horizontal transfer events between distantly related members of the species B. cepacia, leading to the .... Sarcosine-free acid (Sigma) was ...... 2713â2720. 15.
JOURNAL OF BACTERIOLOGY, Feb. 1995, p. 1039–1052 0021-9193/95/$04.0010 Copyright q 1995, American Society for Microbiology
Vol. 177, No. 4
Structurally Variant Classes of Pilus Appendage Fibers Coexpressed from Burkholderia (Pseudomonas) cepacia RICHARD GOLDSTEIN,1* LI SUN,1 RU-ZHANG JIANG,1 UMA SAJJAN,2 JANET F. FORSTNER,2 AND CRAIG CAMPANELLI1 Section of Molecular Genetics, Maxwell Finland Laboratory for Infectious Diseases, Boston University School of Medicine and Boston City Hospital, Boston, Massachusetts 02118,1 and Division of Biochemistry, Research Institute, The Hospital for Sick Children, Toronto, Ontario M5G 1X8, Canada2 Received 20 July 1994/Accepted 13 December 1994
One or more of five morphologically distinct classes of appendage pili were determined to be peritrichously expressed by Burkholderia (formerly Pseudomonas) cepacia isolated from disparate sources. B. cepacia-encoded cblA pilin gene hybridization-based analysis revealed that one associated class, cable (Cbl) adhesin type IIB. cepacia pili, correlates with epidemically transmitted strains from a single cystic fibrosis (CF) center. When only phenotypic assays were available, correlations between the source and the pilus type were nonetheless observed: filamentous (Fil) type IIIB. cepacia pili correlated with CF-associated nonepidemic isolates, spine (Spn) type IVB. cepacia pili correlated with clinical (non-CF) isolates, and spike (Spk) type VB. cepacia pili correlated with environmental isolates. Further, Cbl, Fil, or Spk pili typically appear as an internal framework for constitutively coexpressed, peritrichously arranged dense mats of fine, curly mesh (Msh) type IB. cepacia pili. Constitutive coexpression of dense mats of Msh type IB. cepacia pili in association with a labyrinth of either Cbl, Fil, or Spk pili suggests possible cooperative pilus interactions mediating adhesion-based colonization in the differing environments from which the strains were isolated. Despite such correlations, phylogenetic analyses indicate that with the exception of the epidemically transmitted clusters of isolates, the remaining B. cepacia strains from the other three sources exhibited an equal degree of genetic relatedness independent of origin. As previously found for Escherichia coli, this discrepancy could be accounted for by selection-driven, in vivo horizontal transfer events between distantly related members of the species B. cepacia, leading to the genetic acquisition of environmentally appropriate adhesion-based colonization pilus operons. of 63 isolates from these four sources and likewise determined the phylogenetic relationships both among isolates from a common source and between isolates from different sources. Given the established precedent for the role of appendage pili as mediators of adhesion to the eukaryotic cell surface (5, 11, 19, 26–28, 36, 38, 45), the findings presented in this report suggest that future studies characterizing pilus adhesion-based colonization by B. cepacia need to account for which of these five specific classes, or others as yet unidentified, mediate a particular interaction. Our results also demonstrate that for the species B. cepacia, standard methods of sample preparation for electron microscopy are inadequate to resolve most of the structurally variant classes of pili described in this report. The artifacts associated with such methods typically lead to the appearance of polar pili when involved appendages are actually peritrichously arranged, or they may result in the total loss of the more fragile classes of pili.
In a companion report (46), we describe the purification and immunolocalization of a 17-kDa major subunit pilin protein of a cystic fibrosis (CF)-associated Burkholderia (formerly Pseudomonas) cepacia isolate and reveal that it is peritrichously assembled into novel, giant cable-like appendage (Cbl) pilus fibers. Also described are the sequence of the encoding structural gene (cblA) and the deduced amino acid sequence of the encoded CblA major subunit pilin protein, and the observation that this protein is chromosomally encoded among B. cepacia isolates is made. While carrying out the above-mentioned study (46), we observed peritrichous coexpression of a second, morphologically distinct class of extremely fine, steel wool-like mesh (Msh) pilus fibers from all 14 B. cepacia isolates expressing Cbl pili. Because all of the involved isolates were obtained from patients at a particular CF center in Toronto, we expanded the survey, using high-resolution transmission electron microscopy to determine whether Cbl and Msh pili were likewise coexpressed under identical growth conditions by B. cepacia isolates from patients at five other CF centers in North America and Europe as well as by clinical (non-CF) and environmental isolates. An initial survey of isolates from CF epidemic, CF nonepidemic, clinical, and environmental sources revealed that at least five distinctly different structural classes of peritrichous appendage pilus fibers are expressed by members of the species B. cepacia. Because different pilus types appeared to correlate with particular isolate sources, we examined a larger set
MATERIALS AND METHODS Bacterial isolates. A total of 63 Burkholderia (Pseudomonas) cepacia isolates were cultured by standard methods (18): 47 CF-associated (14 epidemically transmitted and 33 nonepidemic) isolates plus 6 from clinical sources and 10 from environmental sources. Single colonies picked from solid agar plates were grown to mid-log phase, and cultures were frozen at 2708C in Luria broth (LB) plus 10% glycerol. Table 1 presents information as to sources of the isolates and their relevant phenotypes and genotypes. Growth conditions. As described in Results, strains were analyzed by electron microscopy following both growth in liquid medium and growth on solid medium. LB with and without 0.4% glucose and brain heart infusion were tested for each condition. Testing of growth in liquid involved both mid-log-phase and stationary-phase cultures. Electron microscopy. High-resolution electron microscopy was carried out by using 400-mesh copper grids coated only with a thin layer of carbon. Grids were glow discharged immediately before use to ensure uniform spreading of bacterial
* Corresponding author. Mailing address: Section of Molecular Genetics, Maxwell Finland Laboratory for Infectious Diseases, 774 Albany St., Boston, MA 02118. Phone: (617) 638-5328. Fax: (617) 5345280. 1039
1040
GOLDSTEIN ET AL.
J. BACTERIOL.
TABLE 1. B. cepacia isolates Strain no.
