Blood Journal
Leading the way in experimental and clinical research in hematology

The prototype endothelial marker PAL-E is a leukocyte trafficking molecule

  1. Johannes Keuschnigg1,2,
  2. Tiina Henttinen3,
  3. Kaisa Auvinen1,
  4. Marika Karikoski1,2,
  5. Marko Salmi1, and
  6. Sirpa Jalkanen1
  1. 1MediCity Research Laboratory and Department of Medical Microbiology, University of Turku and National Public Health Institute, Turku;
  2. 2Turku Graduate School of Biomedical Sciences, Turku; and
  3. 3Department of Biology, University of Turku, Turku, Finland


Pathologische Anatomie Leiden-endothelium antibody has been used for more than 20 years as a marker for vascular endothelium. Despite its widespread use, the target of this antibody was only recently identified as plasmalemma vesicle–associated protein-1 (PV-1). However, no function has been identified for this molecule. Here we report that activation of human umbilical vein endothelial cells with tumor necrosis factor-α resulted in a remarkable redistribution of PV-1 toward the peripheral areas of the cells. Furthermore, in vitro endpoint transmigration experiments showed that transcellularly migrating lymphocytes are surrounded by rings containing PV-1 and caveolin-1. Moreover, PV-1 associates physically with vimentin. In addition, administration of anti–PV-1 antibody during capillary flow assays resulted in a significant inhibition of lymphocyte transmigration through the endothelial cell layer, whereas rolling and adhesion were unaffected. In vivo blockage of PV-1 by an antibody in acute peritonitis and air pouch model resulted in a significant decrease in the number of migrating leukocytes. Here we thus define leukocyte transendothelial migration as the first known function for PV-1.


Discrimination between vascular and lymphatic endothelium is crucial but challenging in cell biology and pathology. An antigen defined by the monoclonal antibody Pathologische Anatomie Leiden-endothelium (PAL-E) has been a “gold standard” to define vascular endothelium. In contrast to the other established markers for vascular endothelium, such as factor VIII, CD31, and endoglin, PAL-E is completely absent from lymphatic endothelial cells. It is now known that the PAL-E antibody recognizes plasmalemma vesicle–associated protein-1 (PV-1).1

PV-1 was originally discovered in rat lung endothelium.2 The protein has a long extracellular domain of 380 amino acids, a single span transmembrane domain and a short 27 amino acids long intracellular N-terminal domain. The amino acid sequence of any of the domains gives no hints about the possible function of this protein. PV-1 forms homodimers in situ, and its extracellular domain contains 4 N-glycosylation sites.3 In human, the protein has a molecular weight of approximately 60 kDa depending on the glycosylation. The subcellular localization of PV-1 is restricted to stomatal diaphragms of endothelial caveolae, containing high concentrations of caveolar proteins, such as caveolin-1. Furthermore, PV-1 was found in transendothelial channels and diaphragms of fenestrae in rat lung,2 all structures that have been implicated in the transcellular exchange of liquids and macromolecules.4,5

The migration of leukocytes from the bloodstream through the endothelium into the tissue is one of the central paradigms of inflammation and immunity. This process requires a sequence of events, starting with the retardation of leukocyte flow and leading via firm adhesion to transendothelial migration.6,7 Leukocytes have been shown to cross endothelia via the paracellular and transcellular pathway.8,9 Even though in vivo studies in the 1960s collected evidence that leukocytes can indeed extravasate from blood vessels via a transcellular pathway,10,11 this view was largely disregarded. Only in the past few years have new in vitro studies revived this concept.8,1214 Thus, although the recruitment of leukocytes from the bloodstream and the paracellular migration are well understood, the molecular mechanisms of transcellular migration are still incompletely resolved. Because the expression pattern of PV-1 would be compatible with its involvement in cell migration, this work was designed to analyze the behavior and role of PV-1 in the transmigration process both in vitro and in vivo.


Cell isolation and culture

Human umbilical vein endothelial cells (HUVECs) were isolated and cultured as previously described.15 Human dermal microvascular endothelial cells (HDMECs) were purchased from PromoCell. For isolation of lymphocytes, peripheral blood was collected from healthy volunteers and cells were isolated using Ficoll-Paque Plus (GE Healthcare) according to the manufacturer's instructions. The use of human material was approved by the Ethical Board of Turku University Hospital.


