Dissecting Integrin Expression and Function on Memory B Cells in Mice and Humans in Autoimmunity

Immunological memory ensures life-long protection against previously encountered pathogens, and in mice and humans the spleen is an important reservoir for long-lived memory B cells (MBCs). It is well-established that integrins play several crucial roles in lymphocyte survival and trafficking, but their involvement in the retention of MBCs in secondary lymphoid organs, and differences between B cell subsets in their adhesion capacity to ICAM-1 and/or VCAM-1 have not yet been confirmed. Here, we use an autoimmune mouse model, where MBCs are abundant, to show that the highest levels of LFA-1 and VLA-4 amongst B cells are found on MBCs. In vivo blockade of VLA-4 alone or in combination with LFA-1, but not LFA-1 alone, causes a release of MBCs from the spleen into the blood stream. In humans, we find that in peripheral blood, spleens, and tonsils from healthy donors the highest expression levels of the integrins LFA-1 and VLA-4 are also found on MBCs. Consistent with this, we found MBCs to have a higher capacity to adhere to ICAM-1 and VCAM-1 than naïve B cells. In patients with the autoimmune disease rheumatoid arthritis, it is the MBCs that have the highest levels of LFA-1 and VLA-4; moreover, compared with healthy donors, naïve B and MBCs of patients receiving anti-TNF medication have enhanced levels of the active form of LFA-1. Commensurate levels of the active αL subunit can be induced on B cells from healthy donors by exposure to the integrin ligands. Thus, our findings establish the selective use of the integrins LFA-1 and VLA-4 in the localization and adhesion of MBCs in both mice and humans.


INTRODUCTION
Interactions between integrins and their ligands are essential for maintaining the location of leukocytes in general, and mediate the extravasation of cells into tissues as well as their retention in peripheral lymphoid organs (1). In particular, marginal zone (MZ) B cells not only express higher levels of the LFA-1 (αLβ2) and VLA-4 (α4β1) integrins than do their neighbors the follicular (FO) B cells (2), but also depend on these receptors for their retention in the splenic marginal zone; MZ B cells retained in the spleen are released to the circulation when the interaction with ICAM-1 and VCAM-1 is blocked, both in wild-type and in autoimmune mice (2,3). Appropriately, the LFA-1 and VLA-4 ligands, ICAM-1 and VCAM-1, are expressed in the splenic marginal sinus in mice (2), and VCAM-1 is expressed in the perifollicular and marginal zone in the human spleen (4).
Memory B cells (MBCs) are key components of immunological memory, and consequently serve as mediators of disease when they react to self-antigens in autoimmunity. In humans, unswitched MBCs (also referred to as IgM MBCs or MZ-like B cells) express high levels of the LFA-1 subunit β2 (5), and the VLA-4 integrin is involved in the activation of MBCs (6). As well as harboring MZ B cells, the spleen is an important reservoir for MBCs (7)(8)(9); although to date the mechanism of their retention there has not been identified, LFA-1 and/or VLA-4 integrins are obvious candidates for that role. Here we investigate this possibility using SLC −/− (surrogate light chain-deficient) mice, which share features with autoimmune mouse models, including spontaneous formation of germinal centers (GCs) and MBCs (3,(10)(11)(12). Furthermore, in humans we examine the expression pattern of the LFA-1 and VLA-4 integrins in B-cell subsets from secondary lymphoid organs as well as functional properties of these integrins in PB of healthy donors and patients with rheumatoid arthritis.

Human Samples
Peripheral blood mononuclear cells (PBMCs), tonsils and spleens were used for the staining of integrins. The spleens were from children (6-10 years of age) undergoing surgery due to vessel malformations. The study on the splenic samples was approved by the Ethical Committee of Ospedale Pediatrico Bambino Gesù in Rome. Informed consent was obtained from the children's parents. Tonsils were collected from children (1-9 years of age) undergoing tonsillotomy. As no personal information or identity was recorded, no written consent or approval by the (57 ± 12 years of age) positive for both rheumatoid factor and antibodies to citrullinated proteins, except two patients that were positive only for rheumatoid factor (Supplementary Table 1),  were obtained from the Sahlgrenska University Hospital (660-11, approved 2011-09-07; T996-13, approved 2013-12-12 and  334-15, approved 2015-06-11 all by the Ethical Committee of Gothenburg). In terms of the controls for PBMCs, as we did not have sufficient numbers of age-and sex-matched controls we used additional controls that were not matched. Comparing the two control groups used in our study, showed no significant differences in integrin expression. The study was performed following the guidelines of the Declaration of Helsinki.

