Azole antifungals are potent inhibitors of cytochrome P450 mono-oxygenases and bacterial growth ...

13 Pages·2002·0.42 MB·English

Microbiology(2002),148,2937–2949 PrintedinGreatBritain
Azole antifungals are potent inhibitors of
cytochrome P450 mono-oxygenases and
bacterial growth in mycobacteria and
streptomycetes
Kirsty J. McLean,1 Ker R. Marshall,1 Alison Richmond,2 Iain S. Hunter,2
Kay Fowler,3 Tobias Kieser,3 Sudagar S. Gurcha,4 Gurydal S. Besra4
and Andrew W. Munro1
Authorforcorrespondence:AndrewW.Munro.Tel:›441162523464.Fax:›441162523630.
e-mail:awm9!le.ac.uk
1Departmentof ThegenomesequenceofMycobacterium tuberculosishasrevealedthe
Biochemistry,Universityof presenceof20differentcytochromeP450mono-oxygenases(P450s)withinthis
Leicester,TheAdrian
organism,andsubsequentgenomesequencesofothermycobacteriaandof
Building,UniversityRoad,
LeicesterLE17RH,UK Streptomycescoelicolorhaveindicatedthattheseactinomycetesalsohave
2DepartmentofBiosciences, largecomplementsofP450s,pointingtoimportantphysiologicalrolesfor
UniversityofStrathclyde, theseenzymes.TheactinomyceteP450sincludehomologuesof14a-sterol
TheRoyalCollege,204 demethylases,thetargetsfortheazoleclassofdrugsinyeastandfungi.
GeorgeStreet,Glasgow
Previously,thistypeofP450wasconsideredtobeabsentfrombacteria.When
G11XL,UK
presentatlowconcentrationsingrowthmedium,azoleantifungaldrugswere
3DepartmentofGenetics,
showntobepotentinhibitorsofthegrowthofMycobacterium smegmatisand
JohnInnesCentre,
NorwichNR47UH,UK ofStreptomycesstrains,indicatingthatoneormoreoftheP450sinthese
4SchoolofBiosciences,The bacteriawereviabledrugtargets.Thedrugseconazoleandclotrimazolewere
UniversityofBirmingham, mosteffectiveagainstM. smegmatis(MICvaluesofT0<2and0<3lM,
Edgbaston,Birmingham, respectively)andweresuperiorinhibitorsofmycobacterialgrowthcompared
B152TT,UK torifampicinandisoniazid(whichhadMICvaluesof1<2and36<5lM,
respectively).Incontrasttotheireffectsontheactinomycetes,theazoles
showedminimaleffectsonthegrowthofEscherichiacoli,whichisdevoidof
P450s.AzoledrugscoordinatedtightlytothehaemironinM. tuberculosis
H37RvP450sencodedbygenesRv0764c(thesteroldemethylaseCYP51)and
Rv2276(CYP121).However,theazoleshadahigheraffinityforM. tuberculosis
CYP121,withK valuesbroadlyinlinewiththeMICvaluesforM. smegmatis.
d
ThissuggestedthatCYP121maybeamorerealistictargetenzymeforthe
azoledrugsthanCYP51,particularlyinlightofthefactthatanS.coelicolor
DCYP51strainwasviableandshowedlittledifferenceinitssensitivitytoazole
drugscomparedtothewild-type.Iftheazoledrugsprovetoinhibitanumber
ofimportantP450sinM. smegmatisandS.coelicolor,thenthelikelihoodof
drugresistancedevelopinginthesespeciesshouldbeminimal.Thissuggests
thatazoledrugtherapymayprovideanovelantibioticstrategyagainststrains
ofM. tuberculosisthathavealreadydevelopedresistancetoisoniazidand
otherfront-linedrugs.
Keywords: tuberculosis,azoledrugs,CYP121,P450inhibitors
.................................................................................................................................................................................................................................................................................................................
Abbreviations:g.t.,Generationtime;K,bindingconstant;Mtb,Mycobacteriumtuberculosis;s.g.r.,specificgrowthrate;P450,cytochromeP450mono-
d
oxygenase.
0002-5733#2002SGM 2937
K.J.McLeanandothers
INTRODUCTION and Bacillus subtilis, with seven P450s, had the next
largest bacterial complement of P450s prior to de-
There-emergenceofMycobacteriumtuberculosis(Mtb)
terminationoftheMtbgenomesequence(Blattneretal.,
as a worldwide threat to human health has been
1997; Kunst et al., 1997). This contrasts with the
describedbytheWorldHealthOrganization(WHO)as
situationineukaryotes,withthehumangenomeprom-
having the capacity to cause a ‘global catastrophe’
ising to reveal up to 100 P450s, and the Arabidopsis
(WHO fact sheet on tuberculosis at http:}}www.who. thaliana genome encoding "250 P450s (Lander et al.,
int}inf-fs}en}fact104.html). There are many reasons 2001;A.thalianaP450ssiteathttp:}}drnelson.utmem.
underlyingtheresurgenceinMtbinfectionratesaround edu}Arab.genes.chr.html). The relatively large number
the world, but major factors include the synergy with
ofP450sintheMtbgenomeindicatesimportantphysio-
the human immunodeficiency virus (HIV) (a large
logical roles for these enzymes and, given that the
proportion of HIV-infected individuals, around 15%,
substrate preference of the majority of P450s is for
die ultimately from tuberculosis) and the development
hydrophobicmolecules,mostarelikelytobeinvolvedin
ofdrug-resistantandmulti-drug-resistantstrainsofthe
lipidmetabolism.Anunexpectedfindingwasthatoneof
pathogen, often as a result of poor or incomplete drug
the Mtb P450s (the product of gene Rv0764c) is a
regimens in third-world countries (Heym et al., 1996). homologue of eukaryotic 14a-sterol demethylases–
Regardlessofthecauseofitsre-emergence,itisobvious
known as CYP51 (cytochrome P450 family 51) in the
thatnoveltreatmentsaredesperatelyrequiredtoprevent
officialP450nomenclaturesystemdevisedbyNelsonet
thefurtherrapidspreadoftuberculosis,whichcurrently
al. (1996). Previously, this type of P450 and the sterol
kills approximately 2 million people each year. It is
biosynthetic pathway in which it participates were
estimatedthatbetween2000and2020nearlyonebillion
considered to be found exclusively in eukaryotes
peoplewillbecomeinfected withMtb,200millionwill
(Yoshida et al., 1997). However, while it remains
becomesickbecauseofitandaround35millionwilldie
uncertain as to whether Mtb actually has a sterol
from tuberculosis.Atpresent, approximatelyone-third
biosynthetic pathway, cholesterol has been reported to
of the world’s population is infected with tuberculosis,
bepresentinthebacterialenvelope(Lambetal.,1998),
and the worldwide rate of new infections runs at one
and the Mtb CYP51 protein has been purified and
individualper second(WHO factsheetontuberculosis
shown to catalyse sterol demethylation, albeit at low
athttp:}}www.who.int}inf-fs}en}fact104.html;Young
rates (Bellamine et al., 1999). The genome sequence of
& Cole, 1993).
