Polyclonal T-Cells Express CD1a in Langerhans Cell Histiocytosis (LCH) Lesions

Langerhans cell histiocytosis (LCH) is a complex and poorly understood disorder that has characteristics of both inflammatory and neoplastic disease. By using eight-colour flow cytometry, we have identified a previously unreported population of CD1a+/CD3+ T-cells in LCH lesions. The expression of CD1a is regarded as a hallmark of this disease; however, it has always been presumed that it was only expressed by pathogenic Langerhans cells (LCs). We have now detected CD1a expression by a range of T-cell subsets within all of the LCH lesions that were examined, establishing that CD1a expression in these lesions is no longer restricted to pathogenic LCs. The presence of CD1a+ T-cells in all of the LCH lesions that we have studied to date warrants further investigation into their biological function to determine whether these cells are important in the pathogenesis of LCH.


Introduction
Langerhans cell histiocytosis (LCH) is a complex disease with unpredictable progression and no known cause [1,2]. LCH occurs predominantly in children but also occurs in adults. Lesions are most common in bone (eosinophilic granuloma) and skin, but may occur in other organs. LCH may be confined to single sites, have multifocal involvement or become disseminated. The clinical course varies from lesions that spontaneously resolve, to a chronic disease, or can be disseminated and life-threatening [1]. The severity and prognosis are dependent on the type and extent of organ involvement, with children under two most at risk of lifethreatening complications.
LCH was originally defined by the Writing Group of the Histiocyte Society in 1987 [3], and was more recently revised [4]. LCH lesions are diagnosed by an accumulation of cells, long presumed to be pathogenic Langerhans cells (LCs) [5,6]. The standard identification of LCs is by their morphology (including Birbeck granules) or with the immunological marker CD1a [7,8]. Langerin (CD207) has also been used as a marker of LCs [9]. Other inflammatory cells typically found in LCH lesions include T-cells, eosinophils, plasma cells, neutrophils, basophils, macrophages and giant cells [10].
The recognition of close similarities between normal epidermal LCs and pathogenic LCs has led to the concept that epidermal LCs are the precursor cells in LCH [5]. LCH combines in one nosological category, a group of disorders that have differing clinical manifestations, but all are characterized by an accumulation of cells with features of cutaneous LCs and inflammatory cells. The presence of inflammatory cells in all LCH lesions implies that a better understanding of these cells may lead to improvements in the management of this group of disorders.
Most LCH research has focused on pathogenic LCs. While the accumulation of pathogenic LCs within LCH lesions is considered to be a defining characteristic of LCH [3], their role in the causation or impact of the disease remains unclear. LCH has pathological features of both cancer and chronic inflammation, and whether it is a true malignancy remains contentious [11][12][13]. The CD1a-expressing cells have been reported to be clonal and pathogenic [14,15], and this is supported by the detection of BRAF mutations in LCH lesions [16,17].
The unresolved role of T-cells in LCH is indicated by the number of conflicting reports relating to the types of T-cells within lesions [6,[18][19][20][21][22]. Debate also surrounds the detection of high serum levels of IL-17A during active LCH and IL-17A synthesis by dendritic cells (DCs) in LCH lesions [20]. Subsequent studies did not support these findings [21,22]. Expansion of regulatory Tcells has been associated with the accumulation of LCs in LCH lesions [6,18], although more research is warranted to elucidate the role of T-cells in these lesions.
CD1a expression on pathogenic Langerhans cells (LCs) in LCH is regarded as a hallmark of this disease, but the role of this highly restricted molecule in LCH has remained uncertain. Normally, CD1a molecules on the surface of LCs present glycolipid antigens to specialized T-cells as part of their role in immunosurveillance [23,24]. Stimuli from pathogens, tumors or host immune responses that are capable of modifying CD1a expression have demonstrated similar results in vitro [23]. Expression of CD1a is thought to be partly controlled by cytokines, as demonstrated by the induction of CD1a in cultured myeloid progenitor cells [25,26].
While CD1a is expressed predominantly on LCs, it can be expressed by T-cells under certain conditions. Immature thymocytes express CD1a [27], as do some T-cells involved in neoplastic conditions such as pre-lymphoblastic lymphomas/leukemias [28][29][30][31] and follicular dendritic cell sarcoma [32]. T-cells that express CD1a have recently been identified within the tonsil suggesting extrathymic T-cell development [33]. To date, there are no reports of CD1a expression on mature T-cells.
Historically, the cellular composition of LCH was determined using archival tissue sections. This approach formed the foundation of what is currently known about this disease, however, these earlier studies were greatly restricted by the number of CD markers that could be detected simultaneously. LCH tissues have been examined by flow cytometry in a limited number of studies [34,35] and in some cases, specific cell populations have been sorted for further analysis [6,36]. These studies confirmed the presence of pathogenic LCs using CD1a and/or CD207 [34]. Sorted CD1a + cells from LCH lesions expressed CD207 in more than 75% of the cells, and these were deficient in stimulating allogeneic T-cell proliferation in vitro [34].
We set out to better define the role of CD1a and T-cells in LCH lesions, using multi-parameter flow cytometry with up to eight cell markers simultaneously. Here we report for the first time, to our knowledge, the expression of CD1a on polyclonal T-cells in LCH lesions. The results provide new insights into the composition of LCH lesions and we anticipate that future studies into the biological function of these cells may lead to an improved understanding of LCH pathogenesis.

