Published online before print March 21, 2006
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Inflammation Repair Group, MRC Centre for Inflammation Research, Queen’s Medical Research Institute, Edinburgh, United Kingdom
1Correspondence: Inflammation Repair Group, MRC Centre for Inflammation Research, C2.05 Queen’s Medical Research Institute, Little France Crescent, Edinburgh EH16 4TJ, UK. E-mail: simon.brown{at}ed.ac.uk
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
5ß1 integrin and the engulfment of apoptotic Jurkat T lymphocytes via a fibronectin (Fn) "bridge." Although Fn and serum served as an opsonin for ß1 integrin-dependent phagocytosis of apoptotic leukemic T cells, they failed to do so for neutrophils. Given the complexities and inherent variability of working with primary cells, we have refined our model to show that ligation of CD31 on THP-1 macrophages also regulates ß1 integrin-dependent phagocytosis of Fn-coated Latex beads. Thus, selective "tethering" of apoptotic leukocytes by phagocyte CD31 not only discriminates dying from viable cells but also selectively activates phagocyte integrins for the engulfment of apoptotic cells.
Key Words: acute-phase reactants • macrophages • adhesion molecules
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
We recently described a novel discriminatory role for interactions between "target" leukocytes and macrophages involving homophilic ligation of CD31, also known as platelet-endothelial cell adhesion molecule-1 [7 ]. By using a flow chamber system, we examined the "tethering" by macrophage or CD31 monolayers of viable or apoptotic leukocytes bearing or lacking CD31. Although viable and apoptotic leukocytes were bound under low shear at 20°C (but where apoptotic cells were not engulfed), there was active detachment of viable leukocytes at 37°C. This CD31-directed detachment was disabled completely in apoptotic leukocytes with the result that apoptotic cells were engulfed. The use of shear was not to mimic any specific physiological process as such but rather to provide a mechanism by which to remove unbound cells. Antibody blockade experiments confirmed that CD31-mediated tethering promoted selective phagocytosis of apoptotic cells by mechanisms we did not define. Thus, CD31 expressed by live target cells effected detachment and motility, and CD31 expressed by macrophages promoted the engulfment of apoptotic cells. These studies focus on the latter and how CD31 signals within phagocytes to promote the engulfment of apoptotic cells.
CD31 is associated with cell motility [7
8
9
] and the transmigration of leukocytes across endothelial cell junctions [10
] through the perivascular basement membrane [11
, 12
]. In transmigration, homophilic ligation of CD31 results in the up-regulation and activation of integrins on leukocytes, where in vivo evidence strongly implicates the importance of the laminin 6ß1 integrin receptor [12
]. In vitro, antibody ligation of leukocyte CD31 has been shown to lead to the activation of ß1, ß2, and ß3 integrins in the presence of their ligands [13
], a process thought to be dependent on activation of the small GTPase Rap1 [14
]. More recently, induced oligomerization of CD31, independent of homophilic or antibody binding, was found to promote 5ß1-dependent adhesion of transfected epithelial cells to immobilized fibronectin (Fn), whereas dimerization led to homophilic cell interactions [15
]. In this report, we describe new data indicating that CD31 homophilic interactions between phagocyte and target cells can lead to the activation of phagocyte 5ß1 integrins and the engulfment of apoptotic Jurkat T lymphocytes opsonized with plasma Fn. By extending our observations to the ingestion of Fn-coated beads, we have established a more robust model of CD31-dependent engulfment, which in the future, will allow a more detailed analysis as to the mechanism by which CD31 regulates integrins to promote engulfment.