the actinomycete Streptomyces coelicolor also shows a
highnumberofP450-encodinggenes,aswellasmarked
ThedeterminationofthegenomesequenceofMtb(Cole
similarities in genetic context (i.e. synteny) for a large
et al., 1998) has provided a much needed boost for
proportionofthegenome.However,specificbiological
researchintonewdrugtargetsagainstthispathogen.In
rolesfornearlyalloftheP450sinMtbandS.coelicolor
particular, the vast array of genes involved in lipid
remain to be defined [S. coelicolor genome sequence
metabolism(thereareapproximatelyfive-foldthenum-
homepageattheSangerInstitute(http:}}www.sanger.
ber of genes involved in lipid metabolism in Mtb
ac.uk}Projects}Sjcoelicolor})].
compared with Escherichia coli, a bacterium with a
similar genome size) is testament to the importance of
Recently,wehaveclonedandexpressedtheMtbCYP51
thisaspectoftheorganism’sbiochemistrytoitsviability
enzyme and a second P450 from Mtb–CYP121, the
andpathogenicity,particularlywithregardtosynthesis
productoftheRv2276gene(Souteretal.,2000;McLean
of the complex fatty acid, polyketide (and possibly
et al., 2002). We, and others, have demonstrated that
sterol) components of its intricate outer envelope
CYP51bindsazoleantifungaldrugstightly,asmightbe
(Brennan & Nikaido, 1995; Daffe & Draper, 1998).
expectedforatruehomologueofthesteroldemethylase
These data suggest that targeting the lipid metabolism
class of P450. However, as we show here, it is also the
pathways of Mtb may provide an excellent route to
case that CYP121 binds these azole drugs. Indeed,
attenuating or killing the bacterium. Indeed, only
CYP121 binds these drugs with a higher affinity than
relatively recently has the biological target of the lead
CYP51,suggestingthatCYP121maybetargetedbythis
anti-Mtb drug isoniazid (isonicotinic acid hydrazide)
type of drug in vivo. In studies of the effects of azole
been identified as enoyl-acyl carrier protein reductase,
drugs on the growth of Mycobacterium smegmatis
whichisinvolvedinfattyacidandmycolicacidsynthesis
(which possesses a near identical complement of CYP
(Blanchard, 1996). Isoniazid requires activation by the
Mtb catalase}peroxidase enzyme to form a radical genes to Mtb), we show that the azole antifungals are
highly potent growth inhibitors, with very low MIC
speciesthatbindsandcovalentlyinactivatestheNADH-
values. We also demonstrate that these drugs are
binding site of the reductase (Saint-Joanis et al., 1999).
effectiveasinhibitorsofthegrowthofS.coelicolorand
An unexpected finding from the Mtb genome-sequen- Streptomyces lividans, and compare their relative anti-
cing project was the unusually large component of bioticeffectswithtwoCYPgeneknockoutvariantsofS.
cytochromeP450mono-oxygenases(P450s)inthepath- coelicolor.Itappearsthattheactinomycetesmayhaveat
ogen (Souter et al., 2000). Mtb encodes 20 P450s, leasttwoviableP450targetsfortheazoleantibioticclass
whereas previous bacterial genomes had indicated that (thus minimizing the opportunity for drug resistance),
these haem-containing mono-oxygenase enzymes were and that an azole-based strategy may be a new and
relativelyrareinbacteria.Forinstance,E.colihasnone effective means to combat Mtb.
2938
Azoleantifungalsinhibitmycobacterialgrowth
METHODS (5‹25cm)thathadbeenpre-equilibratedinthesamebuffer.
Proteinwaselutedinalineargradient(500ml)ofbufferCto
CloningandexpressionofCYP121.TheCYP121gene(Rv2276 500mM potassium phosphate (pH6–5). The fractions of
fromtheMtbgenomeproject)encodingtheCYP121P450was highest purity (A }A "1–5) were retained, pooled, con-
%"’ #)!
amplified from the Mtb chromosomal DNA cosmid library
centrated by ultrafiltration (Centriprep 30; Millipore) to
(cosmidnumberMTCY339)obtainedfromProfessorStewart !2mlandexchangedintobufferAplusglycerol(50%,v}v)
ColeatthePasteurInstitute,Paris,France.Primersusedinthe bydialysis,priortostorageofthepureenzymeatfi80(cid:176)C.
reactionweredesignedfromtheMtbgenomicsequence(Cole
etal.,1998):Forwardprimer,5«-TATGACTCATATGACC- Analysis of CYP121 purity, and drug binding to CYP121.
GCGACCGTTCTGCTCG-3«; Reverse primer, 5«-AAGAC- Purity of the CYP121 protein was assessed by spectral
GGATCCTACCAGAGCACCGGAAGG-3«. In the forward properties (ratio of P450-specific absorption at 416–5nm
primer, the bold letters indicate an engineered site for comparedwithprotein-specificabsorptionat280nm),andby
restriction enzyme NdeI, incorporating the ATG initiation SDS-PAGE of protein samples (on 10% polyacrylamide
codon of CYP121. In the reverse primer, the bold letters denaturinggels)atdifferentstagesinthepurificationprocess.
indicate a BamHI site incorporating the TAG stop codon of ThedeterminationofthecytochromeP450concentrationwas
CYP121. The PCR fragment generated from Rv2276 was donefromtheSoretmaximumoftheoxidizedenzymeinthe
cplloansmedidicnltoonepplaKsmMi2da,vaencdtoirtspDGNEAMs-eTqu(ePnrcoemwegaas)v,ercirfieeadtibnyg uressintigngth(aelmexotsintcctoiomnpcleoteeffilyclioenwt-sep%"i’n–&)flst9a5tem(AMm−a"xflcm4−1"6,–5asnmde)-,
automatedsequencing(AppliedBiosystemsDNAsequencer). scribedpreviously(Nobleetal.,1999).Integrityoftheprotein
TheCYP121genewasexcisedfrompKM2abydigestionwith was also verified by electrospray MS (Micromass platform
NdeIandBamHI,andligatedintotheT7lacpromotervector quadruple mass spectrometer equipped with an electrospray
pET11a(Novagen)thathadbeenpre-digestedwiththesame ionsource).
enzymes. Correctly ligated plasmids were selected by trans- UV-visibleabsorptionspectrawererecordedonaCaryUV-50
formation of E. coli TG1 (Gibson, 1984) and restriction BioUV-visiblescanningspectrophotometerusing1cmpath-
digestsofplasmidpreparationsweremadefromtransformant length quartz cells. Unless otherwise stated, spectra were
cultures. The expression plasmid was named pKM2b. All recorded using approximately 5–10lM CYP121 in 50mM
molecular techniques were performed using standard proto- Tris}HCl(pH7–2).BindingofazoleinhibitorstoCYP121was
cols(Sambrooketal.,1989). measuredat30(cid:176)C.TheP450 inhibitorsclotrimazole,econa-
zole,ketoconazole,fluconazole,miconazoleand4-phenylimi-
TheCYP121enzymewasproducedintheT7RNApolymerase
dazolewerepreparedasstocksolutions(typically10–25mM)
lysogenstrainHMS174(DE3)(Novagen).Conditionsusedfor
in DMSO. Small aliquots (0–1–0–4ll) corresponding to final
CYP121productionweregrowthofHMS174(DE3)(pKM2b)
transformants at 30(cid:176)C in Terrific Broth (Sambrook et al., concentrations of 1–100lM of inhibitor were added to a
protein solution (typically 1–5lM P450), with the total
1989) with vigorous agitation (250r.p.m.) until the mid-
exponentialphaseofgrowth(OD C0–5).Thetemperature addition lessthan 10llinhibitor. Imidazole was madeup in
wasthendecreasedto18(cid:176)C,andth’!e!culturewasinducedwith 50mM Tris}HCl (pH7–2) as a 3M stock solution and
additions of up to 100ll were made to saturate CYP121.
