Genetic Diversity among Strains of Moraxella catarrhalis: Analysis Using Multiple DNA Probes and a Single-Locus PCR-Restriction Fragment Length Polymorphism Method

Abstract

Moraxella (Branhamella) catarrhalis, a causative agent of otitis media, sinusitis, and exacerbation of bronchitis, has acquired widespread ability to produce β-lactamase and can be nosocomially transmitted. The typing methods used in epidemiological analyses of M. catarrhalis are not optimal for genetic analyses. Two methods, a multiple-locus Southern blot (SB) method and a single-locus PCR-restriction fragment length polymorphism (RFLP) method, were developed and used to assess genetic diversity and potential clinical and geographic relationships in M. catarrhalis. Nine randomly cloned M. catarrhalis DNA fragments were used as probes of SBs containing DNA from 54 geographically and clinically diverse strains. For comparison, a PCR-RFLP method was developed as a quick, inexpensive, and discriminating alternative. A highly variable 3.7-kb genomic region (M46) was cloned and sequenced, and 3.5 kb of the cloned DNA was targeted for PCR amplification. DNAs from the 54 strains were subjected to PCR-RFLP. SB analysis distinguished all strains that had no apparent epidemiological linkage (40 of 54), and PCR-RFLP distinguished fewer strains (21 of 54). Epidemiologically linked strains appeared genetically identical by both methods. PCR-RFLP was compared to pulsed-field gel electrophoresis (PFGE) for 8 of the 54 strains and 23 additional strains. PCR-RFLP distinguished fewer strains than PFGE typing (16 of 31 versus 20 of 31 strains), but PCR-RFLP was more useful for inferring interstrain relatedness. Separate cluster analyses of multilocus SB and single locus PCR-RFLP data showed high genetic diversity within and across geographic locations and clinical presentations. The resultant dendrograms were not entirely concordant, but both methods often gave similar strain clusters at the terminal branches. High genetic diversity, nonconcordance of cluster analyses from different genetic loci, and shared genotypes among epidemiologically linked strains support a hypothesis of high recombination relative to spread of clones. Single-locus PCR-RFLP may be suitable for short-term epidemiological studies, but the SB data demonstrate that greater strain discrimination may be obtained by sampling variation at multiple genomic sites.

Moraxella (Branhamella) catarrhalis is a gram-negative diplococcus with pathogenic potential. Frequently found as a commensal in the upper respiratory tract (7, 9, 17). M. catarrhalis can cause sinusitis and exacerbation of chronic bronchitis in adults and otitis media in children (9, 18, 20). The population genetics of M. catarrhalis are not known (7), but the species is naturally competent for transformation (5). Nosocomial transmission of M. catarrhalis has been documented (1, 15), and patterns of transmission, colonization, and reinfection have been examined by using protein or nucleic acid markers for differentiating strains (4, 7, 11, 12, 14–16).

DNA-based typing methods applied to M. catarrhalis include small-fragment restriction enzyme analysis (4, 6, 8, 15), large-fragment restriction enzyme analysis with pulsed-field gel electrophoresis (PFGE) (11, 12), and Southern blot (SB) analysis (4). Each method has provided valuable information on the epidemiology of M. catarrhalis; however, each approach is time-consuming, and genomic DNA from some strains may be reproducibly refractory to enzymatic digestion (6, 8). Consequently, none of the reported DNA-typing methods for M. catarrhalis are optimal for long-term or population-based studies.

In this study, we report the development of two DNA-typing methods and assess each method for the ability to detect genetic diversity in M. catarrhalis. With multiprobe SB analysis, we could distinguish all epidemiologically unlinked strains tested. Moreover, cluster analysis based upon multiple genomic loci did not reveal a strong association of genotype with clinical presentation or geographic location. Single-locus PCR-restriction fragment length polymorphism (RFLP) did not differentiate unrelated strains as finely as multiprobe SB, but ease, rapidity, and ability to localize the source of variation point to PCR-RFLP as a promising genotyping method for M. catarrhalis.

MATERIALS AND METHODS

Bacterial strains.

