pneumocystis carinii.dihydropteroate synthase mutations and treatment with sulfa or sulfone...

1969

Pneumocystis carinii Dihydropteroate Synthase but Not DihydrofolateReductase Gene Mutations Correlate with Prior Trimethoprim-Sulfamethoxazole or Dapsone Use

Liang Ma, Luciana Borio, Henry Masur,and Joseph A. Kovacs

Critical Care Medicine Department, Clinical Center,National Institutes of Health, Bethesda, Maryland

Recent studies of the human Pneumocystis carinii dihydropteroate synthase (DHPS) genesuggest that P. carinii is developing resistance to sulfamethoxazole (SMX) and dapsone. Toexplore whether P. carinii is also developing resistance to trimethoprim (TMP), the human P.carinii dihydrofolate reductase (DHFR) gene was cloned, DHFR and DHPS genes in 37 P.carinii isolates from 35 patients were sequenced, and the relationship between TMP-SMX ordapsone use and gene mutations was analyzed. The DHFR gene sequences were identical inall isolates except 1 with a synonymous substitution. In contrast, the DHPS gene sequencesshowed mutations in 16 of the 37 isolates; prior sulfa/sulfone prophylaxis was associated withthe presence of these mutations ( ). In addition to suggesting that there is less selectiveP ! .001pressure on DHFR than on DHPS, this study reinforces the hypothesis that mutations in theDHPS gene are likely involved in the development of sulfa resistance in P. carinii.

The combination of trimethoprim (TMP) and sulfamethox-azole (SMX) serves as the first-line therapeutic and prophylacticregimen for pneumonia caused by Pneumocystis carinii, whichremains a major opportunistic agent in patients infected withhuman immunodeficiency virus (HIV) and in other immuno-compromised patients. This combination inhibits 2 key enzymesin folate metabolism. Dapsone is another commonly used pro-phylactic agent, and, like SMX, it targets the enzyme dihydrop-teroate synthase (DHPS), which catalyzes the condensation ofpara-aminobenzoic acid with 7,8-dihydropterin-pyrophosphate(DHPPP), forming 7,8-dihydropteroate. TMP acts on dihydro-folate reductase (DHFR), which catalyzes the reduction of di-hydrofolate to tetrahydrofolate and is dependent on the reducedform of nicotinamide adenine dinucleotide phosphate.

The widespread use of DHFR and DHPS inhibitors in an-timicrobial chemotherapy has resulted in the emergence of an-tifolate drug resistance in numerous bacteria and some protozoa[1–10], raising concerns that similar resistance would develop

Received 29 April 1999; revised 23 August 1999; electronically published12 November 1999.

Presented in part: 36th Annual Meeting of the Infectious Diseases Societyof America, Denver, Colorado, November 1998 (abstract 417); 6th Inter-national Workshops on Opportunistic Protists, Raleigh, North Carolina,May 1999 (abstract W77).

Experimentation guidelines of the US Department of Health and HumanServices and the National Institutes of Health were followed in conductingthis research.

The nucleotide sequences reported in this study have been submitted tothe GenBank databases (accession numbers AF090368 and AF139132).

Reprints or correspondence: Dr. Joseph A. Kovacs, Bldg. 10, Rm. 7D43,National Institutes of Health, 10 Center Dr., MSC 1662, Bethesda, MD20892-1662 ([email protected]).

The Journal of Infectious Diseases 1999;180:1969–78q 1999 by the Infectious Diseases Society of America. All rights reserved.0022-1899/1999/18006-0027$02.00

in P. carinii. The molecular mechanisms of antifolate resistancehave been extensively studied, particularly in malaria. In Plas-modium species, it has been well established that DHFR pointmutations are responsible for differential resistance to DHFRinhibitors [4–7], whereas DHPS point mutations are associatedwith sulfa resistance [8–10].

TMP-SMX has been used for the treatment and preventionof P. carinii pneumonia (PCP) for 120 years. Despite the provenefficacy of TMP-SMX in both AIDS and non-AIDS patients,studies have shown that failure to respond clinically to thiscombination is ultimately observed in 10%–40% of patients[11–13]; the factors leading to this failure are not well under-stood. Recently, mutations in the P. carinii DHPS gene havebeen identified [14], primarily in patients who had previouslyreceived TMP-SMX or dapsone for prophylaxis, which suggeststhat the P. carinii DHPS gene is evolving under selective pres-sure because of the use of sulfa drugs. Further studies fromour group [15] and others [16] have indicated that point mu-tations of the P. carinii DHPS gene were associated with thefailure of sulfa or sulfone prophylaxis in AIDS patients, sug-gesting that these mutations confer sulfa resistance in P. carinii.These mutations are at one of the active sites of the enzyme[17] and correlate with mutations that confer resistance to sulfadrugs in other organisms. Because TMP is used together withSMX and dapsone, it is of particular interest to ask whetherP. carinii may simultaneously be developing mutations that sug-gest resistance to TMP.

Although the rat-derived P. carinii DHFR gene has beensequenced [18] and primers based on this sequence have beenreported to amplify the human P. carinii DHFR gene [19–23],we could not successfully amplify a region of this gene withthese primers. Given that homologous genes such as the DHPSgene [14] and those of major surface glycoproteins [24] are not

by guest on March 21, 2016

http://jid.oxfordjournals.org/D

ownloaded from

http://jid.oxfordjournals.org/

1970 Ma et al. JID 1999;180 (December)

Table 1. Oligonucleotide primers used for polymerase chain reaction analysisin the study.

