Differential Mucoid Exopolysaccharide Production by Members of the Burkholderia cepacia Complex

Abstract

We demonstrate that all nine species of the Burkholderia cepacia complex can express the mucoid phenotype. A survey of clinical isolates showed that strains of B. cenocepacia, the most virulent species of the complex, are most frequently nonmucoid. Additionally, isolates from patients with chronic infections can convert from mucoid to nonmucoid.

The Burkholderia cepacia complex (BCC) is a group of at least nine closely related species, each of which is capable of causing infection in cases of cystic fibrosis (CF) and chronic granulomatous disease (13). Patients infected with BCC are at a proportionally higher risk of death than those who are uninfected (6). Additionally, patients infected with BCC can follow a number of courses, ranging from chronic infection with little or no impact on lung function to rapid deterioration with bacteremia and death (a condition known as cepacia syndrome) (10). The factors mediating these different clinical courses are not understood. Furthermore, patients carrying identical CF genetic mutations and infected with the same clonal bacterial strains can experience highly different clinical outcomes.

Pseudomonas aeruginosa ultimately infects most people with CF; therefore, much research has focused on possible mechanisms of pathogenicity. Typically, patients are initially infected with nonmucoid strains that during the course of chronic infection convert to the grossly mucoid “CF phenotype” that is due to the elaboration of a viscous alginate-like exopolysaccharide (EPS). The isolation of mucoid P. aeruginosa from respiratory tract secretions in CF infections is so common that it has been described as a pathognomonic observation and is linked to increased morbidity and mortality (8). Whereas the mucoid phenotype in bacteria from the BCC has not been widely described, as it is not apparent on routine growth media, a few recent reports have suggested that this morphotype may be underrecognized (2, 7). BCC bacteria can produce at least four different EPSs, and the biosynthetic genes involved in the synthesis of cepacian (the most widely expressed EPS) have been identified previously (14). Data from liquid culture, in S liquid medium, have shown that EPS is produced upon the entry of B. cepacia into stationary phase, and EPS production appears to be a stable phenotype (15). The EPSs of bacteria from the BCC appear to affect the outcomes of experimental and human infections. Both shiny and mucoid variants of B. cenocepacia persist longer in animal models of infection than their nonmucoid isogenic variants (3). Herein we report the largest survey to date of the mucoid phenotype in BCC isolates from clinical and environmental sources.

Given that EPS production on standard laboratory media, including Columbia blood agar, B. cepacia selective agar, and Luria-Bertani agar, is not readily observable, we used yeast extract medium (YEM; 0.5 g of yeast extract liter⁻¹ and 4 g of mannitol liter⁻¹ supplemented with 15 g of agar liter⁻¹) to assess the capacities of BCC isolates to elaborate EPS (16). A simple scoring method was developed for describing the extent of EPS production by each isolate. Bacteria were subcultured from a frozen stock by using a cotton swab to inoculate one-third of a 25-ml YEM plate, streaked to yield individual colonies, and grown at 35°C for 48 h. When bacteria were grown on this medium, the capacities of isolates to elaborate EPS could be clearly observed by visual inspection as opposed to an examination of the appearance of the isolates on conventional diagnostic microbiological media, such as Luria-Bertani agar and blood agar. We scored EPS production as − to +++d, and the criteria for the scoring are described in the legend to Fig. 1. Mucoidy was defined as follows: nonmucoid (−) or partially mucoid (+) isolates showed no or little EPS production, and among frankly mucoid (++, +++, and +++d) isolates, EPS production was evident in all colonies.

FIG. 1.

B. cenocepacia mucoid phenotypes on YEM agar plates after 48 h of growth at 35°C. The scoring of nonmucoid and partially mucoid phenotypes was as follows: −, growth shows no evidence of EPS production and colonies are dry with a matte finish, and +, some evidence of EPS production in the confluent growth region is seen but the plate contains predominantly nonmucoid bacteria. The scoring of frankly mucoid phenotypes was as follows: ++, both the confluent area and the streaked-out region are mucoid in appearance and the growth is flat; +++, EPS production overwhelms the streaked-out area, making separate streaks hard to see, and instead of level growth across the plate there are raised areas; and +++d, same as +++ except that EPS drips onto the lid of the plate.