Liver, lung, heart, and mesenteric lymph nodes were retrieved from 3 vim−/− and 3 wild-type mice (all sv/129 background; 10 weeks old) and stained with Meca-32 rat anti–mouse PV-1 primary antibody16 and Alexa 488 goat anti–rat (Invitrogen) secondary antibody. Samples were analyzed on a Zeiss LSM 510 Meta laser-scanning confocal microscope (Carl Zeiss) using Zeiss LSM Imaging Software. Fluorescence intensity values and surface areas were determined with the Histogram function of the program, and numerical values were statistically analyzed using Microsoft Excel.

Endpoint transmigration and immunocytochemistry

HUVECs and HDMECs were seeded on glass coverslips (Menzel-Gläser) and grown to confluence. Endothelial cell monolayers were activated with 100 U tumor necrosis factor-α (TNF-α)/mL for 4 hours before migration. A total of 106 lymphocytes were added to the coverslips and allowed to transmigrate for 30 to 60 minutes at 37°C. Cells were subsequently fixed with 4% paraformaldehyde and permeabilized using a 0.2% saponin solution. For double stainings, cells were first exposed to anti–PV-1 monoclonal antibody (mAb) 174/2 (mIgG1),1 followed by Alexa 488 conjugated goat anti–mouse IgG1 secondary antibody (all secondary antibodies were from Invitrogen). Thereafter, samples were incubated either with chicken anti–vimentin antibody (a kind gift of J. Eriksson Åbo Akademi University, Turku, Finland) or mouse anticaveolin-1 (mIgG2a) antibody (BD Biosciences) followed by appropriate Alexa 546 goat anti–chicken or mouse IgG2a secondary antibodies.

For analysis of surface expression of PV-1 on HUVECs, cells were incubated with 20 μg/mL of either anti –PV-1 (174/2) or isotype-matched negative control antibodies (NS-1 or 3G6) for 40 minutes at 37°C, 5% CO2. Thereafter, cells were placed on ice and incubated with Alexa 488 goat anti-mIgG secondary antibodies for 40 minutes. Subsequently, samples were fixed with 4% paraformaldehyde. Samples were analyzed on a Zeiss LSM 510 Meta laser-scanning confocal microscope.


HUVECs were isolated, put directly into 6-well plates, and grown until confluence. Precipitations were performed from nonactivated and TNF-α–activated HUVECs (100 U TNF-α/mL medium for 4 hours) using M-270 Epoxy Dynabeads (Dynal). Coating of beads and coimmunoprecipitation were performed as previously described.17 Beads were coated with either V9 (mouse anti–human Vimentin IgG1; Sigma-Aldrich) or an isotype-matched negative control AK1 (In Vivo Biotech Services GmbH). Dried samples were suspended in Laemmli sample buffer containing 5% mercaptoethanol and analyzed on a 6% to 12% gradient sodium dodecyl sulfate–polyacrylamide gel electrophoresis.


The proteins in the gel were transferred to nitrocellulose membranes; and after blocking of nonspecific binding with 5% milk in phosphate-buffered saline-Tween20 for 1 hour, the membranes were probed with the same antibodies used for coimmunoprecipitation for 1 hour. The washed membranes were incubated with horseradish peroxidase–conjugated rabbit anti–mouse IgG immunoglobulins (Dako Denmark) for 1 hour and subsequently visualized using Immobilon Western Chemiluminescent horseradish peroxidase substrate (Millipore).

Live cell imaging

Full-length human PV-1 was fused with its C-terminus to enhanced green fluorescent protein (EGFP) in a pFP-N1 vector (Clontech). HUVECs from passages 2 to 6 were transfected with the PV-1–EGFP construct using an Amaxa Nucleofector Kit in combination with an Amaxa Nucleofector II (Amaxa) according to the manufacturer's instructions. Transfected cells were transferred to the central glass part of glass-bottom petri dishes (MatTek), coated with fibronectin, and allowed to recover for 24 hours. Subsequently, the cells were activated with 100 U TNF-α/mL medium for 4 hours and 106 lymphocytes were added. Cells were allowed to transmigrate for 30 to 60 minutes before time-lapse confocal microscopy was started.

