N-Glycolylated Peptidoglycan Contributes to the Immunogenicity but Not Pathogenicity of...

M A J O R A R T I C L E

N-Glycolylated Peptidoglycan Contributes to theImmunogenicity but Not Pathogenicity ofMycobacterium tuberculosis

Jesse M. Hansen,1 Solmaz A. Golchin,1 Frédéric J. Veyrier,2,3,4 Pilar Domenech,1 Ivo G. Boneca,3,4 Abul K. Azad,5

Murugesan V. S. Rajaram,5 Larry S. Schlesinger,5 Maziar Divangahi,1,6 Michael B. Reed,1,6 and Marcel A. Behr1,6

1Department of Medicine, Research Institute of the McGill University Health Center, Montreal, Canada; 2Département Infection et Epidémiologie,Infection Bactériennes Invasives, and 3Département de Microbiologie, Biologie et génétique de la paroi bactérienne, Institut Pasteur, Paris, 4INSERM,Groupe Avenir, Paris, France; 5Center for Microbial Interface Biology, Department of Microbial Infection and Immunity, The Ohio State University,Columbus; and 6McGill International TB Centre, Montreal, Canada

Mycobacteria produce an unusual, glycolylated form of muramyl dipeptide (MDP) that is more potent and effi-cacious at inducing NOD2-mediated host responses. We tested the importance of this modified form of MDPin Mycobacterium tuberculosis by disrupting the gene, namH, responsible for this modification. In vitro, thenamH mutant did not produce N-glycolylated muropeptides, but there was no alteration in colony morpholo-gy, growth kinetics, cellular morphology, or mycolic acid profile. Ex vivo, the namHmutant survived and repli-cated normally in murine and human macrophages, yet induced diminished production of tumor necrosisfactor α. In vivo, namH disruption did not affect the bacterial burden during infection of C57BL/6 mice or cel-lular recruitment to the lungs but modestly prolonged survival after infection in Rag1−/− mice. These results in-dicate that the modified MDP is an important contributor to the unusual immunogenicity of mycobacteria buthas a limited role in the pathogenesis ofM. tuberculosis infection.

Keywords. tuberculosis; NOD2; N-glycolyl MDP; N-acetyl muramic acid hydroxylase, NamH.

The intracellular receptor NOD2, a member of theNod-like receptor family, mediates the host innate re-sponse to bacterial peptidoglycan (PGN)–derivedmuramyl dipeptide (MDP). Studies by different groupshave shown that the most commonly encountered formof MDP, called N-acetyl MDP, acts upon NOD2,leading to polyubiquitination of the RIP2 kinase, acti-vation of the NF-kB pathway, and ultimately the pro-duction of proinflammatory cytokines, such as tumornecrosis factor alpha (TNF-α) [1, 2]. Recently, direct in-teraction between NOD2 and MDP was shown using

surface plasmon resonance, solidifying the notion thatMDP is the ligand for the NOD2 receptor [3].

Mycobacterial PGN is unique in that the N-acetylgroup at the carbon 2 of the muramic acid is preferen-tially hydroxylated to an N-glycolyl group throughaction of the enzyme N-acetyl muramic acid hydroxy-lase (NamH) [4]. Because the organisms that have thisunusual PGN modification were those used by JulesFreund to produce his eponymous adjuvant [5], wepreviously tested whether this modification contributesto the immunogenicity of the mycobacterial cell wallcomponents, by disrupting the namH gene in Myco-bacterium smegmatis. NOD2-mediated host responseswere blunted in the absence of bacterial namH and ad-ditionally, synthetic N-glycolyl MDP was more potentand efficacious at inducing NOD2-mediated responses,both in vitro and in vivo [6].

Whereas these previous studies in part explain theimmunogenicity of inactivated mycobacteria, the use ofM. smegmatis, an avirulent saprophyte, precluded usfrom understanding the importance of this modifica-tion for Mycobacterium tuberculosis, a professional

Received 12 June 2013; accepted 3 October 2013; electronically published 21November 2013.

Presented in part: Keystone Meeting on Tuberculosis, Vancouver, British Colum-bia, Canada, January 2011. Abstract 230.

Correspondence: Marcel A. Behr, MD, Department of Medicine, McGill Universi-ty Health Centre, A5.156, 1650 Cedar Ave, Montreal, Quebec H3G 1A4, Canada([email protected]).

The Journal of Infectious Diseases 2014;209:1045–54© The Author 2013. Published by Oxford University Press on behalf of the InfectiousDiseases Society of America. All rights reserved. For Permissions, please e-mail:[email protected]: 10.1093/infdis/jit622

NamH andM. tuberculosis • JID 2014:209 (1 April) • 1045

at Ohio State U

niversity Prior Health Sciences L

ibrary on January 16, 2015http://jid.oxfordjournals.org/

Dow

nloaded from

mailto:[email protected]

mailto:[email protected]

http://jid.oxfordjournals.org/

pathogen that causes tuberculosis. The pathogenesis of tubercu-losis requires a fine balance between bacterial survival (oftenconceived of as immune evasion) and bacterial transmission(which requires bacterial-induced immunopathology). Becausethe MDP modification results in enhanced early immune rec-ognition via NOD2, one can envision that a more potent MDPmight result in impaired bacterial persistence in the host cell.Yet, an investigation of the conserved genetic elements in 100different M. tuberculosis clinical isolates revealed that namHwas intact in all cases [7], suggesting selection in favor of thismodification. Alternatively, the modification of PGN to favorhost immune recognition is advantageous for the bacterium,either early in infection (when the bacteria induces recruitmentof host cells to the site of infection) or at the end (when the bac-teria converts the lung into an atomizer for transmission to thenext host).