Source and relevant phenotypic and/or genotypic information
PFGE typeb
Ribotypea
Source and/or reference(s)
BCT-5c, -7c, -18, -19, -21, -23, -24, -25, CF-associated isolates, epidemically transmitted, carrying the 1 -29, -36, -37, -38, -42, -61 cblA gene, from a Toronto CF center (The Hospital for Sick Children)
1
46, 53, 54
BCNC-566, -792, -1711, -1845, -1823, -1873, -1903, -1910, -1948, -1963c, -1985, -2008c, -2034, -2125, -2211, -2221, -2297, -2299, -2308, -21618
CF-associated isolates, nonepidemic (i.e., low-level transmis- 11–29 sibility), from a Chapel Hill, North Carolina, CF center (University of North Carolina); each with a unique ribotype and PFGE type, except for BCNC-792 and -2299, which had the same ribotype but significantly different (D , 0.3) PFGE types
11–30
51
BCMS-2323c
CF-associated isolate, carrying the cblA gene, from a Jackson, Mississippi, CF center (University of Mississippi)
50
R. L. Nolan and P. Gilligan; unpublished data
BCOH-521, -523, -524c, -525c
CF-associated isolates, nonepidemic, from a Cleveland, 70–72 Ohio, CF center (RainBow Babies & Children’s Hospital) (isolates BCOH-523 and -524 have identical ribotypes and PFGE types)
70–72
R. Stern; references 32 and 33 and unpublished data
BCPA-S2, -BC213
CF-associated isolate, nonepidemic, from a Philadelphia, 90 Pennsylvania, CF center (St. Christopher’s Hospital), with polar piliated derivative BCPA-BC213 selected by serial passage of BCPA-S2
90
M. Cervin, A. Smith, and P. Gilligan; reference 29 and unpublished data
BCPA-520, -542, -543, -544
CF-associated isolates, nonepidemic, from a Philadelphia, Pennsylvania, CF center (St. Christopher’s Hospital) (isolates BCPA-520 and -543 have identical ribotypes and PFGE types
91–93
91–93
References 32 and 33 and unpublished data
BCE-SB27, -SB29c
CF-associated isolates, nonepidemic, from CF center in Edinburgh, United Kingdom
120–121 120–121 J. R. W. Govan and J. Forstner; unpublished data
BCMA-2710c, -3129
Clinical isolates, pneumonia associated, with identical ribotypes and PFGE types; nosocomial transmission at Boston City Hospital, Boston, Massachusetts
200
200
Daniel Shapiro; unpublished data
BCATCC-13945c
Clinical isolate, human septic endocarditis, ATCC
210
210
ATCC; 51
BCATCC-17765
Clinical isolate, human urinary tract infection, ATCC
211
211
ATCC; 51
BCATCC-27515
Clinical isolate, human tibial fracture infection, ATCC
212
212
ATCC; 51
BCATCC-25609
50
Clinical isolate, human bronchial lavage sample, ATCC
213
213
ATCC; 51
BCATCC-25416
c
Environmental isolate, onion, ATCC B. cepacia type strain
300
300
ATCC; 51
BCATCC-39277
c
Environmental isolate, cornfield soil, ATCC
301
301
ATCC; 51
BCATCC-17759
Environmental isolate, forest soil, ATCC
302
302
ATCC; 51
BCATCC-29352
Environmental isolate, soil, ATCC
303
303
ATCC; 51
BCE-J2395
Environmental isolate, plant, Edinburgh, United Kingdom
315
315
J. R. W. Govan and J. Forstner; 20
BCE-J759
Environmental isolate, onion, Edinburgh, United Kingdom
325
325
J. R. W. Govan and J. Forstner; 20
BCMA-On1, -On2
Environmental isolates, onions, Boston, Massachusetts
330–331 330–331 Unpublished data
BCPA-On1, -On2
Environmental isolates, onions, Philadelphia, Pennsylvania
335–336 335–336 Unpublished data
a
According to standard criteria for B. cepacia (40), a ribotype RFLP is considered unique when comparison of sizes of hybridizing fragments reveals four or more bands differing between two EcoRI-generated rrn RFLP patterns under comparison. For rrn RFLPs, given that EcoRI analysis of B. cepacia strains reveals 7 to 10 total hybridizing bands (Fig. 6), the unique ribotype distinction would correspond to a D (coefficient of similarity) of #0.79 for two strains compared. Typically, among the collection of independently isolated strains analyzed, RFLP comparison yielding D values of $0.80 represents closely related strains while distantly related strains commonly have D values of #0.50 with intervening values rarely observed. See ‘‘Statistical analysis’’ in Materials and Methods. b According to standard criteria for B. cepacia (51), a PFGE type is considered unique when comparison of sizes of SpeI-generated fragments yields a D of #0.60. Typically, a D of $0.90 was found for closely related strains, with intervening values not being observed and values between 0.5 and 0.6 being rare. Given the presence of 11 to 19 bands in the typical B. cepacia RFLP with SpeI digestion (Fig. 7), 10 of 11 bands or 17 of 19 bands are necessarily shared by two compared strains to produce a D of $0.90. See ‘‘Statistical analysis’’ in Materials and Methods. c Isolate for which ribotype and PFGE type RFLP profiles are displayed in Fig. 6 and 7, respectively.
VOL. 177, 1995 suspensions. For every isolate to be examined, grids with sample were prepared by each of two different methods: method a, application of 5 ml of sample suspension and negative stain (0.01% phosphotungstic acid), after which excess liquid was removed via evaporation at 378C, thus avoiding structural artifacts associated with surface tension shear forces generated by standard blotting methods for removal of excess liquid, and method b, nebulized spray drop deposition of sample and phosphotungstic acid stain (0.1%), a technique that spreads away loosely attached surface matter from the bacterial cell, leaving more evident underlying appendages (5). Specimens were then examined by using a minimal beam technique with a JEOL 100 CX transmission electron microscope at 40 to 80 kV with sample grids stabilized by using a liquid nitrogen cold finger apparatus. Chromosomal DNA isolation for hybridization analysis. Overnight cultures were diluted 10-fold in 10 ml of LB and incubated at 378C until they reached mid-log phase. Cells were pelleted, washed twice with 0.9% NaCl, and then resuspended in cold TE (10 mM Tris-HCl, 10 mM EDTA [pH 8.0]). Lysozyme was added to a final concentration of 20 mg/ml, and the solution was incubated at 378C for 30 min. Proteinase K (20 mg/ml in TE; Sigma) and sodium dodecyl sulfate were added to final concentrations of 20 mg/ml and 0.1%, respectively. The lysates were incubated overnight at 508C. Sarcosine-free acid (Sigma) was added to a final concentration of 2%, and the solution was mixed gently. The DNA was then purified by cesium chloride-ethidium bromide equilibrium density gradient centrifugation (56). DNA probes and hybridization. EcoRI rRNA (rrn) restriction fragment length polymorphisms (RFLPs) for phylogenetic analyses were analyzed by using an rrnB probe spanning the entire rrnB operon of Escherichia coli K-12 (6) as we previously reported for E. coli (3), Pseudomonas aeruginosa (25, 55), and B. cepacia (50, 51). Plasmid DNA was purified by the alkaline lysis method (7), followed by cesium chloride-ethidium bromide gradient centrifugation (56). The DNA restriction fragment used to generate the probe was separated by horizontal slab gel electrophoresis in 0.8% low-melting-point agarose. The relevant restriction fragment from a slice of the gel was radiolabeled in the agarose by random oligonucleotide priming with [a-32P]dCTP (Dupont, NEN Research Products) (15). Southern blot analysis was performed as described previously (3, 51). Dot blot-based epidemiological analyses (9) for the presence of the cblA pilin subunit gene are described in the companion paper (46). Probes for pap/prs, pil (fim), and afa adhesin operons of E. coli and the pil operon of P. aeruginosa were as we described previously (5, 25). PFGE methodology. Pulsed-field gel electrophoresis (PFGE) methods were modified from those which we previously reported (2, 51, 55). Cells were grown to early log phase in LB, harvested by centrifugation, washed with 1:1 TE buffer, resuspended in 1 ml of 10:1 TE buffer, and mixed with 1.2 volumes of melted 1% InCert agarose (Bio-Rad) in TE buffer. The mixture was dispensed into 120-ml insert molds (Pharmacia) and allowed to solidify on ice. Plugs were sliced and incubated in 20 mg of lysozyme per ml at 378C for 12 h. The lysozyme buffer was replaced with ESP buffer (0.5 M EDTA [pH 9], 1% sarcosyl, 200 mg of proteinase K [Sigma] per ml), and the plugs were incubated at 378C for 5 h and then washed with TE buffer for 4 h at 378C. Single plug slices were incubated with SpeI (Boehringer Mannheim) in restriction enzyme buffer for 2 h. Restriction fragments were separated by PFGE using a CHEF Mapper system (Bio-Rad) through 1% agarose (Bio-Rad; Molecular Biology Certified grade) in 0.53 TBE (13 TBE is 0.9 M Tris-HCl, 0.9 M boric acid, and 1.0 mM EDTA). Electrophoresis was performed for 24 h in 0.53 TBE buffer at 148C, with a field strength of 200 V (6 V/cm) and initial and final pulse times of 1.2 and 54 s, respectively. Lambda concatemers were used as DNA size standards. The gels were stained with ethidium bromide and photographed under UV transillumination with a Polaroid camera. Fragment sizes were determined, and the computer-generated translation of chromosomal fingerprint profiles was made into bar code format by using a Macintosh Quadra 950 and Gene Construction Kit (Texto) software. Statistical analysis. PFGE chromosomal fingerprints were initially compared according to the criteria of Prevost et al. (39). According to standard criteria for B. cepacia (40), ribotypes were considered equivalent when comparison of sizes of hybridizing fragments revealed three or fewer bands differing between the two patterns under comparison; subribotypes correspond to patterns differing by no more than one band. Quantitative pairwise comparison of RFLP patterns was accomplished by using the Dice coefficient of similarity (D) calculated by the formula D 5 2nxy/(n1 1 n2), where n1 is the total number of DNA fragments from strain X, n2 is the total number from strain Y, and nxy is the number of fragments identical in the two strains (10, 34). A coefficient of similarity for two PFGE RFLPs (D) of $0.90 represents closely related strains (51). Remarkably, values between 0.9 and 0.6 have not been observed, and values between 0.6 and 0.5 are rare, with distantly related strains typically showing D values of #0.40 (51). For rrn RFLPs, the shared ribotype distinction (see above and reference 40) typically corresponds to a D of $0.79, while distantly related strains show D values of #0.5. Comparisons of mean values were performed by Student’s t test by using a Systat program (Systat Inc.). Phylogenetic analysis. On the basis of the RFLP profiles of chromosomal regions encoding the five rRNA (rrn) operons of B. cepacia (52a), phylogenic relationships of the isolates were examined by the neighbor-joining method of Saitou and Nei (42). Confidence intervals on the tree topology were estimated by bootstrapping analysis (16).
