Introduction
Leukocyte migration requires a chain of molecular interactions between the vascular endothelium and white blood cells that initially retards leukocyte flow and ultimately leads to leukocyte transmigration through the endothelium1, 2. The least well-defined part of this process is the passage of leukocytes through the endothelium and the mechanisms involved in this transmigration. However, it has been established that leukocytes use both paracellular and transcellular routes for transendothelial migration3, 4. When the transmigration of PBMCs and polymorphonuclear leukocytes (PMNs) through TNF-
-activated human umbilical vein endothelial cells (HUVECs) was studied using in vitro capillary flow assays, PMNs (consistent with previous reports5, 6, 7) paused close to their site of transmigration at junctional areas (see Supplementary Information, Fig. S1a, b and Movie 1), and then rapidly (<5 min) transmigrated using the paracellular route7. In contrast, most of the PBMCs paused in the vicinity of the perinuclear region rather than at the cell borders (see Supplementary Information, Fig. S1c–f and Movie 2). After a longer initial phase of adherence, transmigration occurred at the site of adhesion by transcellular migration straight through the endothelium (see Supplementary Information, Fig. S1c and Movie 2). Thus, non-activated PBMCs and PMNs have a preference for different routes for transmigration.
The first contact between a PBMC and its target endothelial cell was established by a pseudopodium-like process emerging from the cell body of the PBMC (10–20 min; see Supplementary Information, Fig. S1e) and involved a remarkable reorganization and polarization of the vimentin intermediate filament network into the uropod of endothelial cell-attached PBMCs (Fig. 1a). This reorganization is associated with adhesive events, as PBMCs seeded on gelatin-coated coverslips without any endothelial cells also displayed similar reorganization (Fig. 1b). Given the prominent effect on vimentin, whether or not the vimentin intermediate filaments have an active role in the transmigration of the PBMCs, and whether the absence of intermediate filaments had an effect on endothelial cell–lymphocyte interactions, was examined. The adhesion of mouse lymphocytes to endothelial cells isolated from both wild-type and vim-/- mice8 was compared. Strikingly, the absence of vimentin in migrating cells markedly reduced adhesion between lymphocytes and endothelial cells (Fig. 1c). The reduction in adhesion was detected not only in vim-/- PBMCs, but also when wild-type PBMCs were seeded on vim-/- endothelial cells (Fig. 1c). To assess whether the observations in the in vitro model are relevant to endothelial cell–lymphocyte interactions in vivo, homing of wild-type and vim-/- lymphocytes in wild-type and vim-/- mice was studied. When compared with wild-type lymphocytes, there was a marked reduction in the homing of vim-/- lymphocytes in both types of mice (Fig. 1d). It is likely that the in vivo effect is partly masked by the analysis of a mixed cell population and because only a fraction of the injected cells employ the transcellular pathway. In contrast, both wild-type and vim-/- lymphocytes homed more efficiently to the mesenteric lymph nodes and spleen of the knockout mice than the wild-type mice, suggesting that the integrity of the endothelium is compromised in knockout mice (see Supplementary Information, Fig. S2a, b). This was further verified in a mouse model of acute peritonitis, in which a slightly increased proportion of incoming neutrophils was observed when compared with resident macrophages in vim-/- mice (see Supplementary Information, Fig. S2c). It seems that the leakiness of the endothelium exceeds the effect of the reduced adhesion observed in the static assays. This result implies that in vim-/- mice lymphocytes primarily (if not exclusively) choose the paracellular pathway.
Figure 1: Vimentin intermediate filaments are involved in PBMC transcellular migration and in vivo homing.
