Molec.AspectsMed.Vol. 12, pp. 107-119, 1991
0098-2997/91 $0.00 + .50 © 1991 Pergamon Press plc.
Printed in Great Britain. All rights reserved.
EFFECTS OF REACTIVE OXYGEN SPECIES ON IMMUNOGLOBULIN G FUNCTION Helen R. Griffiths and Joseph Lunec Wolfson Research Laboratories, Queen Elizabeth Hospital Birmingham, B15 2TH, U.K.
Introduction Much of the pioneering work on antibody structure was performed by Porter (1959) using the proteolytic enzymes pepsin and papain. Figure I illustrates the fragments formed by these enzymes and these have been of enormous value in elucidating the structure/function relationships within IgG. The primary function of IgG is regarded as being the interaction with and clearance of antigen, the functional site residing in the Fab arms. However, each immunoglobulin class also participates in equally important class-restricted effector functions (reviewed by Burton, 1985). These effector sites are typically located within the Fc portion of the molecule. Over the last ten years, considerable advances have been made in determining the structure of Fc and computer modelling systems are now available based on the original crystallographic work of Deisenhofer etal. (1976; 1981). This has facilitated further studies localising specific sites of functional interaction with Fc. Atassi (1975) has also made some important contributions towards our current theories on the nature of antigenic determinants, based on his work on determining the antigenic sites within myoglobin. From these studies, he has proposed that for interaction with antibody, antigenic sites should occupy accessible surface locations and that hydrophobic sites are likely to be of particular importance: the energy required to drive subsequent interaction coming from the shielding of hydrophobic residues from solvent. These principals are equally applicable to the study of any protein-protein interaction, such as the binding of IgG to protein A, rheumatoid factor, complement component Clq or cellular receptors: an approach which has been pursued in detail by Burton (1985). In favour of this hypothesis it has been found that within IgG-Fc, there are hydrophobic patches which are highly conserved, implicating them in a functional role, whether it be molecular stabilisation or interaction with other proteins.
IgG Dysfunction in Rheumatoid Arthritis A characteristic feature of rheumatoid arthritis is an abnormality in the humoral response, as illustrated by the presence of circulating autoantibodies; immunoglobulins directed against intra
107
108
H.R. Griffiths and J. Lunec
and extra-cellular self components such as DNA, collagen and IgG. Autoantibodies to IgG, termed rheumatoid factors (RFs), were first described 50 years ago by Waaler (1939), who demonstrated the agglutination of IgG sensitised sheep red blood ceils in the presence of rheumatoid serum. However, these IgM-FRs are not strictly true autoantibodies since they have a poor reactivity with native IgG (Ka = 104-105), but have a higher affinity for heterologous IgG such as rabbit or baboon (Eisenberg, 1976; Rialdi etal, 1984). Indeed, these allogenic lgGs are frequently used for measurements of true RFs since they show less cross-reactivity with RFs detected in nonrheumatoids. This suggests that there is a fundamental difference in th nature of the antigen recognised by rheumatoids RFs.
H
H
peps,n
Papain
341"
433.
I
=
2* cleavage point
=
2* cleavage point
234 I
'
~pFc
Fig. l. The definition of the major proteolyticsubfragments of IgG. Several mechanisms have been postulated for the generation of rheumatoid factors and genetic studies on the V kappa gene have established that the production of the diverse RF population is an antigen driven T-cell dependent process (Schlomick et al., 1986; Carson et al., 1987). However, the antigen remains elusive. IgG is reported to occur as an aggregated form in the synovial fluid of rheumatoid patients and it is well documented that oxygen radicals can induce protein aggregation. We have therefore investigated the hypothesis that reactive oxygen species (ROS) generated by activated phagocytic
Reactive Oxygen Species and IgG Function
109
cells at inflammatory sites may denature IgG and thereby produce a more antigenic, autorcactive form of IgG. The primary ROS produced by activated PMNs in the respiratory burst is the superoxide anion, which may spontaneously dismutate to H202", a reaction catalysed by superoxide dismutase. In the presence of catalytic metal ions or myeloperoxidase, H202 can be further converted to the ROS, hydroxyl radical (OH') or hypochlorite anion (OCI') respectively. Of these OH" is believed to be the most reactive, however, the ultimate toxicant associated with protein denaturation in vivo remains uncertain. In the following paper we will summarise a detailed study of the effects of ROS on some of the functional interactions of the IgG molecule. In this way we are attempting to define whether particular regions of the IgG molecule are more susceptible to denaturation by ROS and whether one or more specific ROS can induce alterations in IgG typical of those seen in rheumatoid arthritis (RA). To achieve this aim, both the physiochemical and functional properties of ROS-denatured IgG have been compared to those of IgG isolated from normal sera, RA sera and hyaluronidase treated RA synovial fluids (SFs) according to the method of McKinney et al. (1988). Monomeric IgG was further purified from high and low molecular weight contaminants by gel permeation chromatography. In Vitro Denaturation of IgG A variety of different methods are available for the generation of oxygen radicals in vitro, which differ in both their complexity and relevance to biological systems. Whilst the simpler chemical systems allow the generation of defined radical species in solution, they bear little resemblance to ongoing free radical events in inflammation. Therefore several systems have been employed to compare the effects of pure radiolytically generated radicals in solution, when compared to other complex, but more physiological and pathological relevant systems, such as activated peripheral blood polymorphs (PMNs). In the text, denatured IgG will be referred to as IgG*.
