(Received for publication, May 5, 1997)
From the Department of Medical Biochemistry,
Göteborg University, Medicinaregatan 9A, S-413 90 Göteborg,
Sweden, the ¶ Bacterial Enteropathogens Laboratory, Veterans
Affairs Medical Center and Baylor College of Medicine, Houston, Texas
77030, and the
Department of Clinical Microbiology and
Immunology, Örebro Medical Center,
S-701 85 Örebro, Sweden
The possible interaction of the
neutrophil-activating protein of Helicobacter pylori with
target cell glycoconjugates was investigated by the binding of
125I-labeled recombinant protein to glycosphingolipids from
human neutrophils in solid phase assays. Thereby, a distinct binding of
the neutrophil-activating protein to four bands in the acid glycosphingolipid fraction from human neutrophils was detected, whereas
no binding to the non-acid glycosphingolipids or polyglycosyl ceramides
from these cells was obtained. When using glycosphingolipids not
present in the cell membrane of human neutrophils, it was found that
the neutrophil-activating protein also bound to sulfated glycosphingolipids as sulfatide and sulfated gangliotetraosyl ceramide. Comparison of the binding preferences of the protein to
reference glycosphingolipids from other sources suggested that in human
granulocytes, the neutrophil-activating protein of H. pylori preferentially recognizes glycoconjugates with a
terminally unsubstituted NeuAc3Gal
4GlcNAc
3Gal
4GlcNAc
sequence.
The human pathogen Helicobacter pylori is the causative agent of active chronic gastritis (1, 2). Infection with H. pylori is also strongly associated with development of gastric and duodenal ulcers (3). A causal connection between H. pylori infection and gastric malignancies such as gastric adenocarcinoma (4) and lymphoma (5-7) has also been suggested.
A prominent feature in H. pylori-induced gastritis is infiltration of the gastric lamina propria by neutrophil granulocytes, and neutrophils are also seen permeating between the epithelial cells (8). The mechanisms involved in the gastric inflammatory response have been the subject of numerous studies, and it has been demonstrated that H. pylori or cell-bound compounds released from the bacterium can cause chemotaxis (9-13), activate neutrophils to a strong oxidative burst (14), and induce an increased adhesion of neutrophils to endothelial cells (15). Furthermore, neutrophils activated by water extracts of H. pylori promote detachment of endothelial cells (16).
A neutrophil-activating protein from H. pylori was recently identified (17). This 150-kDa protein was isolated from water extracts of the bacterium. The neutrophil-activating protein induces an increased expression of CD11b/CD18 on neutrophils and promotes the adhesion of neutrophils to endothelial cells. By polymerase chain reaction amplification, the gene for the neutrophil-activating protein was found in all H. pylori isolates tested.
Since the activation of human neutrophils might involve
lectin-carbohydrate interactions, the binding of radiolabeled
neutrophil-activating protein of H. pylori to
glycosphingolipids from human neutrophils was investigated using solid
phase assays. A selective binding of the neutrophil-activating protein
to four compounds in the acid glycosphingolipid fraction of human
neutrophils was detected. In addition, the protein bound to sulfatide
and sulfated gangliotetraosyl ceramide. Comparison of the binding
preferences of the protein to reference glycosphingolipids from other
sources suggested that the neutrophil-activating protein of H. pylori preferentially recognizes glycoconjugates with a
terminal linear
NeuAc3 Gal
4GlcNAc
3Gal
(4GlcNAc
)1
sequence present in the acid glycosphingolipid fraction of human neutrophils (18, 19).
Neutrophil leukocytes were isolated from pooled outdated human blood from healthy donors (blood group A) by Ficoll-Paque (Pharmacia, Uppsala, Sweden) centrifugation as described by Bøyum (20), with minor modifications (21).
Acid and non-acid glycosphingolipid fractions were isolated by the
reported method (22). Briefly, the material was lyophilized and then
extracted in two steps in a Soxhlet apparatus with chloroform and
methanol (2:1 and 1:9, by volume). The extract was subsequently subjected to mild alkaline hydrolysis and dialysis, and nonpolar lipids
were removed by silicic acid column chromatography. Acid and non-acid
glycosphingolipids were separated on a DEAE-cellulose column. The
non-acid fraction was acetylated (23), separated from alkali-stable
phospholipids on a second silicic acid column, and then deacetylated.
