Published online before print July 25, 2007
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
* Laboratoire des Interactions Muqueuses-Agents transmissibles LIMA, UPR562, UFRs Médecine et Pharmacie, IFR 100 Santé-STIC, Université de Bourgogne, Dijon, France; and
Virologie Moléculaire et Structurale UMR CNRS 2472 INRA 1157 Gif/Yvette, France
1 Correspondence: LIMA, UFRs Médecine et Pharmacie, 7 Boulevard Jeanne d'Arc, 21079 Dijon, France. E-mail: evelyne.kohli{at}u-bourgogne.fr
ABSTRACT
Selection of mucosal sites is an important step in mucosal vaccine development. The intrarectal (IR) route represents an alternative to the oral route of immunization; nevertheless, immune responses induced by this route are not well defined. Here, we studied the early primary B cell response (induction, homing, and phenotype) induced by IR immunization with rotavirus (RV)-2/6 virus-like particles (VLP). Using flow cytometry, we traced RV-specific B cells in different lymphoid tissues and analyzed the expression of 
4ß7 and CCR9, which are important receptors for homing to the gut, as well as CD5, a marker expressed by B1-a cells, which are a major source of natural antibodies. We observed a massive, specific B cell response in rectal follicles, lumbar, and mesenteric lymph nodes but not in Peyer's patches or cervical lymph nodes. A minority of cells expressed 4ß7, suggesting a probable lack of migration to the gut, whereas CCR9 and CD5 were expressed by 30–50% and 30–75% of specific B cells, respectively. Then, we compared the intranasal route of immunization and observed similar B cell frequency and phenotype but in respiratory lymphoid tissues. These results confirm the high compartmentalization of B cell responses within the mucosal system. They show that CCR9 expression, conversely to 4ß7, is not restricted to B cells induced in the gut. Finally, an important part of the RV-specific B cell response induced at the mucosal level during the primary response to VLP is most likely a result of B1-a cells.
Key Words: B-1a cells • mucosa • cell trafficking • vaccination • rodent
INTRODUCTION
Most pathogens use mucosae as portals of entry making mucosal delivery of antigens an important goal of current vaccine development. Selection of mucosal sites represents an important step in mucosal vaccine development, as the mucosa-associated lymphoid tissue is assumed to be compartmentalized—lymphocytes activated in one mucosa-associated lymphoid tissue homing preferentially to mucosal effector sites related to this mucosa [1
, 2
]. Nevertheless, mucosal immune responses are not completely elucidated; moreover, some mucosa such as the rectal mucosa have been studied little. Tracing specific B lymphocytes and analyzing their phenotype from inductive sites to local and distant lymphoid sites is an interesting way to access the early events induced after antigen encounter at the mucosal level. Using this approach and the rotavirus (RV) model of infection in mice, Youngman et al. [3
] have contributed to a better knowledge of B cell responses in intestinal lymphoid tissues. RV infects mature enterocytes of the small intestine causing gastroenteritis and represents a good viral model of mucosal infection and thus of mucosal immunization. In a previous work [4
], we used the same method as Youngman et al. [3
], based on RV GFP-virus-like particle (VLP) binding, to trace RV-specific B cell responses in respiratory and intestinal lymphoid tissues after intranasal (IN) immunization of mice with VLP in the presence of mucosal adjuvant. This strategy has been shown to confer protection against RV infection [5
]. We found a massive, specific B cell response in respiratory lymphoid tissues [nasopharynx-associated lymphoid tissue (NALT) and cervical lymph nodes (CLN)] but not in intestinal lymphoid tissues together with a low expression of the integrin 4ß7, which binds to the mucosal addressin cell adhesion molecule 1 expressed in the gastrointestinal tract [6
] and plays an essential role in homing to the small intestine. These results indicated that B cells probably do not play an essential role in protection in this model. Intrarectal (IR) immunization with RV VLP has been shown also to induce protection [7
, 8
], but the correlates of protection have not been elucidated. In fact, there are few reports about immune responses induced after IR immunization in mice, notably the inductive sites where specific responses are induced, the regional lymph nodes as well as the phenotype of B cells have not been studied. In humans, immunization by the IR route has shown no difference in the expression of 4ß7 on circulating antibody-secreting cells (ASC) compared with oral immunization [9
, 10
], suggesting that this route may be efficient in inducing B cell responses in the gut.
