Functional genomics in sarcoidosis – reduced or increased apoptosis ? 1

Sarcoidosis is a chronic systemic disorder characterised by the presence of non-caseating granulomas and accumulation of T-lymphocytes and macrophages in multiple organs [1]. Although the most important enigma of sarcoidosis, ie, its aetiology, remains an unsolved problem, the past few years have seen remarkable advances in understanding general immunological and molecular aspects of the mechanisms leading to granuloma formation in sarcoidosis. Accumulation of macrophages and, particularly, CD4-positive T-lymphocytes are present at sites of disease activity, later conglomerating to form Background: A variety of studies have stressed the importance of the control of inflammatory cell longevity and the balance of pro-survival and proapoptotic signaling pathways. The aim of the study was to investigate the systemic activation of apoptosis pathways using cDNA array technology in patients with acute onset sarcoidosis. Method: We have performed a comprehensive genomic analysis, applying high-density human GeneChip  probe arrays (HGU95A, Affymetrix) for RNA expression profiling from peripheral blood mononuclear cells from patients with acute pulmonary sarcoidosis and matched healthy controls. Twelve patients and 12 controls were assessed, mean age 36 ± 12 and 33 ± 10 years respectively. Results focus on apoptosis-related gene products. Group differences were assessed with the Mann-Whitney U-test. Results: Seven patients had self-limited disease (all type I sarcoidosis) and 5 progressive disease requiring immunosuppression (all type II or III sarcoidosis). We found 53 of 112 (47%) apoptosisrelated gene products dysregulated in sarcoidosis compared to controls. Particular growth factors, especially heparin-binding EGF-like GF, EGF, PDEGF, SISPDGF2 and VEGF, were upregulated in patients consistent with a pro-survival profile. The Bcl-2 family of genes also showed a net pro-survival profile in sarcoidosis patients. In contrast, alterations in the TNF-pathway were compatible with increased apoptosis signals in both, type I and type II/III sarcoidosis patients. Other cell death receptors were equally expressed, as were caspases and p53-associated genes. In contrast to patients with type I-sarcoidosis, patients with progressive type II or III disease showed an upregulation of NFKB and a leak of downregulation of inhibitor of apoptosis 1. Conclusion: Significant differences in the expression of apoptosis-related genes were found in peripheral blood of patients with acute onset sarcoidosis. Gene expression did not show a definite pattern that was suggestive of pro-survival or proapoptosis. However, the number of genes whose altered expression would be predicted to favour increased survival exceeded that of genes likely to reduce survival. Protein-based confirmation of the differences in the activity of apoptosis-pathways needs to be done in further studies.

