Original article

Mutation in mce operons attenuatesMycobacterium tuberculosis virulence

Andrea Gioffré a, Eduardo Infante c, Diana Aguilar c, María De la Paz Santangelo a,Laura Klepp a, Ariel Amadio b, Virginia Meikle a, Ignacio Etchechoury a, María Isabel Romano a,

Angel Cataldi a, Rogelio Pando Hernández c, Fabiana Bigi a,*a Institute of Biotechnology, CICVyA-INTA, Los Reseros y Las Cabañas, 1712 Castelar, Argentina

b Institute of Agricultural Microbiology and Zoology, CICVyA-INTA, Castelar, Argentinac Experimental Pathology Section, Department of Pathology, National Institute of Medical Sciences and Nutrition, ‘Salvador Zubirán’, México

Received 20 April 2004; accepted 4 November 2004

Available online 01 February 2005

Abstract

On the Mycobacterium tuberculosis genome there are four mce operons, all of which are similar in sequence and organization, and code forputatively exported proteins. To investigate whether Mce proteins are essential for virulence, we generated knock-out mutants in mce1, mce2and mce3 operons of M. tuberculosis and evaluated their ability to multiply in a mammalian host. The allelic replacement was confirmed ineach mutant strain by Southern blotting. RT-PCR experiments demonstrated the lack of in vitro expression of mutated genes in Dmce1 andDmce2 mutants. On the other hand, no expression of mce3 was detected in either the wild-type or mutant strains. Similar doubling time andgrowth characteristics in in vitro culture were observed for mutants and parental strains. The intratracheal route was used to infect BALB/cmice with the Dmce3, Dmce2 and Dmce1 mutants. Ten weeks after infection, all mice infected with the Dmce mutants survived, while thoseinfected with the wild-type strain died. This long survival correlated with very low counts of colony-forming units (CFU) in the lungs.Dmce1-infected mice developed very few and small granulomas, while animals infected with Dmce3 or Dmce2 mutants showed delayedgranuloma formation. Mice infected with Dmce1 did not develop pneumonia, while animals infected with Dmce3 and Dmce2 mutants showedsmall pneumonic patches. In spleens, bacterial counts of mutant strains were less reduced than in lungs, compared with those of wild-type. Incontrast, no such attenuation was observed when the intraperitoneal route was used for infection. Moreover, Dmce1 mutants appear to be morevirulent in lungs after intraperitoneal inoculation. In conclusion, mce operons seem to affect the virulence of M. tuberculosis in mice, depend-ing on the route of infection. Hypotheses are discussed to explain this last issue. Thus, mutants in these genes seem to be good candidates forvaccine testing.© 2005 Elsevier SAS. All rights reserved.

Keywords: Mycobacterium tuberculosis; Attenuation; Virulence

1. Introduction

Tuberculosis (TB), a chronic illness caused by Mycobac-terium tuberculosis, is still a major worldwide disease.According to the World Health Organization, TB is a consid-erable public health problem in Latin America, Asia andAfrica. Over the last few years, an increase in the incidenceof TB has been observed, attributed to weak control pro-

grams, the AIDS pandemic, which predisposes individuals todevelop TB, and to the appearance of M. tuberculosis strainsresistant to first-line antibiotics [1].

An important first step in bacterial infection is the pres-ence of specific bacterial surface components that provokephagocytic ingestion of the bacteria by host cells. M. tuber-culosis has been shown to have ligands that bind to extracel-lular matrix proteins like fibronectin [2] and proteoglycans[3]. The third gene of mce1 operon codes for a protein that,after transformation of a non-pathogenic strain of Escheri-chia coli, confers the ability to invade macrophages and HeLacells [4]. Paralogous genes are present in the mce2, 3 and 4

* Corresponding author. Tel.: +54 11 4621 1447x0199;fax: +54 11 4481 2975.

E-mail address: fbigi@cicv.inta.gov.ar (F. Bigi).

Microbes and Infection 7 (2005) 325–334

www.elsevier.com/locate/micinf

1286-4579/$ - see front matter © 2005 Elsevier SAS. All rights reserved.doi:10.1016/j.micinf.2004.11.007

operons. This gene family was termed mceA (mce1A, mce2A,mce3A and mce4A). Upstream of mceA, there are two genes(yrbEA and yrbEB) that code for integral membrane proteins,and downstream of it, there are five genes (mceB to mceF)that code for proteins with signal sequences or hydrophobicstretches at the N-terminus (http://genolist.pasteur.fr/TubercuList). These features are consistent with cell surfacelocalization and the proposed role of Mce proteins in the inva-sion of host cells or in host–pathogen interactions. The mcegenes are conserved in all members of the M. tuberculosiscomplex, mce3 being absent in Mycobacterium bovis, a sub-group of Mycobacterium africanum, Mycobacterium microtiand Mycobacterium pinnipedii [5]. Orthologous mce genesare widely distributed throughout the genus Mycobacterium[6], even in non-pathogenic species.

Mce1 proteins are recognized by TB patients’ antibodies[7], indicating the expression of genes and antigenicity ofMce proteins. Harboe et al. [8] demonstrated expression ofthe mce1 operon proteins using cell extracts from M. tuber-culosis and M. bovis BCG. On the other hand, Panigada et al.[9] identified the existence of a promiscuous T-cell epitope inMce proteins.

