Differential Influence of Nutrient-Starved Mycobacterium tuberculosison Adaptive Immunity Results in Progressive Tuberculosis Diseaseand Pathology

Jes Dietrich,a Sugata Roy,a* Ida Rosenkrands,a Thomas Lindenstrøm,a Jonathan Filskov,a Erik Michael Rasmussen,c Joseph Cassidy,b

Peter Andersena

Department of Infectious Disease Immunology, Statens Serum Institut, Copenhagen, Denmarka; Pathobiology Section, School of Veterinary Medicine, University CollegeDublin, Belfield, Dublin, Irelandb; International Reference Laboratory of Mycobacteriology, Division of Diagnostics & Infection Control, Statens Serum Institut, Copenhagen,Denmarkc

When infected with Mycobacterium tuberculosis, most individuals will remain clinically healthy but latently infected. Latentinfection has been proposed to partially involve M. tuberculosis in a nonreplicating stage, which therefore represents an M. tu-berculosis phenotype that the immune system most likely will encounter during latency. It is therefore relevant to examine howthis particular nonreplicating form of M. tuberculosis interacts with the host immune system. To study this, we first induced astate of nonreplication through prolonged nutrient starvation of M. tuberculosis in vitro. This resulted in nonreplicating persis-tence even after prolonged culture in phosphate-buffered saline. Infection with either exponentially growing M. tuberculosis ornutrient-starved M. tuberculosis resulted in similar lung CFU levels in the first phase of the infection. However, between week 3and 6 postinfection, there was a very pronounced increase in bacterial levels and associated lung pathology in nutrient-starved-M. tuberculosis-infected mice. This was associated with a shift from CD4 T cells that coexpressed gamma interferon(IFN-�) and tumor necrosis factor alpha (TNF-�) or IFN-�, TNF-�, and interleukin-2 to T cells that only expressed IFN-�. Thus,nonreplicating M. tuberculosis induced through nutrient starvation promotes a bacterial form that is genetically identical toexponentially growing M. tuberculosis yet characterized by a differential impact on the immune system that may be involved inundermining host antimycobacterial immunity and facilitate increased pathology and transmission.

Mycobacteria are known to adapt to harsh environments byregulating their metabolism, protein expression, and repli-

cation. Mycobacterium smegmatis starved of carbon, nitrogen, orphosphorus has been shown to remain viable for over 650 daysafter an initial 2- to 3-log drop in number of CFU and has dis-played increased stress resistance, increased mRNA stability, andan overall decrease in protein synthesis (1). Another example islatent Mycobacterium tuberculosis, which represents a continuumof M. tuberculosis in different growth stages, some of which in-volve M. tuberculosis in a low-replicating or nonreplicating dor-mant condition (dormant M. tuberculosis is defined as “nonrepli-cating bacilli maintaining full viability at a very low metabolicrate” [2]). As one-third of the world’s population is infected withlatent tuberculosis (TB), these individuals therefore constitute anenormous reservoir for TB disease and transmission (3).Thus,new strategies for the eradication of the low-replicating or non-replicating dormant mycobacteria that are present during a latentinfection are required. We believe that such strategies will arisefrom a better understanding of the immune response against thisparticular form of M. tuberculosis, as well as increased insight intothe process of reactivation. It has been suggested that dormantmycobacteria reside in phagosomes inside macrophages withoutfree access to oxygen and nutrients. To understand the biology ofthese bacteria, conditions have been developed in vitro to simulatethe in vivo environment which probably involves hypoxia pro-duced in avascular calcified granulomas (4), NO (5) or carbonmonoxide (CO) (6) produced by activated immune cells, or aphosphate-limited environment within macrophage phagosomes(7). A widely used model is the “Wayne model” (8), in which alow-inoculum culture is sealed in a tube with stirring and allowed

to slowly consume oxygen until the culture is anaerobic, therebyachieving a nonreplicating and apparently dormant state (9). Thisstate was found to be accompanied by restriction of biosyntheticactivity to conserve energy, induction of alternative energy path-ways, stabilization of essential cell components, and expression ofgenes in particular within the DosR regulon (10), which is believedto be important for rapid resumption of growth once M. tubercu-losis exits from an anaerobic or nitric oxide-induced nonrespiringstate (11). Loebel et al. (12, 13) investigated the effect of the nu-trient limitations predicted to be available in a granuloma on themetabolism of M. tuberculosis and found that nutrient starvationresulted in a gradual shutdown of respiration to minimal levels,but the bacilli remained viable and were able to recover when

Received 18 August 2015 Returned for modification 14 September 2015Accepted 17 September 2015

Accepted manuscript posted online 28 September 2015

Citation Dietrich J, Roy S, Rosenkrands I, Lindenstrøm T, Filskov J, Rasmussen EM,Cassidy J, Andersen P. 2015. Differential influence of nutrient-starvedMycobacterium tuberculosis on adaptive immunity results in progressivetuberculosis disease and pathology. Infect Immun 83:4731–4739.doi:10.1128/IAI.01055-15.

Editor: S. Ehrt

Address correspondence to Jes Dietrich, jdi@ssi.dk.

* Present address: Sugata Roy, Omics Molecular Interaction Research Unit, LSATechnology Development Unit, RIKEN Yokohama, Yokohama, Japan.

Supplemental material for this article may be found at http://dx.doi.org/10.1128/IAI.01055-15.

Copyright © 2015, American Society for Microbiology. All Rights Reserved.

December 2015 Volume 83 Number 12 iai.asm.org 4731Infection and Immunity

returned to rich medium. Using this model, it was shown bymicroarray profiling that transfer to phosphate-buffered saline(PBS) led to the induction of specific genes, downregulation ofaerobic respiration, translation, cell division, and lipid biosynthe-sis (14, 15). Some of the starvation-induced genes are now beingtested as vaccine candidates against a latent infection with M. tu-berculosis (16, 17).

Although gene and protein expression profiling analyses com-paring stressed (dormant) and nonstressed M. tuberculosis bacte-ria have been initiated (10, 14), the impact of dormant M. tuber-culosis on the immune system has not been the subject of detailedstudies. For other bacteria, such as Coxiella burnetii, Leishmania,Legionella pneumophila, and Chlamydia trachomatis, it has beenfound that they can exist in replicative and nonreplicative forms,which differ markedly in their interactions with the phagocyticcells of the immune system (18–21). Based on this, the aims of thisstudy were to analyze the immune response to nonreplicating M.tuberculosis (induced by nutrient starvation, termed StarvM.tb)and to compare that to the response promoted by replicating(exponentially growing) M. tuberculosis (termed LogM.tb). Wefound that these two forms were genetically identical but hadmarkedly different impacts on the immune system and that infec-tion with StarvM.tb, despite resulting in similar bacterial levels inthe first phase of the infection as infection with LogM.tb, led to adramatic breakdown of protective antimycobacterial immunity inthe late stage of the infection.

