SEARCH

SEARCH BY CITATION

Keywords:

  • ANCA;
  • IgG subclass;
  • lymphocyte;
  • neutrophil;
  • vasculitis

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Disclosure
  9. References

Immunoglobulin G (IgG) is a potent neutrophil stimulus, particularly when presented as anti-neutrophil cytoplasm antibody (ANCA) in ANCA-associated vasculitis. We assessed whether IgG subclasses had differential effects on neutrophil activation and whether differences were dependent on specific Fc-receptor engagement. Using a physiologically relevant flow model, we compared adhesion of neutrophils to different subclasses of normal IgG coated onto solid surfaces, with adhesion of neutrophils treated with different subclasses of soluble ANCA IgG to P-selectin surfaces or endothelial cells (EC). Normal IgG captured flowing neutrophils efficiently in the order IgG3 > IgG1 > IgG2 > IgG4. Fc-receptor blockade reduced capture, IgG3 being more dependent on CD16 and IgG1/2 on CD32. Blockade of the integrin CD18 reduced neutrophil spreading, while inhibition of calcium-dependent signalling reduced both capture and spreading, suggesting that both were active processes. Neutrophils treated with ANCA IgG subclasses 1, 3 and 4 showed stabilization of adhesion to P-selectin surfaces and EC. ANCA changed neutrophil behaviour from rolling to static adhesion and the potency of the subclasses followed the same pattern as above: IgG3 > IgG1 > IgG4. Blockade of Fc receptors resulted in neutrophils continuing to roll, i.e. they were not ANCA-activated; differential utilization of Fc receptor by particular IgG subclasses was not as apparent as during neutrophil capture by normal IgG. IgG3 is the most effective subclass for inducing neutrophil adhesion and altered behaviour, irrespective of whether the IgG is surface bound or docks onto neutrophil surface antigens prior to engaging Fc receptors. Engagement of Fc receptors underpins these responses; the dominant Fc receptor depends on IgG subclass.


Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Disclosure
  9. References

Anti-neutrophil cytoplasm antibodies (ANCA) are found in certain small-vessel vasculitides [termed ANCA-associated vasculitis (AAV)] which predominantly affect the kidney and lung, but if untreated can affect most systems in the body causing severe illness and death [1]. From animal models of the disease, myeloperoxidase (MPO)-ANCA are thought to be pathogenic [2–4]. There are two principle forms of pathogenic ANCA that are either directed against the neutrophil serine protease proteinase-3 (PR3) or against MPO [5].

Human IgG has four subclasses, and these have different functions: IgG1 is the predominant class, is directed against proteins and is important in bacterial infection and activation of complement [6]. It binds to the constitutively expressed Fc receptors (FcR) of neutrophils, CD16 (FcγRIIIb) and CD32 (FcγRIIa) and to CD64 (FcγRI) after neutrophil activation [7,8]. CD64 is the only receptor that can bind monomeric soluble IgG [7,8]. IgG2 is directed against polysaccharides and binds via CD32. IgG3 has the largest molecular weight and is distinguished by its long hinge region that allows this protein flexibility; it is directed against proteins and is important in complement activation and can bind CD16, CD32 and CD64 [6]. IgG4 is a smaller protein and is generally thought to bind to neutrophils via CD64 only [7,8], although there is some contrary evidence emerging suggesting that IgG4 can also bind the constitutively expressed receptors [9]. ANCA are predominantly IgG and are found in all four human IgG subclasses. Several papers have discussed the importance of IgG subclass in vasculitis. IgG3 has been proposed to be more pathogenic than the other subclasses in AAV [10], to fall in remission [11] or to predominate in renal-limited AAV [12], and IgG3 has been reported to be predominant in PR3-ANCA patients [13,14]. However, these findings are contradicted by other studies, where IgG1 and IgG4 were dominant in patients with cytoplasm ANCA (largely PR3-ANCA) [15] or overall in AAV [12,13]. In consequence, it is not clear from the published studies which subclasses of ANCA predominate or which are the most pathogenic. Thus the present studies were undertaken to explore the functional and mechanistic effects of IgG subclasses and of ANCA IgG subclasses in particular, on neutrophil adhesion and activation under flow conditions.