Flow Cytometry
Single cell suspensions were stained with a cocktail of antibodies using a PBS buffer containing 3% (mouse) or 10% (human) FCS and 1 mM EDTA (see Supplementary Tables 2, 3 for a complete list of Abs recognizing mouse and human antigens) following standard techniques, and analyzed on LSRII TM , FACSVerse TM , FACSLyric TM , or LSRFortessa TM X-20 (all BD Biosciences) flow cytometers. Data were analyzed using FlowJo software version 10 (Treestar Inc.).

Adhesion Assay
CD19 + B cells were purified from PBMCs by negative immunomagnetic sorting (Dynabeads R Untouched TM Human B Cells Kit, Invitrogen, Life Technologies). After washing, cells were resuspended in RPMI with 10% FCS at a concentration of 5 x 10 5 cells/ml. High absorbance plates (Immulon 4 HBX, Extra High Binding, 96 wells, Flat-bottom, VWR) were coated with the ligands (VCAM-1 5 µg/ml; ICAM-1 10 µg/ml) at 37 • C for 2 h in 50 µl carbonate buffer. The plate was decanted and then incubated with 200 µl PBS with 1% BSA (1 g BSA in 100 ml PBS) for 1 h at 37 • C to block non-specific adhesion. The plate was washed once with RPMI before use. 5 x 10 4 cells (in 100 µl RPMI with 10% FCS) were added to the wells and incubated for 30 min at 37 • C. Non-adherent cells were removed by gentle washing. Subsequently, adherent cells were removed by incubating for 15 min on ice in RPMI with 5 mM EDTA. Both adherent and non-adherent cells were washed once and then stained for flow cytometry.
Pre-treatment of PBMCs With ICAM-1 and VCAM-1 and Staining for Active Conformation of αL Integrin Subunit PBMCs from HDs and patients with RA under anti-TNF medication were pre-treated with ICAM-1 (20 µg/ml) and VCAM-1 (10 µg/ml) at 37 • C for 1 h in RPMI with 10% FCS.
Thereafter the cells were stained for the active form of αL with the standard panel of antibodies (Supplementary Table 3).

Statistics
Outliers were removed by using the ROUT method with Q set to 1%. Shapiro-Wilk test was used to evaluate normality. The statistical significance of differences between tissues and treatments was determined using tests as appropriate: an unpaired or paired two-tailed t-test or ordinary one-way ANOVA with Tukey's multiple comparisons, using Graph-Pad Prism version 7 (La Jolla, CA, USA). Significance levels are represented on figures by asterisks: * p < 0.05; * * p < 0.01; * * * p < 0.001; * * * * p < 0.0001.

Sustained Treatment With Anti-integrin Antibodies Depletes MBCs in the Spleen
The integrins of interest in this study are LFA-1 and VLA-4, and their ligands ICAM-1 and VCAM-1 ( Figure 1A).
Starting with a population of mature B cells identified as CD19 + CD93 − CD43 − GL7 − lymphocytes, MBCs were defined as CD80 + CD73 +/− PDL2 +/− based on the differential expression of the CD80, CD73, and PDL2 surface markers (15), (Supplementary Figure 1; Figure 1B). These are to be described in more detail elsewhere (Aranburu et. al. in preparation); here it suffices to note that the MBCs in SLC −/− mice contain mainly IgM-expressing cells ( Figure 1C).
To determine whether the adhesion of mouse MBCs in the spleen depends on integrins, we treated SLC −/− mice with antibodies against LFA-1 and VLA-4. After a 2-week period, the presence of MBCs was significantly reduced (Figure 1D), showing that MBCs rely on the interaction with ICAM-1 and VCAM-1 for their retention in the spleen.

Acute Treatment With anti-VLA-4 Antibodies Induces the Release of MBCs Into PB
To investigate whether the observed integrin-mediated loss of MBCs from the spleen resulted in their accumulation in the circulation, we started by looking at the number of leukocytes in the PB of SLC −/− mice soon (5 h) after the injection of the blocking antibodies. Compared to the injection of control antibodies, leukocyte number more than doubled after the injection of antibodies against both LFA-1 and VLA-4 together, but did not alter significantly when each antibody was used alone (Figure 2A). This is to be contrasted with the situation for the MBCs, where the anti-VLA-4 acted selectively, increasing their release into the blood (Supplementary   feature that is likely to be pertinent to the autoimmune status of the SLC −/− mice.

