| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Published online before print August 7, 2007
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
* MRC Centre for Inflammation Research and
Centre for Integrative Physiology, College of Medicine and Veterinary Medicine, University of Edinburgh, Edinburgh, United Kingdom
2 Correspondence: Centre for Inflammation Research, Queen’s Medical Research Institute, 47 Little France Crescent, Edinburgh EH16 4TJ, UK. E-mail: simon.brown{at}ed.ac.uk
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Key Words: apoptosis • phagocytosis • integrins • ether-a-go-go related gene (ERG)
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
The cross-linking of CD31 with antibodies is also known to activate β1, β2, and β3 integrins in vitro to promote leukocyte cell adhesion [6
], in which the small GTPase Rap1 is implicated [13
]. More recently, oligomerization of CD31, independent of homophilic ligation or antibody binding, was found to promote 
5β1-dependent adhesion of transfected epithelial cells to immobilized fibronectin (Fn), whereas dimerization led to homophilic cell–cell interactions [14
]. Although initial reports suggested that vβ3 was a heterotypic ligand for CD31 [9
, 15
], it is now generally accepted that CD31 and vβ3 associate in cis [16
]. No evidence exists, however, to suggest that CD31 associates physically with β1 or β2 integrins. Regardless, the mechanism by which CD31 regulates integrin function remains unknown [6
].
Homophilic CD31 interactions have analogous consequences for macrophage discrimination of living from dying cells [17 18 19 20 ]. Healthy, CD31-positive leukocytes disengage actively from macrophages in a CD31-dependent manner, whereas apoptotic cells bearing CD31 are tightly bound, leading to their engulfment. The ability of live cells to disengage from macrophages was dependent on a functional ITIM motif within CD31, expressed by the detaching leukocyte. In contrast, the ability of macrophage CD31 to promote engulfment was ITIM-independent, in which a role for macrophage integrins in the phagocytic recognition and clearance of apoptotic cells is well described [21 ], including CD31-directed regulation of β1 integrins [19 ]. An alternative interpretation of our work is the possibility that by ligating CD31 on macrophages, live leukocytes inhibit macrophage engulfment in much the same manner proposed for CD47-dependent ligation of macrophage signal regulatory protein a (SIRPa) [22 23 24 ]. As attractive a possibility as this is, we have never been able to demonstrate such a function for CD31, despite obvious structural similarities between SIRPa and CD31 as inhibitory receptors [25 , 26 ]. Nevertheless, the key to how CD31 functions will be a better understanding as to the exact role of the ITIM contained within the cytoplasmic tail of CD31, which is implicated in adhesion [13 , 27 ] and migration [17 , 28 , 29 ].
Ether-à-go-go-related gene (ERG) is a voltage-gated potassium channel, which is perhaps best known as a major component of the channel responsible for the rapidly activated delayed rectifier current, promoting repolarization of the cardiac action potential, in which inherited mutations are associated with arrhythmias and a long QT syndrome [30 ]. ERG has also been described in microglial and leukemic cell lines but where ERG expression is not thought to be present in their normal counterparts, including monocytes [31 32 33 34 ]. Nevertheless, we have now described a role for ERG in regulating leukocyte recruitment to an infected wound using zebrafish [35 ], establishing an important role for ERG in leukocyte transmigration, a defined, CD31-dependent process. In this report, we describe how we first identified ERG and how it functions in macrophages and an erythroleukemic cell line to re-establish a resting membrane potential following membrane depolarization. We describe further how cross-linking CD31 by homophilic ligation or antibody binding inhibited ERG current in an ITIM-independent manner. It is important that prolonged depolarization of the macrophage favored the firm binding of phagocytic targets in an integrin-dependent manner.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
vβ3 (Kav100; a gift from Scott D. Blystone, SUNY Upstate Medical University, Syracuse, NY, USA), was maintained in DMEM-F12 + Glutamax supplemented with 10% FBS. Stable lines expressing variant CD31 forms were generated by transfection following electroporation (750 V, 25 µF, BioRad Gene Pulser, BioRad, Hercules, CA, USA). In brief, 5 x 106 cells were resupended in HBSS, pH 6.7, containing 10 mM HEPES, 1 nM ATP, and 2 mM glutathione with 20 µg plasmid DNA. All CD31 plasmid constructs were subcloned from pCDNA3 (gifts from Chris D. Buckley, University of Birmingham, Birmingham, UK) into pCMV-Hyg with HindIII/NotI and polyclonal cultures selected by supplementing medium with 250 µg/mL hygromycin. Kav100 lines were passaged routinely in the presence of 100 U/ml penicillin and 100 µg/mL streptomycin including G418 (1.2 mg/mL) and hygromycin (250 µg/mL) for vβ3 and CD31 transfectants, respectively. Kav100 lines were used typically after removal from antibiotic selection for at least two passages. CD31-coated ThermanoxTM coverslips, when used to culture Kav100 lines, were prepared as described previously [17
]. Human monocyte-derived macrophages (HMDMs) and PMNs were prepared and cultured as described previously, in which HMDM monolayers were typically washed of unbound cells after 1 h and 1 and 3 days [36 ]. PMNs underwent spontaneous apoptosis when cultured in the presence of 10% autologous, platelet-rich, plasma-derived serum. Human leukemic Jurkat T cell lines (JKTs) were maintained in RPMI 1640, supplemented with 10% FBS, 2 mM L-glutamine, 100 U/ml penicillin, and 100 µg/ml streptomycin, and were induced to undergo apoptosis following Fas ligation (CH-11, 100 ng/ml, 24 h). Apoptosis and necrosis were assessed by flow cytometry using annexin-V-FITC and propidium iodide (10 µg/ml).
