BabA dependent binding of Helicobacter pylori to human gastric mucins cause aggregation that inhibits proliferation and is regulated via ArsS

Mucins in the gastric mucus layer carry a range of glycan structures, which vary between individuals, can have antimicrobial effect or act as ligands for Helicobacter pylori. Mucins from various individuals and disease states modulate H. pylori proliferation and adhesin gene expression differently. Here we investigate the relationship between adhesin mediated binding, aggregation, proliferation and adhesin gene expression using human gastric mucins and synthetic adhesin ligand conjugates. By combining measurements of optical density, bacterial metabolic activity and live/dead stains, we could distinguish bacterial aggregation from viability changes, enabling elucidation of mechanisms behind the anti-prolific effects that mucins can have. Binding of H. pylori to Leb-glycoconjugates inhibited the proliferation of the bacteria in a BabA dependent manner, similarly to the effect of mucins carrying Leb. Furthermore, deletion of arsS lead to a decrease in binding to Leb-glycoconjugates and Leb-decorated mucins, accompanied by decreased aggregation and absence of anti-prolific effect of mucins and Leb-glycoconjugates. Inhibition of proliferation caused by adhesin dependent binding to mucins, and the subsequent aggregation suggests a new role of mucins in the host defense against H. pylori. This aggregating trait of mucins may be useful to incorporate into the design of adhesin inhibitors and other disease intervention molecules.

We recently reported that the proliferation of H. pylori can be modulated differently by mucins from different individuals and disease states 21 . Mucins from some of the patients stimulated the proliferation of H. pylori, while mucins from other patients tended to have a growth repressing effect. The latter could not be explained solely by the presence of terminal α 1,4-linked N-acetylglucosamine (α 1,4-GlcNAc), which is a carbohydrate structure with antimicrobial activity present on mucins in the gastric glands 22 , as this structure was not detected in all mucin samples that repressed H. pylori proliferation. The binding ability of H. pylori to mucins is also of importance for proliferation, as H. pylori strains with different mucin binding abilities vary in their proliferative response to mucins 21,23 . Furthermore, interaction with mucins can modify the H. pylori expression of adhesin genes 21 . Taken together, this demonstrates that mucins can regulate the behavior of H. pylori beyond acting merely as a physical barrier and attachment site.
In the present study, we further investigated the relationship between binding and response to mucins. We used H. pylori strains with different binding abilities, including deletion mutants of the babA and sabA adhesin genes, and studied their binding in relation to proliferation and gene expression in response to mucins and glycoconjugates with the ligands Le b and SLe x . In addition, we included an isogenic mutant lacking the histidine kinase sensor protein ArsS, as an arsS deletion mutant has been shown to have an increased expression of sabA 24,25 , and this mutant thus represents a model with potentially increased SabA dependent binding. Combining measurements of optical density, bacterial metabolic activity and live/dead stains enabled elucidation of mechanisms behind the anti-prolific effects of mucins, and we can conclude that H. pylori ligand binding via BabA causes a decrease in proliferation, which appears related to the extent of aggregation caused by binding.