Sarcoidosis is a chronic systemic disorder characterised by the presence of non-caseating granulomas and accumulation of T-lymphocytes and macrophages in multiple organs [1].Although the most important enigma of sarcoidosis, ie, its aetiology, remains an unsolved problem, the past few years have seen remarkable advances in understanding general immunological and molecular aspects of the mechanisms leading to granuloma formation in sarcoidosis.Accumulation of macrophages and, particularly, CD4-positive T-lymphocytes are present at sites of disease activity, later conglomerating to form Background: A variety of studies have stressed the importance of the control of inflammatory cell longevity and the balance of pro-survival and proapoptotic signaling pathways.The aim of the study was to investigate the systemic activation of apoptosis pathways using cDNA array technology in patients with acute onset sarcoidosis.
Method: We have performed a comprehensive genomic analysis, applying high-density human GeneChip ® probe arrays (HGU95A, Affymetrix) for RNA expression profiling from peripheral blood mononuclear cells from patients with acute pulmonary sarcoidosis and matched healthy controls.Twelve patients and 12 controls were assessed, mean age 36 ± 12 and 33 ± 10 years respectively.Results focus on apoptosis-related gene products.Group differences were assessed with the Mann-Whitney U-test.
Results: Seven patients had self-limited disease (all type I sarcoidosis) and 5 progressive disease requiring immunosuppression (all type II or III sarcoidosis).We found 53 of 112 (47%) apoptosisrelated gene products dysregulated in sarcoidosis compared to controls.Particular growth factors, especially heparin-binding EGF-like GF, EGF, PDEGF, SISPDGF2 and VEGF, were upregu-lated in patients consistent with a pro-survival profile.The Bcl-2 family of genes also showed a net pro-survival profile in sarcoidosis patients.In contrast, alterations in the TNF-pathway were compatible with increased apoptosis signals in both, type I and type II/III sarcoidosis patients.Other cell death receptors were equally expressed, as were caspases and p53-associated genes.In contrast to patients with type I-sarcoidosis, patients with progressive type II or III disease showed an upregulation of NFKB and a leak of downregulation of inhibitor of apoptosis 1.
Conclusion: Significant differences in the expression of apoptosis-related genes were found in peripheral blood of patients with acute onset sarcoidosis.Gene expression did not show a definite pattern that was suggestive of pro-survival or proapoptosis.However, the number of genes whose altered expression would be predicted to favour increased survival exceeded that of genes likely to reduce survival.Protein-based confirmation of the differences in the activity of apoptosis-pathways needs to be done in further studies.
However, the mechanism leading to the persistent accumulation of inflammatory cells is not fully understood.Apoptosis, a dynamic process involved in the control of the "tissue load" of immune effecter cells at inflamed sites, limits inflammatory tissue injury and promotes resolution of inflammation [15,16].Whether or not reduced apoptosis is involved in the pathogenesis of sarcoidosis is unclear.Studies looking at Fas and/or TNF-receptor 1 found increased levels of expression on T-lymphocytes in BAL-fluid [17,18] suggesting increased apoptosis and, similarly, Kunitake [19] found increased numbers of cells going into apoptosis in lung tissue of patients with sarcoidosis compared to controls.These findings are in agreement with increased apoptosis in sarcoidosis.On the other hand it has been postulated as early as 1987 by Cree et al. [20] that apoptosis phenomena might correlate with disease activity and, thus, may play an important role in disease outcome in sarcoidosis.
Many different pathways are involved in apoptosis, the main ones being signaling through death receptors (e.g.FAS, TNFR1) [21], the p53 pathway [22], the Bcl-2 family of genes [23] and the cascade of effector proteases -the caspases [24].These pathways involve multiple components, which can be pro-and anti-apoptotic.Reduced apoptosis can, therefore, be the net result of reduced expression of pro-apoptotic factors or increased expression of prosurvival products.The large number of genes, whose differential expression could affect cell survival in particular pathological situations, creates two major problems in attempting to analyse the basis for altered cell survival by measuring gene expression on an individual basis.First, it is likely that only a small proportion of involved genes will show altered expression, making identification of those genes difficult.Secondly, and more importantly, the identification of differential expression of a single gene will not necessarily reflect the whole balance of changes in pro-apoptotic and pro-survival gene expression.
With cDNA or oligonucleotide arrays several thousand gene products can be assayed in a single experiment opening a new dimension to gene expression studies (fig. 1) [25][26][27][28].With this technique, target RNA can be copied into labeled cDNA with reverse transcriptase so that the relative abundance of individual mRNAs is reflected in the cDNA product.For gene expression studies single-stranded probe cDNA or oligonucleotides [29,30] derived from sequences of known genes, cDNA libraries or ESTs can be fixed on filter/glass slide arrays and hybridised with the target cDNA.
To date, there have been few studies of apoptosis in sarcoidosis and these have been limited to the examination of a small number of genes.Using high-throughput arrays we have sought to shed further light on apoptosis signals in the peripheral blood of sarcoidosis patients with self-limited and progressive sarcoidosis.

