This study expands our earlier work with por VR typing of culture samples by applying this method to noncultured clinical samples. Adapting the method for use with direct clinical samples allowed us to investigate the possible occurrence of coinfection with multiple gonococcal strains in more detail than previously reported as well as to examine the range of por types identified in two very divergent populations.
The use of NAAT in conjunction with the ability to characterize strains from direct clinical samples may have an important impact on the study of N. gonorrhoeae
. Turner et al. (28
) estimated that the prevalence of undiagnosed gonococcal infections in a young adult population in Baltimore was 5.3% by using NAAT, much higher than the 2.6% prevalence estimate based on diagnosed and reported infections in the same population. In Australia, a study using NAAT found that the estimated number of women infected was over four times the number of officially reported cases (3
). The characteristics of isolates causing infections identified by NAAT have not been routinely examined, and epidemiologic studies utilizing NAAT will be enhanced by the use of methods that provide molecular characterization of strains.
In addition to potential application with current NAAT, por
VR typing is based on characterization of a single-copy gene and is therefore able to distinguish single, multiple, or hybrid por
types present in a single sample. Our previous data demonstrated that the oligonucleotide probes do not cross-react with sequences that differ by more than 2 bp and are highly specific for the sequences found in the variable regions of porB
). Sequence analysis of neisserial porB
genes has shown that while genetic exchange occurs between genes, exchange does not occur between VRs within a single porB
). For example, sequences found in VR1 are unique to that VR and are not found in any other VR in the gene. Given the specificity of the oligonucleotide probes, hybridization of distinctly different probes for the same VR region strongly suggests that DNA from more than one strain was present in the sample. Mixed gonococcal infections may be significant in many aspects of gonococcal research, including studies of natural immunity, transmission, spread of antimicrobial resistance, and genetic diversification. The extensive mechanisms present in N. gonorrhoeae
for genetic exchange and recombination have been recently reviewed (13
). Since the only ecologic niche occupied by gonococci is the human host, mixed infections are central to the evolution of this pathogen.
Typing directly from clinical specimens identified multiple por
types from a remarkable 40% of the women in the Baltimore study and from 22% of specimens from SWs from Madagascar. While the specimens that were used for this exploratory study do not allow estimates of the prevalence of mixed infections in any population, the data suggest that mixed infections are far more common than previously appreciated. Knapp et al. had previously reported a rate of approximately 21% when isolating N. gonorrhoeae
from multiple sites in men and women and identifying different auxotype/serovar classes from each site (14
). However, the identification of multiple auxotype/serovar types in a single mixed isolate would be difficult, as the serovar system uses a panel of monoclonal antibodies and unusual binding patterns may be due to either usual serovars (such as those seen in hybrid or rare porin types) or mixed infections. Martin and Ison (18
) recently raised this issue by comparing the opa
type of direct urethral specimens with the opa
type of the primary culture. In that study, 19 samples (14 from men and 5 from women) were examined, and four men (21%) had evidence of a mixed gonococcal infection. Those authors did not detect multiple strains in cultures, even by typing multiple individual colonies of the primary isolation plate. In our study, DNA purified from stored culture samples also did not contain evidence of multiple strain types. We did not attempt to identify por
types of single colonies grown from the culture samples, as we avoided possible secondary selection by purifying DNA directly from an aliquot of the stock prepared from the primary culture plate. Since amplification and identification of one porB
in the presence of a 100-fold excess of another porB
was possible, this approach should be more sensitive than screening several hundred colonies.
The discrepancy between the ability to detect mixed infections in stored cultures and the ability to detect mixed infections in stored clinical specimens could have several plausible explanations. Nonviable and live gonococci would be detected by DNA amplification from clinical samples, whereas only viable gonococci would be recovered by culture. Differences in in vitro growth kinetics might bias detection towards more robust strains that grow well on standard culture media. Even if two strains were present on a culture plate, they would not necessarily be distinguishable by colony morphology, and multiple single-colony analyses would be necessary to identify and characterize both strains. So although not a factor in this study, isolation of a single colony to characterize an infecting strain may have underestimated mixed infections in previous studies. The use of NAAT diagnostics combined with molecular methods of strain characterization may greatly improve our understanding of the frequency and consequences of mixed gonococcal infections.
Although comparisons of the diversity of por
types between Baltimore and Madagascar are limited by the differences in study design, some general observations can be made. A high proportion of individuals for whom typing data were obtained had a por
type common to both geographic areas. Our earlier study of Baltimore isolates collected over 10 years (19
) suggested that approximately 10 to 33% of cultures were PIA, which is consistent with the findings of the present study as well as those of others (14
). The unusually high number of PIA strains in the Madagascar SWs, in which 66% of women had at least one PIA strain, may be a consequence of factors including the number of exposures, general physical condition, and immune status. The differential ability of PIA compared to PIB strains to be isolated outside the genital tract, survive in the harsh environment of urine, and remain intact during storage and handling may affect the detection of different por
types. Additionally, we were not able to amplify and type all of the samples that were LCR positive, and we cannot exclude the possibility of sample bias affecting the relative prevalence of PIA- compared to PIB-expressing strains observed in this study. However, given that more serious, invasive, and disseminated disease has been associated with strains expressing PIA, the predominance of PIA in this population deserves further investigation.
The PCR primers used in this study were specifically designed to amplify both PIA and PIB alleles of porB
. Due to the conserved nature of the of neisserial porins, the primers also amplified DNA from some commensal Neisseria
species. Other NAAT systems also detect nongonococcal genes (8
). Our data suggest, however, that gonococcal porB
probes do not hybridize to porin DNA amplified from commensal Neisseria
, allowing accurate characterization of the gonococcal porin even in the presence of commensal Neisseria
We were unable to amplify porB
from many of the LCR-positive urine samples. Experimentally, we demonstrated the loss of an order of magnitude in sensitivity associated with a single, brief freeze-thaw cycle of whole urine, and additional losses may be associated with extended storage time. The Abbott GC LCR targeted the multicopy gene opa
rather than the single-copy gene porB
targeted by por
VR typing. LCR may be more sensitive than PCR; no PCR-based NAAT is widely available for the detection of gonococci in urine from women. por
VR typing is not designed to be a diagnostic test per se but rather a method that can be used in conjunction with NAAT to characterize strains in individuals who have been identified as having gonococcal infections. Several studies have shown that endocervical and vaginal samples are more sensitive than urine for detecting gonococci in women (6
), and the use of self-collected vaginal swabs might be a viable alternative (17
). In studies where urine samples are collected, centrifuging the samples and storing the pellets alone may improve subsequent genotypic analyses.
In these studies, we demonstrated the utility of molecular characterization of the gonococcal outer membrane Por protein from non-culture-based clinical samples. We have also shown that evidence of infections with multiple strains can be easily identified in direct clinical specimens. Additionally, our data support our earlier findings that some por types and some VR types are common among geographically diverse populations. The combination of a molecular typing method with NAAT will enhance our understanding of the molecular epidemiology of N. gonorrhoeae. Our understanding of the significance of gonococcal por type and mixed infections in studies of protective immunity, strain transmission, reinfection, and antimicrobial resistance will be substantially enhanced by using direct clinical specimens and methods that provide specific and accurate information about porB.