Results
Accurate interpretation of H. pylori proliferation and viability requires careful method selection. Optical density is a widely used measure of bacterial growth, and has been used both by us and other research groups previously with the interpretation that the OD value of bacteria cultured with mucins is directly related to the cell count 21,22 . Although we have verified by CFU counts that for a subset of mucin-bacterial samples this appears accurate, we have now found that several cultures of bacteria with strong binding to the added mucins resulted in high OD 560 values (reading performed after vigorous manual shaking), despite low CFU counts (Fig. 1A, B and E). Therefore, we investigated the relationship of these parameters to develop a method for accurate assessment of proliferation and viability. We found that adhesin dependent bacterial binding to mucins isolated from a gastric tumor (positive for both Le b and SLe x ) and from surface mucosa of normal gastric tissue (normal mucin 3, positive for Le b ) results in the formation of aggregates (Skoog et al. 21 and Fig. 1D) along with high OD 560 readings ( Fig. 1A and C). Dispersion of the aggregates by thorough pipette mixing reversed the increased OD in the H. pylori culture with the tumor mucin, thereby demonstrating that the aggregates were the cause of the increased OD (Fig. 1C). In contrast, when H. pylori J99 were cultured with a mucin rich in the anti-microbial glycan structure α 1,4-GlcNAc, the OD 560 correlated with the CFU count and indicated fewer bacteria than without mucins, even though cultures were only mixed by shaking before OD 560 measurement (Fig. 1C). Microscopy analysis of bacteria cultured with mucins indicated that aggregation is caused by specific adhesin-ligand interaction, as most H. pylori J99 wt were present in large clusters whereas the vast majority of J99Δ babAΔ sabA were unassociated (Fig. 1D). Aggregates may cause inaccurate results if not fully dispersed prior to the plating for CFU, as several bacteria in one spot would appear as one colony. Thus, as a measure of bacterial count, aggregation may cause an enhancing error in the OD 560 measurements, but a diminishing error by the CFU counting method. The metabolic activity (which can be used as a measure of cellular viability) can be measured by adding alamarBlue to the cultures 26 . Both the metabolic activity (alamarBlue signal) and the OD 560 readings after vigorous pipetting were lower after culture with the tumor mucin compared to culture without mucins (p < 0.05, Fig. 1C), and the alamarBlue thus seems to accurately measure H. pylori proliferation even though cultures were only mixed by shaking before and after the addition of alamarBlue. Using a mucin that was positive for Le b and devoid of α 1,4-GlcNAc (normal mucin 5), the OD 560 increased, but was reversed after vigorous pipetting, and both the alamarBlue signal and the CFU counts decreased in cultures with J99 wt, whereas none of these parameters were affected in cultures with J99Δ babAΔ sabA (Fig. 1E). The relationship between the alamarBlue signal and CFU counts is similar to the relationship between OD 560 and CFU counts in the absence of mucins (Fig. 1F): within the concentrations the experiments were performed, a % change in alamarBlue approximately equals the same % change in CFU (Y = 0.9399*X + 1.774, Fig. 1G). In summary, both OD and CFU counts, as a measure of bacterial proliferation, can be misleading when molecules that aggregate bacteria are present. Extrapolating metabolic activity into viability and proliferation appears more accurate in this context although one cannot separate proliferation and viability from each other. Microscopic examination of stains determining the number of live vs. dead cells can add additional information, but are time consuming and impractical if large sample numbers are to be analyzed simultaneously. Therefore, we mainly used alamarBlue to assay proliferation of H. pylori throughout the rest of this study, with microscopic examination on a subset of samples to visualize aggregation.
Binding of H. pylori to Le b -glycoconjugates inhibits the proliferation of the bacteria, similarly to the effect of mucins carrying Le b . H. pylori binding to mucins that carry Le b and SLe x , can inhibit proliferation (Fig. 1B,C and E) 21 . However, mucins carry a large repertoire of glycans and may also carry structures that can enhance growth 21 . Furthermore, strains may potentially vary in how they respond to glycan elements due to differences in genetic composition and thereby glycan response machinery. To study proliferation of H. pylori in response to individual glycans, we focused on strains J99 and P12. Both strains carry the babA gene, and J99 also carry the sabA gene. Strain P12 carries two sabB alleles 27 , which is a result from gene conversion with sabA 28 . SabB has an unknown function and thus P12 lacks the SabA mediated binding. Binding of both J99 and P12 was more pronounced to Le b than to SLe x (p < 0.001, Fig. 2). Furthermore, both J99 and P12 bound better to the mucin derived from a healthy stomach that carried Le b (but not SLe x ) than to the tumor derived mucin Scientific RepoRts | 7:40656 | DOI: 10.1038/srep40656 that carried both Le b and SLe x (p < 0.001, Fig. 2A). To study proliferation in response to individual glycans, we cultured strains J99 and P12 with Le b and SLe x conjugated to human serum albumin. Since adhesion to the SLe x -conjugate was either low or absent, the SLe x -glycoconjugate represents a glycan attached to the same carrier as the Le b -glycoconjugate but that does not cause major aggregation. Measurements of the metabolic activity of H. pylori by alamarBlue reduction demonstrated that Le b -conjugates decreased the metabolic activity of J99 and P12 (p < 0.05, Fig. 2B). Thus, binding of H. pylori to Le b -glycoconjugates inhibits the proliferation of the bacteria, similarly to the effect of mucins carrying these glycans. In contrast, the metabolic activity of P12 was enhanced in the presence of SLe x , which might reflect growth stimulation in response to glycans via an unknown mechanism that is independent of SabA and adhesion, as a similar growth enhancement was obtained in response to mucins in the absence of binding with this strain (below).
