MICROBIAL DRUG RESISTANCEVolume 6, Number 4, 2000Mary Ann fiebert. Inc.

iTMCorrelation of pncA Sequence with

Pyrazinamide Resistance Level in BACTEC,M for21 Mycobacterium tuberculosis Clinical Isolates

MARIAN MESTDAGH,1 LAURENCE REALINI,2 PIERRE-ALAIN FONTEYNE,2 RUDI ROSSAU,3GEERT JANNES,3 WOUTER MUS,3 KOENRAAD A.L. DE SMET,3 FRANÇOISE PORTAELS,2

and ELFRIDE VAN DEN EECKHOUT1

ABSTRACT

Mutations in the pncA gene, encoding pyrazinamidase, are considered the major mechanism of pyrazinamide(PZA) resistance in Mycobacterium tuberculosis, but resistant strains containing the wild-type gene have beendescribed. The correlation of pncA sequence with PZA resistance level was examined for 21 M. tuberculosisclinical isolates. Susceptibility patterns were determined for 100, 300, and 900 pg/m\ concentrations of thedrug in BACTEC™. Insertions and deletions and a substitution in the putative promoter region led to high-level resistance, whereas substitutions within the open reading frame seemed to confer variable levels of re-sistance. Variable resistance levels and PZase activities were also observed among isolates lacking pncA mu-tations. The high-level resistance (900 /u.g/ml) in pncA wild-type isolates highlights the clinical significance ofthese isolates. These data also suggest that there may still be more than one alternative mechanism leading toPZA resistance in M. tuberculosis isolates.

INTRODUCTION

Pyrazinamide (PZA) is a nicotinamide analogue of whichthe antituberculous activity was discovered in 1952.4 As a

result of its unique action against semidormant bacilli, adminis-tration of this drug in combination with isoniazid, rifampin, andoften ethambutol allowed shortening of tuberculosis treatmentfrom 2 years to 6 months.1,7 Except for the assumption that PZAis a prodrug that is converted by the bacterial enzyme pyrazi-namidase (PZase) to the active pyrazinoic acid (POA), little isunderstood of its mechanism of action.4 The observation that themajority of PZA-resistant Mycobacterium tuberculosis strainslost their PZase activity, led to the characterization of the en-

coding gene, pncA.9 Single mutations, dispersed over virtuallythe entire length of this gene, are considered the major mecha-nism of PZA resistance in M. tuberculosis,9'10'12 but absence ofpncA mutations in resistant isolates has been observed.1012 How-ever, little is known about the impact of specific pncA mutationson the level of resistance to PZA. To examine the correlationbetween specific pncA mutations and corresponding PZA resis-

tance levels, susceptibility patterns in BACTEC™ PZA test vialscontaining 100, 300, and 900 pg/ml of the drug, respectively,were determined for 16 M. tuberculosis isolates, of which pncAmutations have been published elsewhere.6

In an attempt to identify high-level resistance, suggestive ofa clinically relevant secondary PZA resistance mechanism, sus-

ceptibility patterns were determined in a similar way for a setof five M. tuberculosis isolates known to contain the wild-typepncA sequence, and found resistant to 100 /xg/ml of PZA inBACTEC.6

MATERIALS AND METHODS

Bacterial strains

Fourteen multidrug-resistant M. tuberculosis strains were

isolated from clinical specimens, mostly sputa, collected inAzerbaijan (five), Bangladesh (four), Siberia (two), Rwanda(two), and Congo-Brazzaville (one). Six isolates were obtained

'Laboratory for Pharmaceutical Biotechnology, FFW, University of Ghent, Ghent, Belgium.2Mycobacteriology Unit, Institute of Tropical Medicine, Antwerp, Belgium.3Innogenetics n.v., Ghent, Belgium.

283

284 MESTDAGH ET AL.

from the Scottish Mycobacteria Reference Laboratory (SMRL),whereas one isolate was a quality control strain.5 M. tubercu-losis H37Ra (NCTC 7417) was included as a PZA-susceptiblecontrol strain. One M. bovis isolate from Belgium and one fromEgypt from the collection of the Institute of Tropical Medicinewere used as PZA-resistant control organisms. M. tuberculosisstrains were grown to confluence on Löwenstein-Jensenmedium and M. bovis strains on Stonebrink medium.