In this study, we used the same model as previously described for the IN route [4
] to trace RV specific B lymphocytes during the early primary immune response induced by IR immunization with VLP with the aims to: 1) verify whether the IR route was more efficient in inducing B cell responses in the small intestine; 2) better delineate inductive and regional sites after IR immunization; and 3) compare the global distribution of specific B cells as well as expression of homing and chemokine receptors 4ß7 and CCR9. In fact, it has been demonstrated that the chemokine receptor CCR9 is specifically expressed on IgA antibody-secreting cells (ASC) expressing 4ß7 and that its ligand CCL25 is produced in the small intestine, suggesting that the pair CCR9/CCL25 could also play an essential role in homing these cells into intestinal Lamina propria (ILP) [11
12
13
].
Finally, to be more precise with the phenotype of RV-specific B cells induced in these conditions, we studied CD5 expression. Indeed, it has been suggested that a part of RV-specific B cells could be innate-like B cells recognizing 2/6 VLP via natural antibodies [4 , 14 ]. CD5+-expressing B cells (B1-a cells) produce natural polyreactive IgM but also secretory IgA (sIgA) antibodies at the mucosal level, which provide early protection against pathogens [15 , 16 ]. CD5+ B cells binding RV double-layered particles have been reported in peripheral B cells from RV acutely infected children [17 ], but information about RV-specific CD5+ B cells at the mucosal level is lacking.
Thus, in this study, we quantified specific B cells after immunization of mice with 2/6 VLP by both routes and studied 4ß7, CCR9, and CD5 expression using flow cytometry (FCM) in different intestinal and respiratory lymphoid tissues: rectal follicles, lumbar and mesenteric lymph nodes (LLN, MLN, respectively), Peyer's patches (PP), CLN, the NALT, and the spleen.
MATERIALS AND METHODS
VLP preparation
Two different VLP, containing VP2 and VP6 (2/6 VLP) or GFP-
92VP2 and VP6, were produced in the baculovirus system as described previously [18
]. Briefly, Sf9 insect cells were coinfected with two recombinant baculoviruses expressing the protein VP6 of the bovine RF strain and an authentic or a modified GFP-VP2 at a multiplicity of infection higher than 5 PFU/cell, collected 5–7 days postinfection and purified by density gradient centrifugation in CsCl.
Immunization and sample collection
Pathogen-free, adult, female BALB/c mice (6–8 weeks of age) were obtained from Iffa-Credo (L'Arbresle, France). Study protocols were approved by the local institutional animal care committee. No mouse had evidence of previous RV infection, as determined by serum antibody titers. Mice were immunized IN or IR with 2/6 VLP on Day 0 with 10 µg 2/6 VLP mixed with 10 µg LT-R192G, a nontoxic mutant of LT, the thermolabile enterotoxin of Escherichia coli (kindly supplied by Dr. John D. Clements, Tulane University Medical Center, New Orleans, LA, USA; ref. [19
]), per dose in a volume of 20 µl. Control mice were inoculated with PBS in the presence of LT-R192G. The same experiments were also carried out with 2/6 VLP alone. Prior to immunization, mice were anesthetized by i.p. administration of a mixture of ketamine (80 mg/kg) and xylazine (16 mg/kg). Mice were killed on Day 7 postimmunization (based on previous results; ref. [4
]), and the different lymphoid tissues, rectal follicles, LLN, MLN, PP, NALT, CLN, and the spleen were removed.
Preparation of rectal follicles, LLN, MLN, PP, NALT, CLN, and spleen cells
Single-cell suspensions were prepared by mechanical dissociation, filtered on 100 µm pore-size nylon meshes and washed with incomplete medium (RPMI-1640 supplemented with 2 mM L-glutamine, 1 mM sodium pyruvate, 0.3% glucose, 100 U penicillin per mL, and 100 µg streptomycin per mL). After lysis of splenic erythrocytes, cells were counted.