Flow cytometry

HUVECs were activated for 4 hours with 500 U TNF-α/mL medium. Thereafter, surfaces of cells were stained for PV-1 using 174/2 and 2 class-matched negative controls (NS-1 and 3G6), respectively (10 μg/mL each) for 30 minutes on ice without permeabilization followed by 30 minutes incubation with fluorescein isothiocyanate anti-mIgG secondary antibody (Sigma-Aldrich) and antifluorescein Alexa 488 goat IgG fraction (Invitrogen). Stainings were analyzed on a FACSCalibur System using CellQuest Pro software (both BD Biosciences).

Capillary flow assay

The experiment was performed as previously described.18 In brief, HUVECs were plated into fibronectin-coated perpendicular glass capillaries and grown until confluence. The endothelial cells were induced with 500 U TNF-α/mL medium for 4 hours. For the last 20 minutes, anti-PV-1 mAb 174/2 or negative control mAb 3G6 was added at a concentration of 10 μg/mL. The assay started with a one-minute stabilization period with binding buffer (Dulbecco phosphate-buffered saline containing 0.1% human serum albumin) at 1 dyne cm−2. The peripheral blood mononuclear cells (PBMCs; 106 cells/mL in Dulbecco phosphate-buffered saline containing 0.1% human serum albumin) were then perfused through the capillary for 7 minutes at 1 dyne cm−2 after the perfusion of binding buffer for 33 minutes at equal laminar shear, allowing adherent cells to transmigrate.

The rolling and adherent cells were analyzed 3 minutes after PBMCs were drawn into the capillary by recording 10 fields (0.3 mm2 per field) for 15 seconds each. The transmigrating and adherent cells were analyzed at 20 and 40 minutes after the perfusion of PBMCs by recording again the same 10 fields. Assays were performed with an Olympus IX70 inverted microscope (Olympus) using 100× original magnification. Analyses were carried out off-line by manual counting. Independent assays were made using 3 separate HUVECs and PBMCs isolated from 3 different donors. Five fields per capillary were analyzed for rolling cells and 10 fields for adherent and transmigrated cells. Cells were defined as rolling if they moved slowly into the direction of flow and as adherent if they stayed stationary. In later time points, phase bright cells were counted as adherent and phase dark cells as transmigrated underneath the endothelium. The percentage of transmigrating cells was calculated as [migrating cells]/([adherent cells] + [migrating cells]). The number of counted cells (mean ± SEM) were 137 ± 16 cells rolled and 241 plus or minus 38 adhered in negative control-treated endothelium at the 3-minute time point. After 20 minutes, 587 ± 104 cells adhered and 166 ± 35 cells transmigrated and at 40 minutes, 556 ± 70 adhered and 170 ± 48 transmigrated in negative control-treated capillaries. To study the effect of PV-1 on polymorphonuclear cell (PMN) transmigration, 3 independent experiments were performed with different HUVECs and PMNs as described for PBMCs. Adherent and transmigrating cells were analyzed after 13 minutes.


Mild inflammation was induced in the peritoneal cavities of age-matched (6-8 weeks old) wild-type Balb/C mice by intraperitoneal injection of 1 mL phosphate-buffered saline containing 5% proteose-peptone (BD Difco) and 10 ng interleukin-1β (IL-1β; R&D Systems). One hour later, antibodies against PV-1 (Meca-32; n = 13) or human HLA-DR5 (negative control; n = 11) were injected into the tail vein. Cells were collected from the peritoneal cavities 18 hours after injection by washing with 10 mL RPMI containing 5 U heparin/mL (Løvens Kemiske Fabrik). To ensure that antibodies did not induce leukopenia, whole-blood samples were collected and analyzed using a cell counter (Nihon Kohden MEK-6108K).

Leukocyte subtypes from lavage fluid smears were analyzed after staining with Reastain Quick-Diff (Reagena). The experiments were approved by the Animal Care Committee of the University of Turku (Turku, Finland) and conformed to the guidelines established by the European Union.