VARIANT TYPES OF COEXPRESSED B. CEPACIA PILI
1041
RESULTS Pili of CF-associated, epidemically transmitted B. cepacia isolates revealed by electron microscopy. Initially, by using sample preparation method a (see ‘‘Electron Microscopy’’ above), epidemically transmitted, clonally related B. cepacia isolates from 14 patients at a Toronto CF center (46, 53) (Table 1) were examined by high-resolution transmission electron microscopy. By evaporating excess liquid (sample suspension and negative stain) from the specimen grid prior to examination with the electron microscope, shearing forces normally associated with the removal of excess liquids by standard blotting techniques are avoided, allowing resolution of fragile appendage structures otherwise sheared from the bacterial surface (5, 22, 49). As prepared by this method, all 14 of the B. cepacia isolates associated with the Toronto CF center epidemic were found to be covered with a novel, extremely dense matrix of very fine curly fibers in a mesh-like configuration (Fig. 1A) that peritrichously extended in all directions for vast distances (2 to 4 mm) from the bacterial cells. We refer to the proteinaceous, fibrillar structures making up this labyrinth as mesh (Msh) pili or B. cepacia type I (type IB. cepacia) pili. Occasionally, within this fine mesh could be seen significantly larger pilus fibers (Fig. 1B) that will be further described below. By using standard blotting procedures, .99% of the Msh pili were sheared from the bacterial surfaces, appearing instead at the peripheral regions of the specimen grid (results not shown). At the typically higher beam voltages (60 to 80 kV) used for electron microscopic identification of bacterial appendages (5, 19, 27, 29) and/or in the absence of thin carbon films and minimal beam analysis, the extremely narrow (ca. 0.5-nm diameter) Msh pili proved to be labile, preventing resolution of the individual pilus fibers making up the meshwork. When the alternative method b for sample preparation, that involving nebulized spray deposition of bacteria in suspension onto the electron microscope grid followed by evaporation of excess liquid, was used, loosely attached type IB. cepacia Msh pili spread away from the bacterial cells, revealing a previously occluded, underlying secondary network of significantly larger pili (Fig. 2). We refer to this second, novel class of appendages (type IIB. cepacia) as cable (Cbl) pili on the basis of the typical intertwining of individual fibers that can also be seen to radiate peritrichously for great distances (2 to 4 mm) from the bacterial cells. By using standard blotting procedures, .90% of the Cbl pili were sheared from the bacterial surfaces (results not shown). Characterization of the 17-kDa major subunit pilin protein of the type IIB. cepacia Cbl pilus and its encoding DNA sequence are described in the companion article (46). On the basis of a similar strategy, we are attempting to clone for sequencing purposes the encoding gene for what we have determined to be the 37-kDa type IB. cepacia Msh pilin subunit (57). Pili of CF-associated, non-epidemically transmitted B. cepacia isolates revealed by electron microscopy. To determine whether CF-associated B. cepacia isolates from geographically distant locales express novel classes of appendage pilus fibers similar to those resolved for the Toronto epidemic cluster (see above), we examined 33 nonepidemically related strains (51, 53) (Table 1) provided by four CF centers in North America (University of North Carolina [UNC] Medical Center, Chapel Hill, N.C.; University of Mississippi Medical Center, Jackson, Miss.; RainBow Babies & Children’s Hospital, Cleveland, Ohio; and St. Christopher’s Hospital, Philadelphia, Pa.) and one CF center in the United Kingdom (Western General Hospital, Edinburgh, Scotland). By using sample preparation
1042
GOLDSTEIN ET AL.
J. BACTERIOL.
FIG. 1. High-resolution electron micrographs showing Msh pili. (A) High-resolution electron micrograph of B. cepacia BCT-7, an epidemically transmitted isolate from a Toronto CF center (46, 53, 54). Radiating from the bacterial surface is a dense matrix of extremely fine, curly type I mesh (Msh) pili, typical of all 14 of the characterized isolates from this CF center. Bar, 0.1 mm. (B) Wide-field electron micrograph of B. cepacia BCT-7, an epidemically transmitted isolate from a Toronto CF center (46, 53, 54). Peritrichously expressed from the surfaces of the bacteria and carpeting spaces between these cells is a dense matrix of steel wool-like type I mesh (Msh) pili. In the central region of the field, partially buried within this meshwork, can be seen significantly larger fibers. Bar, 0.1 mm.
method a, all CF-associated isolates from these five distantly separated locations were found to express an extremely dense matrix of very fine curly fibers in a mesh-like configuration (data not shown), structurally equivalent to that displayed in Fig. 1 for the Toronto isolates. When these mats of Msh pili were deliberately released from the bacterial surface on the
basis of sample preparation method b (spray drop), underlying peritrichously arranged pilus fibers were revealed. With the exception of the single Jackson, Mississippi, CF patient isolate (BCMS-2323), for the remaining 32 isolates, these underlying appendages appeared to be morphologically quite different from the Cbl pili found beneath the Msh pilus mat of the
VARIANT TYPES OF COEXPRESSED B. CEPACIA PILI
VOL. 177, 1995
1043
FIG. 1—Continued.
epidemically related Toronto isolates (see above). No evidence of lengthy, intertwined cabling was seen. Rather, this third category of B. cepacia-encoded appendage pili (type IIIB. cepacia) consists of distinct, relatively rigid filaments (Fig. 3). These filamentous (Fil) pili are morphologically similar in length and width to the pap/prs-encoded pili of uropathogenic E. coli (5, 35). When the two above-described methods for sample preparation were used immediately prior to electron microscopy, all 32 independent isolates from four of the involved CF centers (Chapel Hill, N.C.; Cleveland, Ohio; Philadelphia, Pa.; and
Edinburgh, United Kingdom; Table 1) were found to coexpress both Msh (type IB. cepacia) and Fil (type IIIB. cepacia) pili but not Cbl (type IIB. cepacia) pili. Included here was a CFassociated B. cepacia isolate (BC213) that had been previously described as expressing only polar, filament-type pili (29). When this strain was prepared for electron microscopy by using standard blotting procedures to remove excess sample and stain, .99% of the Msh pili were sheared away from bacterial surfaces, appearing to the far side of the electron microscope grid from which excess sample had been blotted. Likewise, with this latter procedure .95% of the Fil pili
1044
GOLDSTEIN ET AL.
