Dr. Ilowite has received consulting fees, speaking fees, and/or honoraria (less than $10,000 each) from Amgen and Xoma.
Gene expression profiling of peripheral blood from patients with untreated new-onset systemic juvenile idiopathic arthritis reveals molecular heterogeneity that may predict macrophage activation syndrome
Article first published online: 29 OCT 2007
Copyright © 2007 by the American College of Rheumatology
Arthritis & Rheumatism
Volume 56, Issue 11, pages 3793–3804, November 2007
How to Cite
Fall, N., Barnes, M., Thornton, S., Luyrink, L., Olson, J., Ilowite, N. T., Gottlieb, B. S., Griffin, T., Sherry, D. D., Thompson, S., Glass, D. N., Colbert, R. A. and Grom, A. A. (2007), Gene expression profiling of peripheral blood from patients with untreated new-onset systemic juvenile idiopathic arthritis reveals molecular heterogeneity that may predict macrophage activation syndrome. Arthritis & Rheumatism, 56: 3793–3804. doi: 10.1002/art.22981
- Issue published online: 29 OCT 2007
- Article first published online: 29 OCT 2007
- Manuscript Accepted: 20 JUL 2007
- Manuscript Received: 7 MAY 2007
- NIH. Grant Number: P01-AR-048929
- Children's Hospital Research Foundation of Cincinnati
Systemic juvenile idiopathic arthritis (JIA) is frequently associated with the development of macrophage activation syndrome. This study was undertaken to better understand the relationship between systemic JIA and macrophage activation syndrome.
Gene expression profiles were examined in 17 patients with untreated new-onset systemic JIA, 5 of whom showed evidence of subclinical macrophage activation syndrome (of whom 2 eventually developed overt macrophage activation syndrome). Peripheral blood mononuclear cells (PBMCs) were separated on Ficoll gradients, and purified RNA was analyzed using Affymetrix GeneChip expression arrays. A fraction of the PBMCs were used for flow cytometry to define the cellular composition of the samples.
Two hundred twenty-five differentially expressed genes (P < 0.05) that distinguished patients with systemic JIA from healthy controls (n = 30) were identified. Clustering analysis indicated that expression patterns correlated with serum ferritin levels. Three main clusters distinguished systemic JIA patients with highly elevated ferritin levels (including those with subclinical macrophage activation syndrome) from those with normal or only moderately elevated ferritin levels. The first cluster comprised genes involved in the synthesis of hemoglobins and structural proteins of erythrocytes. This transcriptional profile was consistent with immature nucleated red blood cells, likely reflective of high red blood cell turnover. Also included were transcripts indicating immature granulocytes. The second cluster was enriched for genes involved in cell cycle regulation. The third cluster was enriched for genes involved in innate immune responses, including those involved in the negative regulation of Toll-like receptor/interleukin-1 receptor–triggered inflammatory cascades and markers of the alternative pathway of macrophage differentiation. Additional differentially expressed genes of interest were those involved in the cytolytic pathway, including SH2D1A and Rab27a.
These data indicate that gene expression profiling can be a useful tool for identifying early macrophage activation syndrome in patients with systemic JIA.
At onset, systemic juvenile idiopathic arthritis (JIA), which constitutes ∼10% of all cases of JIA, is distinguished from other forms of JIA by the prominence of extraarticular features, such as spiking fevers, typical fleeting pink macular rash, generalized lymphadenopathy, hepatosplenomegaly, and occasionally, polyserositis (1). The typical laboratory features include marked polymorphonuclear leukocytosis and thrombocytosis.
The clinical course during the later stages of systemic JIA is highly variable (1). The systemic features tend to subside during the initial months to years of the disease. Approximately half of children with systemic JIA recover almost completely. The other half continue to have progressive involvement of more joints. The joint disease seen in systemic JIA is in some respects quite different from that in the other JIA subtypes. The distinctive features include early destructive changes in the joints and ankylosis involving the cervical spine, wrists, and midfoot (2).
The pathogenetic mechanisms underlying the heterogeneity of systemic JIA are not well understood. Several lines of evidence suggest that the role of the adaptive immune response in systemic JIA may be rather limited compared with its role in other clinical forms of the disease. In contrast, the contribution of innate immunity may be much more prominent (3). On this basis, many view systemic JIA as an autoinflammatory disorder. The relatively good clinical response of systemic JIA to inhibition of interleukin-6 (IL-6) (4, 5) and/or IL-1 (6–8), therapeutic interventions that are effective in autoinflammatory diseases with a known genetic cause (9, 10), is certainly consistent with this view.
