|Home | About | Journals | Submit | Contact Us | Français|
Langerhans-cell histiocytosis (LCH) is a rare disease characterized by heterogeneous lesions containing CD207+ Langerhans cells and lymphocytes that can arise in almost any tissue and cause significant morbidity and mortality. After decades of research, the cause of LCH remains speculative. A prevailing model suggests that LCH arises from malignant transformation and metastasis of epidermal Langerhans cells. In this study, CD207+ cells and CD3+ T cells were isolated from LCH lesions to determine cell-specific gene expression. Compared to control epidermal CD207+ cells, the LCH CD207+ cells yielded 2113 differentially-expressed genes (FDR<0.01). Surprisingly, expression of many genes previously associated with LCH, including cell-cycle regulators, pro-inflammatory cytokines and chemokines were not significantly different from control LCs in our study. However, several novel genes whose products activate and recruit T cells to sites of inflammation, including SPP1 (osteopontin), were highly over-expressed in LCH CD207+ cells. Furthermore, several genes associated with immature myeloid dendritic cells were over-expressed in LCH CD207+ cells. Compared to the peripheral CD3+ cells from LCH patients, the LCH lesion CD3+ cells yielded only 162 differentially-regulated genes (FDR<0.01), and the expression profile of the LCH lesion CD3+ cells was consistent with an activated regulatory T cell phenotype with increased expression of FOXP3, CTLA4 as well as SPP1. Results from this study support a model of LCH pathogenesis in which lesions do not arise from epidermal Langerhans cells, but from accumulation of bone-marrow derived immature myeloid dendritic cells that recruit activated lymphocytes.
Langerhans-cell histiocytosis (LCH) is a potentially fatal disease characterized by invasive lesions infiltrated with multiple cell types including CD1a+/CD207+ cells presumed to be pathologic Langerhans cells (LCs). The incidence of LCH is approximately 5 cases per million children and 1/10,000 live births per year (1). Approximately 30% as many adults are afflicted, though this incidence is likely underestimated (2). LCH includes a spectrum of clinical presentations from single system involvement in skin or bone, to diffuse multisystem involvement of liver, lungs, bone marrow, CNS and other organ systems (Reviewed in 3,4). Patients with limited organ system involvement have a very good prognosis, and patients with multisystem disease have survival rates of approximately 80% (5). However, survival is poor in patients with high risk multisystem disease who fail to respond to induction therapy (6). Chemotherapy for LCH is based on a lymphoma model of general immune suppression and cytotoxicity to rapidly proliferating cells (Reviewed in 7).
The etiology of LCH is not known. Scientific debate has focused on LCH resulting from malignant transformation or from functional proliferation of epidermal Langerhans cells in response to external stimuli (8–12). Regardless of whether or not clonal proliferation is the initiating factor in LCH, the CD207+ cells from these lesions require multiple interactions with other cell types including T cells, eosinophils and macrophages (13). LCs will not grow in isolation in vitro or as xenografts in immune-deficient mice. The mainstay of research on LCH tissues has been immunohistochemical analysis of biopsy samples. While this approach has been useful, it is also limited by various difficulties such as testing multiple antibodies simultaneously, variable sensitivity of antibodies, inability to quantitatively interpret results and lack of control tissues. RNA and protein studies from whole biopsy samples are also difficult to interpret due to the heterogeneous composition of LCH lesions. In order to overcome some of the experimental challenges in studying LCH, we have devised a robust procedure to study cell-specific gene expression profiling in the cells that most likely contribute to pathology in LCH patients: LCs (CD207+) and T cells (CD3+).
LCH diagnosis was established by the presence of CD1a+ or CD207+ histiocytes in clinical biopsy specimens. Samples from the 15 individuals with LCH in this study included patients with relapsed disease and high risk multi-system disease (Supplemental Table IA). Control epidermal Langerhans cells were isolated from discarded skin, primarily elective circumcisions, from patients under 18 years of age. Control tonsil CD3 cells were isolated from discarded samples from elective tonsillectomy in patients under 18 years of age. Studies were performed according to protocols approved by the Institutional Review Board of Baylor College of Medicine.
