Cloning and Characterization of a Novel Trimethoprim-Resistant Dihydrofolate Reductase from a Nosocomial Isolate of Staphylococcus aureus CM.S2 (IMCJ1454)

Abstract

A novel gene, dfrG, encoding a trimethoprim (TMP)-resistant dihydrofolate reductase (DHFR, designated S3DHFR) was cloned from a clinical isolate of methicillin-resistant Staphylococcus aureus. Escherichia coli expressing dfrG was highly resistant to TMP. Recombinant S3DHFR exhibited DHFR activity that was not inhibited by TMP.

Trimethoprim (TMP) is a potent inhibitor of bacterial dihydrofolate reductase (DHFR) and is effective in vitro against methicillin-resistant Staphylococcus aureus (MRSA). In combination with sulfamethoxazole, TMP has been used successfully to treat patients infected with MRSA and is effective at eradicating carriage (10, 16). Resistance of S. aureus to TMP was first reported in the 1980s (12) and was found to be due to plasmid-mediated production of an additional DHFR that was less sensitive to TMP than intrinsic DFHR (S. aureus DHFR [SaDHFR]) encoded by the dfrB gene on the chromosome (1, 12). Plasmid-mediated production of an additional TMP-resistant DHFR is one of the most common mechanisms of resistance to TMP in bacterial organisms. At least 14 different types of TMP-resistant DHFRs in gram-negative bacteria have been reported (10); however, only a limited number of TMP-resistant DHFRs in gram-positive bacteria have been reported (10).

A total of 43 clinical isolates of MRSA from Chiang Mai, Thailand, and 244 clinical isolates of MRSA from Tokyo, Japan, were analyzed in this study. All isolates were positive for dfrB by PCR and also positive for femB encoding coagulase and for mecA associated with methicillin resistance. All isolates from Chiang Mai, Thailand, were resistant to TMP, whereas all those from Tokyo, Japan, except one, S. aureus IMCJ934, were sensitive to TMP (Table 1). Crude extracts prepared from a TMP-resistant isolate from Chiang Mai, S. aureus CM.S2 (IMCJ1454), showed DHFR activity, and K_m values of the extract for DHF and NADPH were similar to those of crude extracts from TMP-sensitive strain ATCC 25923 (Table 2); however, the 50% inhibitory concentration (IC₅₀) of TMP for the crude extract of strain CM.S2 was more than 15,000-fold greater than that of ATCC 25923.

TABLE 1.

MICs of trimethoprim in S. aureus and E. coli strains

Strain	MIC of TMP (μg/ml)	Characteristic(s) or genotype
S. aureus CM.S2 (IMCJ1454)	>512	Clinical isolate from Chiang Mai, Thailand, in 2003
S. aureus IMCJ934	>512	Clinical isolate from Tokyo, Japan, in 2001
S. aureus ATCC 29213	4	Quality control strain for antimicrobial susceptibility testing
E. coli DH5α(pSA1)	>512	Transformant harboring a 3.5-kb BamHI fragment with dfrG ligated to pHSG398
E. coli DH5α(pHSG398)	<2	Transformant harboring pHSG398
E. coli DH5α(pT7dfrG)	>512	Transformant harboring PCR-amplified dfrG ligated to pCRT7/NT
E. coli DH5α(pT7dfrB)	128	Transformant harboring PCR-amplified intrinsic dfrB ligated to pCRT7/NT
E. coli DH5α(pCRT7/NT)	<2	Transformant harboring pCRT7/NT
E. coli DH5α	<2	supE44 hsdR17 recA1 gyrA96 endA1 thi-1 relA1

TABLE 2.

Enzyme kinetic and inhibitory properties of staphylococcal DHFRs

DHFR	Origin	Km (μM)		IC50 of TMP (μM)
		DHF	NADPH
Crude enzyme	S. aureus CM.S2 (IMCJ1454)	5.83 ± 2.09	15.17 ± 1.73	214
	S. aureus ATCC 25923	3.16 ± 1.99	14.78 ± 2.73	0.013
TMP-resistant DHFRs
S3DHFR	S. aureus CM.S2 (IMCJ1454)	2.68 ± 1.09	2.38 ± 1.97	254
S2DHFRa	S. haemolyticus MUR313	5.1	1.7	127
S1DHFRa	S. aureus	6.6	12.4	9.8
TMP-sensitive DHFRs
SaDHFRCM.S2	S. aureus CM.S2 (IMCJ1454)	3.01 ± 1.40	2.97 ± 0.57	0.014
SaDHFRa	S. aureus ATCC 25923			0.012

^aData from references 6, 7, and 8.