A perplexing feature of systemic JIA is its association with macrophage activation syndrome. Macrophage activation syndrome is a severe, potentially life-threatening complication characterized by the excessive activation of well-differentiated macrophages, resulting in fever, hepatosplenomegaly, lymphadenopathy, severe cytopenias, liver disease, and intravascular coagulation (11). Most notably, the activated macrophages in this syndrome exhibit hemophagocytic activity: they phagocytose normal hematopoietic elements, a phenomenon that leads to the development of life-threatening cytopenias. Macrophage activation syndrome is seen usually in association with systemic JIA, and very rarely with other rheumatic diseases. It accounts for much of the significant morbidity and mortality associated with systemic JIA. At least 10% of patients with systemic JIA develop macrophage activation syndrome at some time during the course of the disease (12). Two recent studies demonstrated the existence of “subclinical macrophage activation syndrome” in approximately one-third of patients with new-onset systemic JIA (13, 14).
It is now increasingly recognized that macrophage activation syndrome bears close resemblance to a group of histiocytic disorders collectively known as hemophagocytic lymphohistiocytosis (15, 16). Hemophagocytic lymphohistiocytosis can be primary (familial) or secondary (associated with infection or malignancy) (17). One of the early events in hemophagocytic lymphohistiocytosis/macrophage activation syndrome is an uncontrolled and persistent expansion of activated T lymphocytes and hemophagocytic macrophages. These macrophages express CD163 (13, 14, 18), a scavenger receptor that recognizes hemoglobin–haptoglobin complexes (19). CD163 appears to be expressed mainly on alternatively activated phagocytic macrophages performing “scavenger” functions (20, 21). Increased uptake of hemoglobin–haptoglobin complexes by macrophages leads to increased synthesis of ferritin, an acute-phase response protein that binds free iron to prevent oxidative damage (18). A highly elevated level of serum ferritin is an important diagnostic feature of hemophagocytic syndromes (16, 17).
The most consistent underlying immunologic defect reported in patients with hemophagocytic lymphohistiocytosis has been impairment of cytotoxic functions (16, 17). In ∼40% of patients with primary hemophagocytic lymphohistiocytosis, these immunologic abnormalities have been linked to mutations in the gene encoding perforin, a protein that mediates cytotoxic activity of natural killer (NK) and T cells (22). Other mutations recently implicated in primary hemophagocytic lymphohistiocytosis affect some other proteins involved in the cytolytic pathway (23). Of note, similar immunologic abnormalities, i.e., poor NK cell cytolytic activity, often associated with abnormal levels of perforin expression, have also been reported to distinguish systemic JIA from other forms of childhood arthritis (24).
Identification of the immunologic pathways leading to the increased incidence of macrophage activation syndrome in systemic JIA may help not only with early diagnosis of macrophage activation syndrome, but also with identification of children at risk for macrophage activation syndrome at the time of systemic JIA diagnosis, when different management strategies might avert the development of the syndrome. The microarray technology that allows simultaneous assessment of the expression of thousands of genes has provided a new approach to better understand pathogenesis and to define the molecular basis for clinical heterogeneity of many diseases, including JIA (25) and systemic lupus erythematosus (26). We hypothesized that patients with systemic JIA who have early-stage macrophage activation syndrome could be identified based on their gene expression profiles in peripheral blood, and the present study was conducted to investigate this.
PATIENTS AND METHODS
After written informed consent was provided by their legal guardians, the patients described herein were enrolled in an institutional review board–approved prospective multicenter study of gene expression profiles in childhood arthritis. Systemic JIA was diagnosed based on the International League of Associations for Rheumatology diagnostic criteria (27). The clinical phenotypes of the majority of the patients with systemic JIA have been described previously (13). Nine of the 17 patients with new-onset JIA were male and 8 were female; the median age was 4.5 years. The duration of disease at the time of enrollment ranged from 1 week to 16 weeks (Table 1). Peripheral blood samples were obtained during routine clinic visits prior to the initiation of treatment with disease modifying antirheumatic drugs or steroids. As previously described (13), 5 patients (patients 1–5 in Table 1) had suspected subclinical macrophage activation syndrome based on mild relative cytopenias, hyperferritinemia, and increased serum levels of 2 potential macrophage activation syndrome biomarkers, soluble CD163 (sCD163) and soluble IL-2 receptor (sIL-2R) α-chains (13). Two of these 5 patients eventually developed overt macrophage activation syndrome. The control group consisted of 30 healthy children (11 male and 19 female, median age 11.4 years).
