). 330

The median survival time of fish infected with the fadD28::Tn strain complemented with 331

fadD28 and mmpL7 is 19 days, which is significantly longer than the median survival time of fish 332

(12 days) infected with wild type M. marinum (P < 0.001, Log-rank test). This is consistent with 333

the partial restoration of the PDIM/PGL production in the complemented strain (Figs. 3A, 3C 334

&3D). 335

The fadD26::Tn mutant was also avirulent in zebrafish (Fig. 5). Surprisingly, we found 336

that the fadD26::Tn strain complemented with intact fadD26 remained avirulent in infected 337

zebrafish (Fig. 5). Since the fadD26::Tn mutant produced little PDIMs but intermediate level of 338

PGLs, we initially thought that the loss of PDIMs was the main cause for its attenuation. We 339


http://iai.asm.org/

Dow

nloaded from

http://iai.asm.org/

17

anticipated that complementation with fadD26, which partially restored the PDIM production, 340

would at least partially restore virulence. However, two independent zebrafish infection 341

experiments repeatedly showed that the complemented strain remained avirulent. One 342

plausible explanation is that PDIMs and PGLs are both required for M. marinum virulence. Thus 343

in the complemented strain, although the level of PDIMs was substantially restored, the PGL 344

level was further reduced (Fig. 3D), and consequently, the complemented strain still remained 345

avirulent. 346

347

PDIMs and PGLs play a role in cell wall permeability barrier 348

Since PDIMs/PGLs are components of the cell wall, the loss of PDIMs/PGLs could potentially 349

compromise the cell wall integrity. As such, we next examined whether the PDIM/PGL mutants 350

exhibited increased sensitivity to various antibiotics. We chose two mutants, fadD28::Tn and 351

fadD26::Tn, and their complemented strains for this experiment. Interestingly, we found that 352

both mutants were more sensitive to various hydrophobic drugs including chloramphenicol, 353

rifampicin, tetracycline, erythromycin, ciprofloxacin and ofloxacin, as well as several β-lactams 354

(penicillin G, cephaloridine, and cefazolin) which are hydrophilic (Table I). However, the 355

sensitivity to other hydrophilic drugs, isoniazid, ethambutol, and streptomycin, remained 356

unchanged (Table I). 357

To examine whether the increased drug sensitivity of the above mutants is caused by a 358

general increase in cell permeability, we used fluorescence spectroscopy to measure whole cell 359

accumulations of ethidium bromide and Nile Red, representatives of the hydrophilic and 360

hydrophobic compounds, respectively. The results showed that both compounds accumulated 361


http://iai.asm.org/

Dow

nloaded from

http://iai.asm.org/

18

more rapidly and to higher levels in mutant cells compared to wild type cells, indicating an 362

increase in cell wall permeability (Fig. 6). The two complemented strains were not 363

complemented in drug sensitivity and permeability assay, although there is partial 364

complementation in their ability to exclude ethidium bromide, which is consistent with the only 365

partial restoration of PDIM/PGL production in these strains (Fig. 3). 366

367

DISCUSSION 368

Despite intensive studies in recent years, the biosynthesis of PDIMs and PGLs and their role in 369

mycobacterial pathogenesis remain incompletely understood. For example, while there is 370

substantial evidence supporting the involvement of PDIMs in M. tb virulence, a recent study 371

performed whole genome sequencing of multiple lab stocks of two representative strains of 372

H37Rv (ATCC 25618 and ATCC 27294), which were isolated from the same patient in different 373

years, and found that strains of ATCC 25618 maintained in different laboratories contained a 374

frameshift mutation in mas (22), and therefore cannot produce PDIMs (23). However, these 375

strains of ATCC 25618 are fully virulent in mice (23, 29). It is not clear whether the ATCC 25618 376

strains contain compensating mutations that retain virulence despite the loss of PDIMs. 377

Similarly, the role of PGLs in M. tb virulence remains controversial (12, 35) and is complicated 378

by the fact that many laboratory and clinical strains of M. tb are naturally deficient in PGLs. 379

Because of this, we believe it is important to study the biosynthesis and biological function of 380

PDIMs and PGLs in other pathogenic mycobacteria, which may help to clarify some of the issues 381

associated with M. tb studies. In this study, we examined the biosynthesis and function of 382


http://iai.asm.org/

Dow

nloaded from

http://iai.asm.org/

19

PDIMs and PGLs in M. marinum, which is the closest genetic relative of the M. tb complex. 383