Phagocytosis of Fn-coated LatexTM beads (Fn-beads)
Fn-beads were prepared with 3.0 µm Polybead Amino Microspheres (Polysciences Inc., Warrington, PA, USA), modified with glutaraldehyde, and reacted with human plasma Fn as described [37
]. Kav100 cells were typically washed with DMEM (no serum), and 
4 x 107 beads were added per 48-well assay. Fn-beads were allowed to settle and bind for 30 min before gently removing all media and adding trypsin-EDTA for a further 15 min to remove loosely bound Fn-beads, and Fn-bead ingestion was scored by light microscopy and expressed as the number bound into phagocytic cups per 100 cells (phagocytic index). Arg-Gly-Asp-Ser (RGDS; 1 mM) and all antibodies, including control IgG mAb, were added at 10 µg/mL for 15 min at room temperature prior to the addition of Fn-beads.
Protein chemistry and cross-linking studies
Fab fragments of anti-CD31 mAb (P2B1, 9G11, YRI31.2, YRI31.8, YRI31.9, YRI31.12) and F(ab')2 of P2B1 were generated with immobilized Ficin, according to the manufacturer’s instructions (Pierce Biotechnology, Rockford, IL, USA). Fab fragments may have also been modified further with sulfosuccinimidyl[2-6-(biotinamido)-2-(p-azidobenzamido)-hexanoamido]-ethyl-1, 3'-dithiopropionate (Sulfo-SBED; 1 µg Sulfo-SBED/10 µg Fab), according to the manufacturer’s instructions (Pierce Biotechnology). Specific binding of the Fab reagent was confirmed by indirect epifluorescent microscopy using an Alexa-488-conjugated streptavidin. After binding for 30 min at room temperature and washing away the unbound reagent with a Phenol Red-free HBSS, the cells were exposed to a short UV wavelength (UVC; 100 Jm–2x3). Cells were then lysed in 10 mM Hepes buffer containing 1% Triton X-100 and a cocktail of protease inhibitors (Calbiochem, San Diego, CA, USA), supplemented with 1 mM Na3VO4 and 1 mM NaF. Biotinylated proteins and coassociating proteins were recovered by immunoprecipitation with an anti-biotin mAb. The immunoprecipitate was then reduced with 100 mM ME before depleting CD31 with 9G11 immobilized to DynabeadsTM.
Flash luminescence plate reader (FLIPR)-Blue as a membrane potential reporter dye
FLIPR-Blue reagent (Molecular Devices, Sunnyvale, CA, USA) is a membrane potential-reporting dye, which shows good correlation with patch clamping and is amenable to fluorescence imaging [38
]. Kav100 cell lines were incubated with FLIPR-Blue according to the manufacturer’s instructions, in which the reconstituted dye was added directly to the culture medium at a 1:5 dilution, and cells were maintained at room temperature. Following a minimum, 5-min loading period, cells were visualized by fluorescence microscopy. For HMDMs, the FLIPR-Blue dye was reconstituted in DMEM/10% FBS and used at a 1:1 dilution, as this was found to give a superior signal-to-noise ratio. For experiments involving dofetilide (DOF) or E4031, cells were always maintained in the presence of the inhibitors, including any wash steps, so as to avoid washout. For CD31 cross-linking, cells were preincubated with P2B1 (10 µg/mL) for 15 min prior to the addition of FLIPR-Blue and KCl, before replacing the depolarizing media with fresh media containing FLIPR-Blue (see 
Figs. 4
and 5
). In all instances, depolarization was affected with 1 M KCl to give a final added concentration of 50 mM. Repolarization was affected by replacing the medium with normal saline media. For HMDM binding of apoptotic cells, all experiments were performed on the microscope stage by titrating in apoptotic cells to approximately a 1:1 ratio, where HMDMs pretreated with P5D2 or a control IgG (i.e., see Figs. 6J
and 6K
and 7D
) were washed of unbound mAb first before addition of apoptotic cells.