|Disease duration, weeks||12||6||11||3||5||>2||5||12||1||8||16||8||>12||6||9||>12||6|
|Serum ALT, units/liter||NA||16||8||9||29||NA||NA||NA||NA||NA||NA||NA||NA||NA||NA||NA||NA|
|Serum AST, units/liter||NA||60||43||36||21||NA||NA||NA||NA||NA||NA||NA||NA||NA||NA||NA||NA|
|Serum ferritin, ng/ml||3,471||6,228||2,950||1,541||2,470||48||938||1,896||1,247||1,124||1,300||84||63||137||1,339||39||209|
Of the 17 patients with systemic JIA, 12 were from Cincinnati Children's Hospital Medical Center (CCHMC), 2 from Milwaukee, 1 from Children's Hospital of Philadelphia, and 2 from Schneider Children's Hospital. Whole blood was collected using sodium heparin as an anticoagulant. In all participating centers, peripheral blood mononuclear cells (PBMCs) were separated by Ficoll gradient centrifugation, placed in TRIzol within 4 hours of blood collection, frozen, and stored at −80°C. Special effort was made to ensure the uniformity of sample collection and processing in all participating centers. Personnel from all participating centers received training at CCHMC, RNA samples were prepared by a single technician at CCHMC, and all control samples were collected at CCHMC. For immunophenotyping of the samples, a fraction of the PBMCs was placed in freezing media and stored at −80°C. A separate serum sample for ferritin assessment was collected and also stored frozen at −80°C. All samples were shipped overnight on dry ice to CCHMC.
All RNA samples were prepared at the coordinating center (CCHMC) and sent to the CCHMC Affymetrix core for processing. RNA quality was assessed on a Bioanalyzer (Agilent, Wilmington, DE) prior to labeling. For microarray analysis, labeled complementary DNA (cDNA) was synthesized from total RNA using the Ovation Biotin RNA Amplification and Labeling System (NuGen, San Carlos, CA). Labeled cDNA was hybridized to Affymetrix U133plus2.0 GeneChip. Data quality was assessed using the standard metrics of the CCHMC Affymetrix core, which include assessment of positive and negative control features on the arrays. Expression values were derived using the Robust Multiarray Average (RMA) preprocessing method as implemented in GeneSpring GX 7.3 (Agilent). The RMA preprocessing was performed on a large data set that included patients with other diseases. The data discussed in this report have been deposited in NCBI Gene Expression Omnibus (http://www.ncbi.nlm.nih.gov/geo/ [accession no. GSE7753]).
Flow cytometric analysis.
Single-cell suspensions of PBMCs from patients with systemic JIA and healthy controls were stained with fluorescein isothiocyanate (FITC)–conjugated anti-CD16, phycoerythrin (PE)–conjugated anti-CD15, allophycocyanin (APC)–conjugated anti-CD56, peridin chlorophyll protein (PerCP)–conjugated anti-CD3, FITC-conjugated anti-CD8, APC-conjugated anti-CD4, PE-conjugated anti-CD25, FITC-conjugated anti-CD34, APC-conjugated anti-CD19, PerCP-conjugated anti-CD45, and APC-conjugated anti-CD33 (all from BD Biosciences, San Jose, CA). Cells were gated on forward and side scatter parameters and analyzed on a FACSCalibur flow cytometer (BD Biosciences) using CellQuest software. Each cell population is presented as percentage of the total PBMCs.
Microarray expression values were analyzed using algorithms available in GeneSpring GX 7.3. RMA preprocessing was used to control for chip-to-chip variation. Each probe set was then normalized to the median expression of that probe set in the control samples. Differential expression values were identified by analysis of variance and/or Student's t-test. In the Affymetrix U133plus2.0 GeneChip, a number of genes are represented by >1 oligonucleotide probe set. In some analyses, when the list of differentially expressed genes included multiple probe sets representing the same gene, only the probe with the lowest P value was included. Hierarchical clustering using complete linkage was used to group genes and samples by expression patterns.