We first tested this semiquantitative method for scoring mucoidy with samples of bacteria, the identities of which were concealed, from the BCC experimental strain panel (4, 12). These data (Table 1) demonstrated that all species of the BCC included mucoid strains. However, among strains of B. cenocepacia (apparently the most virulent species in CF), 8 of 10 were non- or only partially mucoid. This observation was mirrored in a survey of environmental isolates from LMG and American Type Culture Collection strain banks and other sources (Table 2). These findings are in contrast to those for P. aeruginosa, mucoid strains of which are rarely isolated from the environment. Our data suggest that the capacity to elaborate EPS may be critical for survival in the environment, the natural niche of BCC bacteria. However, as demonstrated with the experimental strain panel, the proportion of frankly mucoid isolates among B. cenocepacia isolates, specifically among the recA IIIA lineage, was lower than those among the other species.

TABLE 1.

Distribution of EPS production in the BCC experimental strain panel

Species	Strain(s) with EPS score of:
	Non- or partially mucoid		Frankly mucoid
	−	+	++	+++/+++d
B. cepacia		LMG 17797	CEP509	ATCC 17759, ATCC 25416T
B. multivorans	LMG 13010T	249-2		C5393, C1576, CF-A1-1, JTC, C1962, ATCC 17616
B. cenocepacia	J2315, BC7, K56-2, C5424, C6433, CEP511	C1394, PC184		J415, ATCC 17765
B. stabilis	LMG 14294, C7322		LMG 14086, LMG 18888
B. vietnamiensis			PC259	LMG 16232, FC441, FC0369
B. dolosa		CEP021	E12	AU0645, STM1441
B. ambifaria	CEP0996			AMMDT, ATCC 53266
B. anthina			AU1293	W92T, C1765, J2552
B. pyrrocinia		BC011		ATCC 15958T, ATCC 39277

TABLE 2.

Prevalence of nonmucoid and mucoid isolates from the environment and from cultures of respiratory specimens from CF patients at Vancouver children's and adult CF clinics from 1981 to 2007

Species	No. of environmental isolates	No. of environmental isolates with score of:				% Of frankly mucoid environmental isolatesa	No. of clinical isolatesb	No. of clinical isolates (no. of RAPD groups) with score of:				% Of frankly mucoid clinical isolatesa	No. of cases of infectionc	No. of patients whose isolates converted from mucoid to nonmucoid	No. of patients whose isolates converted from nonmucoid to mucoid
		−	+	++	+++/+++d			−	+	++	+++/+++d
B. cepacia	28		1	3	24	96.4	4	4 (1)				0	1	0	0
B. multivorans	12			1	11	100	209	25 (12)	11 (8)	64 (24)	109 (34)	82.8	45	9	0
B. cenocepacia
IIIA	5	2	1	1	1	40	276	146 (6)	28 (3)	61 (4)	41 (6)	37.0	45	3	1
IIIB	4				4	100	16			2 (2)	14 (5)	100	6	0	0
B. stabilis	3		1	1	1	66.6	10	10 (1)				0	2	0	0
B. vietnamiensis	10	1	1	6	2	80	36	8 (2)	5 (3)	19 (4)	4 (3)	64	9	1	1
B. dolosa	0						0						0
B. ambifaria	13		1		12	92.3	0						0
B. anthina	15			3	12	100	0						0
B. pyrrocinia	2				2	100	0						0

^aFrankly mucoid isolates were those with a score of ++ or +++/+++d.

^bAdditionally, there were three B. multivorans and three B. cenocepacia isolates that did not grow on YEM agar and a further three B. cenocepacia isolates that appeared to have a mix of mucoid and nonmucoid colonies.

^cAmong 100 individuals, including seven that had both B. cenocepacia and B. multivorans isolates during their isolate history and one that had B. cenocepacia IIIA and B. cenocepacia IIIB isolates.

