Dp

PPMAa

b

c

d

Ce

f

a

ARAA

KTBPSLPG

1

tm8ctahmao

F

h0

Vaccine 33 (2015) 1353–1359

Contents lists available at ScienceDirect

Vaccine

j our na l ho me page: www.elsev ier .com/ locate /vacc ine

eletion of zmp1 improves Mycobacterium bovis BCG-mediatedrotection in a guinea pig model of tuberculosis

eter Sandera,b,∗, Simon Clarkc, Agnese Petreraa,1, Cristina Vilaplanad, Michael Meulia,etra Selchowa, Andrea Zelmere, Deepa Mohananf, Nuria Andreue, Emma Raynerc,ichael Dal Molina, Gregory J. Bancrofte, Pål Johansenf, Pere-Joan Cardonad,

nn Williamsc, Erik C. Böttgera,b

Institute of Medical Microbiology, University of Zurich, Zurich, SwitzerlandNationales Zentrum für Mykobakterien, Gloriastrasse 30/32, 8006 Zurich, SwitzerlandMicrobiology Services, Public Health England, Porton Down, Salisbury, Wiltshire, United KingdomUnitat de Tuberculosi Experimental, Fundació Institut d’Investigació en Ciències de la Salut Germans Trias i Pujol, Universitat Autonoma de Barcelona,IBERES, Badalona, Catalonia, SpainDeptartment of Immunology and Infection, London School of Hygiene and Tropical Medicine, London WC1E 7HT, United KingdomDepartment of Dermatology, University Hospital Zurich, Zurich, Switzerland

r t i c l e i n f o

rticle history:eceived 10 December 2014ccepted 14 January 2015vailable online 2 February 2015

eywords:

a b s t r a c t

Having demonstrated previously that deletion of zinc metalloprotease zmp1 in Mycobacterium bovis BCGincreased immunogenicity of BCG vaccines, we here investigated the protective efficacy of BCG zmp1deletion mutants in a guinea pig model of tuberculosis infection. zmp1 deletion mutants of BCG providedenhanced protection by reducing the bacterial load of tubercle bacilli in the lungs of infected guineapigs. The increased efficacy of BCG due to zmp1 deletion was demonstrated in both BCG Pasteur andBCG Denmark indicating that the improved protection by zmp1 deletion is independent from the BCG

uberculosisCGrotectionafetyive vaccinerotease

sub-strain. In addition, unmarked BCG �zmp1 mutant strains showed a better safety profile in a CB-17SCID mouse survival model than the parental BCG strains. Together, these results support the furtherdevelopment of BCG �zmp1 for use in clinical trials.

© 2015 Elsevier Ltd. All rights reserved.

uinea pig

. Introduction

Tuberculosis (TB) caused by the bacterial pathogen Mycobac-erium tuberculosis, is a major cause of human morbidity and

ortality. It is estimated that 1.5 million people, among them0 000 children, died of the disease in 2013. About 9 million newases of TB, including 550 000 in children, occur each year and onehird of the world’s population, 2 billion people, is latently infectedccording to WHO (Global Tuberculosis Report 2014, WHO;ttp://www.who.int/tb/publications/global report/en/). Approxi-

ately 5% of these latently infected individuals will develop

ctive TB at a later time point. WHO has declared eliminationf TB by the year 2050 a global target [1]. The low decline in

∗ Corresponding author. Tel.: +41 44 634 2684; fax: +41 44 634 4906.E-mail address: psander@imm.uzh.ch (P. Sander).

1 Present address: Institut für Molekulare Medizin und Zellforschung, Universitätreiburg, Freiburg, Germany.

ttp://dx.doi.org/10.1016/j.vaccine.2015.01.058264-410X/© 2015 Elsevier Ltd. All rights reserved.

the incidence of TB observed in the past years is insufficientto reach this goal. Even the most stringent implementation ofdiagnostic techniques and medication currently available will notsuffice.

Continuous in vitro passaging of Mycobacterium bovis, thecausative agent of bovine tuberculosis, gave rise to the currentlyavailable attenuated TB live vaccine. Deletion of the Region of Dif-ference 1 (RD1), encoding the specialized protein secretion systemESX-1, is considered the primary genetic alteration responsible forattenuation [2]. Named after its inventors, the live vaccine M. bovisbacille Calmette-Guérin (BCG) was first applied to humans in 1921.Still today, BCG is the only licensed TB vaccine and more than 4 bil-lion doses have been applied [3]. BCG confers consistent protectionagainst TB meningitis and disseminated TB in children. However,protective efficacy of BCG against the most prevalent form of TB,

adult lung TB, is highly variable. Protective efficacy varied between0% and 85% in different trials [4,5]. According to WHO, improve-ment of TB vaccines has high priority and is a corner stone in TBelimination.

