In Vivo Roles of Integrins During Leukocyte Development and Traffic: Insights from the Analysis of Mice Chimeric for α₅, α_v, and α₄ Integrins¹

Interactions of hemopoietic progenitors with the stromal environment are crucial for normal development of the different blood lineages (1). Integrin receptors have been suggested to be important in regulating these processes. α₅, α_V, and α₄ integrins recognize different sites on fibronectin, and they also bind other specific ligands, including vitronectin for α_V integrins and VCAM-1 and mucosal addressin cell adhesion molecule-1 (MAdCAM-1)⁶ for α₄ integrins (2, 3). Expression studies have shown the presence of integrins on leukocyte progenitors and their differential regulation among lineages (1, 4). In particular, α₄ and α₅ integrins are expressed early during hemopoietic development, but they are down-regulated during neutrophil development. They are, however, expressed on mature lymphocytes, monocytes, and granulocyte subsets. In contrast, α_V integrins are expressed on specific T lymphocyte subsets (γδ T cells) and on different leukocyte precursors (data not shown), and α_V is up-regulated during myeloid differentiation (5). The functions of integrin receptors during leukocyte development in vivo, however, have been unclear. In vitro studies have previously suggested that α₄ integrins participate in B lymphocyte and myeloid development (6, 7). The expression of α₄ and α₅ integrins on thymocytes is finely regulated, and this has also suggested a role for these receptors during T lymphocyte differentiation (8). However, little was known about the potential roles of α_V integrins during these processes.

Lymphocytes exert their surveillance and activation functions thanks to their continuous recirculation through lymphoid organs. The molecular mechanisms involved in regulating these homing pathways are becoming understood (9). The gut-associated lymphoid tissue in the intestinal mucosa represents an integrated system of secondary lymphoid organs, Peyer’s patches and mesenteric lymph nodes, and mucosal effector sites, lamina propria (LP) and the intraepithelial lymphocyte (IEL) compartment located above the villus basement membrane (10). α₄β₇ integrin is thought to confer gut tropism to lymphocytes because it is expressed at high levels on a small subset of circulating memory T cells (11, 12) that preferentially localize to Peyer’s patches and to the intestinal LP (9). This selective localization is mediated by its interaction with MAdCAM-1 (13), a molecule selectively expressed on intestinal endothelial cells (14). The integrin α_Eβ₇, expressed on a subset of the α₄β₇^high circulating T cells, on most IEL and on a substantial fraction of LP T lymphocytes, mediates adhesion to E-cadherin expressed on epithelial cells (15). Analyses of knockout and chimeric mice, deficient in the expression of β₇ integrin or its associated α-chains (α_E and α₄) have recently dissected some of the roles of these adhesion receptors in constituting the lymphoid gut compartment (15, 16, 17 ; for review, see Ref. 3). In mice deficient for β₇ integrins, Peyer’s patch formation was impaired, and reduced numbers of IEL and LP T cells were observed. α₄-integrin chimeric mice confirmed the critical role of this subunit for homing of lymphocytes to Peyer’s patches. In contrast, α_E-deficient mice had normal Peyer’s patch size but reduced numbers of IEL and LP T cells, suggesting its involvement in recruiting to these sites. The roles of other integrin receptors, such as α₅ and α_V integrins, have not been reported.

During the inflammatory response, lymphocytes initiate contact with the activated endothelium, first by tethering and then by firm adhesion and subsequent transmigration. Different adhesion receptors are involved in regulating these steps, including selectins and integrins, particularly α₄ and β₇ integrins, and their ligands (3, 18). The role of α₄ integrins in diverse inflammatory pathologies has largely been investigated by blocking α₄-mediated interactions with Abs or peptides, and this approach is currently being used in clinical trials (19). However, the direct in vivo roles that α₄ integrins and other integrins, such as α₅ and α_V, might play in leukocyte transmigration during inflammatory processes have not yet been analyzed.

Finally, signaling by integrins and other receptors is likely to play a role in lymphocyte activation (for review, see Refs. 20 and 21). Thus, integrin function is regulated by lymphocyte activation (4), and it has also been shown that several integrin ligands, including fibronectin and VCAM-1, can act as coactivation signals for lymphocytes (for review, see Ref. 21). Moreover, integrin receptors in lymphocytes can trigger multiple different signaling pathways. All these reports suggest that integrins might participate in lymphocyte activation. However, their roles in vivo are poorly defined.

To investigate the roles of α₅, α_V, and α₄ integrins in the development, homing, migration, and activation of lymphocytes in vivo, analysis of mice lacking these receptors has been performed. To circumvent the early lethality of α₅-, α_V-, and α₄-null embryos, chimeric mice were generated by injecting α₅-, α_V-, or α₄-null ES cells into blastocysts from RAG-2^−/− or C57BL6 mice.