Oligonucleotide Sequence (5′r3′) Template

FR01a ATGAATCAGCAAAAGTCTTTAACATTGATTGTT DHFRFR634a TTATAAATCTCTTGTCCACATTTCGAATTC DHFRHu172 ACTTTCCCATGTCTTACGACC DHFRHu528 TTCTCAGTCATGTTTGCCTTG DHFRFR208 GCAGAAAGTAGGTACATTATTACGAGA DHFRFR242 GTTTGGAATAGATTATGTTCATGGTGTACG DHFRFR1018 GCTTCAAACCTTGTGTAACGCG DHFRFR1038 AACCAGTTACCTAATCAAACTATATTGC DHFRPK2a CCTCATCCCCGAGTACTTGAACGTTCTTTTGTTTTA DHPSPSB1a TCTTGAAACTTTATACATTTCATAAACA DHPSPS634 GCAACCAAGGCAAACCATTAAAGC DHPSPS747 GCATCTGTTTTAGGAGGCTGTGA DHPSPK95 GCGTATCGAATGACCTTGTTCATCCT DHPSPK160 GTTAATCCTGGTATTAAACCAGTTTTGCCATT DHPSPS876 ATAAATTGCAAAATTACAATCAACCAAAGT DHPS

NOTE. DHFR, dihydrofolate reductase; DHPS, dihydropteroate synthase.a FR01 and FR634 were based on the rat-derived Pneumocystis carinii DHFR gene [18].

PK2 and PSB1 were based on the rat-derived P. carinii fas gene [26] and the human-derivedP. carinii DHPS gene (GenBank accession number U66282), respectively.

well conserved in P. carinii isolates from different hosts, it ap-peared likely that the DHFR gene was also poorly conserved.

In the present study, we have cloned and characterized thehuman P. carinii DHFR gene. We have also examined the entirecoding sequences of the DHFR and DHPS genes from 37 P.carinii isolates obtained from 35 patients and have attemptedto define the relationship between TMP-SMX or dapsone useand mutations in their therapeutic targets.

Materials and Methods

P. carinii organisms and DNA extraction. Between 1985 and1998, 37 human P. carinii samples were obtained from 35 patients,either during diagnostic procedures or at autopsy. Of the 35 pa-tients, 26 had AIDS, and 6 had other diseases (i.e., lymphocyticleukemia, non-Hodgkin’s lymphoma, breast cancer, aplastic ane-mia, metastatic neuroectodermal tumor, or rheumatoid arthritis);for 3 patients, no clinical data were available. The partial DHPSsequences for 2 patients have been reported elsewhere [15]. Cate-gorization of prior exposure to specific anti-Pneumocystis medi-cations was based on information available by chart review. Pa-tients were classified as having had “prior prophylaxis” if theadministration of drugs to prevent PCP was documented, regard-less of duration, and as having had “prior therapy” if therapy fora prior episode of PCP was documented. The “prior exposure”category included patients with documented prior therapy or pro-phylaxis with specific drugs. The sources of samples and the numberof samples from each source were as follows: bronchoalveolar la-vage fluid, 18; induced sputum, 14; lung specimen at autopsy, 4;and oral wash, 1. Two patients had 2 samples obtained 1 monthand 2.5 months apart, respectively. The presence of P. carinii in thesamples was confirmed by microscopic detection of the organism(35 samples) or by diagnostic polymerase chain reaction (PCR) (2samples). DNA was isolated from samples by treating cell pelletsor lung homogenates with proteinase K and sodium dodecyl sul-

fate, followed by phenol/chloroform extraction and ethanol pre-cipitation [25].

Cloning of the DHFR gene. Oligonucleotides based on regionsof the rat-derived P. carinii DHFR gene [18] encoding amino acidsconserved in the DHFR of other organisms (table 1) were used asprimers in a conventional PCR procedure, to amplify the centralcoding region of the human P. carinii DHFR gene. The PCR mix-ture (100 mL) contained 0.2 mg of DNA extracted from an autopsylung sample of a patient with PCP, 0.25 mM of each primer, 13

PCR buffer (10 mM Tris-HCl [pH 8.3], 50 mM KCl, 2.5 mMMgCl2), 0.2 mM dNTPs, and 2.5 U AmpliTaq Gold DNA poly-merase (Perkin-Elmer, Foster City, CA). Amplification was per-formed in a PTC-100 programmable thermal controller (MJ Re-search, Inc., Watertown, MA) with cycling parameters as follows:957C for 9 min and then 35 cycles at 947C for 1 min, 457C for 2min, and 727C for 3 min, with a final extension at 727C for 10 min.PCR products were subcloned into pCR 2.1 vector (Invitrogen,Carlsbad, CA) by use of the TA Cloning kit (Invitrogen). To am-plify the remainder of the DHFR gene, inverse PCR [27, 28] wasused. The template was prepared essentially as described by Och-man et al. [27]. In brief, human P. carinii genomic DNA (2.4 mg)was digested separately with restriction enzymes EcoRI, HindIII,SspI, and BstUI according to the manufacturer’s instructions (NewEngland Biolabs, Beverly, MA). Digested products were ligatedwith T4 DNA ligase (New England Biolabs), extracted with phenol/chloroform, precipitated with ethanol, and resuspended in steriledistilled water. The reaction mixture for the inverse PCR with prim-ers Hu172 and Hu528 was the same as for the conventional PCRdescribed previously, except that 50 ng of ligated template was used.After preincubation at 957C for 9 min, a touchdown protocol wasused in the amplification reaction, with the following cycling profile:10 cycles of denaturation at 947C for 1 min, annealing at 607C for2 min, with a 17C decrease per cycle (to 507C), and extension at727C for 3 min, followed by 25 cycles of 947C for 45 s, 507C for1 min, and 727C for 3 min. The reaction was terminated at 727Cfor 10 min. The PCR products were isolated on 1.2% agarose gels(0.9% NuSieve GTG/0.3% SeaKem GTG; FMC BioProducts,



ownloaded from


JID 1999;180 (December) Mutations in DHPS and DHFR of P. carinii 1971

Rockland, ME), extracted with the Ultrafree-MC columns (Mil-lipore, Boston, MA ), and then subjected to direct sequencing, asdescribed later.

