Adherence of Streptococcus pneumoniae to Respiratory Epithelial Cells Is Inhibited by Sialylated Oligosaccharides

Abstract

To study carbohydrate-mediated adherence of Streptococcus pneumoniae to the human airway, we measured binding of live S. pneumoniae organisms to a cultured cell line derived from the lining of the conjunctiva and to primary monolayers of human bronchial epithelial cells in the presence and absence of oligosaccharide inhibitors. Both encapsulated and nonencapsulated strains of S. pneumoniae grown to mid-logarithmic phase in suspension culture adhered to cultured primary respiratory epithelial cells and the conjunctival cell line. Adherence of nine clinically prevalent S. pneumoniae capsular types studied was inhibited preferentially by sialylated oligosaccharides that terminate with the disaccharide NeuAcα2-3(or 6)Galβ1. Adherence of some strains also was weakly inhibited by oligosaccharides that terminate with lactosamine (Galβ1-4GlcNAcβ1). When sialylated oligosaccharides were covalently coupled to human serum albumin at a density of approximately 20 oligosaccharides per molecule of protein, the molar inhibitory potency of the oligosaccharide inhibitor was enhanced 500-fold. The above-mentioned experiments reveal a previously unreported dependence upon sialylated carbohydrate ligands for adherence of S. pneumoniae to human upper airway epithelial cells. Enhanced inhibitory potencies of polyvalent over monovalent forms of oligosaccharide inhibitors of adherence suggest that the putative adhesin(s) that recognizes the structure NeuAcα2-3(or 6)Galβ1 is arranged on the bacterial surface in such a manner that it may be cross-linked by oligosaccharides covalently linked to human serum albumin.

Streptococcus pneumoniae is an important pathogen in chronic bronchitis, pneumonia, meningitis, otitis media, and sinusitis (8). The rising incidence of respiratory infections caused by multiple-antibiotic-resistant strains of S. pneumoniae presents an ever-increasing therapeutic challenge (12). Agents that prevent or disrupt adhesion of S. pneumoniae to the airway and thereby permit S. pneumoniae to be efficiently cleared by mucociliary action together with other nonadherent organisms are potentially intriguing alternatives or adjuncts to standard antibiotic therapies (30).

S. pneumoniae, like many other bacterial respiratory pathogens, may colonize the nasopharynx without causing disease. Viral infection of the upper respiratory tract may enable virulent strains of S. pneumoniae to progress to clinical infection (1, 6, 29), whereas less virulent strains may remain in the carrier state.

Evidence for adherence of S. pneumoniae to the human airway via carbohydrate receptors on respiratory epithelial cells was first presented by Andersson et al. (2, 3), who showed that the human milk oligosaccharide lacto-N-neotetraose (LNnT) (Galβ1-4GlcNAcβ1-3Galβ1-4Glc) could effectively inhibit binding of S. pneumoniae to desquamated cells of the human nasopharynx and oropharynx. Krivan et al. (18) described a second carbohydrate receptor, one containing GalNAcβ1-4Galβ1, that occurs in the carbohydrate chains of the glycolipids asialo-GM1 and asialo-GM2 and is recognized not only by S. pneumoniae but also by many other human respiratory pathogens. Finally, Cundell et al. (12, 13) found that the glycolipid globoside (GalNAcβ1-3Galα1-4Galβ1-4Glc-Cer), in addition to asialo-GM1 and asialo-GM2, could competitively inhibit adherence of S. pneumoniae to lung and endothelial cells in vitro.

To further define the role of carbohydrates as receptors for adherence of S. pneumoniae to the human airway, and with the goal of developing possible therapeutic uses of soluble carbohydrate receptors as antiadhesive agents for respiratory pathogens, we have tested oligosaccharides as inhibitors of S. pneumoniae binding to monolayers of human cells derived from the upper respiratory tract and from human conjunctival epithelium. In addition, we tested polyvalent forms of two inhibitors to determine whether constructs that could bridge multiple identical sites might exhibit enhanced inhibitory potency, as has been described for polyvalent inhibitors of viral adhesion (26).