When bacterial antigens are opsonized by specific IgG, the antibody Fab portions dock onto the bacterial antigen such that the antibody Fc can be presented to neutrophil FcRs. Binding of IgG to FcR allows downstream signalling and activation of neutrophil phagocytosis. Neutrophil activation requires cross-linking of more than one FcR [16]. In the present studies, antibody Fc was presented to neutrophil FcR as normal IgG coated to solid surfaces. In contrast, ANCA IgG Fab portions dock onto antigens that are actually expressed on neutrophil membranes. There is evidence that concurrent binding of the antibody Fc portions to FcR occurs, thereby cross-linking antigen and FcR on the neutrophils themselves, resulting in dysregulated downstream signalling and activation of the neutrophil [17].

One proposed mechanism for pathogenesis of AAV is that ANCA activates neutrophils and causes or potentiates adherence to endothelium and subsequent damage [18]. In order for flowing neutrophils to be captured onto an endothelial cell (EC) surface, the neutrophil is engaged via selectin molecules enabling rolling on the EC. Provided that an activating stimulus, normally a chemokine but also ANCA in vasculitis, is passed to the neutrophils, stable, static adhesion is mediated through neutrophil integrins (CD18/CD11a and CD18/CD11b) which bind to endothelial intracellular adhesion molecule-1 (ICAM-1) [19]. Transendothelial migration can then take place via a series of molecules such as junctional adhesion molecules (JAM) [20].

Our hypothesis based on the above was that ANCA IgG subclasses have different potencies and consequently different pathogenic effects; moreover, that the subclasses may activate through different FcRs. We examined the effects of ANCA IgG subclasses on neutrophil adhesion to and behaviour on P-selectin or endothelium under conditions of flow. We compared effects of ANCA IgG subclasses with neutrophil capture by purified normal IgG subclasses, whereby both models necessitate engagement of neutrophil FcR. Previous studies have shown that immobilized IgG can capture flowing neutrophils, and have explored the roles of FcRs and integrins, but effects of IgG subclass were not examined [21–23].

Materials and methods

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Disclosure
  9. References

Preparation of chimeric anti-PR3 ANCA and anti-4-hydroxy-3-nitrophenacetyl (NP) controls

Human/mouse chimeric antibodies were generated, of the same PR3 epitope specificity bearing the same mouse V regions and with human IgG1, 3 or 4 subclass constant regions, and characterized as described previously [24]. This anti-PR3 chimeric ANCA recognizes an epitope found in 50% of patient PR3 ANCA in AAV, and is therefore of direct pathological relevance. Anti-NP chimeric antibodies (against an irrelevant antigen) were developed in the same way and used as controls [24,25]. IgG2 ANCA was not developed, therefore was not available for testing. All antibodies used in these experiments were used at 60 µg/ml.

Neutrophil isolation

Blood from healthy volunteers was collected with informed consent and neutrophils were isolated as described previously using a two-step Histopaque density gradient [26]. Following isolation, neutrophils were then resuspended in 0·1% phosphate-buffered saline/bovine serum albumin (PBS/BSA) to a concentration of 1 × 106 cells per ml.

Culture of endothelial cells

Human umbilical vein endothelial cells (HUVEC) were collected from umbilical cords with informed consent, isolated and cultured as described previously [27].