Pioneering works by Ramakrishnan and co-workers have established M. marinum as an 384

excellent model to understand various aspects of host-pathogen interactions in M. tb 385

pathogenesis [reviewed in (37)]. M. marinum and M. tb share many virulence determinants, 386

such as the ESX-1 secretion system (37). As such, knowledge we gain on PDIMs/PGLs in M. 387

marinum may be applicable to M. tb. 388

We found that disruption of multiple genes in the presumed PDIM/PGL biosynthetic 389

locus in M. marinum abolished the production of PDIMs and PGLs, and complementation of one 390

of the mutants (fadD28::Tn) restored PDIM/PGL synthesis, thus providing direct genetic 391

evidence for the involvement of this locus in PDIM/PGL synthesis. This is consistent with two 392

recent independent studies which showed that disruption of tesA, the gene upstream of 393

fadD26, also abolished the PDIM/PGL synthesis in M. marinum (2, 9). Together, these studies 394

demonstrate that although there are some differences in the chemical structures of PDIMs and 395

PGLs between M. marinum and M. tb (28), these two organisms employ similar biosynthetic 396

machinery to produce PDIMs and PGLs. In addition, our study provides evidence for the roles 397

of ppsA and ppsE in PDIM/PGL synthesis. 398

One intriguing difference between structures of PDIMs/PGLs in M. marinum and M. tb is 399

that in M. marinum (and closely related M. ulcerans), the mycocerosates are deoxtrorotary as 400

opposed to levotroroary found in M. tb and all other mycobacterial species studied (28). We 401

found that the deletion of mas in M. marinum abolished the production of PDIMs and PGLs, 402

confirming that its role in mycocerosate synthesis. Sequence alignment revealed that Mas 403

homologs from M. marinum and M. ulcerans are nearly identical (99.5% sequence identity) and 404


http://iai.asm.org/

Dow

nloaded from

http://iai.asm.org/

20

they are more distantly related to Mas from other mycobacteria such as M. tb (78.14%), M. 405

bovis including BCG (78.09%) and M. leprae (75.9%). This is consistent with the notion that Mas 406

in M. marinum and M. ulcerans synthesizes the R enantiomer of mycocerosates, while its 407

homologs in M. tb and other mycobacteria are involved in synthesizing the S enantiomer of 408

mycocerosates. Future mutagenesis study of these Mas homologs will help to identify residues 409

important for the stereochemistry of mycocerosates. 410

Deletion of fadD28 and mas in M. marinum abolished the production of both PDIMs and 411

PGLs, which is consistent with previous studies of the fadD28::Tn mutant of M. tb (7, 12). 412

However, in M. bovis BCG, fadD28::Tn and mas::Tn mutants still produced PDIMs but with 413

shorter chain mycocerosates (6, 16). It was suggested that in M. bovis BCG, another PKS 414

catalyzed the synthesis of the shorter chain mycocerosates (14). However, the identity of this 415

enzyme has not been described and the identity of these ‘short chain mycocerosates’ has been 416

questioned by others (28). Our results suggest that in M. marinum, like in M. tb, Mas is the only 417

PKS involved in mycocerosate synthesis, and that fadD28 is also involved in this pathway (Fig. 1). 418

In addition to fadD28, there are three other fadD genes in the PDIM/PGL locus (fadD26, 419

fadD22, and fadD29, Fig. 1). These genes encode fatty acyl-AMP ligases that convert long-chain 420

fatty acids to acyl-adenylates (38). Recent studies in M. bovis BCG found that fadD22 and 421

fadD29 are required for PGL synthesis only, and fadD26 is specifically involved in PDIM 422

synthesis (15, 34). Consistently, we found that fadD26 is involved in the synthesis of PDIMs but 423

not PGLs in M. marinum. FadD26 and FadD29 are thought to transfer different fatty acyl 424

substrates to the same set of PKSs, PpsA-E, for reiterative elongation, leading to phthiocerols 425

and phenolphthiocerols, respectively (Fig. 1). We found that the expression level of fadD26 426


http://iai.asm.org/

Dow

nloaded from

http://iai.asm.org/

21

affects the synthesis of PDIMs and indirectly the production of PGLs, suggesting that the 427

production of PDIMs and PGLs is connected. The ratio of these two lipids may be maintained 428

by the relative abundance of enzymes specific for each pathway (FadD26 and FadD29) and 429

enzymes common to both pathways (PpsA-E). In the wild type cells, since fadD26 and ppsA-E 430

are transcriptionally coupled, the induction of fadD26 expression will also increase the levels of 431

