Capillary Electrophoresis–Single-Strand Conformation Polymorphism Analysis for Rapid Identification of Pseudomonas aeruginosa and Other Gram-Negative Nonfermenting Bacilli Recovered from Patients with Cystic Fibrosis

Abstract

We used capillary electrophoresis–single-strand conformation polymorphism (CE-SSCP) analysis of PCR-amplified 16S rRNA gene fragments for rapid identification of Pseudomonas aeruginosa and other gram-negative nonfermenting bacilli isolated from patients with cystic fibrosis (CF). Target sequences were amplified by using forward and reverse primers labeled with various fluorescent dyes. The labeled PCR products were denatured by heating and separated by capillary gel electrophoresis with an automated DNA sequencer. Data were analyzed with GeneScan 672 software. This program made it possible to control lane-to-lane variability by standardizing the peak positions relative to internal DNA size markers. Thirty-four reference strains belonging to the genera Pseudomonas, Brevundimonas, Burkholderia, Comamonas, Ralstonia, Stenotrophomonas, and Alcaligenes were tested with primer sets spanning 16S rRNA gene regions with various degrees of polymorphism. The best results were obtained with the primer set P11P-P13P, which spans a moderately polymorphic region (Escherichia coli 16S rRNA positions 1173 to 1389 [M. N. Widjojoatmodjo, A. C. Fluit, and J. Verhoef, J. Clin. Microbiol. 32:3002–3007, 1994]). This primer set differentiated the main CF pathogens from closely related species but did not distinguish P. aeruginosa from Pseudomonas alcaligenes-Pseudomonas pseudoalcaligenes and Alcaligenes xylosoxidans from Alcaligenes denitrificans. Two hundred seven CF clinical isolates (153 of P. aeruginosa, 26 of Stenotrophomonas maltophilia, 15 of Burkholderia spp., and 13 of A. xylosoxidans) were tested with P11P-P13P. The CE-SSCP patterns obtained were identical to those for the corresponding reference strains. Fluorescence-based CE-SSCP analysis is simple to use, gives highly reproducible results, and makes it possible to analyze a large number of strains. This approach is suited for the rapid identification of the main gram-negative nonfermenting bacilli encountered in CF.

Pseudomonas aeruginosa and other gram-negative nonfermenting bacilli such as Burkholderia cepacia, Stenotrophomonas maltophilia, and Alcaligenes xylosoxidans are a major cause of chronic lung infection in patients with cystic fibrosis (CF) (5). It is essential to identify these organisms accurately to species level because they differ in clinical significance. P. aeruginosa is associated with progressive pulmonary deterioration and poor clinical prognosis (9, 10). Its isolation from CF patients is therefore a major prognostic indicator. The isolation of B. cepacia from CF patients also has major medical and infection control implications because this organism may cause rapidly fatal, necrotizing pneumonia (18) and is resistant to multiple antibiotics (12) and at least some of its clones spread efficiently from patient to patient in hospitals and via social contact (13, 16, 17). In contrast, S. maltophilia and A. xylosoxidans are mostly isolated after lengthy colonization by P. aeruginosa and are generally considered to be harmless colonizers selected after repeated or prolonged treatment for P. aeruginosa infection (3, 5).

The phenotypic identification of P. aeruginosa and other gram-negative nonfermenting bacilli recovered from chronically colonized CF patients is difficult. A large proportion of strains are atypical in terms of the appearance of colonies, metabolism, antibiotic susceptibility, and immunoreactivity (4, 7, 14, 15). Several commercial kits are available for the identification of gram-negative nonfermenting bacilli. Unfortunately, the overall accuracy of these systems is poor for strains isolated from CF patients. A recent study found that the accuracy of four commercial systems for identifying 150 clinical isolates of nonfermenting gram-negative bacilli recovered from CF patients was only 57 to 80% (11). Therefore, the identification of gram-negative nonfermenting bacilli recovered from CF patients is still based on conventional biochemical testing and remains a fastidious, difficult, and time-consuming procedure. Even this approach is not totally reliable. A recent paper reported the misidentification of S. maltophilia as B. cepacia by diagnostic laboratories with experience in working with organisms isolated from CF patients (2).

