Research ArticleImmunology

The Tetraspanin CD37 Orchestrates the α4β1 Integrin–Akt Signaling Axis and Supports Long-Lived Plasma Cell Survival

See allHide authors and affiliations

Sci. Signal.  13 Nov 2012:
Vol. 5, Issue 250, pp. ra82
DOI: 10.1126/scisignal.2003113

Abstract

Signaling by the serine and threonine kinase Akt (also known as protein kinase B), a pathway that is common to all eukaryotic cells, is central to cell survival, proliferation, and gene induction. We sought to elucidate the mechanisms underlying regulation of the kinase activity of Akt in the immune system. We found that the four-transmembrane protein CD37 was essential for B cell survival and long-lived protective immunity. CD37-deficient (Cd37−/−) mice had reduced numbers of immunoglobulin G (IgG)–secreting plasma cells in lymphoid organs compared to those in wild-type mice, which we attributed to increased apoptosis of plasma cells in the germinal centers of the spleen, areas in which B cells proliferate and are selected. CD37 was required for the survival of IgG-secreting plasma cells in response to binding of vascular cell adhesion molecule 1 to the α4β1 integrin. Impaired α4β1 integrin–dependent Akt signaling in Cd37−/− IgG-secreting plasma cells was the underlying cause responsible for impaired cell survival. CD37 was required for the mobility and clustering of α4β1 integrins in the plasma membrane, thus regulating the membrane distribution of α4β1 integrin necessary for activation of the Akt survival pathway in the immune system.

Introduction

The importance of signaling by the kinase Akt (also known as protein kinase B) to cell survival, proliferation, and gene expression is well established across many cell types, including immune cells (13). Long-lasting protective humoral immunity is dependent on the generation and selection (survival) of B cells that generate high-affinity antibodies. The typical immune response to T cell–dependent antigens starts with the formation of short-lived antibody-secreting cells (ASCs) in the extrafollicular region of the white pulp in the spleen. These foci appear rapidly, peak in size within 7 to 10 days, and then dissipate as a result of in situ apoptosis, which coincides with the development of germinal centers (GCs) reaching a maximum of about 14 days after immunization (4, 5).

GCs are the principal sites of somatic hypermutation, which involves a programmed process of mutation affecting the variable regions of the gene encoding the immunoglobulin (Ig) heavy chain (VH gene). B cells with the highest affinity Ig receptors are selected to proliferate and differentiate into memory B cells and long-lived ASCs (known as plasma cells), which then accumulate in the bone marrow (6, 7). These long-lived plasma cells secrete protective antibodies over a lifetime, thus contributing to “immunological memory.” Despite their importance, the precise molecular mechanisms and signaling pathways by which GC B cells survive and are selected to become plasma cells remain poorly understood (7). In particular, how plasma membrane proteins efficiently integrate incoming signals into regulated membrane–proximal signaling complexes that are responsible for the survival of ASCs is ill defined. ASCs contain large amounts of the integrin α4β1 (also known as very late antigen–4) (8, 9), and this adhesion molecule is thought to be important for the development of an optimal T cell–dependent humoral immune response. B cells receive antiapoptotic signals from follicular dendritic cells (FDCs) through interactions between the α4β1 integrin and its ligand vascular cell adhesion molecule–1 (VCAM-1) in GCs (10), and these signals are implicated in facilitating B cell activation (11). Loss of VCAM-1 results in a diminished high-affinity antibody response to T cell–dependent immunization (12).

One important mechanism that regulates integrin function is the clustering of integrins and their association with signaling molecules in functional membrane complexes (13, 14). The integrin-mediated signaling pathways (outside-in signaling) that are important for cell survival include the focal adhesion kinase–phosphatidylinositol 3-kinase (PI3K)–Akt pathway and the mitogen-activated protein kinase pathway (15). Tetraspanin proteins, which are characterized by their four highly conserved transmembrane domains, are implicated in organizing integrins, immune receptors, and signaling molecules into functional complexes, which are known as tetraspanin microdomains (1619). Although the molecular mechanisms underlying tetraspanin function are not well understood, tetraspanins are critically important in fundamental cellular processes, including proliferation, migration, differentiation, and, when deregulated, cancer (20, 21). In contrast to most other tetraspanins, the tetraspanin CD37 (which is encoded by the gene Tspan-26 or GP52-40) is found exclusively in the immune system and is most highly abundant on B cells (22). CD37-deficient (Cd37−/−) mice show aberrations in various arms of the immune system, including impaired T cell–dependent B cell responses, T cell hyperproliferation, and altered antigen presentation (2227); however, the role of the tetraspanin CD37 on B cells remains elusive. Here, we report that CD37 is essential for the generation of long-lived IgG1+ plasma cells by controlling α4β1 integrin–mediated activation of the Akt signaling pathway. This study provides insight into the activation of the integrin-Akt survival pathway in the immune system, which is tightly controlled at the plasma membrane by the tetraspanin protein CD37.

Results

CD37-deficient mice display a defective T cell–dependent antibody response that is intrinsic to B cells

The high abundance of CD37 in B cells led us to study humoral immunity in CD37-deficient (Cd37−/−) mice in detail. We studied antibody responses after immunization with the classical T cell–dependent antigen 4-hydroxy-3-nitrophenylacetyl (NP)–keyhole limpet hemocyanin (KLH) in Cd37−/− mice and in wild-type C57Bl/6 littermates. We analyzed sera for the presence of high-affinity anti-NP antibodies, which were reactive with NP3 bound to bovine serum albumin (BSA), and of total anti-NP antibodies, which were reactive with NP20-BSA. In C57Bl/6 mice, IgG1 represents the major antibody isotype produced after immunization with NP-KLH in alum. We observed a marked impairment in the amount of IgG1 produced by Cd37−/− mice 14 days after immunization compared to that in control mice (Fig. 1, A and B). Anti-NP antibody titers of the IgG2 and IgG3 isotypes were also decreased in Cd37−/− mice compared to those in control mice, whereas IgM responses were normal (fig. S1). Nevertheless, the major B cell subsets in the spleen, transitional 1 (T1) and T2, follicular, and marginal zone B cells were found in normal numbers and percentages in Cd37−/− mice (fig. S2).