100lM IPTG. Growth of the culture was continued for a
Spectra were recorded between 300 and 750nm after each
further 20–24h. After this time, cells were harvested by
centrifugation(7500g,30min,4(cid:176)C),washedbyresuspension addition of substrate, and absorption shifts of the haem
in ice-cold 50mM Tris}HCl (pH7–2) plus 1mM EDTA (Soret) absorption maximum from approximately 416–5nm
(azole ligand-free) to 423–5nm (azole-saturated) were mea-
(bufferA),pooledandrecentrifuged.Thefinalcellpelletwas
frozen at fi20(cid:176)C until use. CYP121 was purified from the sured.Kd(bindingconstant)valuesforthebindingoftheazole
ligandsweredeterminedbyfirstgeneratingdifferencespectra
culture(typically5–8l)bystandardmethods.Forcellbreak-
(by the subtraction of the original inhibitor-free spectrum
age,cellswerethawedandresuspendedinasmallvolumeof
from each subsequent inhibitor-bound spectrum). At this
buffer A containing protease inhibitors PMSF and benzami-
point,spectrawerealsomultipliedbyappropriatefactorsto
dinehydrochloride(bothat1mM)anddisruptedbysonica-
tionusing aBandelin SonoplusGM2600 sonicator(10‹20s accountfordilutionthroughadditionoftheazoles.Next,the
maximum overall absorption change at each [azole] (peak
bursts at full power on ice, with appropriate cooling time
absorption minus trough absorption in each difference spec-
between bursts). The extract was then passed through a
Frenchpress[950p.s.i.(6–55MPa),3passes]tocompletethe trum) was plotted against the [azole]. The data points were
thenfittedtoarectangularhyperbolatogeneratetheK value,
cfoelrl3b0remakinag(eatpr4o(cid:176)cCes)s.aTndhethlyesastoeluwbales ceexntrtaricftugweadsadte1c8a0n0te0dg. using the equation AflAmax [I]}(Kd›[I]), where Adis the
observed shift in absorption at any given concentration of
Proteinwasfractionatedbyammoniumsulfate precipitation
azole ligand [I], and A is the maximum overall shift in
(2 steps: 0–30% and 30–70%), and the 30–70% pellet max
absorptionatazoleligandsaturation.Fortitrationsinwhich
(containing CYP121) was retained. The pellet was resus-
pended in a minimal volume of buffer A containing 0–5M theKdvalueoftheazoledrugwasnotmarkedlyhigherthan
the [P450] used, data were fitted instead to a quadratic
ammonium sulfate. The solution was then loaded onto a
Phenyl-Sepharose column (5‹30cm), which had been pre- equation, which describes better the tight binding process
equilibratedinbufferAcontaining1–5Mammoniumsulfate observedandwhichaccountsforthequantityofligandcom-
plexed with the P450. The equation is Afl((A }(2E))‹
(bufferB).Theproteinwaselutedinalineargradient(500ml) (I›E›K ))fi((I›E›K )#fi(4IE))!–&, where A misaxthe ob-
of buffer B to buffer A. CYP121-containing fractions were d d
served absorption difference at each azole addition, A is
dialysedagainst5lofbufferA,andloadedontoaQ-Sepharose max
column(5‹25cm)thathadbeenpre-equilibratedinbufferA. themaximalabsorptiondifferenceatazolesaturation,Eisthe
total enzyme concentration and I is the azole inhibitor
Proteinwaselutedinalineargradient(500ml)ofbufferAto
concentrationused.
buffer A plus 500mM potassium chloride. The CYP121-
containing fractions of highest purity were pooled and Determination of MIC values for M. smegmatis. M. smeg-
dialysed against 5l of 25mM potassium phosphate (pH6–5, matis mc#155 (Snapper et al., 1990) was used for testing the
bufferC)beforebeingloadedontoahydroxyapatitecolumn efficiency of the azole antifungal drugs. The MIC values for
2939
K.J.McLeanandothers
theazoleantifungaldrugswithM.smegmatisweremeasured 3F60.Therecombinationplasmidcarriesgam,theproductof
by serial dilution of M. smegmatis cultures onto LB agar which inhibits host RecBCD exonuclease V, thus allowing
media (Sambrook et al., 1989) containing azole antifungal introduction of the linear disruption cassette into BW25113.
drugs (econazole, miconazole, fluconazole, ketoconazole or The disruption cassette was allowed to recombine with
clotrimazole) in the concentration range 0–01–250lgml−". the 3SCF60.06c gene on the 3F60 cosmid at 30(cid:176)C, the
MICswerealsodeterminedforthevalidatedanti-Mtbdrugs recombination plasmid was then removed by incubation
isoniazid (1–15lgml−") and metronidazole (4–100lgml−"). overnight at 37(cid:176)C with concomitant selection of disrupted
The MIC was defined as the minimal concentration of the cosmid by incubation with 50lg apramycinml−" and 25lg
drugrequiredtocompletelyinhibitthegrowthof99%ofthe kanamycinml−".
cells,asdescribedpreviously(Cookseyetal.,1993;Phetsuksiri
PriortoconjugativetransferofthemutantcosmidintoS.co-
etal.,1999).Allassayswereperformedintriplicate.
elicolor, the cosmid was first transformed into the methyl-
Theeffectsoftheazoleantifungaldrugsonthegrowthrateof ation-deficient E. coli host ET12456, to prevent methyl-
M. smegmatis mc#155 and E. coli TG1 were determined by specific restriction of the cosmid by S. coelicolor. ET12456
growing overnight cultures of the bacteria in LB medium at alsocontainsthenon-transmissibleplasmidpUZ8002,which
37(cid:176)C with agitation at 180 r.p.m. The following morning, encodes the tra gene product that enables mobilization of
50llaliquotsoftheE.colicultureand500llaliquotsofthe vectorDNAcontainingtheRK2oriT.Followingintergeneric
M.smegmatiscultureweretransferredintotubescontaining conjugative transfer into S. coelicolor, and selection in the
5mloffreshLBmedium,andreturnedtotheorbitalincubator presenceof50lgapramycinml−"and25lgnalidixicacidml−"
untiltheculturereachedanOD valueofC0–1–thistook toselectivelykillE.coli,colonieswerescreenedtwiceforthe
’!!
approximately1–5hforE.coliand3hforM.smegmatis.At presence of double-crossover recombinants by testing for
this stage, the antibiotics clotrimazole, econazole and isoni- resistance to apramycin and sensitivity to kanamycin on
azid were added from the DMSO stock solutions (at final replicaDifconutrientagarplates.Finally,3SCF60.06cknock-
concentrationsof10,1and0–1lM,withDMSOat%0–1%, outmutantswerestreakedontomannitolsoyaflourmedium
v}v) to separate tubes, and controls without drug addition, plates (Hobbs et al., 1989) and stocks were prepared from
plusrifampicin(5lM)orplusDMSO(0–1%,v}v)wereadded singlecolonies.