1 cine 33

cdnMbefcAiaaaCofa

zaecaineenmtstdss

2

2

aNc[rpitrssctsneaicrat

354 P. Sander et al. / Vac

Vaccine development from bench to field is a long lasting pro-ess and the last decade witnessed considerable research efforts toevelop more protective TB vaccines. Candidates include recombi-ant BCG [6], attenuated strains of M. tuberculosis [7], recombinantycobacterium smegmatis [8], subunit vaccines [9] and recom-

inant viruses [10]. Many more candidate vaccines have beenxplored, however, the attrition rate proved to be high [11,12]. Aew advanced candidates are currently tested for safety in clini-al trials. MVA85A, a recombinant strain of modified vaccinia virusnkara expressing antigen 85A from M. tuberculosis was tested for

ts protective efficacy in a large scale clinical trial. MVA85A, givens a booster after primary vaccination with BCG, was well toler-ted but failed to show a beneficial effect over BCG vaccinationlone [13]. Clinical trials with recombinant BCG strains expressinglostridium perfringens perfringolysin [14] were stopped becausef adverse side effects [11]. Therefore, there is a continuing needor new TB vaccines to be developed which are both effectivend safe.

M. tuberculosis open reading frame Rv0198c encodes theinc metallopeptidase Zmp1. Rv0198c has considerable sequencend structural homology with human peptidases neprilysin andndothelin converting enzyme (ECE) [15]. Zmp1 preferentiallyleaves small peptides and has an optimal activity at a slightlycidic pH [16]. Deletion of zmp1 attenuates M. tuberculosis, inducesnnate, inflammasome- and IL-1�-dependent immune mecha-isms and increases the ratio of mycobacteria localizing in latendosomal compartments of infected macrophages instead of inarly endosomal compartments [17]. In vitro and in vivo immu-ization experiments in mice indicated that a BCG zmp1 deletionutant is more immunogenic than its parental strain [18]. In cat-

le, BCG zmp1 deletion mutant strains induced an enhanced antigenpecific immune response [19] considered to be a correlate of pro-ection [20]. Here, we report on the protective efficacy of BCG zmp1eletion mutants in the guinea pig model of TB infection and theafety of the BCG zmp1 deletion mutants in a CB-17 SCID mouseurvival model.

. Materials and methods

.1. Recombinant BCG zmp1 mutant strains

BCG Pasteur SmR zmp1::aph is a zmp1 deletion mutant of streptomycin resistant BCG Pasteur strain. Two consecutiveheI fragments within zmp1 were substituted by an aph cassetteonferring kanamycin resistance. The strain has been described17]. BCG Denmark �zmp1 is a derivative of BCG Denmarkesulting from transformation with zmp1 replacement vectorzmp1 sacB aph hyg. Besides the NheI deletions, the zmp1 allele

n pzmp1 sacB aph hyg carries a triple translation stop. In the vec-or, the �zmp1 allele is flanked by a hygromycin and a kanamycinesistance cassette for (transient) positive selection and levanucrase gene sacB [21] for negative selection. Integration andubsequent removal of vector backbone from transformants washecked by Southern blot analyses and antibiotic susceptibilityesting [19]. Vector pzmp1 sacB aph hyg was also used to con-truct an unmarked �zmp1 mutant in BCG Pasteur SmS strainamed BCG Pasteur SmS �zmp1. BCG Pasteur SmS was madelectrocompetent, transformed with plasmid pzmp1 sacB aph hygnd transformants were selected on 7H10 OADC plates contain-ng hygromycin (25 mg L−1). Following counter selection of single

ross over transformants on sucrose, deletion of the zmp1 locus andemoval of the vector backbone was confirmed by Southern blotnalysis and phenotypic (hygromycin, kanamycin) susceptibilityesting.

(2015) 1353–1359

2.2. Growth of mycobacteria

BCG strains were grown in Middlebrook 7H9 supplementedwith oleic acid albumin dextrose catalase (OADC) (10% v/v) untilOD600 = 0.4 to 0.6. Bacterial aliquots were frozen and stored at−80 ◦C. Serial dilutions of thawed aliquots were plated on 7H10OADC agar plates and the number of colony forming units (CFU) wasdetermined after 3–4 weeks of incubation at 37 ◦C. The M. tuber-culosis H37Rv (National Collection of Type Cultures (NCTC) 7416)challenge stock was generated from a chemostat culture grownto steady state in a defined medium which has been previouslydescribed [22].