When the overall number of isolated PBMCs was sufficient, a fraction of the cells was used for flow cytometry to define the cellular composition of the samples to be used in microarray experiments. As shown in Figure 1A, the relative cellular composition of PBMC preparations from patients with systemic JIA (n = 9) was similar to that of controls (n = 30), with the exception of a mild decrease in the proportion of NK cells in the group of patients with systemic JIA (P < 0.05 by Student's t-test). This difference was even stronger when the population of CD56dim NK cells was analyzed separately (P < 0.01 by Student's t-test). There was a trend toward an increased proportion of monocytes in patients with systemic JIA, but the difference from controls did not reach statistical significance. Also worth noting was a small but statistically significant increase in the proportion of immature CD34+ hematopoietic cells in PBMC preparations from patients with systemic JIA (P < 0.05 by Student's t-test) (Figure 1B). There were no statistically significant differences between patients and controls in the numbers of NK T cells, B lymphocytes, CD4+ T lymphocytes, CD8+ lymphocytes, or Treg (CD4+, CD25+) cells.
Genes differentially expressed in patients with new-onset systemic JIA compared with healthy controls.
Microarray-generated gene expression levels were analyzed as described in Patients and Methods. To identify genes that were differentially expressed between healthy controls and patients with untreated new-onset systemic JIA, a rather stringent statistical approach (P < 0.05 by t-test, with Bonferroni correction for multiple comparisons) yielded a list of 284 differentially expressed probe sets. Elimination of the redundant probe sets as described in Patients and Methods yielded a list of 225 differentially expressed genes. The expression of 147 genes was increased, and the expression of 78 genes was decreased, in systemic JIA patients compared with healthy controls.
The median age of the control group (11.4 years) was higher than that of the systemic JIA patients (4.5 years). To ensure that the age difference did not contribute to the differences in gene expression, a further analysis was performed. The control group was divided into 2 subgroups: those above and those below the median age. Subgroups were compared to determine genes whose expression varied with age. This analysis yielded a list of genes (P < 0.05 by t-test, with Bonferroni correction) that did not include any of the genes that had been found to be differentially expressed in patients with systemic JIA compared with controls, suggesting that the age difference did not contribute to the observed differences in gene expression between patients and controls.
Systemic JIA heterogeneity based on clustering analysis of differentially expressed genes.
The 225 differentially expressed genes were analyzed by hierarchical clustering using the complete linkage clustering algorithm, in which distance between samples is inversely proportional to similarity (Figure 2). The systemic JIA patients clustered together, with the exception of 1 individual. Interestingly, this patient (patient 17 in Table 1) was asymptomatic at the 9-month followup, after treatment with only nonsteroidal antiinflammatory drugs. In contrast, all other patients with new-onset systemic JIA had persistent disease. Although these systemic JIA patients clustered, heterogeneity in gene expression was noted, and was reflected in higher-level “branching” of the clustering tree (Figure 2).
At least 5 distinct gene clusters contributed to the heterogeneity of systemic JIA (Figure 2). Two clusters included genes whose expression was lower in systemic JIA patients compared with healthy individuals. Cluster 1 comprised various genes. Cluster 2 was enriched for genes involved in cell physiology, genes encoding NK receptors (including several killer cell lectin- and immunoglobulin-like receptors), and genes involved in the regulation of cell cycle and apoptosis.
The genes in the remaining 3 clusters exhibited higher expression in patients with systemic JIA compared with healthy controls. Cluster 3 was enriched for genes implicated in the innate immune response to pathogens, most notably genes involved in negative regulation of such responses (e.g., SOCS3 [the gene for suppressor of cytokine signaling protein 3, or SOCS-3] and ADM [the gene for adrenomedullin]). It also included markers of the monocyte/macrophage pathway of cell differentiation (CD64, MS4A4A [the gene for membrane-spanning 4 domain, subfamily A, member 4], and GPR84 [the gene for orphan G-protein–coupled receptor 84]). Several genes involved in the regulation of coagulation (genes for protein S, hemostatic tissue factor inhibitor, and thrombospondin) were also present in this cluster. Cluster 4 was enriched for genes involved in the regulation of cell cycle, chromatin/nucleosome assembly, regulation of transcription, and apoptosis. Cluster 5 was highly enriched for genes involved in the processes of hemoglobin synthesis and oxygen transport, genes encoding structural proteins of red blood cells (RBCs), as well as genes encoding rare hemoglobins, suggesting that this cluster of genes might represent the signature of the immature nucleated RBCs that can copurify with PBMCs isolated on Ficoll gradients.
