Is There a Link between the Lipopolysaccharide of Helicobacter pylori Gastric MALT Lymphoma Associated Strains and Lymphoma Pathogenesis?

The aim of this study was to investigate the Lewis antigen expression in Helicobacter pylori gastric MALT lymphoma associated strains in comparison to chronic gastritis only strains. Forty MALT strains (19 cagPAI (−) and 21 cagPAI (+)) and 39 cagPAI frequency-matched gastritis strains (17 cagPAI (−) and 22 cagPAI (+)) were included in this study. The lipopolyssacharide for each strain was extracted using a hot phenol method and the expression of Lex and Ley were investigated using Western Blot. The data were analyzed according to the strains' cagPAI status and vacA genotype. Lex was identified in 21 (52.5%) MALT strains and 29 (74.3%) gastritis strains. Ley was identified in 30 (75%) MALT strains and 31 (79.5%) gastritis strains. There was an association between cagPAI positivity and Lex expression among MALT strains (p<0.0001), but not in gastritis strains (p = 0.64). Among cagPAI (−) strains, isolates expressing solely Ley were associated with MALT with an odds ratio of 64.2 (95% CI 4.9–841.0) when compared to strains expressing both Lex and Ley. vacA genotypes did not modify the association between Lewis antigen expression and disease status. In conclusion, cagPAI (−) MALT strains have a particular Lewis antigen profile which could represent an adaptive mechanism to the host response or participate in MALT lymphomagenesis.


Introduction
Helicobacter pylori was the first bacterium classified as a type I carcinogen (maximum level) [1]. Since its discovery, many research projects have focused on virulence factors or genetic markers but few studies have included H. pylori strains associated with gastric mucosa associated lymphoid tissue (MALT) lymphoma [2][3][4][5]. Gastric lymphoma is considered to be the classic lymphoma of MALT-type of the digestive tract [6]. It is a B lymphoma with a very unusual pathogenesis and evolution which evolves very progressively and stays localized in the stomach for a long time. The development of the lymphoma is directly linked to the H. pylori infection although it is not known why this evolution occurs in only a very small number of infected subjects. A large number of molecular events participating in the lymphomagenesis of MALT lymphomas have already been described, in which a chronic antigenic stimulation plays a pivotal role. Moreover, as the incidence of MALT lymphoma may correlate with different inflammatory cytokines and gene polymorphisms, the role of the host immune response has not been clearly defined yet [7].
The role of H. pylori strains on B-cell proliferation in low-grade MALT lymphoma is well established [8][9][10]. However, some questions remained unanswered. What is the nature of the H. pylori antigens recognized by lymphocytes? How does this recognition occur? Dendritic cells (DC) could play a role in antigen recognition and in inflammatory response in this disease. Indeed, the response of DC to a specific organism depends on the pathways activated in response to the microbial agent and to the cells present in the environment [11]. H. pylori lipopolyssacharide (LPS) is one of the key effectors of Toll Like Receptor-4 and it has been shown that the nature of the Lewis-type antigens expressed on the surface of the LPS of H. pylori determines the interaction with DC via the recognition by a C-type lectin called DC-SIGN at the surface of DC [12]. It has been also suggested that interaction with DC-SIGN could influence the pro-inflammatory response. Lewis negative strains escape binding to DC and induce a strong Th1cell response. In contrast, H. pylori strains that express Le x and/or Le y can bind to DC-SIGN and enhance the production of IL-10 which promotes a Th2-cell response and blocking of Th1-cell activation.
In the context of gastric MALT lymphoma, the chronicity of H. pylori infection is believed to be crucial [6]. According to Suarez et al., the chronic microbial antigenic stimulation observed during persisting H. pylori infection constitutes an antigenic source in autoimmunity which leads to sustained B-cell stimulation, thus favoring lymphoid transformation and lymphoma development [13]. Therefore, in this chronic disease a molecular mimicry, which implies the expression of microbial pathogen motifs shared with the host, could be mandatory to favor H. pylori persistence. Lewis determinants resemble autoantigens because of the molecular mimicry with the fucosylated Lewis antigens [14] expressed by the mucosal chief and parietal cells of the gastric glands as well as on the surface and foveolar epithelia [15]. Le x and Le y (type 2 carbohydrates) are the dominant Lewis antigens in H. pylori LPS expressed by 80-90% of clinical isolates whereas Le a , Le b , H-1 (type 1 carbohydrates), Lewis c, and sialyl Le x are rarely expressed (less than 5%) [16,17]. Three fucosyltransferases FutA, FutB and FutC are involved in Lewis antigen synthesis. FutA and FutB which have an a1,3 and/or a1,4fucosyltransferase activity are required for Le x or Le a antigen synthesis, respectively. FutC which has an a1,2 fucosyltransferase activity transfers an additional fucose to produce Le y or Le b antigens. The three corresponding genes, ie, futA (HP0379), futB (HP0651) and futC (HP0093-94), contain poly-C tracts at the 59 end that regulates their expression by a slipped strand mispairing mechanism [18].
The Le x and Le y antigens expressed on the LPS of H. pylori gastric MALT lymphoma strains has not yet been investigated. Therefore, the aim of this project was to study these LPS structures of H. pylori gastric MALT lymphoma strains in comparison to strains solely associated with chronic gastritis.

