Predicted Configurations of Oligosaccharide Extensions in the Lipooligosaccharide of Nontypeable Haemophilus influenzae Isolates

Abstract

Lipooligosaccharide configurations were predicted in nontypeable Haemophilus influenzae isolates based on the presence of seven oligosaccharide extension-initiating genes (or alleles). Predicted configurations with 2 to 3 oligosaccharide extensions were more prevalent among middle ear than throat strains. In addition, strains with these configurations averaged higher levels of serum resistance than strains with other configurations.

TEXT

Nontypeable Haemophilus influenzae (NTHi) is a pharyngeal commensal and also a primary cause of otitis media in children. Lipooligosaccharide (LOS) is a major virulence factor of NTHi that contributes to antigenic diversity, molecular mimicry, host-cell adherence, and resistance to complement-mediated and antimicrobial peptide killing (1). Structurally, LOS consists of lipid A linked to an invariant, glucose-substituted triheptosyl inner core; its heptose residues provide initiation sites for synthesis of oligosaccharide (OS) extensions and noncarbohydrate substituents (comprising the outer core). The LOS outer core is highly variable (2) and the variation affects persistence in host colonization or infection (3). LOS heterogeneity between NTHi strains is largely driven by the presence or absence of LOS genes and alleles that initiate OS extensions from the three conserved inner core heptose residues (heptose I, II, and III) (4). We sought epidemiologic evidence supporting a potential difference in LOS structures between disease and commensal isolates by examining the distribution of OS extension-initiating genes and the predicted configurations of OS extensions based on combinations of these genes in a collection of NTHi strains.

A total of 82 NTHi strains, 44 from middle ears of children with otitis media and 38 from throats of healthy children isolated in previous studies (5–7) were used. The presence of the LOS genes lex2B, losB1, losB2, lic2C, and lpsA were determined by dot blot hybridization (6) using internal gene probes prepared with the primer pairs 5′-ACAGATCGAAAAGCTTATATGCAAGCACAG and 5′-TCAATGGAATAAGTTATAAATACGTC, 5′-GGTTAAAAGTGGGAAAATGGATGCTCG and 5′-GCCACTCTGGTTTTATTTCTTTCGGAGAG, 5′-CAAATGTTAAGGTTGGGTCTAGGC and 5′-GTCAATTTGTTCAGGTGAGAGCTC, 5′-AATTTCAATACGGTGGAGGTATGGAACG and 5′-GGATTGAGAATACGGTGTAGCTGTTCAGATGC, and 5′-CAGCACAATTATGTTATYAGTTTAACKACTG and 5′-CTTGTTTTCCAAACATTTTTATTTTTACTCGTCC or 5′-GGAACTTGTTTTCCGAATATCCGTYTTTTTATTC, respectively. DNA sequence analysis of the PCR products was used to identify lex2B and lpsA alleles. The results of individual OS extension genotypes among disease and commensal isolates are presented in Table 1. Heptose I of the triheptosyl inner core is always substituted with glucose (glucose I) in NTHi via the LgtF glycosyltransferase (3, 4). Hybridization using an lgtF probe confirmed its presence in all our isolates. Further extension from glucose I occurs at its O-4 position in strains containing lex2B encoding a glucosyl- or galactosyl-specific transferase (8) and at its O-6 position in strains with losB1 or losB2 encoding a heptosyltransferase (heptose IV) (9). lex2B was present in 73% of our NTHi strains and the majority (47/60) were predicted to express a glucose-specific allele. Forty-one percent of all strains possessed either losB1 or losB2, suggesting the potential of these strains to add a fourth heptose at the O-6 position of glucose I, allowing for further OS extension. Stratifying by strain isolation site, lex2B was present slightly more frequently in middle ear strains (82%) than in commensal throat strains (63%) (prevalence ratio [PR], 1.30; P = 0.0572).

TABLE 1.