PZA susceptibility testingIsolates were tested for PZA susceptibility with the

BACTEC™ radiometric method2 in BACTEC™ PZA TestMedium Culture Vials (Becton Dickinson Diagnostic Instru-ment Systems, Townson, MD) containing 100, 300, and 900pg/ml of PZA, respectively, according to the manufacturer'sinstruction. ' '

PZase assayPZase activity of all isolates was assayed in a previous

study.6Colorimetric detection of PZA conversion

This assay was adapted from a test described by Wayne.13Briefly, isolates were subcultured in Dubos Tween-albumin liq-uid medium and resuspended in Dubos Tween-albumin liquidmedium containing 200 pg/ml of PZA, to yield a final con-

centration of bacteria of about 100 mg (wet weight) per ml.Aliquots ( 1,400 pi) of the resuspended cultures were taken im-mediately, and 24, 48, 72, 96, 144, and 312 hr after resuspen-sion. 400 pi of a 25% trichloroacetic acid solution was added,

and mixtures were centrifuged directly afterwards. One milli-liter of the supernatants was added to 2.5 ml of a 0.1 M phos-phate buffer at pH 7.0, and optical density at 470 nm was mea-

sured to obtain a blank value. After addition of 100 pi of a

freshly prepared 20% ferrous(II) ammonia sulfate (Acros Chim-ica, Geel, Belgium) solution, POA formation was detected mea-

suring the optical density at 470 nm again. Optical densities ofa series of POA solutions with known concentrations were mea-

sured identically to construct a standard curve for conversion.M. tuberculosis H37Ra (NCTC 7417) and two PZA-suscepti-ble M. tuberculosis clinical isolates (97-12 and 97-38) were in-cluded as PZase-positive control strains, and M. bovis BCG(98-8269) and one M. bovis isolate (96-1661) were used as

PZase-negative control strains, respectively.

RESULTS

The results of the BACTEC™ PZA susceptibility tests for 21M. tuberculosis strains with known pncA sequences6 are listedin Table 1. Fifteen isolates showed resistance to 900 pg/ml ofPZA, three were resistant to 300 pg/ml PZA concentration, andthe remaining three isolates only confirmed resistance to the100 pg/ml breakpoint concentration, recommended for PZAsusceptibility testing by the radiometric method, the "gold stan-dard" for PZA susceptibility testing.

Of the five isolates lacking any pncA mutation, two were re-sistant to the 900 pg/ml concentration of PZA (97-58 and 97-69), two showed resistance to 300 pg/ml (94-735 and 97-1052),and the remaining isolate (97-983) was only resistant in the 100/xg/ml test concentration. Isolate 94-735 was the only one stillpositive on PZase testing.

Table 1. Pyrazinamide Susceptibility Patterns, Changes in pncA Nucleotide Sequence, and CorrespondingAmino Acid Changes for 21 Pyrazinamide-Resistant Mycobacterium tuberculosis Isolates

Isolate OriginPZA resistance level*

(pg/ml)Changes in

Nucleotide sequence Amino acid sequence

97-983 Bangladesh97-854 Bangladesh97-553 Ruanda94-735 Congo97-1052 Azerbaijan97-826 Quality control strain97-58 Siberia97-69 Siberia98-31 Scotland97-1174 Azerbaijan97-718 Azerbaijan98-34 Scotland97-1006 Bangladesh97-578 Bangladesh98-32 Scotland98-36 Scotland98-29 Scotland98-30 Scotland97-720 Azerbaijan97-1069 Azerbaijan8251 Rwanda

100100100300300300900900900900900900900900900900900900900900900

NoG -> A at 511

G at 514 andT -> C at 515NoNo

T -> C at 307NoNo

C ->Tat 151C -> A at 185T -> G at 254G -> C at 406A -> C at 410C -> T at 425C -> T at 512A -> G at -11

G insertion 221-222G insertion 221-222C insertion 407-408

A 68-bp 195-263A C at 512

AL

No171 T172 ANoNo

Y 103 HNoNo

H 51 YP62HL85R

D 136 HH 137 PT 142 MA 171 V

NoFrameshiftFrameshiftFrameshiftFrameshiftFrameshift

'Highest concentration of PZA to which the isolate was resistant in BACTEC™ PZA test vials.