2.3. Vaccination

Groups of 8 pathogen-free Dunkin–Hartley guinea pigs, weigh-ing between 250 and 300 g, purchased from a commercial supplier(Harlan, UK), were used to evaluate the relative efficacy ofvaccination with BCG zmp1 mutants and standard BCG Danish1331 (Statens Serum Institute, Copenhagen, Denmark). BCG wasadministered subcutaneously in a single dose (5 × 104 CFU). Onevaccination group received phosphate buffered saline (PBS) as anegative control. Individual animals were identified using sub-cutaneously implanted microchips (PLEXX BV, The Netherlands).Guinea pig experimental work was conducted according to UKHome Office legislation for animal experimentation and wasapproved by the local ethics committee. All animals were weighedweekly and observed daily in order to monitor any adverseeffects.

2.4. Aerosol challenge

Guinea pigs were infected with a low aerosol dose (10–50 CFUretained dose in the lung) of M. tuberculosis H37Rv [23] 12 weeksafter vaccination. Aerosol challenge was performed using a fullycontained Henderson apparatus as previously described [24,25]in conjunction with the AeroMP (Biaera) control unit [26]. Fineparticle aerosols of M. tuberculosis H37Rv, with a mean diame-ter of 2 �m (diameter range 0.5–7 �m) [24] were generated usinga Collison nebulizer and delivered directly to the animal snout.The aerosol was generated from a water suspension containing5 × 106 CFU/ml M. tuberculosis H37Rv. The Henderson apparatusallows controlled delivery of aerosols to the animals and the repro-ducibility of the system and relationship between inhaled CFU andthe concentration of organisms in the nebulizer has been describedpreviously [25].

2.5. Post mortem procedures

Four weeks after challenge, guinea pigs were euthanised by anintraperitoneal injection of pentabarbitone sodium (Euthatal). Apost-mortem examination was performed immediately. The tho-racic “pluck” (comprising lung, trachea, heart and mediastinum)was dissected and removed. The left cranial, right cranial and rightcaudal lung lobes were placed in sterile containers for bacterio-logical analyses. The remaining tissue, including left middle, leftcaudal, right middle and right accessory lobes were fixed in 10%(v/v) neutral-buffered formalin (NBF) for histological examination.Sections of spleen were sampled separately either for bacteriolog-ical analysis or placed in 10% NBF for histological examination.

2.6. Histopathological examination

A representative sample from each lung lobes, sampled con-sistently between animals, and spleen, were processed routinelyand embedded in paraffin wax. Sections (approximately 5 �m)

ine 33

wsg

2

lV

2

s(p1Unapwsaa

2

l�tptacpt

ortlw

2

no1eb(mwe(

3

3p

s

P. Sander et al. / Vacc

ere stained with haematoxylin and eosin. Selected sections weretained with Alizarin red for the presence of calcium. The patholo-ist was blinded to the treatment groups.

.7. Morphometric analyses of lung and spleen histology sections

For each animal, the area of lung granulomas relative to the totalung area was analysed with an image analysis software (Matlabersion 7.9.0.529, The MathWorks TM).

.8. Bacterial analysis

Tissues for determination of CFU were homogenized in 5 ml ofterile deionized water using a rotating blade macerator systemYstral, UK). Viable counts were performed on the macerate byreparing serial dilutions in sterile deionized water and plating00 �l aliquots onto Middlebrook 7H11 OADC agar (BioMerieux,K). Plates were incubated at 37 ◦C for 3 weeks before counting theumber of M. tuberculosis colonies (CFU). Protection was primarilyssessed by measuring bacterial load in lungs and spleen com-aring the groups of animals vaccinated with BCG zmp1 mutantsith the control groups (BCG Danish 1331 and buffered saline). A

tatistical significant reduction in bacterial load in the vaccinatednimals when compared with the control groups was considereds a protective effect of the vaccine.

.9. CB-17 SCID mouse infection

Groups of 7–8 female CB-17 SCID mice were intravenously chal-enged with a single dose of the candidate vaccines Pasteur SmS

zmp1, BCG Denmark �zmp1 or the parental strains BCG Pas-eur SmS or BCG Denmark, in a total volume of 200 �l diluted inyrogen-free saline. Titrated frozen stocks of mycobacteria werehawed, centrifuged and resuspended in saline and injected. In par-llel aliquots from the prepared inoculum was plated for colonyounts. Pellets of centrifugations were compact and looked com-arable. The supernatants were clear. Bacterial suspensions wereested for lack of clumping prior to use in animals.

All animals included in the experiments were weighed at leastnce a week and observed daily in order to comply with ethicalequirements and to monitor any potential adverse effects relatedo BCG administration. In line with the UK Home Office guide-ines for welfare of experimental animals, animals were euthanized

hen they reached the defined endpoint of 20% weight loss.

.10. Statistics

The CFU data from the protection experiments were analysed byon-parametric Mann–Whitney U-test to compare median valuesf the various experimental groups using Minitab software version5. A p value of <0.05 was considered significant. Lung histom-try data were represented graphically, compared and evaluatedy Mann–Whitney U-test for statistically significant differencesp < 0.05) with the GraphPad Prism software. Survival curves and

edian survival times (MST) of each group of CB-17 SCID miceas compared using the Kaplan–Meier method. Statistical differ-

nces were assessed using GraphPad Prism using the Log RankMantel–Cox) test.