Preparation of adhesive surfaces

Glass microslides (rectangular cross-section 0·3 × 3 mm; length 50 mm; Camlab, Cambridge, UK) were prepared with 3-aminopropyltriethoxysilane (APES) coating [28]. One of the following was then performed:

  • 1
    Coating with normal human IgG of subclasses 1, 2, 3 or 4 (Binding Site, Birmingham, UK) at a concentration of 100 µg/ml, as determined by previous studies, which were left on the slides for 1 h. Any remaining exposed APES was then blocked with 1% albumin (Sigma, Poole, UK) for 2 h.
  • 2
    P-selectin (R&D, Abingdon, UK) coating of microslides at 2 µg/ml was undertaken in a similar manner.
  • 3
    EC were cultured in glass microslides for 24 h prior to use, as described previously [28]. The EC were pretreated with tumour necrosis factor (TNF)-α at 2 units/ml for 4 h and washed fully prior to flowing neutrophils across the monolayer.

Analysis of neutrophil adhesive behaviour under conditions of flow

Microslides were attached to a perfusion system and their surfaces viewed by phase-contrast video-microscopy during perfusion of neutrophils, as described for surfaces coated with IgG [23], P-selectin [29] or TNF-treated EC [27].

Capture and activation of flowing neutrophils by immobilized human IgG of different subclasses.

Neutrophils were perfused for 4 min at a wall shear stress of 0·05 Pa followed by a washout of 1 min. Video recordings of a series of fields along the centre-line were made at 2 min intervals up to 15 min after initial neutrophil inflow. In each recording the number of neutrophils adhered, and the proportion that were spread (becoming phase-dark), or remained spherical and phase-bright were measured.

Activation of neutrophils rolling on P-selectin by soluble ANCA chimeras of different subclasses.

Neutrophils were perfused across the P-selectin surface at a wall shear stress of 0·1 Pa for 4 min. After washout, ANCA IgG of different subclasses were perfused over the rolling neutrophils and video-recordings made as above. The number of neutrophils adhered, and the proportion rolling or stationary were counted at 2-min intervals for 15 min.

Activation of neutrophils adherent to TNF-α-treated HUVEC by soluble ANCA chimeras of different subclasses.

PR3-ANCA chimeric IgG or control anti-NP was added to neutrophils just prior to perfusion for 4 min at 0·1 Pascals. Video recordings were made as above and the number of neutrophils adhered, and the proportion that were rolling, stationary on the endothelial surface (phase-bright) or migrated through the endothelium (spread and phase-dark) were measured.

In some studies, neutrophils were pretreated with intracellular calcium-chelator BAPTA-AM or function-blocking antibodies against CD18 (6·5E murine IgG1; a gift from Dr M. Robinson at Celltech); CD32 (clone IV·3, mouse IgG2a, Stemcell Technologies, Manchester, UK); CD16 (clone 3G8 mouse IgG1, BD Pharmingen, Oxford, UK). Antibodies were added to the neutrophils at 10 µg/ml for 20 min at 37°C prior to perfusion. BAPTA-AM (Sigma, UK) was added at 250 µM to neutrophils for 30 min at 37°C. This concentration was chosen based on the concentration required to inhibit response of rolling neutrophils to N-formyl-methionyl-leucyl-phenylalanine (fMLP). The neutrophils were washed, resuspended and used immediately. We checked the viability of neutrophils after the addition of BAPTA-AM with Trypan blue. In some experiments both anti-CD16 and anti-CD32 were used together. All blocking antibodies were compared with an isotype control antibody.

Statistical analysis

Effects of time and treatment were tested by analysis of variance (anova). Where appropriate, post-hoc comparisons to untreated controls were performed with Dunnett's test (Minitab software programme 13; Bradford, UK).

Neutrophil superoxide production

Superoxide production from neutrophils was assessed as described previously [24].

Results

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Disclosure
  9. References

Capture of flowing neutrophils by normal IgG subclasses presented on a solid-phase support

These experiments explored neutrophil binding to normal polyclonal human IgG under conditions of flow, where capture is supported by neutrophil FcR interacting with IgG Fc that is presented on a surface [23]. ANCA IgG was not used for these experiments, as we wished to explore the effects of Fc binding in the absence of any confounding antigen binding by Fab.