|
|
|
|
|
|
|
Macrophage binding of apoptotic leukocytes
HMDMs, cultured for 5–6 days in 48-well plates, were washed with serum-free DMEM before adding 1 x 106 PMNs or 1 x 106 JKTs, in which both had undergone >70% apoptosis (data not presented). The 48-well plates were then centrifuged at 100 g for 2 min before removing immediately from the centrifuge, submerging in a small tank of serum-free DMEM and inverting for 20 min to allow unbound PMNs to fall away from the HMDM monolayer under gravity, in which trapped air bubbles were avoided. Plates were removed, fixed with 4% paraformaldehdye, and counted by light microscopy. In addition and prior to incubation with apoptotic cells, HMDMs may have been depolarized with 50 mM KCl or depolarized and repolarized by replacing the added K+ with serum-free DMEM, in which DOF may have also been present at 10 µM. For β1-integrin blockade studies, washed HMDMs were preincubated with P5D2 (10 µg/mL) or a control IgG for 15 min before washing free of unbound antibody. All steps were performed at room temperature.
Electrophysiology
Whole cell currents were recorded in the conventional whole-cell mode of the patch-clamp technique at room temperature (21–23°C). Outward K+ currents were determined in physiological K+ gradients. The bath solution (extracellular) was contained in mM: 140 NaCl, 5 KCl, 2 MgCl2, 1 CaCl2, 10 HEPES, 20 glucose, pH 7.4. The patch pipette (intracellular) was contained in mM: 140 KCl, 2 MgCl2, 10 HEPES, 30 glucose, 1 BAPTA, 1 ATP, pH 7.3, with intracellular-free calcium buffered to 100 nM. Cells were voltage-clamped at –50 mV and depolarized to the respective potentials for 100 ms. Steady-state, outward current was determined 80 ms into the pulse and was stable for >30 min under these conditions. Resting membrane potential (Vrest) was determined in the conventional, whole-cell current clamp mode under the same ionic conditions. Vrest was monitored for 5–10 min following establishment of the whole-cell configuration to ensure stability, and the Vrest was averaged over 5 min. All data acquisition and voltage protocols were controlled by an Axopatch 200B (or 200A) amplifier and pCLAMP9 software (Axon Instruments Inc., Foster City, CA, USA). All data were sampled at 10 kHz and filtered at 2 kHz. Pipettes were manufactured from Garner No. 7052 glass, with resistances of 1–3 M
in physiological saline after fire polishing. DOF (10 µM), mAb P2B (10 µg/mL), or KCl (54.5 mM) were applied in bath solution by gravity-driven perfusion at a flow of 3–5 mL/min or by direct application to the bath.
Data analysis
All data were analyzed by single-factor ANOVA (Microsoft Excel), and all P values were determined by Tukey’s post-hoc analysis. All errors are expressed as a 95% confidence interval (c.i) unless stated otherwise.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
vβ3, in which 5β1 is the only integrin expressed endogenously. It is important that K562 cells are devoid of many receptors implicated in the phagocytic clearance of apoptotic cells, including CD31, and therefore, represent a reductionist but genetically tractable system to model β1-integrin-dependent, adhesive events involved in phagocytosis. K562 cells are an erythroleukemic cell line, which exists as a single cell suspension. Ectopic expression of WT CD31 results in the formation of large cellular aggregates, which are difficult to maintain as a single cell suspension. In contrast, Kav100 cells are adherent and can be plated out readily as single cell colonies, regardless of CD31 expression. This was important, as it allowed us to control better for the effects of homophilic CD31 interactions between juxtaposed cells, which affected the ability of CD31 to modulate the binding of Fn-beads (Fig. 1 ). Stable expression of CD31 by Kav100 cells (Kav31WT) at HCD (160,000 cells/cm2; >80% confluent) promoted the binding of Fn-beads into distinct phagocytic cups (Fig. 1A 1B 1C 1D) at levels greater than that seen for Kav100 or a hygromycin vector control, KavHygro (Fig. 1E) . The ability of CD31 to promote Fn-bead binding at HCD was independent of an ITIM motif found within the cytoplasmic tail of CD31 but was dependent on an intact cytoplasmic tail (Fig. 1E) . The inability of CD31 containing a K89A point mutation (Kav31K89A) to promote Fn-bead binding at HCD further indicated that homophilic interactions between juxtaposed cells were required (Fig. 1E) [39 ]. It is important that the expression of CD31 by all lines was comparable (supplemental data) in which β1-integrin surface expression remained unaffected (data not shown).