J. BACTERIOL.
FIG. 2. Electron micrograph of B. cepacia BCT-7, an epidemically transmitted isolate from a Toronto CF center (46, 53, 54). Peritrichously radiating from the surfaces of the bacterial cells are type II adhesin cable (Cbl) pili, those typical of all 14 of the characterized isolates from this CF center. Lower magnification indicates that cable appendages extend for vast distances (2 to 4 mm) on the microbial scale and likewise that they appear to tether together large aggregates of Cbl pilus-expressing bacteria. Cloning and sequencing of the cblA pilin-encoding gene are described in an accompanying article (46), and the mucin-binding properties of Cbl pili were described by Sajjan et al. (44). As presented in Results and Materials and Methods, resolution of Cbl pili was made possible by removal of the type I Msh pili shown in Fig. 1. Bar, 0.1 mm.
VOL. 177, 1995
VARIANT TYPES OF COEXPRESSED B. CEPACIA PILI
1045
FIG. 3. Electron micrograph of B. cepacia BCNC-1963, a UNC Medical Center CF isolate of low transmissibility (51, 53). Peritrichously arranged type III filamentous (Fil) pili radiate from the bacterial surface. Type III pili were typically associated with CF isolates of low transmissibility. Two significantly larger flagella can be seen in the upper portion of the micrograph. As described in Results and Materials and Methods, resolution of the type III Fil pili was made possible by removal of the type I Msh pili shown in Fig. 1. Bar, 0.1 mm.
proved to be susceptible to surface tension shearing forces associated with blotting, leaving the remaining attached filaments in an artifactual polar arrangement (results not shown) similar to that which was reported previously (29).
Pili of clinical (non-CF) B. cepacia isolates revealed by electron microscopy. In order to determine if Msh-, Cbl-, and/or Fil-type pili (see above) are also typically expressed by nonCF-associated clinical B. cepacia isolates, we applied the two
1046
GOLDSTEIN ET AL.
FIG. 4. Electron micrograph of endocarditis-associated B. cepacia BCATCC13945 expressing peritrichously arranged type IV spine (Spn) pili. Type IV pili were typically associated with clinical (non-CF) isolates. Bar, 0.1 mm.
methods of sample preparation for analysis of six strains obtained from diverse human infections or sources, including a urinary tract infection, a bronchial lavage sample, endocarditis, osteomyelitis, and pneumonia (Table 1). With the exception of the human tibia fracture isolate (BCATCC-27515), for each of the other non-CF clinical isolates, when either preparative method a or b was used, only a single structural category of peritrichously arranged appendage pili was observed. These pili proved to be distinct from any of the morphological pilus types correlated with either of the categories of CF-associated isolates. We refer to these peritrichously arranged, densely packed type IVB. cepacia appendages as spine (Spn) pili on the basis of their appearance as relatively rigid, sea urchin-like fine spines (Fig. 4). The osteomyelitis isolate BCATCC-27515 displayed no pili when prepared by method a or b. Alternatively, when the other clinical type IVB. cepacia pilus-expressing strains were initially prepared for electron microscopy by the standard blotting method, .90% of the type IVB. cepacia Spn pili sheared from bacterial surfaces (results not shown).
J. BACTERIOL.
Pili of environmental B. cepacia isolates revealed by electron microscopy. For comparison with the results obtained for B. cepacia isolates from CF and other clinical sources (see above), environmental strains were also analyzed via electron microscopy. As indicated in Table 1, these included three independently isolated American Type Culture Collection (ATCC) B. cepacia strains from different soil sources and one strain isolated from a plant. In addition, because onions and other vegetables commonly carry B. cepacia and have been alluded to as a potential source for cross-spread to the CF population (17), we also examined the ATCC B. cepacia type strain (an onion isolate) plus other onion isolates from Boston, Philadelphia, and Edinburgh (Table 1). When sample preparation method a was used, all environmental isolates but BCATCC-39277 (cornfield soil) and BCATCC-25416 (onion; the ATCC type strain) were found to be covered by a dense meshwork of fine, curly fibers structurally equivalent to that described above for type IB. cepacia Msh pili (Fig. 1). When observed following the performance of sample preparation method b, these two isolates likewise displayed no resolvable appendages. However, when the other eight isolates were examined by electron microscopy following the performance of sample preparation method b, removal of the dense mats of Msh pili revealed an underlying single structural category of peritrichously arranged appendage pili. The involved class of appendages (type VB. cepacia) associated with these environmental isolates was structurally different from those described above, with a relatively short, broad, spikelike morphology (Fig. 5). On the basis of their morphology, we refer to these seemingly thatched, densely packed appendages as spike (Spk) pili. Unlike the case with type I through IVB. cepacia pili (see above), standard preparative procedures involving blotting of excess sample and stain from the electron microscope grid did not shear the type VB. cepacia pili from the bacterial surface. Effect of growth conditions and media on categories of expressed pili. Because regulation of the expression of pilus appendages of other bacterial species such as E. coli has been found to be dependent on growth conditions and media (1, 12, 13, 24), the four categories of B. cepacia described above were grown at 378C both on solid agar (1.2%) and in liquid broth (plus and minus aeration). Both LB minimal (LB min) medium (with and without 0.4% glucose) and brain heart infusion were tested as liquid and solid media. For 61 of the 63 isolates, no change in the piliation profiles described above was detected when strains were analyzed by electron microscopy following repeated growth on either type of solid medium. Only a single Toronto CF center isolate (BCT-61) proved to be sensitive to a change of solid medium, expressing both Msh and Cbl pili when grown on brain heart infusion but not LB agar, as did a single UNC isolate (BCNC1910) expressing Msh and Fil pili when grown on brain heart infusion but not LB agar. In contrast, for the isolates expressing type IB. cepacia Msh-type pili, aerated growth to mid-log phase by agitation in liquid medium of either type yielded cultures of bacteria with ,5% of the cells covered with mats of Msh pili. When supernatants from such cultures were pelleted and examined by electron microscopy, mats of Msh pili were resolved (results not shown). Similar significant fracturing of surface appendage fibers was found when strains expressing type IIB. cepacia Cbl or type IVB. cepacia Spn pili were grown with agitation in liquid media. While the presence of glucose did not affect the state of piliation, it did, however, in all cases repress expression of polar flagella otherwise observed on ca. 40% of the isolates when grown on solid media and on 95% of the isolates when grown in aerobic liquid media.
VOL. 177, 1995
VARIANT TYPES OF COEXPRESSED B. CEPACIA PILI
1047
FIG. 5. Electron micrograph of environmental (soil) B. cepacia isolate BCATCC-29352 expressing peritrichously arranged type V spike (Spk) pili. Type V pili were typically associated with environmental isolates. As described in Results and Materials and Methods, resolution of the type V Spk pili was made possible by removal of the type I Msh pili shown in Fig. 1. Bar, 0.1 mm.
Hybridization-based surveys for the cblA major pilin subunit gene and for homologies to pilus operons of P. aeruginosa and E. coli. In the companion article (46) we describe both the isolation and the DNA sequence of the cblA major subunit gene of the type IIB. cepacia pilus. Use of this gene as a probe in a dot blot epidemiological survey of all strains phenotypically analyzed as described above by electron microscopy revealed a
100% correlation with the Cbl pilus phenotype results in the present report, i.e., among the 63 B. cepacia isolates so surveyed, only and all of the 14 Toronto CF center epidemically associated strains plus the single Jackson, Mississippi, CF center isolate (BCMS-2323) contained the cblA gene (reference 46 and unpublished results). To determine whether there existed DNA sequence homol-
1048
GOLDSTEIN ET AL.