Given these competing possibilities, we have investigated theimportance of this PGN modification in M. tuberculosisthrough the disruption of the namH gene and its complemen-tation. Through a combination of in vitro, ex vivo, and in vivoexperiments, we confirm that the NOD2-mediated responsesare critically dependent on the presence of N-glycolylatedPGN, yet, in a whole animal model, the effect of this disruptionon the course ofM. tuberculosis is modest, at best.

MATERIALS ANDMETHODS

Bacterial Strains, Growth Conditions, ElectroporationTechniques, and ReagentsThe stock of M. tuberculosis H37Rv was derived from a strainoriginally obtained from the Pasteur Institute (Paris, France).All strains of M. tuberculosis were grown at 37°C in Middle-brook 7H9 medium (Difco) supplemented with 10% albumin-dextrose-catalase (ADC; BD Biosciences), 0.05% tween 80(Sigma-Aldrich), and 0.2% glycerol, or on 7H11 agar (Difco)supplemented with 10% oleic acid–supplemented ADC (BDBiosciences) and 0.2% glycerol. Hygromycin B (50 μg/mL;Multicell), kanamycin monosulphate (25 μg/mL; Multicell), or2% sucrose (Sigma-Aldrich) was added where indicated. Elec-trocompetent M. tuberculosis was prepared and electroporatedas previously described [8]. For plasmid DNA cloning, theElectroMAX DH5α-E competent Escherichia coli cells (Invitro-gen) were used. All E. coli was grown in Luria-Bertani (LB)broth or on LB agar (BD Biosciences) with hygromycin (200μg/mL) or gentamicin (5 μg/mL) (Multicell) where indicated.

Nucleic Acid TechniquesMycobacterial DNA was extracted as previously described [9]and Southern blot and hybridization technique performed aspreviously described [10]. For confirmation of gene knockoutby Southern blot, genomic DNA was digested with ClaI over-night and transferred to a Hybond-N nylon membrane

(Amersham Biosciences) where it was hybridized with a 1.8-kbp polymerase chain reaction (PCR) amplicon probe of thenamH gene.

Construction and Complementation of theM. tuberculosisnamHMutantGene knockout of namH was accomplished by a 2-step allelic-exchange strategy using a thermosensitive sacB replicativevector as previously described [9]. In brief, a 210-bp fragmentin the 5′ region of the namH gene was replaced by a hygromy-cin-resistance cassette. Further technical details and primersused are presented in Supplementary Figure 1C. The mutantswere complemented with the functional namH gene undercontrol of either the native or hsp60 promoter (see Supplemen-tary Figure 1A for details).

Lipid AnalysisThe extraction and analysis of polar lipids and mycolic acidcontaining glycolipids was performed following an adaptedprotocol from Slayden and Barry [11]. The mycolates werevisualized by thin layer chromatography (TLC) analysis, byrunning the samples 3 times in a solvent system of hexanes anddiethyl ether (Sigma-Aldrich) in a ratio of 85:15 (v/v).

Peptidoglycan Purification and High-Performance LiquidChromatography AnalysisMycobacterium tuberculosis strains were grown in a 500-mLvolume of standard glycerol-supplemented 7H9 growth medium.Cultures at optical density readings at 600 nm (OD600) 0.6–0.8were centrifuged at 4°C and resuspended in 10 mL ice-coldH2O. The suspension was then added, drop by drop, to 10 mL ofsterile 8% sodium dodecyl sulfate warmed to 100°C and then auto-claved at 121°C for 60 minutes to export them from the contain-ment level 3 facility for further study. Samples were thenprocessed for PGN extraction following a modified version of theprotocol published by McNeil et al [12]. The HPLC fractionationand mass spectrometry were carried out as previously describedby Veyrier et al [13].

In Vitro Growth and Characterization of M. tuberculosisBacterial cultures were grown to mid–exponential phase andadjusted to an OD600 of 0.1, diluted 1:100 in 7H9 in a finalvolume of 50 mL, and allowed to grow for 13 days at 37°C withconstant rolling. OD600 measurements and plating for colony-forming units (CFU) was done every 2–3 days. On days 8, 11,and 13 the cultures were diluted 1:4 and later multiplied by afactor of 4 to obtain the undiluted OD600 value.

Heat-killed bacteria from the mid–exponential growth phasewere subject to 3 types of imaging using previously establishedprotocols: Ziehl–Neelsen staining for the mycolic acids, fluorescein-labeled vancomycin staining for the PGN [14], and scanningelectron microscopy [13] to assess the overall cell architecture.

1046 • JID 2014:209 (1 April) • Hansen et al

at Ohio State U



Dow

nloaded from

http://jid.oxfordjournals.org/lookup/suppl/doi:10.1093/infdis/jit622/-/DC1




MiceThe Nod2−/− breeding colony and Nod2+/+ littermate controlswere established as previously described [15]. For survivalstudies, we used C57Bl/6-DBA2 F1 and Rag1−/− mice(B6.129S7-Rag1tm1Mom/J) from Jackson Laboratory. All micewere 6–12 weeks old, and experiments were conducted in ac-cordance with the guidelines and regulations of the animal re-search ethics board of McGill University.