Cloning of the DHPS gene. Because the reported human P.carinii DHPS gene [14] is incomplete in the 3′ end of the codingregion and because primers used in examining mutations in theDHPS gene are in the coding region, we sequenced the remainingcoding regions as well as upstream and downstream regions. Toamplify the 5′ region, an upstream primer (PK2) was selected froma conserved region in the hydroxymethyldihydropterin pyrophos-phokinase (PPPK) domain of the rat-derived P. carinii fas gene[26], and a downstream primer (PSB1) was taken from a previouslyobtained partial sequence of the human P. carinii DHPS gene [14].PCR amplification was carried out with the touchdown procedure,as described previously. To amplify the 3′ region of the DHPS gene,an inverse PCR was performed as described previously, by use ofprimers PS634 and PS747, which were designed from the 3′ end ofthe reported human P. carinii DHPS sequence [14].

Amplification of the P. carinii DHFR and DHPS genes from clin-ical samples. Once the complete sequences of the DHFR andDHPS genes from human-derived P. carinii were determined, wedesigned PCR primers to allow amplification and sequencing ofthe entire DHFR and DHPS coding domains from clinical isolates.To obtain DNA in sufficient quality and quantity for sequencing,we performed a primary PCR for all samples and a secondarynested or heminested PCR for those samples for which the initialPCR products were not visible or were very faint on agarose gels.In the primary PCR, 2–5 mL of DNA were used in a final volumeof 50 mL. In the secondary PCR, 1–5 mL of the primary PCRproducts were used in a final volume of 100 mL. To amplify theDHFR gene, the primers used in the primary PCR were FR208and FR1038, which covered the DHFR domain from base 2123to base 735, and the primers used in the secondary PCR wereFR242 and FR1018, which encompassed the DHFR domain frombase 289 to base 709. To amplify the DHPS gene, the primersused in the primary PCR were PK95 and PS876, which spannedthe DHPS domain from base 126 upstream of the first codon tobase 79 beyond the stop codon TGA, and the primers used in thesecondary PCR were PK160, which flanked the DHPS domainfrom base 61 upstream of the first codon, and PS876. The thermalcycling conditions for both the primary and secondary PCRs wereas follows: 957C for 9 min and then 35 cycles of 947C for 1 min,607C for 2 min, and 727C for 3 min, with a final cycle of 727C for10 min. PCR products were purified either on 1.2% agarosegels, as described previously, or by silica gel separation with theQIAquick PCR purification kit (QIAGEN, Santa Clarita, CA).The PCR products were then directly sequenced.

DNA sequencing and analysis. DNA sequencing with universalor sequence-specific primers was carried out by use of the dideoxychain termination reaction and the ABI PRISM dye terminatorcycle sequencing kit (Perkin-Elmer) and an ABI PRISM 377 au-tomated DNA sequencer (Perkin-Elmer). Sequencing data wereanalyzed by use of Sequencher 3.0 software (Gene Codes Corp.,Ann Arbor, MI). Amino acid sequences of both DHFR and DHPSwere deduced from the nucleic acid sequences and analyzed withMacVector 6.0.1 software (Oxford Molecular Group, Oxford, UK).

Statistical analysis. Statistical analyses were performed withSSPS 6.1 software for Macintosh (SPSS Inc., Chicago) or the

Primer of Biostatistics v3.0 software (McGraw-Hill, New York).Associations between DHPS mutations and sulfa/sulfone use wereanalyzed by use of Fisher’s exact test. was consideredP! .05significant.

Results

Cloning and sequencing of the DHFR gene. ConventionalPCR amplification of human P. carinii genomic DNA by useof DHFR-specific primers FR01 and FR634 yielded a singlefragment of ∼600 bp, which was absent in normal human ge-nomic DNA. Sequencing identified a 597-bp insert that ap-peared significantly homologous to the rat P. carinii DHFRgene as well as to other DHFR genes available in GenBank.Because no adequate genomic or cDNA library for human P.carinii is currently available, we used an inverse PCR to obtainthe 5′ and 3′ ends of the gene. When SspI-digested and circu-larized human P. carinii DNA was used as a template, the PCRamplification produced a strong band of ∼1.3 kb that was notpresent in the controls. Direct sequencing of this fragment re-vealed a 1281-bp sequence, which included the 5′ and 3′ endsof the DHFR coding sequence.

Combining the sequences of the conventional PCR and in-verse PCR fragments produced a 1616-bp sequence (GenBankaccession number AF090368). Alignment with the rat-derivedP. carinii DHFR gene [18] revealed a 618-bp DHFR codingsequence interrupted by a 42-bp intron, an 846-bp 5′ flankingsequence, and a 110-bp 3′ flanking sequence. The A1T contentis 67.2% in the entire 1616-bp sequence and 63% in the codingregion, which is similar to that of other reported P. carinii genes.Although the 42-bp intron located at bases 267–308 is identicalin location to that in the rat-derived P. carinii DHFR gene, itis 1 bp smaller than that intron and has a different sequence[18]. It contains a 5′-GT splice donor site and a 3′-AG spliceacceptor site, which are similar to those in other introns iden-tified in P. carinii [18, 29–31]. Inspection of the 5′ noncodingregion near the ATG start codon revealed the presence of aputative TATA box (TATAATT) at base 246, and a putativeCAAT box (GGACAATCG) at base 217. Presumably, thesequence AATATA or ATAAAA, located beyond the stop co-don at base 48 or 77, respectively, may serve as the polyaden-ylation signal.

Examination of the sequence upstream of the DHFR domainrevealed a short open reading frame of 156 bp starting at base2600; translation yields a 52–amino acid polypeptide with amolecular mass of 5774 Da. A GenBank search revealed thatthe deduced polypeptide was most similar (34% identity) to ahypothetical protein in Schizosaccharomyces pombe (GenBankaccession number O14010).