PpsA-E, which may lead to higher productions of both PDIMs and PGLs. Conversely, the 432

repression of fadD26 will lead to lower levels of both PDIMs and PGLs because of the decreased 433

ppsA-E expression. As such, the bacteria can efficiently regulate the amounts of PDIMs and 434

PGLs produced by controlling the expression of fadD26. This also implies that the production 435

of PGLs is dictated by the level of PDIMs, such that wild type cells cannot produce exceedingly 436

higher level of PGLs without a concurrent increase in PDIM production. It is tempting to 437

suggest that bacteria may employ this mechanism to maintain a constant ratio of PDIMs/PGLs 438

to maximize their effect in virulence. 439

Our results suggest that in M. marinum, not only are both PDIMs and PGLs essential for 440

virulence, they need to be expressed at appropriate levels such that insufficient production of 441

either lipid will compromise virulence and result in attenuation. For example, the low levels of 442

either PDIMs in the fadD26::Tn mutant or PGLs in the complemented strain resulted in 443

complete attenuation. Furthermore, the amounts of PDIMs and PGLs appear to correlate with 444

the level of virulence, as demonstrated by the partial recovery of virulence in the fadD28 445

complemented strain that only had partial restoration of PDIMs and PGLs. 446

In other mycobacteria, the relative role of PDIMs and PGLs in virulence is less clear. It 447

has been demonstrated that loss of PGLs in M. tb or M. bovis strains, which still produce PDIMs, 448


http://iai.asm.org/

Dow

nloaded from

http://iai.asm.org/

22

compromise their virulence (20, 31). Conversely, Sinsimer et al. showed that a recombinant 449

H37Rv strain that expresses intact pks15/1 thus producing PGLs was not more virulent than the 450

parental strain (22). The effect of removing PDIMs from a M. tb strain that retains PGLs has yet 451

to be evaluated. Previous studies by Sinsimer et al. (35) and Ioerger et al. (22) did not find a 452

good correlation between the production of PDIMs or PGLs with virulence among M. tb strains. 453

However, it should be noted that these comparative studies were performed on strains that are 454

not isogenic and alternative explanations cannot be excluded. Together, these studies suggest 455

that PDIMs may play a more critical role than PGLs in M. tb virulence. PGLs are probably not 456

required for but will augment virulence presumably in combination with PDIMs. 457

Previous studies of PDIM/PGL biosynthesis and their role in virulence have been 458

complicated by recent findings that laboratory H37Rv strains including the ATCC 27294 strain 459

frequently undergo spontaneous mutations leading to loss of PDIMs, which has caused errors in 460

identifying genes involved in PDIM synthesis and virulence (3, 13). This stresses the importance 461

of performing genetic complementation experiments to validate that a specific phenotype is 462

actually caused by a given mutation, as well as performing analysis of PDIM production in all 463

strains used in studies. Unfortunately, except for some genes in the PDIM/PGL biosynthetic 464

locus, complementation experiments have not been carried out for a number of other genes in 465

M. tb reported to be involved in PDIM synthesis (28), and their involvements in PDIM synthesis 466

are uncertain. 467

The mechanisms by which PDIMs and PGLs mediate virulence remain unclear. PGLs are 468

thought to inhibit the proinflammatory cytokine responses (31). PGLs of M. leprae are involved 469

in bacterial attachment and invasion of Schwann cells (26). A recent study found that PDIMs 470


http://iai.asm.org/

Dow

nloaded from

http://iai.asm.org/

23

may mediate a receptor-dependent phagocytosis of M. tb, allowing it to create a protective 471

niche by preventing phagosomal maturation (4). Elevated cell wall permeability and increased 472

antibiotic sensitivity have also been previously described in PDIM deficient mutant of M. tb (7, 473