Fig. 1

CD37-deficient mice display defective T cell–dependent antibody responses. (A and B) Wild-type (WT) (filled circles) and Cd37−/− (open circles) mice were immunized with NP-KLH in alum, and serum was assayed by enzyme-linked immunosorbent assay (ELISA) against (A) NP20-BSA to measure total NP-specific IgG1 and (B) NP3-BSA to measure high-affinity NP-specific IgG1. (C) Sublethally irradiated WT mice were reconstituted with bone marrow from B cell–deficient (μMT) mice and either WT or Cd37−/− mouse B cells to generate WT and chimeric Cd37−/− mice with a CD37-deficient B cell compartment. BM, bone marrow. (D and E) Chimeric WT (filled circles) and Cd37−/− (open circles) mice were immunized, and serum was analyzed for the presence of (D) total and (E) high-affinity NP-specific IgG1, as in (A) and (B). Antibody titers are expressed in arbitrary units and presented as means ± SEM (n = 6 mice). Experiments were performed five times, yielding similar results. *P < 0.03, **P < 0.001.

Because CD37 is also functional on T cells and dendritic cells (DCs) (24, 26), we investigated whether CD37 deficiency in the B cell population was solely responsible for the impaired antibody response. We created chimeric mice by reconstituting the bone marrow of sublethally irradiated C57Bl/6 recipient mice with bone marrow from B cell–deficient mice (μMT mice). In these experiments, wild-type or Cd37−/− mice were used as a source for B cells to generate wild-type and chimeric Cd37−/− mice with a CD37-deficient B cell compartment (Fig. 1C). Analysis of specific antibody production in these mice after immunization revealed that chimeric Cd37−/− mice were ineffective at producing total and high-affinity NP-specific IgG1 titers in serum (Fig. 1, D and E) in a manner comparable to that in intact Cd37−/− mice (Fig. 1, A and B). From these experiments, we conclude that the impaired antibody response in Cd37−/− mice is a B cell–intrinsic defect.

The GC output of ASCs and memory B cells is defective in Cd37−/− mice

The reduced serum IgG1 titers in Cd37−/− mice (Fig. 1) inferred a deficit in the GC response. To determine whether this was the case, we first investigated the early immune response, which is characterized by the production of short-lived ASCs that are usually clustered in extrafollicular foci of the spleen within 1 week of immunization. Histological analysis revealed a normal immune architecture in the spleens of immunized Cd37−/− mice and showed that the IgG1-secreting cells present at day 7 after immunization were indeed aggregated into foci in both wild-type and Cd37−/− mice (Fig. 2A). Thus, the production of short-lived plasma cells in the extrafollicular phase of the immune response appeared normal in Cd37−/− mice.

Fig. 2

CD37 is essential for the production of long-lived IgG1-secreting cells. (A) Immunohistochemical analysis of spleens from WT (left) and Cd37−/− (right) mice on day 7 after immunization with NP-KLH. Spleens were stained with antibodies specific for IgG1 (dark) and IgD (light). Arrows indicate representative foci of IgG1+ ASCs. F, follicles. Scale bar, 200 μm. (B) GCs in spleens from WT and Cd37−/− mice were visualized with antibody against B220 (blue) and the GC marker GL7 (red) on day 14 after immunization. Arrows indicate representative GCs. Scale bar, 200 μm. (C) Quantification of the number of GCs (GL7+) in each B cell follicle in spleens from WT and Cd37−/− mice on day 14 after immunization. Data are means ± SEM (n = 3 mice). Experiments were performed three times, yielding similar results. (D) Frequency of high-affinity, NP-specific, IgG1-secreting cells at days 7, 14, and 35 of the immune response in the spleens (left) and bone marrow (right) of WT (filled bars) and Cd37−/− (open bars) mice as assessed by ELISPOT. Data are means ± SEM (n = 6 mice). *P < 0.01, **P < 0.03. (E) Immunohistochemical analysis of spleens from naïve and immunized (day 14) WT and Cd37−/− mice stained for IgG1-secreting cells (red, arrows) or with an isotype control antibody. Slides were counterstained with hematoxylin (blue). Scale bar, 50 μm. Experiments were performed three times, yielding similar results.

Next, we used immunohistochemistry to analyze the formation of GCs in Cd37−/− mice 14 days after immunization, which is the peak of the GC response. We confirmed the existence of GCs in the spleens of Cd37−/− and wild-type mice at equal frequency and size, indicating that CD37 was not essential for the formation of GC structures (Fig. 2, B and C). The frequency of GCs in spleens at later time points (21 to 35 days after immunization) was also similar between wild-type and Cd37−/− mice (fig. S3A).

Although GCs developed in Cd37−/− mice, this does not mean that all aspects of their function are normal. We therefore studied the generation of IgG1-secreting cells and memory B cells in the spleens of wild-type and Cd37−/− mice after immunization. The frequency of NP-specific IgG1-secreting cells in the spleens and bone marrow was examined by enzyme-linked immunospot (ELISPOT) assays at different days after immunization. Although the production of short-lived, NP-specific, IgG1-secreting cells at day 7 was normal in Cd37−/− mice, a substantial deficit in the number of IgG1-secreting cells became apparent in Cd37−/− mice by day 14 after immunization (Fig. 2D). In the spleen, the frequencies of total and high-affinity ASCs were reduced to about fivefold when compared to those of wild-type mice, and this difference was still apparent at day 35. Measuring the frequency of long-lived ASCs in the bone marrow revealed an even more marked deficit at day 14 (with an 8-fold difference compared to those in wild-type mice), and by day 35, this difference had increased to an ~10-fold reduction in the numbers of total and high-affinity ASCs (Fig. 2D). Histological analysis of spleens confirmed the reduced number of IgG1-secreting cells in Cd37−/− mice compared to those in wild-type mice. Wild-type mice harbored a distinct ASC population that expressed cytoplasmic IgG1 in GCs and splenic white pulp areas, whereas the number of IgG1+ ASCs was reduced in spleens from Cd37−/− mice 14 days after immunization (Fig. 2E).

To investigate whether CD37 was also required for the formation of memory B cells, we quantified NP-specific IgG1+ memory cells by flow cytometry. In wild-type spleens, immunization induced an NP-specific memory population (characterized as B220+, IgM, IgD, and surface IgG1+) of 0.05% of the total splenocytes on day 14 and 0.02% on day 35, consistent with the normal kinetics of the GC reaction in response to a nonreplicating antigen. In the spleens of Cd37−/− mice, the frequency of these cells was about half that in wild-type mice at both time points (fig. S3, B and C). Together, these data suggest that the GC output of both IgG1-secreting plasma cells and memory B cells is substantially impaired in Cd37−/− mice.