MATERIALS AND METHODSAnimal handling. Studies were performed with 6- to 8-week-old femaleCB6F1 mice (BALB/c � C57BL/6) from Harland Netherlands. Nonin-fected mice were housed in cages in appropriate animal facilities at StatensSerum Institut. Infected animals were housed in cages contained withinlaminar-flow safety enclosures (Scantainer; Scanbur, Denmark) in a sep-arate biosafety level 3 facility. All mice were fed radiation-sterilized 2016global rodent maintenance diet (Harlan, Scandinavia) and water ad libi-tum. All animals were allowed a 1-week rest period after delivery beforethe initiation of the experiments. The handling of mice was conducted inaccordance with the regulations set forward by the Danish Ministry ofJustice and animal protection committees in Danish Animal ExperimentsInspectorate permit 2004-561-868 of 01-07-2004. This was in compliancewith European Community Directive 86/609 and the U.S. Association forLaboratory Animal Care recommendations for the care and use of labo-ratory animals. All animal handling was done at Statens Serum Institut byauthorized personnel.

Bacteria. All bacteria were stored at �80°C in growth medium at�5 � 108 CFU/ml. Bacteria were thawed, sonicated, washed, and dilutedin PBS. All bacterial work was performed at the Statens Serum Institut byauthorized personnel. M. tuberculosis H37Rv was grown at 37°C in sus-pension in Sauton medium (BD Pharmingen) enriched with 0.5% glu-cose. StarvM.tb bacteria were established from H37Rv (ATCC 27294)bacteria from Sauton medium. The bacteria were washed twice and trans-ferred to PBS as previously described (22). Bacteria were cultured for 6weeks to 2 months in PBS at 37°C in a shaker incubator. Determinationsof CFU and most probable number (MPN) were performed prior to theinfection to count and test the viability of the bacteria. Assessment ofMPN was done as described previously (23). Briefly, bacterial suspensionswere serially diluted in growth medium. Tubes with visible bacterialgrowth were counted as positive, and MPN values were calculated usingstandard statistical methods (24). Prior to infection, the bacteria weresonicated and passaged several times through a small (27-gauge) syringethree times to reduce clumping.

Sequencing of M. tuberculosis. Genomic DNA of isolates from theLogM.tb bacteria and two batches of StarvM.tb bacteria was sequenced on

an Illumina NextSeq 500 instrument at Statens Serum Institut (Copenha-gen, Denmark). Procedures were performed according to manufacturer’smanual, except that 1.5 ng of input DNA was used instead of the 1 ng asdescribed. Reads were mapped to the M. tuberculosis H37Rv genome(GenBank accession no. NC_000962.2) by employing the BWA 0.7 exactalignment program (25). Reads were sorted and duplicates were removedusing Picard 1.127, Freebayes call variants, BCFtools, vcfutils variant fil-tering, and Plink to transform VCF files to tped files. For all isolates, morethan 96% of the M. tuberculosis H37Rv genome was covered with at leastone read.

Antigens for in vitro stimulation. Synthetic overlapping peptides (9-and 18-mers) covering the complete primary structure of Ag85B, TB10.4,and ESAT-6 were synthesized by standard solid-phase methods on a SyRopeptide synthesizer (MultiSynTech GmbH, Witten, Germany) at the JPTPeptide Technologies (Berlin, Germany) or at Schafer-N (Copenhagen,Denmark). Peptides were lyophilized and stored dry at �20°C until re-constitution in PBS.

Experimental infections. Upon challenge by the aerosol route, theanimals were infected with approximately 100 to 150 CFU of M. tubercu-losis H37Rv/mouse with an inhalation exposure system (Glas-Col, TerraHaute, IN). For intravenous (i.v.) infection, the dose was 5 � 106 bacteria/dose. The numbers of bacteria in the spleen or lung were determined byserial 3-fold dilutions of individual whole-organ homogenates in dupli-cate on 7H11 medium supplemented with polymyxin B, amphotericin B,nalidixic acid, trimethoprim, and azlocillin (PANTA) (Becton Dickinson,San Diego, CA). Colonies were counted after 2 to 3 weeks of incubation at37°C.

Lymphocyte cultures. Splenocyte cultures were obtained by passageof spleens through a metal mesh, followed by two washing proceduresusing RPMI medium. Lung lymphocytes were obtained by passage of thelungs through a 100-�m nylon cell strainer (BD Pharmingen, USA), fol-lowed by two washing procedures using RPMI medium. The cells in eachexperiment were cultured in sterile microtiter wells (96-well plates; Nunc,Denmark) containing 2 � 105 to 10 � 105 cells in 200 �l of RPMI mediumsupplemented with 1% (vol/vol) premixed penicillin-streptomycin solu-tion (Invitrogen Life Technologies), 1 mM glutamine, and 10% (vol/vol)fetal calf serum (FCS) at 37°C and 5% CO2.

Flow cytometric analysis. For the intracellular cytokine (IC) stainingprocedure, cells from the spleen or lungs of mice were stimulated for 1 to2 h with 2 �g/ml antigen at 37°C and subsequently incubated for 5 h at37°C with 10 �g/ml brefeldin A (Sigma-Aldrich, Denmark) at 37°C. Fcreceptors were blocked with 0.5 �g/ml anti-CD16/CD32 monoclonal an-tibody (MAb) (BD Pharmingen, USA) for 10 min, after which the cellswere washed in FACS (fluorescence-activated cell sorting) buffer (PBScontaining 0.1% sodium azide and 1% FCS) before being stained with acombination of the following rat anti-mouse antibodies: phycoerythrin(PE)-Cy7– and peridinin chlorophyll protein (PerCP)-Cy5.5–anti-CD8�(53-6.7, RM4-5), allophycocyanin (APC)-Cy7–anti-CD4 (GKI.5), fluo-rescein isothiocyanate (FITC)– and PE-Cy5.5–anti-CD44 (IM7), all pur-chased from BD Pharmingen (San Diego, CA), R&D Systems (Minneap-olis, MN), or eBiosciences (San Diego, CA). Cells were washed with FACSbuffer before fixation and permeabilization using the BD Cytofix/Cy-toperm kit (BD, San Diego, CA) according to the manufacturer’s protocoland then stained intracellularly with PE–, PE-Cy7–, or APC–anti-gammainterferon (IFN-�) (XMG1.2), PE–anti-tumor necrosis factor alpha(TNF-�), and/or PE– and APC–anti-interleukin 2 (IL-2) (JES6-5H4). Af-ter washing, cells were resuspended in formaldehyde solution (4%, wt/vol), pH 7.0 (Bie & Berntsen, Denmark), and samples were analyzed on asix-color FACSCanto flow cytometer (BD Biosciences, USA). Data anal-ysis was done with FACS Diva software (Becton-Dickinson, San Diego,CA) and Flowjo software (Tree Star, Ashland, OR). TB10.4 pentamerstaining was performed as previously described (26).

IFN-� ELISA. Microtiter plates (96 well; Maxisorb; Nunc, Denmark)were coated with 1 �g/ml monoclonal rat anti-murine IFN-� (clone R4-6A2; BD Pharmingen). Free binding sites were blocked with 2% (wt/vol)

Dietrich et al.