33), and recently, in PDIM/PGL deficient mutant of M. marinum (2, 9). Consistently, we found 474

that the PDIM/PGL deficient mutants of M. marinum exhibited increased cell wall permeability 475

and antibiotic sensitivity. The weakening of the permeability barrier of mycobacteria cell wall 476

may also compromise the intracellular survival of the PDIM/PGL deficient mutants, contributing 477

to attenuation in virulence. 478

479

480

Acknowledgement: 481

We thank E.J. Rubin, Harvard University for providing the MycoMarT7 mariner transposon 482

phage and D.B. Young, Imperial College London for the pSMT3 plasmid. This work was 483

supported by the National Natural Science Foundation of China Grants 30872259 and the 484

international cooperation project of Ministry of Science and Technology (2010DFA34440) (to 485

Q.G.), and Canadian Institutes of Health Research (CIHR) Grants CCI-85667, MOP-15107, and 486

MOP-106559 (to J.L.). 487

488


http://iai.asm.org/

Dow

nloaded from

http://iai.asm.org/

24

References: 489

1. Alexander, D. C., J. R. Jones, T. Tan, J. M. Chen, and J. Liu. 2004. PimF, a 490 mannosyltransferase of mycobacteria, is involved in the biosynthesis of 491 phosphatidylinositol mannosides and lipoarabinomannan. J Biol Chem 279:18824-18833. 492

2. Alibaud, L., Y. Rombouts, X. Trivelli, A. Burguiere, S. L. Cirillo, J. D. Cirillo, J. F. 493 Dubremetz, Y. Guerardel, G. Lutfalla, and L. Kremer. 2011. A Mycobacterium marinum 494 TesA mutant defective for major cell wall-associated lipids is highly attenuated in 495 Dictyostelium discoideum and zebrafish embryos. Mol Microbiol 80:919-934. 496

3. Andreu, N., and I. Gibert. 2008. Cell population heterogeneity in Mycobacterium 497 tuberculosis H37Rv. Tuberculosis (Edinb) 88:553-559. 498

4. Astarie-Dequeker, C., L. Le Guyader, W. Malaga, F. K. Seaphanh, C. Chalut, A. Lopez, 499 and C. Guilhot. 2009. Phthiocerol dimycocerosates of M. tuberculosis participate in 500 macrophage invasion by inducing changes in the organization of plasma membrane 501 lipids. PLoS Pathog 5:e1000289. 502

5. Azad, A. K., T. D. Sirakova, N. D. Fernandes, and P. E. Kolattukudy. 1997. Gene 503 knockout reveals a novel gene cluster for the synthesis of a class of cell wall lipids 504 unique to pathogenic mycobacteria. J Biol Chem 272:16741-16745. 505

6. Azad, A. K., T. D. Sirakova, L. M. Rogers, and P. E. Kolattukudy. 1996. Targeted 506 replacement of the mycocerosic acid synthase gene in Mycobacterium bovis BCG 507 produces a mutant that lacks mycosides. Proc Natl Acad Sci U S A 93:4787-4792. 508

7. Camacho, L. R., P. Constant, C. Raynaud, M. A. Laneelle, J. A. Triccas, B. Gicquel, M. 509 Daffe, and C. Guilhot. 2001. Analysis of the phthiocerol dimycocerosate locus of 510 Mycobacterium tuberculosis. Evidence that this lipid is involved in the cell wall 511 permeability barrier. J Biol Chem 276:19845-19854. 512

8. Camacho, L. R., D. Ensergueix, E. Perez, B. Gicquel, and C. Guilhot. 1999. Identification 513 of a virulence gene cluster of Mycobacterium tuberculosis by signature-tagged 514 transposon mutagenesis. Mol Microbiol 34:257-267. 515

9. Chavadi, S. S., U. R. Edupuganti, O. Vergnolle, I. Fatima, S. M. Singh, C. E. Soll, and L. E. 516 Quadri. 2011. Inactivation of tesA Reduces Cell Wall Lipid Production and Increases Drug 517 Susceptibility in Mycobacteria. J Biol Chem 286:24616-24625. 518

10. Chen, J. M., G. J. German, D. C. Alexander, H. Ren, T. Tan, and J. Liu. 2006. Roles of Lsr2 519 in colony morphology and biofilm formation of Mycobacterium smegmatis. J Bacteriol 520 188:633-641. 521