MATERIALS AND METHODS

Materials.

To render tryptic soy broth (Difco) lysine deficient, lysine decarboxylase (0.4 U/ml; Sigma) was added and the solution was sterile filtered, incubated overnight at 37°C, and then autoclaved for 15 min. N-Acetyllactosamine (LacNAc) and low-endotoxin bovine serum albumin (BSA) were obtained from Sigma (see Table 1 for the complete structures of the oligosaccharides). Some samples of 6′-sialyllacto-N-neotetraose (6′SLNnT) were from Oxford Glycosystems. 3′-Sialyllactose (3′SL), 3′-sialyllactosamine (3′SLn), 3′-sialyllacto-N-neotetraose (3′SLNnT), 6′SLn, LNnT, lacto-N-triose (LNTII), GalNAcβ1-3LacNAc, 6′SL conjugated to human serum albumin (6′SL-HSA), 3′SL-HSA, and some samples of 6′SLNnT were from Neose Technologies, Inc. The HSA glycoconjugates, prepared according to the method of Smith et al. (25), were kindly provided by J. McCauley, Neose Technologies, Inc. Anthrone and ninhydrin reactions were used to determine an oligosaccharide/protein molar ratio of approximately 20 for both polyvalent constructs studied.

TABLE 1.

Oligosaccharide structures

Oligosaccharide	Structure
LacNAc	Galβ1-4GlcNAc
LNnT	Galβ1-4GlcNAcβ1-3Galβ1-4Glc
LNTII	GlcNAcβ1-3Galβ1-4Glc
3′SL	NeuAcα2-3Galβ1-4Glc
3′SLn	NeuAcα2-3Galβ1-4GlcNAc
3′SLNnT	NeuAcα2-3Galβ1-4GlcNAcβ1-3Galβ1-4Glc
6′SL	NeuAcα2-6Galβ1-4Glc
6′SLn	NeuAcα2-6Galβ1-4GlcNAc
6′SLNnT	NeuAcα2-6Galβ1-4GlcNAcβ1-3Galβ1-4Glc
GalNAcβ1-3LacNAc	GalNAcβ1-3Galβ1-4GlcNAc

Bacteria.

Strains R-6 (unencapsulated, derived from type 2) and SIII (type 3) of S. pneumoniae were obtained from Elaine Tuomanen, Rockefeller University. Clinical isolates of S. pneumoniae were obtained from Jeffrey Weiser and Robert Austrian at the University of Pennsylvania. Bacteria were maintained as frozen stocks and passaged on blood agar plates kept at 37°C and 5% CO₂. For each radioisotope experiment, an inoculum was taken from a 1- to 2-day plate culture, added to lysine-deficient tryptic soy broth containing 70 μCi of [³H]lysine (80 to 100 Ci/mmol) per ml, and incubated at 37°C in 5% CO₂. The growth of each culture was monitored by counting with a Petroff-Hausser chamber and/or by light scattering (absorbance at 595 nm [A₅₉₅]) changes. For the visual adherence assay, the R-6 strain of S. pneumoniae was grown in normal tryptic soy broth, without an isotope. The clinical isolates of S. pneumoniae were cultured in 175-cm² tissue culture flasks containing 30 ml of Columbia broth supplemented with 1 mg of sodium ascorbate per ml, and the flasks were mixed by inversion every hour when a sample was taken to monitor the A₅₉₅.

Epithelial cells and cell lines.