Identification of CD1a + /CD3 + cells by flow cytometry
Using flow cytometry we characterized the cell marker profiles of LCH lesions from six patients ( Table 1, 2). Within the LCH lesion from patient #1, we identified an unexpected population of CD1a + /CD3 + cells that constituted 23.8% of the live-gated cells (Table 3). A representative plot of these CD1a + /CD3 + cells from a single experiment is shown in Figure 1. Although the CD1a + population represented 44.7% of the cells in this LCH sample, over half of the CD1a + cells (53.2%) also co-expressed CD3 (Table 3). This represented 56.7% of the total CD3 + cells. These CD1a + /CD3 + cells were absent in the peripheral blood of this patient and a tonsil control tissue ( Figure 1). CD1a + /CD3 + cells were subsequently identified in a further five LCH patients (Table 3).
CD1a + /CD3 + cells were not detected in normal peripheral blood from six volunteers, in peripheral blood from nine LCH patients or in single cell suspensions (SCS) prepared from the epithelial layer of five tonsils (Table 2, 4). Two acute myeloid leukemia (AML) samples, one chronic lymphoid leukemia and one normal lymph node were profiled and no CD1a + /CD3 + cells were identified. A Kruskal-Wallis test was conducted to compare the mean ranks of the percentage of CD1a + /CD3 + cells for LCH lesions (n = 6, 8.368.8), LCH peripheral blood tissues (n = 9, 0.060.0), and control tissues (n = 15, 0.060.0). There was a significant difference across the three groups (H(2) = 28.54, p, 0.0001) with a mean rank of 27.5 for LCH lesions and 12.5 for the other groups. Dunn's multiple comparisons test revealed significant differences between LCH lesions and LCH peripheral blood tissues (p,0.0001) and LCH lesions and control tissues (p, 0.0001). One T-cell lymphoma contained 0.4% CD1a + /CD3 + cells. This was not unexpected, as immature T-cells have been reported to express both CD1a and CD3 [32].
CD1a + /CD3 + cells are morphologically and phenotypically polyclonal T-cells Tight doublet gates were applied to ensure that CD1a + /CD3 + cells were not two adjoined cells (Figure 2). No doublets were observed when sorted cells were examined under the microscope. In flow cytometry plots, the relative size and mean forward scatter intensity of the CD1a + /CD3 + cells was more typical of T-cells than LCs ( Figure 3A). Similarly, the morphology of the sorted CD1a + /CD3 + cells, as seen microscopically, was more typical of lymphocytes than LCs ( Figure 3B). When fluorescent immunocytochemistry was performed on lesional cells, some individual cells clearly expressed both CD1a and CD3 ( Figure 3C).