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Leukocyte isolation and culture
Human polymorphonuclear neutrophils (PMNs) and monocytes were isolated from freshly drawn venous blood and separated following citration, dextran sedimentation, and discontinuous PercollTM density gradient centrifugation as described previously [16
]. PMNs (typically maintained at 5x106 cells/mL in Teflon-lined receptacles) were cultured overnight at 37°C in Iscove’s modified Dulbecco’s medium (IMDM) to allow spontaneous apoptosis, further supplemented with 10% autologous platelet-rich plasma-derived serum (PRPDS) when required. Human monocyte-derived macrophages (HMDMs) were obtained by culturing adherent monocytes to plastic in IMDM with 10% autologous PRPDS for 5 days. THP-1, a human premyelomonocytic leukemic cell line, and human leukemic JKT lines were maintained in RPMI 1640 supplemented with 10% fetal bovine serum, 2 mM L-glutamine, 100 U/ml penicillin, and 100 µg/ml streptomycin. THP-1 were differentiated to macrophages with 12-O-tetradecanoylphorbol 13-acetate (10 ng/ml, 72 h). In some instances, apoptotic leukocytes were washed and resuspended in nonautologous serum, which represents pooled PRPDS from a minimum of 10 donors.
Induction of apoptosis
JKTs were induced to undergo apoptosis following Fas ligation (CH-11, 100 ng/ml, 24 h) in the absence of serum, whereas PMNs underwent spontaneous apoptosis when cultured in the presence or absence of 10% autologous PRPDS. Apoptosis and necrosis were assessed by flow cytometry using annexin-V-FITC and propidium iodide (PI; 10 µg/ml).
Phagocytosis of apoptotic leukocytes
PMNs and JKT lines were labeled with ((4-chloromethyl)benzoyl)aminotetra-methylrhodamine (Molecular Probes, Eugene, OR) in serum-free IMDM or Dulbecco’s modified Eagle’s medium (DMEM), respectively, at 2 µg/20 x 106 cells/ml for 20 min at 37°C before inducing apoptosis. Apoptotic cells, washed but unsorted, were incubated with THP-1 or HMDM macrophages at 37°C in the presence or absence of 10% PRPDS (autologous for PMN) for periods of time defined elsewhere. For antibody-blockade studies with THP-1, the ß1 integrin functional-blocking mAb P5D2 and P4C10 or a control IgG1 mAb were added at 10 µg/ml to the macrophages at room temperature for 15 min prior to the addition of apoptotic cell feed. For HMDMs, the mAb 23C6, LM609, TS1/18, P4C10, and P5D2, all at 100 µg/mL, were added to macrophages at room temperature for 15 min before washing and then adding apoptotic cell feed. Apoptotic cells induced in the absence of serum may also have been preincubated with Fn at 100 µg/ml for 15 min prior to their unwashed addition to macrophages. Unengulfed cells were removed, and adherent macrophages were collected for flow cytometric analysis (Becton Dickinson FACSCalibur, Becton Dickinson, San Jose, CA) as described previously [7
, 17
]. Alternatively, apoptotic-labeled PMNs were incubated with COS-7 cells in serum-free DMEM and processed as above.
Phagocytosis of LatexTM beads or DynabeadsTM
Fn-coated Latex beads were prepared with 3.0 µm polybead amino microspheres (Polysciences Inc.), modified with glutaraldehyde, and reacted with human plasma Fn as described [18
]. P2B1, 9G11, DF-T1, and DU-HL60-3 mAb were bound to 4.5 µm Dynabeads using the CELLectionTM pan mouse IgG kit, according to the manufacturer’s instructions (Dynal Biotech, Bromborough, UK). THP-1 macrophages were typically washed with DMEM (no serum), to which 
4 x 107 beads were added per 48-well assay. Beads were allowed to settle and bind for 30 min before gently removing all media and adding trypsin-EDTA (Fn-coated Latex beads) or DNase (Dynabeads) for a further 15 min to remove unengulfed, bound beads. Bead ingestion was scored by light microscopy and expressed as the number engulfed per 100 cells (phagocytic index). Antibody blockade or stimulation studies were as described for phagocytosis of apoptotic leukocytes.