FCM assay
To detect RV-specific B cells, which express 4ß7 integrin, CCR9, or CD5 in immunized mice, we used a FCM assay based on the technique described by Youngman et al. [3
]. Cells from the different lymphoid tissues were washed once with PBS-1% BSA-0.1% sodium azide buffer. Pellets containing 2 x 106 cells were resuspended and incubated with a mixture of PE Cy-chrome-labeled anti-B220 (Clone RA3-6B2, PharMingen, San Diego, CA, USA), biotinylated anti-IgD (Clone AMS9.1, PharMingen), PE-labeled anti-4ß7 (Clone DATK32, PharMingen), or anti-CCR9 (Clone 248918, R&D Systems, Minneapolis, MN, USA) or anti-CD5 (Clone 53-7.3, PharMingen) and GFP-2/6 VLP (to detect RV-specific B lymphocytes) for 30 min in the dark at 4°C. Cells were washed and then labeled with streptavidin-RED613 (Gibco-BRL, Scotland, UK). After incubation for 30 min in the dark at 4°C, cells were washed, resuspended in buffer containing PBS-1% BSA-0.1% sodium azide, and analyzed on a flow cytometer (LSR II, Becton Dickinson, San Jose, CA, USA). Approximately 3 x 105 cells were acquired. Analysis was done as described previously [4
]. Small lymphocytes were discriminated from large lymphocytes by their light-scatter profile. Three types of B cell subsets were analyzed: a large B cell subset B220int IgD– representing extrafollicular B cells, a large B cell subset B220high IgD– representing germinal center B cells, and a small B220high IgD– lymphocyte subset consisting of memory and germinal center B cells [3
]. To delineate RV-positive and -negative cell populations and to control for the specifity of the staining, cells were stained, omitting GFP-VLP, and the quadrant position was fitted eventually, after comparison with identical B cell subsets from nonimmunized mice, as described previously. An isotype-matched, irrelevant antibody, PE-rat IgG2a isotype (Clone R35-95, PharMingen) or PE-rat IgG2b isotype (Clone A95-1, PharMingen) was used to draw quadrants for 4ß7– versus 4ß7+, CD5– versus CD5+, or CCR9– versus CCR9+ cells.
Statistics
Results were expressed as means with SEM. Statistical analysis was performed using SigmaStat software. Pair-wise comparisons between groups of mice were performed using Mann-Whitney nonparametric U-test. For all the tests, a P value <0.05 was considered statistically significant.
RESULTS
Distribution of RV-specific B cells induced after one immunization by the IR route with 2/6 VLP and LT-R192G
Day 7 was chosen to analyze B cells on the basis of preliminary results, which confirmed a kinetics similar to that observed with the IN route [4
]. A massive, RV-specific B cell response (7–10%) was observed within the large B220high B cells in rectal follicles, LLN, and MLN but not in PP, CLN, or the spleen (Fig. 1A
). A lower response was observed within the small lymphocytes (2–5%). In the large B220int subpopulation, a significant but low response was observed in rectal follicles (1.2%) and LLN (1.5%) but not in MLN.
 View larger version (18K): [in a new window] |
Figure 1. (A) Frequency of RV-specific B cells (%) within phenotypically defined B220+ lymphocyte subsets in rectal follicles (RF), LLN, MLN, PP, CLN, and spleen on Day 7 after IR immunization with RV 2/6 VLP and LT-R192G. The results are plotted as the means, and error bars represent 1 SEM (n=6–16 mice/group, depending on the lymphoid tissue). *, Data points statistically different from unimmunized mice after one immunization. (B) 4ß7, CCR9, and CD5 expression within RV-specific B220+ lymphocyte subsets in rectal follicles, LLN, and MLN after IR immunization with RV 2/6 VLP and LT-R192G. The results are plotted as the means, and error bars represent 1 SEM (n=3–8, depending on the subpopulation and the lymphoid tissue).
|
4ß7, CCR9, and CD5 expression by RV-specific B cells induced after one immunization by the IR route with 2/6 VLP and LT-R192G4ß7, CCR9, and CD5 expression was analyzed in tissues, where the specific B cell response was found to be >1%, except in rectal follicles for the B220int subpopulation because of the low number of cells analyzed (Fig. 1B)
. The homing receptor 4ß7, which is essential in driving intestinal homing, was not expressed by the large B220high or the small RV-specific lymphocytes and was expressed weakly by the large B220int B cells in LLN (15%). The chemokine receptor CCR9, which is also assumed to play an important role in intestinal homing, was expressed by 
1/3 of RV-specific B cells, except in two cases: in rectal follicles, where CCR9 expression was lower (11%) within the B220high subpopulation, and in LLN, where it was higher (49%) within the B220int subpopulation. Finally, considering the massive expansion of RV-specific B cells and the fact that it has been suggested that a part of this response could be attributed to an innate-like response involving natural antibodies expressing B cells, we also studied CD5 expression. We found a high frequency of CD5-expressing B cells among RV-specific B cells, 50–76% within the large B cells and 30–50% within the small lymphocytes. Figure 2 shows representative examples of FCM histograms used to quantify specific B cells and CCR9 and CD5 expression.
