Nontypeable Haemophilus influenzae (NTHI) are a major cause of human infections. We previously demonstrated high affinity and high specificity binding of NTHI to minor gangliosides of human respiratory (HEp-2) cells and macrophages, but not to brain gangliosides. We further identified the NTHI-binding ganglioside of human macrophages as α2,3-sialylosylparagloboside (IV3NeuAc-nLcOse4Cer, nLM1), which possesses a neolacto core structure that is absent in brain gangliosides. This supported a hypothesis that lacto/neolacto core carbohydrates are critical for NTHI-ganglioside binding. To investigate, we determined the core carbohydrate structure of NTHI-binding gangliosides of HEp-2 cells, through multiple approaches, including specific enzymatic degradation, mass spectral analysis and gas–liquid chromatography. Our analyses denote the following critical structural attributes of NTHI-binding gangliosides: (1) a conserved lacto/neolacto core structure; (2) requisite sialylation, which may be either internal or external, with α2,3 (human macrophages) or α2,6 (HEp-2 cells) anomeric linkages; (3) internalized galactose residues. Mass spectral and gas chromatographic analyses confirm that NTHI-binding gangliosides of HEp-2 cells possess lacto/neolacto carbohydrate cores and identify the structure of the major peak as NeuAcα2–6Galβ1–4GlcNAcβ1–3Galβ1–4Glcβ1–1Cer (α2,6-sialosylparagloboside, nLM1). Collectively, our studies denote NTHI-binding gangliosides as lacto/neolacto series structures.
Gangliosides are sialylated glycosphingolipids of the cell membranes of virtually all eukaryotic cells that have emerged as important mediators of diverse immunoregulatory functions [1,2]. Among their immunologic roles, specific gangliosides serve as potential receptors for a variety of bacteria and bacterial products [3,4]. Earlier studies implicated selected gangliosides as receptors for Haemophilus influenzae. However, results were reliant upon ganglioside inhibition of H. influenzae-induced hemagglutination and thus were based on indirect evidence . Conflicting data suggested that nonganglioside glycosphingolipids might also serve as receptors for H. influenzae[6,7]. Moreover, most preceding studies relied upon brain glycosphingolipids from commercial sources [5,7].
Studies from our laboratory determined that gangliosides of human macrophages have unique immunoregulatory properties, not found in gangliosides of brain tissue. For example, macrophage gangliosides exerted a reversible downregulation on mitogen-induced T cell proliferation, that was 100-fold more potent than brain gangliosides [8,9]. These data supported a paradigm that gangliosides of different tissues possess distinct biological properties. This was borne out in our studies of nontypeable H. influenzae (NTHI) binding to gangliosides. We found that NTHI bound with high affinity and specificity to minor gangliosides of human respiratory epithelial (HEp-2) cells and of human macrophages, but did not bind at all to murine brain gangliosides . Furthermore, sialic acid was a critical component for NTHI-ganglioside binding. Recently, we identified the NTHI-binding ganglioside of human macrophages as sialosylparagloboside (SPG) (IV3NeuAc-nLcOse4Cer) . Unlike ganglio carbohydrate cores, found in most commercial brain gangliosides, SPG possesses a neolacto carbohydrate core. In fact, ganglio carbohydrate structures are absent in human macrophage gangliosides . Furthermore, the sialic acid residue of the NTHI-binding SPG of human macrophage gangliosides had an α2,3 anomeric linkage [10,11]. We theorized that the high affinity ganglioside binding structure for NTHI was contained within the neolacto core structure. To test this hypothesis and to identify the structure of domains for NTHI-ganglioside binding, we performed studies to determine the structure of NTHI-binding gangliosides of HEp-2 cells.
2Materials and methods
Eagle's minimum essential medium (MEM), fetal bovine serum (FBS), and trypsin-EDTA were purchased from Gibco (Grand Island, NY).
All organic solvents were of standard analytical high performance liquid chromatographic grade (Baker Chemical Co., Phillipsburg, NJ). All chemicals were standard analytical reagent quality. High performance silica gel 60 thin layer chromatography (TLC) plates were purchased from E. Merck, Darmstadt, Germany. Human laryngeal cancer (HEp-2) cells were purchased from the American Type Culture Collection (Rockville, MD).