Binding of proteoliposomes to HMDMs
Unsorted, apoptotic PMNs were resuspended in phosphate-buffered saline (PBS) and surface-biotinylated with NHS-LC-biotin (1 mg per 108 cells) [7
]. Biotinylated cells were then washed and resuspended in ice-cold lysis buffer (10 mM Hepes, pH 7.4/1 mM EDTA/1 mM benzamidine/1 mM 1,10-phenanthroline/1 mM phenylmethylsulfonyl fluoride/10 µM pepstatin/10 µM leupeptin/10 µM antipain) at 20 x 106 cells/ml before pelleting at 1600 g. The cell pellet was resuspended in lysis buffer before sonication and sequential centrifugation at 10,000 g (10 min) and 500,000 g (10 min). The microsomal-enriched pellet at 500,000 g was sonicated and washed twice before resuspending by sonication in DMEM. The proteoliposomes were then incubated with HMDMs at 20°C before extracting with 1% Triton X-100 followed by radioimmunoprecipitation assay (RIPA) buffer, and the RIPA extract was then analyzed following sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and Western blot for biotinylated proteins under nonreducing and reducing conditions. HMDMs may also have been preincubated for 10 min with a ß1 (P4C10), ß2(TS1/18), and vß3(LM609) integrin-blocking mAb at 100 µg/ml and 20°C.
Binding of Fn by target leukocytes
Viable and apoptotic JKTs or PMNs, cultured for 24 h in the presence or absence of 10% fetal calf serum (FCS) or 10% PRPDS, respectively, were washed in PBS before resuspending at 4 x 106 cells/ml in serum-free RPMI 1640 containing Fn-FITC (40 µg/ml) for 15 min at room temperature. Fn was conjugated with FITC as described in ref. [19
]. Alternatively, leukocytes were incubated with Fn-FITC as described above but with varying concentrations of unlabeled Fn as detailed elsewhere. Cells were not washed prior to flow cytometric analysis.
Microscopy
THP-1 macrophages were incubated with P5D2 mAb or an isotype control at 10 µg/ml for 20 min before fixing with 4% paraformaldehyde, all at room temperature. Fixation prior to P5D2 binding destroyed mAb recognition of the ß1 integrin. Immunodetection was with an Alexa488-conjugated, secondary polyclonal antibody (pAb; Molecular Probes). Epifluorescent images were captured on an inverted microscope (Axiovert S100, Carl Zeiss MicroImaging, Inc., Thornwood, NY), equipped with a CoolSNAP charged-coupled device camera and OpenLab 3.0 image analysis software (ImproVision, Coventry, UK).
Data Analysis
All summary data are presented as the mean ± SD, and statistical significance was determined by ANOVA using a Tukey or Scheffé post-hoc test. P 
0.05 was considered statistically significant.
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
vß3 (LM609) and ß1 (P4C10) but not by a blocking ß2 mAb (TS1/18). Given that the 140-kDa protein was subsequently identified as CD31 [7
], these results indicated that CD31 might function by influencing integrins in the recognition and clearance of apoptotic PMNs.
 View larger version (18K): [in a new window] |
Figure 1. CD31 associates with ß1 and ß3 integrins. (A) Proteopliposomes from apoptotic PMNs, surface biotinylated with NHS-LC-biotin, were incubated with HMDM in the presence or absence of Arg-Gly-Glu-Ser (RGES; 1 mM), RGD-Ser (RGDS; 1 mM), P4C10 (100 µg/ml), TS1/18 (100 µg/ml), or LM609 (100 µg/ml) for 30 min at 20°C before extracting with Triton X-100 followed by RIPA. The RIPA extracts were analyzed by Western blot for biotinylated proteins following separation on a 9% SDS-PAGE under reducing conditions, where we have previously identified the p140 band as CD31. All lanes were controlled for the number of HMDM extracted. (B) The effect of various integrin-blocking reagents at concentrations defined in Materials and Methods on the phagocytosis of aged PMNs by HMDMs, which were pretreated with blocking reagents for 15 min at room temperature before washing and coculturing with aged PMNs in the absence of serum for 30 min.
|
vß3 in the phagocytosis of apoptotic PMNs using the blocking mAb LM609 when compared with an IgG (P<0.1) but not an untreated control. These results underline the problematic nature of working with HMDMs and apoptotic PMNs, especially when attempting to dissect a mechanism specific for a particular receptor, e.g., CD31. The problem arises foremost in our view from the variability in primary macrophage preparations and the level of apoptosis obtained within aged PMN cultures (47.0±15.3; Supplemental Fig. 2). The problem of variability was also manifest in antibody-blockade studies of CD31.