Integrin Expression Is Elevated on Human MBCs in Peripheral Blood and Secondary Lymphoid Organs
Looking for relevance of our data from SLC −/− mice to human immunity, we investigated the expression patterns of the integrin subunits of interest in human tissues: PBMC, tonsil and spleen (for gating strategy see Supplementary Figure 3). In all three tissues, MBCs (unswitched and switched) expressed higher levels of all four integrin subunits of interest than did their naïve counterparts; these elevations were statistically significant in all cases except for β2 (naïve vs. switched MBCs) and α4 (naïve vs. unswitched MBC) in the spleen (Figures 4A,B,D). Further, both VLA-4 subunits were significantly more highly expressed on switched than on unswitched MBCs from PBMCs and tonsils (Figures 4A,B). Gating on GC B and plasma cells in the tonsils, we also noticed that the levels of all four integrin subunits were higher on the latter (Figure 4C). Thus, in humans as in mice, the integrins of interest here are expressed at higher levels on MBCs than on naïve B cells.

Greater Adhesion Capacity to ICAM-1 and VCAM-1 in Human MBCs Than in Naïve B Cells
To determine whether the enhanced expression of LFA-1 and VLA-4 integrins on human MBCs compared to naïve B cells was related to adhesion capacity, we performed an ex vivo adhesion assay (2) on human B cell subsets. A higher proportion of the naïve B cells were found in the non-adherent fraction (before EDTA) whereas the unswitched and switched MBCs showed the opposite pattern (Figure 5), thus demonstrating a higher adhesion capacity to ICAM-1 and VCAM-1 than the naïve B cells. The ex vivo demonstration of MBC adhesion to the integrin ligands is complementary support of the in vivo data from our mouse model.

Higher Than Normal LFA-1 Expression on MBCs in Patients With RA
To test our observations on autoimmune mice in humans, we chose to investigate patients with RA, following studies by others in which antibody blockade of ICAM-1 in such patients was reported to have beneficial effects on various clinical parameters, although no data on the effects on B cells were published (16). As for the MBCs from healthy donors, we found that the MBCs in peripheral blood from patients with RA expressed higher levels of all four integrin subunits of interest than the naïve B cells ( Figure 6A). Additionally, and as observed in PBMCs and tonsils in healthy donors, both VLA-4 subunits (α4 and β1) appeared at significantly higher levels on switched than unswitched MBCs.
Comparing B cells in peripheral blood from healthy donors and patients with RA revealed some significant elevations of LFA-1 subunits expression in the RA samples: unswitched MBCs had higher levels of both αL and β2, and naïve B cells had a higher level of β2 than their counterparts from HDs ( Figure 6B). As the patients appeared to split into two groups, we looked for an association between the number of swollen/tender joints and integrin expression, but found none (data not shown). In contrast however, dividing the patients into those that receive anti-TNF treatment and those that do not revealed significant differences in LFA-1 expression (Figure 6C). Those treated with anti-TNF had higher expression than HDs of the αL subunit on unswitched MBCs, and of the β2 subunit on all three B cell populations identified. The Expression of the Active Form of αL can be Induced by Pre-treatment With ICAM-1 and VCAM-1 The antibody that was used to stain for β2 integrin expression recognizes the active form of this subunit (17). We decided, therefore, to investigate also the expression levels of the active form of αL using an antibody known to recognize this form, selecting RA patients treated with anti-TNF medication. Our analysis showed that expression of the active form of αL was elevated relative to HDs on naïve, unswitched and switched MBCs from patients with RA, amongst which active αL expression was invariant across the B cell populations ( Figure 7A). Thus, circulating naïve B cells as well as unswitched and switched MBCs in RA patients on anti-TNF medication express the active form of both subunits of the LFA-1 integrin.
It has previously been shown that patients with RA have higher levels of circulating ICAM-1 (18). To see whether the higher expression levels of the active form of αL seen in patients with RA could be induced in healthy donors in the presence of the ligand, we pre-treated PBMCs from both HDs and patients with ICAM-1 and VCAM-1. In consequence, the level of αL expression on naïve, unswitched and switched MBCs from HDs increased, matching those of their counterparts from the RA patient group (Figures 7B,C), which remained unaffected by the pre-treatment. Thus, the presence of ICAM-1 and VCAM-1 can trigger an increase in the activated form of αL expressed on the surface of PB B cells in healthy donors, to levels already expressed on the B cells of patients with RA.