The proliferation of the babA deletion mutants is not inhibited by Le b -positive mucins and glycoconjugates. To verify that H. pylori proliferation is inhibited by the glycan binding event specifically, we used adhesin mutants. These were cultured with glycoconjugates as well as with mucins to provide a more complex, in vivo like, glycan environment. In line with the genetic modification, the J99 and P12 babA deletion mutants (Δ babA) bound less to Le b -glycoconjugates and less to mucins isolated from a healthy stomach than the isogenic wt strains (Fig. 3). P12 binding to SLe x was very low as expected due to the lack of sabA, and no   Values are mean ± S.E.M. Stars indicate statistical relationship of mutant binding compared to the binding of the isogenic wt to the same ligand, One-way ANOVA with Bonferroni's post hoc test, *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, whereas the letter a indicates that the binding (depicted after subtraction of the background signal) is statistically different (p < 0.05) from the background signal. Binding of strain J99 (C) and P12 (D) to Le b -and SLe x -glycoconjugates in solution using RadioImmuno assay (n = 9-10). Values are median ± interquartile range. Stars indicate statistical relationship of mutant binding compared to the binding of the isogenic wt to the same ligand, Kruskal-Walllis One-way ANOVA on ranks with Dunn's multiple comparisons test, *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001. difference in binding to SLe x was detected with a P12 mutant where the sabB allele in the sabA locus had been deleted ( Fig. 3B and D). Binding of J99 wt to SLe x was also very low, but although no decrease in binding to SLe x was detected with the J99 sabA deletion mutant (Δ sabA) using the microtiter based assay, a decrease was detected when binding was measured in solution in the presence of isotopes (RIA) ( Fig. 3A and C). Thus, both strains mainly interact with mucins and glycoconjugates via BabA, but we still kept the J99Δ sabA in the below assays as an additional control to verify that the differences between wt and Δ babA strains were not due to procedures associated with the genetic manipulations or clone selection.
J99Δ sabA as well as J99 wt formed aggregates in the presence of both the Le b -glycoconjugate ( Fig. 4) and the tumor mucin sample (tumor mucin referred to as P1 TS in Skoog et al. 21 ) resulting in an increase in OD (Figs 1C and 5B). When the bacteria were mixed by pipetting to break aggregates, the ODs at the end point of these cultures were instead decreased (Figs 1C and 5B), which corresponded to a decreased metabolic activity as measured by alamarBlue reduction (p < 0.001, Fig. 5C and E). The loss of proliferation in response to the Le b -glycoconjugate and the tumor mucin sample was reversed by deletion of babA, which also resulted in a decrease in formation of aggregates (Figs 4 and 5A,C,E: an image of the aggregates formed after culture of J99 with the tumor mucin has been published ref. 21). These results are in line with the OD and metabolic activity of J99 wt and J99Δ babAΔ sabA after culture with the normal mucin 5 in Fig. 1E. Similarly, the decreased metabolic activity of strain P12 wt in the presence of Le b , was reversed by deletion of babA (Fig. 5D). The tumor mucin did not have negative effects on the growth on strain P12, however, deletion of babA lead to enhanced growth in the presence of the mucin and Le b (Fig. 5F), further indicating that this strain has a positive growth response to mucin glycans, which is suppressed by the BabA dependent binding to the mucin. Thus, this mucin has no negative effect on proliferation other than that caused by BabA dependent binding to Le b (and similar structures) present on the mucin.