CD31-directed macrophage clearance of apoptotic PMNs is obscured by apoptotic cell debris
Although we initially considered proteoliposomes as surrogates for apoptotic cells in identifying CD31, it is perhaps more appropriate to consider them as apoptotic cell debris. Nevertheless, antibody blockade studies with PMNs and the use of Jurkats, which failed to express CD31, confirmed a role for CD31 in the clearance of apoptotic cells [7
]. In addition, we have since noticed that the ability of antagonistic mAb to block the engulfment of apoptotic PMNs by HMDMs demonstrates a reliance on the level of apoptosis within the feed population, and inhibition is greatest when the level of apoptosis was low (Fig. 2A
and 2B
). Furthermore, this effect was independent of whether the engulfment assay was performed in the presence of serum or not. In contrast, phagocytic removal of lysed apoptotic PMNs was insensitive to blocking mAb directed against CD31 (Fig. 2C
and 2D)
. It therefore appeared that CD31 better discriminated early rather than late apoptotic cells, and in contrast to what one might have predicted, the presence of apoptotic cell debris appeared to mask CD31-dependent engulfment, presumably as a result of more efficient, serum-dependent clearance mechanisms (see also Supplemental Fig. 1).
 View larger version (27K): [in a new window] |
Figure 2. CD31-dependent engulfment of apoptotic PMNs is serum-independent and masked by apoptotic cell debris. (A) Combined data for the phagocytosis of apoptotic PMN preparations by HMDMs performed in the absence (n=43) or presence of autologous (n=34) or nonautologous (n=6) serum (PRPDS). (B) HMDMs, pretreated with 10 µg/mL of the CD31-blocking mAb Hec7.2, were incubated for 30 min with intact, washed PMNs in the absence or presence of PRPDS before assessing engulfment. The inhibitory effects of Hec7.2 on phagocytosis have been subgrouped further depending on the level of apoptosis within the aged PMN cultures. LOW, <35% apoptosis; MED, 35–55% apoptosis; HIGH, >55% apoptosis. (C and D) Alternatively, aged, washed PMNs were lysed with one cycle of freeze-thaw to render cells permeable to PI before coculturing for 5 min with HMDMs in the absence or presence of PRPDS, pretreated (D) or not (C) with Hec7.2 (10 µg/mL).
|
 View larger version (35K): [in a new window] |
Figure 3. Fn is an opsonin for apoptotic Jurkat cells and augments their phagocytic clearance. (A) Serum augments the phagocytic clearance of apoptotic Jurkat cells by THP-1 macrophages, regardless of whether the JKTs express CD31 (JKT POS) or not (JKT NEG). Phagocytosis assays were performed further in the absence (–PRPDS) or presence (+PRPDS) of nonautologous serum. (B) Addition of Fn (100 µg/ml) promotes ß1-dependent phagocytosis of apoptotic JKTs, expressing CD31 by THP-1 macrophages. An IgG isotype control for the ß1-blocking mAb P4C10 and P5D2 was found to have no effect and for clarity, has not been presented. (C) Apoptotic JKTs bound FITC-conjugated Fn, as assessed by flow cytometry when cultured in the absence of 10% FCS and was competed with unlabeled Fn at the defined concentrations. In contrast, apoptotic PMNs, cultured in the absence of autologous PRPDS, failed to bind exogenous Fn. (D) The binding of Fn to viable (VC) or apoptotic (AC) JKTs and PMNs, cultured in the presence or absence of PRPDS, was confirmed by Western blot with the anti-Fn mAb IST-4.