Fifteen fresh LCH biopsy samples were collected. They were transported in RPMI media (Invitrogen, Carlsbad, CA) and processed within 24 hours. All samples were processed into single cell suspension over a 70 uM mesh filter. Cells were washed twice with RPMI supplemented with 10% fetal bovine serum (FBS), then incubated with conjugated antibodies, CD207-PE (Beckman Coulter, Fullerton, CA) and CD3-FITC (BD Bioscience, San Jose, CA) for 30 minutes on ice. Cells were washed again and resuspended in RPMI/FBS with 2 ug/ml propidium iodide (Molecular Probes, Eugene, OR). Cells were then separated by flow cytometry by gating on the propidium iodide-negative population and antibody-specific fluorescence. Cells were sorted with a MoFlo Sorter (Fullerton, CA) directly into PicoPure RNA Extraction Buffer (Molecular Devices, Sunnyvale, CA) (Supplemental Table IB). Select samples were re-analyzed by flow cytometry for purity (Figure 1).
Control LCs were isolated from 12 skin samples that were transported in RPMI media, and processed within 24 hours. Tissue was incubated in RPMI with 5 units/ml dispase II (Roche, Indianapolis, IN) at 4°C for 8 hours prior to separation of the epidermal layer. The epidermal layer was further treated with 0.25% trypsin-EDTA (Invitrogen, Carlsbad, CA) for 15 minutes at 37°C, then Langerhans cells were isolated with CD207-PE conjugated antibody (Beckman Coulter, Fullerton, CA) as described above. Patient details from the individual skin samples were not available. However, review of elective circumcisions performed at Texas Children’s Hospital in an operating room shows a range of ages from 1 month to 18 years.
Peripheral T cells were isolated from 7 patients with active LCH prior to chemotherapy (Supplemental Table I). Peripheral blood was collected and stored in EDTA+ tubes, then processed within 24 hours. Peripheral blood mononuclear cells were isolated after centrifuging blood over a Histopaque-1077 (Sigma-Aldrich, St. Louis, MO) gradient at 450G for 30 minutes. PBMCs were washed twice in RPMI, and then T cells were isolated with CD3-FITC (BD Bioscience, San Jose, CA) conjugated antibodies as described above.
T cells were isolated from 20 tonsil samples from children who underwent elective tonsillectomy. Tonsils were processed in an identical fashion to the LCH lesions, and CD3+ cells were isolated by flow cytometry with CD3-FITC antibody as described above.
Total RNA was processed and cDNA amplified for 13 LCH-CD207 samples, 12 control-CD207 samples, 12 LCH lesion-CD3 samples, 7 LCH-peripheral CD3 samples and 4 pooled tonsil-CD3 samples (5 individual RNA samples/tonsil pool). Total RNA was processed from the sorted cells according to the PicoPure RNA Isolation Kit protocol (Molecular Devices, Sunnyvale, CA). RNA concentration and quality was verified using the 6000 Pico Chip (Agilent, Santa Clara, CA) at the Baylor College of Medicine Microarray Facility. Any samples with a detectable RNA integrity number (14) (RIN)<5 were excluded from the study. cDNA amplification was performed with the WT-Ovation Pico System according to manufacturer’s protocol (NuGen, San Carlos, CA). This is a whole-transcriptome amplification system with which we were able to generate 4–6 μg of cDNA from 700 pg-25 ng of input RNA. Fragmented and biotinylated cDNA samples for the gene chip studies were generated using the FL-Ovation cDNA Biotin Module v2 (Nugen, San Carlos, CA).
Fragmented and biotinylated cDNA was hybridized to Affymetrix U133 Plus 2.0 chips (Affymetrix, Santa Clara, CA). cDNA was generated from each of the LCH-CD207, control-CD207, LCH lesion-CD3, and peripheral-CD3 samples and was used to hybridize onto individual chips. Four tonsil-CD3 cDNA were generated, each from 5 pooled cDNA samples. Chips were hybridized overnight at 45°C with 60 revolutions per min for 16 hours. The chips were processed in the fluidic station using the FS450_004 wash and stain protocol according to manufacturer’s instruction. The stained chips were scanned using the Affymetrix GeneChip scanner 3000. The datafiles have been deposited in the Gene Expression Omnibus website (http://www.ncbi.nlm.nih.gov/geo/; Accession number: GSE16395).