. Results

.1. Protective efficacy of BCG Pasteur SmR zmp1::aph in guinea

igs

We have previously characterized a zmp1 deletion mutant of atreptomycin resistant derivative of BCG Pasteur [17], referred here

(2015) 1353–1359 1355

to as BCG Pasteur SmR zmp1::aph. This mutant proved to be moreimmunogenic than its parental strain in various in vitro and in vivoimmunological murine assays as indicated by increased presenta-tion of MHC class II-restricted antigens, increased and accelerateddelayed-type hypersensitivity (DTH) response, antigen-specificlymphocyte proliferation and the frequency of antigen-specificinterferon-� producing lymphocytes [18]. In the absence of cor-relates of protection [27], vaccine candidates such as the zmp1mutants described here, have to be tested in animal models todemonstrate their ability to protect against virulent challengeprior to further testing in clinical trials. The guinea pig has beenused extensively for this purpose and has aided in identifyingpromising vaccine candidates [28]. To assess the protective effi-cacy of the BCG Pasteur SmR zmp1::aph strain in the guinea pigmodel a proof of concept study was conducted. In a head tohead comparison, guinea pigs were vaccinated subcutaneouslywith BCG Pasteur SmR zmp1::aph, BCG Danish (standard posi-tive control) or saline (negative control). The vaccine candidatewas well tolerated, as was the comparator. The weight of thevaccinated animals increased similar as the weight of the ani-mals in the non-vaccinated control group. Twelve weeks later,the guinea pigs were challenged with a low aerosol dose (10–50CFU) of M. tuberculosis H37Rv. Four weeks post challenge, guineapigs were euthanized and the bacterial load of M. tuberculosis inthe lungs and the spleens was determined. The median bacte-rial load in the lungs of unvaccinated guinea pigs was 6.33 log10,compared to 4.53 log10 in BCG Danish vaccinated animals and4.11 log10 in BCG Pasteur SmR zmp1::aph vaccinated animals(Fig. 1A). In the lungs, vaccination with both BCG Danish and theBCG Pasteur SmR zmp1::aph mutant resulted in a statistically sig-nificant reduction in bacterial load compared to non-vaccinatedcontrol (1.80 log10, p = 7.8 × 10−5 and 2.22 log10, p = 7.8 × 10−5,respectively) and the mutant conferred a statistically significantreduction of 0.42 log10 as compared to the BCG Danish control(p = 0.019).

Microscopic-examination of lung parenchyma from animals inthe saline group revealed numerous, often coalescing tubercu-lous granulomas, comprising abundant epithelioid macrophages,scattered lymphocytes, often around the periphery of the lesion,neutrophils and central necrosis. Large areas of lung parenchymawere affected (median = 16.39%) (Fig. 1B and D). By contrast,the lungs of BCG Danish and BCG Pasteur SmR zmp1::aphvaccinated animals had smaller, primarily non-necrotic granu-lomas, with fewer neutrophils than those lesions in the lungsof the control animals (Fig. 1D). Whist the lung of animalsvaccinated with BCG Danish contained fewer, large confluentlesions and scattered, smaller lesions with central, necrotic foci,necrosis was not observed in animals vaccinated with BCG Pas-teur SmR zmp1::aph. The affected area in the lungs of BCGDanish vaccinated animals was slightly but not significantly(p = 0.16) larger than the affected area in the lungs of BCGPasteur SmR zmp1::aph vaccinated animals (median 4.48% vs.2.99%) (Fig. 1B).

BCG vaccine strains reduced the bacterial load in the spleensimilarly (3.95 log10 for BCG Danish and 3.90 log10 for BCG Pas-teur SmR zmp1::aph, p = 0.48); the median CFU in the spleens ofnon-vaccinated guinea pigs was 5.61 log10, compared to 1.66 log10(p = 0.00045) and 1.71 log10 (p = 0.00042) in BCG Danish andBCG Pasteur SmR zmp1::aph vaccinated guinea pigs, respectively(Fig. 1C). Tuberculous granuloma were observed in the spleen of allanimals in the saline control group. By contrast, only one animalin the BCG Danish-vaccinated group, and two animals in the BCGPasteur SmR zmp1::aph-vaccinated group contained lesions in thespleen. A decrease in severity of lesions (or any at all) was present

in the spleen of animals vaccinated with either BCG Danish or BCGPasteur SmR zmp1::aph.