2.2Purification of gangliosides
HEp-2 cells were plated at 5 ×104 cells ml−1 on glass petri dishes and grown to confluence using Eagle's MEM supplemented with 10% FBS. Nonadherent cells were removed with two washes of 10 ml of phosphate buffer solution (PBS) (Whittaker MA Bioproducts, Walkersville, MD) then once with 5 ml of 0.31 M of pentaerythritol (Sigma Chemical Co., St. Louis, MO). Total lipid was extracted by sonication of cells in chloroform:methanol (1:1, v/v). Gangliosides were purified from the total lipid extract as previously described [10,11]. In brief, ganglioside-containing acidic lipids were eluted through a 3 ml column of DEAE-Sephadex A-25 (Sigma Chemical Co.), dried by rotary evaporation and hydrolyzed with 0.1 N NaOH at 37 °C for 90 min. Samples were neutralized with 0.1 N HCl and desalted on a reverse phase silica gel column (SepPak, Waters Assoc., Waltham, MA). Samples were applied to a 2–3 ml column of Iatrobeads 6RS-8060 (Iatron Laboratories, Tokyo, Japan) in chloroform–methanol (85:15, v/v). After elution of less polar lipids, the total ganglioside fraction was eluted with chloroform–methanol (1:2, v/v), and dried by rotary evaporation. Total lipids were extracted and gangliosides were purified from human macrophages in culture by the same methods. Purified gangliosides were run in two dimensions on TLC plates in chloroform–methanol–0.25% KCl (50:45:10) (Solvent 1) and, after rotating the TLC plate 90° counterclockwise, in chloroform–methanol–0.25% KCl in 2.5 N NH4 (50:40:10) (Solvent 2). Distinct ganglioside peaks were visualized by heating the TLC plates to 92–94 °C after spraying with resorcinol reagent .
2.3Enzymatic degradation of HEp-2 cell gangliosides
2.3.1Clostridium perfringens sialidase treatment
HEp-2 cell gangliosides containing 5–10 μg sialic acid were incubated with Clostridium perfringens sialidase (2 U ml−1) (Sigma Chemical Co., St. Louis, MO) in 0.5 ml of 50 mM sodium citrate-phosphate buffer, pH 5.5, at 37 °C, for 2 h, as previously described . A duplicate sample of equal quantity of gangliosides was incubated in buffer alone. Reactions were terminated with addition of 0.1 M NaOH, and neutralized with 0.1 M HCl. Solutions were desalted on SepPak (Waters Assoc., Milford, MA) columns and tested for hydrolytic products by TLC. TLC plates were run in two dimensions, as described earlier, and then were sprayed with resorcinol and heated (92–94 °C). Resorcinol-positive intensity was quantitated by scanning densitometry and desialylation was determined by loss of resorcinol-positivity, compared with untreated samples. The presence of resorcinol-negative spots on sialidase-treated samples was confirmed by TLC, by reversible staining with iodine vapor, prior to resorcinol spraying .
To verify efficacy of enzymatic activity, positive and negative control gangliosides were concomitantly treated with sialidase, including gangliosides with sialidase-susceptible external sialic residues (GM3) and gangliosides with sialidase-resistant internal sialic acid residues (GM1a, GM2), respectively (Matreya, Inc., Pleasant Gap, PA).
