Analysis of Cytokine Expression in Dermatology FREE

Cytokines play an essential role in the intercellular communication network. They are essential for the mediation and regulation not only of inflammatory reactions but also of specific immune reactions and nonimmunologic processes (hematopoiesis and bone and cartilage metabolism). The importance of cytokines in medicine is increasing dramatically. They are far beyond the stage when they were of interest only to the research sector; cytokine therapy of various diseases has long been used clinically.¹ The use of cytokine therapy in dermatology has recently been reviewed.²Table 1 gives an overview. Moreover, the diagnostic determination of cytokine expression is being used clinically as well. Before addressing the possibilities and limits of the different approaches for cytokine determination, an overview of nomenclature and receptor binding and signaling and their functional consequences will be given.

The nomenclature of cytokines is confusing and has changed over time. Initially, cytokines were generally given a name describing their function (eg, tumor necrosis factor [TNF], which induces necrosis of some tumors). However, it was soon discovered that the functional behavior of cytokines is often redundant (several cytokines have the same effects) or pleiotropic (one cytokine has several different effects) (Figure 1 and Figure 2, respectively). Based on these facts, new cytokines are not named according to their properties. An international nomenclature, at least for the immunoregulatory cytokines, has been developed. The relevant cytokines were designated as interleukins (ILs) and provided with numbers (currently, IL-1 to IL-24); the IL-1 receptor antagonist is a relevant cytokine as well. However, numerous cytokines were omitted from this nomenclature. They include interferons (IFNs) (IFN-α, IFN-β, and IFN-γ), colony-stimulating factors (CSFs) (macrophage CSFs, granulocyte-monocyte CSFs, granulocyte CSFs, stem cell factors, erythropoietin, and thrombopoietin), TNFs (TNF-α, TNF-β, and lymphotoxin β), chemotactic factors (macrophage inflammatory protein and monocyte chemoattractant protein), and soluble receptors (CD23, p55–IL-2 receptor [IL-2R], IL-4 receptor, TNF receptor, and IL-1 receptor).

The diversity of cytokine effects makes a didactically sensible nomenclature difficult. In principle, cytokines can be divided into 3 groups, based on their origin and effect: (1) those produced by many cell types that act on many cell types (eg, transforming growth factor β); (2) those produced by few cell types that act on few cell types (eg, IL-2); and (3) those produced by few cell types that act on many cell types (eg, TNF-α and IFN-γ).

From the viewpoint of their effect on inflammation processes, cytokines can be divided into proinflammatory (eg, IL-1, IL-2, INF-γ, and TNF-α) and anti-inflammatory (eg, IL-1 receptor antagonist, IL-10, and transforming growth factor α). Certain soluble receptors are naturally occurring antagonists of cytokines.

The cytokines of immunologic relevance are primarily those that are formed by lymphocytes and antigen-presenting cells and influence their function. For many years, cytokine production by lymphocytes has been the object of intense investigation. Based on the results of a study by Mosmann et al,³ published in 1986, murine CD4⁺ T cells can be divided into subclasses. Helper T (T_H) 1 cells are characterized by the production of IL-2, TNF-β, and INF-γ, whereas Th2 cells produce mainly IL-4, IL-5, IL-6, IL-9, IL-10, and IL-13. Whether a T_H0 cell (a cell with no preferred cytokine secretion pattern) differentiates into a T_H1 or a T_H2 cell under antigen influence depends essentially on the local cytokine milieu. Interleukin 12 and INF-γ encourage development of the T_H1 cell response, whereas IL-4 favors the T_H2 cell response (Figure 3). An essential role in the origin of the local cytokine milieu is played by non–T cells too (eg, natural killer cells [INF-γ], monocytes or macrophages [IL-12 or IL-10, according to stimulus], and mast cells [IL-4]). CD8⁺ T cells usually secrete a T_H1 cell–like pattern. Consequently, because not only CD4⁺ T_H cells but also other cell populations contribute to local cytokine expression, the cytokine pattern is often designated as type 1 or type 2 accordingly.