—•—96-1661 M. bovis

-

m- -

98-269 M. bovis BCG

—

A- M. tuberculosisH37Ra

200Time of incubation in Dubos tween-albumin medium containing 200 ug per ml.

of PZA (Hours)

50 100 150 200 250 300Time of incubation in Dubos tween-albumin medium containing 200 ug per mL of PZA (Hours)

50 100 150 200 250 300Time of incubation in Dubos tween-albumin medium containing 200 ug per mL of PZA

(Hours)

350

FIG. 1. Detection of remaining pyrazinamidase activity in pyrazinamide-resistant M. tuberculosis isolates.

286 MESTDAGH ET AL.

Thirteen of the 16 isolates (81%) containing pncA mutationswere resistant to the highest concentration of PZA used, namely900 /xg/ml. Lack of pyrazinamidase activity was correlated withmutations resulting in amino acid substitutions in seven of them,with frameshifts by single nucleotide insertion or deletion infour of them, a large deletion in one isolate, and with a pro-moter mutation in another isolate. One pncA-mutated, PZase-negative isolate bearing a point mutation resulting in an aminoacid substitution, showed resistance to 300 /xg/ml (97-826). Sur-prisingly, two of the 16 isolates with pncA mutations turned outnot to be resistant to any of both concentrations higher than the100 /xg/ml normal test concentration (97-553 and 97-854). Bothisolates were PZase-negative and their pncA mutations con-

sisted of single base substitutions, resulting in amino acidchanges.

For eight of these PZA-resistant isolates, remaining PZaseactivity was determined by colorimetric detection of POA for-mation using a slight modification of a method described byWayne.13 In Fig. 1, A, B and C, data are presented for the con-

trol strains (M. tuberculosis H37Ra, M. bovis, and M. bovisBCG), M. tuberculosis clinical isolates containing the wild-typepncA gene (97-12, 97-38, 94-735, 97-983, and 97-1052), andM. tuberculosis clinical isolates with pncA mutations (8251, 97-826, 97-1069, 983-32, and 98-36), respectively. Isolate 94-735(see Fig. 1C) was the only one still producing an amount ofPOA, comparable to that produced by the PZA-susceptible con-

trol strains M. tuberculosis H37Ra, 97-12, and 97-38. This isin total agreement with the PZase test results.

All seven other PZA-resistant isolates tested converted PZAto yield only lower concentrations of POA. In Fig. 2, POA pro-duction at the 312-hr time point is shown in function of PZAresistance level for the eight isolates. No correlation was foundbetween the level of PZA resistance, as tested in BACTEC™,and the amount of PZase activity remaining in the PZA-resis-

tant isolates tested, as the isolate 97-826, which was resistantto only 300 /xg/ml, and the isolate 97-1069, which showed re-

sistance to 900 /xg/ml, present identical curves of POA con-

version (see Fig. 1C). Moreover, isolates 98-32 and 98-36,which showed resistance to 900 /xg/ml of PZA, have greaterPZase activity than isolate 97-983, which was resistant to only100 /xg/ml of PZA in BACTEC™.

DISCUSSION

Routine testing for susceptibility to PZA in the clinical lab-oratory is rarely done, because the test has to be performed atan acid pH (±5.5), test conditions at which many strains failto grow.3 Because mutations in the pncA gene are consideredthe major mechanism of PZA resistance in tubercle bacilli, se-

quence analysis has been proposed as an alternative method fordetection of resistance. Early studies showed a good correla-tion between PZA resistance and pncA mutations, but additionalinvestigations including more strains revealed a significantnumber of PZA-resistant M. tuberculosis lacking pncA muta-tions.8 In preceding studies,9,10 these kinds of strains were foundto be falsely resistant upon retesting or defined borderline re-

sistant (MIC 200-300 fig/ml), but considered to be still re-

sponsive to PZA treatment after testing in mice. In the presentstudy, we intended to verify the level of resistance and, conse-

quently, the clinical relevance of these PZA-resistant M. tu-berculosis lacking pncA mutations. For four of the five isolatesanalyzed here, the MIC in BACTEC™ was greater than 300/xg/ml of PZA; two of these isolates showed an MIC even

greater than 900 /xg/ml of PZA. These findings demonstrate theexistence of high-level, clinically relevant PZA resistance in M.tuberculosis due to other mechanisms.