Macromolecular Modifications to ROS Denatured IgG All proteins exhibit a native fluorescence spectrum in the ultraviolet region, conferred largely by the presence of aromatic amino acids. Following mixed oxygen radical attack, a decrease in the UV fluorescence has been reported for several proteins, including, IgG gamma lens crystallins, transferrin and superoxide dismutase (Wickens et al., 1983; Bodanes et al., 1985; Davies, 1987). Associated with this loss, a dose dependent increase in visible fluorescence has been described (Jones and Lunec, 1987). The induction of visible fluorescence is accompanied by significant physical changes to the protein, the nature of these changes reflecting both the radical species and the characteristic of different protein biomolecules. Since the consequence of oxidative damage varies with different ROS and because activated PMNs generate several different ROS, we have studied the effects of: (1)
OH" generation by H202 in the presence of CulI ions according to Fenton chemistry H202 + Cull ---) CuI + HO 2" + H + CuI + H202 ~ Cull+ OH- + OH"
(2)
radiolytic generation of OH" and O~ by the radiolysis of water in the presence of oxygen, H20 ---)OH" + H30 + eaq
110
H.R. Griffiths and J. Lunec
02 + eaq --~ O~ or racliolytic generation of O~ alone in the presence of 20raM formate, OH" + HCO 2- --) H20 + CO 2" CO 2" + 02 ~ O~ + CO 2
and (3)
peripheral blood PMNs activated in vitro by phorbol myristate acetate, on the macromolecular and fluorescence properties of IgG. The yields of ROS generated by each system are summarised in the Table below:
Dose
OH'(IxM)
O~ (laM)
5p.M Cu/50 IxM H202
50
100krad water
400
200
100krad formate
0
600
PMNs (106 cells) (60 mins)
200
400
We have previously described (Griffiths etal., 1988 a,b) that OH" but not O~ can induce autofluorescence and aggregation in IgG. In the presence of CulI ions, IgG does not undergo any physiological changes, but following the addition of H20 2 autofluorescence and fragmentation occur; fragmentation being observed in the presence of as low as 0.5 gM Cu and 501xM H20 2 (see Fig. 2). Following exposure to activated polymorphs, autofluorescence characteristic of OH" attack can be seen and fragmentation of IgG is observed in a manner similar to that seen after Cu/H20 2 attack (Griffiths, 1989). The importance of metal ions in IgG denaturation by activated PMNs was confirmed by inhibition experiments; in the presence of desferrioxamine, a high affinity metal ion chelator (Ka = 1030), fluorescence formation was reduced by 46%. For a free radical mechanism to be biologically relevant, it must proceed at a physiologically significant rate. Using pulse radiolysis to determine the rate constant of reaction between OH" and IgG, we have calculated K = 2x1011 1/mol/sec (Griffiths, 1989); i.e. as fast as the rate of diffusion. Whilst there is debate about the relevance of such high rate constants for biological molecules, it is apparent from other published data, that there is a linear relationship between protein MW and rate of reaction with OH'. One possible interpretation for this is that OH" produced at a site of inflammation will react immediately with the largest protein present in highest concentration, and owing to their high reactivity, they rarely diffuse more than 14A from their site of generation. In
Reactive Oxygen Species and IgG Function
111
the inflamed joint, IgG is the major protein; up to 90 mg can be synthesised daily by actively secreting plasma cells (Smiley et al., 1968). Thus IgG may be a selective target for radicals produced within the rheumatoid synovium.