tofurtherE.coli and M.smegmatiscultures. Growthofthe
An S. coelicolor knockout of the CYP51-encoding gene
cell cultures was then followed by OD measurements at
’!! (SC7E4.20) was generated by transposon mutagenesis using
30–60min intervals over the next 5–6h (E. coli) or 10–12h
Tn4560, as described previously (Bhatt et al., 2002). Briefly,
(M. smegmatis). The viability of the bacteria following
the strain was isolated from a screen of random transposon
antibioticdrugtreatmentwasalsoassessedbymeasuringthe
mutantsgeneratedusingTn4560,aderivative(vph,viomycin
growthratesofculturesofE.coliandM.smegmatisthathad
resistance) of the Streptomyces fradiae Tn4556 for muta-
been pre-treated with econazole and isoniazid (at 10, 1 and
genesisofS.coelicolorA3(2)(Bhattetal.,2002).Tn4560was
0–1lM)andrifampicin(5lM)andrestoredtoantibiotic-free
introducedintoS.coelicolorbyintergenericconjugationfrom
medium. Inoculates (50ll for E. coli, 500ll for M. smeg-
E. coli [the non-methylating strain ET12567 containing the
matis)weretakenfromtherelevantdrug-treatedcultures12h
mobilizing plasmid pUZ8002 (Kmr) and the transposon
after addition of the antibiotics, centrifuged briefly to pellet
deliveryvector pKay1, includingTn4560,vph,plasmidmar-
the cells, resuspended in the same volume of antibiotic-free
kers tsr and aac (3)IV, and oriT (RK2)]. Conjugation is
medium and transferred to fresh LB medium for culture as
described in detail elsewhere (Kieser et al., 2002). Random
before. The OD values of the two cultures were followed
’!! viomycin-resistantcolonieswereselectedfromthetransposon
over the next several hours to assess bacterial recovery. All
library and the Tn4560 insertion site was determined by
growth-ratemeasurementsweredoneintriplicate.
ligation-mediatedPCR amplification (Bhatt et al., 2002) and
GenerationofCYPgeneknockoutmutantsinS.coelicolor.A dideoxyDNAsequencing.TofacilitatePCR,theS.coelicolor
knockoutmutantoftheS.coelicolorA3(2)gene3SCF60.06c DNAwasdigestedcompletelyusingrestrictionenzymeEagI,
(encodingputativeP450CYP105D5)wasgeneratedusingthe which does not cut the Tn4560 ends. Then a non-phos-
‘Redirect–Technology’ PCR targeting system developed by phorylated, partially double-stranded adaptor was ligated
Gust et al. (2002) under licence from Plant Bioscience Ltd, (onestrandonly)totheEagI-generatedfragments.PCRwas
Norwich Research Park, Colney, Norwich, UK. This P450 thendoneusingaTn-specificprimerandaprimerspecificfor
(CYP105D5) shows closest similarity to the Mtb CYP121 one strand of the adaptor, as described elsewhere (Fowler,
enzymeattheaminoacidlevel(28%),asdeterminedusingthe 2002). DNA sequencing for one of the isolates showed that
clustalwsoftwareattheEuropeanBioinformaticsInstitute transpositionhadoccurredwithintheSC7E4.20gene,andthis
(http:}}www2.ebi.ac.uk}clustalw}). mutantwascharacterizedfurther.
Acassettecomprisingtheapramycin-resistancegeneaac(3)IV, AssessmentofeffectsofazolesongrowthofStreptomyces
the origin of transfer (oriT) from RK2 and flanking FLP strains. S. coelicolor A3(2), the two CYP knockout strains
recombinase recognition targets was modified by PCR to (DCYP51 and DCYP105D5) and S. lividans TK24 were
generate a gene-specific disruption cassette, using primers maintainedon soya}mannitol agarslopesorasfrozen spore
with 20nt identities to the ends of the cassette and 39nt suspensions(indistilledwater)(Kieseretal.,2002).Toassess
identities to genomic regions flanking the 3SCF60.06c gene. theefficiencyoftheazoledrugsonthegrowthofthesestrains,
Theupstreamprimerwas5«-GTCCTTGTCGATGCCGAT- sporesuspensionswerethawedandspreadoverthesurfaceof
GTGCATGCCGGGAGCCTCCTATGTAGGCTGGAGC- tryptone}soyaagar(TSA;Oxoid)plates.Atthisstage,6mm
TGCTTC-3«andthedownstreamprimerwas5«-ATCGATC- antibioticdiscsthathadbeenpre-soakedintherelevantazole
CGACGCACTCCTCGTCTGCGAGGTCTTCCATGAT- drug (or in isoniazid or rifampicin) or in DMSO or water
TCCGGGGATCCGTCGACC-3«. S. coelicolor homologous (controls) were placed onto the surface of the plates. Cells
sequences are shown in bold in both primer sequences. The weregrownfor3daysat30(cid:176)C,priortodeterminationofthe
linear disruption cassette DNA was then introduced into E. antibiotic efficiency through measurement of the zone of
coliBW25113.Thisstraincontainedthetemperature-sensitive cellular-growthinhibitioncreatedbyeachdrug.Asimilarset
k red recombination plasmid and the S. coelicolor cosmid ofcontrolexperimentswasperformedusingE.coliTG1.Cells
2940
Azoleantifungalsinhibitmycobacterialgrowth
weregrowntoanOD’!!valueofbetween0–7and0–8,plated T7 promoter lac operator RBS
ontoTSAmediumand grown for 16–18h at30(cid:176)C,prior to
analysis of growth inhibition. The incubation times allowed
confluent bacterial cellular growth on the plates outside the
NdeI
zonesofinhibition.MICvaluesinthissystemweredefinedas
the minimum concentration of the drug required to give a EcoRI
measurable zone of inhibition around the disc. MIC values
were also determined by serial dilution of S. coelicolor and
S. lividans spore suspensions, and spotting these onto Difco bla
Rv2276
Nutrient agar medium (Kieser et al., 2002). As with M.
BamHI
smegmatis, the MIC here was defined as the minimum drug
concentrationrequiredtopreventgrowthof99%ofthecells.
Allassayswereperformedintriplicate. EcoRI
pKM2b
Materials. Oligonucleotide primers for PCR were obtained 6832 bp
fromPerkinElmerAppliedBiosystemsorweremanufactured
attheProtein and Nucleic AcidLaboratory(PNACL) at the
UniversityofLeicester,UK.AllrestrictionenzymesandDNA-
modifying enzymes were from New England Biolabs. Other
lacI
modifying enzymes, Taq and Pfu DNA polymerase and T4
DNA ligase were obtained from Promega. Unless otherwise
ori
stated,allreagentsusedwereobtainedfromSigmaandwere
ofthehighestgradeavailable.Mediaandmostsolutionswere
madeaccording tostandard recipes(Sambrooket al., 1989). .................................................................................................................................................
Tryptone}soyamediumforthegrowthofStreptomycesspp. Fig.1. Cloning ofthe CYP121geneandcreationofexpression
was from Oxoid. Econazole, ketoconazole and fluconazole vectorpKM2b.TheCYP121genewasclonedbyPCRfromaMtb
werefromICN.Metronidazole,clotrimazole,miconazoleand cosmid library, as described in Methods. The PCR product was
cloned directly into pGEM-T (creating plasmid clone pKM2a),
otherazoleswerefromSigma.IsoniazidwasfromAldrich.
and then as a NdeI–BamHI fragment into plasmid vector
pET11a,creatingtheCYP121-expressingplasmidpKM2b.