11. Chen, J. M., S. T. Islam, H. Ren, and J. Liu. 2007. Differential productions of lipid 522 virulence factors among BCG vaccine strains and implications on BCG safety. Vaccine 523 25:8114-8122. 524

12. Cox, J. S., B. Chen, M. McNeil, and W. R. Jacobs, Jr. 1999. Complex lipid determines 525 tissue-specific replication of Mycobacterium tuberculosis in mice. Nature 402:79-83. 526

13. Domenech, P., and M. B. Reed. 2009. Rapid and spontaneous loss of phthiocerol 527 dimycocerosate (PDIM) from Mycobacterium tuberculosis grown in vitro: implications 528 for virulence studies. Microbiology 155:3532-3543. 529


http://iai.asm.org/

Dow

nloaded from

http://iai.asm.org/

25

14. Fernandes, N. D., and P. E. Kolattukudy. 1998. A newly identified methyl-branched 530 chain fatty acid synthesizing enzyme from Mycobacterium tuberculosis var. bovis BCG. J 531 Biol Chem 273:2823-2828. 532

15. Ferreras, J. A., K. L. Stirrett, X. Lu, J. S. Ryu, C. E. Soll, D. S. Tan, and L. E. Quadri. 2008. 533 Mycobacterial phenolic glycolipid virulence factor biosynthesis: mechanism and small-534 molecule inhibition of polyketide chain initiation. Chem Biol 15:51-61. 535

16. Fitzmaurice, A. M., and P. E. Kolattukudy. 1998. An acyl-CoA synthase (acoas) gene 536 adjacent to the mycocerosic acid synthase (mas) locus is necessary for mycocerosyl lipid 537 synthesis in Mycobacterium tuberculosis var. bovis BCG. J Biol Chem 273:8033-8039. 538

17. Gao, L. Y., F. Laval, E. H. Lawson, R. K. Groger, A. Woodruff, J. H. Morisaki, J. S. Cox, M. 539 Daffe, and E. J. Brown. 2003. Requirement for kasB in Mycobacterium mycolic acid 540 biosynthesis, cell wall impermeability and intracellular survival: implications for therapy. 541 Mol Microbiol 49:1547-1563. 542

18. Glickman, M. S., J. S. Cox, and W. R. Jacobs, Jr. 2000. A novel mycolic acid cyclopropane 543 synthetase is required for cording, persistence, and virulence of Mycobacterium 544 tuberculosis. Mol Cell 5:717-727. 545

19. Hall-Stoodley, L., O. S. Brun, G. Polshyna, and L. P. Barker. 2006. Mycobacterium 546 marinum biofilm formation reveals cording morphology. FEMS Microbiol Lett 257:43-49. 547

20. Hotter, G. S., B. J. Wards, P. Mouat, G. S. Besra, J. Gomes, M. Singh, S. Bassett, P. 548 Kawakami, P. R. Wheeler, G. W. de Lisle, and D. M. Collins. 2005. Transposon 549 mutagenesis of Mb0100 at the ppe1-nrp locus in Mycobacterium bovis disrupts 550 phthiocerol dimycocerosate (PDIM) and glycosylphenol-PDIM biosynthesis, producing 551 an avirulent strain with vaccine properties at least equal to those of M. bovis BCG. J 552 Bacteriol 187:2267-2277. 553

21. Humphreys, I. R., G. R. Stewart, D. J. Turner, J. Patel, D. Karamanou, R. J. Snelgrove, 554 and D. B. Young. 2006. A role for dendritic cells in the dissemination of mycobacterial 555 infection. Microbes Infect 8:1339-1346. 556

22. Ioerger, T. R., Y. Feng, K. Ganesula, X. Chen, K. M. Dobos, S. Fortune, W. R. Jacobs, Jr., 557 V. Mizrahi, T. Parish, E. Rubin, C. Sassetti, and J. C. Sacchettini. 2010. Variation among 558 genome sequences of H37Rv strains of Mycobacterium tuberculosis from multiple 559 laboratories. J Bacteriol 192:3645-3653. 560

23. Kana, B. D., B. G. Gordhan, K. J. Downing, N. Sung, G. Vostroktunova, E. E. Machowski, 561 L. Tsenova, M. Young, A. Kaprelyants, G. Kaplan, and V. Mizrahi. 2008. The 562 resuscitation-promoting factors of Mycobacterium tuberculosis are required for 563 virulence and resuscitation from dormancy but are collectively dispensable for growth in 564 vitro. Mol Microbiol 67:672-684. 565