Results
Description of the cag pathogenicity island (cagPAI) status and vacuolating cytotoxin gene A (vacA) genotypes of the strains included in the present study As indicated in material and methods, 40 MALT strains (19 cagPAI (2) and 21 cagPAI (+)) and 39 gastritis strains (17 cagPAI (2) and 22 cagPAI (+)) were included in this study.
The 39 gastritis strains were classified according to the signal region as either s1 or s2 or according to the middle region as m1 or m2 of the vacA gene. The distribution of the different alleles, s1m1, s1m2 and s2m2 of the gastritis strains were 16 (41%), 15 (38.5%) and 8 (20.5%), respectively.
Distribution of the Le x and Le y antigens among the strains according to the disease status and virulence factors Le x was identified in 21 MALT strains (52.5%) and Le y in 30 strains (75%). Le x was identified in 29 gastritis strains (74.3%) and Le y in 31 strains (79.5%). Depending on the Lewis antigens expressed, the strains were divided into four groups. Five gastritis strains (12.8%) and 5 MALT strains (12.5%) were Le x/y negative. Three gastritis strains (7.7%) and 5 MALT strains (12.5%) only expressed Le x . Five gastritis strains (12.8%) and 14 MALT strains (35%) solely expressed Le y . Finally, 26 gastritis strains (66.7%) and 16 MALT strains (40%) were both Le x/y positive.
There was a significant association between cagPAI status and Le x expression among MALT strains (p,0.0001), but not in gastritis strains (p = 0.64, NS).
There was a significant interaction between Lewis antigens and cagPAI status in relation to the disease status (P value for interaction = 0.028), therefore we stratified the analysis.
Among cagPAI negative strains, Le x were expressed in 12 gastritis strains (70.6%) and only in 3 MALT strains (15.8%); and Le y in 13 gastritis strains (76.5%) and in 13 MALT strains (68.4%). In a regression model, cagPAI negative strains expressing solely Le y were associated with MALT with an odds ratio (OR) of 64.2 with 95% confidence intervals (95% CIs) 4.9-841.0 when compared to Le x/y positive strains (Table 1).
vacA genotypes did not modify the association between Lewis antigens and disease status.