(a) Reorganization of vimentin intermediate filaments (green) of two migrating PBMCs towards the uropods (white arrows). Anti-CD44 (red), a membrane marker for PBMCs, outlines the borders of the PBMCs. The focal plane is at the upper level of the endothelial cell, so only parts of the endothelial-cell vimentin intermediate filaments can be seen around the migrating cell. (b) A similar polarization and reorganization of vimentin intermediate filaments (green) also occurred when the cells were incubated for 30 min on gelatin-coated coverslips. z-axis side views are shown from two different planes (coloured lines on the top and on the right-hand side of the large image show the maximum projection). (c) Quantification of PBMC adhesion to endothelial cells. The number of adherent PBMCs in wild-type (WT) endothelium or vim-/- endothelium was calculated as an average number of adherent lymphocytes in randomly chosen microscopic fields. Values are mean 
s.e.m. of three different experiments (*** indicates P < 0.001 as analysed by ANOVA). (d) Lymphocytes from vim-/- mice migrate less efficiently than lymphocytes from wild-type mice to both the mesenteric lymph nodes (MLNs) and the spleen of wild-type and knockout mice.
To further analyse leukocyte–endothelial cell interactions and possible vascular leakage, intravital video microscopy (a technique that allows direct observation of individual cells inside live animals) was used to examine inflamed cremaster muscle. This muscle covers the testis and its function is to raise and lower the scrotum to regulate the temperature of the testis. Models utilizing the cremaster muscle are widely used in intravital microscopy studies as the muscle is transparent and, therefore, leukocyte trafficking within its vasculature is relatively easy to analyse. In this model, >90% of interacting leukocytes are PMNs. Parameters of the analysed vessels are presented in Fig. 2a. There was no difference in the rolling flux (defined as the number of leukocytes rolling in close contact to the vessel wall across a fixed line over a period of 1 min), but velocity of the rolling cells (the time cells needed to roll through a 100-
m segment of the vessel) was approximately three times higher in vim-/- mice than in wild-type mice (see Methods for details). A leukocyte was scored as 'firmly adherent' if it was stably bound to the endothelial cells for at least 30 s, and as 'transmigrated' if it was in the perivenular tissue within 50 m of the vessel segment under observation. The number of adherent cells was significantly lower in vim-/- mice; however, the number of transmigrated cells was significantly higher (Fig. 2b). The numbers translate to a 2.6-fold higher transmigration efficiency of adherent cells in vim-/- mice. To determine whether this is a result of compromised integrity of the endothelial cell barrier in vim-/- mice, fluorescein-isothiocyanate (FITC)-conjugated–dextran was injected into wild-type and vim-/- mice and its in vivo distribution was examined. The inflammation model used was relatively mild, hence there was no marked leakiness of FITC–dextran outside the vasculature in wild-type mice. In contrast, in vim-/- mice FITC–dextran accumulated in protrusions within the endothelial-cell layer indicating that the endothelial-cell barrier integrity is severely compromised (Fig. 2c, d).
Figure 2: Leukocyte–endothelial cell interactions, and endothelial integrity are compromised in vim-/- mice.
![Figure 2 : Leukocyte|[ndash]|endothelial cell interactions, and endothelial integrity are compromised in vim|[minus]|/|[minus]| mice.](/ncb/journal/v8/n2/thumbs/ncb1355-f2.gif)
(a) Characteristics of the analysed vessels. (b) The velocity of rolling cells and the number of rolling cells, firmly bound cells and transmigrating cells was determined using intravital microscopy in wild-type and vim-/- mice. The vasculature was inflamed with intrascrotal TNF- injections. The results are mean s.e.m. of 90 and 60 vessels. (c) Leakiness was measured using intravenous injections of FITC–dextran. The number of FITC–dextran-filled protrusions within the endothelial cell layer were counted and expressed as mean s.d. of 181–188 vessels. (d) Examples of the FITC–dextran filled protrusions in vim-/- mice (arrows). Protrusions were not observed in wild-type animals. Scale bar represents 20 m.