Further phenotyping of CD1a + T-cells
Since the morphology of CD1a + /CD3 + cells was more typical of lymphocytes than LCs, further experiments were performed to determine their phenotype. The T-cell receptor (TCR) Vb repertoires of the T-cells within lesions from three LCH patients were examined. Most of the 25 TCR Vb subsets were detected in lesional CD1a + T-cells without any obvious bias (data not shown).
There was insufficient sample to analyze the remaining three LCH lesions. Because there were multiple TCR subtypes, CD1a + T-cells could not have arisen from a single parent cell, and were therefore not clonal in origin.
The T-cell subsets within five LCH lesional samples were examined to differentiate between CD4 + helper (T H ) T-cells and CD8 + cytotoxic (T C ) T-cells. One sample was excluded due to insufficient data. CD1a expression on CD1a + /CD3 + cells was not restricted to T H or T C subsets ( Figure 4, Table 5). For LCH lesions, Mann-Whitney tests were conducted to compare the median percentage of CD1a + T H cells (n = 5, 60.1626.6) with CD1a 2 T H cells (n = 5, 50.2617.9), and the percentage of CD1a + T C cells (n = 5, 12.9611.4) with CD1a 2 T C cells (n = 5, 19.5610.6). No statistically significant differences were detected amongst the CD1a + and the CD1a 2 T H groups (U = 9, p = 0.53) or the CD1a + and the CD1a 2 T C groups (U = 7, p = 0.31). These results indicate that the proportions of T H and T C cells are not unusual in the CD1a + T-cells when compared to other (CD1a 2 ) T-cells.
Additional CD markers were used where possible, to further phenotype the CD1a + T-cells (Table 2, 6). As expected, these cells co-expressed the common leukocyte marker CD45. There was also variable, but low expression of T-cell markers including CD25 (expressed on T-cells during activation), and CD16 and CD56 (found on NKT-cells). Whilst the CD1a + T-cells displayed some typical T-cell markers, most of the non T-cell markers examined were absent, or, present at negligible levels. These included CD19 (a B-cell marker), CD138 (a plasma cell marker), CD14 (a monocyte/macrophage marker) and CD34 (found on endothelial cells, DCs and progenitor cells). CD123 is not expressed on T-cells and was absent on the CD1a + T-cells. Low levels of CD207 expression were detected in the two LCH samples examined ( Table 6).
It is known that upon recognition of specific antigen, naïve Tcells (CD45RA + /CD45RO 2 ) reduce their expression of CD45RA and increase their expression of CD45RO, producing memory Tcells (CD45RA 2/ CD45RO + ) [37]. We looked at CD45RA/ CD45RO co-expression on CD1a + T-cells and our results show that CD1a expression is not restricted to either naïve or memory T-cells ( Figure 5).

Sorted CD1a + T-cells express CD1a mRNA
PCR amplification of CD1a and CD3 confirmed our hypothesis that the CD1a + /CD3 + sorted cells from LCH lesions expressed both CD1a and CD3 mRNA in all four LCH lesions examined ( Figure 6). b-actin could not be successfully amplified in the remaining two lesions, hence they were excluded. Amplification of CD1a from patient #3 was successful but faint. Using cDNA from CD1a + /CD3 + sorted cells; CD1a was successfully amplified from all four LCH lesions with a second pair of CD1a primers directed towards a different region of the CD1a sequence (data not shown).

Discussion
Here we demonstrate for the first time, to our knowledge, the presence of CD1a + T-cells in all LCH lesions we have studied to date. These cells were not detected in the peripheral blood from LCH patients. Cellular morphology was typical of lymphocytes, and these cells expressed many of the typical T-cell markers. We have shown that sorted CD1a + /CD3 + cells from LCH lesions also expressed both CD1a and CD3 mRNA, indicating that the CD1a was generated within these cells and not transferred as protein by T-cell interaction with LCs.
Previous studies have confirmed that T-cells within LCH lesions are polyclonal [15,18]. Our data also demonstrates that CD1a + Tcells are polyclonal, and are comprised of a wide range of T-cell subtypes. Furthermore, a number of studies have used flow cytometry to identify cells in LCH lesions; these investigations were limited to two or three parameters and predominantly focused on pathogenic LCs [34][35][36].
Multi-parameter flow cytometry and cell sorting have aided our capacity to more precisely determine cell phenotypes within LCH lesions. The heterogeneous populations of cells identified in this study clearly show that the cellular composition in LCH lesions is more complex in regards to CD1a expression than was previously recognized [14,34].
More recent work using cell-specific gene expression profiling compared T-cells from LCH lesions to those of peripheral blood of LCH patients [6]. Although not commented on by the authors, examination of the supplemental data revealed a 48.95-fold increase in CD1a expression in lesional T-cells. These data appear to support our observation of a distinct population of CD1a + Tcells within LCH lesions.
It is evident from our data that CD1a expression has been induced across a range of T-cell subsets in LCH lesions. The coexpression of T-cell markers and the lack of DC marker expression suggests it is unlikely that CD1a + T-cells originate from pathogenic LCs. In light of these data, the question must be posed as to how and why this might occur. Peripheral T-cells do not normally express CD1a; however CD1a expression on cortical thymocytes is well documented [27]. During thymocyte maturation, CD1a is down-regulated so it is not normally expressed by T-cells in the peripheral circulation [27,38]. Since our studies show that CD1a + T-cells were not detected in the peripheral blood from LCH patients, it is unlikely that these cells represent circulating immature cortical thymocytes.
T-cells expressing CD1a were recently identified in very low numbers in the lymphoid tissue of tonsils [33]. Although we  in vitro could express CD1a as well as CD207 and E-cadherin [25]. Studies have demonstrated that a subset of CD34 + progenitor cells proliferated in vitro in the presence of GM-CSF and TNFa, to produce CD1a + cells [39], and that the lipid microenvironment can modulate CD1a expression and differentiation of monocyte-derived DCs [26]. Little is known, however, regarding the modulation of CD1a expression on lymphoid cells. Although PHA-stimulated normal T-cells have demonstrated intra-cellular expression of CD1a [40], there is no expression of CD1a molecules on the cell surface. Cultures of both AML and acute lymphoblastic leukemia blasts in vitro with various cytokine preparations have led to the expression of CD1a on the blast cells [41]. It is known that the LCH micro-environment contains numerous cytokines including GM-CSF and TNFa [19,42,43], and taken into conjunction with the above studies it would be tempting to postulate that the cytokine storm within LCH lesions may be responsible for inducing CD1a expression in T-cells. Against this hypothesis is the observation, to the best of our knowledge, that similar inflammatory conditions such as familial hemophagocytic lymphohistiocytosis, rheumatoid arthritis and inflammatory bowel disease are not characterized by a predominance of CD1a + cells. CD1a expression on cells that would not normally express CD1a appears to be unique to LCH and not evident in any other inflammatory conditions associated with cytokine storms.