24. Kirksey, M. A., A. D. Tischler, R. Simeone, K. B. Hisert, S. Uplekar, C. Guilhot, and J. D. 566 McKinney. 2011. Spontaneous Phthiocerol Dimycocerosate-Deficient Variants of 567 Mycobacterium tuberculosis Are Susceptible to Gamma Interferon-Mediated Immunity. 568 Infect Immun 79:2829-2838. 569

25. Middlebrook, G., R. J. Dubos, and C. Pierce. 1947. Virulence and Morphological 570 Characteristics of Mammalian Tubercle Bacilli. J Exp Med 86:175-184. 571


http://iai.asm.org/

Dow

nloaded from

http://iai.asm.org/

26

26. Ng, V., G. Zanazzi, R. Timpl, J. F. Talts, J. L. Salzer, P. J. Brennan, and A. Rambukkana. 572 2000. Role of the cell wall phenolic glycolipid-1 in the peripheral nerve predilection of 573 Mycobacterium leprae. Cell 103:511-524. 574

27. Onwueme, K. C., J. A. Ferreras, J. Buglino, C. D. Lima, and L. E. Quadri. 2004. 575 Mycobacterial polyketide-associated proteins are acyltransferases: proof of principle 576 with Mycobacterium tuberculosis PapA5. Proc Natl Acad Sci U S A 101:4608-4613. 577

28. Onwueme, K. C., C. J. Vos, J. Zurita, J. A. Ferreras, and L. E. Quadri. 2005. The 578 dimycocerosate ester polyketide virulence factors of mycobacteria. Prog Lipid Res 579 44:259-302. 580

29. Parish, T., D. A. Smith, S. Kendall, N. Casali, G. J. Bancroft, and N. G. Stoker. 2003. 581 Deletion of two-component regulatory systems increases the virulence of 582 Mycobacterium tuberculosis. Infect Immun 71:1134-1140. 583

30. Perez, E., P. Constant, A. Lemassu, F. Laval, M. Daffe, and C. Guilhot. 2004. 584 Characterization of three glycosyltransferases involved in the biosynthesis of the 585 phenolic glycolipid antigens from the Mycobacterium tuberculosis complex. J Biol Chem 586 279:42574-42583. 587

31. Reed, M. B., P. Domenech, C. Manca, H. Su, A. K. Barczak, B. N. Kreiswirth, G. Kaplan, 588 and C. E. Barry, 3rd. 2004. A glycolipid of hypervirulent tuberculosis strains that inhibits 589 the innate immune response. Nature 431:84-87. 590

32. Ren, H., L. G. Dover, S. T. Islam, D. C. Alexander, J. M. Chen, G. S. Besra, and J. Liu. 591 2007. Identification of the lipooligosaccharide biosynthetic gene cluster from 592 Mycobacterium marinum. Mol Microbiol 63:1345-1359. 593

33. Rousseau, C., N. Winter, E. Pivert, Y. Bordat, O. Neyrolles, P. Ave, M. Huerre, B. 594 Gicquel, and M. Jackson. 2004. Production of phthiocerol dimycocerosates protects 595 Mycobacterium tuberculosis from the cidal activity of reactive nitrogen intermediates 596 produced by macrophages and modulates the early immune response to infection. Cell 597 Microbiol 6:277-287. 598

34. Simeone, R., M. Leger, P. Constant, W. Malaga, H. Marrakchi, M. Daffe, C. Guilhot, and 599 C. Chalut. 2010. Delineation of the roles of FadD22, FadD26 and FadD29 in the 600 biosynthesis of phthiocerol dimycocerosates and related compounds in Mycobacterium 601 tuberculosis. FEBS J 277:2715-2725. 602

35. Sinsimer, D., G. Huet, C. Manca, L. Tsenova, M. S. Koo, N. Kurepina, B. Kana, B. 603 Mathema, S. A. Marras, B. N. Kreiswirth, C. Guilhot, and G. Kaplan. 2008. The phenolic 604 glycolipid of Mycobacterium tuberculosis differentially modulates the early host 605 cytokine response but does not in itself confer hypervirulence. Infect Immun 76:3027-606 3036. 607

