8 Cytokine abnormalities in inflammatory arthritis P. MIOSSEC
Cells interact through direct cell-to-cell contact and through soluble factors acting as local and systemic signals. Cytokines are soluble molecules which control communication between cells in a non-antigen-specific manner. They act locally as paracrine or autocrine factors, and some are also detected systematically, where they act as endocrine factors. They contribute to many biological activities and defence mechanisms. The specificity of their activity is related to their binding to specific membrane receptors. Research in the field of cytokines has been growing extensively over recent years because of advances in immunology and cellular and molecular biology. These methods have allowed the production of large amounts of pure recombinant products for research and, more recently, for clinical applications. Similarly, the number of studies on the role of cytokines in the pathogenesis of inflammatory arthritis expanded with the discovery of new cytokines and the development of assays for their detection (Arend and Dayer, 1990; Harris, 1990; Ziff, 1990). The rationale for the use of cytokines and for drug targeting of the cytokine network in conditions such as rheumatoid arthritis lies in the central role they play in inflammation and autoimmunity. This review will summarize the methods of cytokine detection and then concentrate on the complex balance of the cytokine network and its dysregulation in rheumatoid arthritis.
DETECTION OF CYTOKINES From basic research on cytokines, assays for cytokine detection have been developed for clinical studies. Indeed, results obtained in vitro suggest that these measurements might have both diagnostic and prognostic value (Whicher and Ingham, 1990; Feldman et al, 1991). However, with few exceptions, the usefulness of routine cytokine measurements is still under investigation. Unfortunately, it remains difficult to compare results from the literature, in part because of large variations in sensitivity and specificity between these assays. Synthesis of a cytokine includes numerous steps from its gene activation to the release of the active protein. Each of these steps can now be studied in Baillikre' s Clinical Rheumatology--
Vol. 6, No. 2, June 1992 ISBN 0-7020-1636-5
373 Copyright9 1992,byBailli6reTindall Allrightsofreproductionin anyformreserved
374
P. MIOSSEC
detail with tools derived from basic research. However, the most important factor remains the release of the active protein. Indeed, detection of messenger RNA (mRNA) is not always associated with cytokine secretion when post-transcriptional regulation is present. In addition, following its release, a cytokine can also interact with various regulatory mechanisms before reaching an active membrane receptor.
Assays for cytokines Assays for cytokines can be classified into two major groups (Whicher and Ingham, 1990). Biological assays measure the activity of a cell system. They were the first type of assay to be described, as cytokines were discovered on the basis of their biological activity. They are sensitive and demonstrate the presence of biological activity. Their specificity is usually limited because of the ple!o[_ropi c act!.vity of cytokines, but this limitation can be improved with the use of specific antibodies. The mouse thymocyte proliferation assay was first described as an interleukin (IL)-I specific assay, but was later also shown to respond to TNFo~ (tumour necrosis factor), IL-6, IL-2 and IL-4. Similarly, the CTTL-2 cell line used in the first IL-2 biological assay can also respond in some conditions to IL-4. Cytokines can also act indirectly through the release of other factors which are part of a cascade of secretory events. For example, in the thymocyte proliferation assay, IL-1 induces the release of IL-2, which is responsible for thymocyte proliferation. Furthermore, complex biological fluids can induce specific interference with anticytokine antibodies and soluble receptors. In addition, potent non-specific inhibitory factors are often present in human biological fluids, some of these being toxic for mouse cell lines. Purification procedures are thus necessary to demonstrate the presence of a biologically active molecule (Miossec et al, 1986). Immunoassays detect cytokines as antigenic proteins recognized by specific polyclonal and/or monoclonal antibodies. The most common combination is the double sandwich enzyme-linked immunosorbent assay (ELISA), but the principle is virtually the same for radioimmunoassay. The first 'capture' antibody, which recognizes the first cytokine epitope, is bound to the bottom of a plastic well. The interaction is demonstrated with the use of a second antibody which recognizes a different cytokine epitope. An enzymatic colorimetric assay is used to measure the reaction. The optical density is read and concentrations are calculated by extrapolation of standard curves. Specific interference in these assays also includes anticytokine antibodies and soluble receptors. These methods are sensitive, are easy to use and have a high specificity. However, they do not measure the biological activity of the detected protein. In addition, antibodies against some cytokines have been difficult to obtain. For instance, the molecule can be highly conserved in various species, as for transforming growth factor (TGF) [3, or affinity may be higher for the recombinant molecule used for immunization rather than the natural one, as is the case for IL-4. For the reasons outlined above, it is useful to combine the two methods:
CYTOKINE A B N O R M A L I T I E S
375
an immunoassay for detection and a b i o l o ~ a l assay for the demonstration of a~-ivit~i[-n-addition, results should be expTessed-lh concentrations using international standards. Indeed, it is difficult to compare published results when undefined biological units are used.