Three isolates resistant to 100 /xg/ml of PZA, two of which

200

180

160 +

5 140|3 120co

I 100 i

O 80u«tO0. 60

40

20--

300PZA resistance level (ug/mL)

FIG. 2. POA produced by M. tuberculosis isolates in function of their PZA resistance level.aIsolate containing the wild-type pncA gene.

PYRAZINAMIDE RESISTANCE IN M. TUBERCULOSIS 287

having acquired pncA mutations, were susceptible to 300 /ig/mlof PZA in BACTEC™ in this study. This confirms the 100¡xglml of PZA concentration to be the accurate concentrationfor PZA susceptibility testing in BACTEC™. Using a higherconcentration of PZA in BACTEC™ PZA Test Vials might thuslead to false-susceptible test results, which seem to have even

worse consequences, like inclusion of PZA in the patient's treat-ment, than false-resistant readings.

One could assume that mutations leading to high-level ré-sistance, to 900 pg/ml of PZA in this study, are related with a

total loss of PZase activity. For all four isolates resistant to 900pg/ml of PZA, and tested for eventual remaining PZase activ-ity, a substantial reduction of enzyme activity had occurred. Theisolate 98-32, however, which is bearing a C —» T substitu-tion at position 512 of the open reading frame and showed re-

sistance to 900 pg/ml in BACTEC™, produced POA in a

slightly higher amount than the six other PZase-negative iso-lates. Repeated testing of this isolate for PZase activity sug-gested the isolate to be rather borderline (PZase-negative), in-stead of PZase-positive, as listed before.6 This prompted us toassume that the breakpoint for positivity in the PZase assay cor-

responds to production of POA at a concentration of about55-60 pg/ml (optical densities of about 0.035-0.040) under thePZA conversion test conditions used in this study.

Production of POA from PZA appeared to be not linear over

time in all samples. This lack of linearity could be explainedmainly by variability in viability of bacteria during this 13 days'test period, which is consequently causing changes in enzymaticactivities of the processed samples. However, a delay in PZAuptake or a decreased uptake of PZA could also play a role.When comparing PZA conversion rates of both PZA-suscepti-ble (multidrug-resistant) clinical isolates (97-12 and 97-38) andof the control strain M. tuberculosis H37Ra, differences can beexplained by the lower bacterial fitness encountered with mul-tidrug-resistant mycobacteria, as experienced in measuringdaily growth indexes in BACTEC™ (data not shown).

From this PZA conversion test, it can also be seen that twodifferent isolates, both resistant to 300 ^ig/ml of PZA inBACTEC™, and both containing the wild-type pncA sequence,one (94-735) showed a level of PZase activity comparable tothat of the PZA-susceptible control isolates, whereas the sec-

ond one (97-1052) produced POA only in a very small amount.Therefore, we suggest that at least two different unrevealed PZAresistance mechanisms exist among M. tuberculosis clinical iso-lates. Resistance to PZA for the lone PZase-positive isolate (94-735) lacking pncA mutations may be due to mutations in thePOA drug target, not yet been found, whereas PZase-negativestrains lacking pncA mutations could have undergone down-regulation of the pncA gene, or they might display a defect inPZA uptake. Further experiments including PZA uptake stud-ies, and pncA expression studies, are needed to clarify these hy-potheses.

ACKNOWLEDGMENTS

We thank Becton Dickinson for use of the BACTEC 460 TBinstrument. The unit of Mycobacteriology was supported by a

grant of the FWO (Fonds voor Wetenschappelijk Onderzoek),and M. Mestdagh was supported by the I.W.T. (Flanders' Gov-ernment). We also kindly thank M. Fauville-Dufaux (Institut

Pasteur, Brussels), A. Laszlo (CDC), and J. Norton (MurexBiotech Limited) for providing us with M. tuberculosis isolates.