As both in vitro and in vivo experiments indicated that vimentin functions in both leukocytes and endothelial cells during their interaction, the distribution of intermediate filaments in endothelial cells through which PBMCs were migrating was examined. HUVECs were grown on glass coverslips or transwell inserts and the lymphocytes were allowed to interact with the HUVECs. Cells were fixed and labelled with either anti-vimentin or an antibody recognizing keratin 8–18 at the indicated times and analysed using confocal laser microscopy. CD44, a membrane marker for PBMCs, was used to visualize the cell borders of migrating cells. Initially, most PBMCs were still round and topologically distributed along the endothelial cells, and no realignment or redistribution of endothelial cell intermediate filaments was observed (data not shown). After 20 min of co-culture, confocal images showed that the PBMCs were clearly inside the endothelial cells, as they were obviously entangled by the vimentin network (Fig. 3a). Surprisingly, the vimentin intermediate-filament polymers seemed to be actively reorganized towards the entrance site of the lymphocyte — the filamentous network was clustered towards the site of contact of the endothelial cell and the PBMC, and formed a cup-shape structure around the PBMC (Fig. 3a).
Figure 3: Reorganization of vimentin and keratin intermediate filaments during PBMC transmigration.
(a) A cup-like intermediate filament structure formed around the migrating PBMC (white arrow). (b) Keratin intermediate filaments (green) were immunolabelled for keratin 8–18. A CD44-labelled PBMC (red; arrowhead) was captured in the midst of the intermediate filaments. The image shows the maximum projection of a stacked series of confocal z-scans through the full thickness of an endothelial cell (for a three-dimensional reconstruction see Supplementary Information, Movie 3). (c) The clustering of GFP-tagged vimentin during PBMC adhesion and transmigration through HUVECs. GFP-transfected HUVECs were placed in a plate-flow chamber device and PBMCs were allowed to interact and migrate across the TNF--activated HUVEC monolayer at a shear stress of 1.0 dyn cm-2. At 15 min, endothelial GFP–vimentin began to redistribute to the site of adhesion. By 30 min, endothelial cell GFP–vimentin filaments were concentrated at the site of PBMC–endothelial contact (see also Supplementary Information, Movie 4).
The endothelial keratin filament bundles were organized around the PBMCs (Fig. 3b; see Supplementary Information, Movie 3). The organization of the other major cytoskeletal network systems, microtubules and microfilaments, was not altered during the redistribution of the intermediate filaments (data not shown). In agreement with the assumed different routes of migration, no such clustering of endothelial-cell intermediate filaments or PMN vimentin intermediate filaments was observed during PMN adhesion and transmigration (data not shown).
As it was not possible to determine the exact sequence of events and kinetics during transmigration in an end-point-type assay, the altered distribution of vimentin intermediate filaments was analysed using live-cell, time-lapse confocal microscopy. This method allows direct visualization of endothelial-cell intermediate filament movement during leukocyte binding and transmigration under physiological fluid-stress conditions. HUVECs were transiently transfected with green fluorescent protein (GFP)-tagged vimentin and placed in a plate flow chamber device. PBMCs were allowed to interact and transmigrate across activated HUVECs at a shear stress of 1.0 dyn cm-2. Following a 10-min stabilization period, images were taken every 45 s. The attachment of PBMCs to HUVECs induced a marked redistribution of the vimentin intermediate filaments: at 15 min, the endothelial GFP–vimentin began to redistribute to the site of contact and at 30 min, the endothelial GFP–vimentin was concentrated at the site of contact between the PBMCs and endothelial cells (Fig. 3c; see Supplementary Information, Movie 4).
Together, these results clearly demonstrate that the intermediate filament networks of both lymphocytes and endothelial cells contribute to the migration process. As a recent study suggested that ICAM-1 and VCAM-1 were important for the transmigration process3, whether or not the distribution of adhesion molecules was affected by the absence of vimentin was analysed. Both the distribution, and particularly the expression level of ICAM-1, were markedly affected by the absence of vimentin in the endothelial cells — vim-/- cells displayed very low cell-surface expression of this molecule (Fig. 4a, b). Similar results were obtained for VCAM-1 (data not shown). Furthermore, the surface expression of a primary lymphocytic adhesion molecule, integrin-
1, was very low in vim-/- lymphocytes (Fig. 4c, d) and the normally polarized distribution of this surface molecule was completely lost. To quantify the effects of vimentin deficiency on these critical surface molecules, protein levels were analysed by western blotting of extracts from endothelial cells. PBMCs were analysed by flow cytometry. The results show that the protein levels of both ICAM-1 and VCAM-1 were clearly reduced in vim-/- endothelial cells (Fig. 4e). Also, the surface-protein expression levels of integrin-1 showed a remarkable reduction on both mesenteric and splenic lymphocytes from vim-/- mice (Fig. 4f).