Study of cytokine genes Methods for the study of cytokine mRNA have recently improved, allowing their use with clinical samples such as synovium pieces. These methods now allow the quantification or semi-quantification of cytokine mRNA expression and its topographic distribution in a cell population or tissue (Dallman et al, 1991). , Methods for mRNA quantification were limited by the amount of total RNA necessary for characterization. High quantities of total RNA and high expression Of specific mRNA were needed for the first Northern blot methods. Amplification techniques with the use of the polymerase chain reaction (PCR) can now be performed with a very small number of cells and minute amounts of extracted RNA, including material from arthroscopic biopsies or synovial fluid. However, this very sensitive method can amplify any contamination, stressing the need for perfect internal and external controls. For these reasons, the diagnostic value of these procedures for routine clinical application is still being assessed. In addition, precise quantification of mRNA expression remains to be improved. This will allow the use of these tools for diagnostic purposes as well as for quantitative followup of any changes with treatment. , Methods of mRNA localization such as in situ hybridization allow determination of the percentage of cells expressing specific mRNA for a given cytokine in a purified cell population and also allow demonstration of the histological distribution of the expressing cells (Dallman et al, 1991). They have been used for the detection of cells producing cytokines in rheumatoid synovial fluid and synovium (Firestein et al, 1990). Again, quantification is difficult, as mRNA is not expressed in large amounts. Furthermore, the method only allows for relative quantification with respect to background expression. The study of cytokine DNA includes the use of restriction enzymes and gene sequencing. Restriction enzymes cut the DNA at specific sites. Variations in sequences can induce variations in the number of restriction sites, leading to variations in the length of the fragments, which are identified by electrophoresis. Restriction fragment length polymorphism (RFLP) has been used in various fields, and studies have just started on cytokines in autoimmune conditions. A polymorphism in the TNFe~ gene has been detected by RFLP in mouse and human lupus and correlates with the production of TNFoL (Jacob and McDewitt, 1988; Jacob et al, 1990). The TNFoL and TNF[3 genes are closely linked and are located within the major histocompatibility complex on human chromosome 6. Direct sequencing of cytokine genes allows detection of variations between individuals. At the present time, the significance of these variations remains to be established.
376
P. MIOSSEC
Technical aspects and limitation
Cytokine levels can be measured in plasma, serum, various biological fluids and culture supernatants (Whicher and Ingham, 1990). Synovial fluid can be the source of many artefacts. Some of them relate to increased viscosity, which can be reduced with hyaluronidase treatment. Dilution curves are necessary because of non-specific inhibition found in biological as well as in immunoassays. When using ELISAs, additional saturation steps with proteins such as bovine serum albumin to block non-specific binding sites are often necessary. Inhibitors of cell proliferation and cytotoxic factors for mouse cells interfere with biological assays. A recovery test with the addition of purified cytokines to biological samples is the simplest way to detect these potential problems. Interaction with rheumatoid factors leading to increased values, and even to false positive results, is more difficult to resolve. An improvement can sometimes be obtained with reduction and alkylation treatment of the samples (Malyak et al, 1991). The end-point of these measurements is the demonstration of a correlation between cytokine levels and the pathogenesis and evolution of a disease. However, various groups of cytokines can be defined with discrete major function. The local and systemic role of proinflammatory cytokines is different from the mostly local and regulatory role of T cell derived cytokines. Only a limited number of cytokines (IL-6, TNFc~, IL-1) have circulating levels which correlate with the activity and prognosis of various acute and chronic diseases, including rheumatoid arthritis (Eastgate et al, 1988; Briolay et al, 1992). However, paracrine and autocrine secretion play a major regulatory role at the site of the inflammatory reactior~. Measurement of the local production of cytokines remains more difficult.qndeed, the absence of circulating levels does not imply the absence of secretion.)Local concentrations of these factors can be very high at the interface between cells without diffusion into the circulation or action may be brought about by membrane-bound cytokines. At the present time, in situ hybridization is being used to detect local cytokine gene activation. Moreover, ELISPOT techniques are now available to measure cytokine production (at the protein level) by an isolated cell (Czerkinsky et al, 1991). These techniques should allow for more direct investigation of this crucial question. EXPERIMENTAL FINDINGS
The number of studies on cytokines in inflammatory arthritis has been growing extensively as the list of cytokines increases. The first studies were limited to demonstration of the presence of a single cytokine in synovial fluid using biological assays. In fact, these studies provide limited information as cytokines interact with each other and with factors from other systems. These factors are associated in various cascades with positive and negative regulatory feedback (Figure 1) (Balkwill and Burke, 1989). Levels of cytokines in synovial fluid are highly variable, ranging from undetectable levels to concentrations measured in micrograms per millilitre (Table 1). The major factors contributing to these levels are the number of producing
