Published online before print April 1, 2008
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Departments of Physiology and Pharmacology and Molecular Medicine and Surgery, Karolinska Institute, Stockholm, Sweden
1Correspondence: Dept. of Physiology and Pharmacology, Karolinska Institutet, S-171 77 Stockholm, Sweden. E-mail: einar.eriksson{at}ki.se
ABSTRACT
L-selectin is important in mediating leukocyte recruitment in inflammation. The role of L-selectin was for long believed to be influenced by an inducible endothelial ligand; however, L-selectin ligand activity was recently shown to be mediated by leukocytic P-selectin glycoprotein ligand 1 (PSGL-1). Still, it is unknown whether PSGL-1 is deposited on the endothelium or whether leukocyte fragments or leukocytic uropods are presented on the venular surface. Moreover, it is unclear whether ligands for L-selectin other than PSGL-1 are present in inflammation. Overall, this has complicated understanding of the mechanisms that guide recruitment of inflammatory cells. Here, I used intravital microscopy on mouse cremaster muscle venules to show that L-selectin influences leukocyte rolling in inflammation exclusively by mediating L-selectin/PSGL-1-dependent, secondary capture to rolling and adherent leukocytes. I show that leukocyte primary capture in inflammation is mediated almost entirely by P-selectin, whereas the capacity of E-selectin to mediate capture appears to be minimal. In parallel, primary capture remaining after function inhibition of P-selectin is not decreased by blockage or absence of L-selectin. Rolling along the endothelium in venules following a number of inflammatory treatments was abolished by simultaneous blockage of P-selectin, E-selectin, and VCAM-1, indicating that there is no additional adhesive pathway involving L-selectin or any other molecule that can mediate leukocyte rolling in inflamed cremaster muscle venules in response to the used stimuli. Moreover, in vivo staining failed to detect any L-selectin ligand activity on the endothelium. These data demonstrate that expression of L-selectin on leukocytes is insufficient for mediating rolling and efficient recruitment of leukocytes in inflammation.
Key Words: rolling • integrin • capture
INTRODUCTION
The multistep recruitment cascade guides leukocyte trafficking and immune function [1 , 2 ]. The leukocyte cell adhesion molecule (CAM) L-selectin is a key molecule in this process through its capacity to mediate leukocyte rolling along high endothelial venules (HEV) and subsequent homing of lymphocytes to secondary lymphoid organs [3 , 4 ]. Important roles of L-selectin have also been demonstrated in peripheral inflammation [5 6 7 8 ]. These roles have largely been attributed to the presence of an endothelial L-selectin ligand that mediates rolling along inflamed endothelium. However, although several endothelial molecules have been suggested to mediate adhesive interactions with L-selectin, the identity of this ligand(s) was for long unclear.
In lymphoid tissue, carbohydrate structures capable of mediating rolling by interacting with L-selectin on leukocytes are presented on the surface of specialized high endothelial cells [9 ]. These structures include the peripheral node addressin, which is presented on CD34 [10 11 12 ] and other molecules as protein backbones. In gut-associated lymphoid tissue, mucosal addressin CAM-1 likely acts as the main endothelial ligand for L-selectin that mediates initial attachment and subsequent homing of lymphocytes [13 14 15 ]. In peripheral inflammation, P-selectin was pointed out early as having L-selectin-binding capacity by experiments carried out in vitro [16 ]. However, such interaction has not been demonstrated in other models. In vitro work has also indicated that E-selectin may function as an L-selectin ligand, at least in humans [17 , 18 ]. In contrast, ICAM-1 influences L-selectin-dependent rolling in vivo [19 20 21 ]; still, the mechanism responsible for the interplay between ICAM-1 and L-selectin has not been determined. L-selectin has also been demonstrated to bind other molecules on the endothelium [22 , 23 ], but no evidence for any role of these molecules in leukocyte recruitment has been presented. Apparently, the roles for suggested ligands for L-selectin expressed by inflamed endothelium are unclear.
An alternative function of L-selectin in leukocyte recruitment is the mechanism of secondary capture in which rolling and adherent leukocytes present an adhesive surface that could redirect leukocytes in free flow toward the vessel wall and induce their capture and subsequent rolling along the endothelium [24 , 25 ]. Secondary capture is mediated by L-selectin interacting with P-selectin glycoprotein ligand 1 (PSGL-1), the main selectin ligand on leukocytes [26 27 28 ], and has been shown to have an important function by increasing capture and rolling in inflammation [29 ]. Nonetheless, the view that an L-selectin ligand is expressed on inflamed endothelium has persisted in the literature [30 ].
Recently, Sperandio et al. [31 ] showed that PSGL-1 is responsible not only for secondary capture but also for L-selectin-dependent rolling in inflamed venules. The authors postulated that although most of the activity of PSGL-1 is related to interactions between leukocytes, they also suggested that PSGL-1 is presented on cell fragments deposited on the endothelium by rolling and adherent cells. Thus, the current view is that PSGL-1 acts as an L-selectin ligand on the endothelial surface that could mediate capture and rolling in inflammation. However, it is still unknown whether L-selectin can interact with other molecules on the endothelium, such as those being demonstrated to have L-selectin-binding capacity in vitro. Moreover, deposition of leukocyte fragments on the endothelium would also likely influence other properties of inflamed endothelium, not only those involved in leukocyte adhesion. Therefore, the details of L-selectin-dependent rolling in inflammation and the possibility of leukocyte fragments being involved in this process need to be elucidated.