Genotypic identification methods overcome these problems. A new approach based on single-strand conformation polymorphism (SSCP) analysis of PCR-amplified 16S rRNA gene fragments has been developed (20, 21). SSCP patterns were initially analyzed by nondenaturing polyacrylamide gel electrophoresis and silver staining (20). This protocol was then modified to include fluorescence-based PCR-SSCP coupled to an automated DNA sequencer (21). A single nucleotide difference in the amplified region was sufficient to obtain different patterns with fluorescence-based SSCP analysis. This made it possible to identify a broad range of gram-positive and gram-negative bacteria accurately, including P. aeruginosa and S. maltophilia (21). Capillary electrophoresis (CE) is a new electrophoretic technique in which slab gels are replaced by capillaries. CE-SSCP analysis is particularly useful for the detection of point mutations associated with inherited diseases (1, 8). In this study, we used CE-SSCP analysis for the rapid identification of P. aeruginosa and other nonfermenting gram-negative bacilli recovered from CF patients.

Test organisms.

The strains used in this study were 34 reference strains (Table 1) and 207 clinical isolates of gram-negative nonfermenting bacilli (gram-negative, rod-shaped, cytochrome c oxidase-positive organisms utilizing glucose oxidatively) recovered from the respiratory tracts of CF patients. Clinical isolates were obtained from the frozen strain collections of the microbiology laboratories of Necker-Enfants Malades Hospital and Robert Debré Hospital, Paris, France, and Centre Hospitalier Régional et Universitaire, Lille, France. They included 153 P. aeruginosa, 26 S. maltophilia, 11 B. cepacia, 4 Burkholderia gladioli, and 13 A. xylosoxidans isolates. Multiple isolates of the same species from an individual patient were not included, except for P. aeruginosa isolates with colonies differing in appearance (e.g., fried-egg pigmented and mucoid nonpigmented P. aeruginosa). Clinical isolates were identified by standard methods (6, 19). P. aeruginosa isolates were serotyped by the slide agglutination technique, with commercially prepared antisera against 16 somatic O antigens (Sanofi Diagnostics Pasteur, Marnes-la-Coquette, France). Strains that were not agglutinated or were agglutinated by more than one typing serum were classed as nontypeable. The mucoid aspect of the colonies was also noted.

TABLE 1.

Reference strains used in this study

Species	Strainb	Serotype
Pseudomonas aeruginosa	ATCC 10145T	O6
Pseudomonas aeruginosa	CIP 59.33	O1
Pseudomonas aeruginosa	CIP 59.34	O2
Pseudomonas aeruginosa	CIP 59.35	O3
Pseudomonas aeruginosa	CIP 59.36	O4
Pseudomonas aeruginosa	CIP 59.38	O7
Pseudomonas aeruginosa	CIP 59.39	O6
Pseudomonas aeruginosa	CIP 59.41	O9
Pseudomonas aeruginosa	CIP 59.43	O10
Pseudomonas aeruginosa	CIP 59.44	O11
Pseudomonas aeruginosa	CIP 59.45	O12
Pseudomonas aeruginosa	CIP 60.92	O13
Pseudomonas alcaligenes	ATCC 14909T	NAc
Pseudomonas fluorescens	CIP 69.13T	NA
Pseudomonas pseudoalcaligenesa	ATCC 17440T	NA
Pseudomonas putida	CIP 52.191T	NA
Pseudomonas stutzeri	CIP 103022T	NA
Brevundimonas diminuta	ATCC 11568T	NA
Brevundimonas vesicularis	ATCC 11426T	NA
Burkholderia andropogonis	LMG 2129T	NA
Burkholderia caryophylii	LMG 2155T	NA
Burkholderia cepacia	CIP 80.24T	NA
Burkholderia cocovenenans	LMG 11626T	NA
Burkholderia gladioli pv. gladioli	CFBP 2427T	NA
Burkholderia gladioli pv. alliicola	CFBP 2422T	NA
Burkholderia glumae	CFBP 2430T	NA
Burkholderia plantarii	LMG 9035T	NA
Ralstonia pickettii	ATCC 27511T	NA
Comamonas acidovorans	ATCC 15668T	NA
Stenotrophomonas maltophilia	CIP 60.77T	NA
Alcaligenes xylosoxidans	CIP 61.20	NA
Alcaligenes faecalis	CIP 67.23	NA
Alcaligenes denitrificans	CIP 60.83	NA
Alcaligenes piechaudii	CIP 101223	NA

^aSubspecies pseudoalcaligenes. 