Previous studies have demonstrated the in vivo expressionof Mce proteins in mycobacterial species: using the SCOTStechnology (selective capture of transcribed sequences), Houet al. [10] found that genes similar to two mce 4 genes areexpressed by Mycobacterium avium during growth in mac-rophages; whereas Graham and Clark-Curtiss [11] identifiedmceD1 (from mce1 operon) as a differentially expressed genein M. tuberculosis infecting macrophages. However, littleprogress has been made concerning the physiological func-tion of mce operons. A vast majority of mce genes are notessential for bacterial growth in vitro, as defined by Sassettiet al. [12].

In most bacteria, directed mutation of a gene is a standardstrategy to define its function and pathogenic relevance. How-ever, the extreme difficulty in creating defined mutants of M.tuberculosis has delayed identification of its virulence fac-tors. Improved methods for site-directed mutagenesis by inser-tion of antibiotic-resistant cassettes in slow-growing myco-bacteria have been created [13,14]. Pelicic et al. [15] havedeveloped an efficient system for targeted genetic disruptionthat uses both the counterselective properties of the sacB geneand a mycobacterial thermosentitive origin of replication, bothof which enable the positive selection of insertion mutants.

To investigate whether Mce proteins are essential for invitro and in vivo growth of M. tuberculosis, we generatedinsertional mutants in mce1, mce2 and mce3 operons of M.tuberculosis and evaluated their ability to multiply a in mam-malian host.

2. Materials and methods

2.1. Bacterial strains and culture conditions

All cloning steps were performed in E. coli DH5a E. coliwas grown on liquid or solid Luria–Bertani medium. M. tuber-culosis H37Rv was grown either in Middlebrook 7H9 liquidmedium supplemented with ADC (albumin 0.5%, glucose0.4%), 0.05% Tween-80 and 0.5% glycerol (M7H9-ADC-GT) or Middlebrook 7H11 medium supplemented with ADCand 0.5% glycerol (M7H11-ADC-G). When necessary, kana-mycin (20 µg ml–1), hygromycin (50 µg ml–1) or sucrose (2%)was added.

2.2. Electrocompetent M. tuberculosis H37Rv cells

M. tuberculosis H37Rv was prepared for electroporationaccording to Parish and Stocker [16]. Briefly, cells were grownin M7H9-ADC-GT without shaking, at 37 °C, to an opticaldensity at 600 nm of 0.5–0.8, then washed three times with1 volume of cold 10% glycerol (made up with HPLC gradewater) and concentrated 1/100.

2.3. PCR conditions

Sequences from the M. tuberculosis genome database(http://genolist.pasteur.fr/TubercuList) were used to design theprimers (Table 1). The PCR conditions were as follows: dena-turation for 3 min at 94 °C, 35 cycles of 1 min at 94 °C, 1 minof annealing step, extension at 72 °C for 1 min kb–1 of theexpected amplified products, and one final extension of 10 minat 72 °C. A touchdown method, in which the annealing tem-perature of each subsequent cycle was run at 1 °C less thanthe preceding cycle, was used for thermal cycling [17].

2.4. Allelic replacement and construction of mce mutantsof M. tuberculosis

Three genomic regions containing yrbE1B (second ORFof mce1 operon), mce2A (third ORF of mce2 operon), mce3A

Table 1Primers and plasmids used in this study

Primer pair Sequencea Annealingtemperatureb (°C)

Size of PCRproduct (bp)

Cloned into

Mce1 up Mce1 low tctagactggtcggtggctttctcttc tctagaggttcgggttgtcgttgata 58–50 4.34 PGemTMce1Mce2 up Mce2 low tctagaggtggtttcgtcttcagtgtc tctagacagcgttctcattgcggtgtg 58–50 3.43 pGemTMce2U 0.4K12 6Lmbss atgaaggcaaacaccacg ccagtcctcgctgtaggt 61–53 5.57 pGemTMce3RTmce1J2f Rtmce1J2r tacctggacgctattcagc tcggagaacttagccacc 56–55 0.78 Used for RT-PCR of mce1RTmce2f Rtmce2r cgacatggctttcacctctg ccgaccccacatcaatcac 59–54 0.75 Used for RT-PCR of mce2RTmce3f Rtmce3 caacacccgcgagattcag gttcttcgaatgcagtacc 58–53 0.95 Used for RT-PCR of mce3

a Restriction enzyme site added at the end of each primer is underlined.b Indicates the range of annealing temperatures used during touch-down cycling program (see Section 2).