In this study, I present data that indicate that all effects on leukocyte rolling in acute inflammation that are mediated by L-selectin can be attributed to secondary capture to rolling or adherent leukocytes. I demonstrate that no endothelial ligand for L-selectin, expressed by the endothelium or deposited by leukocytes, is capable of mediating capture or rolling in cremaster venules after several stimuli. These findings clarify the roles for the three selectins in leukocyte recruitment and suggest that fragments of leukocytes play minor roles in inflammation.
MATERIALS AND METHODS
Antibodies and reagents
mAb RB40.34 against mouse P-selectin (30 µg per mouse), mAb 6C7 against mouse VCAM-1 (30 µg per mouse), and mAb 4RA10 against mouse PSGL-1 (40 µg per mouse) were kindly provided by Dietmar Vestweber, University of Münster (Germany). mAb 9A9 (20 µg per mouse) against mouse E-selectin was kindly provided by Barry A. Wolitzky (previously at Hoffmann-LaRoche, Inc., Nutley, NJ, USA). The hybridoma for mAb MEL14 against mouse L-selectin (American Type Culture Collection, Manassas, VA, USA) was cultured, and antibody was prepared from supernatants on a protein L column. The antibody was cleaved by papain (Sigma Chemical Co., St. Louis, MO, USA) to release Fab fragments, which were again purified by protein L. MEL14 Fab fragments were used at 50 µg per mouse. mAb PS2 against mouse 
4 integrin (50 µg per mouse) was obtained from Serotec (Oxford, UK). The efficiencies and specificities of all antibodies used have been tested in previous studies. Recombinant human (h)TNF- and the mouse L-selectin/hIgG Fc chimera were obtained from R&D Systems (Minneapolis, MN, USA). Rhodamine 6G came from Sigma Chemical Co. FITC-labeled microspheres were obtained from Molecular Probes (Eugene, OR, USA), and beads coated with protein G were obtained from Polysciences Inc. (Warrington, PA, USA).
Animals
Male wild-type (WT) C57BL/6 mice were obtained from B&K (Stockholm, Sweden). Selectin-deficient mice were obtained from Jackson Laboratory (Bar Harbor, ME, USA) as B6/129F2 and backcrossed for five generations to C57BL/6. Animals were fed normal chow and water ad libitum. All experiments were approved by the regional ethical committee for animal experimentation.
Cytokine stimulation
Cytokines were given through intrascrotal injections. The doses were 0.5 µg hTNF- 4–6 h prior to experiments or 0.2 µg hTNF- for 7 sequential days before intravital microscopy.
Experimental procedure
Mice were anesthetized by an i.p. injection of ketamine (Ketalar®) and xylazine (Narcoxyl vet®). Catheters were placed in the right femoral artery and in the left jugular vein. Blood pressure was monitored and ranged between 60 and 100 mmHg. Temperature was kept at 37°C with a heating pad and an infrared heat lamp. The exposed tissue was superfused with a buffered saline solution at pH 7.4 and 37°C. Blood samples (10 µl) were taken through the femoral catheter and later analyzed for systemic leukocyte count (WBC) in a Bürker chamber. The cremaster muscle was prepared as described previously. Briefly, an incision of the skin and fascia ventrally on the right scrotum was made, and the tissue was retracted to expose the cremaster muscle. The muscle was incised and spread on a transparent pedestal to allow transillumination. The testis was then pinned to the side. Abdominal lymph nodes were visualized by a midline incision, and the intestines were retracted to expose the dorsal peritoneum. Lymph nodes were identified to the left of the aorta, and the peritoneum was incised to expose the lymphatic microcirculation. Microscopic observations were made using a Leitz Orthoplan microscope with water immersion objectives (Leitz SW55/0.80, Leitz SW25/0.60, Nikon WI10/0.3, or Leitz 3,5/0.10). Observation was performed by light microscopy or epi-illumination fluorescence microscopy [Leitz Ploem-o-pac (filter block M2 or I3)]. In some experiments, labeling of circulating leukocytes was achieved by an i.v. injection of rhodamine 6G (0.3 mg/ml, 0.67 mg/kg). Images were televised and recorded on videotape using Panasonic WV-1550 or VNC-703CCD video cameras.