RESULTS
ProductionandpurificationofCYP121
Typically, CYP121 was purified to homogeneity using
TheRv2276gene(hereafterreferredtoasCYP121)was
ammonium sulfate fractionation followed by column
amplified by PCR using Taq DNA polymerase as
chromatography using Phenyl-Sepharose, Q-Sepharose
described in Methods, and verified to be of the correct
size (1–188kb) by gel electrophoresis alongside DNA andhydroxyapatite.TherelativepurityofCYP121was
measured during its purification, by comparing the
size standards (1kb ladder; Gibco-BRL). The PCR
haem-specific absorption (at 416–5nm) with the total
product was cloned into pGEM-T (Promega), and
protein absorption at 280nm at different stages of
successful clones were selected by a-complementation
purification.Thesemeasurementstypicallyshowedthat
on X-Gal- and IPTG-containing LB agar medium. The
soluble CYP121 was purified approximately 50-fold
CYP121 gene was then excised from a verified clone
from the original cell extract, and was recovered in
(pKM2a), using NdeI and BamHI, and cloned into
yieldsofapproximately4mgpureP450(lcellculture)−".
expressionvectorpET11a(Novagen)underthecontrol
The overall yield of CYP121 was C20%, based on
ofanIPTG-inducibleT7lacpromoter,creatingCYP121
recovery of haemoprotein. However, given the afore-
expression plasmid pKM2b (Fig. 1).
mentionedtendencyoftheproteintomisfoldandform
PreliminaryexpressiontrialsindicatedthatCYP121was inclusion bodies, we suspect that much of the CYP121
expressed to high levels in a variety of E. coli T7 RNA present in the initial extract is haem-bound denatured
polymeraselysogenstrains.However,largeproportions enzyme,andthatthisisseparatedfromthenativeform
of the CYP121 enzyme were found to form inclusion by ion exchange chromatography. The pure CYP121
bodies.ToachieveahighyieldofsolubleCYP121,theE. bound carbon monoxide in its (dithionite-reduced)
colistrainHMS174(DE3)wasused(Studier&Moffat, ferrousform,givinganabsorptionmaximumat448nm,
1986).AHMS174(DE3)(pKM2b)transformantculture typicaloftheP450enzymeclass(notshown).Theratio
(typically of 5–10l) was grown at 30(cid:176)C until early- ofhaem-specifictototalproteinabsorption(i.e.A }
%"’–&
exponential phase. Thereafter, mild IPTG induction A ) gives a measure of purity, with a value of
#)!
(100lM) of the culture was used, and the growth approximately1–9indicatinghomogeneousCYP121,as
temperaturewasreducedto18(cid:176)C.Cultivationwasthen verified by SDS-PAGE. Electrospray ionization MS
continuedforafurther24h.Undertheseconditionsthe indicated a single, intact species of molecular mass
overall expression level of CYP121 was reduced by 43128Da (not shown). This correlated almost exactly
approximately twofold (compared to growth and in- withthepredictedmassofCYP121basedontranslation
duction at 37(cid:176)C), but production of soluble CYP121 of its gene sequence, once the mass of the initiator
wasgreatlyincreased,allowingmuchhigherrecoveryof methioninehadbeensubtracted(43126Da).Adetailed
soluble P450 from the cells. descriptionofthespectroscopicandbiophysicalproper-
2941
K.J.McLeanandothers
0·4
0·3
0·10 AD 0·2
0·1
0·0
0·05
0 5 10 15 20
[Ketoconazole] (lM)
A0·00
D
–0·05
–0·10
375 400 425 450 475 500 525
Wavelength (nm)
.................................................................................................................................................
Fig.3.BindingofketoconazoletoMtbCYP121.Thebindingof
theazoledrugtotheP450haeminducesshiftsintheUV-visible
absorption spectrum, due to replacement of the sixth water
ligandtotheferrichaemironbytheazolemoietyofthedrug.
Themainsectionofthefigureshowsacollectionofdifference
spectra induced by the addition of the azole antifungal drug
ketoconazole to oxidized CYP121 (6–5lM), with difference
spectra generated by subtraction of the starting (azole-free)
CYP121 absolute spectrum from those collected following
addition of 0–2, 1–2, 2–2, 3–2, 5–2, 7–2, 13–2 and 23–1lM
ketoconazole.Thereisaprogressiveincreaseinthemagnitude
of the difference spectra with increasing ketoconazole
concentrations.Theazoledrugcomplexisvirtuallyfullyformed
by 23–1lM, with the Soret absorption maximum shifted to
................................................................................................................................................. 423–5nm in the absolute spectrum. In the difference spectra,
Fig. 2. Chemical structures of the azole antifungal drugs used minima and maxima resulting from azole addition are located
in this study. In all cases, enzyme inhibition is achieved by atapproximately 395/408nm and430 nm.The inset shows the
ligationofbasicnitrogenatomsfromtheinhibitorstotheferric plot of the maximal shifts in absorption data (A minus A )
430 408
haemironofP450. for binding of ketoconazole to CYP121 versus the relevant
ketoconazole concentrations. Fitting the data to a quadratic
functiongeneratesaK valueof3–3‡0–3lMforthedrug.
d
tiesofCYP121ispresentedelsewhere,andwehavealso
determined the atomic structure of this P450 (McLean
at low concentrations of CYP121 (1–2lM). For these
et al., 2002; D. Leys, C.G. Mowat, K.J. McLean,
three azoles, the optical change associated with azole
A.Richmond,S.K.Chapman,M.D.Walkinshaw,M.J.
ligation occurred linearly with azole concentration,
Sutcliffe & A.W. Munro, unpublished data).
sharply reaching a plateau and indicative of stoichio-
metricbindingtotheP450.Evenquadraticfitscouldnot
Spectrophotometricanalysisofazoledrugbindingto
describe the K with sufficient accuracy. Evidently, the
CYP121 d
K valuefortheseazolesisverylow,andislikelytobe
d
UV-visible absorption spectroscopy provides a simple !0–2lM. Optical titrations with fluconazole revealed
and accurate method for the determination of the slightlyweakerbindingofthisazoletoCYP121,witha
binding of substrates and inhibitors to P450s. The K value of 9–7‡0–2lM, as determined from a hy-
d
binding of azole inhibitors to P450s involves displace- perbolic fit of the absorption change versus the flucon-
ment of a water molecule (weakly ligated as an axial azole concentration. Ketoconazole bound to CYP121
ligand to the haem iron) by a basic nitrogen from the withaslightlyhigheraffinitythanfluconazole,withaK
d
azole ring of the inhibitor. This leads to a shift in value of 3–3‡0–3lM, as determined from the fit to the
the visible absorption spectrum of the P450, most quadratic equation (Fig. 3).
notablethroughtheshiftofthemajor(Soret)absorption The binding of imidazole (K fl64‡3 mM) and 4-
maximum to a longer wavelength. In the case of phenylimidazole (K fl53‡3lMd ) to CYP121 was con-
CYP121, the Soret band shifted from 416–5nm (inhibi- d
siderablyweakerthanthatofclotrimazole,econazoleor
tor-free) to 423–5 nm (azole-saturated).
miconazole,indicatingthatthehighaffinityoftheazole
Eachoftheazoleantifungaldrugstestedhere(clotrim- antifungals for CYP121 is determined primarily by
azole,econazole,fluconazole,ketoconazoleandmicon- favourable interactions between the bulky, polycyclic
azole)wasfoundtobindtightlytoCYP121,inducinga azole antifungals and the hydrophobic residues in the
shiftoftheSoretspectralmaximumto423–5nm(Fig.2). largely apolar active site of CYP121, rather than being
Thebindingofclotrimazole,econazoleandmiconazole drivenbyligationoftheazolegrouptotheferrichaem
toCYP121provedtootighttoanalyseaccurately,even iron.Itisinterestingtonotethateventheadditionofa
2942
Azoleantifungalsinhibitmycobacterialgrowth
Table1.ComparisonofM.smegmatisMICvaluesforselectedazoleantifungaldrugs
withtherelevantK valuesforMtbCYP121andCYP51
d
.....................................................................................................................................................................................................................................