The Wong-Kilbourne clone of Chang conjunctival cells (derived from human conjunctival epithelial carcinoma; American Type Culture Collection) were grown in Dulbecco’s modified essential medium with 10% fetal calf serum, penicillin, and streptomycin. Cells to be used as targets in bacterial binding studies were first cultured in tissue culture flasks and then seeded in 96-well plates (Wallac 1450-405) coated with 70 μg of human placental collagen (Sigma) per cm² (10) at a density sufficient to form a confluent monolayer after overnight culturing. For the visual adherence assay, the cells were seeded into eight-well chamber slides (Nunc Inc., Naperville, Ill.) for culturing to confluence in 2 days. Primary normal human bronchial and tracheal cells (NHBE cells) obtained from Clonetics Corp. (San Diego, Calif.) were cultured in the Clonetics medium exactly as recommended in the provider’s accompanying literature.

Isotopic adherence assay.

Progression of S. pneumoniae cultures through the growth cycle was monitored by the A₅₉₅. The bacteria were centrifuged, rinsed three times, and resuspended in L-15 medium (without phenol red) plus 0.1% low-endotoxin BSA (L-15–BSA). The concentration of bacteria was determined by visual counting with a Petroff-Hausser chamber, radioactivity was determined by scintillation counting, and the specific activity of the ³H-labeled cells was calculated. Preparations of bacteria with 7 cpm/1,000 cells or greater were used. The bacteria were diluted to 2 × 10⁸ to 5 × 10⁸/ml, mixed 1:1 with oligosaccharides serially diluted in L-15–BSA, and incubated for 15 min at room temperature, after which 25 μl of each bacterial suspension was transferred to the surface of a monolayer of epithelial cells in a 96-well plate. The 96-well plates were covered and incubated on an orbital platform shaker with agitation at ∼55 rpm for 30 min at room temperature. The bacterial suspension was removed by rapidly flicking the plate, and the monolayers were rinsed free of unbound bacteria with 100 μl of L-15–BSA followed by flicking away the liquid a total of four times. Finally, 50 μl of scintillant (Scintisafe; Fisher) was added to each well, and the plate was sealed and put into a Wallac Microbeta scintillation counter to determine radioactive counts in each well. The concentration of bacteria was adjusted to obtain a minimum of 200 cpm per well in the absence of inhibitor (background, typically 7 cpm). Experiments were performed in triplicate, and the mean value and standard error of the mean were calculated for each experimental condition.

Visual adherence assay.

To each monolayer of cells on eight-well chamber slides was added 130 μl of bacterial suspension containing S. pneumoniae at 10⁹ organisms per ml in L-15–BSA that had been preincubated with or without oligosaccharide inhibitors at room temperature. After incubation of the bacteria with the target cells for 30 min at room temperature, the cells were washed with L-15–BSA, fixed in HistoChoice MB (Amresco Inc., Solon, Ohio), air dried, and then stained with Giemsa stain. Giemsa staining was accomplished by rehydrating the slides in distilled water for 1 min and incubating them with Giemsa stain from Fluka (Buchs, Switzerland), diluted 1:7 in 0.01 M phosphate buffer (pH 6.0). After 30 min of staining, the slides were incubated in distilled water for 2 min, dipped five times in 0.01% acetic acid, again incubated for 2 min in distilled water, air dried, and mounted with balsam. A Zeiss Photomicroscope I with an ocular grid was used to count the bacteria and to photograph the slides. For 5 to 15 fields per condition, all bacteria within the ocular grid were counted, and the data were recorded as the mean number of bacteria per field. On control slides the mean number of bacteria per grid field (0.017 mm²) was 990 for the R-6 strain. For the clinical isolates of S. pneumoniae the adherent bacteria were counted at a magnification of ×550 with a Wild inverted microscope with a Spencer 44× objective. The mean number of bacteria per field (0.073 mm²) was 8 to 70 for the clinical isolates.

RESULTS

Adherence of S. pneumoniae R-6 to Chang cells in the absence of inhibitors was measured for organisms withdrawn from suspension culture at successive stages of the growth cycle. Pilot experiments showed that when bacterial suspensions at a density of 1 × 10⁸ to 2.5 × 10⁸/ml were incubated for 30 min with target cell monolayers, the total available sites for bacterial binding were less than half saturated. The percentage of added bacteria bound to cells was maximal (range = 0.4 to 1%) during lag phase (i.e., just after transfer of bacteria from plates into liquid growth medium) through early exponential phase and decreased gradually thereafter to approximately one-third of the maximum value 3 h after cessation of exponential growth.