J. BACTERIOL.
TABLE 2. Typical piliation phenotypic profiles revealed by electron microscopy Result for pilus structural type (acronym)a: Isolate source
CF epidemic CF nonepidemic Non-CF clinical Environmental a b
I (Msh)
II (Cbl)
III (Fil)
IV (Spn)
V (Spk)
1 1 2 1
1b 2b 2b 2b
2 1 2 2
2 2 1 2
2 2 2 1
1, occurrence of piliation phenotype; 2, no occurrence. Confirmed by hybridization-based genotypic analysis.
ogy between the genes coding for the major pilin subunits of type IB. cepacia (Msh) pili, type IIB. cepacia (Cbl) pili, type IIIB. cepacia (Fil) pili, type IVB. cepacia (Spn) pili, and/or type VB. cepacia (Spk) pili and the pilin genes coding for appendage pili commonly expressed by E. coli and P. aeruginosa, we carried out dot blot hybridization analyses for all 63 of the B. cepacia strains analyzed by electron microscopy (see above). Each proved to be negative for probes that we had previously constructed and used in epidemiological surveys of pil (fim), pap/prs, and afa pilus operons of E. coli (4, 5) as well as the pil operon of P. aeruginosa (25, 47) (data not shown). Genetic relatedness of the isolates expressing different classes of appendage pili. Given the apparent phenotypic correlation of a unique pilus type with each of the four sources of our B. cepacia isolates (Table 2), we asked whether evolutionary lineage correlates with one or another of these types of pili. For the purpose of this investigation, the degree of genetic relatedness among independent isolates within a species such as B. cepacia can be rapidly determined by using either of two commonly applied approaches: (i) ribotyping analyses, i.e., the determination of RFLPs associated with the multicopy rRNA operon (rrn) (33, 40, 50, 51), or (ii) PFGE-based resolution of chromosomal macro-restriction fragment patterns (50, 51). The set of 14 epidemically associated isolates from the Toronto CF center found to chromosomally encode the cblA gene and to constitutively express type IIB. cepacia Cbl pili (see above) proved to be of identical ribotype lineage. Among the remaining 49 isolates, only four other ribotype profiles were found to be shared by geographically associated pairs of independent isolates. With the exception of the epidemically related Toronto strains plus these eight isolates, EcoRI-generated, multicopy rrn hybridization patterns revealed only the predictable, conserved set of two identical, B. cepacia species signaturespecific bands (at 4.2 and 2.6 kbp) (21, 37, 51) occurring among all of the examined RFLP profiles (Fig. 6). The other five to eight hybridization bands per RFLP profile defined for all of the latter isolates relatively unique ribotype patterns, with a mean D (coefficient of similarity) of 0.39 6 0.13 (range, 0.24 to 0.66). A comparative, typical set of prototypically variant profiles for all four of the different isolate sources (CF epidemic, CF nonepidemic, clinical, and environmental) is displayed in Fig. 6. With the cited exceptions, isolates within the CF nonepidemic, clinical, and environmental source groups differed by ribotype to an extent equivalent to that found for differences between members of the different source groups. Thus, for these three groups, while there appeared to be phenotypic correlations between isolate source and expressed pilus type (Table 2), the common degrees of RFLP variability among and between these three diverse isolate source groups indicate that distinct evolutionary lineages correlating with these phenotypes do not exist. On the basis of PFGE-resolved SpeI chromosomal RFLPs,
FIG. 6. Typically variant ribotype (rrn RFLP) profiles obtained for B. cepacia isolates from different sources (CF epidemic, CF nonepidemic, clinical, and environmental). The isolate number appears at the top of each lane. The source appears above isolate number. Lanes 1 and 2, epidemically transmitted isolates from the Toronto CF center (46, 53, 54); lanes 3 through 8, isolates of low transmissibility (51, 53) from four different CF centers (lane 3, Edinburgh, Scotland; lanes 4 and 5, Chapel Hill, North Carolina; lanes 6 and 7, Cleveland, Ohio; lane 8, Jackson, Mississippi); lanes 9 and 10, clinical (non-CF) isolates (lane 9, isolate from a pneumonia patient on a ventilator, Boston, Massachusetts; lane 10, endocarditis isolate from the ATCC); lanes 11 and 12, environmental isolates (lane 11, onion isolate, ATCC B. cepacia type strain; lane 12, ATCC cornfield soil isolate). Cited isolates are described in Results and Table 1.
banding patterns consisted of 11 to 19 DNA fragments. With the exception of the epidemically associated Toronto strains and three of the four pairs of isolates with identical ribotypes mentioned above, all of the CF nonepidemic, clinical, and environmental isolates possessed distinct profiles, with a mean D of 0.2 6 0.09 as determined by pairwise comparison. A comparative, typical set of such prototypically variant profiles for all four of the different isolate sources (CF epidemic, CF nonepidemic, clinical, and environmental) is displayed in bar code format in Fig. 7. Similar variability was revealed by using restriction endonuclease AseI (data not shown). For one of the four strain pairs with identical ribotypes mentioned above, PFGE chromosomal fingerprinting produced a D of ,0.38 6 0.20. Thus, as previously found for isolates of independent origins, PFGE-based chromosomal RFLPs identified strains as distinct, while ribotyping occasionally characterizes two totally unrelated isolates as sharing the same ribotype (51). Notwithstanding this limitation of ribotype analysis, both assays of genetic relatedness demonstrated similar pictures of the heterologous nature of all of the nonepidemic isolates, i.e., despite the apparent correlation between pilus type and isolate source, no correlation with evolutionary lineage was found to exist. DISCUSSION Numerous studies of adhesion-based colonization by different species of gram-negative bacteria have contributed to our current perspective on the common role of otherwise morphologically variant classes of appendage pili (fimbriae) as mediators of adhesion to the assortment of receptors on the sur-
VOL. 177, 1995
FIG. 7. Bar code translation of typically variant PFGE-resolved chromosomal macro-RFLP profiles obtained for B. cepacia isolates from different sources (CF epidemic, CF nonepidemic, clinical, and environmental). The same set of isolates is characterized for rrn RFLPs in Fig. 6. The isolate number appears at the top of each lane. The source appears above the isolate number (see the legend to Fig. 6). Fragments smaller than 100 kbp are not shown. In this range, Toronto isolates (lanes 1 and 2) had two identical fragments (60 and 48 kbp). Other isolates (lanes 3 through 12) had polymorphic sets of two to six fragments in this low range.
faces of plant and animal tissues. Our study, the first broad analysis of B. cepacia isolates from diverse sources, demonstrates that members of this species have the capacity to phenotypically express at least five significantly different structural classes of appendage pilus fibers, several of which may be expressed simultaneously (Table 2). (i) Visualization of the piliated state by electron microscopy. Our results indicate that the standard procedures developed for the spreading of sample suspension and negative stain on the electron microscope grid and for the subsequent removal of excess liquid sheared four of the five structurally distinct classes of B. cepacia pili. They further demonstrate that in the absence of more stringent methods (to minimize surface tension forces), otherwise peritrichously piliated bacteria may appear either solely with polar pili or with no pili at all. This observation explains why we were able to visualize the peritrichous coexpression of both Msh type IB. cepacia pili and Fil type IIIB. cepacia pili from a CF-associated strain reported to solely express polar Fil type IIIB. cepacia appendages in the only previous electron microscopic publication concerning the pili of B. cepacia (29). Observed piliation phenotypes have been found to be affected by environmental signals that regulate expression. Unlike the cases for several different well-characterized pilus operons of E. coli (1, 12, 13, 19, 24), we observed no significant changes in states of expression for any of the five classes of pili described in this report based on different media, liquid versus solid media, or degrees of aeration. In contrast, all four source classes of isolates displayed repression of flagellum expression when grown in the presence of glucose.