 View larger version (39K): [in a new window] |
Figure 2. Representative histograms of FCM detection of RV-specific B cells and CCR9 and CD5 expression on Day 7 post-IR immunization with 2/6 VLP and LT-R192G: examples of large IgD– B220high lymphocytes in MLN. Small and large lymphocytes are first discriminated based on light-scatter, and B cell subsets are then determined by expression of IgD and B220 [4
]. RV specificity is determined by binding of GFP-VLP, and CCR9 and CD5 expression is analyzed using isotype-irrelevant antibodies IgG2b (CCR9) or IgG2a (CD5).
|
4ß7 expression was observed as previously shown, whereas 39–45% and 25–77% (25–35% for the NALT and 56–77% for the CLN and spleen) of specific B cells expressed CCR9 and CD5, respectively. CCR9 and CD5 expression was compared in the NALT and rectal follicles (inductive sites) and in CLN and LLN (regional lymph nodes). No significant difference was observed, except in two cases: CCR9 expression was higher within the large B220high subpopulation in the NALT compared with rectal follicles, and CD5 expression was higher within the small lymphocytes in CLN compared with LLN. Considering the fact that these differences do not concern all the subpopulations nor all the tissues, these results suggest that CCR9 and CD5 expression by RV-specific B cells induced after immunization by the IR and the IN routes does not depend on the route of immunization.
 View larger version (18K): [in a new window] |
Figure 3. (A) Frequency of RV-specific B cells (%) within phenotypically defined B220+ lymphocyte subsets in NALT, CLN, MLN, PP, and spleen after IN immunization with RV 2/6 VLP and LT-R192G. The results are plotted as the means, and error bars represent 1 SEM (NALT, CLN, and spleen: n=5–11; MLN and PP: n=3). *, Data points statistically different from unimmunized mice after one immunization. (B) 4ß7, CCR9, and CD5 expression within RV-specific B220+ lymphocyte subsets in NALT, CLN, and spleen after IN immunization with RV 2/6 VLP and LT-R192G. The results are plotted as the means, and error bars represent 1 SEM (4ß7: n=3; CCR9 and CD5: n=4–8, depending on the subpopulation and the lymphoid tissue). *, Data point statistically different from CCR9 expression in rectal follicles after IR immunization within the same subpopulation; **, data point statistically different from CD5 expression in LLN after IR immunization within the same subpopulation. nd, Not determined.
|
 View larger version (29K): [in a new window] |
Figure 4. CCR9 (A) and CD5 (B) expression within RV-specific and nonspecific B cells after IR and IN immunization with RV 2/6 VLP and LT-R192G. The results are plotted as the means, and error bars represent 1 SEM (n=3–8/group, depending on the subpopulation and the lymphoid tissue). *Statistically different from RV-nonspecific B cells.
|
 View larger version (11K): [in a new window] |
Figure 5. Frequency of RV-specific B cells (%) within phenotypically defined B220+ lymphocyte subsets in LLN and MLN after IR immunization with RV 2/6 VLP alone (A). CCR9 and CD5 expression by RV-specific B cells from LLN (B). (A) • Nonimmunized mice (n=4);  immunized mice (n=11). Horizontal bars represent means. (B) The results are plotted as the means, and error bars represent 1 SEM CCR9 expression: n = 6, 2, 6; CD5 expression: n = 4, 5, 1 for the large B220high, the large B220int, and the small lymphocytes, respectively.