2.3.2Newcastle disease virus sialidase treatment
To further investigate the anomeric sialic acid linkage of NTHI-binding gangliosides, HEp-2 cell gangliosides were incubated with sialidase from Newcastle disease virus (NCDV) (Glyko, Novato, CA), which is specific for removal of α2,3-linked sialic acid . HEp-2 cell gangliosides containing 5–10 μg sialic acid were incubated with 0.2 units of NCDV sialidase in 100 μl of 50 mM sodium acetate buffer containing 2 μg μl−1 sodium cholate (Sigma, St. Louis, MO), pH 5.5, at 37 °C, for 18 h as previously described [11,15]. A duplicate sample of equal quantity of gangliosides was incubated in buffer alone. Reactions were terminated by placement of samples on ice in 2 ml of 0.1 M NaCl. After adjustment of pH to 5.0, samples were desalted on SepPak columns, and re-eluted over Iatrobead columns, as described earlier. The entire content of each sample was loaded onto a TLC plate and run in two dimensions, as described earlier. TLC plates were sprayed with resorcinol and heated (92–94 °C). Resorcinol-positive intensity was quantitated by scanning densitometry and desialylation was determined by loss of resorcinol-positivity, compared with untreated samples. The presence of resorcinol-negative spots on NCDV sialidase-treated samples was again confirmed on TLC plates by reversible staining with iodine vapor, prior to resorcinol spraying .
Conditions for effective desialylation of α2,3 linked gangliosides were established, using known ganglioside (GD1a, GD1b, GD3, GM3) standards (Matreya, Inc., Pleasant Gap, PA). Under the established conditions, gangliosides with external α2,3 linked sialic acids (GM3, GD3) were successfully desialylated, while those with internal α2,3 linked sialic acids (GD1a, GD1b) were not.
HEp-2 cell gangliosides containing 5–10 μg sialic acid were incubated with bovine testes β-galactosidase (0.3 U ml−1) (Sigma Chemical Co., St. Louis, MO) at 37 °C in 100 μl of 50 mM citrate phosphate buffer (pH 4.3), containing Triton X-100 (0.5 μg μl−1), for 18 h, as previously described . A duplicate sample of equal quantity of gangliosides was incubated in buffer alone. Reactions were terminated with 0.1 M NaOH, and neutralized to pH 4–5 with 0.1 M HCl. As with sialidase treatment, solutions were desalted on SepPak columns and tested for hydrolysis by TLC, run in two dimensions as described earlier. TLC plates were then sprayed with resorcinol and heated. Enzymatic degradation was determined by a shift in chromatographic mobility of bands compared with untreated samples and with controls of known structures. To verify efficacy of enzymatic activity, positive and negative control gangliosides were concomitantly treated with β-galactosidase and included gangliosides with galactosidase-susceptible external galactose residues (GM1a, GD1b) and gangliosides with galactosidase-resistant internal galactose residues (GD3, GD1a), respectively (Matreya, Inc., Pleasant Gap, PA).
2.4Mass spectral analyses of HEp-2 gangliosides
Preparations of HEp-2 gangliosides were separated by preparative two-dimensional TLC. Discrete groups of individual peaks were recovered from silica scrapings and pooled from several TLC preparations. The first fraction included peaks 1 and 2; the second fraction included peaks 3, 4 and 5, 6; the third fraction included peaks 7, 8 and 9 (see Fig. 1). Each fraction of HEp-2 cell gangliosides was permethylated and subjected to electrospray ionization mass spectrometry performed on a triple quadrapole mass spectrometer, as described previously for murine macrophage [16,17] and human macrophage glycolipids . Matrix assisted laser desorption ionization-time of flight (MALDI-TOF) mass spectra were recorded using an Axima-curved field reflectron (CFR) mass spectrometer (Shimadzu/Kratos Analytical, Manchester, UK) equipped with a nitrogen laser. The instrument was calibrated externally using two peptide standards. Approximately 0.5 μl sample and 0.5 μl of matrix were spotted onto a stainless steel target plate and the spots were allowed to dry in the air before applied on MALDI. The samples were desorbed with 100 laser shots fired in 10 shot packets as the laser was rastered over the target surface. All the spectra were acquired in positive-ion reflectron mode using 2,5-dihydroxybenzoic acid (DHB) (10 mg ml−1 in 50/50 ACN/0.1%TFA) as matrix. Post source decay (PSD) spectra were acquired by setting the ion gate and increasing 15% laser power. Processing was done with Kratos Launchpad software. Analysis of glycolipids as permethylated structures results in the replacement of hydroxyl groups with an OCH3 group, resulting in a net increased mass of 14 for each substitution. For sialic acid, six permethylations occur, adding 84 mass units. For each hexose, three substitutions occur, adding 42 units. For each ceramide, two substitutions occur.