Discussion
The LPS expressed by H. pylori strains has been suggested to be important for gastric colonization, adherence and immune evasion through a ''camouflage'' mechanism in order to escape the host immune response [19]. The influence of Lewis antigens is also believed to participate in the polarization of the Th1/Th2 inflammatory response. Lewis antigens expressed by H. pylori MALT strains could play a key role in the physiopathology of the disease. We previously described the main virulence factors  Table 2. Signal-sequence coding region of the futA and futB genes and deduced amino acid sequences of 10 Le x/y negative Helicobacter pylori strains. expressed by gastric MALT lymphoma strains except the Lewis antigens [4]. Therefore, in the present study, the Lewis antigen expression was compared between H. pylori MALT strains and the chronic gastritis strains. The gastritis strains were selected on the presence or absence of the cagPAI in order to match the gastric MALT lymphoma strains and to constitute two comparable groups.
Most of the strains (59/69, 85.5%) included in the present study expressed at least one Lewis antigen (Le x and/or Le y ) which is in line with previous reports [17]. Although it is well known that a combination of all several methods (serodot, Western blotting and enzyme-linked immunosorbent assay) may allow a more accurate assessment of Le expression [20], only 10 double Le x/y negative strains were identified and mainly among cagPAI negative strains (60%) as already described. Concerning these double negative strains and according to the number of C repeats, the correlation with the futA and futB sequences was not perfect. Indeed, because 3 of these 10 strains were ''on'' for futA or futB, we cannot exclude that they do not express some of the other less common Lewis antigens [16].
The Lewis antigen patterns were then analyzed according to the cagPAI status and vacA genotypes of the strains. This is an issue that has been extensively discussed previously. A first study published by Wirth et al. showed that Le x expression in H. pylori was correlated with its CagA status [21]. Some studies suggest that such a correlation does not exist [22] or that an inverse correlation exists between CagA positive strains and Le x and Le y expression [23]. Broutet et al. studied the LPS of 155 isolates from atrophic gastritis patients, and identified two main clusters of strains, those which were CagA positive and double Le x/y positive, and a second cluster comprised of CagA negative and solely Le y positive or double Le x/y negative [24]. The authors also showed a correlation with vacA genotypes. However, more recently, Skoglund et al., considering only Le antigen expression, were not able to confirm such a correlation for atrophic gastritis isolates [25]. Nevertheless, CagA positive isolates are believed to be more aggressive and more exposed to the immune system. Therefore, a camouflage by an efficient mimicry process could help them to persist in the stomach of their host. A similar association was found in the present study for Le x and cagPAI positive gastric MALT lymphoma strains. Indeed, the proinflammatory properties of cagPAI positive gastric MALT lymphoma strains have been previously proven using the AGS cell model (most cagPAI positive MALT lymphoma strains harbored a functional cagPAI) but no important proinflammatory factors could be identified in cagPAI negative H. pylori MALT strains [26]. According to Moran, all of the discrepancies between Lewis antigens expression and cagPAI status are believed to be due to the adaptation of H. pylori strains to different populations [19,27]. However, in contrary to the study of Broutet el al., the present study did not identify a correlation between Lewis antigens and vacA genotypes [24].
Concerning the disease status among cagPAI negative MALT strains, a significant association was found with Le y . Although we were not able to quantify the amount of Le y expressed by the strains, the Le y bands obtained by Western Blot were very intense (data not shown). It has been suggested that a high expression of Le y could mask the Le x epitopes [18]. The activity of futC in Le y synthesis is very critical. Indeed, strains that have a relatively high level of futC activity could transform nearly all of the Le x to Le y giving rise to a phenotype expressing solely Le y , whereas relatively lower activity is believed to be responsible for a simultaneous Le x and Le y phenotype. Like futA and futB, futC is regulated by a slipped-strand mispairing mechanism through a polyC tract as well as an imperfect TAA repeat in the mid-region of futC [28].
The futC expression could also be influenced by translational frameshift, transcriptional regulation of certain genes as well as deletion in the promoter region. Therefore, we believe that the high Le y expression in cagPAI negative MALT strains should be studied more extensively on a genetic basis.
What could the role of Le y be in cagPAI negative MALT strains? During H. pylori chronic infection, Le y expression has been implicated in the pathogenesis of atrophic gastritis through the induction of autoreactive antibodies against the gastric mucosa [19]. For example, such antibodies have been found in patients with atrophic gastritis and gastric cancer. Skoglund et al. recently published a paper in which they showed that the HPAG1 strain (which is a Le y positive strain) was able to switch in vivo to both Le x and Le y expression, while the same strain in a mouse model of gastric atrophy remained to exclusively express Le y . The authors concluded that the switch in Lewis antigen expression was linked to different gastric environments, and possibly gastric pH. However, the authors were not able to reproduce this phenomenon in vitro by using culture broth at different pH [25]. Indeed, it has been shown that futC transcription can be regulated by environmental factors such as gastric acid and host immunity [29]. The study of Skoglund et al. indicates that Lewis expression at the time of isolation can be considered as a ''snap-shot'' of a strain at a particular time of the disease status/development reflecting the bacterial adaptation to in vivo conditions. Therefore, the higher expression of Le y in cagPAI negative MALT strains identified in the present study raises the question of the presence of a component of atrophy in patients infected by these types of strains. This particular point has been suggested in long-term infected BALB/c mice [30,31]. The presence of auto anti Le y antibodies in the physiopathology of this disease should also be evaluated.
In the context of gastric MALT lymphoma, intra Th lymphocytes are essentially Th0, and Greiner et al. showed the important role of cytokines produced by Th2 lymphocytes (IL-4, IL-10) [32]. Similarly, animal models for MALT lymphoma (BALB/c mouse) showed a Th2 type response [31,33]. Because most of the MALT strains expressed at least one Lewis antigen, they should be able to polarize a Th2 inflammatory response and to address this major point an in vitro study is currently being performed using DCs.
Our study shows for the first time that gastric MALT lymphoma strains, especially cagPAI negative, can be distinguished from gastritis strains. This particular Lewis profile is probably representative of the adaptation of H. pylori during the microenvironmental changes that the bacterium encounters in the course of infection. Moreover, it could represent an adaptive mechanism to the host response that should be explored in vivo in a murine model of gastric MALT lymphoma, and especially on a long-term survival in the host and modulation of host immune responses (autoimmune antibody production, release of cytokines).