Fifty-four strains of M. catarrhalis were typed by both SB and PCR-RFLP analyses. The sample included representatives from each of five localities in the United States and two in Europe (Table 1). Thirty-one strains, including 13 from a single European locality and 17 from a single U.S. locality, were typed by both PFGE and PCR-RFLP (Table 2). A subset of eight strains were typed by all three methods (SB, PCR-RFLP, and PFGE). Epidemiological links between strains, when available, are indicated in Tables 1 and 2. Strains of M. catarrhalis having strong epidemiological links included 5 isolates from a nosocomial outbreak in a Connecticut veterans hospital (15), pediatric paired isolates (nasopharyngeal and ear effusion) from children in New York with otitis media (5 pairs), 13 isolates from six bronchiectatic patients in the United Kingdom (12), 4 pediatric colonists from two children in the Netherlands, and 5 isolates from Swedish patients with laryngitis.

TABLE 1.

Characteristics of bacterial strains typed by SB and PCR-RFLP analysis

Location and isolatea	Source	DOIb	Genotypec
			SB	PCRRFLP
Veterans hospital, West Haven, Conn. (noso- comial outbreak) (10)
CT1		1987	S1	P1
CT2		1987	S1	P1
CT3		1987	S1	P1
CT4		1987	S1	P1
CT5		1987	S1	P1
Netherlands (pediatric isolates; colonists)
NE1a	Throat		S20	P16
NE1b	Nose		S21	P10
NE2a	Throat		S22	P15
NE2b	Nose		S22	P15
Sweden (adults with laryngitis)
SW1	Pharynx		S23	P10
SW2	Pharynx		S24	P10
SW3	Pharynx		S25	P4
SW4	Pharynx		S26	P17
SW5	Pharynx		S27	P7
New York (children with otitis media)
NY1a	Ear		S28	P15
NY1b	Nasopharynx		S28	P15
NY2a	Ear		S29	P18
NY2b	Nasopharynx		S29	P18
NY3a	Ear		S30	P10
NY3b	Nasopharynx		S30	P10
NY4a	Ear		S31	P19
NY4b	Nasopharynx		S32	P20
MHVA
TN1	TTAd, pneumonia	2-28-83	S2	P2
TN2	TTA, pneumonia	3-27-84	S3	P3
TN3	TTA, pneumonia	5-26-84	S4	P4
TN4a	Eye	4-16-85	S5	P5
TN4b	Sputum	4-18-85	S6	P6
TN5	Eye	2-4-87	S7	P7
TN6	Eye	3-15-87	S8	P8
TN7a	Sputum	4-24-87	S9	P9
TN7b	Sputum	4-27-87	S9	P9
TN8	Pharynx, colonist	3-30-87	S10	P7
TN9	Pharynx, colonist	3-30-87	S11	P7
TN10	Pharynx, tracheitis	4-8-87	S12	P7
TN11	Pharynx, colonist	4-9-87	S13	P2
TN12	Pharynx, colonist	4-20-87	S14	P12
TN13a	Sputum	3-19-88	S15	P11
TN13b	Eye	3-20-88	S15	P11
TN14a	Sputum	12-6-88	S16	P10
TN14b	Sputum	12-7-88	S16	P10
TN15a	Sputum	6-2-89	S17	P13
TN15b	Sputum	6-6-89	S17	P13
TN16a	Sputum	12-26-89	S18	P14
TN16b	Sputum	12-27-89	S18	P14
TN17a	Sputum	2-20-90	S19	P15
TN17b	Sputum	2-21-90	S19	P15
Bloodstream isolates
BL1	Blood, Nebraska	1989	S33	P7
BL2	Blood, Massachusetts		S34	P10
BL3	Blood, Tennessee	3-13-93	S35	P15
Other
ATCC 25238	N/Ae	Before 1964	S36	P7
ATCC 43628	Sputum	Before 1986	S37	P7
MA1	N/A, Massachusetts	1991	S38	P10
MA2	N/A, Massachusetts	1991	S39	P21
MA3	N/A, Tennessee	1991	S40	P4

^aIsolates are sorted by geographical location and epidemiological linkage. Paired isolates from individuals are given the same isolate number followed by “a” or “b.” 

^bDOI, date of isolation (year or month-day-year). 

^cGenotypes are identified by number separately for SB and PCR-RFLP. 

^dTTA, transtracheal aspirate. 

^eN/A, not available. 