Normal IgG that had been purified into subclasses was coated onto glass microslides and neutrophils flowed across. Many neutrophils bound to IgG3, fewer to IgG1 and only a small number to IgG2 and IgG4 (Fig. 1a). The numbers of neutrophils captured onto these surfaces were significantly different from each other except IgG2 compared to IgG4. Once captured onto the immunoglobulin surface the neutrophils behaved similarly, changing from bright spheres to a phase-dark spread morphology with extended pseudopodia stretching across the surface (e.g. Fig. 1b). Movement appeared to be random on the surface, which was covered uniformly with IgG. There was no difference in the percentage of the neutrophils that spread on the different IgG subclass surfaces.

image

Figure 1. Binding and behaviour of neutrophils on different immunoglobulin (Ig)G subclasses and blockade of CD16 and CD32. (a) Neutrophils were perfused over IgG coated in microslides; the number of neutrophils captured were quantified. Data are mean ± standard error of the mean (s.e.m.) from three experiments. (b) Left-hand panel: neutrophils are adhered but round and have not spread. Right-hand panel: neutrophils have spread out and appear dark. (c) Neutrophils were perfused over IgG coated in microslides after blockade of CD16 or CD32. CD32 was more important with IgG1 and 2 and CD16 more important with IgG3. There was no significant difference on IgG4, as binding was very minimal prior to blocking. Data are mean ± standard error of the mean (s.e.m.) from three experiments.

Download figure to PowerPoint

Neutrophil behaviour was studied after treatment with antibodies to block CD32 and/or CD16 (Fig. 1c). On coated IgG3, both antibodies reduced neutrophil binding compared significantly to control antibody, but anti-CD16 was more effective than anti-CD32. On coated IgG1 and IgG2, anti-CD32 was more effective than anti-CD16. IgG4 supported such minimal capture that no effect of blocking FcR could be observed reliably. On IgG3, where there was residual binding after individual FcR blockade, the use of both blocking antibodies together abrogated neutrophil capture almost completely, with 95 ± 3% inhibition of attachment. Blockade of FcRs also changed the behaviour of the neutrophils once captured. Much less spreading was observed, especially with blockade of CD32, with 70–100% of adherent neutrophils remaining spherical.

When the neutrophils were pretreated with anti-CD18, this did not reduce neutrophil binding to the IgG3 but did reduce neutrophil spreading (Fig. 2) Thus, integrin molecules were not involved in capture but were important in subsequent behaviour of the ‘activated’ cells.

image

Figure 2. Neutrophil adhesion and behaviour on immunoglobulin (Ig)G3-coated surface after blockade of CD18 and neutrophil calcium signalling inhibition with BAPTA-AM. Left-hand panel: blockade of CD18 did not reduce binding; treatment with BAPTA-AM reduced neutrophil binding. Right-hand panel: blocking CD18 reduced the number of neutrophils which spread. Use of BAPTA-AM reduced neutrophil spreading. Data are mean ± standard error of the mean (s.e.m.) from three experiments.

Download figure to PowerPoint

To investigate further the requirement for activation, we inhibited cell signalling by using BAPTA-AM, which chelates intracellular calcium. This reduced the ability of the neutrophil to bind to the surface (Fig. 2), and in addition reduced the ability of bound neutrophils to spread (Fig. 2). Thus, while FcR-IgG interaction directly captured flowing neutrophils, it also transduced a signal which stabilized adhesion through β2-integrins.