To examine the sensitivity of bacteria at various phases of growth in suspension culture to inhibition of adherence by oligosaccharides, adherence assays were performed in the presence of 6 mM oligosaccharide with bacteria sampled at hourly intervals during growth at 37°C. Three oligosaccharides, i.e., 6′SLn (NeuAcα2-6Galβ1-4GlcNAc), LNnT (Galβ1-4GlcNAcβ1-3Galβ1-4Glc), and GalNAcβ1-3LacNAc (GalNAcβ1-3Galβ1-4GlcNAc) (Table 1), were observed to inhibit adherence of radiolabeled S. pneumoniae R-6 to Chang cells. Sensitivity of adherence to oligosaccharide inhibitors was observed to change during the bacterial growth cycle in a manner specific to each inhibitor: 6′SLn was a potent inhibitor of S. pneumoniae R-6 adherence to Chang cells at all stages during the bacterial growth cycle (Fig. 1), whereas LNnT exhibited weaker inhibitory activity transiently during mid-exponential phase and GalNAcβ1-3LacNAc was almost inactive at the concentration tested (Fig. 1).

FIG. 1.

Growth phase-dependent adherence of S. pneumoniae to Chang conjunctival cells. At hourly intervals growth of S. pneumoniae R-6 labeled with [³H]lysine in suspension culture was monitored by light scattering (A₅₉₅) (▵) and aliquots of bacteria were withdrawn, washed, and resuspended in L-15–BSA or L-15–BSA plus 6 mM 6′SLn (□), 6 mM LNnT (▿), or 6 mM GalNAcβ1-3LacNAc (○); they were then added to flat-bottom microtiter wells in which Chang cells had been previously grown to confluence. The fraction of bacteria adherent to cells in the presence of each oligosaccharide inhibitor was determined by scintillation counting and plotted as the percent decrease in binding with respect to the control. Each point represents the mean of at least three determinations in one experiment; for visual clarity, the error was not plotted.

Similar results were obtained when several oligosaccharides with related structures were tested over a range of concentrations to determine the concentration of inhibitor required to reduce the number of adherent bacteria by 50% compared to the same bacteria incubated in the absence of inhibitor (defined as the IC₅₀). Adherence of mid-exponential-phase R-6 to Chang cells was inhibited by all sialylated (NeuAcα2-3[or 6]Galβ1-R) oligosaccharides tested, at IC₅₀s ranging from 1.6 to 3 mM (Table 2). Under the same conditions, LNnT (Galβ1- 4GlcNAcβ1-3Galβ1-4Glc) and LacNAc (Galβ1-4GlcNAc) were poor inhibitors (IC₅₀s = 7.1 to 9.5 mM [Table 2]) and LNTII (GlcNAcβ1-3Galβ1-4Glc) and GalNAcβ1-3LacNAc (GalNAcβ1-3-Galβ1-4GlcNAc) were inactive (≥10 mM). Similar results were observed for inhibition of binding of exponential-phase S. pneumoniae SIII to Chang cells (Table 2).

TABLE 2.

Inhibition of adherence of S. pneumoniae to Chang cells (radioisotope assay)

Inhibitora	R-6		SIII
	Mean IC50b	nc	Mean IC50b	nc
LacNAc	7.1	6	9.0	2
LNnT	9.5	10	10	2
LNTII	10.0	1	10.0	1
3′SL	1.7	3	5.9	2
6′SL	2.0	3	4.7	1
3′SLn	1.8	5
6′SLn	3.0	8	3.0	2
3′SLNnT	2.0	3
6′SLNnT	1.6	6
GalNAcβ1-3LacNAc	10.0	2

^aSee Table 1 for structures of oligosaccharides. 