1356 P. Sander et al. / Vaccine 33 (2015) 1353–1359

Fig. 1. Protective efficacy of BCG Pasteur SmR zmp1::aph against tuberculosis in a guinea pig model. Vaccinated (BCG Pasteur SmR zmp1::aph; BCG Danish) or non-vaccinatedguinea pigs were challenged with virulent M. tuberculosis via the aerogenic route. M. tuberculosis bacterial burden within lungs (A) and spleens (C) was determined fourw ng wit ectionp :aph v

3

rpeaztPtc[tbuppw1t�TFlBv

eeks after challenge. M. tuberculosis counts in individual guinea pigs are given alootal lung area) of vaccinated and non-vaccinated guinea pigs four weeks after infigs (unvaccinated [left]; BCG Danish vaccinated [middle]; BCG Pasteur SmR zmp1:

.2. Protective efficacy of BCG Denmark �zmp1

In the proof of concept study described above, the streptomycinesistant BCG Pasteur SmR zmp1::aph mutant showed improvedrotective efficacy in guinea pigs compared to BCG Danish. How-ver, the use of this strain in humans is compromised by itsntibiotic resistance phenotype. The replacement of zmp1+ by themp1::aph allele introduced a kanamycin resistance cassette, andhe presence of an rpsL mutation already present in the parent BCGasteur SmR strain [29] confers streptomycin resistance. Accordingo the “Geneva consensus recommendation for novel live TB vac-ines”, live vaccines must be free of antibiotic resistance markers30]. In addition, differences in protective efficacy in pre-clinicalests and clinical vaccine trials have been attributed to genetic,iochemical and immunological differences in the BCG sub-strainssed [4,5,31,32]. To exclude the possibility that the difference inrotective efficacy of BCG Pasteur SmR zmp1::aph and the com-arator BCG Danish reflects the use of distinct BCG sub-strains,e generated an unmarked zmp1 deletion mutant in BCG Danish

331 background (BCG Denmark �zmp1) [19] and tested the pro-ective efficacy of this mutant as described above. BCG Denmark

zmp1 vaccine candidate was well tolerated by the guinea pigs.hey showed similar weight gain profiles as the control animals.

our weeks after the aerosol M. tuberculosis challenge, the medianung CFU counts were 6.27 log10 in non-vaccinated, 4.57 log10 inCG Danish vaccinated and 3.66 log10 in BCG Denmark �zmp1accinated guinea pigs (Fig. 2A). Importantly, the �zmp1 mutant

th the median and quartiles. (B) Quantitation of granuloma affected lung area (% of with M. tuberculosis. (D) Photomicrographs of lung sections from infected guineaaccinated [right]). Haematoxylin and eosin.

conferred a statistically significant improved protection over BCGDanish against growth of tubercle bacilli in the lung (0.91 log10;p = 0.019). Vaccination significantly reduced granuloma and necro-sis formation in the lung and the �zmp1 vaccinated animals wereslightly but not significantly better protected than BCG Danishvaccinated animals against pathological changes in the lung asassessed by histology (Fig. 2B and D). The median of the affectedarea of the lung was 33.63% for non-vaccinated, 5.62% for BCG Dan-ish and 4.75% for BCG Denmark �zmp1 vaccinated animals.

Median spleen CFU counts were 5.05 log10 for non-vaccinated,1.08 log10 BCG Danish vaccinated (p = 0.00045) and 1.08 log10 forBCG Denmark �zmp1 (p = 0.00045) vaccinated animals (Fig. 2C).The growth of M. tuberculosis in the spleen was not significantlydifferent in animals vaccinated with BCG Danish and BCG �zmp1(p = 0.65).

In summary, vaccination with either of the zmp1 deletionmutants consistently confers better protection in the lung than theBCG Danish control strain against M. tuberculosis challenge in theguinea pig model and this protection is independent of the geneticbackground of the BCG sub-strain.

3.3. Virulence of unmarked �zmp1 mutants in CB-17 SCID mice

Having demonstrated that the improved protection of the can-didate vaccine is a consistent property of zmp1 deletion andindependent of the BCG sub-strain, we constructed a �zmp1 dele-tion mutant without antibiotic resistance marker also in the BCG

P. Sander et al. / Vaccine 33 (2015) 1353–1359 1357

Fig. 2. Protective efficacy of BCG Denmark �zmp1 against tuberculosis. Vaccinated (BCG Denmark �zmp1; BCG Danish) or non-vaccinated guinea pigs were challengedwith virulent M. tuberculosis via the aerogenic route. M. tuberculosis CFU bacterial burden within lungs (A) and spleens (C) was determined four weeks after challenge. M.t and qv rculos[ aema