TheMICvaluesfortheazoleantifungaldrugsweredeterminedasdescribedinMethods.These
valuesarecomparedwiththerelevantK valuesforthesamedrugs(derivedfromspectral
d
titrations)forpurifiedMtbCYP121(fromthisreport)andthesteroldemethylaseCYP51(Bellamine
etal.,1999;Guardiola-Diazetal.,2001).AllMICvalueshavestandarderrorsof!10%.
Drug MICvalueforM. K valueexpressedinlM(andlgml−1)for
d
smegmatisinlM
(andlgml−1) MtbCYP121 MtbCYP51
Econazole* !0–2(!0–1) !0–2(!0–09) 5–4‡1–8(cid:140)(2–4‡0–8)
Clotrimazole* 0–3(0–1) !0–2(!0–07) C5(cid:140)(C1–7)
Miconazole* 2–6(1–25) !0–2(!0–10) nd
Ketoconazole* 38(20) 3–3‡0–3(1–75‡0–2) 11–1‡5–0(cid:140);5(cid:141)(5–9‡2–7;2–7)
Fluconazole* "325("100) 9–7‡0–2(3–0‡0–1) C10(cid:141)(C3–1)
Isoniazid(cid:139) 36–5(5) nc nds
Rifampicin(cid:139) 12(1) nc nd
nd,Notdetermined;nc,nospectralchangeobserved.
*Antifungaldrug.
(cid:139)Anti-Mtbdrug.
(cid:140)K valuesforCYP51arefromGuardiola-Diazetal.(2001).
d
(cid:141)EstimatedK valuesfromBellamineetal.(1999).
d
sSpectralchangeswereobserved,butnoK valuewasobtained(fromGuardiola-Diazetal.,2001).
d
hydrophobicphenylmoietytotheazoles(for4-phenyl- CYP51 sterol demethylase) may be the major site of
imidazole cf. imidazole) is sufficient to reduce the K action of these drugs.
d
value by a factor of 10$. No significant spectral
The effects of the azole drugs on the growth rate of
perturbations were observed on titration of CYP121
M. smegmatis were also determined, and the growth
with the nitroimidazopyran drug metronidazole at
rateswerecomparedwiththoseofE.coli(seeMethods).
concentrations up to 10 mM, suggesting that this drug
ForE.coli,neitherclotrimazolenoreconazole(thetwo
binds more weakly to CYP121 than imidazole does.
most effective azole drugs against M. smegmatis and
S. coelicolor) had any significant effect on the specific
Determinationoftheefficiencyoftheazole
growth rate (s.g.r.) of this organism when applied at
antifungaldrugsagainstM.smegmatis
concentrations of 10, 1 and 0–1lM. Isoniazid (at the
The MIC values for a range of azole antifungal drugs sameconcentrations)andDMSOalsofailedtodecrease
were determined using M. smegmatis, and these were the growth rate of E. coli when compared with the
comparedwiththeMICvaluefortheleadinganti-Mtb control sample. In all cases the s.g.r. was around
drug,isoniazid[isonicotinicacidhydrazide(INH)].The 0–354‡0–044h−"[generationtime(g.t.)fl0–85‡0–10h].
MIC of isoniazid was 36–5lM (5lgml−"). However, However, rifampicin at a concentration of 5lM com-
threeofthefivecommerciallyavailableazoleantifungal pletelypreventedthegrowthofE.coli.Incontrasttothe
drugs proved more effective than isoniazid against M. results with E. coli, the azole drugs inhibited M. smeg-
smegmatis. The MIC values for clotrimazole (0–3lM; matis growth strongly atlow concentrations(Table 2).
0–1lgml−"), miconazole (2–6lM; 1–25lgml−") and The naturally slower-growing control and DMSO-
econazole (!0–2lM; !0–1lgml−") indicated that all treated M. smegmatis samples had identical growth
ofthesedrugsweresuperiortoisoniazidasinhibitorsof rates(s.g.r.fl0–130‡0–008h−";g.t.fl2–32‡0–14h),but
M. smegmatis cell growth, and both clotrimazole and growth was completely terminated by the presence of
econazoleweremoreeffectivethanthepotentanti-Mtb econazoleatconcentrationsof10and1lM.At0–1lM
drug rifampicin (MICfl1–2lM; 1lgml−"). Ketocon- econazole,theM.smegmatiss.g.r.wasreducedby54%.
azole(MICfl38lM;20lgml−")wasratherlesseffective ClotrimazolewasalsohighlyeffectiveagainstM.smeg-
than rifampicin, and fluconazole had a negligible effect matis, but it was slightly less potent than econazole.
on bacterial growth (MIC"325lM; "250lgml−"). At 10lM clotrimazole there was no M. smegmatis
DMSO,usedasthesolventfortheazoles,hadnoeffect growth; at 1 and 0–1lM clotrimazole the s.g.r. of
oncellulargrowthattheconcentrationsused(!0–1%). M.smegmatiswasreducedby72and41%,respectively.
Theorderofpotencyoftheazoledrugscorrelatedwith Theseazoledrugswerefarmoreeffectivethanisoniazid,
theirK valuesforbindingtotheMtbCYP121enzyme which decreased the growth rate of M. smegmatis by
d
(Table 1), suggesting that this P450 (rather than the 26% only at the highest concentration tested (10lM).
2943
K.J.McLeanandothers
Table2.InhibitionofM.smegmatisgrowthbyazoleantifungaldrugscomparedto
inhibitionofgrowthbyanti-Mtbdrugs
.....................................................................................................................................................................................................................................
GrowthofM.smegmatiscultureswasrecordedbymeasuringtheOD value(seeMethods).
’!!
DMSOwasaddedtotheculturesat0–1%(v}v)toactasasolventforthedrugs.Resultsarebased
onthreereplicationsandareshown‡sd.
Drug Finalconcn s.g.r.(h−1) g.t.(h) Decreasein
(lM)* growth(%)
Control(nodrug) 0–130‡0–008 2–32‡0–14 –
Econazole(cid:139) 10 ng ng 100
1 ng ng 100
0–1 0–060‡0–011 5–02‡0–92 54
Clotrimazole(cid:139) 10 ng ng 100
1 0–036‡0–009 8–36‡2–09 72
0–1 0–076‡0–004 3–96‡0–21 41
Isoniazid(cid:140) 10 0–096‡0–020 3–14‡0–65 26
1 0–128‡0–008 2–35‡0–15 0
0–1 0–123‡0–012 2–45‡0–24 0
Rifampicin(cid:140) 5 0–110‡0–011 2–74‡0–27 15
ng,Nogrowth.