The importance of cell–cell contact was also confirmed with Kav31WT and Kav2YF cells when seeded at LCD to give single-cell colonies (typically <20,000 cells/cm2; <10% confluent; Fig. 1F
). At LCD, Kav31 and Kav2YF bound Fn-beads at reduced levels, consistent with that observed for KavHygro cells. In contrast, when Kav31WT and Kav2YF cells but not Kav100 were seeded on CD31-coated coverslips at LCD, Fn-bead binding was restored to levels approaching that observed at HCD. Alternatively, we could augment Fn-bead binding by cross-linking CD31 with P2B1, an anti-CD31 mAb (Fig. 1F)
. In keeping with a previous report [37
], Fn-bead binding was β1-integrin-mediated, being blocked with the mAb P5D2 (Fig. 1G)
and P4C10 (data not presented), which was not augmented with an vβ3-blocking mAb, LM609. Although RGDS blocked Fn-bead binding, it was observed to also have a deleterious effect on cell adhesion. Thus, cross-linking CD31 by homophilic ligation (HCD), antibody binding (LCD), or binding of Kav31WT cells to CD31-coated surfaces regulated β1-integrin-dependent binding of Fn-beads.
CD31 associates with the voltage-gated potassium channel ERG
As CD31-directed β1-integrin function was independent of the ITIM motif within the cytoplasmic tail of CD31, we sought novel protein partners for CD31 in Kav31WT cells, which at HCD, where Fn-bead binding was constitutively high, were exposed to the chemical cross-linking agents disuccinimidyl-suberate (DSS), dithiobis-succinimidyl-propionate (DSP), or N-succinimidyl-3-(2-pyridyldithio)-propionate (SPDP). Unfortunately, this direct chemical approach failed to identify any novel, protein-binding partners; however, an indirect approach using a biotin transfer reagent was more successful. Protein cross-linking, using Sulfo-SBED (Perbio Sciences, UK) conjugated to a Fab antibody recognizing Ig Domain 4 of CD31 (YRI31.8) but not Fabs directed against Ig Domains 1 and 2 (data not presented), was found to incorporate biotin into a 140-kDa protein distinct from CD31. Tandem mass spectrometry (MS/MS) analysis unequivocally identified our biotinylated protein of 140 kDa as the pore-forming -subunit of the voltage-gated potassium channel, ERG1a (supplemental data), after CD31 and YRI31.8 Fab were immunodepleted from the sample following chemical reduction of the disulfide bond within Sulfo-SBED (Fig. 2D
, Lane ii). This protein band was not observed if the cell-bound Fab reagent was not exposed to UVC, which would otherwise activate the aryl azide moiety of Sulfo-SBED for chemical cross-linking (data not presented). Furthermore, no biotinylated proteins were observed when we used Kav100 cells, where immunofluorescence confirmed that YRI31.8 failed to bind (data not presented).
To validate the specificity in biotin tagging of ERG, we repeated the experiment with Kav31.30 cells, which failed to augment Fn-bead binding at HCD. It is interesting that biotinylated ERG could still be recovered but at levels that were reduced substantially compared with Kav31WT cells (Fig. 2E) . The lower recovery of a biotinylated ERG from Kav31.30 cells was not a result of a reduction in CD31 (or ERG) surface expression (supplemental data), in which the absence of a terminal 88 amino acid carboxy tail was confirmed by Western blot using a pAb raised against the carboxy terminus of CD31 (Fig. 2F) . It is important that ERG was also biotinylated with our Fab reagent on HMDMs, where we have confirmed the presence of ERG by Western blot (Fig. 2G) and RT-PCR (data not shown). These results indicate that the ability of our CD31-targeted Fab reagent to transfer biotin efficiently to ERG was dependent on a full-length cytoplasmic tail of CD31.
Consistent with the indirect cross-linking strategy, we also observed that ERG and CD31 frequently colocalized by immunofluorescence at the cell surface for Kav31WT cells (Fig. 3A ) and HMDMs (Fig. 3B) . However, attempts to coimmunoprecipiate CD31 with ERG or vice versa, even following direct chemical cross-linking strategies with DSS, DSP, or SPDP, were unsuccessful, indicating that CD31 and ERG may not interact directly but may form part of a larger macromolecular complex.