HindIII-digested fragments of the S. aureus CM.S2 genome were cloned, transformed into Escherichia coli DH5α cells, and selected on agar medium containing TMP (8 μg/ml). The resultant plasmid, named pSA1, had a 3.5-kb insert containing a complete open reading frame (ORF) surrounded by truncated ORFs (data not shown). The complete ORF consisted of 498 bp encoding a putative protein of 165 amino acids with similarities to TMP-resistant DHFR from Staphylococcus haemolyticus (79% identity) (7), Bacillus anthracis (67% identity) (2), and Bacillus cereus (65% identity) (15) (Fig. 1). The deduced protein is somewhat less similar to the intrinsic TMP-sensitive DHFRs from S. aureus (SaDHFR) (8), S. epidermidis (SeDHFR) (6), and E. coli K-12 (17), with 41%, 40%, and 40% similarity, respectively (Fig. 1). This complete ORF was named dfrG, and the deduced protein was designated S3DHFR. Amino acid sequence alignment of DHFRs suggests that residues involved in the binding of TMP and NADPH in other DHFRs are conserved in S3DHFR (Fig. 1). An ORF downstream of dfrG, designated orfU1, was located in the opposite direction of dfrG and consists of 1,950 bp encoding 650 amino acids, although the deduced amino acid sequence did not show any significant homology to sequences of other previously reported proteins. An ORF upstream of dfrG consisted of 582 nucleotides and was identical to the 3′-flanking region of the SAV0404 gene encoding a hypothetical protein (11). dfrG and orfU1 were flanked by a 28-bp inverted repeat and a 7-bp direct repeat, indicating that the region is in an insertion sequence (IS). The DNA sequence, ranging from 275 bp upstream of dfrG to the 363 bp of the 5′-terminal region of dfrG, was identical to that previously reported for plasmid pMG1 in Enterococcus faecium (18). The dfrG gene may have been acquired from E. faecium via IS-mediated recombination. The ancestral origin of S3DHFR, however, remains unknown; S3DHFR showed little similarity to and considerable phylogenetic distance from intrinsic DHFR of E. faecium (Fig. 2).

FIG. 1.

Multiple-sequence alignment of the amino acid sequence of S3DHFR from S. aureus CM.S2 (IMCJ1454) isolate with those of DHFRs from other bacteria. The amino acid sequence of S3DHFR was compared with that of type S1 from S. aureus, S2 from S. haemolyticus MUR313, and the chromosomal DHFRs from B. anthracis Ames, B. cereus ATCC 10987, B. cereus ATCC 14579, Bacillus subtilis Marburg, E. faecalis V583, a methotrexate-resistant mutant of E. faecium strain A, Streptococcus pneumoniae ATCC 49619, Staphylococcus epidermidis ATCC 14900 (SeDHFR), S. aureus ATCC 25923 (SaDHFR), and E. coli K-12. Sequence comparison was performed by aligning the proteins with the ClustalW program (http://www.ddbj.nig.ac.jp/E-mail/clustalw-e.html). Amino acid positions involved in the binding of trimethoprim (T) and NADPH cofactor (N) are according to studies of the E. coli K-12 enzyme (3, 9, 13, 14). Identical residues are indicated by white letters on black background. Gaps introduced to maximize alignment are indicated by dashes.

FIG. 2.

Dendrogram of S3DHFR and DHFR from a variety of organisms. The dendrogram was created by the ClustalW program. Branch lengths correspond to the number of amino acid exchanges of the DHFR proteins (accession number and species given in parentheses) of types I (X00926, from E. coli), Ib (I40985, from E. coli), IV (A60935, from E. coli), V (X12868, from enterobacterial plasmid pLMO150), VI (Z86002, from Proteus mirabilis), VII (X58425, from E. coli), VIII (U10186, from E. coli), IX (A49788, from E. coli), X (AY123253, from Klebsiella pneumoniae), XII (I41043, from E. coli), E1 (AF028812, from E. faecalis), S1 (X13290, from S. aureus), and S2 (Z50141, from S. haemolyticus MUR313) and the chromosomal DHFRs of B. anthracis Ames (AE017031), B. cereus ATCC 14579 (AE017005), B. cereus ATCC 10987 (AE017271), B. subtilis Marburg (L77246), Enterobacter aerogenes (M26022), E. coli K-12 (P00379), E. faecalis V583 (AE016951), E. faecium mutant strain A (741860A), Haemophilus influenzae R1047 (X84205), Salmonella enterica serovar Paratyphi ATCC 9150 (CP000026), S. aureus ATCC 25923 (SaDHFR; Z16422), S. epidermidis ATCC 14900 (SeDHFR; Z48233), Streptococcus pneumoniae ATCC 49619 (Z74778), and Vibrio vulnificus YJ016 (BA000037).