Air pouch model

The experiment was performed as previously described.19 Briefly, air pouches were created on the backs of Balb/C mice (∼ 10 weeks old) by subcutaneous injection of 5 mL of filtered air (Millex-GV filter unit; Millipore). Injections were repeated the following day with 3 mL of filtered air. On day 3 after establishment of the air pouch, mild inflammation was induced by injection of 1 mL of RPMI containing CCL-21 (1 μg/mL, SLC, mouse 6Ckine; R&D Systems) and bovine serum albumin (50 μg/mL) into the pouches. At 5 and 9 hours after the last injection, the mice were treated with 100 μg of antibodies Meca-32 (rat IgG2a anti mouse PV-1; n = 8) or negative control MJ7/18 (rat IgG2a anti mouse endoglin; n = 6). Cells were collected from the air pouches 9 hours after the last antibody injection with 5 mL RPMI containing 5 U/mL of heparin and counted. Blood samples were analyzed to rule out induction of leukopenia by antibodies.

Statistical analyses

Results are expressed as mean plus or minus SEM. Two-tailed Student t test with unequal variance was used to evaluate the statistical significance of the data.


Expression of PV-1 is reduced in vim−/− mice

We have earlier demonstrated a severe impediment in the capacity of leukocyte adhesion and transcellular migration in vimentin-deficient (vim−/−) mice.13 Therefore, we first investigated possible consequences of the absence of vimentin on the expression of PV-1. We found that the expression levels of PV-1 were reduced approximately 20% in the heart and spleen vasculature of vim−/− mice (Figure 1A-B), compared with wild-type mice, but no detectable differences were seen in the staining pattern of PV-1 between these mouse strains (supplemental Figure 1, available on the Blood website; see the Supplemental Materials link at the top of the online article).

Figure 1

Expression of PV-1 is reduced in vim−/− mice, and its distribution changes significantly on activation with TNF-α. (A-B) Frozen sections of heart and spleen from wt and vim−/− mice were stained for PV-1 with Meca-32 (rat anti–mouse PV-1) or negative control antibody. Vascular areas were digitally determined and the (A) fluorescence intensity values and (B) surface area of the analyzed vasculature were determined using the histogram function of the Zeiss LSM software. Two sections per mouse of 3 wild-type and 3 vim−/− mice were analyzed. The data are shown as mean ± SEM. (C) Double-staining of nonactivated and TNF-α–activated HUVECs with anti–PV-1 and antivimentin antibodies. The arrows point to the PV-1 pool in peripheral areas of the cells, where a partial colocalization with vimentin can be seen. (D) Double staining of nonactivated and activated HUVECs with anti–PV-1 and anti–caveolin-1 antibodies. The arrows point to the peripheral areas of the cells, where PV-1 and caveolin-1 colocalize. Nuclear counterstaining was performed with 4,6-diamidino-2-phenylindole (DAPI). Scale bars represent 10 μm.

Because the result suggested a relationship between PV-1 and vimentin, we investigated the localization of these 2 proteins at the cellular level in HUVECs. In nonactivated cells, PV-1 and vimentin were evenly distributed over the cell with occasional colocalization (Figure 1C). However, on activation for 4 hours with 100 U TNF-α/mL, the distribution of PV-1 changed significantly (Figure 1C). PV-1 was now concentrated in peripheral areas of the cells. In addition, a partial colocalization with a fraction of vimentin could be detected at these locations (Figures 1C, supplemental Figure 2). Consistent with previous studies,20 PV-1 highly colocalized with the caveolar marker caveolin-1; interestingly, both proteins showed an identical redistribution on activation of the cells (Figure 1D). These stainings thus indicate a partial colocalization of PV-1, vimentin, and caveolin-1 in endothelial cells.

PV-1 physically associates with vimentin

To test whether PV-1 and vimentin are physically binding to each other, we coimmunoprecipitated these 2 proteins from resting and activated HUVECs. Consistent with the immunohistologic data, PV-1 and vimentin could be coimmunoprecipitated from nonactivated (Figure 2A) and TNF-α–activated HUVECs (Figure 2B). The 2 bands in lane 1 of Figure 2B most probably represent 2 differentially glycosylated isoforms of PV-1.5

Figure 2

PV-1 and vimentin can be coimmunoprecipitated from HUVECs. (A) Coprecipitations from nonactivated HUVECs. (B) Coprecipitations from TNF-α–activated HUVECs. HUVECs were lysed and the lysate incubated with magnetic beads coupled to primary antibodies against vimentin and an isotype-matched negative control antibody (AK1). Detection was performed with antibodies against the suspected binding partner indicated in the figure. Arrows point to nonspecific bands deriving from the heavy chains of the immunoglobulins. PV-1 bands of approximately 62 kDa can be seen in lane 1 (A-B), where precipitation was performed with anti-vimentin antibody and detection with anti–PV-1 antibody. In addition, in panel B lane 1, a weaker PV-1 band of approximately 55 kDa can be seen.