TABLE 2.

Isolates used for comparison of PCR-RFLP with SpeI PFGE for discriminating M. catarrhalis strains

Location and isolatea	Source	Timeb	Genotype
			PFGE	PCR-RFLP
England (bronchiectasis)
EN1	Sputum		F13	P1
EN2a	Sputum		F4	P22
EN2b	Sputum	2 mo	F5	P7
EN3a	Sputum		F8	P23
EN3b	Sputum	2 mo	F8	P23
EN3c	Sputum	>12 mo	F3	P18
EN4a	Sputum		F10	P19
EN4b	Sputum	1 wk	F10	P19
EN4c	Sputum	2 mo	F13	P1
EN5	Sputum		F10	P19
EN6a	Sputum		F1	P24
EN6b	Sputum	8 mo	F13	P1
EN6c	Sputum	6 mo	F2	P25
New York (children with acute otitis media)
NY1a	Ear		F12	P15
NY1b	Nasopharynx		F12	P15
NY2a	Ear		F11	P18
NY2b	Nasopharynx		F11	P18
NY3a	Ear		F6	P10
NY3b	Nasopharynx		F6	P10
NY4a	Ear		F10	P19
NY4b	Nasopharynx		F14	P20
NY5a	Ear		F10	P19
NY5b	Nasopharynx		F10	P19
NY6	Ear		F15	P26
NY7	Ear		F7	P23
NY8	Ear		F9	P27
NY9	Ear		F18	P1
NY10	Nasopharynx		F19	P28
NY11	Ear		F16	P1
NY12	Ear		F17	P15
Other
XX1			F20	P29

^aIsolates are sorted by geographic location and epidemiological linkage. Serial isolates from a single individual are designated by an “a,” “b,” or “c” following the isolate number. 

^bTime, the time span separating an isolate from the previous isolate in a series. 

The ability of SB and PCR-RFLP to detect genetic diversity at the local level was assessed by including 24 strains collected from patients over a 7-year span (1984 to 1990) at a single veterans hospital (Mountain Home Veterans Affairs Medical Center, Johnson City, Tenn.) (referred to as MHVA strains). MHVA strains included paired isolates from seven patients, where two isolates collected within 3 days from the same individual comprise a pair. Four MHVA strains were isolated from transtracheal aspirates from patients with a diagnosis of pneumonia. Nine MHVA isolates recovered within a 10-week span included one pair of sputum isolates, four pharyngeal colonists, two isolates from eyes, and an isolate from a patient with tracheitis. The remainder of the MHVA isolates were collected from sputum specimens, for which Gram stains revealed fewer than 10 epithelial cells and greater than 25 leukocytes per low-power field.

Experimental design.

Random HindIII DNA fragments of 1.0 to 4.0 kb were size selected and then cloned from M. catarrhalis ATCC 25238. DNA from each cloned fragment was labeled with digoxigenin and used individually as a probe in SB analysis of M. catarrhalis strains of diverse origins. Fifteen cloned M. catarrhalis DNA fragments were tested in a preliminary SB analysis of seven pairs of isolates and three American Type Culture Collection (Rockville, Md.) M. catarrhalis strains. Nine of the fragments, revealing different levels of variation in the preliminary screen, were used as probes of the larger sample of strains subjected to SB analysis (Table 1). SB-RFLPs revealed by individual hybridization of these nine probes were combined into a single data set for cluster analysis to generate inferences of relatedness among strains. One clone, M46, detected high diversity upon Southern hybridization, and this genomic region was targeted for PCR-RFLP to develop a more time-efficient, less costly typing system.

Cloning strategy.

Prior work has shown that the restriction enzymes HindIII and HaeIII reveal considerable diversity among strains of M. catarrhalis (4, 15, 16). Because HaeIII has a 4-bp recognition site and HindIII has a 6-bp recognition site, HindIII can be expected to produce fewer and larger genomic DNA restriction fragments than HaeIII. Therefore, detection of restriction site polymorphisms was enhanced by using relatively large cloned HindIII fragments as the probes of genomic DNA digested with HaeIII.

DNA extraction.