Reactivity of IgG with Protein A Protein A (SPA) is a major cell wall component of the bacterial strain Staphylococcus aureus, which binds immunoglobulins from a variety of different species with differing avidities. Whilst the physiological importance of the interaction between these proteins is not understood, it nonetheless provides a useful tool for studying functional changes in IgG (Goding, 1978; Langone, 1982). Sequence analysis of the contact residues involved in the binding of human IgG subclasses to SPA has revealed a surprising homology and conservation of hydrophobic residues into two patches. Further supportive evidence for the importance of these particular regions has come from the work of Nik Jafaar et al. (1984) who defined the topographical distribution of Fcy receptors using a panel of 24 monoclonal antibodies and defined Histidine 435 as an essential residue for the expression of CH2-CH3 interdomaln epitopes. Secondly, the importance of hydrophobic patch 3 of the CH 3 domain is implied by the non-reactivity of IgG 3m (5) and (21) with SPA; in these genetic allotypes of the IgG3 subclass, Arg is substituted for His at position 435, and Phe for Tyr at position 436. This latter allotype is very closely related to the RF binding site and can successfully compete with the interaction between monoclonal RF and IgG (Sasso et al., 1988).
/
o F/UVratio xl00
100 I-
I >
x Monomer
90I-
+ Breakdown ~ Aggregate
701--~x -
^~"--1 u
20
I
40
I
60
I
80
I
100
I
120
I
140
I
160
I
180
I
200
H 2 0 2 concentration (pM)
Fig. 2. Dose dependent changes in the physicochemicalpropertiesof IgG following exposureto Cu-H202. The CH2-CH3 interface and CH 3 domain are comparatively rigid parts of the IgG molecule and their functions (unlike the CH 2 domain) are unaffected by changes in the Fab arms. Thus any ROS induced functional changes are likely to be due to local modification to essential residues. The ability of IgG* to bind to SPA was assessed using a modified ELISA procedure, which was validated by studying the effects of heat aggregation on IgG-SPA interaction; using this system, we were able to confn'rn the findings of Goding (1978), that heat aggregation facilitates IgG-SPA binding. The SPA reactivity of IgG isolated from biofluids was found to be related to the source of the protein; rheumatoid serum I g G was slightly but not significantly more reactive with SPA than IgG from elderly control sera. However, rheumatoid synovial fluid IgG showed a significant 25%
112
H. R. Griffiths a n d J. L u n e c
increase in binding (p
0098-2997/91 $0.00 + .50 © 1991 Pergamon Press plc.
Printed in Great Britain. All rights reserved.
EFFECTS OF REACTIVE OXYGEN SPECIES ON IMMUNOGLOBULIN G FUNCTION Helen R. Griffiths and Joseph Lunec Wolfson Research Laboratories, Queen Elizabeth Hospital Birmingham, B15 2TH, U.K.
Introduction Much of the pioneering work on antibody structure was performed by Porter (1959) using the proteolytic enzymes pepsin and papain. Figure I illustrates the fragments formed by these enzymes and these have been of enormous value in elucidating the structure/function relationships within IgG. The primary function of IgG is regarded as being the interaction with and clearance of antigen, the functional site residing in the Fab arms. However, each immunoglobulin class also participates in equally important class-restricted effector functions (reviewed by Burton, 1985). These effector sites are typically located within the Fc portion of the molecule. Over the last ten years, considerable advances have been made in determining the structure of Fc and computer modelling systems are now available based on the original crystallographic work of Deisenhofer etal. (1976; 1981). This has facilitated further studies localising specific sites of functional interaction with Fc. Atassi (1975) has also made some important contributions towards our current theories on the nature of antigenic determinants, based on his work on determining the antigenic sites within myoglobin. From these studies, he has proposed that for interaction with antibody, antigenic sites should occupy accessible surface locations and that hydrophobic sites are likely to be of particular importance: the energy required to drive subsequent interaction coming from the shielding of hydrophobic residues from solvent. These principals are equally applicable to the study of any protein-protein interaction, such as the binding of IgG to protein A, rheumatoid factor, complement component Clq or cellular receptors: an approach which has been pursued in detail by Burton (1985). In favour of this hypothesis it has been found that within IgG-Fc, there are hydrophobic patches which are highly conserved, implicating them in a functional role, whether it be molecular stabilisation or interaction with other proteins.