|
The IR route of immunization is effective in inducing immune responses and protection against different pathogens in mice [8
, 20
21
22
23
]; however, little has been reported about the sites for local initiation of immune responses, the phenotype, and homing patterns of specific lymphocytes. Rectal follicles have been proposed to be sites for local induction of immune responses in BALB/c mice [24
] and the LLN to drain the rectum and the reproductive tract [25
, 26
]. In this work, we traced RV-specific B lymphocytes induced early in BALB/c mice during the primary response to 2/6 VLP, administered by the IR route in the presence of the mucosal adjuvant LT-R192G. Despite the low number of cells obtained per mouse (
6.105), we could study the response in rectal follicles and found a high frequency of RV-specific B cells (Fig. 1A)
. A high response was also observed in LLN, thus confirming that both sites are involved in immune responses after rectal encounter of antigen. From there, the cells could migrate to the MLN, where an important response was also detected. To our knowledge, this is the first report of a specific B cell response in rectal follicles and LLN in mice.
We did not observe any response in PP, spleen, or CLN. The absence of response in CLN was expected as the consequence of the compartmentalization in the mucosal immune system. The lack of response in the spleen is in accordance with the low serum antibody response observed after one immunization (data not shown) and indicates that a single immunization by the IR route induces a weak systemic B response. The lack of response in PP suggests that B lymphocytes do not migrate to this intestinal-inductive site. Results obtained on Day 17 (data not shown) as well as the kinetics of B cell responses observed previously for the IN route [4 ] do not suggest that Day 7 could be too early. Others have reported the presence of ASC in PP from mice IR-immunized with hepatitis A virus vaccine [21 , 22 ] or with fixed trophozoites of Entamoeba histolytica [23 ]. The antigen, the number of immunizations, and/or the method may explain this discordance. Moreover, in the hepatitis A virus model, the ASC response was not observed in all the mice tested.
Cellular receptors important for homing in the small intestine, where RV replicates, are the integrin 4ß7 and the chemokine receptor CCR9. It is notable that an important fraction of RV-sIgB cells expresses 4ß7 during RV infection in humans and in mice [3
, 27
, 28
]. We showed here that IR immunization with 2/6 VLP induces a low frequency of B cells expressing 4ß7, as observed after IN immunization with RV 2/6 VLP [4
], suggesting that the frequency of ASC in the ILP should probably be low, as shown after IN immunization. Finally, 4ß7 expression in mice may be associated only with B cells induced in the small intestine, and as shown for the IN route of immunization and the VP6 protein [29
], protection against RV infection after immunization with 2/6 VLP by the IR route probably is not dependent on B cells.
CCR9 and its ligand CCL25 have been proposed to play an important role in homing of IgA ASC to the gut. CCL25 is predominantly restricted to the small intestine and thymus [12
]. Also, CCR9 expression and migration to CCL25 are associated specifically with small intestinal IgA ASC [13
, 15
], and IgA ASC, arising in nonintestinal mucosal tissues such as the tonsils and adenoids, largely lack CCR9 [30
]. During acute RV infection in children, Jaimes et al. [28
] have reported that most circulating B cells with RV-sIg express 4ß7 and CCR9 [31
]. In this work, we showed that 30–50% of RV-specific IgD– B cells expressed CCR9 after IR immunization (Fig. 1B)
. Moreover, we found the same percentage after immunization by the IN route (Fig. 3B)
, a nonintestinal site. Indeed, we repeated previous IN immunization experiments with 2/6 VLP [4
] to better define the phenotype of B cells induced and showed no significant difference in CCR9 expression between the two routes. In addition, CCR9 expression was not restricted to the B220int subpopulation as previously shown [12
] but was observed among the three B cell subsets. Jaimes et al. [28
], who observed CCR9 expression by sIgM B cells, already outlined conflicting results concerning CCR9 expression, whereas it had been proposed to be restricted to sIgA B cells [30
]. One explanation may be that RV up-regulates CCR9 expression in RV-specific B cells but also in RV-nonspecific B cells, as the frequency of CCR9-expressing cells was not different between RV-specific and nonspecific B cells (Fig. 4A)
. However, a similar expression was also observed in nonimmunized mice (data not shown). Another explanation to these discordant results may be that different methods were used: functional methods based on migration of cells to the chemokine or B cell labeling with CCL25-Fc or antibodies. However, Jaimes et al. [28
] confirmed CCR9 expression by sIgM B cells using migration experiments. Finally, CCR9 expression may not be as restricted as proposed and may not depend on the site of antigen exposure, the chemokine expression pattern being the limiting factor.