2.5Gas–liquid chromatography of HEp-2 ganglioside monosaccharides
Preparations of HEp-2 gangliosides were separated by preparative two-dimensional TLC and pooled from several TLC preparations, as described in the previous section. Gangliosides were recovered by eluting silica scrapings with chloroform:methanol:water – 30:60:8 (v/v/v). Solvent was removed by evaporation under nitrogen. The samples were redissolved three times in 0.3 ml chloroform:methanol:water – 60:30:8 (v/v/v) and the entire volume placed into 1 ml Reacti-vials (Pierce Chemical Co., Rockville, IL). The samples were dried under nitrogen and then by lyophilization to insure the removal of all traces of water. The monosaccharide components of each fraction were then determined by gas–liquid chromatography (GLC) as trifluoroacetyl derivatives of the methyl glycosides . First, acidic methanol was prepared by slowly adding 3 ml of acetyl chloride dropwise into 50 ml of anhydrous methanol. The dried ganglioside samples were then methanolized with 200 μl of the acidic methanol at 80 °C for 3 h using a block heating apparatus. After cooling to room temperature, the fatty acid methyl esters were removed by extracting the acidic methanol mixture three times with 0.5 ml of hexane. The remaining methanolic mixture was then evaporated under nitrogen at room temperature and again by lyophilization. Methyl glycosides were converted to trifluoroacetyl derivatives by treating them for 40–45 min at 35 °C in 10 μl of a reagent prepared by thoroughly mixing 1 ml of trifluoroacetamide, 4 ml of methylene chloride, and 0.4 ml of pyridine. The carbohydrate components were then analyzed using an Autosystem XL Gas Chromatograph (Perkin Elmer; Shelton, CT) fitted with a 30 m Zebron ZB-5 capillary column (Phenomenex; Torrance, CA). The instrument was programmed from 105 to 130 °C at 2 °C min−1 and then at 4 °C min−1 to 180 °C where it was held for 2.5 min. Samples of 1–2 μl were injected at a 1:50 split ratio and the samples swept through the column using helium carrier gas at 1 ml per min. The column conditions and carbohydrate retention times were verified using brain ganglioside and individual monosaccharide standards .
3.1Susceptibility to Clostridium perfringens sialidase
Gangliosides of HEp-2 cells previously identified as high affinity NTHI-binding glycolipids are shown in Fig. 1. Each is identified by numbering in the schematic diagram. To determine if any NTHI-binding gangliosides of HEp-2 cells possess external sialic acid residues, gangliosides were analyzed for their susceptibility to C. perfringens sialidase. Conditions were first tested using control samples of known structures (see Fig. 2). These experiments demonstrated that GM3, which has an external sialic acid, was no longer resorcinol-positive after being treated with C. perfringens sialidase. GM1a and GM2, which both possess internal sialic acids, were sialidase-resistant (Fig. 3, Panel C).
Under the same conditions, NTHI-binding gangliosides of HEp-2 cells were evaluated for susceptibility to C. perfringens sialidase. Ganglioside peak 3/4 remained resorcinol-positive after enzymatic treatment and was therefore sialidase-resistant. Ganglioside peak 5/6 was no longer resorcinol-positive after enzymatic treatment and thus was sialidase-susceptible (Fig. 3). Therefore, although sialic acid is requisite for NTHI binding , both internal and external sialylation are permissive for NTHI-ganglioside interaction.
Major peaks 1 and 2 were susceptible to sialidase. Minor ganglioside peaks 7, 8 and 9 were also susceptible to C. perfringens sialidase. However, with desialylation, a faint new spot appeared that comigrates with GM1 standards (Fig. 3, Panel B), consistent with a disialylated ganglio carbohydrate structure.