Our observation of CD1a + T-cells in LCH lesions may open up
further opportunities for functional studies. Experiments to determine function were beyond the scope of this study. Due to the limited availability of live cells for analysis, any potential mechanisms must remain theoretical. It has been reported that pathogenic LCs in LCH poorly present alloantigens [24,34] and therefore have a defect in antigen presentation. This was corrected by the addition of CD40L in vitro [34] indicating a possible in vivo defect in this area that could be examined further in future studies.
Debate around the classification of LCH as a neoplastic disease or immune dysfunction has been ongoing [6,[13][14][15][16]18,[44][45][46][47]. Numerous investigators have suggested an immunological dysfunction due to the altered expression of cytokines [42,43,48,49]. Alternatively, support for LCH as a neoplasm is based on publications that have shown clonality of the CD1a + cells in LCH using the human androgen receptor assay and BRAF mutation studies [14][15][16][17]45,50]. Our studies indicate that the CD1a + cells in LCH are comprised of pathogenic LCs and polyclonal T-cells and as such, cast doubt on the concept that all CD1a + cells in LCH are clonal [14,15]. Our results do not discount that pathogenic LCs are clonal. If anything should be taken from this long standing debate, it is that LCH is an unusual disease where perhaps immune dysfunction and neoplasia are interconnected.
We demonstrate that CD1a is not a unique marker for pathogenic LCs in LCH and a new definition to include CD3 2 and CD1a + and/or CD207 + cells or the presence of Birbeck granules is required for the identification of pathogenic LCs in LCH. We hypothesize, that defective cycling of CD1a molecules to the surface of pathogenic LCs may lead to redundancy within the immune system, whereby T-cells can be induced to express CD1a and present antigen within LCH lesions. In summary, our studies have identified for the first time, to our knowledge, the presence of polyclonal CD1a + T-cells in LCH lesions and the presence of these cells is specific to LCH lesions.

Patient samples
We obtained peripheral blood from nine LCH patients and lesional tissue samples from six LCH patients. Four LCH patients provided both peripheral blood and tissue (Table 1). Non-LCH control tissues were collected including five tonsils, one normal lymph node (non-involved), peripheral blood from six healthy    volunteers, two AML samples, one chronic lymphoid leukemia and one T-cell lymphoma sample (Table 4).

Sample preparation
Samples were prepared from fresh tissue then stored in liquid nitrogen in 10% DMSO (v/v) or used fresh. LCH lesions, except the dermal sample from patient #3, were prepared as SCS in media (IMDM (Gibco) supplemented with 2 mM L-glutamine, 50 mg/ml kanamycin (Invitrogen) and 10% (v/v) heat-inactivated fetal calf serum (Gibco)). The dermal sample was prepared from a skin biopsy of an LCH lesion as previously described [51] except cells were resuspended in media. Tonsil epithelium was prepared by removing thin strips from the external surface. The central lymphoid tissue was discarded. The epithelial strips were incubated for 45 min at 37uC in 5 ml of 10 mg/ml dispase II (Roche) prior to the removal of any remaining lymphoid tissue. The strips were further incubated in 4 ml of 0.3% trypsin for 10 min at 37uC then added to 4 ml of media. The suspension was filtered then incubated for 10 min at 37uC with 1 mg/ml DNAse-1. Following centrifugation, pelleted cells were resuspended in media. White blood cells were isolated from peripheral blood from both normal and diseased samples. Red blood cell lysis buffer (155 mM NH 4 Cl; 10 mM KHCO 3 and 0.1 mM ethylenediaminetetraacetic acid) was added to peripheral blood, incubated at room temperature for 10 min, re-centrifuged then washed twice by resuspending the pellet in phosphate buffered saline pH 7.4.