|
Macrophage ß1 integrin-binding is enhanced by ligation of macrophage CD31
Our priority, however, was to define the mechanism by which CD31 on the phagocyte promoted the engulfment of apoptotic cells, in which the preceding data suggested a functional link with ß1 integrins and the binding of Fn presented by the target feed. To refine our phagocytic model further and in an attempt to accentuate a reliance on CD31, we turned to the ß1 integrin-dependent clearance of Fn-coated Latex beads [19
]. Specifically, when THP-1 macrophages were coincubated with the agonistic anti-CD31 mAb P2B1, we observed an almost two-fold increase in the level of engulfment, which was not seen when a neutral anti-CD31 mAb 9G11 was used (Fig. 4A
). Furthermore, constitutive and CD31-inducible engulfment was ß1 integrin-dependent. Thus, CD31 ligation with an agonistic mAb promoted ß1 integrin-dependent engulfment of a particle opsonized with plasma Fn. It is important that P2B1 did not result in the up-regulation of THP-1 cell surface ß1 integrin expression or its activation, as monitored with the conformation-sensitive mAb 12G10 (data not presented). Instead, THP-1 macrophages stained uniformly for the ß1 integrin, which appeared to concentrate within the phagocytic cup after the capture of uncoated (data not presented) and Fn-coated Latex beads (Fig. 4C
and inset). HMDMs were excluded as the phagocyte as a result of the avidity with which they ingested beads in a manner that was insensitive to CD31 and ß1 integrin manipulation.
 View larger version (54K): [in a new window] |
Figure 4. CD31 ligation augments ß1 integrin-dependent clearance of Fn-coated microparticles. (A) THP-1 macrophages ingest Fn-coated Latex beads readily, which is augmented by coligating CD31 on THP-1 macrophages with the agonistic mAb P2B1 but not the neutral mAb 9G11. Ingestion in all cases was inhibited with the ß1 integrin-blocking mAb P5D2. The inset pictograph is representative and shows a concentration of ß1 integrin surrounding an unengulfed 3.0 µm Latex particle bound within a phagocytic cup. (B) Alternatively, CD31-directed binding of Dynabeads to THP-1 macrophages also augments P5D2-sensitive engulfment. (C) A series of sectional views (i–v) of a THP-1, showing a concentration of ß1 integrin adjacent to a surface-bound, Fn-coated Latex bead. Also shown (vi) is a typical staining for an isotype control mAb.
|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Our finding that CD31 activates phagocyte ß1 integrins and that they in turn bind Fn presented by apoptotic or particulate matter is perhaps not surprising. First, antibody ligation of leukocyte CD31 is known to activate ß1, ß2, and ß3 integrins in the presence of their ligands, and homophilic ligation of CD31 results in the up-regulation and activation of ß1 integrins on transmigrating leukocytes [12
, 13
]. More recently, enforced oligomerization of CD31 was found to promote 5ß1-dependent adhesion of transfected epithelial cells to immobilized Fn [15
]. In keeping with our own results, CD31 also interacts with vß3, and although this was originally reported to occur in trans [26
], it now appears to be a cis interaction [27
]. Thus, CD31 is intimately associated with integrins.
Second, although Fn is better appreciated as a component of the extracellular matrix, it also has important roles in promoting resolution of sepsis by activating C3 and Fc receptors on macrophages to bind complement-fixed pathogens [28 ]. Thus, Fn joins an ever-increasing list of acute-phase plasma proteins normally associated with the clearance of pathogens, such as collectins and pentraxins, which bind apoptotic cells to modulate engulfment [20 , 29 30 31 32 33 34 35 ]. However, Fn did not bind apoptotic PMNs or augment their clearance, perhaps explaining the lack of a serum effect on PMN clearance. Nevertheless, this reminds us that not all apoptotic cells express the same "array" of proteins, and if they did, they may not engage the same array of phagocytic receptors. For instance, although apoptotic PMNs bind iC3b [23 ], which we have confirmed (unpublished observations), we have failed to observe a role for serum complement or ß2 integrins in promoting macrophage clearance of apoptotic PMNs. Although this finding is somewhat disappointing in that it would have provided, by analogy with sepsis, a mechanism by which Fn could augment clearance [28 , 36 ], these results leave open the possibility that by binding phagocyte ß1 integrins, Fn augments phagocytosis through the activation of other receptor-ligand interactions.
A requirement for other receptor-ligand interactions might also explain why Fn on its own was not effective at concentrations normally encountered in plasma (1–2 µg/mL) but was at 100 µg/mL. Although the bioactivity of purified plasma Fn may be less effective than fresh plasma, it remains likely, as with the recognition and clearance of Staphylococcus aureus, that other plasma constituents were required [37 ]. Also of note is the description of purified plasma Fn at concentrations greater than 40 µg/mL, augmenting the phagocytosis of red blood cells, which had been opsonized with components of the alternative but not the classical complement system [36 ].