4732 iai.asm.org December 2015 Volume 83 Number 12Infection and Immunity

milk powder in PBS. Culture supernatants were harvested from lympho-cyte cultures after 72 h of in vitro antigen stimulation and tested in tripli-cate. IFN-� was detected with a 0.1-�g/ml biotin-labeled rat anti-murineantibody (clone XMG1.2; BD Pharmingen, USA) and 0.35 �g/ml horse-radish peroxidase-conjugated streptavidin (Zymed, Invitrogen, USA).The enzyme reaction was developed with 3.3=,5.5=-tetramethylbenzidineand hydrogen peroxide (TMB plus; Kementec, Denmark) and stoppedwith 0.2 M H2SO4. Recombinant IFN-� (rIFN-�; BD Pharmingen, USA)was used as a standard. Plates were read at 450 nm with an enzyme-linkedimmunosorbent assay (ELISA) reader and analyzed with KC4 3.03 Rev 4software.

Histopathological analysis. The right lung of each mouse sampledwas fixed by immersion in 10% neutral buffered formalin and processedfor histological examination. Cut sections were stained with hematoxylinand eosin and by the Ziehl-Neelsen method and were evaluated withoutprior knowledge of treatment group. Lesion morphometry was carriedout using the public-domain, Java-based image processing program Im-ageJ (http://imagej.nih.gov/ij/).

RESULTSIn vitro analysis of exponentially growing and nutrient-starvedM. tuberculosis. To induce a state of nonreplicating M. tubercu-losis, the bacteria were taken from exponentially growing bacteriain Sauton medium and transferred to PBS at 37°C (Fig. 1A). Aftertransfer of the bacteria to PBS, CFU levels were determined atweeks 0 and 6. Although a small reduction was observed in CFUcounts, this was not significant (Fig. 1B). Moreover, althoughsome forms of stress have been shown to induce a nonculturableform of M. tuberculosis (23, 27–29), incubation in PBS did not leadto nonculturable M. tuberculosis since direct viable counting didnot show bacterial numbers that were significantly different fromthose determined by methods such as indirect enumeration bymost-probable-number (MPN) estimation (27, 30) (Fig. 1C).Furthermore, transfer to PBS did not, as expected, lead to muta-

tions in the StarvM.tb bacterial genome, as full genome sequenc-ing of two StarvM.tb batches and one LogM.tb batch (methoddescribed in Materials and Methods) did not reveal any significantdifferences between the samples (data not shown). Transfer ofStarvM.tb bacteria back to Sauton medium showed that after a lagperiod of 4 days, these bacteria resumed growth with a growth ratesimilar to that observed with LogM.tb (Fig. 1D). As stress induc-tion has been shown to reduce the acid-fast staining of the stressedM. tuberculosis cultures (31, 32), we also tested whether the tran-sition to nutrient starvation-induced nonreplication/dormancywould decrease the percentage of acid-fast staining in starved M.tuberculosis bacteria. After 6 weeks, we noticed no significant de-crease in acid-fast-positive bacteria. We did however notice(slightly) less intensive acid-fast staining in StarvM.tb bacteria,reflecting changes in cell wall lipid composition as previously de-scribed (31) (Fig. 1E). Thus, even prolonged culture of bacteria inthe absence of nutrients did not result in marked reduction ofbacterial numbers measured by either direct plate counting orMPN estimation, and the bacteria exhibited only a minor mor-phological change. Consequently, in all subsequent infectionstudies, bacteria were counted on agar plates prior to infection toestablish that the mice had received the same inocula of nonrep-licating M. tuberculosis (StarvM.tb) and exponentially growing M.tuberculosis (LogM.tb).

In vivo growth and immunopathology of LogM.tb andStarvM.tb. We next compared the in vivo growth of StarvM.tb andLogM.tb. We chose CB6F1 mice, as they are well characterized inour laboratory in terms of immune recognition of TB antigensduring an infection. Mice were infected by the aerosol route withthe same number of either StarvM.tb or LogM.tb bacteria (ap-proximately 200 CFU). Thereafter, the mice were sacrificed at dif-ferent time points and the bacterial numbers in the lungs were

FIG 1 Generating nutrient-starved mycobacteria. (A) Overview of the procedure to generate nutrient-starved mycobacteria. (B) StarvM.tb and LogM.tbbacteria were transferred to Sauton medium, and the numbers of CFU were determined at the time points indicated. (C) StarvM.tb and LogM.tb bacteria werecounted either by plating on agar or by determining the MPN. (D) Growth measured by absorbance of StarvM.tb and LogM.tb in liquid culture over a period of15 days. (E and F) LogM.tb (E) or StarvM.tb (F) bacteria were stained with auramine-rhodamine. Magnifications are indicated on the panels.

Starved M. tuberculosis Induces Altered T Cell Immunity

December 2015 Volume 83 Number 12 iai.asm.org 4733Infection and Immunity

determined. StarvM.tb did not show an extended lag period ingrowth compared to LogM.tb, as we did not observe any signifi-cant differences in the bacterial numbers in the StarvM.tb- andLogM.tb-infected mice at days 4 and 11 (Fig. 2A). However, fromweek 2 onward, the bacterial number of StarvM.tb (7.88 � 0.06log10 CFU/lung at week 3) significantly exceeded that of LogM.tb(6.89 � 0.13 log10 CFU/lung at week 3) (Fig. 2A). This differencebecame very pronounced after week 3, at which time the bacterialnumber of LogM.tb declined to 5.23 � 0.18 log10 CFU/lung (week6), whereas bacterial numbers in StarvM.tb instead continued toincrease to 8.61 � 0.23 log10 CFU/lung. The same pattern wasobserved in an additional experiment illustrated in Fig. 2B. Again,in the initial phase of the infection (until week 3 postinfection),the bacterial numbers were similar between the groups, but be-tween weeks 3 and 6, StarvM.tb-infected mice showed in-creased CFU levels (Fig. 2B). Significantly higher bacterial levelsin StarvM.tb-infected mice in the late stage of the infection wasobserved repeatedly in individual experiments, using differentStarvM.tb batches, and in both the aerosol and i.v. infection mod-els. Results from an experiment using the i.v. infection model areshown in Fig. S1 in the supplemental material.

At week 3 (in the experiment shown in Fig. 2A), macroscopicexamination showed larger numbers of granulomas in the lungs ofStarvM.tb-infected mice than in LogM.tb-infected mice (Fig. 3A).Histopathological analysis indicated that while granulomas inboth groups consisted of dense unencapsulated aggregates of ad-mixed intact and degenerate macrophages and neutrophils withsmaller numbers of accompanying lymphocytes, neutrophil infil-trates were larger and denser and were associated with extensivefoci of necrosis and greater numbers of acid-fast bacteria in the

FIG 2 In vivo growth of StarvM.tb and LogM.tb bacteria. (A and B) Bacterialload (shown as log10 CFU) in the lungs of mice infected with StarvM.tb orLogM.tb via the aerosol route. Two experiments are shown in panels A and B,respectively. Data represent the mean � SD for a minimum of four mice pergroup. *, P 0.05.