From 1.6 × 1010 cells of pooled neutrophils, 24.4 mg
of acid glycosphingolipids (1.5 × 109 mg/cell) and
41.2 mg of non-acid glycosphingolipids (2.6 × 10
9
mg/cell) were obtained.
Polyglycosyl ceramides were isolated as described (24). In brief, the residue from extraction with chloroform and methanol was acetylated and then extracted with chloroform. The extract was filtered, and the residue was washed with water. The material in the chloroform phase was purified by Sephadex LH-20 and Sephadex LH-60 (Pharmacia) chromatography. The main carbohydrate-containing fraction was deacetylated and, after dialysis, suspended in 2-propanol/hexane/water (55:25:20, by volume) and centrifuged. Thus, the polyglycosyl ceramides were obtained in the supernatant, and the yield was approximately 0.3 mg/g of dry weight.
Reference GlycosphingolipidsAcid and non-acid
glycosphingolipid fractions were obtained from the sources given in the
legend to Fig. 1 and in Table I by standard procedures (22). The
individual glycosphingolipids were isolated by repeated chromatography
of the native glycosphingolipid fractions or their acetylated
derivatives on silicic acid columns. The identity of the purified
glycosphingolipids was confirmed by mass spectrometry (25), proton NMR
spectroscopy (26-29), and degradation studies (30, 31). For removal of
sialic acid, the acid glycosphingolipids were treated with 1% acetic
acid at 100 °C for 1 h.
|
The napA gene,
GenBankTM accession number U16121 (32), was cloned into the
BamHI site of plasmid pTrxFus in Escherichia coli
host strain Gi724 so that a
thioredoxin-HPNAP2 fusion protein was
expressed under control of the bacteriophage promotor. Expression
of the tryptophan-inducible thioredoxin HPNAP fusion protein and its
recovery in lysozyme EDTA whole cell lysates were done using the
ThioFusion Expression System (Invitrogen Corp., San Diego, CA 92121).
The fused protein was purified by the same method as described for the
native neutrophil-activating protein of H. pylori (17). The
N-terminal thioredoxin did not diminish the biological activity of the
neutrophil-activating protein, as determined using the nitro blue
tetrazolium dye reduction assay (17).
The protein was concentrated to 1 mg/ml on a Filtron 10K Microsep membrane (Filtron Scandinavia AB, Bjärred, Sweden). Aliquots of 100 µl were labeled with 125I by the Iodogen method (33), yielding on average 5 × 103 cpm/µg.
Thin-layer ChromatographyThin-layer chromatography was performed on glass- or aluminum-backed silica gel 60 HPTLC plates (Merck, Darmstadt, Germany) using chloroform/methanol/water (60:35:8, by volume) as the solvent system. For separation of gangliosides, a solvent system composed of chloroform/methanol/water with 0.25% KCl (50:40:9, by volume) was used, and chloroform/methanol/water with 0.2% CaCl2 (50:55:19, by volume) was utilized for separation of polyglycosyl ceramides. Chemical detection was done with anisaldehyde (34) or the resorcinol reagent (35).
Chromatogram Binding AssayBinding of radiolabeled recombinant neutrophil-activating protein from H. pylori to glycosphingolipids separated on thin-layer plates was done as described previously for the binding of bacteria (36). Mixtures of glycosphingolipids (20-40 µg/lane) or pure compounds (1-4 µg/lane) were separated on aluminum-backed silica gel 60 PHTLC plates. After drying, the chromatography plates were treated with 0.5% (w/v) polyisobutylmethacrylate (Plexigum P28, Röhm, GmbH, Darmstadt, Germany) in diethyl ether for 1 min. The plates were then soaked in phosphate-buffered saline (PBS), pH 7.3, containing 2% (w/v) bovine serum albumin and 0.1% (w/v) NaN3 (Solution 1) for 2 h at room temperature. A suspension of 125I-labeled protein (diluted in Solution 1 to approximately 2 × 103 cpm/µl) was sprinkled over the plates, and the plates were incubated for 2 h at room temperature and then washed six times with PBS. Autoradiography was performed for 6-24 h using XAR-5 x-ray film (Eastman Kodak Co.) with an intensifying screen.
Microtiter Well AssayThe microtiter well binding assay was performed as described previously (37). In short, serial dilutions (each dilution in triplicate) of pure glycosphingolipids in methanol were applied in microtiter wells (Cooks M24, Nutacon, Holland). When the solvent had evaporated, the wells were blocked for 2 h with 200 µl of Solution 1. Then 50 µl of radiolabeled neutrophil-activating protein diluted in Solution 1 (approximately 2 × 103 cpm/µl) was added per well and incubated for 4 h at room temperature. After washing six times with Solution 1, the wells were cut out, and the radioactivity was counted in a gamma counter.