Somatic mutation is decreased in Cd37−/− memory B cells

To further investigate the function of GCs in Cd37−/− mice, we quantified VH gene somatic mutation and affinity maturation in B cells in T cell–dependent immune responses. These processes can be directly monitored by assessing the frequency and location of mutations in the VH genes of antigen-specific B cells, which in case of the immune response of C57Bl/6 mice to NP is dominated by a single gene, VH186.2 (28). We therefore sorted single antigen–specific IgG1+ B cells at day 35 of the immune response, recovered their RNA, and generated complementary DNA (cDNA), which was amplified by polymerase chain reaction (PCR) of the VH186.2 gene. These PCR products were then sequenced, and those that were identified as VH186.2 were compared to the germline counterpart over their entirety. This analysis of sequences from Cd37−/− NP-specific B cells revealed a substantially reduced frequency of mutations in each VH gene, which was about half of that of controls [3.2 ± 1.3 (n = 14 sequences) compared to 6.7 ± 3.6 (n = 16 sequences); Table 1]. The proportion of VH sequences containing a mutation at position 33 (tryptophan-leucine) was equal in wild-type and Cd37−/− mice (57 versus 56%), indicating that affinity maturation in itself was not defective. Thus, although GCs developed in the absence of CD37, the reduced frequency of somatic mutation in the memory B cell compartment demonstrated that the function of GCs in Cd37−/− mice was impaired.

Table 1

Distribution of somatic mutations in the VH genes of wild-type and Cd37−/− antigen-specific B cells 35 days after immunization. VH gene cDNA sequences, derived from sorted single antigen–specific IgG1+ B cells (fig. S3), were compared to the germline VH186.2 sequence. DNA mutations were defined as replacement (R) or silent (S) amino acid exchanges and as being within the coding region (CDR) or the framework (FW) region (55). Random mutagenesis of VH186.2 would produce an R/S ratio of 6.2:1 within CDR1–2 and of 3.1:1 within FW1–3 (28). The overall frequency of mutations in wild-type and Cd37−/− B cells was statistically significantly different.

View this table:

B cell apoptosis is increased in the GCs of Cd37−/− mice

To investigate the underlying cause of the defective IgG1 response in Cd37−/− mice, we studied both the proliferation and the survival of B cells in the GCs of Cd37−/− mice, both of which determine the GC output of ASCs and memory B cells. Whereas the proliferation of purified Cd37−/− B cells stimulated with anti-CD40 or anti-IgM F(ab)2 fragments in vitro was normal (fig. S4), we noticed a substantial reduction in B cell survival (Fig. 3). Apoptosis, which selects the high-affinity B cells during the GC response, was analyzed by terminal deoxynucleotidyl transferase–mediated deoxyuridine triphosphate nick end labeling (TUNEL) staining of spleens 14 days after immunization (Fig. 3A). We observed clear TUNEL-positive clusters in GCs (visualized by peanut agglutinin staining) of wild-type and Cd37−/− spleens. Wild-type and Cd37−/− NP-specific ASCs (characterized as CD138int and IgG1+) were stained with annexin V and quantified by flow cytometry to determine the frequency of apoptotic cells. The percentage of apoptotic IgG1-secreting cells in Cd37−/− spleens was substantially increased compared to that in wild-type ASCs on different days after immunization, whereas total B220+ cells from the spleens of immunized wild-type and Cd37−/− mice did not differ in their viability (Fig. 3, B and C). Thus, the development of GCs in Cd37−/− mice appeared normal, as did B cell proliferation, but the survival of Cd37−/− IgG1-secreting cells in spleens was markedly impaired.

Fig. 3

Increased apoptosis in GCs of Cd37−/− mice. (A) Mice were immunized with NP-KLH, and 14 days later, spleens were stained for apoptotic cells (brown) with the TUNEL assay. Scale bar, 50 μm. Experiments were performed four times, yielding similar results. (B) WT and Cd37−/− splenocytes were stained with antibodies against CD138, B220, IgG1 and NP and with annexin V on different days after immunization and were analyzed by flow cytometry. Experiments were performed three times, yielding similar results. (C) Histogram shows the percentages of PI annexin V+ cells in the total B220+ population and in the NP-specific plasma cell population (CD138+ B220int). WT cells are shown as black bars, and Cd37−/− cells are shown as white bars. Data are means ± SEM (n = 3 mice). *P < 0.03. Experiments were performed three times with three mice of each genotype for each time point.

CD37 is essential for α4β1 integrin function and optimal VCAM-1 binding

In an attempt to unravel the mechanism underlying the increased apoptosis of IgG1-secreting cells in the GCs of Cd37−/− mice, we noticed that the impaired antibody production by Cd37−/− mice showed similarities to the impaired humoral responses observed in mice with a postnatal deletion of VCAM-1 (12), the ligand for the α4β1 integrin. These findings, together with the documented interactions of tetraspanins with β1 integrins (29), prompted us to investigate whether CD37 regulated the function of the α4β1 integrin in B cells.

We exploited the α4β1 ligands fibronectin and VCAM-1, which are abundant in the spleen and bone marrow, to investigate α4β1 function in the absence of CD37 with a dynamic flow system. B cells were purified from wild-type and Cd37−/− spleens and were perfused at various shear flow rates through a chamber coated with ligand for α4β1 integrin. Video recordings of individual cells were assessed, and the rolling velocity and the numbers of rolling cells were quantified. We found that Cd37−/− B cells were substantially impaired in their capacity to roll on α4 integrin ligands compared to wild-type B cells (Fig. 4A and movies S1 and S2). Next, we quantified the number of B cells that firmly adhered to α4 integrin ligands under flow conditions. It appeared that Cd37−/− B cells were also defective in their adhesion to VCAM-1 as well as fibronectin under dynamic flow (Fig. 4B and movies S1 and S2). In blocking experiments in which B cells were incubated with α4 integrin–blocking antibodies, we confirmed that rolling and adhesion were indeed dependent on the α4 integrin (fig. S5).

Fig. 4

CD37 is essential for α4β1 function and optimal VCAM-1 binding. (A) Purified splenic WT and Cd37−/− B cells were perfused through a fibronectin-coated chamber. Video recordings of individual cells were read, and rolling velocities and percentages of rolling cells were quantified at different shear flow rates (1 or 2 dyne/cm2). In blocking experiments, cells were incubated with α4 integrin–blocking antibodies (fig. S5). Data are means ± SEM (n = 3 mice for 1 dyne, n = 4 mice for 2 dyne). *P < 0.03. (B) Quantification of the number of WT (black) and Cd37−/− (white) B cells that adhered to fibronectin under flow. Data are means ± SEM (n = 4 mice). *P < 0.03. (C) Binding of soluble VCAM-1–Fc (20 μg/ml) by B cells (CD138 B220+) and ASCs (CD138+ B220int) of WT and Cd37−/− mice 14 days after immunization as assessed by flow cytometry and depicted as the percentage of positive cells. (D) Quantification of the binding of soluble VCAM-1–Fc (20 μg/ml) by splenic T cells (CD3+), B cells (CD138 B220+), monocytes (CD14+), and ASCs (CD138+ B220int) of WT and Cd37−/− mice 14 days after immunization. Data are means ± SEM (n = 4). *P < 0.03. (E) Binding of different concentrations of soluble VCAM-1–Fc to ASCs (CD138+ B220int) from WT and Cd37−/− mice. Data are means ± SEM (n = 4). Experiments were performed three times with three or four mice of each genotype per experiment, yielding similar results.