Quality control of 48 Affymetrix U133 plus 2 gene expression chips was done using the BioConductor package affyQCReport (15). Beta actin and GAPDH ratios as well as signal distribution were assessed to determine the outlier cases. Normalization and probe set summarization was done in BRB-Arraytools (available at http://linus.nci.nih.gov/BRB-ArrayTools.html) using the Robust Multichip Average algorithm. Hierarchical clustering was performed using centered correlation and average linkage. The Significance Analysis of Microarrays algorithm (16) was used for analysis of differential expression with a False Discovery Rate (FDR) of less than 0.01. Samples were split into three groups for analysis. Group 1 consisted of 13 CD207 tumor samples and 12 CD207 controls obtained from normal skin. Group 2 consisted of 7 CD3 LCH lesion samples and 7 CD3 matched paired samples obtained from patients’ peripheral blood. Group 3 consisted of 12 CD3 LCH lesion samples and 4 CD3 pooled controls obtained from tonsil. Heatmaps and other graphics were created using MultiExperiment Viewer, part of the TM4 Microarray Software Suite (17).
Real-time PCR reactions were performed with TaqMan Gene Expression Assays (Applied Biosystems, Carlsbad, CA), which includes a mix of 2 unlabeled PCR primers (900 nM final concentration) and 1 FAM dye-labeled TaqMan MGB probe (250 nM final concentration). Single-stranded cDNA was generated by RNA amplification as described above in experiments independent from the amplifications used to generate cDNA probe for microarray studies. For the CD207+ analysis, one cDNA pool was made from equal contribution from the 13 LCH CD207+ samples and another from equal contribution from the 12 control CD207+ samples. For the CD3+ analysis, one cDNA pool was made from equal contribution from the 7 peripheral LCH CD3+ samples and another from equal contribution from the 7 matched LCH lesion CD3+ samples. Each reaction included 20 ng cDNA. TaqMan Fast Universal PCR Master Mix (Applied Biosystems) was used in 25 μl reactions in 96 well plates on a iQ5 Real Time PCR Detection System (Bio-Rad, Hercules, CA). Assays were plated in triplicate in two independent experiments. Thermal cycling conditions were set at 2 minutes at 50°C and 10 minutes at 95°C, then 40 cycles of 95°C and 60°C for 1 minute. Relative mRNA levels were determined using standard ddCt calculations, with expression levels of experimental samples normalized to GAPDH (18). Taqman probesets (Applied Biosytems) included AFF3, Hs00289335_s1; CCL5, Hs00174575_m1; CCR1, Hs00174298_m1; CD36, Hs01567188_g1; CD74, Hs00269961_m1; CDH1, Hs00170423_m1; CEACAM6, Hs00366002_m1; CTLA4, Hs00175480_m1; DCAL1, Hs00416849_m1; DUSP4, Hs01027785_m1; FOXP3, Hs00203958_m1; GAPDH, Hs02758991_g1; HLA-DRA, Hs00219578_m1; HOXB7, Hs00270131_m1; HSPA1A, Hs00359147_s1; IL17A, Hs00174383_m1; IL2, Hs00174114_m1; IL8, Hs00174103_m1; MMP1, Hs00899660_g1; MMP9, Hs00234579_m1; NRP1, Hs00826128_m1; PERP, Hs00751717_s1; S100A7, Hs00162488_m1; S100A8, Hs00374263_m1; SMYD3, Hs00224208_m1; SPP1, Hs00959010_m1; TACSTD1, Hs00158980_m1; TNFRSF9, Hs00155512_m1; and VNN1, Hs01546812_m1.
Fresh LCH biopsy tissue, skin samples and tonsil samples were snap frozen in embedding medium (Tissue-Tek O.C.T. Compound; Sakura Finetech USA Inc, Torrance, CA) for frozen tissue specimens and stored at −80°C within 8 hours from the time of collection. Tissue was cut at a thickness of 4–8μm from frozen blocks and adhered to microscope slides. Slides were stored at −80°C until use. After removal from storage, slides were air-dried for one half-hour and tissue was subsequently fixed by incubating slides in ice-cold acetone for 15 minutes. After being air-dried for 10 minutes, the tissue was then rehydrated by incubation in phosphate-buffered saline (PBS) for 10 minutes.