Chimeric mice were generated by injection of α₅-, α_V-, or α₄-null or wild-type ES cells into blastocysts from C57BL6 or RAG-2 deficient mice as previously described (17). (For α₅- and α₄-null ES cells see Refs. 22 and 17 , respectively.) α_V-null ES cells were obtained from heterozygous ES cell clones (23) by selection in high concentration of G418 (3–4 mg/ml) as previously described (24). Several independent α_V-null ES cell clones were further analyzed and used for chimeric studies and germline transmission experiments (25). Mice were kept in the Massachusetts Institute of Technology animal facility under clean conditions.

Anti-mouse mAbs from PharMingen were used for staining: PE-conjugated anti-CD3-ε (145-2C11), PE-conjugated anti-CD4 (L3T4, RM4-5), Cy-Chrome-conjugated anti-CD8 (Ly-2, 53-6.7), PE-conjugated anti-CD45R/B220 (RA3-6B2), FITC-conjugated anti-IgM (R6-60.2), PE- conjugated anti-CD11b (Mac-1 α-chain, M1/70), PE-conjugated anti-Ly-6G (Gr-1, RB6-8C5), FITC-conjugated anti-CD49d (integrin α₄ chain, R1-2), FITC-conjugated anti-CD45.2 (Ly-5.2, 104), FITC-conjugated anti-Ly-9.1 (30C7), PE-conjugated anti-CD25/IL2-Rα (3C7), and PE-conjugated anti-CD69 (H1.2F3). Single-cell suspensions obtained from thymus, bone marrow, spleen, peritoneal lavage, and blood by routine dissection techniques were incubated with purified anti-CD32/CD16 to block Fc receptors and with an appropriate dilution of the different Abs at 4°C or room temperature (blood samples). Samples were washed twice with PBS and resuspended in PBS. Dead cells were excluded by propidium iodide staining. Samples and data were analyzed in a FACScan using CellQuest software (Becton Dickinson, Mountain View, CA).

Tissue specimens were fixed overnight in 4% formaldehyde and embedded in paraffin. Blocks were cut, and slides were processed and stained with hematoxylin-eosin by routine techniques. Slides were examined and photographed (Ektachrome 160T film; Eastman Kodak, Rochester, NY) with an Axiophot microscope (Carl Zeiss, New York, NY).

For immunohistochemistry, 6-μm cryostat-cut jejunal tissue sections were stained using the avidin-biotin-immunoperoxidase method according to the manufacturer’s instructions (Vector Laboratories, Burlingame, CA) using the anti-CD3 mAb 145-2C11 (Armenian hamster anti-mouse) or a nonbinding control Armenian hamster IgG, biotinylated goat anti-Armenian hamster (Jackson ImmunoResearch Laboratories, West Grove, PA), and then the avidin-biotin-immunoperoxidase elite reagent with 3-amino-9-ethylcarbazole as chromogen. Sections were counterstained with hematoxylin and anti-CD3-stained cells were quantitated as IEL per 1000 epithelial cells or as LP T cells per 0.5 mm villus length. The observer was blinded with respect to the tissue source during cell number quantitation.

Injection of 3% thioglycolate broth into the peritoneum of 8- to 12-wk-old α₄- or control-C57BL6 chimeric mice was performed. Eye bleeds and peritoneal lavages were obtained at 24 or 48 h after injection. Lavage was recovered by i.p. injection of cold PBS containing 5 mM EDTA and 1% BSA and by gently massaging the abdomen of the animal. Samples were processed for flow cytometry as described.

Spleens from different chimeric mice were dissected under sterile conditions. Mononuclear cells were obtained by density gradient centrifugation on Lympholite (Cedarlane Laboratories, Hornby, Ontario, Canada). Cells were resuspended in complete medium (RPMI plus 10% FBS). Cells (5 × 10⁵/well) were dispensed in 96-well U-bottom plates, and the various stimuli were added (anti-CD3 from PharMingen (San Diego, CA); PMA, ionomycin, LPS, and Con A from Sigma (St. Louis, MO)). For mixed lymphocyte reaction, allogeneic irradiated splenocytes from BALB/c mice were used as targets at different ratios. Proliferation was quantitated by adding Alamar Blue for 24 h before reading the plates at 570 nm (26).

Data were analyzed and compared for statistically significant differences using Student’s t test.