The T_H1/T_H2 or type 1/2 cytokine model has made a major contribution to our understanding of immunologic processes. To simplify, one may say that a type 1 cytokine pattern is necessary for an effective cellular immunologic reaction to antigen, such as the immune response to intracellular pathogens,⁴ although the secretion of immunoglobulin subclasses IgG1 and IgG3, which have a high capacity to interact with immune cells via Fc and C receptors, is stimulated by type 1 cytokines too. Type 2 cytokines, on the other hand, are mainly responsible for effective humoral immunologic mechanisms. Thus, these cytokines support IgE, IgA, IgG2, and IgG4 synthesis and eosinophil and mast cell growth. Consequently, they play a particularly decisive role in defense against intestinal parasites, neutralization of bacterial toxins, and local mucosal defense.⁵ Type 1 and type 2 responses are essential for an adequate immunity.

In recent years, cytokine patterns have been investigated in diseases from nearly all specialties. This has made a major contribution to today's better understanding of pathophysiological processes. An attempt has been made to identify classic type 1 or type 2 diseases. In fact, in many diseases, an immune deviation toward a type 1 or a type 2 cytokine pattern has been observed and might be of pathophysiological relevance⁵ (Figure 3). However, the limitations of the type 1/type 2 system are becoming increasingly clear and prevent dogmatic application. Thus, several investigations⁶ have revealed a remarkable variety of cytokine patterns, indicating a broad spectrum of cytokine expression in which the classic type 1 and type 2 patterns are merely the extremes. Moreover, the existence of another important T-cell population with regulatory properties (T-regulatory and T_H3 cells) has been demonstrated recently.^7- 8 These cells predominantly act immunosuppressively via secretion of cytokines like IL-10 and transforming growth factor β and via cell-cell interaction. Meanwhile, the complexity of cytokine expression has also been demonstrated in vivo—taking dermatology as an example—in patients with psoriasis,⁹ atopic dermatitis,¹⁰ and mycosis fungoides.¹¹ For example, stage, dependent shifts (for atopic dermatitis, early phase type 2 and late phase type 1), and untypical cytokine patterns (the advanced stage of mycosis fungoides is like type 2 but shows weak IL-4 expression) can be observed.

Class 1 receptors bind factors designated variously as ILs, hormones, or CSFs (Table 2), but, despite the different designations, the ligands all have a common 4-α–helical structure.^12- 13 These receptors are transmembrane proteins. Intracellularly, they have a conserved membrane–proximal domain that serves to bind Janus kinases (Jaks).¹⁴ The conserved features of the ligands and receptors permit computational database mining and the recent identification of new family members.^15- 17 Class 2 receptors are structurally related to class 1 receptors and bind IFNs and IL-10 family cytokines; this family is rapidly expanding.¹⁸ Recently, researchers discovered the first soluble class 2 receptor.¹⁹

Sharing of receptor subunits is an important feature of cytokines binding class 1/2 receptors that defines subfamilies (Table 2).¹² For instance, IL-2, IL-4, IL-7, IL-9, and IL-15 use the common γ chain in conjunction with ligand-specific chains, which explains the specific and redundant actions of these cytokines. Moreover, some cytokines, like IL-2 and IL-15, share even 2 receptor subunits (IL-2Rβ and the common γ chain), but have an additional ligand-specific binding subunit (IL-2Rα or IL-15 receptor α). Another level of complexity is that more than one type of receptor may exist, as in the case of IL-4. One IL-4 receptor consists of the IL-4 receptor associated with the common γ chain, whereas a second comprises IL-4 receptor with IL-13 receptor. Finally, a third level of complexity has been illustrated by IL-12 and IL-23.¹⁶ The p40 subunit is not homologous to other cytokines but rather to cytokine receptors. It was first identified in association with p35 to form IL-12. In contrast, IL-23 comprises p40 and a novel subunit, p19; the latter is homologous to other cytokines. Like IL-12, IL-23 binds to the IL-12 receptor β1 chain and induces IFN-γ production. However, IL-12 induces IFN-γ in naive and memory T cells, and IL-23 acts on memory T cells only. This may be because IL-23, in contrast to IL-12, does not bind to the IL-12 receptor β2 chain.¹²