326 A. Gioffré et al. / Microbes and Infection 7 (2005) 325–334

(third ORF of mce3 operon) and about 2 kb flanking 5´ and3´ regions, were obtained by PCR from M. tuberculosisH37Rv total DNA. The amplified fragments were cloned inpGem-T plasmid (Promega), and the plasmids obtained werepGEMyrbE1BR, pGEMmce2AR, and pGEMmce3AR, re-spectively. To eliminate an additional PstI site frompGEMmce3AR insert, the plasmid was digested by AgeI andNdeI, blunted and religated. Mutant alleles of mce3A andmce2A were generated by insertion of a kanamycin cassettefrom pUC4K (Amersham Biosciences) into a unique PstI siteinternal to mce3A and mce2A in pGEMmce3AR andpGEMmce2AR, respectively. yrbE1B was interrupted byinserting a cassette conferring hygromycin resistance frompUC-Hy into a unique KpnI site internal to yrbE1B inpGEMyrbE1BR. To generate transforming plasmids,SphI/SacI mce3A3::km, XbaI-mce2A::km and XbaI-yrbE1B::hyg inserts were released from pGem-T derivativevectors. SphI/SacI mceA3::km insert was blunted by filling inwith Klenow DNA polymerase (Promega). Fragments weresubcloned separately into blunt-ended BamHI linearizedpPR27 shuttle vector [15] to give pPR27Dmce3A, and at theXbaI site of pPR27, generating pPR27Dmce2A andpPR27Dmce1.

For transformation, 1 µg of each construction was addedto 200 µl of M. tuberculosis H37Rv electrocompetent cellsby electroporation using a Bio-Rad Gene Pulser, as describedpreviously [15]. Five milliliters of M7H9-ADC broth with-out antibiotic was added immediately after electroporation,and the bacteria were incubated at 32 °C for 48 h before plat-ing on M7H11-ADC-G media containing kanamycin for bothM. tuberculosis (pPR27D mceA3) and (pPR27D mceA2) trans-formants and hygromycin for M. tuberculosis (pPR27DyrbE1B) transformant. Recombinant colonies were observedafter 5 weeks. Four different sucrose-sensitive colonies werepicked up from each transformation and cultivated in liquidmedium with adequate antibiotic to near saturation at 32 °C.Individual cultures were plated on M7H11-ADC-G plus 2%sucrose antibiotic-containing medium and cultivated at 39 °C.Mycobacterial colonies were observed after 4 weeks. Tensucrose-resistant isolates from each experiment were screenedby colony PCR to observe whether allelic replacementoccurred. DNA samples were prepared by re-suspending bac-teria in 100 µl of water and boiling for 30 min. The primersand PCR conditions were as described above.

Selected clones were subcultured onto fresh 7H11-ADC-G-antibiotic plates and grown for an additional 3–4 weeksprior to further evaluation. To confirm allelic replacement,chromosomal DNA was prepared from these cultures accord-ing to van Soolingen et al. [18] and digested with XhoI foryrbE1B::hyg, SalI for mce2A::km and NotI for mce3A::kmclones and then analyzed by Southern blotting, using the wild-type gene as probes.

2.5. RNA isolation

The pellet from a 20-ml culture was re-suspended in 1 mlof Tri reagent (Sigma-Aldrich), transferred to 2-ml screw-

cap tubes containing glass beads and agitated in a ribolyserzirconium bead beater (Bio 101). The aqueous phase was sub-sequently extracted twice with chloroform and precipitatedwith isopropanol. Remaining DNA in RNA samples wasdigested with Dnase I (Invitrogen) for 1 h at room tempera-ture, followed by Dnase I inactivation at 65 °C for 5 min.RNA samples were re-purified by using an RNA purificationkit according to the manufacturer’s instructions (RNAesyQiagen). Examination of the purified total RNA by 1% aga-rose gel electrophoresis revealed prominent 23S, 16S and 5Sribosomal bands.

2.6. RT-PCR

Synthesis of the first strand of cDNA was performed using1 µg total RNA from M. tuberculosis H37Rv and mutantstrains as template and randomhexamers (Invitrogen) as prim-ers, following the indications in the SUPERScript Preampli-fication System for First Strand cDNA Synthesis kit (LifeTechnologies). A 5-µl aliquot of the cDNA synthesis reactionwas amplified with primers RTmce1J2f and RTmce1J2r formce1 operon, primers RTmce2f and RTmce2r for mce2operon, and primers RTmce3f and RTmce3r for mce3 operon.The mce1 primers target mce1B and mce1C genes, mce2 prim-ers target yrb2B and mce2A genes, and mce3 primers targetyrb3B and mce3A. These primers were designed by Kumar etal (2003), and the sequences are shown in Table 1. Amplifi-cation conditions were as follows: denaturation at 94 °C for2 min, followed by 30 cycles at 94 °C for 1 min, annealingtemperatures depending on the template (Table 1) of touch-down type, and extension at 72 °C for 1 min. Amplificationproducts were detected in 1% agarose gels.

2.7. In vitro growth studies

Cultures of the parent strain H37Rv and the Dmce3, Dmce2and Dmce1 mutants were washed with 1× phosphate-bufferedsaline (PBS) and stored at –70 °C. Colony-forming units weredetermined in each bacterial stock and used to inoculate freshM7H9-ADC-GT at 104 CFU ml–1. Cultures were incubatedwithout shaking at 37 °C for 20 days. At days 1, 3, 6, 8, 9 and13, aliquots were removed from each culture, and optical den-sity was determined at 600 nm.