PpitBTsStt(wmvht

TB

entered in clinical trials. It has been hypothesized that BCG, derivedfrom virulent M. bovis is over-attenuated and therefore only con-fers limited protection [33]. However, an episode of tuberculosis

uberculosis CFU counts in individual guinea pigs are given along with the medianaccinated and non-vaccinated guinea pigs four weeks after infection with M. tubeleft]; BCG Danish vaccinated [middle]; BCG Denmark �zmp1 vaccinated [right]). H

asteur SmS background, BCG Pasteur SmS �zmp1. We addressedossible unexpected adverse effects of unmarked zmp1 deletion by

nfecting immune-compromised CB-17 SCID mice and monitoringheir survival. CB-17 SCID mice were infected with BCG Denmark,CG Denmark �zmp1, BCG Pasteur SmS and Pasteur SmS �zmp1.he input doses (CFU/mouse) are given in Table 1. The medianurvival time of CB-17 SCID mice infected with the BCG PasteurmS �zmp1 deletion mutant was 186 days and significantly longerhan the median survival time of CB-17 SCID mice infected withhe same dose of the corresponding BCG Pasteur parental strainFig. 3; Table 1; p < 0.0001). Likewise, all CB-17 SCID mice infectedith BCG Denmark reached their humane endpoint before the firstouse infected with BCG Denmark �zmp1 did, the median sur-

ival times being 127 and 278 days (Fig. 3; Table 1; p < 0.0001),

ence demonstrating enhanced safety of the �zmp1 mutants overheir respective parental BCG strains.

able 1CG inoculation doses and median survival times of CB-17 SCID mice.

Strain Actual input(CFU/mouse)

Mediansurvival time(days)

Significance(p value)

BCG Pasteur SmS 1.0 × 106 135.5 <0.0001BCG Pasteur SmS �zmp1 0.98 × 106 186.5BCG Denmark 1.2 × 105 127.0 <0.0001BCG Denmark �zmp1 1.8 × 105 278.0

uartiles. (B) Quantitation of granuloma affected lung area (% of total lung area) ofis. (D) Photomicrographs of lung sections from infected guinea pigs (unvaccinatedtoxylin and eosin.

4. Discussion

BCG is still the only licensed TB vaccine despite its limitedprotective efficacy against adult lung TB [3]. Even a modestimprovement in potency over BCG would translate into hundredsof thousands of saved lives in the next years. The past decadehas seen numerous efforts to improve vaccine-mediated pro-tection against tuberculosis and several vaccine candidates have

Fig. 3. Virulence of BCG �zmp1 in CB-17 SCID mice. Kaplan–Meyer survival curves ofCB-17 SCID mice infected with a defined dose of BCG Pasteur SmS, BCG Pasteur SmS�zmp1, BCG Denmark and BCG Denmark �zmp1 (for details of statistical analysissee: Table 1).

1 cine 33

lftuttzimziatscmobep1tmpBpwBit0

acTveBmpltaBtlwi

gmTasilmo

C

A

(

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

358 P. Sander et al. / Vac

ikewise only confers limited protection against subsequent rein-ection [34]. The limited post-infective immunity may indicatehat M. tuberculosis escapes immune surveillance and that thenderlying mechanisms are preserved in BCG [18]. Mutants of M.uberculosis and M. bovis BCG deficient in zmp1 zinc metallopep-idase are unable to arrest phagosome maturation [17] and a BCGmp1 mutant has been demonstrated to enhance immunogenic-ty of mycobacterial antigens [18,19]. Our previous investigations

ainly focused on virulence, cellular and immunological aspects ofmp1 mutant strains and on the characterization of the biochem-cal and structural properties of the enzyme [15,16,35]. We nowdd an important aspect towards exploitation of BCG zmp1 dele-ion mutants as TB vaccine by characterizing the protective andafety properties of corresponding BCG mutants. Protective effi-acy of BCG zmp1 deletion mutants was tested in the guinea pigodel which is considered the most relevant small animal model

f pulmonary tuberculosis [36]. Guinea pigs are highly suscepti-le towards a small number of virulent M. tuberculosis bacteria andventually succumb to infection due to weight loss and decreasedulmonary function. BCG confers significant protection (approx.00-fold reduction in CFU counts) in this model and improved pro-ection over BCG is a most stringent criterion, which has only been

et by few vaccine candidates [28,37–40]. Two completely inde-endent experiments with zmp1 mutant strains in two differentCG genetic backgrounds were conducted in a head-to-head com-arison. Each experiment included a positive control (vaccinationith BCG Danish) and a negative control (saline). Compared toCG Danish, which conferred approximately 1.8 log10 protection

n terms of CFU counts in the lung, both zmp1 mutants of BCG Pas-eur and BCG Denmark further improved the level of protection by.4–0.9 log10.