We next examined whether CD37 also affected α4β1 integrin function in ASCs, which might have explained the defective production of long-lived ASCs in Cd37−/− mice (Fig. 2). We first investigated CD37 abundance in plasmablasts, plasma cells, and mature B cells from the spleens of C57Bl/6 wild-type mice by quantitative reverse transcription PCR because antibodies against murine CD37 are not available. CD37 mRNA was highly abundant in mature B cells (B220+, CD138) up to the plasmablast stage (CD138+, CD43+), whereas plasma cells (CD138+, CD43) had a lower amount of CD37 mRNA (fig. S6).

To determine the affinity status of the α4β1 integrin, we investigated the ability of wild-type and Cd37−/− ASCs to bind to soluble VCAM-1, which is known to bind only to high-affinity α4β1 integrins (30). We analyzed T cells, B cells, monocytes, and ASCs (CD138+, B220int) from wild-type and Cd37−/− mice for their ability to bind to VCAM-1 and found that only the wild-type ASCs showed potent soluble VCAM-1 binding, whereas the percentage of ASCs with high-affinity α4β1 integrin was substantially decreased in the absence of CD37 (Fig. 4, C and D). Although α4β1 integrin was present on both the CD138 and CD138+ B cell populations, only the CD138+ B220int ASCs, but not the CD138 B220+ B cells, displayed high-affinity α4β1 molecules, which was demonstrated by their ability to bind to soluble VCAM-1, whereas B220+ CD138 B cells (from wild-type and Cd37−/− mice) were not effective in VCAM-1 binding (Fig. 4, C and D). These results were confirmed by quantifying the binding of a range of different concentrations of VCAM-1 to wild-type and Cd37−/− ASCs (Fig. 4E). Together, these data demonstrate that CD37 is essential for the function of the α4β1 integrin.

α4β1 integrin–mediated antiapoptotic signaling through Akt is dependent on CD37

To verify whether the importance of CD37 in protecting ASCs from apoptosis (Fig. 3) was indeed dependent on α4β1 integrin, we investigated the survival of wild-type and Cd37−/− ASCs ex vivo in the absence and presence of VCAM-1. Splenocytes from immunized wild-type and Cd37−/− mice were labeled with antibodies against CD138 and B220, incubated with VCAM-1, and analyzed for annexin V and propidium iodide (PI) staining to measure apoptosis. In the absence of VCAM-1, the percentages of annexin V+ wild-type and Cd37−/− cells were comparable, demonstrating that Cd37−/− ASCs were not intrinsically more prone to apoptosis than wild-type cells (Fig. 5A). However, in the presence of VCAM-1, Cd37−/− ASCs were substantially less protected from apoptosis than were wild-type ASCs, confirming that CD37 was essential for α4β1 integrin function and the survival of long-lived ASCs.

Fig. 5

CD37 is essential for α4β1 integrin–dependent survival of long-lived ASCs and for optimal signaling through the Akt pathway. (A) ASCs (CD138+ B220int) from WT and Cd37−/− mice 14 days after immunization with NP-KLH were incubated with VCAM-1–Fc (1 μg/ml) or human IgG (control) for 3 hours at 37°C, labeled with annexin V and PI, and analyzed by flow cytometry. The bar graph shows quantification of the percentages of apoptotic (AV+PI) and dead (AV+PI+) cells. Data are means ± SEM. Experiments were performed three times with three or four mice of each genotype per experiment. WT cells are significantly (*P < 0.03) protected from apoptosis by VCAM-1 in contrast to Cd37−/− cells. (B) Fourteen days after immunization, splenocytes from WT and Cd37−/− mice were stimulated ex vivo with α4 integrin–cross-linking antibodies for 10 min at 37°C, after which cells were lysed, and pAkt (Ser473), total Akt (60 kD), and actin (42 kD) were detected by Western blotting analysis. Each lane contains the lysate from 5 × 106 cells of an individual mouse. Stimulation with pervanadate served as the positive control (+). Data are representative of at least three independent experiments. (C) Quantification of the phosphorylation experiments depicting the average ratios of pAkt to total Akt protein and of pAkt to total actin protein in control and α4-stimulated WT and Cd37−/− cells. Data are means ± SEM. *P < 0.05. (D) Confocal microscopic imaging of pAkt localization in WT and Cd37−/− IgG1-secreting cells labeled with antibodies specific for IgG1 (green) and pAkt-Ser473 (blue). Scale bar, 10 μm. Note the impaired activation of Akt in Cd37−/− IgG1-secreting cells upon stimulation with VCAM-1. Experiments were performed three times with three mice of each genotype, yielding similar results. (E) Fourteen days after immunization, splenocytes from WT and Cd37−/− mice were stimulated ex vivo with α4 integrin–cross-linking antibodies for 10 min at 37°C followed by intracellular staining for pBad. Quantification of the percentages of cells containing pBad as assessed by flow cytometry. Thick line, pBad; thin line, isotype control. (F) α4-stimulated WT and Cd37−/− splenocytes were lysed and analyzed by Western blotting for pBad-Ser136 and tubulin as a loading control. Each lane contains the lysate of 5 × 106 cells from an individual mouse. Quantification of the phosphorylation experiments depicting the average ratio of pBad to tubulin protein in α4-stimulated WT and Cd37−/− cells. Data are means ± SEM. *P < 0.01. Experiments were performed twice with three mice of each genotype.