3.2Susceptibility to Newcastle disease virus sialidase
Macrophage gangliosides that bind NTHI1479 possess an α2,3 anomeric linkage [10,11]. To determine if NTHI-binding gangliosides of HEp-2 cells also possess an α2,3 anomeric linkage, HEp-2 gangliosides were subjected to hydrolysis with NCDV sialidase, which selectively cleaves α2,3 linked sialic acid [14,15]. Conditions were first established using control α2,3 sialylated ganglioside samples of known structures, including GD3, GD1a, and GD1b (see Fig. 2). Under the established conditions, the majority of each was desialylated successfully (Fig. 4). However, a faint new band that co-migrated with GM1 standards appeared in sialidase-treated samples (Fig. 4, Panel C, Lane 1). GM1a is the sialylated core structure of partial desialylation of each of these disialogangliosides (Fig. 2). Thus, the internal α2,3-linked sialic acid of GM1a is not fully susceptible to NCDV sialidase under the conditions tested.
Under identical conditions, HEp-2 gangliosides were evaluated for susceptibility to NCDV sialidase. Major ganglioside peaks 1 and 2 were partially desialylated (Fig. 4, Panel B), relative to remaining peaks, indicating at least partial α2,3 anomeric linkage of this doublet. However, NTHI-binding ganglioside peaks 3/4 and 5/6 of HEp-2 cells were all resistant to NCDV sialidase, consistent with possession of an α2,6 anomeric linkage. Neither peak co-migrated with GM1 standards and thus are unlikely to be GM1.
3.3Susceptibility to β-galactosidase treatment
Macrophage gangliosides that bind NTHI1479 possess an external galactose [10,11]. To determine if an external galactose is critical to NTHI-ganglioside interactions, HEp-2 gangliosides were subjected to treatment with β-galactosidase. Conditions were first established using control ganglioside samples of known structures, including GD3, GD1a and GD1b (see Fig. 2). Susceptibility to β-galactosidase was determined by a shift in ganglioside mobility compared with buffer-treated controls. Under experimental conditions, GD3 and GD1a, both of which have internal galactose residues, had no change in relative chromatographic mobilities. GD1b, which has an external galactose, had a distinct shift in mobility and was thus susceptible to β-galactosidase (Fig. 5, Panel C).
All gangliosides of HEp-2 cells had no change in chromatographic mobilities, relative to each other and relative to ganglioside standards, after treatment with β-galactosidase, and were therefore resistant to hydrolysis by β-galactosidase (Fig. 5). Included are NTHI-binding ganglioside peaks 3/4 and 5/6. Therefore, all NTHI-binding gangliosides identified thus far, including those of HEp-2 cells and human macrophages, possess internal galactose residues.
3.4Mass spectral analysis of HEp-2 cell gangliosides
HEp-2 cell gangliosides were recovered by preparative TLC, as separate fractions, which were analyzed by tandem mass spectrometry. The first fraction included the major ganglioside component, peaks 1 and 2. The parent ions and subsequent serial collision spectra for ganglioside peaks 1 and 2 are shown in Fig. 6 and the ion identifications are given in Table 1. The major parent ions were identified as having a mass to charge ratio of m/z 1370 and m/z 1480 (1482). Subsequent collision analyses (Fig. 6) showed that ions containing only carbohydrate are the same for both parent ions. Serial collisions identify peaks 1 and 2 of HEp-2 gangliosides as species of GM3 (II3NeuAcLacCer), differing in their respective ceramide structures. For example, m/z 823 and m/z 448 appear for both parent ions and are identified as containing sialic acid–hexose–hexose and hexose–hexose, respectively. However, ceramide peaks differed by m/z 110 (112) and account entirely for the m/z difference seen for the parent ions.
Table 1. Example of mass/charge (m/z) spectral analysis derived from HEp-2 ganglioside peaks 1/2a
Parent ion 1370.2 (NeuAc–hex–hex–cer) (m/z)
Parent ion 1479.5 (1481.6) (NeuAc–hex–hex–cer) (m/z)
aPeaks 1/2 were readily identified as heterogeneous species of GM3, differing in ceramide composition. m/z for hexosamines present in higher mass structures are conspicuously absent. All m/z given are for Na+ adducts with the exception of ceramide and all are derived from permethylated structures. Data correspond with mass spectra shown in Fig. 6.
bAll sialic acids are NeuAc.
cCeramide with m/z– 547.1 is comprised of C18 sphingosine and C16 fatty acid. Ceramide with m/z– 657.5 (659.5) is comprised of C18 sphingosine and C24:1 (C24:0) fatty acid.