Bacterial strains
A subset of the French gastric MALT lymphoma strain collection provided by the French National Reference Center for Campylobacters and Helicobacters was included in this study: 40 strains from gastric MALT lymphoma patients (24 men and 16 women), mean age 58.4 (13.8)). For all these strains the cagPAI status and vacA genotypes have been previously published: 19 cagPAI (2) and 21 cagPAI (+) [4].
A selection of 39 strains isolated from chronic gastritis only patients (16 men and 23 women, mean age 48.3 (13.2)) were included as a population control. All of these strains were isolated from patients included in the Swedish Kalixanda study which was performed on a general adult population from northern Sweden. The strains included in this study were selected according to their cagPAI status (17 cagPAI (2) and 22 cagPAI (+)) in order to be comparable with the MALT strain collection [34].

LPS purification
All of the strains (previously obtained from single colonies) were defrosted on G plates before LPS extraction. Then H. pylori was grown in brucella broth supplemented with 5% fetal bovine serum and 1% IsoVitaleX Enrichment to exponential phase (OD 600nm between 0.3 and 0.6). An equal amount of bacteria from each strain was harvested by centrifugation and washed once in PBS. The LPS from each bacterial pellet were extracted twice using a hot phenol protocol as already described [18]. After an overnight precipitation in ethanol and sodium acetate, the LPS were dried in air and resuspended in water.

PCR and sequencing
H. pylori DNAs were purified with DNeasy Tissue Kit (Qiagen, Hilden, Germany) according to the manufacturer's recommendations from bacterial pellets obtained from the same brucella broth as those used for LPS extraction. For gastritis strains, the vacA alleles (s and m regions) were detected by PCR as previously described [4,36]. Finally, the 59end parts of genes encoding FutA and FutB were amplified and sequenced for Le x and Le y negative gastritis and MALT strains identified in this study. The sequences of the primers used for futA and futB amplification and sequencing are indicated in Table 3.

Statistical methods
All analyses were conducted using the SAS 9.2 package (SAS Institute, Cary, NC). Correlations between genes were tested by the Spearman method. ORs with 95% CIs derived from unconditional logistic models were used to assess associations between Lewis antigens and disease status, and vacA genotypes and disease status, with adjustments for gender and age. Interaction effects between and among genotypes and Lewis antigens were tested by inclusion of product terms in regression models. Table 3. Sequences of the primers used to amplify and/or sequence futA and futB genes isolated from Helicobacter pylori gastric MALT lymphoma and chronic gastritis strains. indicates that the primer binding sequence is located outside of the coding region of the genes. 2 indicates that the primer binding sequence is located inside of the coding region of the genes. F for a forward primer; R for a reverse primer. doi:10.1371/journal.pone.0007297.t003