^bT, type strain; ATCC, American Type Culture Collection, Manassas, Va.; CIP, Collection de l’Institut Pasteur, Paris, France; LMG, Laboratorium voor Microbiologie, State University of Ghent, Ghent, Belgium; CFBP, Collection Française de Bactéries Phytopathogènes, Institut National de la Recherche Agronomique, Station de Pathologie Végétale et Phytobactériologie, Angers, France. 

^cNA, not applicable. 

DNA amplification.

Bacterial strains were cultured overnight at 35°C on tryptic soy agar, scraped from the plates, and suspended in distilled water to give an optical density of 1.2 to 1.3 at 600 nm. Bacterial cells were lysed by heating at 100°C for 10 min. They were centrifuged for 2 min, and 5 μl of the supernatant was then directly used for PCR. The PCR mixture (final volume = 100 μl) contained 0.1 μM (each) primer, 200 μM (each) deoxynucleoside triphosphates, and 2.5 U of Taq DNA polymerase (ATGC Biotechnologies, Noisy-le-Grand, France), in 1× amplification buffer (10 mM Tris-HCl [pH 8.3], 50 mM KCl, 1.5 mM MgCl₂). Three sets of primers, purchased from GENSET SA (Paris, France), were used for amplification of the 16S rRNA gene regions. The first set of primers was P11P (5′-GAG GAA GGT GGG GAT GAC GT) and P13P (5′-AGG CCC GGG AAC GTA TTC AC), to amplify a 217-bp fragment (Escherichia coli 16S rRNA positions 1173 to 1389) (20, 21). The second set of primers was ER14 (5′-GCT AAC TCC GTG CCA GCA) and ER15 (5′-GCG TGG ACT ACC AGG GTA TC), to amplify a 305-bp fragment (E. coli 16S rRNA positions 506 to 810) (21). The third set of primers (this study) was MM3 (5′-GCA GCA GTG GGG AAT TTT GG) and MM4 (5′-TTA CGC CCA GTA ATT CCG AT), to amplify a 213-bp fragment (E. coli 16S rRNA positions 359 to 571). The forward primers (P11P, ER10, and MM3) were labeled with the fluorescent dye TET (green), and the reverse primers (P13P, ER11, and MM4) were labeled with the fluorescent dye FAM (blue). The PCR was performed with 25 cycles of 1 min at 94°C, 10 s at 72°C, and 1 min at 55°C. Five microliters of the amplified product was loaded onto an agarose gel and subjected to electrophoresis, and the gel was viewed under UV light to check for DNA amplification.

CE-SSCP analysis.