Hexose–hexose–ceramide or parent ion minus NeuAcb
Parent minus sialic acid (NeuAc) and a O–CH3
NeuAc–hexose–hexose or parent ion minus ceramide
Parent minus NeuAc and a hexose
Ceramide (–H2O, H+)c
Hexose–hexose or parent ion minus NeuAc and ceramide
Sialic acid (Na+ ion adduct)
The second fraction obtained by preparative TLC contained NTHI-binding ganglioside peaks 3/4 and 5/6 and also contained a fraction of the major ganglioside peaks 1 and 2. The parent ion spectrum is shown in Fig. 7. Following collision analyses, the parent ions were identified as the structures given in Table 2. As indicated in Table 2, spectra demonstrated the presence of three groups of components, which collision analyses again indicated differed within each structure only by heterogeneity of the fatty acid portions of the ceramides. The fatty acids of ceramides of gangliosides 3/4 and 5/6 were C16, hydroxy-C16 (C16–OH), C22 and C24:1, 24:0 (Fig. 8). Permethylation results in two methyl (CH3) group substitutions on each ceramide. The replacement of a hydroxyl group with an OCH3 group for one, results in a net increased mass of 14. The second substitution, at the fatty acyl-amide linkage, results in replacement of H+ with a CH3 group, again resulting in a net increased mass of 14. Ceramide comprised of C-16-OH fatty acid (Fig. 8, Panel A) has a third OCH3 substitution.
Table 2. Structural identification of all mass/charge peaks derived from preparative TLC of HEp-2 ganglioside peaks 3/4 and 5/6a
Ganglioside peak number
Ceramide (sphingosine/fatty acid)
aData correspond with parent ion m/z spectra shown in Fig. 7. Structures of heterogeneous species of each ganglioside, based on ceramide structure, are indicated by sphingosine chain length, followed by fatty acid chain length. Structural formulae are shown in Fig. 8. All m/z are derived from permethylated structures.
bm/z value indicates the presence of ceramide with a C18 sphingosine and a C16–hydroxyfatty acid (C16–OH). Dihydrosphingosine with a C18 fatty acid chain is an alternative, but less probable structure, consistent with these m/z values.
As above, mass/charge ions at 1372, 1402, 1456 and 1482, 1484 were identified as sialosyl lactosylceramide (GM3, II3NeuAcLacCer) whose ceramide contained C16, C16–OH, C22 and C24:0, 24:1 fatty acids. Mass/charge ions of 1821, 1905, and 1931, 1933 were identified as sialosylneolactotetraosylceramide (nLM1) whose ceramide contained C16, C22 and C24:0, 24:1 fatty acids. The absence of detectable ceramides with C16–OH fatty acid in nLM1 may be due to limited sample amount. The relatively low sensitivity as a peak suggests that this isomer was present just above the background for detection. Mass/charge ions of 1617, 1647 1701 and 1727, 1729 were identified as a sialyltriosylceramide structure whose ceramide contained C16, C16–OH, C22 and C24:0, 24:1 fatty acids (Fig. 8).
3.5Gas–liquid chromatographic analysis of HEp-2 cell gangliosides
GLC analysis verified that the major component (peaks 1 and 2), identified by mass spectra as GM3, contained no hexosamine, but only the hexoses glucose, galactose and the sialic acid N-acetylneuraminic acid (Fig. 6). Mass spectra of peaks 3/4 and 5/6 (Fig. 7) demonstrated the presence of triosyl and tetraosyl entities containing a hexosamine, identified as N-acetylglucosamine (GlcNAc). The identification of the hexosamine of peaks 3/4 and 5/6 as GlcNAc is consistent with the fragmentation analysis for the presence of a hexosamine in the mass spectral data and confirms that ganglioside peaks 3/4 and 5/6 are lacto/neolacto series glycosphingolipids.