Immunohistochemical staining was performed using the Dako EnVision + Dual Link Kit (Carpinteria, CA). Endogenous peroxidase was blocked using Dako Block Buffer for 10 minutes. Primary antibodies were diluted in Dako Antibody Diluent and tissue incubated at room temperature for 3 hours or at 4°C overnight. Antibodies against CD11b and CD11c were diluted 1:10. All other antibodies were diluted 1:50. Mouse and rabbit antibodies were detected using the previously diluted HRP-linked secondary antibody included in the kit. Goat antibodies were detected using anti-goat/horseradish peroxidase (HRP) antibody (Dako) diluted 1:100. Slides were washed 3 times for 5 minutes each in Dako Wash Buffer between incubations. HRP activity was detected using Dako Liquid DAB+ Substrate Chromagen System. Tissue was then stained with hematoxylin, coverslipped and stored at room temperature. Antibodies used: CD207 – mouse monoclonal – ab49730 (Abcam, Cambridge, MA); CD3 – mouse monoclonal – ab699 (Abcam); HLA-DR+DP+DQ – mouse monoclonal sc59251 – (Santa Cruz Biotechnology, Santa Cruz, CA); CTLA4 – goat polyclonal – sc1628 (Santa Cruz Biotechnology); Osteopontin – mouse monoclonal – sc21742 (Santa Cruz Biotechnology); CD44 – rabbit monoclonal ab51037 – (Abcam); Neuropilin1 – mouse monoclonal – sc5307 (Santa Cruz Biotechnology); Vanin1 – goat polyclonal – sc16776 (Santa Cruz Biotechnology); CEACAM6 – rabbit polyclonal – ab56234 (Abcam); MMP1 – rabbit monoclonal – ab52631 (Abcam); MMP9 – goat polyclonal – sc6840 (Santa Cruz Biotechnology); CD13 – goat polyclonal –sc6995 (Santa Cruz Biotechnology); CD33 rabbit polyclonal – ab59940 (Abcam); CD1d – mouse monoclonal – ab11076 (Abcam); ICAM1 – mouse monoclonal – ab20 (Abcam); CD11b – mouse monoclonal–MAB1699 (R&D Systems, Minneapolis, MN); CD11c – mouse monoclonal – MAB1777 (R&D Systems). Images were magnified using the Olympus BX51 microscope, 40X objective. Images were captured using the Olympus DP71 digital camera with Olympus DP Controller and DP Manager software.
Validity of the data derived from these gene expression profiling experiments is predicated on fidelity of the methods used to isolate cells and generate cDNA probe. Cells were sorted using fluorescent markers conjugated to CD3 and CD207 antibodies using standard methods (Figure 1). Cell purity was evaluated in several samples by sorting cells into RPMI, then re-analyzing with the MoFlo Sorter. Purity exceeded 95% in all CD3+ and CD207+ samples tested (Figure 1). The Nugen WT-Ovation Pico System was used to amplify RNA to generate cDNA probe for the gene expression profiling studies. The yield of cDNA consistently ranged between 4 μg and 6 μg. Previous studies have validated this experimental strategy (19,20). In this series of experiments, gene expression datasets generated from technical replicates of amplified probe from 5 ng input RNA had a correlation > 90% (data not shown).
Cell-specific gene expression from LCH biopsy specimens was evaluated by comparing hybridization signal from amplified cDNA on Affymetrix gene chips (U133A Plus 2.0) (Table I). Three sets of comparisons were performed: 1) Thirteen LCH CD207+ samples were compared to 12 control skin LCH CD207+ samples. 2) Seven LCH lesion CD3+ samples were compared to 7 peripheral blood CD3+ samples from the same patients. 3) Twelve LCH lesion CD3+ samples were compared to 4 pooled control tonsil CD3+ samples (5 individual tonsil CD3+ samples/pool). Validity of the data derived from this experimental strategy is evidenced by cell-specific grouping of the datasets in dendrograms constructed with unsupervised clustering (Supplemental Figure 1).