Clinical phenotypes in relation to clustering.
To determine whether clinical phenotypes in these patients with systemic JIA correlated with the gene clusters, we compared the position of individual patients in the clustering tree with their clinical features. All patients had arthritis and active systemic disease (i.e., characteristic spiking fevers and rash) at the time of sampling. Laboratory findings were reflective of high inflammatory activity in all patients (Table 1) and did not correlate with patient clustering. The extent of the joint disease was highly variable, ranging from mild oligoarticular to severe polyarticular disease, and failed to correlate with the observed clustering (data not shown).
Since one of the major diagnostic criteria of hemophagocytic syndromes is extreme hyperferritinemia, we assessed ferritin levels in the serum samples from systemic JIA patients collected at the same time as the PBMC samples for gene expression profiling. As shown in Table 1, although the levels were highly variable, there were 2 distinct subgroups of patients: those with normal or only slightly elevated levels of ferritin (≤209 ng/ml) and those with very high ferritin levels that were comparable with levels observed in hemophagocytic syndromes (≥938 ng/ml). As shown in Figure 2, there was a trend toward separation of these 2 groups of patients in the clustering tree. The 5 patients in whom subclinical macrophage activation syndrome was suspected based on mild relative cytopenias and increased levels of sCD163 and sIL-2R α-chains (2 putative macrophage activation syndrome biomarkers ) clustered within the “high-ferritin” group. Since separation based on the degree of hyperferritinemia may reflect important pathogenetic differences, the subsequent analysis included comparison between “normal-ferritin” and “high-ferritin” groups.
The “innate immune response” cluster.
Some of the genes comprising the innate immune response cluster are listed in Table 2. The list included the monocyte markers CD64 and MS4A4A. There were several genes whose expression is strongly up-regulated in the innate immune response, including those triggered by Toll-like receptor (TLR)/IL-1R signaling (SOCS3, ADM, BPI [the gene for bactericidal/permeability increasing protein], INHBA [the gene for activin A, or inhibin βA], SLPI [the gene for secretory leukoprotease inhibitor], PGLYRP1 [the gene for peptidoglycan recognition protein 1], and RETN [the gene for resistin]). Other genes in the cluster included solute carrier MSCP (the gene for solute carrier family 25 member 37 [also known as mitoferrin]), CEACAM1 (the gene for carcinoembryonic antigen–related cell adhesion molecule 1), and CEACAM6/6. There were 2 neutrophil-derived proteins (ELA2 [the gene for elastase 2] and MPO [the gene for myeloperoxidase]), and FAM20A, a protein that plays a role in myelocytic lineage commitment (28), was included. The cluster also included the genes for haptoglobin and haptoglobin-related protein. In our data set, the levels of expression of 12 of the 16 genes in the cluster were significantly higher in the group of patients with high ferritin levels (Figure 3A).
|Description||Common name||Probe ID||P||Fold change†|
|Solute carrier family 25 member 37||MSCP||226179_at||3.41 × 10−8||1.82|
|Elastase 2, neutrophil||ELA2||206871_at||1.88 × 10−6||2.81|
|Resistin||RETN||220570_at||1.10 × 10−6||2.95|
|Hypothetical protein MGC17301||MGC17301||227055_at||9.92 × 10−5||2.00|
|Eukaryotic translation initiation factor 2c, 2||EIF2C2||225827_at||8.48 × 10−5||2.00|
|Ubiquitin-specific protease 32||USP32||211702_s_at||3.68 × 10−6||2.02|
|Peptidoglycan recognition protein 1||PGLYRP1||207384_at||2.33 × 10−7||2.34|
|Bactericidal/permeability increasing protein||BPI||205557_at||4.82 × 10−5||8.72|
|Ribonuclease, RNase A family 3||RNASE3||206851_at||0.000271||5.83|
|Secretory leukocyte protease inhibitor||SLPI||203021_at||2.82 × 10−5||5.08|
|Carcinoembryonic antigen–related cell adhesion molecule 6||CEACAM6||203757_s_at||8.88 × 10−5||8.16|
|Myeloperoxidase||MPO||203949_at||3.75 × 10−5||4.53|
|Carcinoembryonic antigen–related cell adhesion molecule 1||CEACAM1||209498_at||0.000342||2.68|
|Haptoglobin||HP||206697_s_at||4.43 × 10−6||2.93|
|Haptoglobin-related protein||HPR||208470_s_at||2.02 × 10−6||2.98|
|Suppressor of cytokine signaling 3||SOCS3||206359_at||5.52 × 10−6||1.26|
|Adrenomedullin||ADM||202912_at||5.31 × 10−6||1.85|
|Family with sequence similarity 20, member A||FAM20A||241981_at||6.26 × 10−9||2.35|
|Membrane-spanning 4 domain, subfamily A, member 4||MS4A4A||1555728_a_at||1.04 × 10−7||1.71|
The gene encoding haptoglobin was one of the most highly overexpressed genes not only in the innate immune response cluster, but also in the entire data set. Since haptoglobin is associated with the pathogenetic pathways relevant to macrophage activation syndrome, we then investigated for genes with expression patterns closely correlated with the expression of the haptoglobin gene. In addition to the entire previously described innate immune response cluster, the group of haptoglobin-correlated genes included several genes encoding antiinflammatory molecules, such as IL-1R antagonist and the light chain of ferritin. Pearson correlation coefficients for these genes were ≥0.7, suggesting that their expression was highly coregulated.