Binding to Le b affect adhesin gene expression. We have previously shown that H. pylori adhesin expression can be differentially affected in culture with different mucins 21 . Here we investigated the relationship between binding to mucins and expression of the BabA and SabA adhesins in H. pylori J99 after co-culture with mucins, including mucins isolated from 10 individuals, and found that the Pearson product-moment correlation coefficient tended to be negative (p = 0.068, r = − 0.597 and p = 0.085, r = − 0.571, Fig. 6A). All samples contained MUC5AC and some also MUC6, and all but one sample were positive for Le b , while only two samples were positive for SLe x 21 . The very complex environment provided by the mucins carrying in the order of 100 different carbohydrate structures has potential to affect binding and stimulate the bacteria in multiple ways. Thus, even in this very complex surrounding, the overall effect of binding appears to be downregulating the adhesins. Therefore, we examined if the expression of the adhesins was affected by binding to its ligand after 24 h of culture with the SLe xand Le b -glycoconjugates in strain J99. babA mRNA levels decreased after culture in the presence of Le b (p < 0.05), but not in the presence of SLe x (Fig. 6B), and expression of sabA trended towards being decreased after culture with Le b and SLe x (p = 0.09, Fig. 6B).

Absence of ArsS reduces BabA dependent decreases in viability and aggregation. The level of
SabA expression has previously been shown to be increased in a H. pylori strain with deleted arsS compared to its isogenic wt strain 24 . Consistent with these results, J99Δ arsS binding to SLe x slightly increased compared to that of J99 wt (p < 0.05, Fig. 7A). Unexpectedly, binding to Le b was decreased by 80% in J99Δ arsS (p < 0.0001, Fig. 7A), and by 30% in strain P12Δ arsS (p < 0.01, Fig. 7B), as determined using the microtiter based assays. In line with these results, binding to the mucin derived from the gastric tumor (carrying both Le b and SLe x ) increased in J99Δ arsS compared to J99 wt strain (p < 0.05), and binding to mucins derived from a healthy stomach (carrying Le b ) trended towards a decrease in the P12Δ arsS strain. That the differences are less distinct with the mucins may be explained by the presence of a multitude of other structures that might contribute to binding. The decrease in binding to Le b was accompanied by a slight decrease in babA gene expression in J99Δ arsS compared to wt (p < 0.05, Fig. 8A), but BabA protein expression was slightly increased in J99Δ arsS and tended to be decreased by a similar degree in P12Δ arsS ( Fig. 8B and C, p < 0.05 and p < 0.08). In comparison, the level of SabA protein increased 6-fold in J99 Δ arsS compared to wt ( Fig. 8D and E, p < 0.01). Together with that we found a decreased binding to Le b conjugates in solution with the P12Δ arsS compared to wt (i.e. consistent with the microtiter based assay) but not for J99Δ arsS (data not shown) and that ArsS has been shown to be involved in protein trafficking 29 , the results indicate that it is the topographical localization or presentation of the adhesin that determines the level of BabA-dependent binding on microtiter plates and aggregation in solution (Fig. 5), and not the amount of adhesin present. When cultured with Le b or the tumor or normal mucin sample, the metabolic activity of J99 wt decreased (p < 0.05) whereas that of J99Δ arsS was not affected (Fig. 7C). This could be explained by the lesser formation of aggregates (Fig. 4), causing less hindrance to proliferation. In contrast, neither J99 wt nor J99Δ arsS proliferation was affected by SLe x . Similarly, the metabolic activity of P12 wt, but not of P12Δ arsS, decreased when cultured  with Le b , and proliferation of P12Δ arsS was higher than that of the wt when cultured with mucins too (Fig. 7D). In line with that glycans can act stimulatory on this strain in the absence of babA dependent inhibition (Fig. 5F), the mucins enhanced growth of P12Δ arsS. The amplitude of the effects of deleting arsS on binding and proliferation differed between the strains, but decreasing BabA mediated aggregation via arsS deletion provides a second line of evidence for that adhesion dependent aggregation leads to decreased proliferation.

Discussion
Here, we demonstrated that BabA dependent binding of H. pylori to Le b -glycoconjugates inhibits proliferation of the bacteria due to formation of aggregates, similarly to the effect of human gastric mucins carrying Le b , and that these events are regulated by ArsS. Furthermore, binding to Le b decreases babA gene expression, indicating a negative feedback loop of the process. Thus, the results show that mucin-pathogen binding can have effects beyond merely inhibiting bacterial adhesion to the epithelial cell surface. The inhibition of proliferation caused by mucin binding and the subsequent aggregation suggests a new role of mucins in the host defense against H. pylori.