377
/,+L
CYTOKINE ABNORMALITIES
STIMULUS
IL 6
8 Figure 1. Interactions between proinflammatory cytokines.
cells, the production per cell and the diffusion of the cytokine, as well as the sensitivity of the assay. Thus a cytokine such as IL-6, produced in large amounts by almost any cell type, will be easy to detect, even in serum. In contrast, a cytokine such as IL-4, which is produced in small amounts by a limited T cell subset, will be much more difficult to detect. Cytokines Can be classified for simplicity into three groups according to the cells which produce them: 1. Cytokines produced by almost any cell. 2. Cytokines produced by monocytes/macrophages. 3. Cytokines produced by T lymphocytes.
Table 1. Cytokines in rheumatoid synovial fluid. Cytokine
Origin
Levels
PDGF TGFI3 IL-6
Platelets, macrophages Fibroblasts, macrophages, bone Fibroblasts, macrophages etc
++++ +++ +++
TNFa IL-1
Macrophages Macrophages
+ +
100-1000 pg/ml 10-100 pg/ml
IL-2 IL-4 IFN~/ TNF[3
T T T T
-
Cells interact through direct cell-to-cell contact and through soluble factors acting as local and systemic signals. Cytokines are soluble molecules which control communication between cells in a non-antigen-specific manner. They act locally as paracrine or autocrine factors, and some are also detected systematically, where they act as endocrine factors. They contribute to many biological activities and defence mechanisms. The specificity of their activity is related to their binding to specific membrane receptors. Research in the field of cytokines has been growing extensively over recent years because of advances in immunology and cellular and molecular biology. These methods have allowed the production of large amounts of pure recombinant products for research and, more recently, for clinical applications. Similarly, the number of studies on the role of cytokines in the pathogenesis of inflammatory arthritis expanded with the discovery of new cytokines and the development of assays for their detection (Arend and Dayer, 1990; Harris, 1990; Ziff, 1990). The rationale for the use of cytokines and for drug targeting of the cytokine network in conditions such as rheumatoid arthritis lies in the central role they play in inflammation and autoimmunity. This review will summarize the methods of cytokine detection and then concentrate on the complex balance of the cytokine network and its dysregulation in rheumatoid arthritis.
DETECTION OF CYTOKINES From basic research on cytokines, assays for cytokine detection have been developed for clinical studies. Indeed, results obtained in vitro suggest that these measurements might have both diagnostic and prognostic value (Whicher and Ingham, 1990; Feldman et al, 1991). However, with few exceptions, the usefulness of routine cytokine measurements is still under investigation. Unfortunately, it remains difficult to compare results from the literature, in part because of large variations in sensitivity and specificity between these assays. Synthesis of a cytokine includes numerous steps from its gene activation to the release of the active protein. Each of these steps can now be studied in Baillikre' s Clinical Rheumatology--
Vol. 6, No. 2, June 1992 ISBN 0-7020-1636-5
373 Copyright9 1992,byBailli6reTindall Allrightsofreproductionin anyformreserved
374
P. MIOSSEC
detail with tools derived from basic research. However, the most important factor remains the release of the active protein. Indeed, detection of messenger RNA (mRNA) is not always associated with cytokine secretion when post-transcriptional regulation is present. In addition, following its release, a cytokine can also interact with various regulatory mechanisms before reaching an active membrane receptor.
