In this thesis, using the conserved 16S rRNA sequences as PCR primers, IC patient’s bladder biopsies were analyzed for the presence of bacterial 16S ribosome DNA. Sequence data were analyzed by utilizing two computer databases to determine the species of bacteria. Two databases were searched and similarity was found for some species of bacteria. It appeared that Pseudomonas and Escherichia coli could be possible etiologic agents of IC. However, contaminants can not be excluded and the possibility of the other bacteria should be considered. Furthermore, we should consider several questions about the different results from two computer databases. The 16S rRNA primers used in this study might not PCR amplify all organisms from IC patients’ biopsies. The true etiologic agent might not show on our lists. Another issue that should be addressed is whether there are differences in fidelity based on direct sequencing of PCR products or cloning prior to sequencing. The reproducibility of sequencing, heterogeneity of genes and properties of computer databases also needed to be considered.

Diverse sequences’searches are illustrated in Table 13 and the dendrogram (Fig. 12). The sequences of clone 129 and 131 (P. aeruginosa and E. coli) have the closest relationship in the dendrogram (Fig. 12). This result coincides with the studies of Olsen et al. (1993). It is also similar to the result of Pettersson et al. (1994) that MG (Mycoplasma) is distantly related to clone 131 (E. coli). In contrast, clones P5 and 63 appeared to be derived from Bradyrhizobium japonicum, but their dendrogram analysis appears distant. The sequences of seven clones (clones 83-35, 135-17, 135-19, 175-2, 175-4 and clones 182-1, 182-5) matched the 16S rRNA sequences of Acidiphilium facilis, based on both computer database searches. Five of the clones (83-35, 135-17, 135-19, 175-2, 175-4) appeared to be closely related, but were only distantly related to clones 182-1 and 182-5 on the dendrogram. This diversity points out that some problems existed in these sequences.

Why were the two computer searches not coincident? Why did the dendrogram illustrate a non-significant sequence relationship? There are several possible reasons for these results. First, sequence conservation is important in deciding which region is most likely to be successful for primer construction. Are the 16S rRNA primers used in these studies universal? The 16S rRNA primer might not amplify all organisms from IC patient’ biopsies. For instance, the 16S rRNA universal primers can not amplify Mycoplasma genitalium rRNA (MG) for unknown reasons. Primers from 23S rRNA sequences are also very conserved gene among various organisms and can be used to detect unknown agents (Gutell et al. 1994). Recently, Gurtler et al. (1996) has suggested that the flanking 16S-23S spacer region may be a better conserved region to serve in primer recognition. The computer searches would show the irrelevant species of bacteria, if the true organisms were missed by PCR using 16S rRNA primers.

Second, contaminant agents could also be amplified by this 16S rRNA primers and appear on the lists at the same time. For this reason, it is very difficult to decide which one is the true etiologic agent of IC. Several specific 16S rRNA primers were designed to detect high frequency contaminant organisms. To understand whether these biopsies contained contaminant organisms, the same primers could be used to PCR amplification of new biopsies from the same IC patients in the future.

Third, IC could be caused by multiple agents or superinfection. The computer search lists could show all agents or just one of them. The results differ according to the sites of biopsies and timing of pathogenesis. Bacterial postinfection is another possibility. Finally, the etiologic agent of IC could be a new species that does not exist in current computer databases.

In this work, DNA amplified from patients’ biopsies were cloned. Cloning could accumulate mutations in each step. Finally, we would lose the original sequence from patients’ biopsies. Therefore, an alternative method to derive sequence data involves direct sequencing of PCR products. Direct sequencing can reveal the sqeuence of the starting molecule accurately. Mutations introduced during amplification are not apparent from direct sequencing because each mutation is present in a minority of the reaction products. Cloning, however, selects a single amplified molecule and any mutations present are manifested in the sequence of that clone. Experimentally, cloning can be advantageous because it can preserve the PCR products from a limited amount of biopsies. Also, heterogeneous DNA can be stored in vectors and is easy to access after originally taking patients’samples. Therefore, PCR products were cloned into E. coli expression vectors (pUC18) in this thesis. This is a possible explanation for some of my various results. Next, the sequencing strategy is another key point to cause the differences in sequences.