H. pylori have been observed by others to grow in aggregates within the human gastric mucus layer 30 . Growth limiting aggregation caused by mucin binding provides an additional manner for the host to control pathogen numbers alongside with that binding facilitate washing away the pathogen with mucus shedding. Formation of H. pylori aggregates in liquid cultures with mucins has also been reported by others 31 , who discuss that bacteria in aggregates may be protected from the outside environment. Similarly, Pseudomonas aeruginosa, has been shown to grow in aggregates within mucin gels, where aggregated bacteria are more resistant to antibiotics than non-aggregated bacteria of the same strain 32 . Thus, aggregation may also be beneficial for the bacteria protecting it from antibiotics, gastric acid and other harmful factors in their surroundings. Antibiotic resistance of H. pylori is a pressing issue 33 , and bacteria in aggregates may be exposed to subtherapeutic concentration of antibiotics, preventing fast eradication and enabling time to develop resistance. Disrupting aggregates formed in the gastric mucosa during eradication therapy might lead to a higher sensitivity to antibiotics and slower development of resistance.
As the majority of the bacteria were alive after culture with aggregate-causing mucins and glycoconjugates, as demonstrated by live/dead staining, there appears to be little or no direct antimicrobial activity caused by binding. The explanation for this might be that the binding causes formation of aggregates that slow down the proliferation due to physical hindrance or inter-bacterial communication. However, mucins are highly complex molecules, carrying in the order of 100 different glycan structures with potential to cause a multitude of effects. e.g. the ratio of live/dead bacteria was reduced in cultures with some mucins samples independent of aggregation, accompanied by a lower CFU count and reduced OD. These mucin samples are thus able to inhibit the proliferation via antimicrobial mechanisms, in analogy with the previously described "natural antibiotic" effect of terminal α 1,4-linked N-acetylglucosamine 22 , which is present on some mucins. In addition, some mucin samples used in our previous studies increased the OD in the absence of noteworthy aggregation formation 21 . These samples are likely able to stimulate proliferation as suggested, similarly to the effect on strain P12, that we here show grows faster in the presence of glycoconjugates and mucins, when the growth inhibiting BabA dependent binding to Le b is absent. The enhancement in growth could possibly be explained by the use of mucin glycans as a nutrient source or by activation of growth stimulating signaling pathways, although no such abilities have yet been described for H. pylori. H. pylori can thus be differently affected by mucins with different glycosylation. The majority of the patient mucins that we have investigated cause effects on growth ranging from − 40 to + 40%. Although these may appear to be marginal changes compared to for example antibiotics, H. pylori is a slow growing bacteria, present in relatively low numbers in most stomachs. Since infection often is present for 15-20 years before symptoms arise, these effects are likely to have major impact over time, in analogy to compound interest. Stimulation and inhibition of H. pylori proliferation by mucins from different individuals might be contributing factors to the outcome of the infection.
Another novel finding was that deleting arsS lead to a decreased aggregate formation. Growth of the arsS mutants were not inhibited in the presence of Le b -positive mucin and glycoconjugate compared to the corresponding isogenic wt strains, further supporting that it is the aggregation per se that causes the inhibition. There was only a 15% decrease in babA expression in J99Δ arsS compared to wt (p < 0.05), which does not seem proportional to the loss in aggregation, and similarly there were only marginal effects on BabA protein levels. Alternative mechanism for the decrease in binding and aggregation remains to be investigated, but our opinion is that altered surface presentation or accessibility due to changes in outer membrane protein composition (such as the increased expression of SabA) are the likely causes for the decreased aggregation. arsS deletion may also interfere with downstream signaling in response to aggregation.