Methods
Twelve consecutive symptomatic patients presenting with acute onset sarcoidosis were included from August 1999 to May 2000 for this prospective controlled trial.Patients and 12 healthy controls, matched for age and sex, were non-smokers, non-atopic, with no prior history of malignancy, chronic inflammatory disorder or treatment with corticosteroids.All included subjects were white Caucasians and gave informed written consent.Non-atopic state was confirmed by history and negative skin prick test to house dust mite, grass pollen, cat and dog dander.The diagnosis of sarcoidosis was confirmed histologically in 9 patients by transbronchial lung biopsy and in 3 patients, presenting with erythema nodosum and bilateral hilar lymphadenopathy, by a bronchoalveolar lavage CD4/CD8 ratio >3.5.Chest radiographs were staged according to the Silzbach classification [31].All patients had a clinical follow-up at 6 months including chest radiograph and pulmonary function tests.Seven of 7 patients presenting with type I sarcoidosis fully recovered during the 12-month follow-up period, whereas all 5 subjects with type II/III disease had persistent symptoms and all requiring immunosuppressive therapy.Spirometry and diffusion capacity were measured with a SensorMedics Vmax 22 series.
Peripheral blood mononuclear cells (PBMCs) were separated from 50 mL heparinised whole blood, drawn at the baseline visit, using gradient centrifugation (Ficoll ® Paque, Pharmacia, Uppsala Sweden).The buffy layer was carefully recovered and washed 3 times in culture medium (AIM V ® , Life Technologies Paisley U.K.).Cell pellets were frozen and stored at -80 °C.
For GeneChip ® -experiments extracted RNA from PBL was reverse transcribed using Superscript ® Choice System (Life Technologies).The cDNA was then in vitro transcribed (BioArray HighYield RNA Transcript Labelling Kit, Enzo) to form biotin labeled cRNA.Probe hybridisation to the GeneChip ® probe arrays (HGU95A, Affymetrix), washing, staining and scanning was done according to the instructions of the manufacturer.For validation of specific genes with altered expression validation experiments were done using real-time RT-PCR (TaqMan ® ).
Power analysis based on assumed expression variability and expectable intergroup differences was done.However, it has to be appreciated that every cytokine has a distinct variability of expression, which for most of them is not yet known.Generally, sample sizes of 12 probes for each phenotype tested are in accordance with published articles and current understanding.The primary end point being the differential expression of all tested gene products.For data analysis and mining we used GeneSpring ™ (version 4.0.0,Silicon Genetics) and SPSS ® (version 10.02, SPSS ® Inc) software packages.Two-group gene expression comparison between sarcoidosis patients and healthy controls was done using the Mann-Whitney Utest.A three-group comparison of healthy controls subjects, type I and type II/III-sarcoidosis patients (Kruskal-Wallis-test), as well as other two-group comparisons between two phenotypes were only performed in those cases with a p-value <0.05 [32,33].Due to low n-numbers in the two separated phenotypes of sarcoidosis, the interpre-tation of significant differential expression in these subanalyses requires prudence.Correlation of lung function parameters (DLCO, FEV1, FVC) and gene expression was analysed with Spearman' rank correlations.A conventional significance level of 0.05 was taken, but required prudent interpretation due to multiple significance testing.The study was approved by the local ethical committee.

Results
Patient characteristics are given in table 1. Results of apoptosis-related gene expression in sarcoidosis compared to matched healthy controls are summarised in table 2. Figure 1 shows the comparative expression in sarcoidosis patients and healthy controls of all 12 626 genes and sequence tags tested.A selection of 112 genes with known involvement in apoptosis were selected and discussed below.
We were able to identify significantly altered expression of 53 of 112 (47%) apoptosis-related genes in sarcoidosis compared to the healthy subjects.The alteration in gene expression followed one of three patterns: genes whose expression was altered in both types of sarcoidosis (n = 14), genes with altered expression in type I, but not type II/III sarcoidosis (n = 23) and genes with altered expression in type II/III sarcoidosis only (n = 8).Eight gene products only reached significance when testing both sarcoidosis phenotypes together against controls (n = 8).We were not able to find a gene expressed in type I or type II/III sarcoidosis exclusively.The genes fell into different apoptosis pathways, which are presented below.

Pro-survival cytokine and growth factor genes
The most upregulated growth factors comprise heparin-binding EGF-like growth factor, endothelial cell growth factor 1, platelet-derived endothelial cell growth factor, c-sisplatelet-derived growth factor 2 and vascular endothelial growth factor.Early growth response proteins 1 and 2, as well as the early growth response gene alpha infer increased growth factor receptor signaling.Even though significance was not reached in both sarcoidosis subgroups in all cases the relative expression levels showed a very similar profile in patients with type I and type II/III sarcoidosis.Heparin-binding EGFlike growth factor (DLCO, FEV 1 ) and vascular endothelial growth factor expression (FEV 1 , FVC) correlated with the impairment of lung function.