^bFor individual experiments where 50% inhibition was not achieved at any concentration, the value was recorded as 10 mM (the highest concentration tested) for the purpose of calculating a mean. A mean of 10 mM indicates that the IC₅₀ was 10 mM or that none was reached in any experiment. 

^cNumber of experiments performed. 

To better evaluate whether transformed target cell lines derived from the human respiratory tract accurately model adherence properties of native airway epithelium, we counted under a microscope the number of mid-exponential-phase S. pneumoniae R-6 cells adherent to monolayers of NHBE in the presence or absence of oligosaccharide inhibitors. The sialylated oligosaccharides 3′SLNnT (NeuAcα2-3Galβ1-4GlcNAcβ1-3Galβ1-4Glc) and 6′SLNnT (NeuAcα2-6Galβ1-4GlcNAcβ1-3Galβ1-4Glc) at 5 mM nearly completely inhibited bacterial adherence, whereas LNnT (Galβ1-4GlcNAcβ1-3Galβ1-4Glc) at the same concentration failed to inhibit adherence (Fig. 2). Qualitatively similar results were obtained with the Chang conjunctival cell line.

FIG. 2.

Inhibition of adherence of S. pneumoniae R-6 to NHBE cells by oligosaccharides. S. pneumoniae R-6 sampled from Columbia broth suspension culture at mid-exponential phase was washed in L-15–BSA and suspended at a density of 10⁹ organisms per ml in the same buffer alone (a) or containing 5 mM 6′SLNnT (b), 5 mM 3′SLNnT (c), or 5 mM LNnT (d); bacteria were then incubated with NHBE cell monolayers for 30 min at room temperature. The slides were then washed, fixed, stained with Giemsa stain, and photographed at a magnification of ×800 under bright-field illumination.

By using the same method, a series of S. pneumoniae clinical isolates (serotypes 3, 4, 6A, 6B, 9V, 14, 18C, 19A, 19F, and 23F) in exponential phase were incubated with Chang cells or NHBE cells in the presence or absence of serially diluted 6′SLNnT, 3′SLNnT, or LNnT (Table 3). Adherence of all of the clinical isolates, except one type 23F isolate, was inhibited by 6′SLNnT or 3′SLNnT in at least one experiment. In single experiments with four strains (type 14 strain P40, type 19A strain T19ABrazil, type 19F strain P35, and type 23F strain T23FSF) inhibition failed to reach 50% at the highest concentration of oligosaccharide tested, but the same inhibitor gave >50% inhibition in a repeat experiment, indicating some variability in degree of expression of adhesion characteristics under in vitro culture conditions (Table 3). For 18 experiments where it could be measured, the mean IC₅₀ for inhibition of adherence of various encapsulated S. pneumoniae strains to primary bronchial cells by 3′SLNnT was 1.5 mM (range, <0.5 to 4 mM). The mean IC₅₀ for inhibition of adherence of various S. pneumoniae isolates to Chang cells by 6′SLNnT was 2.3 mM (range, <0.5 to 5 mM) (Table 3). LNnT (Galβ1-4GlcNAcβ1-3Galβ1-4Glc) inhibited the adherence of only three of eight isolates tested with Chang cells, with a mean IC₅₀ of 1.0 mM (range, <0.5 to 2 mM). LNTII (GlcNAcβ1-3Galβ1-4Glc) at concentrations up to 6 mM failed to inhibit the adherence of S. pneumoniae P303 (type 6A) and T18CNJ (type 18C) to human bronchial cells (data not shown).

TABLE 3.