As mucins can affect the proliferation of H. pylori, a rational hypothesis is that mucins also can affect other aspects of the bacteria, such as the expression of genes relevant for colonization and virulence. Indeed, it has previously been shown that the attachment to mucins and host cells can regulate H. pylori virulence genes 21,34,35 . Furthermore, the expression of two virulence factors (cagA and ureA), that were increased in the J99 wt strain after culture with mucins, was not affected in the isogenic mutant lacking the babA and sabA adhesins 21 . In this study we have investigated how the binding to the Le b -and SLe x -glycans affects the gene expression. Gene expression varies in different growth stages of the bacteria 36 , and thus a difference in growth rate between replicates and experiments may give varying results. When studying the response to mucins, and in particularly gene expression, the viability measurement by alamarBlue is useful when comparing different strains of bacteria to ensure that a difference between them is a cause of genetic strain differences and not a cause of occasional differences in growth. Here we showed that the correlation between binding to human gastric mucins with varying glycosylation and babA expression tended to be negative, and in the presence of Le b , the expression of the babA adhesin is reduced. In addition, our previous study showed a change in babA expression in response to mucins, with a lack of upregulation of babA only with mucins to which H. pylori bound via BabA 21 . Neither here nor in the previous study was there any significant change in babA expression level in response to the tumor mucin sample (referred to as P1 TS in ref. 21), to which there is only a low BabA-mediated binding, indicating that the degree of binding may be of importance for the regulation. Although a subset of human mucins may carry other glycan elements that stimulate BabA expression 21 , our results combined imply that BabA mediated binding represses H. pylori babA expression, inhibiting an increase in BabA in response to mucins where otherwise there may have been such increase, and decreasing it in the absence of such elements. The repression of babA expression in response to Le b binding may thus act as a negative feedback loop. An excessive binding to mucins would allow the bacteria to be washed away along with shedding mucus, decreasing the amount of adhesin expressed would be one way to enable long term colonization.
Although presence of babA in the infecting strain has been associated with a more severe clinical outcome, presence of its ligand Le b in the host has been associated with lower density of H. pylori in the stomach of infected individuals and lower level of gastritis, both in human children and in the rhesus monkey infection model 16,37 . These seemingly contradictory outcomes may be explained by a model where on the one hand binding to Le b on mucins decrease H. pylori growth and bound bacteria also are removed and disseminated with the shedding of the mucus, whereas on the other hand, BabA can provide intimate adherence to glycolipids on the epithelial cell surface. Indeed, it has been reported that low producers of BabA are associated with more severe clinical outcomes compared to high producers or BabA-negative strains 38 , indicating a fine tuned balance between these aspects of virulence and host defense.
In conclusion, mucin glycosylation can directly affect H. pylori binding repertoire and H. pylori adhesin expression. Host ligand presentation on mucins appear to undergo a constant host pathogen adaptation and response process; when H. pylori encounters the mucins that build up the mucus layer, the pathogen binding repertoire changes in response to the mucin glycans. The host responds to infection by changing its mucins and mucin glycans in a time dependent manner 15,16,39,40 , which in turn provoke further adaptations of the pathogen. Furthermore, mucins have the potential to influence H. pylori pathogenicity by affecting its growth and expression of virulence factors. The effect of mucins is mediated by its glycan composition, which may inhibit H. pylori proliferation by adhesion and aggregation of bacteria, as well as by antibiotic effects. Since glycosylation of mucins differs between individuals and varies with disease status, these effects may influence the outcome of the infection.

Methods
Ethics statement. One tumor sample was from our well characterized mucin library, and that sample was collected in 1983 at the IMIM-Hospital del Mar, Barcelona, Spain, before the hospital had an ethics committee. The other samples were obtained after written informed consent and approval of the local ethics committee (Lund University Hospital, Lund, Sweden). The methods applied to these samples were performed in accordance with the committee's regulations.

Isolation of mucins.
The main part of the study was performed on mucins isolated from two gastric specimens; one specimen was from a gastric adenocarcinoma tumor (intestinal type, hereafter referred to as tumor sample) and the other one from macroscopically normal antral mucosa of a tumor-affected stomach (hereafter referred to as normal sample), as evaluated by a clinical pathologist. In addition, a series of additional differentially glycosylated mucins were used for the results shown in Figs 1 and 6. Mucins were isolated using isopycnic gradient centrifugation and characterized for mucin and glycan content as previously described 21,23,41 . In this study we only used mucins present in the supernatant, where most of the MUC5AC and MUC6 molecules usually are found. The mucin content and carbohydrate structures present in the samples are summarized in Table 1.