The DHFR coding region was predictive of a 206–aminoacid protein with a calculated molecular mass of 23,407 Da.Comparison with other DHFR sequences available in Gen-Bank revealed that the human-derived P. carinii DHFR is most



ownloaded from




ownloaded from


Fig

ure

1.A

lignm

ent

ofde

duce

dse

quen

ceof

hum

anP

neum

ocys

tis

cari

nii

dihy

drof

olat

ere

duct

ase

(DH

FR

)w

ith

that

ofD

HF

Rfr

omse

lect

edor

gani

sms.

Box

esar

edr

awn

arou

ndid

enti

cal

(dar

ksh

adin

g)an

dsi

mila

r(l

ight

shad

ing)

resi

dues

.Gen

Ban

kac

cess

ion

num

bers

for

thes

ese

quen

ces

are

ratP

.car

inii,

M26

495;

Sch

izos

acch

arom

yces

pom

be(r

esid

ues

206–

461)

,P36

591;

Filo

basi

diel

lane

ofor

man

s,Q

0780

1;C

andi

daal

bica

ns,X

7896

8;P

lasm

odiu

mfa

lcip

arum

,J03

772;

Tox

opla

sma

gond

ii,G

7293

70;E

sche

rich

iaco

li,J0

1609

;Bac

illus

subt

ilis,

M20

012;

and

Hom

osa

pien

s,J0

0140

.Alig

nmen

twas

perf

orm

edby

use

ofC

lust

alW

soft

war

e(M

acV

ecto

r;O

xfor

dM

olec

ular

Gro

up,

Oxf

ord,

UK

).



ownloaded from



Figure 2. Comparison of the rat Pneumocystis carinii dihydrofolatereductase (DHFR) primers [20] and the human P. carinii DHFR prim-ers for polymerase chain reaction (PCR) amplification of a homologousregion (273 bp) of the rat and human P. carinii DHFR genes. A, Se-quence alignment of the sense and antisense primers. Identical nucle-otides are boxed and shaded. B, Analysis of PCR products by a 4%agarose gel containing SYBR green I stain (Molecular Probes, Eugene,OR). Lane M, PCR marker (Sigma Chemical Co., St. Louis); lanes 1,3, 5, and 7, rat P. carinii DHFR primers; lanes 2, 4, 6, and 8, humanP. carinii DHFR primers. DNA templates included rat P. carinii ge-nomic DNA (lanes 1 and 2) [24], human P. carinii genomic DNA (lanes3 and 4), plasmid DNA containing the human P. carinii DHFR domain(lanes 5 and 6), and normal human genomic DNA (lanes 7 and 8).

closely related to the DHFR of rat-derived P. carinii (62% iden-tity), S. pombe (39% identity), and Filobasidiella neoformans(35% identity). Figure 1 shows an alignment of 10 DHFR se-quences from different organisms. In all the sequences, 16 res-idues are identical and 21 are similar. A putative DHFR sig-nature for binding of the folate substrate [32], IGlkndLPW, isidentified at residues 19–27.

Although the human P. carinii DHFR is most similar to therat P. carinii DHFR, there are substantial differences in boththe nucleotide and amino acid sequences. The 5′ and 3′ flankingsequences are 48% different in the overlapping regions, and thecoding sequences are 30% (186 of 618) different. The deducedamino acid sequences are 38% (79 of 206) different.

Some investigators have reported that primers based on therat P. carinii DHFR gene are useful in diagnosing human P.carinii infections [19–23]. To investigate this matter further, weused previously reported PCR conditions [20] to compare thepreviously used rat P. carinii DHFR primers [19, 20] with prim-ers that were based on a homologous region of the human P.carinii DHFR gene in terms of their efficiency in amplifyingthe rat and human P. carinii DHFR genes (figure 2A). As shownin figure 2B, when PCR was performed by use of the rat P.carinii DHFR primers and rat P. carinii DNA, a single bandwith the predicted size of 273 bp was observed; when humanP. carinii DNA was used, a strong band was also observed, butits size was smaller than expected. No product was seen whenplasmid DNA containing the human P. carinii DHFR gene wasamplified with these primers. In contrast, PCR with primersfrom the human P. carinii DHFR gene amplified a band of theexpected size (273 bp) from human P. carinii genomic DNAand plasmid DNA containing the human P. carinii DHFR gene.This band was not seen in rat P. carinii DNA or in normalhuman DNA.

Cloning and sequencing of the DHPS gene. ConventionalPCR amplification of human P. carinii genomic DNA by useof DHPS-specific primers PK2 and PSB1 produced a 987-bpfragment, including 183 bp of the 3′ end of the PPPK domainwith 80% homology to the 3′ end of the PPPK domain of therat-derived P. carinii fas gene [26], and 804 bp identical to the5′ region of the published sequence of the partial human P.carinii DHPS gene (GenBank accession number U66282). The3′ end of the gene was cloned by inverse PCR with primersPS634 and PS747. The 299-bp insert included the 3′ end of thegene and 79 bp of the 3′ untranslated region. The length of thenucleotide sequence assembled from the 2 PCR-generated frag-ments just mentioned is 1099 bp (GenBank accession numberAF139132) and includes the entire DHPS domain (834 bp).The PPPK domain, which is interrupted by a 45-bp intron,deduces a 46–amino acid sequence that shares 71.7% identitywith the C-terminal of the rat P. carinii PPPK domain [26].

Polymorphisms of the P. carinii DHFR and DHPS genes inclinical isolates. PCR was performed with primers fromregions outside the coding region, to amplify the entire coding

sequences of both the DHFR and DHPS genes in 37 P. cariniisamples from 35 patients. The DHFR gene was amplified from17 samples by use of a single round of PCR (858 bp) and from20 samples by use of a nested PCR (798 bp). The DHPS genewas obtained from 14 samples by use of a single round of PCR(1030 bp) and from 23 samples by use of a nested PCR (965bp). All the PCR products were directly sequenced on bothstrands.

All 37 samples showed identical DHFR sequences, except



ownloaded from



Table 2. Polymorphisms of the human Pneumocystis carinii DHPSgene in 37 samples from 35 patients.