Next, we explored the signaling pathway downstream of α4β1 integrin that was responsible for inducing survival in IgG1-secreting cells. Ligation of α4β1 integrin with VCAM-1 stimulates the PI3K-Akt signaling pathway, which promotes ASC survival (7). We therefore analyzed the activation of Akt in wild-type and Cd37−/− IgG1-secreting cells by investigating the extent of Akt phosphorylation (at Ser473) upon stimulation with VCAM-1 with three different read-out systems. First, detection of phosphorylated Akt (pAkt) by Western blotting revealed aberrant phosphorylation in Cd37−/− splenocytes upon stimulation of the α4 integrin (Fig. 5, B and C). Second, we performed intracellular staining by flow cytometry, and we detected impaired phosphorylation of Akt in Cd37−/− IgG1-secreting cells compared to that in wild-type IgG1-secreting cells upon stimulation with VCAM-1 (fig. S7). Third, we imaged pAkt in wild-type and Cd37−/− IgG1-secreting cells by confocal microscopy. As before, the kinase activity of Akt was apparent in wild-type IgG1-secreting cells after stimulation with VCAM-1 but was largely absent in Cd37−/− ASCs (Fig. 5D). Stimulation with pervanadate induced Akt activation in Cd37−/− ASCs (Fig. 5, B and D), which showed that the Akt pathway itself was not defective in the absence of CD37 but that Akt signaling in response to α4 integrin stimulation was impaired. Finally, we investigated downstream signaling events of Akt that are important for promoting cell survival. Akt directly phosphorylates and inhibits the proapoptotic protein Bad (31, 32). We found that wild-type splenocytes contained pBad (at Ser136) upon α4 integrin stimulation, whereas the extent of Bad phosphorylation was consistently lower in Cd37−/− cells (Fig. 5, E and F), which corresponded to impaired Akt activation in the absence of CD37.

The tetraspanin CD37 is essential for clustering of α4β1 integrin and promotes its mobility in the plasma membrane

To further investigate the molecular mechanism underlying the capacity of CD37 to control α4β1 integrin function, we analyzed whether α4β1 integrin associated with CD37. Co-patching experiments demonstrated that CD37 specifically colocalized with the α4 integrin subunit in distinct clusters in B cell membranes, but not with the integrin αLβ2 (LFA-1), which is also abundant on B cells (Fig. 6A). Quantification of the extent of colocalization between CD37 and the α4 subunit revealed a Manders’ overlap coefficient of 0.6 ± 0.06 as compared to 0.23 ± 0.02 for the αLβ2 integrin (Fig. 6B). We used flow cytometry to establish that the total abundance of α4β1 in the plasma membrane was not dependent on CD37 (Fig. 6C). Next, we used confocal microscopy to visualize the distribution of α4β1 in the plasma membrane of wild-type and Cd37−/− B cells adherent to VCAM-1. Imaging showed a marked difference between wild-type and Cd37−/− B cells in the distribution of α4β1 molecules at the VCAM-1 contact surface (Fig. 6D). In wild-type cells, the localization of α4β1 integrin was densely clustered at the contact site, whereas in Cd37−/− cells, the distribution was more uniformly dispersed in small clusters over a larger area. Quantitative data analysis of >50 cells confirmed these observations and showed that the average mean fluorescence intensity (MFI) per pixel at the contact site was significantly higher in wild-type cells than in Cd37−/− cells (83 ± 6 versus 40 ± 3 counts per pixel, respectively; Fig. 6E). The higher MFI observed at the contact site of wild-type B cells was likely a result of the recruitment of α4β1 integrin clusters into the basal membrane area, which did not occur in Cd37−/− B cells. Furthermore, in wild-type cells (n = 54 cells), the average area of α4β1 integrin fluorescence at the contact site was 10 ± 0.6 pixels, compared to 17 ± 1.1 pixels in Cd37−/− (n = 56 cells) (Fig. 6E and fig. S8). Together, these data suggest that CD37 is important for the clustering of α4β1 integrin in the plasma membrane. To confirm this, we investigated the surface distribution of α4β1 by whole-mount electron microscopy studies (Fig. 6F). In these experiments, α4β1 integrin was labeled with 10-nm gold particles, and we calculated the interparticle distances with a nearest-neighbor distance algorithm, as described previously (13), and divided them into three classes: 0 to 50, 50 to 100, and >100 nm. We identified substantially larger nearest-neighbor distances in Cd37−/− cells and less beads per cluster as compared to wild-type cells, demonstrating that α4β1 was indeed present in smaller clusters in the absence of CD37.

Fig. 6

The clustering and mobility of α4β1 integrins in the plasma membrane are controlled by CD37. (A) The co-patching of CD37 (green) and α4β1 integrin (red) was studied on human Raji B cells by confocal microscopy (merge, colocalization in yellow). To exclude nonspecific colocalization, we also performed co-patching of CD37 and the αLβ2 integrin. (B) Quantification of the colocalization by the Manders’ overlap coefficient. Data are means ± SEM (n = 15 cells). (C) Comparable surface abundance of α4β1 integrin as determined by flow cytometric analysis of specific antibody staining (thick) versus isotype control antibody (thin) of B220+ splenic B cells from WT and Cd37−/− mice. (D) WT and Cd37−/− B cells labeled with antibody against α4β1 (Alexa Fluor 488–conjugated 9C10) were analyzed at the VCAM-1 contact site at 37°C by confocal microscopy. Three representative images of WT and CD37−/− cells are shown. (E) The average MFIs and areas of fluorescence of α4β1 labeling were measured in WT (n = 54) and Cd37−/− (n = 56) cells. Cd37−/− B cells contain significantly reduced α4β1 integrin recruitment and reduced clustering. Data represent means ± SEM. **P < 0.01. (F) CD37 deficiency leads to impaired topography and surface clustering of α4β1 molecules. Whole-mount electron microscopy images of WT and Cd37−/− antigen-presenting cells labeled with α4β1-specific 10-nm gold particles. Scale bars, 200 nm. Nearest-neighbor distances (left) and percentages of beads involved in cluster formation (particles <50 nm apart) (right) were calculated as described previously (13). Data are means ± SEM (n = 5 mice). *P < 0.05, **P < 0.01. (G) The mobility of α4β1 integrins was measured in WT and Cd37−/− B cells labeled with antibody against α4β1 (Alexa Fluor 488–conjugated 9C10) by FRAP over time. Half-maximal recovery (τ1/2) was 9.9 ± 1.4 s in WT cells and 5.3 ± 1.0 s in Cd37−/− B cells (n = 10). FRAP curves of α4β1 integrin in WT and Cd37−/− B cells were significantly different (within the 95% confidence interval measured with the two-sample Kolmogorov-Smirnov test; asymptotic P = 2.6252 × 10−97). (H) Quantification of FRAP data revealed that 37 ± 4% of α4β1 integrin molecules were mobile in WT cells versus 10 ± 5% in Cd37−/− B cells. Data shown are pooled from three independent experiments.