Analysis of global gene expression in LCH-CD207 cells shows a consistent and distinct profile from control skin-CD207 cells, with overall correlation < 0.2. While the LCH-CD207 datasets are grouped with unsupervised clustering, the overall inter-group correlation > 30%. By comparison, the control skin-CD207 cells show less overall variability with an inter-group correlation > 50% (Supplemental Figure 1) (excluding Sample #1, which appears to be an outlier).
Expression of 2113 genes (out of >47,000) was significantly different between the LCH-CD207 cells and the control-CD207 cells using stringent statistical criteria (FDR<0.01). Overall, 747 probes (520 genes) showed increased expression and 2153 probes (1593 genes) showed decreased expression in the LCH-CD207 cells (Supplemental Data 1). Several genes involved in cell cycle regulation, inflammation and Langerhans cell development have been evaluated in LCH lesions and implicated in models of LCH pathogenesis. The most widely accepted current model of LCH can be described as the “Activated-Immature” model in which immature epidermal CD207+ Langerhans cells undergo malignant transformation, or pathologic immune stimulation, and subsequently take on some properties of an activated LC including down-regulation of anchoring molecules (E-cadherin) and increased expression of pro-inflammatory cytokines (TNF-α IL-1β), chemokine receptors (CCR6, CCR7), T cell co-stimulatory molecules (CD40, CD80, CD86), and markers of cell proliferation (Ki67, PCNA) (Reviewed in 8,10,21). Furthermore, several studies have described the LCH lesion microenvironment as a “cytokine storm”, with expression of several pro-inflammatory cytokines (10,21–24).
A comprehensive review of PubMed for genes and proteins associated with LCH pathogenesis was performed, and results were compared to cell-specific gene expression in this study (Figure 2, Supplemental Table IIA). Results from our study supported some previous observations, including significantly decreased expression of E-cadherin and relatively increased expression of MMP9 (25,26 22,27,28). However, the lack of differential gene expression in many molecules previously associated with LCH was surprising. In some cases, such as TNF-α and IL1-β, genes were expressed at very highlevels in both control skin CD207+ cells and LCH lesion CD207+ cells. For others, such as IL-2 and IL-17A, we failed to find significant levels of expression in either control epidermal or LCH lesion CD207+ cells. Additionally, some genes, such as CCL20, were not found to be significant despite high levels of expression in some LCH CD207+ samples due to highly variable expression. Specific results of differential expression of genes reportedly associated with LCH in these experiments are detailed below.
Further review of this dataset identified several genes of interest not previously associated with increased expression in LCH including genes involved in regulation of cell cycle (CDC2A, AF4/FMR2 (member 3), SMYD3, HOXB7), apoptosis (BAX, BCL2L1, CFLAR) signal transduction (DUSP4, JAK3, PRKCA, TLR2, TLR4, SOCS3, JAG2), tumor invasion and metastasis/tissue invasion (CEACAM6, MMP1, TGFβ1), myeloid cell maturation (CD1D, CD13, CD14, CD33, ITGA2B, ITGAX, ITGAM, CD300F) and lymphocyte trafficking (SPP1, VNN1, NRP1, CCR1). A large number of the genes with decreased or absent expression in the LCH-CD207 cells are involved in cell-cell adhesion, including TACSTD1 (29). PERP, an apoptosis effector, is one of the genes with most significantly decreased expression in LCH-CD207 cells (30,31). TIMP2, a metalloproteinase inhibitor, also has reduced expression in LCH-CD207 cells (Figure 2, Supplemental Table IIA).
Analysis of global gene expression in the LCH lesion-CD3 cells also shows a consistent and distinct profile from both the peripheral-LCH-CD3 cells and the tonsil control CD3 cells. Therefore, while lymphocytes from peripheral blood may factor into the variability of the gene expression profiles, these results suggest the CD3+ cells isolated from the LCH lesions primarily represent tumor infiltrating lymphocytes. Differential T cell populations within the LCH lesions may also contribute to the variability between the LCH lesion-CD3 datasets (Correlation >20%) compared to the peripheral LCH-CD3 datasets (Correlation>70%) or the control tonsil CD3 datasets (Correlation>70%) (Supplemental Figure 1).