Figure 4: Organization and expression of critical surface adhesion molecules is altered in vim-/- endothelial cells and lymphocytes.
![Figure 4 : Organization and expression of critical surface adhesion molecules is altered in vim|[minus]|/|[minus]| endothelial cells and lymphocytes.](/ncb/journal/v8/n2/thumbs/ncb1355-f4.gif)
(a, b) Endothelial cells from wild-type (a) and vim-/- (b) mouse cells were grown on glass coverslips, immunostained with anti-ICAM-1 and analysed by confocal microscopy. High levels of ICAM-1 immunoreactivity (green) were observed in wild-type cells; weak levels were found in vim-/- cells. (c, d) Integrin-1 staining of wild-type (c) and vim-/- (d) lymphocytes. In addition to changes in the expression level, the reorganization of integrin- 1 to the uropod is greatly diminished in vim-/- lymphocytes. The cell nucleus is stained with DAPI (blue; a–d) and is characteristically deformed in migrating lymphocytes. Scale bar represents 20 m in a and b and 10 m in c and d. (e) Western blot analysis of the total endothelial cell extracts from mouse skin were performed using antibodies against ICAM-1 and VCAM-1. Endothelial cells were isolated from mouse skin and cultured for 2–3 passages. (f) Anti-integrin-1 staining of lymphocytes isolated from wild-type and vim-/- mice. FACS histograms are shown from MLN and spleen lymphocyte labellings. FL1-H, fluorescence intensity on a logarithmic scale.
Our results indicate that intermediate filaments are actively remodelled during adhesion and transcellular migration. They also participate in the formation of an anchoring structure for the involved adhesion molecules, thereby allowing transmigration to occur. Several physiological roles for intermediate filaments have been proposed, including maintenance of cell shape, resistance to mechanical stress, conveyance of cytoskeletal crosstalk and organization of signalling molecules9. Although the intermediate-filament networks were initially regarded as static, they have now been established as highly dynamic structures that undergo constant remodelling and turnover9, 10. We observed active roles for intermediate filaments in both PBMCs and endothelial cells. The live-cell analysis of the endothelium underscored the highly dynamic nature of intermediate-filament organization, as it showed that the endothelial-cell intermediate-filament network is actively organized around the migrating cell.
Although there is very little data on the properties of intermediate filaments during leukocyte transmigration, there are a number of studies on the molecular mechanisms underlying the effects we observed. Our results suggest that the presence or absence of intermediate filaments markedly affects the organization and expression of surface molecules that are critical for adhesion. Therefore, the effects we observed in vitro and in vivo may be due to disturbed functions of critical adhesion molecules. Consistent with a critical role of intermediate filaments in organizing surface molecules, it has been shown that intermediate filaments affect the localization and activity of surface molecules: for example, the activity of the sodium–glucose cotransporter, SGLT1, and its localization to membrane rafts are disturbed in the absence of vimentin11, 12; keratin intermediate filaments have also been shown to participate in protein targeting in polarized epithelia13, 14; furthermore, a recent study showed that vimentin is crucial for PKC
-dependent motility and that PKC-mediated phosphorylation of vimentin is a key process in the trafficking of integrin through the cell15. This function of vimentin may help to explain its role in PBMCs during initial adhesion, the formation of the pseudopod and the concurrent transmigration.
The leukocyte intermediate filaments were focused into uropods during transmigration, suggesting an active function for the intermediate-filament network in controlling leukocyte cyto-architecture during transendothelial migration. Previously it has been reported that the rigidity of circulating lymphocytes is conferred by vimentin intermediate filaments and that their regulated collapse may be an essential element in chemokine-induced transmigration16.