FIG 3 Macroscopic analysis and immunopathology of infected lungs. Mice infected by the aerosol route with StarvM.tb or LogM.tb were sacrificed 3 (A to C)or 6 (D and E) weeks after challenge and were examined macroscopically (A, D, and E) or subjected to histopathological examination (B and C [which shows theweek 3 time point]). For histopathology, the right lung was fixed by immersion in 10% neutral buffered formalin and processed for examination. Cut sectionswere stained with hematoxylin and eosin (HE). Lesion morphometry was carried out using the public-domain, Java-based image processing program ImageJ. Inpanel C, results for lesions in lungs are from three independent experiments (represented by unique symbols).

Dietrich et al.

4734 iai.asm.org December 2015 Volume 83 Number 12Infection and Immunity

StarvM.tb lesions (Fig. 3B and see Fig. S2 in the supplementalmaterial). In both groups, clusters of neutrophils admixed withacid-fast bacteria were detected in bronchiolar lumens. Lesionmorphometry found that the mean cross-sectional surface area ofStarvM.tb granulomas was 3.7-fold larger than that of LogM.tblesions (P 0.0001, Mann-Whitney test) (Fig. 3C).

At week 6, while all LogM.tb-infected animals exhibited only afew macroscopically visible granulomas, all StarvM.tb-infectedmice showed numerous pulmonary granulomas scatteredthroughout all lobes (Fig. 3D), and macroscopic examinationshowed increased numbers of granulomas in the lungs ofStarvM.tb-infected mice (Fig. 3E).

Influence of StarvM.tb on adaptive immunity. To study theinfluence of StarvM.tb on the host immune response during infec-tion, we first focused on the influence on adaptive immunity. Anti-gen-specific cytokine secretion/expression was analyzed by ELISA orIC-FACS staining. To study the antigen-specific T cell response, wechose the mycobacterial antigens ESAT-6, Ag85B, and TB10.4, whichare all known to be recognized following an infection (33). As amarker for the CD8 T cell response, we also used the antigen TB10.4,as it contains well-characterized CD8 epitopes (26, 34, 35).

We first analyzed the cytokine secretion by ELISA. At week 2postinfection, lung T cells from LogM.tb-infected mice showed asecretion of IFN-� that was below detection, whereas T cells iso-lated from the lungs of StarvM.tb-infected mice already secreted

substantial levels of IFN-� in response to all three antigens (Fig.4A; see also Fig. S3 in the supplemental material, which shows theAg85B-specific response).

Interestingly, in contrast to the differences in CFU counts atweek 3, we did not observe any differences in the magnitude oflung T cell IFN-� secretion from LogM.tb- and StarvM.tb-in-fected mice at this time point. Analysis of cytokine expression byIC flow cytometry at week 3 also revealed only minor differencesamong the distribution of T cell subtypes within the pool of anti-gen-specific T cells (with IFN-�/TNF-�-coexpressing T cells asthe dominant subtype in both groups and for both CD4 and CD8T cells) (Fig. 4B).

Three weeks later (week 6), the difference was pronounced,with a significantly higher cytokine secretion in LogM.tb-infectedmice following stimulation with any of the three antigens (TB10.4response, 36,991.6 � 667.0 pg/ml IFN-� in LogM.tb-infectedmice versus 12,493.3 � 2,835.6 pg/ml in StarvM.tb-infected mice;Ag85B response, 6,417.4 � 2,485.3 pg/ml versus 3,136.0 � 879.6pg/ml IFN-�; ESAT-6 response, 9,567.0 � 1,505.1 versus3,376.0 � 1,480.8 pg/ml IFN-�) (Fig. 4A, and see Fig. S3 in thesupplemental material). This correlated with a markedly differentprofile by IC flow cytometry, where the dominating T cell subtypein StarvM.tb-infected mice was T cells that produced only IFN-�,whereas LogM.tb-infected mice maintained polyfunctional T cellsin both the CD4 and CD8 T cell population (Fig. 4B). In line with

FIG 4 T cell-induced IFN-� secretion. (A) Lung cells were stimulated with TB10.4 or ESAT6 in vitro for 72 h, and IFN-� levels in supernatants were assessed byELISA. Data represent the mean � SEM of the results for a minimum of three mice per group. (B) Mice were sacrificed 3 or 6 weeks after the challenge, andlymphocytes from lungs were stimulated in vitro with TB10.4 peptides prior to staining with anti-CD4, -CD8, -CD44, -IFN-�, -IL-2, and -TNF-�. Cytokineprofiles of specific CD4 and CD8 T cells 3 weeks after an aerosol challenge are shown. Frequencies represent cytokine-producing T cells within the subset of T cellsspecific for TB10.4. Data represent the mean � SEM of the results for a minimum of three mice per group (*, P 0.05 [two-way analysis of variance withBonferroni’s posttest]). (C) At week 3 after aerosol infection, lung cells were stimulated with TB10.4 or ESAT6 in vitro for 72 h, and IL-10 and IL-5 levels insupernatants were assessed by ELISA. Data represent the mean � SEM of the results for a minimum of three mice per group.

Starved M. tuberculosis Induces Altered T Cell Immunity

December 2015 Volume 83 Number 12 iai.asm.org 4735Infection and Immunity

this, lung T cells from StarvM.tb-infected mice showed a signifi-cant decrease in not only polyfunctional T cell subtypes but alsothe expression of IFN-� per single antigen-specific CD4 or CD8 Tcell, measured by its mean fluorescence intensity (MFI) (data notshown). Finally, in addition to a lack of IFN-� expression, stainingthe CD8 T cells directly ex vivo with H2-Kb pentamer loaded withthe TB10.4 CD8 epitope IMYNYPAM showed a strong reductionin CD8 T cell numbers (data not shown).

In the i.v. infection model, we observed the same overall im-munological pattern as observed in the aerosol model (see Fig. S1in the supplemental material).

In the aerosol model, week 3 represented a time point at whichboth groups of mice had a strong T cell response and we thereforeexpanded our analysis to also include Th2/Treg cytokines. In con-trast to the similar levels of IFN-�, we measured a significantlyhigher secretion of both IL-5 and IL-10 from lymphocytes ex-tracted from the lungs of StarvM.tb-infected mice than from lungsof LogM.tb-infected mice (Fig. 4C). This was observed not onlyfollowing stimulation with TB10.4 but also after stimulation withAg85B and ESAT-6 (Fig. 4C, and see Fig. S3 in the supplementalmaterial).

Taken together, at week 3, lung T cells from StarvM.tb-infectedmice showed increased bacterial numbers, a substantial IFN-�response, and increased IL-5/IL-10 expression and subsequentlysuffered from a breakdown of antimycobacterial protective im-munity. Most probably reflecting the increased bacterial numbersobserved in the lung, at week 3 to 4 postinfection, StarvM.tb-infected mice also showed significantly increased expression in thelung of proinflammatory cytokines such as CXCL1, IL-12, MCP-1and IP-10 (data not shown).