Gradient fractions containing mucins were pooled together to obtain one sample for each specimen. All samples were extensively dialyzed in phosphate buffered saline (PBS) to remove guanidinium hydrochloride and cesium chloride (CsCl). Mucin concentration in pooled samples was determined by detection of carbohydrate as periodate-oxidisable structures in a microtiter-based assay: Flexible 96-well plates (BD Biosciences, Franklin Lakes, NJ, USA) were coated with mucin sample and left overnight at 4 °C. After washing three times with washing solution (5 mM Tris-HCl, 0.15 M NaCl, 0.005% Tween 20, 0.02% NaN 3 , pH 7.75), the carbohydrates were oxidized by treatment with 25 mM sodium metaperiodate in 0.1 M sodium acetate buffer, pH 5.5 for 20 min in room temperature. The plates were washed again and the wells were blocked with DELFIA blocking solution (50 mM Tris-HCl, 0.15 M NaCl, 90 μ M CaCl 2 , 4 μ M EDTA, 0.02% NaN 3 , 6% sorbitol, 0.1% BSA, pH 7.75) for 1 h. After further washing steps, the samples were incubated for 1 h with 2.5 μ M biotin hydrazide in 0.1 M sodium acetate buffer, pH 5.5, followed by washing again. Europium-labeled streptavidin was diluted 1:400 in assay buffer (50 mM Tris-HCl, 0.15 M NaCl, 20 μ M DTPA, 0.01% Tween 20, 0.02% NaN 3 , 1.5% BSA, pH 7.75) and was added to the wells. After 1 h incubation, the plates were washed six times and then incubated with enhancement solution (0.05 M NaOH, 0.1 M ftalat, 0.1% Triton X-100, 50 μ M TOPO, 15 μ M β -NTA) for 5 min on a shaker. The plates were measured using Wallac 1420 VICTOR 2 plate reader with the Europium label protocol (PerkinElmer, Waltham, MA, USA). The concentrations were calculated from a standard curve of a fusion protein of MUC1, 16TR and IgG2a Fc starting at a concentration of 20 μ g/mL and using seven 1:2 serial dilutions. This method of concentration determination was chosen as all mucins do not come into solution after freeze drying, and determining concentration by freeze drying therefore can contain large errors as well as remove mucin species selectively. Since this study focuses on the effects of carbohydrates, setting the concentration based on the carbohydrate content appear appropriate.
Bacterial strains and culture conditions. H. pylori strains J99 and P12 were cultured on Brucella agar (Brucella Medium Base, Oxoid, Basingstoke, Hampshire, England) supplemented with 10% citrated bovine blood (Svenska Labfab, Ljusne, Sweden), 1% IsoVitox (Oxoid), 4 mg/L amphotericin B, 10 mg/L vancomycin and 5 mg/L trimethoprim in 5% O 2 and 15% CO 2 at 37 °C. Plates or broth were, when required, supplemented with streptomycin (100 mg/L), chloramphenicol (20 mg/L) or kanamycin (25 mg/L). Strain J99Δ arsS was created by transformation of a Δ arsS::rpsLCAT PCR fragment (P38/P43) generated by amplifying regions flanking the deletion by primers P38 with P40 and P41 with P43, the rpsLCAT cassette with P54 and P55 and adjoining these pieces with primers P38 and P43. Insertion and deletion of the 5′ region of the arsS gene was verified by PCR. The Δ arsS::rpsLcat PCR fragment was cloned in pUC19 and used as PCR template for making the other Δ arsS mutant strains described. H. pylori strains J99 wild type (wt), J99Δ babA, J99Δ sabA and J99Δ babAΔ sabA were kindly provided by Prof. Thomas Borén, Umeå University, Sweden. The P12Δ sabAB and Δ babA mutants were constructed by transformation of chromosomal DNA of J99 Δ sabA::Cm or J99 Δ babA::Cm 42 . The mutants were verified by PCR (sabA, babA), Immunoblot analysis (α -SabA and α -BabA) and RIA assay (SLe x and Le b ). Primers used for PCR were the following; SabA-1 and SabA-R for sabA locus, A81 and A19 for babA locus in strain P12. All primer sequences are listed in Table 2. Culture of H. pylori with mucins and glycoconjugates. H. pylori were harvested from agar plates (without chloramphenicol or kanamycin) into PBS and centrifuged at 2500 × g for 3 min. Bacteria were then resuspended and cultured in 6 replicates in 60% Brucella broth, 20% fetal bovine serum (FBS), 20% PBS containing mucins, with a final H. pylori starting concentration of OD 600 0.2 and 50 μ g/mL mucin. A control for normal proliferation was obtained by adding PBS without mucins to the culture medium. A second control was obtained by dialyzing mucin isolation buffer against PBS, parallel to the dialysis of the mucin, and this control gave very similar results to PBS alone. H. pylori were also cultured with 50 μ g/mL of glycoconjugates of the carbohydrate structures Le b and sialyl-Le x coupled to human serum albumin (HSA) to create multivalency (IsoSep AB, Tullinge, Sweden). The control for normal proliferation in this assay was obtained by adding HSA (50 μ g/mL) in PBS to the culture medium, or PBS alone, which gave very similar results. Bacteria were cultured in a total volume of 100 μ L in 96-well plates for 24 or 48 hours at 37 °C under aerobic (20% O 2 , 5% CO 2 ) or microaerobic (5% O 2 , 10% CO 2 ) conditions with similar results. The strains used in this study grow well both under microaerobic and aerobic conditions. The comparisons between growth with different mucins or glycoconjugates were always performed on cultures run simultaneously, using the same starting inoculum, as slight variations in growth rate or length of lag phase may occur between experiments.