Assays for cytokines Assays for cytokines can be classified into two major groups (Whicher and Ingham, 1990). Biological assays measure the activity of a cell system. They were the first type of assay to be described, as cytokines were discovered on the basis of their biological activity. They are sensitive and demonstrate the presence of biological activity. Their specificity is usually limited because of the ple!o[_ropi c act!.vity of cytokines, but this limitation can be improved with the use of specific antibodies. The mouse thymocyte proliferation assay was first described as an interleukin (IL)-I specific assay, but was later also shown to respond to TNFo~ (tumour necrosis factor), IL-6, IL-2 and IL-4. Similarly, the CTTL-2 cell line used in the first IL-2 biological assay can also respond in some conditions to IL-4. Cytokines can also act indirectly through the release of other factors which are part of a cascade of secretory events. For example, in the thymocyte proliferation assay, IL-1 induces the release of IL-2, which is responsible for thymocyte proliferation. Furthermore, complex biological fluids can induce specific interference with anticytokine antibodies and soluble receptors. In addition, potent non-specific inhibitory factors are often present in human biological fluids, some of these being toxic for mouse cell lines. Purification procedures are thus necessary to demonstrate the presence of a biologically active molecule (Miossec et al, 1986). Immunoassays detect cytokines as antigenic proteins recognized by specific polyclonal and/or monoclonal antibodies. The most common combination is the double sandwich enzyme-linked immunosorbent assay (ELISA), but the principle is virtually the same for radioimmunoassay. The first 'capture' antibody, which recognizes the first cytokine epitope, is bound to the bottom of a plastic well. The interaction is demonstrated with the use of a second antibody which recognizes a different cytokine epitope. An enzymatic colorimetric assay is used to measure the reaction. The optical density is read and concentrations are calculated by extrapolation of standard curves. Specific interference in these assays also includes anticytokine antibodies and soluble receptors. These methods are sensitive, are easy to use and have a high specificity. However, they do not measure the biological activity of the detected protein. In addition, antibodies against some cytokines have been difficult to obtain. For instance, the molecule can be highly conserved in various species, as for transforming growth factor (TGF) [3, or affinity may be higher for the recombinant molecule used for immunization rather than the natural one, as is the case for IL-4. For the reasons outlined above, it is useful to combine the two methods:
CYTOKINE A B N O R M A L I T I E S
375
an immunoassay for detection and a b i o l o ~ a l assay for the demonstration of a~-ivit~i[-n-addition, results should be expTessed-lh concentrations using international standards. Indeed, it is difficult to compare published results when undefined biological units are used.
Study of cytokine genes Methods for the study of cytokine mRNA have recently improved, allowing their use with clinical samples such as synovium pieces. These methods now allow the quantification or semi-quantification of cytokine mRNA expression and its topographic distribution in a cell population or tissue (Dallman et al, 1991). , Methods for mRNA quantification were limited by the amount of total RNA necessary for characterization. High quantities of total RNA and high expression Of specific mRNA were needed for the first Northern blot methods. Amplification techniques with the use of the polymerase chain reaction (PCR) can now be performed with a very small number of cells and minute amounts of extracted RNA, including material from arthroscopic biopsies or synovial fluid. However, this very sensitive method can amplify any contamination, stressing the need for perfect internal and external controls. For these reasons, the diagnostic value of these procedures for routine clinical application is still being assessed. In addition, precise quantification of mRNA expression remains to be improved. This will allow the use of these tools for diagnostic purposes as well as for quantitative followup of any changes with treatment. , Methods of mRNA localization such as in situ hybridization allow determination of the percentage of cells expressing specific mRNA for a given cytokine in a purified cell population and also allow demonstration of the histological distribution of the expressing cells (Dallman et al, 1991). They have been used for the detection of cells producing cytokines in rheumatoid synovial fluid and synovium (Firestein et al, 1990). Again, quantification is difficult, as mRNA is not expressed in large amounts. Furthermore, the method only allows for relative quantification with respect to background expression. The study of cytokine DNA includes the use of restriction enzymes and gene sequencing. Restriction enzymes cut the DNA at specific sites. Variations in sequences can induce variations in the number of restriction sites, leading to variations in the length of the fragments, which are identified by electrophoresis. Restriction fragment length polymorphism (RFLP) has been used in various fields, and studies have just started on cytokines in autoimmune conditions. A polymorphism in the TNFe~ gene has been detected by RFLP in mouse and human lupus and correlates with the production of TNFoL (Jacob and McDewitt, 1988; Jacob et al, 1990). The TNFoL and TNF[3 genes are closely linked and are located within the major histocompatibility complex on human chromosome 6. Direct sequencing of cytokine genes allows detection of variations between individuals. At the present time, the significance of these variations remains to be established.