M. catarrhalis cultures were grown at 37°C in Todd-Hewitt broth (Difco, Detroit, Mich.) in 5% CO₂, with shaking for aeration. DNA for cloning experiments was extracted from ATCC 25238 by a hexadecyltrimethylammonium bromide (Sigma Chemical Co., St. Louis, Mo.) technique (3). DNA for SB and PCR-RFLP analyses was extracted with the Puregene DNA isolation kit (Gentra Systems, Research Triangle Park, N.C.).

DNA cloning.

HindIII restriction enzyme digestion of DNA from ATCC 25238 was performed according to the manufacturer’s directions (Promega Corp., Madison, Wis.). Restriction fragments of 1.0 to 4.0 kb were ligated into cloning vector pTZ19U (United States Biochemical Corp., Cleveland, Ohio) and transformed into Escherichia coli dH5α. Transformants were selected on Luria-Bertani broth supplemented with 50 μg of ampicillin/ml. Plasmid DNA from individual transformant colonies was purified by alkaline lysis (Magic Miniprep; Promega Corp.).

Genomic fragments adjacent to the M46 clone (see below) were obtained from a library of ATCC 25238 DNA prepared in the LambdaGEM-11 vector system (Promega Corp.) according to the manufacturer’s protocol. Candidate clones from a genomic library were identified by autoradiography of colony blots with Bluescript plasmids (Stratagene, La Jolla, Calif.) for sequence analysis.

SB analysis.

Genomic DNA from M. catarrhalis strains was digested with HaeIII, subjected to agarose gel electrophoresis, and transferred to positively charged nylon membranes (Boehringer Mannheim Corp., Indianapolis, Ind.) by SB transfer. Probe preparation, hybridization, and detection were performed with a nonradioactive labeling and chemiluminescent detection system (Genius System with Lumi-Phos; Boehringer Mannheim Corp.). Each of nine clones was individually hybridized to blots of M. catarrhalis DNA from test strains.

Restriction enzyme analysis by PFGE.

M. catarrhalis strains were grown overnight at 37°C in brain heart infusion broth, and DNA was purified as described previously (12). For restriction endonuclease digestion, an 80-μl agarose plug containing DNA was placed into restriction enzyme buffer containing 100 mg of acetylated bovine serum albumin and 12 U of SpeI (New England Biolabs, Beverly, Mass.) and incubated overnight at 37°C. Samples were loaded into the wells of a 1.2% agarose gel (Agarose NA; Pharmacia LKB, Piscataway, N.J.). Lambda concatamers (Lambda Ladder PFG marker; New England Biolabs) were used as a molecular weight standard. Electrophoresis was performed with the contour-clamped homogeneous gel electrophoresis device (CHEF DR II; Bio-Rad, Richmond, Calif.). The switch interval was ramped from 1 to 20 s over a 24-h period at 200 V and 10°C. The gels were photographed with UV transillumination, and RFLP patterns were determined by visual inspection of the gels. RFLP patterns differing by one or more DNA fragments were considered different.

DNA sequencing.

The M46 clone was sequenced with an Applied Biosystems (Foster City, Calif.) 373A sequencer with fluorescent terminator chemistry. The M46 insert in pTZ19U was sequenced on both strands by a primer-walking strategy, beginning with universal M13 primers at opposite ends of the insert. Subsequent primers were chosen from newly analyzed sequence at 250- to 300-bp intervals. Sequencing of the genomic regions adjacent to M46 is currently in progress, using primer walking with the clones described above. Sequence data was analyzed with AssemblyLIGN and MacVector software (Eastman Kodak, New Haven, Conn.) and the National Center for Biotechnology Information on-line BLAST sequence comparison program (2).

PCR-RFLP.

Primers targeting the M46 DNA sequence were selected with Oligo 4.0 software (National Biosciences, Plymouth, Minn.). One primer pair (A) was designed to amplify as much of the M46 sequence as possible in one reaction, with a predicted amplimer size of 3,172 bp (Fig. 1). The combination of primer pairs B and C was designed to span the M46 region in two smaller amplimers that would overlap by approximately 100 bp (Fig. 1). Amplimer B has a predicted size of 1,739 bp, and amplimer C has a predicted size of 1,890 bp. Primer sequences are listed 5′ to 3′ as follows: primer A left, TGATTGCCGTGCCATTCACA; A right, TCGCCACCAACAATACAAAC; B left, CGTGCCATTCACATCATCAG; B right, ACCCCCGAGCGTTGGAATAC; C left, GGGCTTTAACACCACTTGAA; and C right, CAGCGACCCAAGCAACATTA.