Dynamic redistribution of PV-1 during transcellular transmigration

Because vimentin is known to participate in the transcellular migration of leukocytes,13 we next analyzed the behavior of PV-1 in endpoint transmigration assays. Lymphocytes from healthy volunteers were allowed to interact with activated HUVEC or HDMEC monolayers for 30 to 60 minutes and subsequently stained for PV-1, vimentin, and caveolin-1 (“Endpoint transmigration and immunocytochemistry”). Emigrating lymphocytes crossed the endothelial cells in the perinuclear area through transendothelial channels. In HUVECs, these channels had a diameter of approximately 7 μm, whereas their size in HDMEC averaged to approximately 13 to 15 μm (Figure 3A; supplemental Videos 1-2). These conduits formed in front of the migrating cells and deepened gradually as the lymphocytes penetrated the endothelial cell. The channels were surrounded by a PV-1– and caveolin-1–containing ring around the emigrating lymphocyte (Figure 3B). Furthermore, a vimentin network embracing the lymphocytes was detected in the endothelial cells (Figure 3C; supplemental Video 3). These results demonstrate a dynamic distribution of PV-1 and partial colocalization of PV-1 with vimentin and caveolin-1 during lymphocyte transmigration.

Figure 3

Lymphocytes cross endothelial cells via large transcellular channels marked by rings of PV-1 and caveolin-1 and are embraced by a vimentin network. (A) Lymphocyte migrating through a HUVEC transfected with a PV-1–EGFP construct. A pore forms in front of the transmigrating cell and widens to a diameter of approximately 7 μm (white arrows). The time course is outlined in minutes in the upper left corner of the pictures. (B) Localization of endogenous PV-1 and caveolin-1 in an endpoint transmigration assay. Lymphocytes were overlaid on top of TNF-α–activated HUVECs, allowed to transmigrate, and fixed. A transendothelial channel with a migrating lymphocyte is depicted. The channel is outlined by colocalizing rings of PV-1 and caveolin-1 (white arrows). (C) Localization of PV-1 and vimentin. HDMECs were stained for vimentin and PV-1 after an endpoint transmigration assay. White arrows point to an accumulation of PV-1 around the migrating cell and to a vimentin network surrounding the lymphocyte. Nuclear counterstaining was performed with DAPI. Scale bars represent 10 μm.

PV-1 is involved in lymphocyte transmigration in vitro

In the next step, we wanted to analyze the function of PV-1 under more physiologic conditions. Thus, we performed capillary flow assays, in which the transmigration of leukocytes under the influence of defined laminar shear stress can be investigated. We first confirmed that PV-1 is expressed on the surface of HUVECs using flow cytometry (Figure 4A) and confocal microscopy (Figure 4B). Both methods demonstrated a significant amount of PV-1 on the surface of HUVECs. For the capillary flow assays, HUVECs were grown in glass capillaries, activated with TNF-α, and incubated with either anti-PV-1 or negative control antibody (3G6). PBMCs were isolated and drawn through the capillaries with a constant laminar shear stress of 1 dyne cm−2. Administration of anti–PV-1 antibody did not significantly affect the numbers of either rolling or adherent (P = .54 and P = .42, respectively) cells (Figure 4C). However, transmigrating cells were significantly reduced at both time points analyzed.

Figure 4

Migration of leukocytes through endothelium is impaired by blockage of PV-1 under flow. (A) Analysis of PV-1 expression on the surface of HUVECs. Cells were incubated with anti–PV-1 or 2 different isotype-matched negative control antibodies (3G6 or NS-1). (B) Cytochemical detection of PV-1 expression on the surface of HUVECs. A representative cell from the brightly stained population is depicted. Side views of a z-stack are shown from 2 different planes, indicated by colored lines in the small pictures on top and on the right. Small inset in the upper right corner represents negative control staining. Arrows point to a PV-1 pool on the surface. Nuclear counterstaining was performed with DAPI. (C) Analysis of rolling, adherent, and transmigrated PBMCs at the indicated time points using capillary flow assay. (D) Analysis of polymorphonuclear cell adhesion and transmigration under flow. The graphs represent mean ± SEM of 3 independent experiments using HUVECs and leukocytes from different individuals. Values of capillaries treated with negative control antibodies have been set as 100%.

