BRAFV600E mutant protein is expressed in cells of variable maturation in Langerhans cell histiocytosis

Langerhans cell histiocytosis (LCH) is a heterogeneous disease affecting patients of different ages, manifesting at different sites and presenting as focal or systemic disease. LCH takes an unpredictable clinical course ranging from spontaneous remission to fatal progression. The lesions in LCH present microscopically as heterogeneous mixtures of cells containing various proportions of lymphocytes, macrophages, giant cells, a prominent population of eosinophilic granulocytes, and histiocytes. Lichtenstein was the first to identify those histiocytes as an abundant and thus defining fraction and chose the term “Histiocytosis X” to point toward the uncertainty about their exact nature.¹ 

The term “Langerhans cell histiocytosis” was coined on recognition of common morphologic and immunohistochemical features with skin Langerhans cells.^2-4  LCH thus was separated from non-Langerhans cell histiocytosis, such as Erdheim-Chester or Rosai-Dorfman disease.

Currently, the diagnostic gold standard for LCH is detection of both CD1a and CD207 in Langerhans cells from LCH lesions (herein referred to as “pathologic LCs”). Expression of S-100 is less specific.^2,5-8  The origin of pathologic LCs has not been determined. LCH has been discussed to be either an inflammatory reactive or a neoplastic disease. The mixed composition of the lesions with a prominent lymphocytic and granulocytic component was compatible with inflammation and reactive change. However, 2 independent studies recently detected frequent BRAFV600E mutations by DNA sequencing in LCH, strongly arguing for neoplastic disease.^9-12  Because of the heterogeneous mixture of cells in LCH lesions, the exact cell type harboring the BRAF mutation so far has not been established. Mutant BRAF is a constitutive activator of the AKT/ERK kinases pathway driving proliferation and is observed in many types of cancer.¹³ 

DNA sequencing of LCH is difficult because of the mixed composition of the lesions, including neoplastic and reactive cells. We have recently developed a mouse monoclonal antibody (clone VE1) with high specificity for mutant BRAFV600E protein. With this study, we demonstrate the feasibility of VE1 to detect BRAFV600E mutation in LCH. Further, we intended to characterize the neoplastic cells in LCH.

Formalin-fixed, paraffin embedded tissue from 89 cases of LCH were obtained from the Institute of Pathology, Heidelberg, the Institute of Pathology at the Medical University of Vienna, the Clinical Institute of Neurology (Neuropathology), Medical University of Vienna, and the Institute of Pathology, Charité-University Medicine, Berlin. Normal skin and tonsil tissues served as controls. Mutant BRAFV600E protein was detected by specific mouse monoclonal antibody VE1.^14,15  Immunohistochemistry included cell conditioning with Ventana (Ventana Medical Systems) cell conditioner 1 for 64 minutes, preprimary peroxidase inhibition, incubation with 1:5 diluted VE1 hybridoma supernatant at 37°C for 32 minutes, incubation with OptiView HQ Universal Linker for 12 minutes, incubation with OptiView HRP Multimer for 12 minutes, OptiView Amplification (setting of OptiView Amplifier and OptiView Amplifier Multimer both for 12 minutes), and incubation with hematoxylin and Bluing reagent for 4 minutes each. Although we used hybridoma supernatant, purified VE1 is now commercially available (Spring Bioscience). Meanwhile, studies applying different detection systems have also been published,^14-16  one discussing advantages of both methods under different conditions in detail.¹⁵ 

Immunohistochemistry for CD207 (1:25; Atlas Antibodies), CD1a (1:25; Thermo Scientific), Ki-67 (1:100; Medac), p53 (1:50; Novocastra, distributed by Leica), S-100 (1:100; Dako Denmark), and pERK (1:100; Cell Signaling), CD34 (1:2; Thermo Scientific) included pretreatment with either Ventana cell conditioner 1 (CD207, CD1a, Ki-67, p53, CD34) or protease for 4 minutes (pERK) and antibody incubation for 32 minutes at 37°C. Incubation was followed by Ventana standard signal amplification, UltraWash, counterstaining with 1 drop of hematoxylin for 4 minutes, and 1 drop of bluing reagent for 4 minutes. For visualization, ultraViewUniversal DAB or FastRed (S-100) Detection Kit (Ventana) was used. Images were acquired on an Axioplan2 microscope (Carl Zeiss) equipped with an AxioCamHR camera and AxioVision Version 4.8 (both Carl Zeiss) software. Magnifications are given in figure legends.