Each clone was sequenced in two directions. Due to limits of the sequencing technique and various GC contents, some clones could not be sequenced in one direction completely by only one primer. Sequences of each clone contain a 5’ region, an overlapped region and a 3’ region. Only the overlapped region was sequenced twice, but sequences of all regions were sent to two computer databases. If we just sent the overlapped region, would the search results be changed? To test it, overlapped sequences of each clone were sent to NCBI database alone. The results of these searches were very similar to the previous searches except for clone 63. For clone 63, two new species have higher similarity than the original top one species (Table 14). Generally, the searches of overlapped regions are insignificantly different from that of all regions. However, each clone should be sequenced twice to confirm the fidelity of the derived sequences and prevent errors from reading sequences. Furthermore, an internal sequencing primer is required extend the sequencing date. Nevertheless, diversity of 16S rRNA gene per se could be another reason for these various results.

Several same species names appeared in the computer search list repeatedly. Presumably, they have different gene numbers because the sequences of same gene come from different sources (locations) and are submitted by different laboratories. It could cause the heterogeneity of 16S rRNA gene. Different sources (locations) would have natural mutation in the same species bacteria and accuracy of sequencing techniques could be different between laboratories. It is noteworthy that our control DNA sequences did not display 100% identity to sequences from computer databases. This could be due to sequencing errors in the current work or the differences in two computer databases per se.

Two databases (NCBI and RDP) that were used in these studies are designed for different purposes. First, the capacity of databases are different. The RDP consists of about 3000 aligned small subunit (SSU) and large subunit (LSU) ribosomal RNA sequences which were retrieved from GenBank and EMBL; prokaryotic sequences predominate. On the other hand, NCBI includes Brookhaven Protein Data Bank, GenBank Release 87.0, GenBank cumulative daily updates to the major release, EMBL Data Library and Release 41.0, EMBL Data Library cumulative daily updates and DDBJ (Databases of Japan). Therefore, NCBI has a wider range and more complete gene database than that of RDP, but RDP has higher specificity to rRNA genes than that of NCBI.

Second, two databases have different statistical strategies for sequences alignment. In the NCBI, the fundamental unit of BLAST algorithm output is the High-Scoring Segment Pair (HSP). An HSP consists of two sequence fragments of arbitrary but equal length whose alignment is locally maximal and for which the alignment score meets or exceeds a threshold or cutoff score. Each HSP consists of a segment from the query sequence (our unknown sequence) and one from a database sequence. The BLAST program first looks for similar segments (HSPs) between the query sequence and a database sequence. Then BLAST evaluates the statistical significance of any matches that satisfy setup threshold of significance. Findings of multiple HSPs involving the query sequence and a single database sequence will be treated statistically in a variety of ways. By default the programs use "Sum" statistics (Karlin et al. 1993), the statistical significance ascribed to a set of HSPs may be higher than that ascribed to any individual member of the set. Only when the ascribed significance satisfies the setup threshold (E parameter) will the match be reported. The parameter E establishes a statistical significance threshold for reporting database sequence matches. E is interpreted as the upper bound on the expected frequency of chance occurrence of an HSP (or set of HSPs) within the context of the entire database search. Any database sequence whose matching satisfies E is reported in the program output. If the query sequence and database sequences follow the random sequence model of Karlin and Altschul, and if sufficiently sensitive BLAST algorithm parameters are used, then E may be thought of as the number of matches one expects to observe by chance alone during the database search. In the RDP, the search algorithm compiles ribosomal sequences and related data and re-distributes them in aligned and phylogenetically ordered form. So it is specific to rRNA rather than all genes. It provides ribosomal probe checking, approximate phylogenetic placement of user-submitted sequences, screening for chimeric nature of newly sequenced rRNA, and automated alignment. MG is a clone for an attachment protein of Mycoplasma; this is very likely the reason that the blind control did not appear on the top of list.

Although all of these bacteria could be etiologic agents of IC, to identify a right one is the main goal of the thesis. How do we choose possible pathogens of IC from the different results in two computer database searches? We analyzed this question in several ways.

First, for each clone some common but differently ranked species were displayed from two databases (Table 11). These common bacteria, which were searched by both computer databases, might have higher possibility to cause IC. Second, we searched the related literature and published cases of reported infections of humans caused occasionally by the bacteria on the lists. This literature search may give us some instances into opportunistic infection especially from UTI-related diseases. At the same time, some of well-known virulence factors and pathogenesis of these bacteria might be found to involve in IC symptoms.