Preventing StarvM.tb-mediated alteration of the T cell re-sponse. We next examined if a vaccine with efficacy againstLogM.tb could prevent the observed StarvM.tb-mediated altera-tion of the T cell response. We used the vaccine antigen Ag85B–ESAT-6 formulated in the adjuvant CAF01, which we previouslyshowed protects against infection with LogM.tb (36). Vaccinationwith Ag85B–ESAT-6/CAF01 reduced the growth of LogM.tb (1.17log10 CFU reduction) in agreement with previous reports on thisvaccine (36). In addition, this vaccine also protected against infec-tion with StarvM.tb with approximately the same difference be-tween vaccinated and unvaccinated animals as seen in theLogM.tb-infected animals. Thus, bacterial numbers in the lungs ofvaccinated animals were reduced to 5.62 � 0.18 log10 CFU com-pared to 6.88 � 0.21 log10 CFU in unvaccinated animals (Fig. 5A).The Ag85B–ESAT-6/CAF01 vaccination prevented the T cell phe-notypic change to T cells only expressing IFN-�, as shown by theimproved quality of antigen-specific T cell phenotypes (Fig. 5B)(which now included polyfunctional T cells and only a minimumof IFN-� only-positive T cells) as well as an increased antigen-specific IFN-� secretion by peripheral blood mononuclear cells(PBMCs) from vaccinated StarvM.tb-infected animals (data notshown). Interestingly, the phenotype of TB10.4-specific T cells(not in the vaccine) was similar for vaccinated and nonvaccinatedanimals. This suggests that the improved ESAT-6/Ag85B CD4 Tcell phenotypes in vaccinated animals is mediated by vaccine-pro-moted expansion of less-differentiated memory T cells, as previ-ously reported (37), and is not due to a general improvement of Tcell quality as a result of the reduced bacterial numbers in vacci-nated animals.

DISCUSSION

Subjecting mycobacteria to nutrient starvation has been previ-ously shown to lead to a slowdown of the transcription apparatus,energy metabolism, lipid biosynthesis, and cell division (14). Inthe present study, we expand these results and show that nonrep-licating and logarithmically growing mycobacteria differ in theirimpacts on host-adaptive immune responses raised during infec-tion. StarvM.tb bacteria cause the development of a functionallyimpaired immune response, leading to IFN-� only-positive effec-tor T cells that are unable to contain infection. The result of this isa breakdown of protective immunity with accelerated bacterialgrowth and an excessive pathology not observed during an infec-tion with replicating M. tuberculosis. Due to the very differentimpact on the host immune system, compared to that of conven-tional M. tuberculosis, these results also highlight the importanceof considering the starved form of M. tuberculosis in future vaccinestudies.

There are several ways to induce M. tuberculosis dormancy invitro, and the methods used in our study (transfer to PBS) pro-duced a bacterial M. tuberculosis phenotype that in several ways

FIG 5 Vaccination against infection with StarvM.tb. (A) Animals were vacci-nated with Ag85B–ESAT-6 in CAF01 adjuvant and infected with StarvM.tb orLogM.tb at week 6 after the last vaccination. CFU levels were determined in thelungs at week 6 postinfection. (B) Cytokine profiles of ESAT-6-, Ag85B-, andTB10.4-specific CD4 T cells 6 weeks after aerosol challenge.

Dietrich et al.

4736 iai.asm.org December 2015 Volume 83 Number 12Infection and Immunity

differed from that of the dormant M. tuberculosis generated inother studies. Thus, in contrast to previous studies (23, 27–29),the StarvM.tb bacteria used in our study did not constitute anonculturable form of M. tuberculosis (Fig. 1). Furthermore,StarvM.tb did not show a longer lag period in mice than LogM.tb,in contrast to a recent study (38) where dormancy was achievedthrough oxygen depletion according to the method used byWayne et al. (9) and to the 4-day lag period upon transfer to liquidculture (Fig. 1). Finally, the increased infectivity of StarvM.tb inthis study is in contrast to another recent study (39) (using theguinea pig TB model), where nonreplicating M. tuberculosis wasestablished by the gradual depletion of nutrients in an oxygen-replete environment. In this study, it was found that the nonrep-licating M. tuberculosis elicited a significantly reduced infectivity(39). Taken together, these contrasting results may reflect the dif-ferent methods used to establish a dormant form of M. tuberculo-sis, and exactly how these different dormant forms of M. tubercu-losis are present in humans during a latent (or chronic) infection isimportant to establish in future studies.

Adaptive immune response against StarvM.tb and LogM.tb.Concerning the impact on adaptive immunity, we observed inter-esting differences between the two groups of infected mice. At theweek 2 time point, we observed only an accelerated T cell responsein StarvM.tb-infected mice (Fig. 4). The week 3 time point repre-sents an interesting turning point, as both groups have similarlevels of both IFN-� responses and T cell quality as measured by ICflow cytometry. However, control over the infection was clearlyfailing in StarvM.tb-infected mice, as shown by increased bacterialnumbers and increased pathology as well as an increase in both thesize and numbers of granolomas in the lung (Fig. 2 and 3). More-over, StarvM.tb-infected mice showed increased levels of Il-5 andIL-10 following stimulation with each of the three antigens used(Fig. 4). It is not fully known how or if IL-5/IL-10-secreting cellscontribute to the control of the infection, but IL-10 expression hasin fact been inversely associated with disease progression (40, 41).Whether the differences observed in this study play a role in pro-viding an advantage for the bacteria in StarvM.tb-infected mice isnot known. It could be speculated that in light of the much higherexpression of IFN-� in both StarvM.tb and LogM.tb bacteria,other factors may be responsible for the different outcomes of theinfection from week 3 onward.

The outcomes of the infection with StarvM.tb and LogM.tb aredifferent, and while the growth of LogM.tb is reduced approxi-mately 100 times after week 4, StarvM.tb bacteria continue togrow exponentially, resulting in more numerous, and on average3.7 times larger, granulomas at week 3 that exhibited extensive fociof necrosis and in which acid-fast bacteria were more numerous.Fulminant pathology on such a scale in the StarvM.tb-infectedmice is consistent with lack of control of infection and has notbeen reported in mice that have been aerosol infected with labo-ratory-adapted strains of log-phase M. tuberculosis. At week 6, thebacterial numbers in the StarvM.tb-infected mice were signifi-cantly higher than those during the conventional infection, andthe immune profile in these animals was dominated by IFN-�only-positive T cells, indicating a change in the T cell response inthese mice. This was supported by a decline in the expression ofIFN-�, either measured by mean fluorescence intensity or ELISA.The presence of polyfunctional T cells at week 6 correlated in-versely with bacterial load, which is in agreement with previousstudies in which polyfunctional T cells have been shown to corre-