*A 1lM drug concentration corresponds to 0–45lgml−" (econazole), 0–35lgml−" (clotrimazole) or
0–14lgml−"(isoniazid).A5lMconcentrationofrifampicincorrespondsto4–1lgrifampicinml−".
(cid:139)Antifungaldrug.
(cid:140)Anti-Mtbdrug.
Table3.MICvaluesforazoleantifungaldrugswithS.coelicolorandS.lividansstrains,as
determinedbythedisc-basedassay
.....................................................................................................................................................................................................................................
DMSO,usedasthesolventfortheazoledrugs,hadnoeffectonthegrowthoftheStreptomyces
strains.TheMICvaluesreportedasmolaritiescorrespondtothefollowingconcentrationsinunits
oflgml−"fortheMICvalueswithS.coelicolor:econazole,2–2;clotrimazole,3–4;miconazole,3–6;
ketoconazole,130;fluconazole,4600;metronidazole,"17000;isoniazid,"14000;rifampicin,
200000.AllMICvalueshavestandarderrorsof!10%.
Drug MICvalues(lM)for*
S.coelicolor S.coelicolor S.coelicolor S.lividans
A3(2) DCYP51 DCYP105D5 TK24
Econazole(cid:139) 5(2) 7–5(2) 7–5(5) 7–5(5)
Clotrimazole(cid:139) 10(5) 10(5) 15(5) 10(5)
Miconazole(cid:139) 7–5(5) 7–5(5) 10(5) 7–5(5)
Ketoconazole(cid:139) 250(100) 250(100) 250(250) 100(100)
Fluconazole(cid:139) 1–5‹10%(1–5‹10%) 2‹10%(1–5‹10%) 2‹10%(2‹10%) 2‹10%(2‹10%)
Metronidazole(cid:140) "10& "10& "10& "10&
Isoniazid(cid:140) "10& "10& "10& "10&
Rifampicin(cid:140) 2–5‹10& 2–5‹10& 2–5‹10& 2–5‹10&
*ValuesinparenthesesareMICvaluesdeterminedfromcolonygrowthonazole-containingmedium.
(cid:139)Antifungaldrug.
(cid:140)Anti-Mtbdrug.
Rifampicin at a concentration of 5lM caused only a Tofurtherexaminetheeffectsoftheazoledrugsonthe
15%decreaseintheM.smegmatisgrowthrate,amuch viabilityofM.smegmatis,thegrowthofthebacterium
lessereffectthanobservedwhenE.coliwasgrowninthe post-azole treatment was measured. This experiment
presence of this drug. wasperformedusingeconazole(themosteffectiveazole
2944
Azoleantifungalsinhibitmycobacterialgrowth
.....................................................................................................
Fig. 4. Inhibition of Streptomyces cell
growth by azole drugs. Azole antifungal
drugsarepotentinhibitorsofthegrowthof
S. coelicolor and S. lividans. The figure
shows zones of inhibition of growth of
S. coelicolor (a) and S. lividans (b) around
discs soaked in the azole drugs miconazole
(Mic), ketoconazole (Ket) and fluconazole
(Flu), along with control discs soaked in
metronidazole (Met; 100mM), DMSO and
water(H O).S.coelicolorgrowthisinhibited
2
by miconazole and ketoconazole (both at
10mM),butnotbyfluconazoleatthesame
concentration. S. lividans growth is also
inhibited by miconazole and ketoconazole
(bothat10mM),andsomeinhibitionisalso
observed for fluconazole at a higher
concentration(100mM).
drug tested) and data for bacterial recovery compared although the MIC values were generally slightly lower
withdataobtainedwhenisoniazidandrifampicinwere whendeterminedbythecolony-countingmethod.This
used as the antibiotics. An identical set of experiments mayhaveresultedfromthelimiteddiffusionoftheazole
was performed on E. coli as a further control. For drugsthroughtheagarmediuminthedisc-basedassay.
E. coli, 5lM rifampicin had a bactericidal effect, with
ThethreeazoledrugsthatboundmosttightlytoCYP121
completeinhibitionofthegrowthofthisorganismupon
andwhichweremosteffectiveagainstM.smegmatisin
thetransferofaninoculumoftreatedcellstoantibiotic-
growth trials (i.e. econazole, clotrimazole and micon-
free medium. In contrast to the rifampicin pre-treat-
azole) were by far the most effective azoles against the
ment,cellsofE.colipre-treatedwitheithereconazoleor
Streptomycesspecies,withMICvaluesof%10lM.The
isoniazid (both at 10lM) resumed growth, with rates
MIC values for the three azoles were all considerably
identicaltothoseoftheuntreatedcontrolcells(s.g.r.fl
0–354‡0–010h−"; g.t.fl0–85‡0–20h). For M. smeg- lower than that for rifampicin. The next most effective
azole,ketoconazole,hadasimilarantibacterialpotency
matiscellspre-treatedwithisoniazid(atconcentrations
to rifampicin against the Streptomyces species. As seen
oinf1g0ro,w1tahndra0te–1sl(Ms.g).rt.hfler0e–1w2a0s‡n0o–0s1ig0nhifi−"c;angt.td.fliff2er–5e1n‡ce with M. smegmatis, fluconazole was !10$-fold as
effectiveaseconazoleandclotrimazoleagainstS. coeli-
0–21h) when compared to the untreated controls
(0–124‡0–014h−"; g.t.fl2–43‡0–27h). In contrast to colorandS.lividans.Isoniazidhadanegligibleeffecton
Streptomyces growth (MIC "100mM), and the Mtb
theresultwithE.coli,cellsofM.smegmatispre-treated
nitroimidazopyran drug candidate metronidazole also
with rifampicin (5lM) recovered fully and resumed
normal growth (0–126‡0–012h−"; g.t.fl2–39‡0–23h). hadaMICvalue"100mMwithalloftheStreptomyces
strains tested. DMSO (the solvent for the drugs) also
However, inM. smegmatis cells pre-treated with econ-
hadnoinhibitoryeffectsonstreptomycetegrowthatthe
azole there was a clear detrimental effect on bacterial
concentrations tested.
growthinthesubcultures,whichwasdependentonthe
initial concentration of the drug. In cells of M. smeg- Inthedisc-basedassaysoftheazoleandanti-Mtbdrugs
matispre-treatedwith10lMeconazole,growthwasre- against E. coliTG1, noneoftheazoleantifungal drugs
establishedatans.g.r.of0–069‡0–012h−"(g.t.fl4–36‡ were effective at concentrations up to 100mM. Simi-
0–76h); at 1lM econazole the s.g.r. was 0–097‡ larly,DMSOhadnoeffectonE.coligrowth.Metroni-
0–009h−" (g.t.fl3–10‡0–29h), and at 0–1lM econazole dazole and isoniazid had only minor effects on the
the s.g.r. was 0–103‡0–008h−" (g.t.fl2–92‡0–23h). growthofE.coliTG1(MICvaluesof75and100mM,
Thus, after pre-treatment with 10lM econazole, the respectively),butrifampicinwashighlyeffectiveagainst
subculturedcellsofM.smegmatisre-establishedgrowth this organism, with an MIC of !5lM.
at only 55% of the rate of the untreated control cells.