FIG. 1.

Map of M46 sequence with predicted coding regions and positions of PCR-RFLP amplimers. The numbers refer to nucleotide positions on the M46 clone. The thick bars indicate coding sequences, with arrows showing the direction of transcription. See the text for primer sequences for the indicated amplimers.

PCR mixtures consisted of 50 mM KCl; 10 mM Tris-HCl (pH 9.0); 1% Triton X-100; 0.2 mM (each) dGTP, dATP, dCTP, and dTTP; 1.5 mM MgCl₂; 1 mM each primer; 20 ng of genomic DNA; and 2.5 U of Taq polymerase per 50-μl reaction (Perkin-Elmer, Foster City, Calif.). Amplification was performed in a Gene Amp 9600 (Perkin-Elmer). The cycling program was 5 min at 94°C and 35 cycles of 94°C for 15 s, 45°C for 15 s, and 72°C for 45 s, followed by 7 min at 72°C. Amplimer size and yield were assayed by agarose gel electrophoresis on 1% Seakem GTG agarose gels (FMC, Rockland, Maine). BioMarker EXT (BioVentures, Murfreesboro, Tenn.) or the READY-LOAD 100-bp ladder (Life Technologies Inc., Gaithersburg, Md.) was used as a size marker.

To detect RFLP variation within PCR-amplified regions encompassed by M46 DNA, 10-μl aliquots of PCR products were subjected to separate restriction enzyme digestion with HaeIII and RsaI (Promega Corp.) according to the manufacturer’s instructions. Restriction fragments were separated on 1% Seakem GTG–0.65% NuSeive GTG agarose gels (FMC) at 75 V for 4 h.

Data collection and cluster analysis.

Unique DNA types identified by each of the three typing methods were assigned different numerical designations, as shown in Tables 1 and 2. PFGE patterns were identified as different if one or more variant DNA fragments were detected by visual inspection. PFGE patterns were readily distinguished within a size range of DNA fragments from 23 to greater than 242 kb. Strains differing by one or more fragments greater than 98 kb were considered different. However, similarity estimates were not feasible because the density of DNA fragments in the 23- to 98-kb size range did not permit reliable differentiation of individual fragments (Fig. 2). The various types correspond to a minimum of a single-band difference. Each restriction fragment was scored as present or absent for each strain. The resultant strain-by-restriction fragment matrix was used in cluster analyses to infer relatedness among strains.

FIG. 2.

Examples of patterns produced by SpeI PFGE restriction digestion of M. catarrhalis strains. The strains are from four children with otitis media. Lane M contains the 48.5-kb ladder formed from concatamers of the bacteriophage lambda genome, with the lowest band a monomer. Note the concentration of restriction fragments smaller than 97 kb. Lanes 1 to 8 contain strains NY1a, NY1b, NY2a, NY3a, NY2b, NY3b, NY4a, and NY4b, respectively.

Small SB-RFLP fragments (≤400 bp), which may not bind efficiently to the membrane, were excluded from the analysis. Since PCR primers and primer artifacts present in restriction reactions may mask small PCR-RFLP fragments, fragments less than 80 to 90 bp were also excluded from the analysis.

PCR-RFLP and SB data sets were analyzed separately to construct two similarity dendrograms. Cluster analysis was performed with MEGA software (19) by using the neighbor-joining method and Euclidean distance, with the results displayed in dendrogram form (Fig. 3).

FIG. 3.

Dendrograms displaying similarities among M. catarrhalis genotypes. The bar indicates the branch length of a single DNA fragment difference. (A) Dendrogram based on SB-RFLP data. (B) Dendrogram based on M46 PCR-RFLP data.

Nucleotide sequence accession number.

The GenBank accession no. for the M46 sequence is U73324.

RESULTS

Cloned DNAs as probes in SB analysis.