Gangliosides function as receptors for a variety of bacteria and bacterial products [19,20]. Specific binding molecules on bacterial cell surfaces adhere to host cell gangliosides via their oligosaccharide chains . Earlier studies, including work of our laboratory, implicated gangliosides as receptors for NTHI [5,10]. In our earlier study, we were intrigued to discover that the gangliosides of human respiratory cells and macrophages that bound NTHI1479 were present in minute quantities, yet bound NTHI with extremely high affinity . Although more abundant gangliosides were present in target cells, only select gangliosides of human respiratory cells and macrophages bound NTHI. Further, no gangliosides of murine brain tissue bound NTHI. Thus NTHI-ganglioside binding was highly specific. The absence of binding to brain gangliosides, all of which possess ganglio core carbohydrate structures (Gal–GalNAc–Gal–Glc), supported an hypothesis that NTHI bound preferentially to gangliosides other than those of ganglio series. This hypothesis was extended with our recent identification of the minor ganglioside doublet of human macrophages that binds NTHI1479 as SPG, implying that sialylated neolacto core structures might contain the key domain required for NTHI-ganglioside binding .
To test this hypothesis, we investigated the identities of the Haemophilus-binding ganglioside doublets (peaks 3/4 and 5/6) of HEp-2 cells by multiple approaches including mass spectroscopy, enzymatic degradation and gas chromatography. Collectively, these data indicate that: (1) Haemophilus binding gangliosides possess a conserved lacto/neolacto core structure. (2) While Haemophilus binding gangliosides possess a requisite sialic acid, internal or external sialylation are both permissive for NTHI binding. Sialylated mucins also bind NTHI1479, primarily through interaction with outer membrane NTHI proteins . (3) The anomeric linkage for sialic acid may be either α2,3, as we found in human macrophages  or α2,6, as we have now found in HEp-2 cells. An α2,6 anomeric linkage was established for HEp-2 gangliosides based on susceptibility to NCDV sialidase, just as an α2,3 anomeric linkage was confirmed for macrophage gangliosides based on resistance to NCDV sialidase . (4) All Haemophilus binding gangliosides identified thus far have only internal galactose residues. Our data identify the structure of the major Haemophilus-binding ganglioside (peak 5/6) of HEp-2 cells as NeuAcα2–6Galβ1–4GlcNAcβ1–3Galβ1–4Glcβ1–1Cer (nLM1, IV6NeuAc-nLcOse4Cer). Furthermore, Haemophilus-binding ganglioside peak 3/4 also possesses a lacto series carbohydrate core. GLC data demonstrated a sialyltriosylceramide structure containing the hexosamine GlcNAc. This NTHI-binding ganglioside displays relatively novel structural features, previously noted in structural characterization of human neutrophil and lymphocyte gangliosides . Results confirm that both Haemophilus-binding ganglioside peaks of HEp-2 cells are lacto/neolacto series gangliosides.
Peaks 1 and 2, despite their relative abundance, provide minimal binding affinity for NTHI . Both are now identified as multiple species of GM3 whose heterogeneity is based upon ceramide fatty acid composition. Peaks 7, 8 and 9, which fail to bind NTHI1479, were present in insufficient quantities to permit mass spectral analyses. However, enzymatic degradations suggest that this triplet is comprised of ganglio series disialogangliosides. This is further consistent with the overall absence of NTHI binding to other ganglio series gangliosides .
NTHI adherence to macrophage gangliosides is likely mediated through surface adhesins of the bacteria . While fimbriae are familiar adhesins, ganglioside-mediated binding of H. influenzae is independent of fimbriae . Two high molecular weight adhesins, HMW-1 and HMW-2, are expressed by 75% of NTHI strains  and likely have a sialoglycoprotein receptor . Another outer membrane protein, Haemophilus adhesin protein (Hap) has homology with bacterial IgA proteases and also promotes surface interactions with epithelial cells . H. influenzae adhesin (Hia) is a 115 kDa protein, with 72% homology to Hsf, an adhesin expressed by type b strains . 90% of NTHI strains that are deficient in HMW-1 and HMW-2 possess Hia, and Hia-expressing strains do not express HMW's . Our preliminary studies have not yet defined the specific adhesin receptor of NTHI for macrophage gangliosides.