Finally, we investigated α4β1 integrin in the plasma membrane of living wild-type and Cd37−/− B cells. The dynamic nature of α4β1 integrin was analyzed by fluorescence recovery after photobleaching (FRAP) microscopy of wild-type and Cd37−/− B cells that were fluorescently labeled for α4β1 integrin and allowed to adhere to VCAM-1. Quantification of the time required for the bleached spot to recover half of its initial integrated intensity (τ1/2) showed that τ1/2 was 9.9 ± 1.4 s for α4β1 integrin molecules in wild-type cells as compared to 5.3 ± 1.0 s in Cd37−/− B cells (Fig. 6G). Calculation of the mobile fraction of α4β1 molecules in the plasma membrane revealed substantially decreased mobility in the absence of CD37 because only 10 ± 5% of the α4β1 molecules were mobile in Cd37−/− B cells compared to 37 ± 4% in wild-type cells (Fig. 6H). Thus, CD37 is important for the mobility of α4β1 molecules in the plasma membrane and for the recruitment of small α4β1 clusters into a large domain at the VCAM-1 contact site.

Discussion

The Akt signaling pathway, a common signal transduction pathway in all eukaryotic cells, is central to cell survival, proliferation, and gene induction. Integrin-dependent activation of the Akt pathway is important for the survival and proliferation of many cell types, including immune cells; however, the molecular mechanisms underlying the regulation of integrin-Akt activity remain incompletely characterized. In particular, it is ill defined how integrins efficiently integrate incoming signals into regulated membrane–proximal signaling complexes responsible for the survival of immune cells. Here, we demonstrated that the tetraspanin protein CD37 on B cells tightly controls the α4β1 integrin–Akt signaling pathway. In the absence of CD37, α4β1 clustering and function were impaired, resulting in a survival defect of IgG+ plasma cells, which provides a previously unrecognized mechanism underlying the regulation of the integrin-Akt pathway at the level of the plasma membrane.

B cell differentiation in GCs is determined by cell proliferation, apoptosis (selection), and somatic hypermutation. Studying these different aspects in Cd37−/− mice revealed that (i) the frequency of somatic mutation in memory B cells was reduced, (ii) the apoptosis of antigen-specific IgG1-secreting cells was substantially increased, and (iii) the GC output of IgG1-secreting cells and memory B cells was defective. These findings collectively demonstrate that the function of GCs is impaired in Cd37−/− mice. This GC defect is rather distinct; in most mutant mouse strains in which the development of long-lived ASCs is impaired, this is often a result of a defect in GC formation (3335). Our bone marrow reconstitution studies demonstrated that the impaired humoral immune response in Cd37−/− mice was B cell–intrinsic (Fig. 1, D and E), excluding CD37 function on follicular helper T cells (36), DCs, or FDCs (37) as the underlying cause. Although the basis of the reduced average frequency of mutations in the VH genes of Cd37−/− B cells remains unclear, we speculate that increased apoptosis among these cells may contribute to this difference. If the probability of apoptosis is proportional with residence within the GC in the Cd37−/− mice, then one would expect increased loss of cells with more mutations, reflecting their longer residence in the GC. Given that the incidence of the position 33 affinity-enhancing mutation that occurs in this response plateaus after about 14 days (38), random but increased loss of persistent GC B cells would explain the distribution that we observed. Clearly, additional experiments are required to resolve the basis of the reduced frequency of mutations in the VH genes of Cd37−/− B cells.

The interaction between B cells and FDCs in GCs is important for providing survival signals for B cells and for driving affinity maturation (5). In particular, the interaction between VCAM-1 and the α4β1 integrin on B cells is critical for B cell survival, selection, and activation (7, 10, 11). These findings, together with the documented role of tetraspanins in integrin function, led us to analyze α4β1 function in Cd37−/− B cells. The α4β1 integrin is distinctive in its capacity to mediate attachment and rolling under flow conditions (39, 40), processes that we observed to be dependent on CD37 (Fig. 4). Our results agree with previous studies on the tetraspanin CD81, which facilitates α4β1 integrin–mediated rolling and arrest under shear flow (41). The α4 integrin is rather exceptional among other immune cell integrins because it preexists in multiple affinity states in the membrane, which enables spontaneous ligand binding (39, 42). Here, we demonstrated that the percentage of ASCs with high-affinity α4β1 integrin was substantially reduced in the absence of CD37, which resulted in less protection from apoptosis in vitro (Fig. 5A) and in vivo (Fig. 3). The defective Akt activation in Cd37−/− IgG1-secreting cells upon stimulation of the α4 integrin, and as a consequence inhibition of the proapoptotic protein Bad (Fig. 5), is consistent with the important role of PI3K-Akt signaling pathway in promoting ASC survival (7). The tetraspanin CD81 is reported to inhibit Akt activity in hematopoietic stem cells (43), which suggests that the coupling of tetraspanins to the Akt signaling pathway is a common cellular mechanism for controlling proliferation and survival. This concept is supported by two studies of human CD37 in malignant B cells (44, 45). Targeting CD37 with engineered antibodies results in activation of both the SHP-1–dependent apoptotic signaling pathway and the PI3K-Akt survival signaling pathway, which warrants clinical development of CD37-specific antibodies for the treatment of B cell malignancies.

Our observations that CD37 is essential for the clustering of α4β1 molecules in the plasma membrane of B cells are of particular interest (Fig. 6). The α4β1 integrin molecules were more dispersed and were present in smaller clusters at the VCAM-1 contact site in Cd37−/− cells than in wild-type cells. This was not a result of a defect in α4β1 integrin protein abundance itself, but CD37 deficiency restricted the mobility of α4β1 and its recruitment into a large cluster at the contact site, which was revealed by molecular analysis of the dynamic nature of α4β1 molecules in living B cells. On the basis of these data, we suggest a model in which CD37 is required for the movement of α4β1 in the plasma membrane, α4β1 clustering, and subsequent outside-in signaling that leads to cell survival (Fig. 7). Given that the lateral mobility of integrins in the plasma membrane is highly dependent on anchorage to cytoskeleton components, it is conceivable that CD37 interferes with this process because the tetraspanin web interacts with the actin cytoskeleton (46, 47). Of particular interest is the observation that paxillin selectively associates with high-affinity α4β1 integrin (48, 49) and regulates tethering and adhesion under flow (50). Studies have indicated that the activation status of αLβ2 integrin is linked to its diffusion characteristics in the plasma membrane, where high-affinity integrin molecules form the mobile fraction on resting lymphocytes (51). Thus, the conformational change of the α4β1 integrin into a high-affinity state may affect its dynamic behavior in the plasma membrane, as was shown for the αLβ2 integrin (51). Furthermore, polarization of integrins into organized microdomains has been implicated in different immunological processes (14), and we demonstrated that loss of CD37 impaired the generation of long-lived IgG+ plasma cells by controlling α4β1 integrin function.