2.8. Mouse infections

To induce pulmonary TB in mice, a previously describedmodel of intratracheal inoculation was used [19–21]. MaleBALB/c mice were anesthetized with 56 mg kg–1 of intrap-eritoneal pentothal. The trachea was exposed via a small mid-line incision, and 100 µl of PBS with 2.5 × 104 suspendedviable bacteria was injected. The incision was then closedwith sterile silk, and the mice were maintained in a verticalposition until the effect of anesthesia passed. The wholeexperiment was repeated, and results of two experiments werepooled. Twenty mice in each group were left undisturbed,

327A. Gioffré et al. / Microbes and Infection 7 (2005) 325–334

and survival was recorded up to day 120. Eight mice (all sur-vivors, if less than 8) were killed by exsanguination at 1, 3, 7,14, 21, 28, 60 and 120 days postinfection (dpi). Lungs of fouranimals were removed and disrupted in a Polytron homog-enizer (Kinematica, Luzern, Switzerland) in isotonic saline.Bacterial loads in lung and spleen homogenates were quan-tified by serial dilution on 7H11-ADC and cultured at 37 °Cfor 28 days. The values are expressed as the mean ± S.D. ofCFU for eight mice. Lungs of the remaining four animalswere used for histopathological analysis.

In addition, and to test another infection model, eightBALB/c male mice per group were infected intraperitoneallywith 106 CFU of parental M. tuberculosis H37Rv or Dmce1,Dmce2 and Dmce3 mutant strains suspended in 100 µl of 1×PBS. Four mice were assessed at 15 and 30 dpi, and colonycounting on spleens and lungs was assessed as describedabove.

2.9. Histopathological analysis of lungs

Lungs of four mice were perfused with 10% formalin dis-solved in PBS via the trachea, immersed for 24 h in the samefixative, and embedded in paraffin. Five-micrometer trans-verse sections, taken through the hilus, were stained withhematoxylin and eosin. Both the area in square micrometersoccupied by granulomas and the percentage of lung surfaceaffected by pneumonia were determined using an automatedimage analyzer (Q Win Leica, Milton Keynes, UK) [20].Granuloma was defined as an area of nodular inflammatorylesions constituted by lymphocytes and macrophages. In turn,pneumonia was defined as the presence of inflammatory cellsand protein exudate in alveolar spaces producing lung con-solidation.

3. Results

3.1. Construction of mce mutants of M. tuberculosis

Site-directed mutant strains of M. tuberculosis in mce1,mce2 and mce3 operons were created by two-step mutagen-esis using the pPR27 shuttle plasmid, which carries the coun-terselectable marker sacB and a thermosensitive origin of rep-lication. We chose yrbE1B, mce2A and mce3A from mce1,mce2 and mce3 operons, respectively, as target genes for singlegenetic disruption in independent bacterial clones (Fig. 1A).As described previously [15], a double-crossover genereplacement event occurred in more than 90% of sucrose-resistant isolates. Allelic replacement was confirmed in eachmutant clone by Southern blotting. As described in (Fig. 1B),the allelic exchange mutants showed a single hybridizing frag-ment about 1.2 or 1.5 kb longer than the wild-type strain, andwhich corresponded to the kanamycin or the hygromycin cas-sette, respectively. The yrbE1B::hyg, Mce2A::km andMce3A::km mutants were designated Dmce3, Dmce2 andDmce1, respectively.

3.2. In vitro characterization of mce mutants

To determine whether mce operon disruptions introducealterations during in vitro growth, growth curves of the paren-tal strain and Dmce1, Dmce2 and Dmce3 mutants were com-pared under standard culture conditions (such as liquid mediaM7H9-ADC-GT and solid media M7H11-ADC-G). Allassayed strains showed similar doubling time and growth char-acteristics throughout culture (Fig. 2).

The expression of mce genes in each mutant strain wasinvestigated by RT-PCR. The primer pairs were designed toamplify a region encompassing two neighboring genesaccording to the operonic nature of mce genes [22]. The resultsobtained demonstrate the lack of in vitro expression of the

Fig. 1A. Schematic drawing of mce1, mce2 and mce3 operons. The insertions of antibiotic-resistant cassettes are indicated (hyg and km).

328 A. Gioffré et al. / Microbes and Infection 7 (2005) 325–334

corresponding genes in mce1 and mce2 mutants (Fig. 3). Thetwo amplifications gave bands of the expected sizes whencDNA from the wild-type strain was used. On the other hand,no expression of mce3 genes was detected in either the wild-type or mutant strains.

3.3. Growth characteristics of the mce mutants in mice

To examine the effect of the Dmce3, Dmce2 and Dmce1mutants in vivo, we used the intratracheal route to infectBALB/c mice [19–21], and determined animal survival, bac-terial growth and lung histopathology. Five weeks after infec-tion, mice infected with H37Rv started to die, and after9 weeks of infection, there were no animal survivors. Con-versely, all mice infected with each of the Dmce mutants sur-vived throughout the experiment (Fig. 4A). This long sur-vival coexisted with very low colony-forming units in thelungs (Fig. 4B). Animals infected with Dmce2 and Dmce3

showed a moderate increase in the bacillary load at day 21,but after this point, bacterial counts of the three mce mutantsdid not progress anymore and remained low. In contrast, miceinfected with H37Rv showed a progressive increment ofcolony-forming units, which at 28 and 60 dpi, were five tosixfold higher than the colony-forming units of mutants. Thebacterial load of mce mutants in spleen was also reduced,although the ratio of mutant/wild-type CFU was higher thanin lungs (Fig. 4C).