We and others [4 , 14 ] have suggested previously that a part of B cells binding RV 2/6 VLP could be innate-like B cells recognizing VLP via natural antibodies. To better understand the importance of these cells, we studied CD5 expression by RV VLP-binding B cells. B1 cells represent the major source of natural antibodies in mice. There currently exists no single marker of B1 cells in mice; however, many of them express CD5 (B1-a cells). Among RV-specific B cells, we showed that a high frequency of RV-specific B cells induced after IR and IN immunization expressed CD5 (Figs. 1B and 3B) , suggesting that an important part of the primary RV B cell response is most likely a result of B1-a cells. This result confirms the importance of B1 cell responses at the mucosal level; indeed, it has been shown that under half of the plasma cells in the gastrointestinal LP is derived from B1 cells, as are many of the IgA+ cells in the respiratory tract [15 ]. Moreover, CD5 expression was observed within the three subpopulations: the large B220high and B220int and the small lymphocytes, i.e., between germinal center and extrafollicular B cells, suggesting that B1-a cells are not excluded from germinal centers. Finally, CD5-expressing B cells in immunized mice were more frequent among RV-specific compared with RV-nonspecific B cells, suggesting that RV induces the preferential expansion of CD5+ B cells. This high frequency of CD5+ cells among RV-specific B cells is in accordance with results reported by Weitkamp et al. [17 ] in children, which show that CD5+ B cells are predominant (60–87%) in peripheral blood from infected children. However, information about RV-specific CD5+ B cells at the mucosal level was not available. B1 cells have also been shown to be reactive with pathogens such as Salmonella and E. coli to ameliorate disease outcome following influenza infection [15 ] and to protect against natural Salmonella typhimurium infection [16 ]. There has been confusion as to whether B1 cells respond to antigen stimulation. Here, we showed an important expansion of B1 cells on Day 7 postimmunization with 2/6 VLP and LT-R192G, but whether proliferation of CD5+ B cells binding RV VLP is a consequence of BCR cross-linking by 2/6 VLP or of other types of B cell stimulation such as TLR activation remains to be determined.
Finally, to determine the role of adjuvant on the expansion, distribution, and phenotype of specific B cells, additional mice were IR-immunized with VLP alone. Of note, this regimen has been shown to be nonimmunogenic and nonprotective [8 ]. The results clearly showed an effect of LT-R192G on the interindividual variability and on B cell migration, as no response was observed in MLN in most mice immunized with VLP alone, even in mice giving a high response in LLN. The same proportion of CD5-expressing B cells was observed within RV-specific B cells in both conditions, suggesting that there is no preferential expansion of B1-a cells in the absence of adjuvant, and LT-R192G modulates B1-a and B2 cell activation if we consider the number of responders and the level of response in the presence of adjuvant. To our knowledge, such effects of mucosal adjuvants have not been reported previously.
In conclusion, these results confirm the high compartmentalization of B cell responses within the mucosal system and do not show any difference in 4ß7, CCR9, and CD5 expression by RV-specific B cells during the early, antigen-driven, primary response induced by IR and IN immunization with 2/6 VLP. It is notable that CCR9 expression may not be restricted to immune responses induced in the gut, and an important part of the RV-specific B cell response after mucosal (IN and IR) immunization with 2/6 VLP is a result of CD5+ B cells, which are most likely B1-a cells producing cross-reactive, innate-like antibody responses.
ACKNOWLEDGEMENTS
This work was supported by a grant from "Program de recherche en Microbiologie ACIM-4-6," by the Université and the Conseil Régional de Bourgogne. We thank Dr. Per Brandtzaeg (Laboratory for Immunohistochemistry and Immunopathology, Rikshospitalet-Radiumhospitalet Medical Centre, Oslo, Norway) for critical reading of the manuscript, Dr. John D. Clements (Tulane University Medical Center, New Orleans, LA, USA) for providing LT-R192G, and Arlette Hammann and Anabelle Sequeira-Le Grand (Plateau technique de cytométrie, INSERM U517, IFR 100 Mort cellulaire et cancer, Dijon, France) for technical assistance with FCM.
Received March 19, 2007; revised June 13, 2007; accepted June 19, 2007.
REFERENCES
4ß7 Virology 305,93-105[CrossRef][Medline]
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