Our static assay indicates that vimentin is required in both the receiving endothelial cell and in the migrating PBMC to stabilize endothelial cell–PBMC interactions. However, the mouse model showed that homing was inhibited only when vimentin was absent from the migrating cells. In contrast, when vimentin was absent from endothelial cells, homing was more active. This effect was explained by the intravital microscopy assay, which demonstrated that the endothelium of vim-/- mice displayed a marked leakiness. Thus, the endothelial integrity was severely compromised in vimentin-deficient mice. In addition to the leakiness, all critical parameters of endothelial cell–PBMC interaction (rolling velocity, adherence and transmigration) were altered in the vimentin-deficient endothelium. Taken together, these results demonstrate a critical role for intermediate filaments in regulating the barrier that governs leukocyte extravasation.
Our data demonstrate, for the first time, that intermediate filaments function in lymphocyte adhesion and transmigration. Using intermediate filaments as illustrative markers for cellular orientation allowed us to observe the transcellular migration pathway. The endothelial K8–K18 keratins and vimentin intermediate filaments were attached to and completely engulfed the PBMC. Although the involvement of a transcellular route in lymphocyte trafficking was suggested more than 40 years ago4, 17, the predominant view has been that all leukocytes use the paracellular route18. However, a number of recent studies have presented strong evidence in favour of the transcellular pathway3, 19, 20. Our data show that not only is the transcellular route used, but that PBMCs seem to have a preference for this route when leukocytes are under physiological conditions (that is, not activated and without the addition of chemokines). In contrast, we noticed that when cells were artificially activated with stimulatory cytokines, as previously described3 (resulting in activated lymphoblasts that are normally almost absent from the bloodstream), cell specificity was lost and both cell types could be directed to use the transcellular route (data not shown). Clearly, leukocytes have an innate capacity to use the transcellular route, but have a preference related to their overall signalling and differentiation status. This differential behaviour is a plausible explanation for the longstanding controversy in the literature — depending on the conditions and cell population used, the outcome is likely to vary significantly between different experiments.
Taken together, the active reorganization of intermediate filaments during transcellular migration (in both migrating PBMCs and receiving endothelial cells) seems to reflect the role of intermediate filaments in forming an anchoring and/or organizing structure for the adhesion molecules on the surface of both cell types. Recent studies of the transcellular pathway have indicated the presence of ICAM-1–VCAM-1-enriched, cup-like, transmigratory traction structures at the site of transmigration3, 20. Our data shows that the cytoplasmic counterpart of this cup-like structure is comprised of intermediate filaments and provides a mechanism for anchoring and organizing the surface molecules required for migration. Furthermore, the link between intermediate filaments, migration and organization of critical adhesion molecules was also demonstrated by the loss of expression and organization of ICAM-1 and VCAM-1 in vim-/- endothelial cells, and integrin-1 in vim-/- lymphocytes. In conclusion, our results identify a previously unknown intermediate-filament-based conveyor assembly that provides a mechanism for anchoring the adhesion molecules required for lymphocyte adhesion and transmigration.
Note added in proof: an accompanying manuscript byMilán , J.et al. (Nature Cell Biol. 8, 114–123 (2005)) is also published in this issue.
Methods
Cell isolation and culture.
HUVECs cells were isolated and cultured as previously described21. Endothelial cells were isolated from fresh mouse skin preparations of wild-type and vim-/- mice8 by treatment with type-I collagenase at 37 °C for 3 h. Transmigration experiments were carried out with confluent endothelial-cell monolayers grown on glass coverslips, transwell membranes or glass microslides (CamLab, Cambridge, UK) using freshly isolated lymphocytes and PMNs. For immunocytochemistry, human PBMCs and mouse lymphocytes were isolated from healthy volunteers and mouse spleen, respectively.
In vitro capillary-flow assays.