Two weeks after infection, mice infected with H37Rvstarted to produce lung granulomas, which increased theirsize progressively, raising a peak at day 28. Then, 60 daysafter infection, granulomas decreased their size seven times(Fig. 5A). In contrast, mice infected with Dmce1 developedvery few and small granulomas, while animals infected withDmce3 and Dmce2 mutants showed a delayed granuloma for-mation and a progressive increase in their size. Animalsinfected with H37Rv strain started to develop progressivepneumonia 3 weeks after infection, reaching a peak at day60 of infection, affecting 60% ± 15% of the total lung surface(Fig. 5B). In comparison, mice infected with Dmce1 did notdevelop pneumonia, while animals infected with Dmce3 andDmce2 showed small pneumonic patches, which appearedmuch later and were significantly smaller than those pro-duced by H37Rv. Representative lung histopathologicalimages are shown in Fig. 5C.

To establish whether the route of infection can affect per-sistence of mutants in mice, groups of BALB/c mice weregiven intraperitoneal injections of 106 CFU of each mutantand parental strain. On days 15 and 30 dpi, four mice foreach group were sacrificed, and viable bacteria present inspleen and lungs were determined. Data from mutant strainswere compared with those of wild-type. At 15 dpi, no differ-

Fig. 1B. Southern blot analysis of chromosomal DNA from Dmce1, Dmce2 and Dmce3 mutants and the parental wild-type strain. Lane 1, parental H37Rv; lane2 Dmce1 (A), Dmce2 (B) and Dmce3 (C). In each panel, open arrows indicate the position of the hybridizing bands corresponding to wild-type H37Rvsequences, whereas solid arrows indicate bands containing the mutant allele. Genomic DNA was digested with XhoI (A), SalI (B) and NotI (C) and hybridizedto a yrbE1B (A), mce2A (B) and 3´region of mce3A (C) probes.

Fig. 2. Effect of mce mutation on in vitro growth. M. tuberculosis H37Rv(--♦ --), Dmce1 (--• --), Dmce2 (--m--) and Dmce3 (--"--) mutants were ino-culated into 7H9-ADC-GT medium, and the OD600 was measured at diffe-rent days.

329A. Gioffré et al. / Microbes and Infection 7 (2005) 325–334

ences were observed in colony-forming unit values of eitherthe parental H37Rv or mutants strains collected from spleens(Fig. 6A). However, a reduction of approximately 70% of thebacterial load was detected on day 30 in the spleen of ani-mals inoculated with Dmce2. In contrast, bacterial counts ofDmce1 showed an increase of around 50%, and the growth ofDmce3 was not modified. In lungs (Fig. 6B), a reduction inbacterial load of Dmce2 was observed at 30 dpi, and bacterial

counts of Dmce1 showed an increase of around 50% and six-fold at 15 and 30 dpi, respectively. Again, the colony-forming units of Dmce3 in lung were similar to those of theparental strain.

4. Discussion

In this study, we produced three mutant strains of M. tuber-culosis by disruption of yrbE1B, mce2A and mce3A genes,and we evaluated the effect of these mutations on M. tuber-culosis growth under in vitro and in vivo conditions. Similargrowth rates of parental and mce mutant strains of M. tuber-culosis were observed when maintained in standard culturemedia. Thus, mutations of the mce3A, mce2A and yrbEB1genes of M. tuberculosis do not appear to compromise theirin vitro growth under the conditions studied. By RT-PCR, itwas observed that expression of mce1 and mce2 genes is elimi-nated in mutant strains. However, it was not possible to detectmRNA of the mce3 genes in either the mutants or wild-typestrains, probably due to the strong repression exerted bymce3R negative regulator [22]. No expression in vivo of mce3was observed by Kumar et al. [23] either.

Our data demonstrated that the inactivation of yrbE1B,mce2A, and mce3A affected the in vivo growth of M. tuber-culosis as assessed by reduction of colony-forming units inthe lungs and spleen of mice inoculated by the intratrachealroute. The three mutant strains showed some differences inthe bacterial load they reached, but 21 dpi, the colony-forming unit load per lung declined strongly. Moreover, thehistopathologic study revealed very limited tissue damage(pneumonia), with delayed granuloma formation. Interest-ingly, mce1 mutant induced occasional and very small granu-lomas, without pneumonia.

Our findings are consistent with previous observations sup-porting the involvement of mce operons in the host–pathogeninteraction. Disruption of the mce1 in M. bovis BCG reducedits ability to invade HeLa cells [24]. The mce1 operon wasmutated by transposon mutagenesis, and the resulting strainswere attenuated in a murine model [12]. In addition, a72-amino-acid region of Mce1 was also shown to be respon-sible for the association of HeLa cells and internalization [25].