ERG current is activated following cell depolarization and is inhibited by CD31 cross-linking
Voltage-gated potassium channels, including ERG [40
, 41
], function to set and maintain a resting membrane potential (Vrest). To determine how ERG functioned in Kav31WT cells at LCD, we used a fluorescent membrane potential reporting dye (FLIPR-Blue), which exhibits good correlation with patch-clamping [38
], and two highly selective inhibitors of the ERG current, DOF and E4031, which possess no known alternative activity. DOF and E4031 had no effect on Vrest (Fig. 4A
), a finding confirmed by whole-cell current clamp recordings, in which Kav31WT cells had a resting membrane potential of –33.6 ± 3.5 mV. Thus, ERG was not responsible for setting Vrest in Kav31WT cells (and KavHygro cells; data not shown).
ERG current is activated by membrane depolarization [30 ]. To control for the efficacy of DOF and E4031 as ERG inhibitors, we deliberately depolarized our cells at LCD with 54.5 mM K+ for 5 min (Vrest=–5.8±4.1 mV by patch clamp) before returning to normal saline conditions. In the absence of DOF or E4031, Kav31WT cells repolarized rapidly (Vrest=–31.4±5.5 mV by patch clamp), where they had returned to control values within 5 min. In the presence of DOF or E4031, Kav31WT cells repolarized to varying degrees in a concentration-dependent manner, which correlated strongly with the number of Fn-beads that bound Kav31WT cells (Fig. 4B) . These results emphasized an important role for ERG in the repolarization of Kav100 cell lines and indicated that the depolarized state may have been permissive for integrin-dependent Fn-bead binding. Indeed, deliberate depolarization of Kav100 and Kav31WT cells by raising the extracellular concentration of K+ ([K+]o) from 4.5 to 54.5 mM enhanced Fn-bead binding, which was inhibited with RGDS peptide or the β1-integrin-blocking mAb P5D2 (Fig. 4C) .
When the above experiments were performed at HCD using FLIPR-Blue, we observed that Kav31WT cells now failed to repolarize as fast as they had at LCD. Kav31WT cells repolarized with an apparent half-time of less than 1 min when at LCD, which increased to 15 min when at HCD, a value we also obtained for Kav31WT cells treated with DOF at LCD (Fig. 4D) . These data suggest that cross-linking CD31 by homophilic ligation might inhibit ERG-dependent repolarization. To test this possibility, we used the anti-CD31 mAb P2B1 to cross-link CD31 on Kav31WT cells at LCD (Fig. 4D) . P2B1 whole IgG and a F(ab')2, but not a Fab fragment, inhibited the rapid repolarization of Kav31WT cells following brief exposure to [K+]o= 54.5 mM (Fig. 4D) . Of significance, the inhibitory effect of P2B1 on repolarization closely paralleled that for DOF. Similarly, whole cell voltage clamp analysis revealed that the cross-linking of CD31 with P2B1 was as effective as DOF at inhibiting outward current at depolarized voltages (Fig. 4E) . The effects of DOF and P2B1 were not additive, synergistic, or antagonistic, suggesting that CD31 and ERG functioned in series (Fig. 4F) . These studies indicate that the cross-linking of CD31 by homophilic ligation or antibody binding was responsible for inhibition of ERG current.
CD31 cross-linking inhibits ERG in macrophages
The above studies identified ERG in a leukemic cell line, where CD31 is not expressed, and where ERG is not present in monocytes [33
]. Our data, however, indicated that ERG was present in macrophages, where it could be cross-linked to CD31 using our Fab reagent (Fig. 2E
and 2G)
. We therefore set out to determine whether CD31 could regulate ERG function in primary macrophages as it had for Kav31WT cells. Our first observation with HMDMs, in contrast with K562-derived cell lines, was the greater heterogeneity in FLIPR-Blue staining for individual cells (Fig. 5A
), in which a higher concentration of FLIPR-Blue reagent was used to give a superior signal-to-noise ratio. Deliberate depolarization of HMDMs with [K+]o= 54.5 mM was observed to increase the overall fluorescence of our HMDMs and where repolarization was again sensitive to inhibition by DOF and a F(ab')2 of P2B1 but not a control IgG (Fig. 5B
and 5C)
. F(ab')2 of P2B1 was used so as to avoid complications with FcR binding. These results demonstrated a significant role for ERG current in the repolarization of HMDMs and indicate, as for Kav31WT cells, that by cross-linking CD31, we could inhibit ERG to prolong the depolarized state.