late with protection in disease models (42–45), such as cancer(43), Leishmania major infection (42), M. tuberculosis infection(16), or HIV infection (45). Taken together, our results demon-strated an accelerated immune response in StarvM.tb-infectedmice that eventually led to an altered CD4/CD8 T cell responseand consequently to a lack of control over the infection. Interest-ingly, the change in T cell quality could be prevented by a vaccine.This was demonstrated with the vaccine Ag85B–ESAT-6/CAF01.Previously we showed that this vaccine is able to maintain a pool ofmulticytokine-expressing central memory T cells in the face of anongoing chronic infection (37). The present study expanded onthese observations and showed that also in the face of an infectionwith StarvM.tb, this vaccine was able to maintain a pool of less-differentiated multicytokine-expressing T cells. However, im-proved T cell phenotypes were observed only with vaccine anti-gen-specific T cells, despite the overall reduction in bacterialnumbers. This was exemplified by TB10.4-specific T cells, whichdisplayed the same phenotype in vaccinated and nonvaccinatedanimals, despite the difference in bacterial levels. This is in agree-ment with the previous study analyzing the phenotype of TB10.4,compared to the vaccine antigens Ag85B and ESAT-6, during achronic infection (37). In this study, only the phenotypes of the Tcells specific for Ag85B or ESAT-6 were improved. Taken to-gether, our results demonstrate that the maintenance of a popu-lation of high-quality T cells during infection with StarvM.tb canbe achieved through a vaccine-mediated expansion of these cellsprior to the infection.

What might explain the increased virulence observed withStarvM.tb in the later stage of infection? A trivial explanationwould be that the dose of StarvM.tb was underestimated. How-ever, infection with a similar, or even slightly lower, dose ofStarvM.tb (determined by both numbers of CFU, MPN, and in-fectious loads in the lung or spleen in the initial phase of theinfection [Fig. 2, and see Fig. S1 in the supplemental material])demonstrated that StarvM.tb-infected animals still had signifi-cantly higher immune responses than LogM.tb-infected animalsat days 14 to 21 postinfection despite similar CFU levels through-out the initial phase of the infection. Taken together, we believethat the CFU levels of StarvM.tb are not underestimated but thatthe higher bacterial loads in the late phase of infection is a conse-quence of a breakdown of the immune system, resulting in uncon-trolled bacterial replication after the first 2 to 3 weeks.

Interestingly, in an attempt to discover the reason for the in-creased virulence of Beijing stains compared to other M. tubercu-losis strains, it was found that all the analyzed Beijing strainssynthesized structural variants of two well-known characteristiclipids of the tubercle bacillus, namely, phthiocerol dimycoceros-ates (DIM) and phenolglycolipids (PGL) (46), which correlatedwith higher CFU values (47, 48). It could be speculated that thecell wall of StarvM.tb has a specific immune-activating lipid com-position. Membrane lipids have been shown to be able to stimu-late both activating and inhibitory pathways (49), and M. tuber-culosis produces a wide variety of bioactive lipids that have beenimplicated in the pathogenesis of tuberculosis (50). Importantly,applying stress to M. tuberculosis has been shown to lead tochanges in the cell envelope lipids (32, 51–54), and in our analysisof the bacterium, we did notice a more uneven staining that indi-cated a change in the membrane lipids (Fig. 1). In addition, theincreased virulence of StarvM.tb may also be explained by differ-ent expression levels of certain proteins (14).

Starved M. tuberculosis Induces Altered T Cell Immunity

December 2015 Volume 83 Number 12 iai.asm.org 4737Infection and Immunity

In summary, these results demonstrate that starving M. tuber-culosis of nutrients promotes fundamental changes in the viru-lence of the bacteria and their impact on the immune system.Importantly, this impact is not mediated through an inherenthigher growth rate of the bacteria but seems to be related to anability to accelerate adaptive immune responses that result in im-mune breakdown, uncontrolled bacterial replication, and pathol-ogy. We hypothesize that this may represent a potential mecha-nism that allows dormant bacteria, as part of its resuscitationprocess, to coordinate the development of excess pathology thateventually facilitates transmission.

ACKNOWLEDGMENTS

The excellent technical assistance provided by Lene Rasmussen and JanneRabech as well as the animal technicians at the Statens Serum Institut isgratefully acknowledged. We also thank Nadine Hoffmann for help withthe MPN analyses. We thank Brian Cloak for excellent technical assistancein the preparation of the photomicrographs. We thank Dorte Bek Folk-vardsen for help with genomic sequencing and Jonas Grauholm for helpwith sequencing and the analysis of the results.

This study was supported by Danish Research Council Project 11-107120, 7th Framework Programme for Research and Technological De-velopment-HEALTH-2009 Single-Stage Contract 241745 (NEWTBVAC),and the Lundbeck Foundation R54-A5572.

The funders had no role in study design, data collection and analysis,decision to publish, or preparation of the manuscript.

REFERENCES1. Smeulders MJ, Keer J, Speight RA, Williams HD. 1999. Adaptation of

Mycobacterium smegmatis to stationary phase. J Bacteriol 181:270 –283.2. Gengenbacher M, Kaufmann SHE. 2012. Mycobacterium tuberculosis:

success through dormancy. FEMS Microbiol Rev 36:514 –532. http://dx.doi.org/10.1111/j.1574-6976.2012.00331.x.

3. Kochi A. 1994. Tuberculosis: distribution, risk factors, mortality. Immuno-biology 191:325–336. http://dx.doi.org/10.1016/S0171-2985(11)80437-7.

4. Tsai MC, Chakravarty S, Zhu G, Xu J, Tanaka K, Koch C, Tufariello J,Flynn J, Chan J. 2006. Characterization of the tuberculous granuloma inmurine and human lungs: cellular composition and relative tissue oxygentension. Cell Microbiol 8:218 –232. http://dx.doi.org/10.1111/j.1462-5822.2005.00612.x.

5. Nathan C, Shiloh MU. 2000. Reactive oxygen and nitrogen intermediatesin the relationship between mammalian hosts and microbial pathogens.Proc Natl Acad Sci U S A 97:8841– 8848. http://dx.doi.org/10.1073/pnas.97.16.8841.

6. Ryter SW, Alam J, Choi AM. 2006. Heme oxygenase-1/carbon monox-ide: from basic science to therapeutic applications. Physiol Rev 86:583–650. http://dx.doi.org/10.1152/physrev.00011.2005.

7. Rifat D, Bishai WR, Karakousis PC. 2009. Phosphate depletion: a noveltrigger for Mycobacterium tuberculosis persistence. J Infect Dis 200:1126 –1135. http://dx.doi.org/10.1086/605700.

8. Wayne LG. 1976. Dynamics of submerged growth of Mycobacteriumtuberculosis under aerobic and microaerophilic conditions. Am Rev Re-spir Dis 114:807– 811.

9. Wayne LG, Hayes LG. 1996. An in vitro model for sequential study ofshiftdown of Mycobacterium tuberculosis through two stages of nonrep-licating persistence. Infect Immun 64:2062–2069.

10. Wayne LG, Sohaskey CD. 2001. Nonreplicating persistence of mycobac-terium tuberculosis. Annu Rev Microbiol 55:139 –163. http://dx.doi.org/10.1146/annurev.micro.55.1.139.