Cytolytic pathway in systemic JIA patients with high versus normal ferritin levels.
The development of the primary hemophagocytic syndromes has been linked to the abnormal cytolytic function of NK cells, NK T cells, and cytotoxic CD8+ T cells. Thus, we were interested in whether the 2 groups of systemic JIA patients could be distinguished based on levels of expression of the genes whose products are involved in the cytolytic pathway. We focused on the genes that are 1) involved in the cytolytic pathway and 2) associated with the primary genetic hemophagocytic syndromes. As shown in Figure 3B, the levels of expression of 2 genes, SH2D1A (X-linked lymphoproliferative disease) and Rab27a (type 2 Griscelli syndrome) differed significantly between the normal-ferritin and high-ferritin groups of patients with systemic JIA (P < 0.05 by Student's t-test).
The current study of gene expression profiles in the peripheral blood of patients with untreated new-onset systemic JIA revealed a strong signature that distinguished patients from healthy controls. Considering the highly distinct clinical phenotype and the systemic nature of the disease, this was not surprising. The more intriguing finding was the heterogeneity of patients with systemic JIA, correlating with some clinical features of macrophage activation syndrome. Thus, hierarchical clustering analysis of the differentially expressed genes revealed at least 2 subgroups of patients with several clusters of genes contributing to the observed heterogeneity. Overall, the expression patterns correlated with serum levels of ferritin, and patients with subclinical macrophage activation syndrome clustered within the high-ferritin group. Since a very high serum level of ferritin is an important diagnostic feature of hemophagocytic syndromes, this distinction may reflect important differences relevant to the pathogenesis of macrophage activation syndrome.
We have termed one group of genes that distinguished high-ferritin versus normal-ferritin systemic JIA an “erythropoiesis” cluster. Based on the genes present in this cluster, it likely represents immature nucleated RBCs that are present in PBMC preparations. Since a fraction of the immature nucleated RBCs are CD34+ (29), these cells probably contribute to the small but statistically significant expansion of CD34 cells in patients with systemic JIA compared with healthy controls, as determined by flow cytometric analysis (Figure 1). The increased numbers of these cells in most patients with systemic JIA probably reflect the highly increased turnover of RBCs. The fact that this signature is particularly strong in patients with high ferritin levels suggests that these patients may have increased destruction of RBCs, perhaps due to subclinical hemophagocytosis. Consistent with this notion, subclinical macrophage activation syndrome and mild hemophagocytosis have been demonstrated in approximately one-third of patients with new-onset systemic JIA (13, 14).
The main trends in another interesting cluster, designated the “innate immune response” cluster, included 1) increased expression of genes associated with monocyte/macrophage lineage (CD64, GPR84, and MS4A4A), 2) increased expression of antimicrobial peptides (e.g., BPI, PGLYRP1), whose expression appears to be induced by TLR/IL-1R signaling pathways (30), and 3) increased expression of genes involved in the negative feedback regulation of innate inflammatory response (e.g., SOCS3, ADM, INHBA, SLPI, HP).