Effect of fluid-phase chimeric PR3-ANCA IgG on neutrophils rolling on P-selectin

We then explored the effects of ANCA IgG of different subclasses on neutrophils already rolling on P-selectin in microslides. Typically, activation of such rolling neutrophils causes them to become stably adherent and migrate on the surface [26,29]. Here, soluble ANCA IgG is believed to bind to target PR3 antigen and then cross-link to FcR to transduce an intracellular signal [17]. We have reported previously the effects of the chimeric PR3-ANCA on neutrophils on a P-selectin surface, but here we also looked at blocking to investigate the role of FcRs. Chimeric PR3-ANCA of subclasses IgG1, 3 or 4 or control anti-NP was flowed across the rolling neutrophils and changes in behaviour observed. reported As previously [24], PR3-ANCA IgG1 and 3 converted neutrophils from predominantly rolling (70–80%) to predominantly static adhesion (70–80%), which was significantly different from treatment with control IgG-anti NP where the neutrophils continued to roll. In addition, as reported previously [25], PR3-ANCA IgG4 also reduced the number of neutrophils rolling compared to its control, although this effect was less dramatic than with the other two subclasses.

We explored the role of CD16 and CD32 in the activation of neutrophils rolling on P-selectin after exposure to the different subclasses of chimeric PR3-ANCA (Fig. 3). Blockade of CD32 and CD16 reduced the effect of IgG1 ANCA on the neutrophils; that is, the neutrophil activation induced by ANCA which changed behaviour from rolling to static adhesion was reduced; with anti-CD16 these effects reached significance. For IgG3 PR3-ANCA, blockade of either CD16 or CD32 reduced the effect of the ANCA significantly, allowing neutrophils to continue to roll even when ANCA was present. The influence of FcR blockade on IgG4 was difficult to ascertain as the numbers of neutrophils undergoing stable adhesion with ANCA were low, but blockade of either receptor did increase the proportion of rolling cells. In summary, both IgG1 and IgG3 ANCA potently activated rolling neutrophils with conversion to stable adhesion, whereas IgG4 ANCA was weakly effective. In all cases, the response to ANCA was reduced by blockade of FcR CD16 or CD32.

image

Figure 3. Blockade of CD32 and CD16 on anti-neutrophil cytoplasm antibody (ANCA)-stimulated neutrophils. After exposure to immunoglobulin (Ig)G1 chimeric proteinase 3 (PR3)-ANCA, CD16 prevented ANCA stopping neutrophils more than control (P = 0·028) and CD32 showed a trend. IgG3 chimeric PR3-ANCA was prevented from stopping neutrophils more than control by both CD16 (P = 0·017) and CD32 (P = 0·012). IgG4 chimeric PR3-ANCA was not blocked by either CD16 or CD32, although these again showed a trend. All experiments performed on P-selectin. Neutrophils were exposed to chimeric PR3-ANCA for 2 min. Data are mean ± standard error of the mean (s.e.m.) from three experiments. IC: isotype control.

Download figure to PowerPoint

Effect of chimeric PR3-ANCA IgG on behaviour of neutrophils binding to endothelial cells

Having verified the ability of the chimeric PR3-ANCA to activate neutrophils rolling across P-selectin, we then proceeded to perform flow adhesion experiments using HUVEC, in which neutrophils were mixed with chimeric PR3-ANCA or control antibodies just prior to perfusion. The HUVEC were pre-activated with a low level of TNF-α (2U/ml). In previous experiments, HUVEC not exposed to TNF-α did not capture neutrophils. All three PR3-ANCA IgG subclasses tested (IgG1, IgG3, IgG4) tended to induce greater numbers of neutrophils to bind to the cytokine-activated EC monolayer compared to their equivalent subclass control antibodies (Fig. 4). PR3-ANCA IgG3 was very potent in altering the behaviour of captured neutrophils, causing 94% of neutrophils to stop rolling and convert to stationary adhesion, facilitating transmigration, and this behaviour was different to control antibody where 29% neutrophils became stationary after exposure to the control antibody (Fig. 4). The same was true of PR3-ANCA IgG1, but the affect was less potent (58% of neutrophils converted to static adhesion compared to 21% with control antibody) (Fig. 4). PR3-ANCA IgG4 also tended to convert rolling to static adhesion, but this change in behaviour did not reach statistical significance (Fig. 4). In general, ANCA therefore increased overall binding and tended to stabilize adhesion. IgG3 ANCA was more potent than IgG1 which was more potent than IgG4.