TNF-and other cell-death receptor pathways
Increased levels for TNFA (type II/III only), TNFR1, TNFA-inducible protein A20 and TSG-14 (type I only) in sarcoidosis patients as a whole indicated increased TNF signaling.TNF-receptor signaling leads either to cell activation/proliferation or induces apoptosis.Critical elements are intracellular adaptor molecules.We found diminished inhibitor of apoptosis (IAP)-levels in type I sarcoidosis, but not type II/III sarcoidosis patients and diminished TRAF5 in both sarcoidosis phenotypes indicating that the cell activation-arm of the TNFsignaling is not activated, especially in the self-limiting type I disease.FADD, a molecule essential for the induction of apoptosis, was not differentially expressed.The cell-death adaptor molecules RAIDD and the apoptosis associated protein GADD34 (in type II/III trend only) were both upregulated indicating increased TNF-related apoptosis.This constellation of intracellular adaptor molecules is compatible with increased apoptosis rather then proliferation signals through TNF-pathways in the sarcoidosis patients as a whole.However, and as the main difference in TNF-signals between both sarcoidosis phenotypes, there was higher expression of TNFA and NFKB in type II and III disease, inferring greater ongoing cell activation and proliferation compared to type I disease.
On the other hand, there was no evidence of differential expression/signaling through Apo3L/ death receptor 3, Fas (CD95)/FasL, TRAIL/ Scanned pseudo-colour image of a hybridised human HUG95A GeneChip ® expression array with enlarged probe set of a gene as an insert top right.This specific microarray shown contains probe sets interrogating approximately 12,000 full-length human genes and some EST clusters from the UniGene database (Build 95).Using reverse transcriptase messenger RNA (mRNA) is by copied into a cDNA population reflecting the relative abundance of the individual mRNAs.In a next step labelled cRNA is generated by in vitro transcription, hybridised to the GeneChip ® expression array, stained with a fluorecent dye and analysed using a laser scanner.The intensity of the hybridisation signal for a given gene is a result of its relative abundance in the mRNA-derived target cRNA.The method is described as providing excellent specificity and reproducibility.Messenger RNA species comprising 1:10,000-100,000 of the mass of the target mRNA, which corresponds to approximately 1 transcript per 100,000, can readily be detected.
Scatter plot of the mean gene expression in sarcoidosis patients compared to controls.Apoptosis-related genes discussed in this publication are highlighted in colour-codes (for abbreviations see table 2), which allows estimation of the extent of the regulated expression in the context of the overall transcriptional activity.Gene expression in both phenotypes is very similar for most of the genes tested resulting in a R-square of 0.98.Also given are cutoffs of 1.3 and 0.7 expression relative to the other phenotype by additional lines.
TRAIL receptor 2 and 3, decoy receptor 2 and TNF-related death receptor 6.The soluble form of the death domain receptor 3 was significantly downregulated in both phenotypes.

Cascade of caspases
The group of caspases did not show differential expression in sarcoidosis compared to controls.