36. Staropoli, J. F., and J. A. Branda. 2008. Cord formation in a clinical isolate of 608 Mycobacterium marinum. J Clin Microbiol 46:2814-2816. 609

37. Tobin, D. M., and L. Ramakrishnan. 2008. Comparative pathogenesis of Mycobacterium 610 marinum and Mycobacterium tuberculosis. Cell Microbiol 10:1027-1039. 611

38. Trivedi, O. A., P. Arora, V. Sridharan, R. Tickoo, D. Mohanty, and R. S. Gokhale. 2004. 612 Enzymic activation and transfer of fatty acids as acyl-adenylates in mycobacteria. Nature 613 428:441-445. 614


http://iai.asm.org/

Dow

nloaded from

http://iai.asm.org/

27

39. Trivedi, O. A., P. Arora, A. Vats, M. Z. Ansari, R. Tickoo, V. Sridharan, D. Mohanty, and 615 R. S. Gokhale. 2005. Dissecting the mechanism and assembly of a complex virulence 616 mycobacterial lipid. Mol Cell 17:631-643. 617

40. Tsenova, L., E. Ellison, R. Harbacheuski, A. L. Moreira, N. Kurepina, M. B. Reed, B. 618 Mathema, C. E. Barry, 3rd, and G. Kaplan. 2005. Virulence of selected Mycobacterium 619 tuberculosis clinical isolates in the rabbit model of meningitis is dependent on phenolic 620 glycolipid produced by the bacilli. J Infect Dis 192:98-106. 621

41. Yu, J., C. Niu, D. Wang, M. Li, W. Teo, G. Sun, J. Wang, J. Liu, and Q. Gao. 2011. 622 MMAR_2770, a new enzyme involved in biotin biosynthesis, is essential for the growth 623 of Mycobacterium marinum in macrophages and zebrafish. Microbes Infect 13:33-41. 624

625

626

627


http://iai.asm.org/

Dow

nloaded from

http://iai.asm.org/

28

Table I. Minimal inhibitory concentrations (MIC, µg/ml) of various antibiotics against M. marinum 628

strains. CEP: cephalodrine, CEF: cefazolin, PEN: penicillin, AMP: ampicillin, CAM: chloramphenicol, ERY: 629

erythromycin, TET: tetracycline, RIF: rifampicin, OFL: ofloxacin, CIP: ciprofloxacin, STREP: streptomycin, 630

INH: isoniazid, EMB: ethambutol 631

CEP CEF PEN AMP CAM ERY TET RIF OFL CIP STREP INH EMB

WT 5 20 64 512 20 32 4 0.96 4 0.5 4 4 1

fadD28::Tn 2.5 10 32 256 10 8 2 0.48 2 0.25 4 4 1

fadD28 C 2.5 10 32 256 10 8 2 0.48 2 0.25 4 4 1

fadD26::Tn 2.5 10 32 256 10 8 2 0.48 2 0.25 4 4 1

fadD26 C 2.5 10 32 256 10 8 2 0.48 2 0.25 4 4 1

fadD28C: fadD28::Tn +pFadD28/MmpL7; fadD26C: fadD26::Tn +pFadD26 632

633

634

635

636


http://iai.asm.org/

Dow

nloaded from

http://iai.asm.org/

29

Figure Legend 637

Figure 1. (A) The PDIM/PGL biosynthetic locus in M. marinum. Each arrow represents the 638

open reading frame of predicted genes. The genes disrupted by transposon insertions are 639

highlighted in red and the transposon insertion sites are indicated by filled triangles. (B) 640

Schematic representation of PDIM/PGL synthesis in M. marinum. Enzymatic steps catalyzed by 641

disrupted genes (fadD26, ppsA-E, fadD28, mas) are emphasized. 642

643

Figure 2. Colony morphology and cording phenotype of PDIM/PGL deficient mutants of M. 644

marinum. The PDIM/PGL deficient mutants exhibited altered colony morphology (A) and 645

were defective in cord formation (B). M. marinum WT strain (a), representative PDIM/PGL 646

deficient mutants, fadD28::Tn (b), mas::Tn (c), and fadD28::Tn mutant complemented with the 647

cloning vector (d), intact fadD28 gene (e) or fadD28+mmpL7 (f). Complementation of 648