Ex Vivo Macrophage Stimulation and Infection StudiesMurine peritoneal macrophages were harvested from naiveNod2+/+ and Nod2−/− mice without prestimulation. A total of105 cells per well were purified by adherence and cultured in96-well plates for bacterial infection or stimulation with the in-dicated bacterial ligands. For bacterial survival studies, 4 hourspostinfection the cells were washed 3 times with RPMI to elimi-nate extracellular bacteria, then macrophages were incubated inRPMI medium until sampling time. Culture supernatants werecollected at designated intervals and stored at −20°C. TNF-αproduction by macrophages was assayed using enzyme-linkedimmunosorbent assay (ELISA; R&D Systems).

To obtain human monocyte-derived macrophages (MDMs),peripheral blood mononuclear cells were isolated, as describedpreviously [16], from heparinized blood from healthy donorsusing an approved protocol by the Ohio State University Insti-tutional Review Board. Triplicate monolayers of 12-day-oldMDMs in RHH medium (RPMI + HEPES + HSA) in a 24-welltissue culture plate were infected with a single cell suspensionof M. tuberculosis strains at a multiplicity of infection (MOI) of1:1, and incubated for 2 hours at 37°C/5% CO2 [17]. Monolay-ers were washed and repleted with RPMI containing 2% serumand incubated for different timepoints (2, 24, 48, and 72hours). Infected macrophages were then lysed and the lysatesplated on 7H11 agar for counting CFUs at approximately 21days as described [18]. Cell-free culture supernatants wereharvested from the above experiment and used to determineTNF-α and interleukin 6 (IL-6) production by using an R&DELISA kit.

Mouse Infection StudiesSeven- to 8-week old C57Bl/6 mice were aerosolized for 10minutes using a Lovelace nebulizer (model 01–100, In-ToxProducts) and 5 mice were killed 24 hours after each infectionto determine the inoculum. Mice were killed at 1, 2, 3, 6, and 12weeks postinfection for processing lungs and spleen, with theinclusion of Nod2−/− mice for analysis at 3, 6, and 12 weeks.The left (larger) lung was put in 10% neutral-buffered formalinfor histopathologic analysis, and the right (smaller) lung andspleen were homogenized, serially diluted, and plated for enu-meration of CFU.

For long-term survival experiments, C57Bl/6-DBA2 F1hybrids and Rag1−/− mice were similarly aerosolized and mice

were weighed weekly and monitored for their overall healthstatus. Mice were humanely killed if they met prespecified crite-ria for weight loss or altered health status. Upon killing, lungswere harvested and kept in 10% neutral-buffered formalin forhistopathologic verification of the cause of death.

HistopathologyImmediately following mouse killing, the left lung was put in10% neutral-buffered formalin for at least 24 hours beforebeing sent for processing (Histology Core Facility, GoodmanCancer Center, McGill University). Tissue was embedded inparaffin and 4- to 5-μm sections were processed for hematoxy-lin and eosin (H&E) staining.

Flow Cytometry AnalysisSingle-cell suspensions were obtained from collagenase- andDNase-treated lungs. Cells were stained as previously described[19] using antibodies specific for mouse CD3, CD4, CD19,CD115, CD11c, F4/80, and Gr.1 (BD Biosciences). Flow cytom-etry was performed using a BD LSR II (BD Biosciences) instru-ment with FACSDiva Software Version 6.1.2 (BD Biosciences)and analysis was performed using FlowJo Software (Tree Star).

Figures and StatisticsSignificant differences in ELISA, fluorescence-activated cellsorting (FACS), and CFU results were analyzed using the un-paired t test. For statistics involving CFU in mouse lungs andspleen, the standard error of the mean (SEM) were plotted anddifferences were deemed significant if P≤ .05. Survival datawere assigned P values based on the log-rank test. Figures weregenerated using GraphPad Prism version 4.0b (GraphPad Soft-ware).

Results

M. tuberculosis namHMutantUpon successful disruption of namH (SupplementaryFigure 2A), multiple independent isolates of the namH mutantwere expanded in liquid media to test for loss of phthiocerol di-mycocerosate (PDIM) (Supplementary Figure 2B), a virulencefactor that can be lost during in vitro culture on selective media[20]. Mutants positive for PDIM were retained for furtheranalysis and complemented with 2 different integrative plas-mids containing kanamycin-resistant cassettes, as confirmedby PCR.

To verify that disruption of namH in our strain affectedPGN composition, we performed a direct analysis of the glyco-lylated and acetylated muropeptide molecules in the PGN. Mu-tanolysin was used to liberate disaccharides with varyingpeptide side-chains attached to the muramic moieties. As thesepeptide side-chains are extremely complex in mycobacteria,due to amidation [21], we obtained a complex chromatogram


at Ohio State U



Dow

nloaded from





(Figure 1A). As expected, the PGN from M. tuberculosisΔnamH did not contain muropeptides composed of MurNGlyc(indicated by 1′, 2′, 3′, and 4′ in Figure 1A), and for this reasonthe proportion of muropeptides composed of MurNAc (indi-cated with in Figure 1A) were overrepresented. This phenotypewas reversed by the addition of namH, under control of eitherthe native or the artificial (hsp60) promoter.