The action of cytokines is not exclusively regulated by the level of ligand and receptor expression, but is also dependent on intracellular signal transduction. At least 3 classes of molecules are of major importance: classic transcription factors (eg, nuclear factor κB and activator protein 1), Jaks, and signal transducers and activators of transcription (Stats). Moreover, other intracellular proteins are involved in determining the biological consequences of cytokine presence. Therefore, 3 families of negative regulatory proteins have been discovered: phosphatases, protein inhibitors of activated Stats, and suppressors of cytokine signaling. Each of these protein families seems to act at a distinct point and at a particular time in the cytokine signaling cascade. It is important to be aware of this high complexity when interpreting results from cytokine expression analyses.¹²

The Jaks constitutively bind the membrane-proximal domains of cytokine receptors and seem to be the major initiators of signaling (Table 2). Four mammalian Jaks have been identified—Jak1, Jak2, Jak3, and Tyk2 (a small family of tyrosine kinases with rather specific functions, as illustrated by gene-targeted mice). After activation, the Jaks phosphorylate receptor subunits on tyrosine residues to recruit proteins with src homology 2 domains or phosphotyrosine-binding domains. These proteins, in turn, are also phosphorylated by Jaks, coupling cytokine stimulation to several pathways, including the Ras/Rat/MAPK and phosphatidylinositol 3′ kinase/Akt pathways.¹² Phosphorylation of cytokine receptors also generates docking sites for a class of src homology 2 domain–containing cytosolic molecules termed "Stats."^20- 23 Receptor-bound Stats are then phosphorylated (Table 2), allowing them to dimerize. This permits translocation to the nucleus and DNA binding. Generally, Stats bind 2 types of DNA motifs: IFN-stimulated response elements and γ-activated sequence elements. For Stat6, there seems to be little specificity in Stat binding to DNA. The other Stats seem to be a relatively small family of transcription factors with highly specific functions; remarkably, it seems that at least 4 of the 6 mammalian Stats have major functions in regulating host defense and immune responses. There is only limited knowledge about cytokine-inducible genes and how Stats work with other transcription factors to regulate the expression of these genes.¹² The transcription factor nuclear factor κB is a main mediator of many proinflammatory cytokines (TNF-α and IL-1). Further investigations, in particular on the role of negative regulatory proteins (eg, suppressors of cytokine signaling), will be important if the physiological and pathophysiological relevant controls on cytokine responses are to be uncovered.

Cytokines exert their biological effects at concentrations in the picomolar range. The biological half-life in vivo is short, generally not exceeding 3 minutes in plasma. An exception is IL-12, whose half-life is several hours. The short in vivo half-life is the result of binding to cells (high-affinity receptors) or plasma proteins, proteolytic degradation, and elimination via the kidneys. Moreover, cells nearly always produce cytokine for only a few hours after appropriate stimulation. This results in limited systemic availability, which makes relevant quantitation in plasma difficult. The physiological task of cytokines consists principally in the regulation of local intercellular communication processes, ie, the effects are largely of an autocrine or paracrine nature, whereas endocrine effects are fairly rare.

In principle, a distinction can be made between in vivo, ex vivo, and in vitro cytokine detection. A positive finding in in vivo determinations indicates that the cytokine is actually present, whereas ex vivo and in vitro methods (such as the whole-blood assay^24- 25) use stimulants and primarily reflect the cytokine expression capacity of the stimulated cells and not necessarily the actual activity in vivo. Endotoxin, for example, is often used as an ex vivo stimulus for monocytes and macrophages. Lectins, anti–T-cell antibodies, or superantigens are used as T-cell mitogens. Cytokine detection can take place on various levels (Figure 4).