Using stringent statistical criteria, 162 probe sets (129 genes) showed significant differences in hybridization between cDNA derived from CD3 cells isolated from LCH lesions and peripheral CD3 cells isolated from peripheral blood from the same at the time of biopsy. Overall, 117 probe sets (84 genes) had increased expression and 45 probe sets (45 genes) showed decreased expression in LCH lesion-CD3 cells (Supplemental Data 2). Differentially expressed genes in this group identify several markers of T cell activation including MHC Class II genes, CD58, CD74 and HSP70. Interestingly, FOXP3, a gene specific to regulatory T cells (Tregs), was over-expressed in the LCH lesion-CD3 cells, supporting a similar observation reported by others (32). CTLA4, an activation marker associated with Tregs and inhibition of inflammation, was also overexpressed. LCH lesion infiltrating T cells also significantly overexpressed several genes involved in leukocyte chemotaxis including SPP1, IL-8, CCL3, CCL16, CCR1, CCR5, Vitamin D Receptor and plasminogen activator/urokinase receptor. SPP1 had the highest relative expression in both the Group 1 (LCH-CD207 vs skin-CD207) and Group 2 (LCH lesion-CD3 vs LCH peripheral blood-CD3) datasets. Dual specificity phosphatase 4 (DUSP4), a negative regulator of MAP kinase, was also highly over-expressed by both CD207 cells and CD3 cells from the LCH lesions (33) (Figure 3A, Supplemental Table IIB).
Comparing the 12 LCH lesion-CD3 samples and the 4 control tonsil-CD3 pooled samples, only 81 probe sets (70 genes) were significantly differentially expressed, all of which had decreased relative expression (Supplemental Data 3). Genes associated with an activated TH1 (or TH17) response, including IL17, IL21, and CD40L, were significantly under-expressed in LCH lesion T cells compared to control tonsil T cells (Figure 3B).
We performed real-time PCR (RT-PCR) on independently amplified cDNA samples to validate the microarray data. While there were some differences in amplitude of differences in relative gene expression determined by chip hybridization and RT-PCR in the Group 1 (LCH lesion CD207+ cells vs control CD207+ samples) and the Group 2 (LCH lesion CD3+ cells vs LCH peripheral CD3+ cells) samples, the trends were consistent. Quantitative differences with the two methodologies were likely due to signal compression with the chips at very high and baseline fluctuation of noise at very low levels of expression (Tables IA–IB).
Immunohistochemistry on a sample of LCH-associated proteins also substantiated the gene expression profiling data identified by the array experiments. Immunohistochemistry was performed on four LCH biopsy samples as well as normal skin and tonsil biopsy samples used as controls for antibody staining. The results shown were reproducible and are representative of all LCH biopsy samples tested. Protein products of all of the candidate genes identified by these experiments as overexpressed in LCH lesions (CD207+ or CD3+ cells) were validated by immunohistochemistry (Figure 4).
The concept that LCH arises from epidermal LCs was first proposed by the landmark observation by Nezelof that the histiocytes in LCH lesions contain Birbeck granules (11). Subsequently, cell-surface expression of langerin (CD207), a protein associated with Birbeck granules in antigen processing in LCs, was identified as pathognomonic of both normal LCs and pathological LCs in LCH. (34,35) Therefore models of LCH have developed around the concept of aberrant activation or malignant transformation of the resident epidermal Langerhans cell, as described above in the “Activated Immature Model” (Figure 5) (Reviewed in 8,10,21). However, results from our cell-specific gene expression analysis prompt us to question the origin of the LCH CD207+ cell. First, global expression patterns between control epidermal LCs and LCH CD207+ cells showed low overall correlation (Supplemental Figure 1). Furthermore, expression levels of specific genes central to the Activated Immature Model were not significantly different between the two groups in our experiments.