Attenuated live vaccines (e.g. BCG or attenuated M. tuberculosis)lone or in combination with subunit booster vaccines are oftenonsidered as the most promising approach for developing a betterB vaccine [30]. However, safety issues are a concern for any liveaccine since a high incidence of HIV is found in areas where TB isndemic. A novel live vaccine should therefore at least be as safe asCG. Safety of BCG Denmark �zmp1 and BCG Pasteur SmS �zmp1utants was tested in immunocompromised CB-17 SCID mice in

arallel with their parental strains. Median survival time was pro-onged for 50–150 days for each of the mutant strains, indicatinghat zmp1 deletion mutants of BCG Denmark and BCG Pasteur arettenuated as compared to their parental strains. Attenuation ofCG Denmark �zmp1 and BCG Pasteur SmS �zmp1 mutants inhis experiment is in agreement with attenuation of M. tubercu-osis zmp1 deletion mutant in immune competent mice [17], while

ild-type like persistence of BCG Pasteur SmR zmp1::aph was seenn immune competent mice [18].

Based on superior TB protective efficacy in the highly strin-ent guinea pig model and the good safety profile, BCG �zmp1utants are promising candidates for further vaccine development.

he initial proof of concept protection study was conducted with streptomycin resistant derivative of BCG Pasteur. For subsequenttudies, unmarked BCG �zmp1 deletion mutants were constructedn order to meet the “Geneva consensus recommendation for novelive TB vaccines” [30]. These strains may now be exploited for “Good

anufacturing production” for the final stages of pre-clinical devel-pment and subsequent clinical trials.

onflict of interest statement

PS and ECB are inventors of BCG zmp1.

cknowledgements

This work was supported by the University of ZurichUZH), the Swiss National Foundation (SNF 31003A 153349), the

[

[

(2015) 1353–1359

European Union (EU-FP7, NewTBVAC No 241 745) and the Tuber-culosis Vaccine Initiative (TBVI). PJ was supported by the SwissNational Foundation (31003A 149323). We gratefully acknowledgethe support from the Biological Investigations Group at PHE, PortonDown and Mr Robert Gilbert and other members of the BSF staff atLSHTM.

References

[1] Dye C, et al. Prospects for tuberculosis elimination. Annu Rev Public Health2013;34:271–86.

[2] Brosch R, et al. Genome plasticity of BCG and impact on vaccine efficacy. ProcNatl Acad Sci USA 2007;104(13):5596–601.

[3] Andersen P, Kaufmann SH. Novel vaccination strategies against tuberculosis.Cold Spring Harbor Perspect Med 2014;4(6).

[4] Fine PEM. The BCG story—lessons from the past and implications for the future.Rev Infect Dis 1989;11:S353–9.

[5] Colditz GA, et al. Efficacy of BCG vaccine in the prevention of tuberculosis.Meta-analysis of the published literature. JAMA 1994;271(9):698–702.

[6] Grode L, et al. Increased vaccine efficacy against tuberculosis of recombinantMycobacterium bovis bacille Calmette-Guérin mutants that secrete listeriolysin.J Clin Invest 2005;115(9):2472–9.

[7] Cardona PJ, et al. Extended safety studies of the attenuated live tuberculosisvaccine SO2 based on phoP mutant. Vaccine 2009;27(18):2499–505.

[8] Sweeney KA, et al. A recombinant Mycobacterium smegmatis inducespotent bactericidal immunity against Mycobacterium tuberculosis. Nat Med2011;17(10):1261–8.

[9] Aagaard C, et al. A multistage tuberculosis vaccine that confers efficient pro-tection before and after exposure. Nat Med 2011;17(2):U189–224.

10] McShane H, et al. Recombinant modified vaccinia virus Ankara expressingantigen 85A boosts BCG-primed and naturally acquired antimycobacterialimmunity in humans (vol 10: pg 1240, 2004). Nat Med 2004;10(12):1397-1397.

11] Ottenhoff TH, Kaufmann SH. Vaccines against tuberculosis: where are we andwhere do we need to go? PLoS Pathog 2012;8(5):pe1002607.

12] Kaufmann SH, Gengenbacher M. Recombinant live vaccine candidates againsttuberculosis. Curr Opin Biotechnol 2012;23(6):900–7.

13] Tameris MD, et al. Safety and efficacy of MVA85A, a new tuberculosis vaccine,in infants previously vaccinated with BCG: a randomised, placebo-controlledphase 2b trial. Lancet 2013;381(9871):1021–8.

14] Sun R, et al. Novel recombinant BCG expressing perfringolysin O and the over-expression of key immunodominant antigens: pre-clinical characterization:safety and protection against challenge with Mycobacterium tuberculosis. Vac-cine 2009;27(33):4412–23.

15] Ferraris DM, et al. Crystal structure of Mycobacterium tuberculosis zinc-dependent metalloprotease-1 (Zmp1): a metalloprotease involved inpathogenicity. J Biol Chem 2011;286(37):32475–82.

16] Petrera A, et al. Functional characterization of the Mycobacterium tuberculo-sis zinc metallopeptidase Zmp1 and identification of potential substrates. BiolChem 2012;393(7):631–40.