HUVECs were cultured in small glass capillaries as previously described22. PBMCs prepared from the peripheral blood of normal volunteers were resuspended (2 
106 cells ml-1) and drawn across recombinant TNF- (100 U ml-1 for 4 h)-activated HUVEC monolayers at a flow rate that resulted in a shear stress of 1.0 dyn cm-2. The time-lapse observation was made with Axiovert 200M microscopy and with Axiovision software.
Static transmigration assay.
Endothelial cells were seeded at 5 104 cells per coverslip and 3 104 cells per transwell insert (3-m pore size; Costar, Cambridge, MA). Endothelial-cell monolayers were stimulated with 100 U ml-1 TNF- 4 h before migration. PBMCs or PMNs (1 106) were added to the coverslips or to the inserts and allowed to migrate for 10–90 min at 37 °C. To observe the pathway of leukocyte diapedesis, optical sections were prepared using confocal laser microscopy (Carl Zeiss, Jena, Germany). Endothelial–leukocyte co-cultures were fixed with 4% paraformaldehyde and permeabilized using a 0.2% saponin solution. For the double immunolabelling experiment, the samples were first exposed to anti-K8–K18 mAb (a kind gift of B. Omary) or anti-vimentin mAb (Sigma-Aldrich, St Louis, MO) followed by FITC-conjugated anti-mouse IgG1. The samples were then incubated with a second monoclonal antibody against CD44 (Hermes-3), followed by phycoerythrin-conjugated anti-mouse IgG2a. The samples were examined using a laser-scanning confocal microscope (Zeiss LSM 510). Endothelial cells and lymphocytes from wild-type and vim-/- mouse and human PBMCs were stained with anti-ICAM-1, anti-integrin-1, and anti-vimentin antibodies followed by FITC-conjugated secondary antibodies. Samples were examined on a Zeiss LSM 510 Meta laser-scanning confocal microscope.
To visualize leukocyte–endothelial interactions by scanning electron microscopy, transwell inserts were fixed with 5% glutaraldehyde and then gradually dehydrated in ethanol. After critical-point drying, samples were coated with gold and endothelial–leukocyte interactions were observed and analysed with JEOL (JSM-5200) SEM and SemAfore4 software.
Static adhesion assay.
Endothelial cells from wild-type and vim-/- mice were seeded on 24-well plates (Costar, Corning, NY). Confluent endothelial-cell monolayers were stimulated with mouse TNF- (125 U ml-1) 4 h before adding the lymphocytes. Naive mouse lymphocytes (1 106) isolated from spleens and lymph nodes of wild-type and vim-/- mice were added to the wells and allowed to adhere for 90 min at 37 °C. Thirty-five microscope fields were randomly selected and images were taken using Zeiss AxioCam CCD camera. The number of adherent lymphocytes was calculated from the images and the data were analysed using ANOVA.
Intravital video-microscopy.
The method was adapted from ref. 23 and performed as described24. Briefly, 1.5 g TNF- was intrascrotally injected into anesthetized mice and after 6 h leukocyte–endothelial cell interactions were analysed using an Olympus microscope equipped with a long-distance water immersion objective. A CCD camera was used to record the experiments on a digital video recorder. The parameters that characterize leukocyte–endothelial interactions were measured off-line from these video recordings. The Newtonian wall shear rates (£) were approximated as £ = 2.12 8 Vb / d (s-1), where 2.12 is an empirical correction factor, d is the mean diameter of the vessel and Vb is the mean blood flow velocity (obtained by multiplying the centerline red-blood-cell velocity by an empirical factor of 0.625), as previously described25, 26. To measure leakiness of the vasculature, FITC–dextran (25 mg kg-1; molecular weight, 70 K) in 100 l of PBS was intravenously injected as a bolus into the mice. Leakiness was counted as the number of protrusions filled by FITC–dextran per 100 m vessel.
Transfections.
HUVECs were transfected with pEGFP–human vimentin (hVIM)–Myc (a kind gift of R. Goldman; 5 g per transfection) using electroporation (300V per 450F; GenePulser II electroporator, BioRad Laboratories, Hercules, CA). Cells were allowed to recover 48 h before the experiments.