Hemagglutination AssayHuman erythrocytes (blood group ARh+ and ORh+) were washed three times with 0.9% NaCl (w/v) and suspended in PBS to 3% (by volume). The erythrocyte suspension (25 µl) was added to serial dilutions (in PBS) of the neutrophil-activating protein from H. pylori in microtiter wells (50 µl/well, starting at 1 mg/ml), and the resulting dilutions were incubated for 30 min at room temperature then visually inspected (38). As a control, the hemagglutination of the lectin from Erythrina cristagalli (39) was tested in parallel.
Radiolabeled HPNAP was incubated
with separated glycosphingolipids on thin-layer chromatography plates,
and bound protein was visualized by autoradiography. A distinct binding
to four bands in the acid glycosphingolipid fraction isolated from
human granulocytes was detected (Fig. 1, lane
1). The binding-active compounds migrated below
sialylneolactotetraosyl ceramide. Desialylation of the acid
glycosphingolipids from human neutrophils abolished the binding of the
neutrophil-activating protein (Fig. 2, lane 6). No binding of the protein to the non-acid glycosphingolipid fraction (Fig. 2, lane 4) or polyglycosyl ceramides (not
shown) from human granulocytes was obtained.
The amount of acid glycosphingolipids obtained from pooled granulocytes
from healthy donors was judged to be too low to permit an isolation of
the binding-active gangliosides. Therefore, a number of reference
glycosphingolipids of known structure was utilized in chromatogram
binding assays in an attempt to circumstantially identify the
structures of the compounds responsible for binding of the
neutrophil-activating protein. The results are summarized in Table
I. The protein bound to sulfatide (Table I, 1; Fig. 1,
lane 5) and sulfated gangliotetraosyl ceramide (Table I, 9) with a detection level at 1 µg and also to neolactotetraosyl ceramide (Table I, 6), gangliotetraosyl ceramide (Table I, 7) and
fucosyl-gangliotetraosyl ceramide (Table I, 8). The binding to the
latter three compounds was, however, sporadic, and the detection level
was 4 µg. Several of the gangliosides tested were consistently
nonbinding. Thus, no binding to GM3 (Table I, 10),
GM1 (Table I, 11), GD1a (Table I, 12),
NeuAc/NeuGc3neolactotetraosyl ceramide (Table I, 13 and 15),
NeuAc
6neolactotetraosyl ceramide (Table I, 14), sialyllactotetraosyl ceramide (Table I, 16), NeuAc
3Lea hexaglycosyl ceramide
(Table, 17), or disialylneolactotetraosyl ceramide (Table, 18), was
detected. However, a consistent binding to the gangliosides
NeuGc
3Gal
4GlcNAc
3Gal
4GlcNAc
3Gal
4Glc
1Cer (Table I, 20; Fig. 1, lane 8) and
NeuAc
3Gal
4GlcNAc
3Gal
4GlcNAc
3Gal
4Glc
1Cer (Table I, 21; Fig. 3, lanes 3 and
4) was obtained, with a detection level at 1 µg.
De-sialylation of these two gangliosides (Table I, 26) abolished the
binding (not shown).
NeuAc
3Gal
4GlcNAc
3Gal
4GlcNAc
3Gal
4GlcNAc
3Gal
4Glc
1Cer (Table I, 22) was also bound by the neutrophil-activating protein (not shown). However, the related gangliosides,
NeuAc
3Gal
4GlcNAc
6(NeuAc
3Gal
4GlcNAc
3)Gal
4GlcNAc
3Gal
4Glc
1Cer (Table I, 23; Fig. 3, lane 5),
Fuc
2Gal
4GlcNAc
6(NeuAc
3Gal
4GlcNAc
3)Gal
4GlcNAc
3Gal
4Glc
1Cer (Table I, 24), and
Gal
3Gal
4GlcNAc
6(NeuAc
3Gal
4GlcNAc
3)Gal
4GlcNAc
3Gal
4Glc
1Cer (Table I, 25), having branches at the 6 position of the terminal Gal of a neolactotetraosyl ceramide core structure, were
nonbinding.
Microtiter Well Assay
To further define the binding
specificities and compare the relative binding affinities, the binding
of radiolabeled HPNAP to selected glycosphingolipids was evaluated in
the quantitative microtiter well assay. As shown in Fig.