CE-SSCP electrophoresis was performed on an ABI PRISM 310 genetic analyzer (Perkin-Elmer). Experimental conditions were optimized by using P11P-P13P PCR products from P. aeruginosa ATCC 10145^T. We evaluated 2 to 4% polymer and 5 to 10% glycerol. The best results were obtained with 4% polymer and 10% glycerol. Perkin-Elmer recommends purifying PCR products by removing salts and excess primers with Amicon Centricon-100 columns (Millipore, Bedford, Mass.). In our hands, this step did not appear to be absolutely necessary, providing that the PCR products were used at dilutions greater than 1/50 for SSCP analysis. Perkin-Elmer also recommends improving DNA heat denaturation by using both formamide and NaOH. Preliminary trials showed that adding NaOH did not improve DNA denaturation and that it affected the mobility of single-stranded DNA molecules. DNA was therefore heat denatured in the presence of formamide alone in our final protocol. The deionization of formamide was essential and was carefully performed. The definitive protocol used with P11P-P13P was as follows. After thermal cycling, 1 μl of the PCR mixture diluted 1:100 was added to 14 μl of sampling buffer (12.5 μl of deionized formamide, 1.0 μl of distilled water, and 0.5 μl of GeneScan 500 standard). GeneScan 500 standard was labeled with the fluorescent dye TAMRA (yellow). Formamide was deionized by using Ag501-X8 resin (Bio-Rad Laboratories, Hercules, Calif.). The sample was heated for 3 min at 95°C and chilled on ice before being loaded on the analyzer. The SSCP analysis gel contained 4% GeneScan polymer, 10% glycerol, and 1× Tris-borate-EDTA. The electrophoresis conditions were as follows: capillary, Lt, 41 cm, and Ld, 30 cm; temperature, 30°C; electric field, 317 V/cm. SSCP patterns were analyzed with GeneScan 672 software (Perkin-Elmer). The conditions established with P11P-P13P were applied to ER10-ER11 and MM3-MM4. Similar results were obtained, but PCR products were diluted only 1/10 for CE-SSCP analysis because the yield of PCRs was lower with these primer sets. Purification of PCR products with Amicon Centricon-100 columns was therefore required. The protocol was otherwise the same as that with P11P-P13P.

Electrophoresis was completed in 25 min, making it possible to analyze 48 samples (one 48-well sample tray) in about 20 h. For each sample, the fluorescence intensity (y axis) was plotted as a function of time (x axis, scan numbers). The values on the x axis were the scan numbers at which the peaks were detected. Electropherograms consisted of two major peaks and several minor peaks. The major peaks corresponded to the principal conformations of the two single-stranded DNAs amplified by PCR. The extra bands corresponded to unincorporated primers, minor conformations of single-stranded DNAs, and hybridized structures. The GeneScan 672 software made it possible to align samples very precisely by using the GeneScan 500 standard (Fig. 1). Lane-to-lane variability was thus perfectly controlled. The values determined on the x axis after alignment were called corrected scan values (as opposed to the absolute scan values determined before alignment). If the same PCR product was tested 20 times in the same run, the differences in mobility of the peaks did not exceed one scan number. Run-to-run variations for individual PCR products were also minor and did not exceed 2.5 scan numbers. PCR products obtained from the same strain and the same primer set in separate PCRs gave very consistent results, with differences of no more than one scan number if analyzed in the same run. Based on these data, mobility shifts of more than two scan numbers were considered to be significant (i.e., reflecting sequence differences), if samples were tested in the same run and aligned by using the GeneScan 500 standard.

FIG. 1.

Correction of lane-to-lane variations with GeneScan 672. Curves: A, internal control (GeneScan 500 standard); B, electropherograms before alignment; C, electropherograms after alignment. Curves B and C consist of the five electropherograms obtained with the same sample passed five times in the same run, superimposed on one another. x axis, scan values; y axis; fluorescence intensity. Before alignment (B), a shift is clearly apparent between samples; after alignment (C), the samples have similar migration patterns.

CE-SSCP analysis of reference strains. (i) Primer set P11P-P13P.