When mutant strains were assayed by intraperitonealinoculation, the general attenuation described previously wasnot observed. In those experiments, Dmce2 mutant showed amoderate reduction in bacterial counts and colony-formingunits of Dmce3 were essentially identical to those of the paren-tal strain. Conversely, the bacterial burden in spleens of miceinoculated/infected with Dmce1 was higher than that of miceinoculated with the parental strain. It should be noted thatdifferences in spleen bacterial burden between mce mutants(Dmce1 and Dmce2) and the virulent parental strain, althoughnot significant, were also observed in two additional indepen-dent experiments. A significant increase in colony-formingunits was observed in lungs of mice inoculated with Dmce1mutant compared with those from mice infected with wild-

Fig. 3. Expression of mce genes as determined by RT-PCR. (A) RT-PCRusing mce1 primers. Lanes 1–2, RT-PCR using RNA extracted from Dmce1mutant; lanes 3–4, RT-PCR using RNA extracted from Dmce2 mutant; lanes5–6, RT-PCR using RNA extracted from Dmce3 mutant; lanes 7–8, RT-PCRusing RNA extracted from wild-type strain; lane 9, control of PCR amplifi-cation using DNA from wild-type strain; lane 10, negative control of rea-gents; and lane 11, molecular weight markers. Lanes 1, 3, 5, and 7, reactionwith reverse transcriptase treatment; lanes 2, 4, 6, and 8, reaction withoutreverse transcriptase treatment. (B) RT-PCR using mce2 primers. Distribu-tion of lanes is the same as in panel A. (C) RT-PCR using mce3 primers.Distribution of lanes is the same as in panel A.

330 A. Gioffré et al. / Microbes and Infection 7 (2005) 325–334

type. This last finding is coincident with observations by Shi-mono et al. [25], who demonstrated that the disruption of mce1operon in M. tuberculosis led to hypervirulence of the mutantstrain and aberrant granuloma formation in mice. Theseauthors described that the number of colony-forming unitsrecovered from different organs was significantly greater formice infected with the mce1 mutant strain than that in miceinfected with the parental strain [25].

Our results indicate that the behavior of mce mutants inboth lung and spleen in mice depends on the routes used toassess their persistence. Our hypothesis is that intratrachealinoculation may deliver the bacilli directly to the lower res-piratory tract and then to alveolar macrophages, where a morehostile milieu should exist. On the other hand, when the bac-teria enter the organism via a systemic route, such as the intra-peritoneal route, or the intravenous route, as used by Shi-

mono et al. [25], the bacteria may reach less stringent niches(for example epithelial cells), where both attenuated and viru-lent bacilli could equally persist or multiply. In connectionwith this idea is the observation that mce mutants are lessrestricted in spleen after intratracheal inoculation. In a recentwork, different virulence levels that depended on the organstudied and the infection route used were observed after inocu-lation with different M. tuberculosis strains [26]. Also, otherauthors observed that certain mutations affect the organ dis-tribution of M. tuberculosis [27].

Based on previous microarray data (unpublished results),which indicated that mutation by insertion of an antibioticcassette exerted a polar effect on downstream mce genes, wewanted to complement mce mutant strains with the entireoperon. Unfortunately, Southern blot experiments showed thatthe operon sequences, cloned in the integrative mycobacte-

Fig. 4. Pathogenicity of mce mutant strains after intratracheal inoculation.(A) Survival of BALB/c mice (20 mice per strain) infected by intratracheal injection of M. tuberculosis H37Rv, Dmce1, Dmce2 and Dmce3 mutants. (B) On 1,3, 7, 14, 21, 28, 60 and 120 dpi, mice were sacrificed, and viable bacteria present in lungs were counted. (C) Bacterial counts in spleens. The results areexpressed as the mean colony-forming units ± standard deviations in four mice.

331A. Gioffré et al. / Microbes and Infection 7 (2005) 325–334

rial plasmid pYUB178, were unstable when introduced in M.tuberculosis (data not shown). However, it is important to con-clude that in our hands, the mce operons seemed to be involvedin M. tuberculosis virulence, in spite of the fact that the con-tribution of every individual gene is not known.

Mce proteins have a predicted membrane or extracellularlocalization, a characteristic that has been experimentallydemonstrated (unpublished data) for some of these proteins.Based on their topology, it is possible to speculate that thefunction of Mce proteins is related to transmembrane trans-port, cell wall synthesis, or other functions involved in therelationship with the environment or the host. In that sense,the loss of one of these structures may result in some form ofattenuation. The bacteria could no longer uptake nutrients ormetabolic precursors, expel host substances that are toxic forthe bacteria, or synthesize complex lipids that protect the bac-teria against the host response. But, although mce operonsare highly homologous, the infection kinetics of mce mutantsin spleen of mice was different. Perhaps these operons playspecific roles during the infection process. Regulatory mecha-

nisms that control mce expression have recently been dem-onstrated [22,27]. Santangelo et al. [27] found that a TetRfamily transcriptional regulator downregulates mce3 operonduring the in vitro growth of M. tuberculosis and that the firstORF of mce2 operon that codes for a protein with homologyto GntR family-regulator protein is also a transcriptionalrepressor of mce2 operon (unpublished results). In addition,a putative transcriptional regulator gene (Rv0165c) is local-ized upstream of mce1 operon. In this respect, we can hypoth-esize that each mce operon is selectively expressed under aparticular signal during host infection. However, at present,the extracellular signals required for mce expression areunknown.