Inhibition of adherence of S. pneumoniae to NHBE and Chang cells (visual assay)

Cell type	Serotype	Strain	IC50a (mM)
			6′SLNnT	3′SLNnT	LNnT
NHBE	Unencapsulated	R-6	0.9		N
	Unencapsulated	R-6	<0.5	<0.5	N
	Unencapsulated	R-6	4.0	1.0	N
	3	SIII	4.0	1.4	N
	4	P15		2.0
	6A	P303		1.0
	6A	P303		1.0
	6A	P305		4.0
	6B	P314		2.0
	6B	P317		0.4
	9V	P62		1.0
	9V	P64		1.0
	14	P31		2.4
	14	P40		N
	14	P40		0.8
	18C	T18CNJ		0.3
	18C	T18CNJ		1.0
	18C	T18CNJ	3.0	1.0	N
	18C	P57		2.4
	19A	T19AKY		2.0
	19A	T19ABrazil		N
	19A	T19ABrazil		3.0
	19A	T19ABrazil		2.0
	19F	P35		0.4
	19F	P35		N
	19F	T19FLn		2.0
	23F	P19		0.6
	23F	T23FSF		N
Chang	3	SIII	<5		<5
	3	SIII	1.0	0.4
	4	P15	1.0		N
	6A	P303	5.0		N
	6A	P305	<0.5		<0.5
	6B	P314	3.0		N
	6B	P317	0.6		N
	14	P31	4.0		N
	18C	T18CNJ	2.0		2.0
	23F	P19	2.0		0.6

^aN, no IC₅₀; inhibition was less than 50% at the maximum concentration tested, 6 mM. 

Multivalent neoglycoconjugates containing 3′SL (NeuAcα2-3Galβ1-4Glc) and 6′SL (NeuAcα2-6Galβ1-4Glc) were tested as inhibitors of adherence of exponential-phase S. pneumoniae R-6 to Chang cells under conditions in which the free monovalent oligosaccharides were already shown to be effective. A remarkable enhancement of activity (calculated on the basis of the molar concentration of the oligosaccharide inhibitor added) was observed for the HSA conjugates, 3′SL-HSA and 6′SL-HSA (IC₅₀s = 2 and 10 μM, respectively) compared with the free 3′SL and 6′SL (IC₅₀s = 1.7 and 2 mM, respectively) (Fig. 3). HSA at the same molar protein concentration as the HSA glycoconjugates did not inhibit the binding of S. pneumoniae R-6 to Chang cells (data not shown).

FIG. 3.

Inhibition of binding of S. pneumoniae R-6 at mid-exponential phase to Chang conjunctival cells by oligosaccharides. S. pneumoniae R-6 sampled from tryptic soy broth suspension culture at mid-exponential phase was washed in L-15–BSA and resuspended in the same buffer containing 3′SL-HSA (▾), 6′SL-HSA (▴), 3′SL (▿), 6′SL (▵), or 6′SLn (○) at various concentrations; the bacteria were then incubated with Chang conjunctival cells previously grown to confluence in microtiter wells. After incubation for 30 min at room temperature, the wells were washed and counted by liquid scintillation. The fraction of bacteria adherent to cells in the presence of each oligosaccharide inhibitor is plotted as a percentage of binding observed in the absence of inhibitor at each time point. HSA alone, at concentrations comparable to those used for the glycoconjugates, did not inhibit the adherence of S. pneumoniae to Chang cells (data not shown).

DISCUSSION

In 1981 Beachey clearly formulated the hypothesis that “… one might apply the isolated and purified bacterial adhesin membrane receptors, or analogues of these substances as competitive inhibitors of bacterial adherence …” to prevent or treat bacterial infection (7). During the intervening years several complex sugar sequences have been proposed as targets to which S. pneumoniae specifically adheres (2, 3, 12, 13, 18). To explore the efficacy of free oligosaccharides as inhibitors of S. pneumoniae adherence to cells of respiratory and conjunctival origin, we tested compounds that included previously identified targets, as well as several other compounds comprised of sugar sequences known to occur on the surfaces of many eukaryotic cells. In the course of this work we discovered that S. pneumoniae relies to a significant extent upon oligosaccharide ligands terminating in NeuAcα2-3(or 6)Galβ1 for adherence to epithelial cells. In addition, we confirmed prior reports describing S. pneumoniae adherence specific for oligosaccharide chains representative of the neolacto (2, 3) and globo (12) series of glycosphingolipids. It should be noted, however, that inhibitory effects of nonsialylated oligosaccharides were weak and transient during growth of S. pneumoniae in suspension culture.