This study is the first to establish the gene expression profile of CD207+ cells isolated from patient LCH lesions, rather than from in vitro models derived from monocytes or monocyte-derived DCs. LCH-associated genes and proteins were identified through a comprehensive literature review, and relative LCH lesion CD207+ and CD3+ expression in this study are detailed in Figures 2–3 and Supplemental Table II. Differences between gene expression results from this study and some previously published observations may be due to the experimental limitations of immunohistochemistry including heterogeneity of LCH lesions, variable specificity of antibodies, and lack of reliable control tissues. It is also possible that tissue processing altered gene expression relative to cells that are fixed or frozen in tissue blocks.
This study was designed to identify gene expression patterns common to all LCH lesions. Datasets from all LCH CD207+ samples clustered in a group distinct from control epidermal LCs and with no distinction between the profiles from High Risk and Low Risk samples. Additional samples will be required to power a study to determine if there are significant cell-specific differences in gene transcription in biopsies from different patient groups.
In addition to testing previous concepts of LCH and associated markers, the experimental approach of cell-specific gene expression revealed some previously unrecognized patterns of gene expression. LCH CD207+ cells had increased expression of early myeloid markers compared to control epidermal LCs. Epidermal LCs are derived from myeloid precursors that take residence in the skin in embryonic life, but may be replenished by peripheral monocytes in inflamed skin (Reviewed in 36) (Figure 5A). Therefore, it has been difficult to understand how LCs, which are normally restricted to the epidermis, could give rise to such a multi-focal disorder. We recently identified in mice, a population of interstitial CD207+ (langerin+) DCs in most tissues including the lung, liver and lymph nodes. In contrast to LCs that are confined to the epidermis and derive from skin resident hematopoietic precursors, interstitial CD207+ DCs are derived from blood-borne precursor cells and are present in most of the tissues affected by LCH (36,37). Therefore, langerin expression is inducible and not exclusive to epidermal Langerhans cells. These observations, together with the results of the LCH gene expression studies presented here, suggest that LCH may not derive from abnormal LCs, rather from abnormal circulating DC precursors. The observation that plasma M-CSF levels and myeloid dendritic cell precursors (lin-HLA-DR+CD11c+) are over-abundant in patients with active LCH also supports bone-marrow derived precursors rather than epidermal LCs as the source of pathologic CD207+ cells in LCH (38).
We propose a model for LCH pathogenesis in which bone marrow-derived myeloid dendritic cell precursors rather than epidermal LCs migrate to sites of LCH lesions and differentiate into CD207+ cells (Figure 5C). In the “Misguided Myeloid Dendritic Cell Precursor” model, blood-derived myeloid dendritic cells are recruited to sites of disease where they accumulate and recruit activated T cells. Together, these cells elaborate mediators of inflammation and tissue destruction. These experiments do not answer the ongoing question of whether LCH arises from intrinsic defects in a transformed clonal population of precursor cells, or whether it is caused by a functional response to a pathologic stimulus. Lack of differential expression of markers of proliferation supports a model in which CD207+ LCH cells arise from accumulation of pathologic cells, which is also supported by other recent studies (32).
The role of T cells in LCH is not well-understood. There is no evidence of a functional interaction between T cells and CD207+ cells in vivo in LCH (39). Results from this study show highly variable ratios of CD207:CD3 cells in LCH lesions (Supplemental Table IB). The CD3 cells are unlikely to arise primarily from contamination of lesions at the time of biopsy since they have unique gene expression profiles when compared to peripheral blood (Supplemental Figure 1). Several genes with significant gene expression in LCH CD207+ play roles in lymphocyte activation, recruitment, and dendritic cell interaction including osteopontin, neuropilin-1 and vanin-1. Osteopontin, expressed by SPP1, also known as early T lymphocyte activation 1 (ETA-1), had the highest differential expression in both the LCH CD207+ vs control LC and infiltrating LCH CD3+ vs peripheral LCH CD3+ datasets. It is a secreted phospoprotein with putative functions including T cell activation, Treg differentiation, homing of macrophages and T cells, cell survival and tumor progression (Reviewed in 40). Osteopontin-deficient mice have deficient type 1 immunity to bacterial and viral infections and are unable to form granulomas. (41) Osteopontin overexpression has been found in a variety of cancers, including breast cancer, lung cancer, colorectal cancer, stomach cancer, ovarian cancer, and melanoma (42). One of the major osteopontin ligands, CD44, is also significantly over-expressed in LCH CD207 cells compared to control LCs. Neuropilin-1, expressed by NRP1, is a membrane-bound co-receptor to a tyrosine kinase receptor for both vascular endothelial growth factor and semaphorin molecules. NRP1 is involved in axonal guidance of nerves. In the immune system, it is important in establishing the T cell-dendritic cell synapse as well as migration and adhesion of thymocytes (43,44). Vanin-1, expressed by VNN1, is a glycosylphosphatidylinositol (GPI)-anchored molecule expressed by perivascular thymic stromal cells that is essential in migration of hematopoietic progenitor cells to the thymus (45). Mice deficient in vanin-1 are unable to form granulomas in response to Coxiella burnetti infection (46). Increased expression of SPP1, NRP1, and VNN1 by LCH CD207+ cells suggest a potentially interesting role for these genes in CD207+ cell- mediated recruitment of T cells to LCH lesions.