These determinations are based on the observation that resting immunocytes express little or no cytokine messenger RNA (mRNA). Within a few hours after stimulation, an increase in mRNA level is seen. Whereas the sensitivity of conventional detection methods is sufficient to detect the mRNA of typical inflammatory cytokines (eg, TNF, IL-6, and IL-8) without amplification of the specific nucleic acid sequence (eg, in situ hybridization and a nuclease protection assay), the mRNA expression of most T-cell cytokines (eg, IL-2 and IL-4) is barely detectable with these methods in in vivo samples. A much higher sensitivity is achieved by cytokine mRNA detection using the polymerase chain reaction (PCR) (Table 3). Nevertheless, quantitation of the degree of expression is much more difficult than in the hybridization procedures without prior amplification of complementary DNA (cDNA). Competitive reverse transcription–PCR,²⁶ in which a DNA competitor fragment added in a known concentration competes with the target cDNA for the primer, is suitable for quantitative determination.^11,27- 28 However, real-time reverse transcription–PCR is the most reliable PCR method. It uses the 5′ nuclease activity of Taq polymerase to cleave a nonextendable hybridization probe during the extension phase of the PCR. The approach uses dual-labeled fluorogenic hybridization probes. One fluorescence dye serves as a reporter, and its emission spectrum is quenched by the second fluorescent dye. Following hybridization of the labeled probe, reporter and quencher are separated, resulting in fluorescence. The reactions are monitored in real time during the log phase of product accumulation. The increase of the reporter dye fluorescence intensity during PCR is proportional to the amplification of the target sequence. The cycle number at which the amplification plot crosses a fixed threshold above baseline is defined as the threshold cycle. These values can be used for comparison of mRNA content. To control variation in cDNA contents throughout different preparations, all results are related to the concentration of a gene whose expression remains relatively constant ("housekeeping gene" [eg, the human hypoxanthine-phosphoribosyl-transferase gene]).²⁹

When using a PCR method to investigate cutaneous cytokine expression, one obvious disadvantage is the lack of information regarding the cellular source of a cytokine found to be overexpressed. Drawing conclusions from the cellular pattern determined simultaneously by histologic investigation requires caution, because small cell numbers can also be responsible for large amounts of cytokines, whereas even high numbers of certain cells types could be present without producing any cytokines. To determine the compartment mainly responsible for the cytokine expression, splitting skin samples and comparing epidermis with dermis can be considered. Such analyses, however, are usually not sufficient to identify the cell type because of the complex cellular composition of epidermis and dermis, which is even more complex under pathophysiological conditions. Also, splitting skin samples may affect the transcripts.³⁰ Therefore, we prefer to analyze whole skin punch biopsy specimens immediately shock frozen after attainment, without any further manipulation.

Hybridization by Northern blotting is a technique less sensitive than PCR but enables the detection of splice variants. Gene chip analysis, in turn, as a modern but expensive recent hybridization technique, has the advantage of delivering information on the expression of up to 12 000 gene products simultaneously (Table 3).

However, although cytokine gene expression is a prerequisite for protein synthesis, it does not necessarily lead to such synthesis because there are still several posttranscriptional regulatory bases. Hence, the detected cytokine gene expression and the protein synthesis do not necessarily correlate with each other. Examples of significantly posttranscriptionally regulated cytokines are IL-1β and IL-18.

Historically, cytokines were initially detected solely with bioassays (Table 4). For example, IL-2 was detected based on its property to stimulate the proliferation of activated T cells in vitro. By adding known quantities of IL-2, a calibration curve can be plotted. This test is still used, using highly sensitive T-cell lines whose growth depends on IL-2. However, its specificity is known to be limited because other cytokines (eg, IL-4 and IL-7) mediate similar effects in this test. This can be differentiated by adding a blocking IL-2R antibody. The redundancy of cytokine effects must be borne in mind in other bioassays too.

Thanks to the use of internationally harmonized cytokine standards and the distribution of correspondingly sensitive cell lines via international tissue banks, the standardization of the bioassays has improved considerably. Even so, the specificity of any detected bioactivity must always be checked by adding specific neutralizing antibodies. Although bioassays are cheaper than enzyme-linked immunosorbent assays (ELISAs) (detailed later), their high intra-assay and interassay variance makes them unsuitable for clinical monitoring.