11. Leistikow RL, Morton RA, Bartek IL, Frimpong I, Wagner K,Voskuil MI. 2010. The Mycobacterium tuberculosis DosR regulonassists in metabolic homeostasis and enables rapid recovery from non-respiring dormancy. J Bacteriol 192:1662–1670. http://dx.doi.org/10.1128/JB.00926-09.

12. Loebel RO, Shorr E, Richardson HB. 1933. The influence of foodstuffsupon the respiratory metabolism and growth of human tubercle bacilli. JBacteriol 26:139 –166.

13. Loebel RO, Shorr E, Richardson HB. 1933. The influence of adverse

conditions upon the respiratory metabolism and growth of human tuber-cle bacilli. J Bacteriol 26:167–200.

14. Betts JC, Lukey PT, Robb LC, McAdam RA, Duncan K. 2002. Evalua-tion of a nutrient starvation model of Mycobacterium tuberculosis persis-tence by gene and protein expression profiling. Mol Microbiol 43:717–731. http://dx.doi.org/10.1046/j.1365-2958.2002.02779.x.

15. Betts JC, Dodson P, Quan S, Lewis AP, Thomas PJ, Duncan K, Mc-Adam RA. 2000. Comparison of the proteome of Mycobacterium tuber-culosis strain H37Rv with clinical isolate CDC 1551. Microbiology 146:3205–3216. http://dx.doi.org/10.1099/00221287-146-12-3205.

16. Aagaard C, Hoang T, Dietrich J, Cardona PJ, Izzo A, Dolganov G,Schoolnik GK, Cassidy JP, Billeskov R, Andersen P. 2011. A multistagetuberculosis vaccine that confers efficient protection before and after ex-posure. Nat Med 17:189 –194. http://dx.doi.org/10.1038/nm.2285.

17. Kaufmann SH, Lange C, Rao M, Balaji KN, Lotze M, Schito M, ZumlaAI, Maeurer M. 2014. Progress in tuberculosis vaccine development andhost-directed therapies—a state of the art review. Lancet Respir Med2:301–320. http://dx.doi.org/10.1016/S2213-2600(14)70033-5.

18. Samuel JE, Kiss K, Varghees S. 2003. Molecular pathogenesis of Coxiellaburnetii in a genomics era. Ann N Y Acad Sci 990:653– 663. http://dx.doi.org/10.1111/j.1749-6632.2003.tb07440.x.

19. Swanson MS, Fernandez-Moreira E. 2002. A microbial strategy to mul-tiply in macrophages: the pregnant pause. Traffic 3:170 –177. http://dx.doi.org/10.1034/j.1600-0854.2002.030302.x.

20. Hammerschlag MR. 2002. The intracellular life of chlamydiae. SeminPediatr Infect Dis 13:239 –248. http://dx.doi.org/10.1053/spid.2002.127201.

21. Molofsky AB, Swanson MS. 2004. Differentiate to thrive: lessons from theLegionella pneumophila life cycle. Mol Microbiol 53:29 – 40. http://dx.doi.org/10.1111/j.1365-2958.2004.04129.x.

22. Albrethsen J, Agner J, Piersma SR, Hojrup P, Pham TV, Weldingh K,Jimenez CR, Andersen P, Rosenkrands I. 2013. Proteomic profiling ofMycobacterium tuberculosis identifies nutrient-starvation-responsivetoxin-antitoxin systems. Mol Cell Proteomics 12:1180 –1191. http://dx.doi.org/10.1074/mcp.M112.018846.

23. Shleeva MO, Bagramyan K, Telkov MV, Mukamolova GV, Young M,Kell DB, Kaprelyants AS. 2002. Formation and resuscitation of “non-culturable” cells of Rhodococcus rhodochrous and Mycobacterium tuber-culosis in prolonged stationary phase. Microbiology 148:1581–1591. http://dx.doi.org/10.1099/00221287-148-5-1581.

24. De Man JC. 1975. The probability of most probable numbers. Eur J ApplMicrobiol 1:2.

25. Li H, Durbin R. 2009. Fast and accurate short read alignment with Bur-rows-Wheeler transform. Bioinformatics 25:1754 –1760. http://dx.doi.org/10.1093/bioinformatics/btp324.

26. Billeskov R, Vingsbo-Lundberg C, Andersen P, Dietrich J. 2007. Induc-tion of CD8 T cells against a novel epitope in TB10.4: correlation withmycobacterial virulence and the presence of a functional region of differ-ence-1. J Immunol 179:3973–3981. http://dx.doi.org/10.4049/jimmunol.179.6.3973.

27. Biketov S, Mukamolova GV, Potapov V, Gilenkov E, Vostroknutova G,Kell DB, Young M, Kaprelyants AS. 2000. Culturability of Mycobacte-rium tuberculosis cells isolated from murine macrophages: a bacterialgrowth factor promotes recovery. FEMS Immunol Med Microbiol 29:233–240. http://dx.doi.org/10.1111/j.1574-695X.2000.tb01528.x.

28. Zhang Y, Yang Y, Woods A, Cotter RJ, Sun Z. 2001. Resuscitation ofdormant Mycobacterium tuberculosis by phospholipids or specific pep-tides. Biochem Biophys Res Commun 284:542–547. http://dx.doi.org/10.1006/bbrc.2001.4993.

29. Sun Z, Zhang Y. 1999. Spent culture supernatant of Mycobacteriumtuberculosis H37Ra improves viability of aged cultures of this strain andallows small inocula to initiate growth. J Bacteriol 181:7626 –7628.

30. Kell DB, Kaprelyants AS, Weichart DH, Harwood CR, Barer MR. 1998.Viability and activity in readily culturable bacteria: a review and discussionof the practical issues. Antonie Van Leeuwenhoek 73:169 –187. http://dx.doi.org/10.1023/A:1000664013047.

31. Nyka W. 1974. Studies on the effect of starvation on mycobacteria. InfectImmun 9:843– 850.

32. Deb C, Lee CM, Dubey VS, Daniel J, Abomoelak B, Sirakova TD, PawarS, Rogers L, Kolattukudy PE. 2009. A novel in vitro multiple-stressdormancy model for Mycobacterium tuberculosis generates a lipid-loaded, drug-tolerant, dormant pathogen. PLoS One 4:e6077. http://dx.doi.org/10.1371/journal.pone.0006077.

Dietrich et al.

4738 iai.asm.org December 2015 Volume 83 Number 12Infection and Immunity

33. Andersen P, Doherty TM. 2005. TB subunit vaccines—putting the piecestogether. Microbes Infect 7:911–921. http://dx.doi.org/10.1016/j.micinf.2005.03.013.

34. Majlessi L, Rojas MJ, Brodin P, Leclerc C. 2003. CD8-T-cell responsesof Mycobacterium-infected mice to a newly identified major histocom-patibility complex class I-restricted epitope shared by proteins of theESAT-6 family. Infect Immun 71:7173–7177. http://dx.doi.org/10.1128/IAI.71.12.7173-7177.2003.