Analysis of in vivo experiments
Vessel diameter (D) was measured from the microscopic image, and radius (r) was calculated as D/2. Flow (q) and wall shear rate [WSR (w)] were calculated from r and by measuring the velocity of i.v.-injected fluorescent beads (2.0 µm in diameter). The fastest beads were approximated to represent axial flow velocity, which was divided by the empiric factor 2 to achieve mean flow velocity (vm). Flow was calculated according to q = vm
r2 and WSR using the formula 
w= 2.12 x 4 q/r3. Rolling leukocyte flux (RLF) was determined as the number of leukocytes passing a reference line perpendicular to blood flow. RLF fraction was calculated as RLF divided by total leukocyte flux estimated from flow and WBC. Leukocytes were regarded as captured if they initiated endothelial contact within the field of vision and had not previously been in contact with the endothelium. Captured leukocytes were regarded as potentially secondary if they attached in contact with or 0–30 µm downstream of a previously rolling cell. To control for leukocytes randomly captured by primary capture downstream of cells, the number of potentially secondary captured cells was subtracted with the number of cells that were captured 0–30 µm upstream of rolling or adherent leukocytes. The outcome was regarded as secondary capture. All other captured leukocytes were regarded as primary. In cremaster muscle venules larger than 70 µm in diameter, leukocytes could be observed only on the side of the vessel facing the objective. In such vessels, parameters of leukocyte capture and rolling were adjusted by multiplying these parameters with a factor 2.
In vivo detection of CAMs
Fluorescent beads coated with protein G were coupled to antibodies or L-selectin hIgG chimera by incubating 50 µl bead solution with 50 µg protein. Beads were washed, resuspended in 1.0 ml HBSS, and sonicated. Mice were given 100 µl of the bead solution through a femoral artery catheter, and binding of beads to the endothelium was analyzed after 5 min. As beads coated with 4RA10 or a mouse L-selectin chimera bound a minority of adherent leukocytes, the number of injected beads was not sufficient to bind all expressed CAMs in cremaster venules. However, a less-than-saturating number of beads was necessary for clarity when using intravital fluorescence microscopy.
Statistical analysis
Data represent mean ± SEM of measurements obtained in the indicated number of experiments. Statistical comparison between animal groups was performed using the Student’s t-test or the Mann Whitney rank sum test, whereas comparison before and after antibody blockage of CAMs was performed using paired t-test or Wilcoxon signed rank test for paired samples. Nonparametric tests were used in the cases where samples were not normally distributed. Statistical significance was set at P < 0.05.
RESULTS
To elucidate the mechanisms by which interactions between L-selectin and PSGL-1 contribute to leukocyte rolling in acute inflammation and whether leukocyte fragments play significant roles, I studied mouse cremaster muscle venules after induction of rolling using various strategies. It is well established that rolling along venular endothelium in the untreated, acutely exteriorized cremaster muscle is entirely dependent on the endothelial P-selectin interacting with leukocytic PSGL-1 and that blockage of L-selectin in this situation decreases rolling by limiting leukocyte capture to the rolling cell pool [32
, 33
]. However, rolling in the absence of functional P-selectin, which is inhibited by blockage of L-selectin, has been reported in untreated cremaster venules >60–120 min after exteriorization and in mice treated with TNF- for 4–6 h [8
, 31
, 34
]. In my experiments, up-regulation of rolling, independent of P-selectin, usually required more than 60 min of exteriorization (data not shown), and such rolling could be induced in virtually all venules after trauma and not only in small areas of the vessels, as previously suggested [31
]. However, the time required after surgery to induce P-selectin-independent rolling differed between mice and could be up to 6 h. Therefore, I initially chose to use intravital microscopy on cremaster muscle venules >120 min after exteriorization or after 4–6 h treatment with TNF- to examine L-selectin-dependent interactions between leukocytes and venular endothelium. WBC, vessel diameter, and WSR in different experimental settings in WT and selectin-deficient mice are shown in Table 1
.
|
View this table: [in a new window] |
Table 1. Basic Parameters in Experimental Mice
|
was 16.6 ± 4.6%, similar to what has been shown previously. i.v. injection of a function-blocking antibody against P-selectin decreased rolling fractions to 3.6 ± 0.62 or 8.1 ± 1.4% in exteriorized and TNF--treated venules, respectively (P<0.01; Fig. 1 A
and 1B
). These fractions were further reduced by blockage of L-selectin or PSGL-1, indicating that residual rolling after functional blockage of P-selectin in these models is dependent on interactions involving these leukocytic CAMs (P<0.01; Fig. 2 A
and 2B
). Blockage of leukocyte rolling after function inhibition of P-selectin could also be accomplished by antibody blockage of E-selectin with the mAb 9A9 (P<0.01; Fig. 2A
and 2B
). As blockage of E-selectin inhibited rolling more potently than blockage of L-selectin or PSGL-1, it is clear that E-selectin mediates rolling of the same population of leukocytes that in parallel experiments requires L-selectin and PSGL-1 to interact with the endothelium. These findings were also apparent in P-sel–/– mice, in which rolling >120 min after exteriorization or after 4–6 h of TNF- was evident and inhibited by blockage of E-selectin, L-selectin, or PSGL-1 (P<0.01; Fig. 2C
and 2D
). In parallel, rolling was lower after P-selectin blockage in mice deficient in E-selectin or L-selectin compared with WT mice (P<0.01; Fig. 2E
and 2F
), further demonstrating that in the absence of functional P-selectin, E-selectin and L-selectin are involved in leukocyte rolling along the endothelium in inflamed venules.