Third, to analyze the possible pathogenesis which could match characteristics of bacteria on our list, the species of possible etiology can be narrowed down. IC is a recurrent UTI disease and most patients suffer from recurrent IC for many years. There are many experimental treatments, no definitive cure has been found. IC could probably share some common characteristics of pathogenesis and virulent factors of UTI that are currently known. Bacterial UTI has two features in common: they arise by the ascending route from organisms originating in the urinary tract and a central phase in the infection is significant bacteriuria which may give rise to pyelonephritis. Pathogenic bacteria possess the means of ascending the urethra to reach the bladder and ultimately the kidney. These uropathogens survive due to the presence of virulence to resist the normally efficient host defense mechanisms: a system of continuous urinary flow including the powerful effect of micturition, a rapid turnover of epithelial cells of mucosal surfaces, a specialized local immune system, and a mucus coating. All of mechanisms maintain sterility of the urinary tract (Reid et al. 1987). To persist, survive and maintain their presence on the mucosal surface, adherence is a prerequisite for colonization and infection in the urinary tract. The pathogenesis of infections on mucosal surfaces involves a number of steps: the attachment to the epithelium, colonization, tissue damage, and in some cases invasion and dissemination. Each stage is dependent both on bacterial virulence properties and on host factors.

After literature search and pathogenesis analysis, E. coli and Pseudomonas spp. have characteristic virulence factors to cause UTI and probably these factors also can cause IC. Therefore, E. coli (clone 131) and Pseudomonas spp. (clones P5, 68, 129) are most likely to be the etiologic agents of IC among bacteria on the computer search lists We will discuss each of them respectively in later sections. Here, we discuss the bacterial species that were found in the database searches from the other clones.

Clone 63 is highly related to Bradyrhizobium japonicum, the soybean symbiont. It is a legume root and stem nodule bacterium. They can live either free in the soil and in laboratory culture or endosymbiotically in infected host cells of the central nodule tissue, where they are capable of fixing N2. Bradyrhizobium japonicum, as far as the biochemistry and molecular genetics of respiration are concerned, is best-studied rhizobial species (Preisig et al. 1994). One prominent feature of these nodules is the microaerobic environment. They utilize a high-affinity terminal oxidase, which is encoded by the fixNOPQ operon, to meet the high-energy demands of nitrogen fixation. The fixNOPQ operon is induced under microaerobic conditions and is essential for symbiosis ( Page et al. 1995). It is possible that standard clinical techniques can not culture this kind of nitrogen fixed bacteria under aerobic conditions and routine nutrition media.

Clones 83-35, 135-17, 135-19, 175-2, 175-4 and clones 182-1, 182-5 revealed high similarity to Acidiphilium facilis from both of computer databases. Acidiphilium bacteria are recovered from samples of water and sediment collected from acidic mine drainage streams. They are rod shaped, motile, mesophilic, Gram-negative and strictly aerobic, utilized citric acid and Tween 82 as sole carbon sources. They are unable to grow at or above pH 6.0. Their special nutrition and pH requirement beyond the routine clinical culture techniques and could not be cultured. Bradyrhizobium japonicum and Acidiphilium facilis have not been shown to cause any human infectious diseases to date.

The sequence of clone 131 is most similar to E. coli rrnH gene. E. coli is the most common organism isolated from all types of UTI. Most uropathogenic E. coli belong to a restricted range of serogroups. Strain K12 is one of these serogroups. It is noteworthy that our results showed E. coli K12 sequences in the DNA from IC patients’ biopsies.

Recognized E. coli virulence factors include adherence to uroepithelial cells; resistance to serum bactericidal activity; the type and quantity of K antigen and the production of hemolysin, siderophores, and colicin (Table 15). Capsular (K) antigens are important E. coli virulence factors. Only a limited range of these acidic polysaccharide antigens (K1, K2, K3, K12,) are found with significant frequency in UTI. Though there appears to be no difference in the frequency with which K antigens are found in cystitis and pyelonephritis, presence of K antigens favors kidney invasion in a mouse model particularly if they are present in large amounts. Similarly strains causing acute pyelonephritis contain more K antigen than those causing cystitis (Sussman M. 1985).