VARIANT TYPES OF COEXPRESSED B. CEPACIA PILI
1049
(ii) Isolate source versus piliation state. Epidemiological analysis for phenotypic expression suggests that four of the five structurally distinct classes of appendage pili uniquely correlate with the source from which strains expressing the appendages were isolated. While an analogous, extensive genotypiclevel survey is necessary to confirm this hypothesis, initial genotypic results presented in the companion paper (46), based on the use of a cblA gene probe, are in accordance with these phenotypic observations (Table 2). (iii) Isolate source versus genetic lineage. While DNA sequence information provides the most powerful approach to defining molecular evolutionary relationships, it is often too laborious a methodology for extensive surveys of bacterial populations. Use of the more practical alternatives, rrn RFLP (i.e., ribotyping) or PFGE-resolved chromosomal macro-RFLP profiles, provides a convenient tool for this purpose, allowing the relatively rapid characterization of the phylogenetic relationships of microorganisms (2, 3, 14, 21, 30, 41, 51, 52). With these methods, differences and similarities in the numbers and the sizes of fragments hybridizing to the rrn probe or differences and similarities in the PFGE-resolved chromosomal RFLP patterns can identify specific phylogenic lineage patterns. Given that our initial results revealed a correlation between the source of the isolate and the phenotypic class(es) of expressed pili (Table 2), we analyzed the genetic relatedness of isolates from each of these four types of sources (CF epidemic, CF nonepidemic, clinical, and environmental). With the exception of epidemically transmitted Toronto strains, rrn RFLP (ribotype) and PFGE-resolved chromosomal RFLP profiles of multiple isolates from a common source or locale typically displayed degrees of polymorphism similar to those found by cross-comparison of random isolates among these groups, a picture matching that previously reported for CF-associated B. cepacia isolates from clinic and lung transplant patients (51). This degree of variation is significantly greater than that found for very closely related or clonal isolates that have undergone minor evolutionary changes, for which typically D (coefficient of similarity) is $0.9 (51). In order to further examine this initial picture, we applied the Nei and Li mathematical model (34), developed for quantitative analysis of nucleotide substitutions from restriction fragment data. This allowed us to employ our ribotype RFLP data to investigate the implication that independently of the piliation phenotype, no particular bacterial genetic lineages could be associated with sources of isolates, i.e., their phylogenetic relationship would be no closer than that for a random sample of B. cepacia. By using the rrn RFLP data, phylogenetic relationships among the isolates were examined by the neighbor-joining method of Saitou and Nei (42). Confidence intervals on the resultant phylogenetic tree topology were estimated by bootstrapping analysis (16). The most compelling feature noted was the unrooted shallowness of this tree, indicating that the isolates from CF nonepidemic, clinical, and environmental sources were as polymorphic within these source categories as they were when comparison between isolates of different source categories was made (40a), a finding resembling that for E. coli strains from various extraintestinal infections (3, 8, 48). This picture of phylogenic heterology, when considered together with the above-described pilus type phenotypic correlation with isolate source, appears to be equivalent to that which we demonstrated for the chromosomally encoded pap/ prs adhesin pilus operon of uropathogenic E. coli, i.e., horizontal transfer under selective pressure likely accounts for the association of specific pilus types with isolates of varying evolutionary lineage (3).
1050
J. BACTERIOL.
GOLDSTEIN ET AL.
(iv) Structure and function of Cbl pili. Results presented in the companion article (46) reveal that the CblA subunit peptide encoded by epidemically transmitted CF-associated isolates is polymerized into peritrichous appendages that are unusual in length (2 to 4 mm) and morphology. Typically, individual fibers are braided into cable-like quaternary configurations, reticulated into a labyrinth of webs composed of entangled cables from adjacent cells (Fig. 2). To our knowledge, type IIB. cepacia Cbl appendage structures of this unusual configuration have not been previously reported. Evolution of such an unusual morphology could be associated with genetic selection for optimal colonization of the equally unusual environment of the lungs of patients with CF (CF lungs). Mucus-glycoprotein-carbohydrate molecules, known to act as receptors for Cbl pili (43–45), are typically found in excess in the CF airway, coating the underlying epithelial cells. Maximal efficiency of binding to receptors presented in this unusual form, as a relatively homogeneous, glycoprotein-mucus biofilm, would likely necessitate some morphological variation of the significantly shorter, noncabled adhesin pilus appendages observed commonly for other bacterial pathogens, those associated with binding to specific receptors located within the heterogeneous repertoire of molecules on the eukaryotic cell surface. Thus, in contrast to that required for binding to unique sites on the complex eukaryotic cell surface, efficient adhesion to a homogeneous, relatively ubiquitous receptor biofilm layer would likely call for variation in the typical adhesin-pilus structure-function paradigm. Our previous results support the proposed model. These studies show that Cbl pilus adhesion does not involve typical binding through an adhesin moiety located uniquely at the appendage tip as is the case for the pap/prs and pil operonencoded pili of E. coli (23, 31). Rather, the adhesin subunit of the Cbl pilus is dispersed along the length of the structure, as is the corresponding binding of the mucin receptor (43, 45). Such a significantly variant structural arrangement, one composed of unusually long appendage pilus fibers within which the adhesin subunit is dispersed, appears to maximize interaction with the homogeneous receptor biofilm layer coating the CF airway. Given this environmental niche, it would not be unreasonable to further postulate that the quaternary cabling of individual fibers, coupled with the further intertwining of cables from adjacent bacterial cells, has an important function in stabilized adhesion through these otherwise potentially fragile appendages of enormous length. Combined results of both molecular epidemiological and electron microscopic surveys likewise strongly support the hypothesis that the novel Cbl pili described in this report have evolved for colonization-based adhesion in the unusual biofilm environment of the CF lung. As described here and in the accompanying article (46), characterization of environmental, clinical, and CF-associated B. cepacia isolates indicates that while all of these strains express one or more structurally variant types of pilus appendages, only highly infectious epidemically transmitted CF-associated strains were found to have the genetic capacity to express Cbl adhesin pili. Whether equally virulent B. cepacia isolates from a similar case of epidemic transmission in a distant Edinburgh (United Kingdom) CF center (20) prove to likewise express type IIB. cepacia Cbl pili should provide an independent in vivo test for the correlation of these pili with strain epidemic transmissibility between CF patients and virulence in the CF lung. This picture for B. cepacia contrasts dramatically with the results of similar studies of the adhesin pili expressed by CFassociated P. aeruginosa isolates. In the latter case, an identical structural type of shorter, noncabled pili was found to be ex-