 View larger version (13K): [in a new window] |
Figure 1. Leukocyte rolling in response to inhibition of P-selectin. Mice were exposed to >120 min of exteriorization of the cremaster muscle (A) or 4–6 h pretreatment with an intrascrotal injection of 0.5 µg TNF- (B). Blockage of P-selectin function was accomplished by an i.v. injection of a saturating dose of mAb RB40.34. *, P < 0.05 compared to baseline.
|
 View larger version (22K): [in a new window] |
Figure 2. Leukocyte rolling in WT mice after blockage of P-selectin (A and B), in P-sel–/– mice (C and D), or in WT, L-sel–/–, or E-sel–/– mice (E and F). Additional blockage of CAMs with saturating doses of antibodies was given as indicated. (A, C, and E) Mice were exposed to >120 min of exteriorization of the cremaster muscle. (B, D, and F) Interactions between leukocytes and endothelium were observed 4–6 h after an intrascrotal injection of 0.5 µg TNF-. Panels demonstrate that rolling of leukocytes after blockage of P-selectin is dependent on all three, L-selectin, E-selectin, and PSGL-1, and E-selectin has the most prominent role. *, P < 0.05 compared to baseline.
|
-treated venules, respectively. Interestingly, leukocyte capture from the free flow increased after blockage of P-selectin [413±66 (exteriorized, P<0.05) and 291±76 (TNF--treated, P<0.01) cells/mm2/min; Fig. 3A
], despite the fact that rolling flux decreased (Fig. 1A
and 1B)
. Thus, the turnover of leukocyte capture to and detachment from endothelium increases after P-selectin blockage, which can be explained by the decreased density of functional CAMs mediating rolling along the endothelium. Interestingly, before P-selectin blockage, 35 ± 7.3% (exteriorized) and 31 ± 7.3% (TNF--treated) of observed capture were determined as being primary, whereas 65 ± 7.3% and 69 ± 7.3% were appreciated as secondary capture (Fig. 3B)
. In contrast, secondary capture made up for 95 ± 2.6% (exteriorized) and 100 ± 0.3% (TNF-) of total capture after antibody-blockage of P-selectin (P<0.01 compared with before blockage; Fig. 3C
). Corresponding results were obtained in P-sel–/– mice in which secondary capture dominated (98±1.1% and 100±0% of total capture in exteriorized and TNF--treated mice, respectively, P<0.01). In addition, in mice deficient of functional P-selectin, capture from the free flow was almost eliminated by blockage of secondary capture by antibody inhibition of L-selectin or PSGL-1 (Fig. 3D)
. Thus, secondary capture mediated by rolling or adherent leukocytes increases capture to venular endothelium after blockage of P-selectin, and the effects of inhibition of L-selectin or PSGL-1 on rolling flux are at least in part mediated by reductions in capture. As rolling after blockage of P-selectin is virtually abolished by blockage of E-selectin, this CAM is responsible for rolling along the endothelium subsequent to capture mediated by L-selectin/PSGL-1. This notion is supported by data comparing rolling velocity in different situations. It has been shown previously that L-selectin is the selectin that mediates rolling with the highest velocity, whereas E-selectin and P-selectin mediate rolling of low and intermediate velocities, respectively [35
36
37
]. Here, rolling velocity in lymph node HEV in WT mice after blockage of P-selectin, in which capture and rolling are entirely dependent on L-selectin (ref. [38
] and data not shown), was on average 95 ± 6.8 µm/s, whereas rolling velocity in venules in E-sel–/– mice <20 min after exteriorization of the cremaster muscle, in which rolling along the endothelium is entirely dependent on P-selectin, was on average 38 ± 2.1 µm/s (P<0.01 compared with HEV). In contrast, rolling velocity in WT mice >120 min after exteriorization or after 4–6 h of treatment with TNF- following antibody blockage of P-selectin was 14 ± 0.92 and 12 ± 1.4 µm/s, respectively (P<0.01 compared with HEV and acute exteriorization; Fig. 4A
4B
4C
4D
). Hence, rolling velocities in the two latter situations resemble velocities previously demonstrated for E-selectin-dependent rolling [35
]. These data demonstrate that in situations of L-selectin-dependent leukocyte rolling in inflammation, the dynamics of rolling along endothelium are distinct from those typically mediated by L-selectin. Instead, as previously hypothesized [31
], the data indicate that rolling along endothelium in cremaster muscle venules after blockage of P-selectin is mediated by E-selectin. For videos of rolling at baseline, after blockage of P-selectin and additional inhibition of L-selectin, see online, supplemental material: video1a.mov, video1b.mov, and video1c.mov.