No. ofsamples

No. ofpatients

Codona

55 57 171

18 (wild-type)b 17 Thr (ACA) Pro (CCT) Ser (TCA)3 3 Thr (ACA) Pro (CCT) Ser (TCG)6 6 Ala (GCA) Pro (CCT) Ser (TCA)

10b 9 Ala (GCA) Ser (TCT) Ser (TCA)

NOTE. DHPS, dihydropteroate synthase.a Nucleotides and amino acids that differ from the wild-type sequence are

underlined.b Including 2 samples from 1 patient.

Table 3. Correlation of the Pneumocystis carinii DHPS mutations atcodon 55 and/or codon 57 with prior sulfa/sulfone use in 32 patients.

Correlation factor

DHPS mutations

PYes No

Prior exposureYes 11 5 .011No 3 12

Prior prophylaxisYes 11 2 !.001No 3 16

Prior therapyYes 5 5 .070No 8 12

NOTE. Two patients, one with wild-type and the other with mutant DHPS,had prior P. carinii pneumonia, but the specific treatment was not recorded, andthey were excluded from the prior therapy group. One of these patients, who didnot receive prophylaxis, was also excluded from the prior exposure group. Theother received dapsone as prophylaxis. DHPS, dihydropteroate synthase.

for 1 sample, which had a single nucleotide change from T toC at position 312. This change is synonymous and does notresult in an amino acid change.

In contrast, the DHPS sequences in the 37 samples exhibitedpolymorphisms at 3 nucleotide positions (table 2). Eighteensamples contained the wild-type sequence, which is character-ized by the presence of Thr-55 and Pro-57. Three samplesshowed a single nucleotide change from A to G at position 513(codon 171), which did not result in an amino acid change. Inthe remaining 16 samples, we observed nonsynonymous mu-tations leading to amino acid changes at codons 55 and 57. Asingle 55ThrrAla mutation was observed in 6 samples, in-cluding 4 samples containing a mixture of Thr-55 and Ala-55.Mutations 55ThrrAla coupled with 57ProrSer occurred in 10samples, including 2 samples that had a mixture of Pro-57and Ser-57 and 1 sample that had mixtures of Thr-55 andAla-55 and of Pro-57 and Ser-57. Mutations 55ThrrAla and57Pror Ser are identical to those identified in previous studies[14–16]. The nucleotide change at position 513 (codon 171) hasnot been previously reported. None of the polymorphisms atcodons 23, 60, 111, or 248 that were reported by Lane et al.[14] were observed in the present study.

As noted previously, 7 samples had both the wild-type andmutant DHPS alleles present at codon 55 and/or codon 57, asdetermined by direct sequencing. To verify these mixed forms,we performed enzyme restriction analysis and subcloning of thePCR products. When the PCR products were subjected to di-gestion with AccI, which recognizes the wild-type sequence atcodon 55, and HaeIII, which cleaves the wild-type sequence atcodon 57, all 7 samples showed fragments that appeared inhomozygous wild-type control samples and in homozygous mu-tant control samples (data not shown), indicating the presenceof both wild-type and mutant sequences. We subcloned thePCR products of 1 sample with a clear mutation at codon 55and a mixture of wild-type and mutant codons at 57 and ofanother sample with both wild-type and mutant codons at 55and 57. In the first sample, we obtained 4 clones containing amutant codon 55 and the wild-type codon 57 and 5 clonescontaining a mutant sequence at both codons 55 and 57. Inthe second sample, we obtained 6 clones containing the wild-

type sequence and 2 clones containing a mutant sequence atboth codons 55 and 57.

Correlation between prior sulfa/sulfone use and DHPS mu-tations at codon 55 and/or codon 57. Clinical data regardingprior therapy or prophylaxis were available for 30 and 32 pa-tients, respectively (table 3). Two patients had a prior episodeof PCP, but the therapy was not reported. Prior therapy withantifolate drugs was documented for 9 patients, of whom 8 hadreceived TMP-SMX and 1 had received TMP-SMX followedby TMP-dapsone. Prior prophylaxis was documented for 13patients, of whom 6 had received dapsone only and 7 hadreceived TMP-SMX or TMP-dapsone. Six patients had re-ceived both prior prophylaxis and prior therapy. Of the 16patients with prior exposure, 12 had prior episodes of PCP, and3 of 16 patients with no documented prior exposure to anysulfa- or sulfone-containing agent had prior episodes of PCP.

Using results from the first isolate from the 2 patients whoeach had 2 isolates, we found a significant association betweenprior exposure to TMP-SMX or dapsone and the presence ofmutations at codon 55 and/or codon 57. Mutations were foundin 11 of 16 patients with documented prior exposure, comparedwith 3 of 15 patients without such exposure ( ). ThisP = .011association was related primarily to prior prophylaxis (P!

); prior therapy was not associated with the presence of.001these mutations ( ). Of note, all 3 patients whose isolatesP = .70showed mutations but who did not have prior exposure toSMX/dapsone had a mixed wild-type and mutant DHPS atcodon 55, with a wild-type sequence at codon 57. For the 23patients whose therapy regimens contained a sulfonamide ordapsone, no correlation was found between survival after ini-tiation of the therapy and the presence of mutations at codon55 and/or codon 57 ( ). Prior exposure to pentamidineP = .36was documented for 12 patients, 11 of whom also had priorexposure to TMP-SMX or dapsone. No correlation was foundbetween prior exposure to pentamidine and the presence ofDHPS mutations (data not shown).

Although no mutations in the DHFR gene were seen in the



ownloaded from



Figure 3. Alignment of dihydropteroate synthase (DHPS) protein sequences from different organisms over the regions of mutations at residues55 and 57 (arrows) in mutant human Pneumocystis carinii DHPS (hPC.mut) and wild-type human P. carinii DHPS (hPC.wild). Conserved residuesare boxed. GenBank accession numbers for these sequences are rat P. carinii (rPC), M86602; mouse P. carinii (mPC), U66283; Saccharomycescerevisiae (SC), X96722; Plasmodium falciparum (PF), U07706; Escherichia coli (EC), L06494; Bacillus subtilis (BS), D26185; Streptococcus pneu-moniae (SP), U16156; and Neisseria meningitidis (NM), A57423.