In conclusion, our studies indicate that the inactivation ofmce genes can attenuate M. tuberculosis virulence in a murinemodel, which could be explained as a reduced ability to invadeand/or persist in host cells. The attenuation of mce mutantstrains makes them interesting for testing as candidate vac-cines. Future studies should address this issue.

Fig. 5. Panels showing the morphometric analysis of the lung surface affected by granuloma (panel A) and pneumonia (panel B) in mice infected by H37Rv andmce mutant M. tuberculosis strains. Data are means from eight mice in two different experiments. Three random fields in each lung lobe were assessed. PanelC, (opposite page) Comparative representative lung histopathology of BALB/c mice after 2.5 months of intratracheal infection with M. tuberculosis H37RV ormce mutants. Control mice infected with H37Rv (1) have extensive tissue injury, manifested by massive pneumonia. In contrast, animals infected with Dmce1(2) mutant only show scarce lymphocyte accumulations without any pneumonic lesion. Mice infected with Dmce3 mutant (3) or Dmce2 mutant (4) show limitedtissue injury, manifested by middle-sized pneumonic patches (all photomicrographs, H/E, 40× original magnification).

332 A. Gioffré et al. / Microbes and Infection 7 (2005) 325–334

Fig. 5. (continued)

Fig. 6. In vivo multiplication of mce mutants after intraperitoneal inoculation. Colony-forming units were counted in the spleen (A) or in the lungs (B) ofBALB/c mice. Mice were inoculated intraperitoneally with 106 CFU of Dmce3, Dmce2 and Dmce1 mutants and the parental wild-type strain. On days 15 and30 postinoculation, mice were sacrificed, and viable bacteria present in spleen and lungs were counted. The results for each time point are the mean ± S.D. ofcolony-forming unit counts performed on four mice. These data are from one of three independent experiments with similar results. * P < 0.01 compared withwild-type.

333A. Gioffré et al. / Microbes and Infection 7 (2005) 325–334

Acknowledgements

The present study was supported by grants from the Cen-tro Argentino Brasileño de Biotecnología (CABBIO) ofArgentina, International Foundation for Science (Sweden) andConsejo Nacional de Ciencia y Tecnologia (CONACyT) ofMexico and CONICET fromArgentina. M.I.R., F.B., andA.C.are CONICET fellows. A.G. and A.A. are recipients of fel-lowships from CONICET.

References

[1] J.L. Murray, K. Stiblo, A. Rouillon, Tuberculosis in developingcountries: burden, intervention and cost, Int. J. Tuberc. Lung Dis. 65(1990) 6–24.

[2] R. Pasula, P. Wisniowski, W.J. Martin, Fibronectin facilitates Myco-bacterium tuberculosis attachment to murine alveolar macrophages,Infect. Immun. 70 (2002) 1287–1292.

[3] F.D. Menozzi, K. Pethe, P. Bifani, F. Soncin, M.J. Brennan, C. Locht,Enhanced bacterial virulence through exploitation of host glycosami-noglycans, Mol. Microbiol. 43 (2002) 1379–1386.

[4] S. Arruda, G. Bonfim, R. Knights, T. Huima-Byron, L. Riley, Cloningof an Mycobacterium tuberculosis associated with entry and survivalinside cells, Science 261 (1993) 1454–1457.

[5] R. Brosch, S.V. Gordon, M. Marmiesse, P. Brodin, C. Buchrieser,K. Eiglmeier, et al., A new evolutionary scenario for the Mycobacte-rium tuberculosis complex, Proc. Natl. Acad. Sci. USA 99 (2002)3684–3689.

[6] Y. Haile, D.A. Caugant, G. Bjune, H.G. Wiker, Mycobacterium tuber-culosis mammalian cell entry operon (mce) homologs in Mycobacte-rium other than tuberculosis (MOTT), FEMS Immunol. Med. Micro-biol. 33 (2002) 125–132.

[7] S. Ahmad, P.K. Akbar, H.G. Wiker, M. Harboe, A.S. Mustafa, Clon-ing, expression and immunological reactivity of two mammalian cellentry proteins encoded by the mce 1 operon of Mycobacterium tuber-culosis, Scand. J. Immunol. 50 (1999) 510–518.

[8] M. Harboe, A. Christensen,Y. Haile, G. Ulvund, S. Ahmad, A.S. Mus-tafa, et al., Demonstration of expression of six proteins of the mam-malian cell entry (mce 1) operon of Mycobacterium tuberculosis byantipeptide antibodies, enzyme-linked immunosorbent assay andreverse transcription-polymerase chain reaction, Scand. J. Immunol.50 (1999) 519–527.

[9] M. Panigada, T. Sturniolo, G. Besozzi, M.G. Boccieri, F. Sinigaglia,G.G. Grassi, et al., Identification of a promiscuous T-cell epitope inMycobacterium tuberculosis Mce proteins, Infect. Immun. 70 (2002)79–85.