Cell Wall CharacterizationWe first looked for differences in growth and morphology ofthe mutant and complemented strains compared to the wild-type (WT) strain. There was no difference between any of the4 strains in terms of growth rate as measured by OD600

(Figure 1B). As the 2 namH-complemented strains displayed

no difference in growth (Figure 1B) or apparent NamH activity(Figure 1A), we used M. tuberculosis complemented withnamH under control of its native promoter for subsequentexperiments.

We next investigated whether loss of the muramic acid mod-ification would lead to gross structural differences in otherparts of the mycobacterial cell wall given that the muramic acidwe had modified serves as the anchor point for the arabinoga-lactan complex, itself linked to mycolic acids [22].When assess-ing gross colony morphology on agar, there were no strikingdifferences (Figure 1C). Likewise, assessment of bacterial mor-phology by scanning electron microscopy did not reveal anyevident effect of NamH mutation (Figure 1D). When stainingwith vancomycin, known to reveal sites of peptidoglycan

Figure 1. In vitro characterization of Mycobacterium tuberculosis ΔnamH . A, Reverse-phase high-performance liquid chromatography analysis of muro-peptides of M. tuberculosis wild-type (Mtb), namH-deficient M. tuberculosis (ΔnamH), and the complemented namH mutant under control of either thenative (ΔnamH:namH) or the hsp60 (ΔnamH:namH hsp60) promoter. Mutanolysin digestion of the peptidoglycan liberated disaccharide units, which werefirst analyzed by mass spectrometry usingMycobacterium smegmatis. The numbered peaks correspond to muropeptides composed of N-acetylglucosamineand either N-glycolylmuramic acid (1, 2, 3, and 4) or N-acetylmuramic acid (1′, 2′, 3′, and 4′). The wild-type and complements have a mixture of both typesof muropeptide, whereas the ΔnamH mutant has only the peaks corresponding to the N-acetylmuramic acid disaccharides. (1: GlcNAc-MurNGlyc-L-Ala-D-Glu-mDAPNH2(-Gly); 1′: GlcNAc-MurNAc-L-Ala-D-Glu-mDAPNH2(-Gly); 2: GlcNAc-MurNGlyc-L-Ala-D-Glu-mDAP-D-Ala; 2′: GlcNAc-MurNAc-L-Ala-D-Glu-mDAP-D-Ala; 3: GlcNAc-MurNGlyc-L-Ala-D-Glu-mDAPNH2-D-Ala; 3′: GlcNAc-MurNAc-L-Ala-D-Glu-mDAPNH2-D-Ala; 4: Tri-Tetra composed of MurNGlycand amidated once; 4′: Tri-Tetra composed of MurNAc and amidated once.) B, Disruption of namH does not affect growth kinetics. The growth curve of the4 strains: Mtb, ΔnamH, ΔnamH::namH (nat), and ΔnamH::namH (hsp60), shows no difference in the growth kinetics between the strains when cultured inMiddlebrook 7H9 medium at 37°C with aeration. Results are representative of 2 independent experiments. C, Colony-forming units of cultures grown onsolid 7H11 agar at 37°C, for 4 weeks, for the Mtb wild-type and ΔnamH mutant. No obvious differences in the colony morphology were observed. Scalebar represents 1 cm. D, scanning electron microscopic imaging shows no observable differences in the size and shape of the bacteria. E, Vancomycin-FLstaining shows no difference in nascent peptidoglycan synthesis between the wild-type and knockout strains, with both strains predominantly stained atthe poles, which is consistent with the V-shaped or “snapping” model of mycobacterial growth previously reported [14].


at Ohio State U



Dow

nloaded from


incorporation, as it targets the D-ala-D-ala motif, bacteria wereprimarily stained at the poles (Figure 1E), consistent with thepolar growth or snapping model of mycobacteria [14]; againthere were no differences in the staining patterns with namHdisruption. Because the mycolic acids are indirectly attached tothe muramic acid that we had modified, we assessed mycolates,by both Ziehl–Neelsen staining (data not shown) and TLCanalysis (Supplementary Figure 2C); for both, there was nophenotypic difference with the namH mutant. Thus, we con-clude that we had altered the muramic acid without generatinga mutant with a grossly abnormal mycobacterial cell wall.

Ex Vivo StudiesEx vivo stimulation of WT and Nod2−/− mouse peritoneal mac-rophages with heat-killed M. tuberculosis, at an MOI of 2:1,demonstrated that NOD2-dependent TNF-α production wasdiminished in the absence of namH, and this could be geneti-cally complemented (P < .005; Figure 2A). To test whether thiseffect would be overcome when presenting a greater quantity ofN-acetyl MDP, we repeated stimulation with an MOI of 100:1(Figure 2B); namH-dependent TNF-α production was againobserved. To test whether namH disruption affects NOD2-mediated responses indirectly, eg, through altered presentationof other agonists to host pattern recognition receptors, we asked

whether the introduction of 1 μg/mL N-glycolyl MDP couldcomplement the phenotype seen with namH disruption. Atboth MOI of 2:1 and 100:1, TNF-α levels were restored withchemical complementation (Figures 2C and 2D), indicatingthat N-glycolyl MDP directly contributes to the immunogenici-ty of killedM. tuberculosis, via NOD2.