In addition to adherence to uroepithelial cells by K antigen of E. coli, adhesin is another factor. In a study on the adhesins of strains isolated from pregnant women with asymptotic bacteriuria it is reported that there is a significant association between infection with strains bearing mannose resistant (MR) adhesins and a past history of infection. The cells from the infection-prone women had a much higher tendency to bind P-fimbriae bacteria. It has been shown that other virulence determinants are likely to be required since pyelonephritis and cystitis strains appear to possess more virulence factors than asymptotic bacteriuria strains. Harber et al. (1982) reported that organism isolated from UTI possessed few adhesins on testing for haemagglutination after subculture and they found no evidence for fimbrial production by strains isolated directly from urine on examination under the electron microscope. This may be explained in part by the use of inappropriate culture conditions for expression of adhesins and by the selection of patients with urinary tract defects, a group shown to have a low percentage of adhesive isolates. Parry et al. (1985) have demonstrated the presence of MR adhesins on bacteria adhering to UEC of patients with cystitis by the use of immunofluorescence with fimbria-specific antisera. Adhesion could also play an important role in IC by contributing to the establishment of bacteria in the normal urinary tract. This is shown by 1) strong association of the adhesive property with urinary isolates compared with E. coli from the healthy bowel, 2) demonstration of adhesion to host cells in vitro, 3) demonstration of the production of MR adhesins in vivo, 4) presence of appropriate receptors on target host tissues and 5) demonstration of enhanced infectivity in vivo by strains bearing adhesins compared to those lacking adhesins (Parry et al. 1985). In our results, E. coli K12 is dominant species of clone 131 in two computer database searches. There is high possibility to be a etiologic agent of IC in this experiments. But E. coli appearance could be a contaminant result from patients who already have other latent UTI disease before this experiment.

Clone P5 is highly similar to Pseudomonas pickettii. Clone 68 is highly similar to Pseudomonas syringae. Clone 129 is highly related to Pseudomonas flavescens and Pseudomonas aeruginosa. Pseudomonas is a saprophyte of water, soil and other moist environments. Their ubiquity stems form their minimal nutritional requirements. Pseudomonas is an infrequent commensal of humans but can proliferate in the oropharynx and gastrointestinal tract of elderly, debilitated, or immunocompromised patients as well as in such sites as burns, ulcer and wounds. These sites are the usual portals of entry for organisms causing generalized infections. In immunocompromised patients, contamination in indwelling intravenous devices or urinary catheters has also been implicated (Raveh et al.1993). Pseudomonas species have been classified by Palleroni into five rRNA homology groups and 12 DNA homology subgroups. Most of Pseudomonds of medical importance

belong to RNA groups I and II. Pseudomonas flavescens (clone 125) and P. aeruginosa (clone 129) belong to P. flavescens group, RNA group I. Pseudomonas pickettii (clones P5) belongs to P. solanacearum group, RNA group II. Pseudomonas syringae (clone 68), belongs to Pseudomonas of plant and is not classified into these five groups (Table 16).

Pseudomonas pickettii (clone P5) is nonfermenting, a Gram-negative rod and rarely pathogenic in RNA group II. P. pickettii is widely distributed in nature, being a frequent contaminant in water supplies. It is increasingly identified as an opportunistic pathogen in nosocomial infections, especially among immunosuppressed patients. It is also a member of the Centers for Disease Control Va group of denitrifying nonfermentative Gram-negative bacilli. P. pickettii is an infrequent cause of bacteremia, meningitis, endocarditis, pneumonia, and osteromyelitis. It has also been implicated in common source nosocomial infection outbreaks due to the addition of contaminated water to parenteral fluids and to medical equipment presumed to be sterile. An Australia-wide outbreak of bacteremia is caused by P. pickettii contamination of a single batch of ampoules of parenteral water commercially produced for injection were used (Dimech et al. 1993). Raveh et al. (1993) described a series of four patients with permanent indwelling intravenous devices (PIIDs) who became infected with Pseudomonas pickettii. Wertheim et al. (1992) also investigated a case of vertebral osteomyelitis and intervertebral discitis caused by P. pickettii in a debilitated patient. P. pickettii may be a more invasive organism than previously noted, particularly in hosts with weakened immunity secondary to underlying disease. P. pickettii is usually resistant to aminoglycosides and colistin.