image

Figure 4. Neutrophils binding to and behaviour on human umbilical vein endothelial cells. Neutrophils were flowed across human umbilical vein endothelial cells which had been stimulated with tumour necrosis factor (TNF)-α at 2 units/ml. Left-hand panel: immunoglobulin (Ig)G1, 3 and 4 anti-proteinase 3 (PR3) ANCA captured neutrophils efficiently and IgG4 captured neutrophils more than its control (P = 0·048). Right-hand panel: IgG3 anti-PR3 ANCA efficiently stopped neutrophils from rolling compared to control; IgG1 stopped neutrophils, but to a lesser extent, and IgG4 stopped neutrophils, but this was not significantly different to its control. Data are mean ± standard error of the mean (s.e.m.) from three experiments.

Download figure to PowerPoint

Neutrophil superoxide production

In order to confirm whether IgG subclass had an effect on neutrophil function other than adhesion, superoxide production was also investigated. Superoxide production was induced by all three IgG ANCA subclasses investigated (compared to control anti-NP antibodies). IgG1 and IgG3 ANCA produced similar levels of superoxide, and this was significantly greater than IgG4 superoxide production (Fig. 5), suggesting that the IgG1 and IgG3 subclasses do indeed have a greater ability to activate neutrophils.

image

Figure 5. Superoxide production by neutrophils after exposure to anti-neutrophil cytoplasm antibody (ANCA) immunoglobulin (Ig)G subclasses 1, 3 and 4 or controls. Neutrophils were exposed to ANCA IgG subclasses 1, 3 and 4 or to control anti-4-hydroxy-3-nitrophenacetyl (NP) antibodies and levels of superoxide production measured over time to 120 min (each bar represents a 10 min reading) All three ANCA caused significant superoxide production above control anti-NP [all P ≤ 0·01 by two-way analysis of variance (anova)]. When compared to each other, IgG1 and 3 were not different, but these were both significantly greater at stimulating superoxide production than IgG4 (both P < 0·01 by two-way anova) (n = 5). *P < 0·05; **P < 0·01 compared to control. IgG1 compared to IgG3 not significantly different; both IgG1 and IgG3 significantly greater than IgG4 (P = 0·01).

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Disclosure
  9. References

This study examined whether different ANCA subclasses have different effects pertinent to pathogenicity. We have shown for the first time that the ability of ANCA to change neutrophil behaviour on an endothelial surface is dependent on IgG subclass, and IgG3 is more potent at activating and stopping rolling neutrophils than IgG1, which is more potent than IgG4. All three ANCA subclasses tested were capable of activating neutrophils. In addition we have shown that both CD16 and CD32 play a role in this activation. We have also demonstrated that flowing neutrophils interacting with IgG-coated surfaces are captured most efficiently onto IgG3, but are also captured onto IgG1. Specifically, freshly isolated neutrophils that had not been activated bound to IgG presented on surfaces, and this interaction followed the conventional paradigm that neutrophils will interact with IgG3 and IgG1 and minimally or not at all with IgG2 and IgG4 [30]. The greater capture of neutrophils by IgG3 that we observed may reflect the enhanced flexibility of the larger IgG3 molecule. It is unlikely that the enhanced neutrophil binding to IgG3 was due to a higher concentration of this subclass on the glass APES-coated slide, as the antibodies were coated at equal concentrations and thus at lower molarity for the larger IgG3 molecules. The differential capture by the antibodies is interesting particularly as, in disease states, IgG3 autoantibodies have been proposed to be more pathogenic [10].