 View larger version (21K): [in a new window] |
Figure 3. Leukocyte capture from free flow in mouse cremaster muscle venules. (A) Total capture per mm2 and minute in WT mice before and after blockage of P-selectin with mAb RB40.34 is displayed together with data on total capture from P-sel–/– mice. (B and C) The contributions of primary and secondary capture in WT mice before (B) and after (C) blockage of P-selectin are demonstrated as percentage of total capture. (D) Pooled data for capture before and after blockage of L-selectin or PSGL-1 in WT mice treated with mAb RB40.34 and P-sel–/– mice exposed to >120 min exteriorization or 4–6 h of TNF- are shown. Left and right panels of bars show capture before and after blockage of of L-selectin or PSGL-1. The figure shows that total and secondary capture is strongly restricted after blockage of L-selectin/PSGL-1, whereas primary capture is not decreased (see also text). *, Significant; n.s., nonsignificant difference compared with WT mice.
|
 View larger version (18K): [in a new window] |
Figure 4. Histograms of leukocyte rolling velocity in different situations. (A) Rolling velocity in abdominal lymph nodes (PLN) after blockage of P-selectin in which rolling is totally dependent on L-selectin is displayed. (B) Rolling velocity in acutely exteriorized (<20 min) mouse cremaster muscle venules in E-sel–/– mice in which rolling is entirely P-selectin-dependent. (C and D) The respective rolling velocity histograms for venules exposed to >120 min of exteriorization or 4–6 h treatment with TNF- following blockage of P-selectin. Mv, Mean rolling velocity ± SEM in each respective situation.
|
-treated venules, respectively). Interestingly, primary capture was not lower in L-sel–/– mice after blockage of P-selectin compared with WT mice (L-sel–/– mice: 59±19 and 24±15 cells/mm2 in exteriorized and TNF--treated venules; P=0.197, and P=0.516 vs. WT mice). Moreover, in WT mice, the absolute number of primary-captured leukocytes after blockage of P-selectin was not different in mice in which additional inhibition of L-selectin or PSGL-1 was performed (three vs. three cells before and after blockage of L-selectin; three vs. four cells before and after blockage of PSGL-1 in exteriorized and TNF--treated mice combined). Hence, the data may suggest that L-selectin or leukocyte fragments are not involved in primary capture in inflamed venules. The data also indicate that a theoretically possible release of leukocyte fragments from endothelium in response to VCAM-1 is not involved in the lack of effect on rolling seen after blockage of L-selectin in mice receiving previous blockage of P-selectin, E-selectin, and VCAM-1.
No endothelial L-selectin ligand activity mediates rolling after several treatment regimens
Having demonstrated that rolling of leukocytes, which was previously demonstrated to be inhibited by blockage of L-selectin, is mediated by E-selectin and L-selectin/PSGL-1-dependent secondary capture, further attempts at detecting a ligand for L-selectin on the endothelium in inflamed venules were conducted. In mice treated with antibodies against P- and E-selectin, leukocytes were occasionally observed to interact transiently with the vessel wall. These interactions often took place on the inner lateral sides of the venules and often appeared as being mediated by interactions between leukocytes and endothelial cells. However, using fluorescence microscopy after administration of rhodamine 6G, which labels all circulating leukocytes, most of these interactions were determined to occur between firmly arrested leukocytes and leukocytes that were initially in free flow. Nonetheless, limited interactions between leukocytes and endothelium still took place (Table 2
). To determine what molecules mediate residual rolling after blockage of P- and E-selectin, I observed venules for extended periods of time after additional administration of an antibody blocking the function of VCAM-1, another endothelial molecule capable of mediating rolling in vivo. As seen in Table 2
, blockage of VCAM-1 eliminated residual rolling in venules exteriorized for >120 min completely, whereas three rolling leukocytes during 107 min of observation, representing 0.13% of rolling before CAM inhibition, were observed in mice treated with TNF-. Interestingly, similar numbers of rolling cells (2/53 min) were observed after additional blockage of L-selectin, making it unlikely that rolling after blockage of P-selectin, E-selectin, and VCAM-1 was mediated by L-selectin interacting with an endothelial ligand.
|
View this table: [in a new window] |
Table 2. Rolling Flux in Response to Intervention
|
per day for 7 sequential days before the cremaster muscle was prepared for intravital microscopy. This regimen would give endothelial cells plenty of time to express an L-selectin ligand and leukocytes opportunity to deposit functional epitopes of PSGL-1 or leukocyte fragments on the surface of the endothelium, as has been postulated to be mechanisms by which an L-selectin ligand is introduced on the endothelial surface [31
]. As seen in Table 2
, rolling was lower after 7 days compared with after 4–6 h of treatment with TNF-, indicating a somewhat lower expression of the selectins. However, firm adhesion was in fact higher after 7 days, also indicating a strong inflammatory activity after long-term treatment (1.3±0.16x103 vs. 2.3±0.43x103cells/mm2 after 4–6 h and 7 days, respectively, P<0.05). Combined inhibition of P-selectin, E-selectin, and VCAM-1 almost eliminated rolling, and this was not decreased further by blockage of L-selectin. Moreover, after blockage of P-selectin alone, primary capture was virtually absent (2.5±2.5 cells/mm2), indicating that similar to mice exposed to exteriorization or TNF- for 4–6 h, leukocyte fragments carrying L-selectin and PSGL-1 are not involved in capture or rolling. Importantly, there was no difference in the expression of L-selectin on leukocytes after a 7-day treatment with TNF- compared with untreated mice (data not shown), indicating that the absence of L-selectin-dependent capture and rolling was not influenced by shedding of L-selectin from leukocytes.
