High-Resolution Genotyping of Chlamydia trachomatis from Recurrent Urogenital Infections

Abstract

A method for Chlamydia trachomatis restriction fragment length polymorphism (RFLP) analysis and complete sequencing of omp1 was developed for use on samples collected at home, and results were compared. Genotyping by sequencing was superior to RFLP analysis. The omp1 gene in 31 clinical strains harbored few mutations compared to the same gene in ATCC reference strains. Follow-up samples obtained during a 24-week period from 31 patients showed recurrence with the same genotype in five cases and a new genotype in one case.

Sexually transmitted infection with Chlamydia trachomatis is the most common treatable urogenital infection in young adults (3). In a recent study (5) recurrent infections occurred in 29% of patients within a follow-up period of 24 weeks. The mechanism of recurrent infections is not known. It could be due to relapse after failed antibiotic treatment, reinfection from an untreated partner, or new infection after successful treatment.

The aim of this study was to establish a method where C. trachomatis could be differentiated at the clonal level in urine and vaginal flush samples collected at home, since such a test strategy dramatically increases the number of C. trachomatis infections identified (1, 11, 12) compared to the number of infections identified with conventional swab sampling. Thus, urine and vaginal flush samples collected at home may be the sampling method of the future. By considering C. trachomatis at the clonal level it might be determined whether recurrent infections are caused by relapse or reinfection or by new infection. The complete omp1 gene of C. trachomatis was studied by using restriction fragment length polymorphism (RFLP) analysis and sequencing. The theory was that RFLP analysis and/or sequencing could be applied to distinguish between relapse or reinfection and new infection if the genotype changed during follow-up. In these cases a new infection was considered to have occurred. In cases of unchanged genotypes the omp1 gene, and its variable domains in particular, possibly could have accumulated mutations during the course of infection. By mapping these mutations, the relatedness of the C. trachomatis strains could be determined. We here present the first epidemiological study on genotyping by RFLP analysis and sequencing of the entire omp1 gene using urine (females and males) and vaginal flush samples (females; 0.9% NaCl) collected at home and mailed to the laboratory.

Forty-two of 141 patients (30 females and 12 males; mean age, 22.3 years) with C. trachomatis infections diagnosed between 30 July and 16 December 1997 participated in the study (5). In Denmark more than 95% of all testing for C. trachomatis is done in general practice. Therefore, all patients in this study were recruited from general practice. Baseline samples (week 0, just before antibiotic treatment) and follow-up samples were taken by the patients themselves in weeks 2, 4, 8, 12, and 24 and were sent to the Department of Clinical Microbiology, Herning County Hospital, by ordinary mail. The omp1 gene was amplified by PCR using nested primers (6) in a total volume of 50 μl that contained 5 μl of urine or vaginal flush sample, 100 pmol of each primer, 50 mM KCl, 10 mM Tris HCl (pH 8.3), 1.6 mM deoxynucleoside triphosphate (PE Biosystems), 5 μl of Amplitaq Gold (PE Biosystems), and 2 mM MgCl₂. The PCR program was 95°C for 6 min and 40 cycles of 95°C for 1 min, 60°C for 1 min, and 72°C for 1 min 45 s, followed by an extension period at 72°C for 10 min. PCR products were genotyped by RFLP analysis and sequenced. All controls were American Type Culture Collection (ATCC) reference strains, which were subjected to two independent rounds of PCR, RFLP analysis, and bidirectional sequencing.

For the RFLP analysis, PCR products were digested with restriction enzymes (7) and analyzed on a precast 10% Tris-borate-EDTA gel (Bio-Rad). References for the RFLP analysis were obtained by digestion of PCR products from the ATCC strains.

Sequencing of the omp1 gene was carried out on an ABI PRISM 310 genetic analyzer (PE Biosystems) using a BigDye DNA sequencing kit (PE Biosystems) according to the manufacturer's instructions. Three sense and three antisense primers were required to cover the approximately 1.1-kb DNA fragment: 1s, TCCTTGCAAGCTCTGCCTGTGGGGAATCCT; 1as, CCGCAAGATTTTCTAGATTTC; 2s, CARAATACATCAAARCGAT; 2as, TATYTGGGATCGYTT; 3s, TTGAGCRTATTGGAAWGAA; and 3as, CCTAAARTMGAAGARTT. In order to cover the nucleotide sequence of all genotypes of C. trachomatis, primers 2s, 3s, 2as, and 3as were constructed with degenerate bases (R, A+G; Y, C+T; M, A+C; W, A+T). Clinical samples were sequenced once, and samples differing from the relevant ATCC reference omp1 sequence were reanalyzed by using a second PCR preparation to rule out the possibility of Taq errors.