We have shown that sialylated oligosaccharides at similar concentrations specifically inhibit adherence of S. pneumoniae to a cell line derived from conjunctival epithelium and to primary explants of NHBE cells. Transformed cell lines do not always faithfully replicate the repertoire of complex sugar epitopes displayed on the surface membranes of the native epithelial cells from which they are derived (15). Nevertheless, data suggest that the transformed cell lines Chang (conjunctival epithelial cells [27]) and Detroit 562 (nasopharyngeal epithelial cells [unpublished data]) retain an adherence phenotype representative of native cells and that sialylated glycoconjugates may function as important targets for S. pneumoniae adherence to the conjunctiva and the nasopharynx. Sialylated epitopes appear to represent possible adherence targets for S. pneumoniae colonization of the upper airway.

Many of our initial studies of S. pneumoniae adherence were performed with the relatively sturdy, fast-growing, and highly adherent strains R-6 and SIII. Further studies tested the general importance of adherence to sialylated glycoconjugates with 17 respiratory clinical isolates of S. pneumoniae representing nine serotypes. The adherence to respiratory and conjunctival epithelial cells of all isolates except one was sensitive to 3′SLNnT (NeuAcα2-3Galβ1-4GlcNAcβ1-3Galβ1-4Glc) and/or 6′SLNnT (NeuAcα2-6Galβ1-4GlcNAcβ1-3Galβ1-4Glc).

Andersson et al. (2, 3) employed oligosaccharide inhibitors to probe for the specificity of attachment of S. pneumoniae to epithelial cells scraped from the oropharynx. Among the natural and synthetic carbohydrates they tested, synthetic molecules containing the disaccharide sequence GlcNAcβ1-3Gal were the most active inhibitors. In our experiments, the oligosaccharide LNTII (GlcNAcβ1-3Galβ1-4Glc) added at a concentration of 10 mM did not inhibit adherence of two strains of S. pneumoniae (R-6 and SIII) to a transformed cell line of conjunctival origin, nor did LNTII at 6 mM inhibit the adherence of two encapsulated strains of S. pneumoniae (T18CNJ and P303) to NHBE cells. The oligosaccharide LNnT (Galβ1-4GlcNAcβ1-3Galβ1-4Glc) did not inhibit adherence to NHBE cells in primary culture, although it did show activity for three of nine S. pneumoniae isolates in the type screen survey. The possibility remains that attachment of S. pneumoniae to cells accessible by scraping the oropharynx, many of which are not viable, may be mediated by some sugar chain containing the GlcNAcβ1-3Gal sequence that is poorly represented or absent in the target cells we tested or that the strain of S. pneumoniae employed by Andersson et al. (2, 3) expressed an adhesin lacking in the bacteria we tested.

When tested as inhibitors of S. pneumoniae adherence, sialylated oligosaccharides terminating in NeuAcα2-3Galβ1 and NeuAcα2-6Galβ1 exhibited similar specific activities. This relative lack of specificity both for the aglycone and for the linkage position of sialic acid on the penultimate galactosyl residue is remarkable, in that many previously described viral and bacterial adhesins exhibit clear preference for either 3- or 6-linked sialic acid. For example, S fimbriae associated with Escherichia coli serotype O18:K1:H7, which causes meningitis and septicemia in newborn infants, specifically bind oligosaccharide receptors containing NeuAcα2-3Gal (21, 22). Similarly, the HMW1 adhesin of Haemophilus influenzae exhibits a preference for α2-3-linked sialic acid (27). Specificity for the structure NeuAcα2-6Gal was demonstrated for influenza virus X-31 hemagglutinin, a protein which could be made specific for the structure NeuAcα2-3Gal after exchange of a single amino acid in the peptide binding site (23). Lectin studies indicate that the surface of respiratory epithelium incorporates glycoconjugates bearing sialic acid, including, but not limited to, the structure NeuAcα2-6Gal (20). The extent to which the putative sialic acid binding adhesin of S. pneumoniae is chemically related to adhesins with similar specificities in other organisms remains to be established.