We proceeded to live bacterial infections to test for differenc-es in survival and growth within primary macrophages. By in-fecting WT mouse peritoneal macrophages with M. tuberculosisat an MOI of 2:1 for a total duration of 7 days, we observed aslight increase in CFUs for both bacterial strains, with no diffe-rence between the WT and namH mutant strain (Figure 3A).Given that bacterial numbers were namH-independent, we as-sessed host TNF-α production. When compared to the findingsfrom the heat-killed stimulations at the same MOI (Figure 2A),liveM. tuberculosis was associated with a markedly reduced cyto-kine response; however, TNF-α production was again namH-dependent. In this case, chemical complementation with 1 μg/mLN-glycolyl MDP led to a significant, albeit incomplete, increasein TNF-α (P < .05; Figure 3B).

To determine whether the namH-mediated alteration inM. tuberculosis PGN is also pertinent during infection ofhuman cells, we repeated the above experiments in humanmonocyte–derived macrophages. As was seen in murine cells,

Figure 2. NOD2-mediated recognition of Mycobacterium tuberculosis (Mtb) by macrophages depends on bacterial namH. We tested the importanceof namH disruption of M. tuberculosis during infection of primary macrophages and monitored the associated host cytokine response. A, Peritonealmacrophages from Nod2 +/+ and Nod2−/− mice were harvested and adhesion purified, then infected with heat-killed wild-type, namH-deficient, or namH-complemented M. tuberculosis at a multiplicity of infection (MOI) of 2:1. After 18 hours, the supernatant was collected and tumor necrosis factor alpha(TNF-α) levels were measured. B, Same experiment as A, with an MOI of 100:1. C, Peritoneal macrophages from Nod2 +/+ mice were infected with heat-killed wild-type M. tuberculosis or namH-deficient M. tuberculosis at an MOI of 2:1 with and without chemical complementation with 1 μg/mL N-glycolylmuramyl dipeptide (MDP). D, Same experiment as C, with an MOI of 100:1. Representative data from 2 independent replicates are shown (mean ± SEM).*P < .05; **P < .005.


at Ohio State U



Dow

nloaded from



growth of M. tuberculosis was namH-independent (Figure 4A),but TNF-α production was diminished in the absence of namH(P < .005; Figure 4B). In addition, IL-6 levels were measured atboth 24 and 48 hours, and although no significant differencewas attained there was a similar trend of reduced host responsefrom the namHmutant strain (Figure 4C).

In Vivo StudiesBecause NOD2 mediates both innate and adaptive immunity toM. tuberculosis [15], we tested whether namH disruptionwould affect the course of in vivo infection. After aerosol infec-tion of C57Bl/6 mice with approximately 30 CFUs of M. tuber-culosis, bacterial numbers increased exponentially and peakedat 3 weeks postinfection (Figure 5A). WT M. tuberculosis,namH-deficient M. tuberculosis, and namH-complemented M.tuberculosis manifest an equivalent bacterial burden, both atthe site of infection (lungs) and upon dissemination to thespleens. As bacterial numbers were namH-independent, weasked whether namH disruption would affect the host in-flammatory response to an equivalent number of bacteria.

H&E stains of mouse lung sections at 3 weeks postinfection(Figure 5B) show no qualitative differences between lungs in-fected with WTM. tuberculosis as compared to namH-deficientM. tuberculosis. To quantify the pulmonary host cellular re-sponses, lungs of mice infected for 3 weeks were subject toFACS analysis. The total lung cellular response, includingCD4+ and CD8+ lymphocytes as well as antigen-presentingcells (CD11c+ dendritic cells and F4/80+ macrophages), werenot significantly different (Figure 5C). In addition, we quanti-fied cells positive for the tuberculosis-specific antigen TB10.4and found no significant difference between mice infected withWT or namHmutant strains (data not shown).

Figure 3. Tumor necrosis factor alpha (TNF-α) cytokine levels are di-minished in the absence of namH during in vitro infection with live bacte-ria. A, Peritoneal macrophages from Nod2+/+ mice were harvested andadhesion purified then infected with wild-type (WT) and namH-deficientMycobacterium tuberculosis (Mtb) at a multiplicity of infection (MOI) of2:1. The bacterial count was enumerated at 4 hours postinfection (day 0)and at days, 2, 5, and 7. B, Peritoneal macrophages from Nod2+/+ micewere infected with live WT M. tuberculosis and namH-deficient M. tuber-culosis, at an MOI of 2:1, and chemically complemented with 1 μg/mLN-glycolyl muramyl dipeptide (MDP). Representative data from 2 indepen-dent replicates are shown (mean ± SEM). *P < .05; **P < .005.

Figure 4. NamH-dependent response to infection by human monocyte–derived macrophages. A, Monocyte-derived macrophages were infectedwith live wild-type Mycobacterium tuberculosis (Mtb), namH-deficientM. tuberculosis, and namH-complemented M. tuberculosis, at a multiplici-ty of infection of 1. The extracellular bacteria were washed away, the cellslysed, and the bacteria enumerated at 2, 24, 48, and 72 hours postinfec-tion. The supernatants were collected and tumor necrosis factor alpha(TNF-α; B) or interleukin 6 (IL-6; C) levels were measured at 24 and 48hours postinfection. Representative data from 2 independent replicates areshown (mean ± SEM). *P < .05; **P < .005.