376
P. MIOSSEC
Technical aspects and limitation
Cytokine levels can be measured in plasma, serum, various biological fluids and culture supernatants (Whicher and Ingham, 1990). Synovial fluid can be the source of many artefacts. Some of them relate to increased viscosity, which can be reduced with hyaluronidase treatment. Dilution curves are necessary because of non-specific inhibition found in biological as well as in immunoassays. When using ELISAs, additional saturation steps with proteins such as bovine serum albumin to block non-specific binding sites are often necessary. Inhibitors of cell proliferation and cytotoxic factors for mouse cells interfere with biological assays. A recovery test with the addition of purified cytokines to biological samples is the simplest way to detect these potential problems. Interaction with rheumatoid factors leading to increased values, and even to false positive results, is more difficult to resolve. An improvement can sometimes be obtained with reduction and alkylation treatment of the samples (Malyak et al, 1991). The end-point of these measurements is the demonstration of a correlation between cytokine levels and the pathogenesis and evolution of a disease. However, various groups of cytokines can be defined with discrete major function. The local and systemic role of proinflammatory cytokines is different from the mostly local and regulatory role of T cell derived cytokines. Only a limited number of cytokines (IL-6, TNFc~, IL-1) have circulating levels which correlate with the activity and prognosis of various acute and chronic diseases, including rheumatoid arthritis (Eastgate et al, 1988; Briolay et al, 1992). However, paracrine and autocrine secretion play a major regulatory role at the site of the inflammatory reactior~. Measurement of the local production of cytokines remains more difficult.qndeed, the absence of circulating levels does not imply the absence of secretion.)Local concentrations of these factors can be very high at the interface between cells without diffusion into the circulation or action may be brought about by membrane-bound cytokines. At the present time, in situ hybridization is being used to detect local cytokine gene activation. Moreover, ELISPOT techniques are now available to measure cytokine production (at the protein level) by an isolated cell (Czerkinsky et al, 1991). These techniques should allow for more direct investigation of this crucial question. EXPERIMENTAL FINDINGS
The number of studies on cytokines in inflammatory arthritis has been growing extensively as the list of cytokines increases. The first studies were limited to demonstration of the presence of a single cytokine in synovial fluid using biological assays. In fact, these studies provide limited information as cytokines interact with each other and with factors from other systems. These factors are associated in various cascades with positive and negative regulatory feedback (Figure 1) (Balkwill and Burke, 1989). Levels of cytokines in synovial fluid are highly variable, ranging from undetectable levels to concentrations measured in micrograms per millilitre (Table 1). The major factors contributing to these levels are the number of producing
377
/,+L
CYTOKINE ABNORMALITIES
STIMULUS
IL 6
8 Figure 1. Interactions between proinflammatory cytokines.
cells, the production per cell and the diffusion of the cytokine, as well as the sensitivity of the assay. Thus a cytokine such as IL-6, produced in large amounts by almost any cell type, will be easy to detect, even in serum. In contrast, a cytokine such as IL-4, which is produced in small amounts by a limited T cell subset, will be much more difficult to detect. Cytokines Can be classified for simplicity into three groups according to the cells which produce them: 1. Cytokines produced by almost any cell. 2. Cytokines produced by monocytes/macrophages. 3. Cytokines produced by T lymphocytes.
Table 1. Cytokines in rheumatoid synovial fluid. Cytokine
Origin
Levels
PDGF TGFI3 IL-6
Platelets, macrophages Fibroblasts, macrophages, bone Fibroblasts, macrophages etc
++++ +++ +++
TNFa IL-1
Macrophages Macrophages
+ +
100-1000 pg/ml 10-100 pg/ml
IL-2 IL-4 IFN~/ TNF[3
T T T T
-