pressed from a chromosomally encoded pilus operon that was commonly carried by P. aeruginosa isolates from diverse sources, including environmental, clinical, and CF-associated strains (47, 55). (v) Structure, function, and coexpression of Msh pili. The type IB. cepacia mesh (Msh) pili discovered in this study (Fig. 1A and B) represent a second, morphologically unusual class of appendages. Apparent convolution of these extremely fragile, proteinaceous fibers into extraordinarily dense, distantly radiating mats may in part be made structurally possible by what we have found to be the consistent coexpression of peritrichously expressed, underlying larger networks of either type IIB. cepacia Cbl, type IIIB. cepacia Fil, or type VB. cepacia Spk pili. The possible association of type IB. cepacia Msh pili with these other appendages is further suggested by the observation that the distance from the cell surface covered by the Msh pilus mat for all three cases correlates with the length of the underlying network of larger fibers, e.g., where type IIB. cepacia Cbl appendages of extraordinary length were present (Fig. 1B and 2), the overlying Msh pilus network extended to an equally distant degree, whereas when far shorter type VB. cepacia Spk pili were present (Fig. 5), the Msh network was always restricted to an equal, significantly more limited distance (results not shown). Using methods analogous to that for characterization of CF-associated type IIB. cepacia Cbl adhesin pili (43, 45, 46), we are currently determining both the receptor specificity of Msh pili and the DNA sequences involved in the assembly of the type IB. cepacia pilus. This should allow us to determine precise function and to genotypically test the above-described phenotypic picture at the level of chromosomal correlation of genes encoding type IB. cepacia Msh pili with genes encoding CFassociated type IIB. cepacia Cbl adhesin pili. Construction of otherwise isogenic sets of strains, one of which carries a cblA pilin gene mutation and another of which contains an mshA pilin gene mutation, will be useful to determine (i) whether fine mats of Msh pili require an underlying support network of Cbl pili, (ii) whether the apparent coexpression of these classes of pili is controlled by a single locus, (iii) whether both classes of pili function in consort, and (iv) how the two structures involved might interact to promote efficient adhesion-based colonization of the CF lung. Preliminary new results strongly suggest that either or both of these appendages likely enhance the infectivity of the highly transmissible, epidemically spread strains. Based on binding studies with human, primary culture cftr2/2 airway epithelial cells (56a), these results reveal that (i) strains coexpressing both Cbl and Msh pili proved to be significantly more adherent than did CF-associated strains expressing both Fil and Msh pili or environmental strains expressing both Spk and Msh pili and (ii) the Cbl and Msh pilus-coexpressing strains adhered to the cilia of airway epithelial cells as well as to their apical surfaces. Given that Cbl pili were previously found to bind CF mucins (43, 45), this latter result implies that strains of B. cepacia expressing such pili may interfere with the pulmonary mucociliary transport system by cross-linking mucin to cilia otherwise responsible for the essential clearance of mucin biofilms. ACKNOWLEDGMENTS We thank Gunna Christiansen, Maggie Cervin, Suzanne Steinbach, Alison Holmes, and James Yankaskas for useful discussions, and we thank Aram Chobanian, Peter Rice, Howard Corwin, and Robert Beall (CF Foundation) for encouragement to initiate these studies. Kathy Brown, Richard Kozak, and Daniel Shaipiro are acknowledged for clinical microbiological assistance, and Heather Goldstein, Ethan Johnson, and Mario Benavides are acknowledged for technical support. P. Gilligan, D. Shapiro, T. Stull, M. Cervin, A. Smith, J. Govan,
VARIANT TYPES OF COEXPRESSED B. CEPACIA PILI
VOL. 177, 1995 S. Lory, and R. Nolan are acknowledged for providing bacterial isolates or DNA probes. This study was funded by a research grant from the Cystic Fibrosis Foundation (USA) and by NIH Biomedical Research Support through Boston City Hospital. REFERENCES 1. Abraham, J., C. Freitag, J. Clements, and B. Eisenstein. 1985. An invertible element of DNA controls phase variation of type 1 fimbriae of Escherichia coli. Proc. Natl. Acad. Sci. USA 82:5724–5727. 2. Arbeit, R., M. Arthur, R. Dunn, R. K. Selander, and R. Goldstein. 1990. Resolution of recent evolutionary divergence among Escherichia coli from related lineages: the application of pulsed field electrophoresis to molecular epidemiology. J. Infect. Dis. 161:230–235. 3. Arthur, M., R. Arbeit, C. Kim, P. Beltran, H. Crowe, S. Steinbach, C. Campanelli, R. Wilson, R. Selander, and R. Goldstein. 1990. Restriction fragment length polymorphisms among uropathogenic Escherichia coli isolates: pap-related sequences compared with rrn operons. Infect. Immun. 58:471–479. 4. Arthur, M., C. Campanelli, R. Arbeit, C. Kim, S. Steinbach, C. Johnson, R. Rubin, and R. Goldstein. 1989. Structure and copy number of gene clusters related to the pap P-adhesin operon of uropathogenic Escherichia coli. Infect. Immun. 57:314–321. 5. Arthur, M., C. E. Johnson, R. H. Rubin, R. D. Arbeit, C. Campanelli, C. Kim, S. Steinbach, M. Agarwal, R. Wilkinson, and R. Goldstein. 1989. Molecular epidemiology of adhesin and hemolysin virulence factors among uropathogenic Escherichia coli. Infect. Immun. 57:303–313. 6. Berg, K., C. Squires, and C. Squires. 1987. In vivo translation of a region within the rrnB 16S rRNA gene of Escherichia coli. J. Bacteriol. 169:1691– 1701. 7. Birnboim, H. 1983. A rapid alkaline extraction method for the isolation of plasmid DNA. Methods Enzymol. 100:243–255. 8. Caugant, D., B. Levin, G. Lidin-Janson, T. Whittam, C. Svanborg-Ede´n, and R. Selander. 1983. Genetic diversity and relationships among strains of Escherichia coli in the intestine and those causing urinary tract infections. Prog. Allergy 33:203–227. 9. Davis, I., M. Dibner, and J. Battey. 1986. Basic methods in molecular biology, p. 388. Elsevier Publishers, New York. 10. Dice, L. 1945. Measures of the amount of ecological association between species. Ecology (New York) 26:297–302. 11. Doig, P., T. Todd, S. Parimi, K. Sastry, K. Lee, R. Hodges, W. Paranchych, and R. Irvin. 1988. Role of pili in adhesion of Pseudomonas aeruginosa to human respiratory epithelial cells. Infect. Immun. 56:1641–1646. 12. Eisenstein, B., and C. Dodd. 1982. Pseudocatabolite repression of type 1 fimbriae of Escherichia coli. J. Bacteriol. 151:1560–1567. 13. Eisenstein, B. I., D. S. Sweet, V. Vaughn, and D. I. Friedman. 1987. Integration host factor is required for the DNA inversion that controls phase variation in Escherichia coli. Proc. Natl. Acad. Sci. USA 84:6506–6510. 14. el-Adhami, W., W. Roberts, A. Gibbs, and P. Stewart. 1991. Epidemiological analysis of a methicillin-resistant Staphylococcus aureus outbreak using restriction fragment polymorphisms of genomic DNA. J. Gen. Microbiol. 137: 2713–2720. 15. Feinberg, A., and B. Vogelstein. 1984. A technique for radiolabeling DNA restriction endonuclease fragments to high specific activity. Anal. Biochem. 137:266–267. 16. Felsenstein, J. 1985. Confidence limits on phylogenies: an approach using the bootstrap. Evolution 39:783–791. 17. Fisher, M., J. LiPuma, S. Dasen, G. Caputo, J. Mortensen, K. McGowan, and T. Stull. 