Segregation of individual OS extension genotypes among NTHi diseases and commensal isolates

Residue and genotype	No. (%) of OS extension genotypes found in:			PR	P value
	Total (n = 82)	Middle ear (n = 44)	Throat (n = 38)
Heptose I
lgtF	82 (100)	44 (100)	38 (100)	1.00	0.9184
lex2B	60 (73.2)	36 (81.8)	24 (63.2)	1.30	0.0572
Glca	47 (57.3)	29 (65.9)	18 (47.4)	1.39	0.0905
Galb	13 (15.9)	7 (15.9)	6 (15.8)	1.01	0.9882
losB1 or losB2	34 (41.4)	15 (34.1)	19 (50.0)	0.68	0.1448
losB1	20 (24.4)	8 (18.2)	12 (31.6)	0.58	0.1589
losB2	23 (28.0)	13 (29.5)	10 (26.3)	1.12	0.7455
losB1 and losB2	9 (11.0)	6 (13.6)	3 (7.9)	1.73	0.4069
Heptose II
lic2C	44 (53.6)	31 (70.5)	13 (34.2)	2.06	0.0010
Heptose III
lpsA	78 (95.1)	41 (93.2)	37 (97.4)	0.96	0.3801
Glca	58 (70.7)	30 (68.2)	28 (73.7)	0.92	0.5850
Galb	20 (24.4)	11 (25)	9 (23.7)	1.06	0.8900
β1-2c	57 (69.5)	25 (56.8)	32 (84.2)	0.67	0.0072
β1-3d	21 (25.6)	16 (36.4)	5 (13.2)	2.76	0.0164

^aGlc, allele resulting in addition of glucose.

^bGal, allele resulting in addition of galactose.

^cAllele resulting in β1-2 linkage to heptose III.

^dAllele resulting in β1-3 linkage to heptose III.

Heptose II OS extensions are initiated in strains that possess lic2C (4), which was present in 54% of all strains and was significantly more prevalent in ear than in throat strains (PR, 2.1; P = 0.001). This result parallels that of a previous study showing that lic2B, a gene adjacent to lic2C and responsible for adding a second glucose on the heptose II extension (10), was more prevalent in ear than throat strains (7). Expression of lic2C might be important for further OS extensions by lic2B, which was shown to be crucial for optimal survival of NTHi in a mouse model of bacteremia and for complement-mediated bacterial killing (11).

Heptose III extensions are initiated in strains possessing an lpsA allele that encodes either a glucosyl- or galactosyl-specific transferase (4, 12). Hybridization and DNA sequence analysis showed that 95% of strains possessed a complete lpsA, with 71% and 24% containing glucose- and galactose-specific LpsA alleles, respectively. Deduced amino acid sequence analysis also predicted that ear strains contained more LpsA alleles predicting β1-3 linkage but fewer predicting β1–2 linkage than throat strains. The significance of these linkages in virulence is not known but they may contribute to LOS-associated serum resistance or molecular mimicry (12).

Sixteen configurations of OS extensions at the three conserved and one nonconserved heptose residues can be postulated to exist among NTHi (Table 2). Criteria for the absence of a potential extended OS (i.e., ≥2 hexose residues) include lack of an LOS gene (Table 1) or the presence of an lpsA allele known to confer a terminal galactose addition. Configuration 1 was 2.8 times more prevalent among middle ear than throat strains and, conversely, configuration 4 was 11 times more prevalent in throat than ear strains (P < 0.05 for each) (Table 2). Although not statistically different, configuration 6 was almost 3.5 times more prevalent in ear than throat strains. Configurations 1 and 6 may be similar in structure as both are predicted to have a single OS extension at each conserved, inner-core heptose but differ only in the site of heptose I-associated initiation. Together, configurations 1 and 6 were significantly more prevalent among middle ear than throat strains with the same configuration (PR, 2.94; P = 0.0094) (data not shown).

TABLE 2.