The presence of MS4A4A in the innate immune response cluster is intriguing. MS4A4A is a member of the family of molecules expressed in hematopoietic cells at different stages of their differentiation (31). Expression of MS4A4A strongly increases in monocytes undergoing differentiation into macrophages, especially those following the alternative pathway of macrophage differentiation (32). GPR84 is another marker whose expression is markedly induced in monocytes differentiating into macrophages (33). In our previous study, we noted highly increased serum levels of sCD163, suggestive of expansion of CD163+ macrophages, in a subgroup of patients with systemic JIA (13). Surprisingly, in the current data set the expression of CD163 itself was not significantly higher in patients with systemic JIA compared with controls. This is likely related to the fact that by the time macrophages up-regulate the expression of CD163, they tend to leave the peripheral circulation and accumulate in tissue. The small expansion of monocytes in PBMC preparations from patients with systemic JIA (Figure 1) might have contributed to the increase in the numbers of transcripts of the monocyte/macrophage markers. However, a very high degree of overrepresentation of these transcripts (≥1.7-fold change, compared with only a mild degree of expansion of monocytes) suggests that the observed expression patterns are related to increased gene expression in these cells, rather than solely to the increase in their numbers.
The activated status of the myeloid cells in systemic JIA is also evident based on the increased numbers of transcripts whose expression was induced during the innate immune responses triggered by TLR/IL-1R signaling. This includes such genes in the immune cluster as ADM (34, 35), BPI (36), INHBA (37, 38), PGLYRP1 (30, 39), and SLPI (40). These findings are consistent with previous reports implicating IL-1 in the pathogenesis of systemic JIA (6–8). Interestingly, products of many genes in the cluster have been shown to participate in the negative feedback loops involved in the down-regulation of the TLR/IL-1R–mediated inflammation. One example is SOCS-3. It belongs to the SOCS family of proteins, which are up-regulated during innate immune responses, in which they are key negative regulators of cytokine signaling (41–43). In our data set, the innate immune response cluster was much more prominent in the group of patients with high ferritin levels (Figure 3A).
The genes for haptoglobin and haptoglobin-related protein were other genes in the innate immune response cluster. Since hemoglobin–haptoglobin complexes are recognized by CD163, we identified genes whose expression patterns closely correlated with the expression of the haptoglobin gene. In addition to the entire innate immune response cluster, the list of the haptoglobin-correlated genes included those for IL-1R antagonist and the light chain of ferritin. This observation suggests that high levels of ferritin seen in a subgroup of patients with systemic JIA may be a part of the antiinflammatory cascade described above. This is consistent with current understanding of the role of haptoglobin and ferritin in prevention of oxidative damage induced by iron-derived reactive oxygen species (Figure 4), a phenomenon that likely accompanies increased RBC turnover.
Results of the clustering analysis, as well as the analysis of haptoglobin-correlated genes, suggest coregulated expression of the marker of the alternative pathway of macrophage differentiation (MS4A4A) (32), haptoglobin, and several negative regulators of innate immune responses, including SOCS-3. SOCS-3 is induced by a number of proinflammatory cytokines, most notably, IL-1 (41, 42) and IL-6 (43). It interferes with the signaling pathways of these cytokines, thus providing a negative feedback loop in inflammatory cascades induced by IL-1/IL-6. SOCS-3 may also alter the responsiveness of monocytes to interferon-γ (IFNγ) (41) and induce switching from Th1 to Th2 immune responses (44).
Coregulated expression of the markers of the alternative pathway of macrophage differentiation and several negative regulators of innate immune responses including SOCS-3 raises the question as to whether in new-onset systemic JIA, the induced antiinflammatory cascade might contribute to the skewing of the differentiation of monocytes toward the development of scavenger macrophages, which exhibit increased phagocytic activity and are CD163+. Consistent with this hypothesis is the observation that dendritic cells (DCs) transduced with SOCS3 exhibit a DC type 2 phenotype that directs Th2 responses (44). Although similar experiments have not been performed on macrophages, SOCS3 overexpression has been associated with the shift toward alternatively activated macrophage responses, at least in some experimental systems (45). Also in accordance with the hypothesis, increased expression of INHBA, another gene in the cluster, has also been associated with the alternative pathway of macrophage differentiation (38).