at Ohio State U



Dow

nloaded from


Figure 5. The effect of namH disruption on the pathogenesis of Mycobacterium tuberculosis (Mtb) infection in vivo. A, C57Bl/6 mice were aerosol in-fected with approximately 30 colony-forming units (CFU) for the wild-type (WT) M. tuberculosis, namH-deficient M. tuberculosis, and namH-complementedM. tuberculosis. The bacteria in the lungs were enumerated 24 hours following infection in the lungs, and again at 1, 2, 3, 6, and 12 weeks postinfection.For the spleen, the bacteria were enumerated at 1, 2, 3, 6, and 12 weeks postinfection. Five mice were used per group. Representative data from 2 indepen-dent replicates are shown (mean ± SEM). B, Hematoxylin and eosin stain of mouse lung sections infected 3 weeks with the WT M. tuberculosis andnamH-deficient M. tuberculosis. Lung pathology was examined to qualitatively describe any differences in lymphocyte recruitment or overall lung pathology.At 3 weeks postinfection we observed no difference in the nature of the gross pathologic response to infection between the 2 strains tested (lens magnificationprovided on each panel: 2.5× lens = ×25 total magnification). C, Lungs of C57BL/6 mice infected 3 weeks with WT M. tuberculosis and namH-deficientM. tuberculosis (KO) subject to fluorescence-activated cell sorting analysis to assess the cell populations present. By multiplying cell fractions againstthe total number of cells determined by hemocytometer counts, we quantified the number of CD4, CD8, CD11c, Gr1, F4/80, and CD115 cells present. Nosignificant differences in cell numbers were observed.


at Ohio State U



Dow

nloaded from


Next, we tested whether the mutation of namH would affecthost outcome in a long-term survival model. First, we used ahost with an intact immune system, selecting the C57Bl/6-DBA2 F1 hybrid mice, as this strain is more susceptible toM. tuberculosis infection than regular C57Bl/6 mice [23]. In thismodel, we observed a trend toward improved survival with thenamH mutant between days 120 and 180, but this trend thenabruptly vanished (Figure 6A). To investigate differences at anearlier time point, when the innate response would predominate,we infected Rag1−/− mice that lack an adaptive immune re-sponse. Here, mice infected with WT M. tuberculosis requiredcompassionate sacrifice at a median time of 36 days postinfec-tion, which was 2 weeks earlier than those infected with namH-deficient M. tuberculosis (log-rank test, P < .001; Figure 6B).Mice infected with the namH-complemented M. tuberculosisstrain had an intermediate phenotype that was significantly dif-ferent from the namH-deficient strain (log-rank test, P < .003).

Thus overall, we conclude that the namH-deficient strain dis-played variable effects on survival in the animal model, with atmost a modest survival phenotype seen with Rag1−/− mice.

DISCUSSION

In the current study, we have demonstrated not only that dis-ruption of the namH gene in M. tuberculosis eliminates thepresence of N-glycolated muropeptides in the PGN, but thatthe bacteria appeared otherwise unaltered and viable, providingus with a tool to study the importance of this modification for ahost-adapted, human pathogen. Our results support and extendpreviously published findings with the avirulent bacteriumM. smegmatis, revealing that the namH-mediated modificationof the M. tuberculosis PGN contributes to its immunogenicity,and that the host response to this modification is mediated byNOD2. Using both genetic and chemical complementation,we conclusively showed that this altered immunogenicity of themycobacterial cell wall is attributable to the presence of N-glycolyl MDP and that even at a high burden of infection, theabsence of this molecule translates into significantly less im-mune stimulation. Similar findings were determined for bothmurine and human macrophages, which confirms that the N-glycolyl MDP/NOD2 interaction is relevant to the natural(human) host ofM. tuberculosis. Interestingly, complete Freund’sadjuvant is based on killed mycobacteria [5] and biochemicalstudies in the 1970s showed that its immunologic activitycould be achieved by replacing the mycobacterial cell wall withMDP [24]. Our data showed that heat-killed M. tuberculosis ishyperimmunogenic compared to live M. tuberculosis, whenusing the same cellular preparations at the same bacterial bur-den. The increased activity of killed vs live mycobacteria mayresult from a greater release of both MDP and other pathogen-associated molecular patterns as a consequence of heat-killingthe bacteria. An alternative hypothesis is that live mycobacteriaactively dampen the host response [25], a possibility supportedby the observation that the addition of 1 μg/mL N-glycolylMDP led to only partial complementation during live bacterialinfection.

These same bacteria allowed us to test whether altered inter-action with host innate immunity affected the outcome ofexperimental M. tuberculosis infection. We did not detectdifferences in bacterial burden, immune responses, or survivalwith a long-term infection model. These findings indicate thateven in the absence of namH, M. tuberculosis is able to directan appropriate immune response and the bacterium is stillcapable of surviving and replicating in a mammalian host.