IgG Dysfunction in Rheumatoid Arthritis A characteristic feature of rheumatoid arthritis is an abnormality in the humoral response, as illustrated by the presence of circulating autoantibodies; immunoglobulins directed against intra
107
108
H.R. Griffiths and J. Lunec
and extra-cellular self components such as DNA, collagen and IgG. Autoantibodies to IgG, termed rheumatoid factors (RFs), were first described 50 years ago by Waaler (1939), who demonstrated the agglutination of IgG sensitised sheep red blood ceils in the presence of rheumatoid serum. However, these IgM-FRs are not strictly true autoantibodies since they have a poor reactivity with native IgG (Ka = 104-105), but have a higher affinity for heterologous IgG such as rabbit or baboon (Eisenberg, 1976; Rialdi etal, 1984). Indeed, these allogenic lgGs are frequently used for measurements of true RFs since they show less cross-reactivity with RFs detected in nonrheumatoids. This suggests that there is a fundamental difference in th nature of the antigen recognised by rheumatoids RFs.
H
H
peps,n
Papain
341"
433.
I
=
2* cleavage point
=
2* cleavage point
234 I
'
~pFc
Fig. l. The definition of the major proteolyticsubfragments of IgG. Several mechanisms have been postulated for the generation of rheumatoid factors and genetic studies on the V kappa gene have established that the production of the diverse RF population is an antigen driven T-cell dependent process (Schlomick et al., 1986; Carson et al., 1987). However, the antigen remains elusive. IgG is reported to occur as an aggregated form in the synovial fluid of rheumatoid patients and it is well documented that oxygen radicals can induce protein aggregation. We have therefore investigated the hypothesis that reactive oxygen species (ROS) generated by activated phagocytic
Reactive Oxygen Species and IgG Function
109
cells at inflammatory sites may denature IgG and thereby produce a more antigenic, autorcactive form of IgG. The primary ROS produced by activated PMNs in the respiratory burst is the superoxide anion, which may spontaneously dismutate to H202", a reaction catalysed by superoxide dismutase. In the presence of catalytic metal ions or myeloperoxidase, H202 can be further converted to the ROS, hydroxyl radical (OH') or hypochlorite anion (OCI') respectively. Of these OH" is believed to be the most reactive, however, the ultimate toxicant associated with protein denaturation in vivo remains uncertain. In the following paper we will summarise a detailed study of the effects of ROS on some of the functional interactions of the IgG molecule. In this way we are attempting to define whether particular regions of the IgG molecule are more susceptible to denaturation by ROS and whether one or more specific ROS can induce alterations in IgG typical of those seen in rheumatoid arthritis (RA). To achieve this aim, both the physiochemical and functional properties of ROS-denatured IgG have been compared to those of IgG isolated from normal sera, RA sera and hyaluronidase treated RA synovial fluids (SFs) according to the method of McKinney et al. (1988). Monomeric IgG was further purified from high and low molecular weight contaminants by gel permeation chromatography. In Vitro Denaturation of IgG A variety of different methods are available for the generation of oxygen radicals in vitro, which differ in both their complexity and relevance to biological systems. Whilst the simpler chemical systems allow the generation of defined radical species in solution, they bear little resemblance to ongoing free radical events in inflammation. Therefore several systems have been employed to compare the effects of pure radiolytically generated radicals in solution, when compared to other complex, but more physiological and pathological relevant systems, such as activated peripheral blood polymorphs (PMNs). In the text, denatured IgG will be referred to as IgG*.
Macromolecular Modifications to ROS Denatured IgG All proteins exhibit a native fluorescence spectrum in the ultraviolet region, conferred largely by the presence of aromatic amino acids. Following mixed oxygen radical attack, a decrease in the UV fluorescence has been reported for several proteins, including, IgG gamma lens crystallins, transferrin and superoxide dismutase (Wickens et al., 1983; Bodanes et al., 1985; Davies, 1987). Associated with this loss, a dose dependent increase in visible fluorescence has been described (Jones and Lunec, 1987). The induction of visible fluorescence is accompanied by significant physical changes to the protein, the nature of these changes reflecting both the radical species and the characteristic of different protein biomolecules. Since the consequence of oxidative damage varies with different ROS and because activated PMNs generate several different ROS, we have studied the effects of: (1)
OH" generation by H202 in the presence of CulI ions according to Fenton chemistry H202 + Cull ---) CuI + HO 2" + H + CuI + H202 ~ Cull+ OH- + OH"
(2)