PCR products obtained from reference strains with the primer set P11P-P13P were analyzed by CE-SSCP. We first tested the main species of the genera Pseudomonas, Brevundimonas, Burkholderia, Comamonas, Stenotrophomonas, and Alcaligenes: P. aeruginosa, Brevundimonas vesicularis, B. cepacia, Comamonas acidovorans, S. maltophilia, and A. xylosoxidans, respectively. A specific pattern was observed for each of the six species tested (Fig. 2). Both forward and reverse strands underwent mobility shifts and were useful for discrimination between species. In addition to the strain ATCC 10145^T, 11 reference P. aeruginosa strains belonging to various serotypes were studied. All strains tested gave the same pattern (data not shown), confirming the reliability of CE-SSCP analysis. The discriminatory power of P11P-P13P for closely related species was evaluated (Table 2). Among Pseudomonas species, P. aeruginosa, Pseudomonas alcaligenes, and Pseudomonas pseudoalcaligenes gave patterns that were indistinguishable from each other. The pattern of Pseudomonas stutzeri was related but significantly different (mobility values for the forward strand differing by four scan numbers). The same pattern was observed for Pseudomonas fluorescens and Pseudomonas putida, but this pattern was clearly different from that of P. aeruginosa (Fig. 2). This was not very surprising, because P. fluorescens, P. putida, and P. aeruginosa are classed together in the fluorescent pseudomonad group but P. fluorescens and P. putida differ substantially from P. aeruginosa in the nucleotide sequence of 16S rRNA. The patterns of Brevundimonas diminuta and B. vesicularis were similar and were distinct from other patterns observed for other genera. The range of mobility values was broader with the reverse strand (scan numbers of 3914 to 3928 versus 3825 to 3830 with the forward strand). However, both strands were informative. For example, only the forward strand differentiated P. aeruginosa from P. stutzeri. Differences in mobility values were limited among species of the genus Burkholderia. However, B. cepacia gave a pattern that was different from the others. The patterns of Burkholderia cocovenenans, B. gladioli pv. alliicola, B. gladioli pv. gladioli, Burkholderia glumae, and Burkholderia plantarii could not be distinguished from one another. Finally, P11P-P13P could not discriminate between Alcaligenes species: the mobility values for the forward strand were similar for the four species tested, and those for the reverse strand differed by only two scan numbers.

FIG. 2.

CE-SSCP analysis of reference strains with P11P-P13P. Curves: A, P. aeruginosa ATCC 10145^T; B, B. cepacia CIP 80.24^T; C, C. acidovorans ATCC 15668^T; D, B. vesicularis ATCC 11426^T; E, S. maltophilia CIP 60.77^T; F, A. xylosoxidans CIP 61.20. x axis, scan values; y axis, fluorescence intensity. Samples were analyzed in the same run and aligned by using the GeneScan 500 standard.

TABLE 2.

CE-SSCP analysis of Pseudomonas, Burkholderia, and Alcaligenes species with the P11P-P13P primer set

Speciesa	Mobility valueb
	Forward strand	Reverse strand
Pseudomonas
P. aeruginosa	3829	3928
P. alcaligenes	3829	3928
P. fluorescens	3828	3914
P. pseudoalcaligenes	3828	3927
P. putida	3828	3915
P. stutzeri	3825	3927
Burkholderia
B. cepacia	3823	3868
B. andropogonis	3837	3877
B. cocovenenans	3830	3870
B. gladioli pv. alliicola	3830	3870
B. gladioli pv. gladioli	3829	3869
B. glumae	3828	3870
B. plantarii	3831	3870
Alcaligenes
A. denitrificans	3830	3893
A. faecalis	3830	3895
A. piechaudii	3830	3894
A. xylosoxidans	3830	3895

^aStrains tested are those listed in Table 1; the type strain ATCC 10145^T was tested as a P. aeruginosa strain. 

^bSamples were analyzed in the same run; mobility values are the scan values of the major peaks obtained after alignment (corrected scan values). 

(ii) Primer sets ER14-ER15 and MM3-MM4.

The ER14-ER15 primer set was used to identify Pseudomonas species. Distinct patterns were obtained with all Pseudomonas species tested. However, the major peaks each consisted of two to three peaks, the relative importances of which differed from one PCR to another (data not shown). We tried without success to overcome this by changing the experimental conditions. The MM3-MM4 primer set was used for the identification of Burkholderia and Alcaligenes species. The discrimination of Burkholderia species was no better than that with P11P-P13P (Table 3). For example, the MM3-MM4 set did not differentiate among B. cocovenenans, B. gladioli pv. alliicola, B. glumae, and B. plantarii. The use of MM3-MM4 was also less effective than was hoped for in identifying Alcaligenes species. In particular, this set of primers did not differentiate between A. xylosoxidans and Alcaligenes denitrificans (Table 3).

TABLE 3.

CE-SSCP analysis of Burkholderia and Alcaligenes species with the MM3-MM4 primer set

Speciesa	Mobility valueb
	Forward strand	Reverse strand
Burkholderia
B. cepacia	3963	3985
B. andropogonis	3950	3987
B. cocovenevans	3956	3971
B. gladioli pv. alliicola	3957	3971
B. gladioli pv. gladioli	3957	3972
B. glumae	3957	3972
B. plantarii	3957	3971
Alcaligenes
A. denitrificans	3950	3932
A. faecalis	3987	3964
A. piechaudii	3980	3958
A. xylosoxidans	3950	3932

^aStrains tested are those listed in Table 1. 