1993. Source of Pseudomonas cepacia: ribotyping of isolates from patients and from the environment. J. Pediatr. 123:745–747. 18. Gilligan, P. H., P. Gage, L. Bradshaw, D. Schidlow, and B. T. DeCicco. 1985. Isolation medium for the recovery of Pseudomonas cepacia from respiratory secretions of patients with cystic fibrosis. J. Clin. Microbiol. 22:5–8. 19. Giron, J., A. Ho, and G. Schoolnik. 1993. Characterization of fimbriae produced by enteropathogenic Escherichia coli. J. Bacteriol. 175:7391–7403. 20. Govan, J. R. W., P. H. Brown, J. Maddison, C. J. Doherty, J. W. Nelson, M. Dodd, A. P. Greening, and A. K. Webb. 1993. Evidence for transmission of Pseudomonas cepacia by social contact in cystic fibrosis. Lancet 342:15–19. 21. Grimont, F., and P. Grimont. 1986. Ribosomal ribonucleic acid gene restriction patterns as potential taxonomic tools. Ann. Inst. Pasteur (Microbiol.). 137B:165–175. 22. Gutmann, L., M. Agarwal, M. Arthur, C. Campanelli, and R. Goldstein. 1990. A phasmid shuttle vector for cloning of complex operons in Salmonella. Plasmid 23:42–58. 23. Hanson, M. S., and C. C. Brinton. 1988. Identification and characterization of E. coli type-1 pilus tip adhesion protein. Nature (London) 332:265–268. 24. Hultgren, S., W. Schwan, A. Schaeffer, and J. Duncan. 1986. Regulation of production of type 1 pili among urinary tract isolates of Escherichia coli. Infect. Immun. 54:613–620. 25. Jiang, R.-Z., L. Sun, A. Agodi, S. Steinbach, and R. Goldstein. 1992. Direct comparison of the epidemiological discriminatory power of ribotyping versus
1051
virulence factor probing of CF-associated P. aeruginosa. Pediatr. Pulmonol. S8:A194. 26. Johnson, K., and S. Lory. 1987. Characterization of Pseudomonas aeruginosa mutants with altered piliation. J. Bacteriol. 169:5663–5667. 27. Klem, P. 1985. Fimbrial adhesins of Escherichia coli. Rev. Infect. Dis. 7:321– 340. 28. Korhonen, T. K., K. Lounatmaa, H. Ranta, and N. Kuusi. 1980. Characterization of type 1 pili of Salmonella typhimurium LT2. J. Bacteriol. 144:800– 805. 29. Kuehn, M., K. Lent, J. Haas, J. Hagenzieker, M. Cervin, and A. Smith. 1992. Fimbriation of Pseudomonas cepacia. Infect. Immun. 60:2002–2007. 30. Lew, A., and P. Desmarchelier. 1993. Molecular typing of Pseudomonas pseudomallei: restriction fragment length polymorphisms of rRNA genes. J. Clin. Microbiol. 31:533–539. 31. Lindberg, F., B. Lund, L. Johansson, and S. Normark. 1987. Localization of the receptor-binding protein adhesin at the tip of the bacterial pilus. Nature (London) 328:84–87. 32. LiPuma, J., S. Dasen, D. Nielson, R. Stern, and T. Stull. 1990. Person-toperson transmission of P. cepacia between patients with cystic fibrosis. Lancet 336:1094–1096. 33. LiPuma, J., J. Mortensen, S. Dasen, and T. Stull. 1988. Ribotype analysis of P. cepacia from cystic fibrosis treatment centers. J. Pediatr. 113:859–862. 34. Nei, M., and W. H. Li. 1979. Mathematical model for studying genetic variation in terms of restriction endonucleases. Proc. Natl. Acad. Sci. USA 76:5269–5273. 35. Normark, S., D. Lark, R. Hull, M. Norgren, M. Båga, P. O’Hanley, G. Schoolnik, and S. Falkow. 1983. Genetics of digalactoside-binding adhesin from a uropathogenic Escherichia coli strain. Infect. Immun. 41:942–949. 36. O’Hanley, P., D. Lark, S. Normark, S. Falkow, and G. Schoolnick. 1983. Mannose-sensitive and gal-gal binding E. coli pili from recombinant strains: chemical, functional and serological properties. J. Exp. Med. 158:1713–1719. 37. Pace, N., G. Olsen, and C. Woese. 1986. Ribosomal RNA phylogeny and the primary lines of evolutionary descent. Cell 45:325–326. 38. Paranchych, W., P. Sastry, K. Vopel, B. Loh, and D. Speert. 1986. Fimbriae (pili): molecular basis of P. aeruginosa adherence. Clin. Invest. Med. 9:113– 118. 39. Prevost, G., B. Jaulhac, and Y. Piemont. 1992. DNA fingerprinting by pulsed-field gel electrophoresis is more effective than ribotyping in distinguishing among methicillin-resistant Staphylococcus aureus isolates. J. Clin. Microbiol. 30:967–973. 40. Rabkin, C., W. Jarvis, R. Anderson, T. Stull, and D. Woods. 1989. P. cepacia typing systems: collaborative study to assess their potential in epidemiologic investigations. Rev. Infect. Dis. 11:600–607. 40a.Riley, M., and R. Goldstein. Unpublished data. 41. Roux, V., and D. Raoult. 1993. Genotypic identification and phylogenetic analysis of the spotted fever group rickettsiae by pulsed-field gel electrophoresis. J. Bacteriol. 175:4895–4904. 42. Saitou, N., and M. Nei. 1987. The neighbor-joining method: a new method for reconstructing phylogenetic trees. Mol. Biol. Evol. 4:406–425. 43. Sajjan, S., and J. Forstner. 1992. Identification of the mucin-binding adhesin of P. cepacia isolated from patients with cystic fibrosis. Infect. Immun. 60: 1434–1440. 44. Sajjan, U., M. Corey, M. Karmali, and J. Forstner. 1992. Binding of Pseudomonas cepacia to normal human intestinal mucin and respiratory mucin from patients with cystic fibrosis. J. Clin. Invest. 89:648–656. 45. Sajjan, U., and J. Forstner. 1993. Role of a 22-kilodalton pilin protein in binding of Pseudomonas cepacia to buccal epithelial cells. Infect. Immun. 61:3156–3163. 46. Sajjan, U. S., L. Sun, R. Goldstein, and J. F. Forstner. 1995. Cable (Cbl) type II pili of cystic fibrosis-associated Burkholderia (Pseudomonas) cepacia: nucleotide sequence of the cblA major subunit pilin gene and novel morphology of the assembled appendage fibers. J. Bacteriol. 177:1030–1038. 47. Samadpour, M., S. Moseley, and S. Lory. 1988. Biotinylated DNA probes for exotoxin A and pilin genes in the differentiation of Pseudomonas aeruginosa strains. J. Clin. Microbiol. 26:2319–2323. 48. Selander, R., D. Caugant, H. Ochman, J. Musser, M. Gilmour, and T. Whittam. 1986. Methods of multilocus enzyme electrophoresis for bacterial population genetics and systematics. Appl. Environ. Microbiol. 51:873–884. 49. Shore, D., G. Deho, J. Tsipis, and R. Goldstein. 1978. Determination of capsid size by satellite bacteriophage P4. Proc. Natl. Acad. Sci. USA 75:400– 404. 50. Steinbach, S., L. Sun, R. Jiang, P. Gilligan, P. Flume, and R. Goldstein. 1993. Epidemiology of P. cepacia infection at the molecular genetic level. I. Clonal progeny of pre-transplant strains as the source of post-surgical infection. Pediatr. Pulmonol. S9:A213. 51. Steinbach, S., L. Sun, R.-Z. Jiang, P. Flume, P. Gilligan, T. Egan, and R. Goldstein. 1994. Transmissibility of Pseudomonas cepacia infection in clinic patients and lung-transplant recipients with cystic fibrosis. N. Engl. J. Med. 331:981–987. 52. Stull, T., J. LiPuma, and T. Edlind. 1988. A broad-spectrum probe for molecular epidemiology of bacteria: ribosomal RNA. J. Infect. Dis. 157:280– 286.
1052
GOLDSTEIN ET AL.
52a.Sun, L., and R. Goldstein. Unpublished data. 53. Sun, L., R. Jiang, S. Steinbach, P. Gilligan, J. Forstner, P. Flume, and R. Goldstein. 1993. Epidemiology of P. cepacia infection at the molecular genetic level. II. Evidence for variable transmissibility. Pediatr. Pulmonol. S9:A214. 54. Sun, L., R. Jiang, S. Steinbach, and R. Goldstein. 1993. Epidemiology of P. cepacia infection at the molecular genetic level. III. Genetic analysis of highly virulent strains. Pediatr. Pulmonol. S9:A215. 55. Sun, L., R.-Z. Jiang, A. Agodi, S. Steinbach, and R. Goldstein. 1992. PFGE
J. BACTERIOL. versus ribo-, pil- and tox-typing for detection of persistence, clonality of infection and cross-spread by CF-associated P. aeruginosa. Pediatr. Pulmonol. S8:A195. 56. Vinograd, J., and J. E. Hearst. 1962. Equilibrium sedimentation of macromolecules and viruses in a density gradient. Fortschr. Chem. Org. Naturst. XX:372–422. 56a.Yankaskas, J., P. Gilligan, and R. Goldstein. Unpublished data. 57. Zhou, H., R.-Z. Jiang, and R. Goldstein. Unpublished data.