Having established that flowing neutrophils could be captured well by IgG3 and IgG1 Fc, the specific neutrophil FcRs supporting this were investigated using blocking antibodies to CD16 and CD32, and differential utilization of these receptors was observed. Thus, CD16 was more important in IgG3-mediated adhesion under flow, while CD32 was more important for IgG1- and 2-mediated adhesion. In previous studies using unfractionated human IgG, blockade of either CD32 or CD16 reduced neutrophil capture but CD16 was more important [23]. This may indicate that IgG3 is predominant in capture in a mixed IgG model. As in the present study, blockade of both FcRs completely abolished capture. Of note, we are confident that these neutrophils do not express inducible FcR CD64, as this receptor is induced only after 14 h treatment with interferon (IFN)-γ[31].

Other studies have found that binding of subclasses to neutrophil CD32 was highest with IgG3, then IgG1 and 2 and then IgG4. This was dependent, however, on allotypic form [32]. The present study has directly compared the relative importance of CD32 and CD16 for neutrophil capture and adherence to IgG subclasses under flow conditions and has shown clearly that CD16 is more important for IgG3 mediated adhesion, while CD32 is more important for IgG1- and 2-mediated adhesion. Interestingly, in a previous study that examined soluble immune complex binding to neutrophils via CD16 and CD32, granule release by neutrophils was induced by all four IgG subclasses in the presence of complement, but in the absence of complement only IgG1 and IgG3 were functionally active. Granule release by binding of IgG3 immune complexes was mediated predominantly by CD16, whereas IgG1 immune complexes induced specific granule release by CD32, suggesting similar engagement of FcRs by IgG subclasses for granule release as was observed for capture and adhesion of neutrophils from flow [33].

We next examined the role of β2 integrin molecules in neutrophil stabilization and adhesion using antibodies to the common CD18 chain [34]. β2 integrins were important in allowing the neutrophils to spread on the IgG surface. This finding is supported by previous published work [23]. There are two possible explanations for this observation; the first is that the FcR ligation induces activation of the integrin which binds to a ligand (albumin in this case [35]) and outside-in signalling results in neutrophil spreading. The second is that the FcR ligation itself causes the neutrophil to spread while simultaneously activating β2 integrins which are required to adhere to the surface. When neutrophil signalling was inhibited by chelating the intracellular calcium with BAPTA-AM, binding and spreading were reduced. The CD18 blocking studies indicate that spreading requires signalling within the neutrophil but binding, too, is clearly a metabolically active process and not simply mechanical; signalling is required to stabilize adhesion.

After establishing that solid-phase normal human immunoglobulin IgG subclasses recruit neutrophils from flow differentially by engagement with their constitutive CD16 and CD32 receptors, we wished to compare this with chimeric PR3-ANCA IgG subclass antibodies that were developed in our laboratory and were all directed against the same epitope of PR3. The ANCA IgG had to be present in the fluid phase to enable antibody Fab and Fc portions to engage concurrently with neutrophil antigen and with FcRs [17]. The question was whether, with this added complexity, neutrophil responses differed from those observed following cross-linking of neutrophil FcRs alone. We believe that both Fc and Fab portions of the antibody are required for ANCA to activate the neutrophil, and this is based on previous evidence from our group [17]. F(ab)2 fragments alone were unable to stabilize rolling neutrophils on EC in previous studies [36]. We have evidence to show that neutrophil activation caused by cross-linking of antigen target and FcR by Fab and Fc, respectively, of a single IgG molecule to be a unique feature of ANCA; individually, ANCA IgG Fab or Fc fragments are unable to activate neutrophils (data has been submitted for publication).