We then decided to test whether PV-1 treatment would have a similar effect on the transmigration of PMNs. Under these conditions (HUVECs, 4-hour activation with TNF-α), PMNs have been reported to extravasate via the paracellular pathway.21 Treatment of capillaries with PV-1 antibody affected neither adhesion nor transmigration of PMNs across the endothelial monolayer (Figure 4D).

Inhibition of PV-1 attenuates inflammation in vivo

To study whether these findings translate into an in vivo situation, we performed experimental peritonitis as a model of acute inflammation. In this model, proteose-peptone mixture containing IL-1β was injected intraperitoneally to wild-type mice. One hour later, antibodies (anti–mouse PV-1 or negative control antibody) were administered to the tail vein; and 18 hours after that, cells from the peritoneal cavity were collected. The 18-hour time point allows the entrance of lymphocytes and monocytes in addition to neutrophils into the peritoneal cavity. Compared with the negative control antibody, Meca-32 (rat anti–mouse PV-116) reduced the migration of leukocytes into the peritoneal cavity by approximately 85% (Figure 5A). When investigating the various leukocyte subsets, numbers of neutrophils, macrophages, and lymphocytes were all found to be significantly decreased after anti–PV-1 antibody treatment (Figure 5B). Numbers of macrophages and lymphocytes were reduced to quantities comparable with those found in mice with noninflamed peritoneal cavities and neutrophil counts decreased by approximately 65%.

Figure 5

Migration of leukocytes to sites of inflammation in mice is impaired by blockage of PV-1. (A) Total number of leukocytes in peritoneal lavages. Peritonitis was induced by intraperitoneal injection of proteose-peptone + IL-1β. One hour later, antibodies were injected to the tail vein and peritoneal lavages collected after 18 hours. Cells of mice without inflammation have been also counted and set as 100%. (B) Analysis of leukocyte subsets in the peritoneal lavages. *P ≤ .05, **P ≤ .01, ***P ≤ .001. N = 11 for negative control–treated mice, n = 13 for PV-1–treated mice, and n = 6 for mice without inflammation. (C) Total numbers of leukocytes retrieved from the air pouches 18 hours after induction of inflammation by injection of CCL-21 and bovine serum albumin. Values of negative controls were set as 100%. (D) Analysis of leukocyte subpopulations. N = 6 for negative control (antiendoglin mAb)–treated mice and n = 8 for anti–PV-1 mAb–treated mice. Bars represent mean ± SEM.

To confirm the role of PV-1 in inflammatory cell trafficking, we performed a second in vivo experimental model. Analysis of cell numbers in the air pouches revealed a 38% reduction after PV-1 treatment, compared with anti-endoglin treatment (Figure 5C). The number of granulocytes decreased by 44% and that of lymphocytes by 56%, whereas macrophages remained unaffected (Figure 5D). Induction of leukopenia by antibodies was ruled out in both models by analyzing whole-blood samples for white blood cell counts (supplemental Figure 3). These results unambiguously show that PV-1 is a functionally important molecule controlling leukocyte transmigration.


This work demonstrates active changes in the subcellular localization of PV-1 during the transcellular transmigration. Moreover, it shows intimate colocalization of PV-1 with a subset of caveolin-1 and vimentin in this process. Most importantly, in vitro and in vivo inhibition studies using anti–PV-1 antibodies reveal a functional role for PV-1 in the process of leukocyte transmigration.

PV-1 dimers are suggested to form the diaphragms of fenestrae, caveolae, and stomata. The intracellular domains of the dimers are proposed to be stabilized by scaffold proteins, and a connection to the cytoskeleton was suggested.4 Thus, it seems possible that PV-1 is connected to vimentin and is stabilized by the vimentin intermediate filaments. The decrease in the PV-1 levels in vim−/− mice might thus be explained with a lack of proper attachment because the absence of vimentin intermediate filaments could lead to a loss of PV-1 from the cytoplasmic membrane and to subsequent degradation.