Individual probes hybridized to one to five HaeIII restriction fragments within a strain and revealed various levels of diversity among test strains. The number of RFLP patterns displayed with a single probe ranged from 1 to 10. Nine of the 15 clones were selected for use as probes in SB analysis of a larger sample of isolates. Clones were chosen to represent genomic regions with differing levels of RFLP variation among strains in preliminary analysis. Clones M5, M8, M16, and M69 hybridized to one to three bands per strain and reveal low-to-modest variation among strains (Fig. 4A, B, C, F, and H); clone M64 hybridized to more fragments per strain, revealing minor variation among strains (Fig. 4G); and clones M18, M46, and M79 hybridized to three to seven fragments per strain and revealed higher levels of variation (Fig. 4D, E, and I). The resultant DNA types (assigned numerical identifiers in Table 1) are based upon combined data from genomic regions of both low and high variation and should provide a representative sampling of total genomic variation.

FIG. 4.

Examples of RFLPs revealed by SB hybridization. RFLP patterns were revealed by hybridization of cloned DNA to HaeIII-digested DNA from seven diverse M. catarrhalis strains. Lanes 1 to 7 contain strains BL2, SW3, NY2a, NY3a, CT1, TN14a, and TN17a, respectively. The positions of size markers (6.6, 4.4, 2.2, 2.0, 1, 0.7, 0.5, and 0.2 kb) are indicated by bars to the right of each panel (not all panels contain the entire size range). For each panel, an arrowhead marks the 2.0-kb fragment for orientation. Cloned DNAs used as probes were M5 (A), M8 (B), M16 (C), M18 (D), M46 (E), M57 (F), M64 (G), M69 (H), and M79 (I).

Clone M46 revealed high HaeIII RFLP diversity among many strains but hybridized to a single 3.7-kb HindIII fragment in 14 strains. These data strongly suggest the M46 DNA sequence is present in a single copy in the test strains and therefore represents a suitable target for PCR-RFLP. To identify appropriate sites for PCR-RFLP primers and to gain insight into the nature of the genomic region expressing high genetic variability detected by the M46 clone, we sequenced the M46 clone and the two regions adjacent to M46 in the M. catarrhalis genome. The sequence contained three open reading frames, two of which were truncated at opposite ends of the M46 clone. Based on high sequence similarities, the proteins predicted by each open reading frame are cognates of previously described proteins from other organisms in the GenBank database. As shown in Fig. 1, the M46 DNA sequence corresponds to approximately 30% of the gene for the beta subunit of glycyl-tRNA synthetase (glyS), the entire gene corresponding to the alpha subunit of glycyl-tRNA synthetase, and approximately 90% of a gene similar to the E. coli nadR (nadI) aporepressor gene. (Coding sequences for the remaining portions of the glyS β and nadR cognate genes, in clones adjacent to M46, remain to be fully analyzed.)

PCR-RFLP analysis.

Primer sites were conserved among strains, as evidenced by PCR products of the predicted sizes in all test strains. The PCR product yield of the 3.2-kb fragment from primer pair A was relatively low, and these primers were not used in further testing. Primer pairs B and C were used in PCR-RFLP analysis because product yield was sufficiently high to detect restriction fragments on ethidium-stained agarose gels, eliminating the need for blotting and/or detection of DNA fragments with probes.

There was no detectable length variation in the PCR amplification products from 77 isolates (representative examples are shown in Fig. 5), and restriction digests of the amplimers revealed 29 different DNA types (Table 1). Length variants in this region of the M. catarrhalis genome, if present, are below the limits of detection by our method.

FIG. 5.

Representative diversity revealed by M46 PCR-RFLP in M. catarrhalis. (A) Amplimer B; (B) amplimer C. Samples were treated with HaeIII (lanes 1 to 7) or with RsaI (lanes 8 to 14) or are undigested amplimers (lanes 15 to 21). Each lane contained 10 μl of PCR product. Lanes 1, 8, and 15, sample ATCC 25238; lanes 2, 9, and 16, sample TN16a; lanes 3, 10, and 17, sample BL2; lanes 4, 11, and 18, sample NY1a; lanes 5, 12, and 19, sample TN17a; lanes 6, 13, and 20, sample TN7a; lanes 7, 14, and 21, sample TN13a. Lanes M are 100-to-1,500-bp ladders, with double-bright bands at 600 bp and additional fragments of 2,000 bp.