As P- and E-selectin have previously been reported to have L-selectin-binding capacities, I investigated whether these CAMs could mediate interactions with L-selectin. In WT mice, I studied rolling immediately after exteriorization (<20 min) to avoid expression of E-selectin on the endothelium. In this situation, rolling is totally dependent on P-selectin. Under these conditions, all rolling was abolished by blockage of PSGL-1 (zero rolling cells, 36 venules, 36 min), indicating that PSGL-1 is the only molecule capable of binding P-selectin and that P-selectin and L-selectin are unable to mediate rolling interactions. This is in accordance with previous data [33
]. In mice exposed to >120 min of exteriorization, in which rolling after blockage of P-selectin was evident, indicating endothelial expression of E-selectin, rolling after additional blockage of PSGL-1 was apparent and not reduced further by blockage of L-selectin (rolling fraction: 0.41±0.14% and 0.53±0.19% before and after blockage of L-selectin, respectively, P=0.497). These data indicate that the PSGL-1-binding epitope on L-selectin does not mediate rolling by binding E-selectin or any other endothelial molecule. Moreover, there was no difference in rolling following blockage of P-selectin and L-selectin in WT mice compared with after blockage of P-selectin in L-sel–/– mice [rolling flux fraction: 1.1±0.88% and 0.91±0.48% after >120 min exteriorization (P=0.372); 1.4±0.72% and 1.5±0.45% after TNF- (P=0.117)], indicating that there is no alternative binding site on L-selectin to which E-selectin can bind and mediate rolling in mice in vivo. In addition, as there was no additional reduction of rolling by blockage of L-selectin in mice pretreated with mAb against P-selectin and PSGL-1, the data also indicate that L-selectin and PSGL-1 influence rolling by acting through the same mechanism, secondary capture.
In vivo detection of CAMs on venular endothelium
As I could not detect any L-selectin ligand activity capable of mediating interactions between leukocytes and endothelium in inflamed venules, I sought to investigate whether such activity could be detected by in vivo staining. To do this, I incubated fluorescent beads coated with protein G with a hIgG–mouse L-selectin chimera, mAb 4RA10 against mouse PSGL-1, or an irrelevant control antibody and injected them into the femoral artery in mice treated with TNF- for 4–6 h. Beads coated with an irrelevant control mAb displayed low binding to venular endothelium (Fig. 5 A
and B
). In contrast, beads coated with mAb 4RA10 readily adhered, and the numbers of adherent beads were higher than in mice receiving control beads (P<0.01). Interestingly, by switching between light and fluorescence microscopy, it was clear that almost all bound beads coated with 4RA10 adhered to leukocytes and not to the endothelium, and the number of beads that bound directly to the endothelium was not higher than in mice receiving control beads (P=0.52). Beads coated with the L-selectin chimera also displayed efficient binding to some, but not all, leukocytes and little binding to the endothelium (P=0.26 compared with control beads; online supplemental material: video2a.mov and video2b.mov). Although there was some binding to the endothelium of beads coated with 4RA10 or the L-selectin chimera that could possibly represent the presence of L-selectin ligands, the similar binding of control beads suggests that this limited accumulation of beads on the endothelium was unspecific. Thus, the data may suggest that there is no L-selectin ligand present on inflamed cremaster venule endothelium. Noteworthy, it was difficult to find an appropriate negative control in the experiments using beads coated with the L-selectin chimera, and such control was thus not included; however, as I sought to identify whether the beads could bind inflamed endothelium, the positive control is the one that is crucial. Here, binding of L-selectin-coated and 4RA10-coated beads to adherent leukocytes serves as such control.
 View larger version (14K): [in a new window] |
Figure 5. Adhesion of fluorescent beads coated with mAb 4RA10 against PSGL-1, an L-selectin IgG chimera, or an irrelevant rat IgG in cremaster venules. The total number of adherent beads per mm2 vessel area is displayed in A. (B) The number of beads adhering directly to the endothelium and not to leukocytes. Note the difference in scale on the y-axis between A and B, demonstrating that a majority of beads coated with mAb 4RA10, the L-selectin IgG chimera, bound leukocytes and not the endothelium. *, Significant difference; n.s., nonsignificant difference compared with control.
|
for 4–6 h, beads coated with 10E9.6 readily bound endothelium. Little binding was seen with control beads. Thus, E-selectin is expressed on venular endothelium in situations where blockage of L-selectin has effects on leukocyte rolling.