A remarkable redistribution of PV-1 and caveolin-1 toward the cell periphery of endothelial cells took place on activation with TNF-α. Furthermore, PV-1 colocalized with vimentin and caveolin-1 in these areas, which could be demonstrated using immunocytochemistry. In addition, coimmunoprecipitations confirmed a physical binding between PV-1 and vimentin. It has been shown that endothelial cells under the influence of flow exhibit an elongated form, placing attaching leukocytes almost always in the periphery of the cells.22 Moreover, several in vivo studies of bone marrow, high endothelial venules of Peyer patches, and blood-brain barrier2325 have demonstrated that leukocyte transcellular migration occurs in the immediate vicinity of interendothelial junctions. Quantitative analyses of migration sites show that 60% of the pores form less than 1 μm away from the endothelial junctions.24 These areas represent the thinnest parts of endothelial cells, often being only 200 nm thick, and thus pose the least resistance to extravasating leukocytes. However, it is often difficult to classify migratory events near interendothelial junctions with certainty as transcellular. Thus, to unambiguously demonstrate transcellular migration, frequently events distant of junctional areas are presented (as was done here in Figure 3A).

Recent in vitro data have shown that leukocytes “probe” the endothelial cells with “podosome-like protrusions”14 in the search of sites permissive for transcellular extravasation. In this context, the redistribution of PV-1 toward the cell borders might be understood as a preparation of the endothelial cell for transcellular migration of the leukocyte, thereby actively contributing to the process by reducing the resistance through regulated membrane fusion. This theory is supported by previous reports, showing an enrichment of caveolar proteins (eg, caveolin-1) as well as soluble-N-ethylmaleimide-sensitive-factor accessory-protein receptor and fusogenic proteins (VAMP2 and VAMP3) in endothelial cells near podosome-like protrusions.12,14

The capillary flow assay allows the identification of the step of leukocyte extravasation, during which PV-1 exerts its effect. Using laminar flow and human endothelial cells, this assay mimics in vivo conditions and its results are thus relevant for in vivo situations. Neither rolling nor adherent cells were affected by the anti–PV-1 antibody. In contrast, the number of transmigrating PBMCs was reduced by 30% to 45% when PV-1 was targeted with the antibody. These results and the behavior of PV-1 during imaging of lymphocyte transmigration (Figure 3) indicate that human PV-1 controls transmigration of leukocytes and most probably migration via the transcellular pathway. This theory is supported by the results of the capillary flow experiment, showing no effect of PV-1 antibody on the transmigration efficiency of polymorphonuclear cells as PMNs have been shown to predominantly cross TNF-α activated HUVEC monolayers via the paracellular pathway.21 Moreover, these results show that anti–PV-1 antibodies can be function-blocking, and this should be kept in mind when using this marker for cell isolation or phenotyping.

Our in vivo data unambiguously demonstrate that PV-1 plays a crucial role in the process of leukocyte migration to sites of inflammation. Blockage of the protein by an antibody in a model of acute peritoneal inflammation reduced the migration of leukocytes by approximately 85%. Neutrophils are among the first cells reaching the sites of inflammation. This neutrophil invasion was inhibited by approximately 65%. The time points in our model of peritonitis were chosen in a way that would also allow the slower-migrating monocytes and lymphocytes to reach the site of inflammation. Their ability to get to the site of inflammation was almost completely blocked after treatment of mice with PV-1 antibody.

To confirm these findings, we performed an air pouch model as form of peripheral inflammation and also included a negative control antibody, which binds to mouse endothelial cells. It is difficult to find a binding negative control antibody that would not potentially interfere with any of the steps of the leukocyte adhesion cascade or the extravasation per se. Finally, we chose the rat anti–mouse endoglin antibody MJ7/18 as a negative control. Targeting of PV-1 reduced the number of invading leukocytes by approximately 38%, compared with endoglin targeting. However, the actual efficiency of the PV-1 inhibition might be even higher, as endoglin has been indirectly implicated in the process of leukocyte transmigration. It has been shown that endoglin is up-regulated during inflammation and is “highest in endothelial cells with an associated inflammatory cell infiltrate.”26 Furthermore, endoglin has been demonstrated to promote transforming growth factor-β–mediated signaling,27 which in turn has been shown to reduce leukocyte adhesion and transmigration via inhibition of E-selectin and IL-8 expression.2831 Indeed, in our case, treatment with antiendoglin antibody reduced the migration of leukocytes to the air pouch by approximately 28% compared with the nonbinding antibody (HB151) treatment, but this decrease did not reach statistical significance (data not shown).