Apoptotic cells depolarize macrophages to promote firm binding
In determining whether homophilic ligation of CD31 on HMDM by the apoptotic cell CD31 could also inhibit HMDM repolarization, we became aware that apoptotic cells, but not viable cells, depolarized HMDMs on cell–cell contact. When apoptotic PMNs were coincubated with HMDMs at approximately a 1:1 cell ratio, we noted that the more hyperpolarized HMDMs [low relative fluorescence intensity (RFI)] depolarized on contact with apoptotic PMNs (Fig. 6A
6B
6C
). In contrast, HMDMs, which were relatively depolarized (high RFI), were unaffected (Fig. 6C)
. By focusing on 15% of the HMDM population with a low RFI (<50), the binding of apoptotic cells was found to maintain HMDM depolarization for up to 15 min at 20°C, a temperature at which apoptotic cells were not engulfed. It is important that the ability of apoptotic cells to sustain depolarization (Fig. 6D)
was blocked with a noncross-linking Fab fragment of P2B1 (Fig. 6E)
but not a control IgG (Fig. 6F)
, which in turn, could be rescued with DOF (Fig. 6G)
. It is interesting that DOF also blocked the number of apoptotic cells, which detached from macrophages after initial depolarization in the presence of P2B1 Fab. In these experiments, P2B1 Fab functioned to block homophilic ligation of macrophage CD31 by apoptotic cell CD31 without cross-linking HMDM CD31.
We also used JKTs, serially selected for the presence (JKTPOS) or absence (JKTNEG) of CD31, to examine the role of CD31 and apoptotic cell binding by HMDMs in prolonging the depolarized state of macrophages. Consistent with the above findings, we again found that for those macrophages with a low RFI, JKTPOS cells but not JKTNEG cells were able to inhibit repolarization following contact-dependent depolarization (Fig. 6H and 6I) . It is important that JKTNEG cells were often observed to depolarize macrophages and then drift away, as seen with apoptotic PMNs in the presence of P2B1 Fab (Fig. 6E) , in which the macrophages then repolarized (11/23). In contrast, JKTPOS cells (1/13) rarely disengaged. We conclude that cross-linking CD31 was required to inhibit membrane repolarization of HMDMs to promote firm binding of apoptotic cells.
Apoptotic cell binding is integrin-dependent
We have shown previously that the phagocytosis of apoptotic JKT cells by THP-1 macrophages is CD31-directed and β1-integrin-mediated [19
]. To explore whether the binding of apoptotic JKTs by HMDMs could also be mediated by β1 integrins, we pre-exposed HMDMs to the β1-integrin functional-blocking mAb P5D2 for 5 min before removing excess antibody, adding FLIPR-Blue and leaving for a further 5 min before addition of apoptotic cells (Fig. 6J
and 6K)
. By again focusing on those macrophages with a low RFI, we observed that the number of JKTPOS (8/25), which bound HMDMs, following initial contact and depolarization, was reduced significantly (P<0.01). It is important that in those instances when apoptotic cells failed to remain bound, HMDMs repolarized quickly (Fig. 6I
6J
6K)
. In all cases, macrophage contact with apoptotic cells, regardless of CD31 expression, was the depolarizing stimulus for HMDMs.
Finally and to explore the effect of deliberate membrane depolarization on apoptotic cell binding, we depolarized HMDMs with [K+]o= 54.5 mM and observed a significant increase in the proportion binding apoptotic PMNs and the number of PMNs each HMDM bound (Fig. 7A 7B 7C ). In addition, repolarization of depolarized HMDMs before the addition of apoptotic PMNs returned binding to levels typically observed for untreated controls, unless DOF remained present. Similar results were obtained with apoptotic JKT cells, regardless of CD31 expression (Fig. 7D) . Furthermore, the binding of apoptotic JKTs could be inhibited by pretreatment of the HMDMs with the β1-blocking mAb P5D2 (10 µg/mL) but not a control IgG. Taken together, these data suggest that macrophage membrane depolarization, on contact with apoptotic cells or following an increase in extracellular [K+]o, favored the binding of apoptotic cells, in which a role for β1 integrins could be discerned for HMDM binding of apoptotic JKTs.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Similarly, ERG current was responsible for rapid repolarization of depolarized macrophages, which could again be inhibited following homophilic ligation of CD31 between macrophages and apoptotic cells. As with Kav31WT cells, the finding that P2B1 and DOF had identical effects on the rate of macrophage repolarization is taken to indicate that ERG inhibition is downstream of CD31 in promoting apoptotic cell binding. However, P2B1 can also block phagocytosis of apoptotic cells [17 , 19 ]. It is important to recognize that we identified CD31 originally in the tethering of apoptotic cells by macrophages, and if tethering is blocked, then regardless of whether the macrophage is depolarized, the efficiency of apoptotic cell binding may be diminished. Thus, CD31 appears to function not only in the initial tethering of apoptotic cells [17 ] but also by inhibiting ERG current, rendering macrophages permissive for the firm binding of apoptotic cells.