NTHI is a major cause of serious respiratory infections worldwide . Infection with NTHI is initiated by adherence to host cells through bacterial surface proteins . Our studies have focussed on NTHI receptors of human respiratory epithelial cells and macrophages because NTHI initiates infection by interaction with each. Our rationale for focussing on human respiratory epithelial cells as potential targets is based upon evidence that NTHI is a strictly human pathogen that causes respiratory infections . Increasing evidence supports the role of macrophages in the pathogenesis of NTHI infections. Macrophages are critical to binding and phagocytosis of unencapsulated H. influenzae[30,32]. In fact, NTHI were found to be more readily phagocytosed by human alveolar macrophages than H. influenzae type b . Moreover, examination of hypertrophied adenoid tissue, removed at adenoidectomy from ten children, revealed intracellular NTHI in all samples. The reservoir for NTHI was large subepithelial mononuclear cells, which contained up to 200 viable and actively dividing intracellular NTHI per cell . Examination of explanted lungs from patients undergoing lung transplantation for COPD revealed NTHI in 8 of 16 individuals. NTHI was widely distributed and was again found in macrophages . The importance of macrophage function to NTHI infection is further illustrated by the increased susceptibility to severe, invasive NTHI infections of adults with AIDS . Macrophages of adults with AIDS also have impaired surface expression of gangliosides, including NTHI-binding gangliosides . Thus, identification of a common potential receptor in human macrophages and respiratory epithelial cells provides an important step toward understanding NTHI-glycolipid interactions.
Related studies have identified nonsialylated glycolipids as potential receptors for select strains of NTHI [7,38]. In one instance, a sialogangliosides derived from commercial sources, that lacked neolacto structures, bound Haemophilus . A broad range of glycolipid epitopes, that included neolacto series glycosphingolipids, were found to bind, not only Haemophilus, but Neisseria species as well . Phosphatidylethanolamine, gangliotetraosylceramide and gangliotriosylceramide have also been identified as potential binding foci for strains of NTHI . Numerous studies have corroborated a high degree of heterogeneity among NTHI, as exemplified by the sequence variability of the outer membrane porin protein P2 [39,40]. This is accompanied by antigenic heterogeneity of strain-specific, elicited antibody responses [5,32]. Moreover, different strains of NTHI express different adhesive factors, each targeting a distinct, specific host cell surface structure [41,42]. Thus, it is not surprising that studies utilizing different strains have identified different Haemophilus-binding glycolipids. The findings of each study, including the current one, are not necessarily applicable to all strains of NTHI. In addition, variations of conditions of cell culture and binding, which occurred among studies, are likely to be critical to Haemophilus-glycolipid interactions, as brief heat shock of cells alters binding affinity of at least some NTHI strains towards sulfatides .
Our studies independently identified species of SPG as NTHI-binding gangliosides in human macrophages (IV3NeuAc-nLcOse4Cer)  and in HEp-2 cells (IV6NeuAc-nLcOse4Cer). Our collective results indicate that all NTHI-binding gangliosides identified thus far possess lacto/neolacto core carbohydrate structures, as requisites for NTHI1479-ganglioside interaction. Gangliosides with lacto/neolacto core carbohydrate structures are not routinely available commercially. Although most common sources contain ganglio core carbohydrates, our findings underscore a grave limitation to the use of exogenous gangliosides for study of immunologic phenomena . Our results support specific lacto/neolacto series gangliosides of human cells as potential receptors for strains of NTHI and further highlight the paradigm that gangliosides of different tissues potentially possess different biological properties.
The authors are grateful for the assistance of Timothy F. Murphy, M.D., for offering scientific advice and for critically reading this manuscript, and for the technical assistance of Ms. Jane M. Smigiera. This work was supported by research grant 1RO1HL6654901 from the National Institutes of Health (CSB) and by Merit Review funding of the Department of Veteran's Affairs (HCY).