Regulation of pneumococcal adherence as a function of the growth cycle has been the subject of numerous recent studies relating opaque-transparent phenotypic variants, expression of peptide permeases, choline binding proteins, and release of bacterial cell wall components responsible for epithelial cell activation (28). Not yet understood is the relationship between virulence mechanisms that induce disease in the lower airway and factors that promote the carrier state in which pneumococci exist for weeks in the upper airways of individuals who have no symptoms. It is interesting that adherence of S. pneumoniae to sialylated epitopes on conjunctival epithelial cells seems not to be strongly dependent upon cell cycle, whereas at least a fraction of S. pneumoniae cells seem to transiently express a lactosamine-adherent phenotype during mid-exponential growth. Further study will be required to determine whether cells of the lower respiratory tree, e.g., type II alveolar cells, are recognized similarly.

Sialylated oligosaccharides made multivalent by covalent coupling to carrier protein have a potency as inhibitors of S. pneumoniae adherence to conjunctival epithelial cells that is increased by more than 2 orders of magnitude. Many investigators have previously reported dramatic increases in the inhibitory potency of multivalent constructs compared with monovalent inhibitors of adherence (19, 24, 26). This effect is generally attributed to cooperativity among multiple, tandem, noncovalent interactions at the bacterial surface. Such interactions require that adhesins be spaced so that the oligosaccharide arms of the multivalent construct may reach and cross-link them. The results imply that the density of expression of sialylated carbohydrate chains on epithelial cell surfaces may be an important determinant of tissue tropism for S. pneumoniae colonization.

It is generally accepted that the natural process of pneumococcal infection begins with asymptomatic colonization of the nasopharynx by organisms that are potential pathogens but that may be carried for many days or weeks without causing disease (17). Intervening respiratory viral infection or another inflammatory event may trigger changes in the nasopharyngeal epithelium that compromise host resistance, leading to enhanced bacterial adherence and proliferation characteristic of infection (11). In infants, the well-documented beneficial effects of breast-feeding for prevention of pneumococcal infection (5), which probably cannot be attributed to effects of secreted immunoglobulin A in milk (4), might well be due in large part to antiadherence effects of sialylated and lactosamine-terminated oligosaccharides and glycoproteins. Carlson (9) found that the concentration of oligosaccharide-bound sialic acid in human milk ranges from 350 to 1,800 mg/liter (1.1 to 5.8 mM) during the first 5 weeks of lactation and by 20 weeks decreases to a plateau value of approximately 200 mg/liter (0.6 mM). Corresponding values found for glycoprotein-bound sialic acid are 100 to 500 mg/liter (0.3 to 1.6 mM) during weeks 1 to 5 and a plateau value of approximately 75 mg/liter (0.24 mM), some fraction of which may be expected to be multivalent. Our findings are consistent with the notion that frequent bathing of the nasopharyngeal mucosa with milk containing sialylated oligosaccharides and glycoproteins at concentrations in the millimolar range might interrupt adherence of S. pneumoniae to epithelial cells of the upper respiratory tract, thereby reducing the load of colonizing organisms and diminishing the risk of infection. In a rabbit model of pneumonia (16), 3′SLNnT (NeuAcα2-3Galβ1-4GlcNAcβ1-3Galβ1-4Glc) protected against infection by S. pneumoniae, and in an infant rat model (16), 3′SLNnT reduced nasopharyngeal colonization by S. pneumoniae. The use of orally or nasally administered milk oligosaccharides as prophylactic and/or therapeutic agents to promote clearance of S. pneumoniae from the nasopharyngeal mucosa may have value as a means of reducing the risk of developing otitis media.