fadD28::Tn with fadD28+mmpL7 restored the WT colony morphology and partially restored the 649

cording phenotype. 650

651

Figure 3. 2D-TLC analysis of PDIMs and PGLs. For analysis of PDIMs, equal amount of apolar 652

lipids were developed with petroleum ether/ethyl acetate (98:2, 3×) in the first dimension and 653

petroleum ether/acetone (98:2) in the second dimension. Lipids were visualized by spraying 654

plates with 5% phosphomolybdic acid followed by gentle charring of the plates (A, B) or 655

phosphorimaging (C). For analysis of PGLs, the apolar lipid extract was developed with 656

chloroform/methanol (96:4, v/v) in the first dimension and toluene/acetone (80:20, v/v) in the 657

second dimension, followed by charring with α-naphthol (A, B) or visualized by 658


http://iai.asm.org/

Dow

nloaded from

http://iai.asm.org/

30

phosphorimaging (C). (A) PDIM and PGL analysis of the fadD28::Tn mutant and its 659

complemented strain. The fadD28::Tn mutant failed to synthesize PDIMs and PGLs. The 660

complemented strain fadD28C, which is the fadD28::Tn mutant complemented with 661

fadD28+mmpL7, partially restored the production of PDIMs and PGLs. (B) PDIM and PGL 662

analysis of the fadD26::Tn mutant and its complemented strain. The fadD26::Tn mutant 663

produced residual PDIMs and intermediate levels of PGLs. Complementation with fadD26 664

(fadD26C) partially restored the production of PDIMs but caused further reduction of PGLs. (C) 665

Quantitative 2D-TLC analysis. Bacterial cultures labeled with [14C]propionate were subjected 666

to 2D-TLC analysis as described for (A) and (B) and visualized by phosphorimaging. (D) 667

Quantitation of the levels of PDIMs and PGLs in different strains. Data were determined from 668

(C) by phosphorimaging analysis using Typhoon9400TM imager. The data were normalized to 669

that in the wild type strain. 670

671

Figure 4. qT-PCR analysis. The expression levels of fadD26 and its downstream genes ppsA and 672

ppsE were determined. Relative expression data (fold change relative to wild type strain) were 673

obtained using specific primers for each gene and normalized to the level of sigA. Results are 674

form three independent experiments. 675

676

Figure 5. Survival of zebrafish infected with M. marinum strains. Separate tanks, each 677

containing 20 fish infected with indicated strain at 104 CFU bacteria per fish, were monitored 678

for mortalities over a 30-day period. A Kaplan-Meier curve was calculated for animals infected 679

with each strain. All fish infected with PDIM/PGL deficient mutants (mas::Tn, fadD28::Tn, 680


http://iai.asm.org/

Dow

nloaded from

http://iai.asm.org/

31

ppsE::Tn, and fadD26::Tn) as well as PBS survived. Complementation of fadD28::Tn with 681

fadD28+mmpL7 (fadD28C2) partially restored virulence (p < 0.001, compared to WT infected 682

fish) while complementation with fadD28 alone (fadD28C1) did not restore virulence. 683

Complementation of fadD26::Tn with fadD26 (fadD26C) did not restore virulence. 684

685

Figure 6. Uptake of ethidium bromide (EtBr) and Nile Red by M. marinum strains. Whole cell 686

accumulations of EtBr and Nile Red by M. marinum were measured by fluorescence 687

spectroscopy. Results are from three independent experiments (mean ± SEM). The fadD26::Tn 688

and fadD28::Tn strains exhibited more rapid and higher levels of accumulation for EtBr and Nile 689

Red than the WT strain. Complementation partially restored the accumulation of EtBr but had 690

no effect on Nile Red uptake. fadD26C: fadD26::Tn+pFadD26, fadD28C: 691

fadD28::Tn+pFadD28/Mmpl7. 692


http://iai.asm.org/

Dow

nloaded from

http://iai.asm.org/


http://iai.asm.org/

Dow

nloaded from

http://iai.asm.org/

Nile Red

fadD28 CfadD26::Tn

WTfadD28::Tn


http://iai.asm.org/

Dow

nloaded from

http://iai.asm.org/

PDIMs and PGLs are both re quired for virulence of...

Documents

Transcript of PDIMs and PGLs are both re quired for virulence of...