BCC isolates were collected from patients with CF attending either a pediatric clinic (British Columbia Children's Hospital) or an adult clinic (Shaughnessy or St. Paul's Hospital) between June 1981 and June 2007 and tested for mucoidy. A total of 560 isolates from 100 patients were obtained (Table 2). The largest number of clinical isolates tested were either B. multivorans or B. cenocepacia, reflecting the predominance of these species in CF patients infected with BCC; there were 45 cases of infection with one or both of these two species. All isolates of B. cepacia (four isolates in one case of infection) or B. stabilis (10 isolates in two cases of infection) were nonmucoid. The distributions of colony morphologies among B. multivorans and B. cenocepacia isolates were markedly different. Of the B. multivorans isolates, 173 (82.8%) of 209 were frankly mucoid, while only 102 (37%) of 276 B. cenocepacia isolates from the IIIA lineage were mucoid. Indeed, 146 (53%) of the B. cenocepacia isolates were absolutely nonmucoid (score, −). Of the isolates of the B. cenocepacia IIIB lineage, 100% were frankly mucoid; however, there were just 16 isolates in six cases of infection. Of the B. vietnamiensis isolates, 64% (23 of 36) were frankly mucoid. Typing was performed for each isolate by random amplified polymorphic DNA (RAPD) analysis as described previously (11). For B. multivorans, B. cenocepacia, and B. vietnamiensis, isolates were of multiple different genetic types, indicating that our observations were not due to strain-specific phenotypes. For B. cenocepacia, the common RAPD groups 1, 4, and 6 contained isolates corresponding to each of the five scores for mucoidy (−, +, ++, +++, and +++d); only B. cenocepacia RAPD group 2, of the ET-12, cable-piliated lineage, was uniformly nonmucoid (data not shown).

We also evaluated the frequency of phenotypic switching during chronic BCC infection of CF patients; no data on this topic have been published previously. A switch from the mucoid to the nonmucoid phenotype was defined by the isolation of a nonmucoid variant (score, −) from a patient from whom a frankly mucoid isolate (score, ++, +++, or +++d) of the same species and strain, as determined by RAPD typing, had previously been isolated on at least two separate occasions; a switch from nonmucoid to mucoid was defined vice versa. Phenotypic switches were observed in sequential isolates from 15 patients: nine mucoid-to-nonmucoid transitions of B. multivorans isolates, three transitions of B. cenocepacia IIIA isolates, and one transition of B. vietnamiensis isolates occurred; there were two nonmucoid-to-mucoid conversions, one each of B. cenocepacia IIIA and B. vietnamiensis isolates.

The data presented here show for the first time that all species of the BCC can express the mucoid phenotype when grown on YEM agar. Given that our previous studies demonstrated that both mucoid and shiny variants persist longer in mouse models and interact more poorly with components of the innate immune systems than their isogenic nonmucoid variants (3, 5), it is possible that the mucoid phenotype endows the bacteria with the tools for persistence during chronic infection in CF. This idea is consistent with the observation that P. aeruginosa converts to mucoidy during chronic infection. The role of BCC mucoid EPS in microbial persistence is further suggested by the capacity of EPS to scavenge reactive oxygen species, key components of the pulmonary host defense system, and its interference with neutrophil chemotaxis (1). Without chemical analysis for each isolate, we cannot specify the chemical composition of the EPS produced by each of the isolates in this study. A recent study demonstrated that the most common polysaccharide among the BCC, isolated from several different species, is cepacian; however, the researchers found an isolate of B. multivorans which produced another polysaccharide, PSI (9). Therefore, it is certainly conceivable that variations in the polysaccharides produced within the strains we studied may be significant.

The disproportionately high frequency of nonmucoid isolates among strains of B. cenocepacia, the most virulent species of the BCC, and the observation that phenotypic switching typically is from mucoid to nonmucoid are novel and intriguing, as they run counter to the observation that the conversion of P. aeruginosa to a mucoid phenotype is linked to an increased risk of morbidity and mortality. The mucoid-to-nonmucoid conversion in B. cenocepacia raises the possibility that nonmucoid isolates are associated with increased disease severity while the mucoid phenotype may be associated with persistence. There are a number of conceivable mechanisms for this situation; it is possible that without the metabolic burden of EPS production, nonmucoid bacteria are simply at a competitive advantage in the lung. Given that the mucoid phenotype of both P. aeruginosa and B. cenocepacia is associated with a reduction in virulence factor production (5, 17), it is also conceivable that the nonmucoid form is more invasive and capable of doing damage to the host. Additionally, it is possible that the lack of the EPS layer around the cells may facilitate interaction with the immune system, permitting invasion by members of the BCC. Therefore, in light of these observations, we will expand our bacterial and clinical databases to determine the significance of microbial phenotypes and phenotypic switching in the context of disease severity and the rate of clinical decline in patients with CF infected with bacteria from the BCC.