37 samples from 35 patients, 15 patients had been previouslyexposed to TMP, with 10 receiving prior therapy and 7 receivingprior prophylaxis with TMP.

Discussion

To better understand the relationship between TMP-SMXor dapsone use and mutations in their target enzymes, wecloned and sequenced the DHFR gene of human P. carinii andexamined sequence polymorphisms of this gene, as well as thoseof the DHPS gene, in 37 clinical isolates of P. carinii. We foundthat the human P. carinii DHFR gene is closely related to butclearly different from the rat P. carinii DHFR gene. Thus, theDHFR gene, like all genes studied to date, is not well conservedin P. carinii isolates from different hosts.

To date, we have sequenced the entire DHFR and DHPSdomains of 37 P. carinii isolates from 35 patients. Surprisingly,we found that despite prior exposure to TMP in 15 patients,the sequences of the DHFR gene in all the isolates were identicalexcept for 1 isolate with a single synonymous substitution. Bycontrast, 16 (43%) of the 37 isolates (15 of 35 patients) hadDHPS mutations at codon 55 and/or codon 57.

Since the sequence polymorphism of P. carinii DHPS wasfirst observed in 6 human isolates [14], efforts have been madeto determine its clinical significance. Kazanjian et al. [16] ex-amined partial P. carinii DHPS nucleotide sequences from 27PCP patients and found that mutations at codons 55 and 57were common in patients who had received sulfa/sulfone pro-phylaxis, affecting 5 of 7 such patients, whereas only 2 of 13patients who had not been receiving sulfa/sulfone prophylaxishad mutant DHPS genes. Similarly, in a previous study [15],these mutations were detected in 2 patients with PCP (alsoincluded in this report) who had failed prophylaxis or treatmentwith TMP-SMX. The present study showed a significant as-

sociation between TMP-SMX or dapsone prophylaxis and mu-tations at codon 55 or 57, supporting the hypothesis that DHPSmutations at these sites reflect the development of sulfa resis-tance by human P. carinii. Prior therapy with TMP-SMX ordapsone for PCP was not associated with the development ofthese mutations, suggesting that long-term exposure to low lev-els of TMP-SMX or dapsone facilitates the development ofthese mutations.

The hypothesis regarding resistance is further supported bycomparative structural and functional data of DHPS fromother organisms. Figure 3 shows an alignment of the affectedregions of the predicted P. carinii DHPS protein with the se-quences of the predicted DHPS protein from other organisms,including both prokaryotic and eukaryotic species. The mu-tations (55ThrrAla and 57ProrSer) are located in a highlyconserved region, which is characterized by the sequence Thr-Arg-Pro, which is identical for all the organisms shown exceptPlasmodium falciparum, in which the sequence is Ser-Ala-Pro(codons 436–438). On the basis of the crystal structure of theEscherichia coli DHPS, the analogous 3 residues (codons 62–64in E. coli) are involved in binding to the substrate DHPPP aswell as to sulfonamidine [17]. Mutations at or very near thesepositions are likely to alter the local structure and thus affectthe binding of the substrate and sulfa. Mutations at these po-sitions have been shown to confer resistance to sulfa drugs inorganisms such as P. falciparum [8, 9, 28], E. coli [3], and Strep-tococcus pneumoniae [1].

Including only the first isolate from the 2 patients who eachhad 2 isolates, 12 of 24 isolates obtained in the 1990s had DHPSmutations at codon 55 and/or codon 57, whereas 3 of 11 isolatesobtained in the 1980s had these mutations. This finding suggeststhat the mutations are more likely to occur in recent patientisolates, which is in accord with a previous observation byKazanjian et al. [16], who sequenced the DHPS gene in 27



ownloaded from



human isolates obtained between 1977 and 1997 and detected7 mutant isolates, all of which were obtained from 1995 through1997.

On the basis of the data from the present study and previousstudies [16], it appears evident that the 55ThrrAla mutationis the most common mutation in the DHPS gene. This mutationcan occur either alone or in combination with the 57ProrSermutation, which is less common and almost always coupledwith the 55ThrrAla mutation, suggesting that the 55ThrrAlamutation may be the first mutation to occur in response to drugpressure. The presence of a mixture of wild-type and mutantDHPS in 7 isolates suggests a transition from the wild-type tothe mutant gene, although a mixed infection cannot be ruledout. Of note, 3 isolates from patients without prior exposureto SMX/dapsone showed a mixed wild-type and mutant DHPSat codon 55, with a wild-type sequence at codon 57. This findingmay represent allelic variations or possible selective pressurefrom unidentified exposure to sulfa drugs.

One of the more striking findings of the study is the lack ofmutations in the P. carinii DHFR gene, because mutations inthe DHPS gene were found in 43% of isolates examined. Al-though further studies are needed because fewer patients wereexposed to long-term TMP than to a sulfa drug, the discordantoccurrence of mutations of the DHFR and DHPS genes inthese isolates suggests that the selective pressures on the 2 genesare not comparable—that is, there might be less selective pres-sure on DHFR than on DHPS. Absence of mutations in theDHFR gene further suggests that the DHPS mutations are notrandom occurrences but rather the result of antibiotic pressure,and it supports the concept that TMP contributes little to theefficacy of the TMP-SMX combination against P. carinii. Thisparticular fixed-combination product is used because it wasshown to be effective in animal and human trials of PCP, but,in fact, it was initially chosen for trials because it had beenformulated for bacterial indications and was readily available.Potent DHFR inhibitors such as trimetrexate and piritreximare active as single agents against both murine and human PCP,whereas TMP is 2 logs less potent in in vitro assays [18, 33–35].Moreover, studies in animal models have found that TMP aloneis ineffective in treating or preventing PCP [36, 37], that sul-fonamides alone, including SMX, are highly effective [37, 38],and that the addition of TMP to sulfonamides does not enhancethe efficacy of sulfonamides alone [37]. Thus, it is not clearwhether TMP truly contributes to the efficacy of TMP-SMXtherapy; in fact, it may add toxicity without substantial benefit.