The unaltered survival of mice possessing an adaptiveimmune response may reflect the limitations of the model. Itis already known that a bacteria with a 5-log reduced attackrate for people (bacillus Calmette Guérin) [26] only producesa 1-log reduction in numbers in the C57BL/6 mouse [27],

Figure 6. The effect of namH disruption on mouse survival after infec-tion. A, C57Bl/6-DBA2 F1 hybrid mice (15 per group) were aerosol infectedas previously described with wild-type (WT) Mycobacterium tuberculosis(Mtb), namH-deficient M. tuberculosis, and namH-complemented M. tu-berculosis. There were no significant differences between the 3 strainsused. B, Rag1−/− mice (8 per group) were aerosol infected as previouslydescribed with WT M. tuberculosis, namH-deficient M. tuberculosis, andnamH-complemented M. tuberculosis. Approximately 34 colony-formingunits (CFU) were implanted into the lungs of the mice infected with theMtb strain, 51 CFU for the ΔnamH strain, and 97 CFU for the ΔnamH:namH strain. There were significant differences between both WTM. tuber-culosis vs namH-deficient M. tuberculosis (P < .001) and namH-deficientM. tuberculosis vs namH-complementedM. tuberculosis (P < .003).


at Ohio State U



Dow

nloaded from


suggesting that subtle shifts in bacterial growth (or virulence)may not be readily detectable in this system. Furthermore, in-vestigations measuring a global phenotype after live M. tuber-culosis infection may be limited by the complexity of numerousinterwoven host–pathogen interactions, such that the absenceof 1 receptor or its ligand, in isolation, may not produce a dis-cernible effect. This possibility is supported by the literature onother pattern recognition receptors and M. tuberculosis, whereisolated receptor–agonist studies have produced indisputableevidence of a host–pathogen dialogue, yet infection of mice dis-rupted for TLR2, MINCLE, and NOD2 has provided weak-variable phenotypes [28–30]. By extension, the effect of namHdisruption might best be appreciated in a more indolent bacte-ria, that is, one that does not produce ESAT-6, punch holes inphagosomal membranes [31], and thereby activate a number ofhost responses such as the cytosolic surveillance pathway [32].Further studies, using different mycobacterial species, may berequired to uncover the biological utility of this cell wall modifi-cation for mycobacteria and related Actinomycetes, and to de-termine the species of mycobacteria where NOD2-mediatedrecognition of modified PGN is most critical.

Supplementary Data

Supplementary materials are available at The Journal of Infectious Diseasesonline (http://jid.oxfordjournals.org/). Supplementary materials consist ofdata provided by the author that are published to benefit the reader. Theposted materials are not copyedited. The contents of all supplementary dataare the sole responsibility of the authors. Questions or messages regardingerrors should be addressed to the author

Notes

Acknowledgments. The authors thank Vladislav Kamenski for veteri-nary assistance and acknowledge the Goodman Cancer Research Centre andhistology core facilities at McGill University for routine histological services.M. A. B. is Chercheur National of the Fonds de la Recherche en Santé duQuébec and a William Dawson Scholar of McGill University. M. B. R. andM. D. were both supported by a Canadian Institutes of Health Research(CIHR) New Investigator Award.Financial support. This work was supported by operating grants

MOP-97813 (M. A. B.) and MOP-115133 (M. B. R.) from CIHR and a Eu-ropean Research Council starting grant (PGNfromSHAPEtoVIR 202283 toI. G. B.).Potential conflicts of interest. All authors: No reported conflicts.All authors have submitted the ICMJE Form for Disclosure of Potential

Conflicts of Interest. Conflicts that the editors consider relevant to thecontent of the manuscript have been disclosed.

References

1. Yang Y, Yin C, Pandey A, Abbott D, Sassetti C, Kelliher MA. NOD2pathway activation by MDP or Mycobacterium tuberculosis infectioninvolves the stable polyubiquitination of Rip2. J Biol Chem 2007;282:36223–9.

2. Hasegawa M, Fujimoto Y, Lucas PC, et al. A critical role of RICK/RIP2polyubiquitination in Nod-induced NF-kappaB activation. EMBO J2008; 27:373–83.

3. Mo J, Boyle JP, Howard CB, Monie TP, Davis BK, Duncan JA. Patho-gen sensing by nucleotide-binding oligomerization domain-containing

protein 2 (NOD2) is mediated by direct binding to muramyl dipeptideand ATP. J Biol Chem 2012; 287:23057–67.

4. Raymond JB, Mahapatra S, Crick DC, Pavelka MS Jr. Identificationof the namH gene, encoding the hydroxylase responsible for the N-glycolylation of the mycobacterial peptidoglycan. J Biol Chem 2005;280:326–33.

5. Freund J, Casals J, Hosmer EP. Sensitization and antibody formationafter injection of tubercle bacilli and paraffin oil. Proc Soc Exp BiolMed 1937:509–13.

6. Coulombe F, Divangahi M, Veyrier F, et al. Increased NOD2-mediatedrecognition of N-glycolyl muramyl dipeptide. J Exp Med 2009;206:1709–16.

7. Tsolaki AG, Hirsh AE, DeRiemer K, et al. Functional and evolutionarygenomics of Mycobacterium tuberculosis: insights from genomic dele-tions in 100 strains. Proc Natl Acad Sci U S A 2004; 101:4865–70.

8. Snapper SB, Melton RE, Mustafa S, Kieser T, Jacobs WR Jr. Isolationand characterization of efficient plasmid transformation mutants ofMycobacterium smegmatis. Mol Microbiol 1990; 4:1911–9.

9. Pelicic V, Jackson M, Reyrat JM, Jacobs WR Jr, Gicquel B, Guilhot C.Efficient allelic exchange and transposon mutagenesis in Mycobacteri-um tuberculosis. Proc Natl Acad Sci U S A 1997; 94:10955–60.