Of the 42 patients initially diagnosed with C. trachomatis infection by enzyme immunoassay (Syva, San Jose, Calif.) and confirmed by ligase chain reaction (LCR) (LCx; Abbott Diagnostics, Chicago, Ill.), 36 were positive at baseline (week 0) and 6 were negative. The reason for the six negative samples may be that the infection was spontaneously eradicated in the period between initial testing (screening) and the baseline sample, or that the initial tests were performed on swab samples and the baseline tests were performed on urine and vaginal flush samples. The sensitivity of PCR and LCR amplification is lower for urine samples than for swab and vaginal flush samples, which are equally sensitive (12). Furthermore, the six negative samples at baseline could be due to hormonal factors (9), specimen variation, or difference in the amount of C. trachomatis shedded depending on whether the infection was in an ascending or a descending phase. Yet we chose vaginal flush and urine because home sampling greatly increases the number of samples obtained from potentially infected patients, and thereby more infections are identified. This is especially true for men, where the number of persons screened in a study increased by a factor of 100 (11).

At baseline, samples from 31 of the 36 patients with C. trachomatis infections were successfully amplified by PCR targeting the omp1 gene. During follow-up, three of seven C. trachomatis-positive samples in week 2, none of six in week 4, one of three in week 8, three of six in week 12, and four of six in week 24 could be amplified by omp1 PCR. Overall, 25 of 53 urine samples (47%) and 31 of 38 vaginal flush samples (82%) could be amplified by the omp1 PCR, indicating superior sensitivity of vaginal flush specimens compared to urine specimens. Only 7 vaginal flush samples out of 38 could not be omp1 PCR amplified. The reason for unsuccessful omp1 PCR amplification of some samples may be the lower sensitivity of the omp1 PCR compared to the plasmid-directed LCR, which is explained by the fact that the omp1 gene is present in one copy only, whereas the plasmid of C. trachomatis is present in approximately 10 copies (8). Differences in the nucleotide sequences of the individual genotypes may lead to differences in PCR performance, i.e., some genotypes may be easier to amplify by PCR (primer preference) than others. This may also lead to inaccurate reports of genotype distribution and may cause mixed infections to be overlooked (4). However, all our sequences were clean and showed no sign of mixed infections. This may be because all of the patients were recruited from general practice and not from sexually transmitted disease clinics. The results obtained by genotyping by RFLP analysis and by sequencing are shown are Table 1. The discrepancies between the genotyping results obtained by the two different methods are notable, in particular an apparently systematic error in the RFLP analysis, where 14 samples of genotype F were mistaken for genotype C and 4 samples could not be identified. In one case (P027), genotype G was mistaken for genotype C, and in another (P032), genotype J was mistaken for genotype H. Finally, genotype K (5 samples) could not be identified by means of RFLP analysis. Genotyping by RFLP analysis alone has often been applied (2, 7, 10, 13–15), but the validity of RFLP analysis results in our study might be questioned in light of the poor concordance with the results obtained by sequencing, which studies the genotype at the single nucleotide level and therefore may be considered the reference standard.

TABLE 1.