Immunocytochemistry
Cytospins of SCS were fixed with methanol. The DakoCytomation EnVision Doublestain system was used for chromogenic immunocytochemistry. Flow cytometry analysis and cell sorting LCH biopsy specimens and controls were examined ( Table 2, 4). The same cytometer settings were applied for all experimental samples to allow for direct comparison. Compensation controls, including single antibody stains and fluorescence minus one controls, were used to determine background and eliminate positive outliers. Prior to sample analysis, we compared antibody binding specificities to validate the use of more than one antibodyfluorophore combination for both anti-CD1a and anti-CD3 antibodies (Figure 7). Analysis of samples included dead cell gating by propidium iodide and the exclusion of doublets by forward/side scatter gating ( Figure 2). Cells were analyzed with two scatter and eight fluorescence parameters and sorted into populations using a BD FACSAria II (BD Biosciences) and BD FACSDiva V6.1 software (BD Biosciences).
SCS were washed and resuspended in media. The cells were incubated with antibodies in the dark for 30 min at 4uC then washed twice. For phenotypic characterization, the following antibodies were used: anti-CD1a-FITC [NA 1/34], anti-CD1a-

RNA isolation and reverse transcription-PCR
Total RNA was isolated using a miRNeasy Mini kit (QIAGEN) according to manufacturer's instructions including the optional on-column DNase digestion. Random hexamer primers were used to synthesize cDNA using the Transcriptor First Strand cDNA synthesis Kit (Roche). PCRs were performed using 0.5-5 ml cDNA with gene-specific primers (Table 7) from PrimerBank (final concentration 0.4 mM) [52][53][54]. Final concentration of b-actin primers [55] was 0.125 mM. All primer pairs were intron spanning. Reactions were performed using 1 unit of AmpliTaq Gold DNA polymerase with GeneAmp PCR Gold Buffer, 1 mM MgCl 2 (Applied Biosystems) and a final concentration of 200 mM for each dNTP (Roche). Thermal cycling was carried out on either an Applied Biosytems GeneAmp PCR system 2700 or a Perkin Elmer GeneAmp PCR system 2400 machine. Reactions were preheated at 95uC for 7 min followed by 40 cycles (95uC for 30 sec, 55uC for 30 sec, 72uC for 30 sec), with a final extension at 72uC for 10 min. As previously published, the annealing temperature used for b-actin reactions was 68uC [55]. Reactions were performed under hot start conditions and a positive control (sample previously shown to contain the amplicon) and negative control (no template) were included with every set of reactions. PCR products were separated on an agarose gel, stained with ethidium bromide, illuminated on an ultraviolet light box and photographed using a digital camera.

Sequencing CD1a
CD1a was sequenced from 39-RACE reactions using RNA from CD1a + /CD3 + sorted cells from LCH patient #1 and the CD1a pair 1 forward primer (Table 7). SMART RACE cDNA Amplification Kit (Clontech) was used (following manufacturers' instructions) with an annealing temperature of 60uC and 35 amplification cycles. A semi-nested reaction was performed using the CD1a pair 2 forward primer (Table 7) with an annealing temperature of 62uC and a further 30 amplification cycles. The 39-RACE product was gel purified (QIAquick gel extraction kit, QIAGEN) and the PCR product (80 ng) directly sequenced using the standard cycle sequence protocol with the CD1a pair 2 forward primer (Micromon Services, Monash University).

Statistical analysis
Data was analyzed in the GraphPad Prism statistical program (GraphPad Software, San Diego, CA). Non-parametric tests were used due to non-normality of small sample sizes. A Kruskal-Wallis one-way analysis of variance was used to detect differences between the percentage of CD1a + /CD3 + cells in LCH lesions, LCH peripheral blood and control tissues. While two LCH patients provided lesional tissue only, and five provided peripheral blood tissue only, four LCH patients provided both lesional and peripheral blood tissues. Being less powerful than a paired test to detect differences, an independent test was the most suitable option despite some pairing. Two-tailed Mann-Whitney U tests were conducted to detect differences between the percentage of T H cells in the CD1a + /CD3 + and the CD1a 2/ CD3 + populations, and between the percentage of T C cells in the CD1a + /CD3 + and the CD1a 2/ CD3 + populations. Data are quoted as means 6 standard deviation. Differences were considered statistically significant at an alpha level of 0.05.