Fig. 7

Model illustrating the importance of CD37 in regulating the mobility and clustering of the α4β1 integrin in the plasma membrane. (Left) In WT cells, CD37 facilitates the mobility of α4β1 integrin molecules in the plasma membrane, which leads to the formation of large clusters of α4β1 integrins and efficient VCAM-1 binding. This induces signaling, which leads to activation of the Akt pathway, which phosphorylates and inactivates the proapoptotic protein Bad and thereby protects ASCs from undergoing apoptosis. (Right) In the absence of CD37, the α4β1 integrin is restricted in its movement, and its topography in the plasma membrane is dispersed. As a consequence, the α4β1 integrin–Akt signaling pathway is defective in Cd37−/− ASCs, which renders these cells more prone to apoptosis.

We previously reported that Cd37−/− mice exhibit substantially increased numbers of IgA+ plasma cells in lymphoid organs compared to those in wild-type mice, which was dependent on interleukin-6 (22). That Cd37−/− IgA-secreting cells are not more prone to apoptosis may be a result of the high abundance of the α4β7 integrin and other adhesion molecules on these cells (9). On the basis of our results, we reason that the molecular interactions between FDCs and IgA-secreting cells as opposed to IgG1-secreting cells are different in GCs. We hypothesize that the interaction between Cd37−/− ASCs and other stromal cells (including reticular cells and fibroblasts) is also impaired because these cells have VCAM-1 (52). Together, our data show that the tetraspanin CD37 is essential for the formation of long-lived, IgG1-secreting plasma cells, which reveals a unique role for CD37 in the late-stage differentiation of B cells. The reduced activity of α4β1 integrin on Cd37−/− B cells led to impaired Akt kinase signaling, increased B cell apoptosis, and defective GC function in vivo. This study provides insight into activation of the integrin-Akt kinase survival pathway in the immune system that is tightly controlled at the plasma membrane by the tetraspanin CD37.

Materials and Methods

Mice

Cd37−/− mice were generated by homologous recombination (23) and backcrossed 10 times to the C57Bl/6J background. Age- and sex-matched C57Bl/6J wild-type mice were obtained from the Walter and Eliza Hall Institute (Melbourne, Victoria, Australia) and Charles River (France). Cd37−/− mice and wild-type littermates were bred at the Austin Research Institute animal facility (Heidelberg, Australia) and the Central Animal Laboratory (Nijmegen, the Netherlands). B cell–deficient μMT mice were bred at the Walter and Eliza Hall Institute. Mice were used at 8 to 12 weeks of age. Animal studies were approved by the Animal Ethics Committee of the Austin and Repatriation Medical Centre and the Nijmegen Animal Experiments Committee.

Immunization

Wild-type and Cd37−/− mice (n = 6) were injected intraperitoneally with 100 μg of NP (NP20/KLH conjugation ratio, 20:1) precipitated in alum. Mice were bled, and sera were collected at days 7, 14, 21, and 35 after immunization. Lymphoid organs (spleens and bone marrow) were frozen in cryoembedding medium (OCT) for immunohistochemical analysis or were processed for flow cytometric analysis.

Reconstitution experiments

Bone marrow chimeras were generated as previously described (53). Briefly, 107 Cd37−/− or wild-type and μMT (Cd37+) bone marrow cells were injected intravenously into sublethally irradiated C57Bl/6 recipients at a ratio of 1 part wild-type or Cd37−/− cells to 4 parts μMT cells. Six weeks after reconstitution, heparinized blood was obtained from chimeric mice, and reconstitution of all major leukocyte populations was monitored by flow cytometry with antibodies against the following markers: CD19 (1D3), GR-1 (RB6-8C5), and F4/80, all of which were obtained from Pharmingen, and CD3 (KT3.1) and CD11b (M1/70), which were generated in-house. The percentages of CD19+, CD3+, GR1+, and F4/80+ cells in peripheral blood demonstrated equally efficient reconstitution in wild-type and chimeric Cd37−/− mice. Two weeks thereafter, mice were challenged with NP-KLH as described earlier.

Detection of antigen-specific Ig and ASCs

After immunization, the titer of NP-specific antibodies in the serum was measured by ELISA with NP20-BSA or NP3-BSA to detect total and high-affinity anti-NP antibodies, respectively. NP-specific monoclonal IgG was used to generate a standard curve from which relative units were derived. The frequency of total and high-affinity NP-specific ASCs in spleens and bone marrow was determined by ELISPOT with NP20-BSA and NP3-BSA used for antibody capture.

Immunohistochemistry and apoptosis in situ

Spleens taken from mice before and after immunization were embedded in OCT. Frozen sections (6 μm) were fixed in acetone. Slides were blocked in 5% goat serum and stained with anti-IgG1 and anti-IgD, anti-CD138, anti-B220, or isotype controls (all of which were obtained from BD Pharmingen), followed by anti-rat biotin (Molecular Probes) and the streptavidin-alkaline phosphatase labeling kit (Vector) with Fast-Red Substrate. GCs were identified with biotinylated antibody against the GC marker GL7 (eBioscience). Slides were counterstained with Mayer’s hematoxylin. TUNEL assays were performed with the ApopTag peroxidase kit according to the manufacturer’s protocol (Chemicon). Splenocytes of immunized mice were cultured in RPMI 1640 medium containing 10% fetal calf serum (FCS) and 50 μM 2-mercaptoethanol for 4 hours and were labeled with IgG1 biotin (streptavidin-Cy5), NP, and annexin V (BioVision), and dead cells were excluded by PI staining. Apoptosis of NP-specific plasma cells was quantified by flow cytometry.

Adhesion experiments under flow

B cells from the spleens of wild-type and Cd37−/− mice were purified (>97%) with B220-MACS beads (Miltenyi Biotec). Cells were resuspended in RPMI containing 1 mM CaCl2, 1 mM MgCl2, and 10% FCS at 0.5 × 106 cells/ml. In blocking experiments, α4 integrin–specific antibody Ps/2 (10 μg/ml) or 1 mM EDTA was included. Cover slips (40 mm; Menzel, Omnilabo) were coated with fibronectin (20 μg/ml; Roche) or recombinant mVCAM-1 (0.7 μg/ml; R&D Systems) and were assembled in a 37°C Focht flow chamber mounted on an inverted phase-contrast microscope (Axiovert 35M, Zeiss) connected to a camera (VarioCAM, PCO Computer Optics GmbH). Cells were perfused at different shear stress rates (1 or 2 dyne/cm2) for 60 s, and video recordings of one microscopic field (0.14 mm2) were analyzed continuously at 10 images/s. Analysis of individual cell tracking was performed with in-house developed software (54), which enabled determination of average rolling velocity, number of rolling cells (defined as movement of 10 to 650 μm/s), and number of firmly adhering cells (movement of 0 to 2 μm/s). Perfusions were performed in duplicate in three separate experiments.