Initially we flowed neutrophils over a P-selectin surface, as we have used previously [24,25]. We verified that most of the captured neutrophils roll on this surface, but with the addition of a second signal such as ANCA IgG the rolling is converted to static adhesion. This was true of IgG1 and IgG3 anti-PR3 ANCA [24], but also was seen with PR3-ANCA IgG4 [25] although to a lesser extent. IgG4 has been thought not to engage CD32 or CD16 [7,8,37], although more recent evidence suggests that this may be an over-simplification [33,38]. In addition, IgG4 does not always behave as a conventional antibody; for example, IgG4 bispecific antibodies can be found in plasma [39]. In our study, flow cytometric analysis demonstrated that CD64 was not present on the surface of freshly isolated neutrophils (data not shown), so PR3-ANCA IgG4 must be able to engage the constitutively expressed receptors. Indeed, previous studies demonstrated that chimeric PR3-ANCA IgG4 stimulation of superoxide release and neutrophil degranulation is dependent on CD32 [25].

Blocking CD16 and CD32 on chimeric PR3-ANCA-stimulated neutrophils demonstrated that both FcRs appeared to play a role when ANCA transduced an activation signal, with a suggestion that CD16 may have a more important role, particularly in the presence of IgG1 ANCA. Thus, receptor usage and the signal transduction that is required to enable neutrophil capture from flow where neutrophils are not rolling may differ from that in ANCA-induced activation of stable adhesion of rolling cells on endothelium.

P-selectin is a simpler stratum for rolling neutrophils than endothelium, where chemokines and adhesion molecules may alter responses. Nevertheless, ANCA could stabilize binding of neutrophils and markedly change the behaviour of neutrophils on a cytokine-activated endothelial surface. The greater potency of the ANCA IgG3 subclasses paralleled the findings on normal coated IgG and on P-selectin but, importantly, shows that they also relevant to a more physiological substratum. Further, there is the implication that IgG3 is a more pathogenic subclass in AAV [10], and that IgG4 may not be as inert as considered previously, although pathogenic outcome will depend on the overall composition of the ANCA IgG subclasses between individuals. IgG3 ANCA activates neutrophils more strongly, causing greater adhesion on HUVEC when compared to IgG1 ANCA. The explanation for this may be that IgG3 is a more flexible molecule by virtue of its long hinge region, thereby allowing more effective cross-linking of antigen and FcR enabling IgG3 to have a greater binding affinity to the neutrophil than IgG1. To explore this possibility further would require molecular engineering of the Fc hinge region. Alternatively, the explanation may lie in differing affinities of the Fc regions for the FcR. When tested previously, the affinities of PR3-ANCA for antigen binding were similar for IgG1 and IgG3 [24], but there is evidence that affinity of Fc regions for FcR may vary according to subclass [40,41] and that affinity of IgG3 Fc for FcR may be higher than of IgG1 [42]. Of note in our experiments, the difference seen between IgG1 and IgG3 ANCA on HUVEC was not seen on P-selectin. This is likely to be because the P-selectin surface presents much more selectin to the neutrophil than an endothelial surface and is consequently more efficient at neutrophil capture. As all the neutrophils stopped on IgG1, it is not possible to see an increased number stopping on IgG3. There is evidence that the adhesion molecule PSGl-1 can operate as a signalling molecule, and this signal is much stronger on a P-selectin surface than on EC [43]. The presence of living EC may alter the signals of engagement, thereby resembling more closely the in vivo situation. These factors may be more conducive to a heightened response from IgG3.

In conclusion, we report that IgG3 is the most potent subclass for neutrophil capture from flow and induction of adhesion, whether considering capture of neutrophils to coated IgG surfaces presenting cross-linked Fc portions alone or the effects of ANCA on neutrophils rolling on activated endothelium. CD16 and CD32 FcR are pivotal in these responses, although the particular receptor engaged depends on the subclass and the context. Targeting particular IgG subclasses or FcRs may be useful in designing therapeutic intervention for AAV.

Acknowledgements

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Disclosure
  9. References

This work was supported by a grant from the Wellcome Trust. We are grateful to Birmingham Women's Hospital NHS Trust for source of umbilical cords.

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Disclosure
  9. References