10. van Soolingen D, Hermans PW, de Haas PE, Soll DR, van Embden JD.Occurrence and stability of insertion sequences in Mycobacteriumtuberculosis complex strains: evaluation of an insertion sequence-dependent DNA polymorphism as a tool in the epidemiology of tuber-culosis. J Clin Microbiol 1991; 29:2578–86.

11. Slayden RA, Barry CE 3rd. Analysis of the lipids of Mycobacterium tu-berculosis. Methods Mol Med 2001; 54:229–45.

12. McNeil M, Daffe M, Brennan PJ. Evidence for the nature of the linkbetween the arabinogalactan and peptidoglycan of mycobacterial cellwalls. J Biol Chem 1990; 265:18200–6.

13. Veyrier FJ, Williams AH, Mesnage S, Schmitt C, Taha MK, Boneca IG.De-O-acetylation of peptidoglycan regulates glycan chain extensionand affects in vivo survival of Neisseria meningitidis. Mol Microbiol2013; 87:1100–12.

14. Aldridge BB, Fernandez-Suarez M, Heller D, et al. Asymmetry andaging of mycobacterial cells lead to variable growth and antibiotic sus-ceptibility. Science 2012; 335:100–4.

15. Divangahi M, Mostowy S, Coulombe F, et al. NOD2-deficient micehave impaired resistance to Mycobacterium tuberculosis infectionthrough defective innate and adaptive immunity. J Immunol 2008;181:7157–65.

16. Beharka AA, Gaynor CD, Kang BK, Voelker DR, McCormack FX,Schlesinger LS. Pulmonary surfactant protein A up-regulates activity ofthe mannose receptor, a pattern recognition receptor expressed onhuman macrophages. J Immunol 2002; 169:3565–73.

17. Schlesinger LS, Bellinger-Kawahara CG, Payne NR, Horwitz MA.Phagocytosis of Mycobacterium tuberculosis is mediated by humanmonocyte complement receptors and complement component C3. JImmunol 1990; 144:2771–80.

18. Olakanmi O, Britigan BE, Schlesinger LS. Gallium disrupts iron metab-olism of mycobacteria residing within human macrophages. InfectImmun 2000; 68:5619–27.

19. Divangahi M, Desjardins D, Nunes-Alves C, Remold HG, Behar SM.Eicosanoid pathways regulate adaptive immunity toMycobacterium tu-berculosis. Nat Immunol 2010; 11:751–8.

20. Domenech P, Reed MB. Rapid and spontaneous loss of phthiocerol di-mycocerosate (PDIM) fromMycobacterium tuberculosis grown in vitro:implications for virulence studies. Microbiology 2009; 155:3532–43.

21. Brennan PJ, Nikaido H. The envelope of mycobacteria. Annu RevBiochem 1995; 64:29–63.

22. Azuma I, Yamamura Y, Misaki A. Isolation and characterization ofarabinose mycolate from firmly bound lipids of mycobacteria. J Bacter-iol 1969; 98:331–3.

23. Reed MB, Domenech P, Manca C, et al. A glycolipid of hypervirulenttuberculosis strains that inhibits the innate immune response. Nature2004; 431:84–7.


at Ohio State U



Dow

nloaded from






24. Souvannavong V, Adam A, Lederer E. Kinetics of the humoral and cel-lular immune response of guinea pigs after injection of the syntheticadjuvant N-acetylmuramyl L-alanyl-D-isoglutamine: comparison withFreund complete adjuvant. Infect Immun 1978; 19:966–71.

25. Geijtenbeek TB, Van Vliet SJ, Koppel EA, et al. Mycobacteria targetDC-SIGN to suppress dendritic cell function. J Exp Med 2003; 197:7–17.

26. Casanova JL, Blanche S, Emile JF, et al. Idiopathic disseminated bacillusCalmette-Guerin infection: a French national retrospective study. Pedi-atrics 1996; 98:774–8.

27. Lewis KN, Liao R, Guinn KM, et al. Deletion of RD1 fromMycobacteri-um tuberculosis mimics bacille Calmette-Guerin attenuation. J InfectDis 2003; 187:117–23.

28. Holscher C, Reiling N, Schaible UE, et al. Containment of aerogenicMycobacterium tuberculosis infection in mice does not require

MyD88 adaptor function for TLR2, -4 and -9. Eur J Immunol 2008;38:680–94.

29. Heitmann L, Schoenen H, Ehlers S, Lang R, Holscher C. Mincle is notessential for controlling Mycobacterium tuberculosis infection. Immu-nobiology 2013; 218:506–16.

30. Gandotra S, Jang S, Murray PJ, Salgame P, Ehrt S. Nucleotide-binding oligomerization domain protein 2-deficient mice controlinfection with Mycobacterium tuberculosis. Infect Immun 2007;75:5127–34.

31. Simeone R, Bobard A, Lippmann J, et al. Phagosomal rupture byMyco-bacterium tuberculosis results in toxicity and host cell death. PLoSPathog 2012; 8:e1002507.

32. Manzanillo PS, Shiloh MU, Portnoy DA, Cox JS.Mycobacterium tuber-culosis activates the DNA-dependent cytosolic surveillance pathwaywithin macrophages. Cell Host Microbe 2012; 11:469–80.


at Ohio State U



Dow

nloaded from


N-Glycolylated Peptidoglycan Contributes to the Immunogenicity but Not Pathogenicity of...

Documents

Transcript of N-Glycolylated Peptidoglycan Contributes to the Immunogenicity but Not Pathogenicity of...