These studies have, however, highlighted major limitations associated with the interpretation of in vitro efferocytosis assays, which we and others have commonly used; specifically, the issue of reproducibility between and sometimes within groups and the presence of necrotic cells. Over the past 15 years, a number of receptors, opsonins, and ACAMPs have been implicated in efferocytosis, in which many are operative within the same system, e.g., HMDMs. It is precisely because of this redundancy perhaps that consistent observations regarding a particular receptor are variable and why progress in defining a mechanism has been slow [4
]. Reproducibility is complicated further by the inherent variability of preparing and working with primary cells in the ability of macrophages to ingest and apoptotic cells to be ingested. Increasingly, we also recognize that not all apoptotic cells are equal, even within the same preparation and that secondary necrosis is impossible to control [4
, 38
39
40
]. Therefore, the vagaries of phagocytosis and dependency on particular recognition pathways are readily understood [18
]. Given our own difficulties in revealing a role for vß3 by HMDMs, it might also explain why we fail to see a dependency on serum complement and ß2 receptors. Further complexity is introduced by variations in how different groups prepare their cells. Thus, inducing apoptosis by culturing PMNs in the absence [23
] or presence [16
] of serum may affect the display of molecular moieties by apoptotic cells, including their opsonization with plasma protein, such that the ability of serum to augment engulfment is perceived differently. However, even when we cultured PMNs in the absence of serum but performed phagocytosis in the presence of autologous or nonautologous serum, we could still find no evidence that serum augmented engulfment other than for apoptotic cell debris, which is not normally encountered in constitutively aged cultures.
In conclusion, we have presented evidence that Fn can act as a selective opsonin to promote the phagocytic uptake of cellular or particulate matter. Augmented phagocytosis as a result of Fn was further shown to be dependent on the ability of phagocyte CD31 to regulate phagocyte ß1 integrins, which were presumably responsible for binding Fn presented by particulate matter. A role for Fn in the clearance of apoptotic cells also complements its role in promoting the phagocytic removal of necrotic cells [41 ] and pathogens [28 , 36 , 37 ]. Ongoing work in our laboratory is aimed at defining the underlying mechanism by which apoptotic, cell-directed cross-linking of CD31 signals ß1 integrins.
Received October 5, 2005; revised January 17, 2006; accepted February 15, 2006.
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
6 ß 1 on transmigrated neutrophils in vivo and plays a functional role in the ability of 6 integrins to mediate leukocyte migration through the perivascular basement membrane J. Exp. Med. 196,1201-1211 v ß 3 differentially regulates adhesive and phagocytic functions of the fibronectin receptor 5 ß 1 J. Cell Biol. 127,1129-1137 v ß 3 as a heterotypic ligand for CD31/PECAM-1 J. Cell Sci. 109,437-445[Abstract] v ß 3 in cis Mol. Biol. Cell 11,3109-3121This article has been cited by other articles:
 |
|
 |
|
  R. Goel, B. R. Schrank, S. Arora, B. Boylan, B. Fleming, H. Miura, P. J. Newman, R. C. Molthen, and D. K. Newman Site-Specific Effects of PECAM-1 on Atherosclerosis in LDL Receptor-Deficient Mice Arterioscler. Thromb. Vasc. Biol., November 1, 2008; 28(11): 1996 - 2002. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 E. F. Vernon-Wilson, F. Aurade, L. Tian, I. C. M. Rowe, M. J. Shipston, J. Savill, and S. B. Brown CD31 delays phagocyte membrane repolarization to promote efficient binding of apoptotic cells J. Leukoc. Biol., November 1, 2007; 82(5): 1278 - 1288. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. B. Brown, C. S. Tucker, C. Ford, Y. Lee, D. R. Dunbar, and J. J. Mullins Class III antiarrhythmic methanesulfonanilides inhibit leukocyte recruitment in zebrafish J. Leukoc. Biol., July 1, 2007; 82(1): 79 - 84. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