17] Master SS, et al. Mycobacterium tuberculosis prevents inflammasome activation.Cell Host Microbe 2008;3(4):224–32.

18] Johansen P, et al. Relief from Zmp1-mediated arrest of phagosome maturationis associated with facilitated presentation and enhanced immunogenicity ofmycobacterial antigens. Clin Vaccine Immunol 2011;18(6):907–13.

19] Khatri B, et al. BCG Deltazmp1 vaccine induces enhanced antigen specificimmune responses in cattle. Vaccine 2014;32(7):779–84.

20] Vordermeier HM, et al. Adjuvants induce distinct immunological phen-otypes in a bovine tuberculosis vaccine model. Clin Vaccine Immunol2009;16(10):1443–8.

21] Pelicic V, Reyrat JM, Gicquel B. Expression of the Bacillus subtilis sacB geneconfers sucrose sensitivity on mycobacteria. J Bacteriol 1996;178(4):1197–9.

22] James BW, Williams A, Marsh PD. The physiology and pathogenicity of Mycobac-terium tuberculosis grown under controlled conditions in a defined medium. JAppl Microbiol 2000;88(4):669–77.

23] Williams A, et al. Comparison of the protective efficacy of bacille Calmette-Guérin vaccination against aerosol challenge with Mycobacterium tuberculosisand Mycobacterium bovis. Clin Infect Dis 2000;30(Suppl. 3):S299–301 (an offi-cial publication of the Infectious Diseases Society of America).

24] Lever MS, Williams A, Bennett AM. Survival of mycobacterial species in aerosolsgenerated from artificial saliva. Lett Appl Microbiol 2000;31(3):238–41.

25] Clark SO, et al. Survival of Mycobacterium tuberculosis during experimentalaerosolization and implications for aerosol challenge models. J Appl Microbiol2011;111(2):350–9.

26] Hartings JM, Roy CJ. The automated bioaerosol exposure system: preclinicalplatform development and a respiratory dosimetry application with nonhumanprimates. J Pharmacol Toxicol Methods 2004;49(1):39–55.

27] Nunes-Alves C, et al. In search of a new paradigm for protective immunity toTB. Nat Rev Microbiol 2014;12(4):289–99.

28] Williams A, et al. Evaluation of vaccines in the EU TB Vaccine Cluster usinga guinea pig aerosol infection model of tuberculosis. Tuberculosis (Edinb)2005;85(1–2):29–38.

29] Weber W, et al. A synthetic mammalian gene circuit reveals antituberculosiscompounds. Proc Natl Acad Sci USA 2008;105(29):9994–8.

ine 33

[

[

[

[

[

[

[

[

[

P. Sander et al. / Vacc

30] Walker KB, et al. The second Geneva consensus: recommendations for novellive TB vaccines. Vaccine 2010;28(11):2259–70.

31] Keyser A, et al. BCG sub-strains induce variable protection against virulent pul-monary Mycobacterium tuberculosis infection: with the capacity to drive Th2immunity. Vaccine 2011;29(50):9308–15.

32] Kozak R, Behr MA. Divergence of immunologic and protective responses ofdifferent BCG strains in a murine model. Vaccine 2011;29(7):1519–26.

33] Behr MA, Small PM. Has BCG attenuated to impotence? Nature1997;389(6647):133–44.

34] van Rie A, et al. Exogenous reinfection as a cause of recurrent tuberculosis aftercurative treatment. N Engl J Med 1999;341(16):1174–9.

35] Mori M, et al. Discovery of the first potent and selective Mycobac-terium tuberculosis Zmp1 inhibitor. Bioorg Med Chem Lett 2014;24(11):2508–11.

[

[

(2015) 1353–1359 1359

36] Grover A, et al. Assessment of vaccine testing at three laboratories using theguinea pig model of tuberculosis. Tuberculosis (Edinb) 2012;92(1):105–11.

37] Martin C, et al. The live Mycobacterium tuberculosis phoP mutant strain is moreattenuated than BCG and confers protective immunity against tuberculosis inmice and guinea pigs. Vaccine 2006;24(17):3408–19.

38] Brandt L, et al. The protective effect of the Mycobacterium bovis BCG vac-cine is increased by coadministration with the Mycobacterium tuberculosis72-kilodalton fusion polyprotein Mtb72F in M. tuberculosis-infected guineapigs. Infect Immun 2004;72(11):6622–32.

39] Horwitz MA, Harth G. A new vaccine against tuberculosis affords greater sur-vival after challenge than the current vaccine in the guinea pig model ofpulmonary tuberculosis. Infect Immun 2003;71(4):1672–9.

40] Williams A, et al. Boosting with poxviruses enhances Mycobacterium bovis BCGefficacy against tuberculosis in guinea pigs. Infect Immun 2005;73(6):3814–6.