Our initial assays identified a functional role for ERG in K562-derived cells. It is therefore reassuring to find that a similar mechanism for CD31-dependent modulation of ERG and β1 integrins occurred in the binding of apoptotic cells by macrophages and validates the experimental approach taken. Our initial difficulty in pursuing CD31-dependent regulation of β1-integrin function, exclusively within macrophages, is perhaps best illustrated by the fact that inhibition of CD31-dependent binding of apoptotic cells only results in approximately a 25% (±15%, SD) reduction in overall engulfment, a finding confirmed by others [20 ]. By identifying ERG in a reductionist model of K562 cells, we were able to then focus on a specific subpopulation of macrophages, where CD31-directed regulation of ERG revealed a distinct role for β1 integrins in the binding of apoptotic cells. This is despite a previous report, where we could not discern a significant role for any integrin in the phagocytosis of apoptotic cells by macrophages [19 ]. We have argued previously that the assortment of recognition mechanisms, which can be used by macrophages, makes it difficult to isolate signaling events for a specific receptor [19 , 21 , 42 ]. K562 cells, therefore, provide a useful model for future studies to identify the nature of a macromolecular complex, which allows CD31 to signal ERG, in which any discoveries will again be validated in macrophages.
However, how does membrane depolarization favor β1-integrin function and macrophage binding of apoptotic cells, which leads to their subsequent engulfment? Although depolarization of excitable cells would favor a calcium influx through the activation of voltage-gated calcium channels to promote, e.g., actinomyosin contractility, this cannot occur in macrophages, where voltage-gated calcium channels are not expressed. This is supported by our observation that cross-linking CD31 with primary and secondary antibodies does not affect intracellular calcium concentrations in resting or depolarized macrophages when using Fura-2 as a reporter. Nevertheless, depolarization will affect ion flux across the plasma membrane in which there are other implications not so well appreciated.
Our data suggest that depolarization will activate an ERG current in K562 cells and HMDMs, which promotes the movement of K+ across the membrane. This will, by definition, generate a local electric field, which could couple with electric dipole moments of surface receptors to affect their function and/or signaling [30 , 43 , 44 ]. The prototypical example for such a mechanism, other than voltage-gated ion channels themselves, is the affinity modulation of muscarinic receptors described recently [45 ]. It is tempting to speculate that integrins may be affected similarly by local electric circuits, where ERG and Kv1.3 voltage-gated potassium channels are known to associate and regulate the activity of β1 integrins [46 47 48 ]. An alternative explanation, however, may be provided by the observation that transmembrane potential, as opposed to surface potential, can regulate phosphatidylinositol 4,5-bisphosphate (PI-4,5-P2) and PI 3,4,5-trisphosphate (PI-3,4,5-P3) levels through the activation of a phosphatase and tensin homologue-like phosphatase, which possesses an intrinsic voltage sensor [49 , 50 ]. PI-4,5-P2 and PI-3,4,5-P3 are key second messengers for phagocytosis [51 ]. In addition, changes to the PI pool will have an impact on surface potential and the recruitment of cytoplasmic signaling proteins involved in phagocytosis to the plasma membrane [52 ]. Thus, sustained or prolonged depolarization through the inhibition of an ERG current could promote phagocytic recognition of apoptotic cells through modulation of receptor function and/or alteration of the PI pool.
The suggestion that electric fields might govern how macrophages interact with leukocytes is all the more attractive when we consider our original identification of CD31 in mediating the detachment of live leukocytes from macrophages. Electrotaxis, i.e., directed cell migration under the influence of an electric field, is a well-described phenomenon [44 , 50 ]. Like CD31, electric fields regulate the speed and direction of leukocyte migration [8 , 50 , 53 ]. It is tempting to speculate that the ability of viable leukocytes to detach and move away from macrophages is in effect an electrotactic response to electric fields generated by macrophages. In contrast, apoptotic cells are functionally inert and therefore unable to escape binding or detachment from macrophages [17 ]. Precisely how macrophages might generate electric fields or the ions and their channels responsible must remain the subject of speculation until the tools to monitor electric fields at the single-cell level become available (Colin McCaig and Min Zhao, University of Aberdeen, Aberdeen, UK, personal communication). Nevertheless, given that live cells detach and move away from macrophages, we can assume that any electric field vector must project away from the apical surface of the macrophage. This could be achieved, for example, if K+ efflux were greater at the apical surface and K+ influx was greater at the basolateral surface, where the Na+/K+ ATPase pump is known to preferentially localize in epithelial cells [54 ]. However, one cannot ignore the contribution of Na+ and Cl– ion fluxes to local field strength.