p53 pathway
There was no evidence for increased responses either to DNA-damage or to oxidative stress in sarcoidosis.The Ataxia Telangiectasia gene ATM, induced by DNA damage, showed reduced expression in sarcoidosis (RE = 0.63).The genes for the p53-neutralising protein MDM2A, D, E and MDMX were equally expressed.
Recently, a role for reduced apoptosis has been postulated as contributing towards the accumulation of inflammatory cells in sarcoidosis [34].However, studies in BAL and lung tissue looking for Fas, TNF-receptor 1, TUNEL and electron microscopy were all indicative of increased apoptosis when compared with specimens from healthy controls in these compartments [17][18][19].Nevertheless, it has been suggested but not proven that reduced apoptosis was present in more severe disease when compared to patients with mild forms of sarcoidosis and, therefore, may contribute to disease severity [20].Applying functional genomics, we were able to detect alterations in the expression of several genes associated with cell survival and/or apoptosis in peripheral blood mononuclear cells of patients with sarcoidosis.These alterations in gene expression could be due to genetic (ie, hereditary) or environmental factors.Although the alterations of the expression of growth factor-related genes showed a concerted pro-survival pattern, the situation regarding apoptosis-related genes was less clear-cut.
The pattern of expression of growth factors and pro-survival related genes was generally balanced in favour of a pro-survival climate in sarcoidosis.Particularly upregulated genes were heparin-binding EGF-like growth factor, endothelial cell growth factor 1, platelet-derived endothelial cell growth factor, c-sisplatelet-derived growth factor 2 and vascular endothelial growth factor.Expression of heparin-binding EGF-like growth factor and vascular endothelial growth factor especially, correlated with the impairment of lung function and could thus serve as markers of disease severity.VEGF, which can be induced by TNFA, IL1 and members of the fibroblast growth factor family [35], was upregulated in both sarcoidosis-phenotypes.VEGF can be inhibited by high doses of dexamethasone [36,37].VEGF, expressed by a variety of cells including endothelial cells, is crucial for fibroblast chemotaxis and angioneogenesis.VEGF is involved in lung tissue remodeling as can be seen in patients with chronic inflammatory lung disease and is markedly upregulated under hypoxic conditions [38].
Tumour necrosis factor can either induce cell activation in a NFKB-dependent manner or apoptosis signaling through a cascade of TNF receptor type I, TRADD and the caspases [21,24].Different intracellular adaptor molecules seem to play a pivotal role in this switch.The expression of TNF and related signaling genes is compatible with increased apoptosis signals in both sarcoidosis phenotypes.However, and as an important difference compared to patients with self-limited disease, patients with type II and III disease also seem to signal cell activation / proliferation as demonstrated by the significant upregulation of NFKB in these patients only.The role of the lacking downregulation of the inhibitor of apoptosis IAP1 in type II/III-sarcoidosis is unclear, but might be a possible explanation for the differences observed between the sarcoidosis phenotypes.Interestingly, this gene also correlates with lung function tests.
In contrast to the clear dysregulation of the TNF-pathway, all other death receptors tested, namely Apo3L/death receptor 3, Fas (CD95)/ FasL, TRAIL/TRAIL receptor 2 and 3, decoy receptor 2 and TNF-related death receptor 6, did not show differential expression.The relevance of the significant downregulation of the soluble form of the death domain receptor 3 is unclear.
Of the 19 measured members of the pro-survival Bcl-2 family, 4 were increased and only BCL9 decreased.This is compatible with a clear net prosurvival effect of the Bcl-2 family of genes.Only 1 of 10 pro-apoptotic Bcl-2 family gene, BAX delta, was slightly upregulated (RE 1.15).However, BAX delta is not able to interact directly with APAF1, which is the usual binding partner of pro-survival Bcl-2 genes, and the downstream regulatory element with inhibitory activity on CASP9.BAX delta exerts its pro-apoptotic effect by binding to pro-survival Bcl-2 proteins, which are then unable to bind APAF1.However, quantitatively the up-Discussion regulation of pro-survival genes was much more prominent.Thus, the Bcl-2 family of genes, clearly had an pro-survival profile.
As far as the p53-pathway is concerned, there was no evidence for increased responses either to DNA-damage or to oxidative stress in sarcoidosis.Apart from the Ataxia Telangiectasia gene ATM, induced by DNA damage, which showed reduced expression, the genes for the p53-neutralising protein MDM2A, D, E and MDMX were all equally expressed.
The finding that the expression of apoptosis effector proteases, the caspases, was not altered in sarcoidosis, is not surprising, given that they are produced in excess well in advance of an apoptotic event and when activated induce apoptosis within minutes without the need for any transcriptional activity.
A potential criticism of our study is analysing peripheral blood when the disease activity predominates in the lung.Indeed, the picture in the circulation might represent a mirror image of the true situation, where "disease-modulating" cells have migrated into tissue.Furthermore, it has been suggested that there is compartmentalisation of the immune process in sarcoidosis [40] and that peripheral blood mononuclear cells are quiescent, although several studies refute this.Sarcoidosis is clearly a systemic disease involving activation of many genes in different organs.It is, thus, likely, and underlined by our results, that sufficient "spillover" of local inflammatory phenomena are present in peripheral blood (which circulates approximately once per minute through the lungs).We, therefore, hypothesise that a potential primary alteration in gene regulation, either genetically anchored or triggered by environmental factors, would also be present distant from the actual site of inflammation -the lung.Cell sequestration and the contribution of the different cell types to a specific gene expression pattern need to be further investigated in gene expression studies comparing lung tissue, bronchoalveolar lavage and peripheral blood samples.In exclusively looking at gene expression a limitation of our approach is that we cannot give any statement on the activity of the apoptotic pathways.To do so the data should be combined with analysis of the proteins involved in apoptosis including enzyme assays to evaluate the activity of caspases.Such confirmatory proteinbased studies need to be done as yet.
In summary, using cDNA technology it was possible to demonstrate that peripheral blood mononuclear cells of patients with acute onset sarcoidosis showed differential expression of a number of genes associated with cell survival and/or apoptosis.Gene expression did not show a definite pattern that was suggestive of pro-survival or proapoptosis.Whereas, the TNF-pathway indicated increased apoptosis signaling, the family of Bcl-2 genes and the group of growth hormones clearly had a pro-survival profile.Interestingly, a distinct difference in TNF-signaling between patients with self-limiting type I-sarcoidosis and progressive type II/III-sarcoidosis was identified, which indicates TNF-related proliferation/cell activation signals in the latter.Taken together, the number of genes whose altered expression would be predicted to favour increased survival exceeded that of genes likely to reduce survival.This suggests that the altered gene expression may be determined, at least in part, by systemic factors so that cells are programmed for increased survival before entry into the inflammatory site where local factors may enhance their survival further.

Table 2
Summary