Comparisons between systemic JIA patients grouped according to high versus normal ferritin levels also revealed statistically significant differences in the levels of expression of several genes whose products are critical to the activation of the cytolytic pathway. This includes Rab27a and SH2D1A (46, 47). Mutations in these genes have been implicated in the development of type 2 Griscelli syndrome and X-linked lymphoproliferative disease, both associated with hemophagocytic syndrome. These observations suggest that the pathways that distinguish high-ferritin versus normal-ferritin systemic JIA may also be relevant to the development of the cytolytic abnormalities seen in patients with the disease. In contrast, perforin expression was low in most patients with systemic JIA, consistent with the trend toward lower numbers of NK cells as determined by flow cytometric analysis. The trend was particularly strong for CD56dim NK cells, a cell population that has been associated with particularly high levels of perforin expression (48).
Another interesting finding was the relative paucity of genes whose expression is induced by IFNγ. The scarcity of IFN-induced genes in the list of genes distinguishing systemic JIA patients from controls is intriguing since the presence of IFNγ in the serum of patients with systemic juvenile idiopathic arthritis has been demonstrated in many studies. One possible explanation is the altered responsiveness of immune cells to IFNγ. Indeed, SOCS3, one of the most highly overexpressed genes in our data set, has been shown to interact with the JAK/STAT signaling pathways, leading to decreased responsiveness of monocyte/macrophages to IFNγ (41).
Taken together, these observations support the hypothesis that the antiinflammatory negative feedback loops induced by innate immune responses involving IL-1 and IL-6 may contribute to conditions that favor the alternative pathway of macrophage differentiation, leading to the development of the CD163+ “scavenger” phenotype. This phenotype is associated with increased phagocytic activity and has been implicated in the development of hemophagocytic syndromes. Interestingly, similar pathways involving mild hemophagocytosis have been described in septic shock syndrome (49, 50), another TLR-triggered phenomenon in which IL-1 and IL-6 play a major role. In the presence of cytotoxic abnormalities, this may create a “set-up” for the development of full-blown macrophage activation syndrome.
In summary, gene expression profiling of PBMCs from patients with untreated new-onset systemic JIA revealed molecular heterogeneity that corresponded to serum ferritin levels, a clinical feature that is important in the diagnosis of hemophagocytic syndromes including macrophage activation syndrome. The 3 main clusters of genes that contributed to the heterogeneity are those involved in erythropoiesis, cell cycle regulation, and negative regulation of the innate immune response (particularly those induced by TLR/IL-1R signaling). Based on these data we propose that the gene expression signature that includes these clusters may help identify patients at risk for the development of full-blown macrophage activation syndrome, in whom closer observation as well as timely treatment modifications may be required.
Dr. Grom had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study design. Fall, Barnes, Thornton, Ilowite, Gottlieb, Griffin, Sherry, Thompson, Glass, Colbert, Grom.
Acquisition of data. Fall, Barnes, Luyrink, Olson, Ilowite, Gottlieb, Griffin, Sherry.
Analysis and interpretation of data. Fall, Barnes, Thornton, Griffin, Thompson, Glass, Colbert, Grom.
Manuscript preparation. Fall, Barnes, Thornton, Luyrink, Olson, Ilowite, Gottlieb, Griffin, Sherry, Thompson, Glass, Colbert, Grom.
Statistical analysis. Fall, Barnes.
- 4Phase II trial of anti-IL-6 receptor antibody (MRA) for children with systemic-onset juvenile idiopathic arthritis [abstract]. Arthritis Rheum 2003; 48 Suppl 9: S429., , , , , , et al.
- 6Treatment of systemic onset juvenile rheumatoid arthritis with anakinra [abstract]. Arthritis Rheum 2004; 50 Suppl 9: S437–8., , , .
- 13The diagnostic significance of soluble CD163 and soluble interleukin-2 receptor α-chain in macrophage activation syndrome and untreated new-onset systemic juvenile idiopathic arthritis. Arthritis Rheum 2007; 56: 965–71., , , , , , et al.
- 16Natural killer cell dysfunction: a common pathway in systemic-onset juvenile rheumatoid arthritis, macrophage activation syndrome, and hemophagocytic lymphohistiocytosis? [review]. Arthritis Rheum 2004; 50: 689–98..
- 39Chemically synthesized pathogen-associated molecular patterns increase the expression of peptidoglycan recognition proteins via toll-like receptors, NOD1 and NOD2 in human epithelial cells. Cell Microbiol 2005; 7: 675–8., , , , , , et al.