Comparison of genotypes determined by RFLP analysis and sequencing

Patient		Genotype ata:
		Baseline		Wk 2	Wk 8	Wk 12	Wk 24
Sex	No.	RFLP	Sequencing
Female	P006	E/E	E/E	−	−	ND	−
	P010	E/E	E/E	+/E, +/E	−	ND	ND
	P015	C/C	F/F	u/u, F/F	ND	ND	ND
	P018	E/E	E/E	ND	−	−	ND
	P022	C/C	F/F	−	+/C, +/F	C/C, F/F	−
	P026	E/E	E/E	−	−	C/+, F/+	−
	P027	+/C	+/G	−	−	−	−
	P031	+/E	+/E	−	−	−	ND
	P032	−/H	−/J	−	−	−	−
	P040	E/−	E/−	−	+/−	+/+	−
	P041	+/C	+/F	−	−	−	+/+
	P048	E/E	E/E	−	−	−	−
	P062	u/u	K/K	−	−	−	−
	P069	E/E	E/E	+/+	−	−	−
	P070	+/E	+/E	−	−	−	−
	P072	E/E	E/E	−	−	−	E/E, E/E
	P075	+/E	+/E	−	ND	ND	ND
	P078	u/u	K/K	−	−	−	−
	P080	C/C	F/F	−	−	−	−
	P081	+/C	+/F	−	−	−	−
	P083	C/C	F/F	−	−	−	u/u, F/F
	P086	+/E	+/E	−	−	−	−
	P113	D/D	D/D	−	−	−	−
	P124	+/C	+/F	−	−	−	−
	P129	+/D	+/D	−	−	−	−
	P131	+/E	+/E	−	ND	ND	ND
Male	P016	E	E	−	−	−	−
	P043	C	F	−	−	−	−
	P057	D	D	−	ND	ND	−
	P071	u	K	−	−	−	−
	P094	E	E	−	−	−	ND
	P105b			+	−	−	−

^a−, negative by LCR; +, positive by LCR but not amplifiable by omp1 PCR; u, unidentified; ND, not done. Results with a slash show result for urine sample/result for vaginal flush sample; where two results are given, the first is the RFLP analysis result and the second is the sequencing result. 

^bThis patient was C. trachomatis negative by LCR at baseline and C. trachomatis positive in week 2. 

Only in one of six cases (16.7%) was a new genotype found during the follow-up period, and thus only in this case is it certain that a new infection occurred. Five infections (83.3%) were possible relapses or reinfections, and thus the data suggest that treatment failure and efficient partner treatment should be addressed.

Mutations in the omp1 gene for the baseline clinical specimens compared to the same gene in the reference ATCC strains are shown in Table 2. No mutations were found in the omp1 PCR-amplifiable samples obtained during follow-up. One or more (13 in total) mutations were found in samples from 8 of the 31 patients. In all cases where both urine and vaginal flush samples were provided, the mutations in both sample types were in concordance. This illustrates the reliability and reproducibility of sequencing. All type K samples harbored the same mutation in variable domain IV (VD IV) (2, 16) (residue 925, A→G), and all other mutations were private. The six mutations in the variable domains were all in VD IV, and the remaining seven mutations were found in constant domains. The small number of changes in the clinical samples compared to the nucleotide sequences of the ATCC reference strains suggests a high degree of conservation, even of the variable domains of the omp1 gene. This may reflect that the major outer membrane protein is crucial to the survival of the organism.

TABLE 2.

Mutations in clinical samples at baseline compared with ATCC reference strains

Patient		Genotype	Mutation in samplea		Amino acid change
Sex	No.		Urine	Vaginal flush
Female	P027	G	ND	174: T→A	Silent
				646: G→C	E→Q
				949: T→Ab	S→T
	P048	E	366: A→G	366: A→G	Silent
			940: A→G	940: A→Gb	S→G
	P062	K	925: A→G	925: A→Gb	T→A
	P078	K	925: A→G	925: A→Gb	T→A
	P075	E	ND	880: G→Ab	A→T
	P113	Dc	100: A→G	100: A→G	T→A
	P131	E	ND	189: A→G	Silent
				192: C→T	Silent
				193: C→G	Q→E
Male	P071	K	925: A→Gb		T→A

^aNucleotide position and substitution. ND, not determined. 

^bMutation found in VD IV (VDs are identified in reference 16). 

^cThis genotype D differed from the ATCC reference genotype at 10 base positions (77, 100, 130, 132, 141, 169, 174, 192, 195, and 582). Nine of these were in total agreement with the sequence of genotype D6 (National Center for Biotechnology Information database, accession no. AF086855); position 100 showed an additional base change. 

We conclude that PCR amplification and direct sequencing of the omp1 gene of samples collected at home could be performed for 31 of 36 (86%) infected patients. Sequencing has a high reproducibility, and in contrast to RFLP analysis it is a highly reliable epidemiological tool that also could be applied to urine and vaginal flush samples collected at home. Sequencing, however, is laborious and expensive, and the information obtained by studying omp1 does not justify routine use. The limited number of mutations we have observed in the omp1 gene of our 31 clinical strains compared to the same gene in the relevant ATCC reference strains (which were established some 30 years ago) implies that other target genes should be studied for the purpose of individual strain identification. We are presently engaged in this work.