Atbest,marginaldifferentialeffectsonthegrowthofthe
two S. coelicolor cytochrome P450 knockout strains
Determinationoftheefficiencyoftheazole
(DCYP51 and DCYP105D5) were observed in com-
antifungaldrugsagainstS.coelicolorA3(2)and
parisontothewild-typeonexposuretotheazoledrugs.
S.lividansTK24
Any small difference in azole sensitivity was clearest in
AsfoundwithM.smegmatis,alloftheazoleantifungal theDCYP105D5strain,whereMICvaluesforthemost
drugs tested were potent inhibitors of cellular growth effective azole drugs (econazole, clotrimazole and
for S. coelicolor and S. lividans (Table 3; Fig. 4). MIC miconazole)wereslightlyhigherthanforthewild-type
values, as determined from zones of inhibition around strain (Table 3). The most important feature to note
azole-soaked discs and from colony growth on azole- regarding these strains was that the deletion of either
containing medium (as described for M. smeg- P450wasnon-lethal,andhadlittleeffectonthegrowth
matis), were similar for both Streptomyces species, of the bacteria under laboratory conditions. Thus,
2945
K.J.McLeanandothers
neither CYP105D5 nor CYP51 is essential for viability Azole drugs are commonly used in the treatment of
under such conditions. fungal and yeast infections, and act primarily by
inhibiting the activity of the P450 CYP51, which is a
lanosterol demethylase involved in the generation of
DISCUSSION
ergosterol,acriticalsterolforthecellularmembraneof
P450saregenerallyconsideredtoberareinbacteria,and yeasts and fungi (Schuster, 1993). Fungal resistance to
to function mainly as components of dispensable cata- azole drugs is frequently associated with CYP51 muta-
bolic pathways for the degradation of unusual carbon tions that reduce affinity of the altered proteins for the
sources(Poulosetal.,1987;Petersonetal.,1992).Such drugs (Aoyama et al., 2000). A rationale for the use of
enzymes would be effective targets for manipulation azolesasantibacterialshasbeenabsent,duetoassump-
only under peculiar growth conditions. However, the tions that bacteria generally do not possess sterol
recent genome-sequencing projects for Mtb (and other biosyntheticpathways.However,aCYP51-likeenzyme
related mycobacteria) and S. coelicolor have indicated hasbeencharacterizedinMtbandthisenzymehasbeen
thattheseactinomycetesencodelargenumbersofP450s. showntocatalysethedemethylationoflanosterolanda
Thereare20differentP450isoformsencodedwithinthe plant sterol (obtusifoliol), as would be expected for a
Mtb genome [Cole et al., 1998; Mtb genome sequence bona fide CYP51 enzyme (Bellamine et al., 1999). The
homepageattheSangerInstitute(http:}}www.sanger. subsequentdeterminationoftheatomicstructureofthe
ac.uk}Projects}Mjtuberculosis})], and the recent se- Mtb CYP51 in complex with the drug fluconazole
quence of S. coelicolor indicates there to be 18 P450s provided the first molecular description ofthe mode of
encoded within its genome (Bentley et al., 2002). The interaction of an azole antifungal drug with a P450
large numbers of P450s found in these actinomycetes enzyme. The structure is thus of great importance in
have not (as yet) been found in other bacterial genera, understanding how the specificity for the azole is
and suggest that the P450s play several important determined, and may enable de novo design of more
physiological roles in mycobacteria and the strepto- selective and potent azole inhibitors of CYP51. How-
mycetes; thus, they may provide novel drug targets ever, azole antifungals are used mainly as topically
againstthepathogenicmycobacteria.InvariousStrepto- appliedantibioticsforthetreatmentofsuperficialyeast
mycesspecies,rolesforP450sarerecognizedinspecific andfungalinfections.Duetotheircross-reactivity(often
hydroxylations for the synthesis of polyketide anti- with micromolar or higher affinity) with human P450
biotics (e.g. P450picK in the synthesis of picromycin in isoforms, many of the azole drugs are not systemically
Streptomyces venezuelae; Betlach et al., 1998). How- tolerated and are toxic when administered orally or by
ever, the roles of Mtb P450s remain to be elucidated, intravenous injection (Zhang et al., 2002). Much re-
with the only possible exception being the putative search has gone into the production of systemically
steroldemethylaseCYP51(Souteretal.,2000;Bellamine tolerated azole drugs, and fluconazole has been among
et al., 1999; Lepsheva et al., 2001). The complexity of the most successful of the generation of triazole drugs
the lipids in the Mtb envelope may provide an ex- that is now in clinical use. Research remains active in
planation for the roles of several of the other P450s in thisarea,andvoriconazoleisagoodexampleofanewly
this organism (Daffe&Draper, 1998). The majority of developed systemically tolerated triazole with an im-
the20P450sinMtbareconservedinM.smegmatisand proved potency and spectrum of antifungal activity
Mycobacterium bovis, with the notable exceptions of compared to fluconazole (Hoffman & Rathbun, 2002).
thehomologuesofthoseP450sencodedbyRv3121and In view of the fact that we have demonstrated that
Rv1256c,whichareinregionsofthegenomedeletedin several azole drugs are powerful inhibitors of myco-
theM.bovisBCGstrain(RD12andRD13,respectively) bacterial (and streptomycete) growth, there would
[TheInstituteforGenomicResearchmicrobialgenomes appeartobegoodevidenceforthedevelopmentanduse
home page (http:}}www.tigr.org}tdb}mdb}mdbin- of these drugs as anti-Mtb agents.
progress.html); M. bovis genome sequence home page
at the Sanger Institute (http:}}www.sanger.ac.uk} Thefactthatazoleantifungaldrugsarehighlyeffective
Projects}Mjbovis})]. Interestingly, the much smaller as inhibitors of the growth of M. smegmatis and of
genomeofMycobacteriumleprae(3–27Mbcf.4–44Mb Streptomycesstrainsdoesnotleadusimmediatelytothe
in Mtb) encodes only one functional P450, although identification of their major intracellular target en-
there are several CYP pseudogenes retained within its zyme(s). However, it would appear unlikely that the
genome [M. leprae genome sequence home page at the drugsdonottargetP450(s)invivo(particularlyinlight
Sanger Institute (http:}}www.sanger.ac.uk}Projects} oftheirtightbindingtoMtbCYP121andCYP51),and
Mjleprae})]. The sole viable CYP gene remaining in their ineffectiveness against E. coli (which is devoid of
M.lepraeisML2088,orCYP118,whoseclosestrelative P450s) suggests that these drugs have negligible effects
in Mtb is the CYP140 protein encoded by the Rv1880c on other cellular enzymes, including the respiratory
gene (34% amino acid sequence identity). Obviously, cytochromes. However, any assumption that the sole
thecharacterizationofthisparticularP450isoformmay P450 target in mycobacteria and streptomycetes is the
shed light on a potentially critical P450-mediated role sterol demethylase homologue CYP51 is clearly invalid
retainedinM.leprae,abacteriumwhichhasundergone basedonthisstudy.Wepresentdatathatshowthatnot
dramatic gene decay during its evolution (Cole et al., onlydoestheMtbCYP121enzymebindtightlytoazole
2001). antifungal drugs, but that its K values for the azole
d
2946