[10] J.Y. Hou, J.E. Graham, J.E. Clark-Curtiss, Mycobacterium aviumgenes expressed during growth in human macrophages detected byselective capture of transcribed sequences (SCOTS), Infect. Immun.70 (2002) 3714–3726.

[11] J.E. Graham, J.E. Clark-Curtiss, Identification of Mycobacteriumtuberculosis RNAs synthesized in response to phagocytosis by humanmacrophages by selective capture of transcribed sequences (SCOTS),Proc. Natl. Acad. Sci. USA 96 (1999) 11554–11559.

[12] C.M. Sassetti, D.H. Boyd, E.J. Rubin, Genes required for mycobacte-rial growth defined by high density mutagenesis, Mol. Microbiol. 48(2003) 77–84.

[13] T. Parish, N.G. Stoker, Use of a flexible cassette method to generate adouble unmarked Mycobacterium tuberculosis tlyA plcABC mutantby gene replacement, Microbiology 146 (2000) 1969–1975.

[14] S. Bardarov, S. Bardarov, M.S. Pavelka, V. Sambandamurthy,M. Larsen, J. Tufariello, et al., Specialized transduction an efficientmethod for generating marked and unmarked targeted gene disrup-tions in Mycobacterium tuberculosis, Microbiology 148 (2002)3007–3017.

[15] V. Pelicic, M. Jackson, J.M. Reyrat, W.R. Jacobs Jr., B. Gicquel,C. Guilhot, Efficient allelic exchange and transposon mutagenesis inMycobacterium tuberculosis, Proc. Natl. Acad. Sci. USA 94 (1997)10955–10960.

[16] T. Parish, N.G. Stocker, Electroporation in mycobacteria, MethodsMol. Biol. 101 (1998) 129–144.

[17] H. Hecker, H.R. Kenneth, High and low annealing temperaturesincrease both specificity and yield in touchdown and stepdown PCR,Biotechniques 20 (1996) 478–485.

[18] D. Van Soolingen, P.W.M. Hermans, P.E.W. De Haas, D.R. Soll,J.D.A. Van Embden, Occurrence and stability of insertion sequencesin Mycobacterium tuberculosis complex strains: evaluation of aninsertion sequence dependent DNA polymorphism as a tool in theepidemiology of tuberculosis, J. Clin. Microbiol. 29 (1991) 2578–2586.

[19] R. Hernandez-Pando, E.H. Orozco, A. Sampieri, L. Pavon,C. Velasquillo, J. Larriva-Sahd, J.M. Alcocer, M.V. Madrid, Correla-tion between kinetics of Th1/Th2 cells and pathology in a murinemodel of experimental pulmonary tuberculosis, Immunology 89(1996) 26–33.

[20] R. Hernandez-Pando, E.H. Orozco, A.K. Arriaga, A. Sampieri, J. Lar-riva-Sahd, V. Madrid-Marina, Analysis of the local kinetics and local-ization of interleukin 1 a, tumor necrosis factor a and transforminggrowth factor b during the course of experimental pulmonary tuber-culosis, Immunology 90 (1997) 607–617.

[21] R. Hernandez-Pando, T. Schön, E.H. Orozco, J. Serafín, I. Estrada-Garcia, Expression of nitric oxide synthase and nitrotyrosine duringthe evolution of experimental pulmonary tuberculosis, Exp. Toxicol.Pathol. 53 (2001) 257–265.

[22] M.P. Santangelo, J. Goldstein, A. Alito, A. Gioffre, K. Caimi,O. Zabal, et al., Negative transcriptional regulation of themce3 operon in Mycobacterium tuberculosis, Microbiology 148(2002) 2997–3006.

[23] A. Kumar, M. Bose, V. Brahmachari, Analysis of expression profile ofmammalian cell entry (mce) operons of Mycobacterium tuberculosis,Infect. Immun. 71 (2003) 6083–6087.

[24] B.N. Flesselles, N. Anand, J. Remani, S.M. Loosmore, M.H. Klein,Disruption of the mycobacterial cell entry gene of Mycobacteriumbovis BCG results in a mutant that exhibits a reduced invasiveness forepithelial cells, FEMS Microbiol. Lett. 15 (1999) 237–242.

[25] N. Shimono, L. Morici, N. Casali, S. Cantrell, B. Sidders, S. Ehrt,et al., Hypervirulent mutant of Mycobacterium tuberculosis resultingfrom disruption of the mce1operon, Proc. Natl. Acad. Sci. USA 100(2003) 15918–15923.

[26] B. López, D. Aguilar, H. Orozco, M. Burger, C. Espitia, V. Ritacco,et al., A marked difference in pathogenesis and immune responseinduced by different Mycobacterium tuberculosis genotypes, Clin.Exp. Immunol. 133 (2003) 30–37.

[27] L. Rindi, L. Fattorini, D. Bonanni, E. Iona, G. Freer, D. Tan, et al.,Involvement of the fadD33 gene in the growth of Mycobacteriumtuberculosis in the liver of BALB/c mice, Microbiology 148 (2002)3873–3880.

334 A. Gioffré et al. / Microbes and Infection 7 (2005) 325–334