Estimation of identity and similarity among strains.

SB analysis provided more discrimination among strains than M46 PCR-RFLP (40 versus 21 types among 54 isolates). SpeI PFGE provides a level of discrimination similar to that of M46 PCR-RFLP (20 PFGE versus 16 PCR-RFLP types among 31 isolates). In every instance, those isolates with identical SB or SpeI PFGE types also shared identical M46 PCR-RFLP types. With PCR-RFLP and PFGE, identical types were found in several isolates, but some of these isolates had no evident epidemiological or geographical links. For example, PCR-RFLP type 15 was found in paired isolates from The Netherlands (Table 1) but it was also found in isolates from New York and Tennessee. Similarly, PFGE type 10 was isolated from England as well as New York (Table 2). In contrast, isolates that shared SB types were invariably epidemiologically linked, as evidenced by the presence of SB type 1 in all five isolates from a nosocomial outbreak in Connecticut and by several instances of two isolates from a pair sharing a SB type (Table 1).

For cluster analysis, the number of restriction fragments scored for each data set was similar to those for the other set, with 46 DNA and 40 restriction fragments revealed by SB and PCR-RFLP, respectively. In both PCR-RFLP and SB dendrograms, geographic and clinical diversity are found within clusters.

DISCUSSION

Epidemiological studies of M. catarrhalis have shown diversity within the species and often revealed shared molecular types among epidemiologically linked isolates (4, 8, 12, 15, 16). Our work confirms and extends these observations. First, strains with strong epidemiological links had identical types by multilocus SB and M46 PCR-RFLP (Table 1) and by PCR-RFLP and SpeI-PFGE (Table 2). Identical strains included five Connecticut outbreak strains (CT1 to CT5 [Table 1]) previously shown to be identical by small-fragment restriction enzyme analysis (15) and SB analysis with two probes (4). Second, the most discriminating method, SB, revealed high genetic diversity within and across geographic locations and clinical presentations.

In regard to assessing the identity of strains comprising a paired isolate, both SB and PCR-RFLP had similar levels of resolution. Both SB and PCR-RFLP were consistent in showing strain identity in epidemiologically linked strains but diversity among strains with similar clinical presentations. For example, both methods showed identity among the Connecticut outbreak strains (Table 1) and between the Dutch paired isolates from NE2a and NE2b, as well as a set of six paired isolates (7a and 7b, 13a and 13b, 14a and 14b, 15a and 15b, 16a and 16b, and 17a and 17b) from MHVA. Both methods showed that some paired isolates were comprised of genetically distinct strains within the pair (NE1a and NE1b, NY4a and NY4b, and TN4a and TN4b). However, SB was able to resolve a greater number of genotypes and to show that some strains that were identical by PCR-RFLP were in fact different, as evidenced by the nine strains that were assigned type 10 by PCR-RFLP and that were resolved into seven different strains by SB (Table 1).

The intensive sampling of strains from a single locality, MHVA, shows a pattern revealed in large-scale population genetics surveys of Neisseria gonorrhoeae (13). There appears to be as much diversity within MHVA as is found throughout the global sample, as evidenced by MHVA strains located throughout the dendrogram, and this pattern was seen with both SB and PCR-RFLP data (Fig. 3). A similar pattern of local diversity and random placement throughout the dendrograms was observed in the sample of five Swedish strains (SB types S23 to S27 [Fig. 3A] and PCR types P4, P7, P10, and P17 [Fig. 3B]). Although the eight NY strains were all placed in the upper major branch of the SB dendrogram, there was only a distant relationship between the group of strains that formed a terminal branch and the remaining strains (SB types S30 to S32 versus SB types S28 and S29 [Fig. 3A]). Thus, the genotypic data from both SB and PCR-RFLP strongly suggest that many localities harbor large amounts of genetic diversity with little evidence for local differentiation.

The MHVA isolates of M. catarrhalis represent a highly diverse spectrum of clinical presentation within a locality. Within this single site, isolates were collected from the pharynxes of healthy individuals and an individual with tracheitis, from the sputa of individuals with bronchitis or pneumonia, and from drainages of individuals with eye infections. Examples of diversity of genotypes within a clinical presentation include three genetically distant isolates (SB types 2 to 4 [Fig. 3A]) from pneumonia patients and four colonizing strains (SB types S10 and S11 and S13 and S14 [Fig. 3A]) recovered within a 3-week span which had nonidentical SB types. Evidence of nonspecificity of genotype is represented by sputum and eye specimens (TN13a and -13b) from the same patient which were indistinguishable by SB or PCR-RFLP (Table 1).