 View larger version (83K): [in a new window] |
Figure 6. Expression of E-selectin on venular endothelium. Panels demonstrate binding of beads coated with control (Ctrl) mAb (A) or mAb 10E9.6 against mouse E-selectin in cremaster venules (B–D) in different experimental settings. (Upper row) Light microscopy image; (lower row) the same image in fluorescence. Note that B and C are taken from the same experiment at different time-points. Original bar = 200 µm.
|
L-selectin is of key importance in the recruitment of leukocytes to inflammatory sites. In the present study, I demonstrate that the contribution of L-selectin to leukocyte rolling in inflamed venules is mediated exclusively by L-selectin/PSGL-1-dependent secondary capture, a mechanism that augments capture of leukocytes to the endothelium on which other CAMs such as P-selectin, E-selectin, or VCAM-1 can mediate rolling interactions. The possible presence of an L-selectin ligand activity expressed by the endothelium or by leukocyte fragments or uropods from transmigrating leukocytes was studied carefully after several treatment regimens. Consistently, the data show that there is no endothelial or leukocyte-derived ligand activity for L-selectin in mouse cremaster muscle venules after several stimuli.
The view that there is an endothelial ligand for L-selectin on inflamed endothelium originated in studies from the early 1990s in which rolling in inflamed venules and recruitment of inflammatory cells were reduced by antibody blockage of this CAM [2
, 6
7
8
]. However, although several molecules were hypothesized as being putative endothelial L-selectin ligands, the identity of the ligand for L-selectin in inflammation was for long elusive. In the mid-1990s, secondary capture emerged from experiments carried out in vitro as a possible mechanism by which L-selectin could enhance rolling in inflammation. However, the importance of secondary capture in leukocyte recruitment in vivo was initially put in doubt by a study that failed to detect the role of capture from free flow [39
]. This conclusion was later challenged by data from our own laboratory, which demonstrated that L-selectin/PSGL-1-dependent, secondary capture is in fact an important feature in the inflammatory recruitment of leukocytes [29
]. Subsequently, the study by Sperandio et al. [31
] from 2003 identified PSGL-1 as being responsible for most L-selectin-dependent rolling in venules. Although the authors could demonstrate that a majority of the effects of PSGL-1 was mediated by interactions between leukocytes, they hypothesized that PSGL-1 or functional fragments of this molecule were deposited on the endothelium by rolling and adherent leukocytes and that such leukocyte-derived material could subsequently mediate functionally significant capture and rolling. Such deposition of nonendothelial cell material on luminal endothelium might also alter endothelial function in several other ways. Here, in experiments using light and fluorescence microscopy, in which I specifically studied the cellular substrate for rolling, no capture or rolling on the endothelium mediated by L-selectin was detected. The present study thus indicates that the paper by Sperandio et al. [31
], as well as several earlier studies, underestimated the role for secondary capture and that secondary capture makes up for almost 100% of P-selectin-independent capture. It is conceivable that this would have been better appreciated if they had used fluorescence microscopy. Following inhibition of P-selectin, primary capture was virtually abolished, and in the rare cases where primary capture was observed, there was no difference in primary capture after blockage of P-selectin between WT and L-sel–/– mice or in WT mice before and after additional blockage of L-selectin. Importantly, I could find no evidence for any endothelial L-selectin ligand activity in inflamed cremaster muscle venules, even in such long-term inflammation as 7-day treatment with TNF-, thus indicating that no fragments of leukocytes are deposited on the endothelium even after longer periods of inflammation. What’s more, in experiments using fluorescent beads coated with an L-selectin–IgG chimera or a function-blocking mAb against PSGL-1, binding of beads occurred on adherent, rolling, or freely flowing leukocytes and not on the endothelial cells. Collectively, these data indicate that inflamed endothelium in the cremaster muscle does not express L-selectin ligands and that leukocyte fragments do not mediate capture or rolling in inflammation.
The concept of secondary capture being responsible for L-selectin-dependent rolling finds indirect support in previous studies. For instance, deficiency in ICAM-1, which is a major receptor mediating leukocyte firm adhesion, has been shown to reduce trauma-induced, L-selectin-dependent rolling in venules [19
]. It is likely that the absence of L-selectin-dependent rolling in the mentioned study was paralleled by an inhibition of leukocyte firm adhesion, which would then reduce secondary capture and subsequent rolling on E-selectin. Similarly, a decrease in leukocyte firm adhesion would also explain the decrease in L-selectin-dependent rolling seen in the study by Sperandio et al. [31
] in experiments where blockage of rolling was performed at the time of exteriorization or injection of TNF- and in which the number of firmly adherent cells was likely reduced at the time of observation a few hours later.
The fraction of all captured leukocytes that initiated rolling by secondary capture is higher in this study compared with what myself and others have reported previously [29 ]. We believe that this is influenced by several factors. For instance, capture varies depending on the size of the vessels, fluid shear, the number of rolling leukocytes, and also likely, the number of leukocytes in peripheral blood. Moreover, in my previous study, we did not focus on secondary capture of freely flowing cells to firmly adherent cells on the endothelium. In the present set of experiments, such capture was clearly visualized, and this likely increases the fraction of cells captured by secondary capture.