The MICs of TMP in E. coli transformants harboring pSA1 or pT7dfrG carrying dfrG were significantly increased than those in control strains (Table 1), indicating that dfrG is responsible for TMP resistance. An E. coli transformant harboring pT7dfrB carrying dfrB also showed increased MIC, but it was not as high as those of E. coli strains expressing dfrG. dfrB is believed to encode a TMP-sensitive DHFR of S. aureus because it was found in all S. aureus strains, regardless of TMP susceptibility. Similar results were reported for dfrE encoding Enterococcus faecalis DHFR (4). The increased MIC for TMP in E. coli carrying dfrB may be explained by the multicopy effects of high expression of the housekeeping protein DHFR.

For functional analysis of S3DHFR and DHFR from S. aureus CM.S2 (SaDHFR_CM.S2), overexpression and purification of these recombinant DHFRs were achieved. Overexpression of S3DHFR or SaDHFR_CM.S2 was accomplished by integration of the respective coding regions downstream of the His-tagged coding region of the pCR/T7NT expression vector and transformation into the E. coli strain BL21-AI. Recombinant protein in soluble extracts was purified by affinity chromatography to determine enzymatic activities. The K_m values of recombinant S3DHFR for DHF and NADPH were 2.68 ± 1.09 μM and 2.38 ± 1.97 μM, respectively (Table 2). The K_m values of DHF and NADPH for S3DHFR did not differ from those of SaDHFR_CM.S2, but the IC₅₀ values of TMP for these DHFRs differed significantly. The IC₅₀ of TMP for S3DHFR was more than 8,000-fold greater than IC₅₀ values for TMP-sensitive SaDHFR and SaDHFR_CM.S2, indicating that S3DHFR and SaDHFR_CM.S2 are indeed DHFRs but that only S3DHFR plays a critical role in TMP resistance. The K_m values of crude extracts for NADPH were sixfold greater than those of recombinant S3DHFR (Table 2). Crude extracts may contain other factor(s) that bind to NADPH.

Detection of dfrG was performed by PCR on isolates from Chiang Mai, Thailand, and Tokyo, Japan. All Chiang Mai isolates were resistant to TMP and contained dfrG, whereas all Tokyo isolates but one were sensitive to TMP and did not contain dfrG (data not shown). The single Tokyo isolate IMCJ934 was resistant to TMP and contained dfrG (Table 1).

Pulsed-field gel electrophoresis (PFGE) analysis revealed 13 patterns of SmaI digestion in the 43 MRSA isolates from Chiang Mai, Thailand (data not shown). Cluster analysis showed that 12 of the 13 PFGE patterns formed a cluster (>75% similarity). The PFGE pattern of S. aureus CM.S2 genomic DNA was identical to that of 18 MRSA isolates. These results suggest that clonal expansion of MRSA carrying dfrG occurred at the hospital in Chiang Mai. The TMP-resistant isolate from Tokyo, Japan, IMCJ934, showed the same PFGE pattern as that of one of the Chiang Mai isolates, S. aureus CM.S2 (data not shown).

dfrG was detected by Southern blotting on fragments of SmaI-digested genomic DNA, but it was not detected on plasmids (data not shown). Conjugal transfer of TMP resistance from S. aureus CM.S2 to recipient strains S. aureus IMCJ565RFP^r or IMCJ644RFP^r was unsuccessful, suggesting that dfrG is located on the chromosome and not on a plasmid of these clinical isolates. It remains to be determined whether dfrG can be transferred by phages or mobile elements.

A single amino acid substitution (Phe to Tyr) at codon 98 of SaDHFR was reported to be associated with TMP resistance in S. aureus (5). Therefore, approximately 390 bp of internal DNA sequence of dfrB encoding SaDHFR was determined. When S. aureus ATCC 29213 was used as a control (5), all isolates from Chiang Mai, Thailand, exhibited three silent mutations: CAT to CAC in codon 77 and TTT to TTC in codons 91 and 118. All isolates from Tokyo, Japan, contained four silent mutations: AAA to AAG in codon 30, CAT to CAC in codon 77, and TTT to TTC in codons 91 and 118. These results indicate that these mutational changes are not associated with TMP resistance in the isolates from Chiang Mai or Tokyo. Other possible mechanisms of TMP resistance, such as overexpression of intrinsic DHFR, efflux, or impermeability, may be involved.

The CM.S2 strain was the dominant clone from Chiang Mai, Thailand. MRSA surveillance is being carried out in the hospital from which these isolates were obtained. S. aureus CM.S2 is resistant to clindamycin, erythromycin, gentamicin, and tetracycline and is less sensitive to arbekacin. Fosfomycin, linezolid, and vancomycin are effective in vitro; quinupristin-dalfopristin and daptomycin were not available for testing. Results of this surveillance will be reported in the future.

Our data strongly suggest that the TMP resistance-associated gene dfrG is prevalent in Thailand, and an isolate harboring this gene was found in Japan. This gene may spread worldwide, and measures against this, such as gene monitoring and adequate use of TMP, should be established.