We also tested whether PV-1 was expressed on leukocytes, thus ensuring that our antibody treatment affected only the endothelial cells. Surprisingly we found PV-1 expression in lymphocytes and monocytes (data not shown). However, this PV-1 population was strictly intracellular and remained so even after activation of leukocytes. Thus, our antibody treatment affected only the endothelial cells expressing PV-1 on their surface.

The inhibition of PMN transmigration in these in vivo models might seem to pose a discrepancy to an earlier report,13 in which PMNs transmigrated via the paracellular pathway. However, the capillary flow assay in this previous study was performed with 4-hour TNF-α–activated HUVECs. These settings have been reported to induce PMN paracellular migration.21 The same study further demonstrates that prolonged activation of endothelial cells over 24 hours results in a switch of PMN transmigration from paracellular to transcellular. Our PV-1 antibody did not have any effect in the capillary flow conditions reported to support paracellular migration. However, after long time activation of endothelial cells (18 hours) in vivo, it seems possible that the route of PMN transmigration is predominantly transcellular. This would explain the effect of the PV-1 treatment on PMN transmigration in vivo. The previously reported change of lymphocyte recruitment in vivo from transcellular to paracellular in vim−/− mice13 can be explained with the leaky vessels caused by the absence of vimentin. In this situation, the barrier function posed by interendothelial junctions is markedly reduced, thus strongly favoring the paracellular pathway.

It is not possible to determine which route of extravasation leukocytes chose in these models. However, several lines of evidence indicate that the inhibited pathway could be the transcellular one: First, it has been shown that PBMCs cross endothelia preferably via the transcellular pathway.13 Second, even polymorphonuclear cells choose the transcellular pathway in vitro after long-term activation of endothelial cells.21 Finally, numerous studies in the literature report the transcellular pathway to be the predominant one in various in vivo settings. These in vivo conditions include, for example, migration of lymphocytes into the lymph nodes in mice,32 neutrophil invasion into inflamed skin in mice33 and humans,34 as well as the crossing of the blood-brain barrier by mononuclear leukocytes25 and lymphocytes35 in mice.

Additional evidence supports this idea as knocking down of caveolin-1 by siRNA significantly reduces transcellular migration,12 and another study reports considerably lower levels of PV-1 in the lung of caveolin-1−/− mice.36 Thus, as the lack of caveolin-1 causes reduced levels of PV-1, it may be envisioned that the impaired capacity for transcellular migration after knockdown of caveolin-1 by siRNA12 is a cumulative effect of the diminished levels of both caveolin-1 and PV-1 or even mediated exclusively by PV-1.

Taken together, the capillary flow assays and in vivo experiments unambiguously show that PV-1 is involved in extravasation of leukocytes both in humans and mice. Interestingly, PV-1 does not affect the first steps of the extravasation cascade (rolling and firm adhesion) at all but exerts its function solely during the transendothelial migration. These data reveal a physiologic function of PAL-E 2 decades after its discovery. At the same time, they indicate that PAL-E/PV-1 is a new potential target for preventing harmful cell trafficking to sites of inflammation.


Contribution: J.K. designed and performed experiments, analyzed data, and wrote the manuscript; T.H. and S.J. designed experiments, designed the study, analyzed data, and edited the manuscript; K.A. and M.K. designed and performed experiments and analyzed data; and M.S. designed experiments, analyzed data, and edited the manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Sirpa Jalkanen, MediCity Research Laboratory, University of Turku, Tykistökatu 6 A, FIN-20520 Turku, Finland; e-mail: sirpa.jalkanen{at}


The authors thank J. Eriksson for vimentin knockout mice and antivimentin antibody, E. Butcher for antiendoglin- and Meca-32 antibodies, and A. Sovikoski-Georgieva for secretarial assistance.

This work was supported by the Finnish Academy, the Sigrid Juselius Foundation, the Finnish Cancer Organization, and the Arvo and Inkeri Suominen Foundation.


  • The online version of this article contains a data supplement.

  • The publication costs of this article were defrayed in part by page charge payment. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734.

  • Submitted November 10, 2008.
  • Accepted April 21, 2009.


View Abstract