^bSamples were analyzed in the same run; mobility values are the scan values of the major peaks obtained after alignment (corrected scan values). 

CE-SSCP analysis of CF isolates.

PCR products from 207 clinical isolates were tested with the P11P-P13P primer set. P. aeruginosa isolates included 47 typeable; 48 nonmucoid, nontypeable; and 58 mucoid, nontypeable isolates. Other isolates included 26 of S. maltophilia, 11 of B. cepacia, 4 of B. gladioli, and 13 of A. xylosoxidans. Control injections were made every 20th injection (three injection controls for a 48-well-plate) to overcome the problem of differences between lanes. Controls included the reference strains P. aeruginosa ATCC 10145^T, B. cepacia CIP 80.24^T, B. gladioli pv. gladioli CFBP 2427^T, S. maltophilia CIP 60.77^T, and A. xylosoxidans CIP 61.20. The CE-SSCP patterns of clinical isolates could be strictly superimposed over those of the corresponding reference strains (mobility values of single-strand DNA peaks differed by less than two scan numbers) (data not shown).

Discussion.

This is the first study using CE-SSCP analysis for bacterial identification. CE-SSCP analysis technology has several advantages over SSCP analysis with slab gels. Multiple fluorescence labeling simplifies the interpretation of electropherograms. The two strands of DNA in the PCR amplicon were labeled with different fluorescent dyes. Having the two strands labeled with different colors simplifies comparison of data from lane to lane and ensures that the same strands are compared for the various samples. In addition, residual double-stranded products are labeled with both colors and are easily distinguished from single-stranded DNA molecules. Another advantage of the technique is the possibility of controlling lane-to-lane variability by using internal lane size standards labeled with a third fluorescent dye. This makes CE-SSCP analysis highly sensitive for detecting minor DNA changes. CE-SSCP analysis is simple and fast. The only step necessary after PCR amplification is heat denaturation in the presence of formamide. Samples are loaded into 48- or 96-well trays that are processed automatically in the analyzer. Electrophoresis takes only 20 min, giving a throughput of up to 48 samples in a 20-h period. In our laboratory, the ABI PRISM 310 genetic analyzer is utilized for CE-SSCP analysis one night per week. The rest of the time it is used for sequencing. Finally, the cost of CE-SSCP analysis is lower than that for other genotypic methods of identification. Basically, it is a PCR with fluorescently labeled primers. Purification of PCR products on columns, as recommended by the manufacturer, is costly. However, this step can be omitted with no problems provided that the PCR products are sufficiently diluted.

We tested primer sets spanning 16S rRNA gene regions with various degrees of polymorphism. P11P-P13P spans a region that is conserved among Pseudomonas and related gram-negative nonfermenting bacilli. ER14-ER15 and MM3-MM4 span much more polymorphic regions. The ER14-ER15 region is the most polymorphic in Pseudomonas. For example, the closely related species P. aeruginosa and P. alcaligenes (identical P11P-P13P region) differ by four bases in the ER14-ER15 region and only one base in the MM3-MM4 region. The MM3-MM4 region is the most polymorphic in Burkholderia and Alcaligenes. For example, B. cepacia and B. gladioli pv. gladioli differ by 2 bases in the ER14-ER15 region and by 12 bases in the MM3-MM4 region. We first thought to use the P11P-P13P primer set for genus assignment and the ER14-ER15 and MM3-MM4 primer sets for species assignment in Pseudomonas and Burkholderia-Alcaligenes, respectively. However, CE-SSCP analysis with P11P-P13P turned out to be more discriminative than expected. P11P-P13P distinguished P. aeruginosa from most other Pseudomonas species tested. The only exceptions were the closely related species P. alcaligenes and P. pseudoalcaligenes, which gave the same pattern as P. aeruginosa. However, P. alcaligenes and P. pseudoalcaligenes are isolated only infrequently from CF patients and are easily differentiated from P. aeruginosa by phenotypic tests. P11P-P13P also differentiated B. cepacia from other Burkholderia species. The P11P-P13P pattern of S. maltophilia was unique among all species tested. Finally, P11P-P13P differentiated A. denitrificans-A. xylosoxidans from the species Alcaligenes faecalis and Alcaligenes piechaudii.