The marked difference between the human- and rat-derivedP. carinii DHFR genes explains the conflicting reports on theuse of the rat P. carinii DHFR primers to diagnose human PCP.Schluger et al. [19, 20] first reported the successful use of primersbased on the rat P. carinii DHFR gene, detecting both rat andhuman P. carinii in limited serum and respiratory specimens.Subsequent studies with these primers by other investigators

[21, 22] showed low sensitivity (23%–71%) and specificity(45.5%). Ortona et al. [23] reported that the amplified productsfrom human P. carinii isolates were unrelated to P. cariniiDHFR or any other form of DHFR. In the present study, wehave demonstrated that the previously described primers donot amplify the human P. carinii DHFR gene (figure 2B). Acomparison of the sequences of primers from a homologousregion of the rat and human P. carinii DHFR genes shows thatthe sense and antisense primers differed by 25% (5 of 20) and20% (4 of 20) of nucleotides, respectively (figure 2A). Of note,the 2 nucleotides at the 3′ ends of the antisense primers weredifferent, which is likely to hamper the annealing of the primerswith heterogeneous templates, thereby blocking PCR amplifi-cation. It is evident from these data that the previously reportedP. carinii DHFR primers are specific for rat P. carinii but arenot useful for the detection of human P. carinii.

Although definitive demonstration that P. carinii is devel-oping resistance to sulfa drugs will require additional studieswith native or recombinant enzyme or in vitro culture systems[39], the mounting molecular epidemiologic evidence reportedhere and elsewhere [14–16] supports the conclusion that P. car-inii is indeed developing such resistance. At present, TMP-SMXremains a highly effective drug for the treatment of PCP. How-ever, if resistance to sulfa drugs becomes more widespread andclinically significant in the future, combining sulfa drugs withinhibitors of P. carinii DHFR that are more potent than TMPwould likely provide better anti–P. carinii activity.

Acknowledgments

We are indebted to Vivian Sorial and Ross E. Turner for technicalassistance and to Dr. Qin Mei for support.

References

1. Lopez P, Espinosa M, Greenberg B, Lacks SA. Sulfonamide resistance inStreptococcus pneumoniae: DNA sequence of the gene encoding dihy-dropteroate synthase and characterization of the enzyme. J Bacteriol1987;169:4320–6.

2. Adrian PV, Klugman KP. Mutations in the dihydrofolate reductase gene oftrimethoprim-resistant isolates of Streptococcus pneumoniae. AntimicrobAgents Chemother 1997;41:2406–13.

3. Vedantam G, Guay GG, Austria NE, Doktor SZ, Nichols BP. Characteri-zation of mutations contributing to sulfathiazole resistance in Escherichiacoli. Antimicrob Agents Chemother 1998;42:88–93.

4. Peterson DS, Walliker D, Wellems TE. Evidence that a point mutation indihydrofolate reductase–thymidylate synthase confers resistance to pyri-methamine in falciparum malaria. Proc Natl Acad Sci USA 1988;85:9114–8.

5. Peterson DS, Milhous WK, Wellems TE. Molecular basis of differential re-sistance to cycloguanil and pyrimethamine in Plasmodium falciparum ma-laria. Proc Natl Acad Sci USA 1990;87:3018–22.

6. Cowman AF, Morry MJ, Biggs BA, Cross GA, Foote SJ. Amino acid changeslinked to pyrimethamine resistance in the dihydrofolate reductase–thymidylate synthase gene of Plasmodium falciparum. Proc Natl Acad SciUSA 1988;85:9109–13.



ownloaded from



7. Sirawaraporn W, Prapunwattana P, Sirawaraporn R, Yuthavong Y, SantiDV. The dihydrofolate reductase domain of Plasmodium falciparum thy-midylate synthase–dihydrofolate reductase: gene synthesis, expression,and anti-folate-resistant mutants. J Biol Chem 1993;268:21637–44.

8. Brooks DR, Wang P, Read M, Watkins WM, Sims PF, Hyde JE. Sequencevariation of the hydroxymethyldihydropterin pyrophosphokinase: dihy-dropteroate synthase gene in lines of the human malaria parasite, Plas-modium falciparum, with differing resistance to sulfadoxine. Eur J Biochem1994;224:397–405.

9. Triglia T, Menting JG, Wilson C, Cowman AF. Mutations in dihydropteroatesynthase are responsible for sulfone and sulfonamide resistance in Plas-modium falciparum. Proc Natl Acad Sci USA 1997;94:13944–9.

10. Triglia T, Wang P, Sims PF, Hyde JE, Cowman AF. Allelic exchange at theendogenous genomic locus in Plasmodium falciparum proves the role ofdihydropteroate synthase in sulfadoxine-resistant malaria. EMBO J1998;17:3807–15.

11. Hardy WD, Feinberg J, Finkelstein DM, et al. A controlled trial of tri-methoprim-sulfamethoxazole or aerosolized pentamidine for secondaryprophylaxis of Pneumocystis carinii pneumonia in patients with the ac-quired immunodeficiency syndrome. AIDS Clinical Trials Group Protocol021. N Engl J Med 1992;327:1842–8.

12. Klein NC, Duncanson FP, Lenox TH, et al. Trimethoprim-sulfamethoxazoleversus pentamidine for Pneumocystis carinii pneumonia in AIDS patients:results of a large prospective randomized treatment trial. AIDS 1992;6:301–5.