The data presented in this study clarify the mechanisms that mediate leukocyte rolling in inflammation. It is clear that P-selectin is the principal molecule responsible for rolling along inflamed endothelium; however, rolling velocity may not be optimal for subsequent firm arrest and extravasation. E-selectin mediates slow rolling, which is initiated by primary capture to P-selectin or secondary capture mediated by L-selectin and PSGL-1. However, the capacity for E-selectin to mediate capture from the free flow by interacting with PSGL-1 appears to be minimal. The data on E-selectin-mediated capture stand in sharp contrast to previous data indicating an important role for capture mediated by E-selectin/PSGL-1 [40 ]. Although there is no apparent explanation for the discrepancy between the data on interactions between E-selectin and PSGL-1 presented in the former study and those shown here, it is possible that at least some of the capture seen after blockage of P-selectin in the former study that was interpreted as primary capture was in fact secondary capture mediated by L-selectin/PSGL-1. More clearly, the data in the present study strongly indicate that L-selectin has no endothelial ligand in inflamed cremaster venules and that this molecule influences rolling only by mediating secondary capture. However, L-selectin may still influence later phases of recruitment, including migration in extravascular tissues [41 ]. As neutrophils and monocytes carry ligands for P- and E-selectin [40 , 42 ], these cell types are efficiently recruited in acute inflammation, where the endothelial selectins are abundantly expressed. However, as rolling mediated by L-selectin occurs exclusively on rolling or adherent leukocytes, it is unlikely that cell types expressing L-selectin and not functional ligands for P- or E-selectin can be recruited to the endothelium in the acute situation. For instance, lymphocytes express functional PSGL-1 only in response to cell activation, after which they can invade inflamed tissues and attack antigens to which they are specifically directed [43 ]. If inflamed endothelium would regularly express ligands for L-selectin, this would provide an adhesive pathway for naïve, nonactivated lymphocytes, which in the presence of the array of chemokines presented in inflammation, could possibly recruit these cells. Clearly, such recruitment of naïve lymphocytes is in most cases unphysiological. Moreover, the role for secondary capture in E-selectin-dependent rolling demonstrated here also helps to clarify the importance of the cutaneous lymphocyte antigen (CLA) in the homing of memory T cells to the skin [44 ]. It has been shown previously that CLA is the epitope carried by PSGL-1, which mediates secondary capture by binding L-selectin [28 , 45 ]. According to my data, CLA expressed on PSGL-1 would enhance recruitment of leukocytes by mediating secondary capture to E-selectin, which is expressed constitutively in skin microvessels [46 ]. In contrast, lymphocytes lacking the CLA epitope would not be recruited as efficiently because of a defect in the capture step.
The absence of an endothelial ligand for L-selectin in cremaster venules demonstrates that expression of most L-selectin ligands on the endothelium likely requires a specific endothelial phenotype. A lymphoid endothelial phenotype is present in lymphoid organs [47
, 48
] and under certain circumstances, in chronically inflamed venules [49
]. Interestingly, recent data indicate that endothelium in some locations may express PSGL-1 [50
], and in vitro experiments have demonstrated that such expression can be induced on HUVEC also in response to stimulation with TNF- [51
, 52
]. However, the current data indicate that PSGL-1 is not expressed on venular endothelium in the cremaster muscle in response to this cytokine. Importantly, previous data also suggest that when transformation to lymphoid endothelium occurs or when PSGL-1 is expressed on the endothelium, it may be paralleled by development of a tissue rich in lymphocytes and APC in which lymphocyte activation may take place [49
, 50
]. Apparently, expression of L-selectin ligands on the endothelium may be restricted to lymphoid organs and lymphocyte-driven inflammation.
Taken together, I present data that indicate that L-selectin influences rolling in inflamed cremaster venules solely by mediating leukocyte secondary capture and that there is no ligand for L-selectin present on acutely inflamed endothelium. These data clarify previous data about the role of L-selectin in inflammation and demonstrate that leukocyte fragments do not influence capture or rolling of leukocytes in acute inflammation.
ACKNOWLEDGEMENTS
This work was supported by the Swedish Heart and Lung Foundation, the Swedish Research Council (Projects 2003-6311 and 2005-6416), the Swedish Society of Medicine, the Swedish Society for Medical Research, the Osterman Fund, the Tore Nilson Foundation, the Lars Hierta Memorial Fund, the AFA Health Fund, AstraZeneca, and Karolinska Institutet. I thank Dietmar Vestweber for supplying reagents and for providing useful comments on the manuscript, Barry Wolitzky for supplying mAb 9A9, and Thomas Graf for supplying lysM-enhanced green fluorescent protein mice.
Received November 10, 2007; revised February 21, 2008; accepted February 21, 2008.
REFERENCES
-activated vascular endothelium under flow in vitro J. Immunol. 157,326-335[Abstract](1,3) fucosyltransferase-IV J. Exp. Med. 198,1301-1312(1,3)-fucosyltransferase-IV and FucT-VII during leukocyte rolling in dermal microvessels Immunity 12,665-676[CrossRef][Medline]
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