4A, the protein bound to
NeuGc3Gal
4GlcNAc
3Gal
4GlcNAc
3Gal
4Glc
1Cer
with half-maximal binding occurring at 2 ng/well. Binding to
SO3-Gal
3GalNAc
4Gal
4Glc
1Cer and sulfatide
(Fig. 4B) was also obtained, with halfmaximal binding occurring at 2 and 30 ng/well, respectively.
Hemagglutination Assay
Serial dilutions of HPNAP in microtiter wells were tested for ability to agglutinate human erythrocytes. However, no agglutination was obtained, even at the highest protein concentration used (1 mg/ml). The lectin from E. cristagalli, used as a control, caused hemagglutination at approximately 10 µg/ml.
The gangliosides from human neutrophils, the target cells of
H. pylori neutrophil-activating protein, are a complex
mixture. The main components are the GM3 ganglioside and
sialylneolactotetraosyl ceramide. The larger gangliosides are based on
repetitive N-acetyllactosamine units and in many cases carry
Fuc residues linked to the 3 position of GlcNAc. In addition, NeuAc
may be
3- or
6-linked to the terminal Gal of neolactotetraosyl
ceramide and the longer glycosphingolipids with repetitive
N-acetyllactosamine units (18, 19, 40). The ceramide part is
composed of sphingosine and mainly nonhydroxy 16:0, 24:0, and 24:1
fatty acids. This distribution of the fatty acids makes each
glycosphingolipid appear as a double band on thin-layer
chromatograms.
In the present study, four bands with HPNAP-binding activity were
detected in the acid fraction of human neutrophils. When a number of
reference gangliosides from other sources were tested for binding
activity, most of them were nonbinding. However, HPNAP bound distinctly
to NeuAc- and
NeuGc3Gal
4GlcNAc
3Gal
4GlcNAc
3Gal
4Glc
1Cer, which comigrated with the uppermost binding-active double
band in the acid fraction of human neutrophils (Fig. 3, lanes
3 and 4), and to
NeuAc
3Gal
4GlcNAc
3Gal
4GlcNAc
3Gal
4GlcNAc
3Gal
4Glc
1Cer, which migrated in the region of the lower binding-active double band
(not shown).
NeuAc
3Gal
4GlcNAc
3Gal
4GlcNAc
3Gal
4Glc
1Cer and
NeuAc
3Gal
4GlcNAc
3Gal
4GlcNAc
3Gal
4GlcNAc
3Gal
4Glc
1Cer have also been identified in human neutrophils (18, 19, 40).
When the terminal sialic acid of
NeuAc3Gal
4GlcNAc
3Gal
4GlcNAc
3Gal
4Glc
1Cer
was removed, the binding of HPNAP was abrogated. In addition,
NeuAc
3neolactotetraosyl ceramide (Table I, 13) was nonbinding. Taken
together, this indicates that NeuAc
3Gal
4GlcNAc
3Gal
, or possibly NeuAc
3Gal
4GlcNAc
3Gal
4GlcNAc
, is minimally
required for binding of the neutrophil-activating protein.
A further notation is that HPNAP bound to both NeuAc- and
NeuGc3Gal
4GlcNAc
3Gal
4GlcNAc
3Gal
4Glc
1Cer and thus
accepted the presence of an acetyl group (-NHCOCH3) as well
as an N-glycolyl group (-NHCOCH2OH) at C-5 of
the terminal sialic acid. However, N-glycolylneuraminic acid
has not been found in normal human tissues (41).
From the results of this study, it cannot be excluded that HPNAP also
recognized
NeuAc6Gal
4GlcNAc
3Gal
4GlcNAc
3Gal
4Glc
1Cer. However, there is conflicting evidence about the presence of this glycosphingolipid in human neutrophils. Although previously
reported as one of the human neutrophil gangliosides (18), this
compound was not found in the recent thorough study of gangliosides of human neutrophils (19, 40).
NeuAc
6Gal
4GlcNAc
3Gal
4GlcNAc
3Gal
4Glc
1Cer is
present in human erythrocytes (42), but this compound was, unfortunately, not available in pure form. A large number of
monosialoganglioside fractions from human erythrocytes were, however,
tested for binding of the neutrophil-activating protein, but the
only binding-active compounds detected were
NeuAc
3Gal
4GlcNAc
3Gal
4GlcNAc
3Gal
4Glc
1Cer and
NeuAc
3Gal
4GlcNAc
3Gal
4GlcNAc
3Gal
4GlcNAc
3Gal
4Glc
1Cer.