radiolytic generation of OH" and O~ by the radiolysis of water in the presence of oxygen, H20 ---)OH" + H30 + eaq
110
H.R. Griffiths and J. Lunec
02 + eaq --~ O~ or racliolytic generation of O~ alone in the presence of 20raM formate, OH" + HCO 2- --) H20 + CO 2" CO 2" + 02 ~ O~ + CO 2
and (3)
peripheral blood PMNs activated in vitro by phorbol myristate acetate, on the macromolecular and fluorescence properties of IgG. The yields of ROS generated by each system are summarised in the Table below:
Dose
OH'(IxM)
O~ (laM)
5p.M Cu/50 IxM H202
50
100krad water
400
200
100krad formate
0
600
PMNs (106 cells) (60 mins)
200
400
We have previously described (Griffiths etal., 1988 a,b) that OH" but not O~ can induce autofluorescence and aggregation in IgG. In the presence of CulI ions, IgG does not undergo any physiological changes, but following the addition of H20 2 autofluorescence and fragmentation occur; fragmentation being observed in the presence of as low as 0.5 gM Cu and 501xM H20 2 (see Fig. 2). Following exposure to activated polymorphs, autofluorescence characteristic of OH" attack can be seen and fragmentation of IgG is observed in a manner similar to that seen after Cu/H20 2 attack (Griffiths, 1989). The importance of metal ions in IgG denaturation by activated PMNs was confirmed by inhibition experiments; in the presence of desferrioxamine, a high affinity metal ion chelator (Ka = 1030), fluorescence formation was reduced by 46%. For a free radical mechanism to be biologically relevant, it must proceed at a physiologically significant rate. Using pulse radiolysis to determine the rate constant of reaction between OH" and IgG, we have calculated K = 2x1011 1/mol/sec (Griffiths, 1989); i.e. as fast as the rate of diffusion. Whilst there is debate about the relevance of such high rate constants for biological molecules, it is apparent from other published data, that there is a linear relationship between protein MW and rate of reaction with OH'. One possible interpretation for this is that OH" produced at a site of inflammation will react immediately with the largest protein present in highest concentration, and owing to their high reactivity, they rarely diffuse more than 14A from their site of generation. In
Reactive Oxygen Species and IgG Function
111
the inflamed joint, IgG is the major protein; up to 90 mg can be synthesised daily by actively secreting plasma cells (Smiley et al., 1968). Thus IgG may be a selective target for radicals produced within the rheumatoid synovium.
Reactivity of IgG with Protein A Protein A (SPA) is a major cell wall component of the bacterial strain Staphylococcus aureus, which binds immunoglobulins from a variety of different species with differing avidities. Whilst the physiological importance of the interaction between these proteins is not understood, it nonetheless provides a useful tool for studying functional changes in IgG (Goding, 1978; Langone, 1982). Sequence analysis of the contact residues involved in the binding of human IgG subclasses to SPA has revealed a surprising homology and conservation of hydrophobic residues into two patches. Further supportive evidence for the importance of these particular regions has come from the work of Nik Jafaar et al. (1984) who defined the topographical distribution of Fcy receptors using a panel of 24 monoclonal antibodies and defined Histidine 435 as an essential residue for the expression of CH2-CH3 interdomaln epitopes. Secondly, the importance of hydrophobic patch 3 of the CH 3 domain is implied by the non-reactivity of IgG 3m (5) and (21) with SPA; in these genetic allotypes of the IgG3 subclass, Arg is substituted for His at position 435, and Phe for Tyr at position 436. This latter allotype is very closely related to the RF binding site and can successfully compete with the interaction between monoclonal RF and IgG (Sasso et al., 1988).
/
o F/UVratio xl00
100 I-
I >
x Monomer
90I-
+ Breakdown ~ Aggregate
701--~x -
^~"--1 u
20
I
40
I
60
I
80
I
100
I
120
I
140
I
160
I
180
I
200
H 2 0 2 concentration (pM)
Fig. 2. Dose dependent changes in the physicochemicalpropertiesof IgG following exposureto Cu-H202. The CH2-CH3 interface and CH 3 domain are comparatively rigid parts of the IgG molecule and their functions (unlike the CH 2 domain) are unaffected by changes in the Fab arms. Thus any ROS induced functional changes are likely to be due to local modification to essential residues. The ability of IgG* to bind to SPA was assessed using a modified ELISA procedure, which was validated by studying the effects of heat aggregation on IgG-SPA interaction; using this system, we were able to confn'rn the findings of Goding (1978), that heat aggregation facilitates IgG-SPA binding. The SPA reactivity of IgG isolated from biofluids was found to be related to the source of the protein; rheumatoid serum I g G was slightly but not significantly more reactive with SPA than IgG from elderly control sera. However, rheumatoid synovial fluid IgG showed a significant 25%
112
H. R. Griffiths a n d J. L u n e c
increase in binding (p