Multiple searches for a common antigen in LCH failed to identify an infectious etiology (Reviewed in 4). A recent study found that Tregs were enriched in LCH lesions as well as in peripheral blood of patients with active LCH. The same study found polyclonal T cell receptors rearrangements in DNA isolated from frozen biopsy samples (32). Regulatory T cells are thought to develop as an antigen-specific response. Similarly, activation of T cells typically arises through antigen-dependent mechanisms in vivo. The expression profile of infiltrating CD3 cells compared to circulating CD3 cells in patients with LCH is consistent with an activated phenotype, and also supports enrichment of Tregs. Therefore, the mechanisms by which T cells are activated remain to be determined. Another recent study suggested LCH arises through expression of IL-17A by pathologic Langerhans cells. In our study, IL-17A expression was not detectable in CD207 cells, as described previously (47). Furthermore, relative expression of IL-17A and IL-21 was significantly decreased in LCH lesion T cells compared to control tonsil T cells, suggesting LCH is not characterized by a TH1/TH17 response.
Improved understanding of the biology is essential to improve treatment of individuals with LCH. However, research in LCH is challenged by rare tissue samples and lack of reliable in vitro or animal models. We are now able to generate cDNA probe for expression studies from as few as 1500 cells The results show surprisingly distinct gene expression profiles between CD207+ cells from LCH lesions and control epidermal LCs, which suggests the LCH CD207+ cells are either highly differentiated, or arise from a distinct cell population. The relative expression of immature myeloid markers from the LCH CD207+ cells further supports a model in which LCH arises from a bone marrow-derived dendritic cell precursor rather than from a transformed resident epithelial Langerhans cell. The results from this study further identify several new genes that have not previously been associated with LCH that may be important diagnostic markers or therapeutic targets. Genes that encode osteopontin-1, neuropilin-1, and vanin-1, all of which are involved in lymphocyte trafficking and lymphocyte-dendritic cell communication, are highly expressed in LCH CD207+ cells. In summary, we propose a new model in which LCH does not arise from epidermal Langerhans cells, but from misguided myeloid dendritic cell precursors. Future studies will determine the definitive reclassification of “Histiocytosis X”.
We would like to thank Dr. Cliona Rooney for her helpful suggestions throughout this project. We would like to acknowledge Christopher Threeton, Michael Cubbage and Tatiana Goltsova of the TXCCC Flow Facility for their excellent technical support. Finally, we would also like to thank our colleagues who helped with collection of the LCH biopsy samples, as well as patients and families for their generous participation in this study.
1This study was supported in part by research funding from the Histiocytosis Association of America (CEA and KLM), the Histiocytosis Association of Canada (CEA and KLM), the American Society of Clinical Oncology (CEA), the Thrasher Research Fund (CEA), the Larry and Helen Hoag Foundation Clinical Translational Research Award (CEA), the US National Institutes of Health Pediatric Oncology Research Training Grant (T32 CA115303-03) (CEA) and (R21 CA 114981-01A2) (KLM). MTP received support from the American Cancer Society Institutional Research Grant (2007-08) #93-034-23.