No binding of HPNAP to human neutrophil polyglycosyl ceramides was
obtained. When using the extraction procedures of this study, the more
complex gangliosides are mainly found in the polyglycosyl ceramide
fraction. In human neutrophils, these complex gangliosides have a
poly-N-acetyllactosamine core structure with terminal
NeuAc3 and
3-linked Fuc at one or several of the internal GlcNAc
residues (19, 40). The absence of binding to these compounds indicates that the
3-linked fucose residues interferes with the binding process.
The non-binding of
NeuAc3Gal
4GlcNAc
6(NeuAc
3Gal
4GlcNAc
3)Gal
4GlcNAc
3Gal
4Glc
1Cer
is probably due to steric hindrance imposed by the
6-linked
branches. This could have implications for the interaction of the
protein with glycoproteins and polyglycosyl ceramides. Although the
NeuAc
3Gal
4GlcNAc
3Gal
4GlcNAc
sequence is often present in
the carbohydrate chains of these glycoconjugates, it is mainly found on
multiply branched structures and thus may be nonbinding. This
suggestion was supported by the observed inability of HPNAP to cause
agglutination of human erythrocytes. HPNAP consists of 10 15-kDa
subunits and thus is multimeric. Still, no agglutination of human
erythrocytes was obtained, indicating that the N-linked polylactosamine chains of band 3 of adult human erythrocytes having terminal NeuAc
3 but with
6-linked N-acetyllactosamine
units at internal Gal residues,
Gal
4GlcNAc
6(NeuAc
3Gal
4GlcNAc
3)Gal
4GlcNAc
3 (43), was not recognized by HPNAP. The binding-active
NeuAc
3Gal
4GlcNAc
3Gal
4GlcNAc
3Gal
4Glc
1Cer and
NeuAc
3Gal
4GlcNAc
3Gal
4GlcNAc
3Gal
4GlcNAc
3Gal
4Glc
1Cer are also present in human erythrocytes (42, 44). However, the
density of these minor gangliosides may not be enough to induce hemagglutination. No binding activity was detected in the total acid
fraction of human erythrocytes as shown in Fig. 1, lane 2 and Fig. 3, lane 1.
HPNAP also bound to sulfatide and
SO3-Gal3-GalNAc
4Gal
4Glc
1Cer. If this
binding is a true recognition process or due to complementary ionic
interactions remains to be investigated. However, since sulfated
glycosphingolipids have not been convincingly identified in human
neutrophils, the biological implications of this binding property
are, at present, elusive.
The results from this study do not exclude the possibility that HPNAP
interacts with some granulocyte glycoproteins. However, when proteins
from human neutrophils were extracted (45) and binding of H. pylori neutrophil-activating protein was tested on blotting
membranes, no specific binding was obtained (data not reproduced).
Still, most of the reported oligosaccharides of human neutrophil
glycoproteins also have repetitive N-acetyllactosamine cores
(46). However, the majority of these glycoproteins have terminal
NeuAc6, and the internal GlcNAc residues are substituted with
Fuc
3. The concluded binding-active
NeuAc
3Gal
4GlcNAc
3Gal
(4GlcNAc
) sequence has, however,
been proposed to exist in unsubstituted form in some glycoproteins of
human neutrophils, as e.g. leukosialin (47).
The molecular mechanisms by which H. pylori colonization of the gastric epithelium leads to diseases of the stomach and duodenum are largely unknown. The pathogenesis of H. pylori-induced gastroduodenal diseases is likely to be a complex process, and no single event or phenomenon may be solely responsible for all its clinical manifestations. The major known virulence factors of the bacterium are the urease (48, 49) and the VacA cytotoxin (50). The newly identified neutrophil-activating protein is likely to be involved in the early stages of the disease process, as in gastritis, which is characterized by an abundant accumulation of neutrophils in the superficial gastric mucosa.
In this study, a HPNAP-binding carbohydrate sequence present in human neutrophils was identified by utilizing reference glycosphingolipids from other sources than the target cells. Current studies aim to isolate the binding-active glycosphingolipids from human neutrophils to conclusively establish their structures and obtain more information on the binding epitope. Inhibition studies have now been initiated, and preliminary results indicate that the phagocytosis of the neutrophil-activating protein by preactivated human neutrophils may be partly inhibited by synthetic carbohydrate analogues of the proposed binding-active sequence.3