VCAM-1 binding and protection from apoptosis

Mice were immunized with NP-KLH as described earlier and were sacrificed at day 14. Wild-type and Cd37−/− splenocytes (0.5 × 106 per 100 μl of RPMI containing 1 mM MgCl2 and 1 mM CaCl2) were incubated with different concentrations (0.1 to 20 μg/ml) of soluble mVCAM-1 (containing all seven Ig-like domains) conjugated to human Fc (R&D Systems) for 30 min at room temperature. Binding of VCAM-1–Fc was detected by incubating the cells with Alexa Fluor 488–conjugated anti-human Fc for 20 min at 4°C, followed by dual staining with Alexa Fluor 647–conjugated antibody against B220 and peridinin-chlorophyll protein complex (PerCp)–conjugated antibody against CD138 for 20 min at 4°C to identify ASCs by flow cytometry. For survival assays, wild-type and Cd37−/− splenocytes were incubated in the presence of VCAM-1–Fc (0.1 to 1 μg/ml) or human IgG for 3 hours at 37°C. ASCs were labeled with fluorescein isothiocyanate–conjugated antibody against B220 and PerCp-conjugated antibody against CD138 as described earlier and were stained with annexin V–Cy5 (BD Biosciences) in RPMI containing 1 mM CaCl2, followed by PI, and were analyzed by flow cytometry.

Detection of pAkt and pBad

Viable splenic lymphocytes from wild-type and Cd37−/− mice (n = 6 mice each) were purified by density gradient separation (Lympholyte, Cedarlane) 14 days after immunization. Lymphocytes (5 × 106) were stimulated ex vivo with VCAM-1 (2 μg/ml) for 10 min at 37°C and then with an α4 integrin–specific antibody (20 μg/ml Ps/2) for 20 min at 4°C, followed by washing and cross-linking with goat anti-rat Fab2 antibody (10 μg/ml) for 10 min at 37°C in RPMI containing 0.5 mM MgCl2 or with pervanadate (200 μM Na3VO4, 2.5 mM H2O2). Cells were fixed with 1% paraformaldehye, permeabilized with methanol, and stained with antibody against pAkt (Ser473; Cell Signaling Technology) or antibody against pBad (Ser136; Cell Signaling Technology) followed by Alexa Fluor 647–conjugated donkey anti-rabbit antibody, biotin-conjugated antibody against CD138 (which was detected with Texas Red–conjugated streptavidin), and Alexa Fluor 488–conjugated anti-mouse IgG1. Cells were analyzed by flow cytometry and confocal microscopy. In addition, stimulated cells were lysed in 150 mM NaCl, 10 mM tris-HCl (pH 7.4), 2 mM MgCl2, 1% Triton, 1 mM phenylmethylsulfonyl fluoride, 1 mM Na3VO4, 5 mM NaF, and protease inhibitors and separated by reducing SDS–polyacrylamide gel electrophoresis. Western blots were incubated with antibodies against pBad, pAkt, total Akt (Cell Signaling Technology), and, as loading controls, tubulin (Novus) or actin (Sigma), followed by IRDye-conjugated secondary antibodies and Odyssey infrared detection (LI-COR). Ratios of the amounts of pAkt to that of total Akt and of pAkt to that of actin were measured with ImageJ software.

Statistical analyses

Statistical differences were determined with the unpaired Student’s t test. Significance was accepted at P < 0.05. FRAP analysis was performed with the two-sample Kolmogorov-Smirnov test with an acceptance level of 95%.

Supplementary Materials

www.sciencesignaling.org/cgi/content/full/5/250/ra82/DC1

Methods

Fig. S1. Production of IgG2b, IgG3, and IgM in Cd37−/− mice.

Fig. S2. Splenic B cell subsets in Cd37−/− mice.

Fig. S3. Decreased population of memory B cells in spleens from Cd37−/− mice.

Fig. S4. Normal proliferation of Cd37−/− B cells.

Fig. S5. Blocking the α4 integrin reduces cell rolling and adhesion.

Fig. S6. Cd37 expression in plasmablasts and plasma cells.

Fig. S7. Impaired Akt phosphorylation in IgG1-secreting Cd37−/− cells as detected by flow cytometry.

Fig. S8. Average area of α4β1 integrin fluorescence at the VCAM-1 contact site.

Movie S1. Rolling and adhesion of wild-type B cells under flow.

Movie S2. Rolling and adhesion of Cd37−/− B cells under flow.

References and Notes

Acknowledgments: We thank R. Woestenenk for the sorting of plasma cells. We greatly appreciate the technical help from the animal house staff of the Central Animal Facility of the Radboud University Nijmegen and the Burnet Institute. Funding: This research was supported by grants from the Dutch Cancer Society (KWF 2007-3917 to A.B.v.S.), the Netherlands Organization for Scientific Research (NWO-Vidi to A.B.v.S., NWO-MEERVOUD to A.C., and NWO-Spinoza to C.G.F.), the European Research Council (ERC Advanced Grant 269019 to C.G.F.), the Human Frontier Science Program (to A.C.), the Fondo de Cooperación Internacional en Ciencia y Tecnología (c0002 2008 ALA 2006 18149 to C.G.F.), the Association of International Cancer Research (to M.D.W.), the Anti-Cancer Council of Victoria (to M.D.W.), and the Australian National Health and Medical Research Council (to D.M.T. and M.D.W.). Author contributions: A.B.v.S. designed the study, did most of the experiments, and wrote the manuscript; S.d.K. performed FRAP, confocal laser scanning microscopy, and α4β1 clustering studies; A.v.d.S. and M.Z. performed immunization and phosphorylation experiments; K.H.G., M.S., and A.L. did bone marrow reconstitution experiments; P.C.L. and J.B.B. did flow chamber studies; I.R.-B. and A.C. performed electron microscopy analysis; and F.M., D.M.T., M.D.W., and C.G.F. designed the study, supervised the work, and wrote the manuscript. Competing interests: The authors declare that they have no competing interests. Data and materials availability: Mus musculus Cd37, GeneID 12493, MGI: 88330; α4 integrin (Itga4), GeneID 16401, MGI: 96603.
View Abstract

Navigate This Article