Pseudomonas syringae (clone 68) is a plant pathogen, a Gram-negative bacterium and a typical biotropic parasite that obtains its nutrients from living host cells. P. syringae encompasses a collection of host-specific strains that generally do not macerate host tissues, cause only delayed necrosis in their hosts, and secrete few if any pectic enzymes. Strains of P. syringae typically caused watersoaked lesion surrounded by chlorotic halos in the leaves and fruits of a limited range of hosts. The halos are caused by a variety of low molecular weight toxins which are not host-specific and generally are essential to pathogenicity, but which can contribute significantly to virulence (Collmer et al.). Until now, it is not to be reported as a human pathogen.

Pseudomonas flavescens (clone 125) B62 16S ribosomal RNA. Hildebrand et al. (1994) identified these strains as belonging to a new Pseudomonas rRNA group I species by an extensive nutritional characterization study, DNA-DNA hybridization and a comparison with 16S rRNA genes of other bacterial species. This strain is aerobic, Gram negative and oxidase positive. The cells are slightly curved rods (0.6-0.7 by 1.6 to 2.3 um) and are motile by means of a single polar flagellum. Previously, it is known to be a pathogen of plants; occurrence in humans has not been reported.

P. aeruginosa (clone 129) causes many serious infections such as bacteremia, endocarditis, pulmonary infection, ear infection, burn wound infections, urinary tract infections (UTI), gastroenteritis, eye infections, musculoskeletional infection. Most Pseuodomonas infections are opportunistic. Pseudomonas urinary tract infections are observed in patients with indwelling urinary catheters (Murray et al. 1990). Antimicrobial therapy for Pseudomonas infections is frustrating because the infected patient with compromised host defenses is unable to augment the antibiotic activity, and Pseudomonas are typically resistant to most antibiotics. It could be the possible reason for the IC patient’s recurrent pathogenesis.

In this thesis, a major factor for success was the application of magnetic separation techniques combined with nucleic acid technology. Though the magnetic particles are expensive, magnetic techniques might well be introduced into most diagnostic laboratories to reduce workload and time for a test result. Meanwhile, certain disadvantages, which limit the technique for diagnostic use, should be overcome or avoided.

First, the whole diagnostic protocol should be standardized because rapid and cost-effective identification is imperative in the clinical situation (Table 17). The methods used for the phylogenetic analysis of bacteria based on the 16S rRNA gene invariably involve the sequence determination of PCR products; this is both costly and time consuming. However, it is a powerful tool to augment the traditional techniques used by the clinical microbiologist. Further automation of the technique and preparation of clinical specimens will undoubtedly increase the use of universal PCR/sequencing for identifying the etiology in difficult cases involving probable bacterial infection. In addition, this technique is valuable for researchers interested in the pathology, natural history and epidemiology of bacterial infectious diseases (Anderson et al. 1994).

The next thing is to monitor the process of sample taking, delivery and prevent contaminants. How do we distinguish IC etiology from contaminant species? Several ways can can be used to determine the true pathogen of IC. First, the animal infection models can be designed to test which bacteria causes similar symptoms to human IC. These bacteria could be identical or closed-related species. These species are inoculated into animal bladder and observed whether their pathogenesis can meet the criteria of IC. Second, some typical IC patients could be treated with antibiotics that are supposed to inhibit or kill these expected pathogens. If patients are cured by specific antibiotics, that would support the interpretation that the particular organism is responsible. However, some antibiotic resistant strains may emerge.

Third, since we narrowed the possible agents of IC down to several species of bacteria, immunological studies may be designed to identify the true agents causing IC. The IgA from patients’ bladders or urine can be tested by ELISA or western blotting with specific antigens from candidate organisms. This specific antigen can be the adhesin or P-fimbriae which are related to UTI.

In this thesis, extracted DNA from IC patients’biopsies was amplified with bacterial 16S rRNA primers and cloned into pUC18 expression vector (pUC18-IC). These DNAs can be preserved and their diversity can be maintained. Biotinylated single-stranded DNA for sequencing can be harvested effectively by magnetic separation technique. Although artifactual explanations for our results are possible, it is very likely that our detection of bacterial 16S rRNA sequences in patients’ biopsies genuinely represents bacterial species. E. coli and Pseudomonas spp. were detected to be possible agents of IC by computer database searches. Since results from two computer database searches were diverse, the definitive etiologic agents of IC are still unknown, Moreover, we could not use traditional Koch’s experiments to address this problem because we had no way to isolate these pathogens. Therefore, this work presents a method to detect the etiologic agents of IC. Identification of the true agent of IC may be addressed by animal models and immunological studies in future studies.