Proliferation assay. OD at 560 nm was measured at time points throughout the culturing of H. pylori with mucins and glycoconjugates. In addition, alamarBlue (Molecular Probes, Leiden, The Netherlands) was added to the wells to measure the metabolic activity of the bacteria. The plates were incubated for 1 hour at 37 °C with 10 μ L of alamarBlue per well and measured at OD 560 or at both OD 540 and OD 600 to calculate the relative amount of reduced alamarBlue according to the manufacturer's instructions. At the end of the incubation, bacteria from a subset of wells were cultured on Brucella plates and CFU were counted after 5 days of incubation.
Fluorescence microscopy. Bacteria from two replicate wells were pooled after 24 and 37 h proliferation with mucin samples or glycoconjugates and stained with a LIVE/DEAD BacLight bacterial viability kit (Molecular Probes) according to the manufacturer's instructions. Stained bacteria were applied to a microscopy slide and studied immediately under a fluorescence microscope for red and green fluorescence simultaneously.

RNA extraction, cDNA synthesis and Real-time PCR.
After 24 h culture with mucins or glycoconjugates, 100 μ L of RNAprotect Bacteria Reagent (Qiagen GmbH, Hilden, Germany) was added to each well and 6 wells per sample were pooled. RNA extraction was continued with Qiagen's RNeasy kit, including DNase treatment with Qiagen's RNase-Free DNase Set. cDNA was synthesized from 400 ng RNA using Quantitect Reverse Transcription kit (Qiagen) and real-time PCR was run as previously described 21 with primer sequences listed in Table 3. Gene expression was calculated with 2 −ΔCt and normalized against the 16S rRNA expression. Binding assay. H. pylori were grown on agar plates for 24-48 h and harvested in PBS. The bacteria were centrifuged at 2500 × g for 3 min and then resuspended in Blocking Reagent for ELISA (Roche). Mucin samples were diluted in 4 M GuHCl and glycoconjugates of the carbohydrate structures Le b and sialyl-Le x coupled to human serum albumin were diluted in PBS to 4 μ g/mL and coated in 4 replicates on 96-well polysorp plates (NUNC A/S, Roskilde, Denmark) overnight at 4 °C. The plates were washed three times with PBS 0.05% Tween and the wells were blocked for 1 hour with Blocking Reagent for ELISA containing 0.05% Tween (blocking buffer). After discarding the blocking buffer, bacteria with an OD 600 of 0.1 were diluted 1:10 in blocking buffer and added to the plates, which then were incubated in a bacterial shaker at 37 °C for 2 hours. Then the plates were washed three times, which was repeated between every subsequent incubation step. The plates were incubated for 1 hour at room temperature with rabbit anti-H. pylori serum diluted 1:1000 in blocking buffer, and then another hour with HRP-conjugated donkey anti-rabbit IgG diluted 1:10,000 in blocking buffer. For detection, 100 μ L of TMB substrate was added and the plates were incubated for 15 min before the reaction was stopped with an equivalent amount of 0.5 M H 2 SO 4 . The absorbance was measured in a microplate reader at 450 nm after color stabilization. Control wells without mucin coating but with all bacterial strains as well as wells with all mucin coatings but omitting the bacteria from the protocol above were included in all experiments. The signal in both types of control wells ranged from 0.06-0.1. Data in the figures are shown after subtracting this background signal.