The role of α₅ and α_V integrins during lymphoid and myeloid development was investigated by flow cytometric analysis of chimeric mice. CD3-positive T lymphocytes are present in blood from α₅- and α_V-null/RAG-2 chimeric mice in percentages similar to those in controls (Fig. 1,A). Moreover, no significant differences were observed in the subsets of T cells (CD4⁻CD8⁻, CD4⁺CD8⁺, CD4⁺CD8⁻, and CD4⁻CD8⁺) present in the thymus of α₅ and α_V chimeric mice compared with those in controls. Percentages of γδ T cells in α_V chimeric mice were also similar to control values (not shown). In contrast, although there are T lymphocytes in the blood of α₄-null/RAG-2 chimeric mice (Fig. 1,A), a severe defect occurs in T cell differentiation within the thymus in the absence of α₄ integrins as previously described for chimeras on other backgrounds (17). In fact, histological analysis of the thymus in α₅ and α_V chimeric mice shows cortical and medullary areas similar to those in controls, whereas in α₄ chimeric mice the thymus appears atrophic a few weeks after birth (Fig. 1 C).

B lymphocyte development was analyzed by staining with the combination of specific markers, B220/IgM or B220/CD5, for B-1 cells. No major differences were found in B lymphocyte populations present in spleen, bone marrow, and peritoneal cavity of α₅- or α_V-null/RAG-2 chimeric mice compared with controls (Fig. 1,B). However, very few B cells were found in the periphery, and almost none were found in the bone marrow of α₄-null/RAG-2 chimeras (Fig. 1,B) in accord with previous findings in other backgrounds (17, 27). Histology of the spleen revealed a normal structure, with lymphoid follicles and germinal centers, in α₅ and α_V chimeras (Fig. 1,C). However, very few scattered lymphoid follicles, with poor germinal centers and an increased ratio of red (erythromyeloid) to white (lymphoid) pulp, were observed in α₄ chimeric mice (Fig. 1 C).

Myeloid development was also analyzed by flow cytometric staining with specific markers for monocyte (Mac-1) and granulocyte (Gr-1) populations. Although the percent chimerism was low because of the presence of host-derived leukocytes, the data showed no significant differences in numbers of Mac-1/Ly5.2⁺ or Gr-1/Ly5.2⁺ cells in blood or bone marrow from α₅- and α_V-null/RAG-2 chimeric mice compared with control values (Fig. 2). In contrast, the analysis of α₄-null/RAG-2 chimeric mice showed that monocytes, granulocytes, and their progenitors were present in blood and bone marrow, respectively, but the numbers were significantly lower than control values (Fig. 2) (27). Similar results for all lineages were obtained when chimeras on a C57BL background were analyzed (not shown).

It is interesting to note that, although α₅, α_V, and α₄ integrins all recognize fibronectin, their requirements during leukocyte development seem to be different, suggesting incomplete redundancy in vivo. Thus, the major conclusion from these data is that α₄ integrins are essential for leukocyte development, but α₅ and α_V integrins are not. This might be due to the nature of the site recognized on fibronectin by α₄ integrins that is arginine-glycine-aspartic acid independent and alternatively spliced, in contrast to that recognized by α₅ and α_V integrins. The specific recognition by α₄ integrins of other ligands, such as VCAM-1 and MAdCAM-1, might also play a role. α₄-null/RAG-2 chimeric mice have a severe defect in lymphocyte and myeloid development, as has previously been shown in other backgrounds (17, 27). Regarding the ligands involved in these defects, because VCAM-1-null mice do not show any defect in leukocyte development (28), it seems more likely that an as yet unknown ligand or the CS-1/V region of fibronectin plays the essential role. In this regard, fibronectin has been shown to affect the proliferation state of hemopoietic progenitors through its recognition by α₄ integrins (29, 30). In fact, further analysis has shown that the defects observed in the absence of α₄ integrins are multilineage and probably due to defects in the proliferation of hemopoietic progenitors (27).

In contrast, it is still not possible to rule out the possibility that α₅ or α_V integrins participate in leukocyte development. In this regard, α₅ integrin has previously been suggested to play a role in transmigration of lymphoid progenitors (31). Because α₅ and α_V recognize the same site on fibronectin, although probably with different avidity depending on the cell type, there might be substitution between these two receptors in their putative roles during leukocyte development, as has been shown for other in vivo and in vitro functions (32). Analysis of inducible tissue-specific knockout mice will provide a more definitive answer. Moreover, the participation of wild-type cells in the chimeric system by, for instance, secreting soluble factors such as cytokines cannot be completely ruled out.