DISCUSSION
Immunological memory depends on the retention of long-lived lymphocytes primed against specific antigens; as a consequence of this requirement, secure retention of memory cells in appropriate locations is essential. It has already been shown that VLA-4 and LFA-1 are crucial for the retention of MZ B cells in the murine spleen (2, 3). These molecules are therefore likely candidates for maintaining a functional MBC population in secondary lymphoid tissues, and as regulators of B cell trafficking. Here, we provide evidence that they are indeed responsible, via their enhanced expression on MBCs, for the selective retention of B cell memory at sites where ICAM-1 and VCAM-1 occur. Our blocking experiments demonstrate that both LFA-1 and VLA-4 have a role to play, but that their differential expression on MBCs relative to MZ B cells provides for selective placement of the subsets in lymphoid tissues, VLA-4 exerting the stronger pull for MBCs, and LFA-1 being the dominant influence on MZ B cells.
Unswitched human MBCs are known to be located in the splenic marginal zone, whereas switched MBCs are found in the perifollicular zone (19,20); our results predict that a significant measure of their positional specificity is mediated by ligation to VCAM-1 expressed in these zones (4), and by inference also ICAM-1. For the unswitched MBCs, which depend on the spleen for their generation and survival (8,21,22), this receptorligand pairing may even be of developmental importance. The enhancement of adhesion capacity of both unswitched and switched MBCs over naïve B cells that we demonstrate is likely to reflect the differential importance for the two subsets of staying in one place.
B cells are one of the key players in RA, as is evident in the beneficial effects of B-cell depletion therapy (23). They are also prominent in a mouse model of RA, collagen-induced arthritis, in which the incidence of arthritis is substantially reduced in the absence of the LFA-1 ligand, ICAM-1 (24). It has been known for some time that ICAM-1 and VCAM-1 are expressed in the synovium and on chondrocytes of patients with RA (25)(26)(27)(28), and for not quite so long that MBCs can be found in RA synovium (29). These previous observations are extended by our findings that levels of the active conformations of the LFA-1 subunits are elevated on the B cells of patients with RA on anti-TNF medication, and that the presence of ICAM-1 and VCAM-1 can induce similarly elevated levels of the active αL subunit on B cells from healthy donors. The implication is that the adhesion capacity of B cells, especially MBCs, may be a mediating factor in the development of sites of inflammation in RA. Our data would predict that the beneficial effects of blocking ICAM-1 in some patients with RA (16) might be due, at least in part, to the dual outcomes of reduction in the ICAM-1 induced elevation of active subunit expression, and prevention of MBCs expressing the active form of LFA-1 from adhering to the synovial tissue.
Integrins expressed on B cells, and in particular MBCs, have been found to be relevant to autoimmune conditions other than RA. In multiple sclerosis, interference with the integrin-mediated trafficking of leukocytes to the brain using a blocking antibody to the VLA-4 subunit α4 (natalizumab) results in B cells being trapped in the circulation (30). In  systemic lupus erythematosus it has recently been shown that a MBC population expresses a transcriptional profile that includes increased levels of the adhesion molecules αL and β1 (31). Simple blocking of integrin molecules as a therapy for such autoimmune states carries the risk of side effects such as infection (32), and efforts are already being directed toward more specific inhibition; for example, inhibitors targeting gut-specific integrins are now under evaluation for Ulcerative Colitis (33).
Collectively, our results demonstrate that LFA-1 and VLA-4 are more highly expressed on MBCs in both mice and human than on the respective circulating naïve B cells, and MBCs in humans adhere better to ICAM-1 and VCAM-1 than do naïve B cells. Murine MBCs are retained in the spleen by both LFA-1 and VLA-4, but VLA-4 holds the dominant influence via higher expression levels. These data constitute significant steps toward the deciphering of the mechanisms behind lymphocyte retention, release, and trafficking. This understanding could reopen the use of a more specific integrin blockade to address several autoimmune diseases; blocking MBCs from entry into as well as forcing their release into the circulation from the spleen and inflammatory sites could break the positive feedback loop that nourishes autoimmune pathology.

DATA AVAILABILITY
All datasets generated for this study are included in the manuscript and/or the supplementary files.

ETHICS STATEMENT
This study was carried out in accordance with the recommendations of the regional ethical review board of Gothenburg and of the local ethical committe of Bambino Gesu Children's Hospital, Rome with written informed consent from all subjects. All subjects gave written informed consent in accordance with the Declaration of Helsinki. The protocol was approved by the regional ethical review board of Gothenburg and by the local ethical committee of Bambino Gesu Children's Hospital, Rome. This study was carried out in accordance with the recommendations of the regional animal ethics committee in Gothenburg, Sweden. The protocol was approved by the regional animal ethics committee in Gothenburg, Sweden.

AUTHOR CONTRIBUTIONS
AC, NG, OG, and I-LM designed the experiments. AC, CF, DC, IG, KT, LSO, NG, OG, RC, TF, and YW carried out or contributed essential reagents and materials for the experiments. AA and SC contributed substantially to the discussions. AC, I-LM, NG, and OG wrote the manuscript with contributions from the co-authors.

FUNDING
This work was supported by the Swedish Science Research Council (2018-03128), the Swedish Cancer Foundation