Prevalence of extended OS configurations among the NTHi strains

Configa	Predicted OS extensionb				No. (%) of extended OS configurations found in:			PR	P value
	Hep Ic	Hep IV	Hep II	Hep IIId	Total (n = 82)	MEe (n = 44)	Throat (n = 38)
1	1	0	1	1	17 (20.7)	13 (30.0)	4 (10.5)	2.81	0.0341
2	1	0	1	0	16 (19.5)	8 (18.2)	8 (21.0)	0.86	0.7451
3	1	1	0	1	11 (13.4)	4 (9.1)	7 (18.4)	0.49	0.2164
4	0	1	0	1	11 (13.4)	1 (2.3)	10 (26.3)	0.09	0.0014
5	1	0	0	1	10 (12.2)	5 (11.4)	5 (13.2)	0.86	0.8045
6	0	1	1	1	5 (6.0)	4 (9.1)	1 (2.6)	3.45	0.2229
7	1	1	0	0	3 (3.6)	3 (6.8)	0 (0)	NDf	ND
8	1	1	1	0	2 (2.4)	2 (4.5)	0 (0)	ND	ND
9	0	0	1	1	2 (2.4)	2 (4.5)	0 (0)	ND	ND
10	1	1	1	1	1 (1.2)	1 (2.3)	0 (0)	ND	ND
11	0	0	1	0	1 (1.2)	1 (2.3)	0 (0)	ND	ND
12	0	0	0	1	1 (1.2)	0 (0)	1 (2.6)	0	ND
13	0	1	0	0	1 (1.2)	0 (0)	1 (2.6)	0	ND
14	0	0	0	0	1 (1.2)	0 (0)	1 (2.6)	0	ND
15	0	1	1	0	0 (0)	0 (0)	0 (0)	ND	ND
16	1	0	0	0	0 (0)	0 (0)	0 (0)	ND	ND

^aConfig, configuration.

^b1 or 0, presence or absence of potential extension, respectively. Hep, heptose.

^cHeptose I is substituted with glucose (glucose I) in all strains via LgtF. Only further extension from glucose I was considered here.

^dAddition of glucose at heptose III was considered with potential for further OS extension. Addition of galactose is terminal and was considered as absence of additional extension.

^eME, middle ear.

^fND, not determined.

LOS structures greatly contribute to the ability of NTHi to resist complement-mediated bactericidal killing (serum resistance) (3). Serum resistance was determined by the percent survival at 25 and 12.5% human serum (Sigma) concentrations. Using a one-sided Mann-Whitney test, we found that strains with configurations 1 and 6 were significantly more resistant to 25% serum (3× higher mean percent survival) than strains with another configuration (P = 0.038); no significant difference was observed using 12.5% serum (data not shown). These results suggest the potential involvement of extended OS configurations in serum resistance.

The number of OS extensions appears to be both important for persistence in the NTHi population and a contributing factor to survival in the middle ear. The mechanism(s) by which higher numbers of OS extensions facilitate survival is not known, but the increased average serum resistance found among strains with configurations 1 and 6, each containing three extensions, suggests that higher numbers of extensions increase the potential for sialylation or antigenic diversity, factors that promote serum resistance (13).

Our observations address the potential for, rather than actual expression of, extended OS. Actual extensions may vary due to the presence of phase-variable expression of genes responsible for initiation (i.e., lex2A and lpsA) or for subsequent monosaccharide additions (14). In addition, uncharacterized gene mutations (such as those found in lpsA genes), unknown factors of transcription and translation, partial biosynthesis, and competition for acceptor residues will also affect the number and/or length of OS extensions (15). Despite these variations in NTHi LOS structure, our study has begun to define the borders of LOS heterogeneity between strains. While configurations predicted by the OS extension-initiating genes need to be confirmed structurally, such predicted configurations allow us to start association studies.

Nucleotide sequence accession numbers.

The lex2B and lpsA allele sequences have been deposited in GenBank with the accession numbers KC171044 to KC171124.