Because lymphocytes can develop without α₅ or α_V integrins, and T lymphocytes and a few B lymphocytes are present in the periphery of α₄-null chimeric mice, we next wanted to investigate the roles of these adhesion receptors in lymphocyte homing. Migration to lymph nodes and mucosal sites was first analyzed by flow cytometric staining. Similar percentages of T and B lymphocytes were found in mesenteric nodes and Peyer’s patches of α₅- and α_V-null/RAG-2 chimeric mice compared with controls (Fig. 3,A). α₄-deficient T and B lymphocytes were present in mesenteric and inguinal nodes in similar percentages as in blood, suggesting no defect in migration to peripheral lymph nodes (Fig. 3,A) (17). However, no Peyer’s patches were observed macroscopically in α₄-null/RAG-2 chimeric mice, indicating a severe defect in trafficking of lymphocytes to mucosal lymphoid sites (Fig. 3,A). Histological analysis of mesenteric nodes revealed a normal structure, with lymphoid follicles and germinal centers, in α₅ and α_V chimeric mice (Fig. 3,B). Although lymphoid follicles were present in α₄ chimeric mice, they were smaller and lacked germinal centers, in concordance with the defect observed in B cell development (Fig. 3,B). Histological sections of gut showed the presence of lymphoid cells constituting Peyer’s patches in α₅, α_V, and control chimeric mice (Fig. 3,B). In α₄ chimeric mice, however, only empty patches were found, with very few or no lymphocytes within (Fig. 3 B), confirming the severe defect in migration to these sites in the absence of α₄ integrins.

Lymphocyte traffic to intestinal compartments was also investigated; immunohistochemical analysis showed T lymphocytes in both intraepithelial and LP compartments of the gut in α₅-, α_V-, and α₄-null/RAG-2 chimeric mice (Fig. 3,C). Almost no CD3⁺ lymphocytes were present in RAG-2-deficient mice. No significant differences were observed in the numbers of lymphocytes in either compartment compared with those in wild-type controls (Fig. 3 C). FACS analysis showed a severe reduction in expression of β₇ on α₄- null lymphocytes (Ly 9.1⁺; data not shown) reflecting the loss of α₄β₇. However, a minority population of α₄⁻β₇⁺ Ly 9.1⁺ lymphocytes was observed in the chimeras. These presumably represent an α₄⁻β₇⁺ population that accounts for the efficient recruitment of T cells to the intraepithelial and lamina propria compartments (15).

The roles of α₅ and α_V integrins in lymphocyte trafficking in vivo have not previously been reported. Herein, we show that lymphocytes can migrate properly to different locations in the absence of α₅ or α_V integrins. This means that α₅ and α_V integrins are not essential for lymphocyte migration to lymphoid organs or that they are able to substitute for each other in this function, as discussed for leukocyte development. Interestingly, α₄ is essential for migration of lymphocytes to Peyer’s patches. The α₄ integrin chain can associate with β₁ or β₇ subunits. The analysis of β₇ knockout mice has shown a critical role for β₇ receptors in migration of lymphocytes to Peyer’s patches and an important role in trafficking to the intestinal intraepithelial compartment and LP (16). The defect observed in migration to Peyer’s patches in the absence of α₄β₇ fits with the presence of its specific ligand, MAdCAM-1, at this site. Previous analysis of α₄-deficient chimeric mice had shown an essential role for these receptors in migration to Peyer’s patches (17); however, conclusive data about their roles in lymphocyte migration to intestinal epithelium and LP were missing. Our current results demonstrate that α₄ integrins are not critical for migration to these gut compartments, suggesting that other β₇ partners might be responsible for the defects. In this regard, α_E-null mice show decreased numbers of IEL and LPL (15). It is possible to conclude from these studies that α₄β₇ integrin is essential for migration of lymphocytes to Peyer’s patches and that α_Eβ₇ is critical for localization of lymphocytes within the intestinal epithelium and LP. Finally, it has been shown that β₂ integrins interacting with ICAM-1 can also participate in the establishment of the intestinal lymphocyte compartment (10).

Because the α₄ integrin ligand, VCAM-1, is up-regulated on endothelium during inflammatory responses, migration of α₄-deficient T lymphocytes was also investigated in an inflammatory model, thioglycolate-induced peritonitis. As shown in Fig. 4, the relative percentages of CD4⁺/Ly9.1⁺ T lymphocytes present in the peritoneal lavage and blood of control/C57BL chimeras at 24 and 48 h are similar, showing no defects in migration (blood at 24 h, 23%; peritoneal lavage at 24 h, 28%; blood at 48 h, 30%; peritoneal lavage at 48 h, 34%; n = 3). However, the relative percentages of T lymphocytes in the peritoneal lavage of α₄ chimeric mice were significantly lower compared with their relative percentages in blood (Fig. 4; blood at 24 h, 20.5%; peritoneal lavage at 24 h, 8.3%; blood at 48 h, 21%; peritoneal lavage at 48 h, 7.8%; n = 3), indicating an impairment of T lymphocyte migration to inflammatory foci in the absence of α₄ integrins. This defect is most likely due to the lack of interaction of α₄-null lymphocytes with VCAM-1 at the inflammatory foci. However, because T lymphocytes in α₄-chimeric mice are developed in the thymus only at very early stages of development, it is conceivable that they could also lack some switch necessary to home properly to these sites, as reported for other homing pathways (33).