The use of a highly discriminating typing system combined with cluster analysis can provide insight into the transmission of strains. For example, nine MHVA isolates were recovered from eight patients in a 3-month time span. A subset of four of these isolates were M46 type 7 (colonist isolates TN1 and TN2, eye isolate TN5, and tracheitis isolate TN10 [Table 1]), and they clustered together by SB data (SB types S7 and S10 to S12 [Fig. 3A]). The high degree of similarity is suggestive of a temporally limited clonal group, but the differences are too great to support a hypothesis of nosocomial transmission.

Two possible associations between SB genotype and clinical presentation were observed. The first is represented by bloodstream isolates BL1 and BL2, collected from Nebraska and Massachusetts, respectively. Given the small sample size and the observation that Tennessee bloodstream isolate BL3 is clearly unrelated, it is not clear whether the clustering of two bloodstream isolates is significant. The second genotype-clinical association comprises two of four MHVA isolates recovered from eye drainage (isolates TN4a, TN5, TN6, and TN13b). Eyes are a less common site for M. catarrhalis. Of the four eye-derived isolates, none had similar PCR-RFLP types, but two (TN4a and TN13b) were members of a terminal cluster by SB (SB types S5 and S15 [Fig. 3A]). These data must be considered inconclusive because of the scarcity of strains derived from more unusual clinical presentations.

Of the methods tested, SB analysis provides the stronger estimate of relationships among strains because it is based upon variation at multiple genetic loci. However, SB is labor intensive and therefore not suited for all applications of genetic typing. SpeI PFGE typing is not amenable to inferring relatedness among strains because of its inability to distinguish all DNA fragments and questions of homology among fragments of similar size. In contrast, the enzymes used in PCR amplification are not hampered by modifications to genomic DNA, which result in nontypeable strains by restriction enzymatic digestion (7, 10). Partial digestion, which can cause problems in interpretation of genomic restriction fragment patterns, can be readily detected with PCR-RFLP, since the cumulative size of fragments from a single restriction reaction should equal the size of the intact amplimer.

M46 PCR-RFLP requires neither high concentration nor high quality of DNA, and HaeIII and RsaI restriction digestion of PCR products is relatively quick and inexpensive and yields simple patterns which can be readily compared among large numbers of strains. Furthermore, levels of variation revealed by M46 PCR-RFLP provide the potential for inferring clonality among strains in short-term epidemiological studies. If needed, the ability to infer relatedness among strains can be strengthened by combining PCR-RFLP data from additional sites in the genome, which would be analogous to using multiple SB probes. PCR-RFLP shows promise as a tool for addressing questions of long-term carriage or strain turnover within an individual and for monitoring long-term trends within a pathogen population.

No prior work has directly addressed the population genetic structure of M. catarrhalis. Based upon preliminary data, Enright and McKenzie suggest that the population structure of M. catarrhalis is similar to that of Neisseria meningitidis (7). Clonality in N. meningitidis is obscured, apparently due to extensive recombination, except for the brief emergence of epidemic clones. In M. catarrhalis, high genetic diversity, nonconcordance of single-locus and multilocus cluster analyses, and shared genotypes among epidemiologically or temporally linked strains support a hypothesis of high recombination relative to spread of clones.

In the past 20 years, both the perception of M. catarrhalis as a pathogen and the genetic makeup of the organism have changed dramatically. Once considered primarily a commensal, M. catarrhalis is now recognized as the third most commonly isolated respiratory tract pathogen (7). Moreover, M. catarrhalis appears to be capable of evolving at a rapid pace; β-lactamase production in the species was unknown prior to 1976, but today 80 to 90% of all strains isolated are β-lactamase producers. Given the prevalence of M. catarrhalis, and the rapid global spread of the β-lactamase gene, an assessment of population genetic structure based upon high-resolution genotypes may provide insights into the transmission and evolution of antibiotic resistance.