In recent years, numerous ELISAs for the detection of a wide range of cytokines have been developed (Table 4). Although they were initially less sensitive than bioassays, the new generation of ELISAs is actually more sensitive than bioassays. Thus, most cytokines are detectable even in the plasma of healthy volunteers. The high specificity of ELISAs and the better standardizability are further advantages to be weighed against the high costs. A limitation of ELISAs that is often overlooked in their clinical evaluation is the fact that only results obtained with the same test are generally comparable. This may result from the use of differing standards, epitope specificity (and sometimes also the affinity) of the antibodies used in the test, and the particular properties of the cytokines. Cytokines do not only occur in free and native form but also bind to numerous carrier proteins and can undergo rapid proteolytic cleavage. A wide variety of cells secrete soluble cytokine receptors after activation. These generally have neutralizing properties. On the other hand, they may prolong activity by inhibiting cytokine elimination. Cytokines can also bind to other plasma proteins. Such proteins cover certain epitopes on the cytokine surface that may or may not still be recognized, depending on the epitope specificity and affinity of the cytokine-complex antibodies used in the ELISA (Figure 5).

The interpretation of cytokine concentrations in body fluids (plasma, urine, and lavage fluid) poses additional problems besides those just mentioned. Storage of the samples can have a considerable influence on the results. False high concentrations can arise from activation following coagulation and contact with the syringe material. To avoid these difficulties, it is advisable to (1) determine cytokine levels in plasma, preferably with calcium and magnesium withdrawal (using EDTA or citrate instead of heparin, but some assays do not work in the presence of EDTA), and (2) separate the cells and fluid rapidly (<2 hours) by centrifugation. False low values are caused by proteolytic degradation of cytokines ex vivo. This can be prevented by cold storage, enzyme blockade, and rapid workup of the sample (freezing the cell-free supernatants or performing the test within 120 minutes if possible). This makes it difficult to send samples to central laboratories (worked-up samples on dry ice). The various cytokines certainly differ in sensitivity. A further important aspect is the expression of the concentrations. In plasma, this is unproblematic, but in other body fluids, it poses a relevant problem. Quoting the urinary concentration of a cytokine in picograms per milliliter makes little sense. Similar considerations apply to other body fluids or when analyzing cytokine concentrations in fluid obtained from cutaneous suction blisters. Moreover, this iatrogenic tissue damage, by inducing the blister itself, certainly leads to an immune response associated with cytokine release. Consequently, the cytokine's pattern detected in suction blisters does not necessarily reflect that what is actually present. However, for comparison between different groups of patients in one trial using this technique under exactly the same conditions, it could be a valuable method. In view of the instability of cytokines, however, all portions would have to be frozen immediately.

Even while remaining alert to the previously mentioned methodological problems, however, one must not forget that cytokines are formed only temporarily and have an extremely short half-life in vivo. Although high values are informative, low values usually tell little. It may simply be that the time at which the sample was taken was unfavorable.

Another important method is the detection of cytokine-producing cells by immunocytology or immunohistology. Tissue samples or cells from fluids (blood, cerebrospinal fluid, and fine-needle aspirates) are fixed on slides, made permeable (not always), and incubated with specific anticytokine antibodies. Cytokine-producing cells are detected using immune enzyme methods (alkaline phosphatase–anti-alkaline phosphatase or the peroxidase technique). Immunohistochemical analyses are often used for analyzing frozen skin biopsy specimens. However, unspecific staining (a false-positive result) is a frequent problem, and antibodies could be washed out during the procedure, leading to false-negative results. Moreover, cytokines are frequently detected at an intercellular, not an intracellular, level.^31- 32 Therefore, one advantage of immunohistochemical analyses, the reliable identification of the cellular source, is not always given, although it might be possible in some cases.³³ Test kits that permit intracellular cytokine detection using flow cytometry are increasingly becoming available. This technique is well established for analyzing blood cells. Recently, researchers³⁴ developed methods that also allow the flow cytometric determination of intracellular cytokine expression in mechanically disaggregated cutaneous cells. Finally, Western blotting is still a valid technique for detecting cytokines from tissue or cell lysates, but may not be as sensitive as ELISA (Table 4).