The role that α₄ integrins might have during inflammatory processes had been investigated previously using Abs against α₄ integrins or specific inhibitory peptides (for review, see Ref. 19). These studies have shown a role for α₄ integrins during allergic asthma, delayed contact hypersensitivity, experimental allergic encephalomyelitis, rheumatoid arthritis, and chronic inflammatory bowel disease. In this report we show that α₄ integrins are required in vivo for the efficient migration of lymphocytes to an inflammatory focus. These data reinforce the therapeutic approaches that are under clinical trial using α₄ inhibitors for the treatment of different chronic inflammatory diseases. However, it is also important to emphasize that the complete blockade of α₄ integrin function might have deleterious effects on the development of the different blood lineages and on the homing of lymphocytes to Peyer’s patches in humans as has been demonstrated for mice (Refs. 17 and 27 and this report).

Lymphocyte activation was also studied in the absence of these integrin receptors. First, it is interesting to note that α₅-, α_V-, and α₄-null/RAG-2 chimeric mice do not show any obvious increases in infectious or inflammatory diseases compared with control chimeras during their life span (up to 2 yr) in contrast to RAG-2-deficient mice, which get sick after 6 mo in the same environment, suggesting an appropriate function of the immune system. To assess integrin function during lymphocyte activation, in vitro approaches were undertaken using integrin-deficient lymphocytes. Splenocytes from α₅-, α_V-, or α₄-null/RAG-2 chimeric mice were used for in vitro proliferation assays in the presence of serum containing fibronectin. As shown in Fig. 5,A, no major differences in the proliferation of T lymphocytes induced by Con A or anti-CD3 were observed in the absence of α₅-, α_V-, or α₄-integrins compared with that of cells from control chimeric mice. Similar results were obtained when LPS was used to stimulate proliferation of B lymphocytes (Fig. 5,A). Intercellular adhesion-mediated activation was analyzed by mixed lymphocyte reaction in the presence of serum. Splenocytes deficient for α₅-, α_V-, or α₄-integrins were cocultured in the presence of different ratios of irradiated allogeneic splenocytes. As shown in Fig. 5,B, no major differences were observed in lymphocyte proliferation in the absence of integrin receptors compared with controls. The role of integrins in the activation of lymphocytes was also investigated by checking their activation state by flow cytometry. The expression of the lymphoid activation markers CD25 (IL-2αR) and CD69 was up-regulated on splenocytes deficient for α₅-, α_V-, or α₄-integrin receptors similarly to that in controls on day 3 after stimulation with PMA and ionomycin (Fig. 5 C). Relative secretion of IgM by α₅-, α_V-, or α₄-deficient splenocytes upon LPS stimulation was also similar to that in controls as assessed by ELISA (data not shown). Together, these data indicate that lymphocytes can proliferate and be activated properly in the absence of α₅, α_V, or α₄ integrins.

Activation of lymphocytes that subsequently leads to their proliferation is a complex process in which several membrane receptors are involved. First, a specific signal through the T or B cell-specific receptors is required, but additional accessory molecules play important roles. In this regard, in vivo studies using knockout mice for different accessory molecules have shown that the LFA-1/ICAM-1 accessory pathway is essential for septic shock development and delayed cutaneous hypersensitivity. Other pathways, such as B7-1 and B7-2/CTL4 interaction, are also important in vivo for normal T cell-dependent Ag responses (34). In contrast, our data suggest that α₅, α_V, or α₄ integrins are not essential for in vitro activation of lymphocytes. However, synergism of these adhesion receptors in coactivating lymphocytes under certain conditions in vivo could still be possible. Moreover, because accessory pathways seem to be multiple, it is not possible to rule out substitution by other accessory molecules. Given the lethality of all three integrin mutations, further analyses in vivo are challenging. Inducible tissue-specific knockout mice will give further clues about the functions of these receptors during the immune response in vivo.

We thank Valerie Evans for help with generation of chimeras. We are also grateful to Colleen Leslie and Susan Dutton for their help in editing the manuscript.