Our finding that apoptotic cells can provide a stimulus for macrophage depolarization, independent of CD31 expression, is a novel but perhaps not an unexpected observation. Although we cannot rule out receptor-ligand interactions as mediating the depolarization of macrophages, it is known that apoptotic cells will leak K+ and other soluble factors, such as ATP, which might contribute to the depolarization of macrophages upon cell–cell contact [55 , 56 ]. Of particular interest to us is the recent report that ATP, by binding the purinergic P2Y6 receptor, can promote microglial uptake of Latex beads, where β1 integrins were again concentrated within the phagocytic cup [56 ]. Thus, the release of soluble factors by apoptotic cells may not only recruit macrophages [57 ] and activate calcium entry for cytoskeletal reorganization [56 ] but also depolarize macrophages for firm binding of the phagocytic target. It is interesting that Fc-mediated phagocytosis, implicated in apoptotic cell clearance [58 ], is also associated with membrane depolarization [59 60 61 62 ]. As phagocytic recognition of apoptotic cells is likely to involve a multiprotein complex [42 ], many members of which are known to regulate or be regulated by membrane electrophysiology, it will be a challenging task to define exactly how CD31 regulates ERG in phagocytes to inhibit membrane repolarization and promote binding of the phagocytic target. Nevertheless, the key finding of the current work establishes ERG as a new effector of CD31 signaling and a molecular target for regulating macrophage binding of phagocytic targets.
|
ACKNOWLEDGEMENTS |
|---|
|
FOOTNOTES |
|---|
Received May 6, 2007; revised June 8, 2007; accepted July 13, 2007.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
6β1 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. 196,1201-1211vβ3 integrin and enhances β1 integrin-mediated adhesion of eosinophils to endothelial cells Blood 94,1319-1329 v β 3 as a heterotypic ligand for CD31/PECAM-1 J. Cell Sci. 109,437-445[Abstract]vβ3 in cis Mol. Biol. Cell 11,3109-3121 (SIRP) regulates Fc
and complement receptor-mediated phagocytosis J. Exp. Med. 193,855-862 v β 3 differentially regulates adhesive and phagocytic functions of the fibronectin receptor 5 β 1 J. Cell Biol. 127,1129-1137(v)β(3) integrin in the preosteoclastic leukemia cell line FLG29.1 J. Biol. Chem. 276,4923-4931 and PTEN Nature 442,457-460[CrossRef][Medline] 2b/ 1 Fc receptor-ligand binding Proc. Natl. Acad. Sci. USA 80,1357-1361
), or Fab (•), controlled further with an irrelevant, whole IgG. DEPOL, Depolarized. (E) Current-voltage (I/V) relationship of the outward current, normalized to cell capacitance, recorded in the physiological voltage range in Kav31WT cells. Pretreatment of cells with P2B1 or DOF for 25 min significantly reduced the outward current determined between –20 and +30 mV (P<0.01, ANOVA). n = 8–12/group. A control IgG had no significant effect on the whole cell I/V (data not presented). (F) FLIPR-Blue fluorescence was recorded for Kav31WT cells grown at LCD and 10 min after removal from 54.5 mM K+ and maintained in varying concentrations of DOF and/or P2B1. Error bars represent SD (B) or 95% c.i. (C–F).
50) are depolarized by apoptotic PMNs. The data were compiled from a total of 71 binding events incorporating five distinct HMDM and eight apoptotic PMN preparations. (D–G) Focusing on individual HMDMs with a RFI of less than 60, a P2B1 Fab (E) but not a control IgG Fab (F), both at 10 µg/mL, was found to inhibit prolonged depolarization of HMDMs induced by apoptotic PMNs compared with control HMDMs (D). It is important that the inhibitory effect of P2B1 could be rescued if cells were cocultured with 10 µM DOF (G). (H–K) In addition, apoptotic JKTs, which were positive (H and J) or negative (I and K) for CD31 expression, were found to depolarize HMDMs on cell–cell contact, but only those that were positive maintained depolarization. (J and K) In contrast, when HMDMs were pretreated with a β1-integrin-blocking mAb, P5D2 (10 µg/mL), there was an increase in the number of individual apoptotic cells, detached, associating strongly with repolarization of the depolarized HMDM. n represents the number of individual HMDMs, which bound an apoptotic cell as a fraction of the total number of binding events monitored, except for I–J, where n represents the number of macrophages, which bound or failed (*) to bind an apoptotic cell, respectively, after depolarization. All error bars represent 95% c.i.