Real-time-flow confocal microscopy.
For live-cell studies pEGFP–vimentin-transfected HUVECs were seeded on gelatin-coated 30-mm glass coverslips. After two days in culture, confluent monolayers were activated with rTNF- (100 U ml-1) for 4 h and mounted in a parallel plate flow chamber device (GlycoTech). PBMCs were suspended at 1 106 cells ml-1 and were drawn over the plate with a defined laminar shear stress (1.0 dyn cm-2). Chambers were placed on a heated plate holder warmed to 37 °C to permit transmigration. After a 15-min stabilization period, images were collected using a laser scanning microscope (Zeiss LSM 510, Zeiss, Germany). Every 45 s, sequential green channel and differential interference contrast (DIG) images were recorded for 60 min.
In vivo homing assay.
Spleens were collected from three wild-type and three vim-/- mice (kindly provided by C. Babinet, Institute Pasteur, Paris, France) and homogenized to obtain single-cell suspensions. After lysis of red cells, the unlysed cells were labelled for 20 min at 37 °C with 0.5 M carboxyfluorescein diacetate succinimidyl ester (CFSE; Molecular Probes, Eugene, OR; wild-type) or 5 M TRITC (Molecular Probes; vim-/-). Cells were washed three times with RPMI 1640 supplemented with 10% FCS, 1% 4 mM L-glutamine and 0.128% penicillin–streptomycin and then resuspended in RPMI. The labelling conditions were optimized and verified by pre-testing so that they did not affect the homing capacity of the lymphocytes. After labelling, wild-type and vim-/- cells were pooled and 1.5 107 cells were intravenously injected into seven vim-/- and seven age-matched wild-type recipients. After 4 h, spleens and mesenteric lymph nodes were harvested from the recipients and passed through a wire mesh to obtain single-cell suspensions. Cell suspensions were analysed by flow cytometry (FACScan® and CellQuest software; BD Biosciences, San Jose, CA). The homing index (HI = [Knockout cells]tissue/[Wild-type cells]tissue:[Knockout cells]input/[Wild-type cells]input) was calculated to correct the ratio between the two injected cell populations.
Peritonitis.
Mild inflammation was induced in the peritoneal cavities of six vim-/- and seven age-matched wild-type mice by intraperitoneal injection of 1 ml PBS containing 5% proteose peptone (BD Difco, Sparks, MD) and 10 ng interleukin-1 (IL-1; R&D Systems, Minneapolis, MN). Cells were collected from the peritoneal cavities 25 h after injection by washing with 10 ml RPMI containing 5 U ml-1 heparin (Løvens Kemiske Fabrik, Ballerup, Denmark). Leukocyte subtypes from lavage-fluid smears were analysed after staining with Reastain Quick-Diff (Reagena, Toivala, Finland).
Western blotting.
Wild-type and vim-/- cells were lysed and analysed by SDS–PAGE. The transfer of separated proteins to nitrocellulose membrane was performed by semi-dry blotting. After overnight blocking with 5% milk in PBST, membranes were probed with polyclonal anti-ICAM-1 (M19; Santa Cruz Biotechnology Inc., Santa Cruz, CA) and polyclonal anti-VCAM-1 (C19; Santa Cruz Biotechnology Inc.) for 2 h. The washed membranes were incubated with horseradish peroxidase (HRP)-conjugated secondary antibody for 1h and then visualized using an enhanced chemiluminescence procedure (Amersham–Pharmacia, Piscataway, NJ).
Flow cytometry.
Lymphocytes were isolated from mesenteric lymph nodes and spleen using a glass homogenizer and were filtrated to obtain single-cell suspensions. They were stained with anti-integrin-1 and then with FITC-labelled anti-rabbit IgG secondary antibodies. The cells were fixed in 1% paraformaldehyde and 1 104 cells were analysed using FACScan® and CellQuest software (BD Biosciences). A non-reactive negative control monoclonal antibody was used to set the level of nonspecific background staining.
Note: Supplementary Information is available on the Nature Cell Biology website.