13. Bozzette SA, Finkelstein DM, Spector SA, et al. A randomized trial of threeantipneumocystis agents in patients with advanced human immunodefi-ciency virus infection. NIAID AIDS Clinical Trials Group. N Engl J Med1995;332:693–9.

14. Lane BR, Ast JC, Hossler PA, et al. Dihydropteroate synthase polymorphismsin Pneumocystis carinii. J Infect Dis 1997;175:482–5.

15. Mei Q, Gurunathan S, Masur H, Kovacs JA. Failure of co-trimoxazole inPneumocystis carinii infection and mutations in dihydropteroate synthasegene [letter]. Lancet 1998;351:1631–2.

16. Kazanjian P, Locke AB, Hossler PA, et al. Pneumocystis carinii mutationsassociated with sulfa and sulfone prophylaxis failures in AIDS patients.AIDS 1998;12:873–8.

17. Achari A, Somers DO, Champness JN, Bryant PK, Rosemond J, StammersDK. Crystal structure of the anti-bacterial sulfonamide drug target dihy-dropteroate synthase. Nat Struct Biol 1997;4:490–7.

18. Edman JC, Edman U, Cao M, Lundgren B, Kovacs JA, Santi DV. Isolationand expression of the Pneumocystis carinii dihydrofolate reductase gene.Proc Natl Acad Sci USA 1989;86:8625–9.

19. Schluger N, Sepkowitz K, Armstrong D, et al. Detection of Pneumocystiscarinii in serum of AIDS patients with Pneumocystis pneumonia by thepolymerase chain reaction. J Protozool 1991;38:240S–242S.

20. Schluger N, Godwin T, Sepkowitz K, et al. Application of DNA amplificationto pneumocystosis: presence of serum Pneumocystis carinii DNA duringhuman and experimentally induced Pneumocystis carinii pneumonia. JExp Med 1992;176:1327–33.

21. Lu JJ, Chen CH, Bartlett MS, Smith JW, Lee CH. Comparison of six differentPCR methods for detection of Pneumocystis carinii. J Clin Microbiol1995;33:2785–8.

22. De Luca A, Tamburrini E, Ortona E, et al. Variable efficiency of three primerpairs for the diagnosis of Pneumocystis carinii pneumonia by the poly-merase chain reaction. Mol Cell Probes 1995;9:333–40.

23. Ortona E, Margutti P, De Luca A, et al. Nonspecific PCR products usingrat-derived Pneumocystis carinii dihydrofolate reductase gene–specificprimers in DNA amplification of human respiratory samples. Mol CellProbes 1996;10:187–90.

24. Kovacs JA, Powell F, Edman JC, et al. Multiple genes encode the majorsurface glycoprotein of Pneumocystis carinii. J Biol Chem 1993;268:6034–40.

25. Davis LG, Dibne MD, Battey JF. Basic methods in molecular biology. NewYork: Elsevier, 1986:310–3.

26. Volpe F, Dyer M, Scaife JG, Darby G, Stammers DK, Delves CJ. The mul-tifunctional folic acid synthesis fas gene of Pneumocystis carinii appearsto encode dihydropteroate synthase and hydroxymethyldihydropterinpyrophosphokinase. Gene 1992;112:213–8.

27. Ochman H, Gerber AS, Hartl DL. Genetic applications of an inverse poly-merase chain reaction. Genetics 1988;120:621–3.

28. Triglia T, Peterson MG, Kemp DJ. A procedure for in vitro amplification ofDNA segments that lie outside the boundaries of known sequences. Nu-cleic Acids Res 1988;16:8186.

29. Edman U, Edman JC, Lundgren B, Santi DV. Isolation and expression ofthe Pneumocystis carinii thymidylate synthase gene. Proc Natl Acad SciUSA 1989;86:6503–7.

30. Sunkin SM, Stringer JR. Transcription factor genes from rat Pneumocystiscarinii. J Eukaryot Microbiol 1995;42:12–9.

31. Lugli EB, Allen AG, Wakefield AE. A Pneumocystis carinii multi-gene familywith homology to subtilisin-like serine proteases. Microbiology 1997;143:2223–36.

32. Bairoch A. PROSITE: a dictionary of sites and patterns in proteins. NucleicAcids Res 1992;20(Suppl):2013–8.

33. Allegra CJ, Kovacs JA, Drake JC, Swan JC, Chabner BA, Masur H. Activityof antifolates against Pneumocystis carinii dihydrofolate reductase andidentification of a potent new agent. J Exp Med 1987;165:926–31.

34. Sirawaraporn W, Edman JC, Santi DV. Purification and properties of re-combinant Pneumocystis carinii dihydrofolate reductase. Protein ExprPurif 1991;2:313–6.

35. Margosiak SA, Appleman JR, Santi DV, Blakley RL. Dihydrofolate reduc-tase from the pathogenic fungus Pneumocystis carinii: catalytic propertiesand interaction with antifolates. Arch Biochem Biophys 1993;305:499–508.

36. Kluge RM, Spaulding DM, Spain AJ. Combination of pentamidine andtrimethoprim-sulfamethoxazole in therapy of Pneumocystis carinii pneu-monia in rats. Antimicrob Agents Chemother 1978;13:975–8.

37. Walzer PD, Kim CK, Foy JM, Linke MJ, Cushion MT. Inhibitors of folicacid synthesis in the treatment of experimental Pneumocystis carinii pneu-monia. Antimicrob Agents Chemother 1988;32:96–103.

38. Hughes WT, Killmar J. Monodrug efficacies of sulfonamides in prophylaxisfor Pneumocystis carinii pneumonia. Antimicrob Agents Chemother1996;40:962–5.

39. Merali S, Frevert U, Williams JH, Chin K, Bryan R, Clarkson AB Jr. Con-tinuous axenic cultivation of Pneumocystis carinii. Proc Natl Acad SciUSA 1999;96:2402–7.



ownloaded from


pneumocystis carinii.dihydropteroate synthase mutations and treatment with sulfa or sulfone...

Documents