35. Radosevic K, Wieland CW, Rodriguez A, Weverling GJ, Mintardjo R,Gillissen G, Vogels R, Skeiky YA, Hone DM, Sadoff JC, van der Poll T,Havenga M, Goudsmit J. 2007. Protective immune responses to a recom-binant adenovirus type 35 tuberculosis vaccine in two mouse strains: CD4and CD8 T-cell epitope mapping and role of gamma interferon. InfectImmun 75:4105– 4115. http://dx.doi.org/10.1128/IAI.00004-07.

36. Agger EM, Rosenkrands I, Olsen AW, Hatch G, Williams A, Kritsch C,Lingnau K, von Gabain A, Andersen CS, Korsholm KS, Andersen P.2006. Protective immunity to tuberculosis with Ag85B-ESAT-6 in a syn-thetic cationic adjuvant system IC31. Vaccine 24:5452–5460. http://dx.doi.org/10.1016/j.vaccine.2006.03.072.

37. Lindenstrom T, Knudsen NP, Agger EM, Andersen P. 2013. Control ofchronic mycobacterium tuberculosis infection by CD4 KLRG1-IL-2-secreting central memory cells. J Immunol 190:6311– 6319. http://dx.doi.org/10.4049/jimmunol.1300248.

38. Woolhiser L, Tamayo MH, Wang B, Gruppo V, Belisle JT, Lenaerts AJ,Basaraba RJ, Orme IM. 2007. In vivo adaptation of the Wayne model oflatent tuberculosis. Infect Immun 75:2621–2625. http://dx.doi.org/10.1128/IAI.00918-06.

39. Bacon J, Alderwick LJ, Allnutt JA, Gabasova E, Watson R, Hatch KA,Clark SO, Jeeves RE, Marriott A, Rayner E, Tolley H, Pearson G, HallG, Besra GS, Wernisch L, Williams A, Marsh PD. 2014. Non-replicatingMycobacterium tuberculosis elicits a reduced infectivity profile with cor-responding modifications to the cell wall and extracellular matrix. PLoSOne 9:e87329. http://dx.doi.org/10.1371/journal.pone.0087329.

40. Redford PS, Murray PJ, O’Garra A. 2011. The role of IL-10 in immuneregulation during M. tuberculosis infection. Mucosal Immunol 4:261–270. http://dx.doi.org/10.1038/mi.2011.7.

41. Feng CG, Kullberg MC, Jankovic D, Cheever AW, Caspar P, CoffmanRL, Sher A. 2002. Transgenic mice expressing human interleukin-10 inthe antigen-presenting cell compartment show increased susceptibility toinfection with Mycobacterium avium associated with decreased macro-phage effector function and apoptosis. Infect Immun 70:6672– 6679. http://dx.doi.org/10.1128/IAI.70.12.6672-6679.2002.

42. Darrah PA, Patel DT, De Luca PM, Lindsay RW, Davey DF, Flynn BJ,Hoff ST, Andersen P, Reed SG, Morris SL, Roederer M, Seder RA. 2007.Multifunctional TH1 cells define a correlate of vaccine-mediated protec-tion against Leishmania major. Nat Med 13:843– 850. http://dx.doi.org/10.1038/nm1592.

43. Imai N, Ikeda H, Tawara I, Shiku H. 2009. Tumor progression inhibits

the induction of multifunctionality in adoptively transferred tumor-specific CD8 T cells. Eur J Immunol 39:241–253. http://dx.doi.org/10.1002/eji.200838824.

44. Forbes EK, Sander C, Ronan EO, McShane H, Hill AV, Beverley PC,Tchilian EZ. 2008. Multifunctional, high-level cytokine-producing Th1cells in the lung, but not spleen, correlate with protection against Myco-bacterium tuberculosis aerosol challenge in mice. J Immunol 181:4955–4964. http://dx.doi.org/10.4049/jimmunol.181.7.4955.

45. Kannanganat S, Ibegbu C, Chennareddi L, Robinson HL, Amara RR.2007. Multiple-cytokine-producing antiviral CD4 T cells are functionallysuperior to single-cytokine-producing cells. J Virol 81:8468 – 8476. http://dx.doi.org/10.1128/JVI.00228-07.

46. Huet G, Constant P, Malaga W, Laneelle MA, Kremer K, van SoolingenD, Daffe M, Guilhot C. 2009. A lipid profile typifies the Beijing strains ofMycobacterium tuberculosis: identification of a mutation responsible fora modification of the structures of phthiocerol dimycocerosates and phe-nolic glycolipids. J Biol Chem 284:27101–27113. http://dx.doi.org/10.1074/jbc.M109.041939.

47. Dormans J, Burger M, Aguilar D, Hernandez-Pando R, Kremer K,Roholl P, Arend SM, van Soolingen D. 2004. Correlation of virulence,lung pathology, bacterial load and delayed type hypersensitivity responsesafter infection with different Mycobacterium tuberculosis genotypes in aBALB/c mouse model. Clin Exp Immunol 137:460 – 468. http://dx.doi.org/10.1111/j.1365-2249.2004.02551.x.

48. Barczak AK, Domenech P, Boshoff HI, Reed MB, Manca C, Kaplan G,Barry CE, III. 2005. In vivo phenotypic dominance in mouse mixedinfections with Mycobacterium tuberculosis clinical isolates. J Infect Dis192:600 – 606. http://dx.doi.org/10.1086/432006.

49. Wymann MP, Schneiter R. 2008. Lipid signalling in disease. Nat Rev MolCell Biol 9:162–176. http://dx.doi.org/10.1038/nrm2335.

50. Brennan PJ. 2003. Structure, function, and biogenesis of the cell wall ofMycobacterium tuberculosis. Tuberculosis (Edinb) 83:91–97. http://dx.doi.org/10.1016/S1472-9792(02)00089-6.

51. Asselineau J. 1966. The bacterial lipids. Hermann, Paris, France.52. Minnikin DE. 1982. Lipids: complex lipids, their chemistry, biosynthesis

and roles, p 95–184. In C Ratledge and J L Stanford (ed), The biology of themycobacteria. Academic Press, London, United Kingdom.

53. Daniel J, Maamar H, Deb C, Sirakova TD, Kolattukudy PE. 2011.Mycobacterium tuberculosis uses host triacylglycerol to accumulate lipiddroplets and acquires a dormancy-like phenotype in lipid-loaded macro-phages. PLoS Pathog 7:e1002093. http://dx.doi.org/10.1371/journal.ppat.1002093.

54. Sirakova TD, Dubey VS, Deb C, Daniel J, Korotkova TA, Abomoelak B,Kolattukudy PE. 2006. Identification of a diacylglycerol acyltransferasegene involved in accumulation of triacylglycerol in Mycobacterium tuber-culosis under stress. Microbiology 152:2717–2725. http://dx.doi.org/10.1099/mic.0.28993-0.

Starved M. tuberculosis Induces Altered T Cell Immunity

December 2015 Volume 83 Number 12 iai.asm.org 4739Infection and Immunity