The ER14-ER15 and MM3-MM4 primer sets were unsatisfactory. A major problem with ER14-ER15 was that each major peak consisted of two to three internal peaks, probably reflecting the various possible conformations for each single-stranded DNA. Differences between strains were therefore difficult to assess, precluding the use of this primer set for diagnostic purposes. Such problems were not reported in a previous SSCP study with ER14-ER15 with slab gel electrophoresis (21). Such differences may be due to the higher resolving power of CE-SSCP. The MM3-MM4 primer set was also of limited value. MM3-MM4 gave no better differentiation of Burkholderia species than did P11P-P13P. The discriminatory power of MM3-MM4 was slightly greater than that of P11P-P13P with respect to Alcaligenes. However, MM3-MM4 did not distinguish between A. denitrificans and A. xylosoxidans, by far the most common Alcaligenes species recovered from CF patients. ER14-ER15 and MM3-MM4 also gave much lower PCR yields than did P11P-P13P. Therefore, MM3-MM4 and ER14-ER15 PCR products could not be used at dilutions greater than 1:10 for SSCP analysis and had to be purified on columns prior to use. This rendered the protocol more difficult and costly.

Pseudomonas isolates recovered from the respiratory tracts of CF patients are phenotypically different from wild-type environmental isolates. P. aeruginosa strains isolated from CF patients are commonly mucoid, deficient in lipopolysaccharide O side chains, nonmotile, and resistant to several antibiotics (4, 7, 14, 15). These phenotypic changes are mostly caused by genetic mutations. It was therefore essential to demonstrate that the 16S rRNA gene CE-SSCP was applicable to bacterial isolates recovered from CF patients. We tested about 160 isolates from CF patients. These isolates belonged to the main species of gram-negative nonfermenting bacilli found in CF patients, i.e., P. aeruginosa, B. cepacia-B. gladioli, S. maltophilia, and A. xylosoxidans. These isolates gave CE-SSCP patterns identical to those of the corresponding reference strains. This suggests that the CF isolates do not have mutations within their 16S rRNA genes, which is not totally unexpected because 16S rRNA is functionally constrained by high selection pressure. The 16S rRNA gene therefore appears to be a good target for the identification of CF isolates by CE-SSCP analysis.

The Perkin-Elmer ABI PRISM 310 apparatus can also be used for automated fluorescence sequencing. This sequencing method is easy to perform and gives excellent results. Rhodamine Dye Deoxy Terminator Cycle sequencing kits are now available. This new sequencing technology is very convenient and makes it possible to obtain reliable sequence data for both strands of a 400- to 600-bp fragment. A sequence-based strategy has the advantage of generating direct, unambiguous data. However, it is costly and is neither as fast nor as simple as CE-SSCP analysis. Sequencing a 400-bp sequence is 50 to 100 times more expensive than CE-SSCP analysis. The ABI PRISM 310 apparatus can process up to 60 CE-SSCP analysis samples in a 24-h period versus only 10 400-bp sequences. Therefore, CE-SSCP analysis is more appropriate for the identification of large numbers of strains. We use CE-SSCP in our laboratory to screen for nonpigmented CF clinical isolates of nonfermenting gram-negative bacilli. Only those isolates that cannot be identified by CE-SSCP are further studied by 16S rRNA gene sequencing.

Thus, 16S rRNA gene CE-SSCP analysis is feasible for the identification of the gram-negative nonfermenting bacilli recovered from the respiratory tracts of CF patients. We recommend using the P11P-P13P primer set because it is convenient and reliable. CE-SSCP analysis of the P11P-P13P region may be used to identify the main pathogenic species found in CF patients. Isolates unidentified by CE-SSCP analysis can be studied by 16S rRNA gene sequencing. This two-step strategy of identification by CE technology is currently under evaluation in our laboratory.