REFERENCES

1. Heifets, L.B., and P. Lindholm-Levy. 1992. Pyrazinamide steril-izing activity in vitro against semidormant Mycobacterium tuber-culosis populations. Am. Rev. Respir. Dis. 145:1223-1225.

2. Heifets, L.B., and R.C. Good. 1994. Current laboratory methodsfor the diagnosis of tuberculosis. In B.R. Bloom (ed.), Tuberculo-sis: Pazthogenesis, Protection and Control. American Society forMicrobiology, Washington, DC, pp. 85-110.

3. Hewlett, D.Jr., D.L. Horn, and C. Alfalla. 1995. Drug-resistanttuberculosis: inconsistent results of pyrazinamide susceptibilitytesting. J. Am. Med. Assn. 273:916-917.

4. Konno, K., F.M. Feldmann, and W. McDermott. 1967. Pyrazi-namide susceptibility and amidase activity of tubercle bacilli. Am.Rev. Respir. Dis. 95:461^169.

5. Laszlo, A., M. Rahman, M. Raviglione, F. Bustreo, andWHO/IUATLD Network of Supranational Reference Labora-tories. 1997. Quality assurance programme for drug susceptibilitytesting of Mycobacterium tuberculosis in the WHO/IUATLDSupranational Network: first round of proficiency testing. Int. J.Tuberc. Lung Dis. 1:231-238.

6. Mestdagh, M., P.A. Fonteyne, L. Realini, R. Rossau, G. Jannes,W. Mijs, K.A.L. De Smet, F. Portaels, and E. Van den Eeck-hout. 1999. Relationship between pyrazinamide resistance, loss ofpyrazinamidease activity, and mutations in the pncA locus in mul-tidrug-resistant clinical isolates of Mycobacterium tuberculosis.Antimicrob. Agents Chemother. 43:2317-2319.

7. Mitchison, D.A. The action of antituberculosis drugs in short-course chemotherapy. 1985. Tubercle 66:219-225.

8. Sander, P., and E.C. Böttger. 1999. Mycobacteria: genetics of re-

sistance and implications for treatment. Chemotherapy 45:95-108.9. Scorpio, A., and Y. Zhang. 1996. Mutations in pncA, a gene en-

coding pyrazinamidase/nicotinamidase, cause resistance to the an-

tituberculous drug pyrazinamide in tubercle bacillus. Nature Med.2:662-667.

10. Scorpio, A., P. Lindholm-Lévy, L. Heifets, R. Gilman, S. Sid-diqi, M. Cynamon, and Y. Zhang. 1997. Characterization of thepncA mutations in pyrazinamide-resistant Mycobacterium tuber-culosis. Antimicrob. Agents Chemother. 41:540-543.

11. Siddiqi, S.H. 1989. In BACTEC TB System, Product and Proce-dure Manual. Becton Dickinson Diagnostic Instrument Systems,Townson, MD.

12. Sreevatsan, S., X. Pan, Y. Zhang, B.N. Kreiswirth, and J.M.Musser. 1997. Mutations associated with pyrazinamide resistancein pncA of Mycobacterium tuberculosis complex organisms. An-timicrob. Agents Chemother. 41:636-640.

13. Wayne, L.G. Simple pyrazinamidase and urease tests for routineidentification of mycobacteria. 1974. Am. Rev. Respir. Dis. 109:147-151.

14. Yeager, R.L., W.G.C. Munroe, and F.I. Dessau. 1952. Pyrazi-namide (Aldinamide) in the treatment of pulmonary tuberculosis.Am. Rev. Tuberc. 65:523-534.

Address reprint requests to:

Prof. dr. E. Van den EeckhoutLaboratory for Pharmaceutical Biotechnology

FFW, University of GhentHarelbekestraat 72, 9000 Ghent, Belgium

E-mail: Elfriede.VandenEeckhout@rug.ac.be