There are several types of cytokine-induced products: soluble surface molecules, including receptors (eg, soluble IL-2R), and adhesion molecules (eg, soluble endothelial leukocyte adhesion molecule 1); soluble factors, including neopterin and C-reactive protein; and cell-bound surface molecules, including HLA-DR and adhesion molecules.

Starting from the thesis that only the effects of cytokines are of any biological relevance, one approach for detecting cytokine production in vivo relies on the measurement of cytokine activity. For example, IFNs induce the release of neopterin from monocytes and macrophages. Neopterin has a longer plasma half-life than cytokines and is excreted via the kidneys. Hence, determination in plasma or urine reflects IFN synthesis in vivo. Neopterin can be quantified using commercially available immunoassays.

Many cytokines induce the synthesis of adhesion molecules and HLA molecules in immunocytes and nonimmunologic cells (eg, endothelial and epithelial cells). These molecules can be easily detected by immunocytologic and immunohistologic methods. However, the increased expression of adhesion or HLA molecules on the cells following the action of cytokines is also associated with increased release of such molecules by the activated cells. Soluble adhesion or HLA molecules can then be measured by ELISA in plasma or other body fluids. Owing to their more persistent production and longer plasma half-life, these cytokine-induced molecules are easier to detect than the cytokines themselves. In view of the redundancy of the cytokine effect (Figure 1), however, no exact correlation between the induction of such a molecule and the production of a specific cytokine is possible.

More than 10 years of intensive cytokine research in clinical samples has opened new insights into the pathogenesis of many diseases. The techniques for detecting cytokines at the mRNA and protein level have been significantly improved. Most important was the standardization of methods. Continuing with such processes will be of crucial importance.

Modern (semi) automatic ELISA systems allow precise quantification of cytokines in the picomolar range, with interassay and intra-assay variance of less than 5% to 8%. Using standardized (semi) automatic systems for cytokine detection and cell culture, functional assays allow a high level of reproducibility (eg, for ex vivo lipopolysaccharide-induced TNF-α release: <20% variance during a 1-year follow-up in healthy patients). The introduction of the real-time PCR and the cDNA array technology is a revolution for quantifying cytokines at the mRNA level. The detection of cytokine mRNA by real-time reverse transcription–PCR is faster and more precise (<20% variance) than traditional techniques. Array technologies allow the simultaneous determination of several gene expression products (gene chips) and, more recently, of proteins (protein-binding arrays) in individual samples. The future will be the consequent use of standardized automatic systems only. This will also include methods for RNA extraction and cDNA synthesis, because sample preparation seems to become the bottleneck for quality and quantity of analyzing cytokine mRNA expression. Further emerging technologies may allow reliable detection of cytokine-producing cells (eg, in situ PCR) or quantification without prior amplification (direct mRNA quantification [eg, the branched DNA method]).

The new technologies will provide a better opportunity for well-designed multicenter trials, which are necessary to evaluate the clinical relevance of cytokine diagnostic methods. Although this will be associated with significant costs, it may lead to major progress. For example, based on the expression pattern of cytokines, it might become possible to determine distinct immunologic subpopulations within one disease, such as psoriasis. This is not just of academic interest. We learned that typically only subgroups of these patients respond well to each of the multiple novel immunotherapies, which are to be launched soon.³⁵ It would be highly desirable (and cost reducing) to identify patients who are likely to respond to each approach before or during the initial phase of therapy.

Accepted for publication April 19, 2002.

We thank Ulrich Zuegel, PhD, for preparing Figure 3; Sabine Goerlich and Garry Willett for editing the manuscript; and Wolf-Dietrich Doecke, MD, for helpful discussions.

Corresponding author and reprints: Khusru Asadullah, MD, Corporate Research Business Area Dermatology, Schering AG, Müllerstrasse 178, D-13342 Berlin, Germany (e-mail: khusru.asadullah@schering.de).