For immunofluorescence, slides were deparaffinized according to standard protocols and underwent antigen retrieval for 30 minutes in cell conditioner 1 (Ventana) in a steam cooker. Primary antibodies VE1, CD33 (1:50; Santa Cruz Biotechnology), CD36 (1:50), CD80 (1:50), CD86 (1:50), CD14 (1:100, all by Abcam), CD3 (1:50; Dako), and CD207 (1:50) were incubated overnight at 4°C. VE1 was detected applying the Tyramide 488 Alexa TSA kit (Invitrogen) according to the manufacturer's instructions. Other antibodies were detected by incubation with AlexaFluor-568 nm (1:500; Invitrogen) for 30 minutes at room temperature. Images were obtained using an Olympus BX50 fluorescence microscope with ×100 objective lens (Olympus), equipped with an AxioCamMR3 camera and AxioVision Version 4.8 software (both Carl Zeiss).

DNA was extracted from 46 samples applying the Invisorb DNA extraction kit (Stratek) according to the manufacturer's instructions. Then, a fragment spanning the catalytic domain of BRAF including codon 600 was amplified and sequenced as previously described.¹⁷  In brief, PCR was performed in a total volume of 15 μL, containing 20 ng of DNA and 2 times PCR Master Mix (Promega) with initial denaturation at 95°C for 60 seconds, followed by 35 cycles with denaturation at 95°C for 30 seconds, annealing at 56°C for 40 seconds, and extension at 72°C for 50 seconds. A total of 2 μL of the amplification product was submitted to sequencing using the BigDye Terminator Version 3.1 Sequencing Kit (Applied Biosystems). Twenty-five cycles were performed using 12 ng of the sense primer, with denaturation at 95°C for 30 seconds, annealing at 56°C for 15 seconds, and extension at 60°C for 240 seconds. Sequences were determined performing bidirectional direct sequencing on a semiautomated sequencer (ABI 3100 Genetic Analyzer; Applied Biosystems) and the Sequence Pilot Version 3.1 (JSI-Medisys) software.

The presence or absence of the BRAFV600E mutation was determined in all 89 lesions by VE1 immunohistochemistry. Mutant protein was detected in 34 of 89 (38%) cases. DNA for direct sequencing of the BRAFV600 region was available from 46 patients. A BRAFV600E mutation was detected in 18 of 46 (39%) lesions, a V600K mutation in 1 of 46 lesions (2%), and a wild-type status in 27 of 46 (59%) lesions. In 1 lesion with DNA available, the BRAFV600E alteration was detected by immunohistochemistry only and not by direct sequencing. The single lesion with the BRAFV600K mutation was not detected by antibody VE1. Thus, sensitivity for detection of the V600 mutations was 95% for immunohistochemistry and 95% for direct sequencing. Concordant results of immunohistochemistry and direct sequencing were seen in 96% of the cases. The tissue set assessed by immunohistochemistry only yielded comparable results, with 15 of 43 (35%) mutant cases (Table 1). Because previous reports observed a germline mutation in the BRAF gene,¹²  we were alert to this possibility. However, no case was observed where VE1 positivity was found in adjacent tissue. Age distribution of patients with mutant versus wild-type LCH lesions revealed no significant difference (P > .65, Student t test). Patient characteristics and BRAF V600 status are compiled in Table 1. As far as retrievable from medical records, 4 patients (cases 3, 16, 17, and 73) had multifocal lesions. This did not allow for statistical analysis of multifocal versus focal disease.

We next assessed the phenotype of the BRAFV600E mutant cells in 3 different LCH lesions (cases 9, 19, and 47). Immunohistochemistry of consecutive sections demonstrated a wide overlap of BRAFV600E mutant cells (Figure 1A-B) with classic antigens expressed by Langerhans cells, such as S-100 (Figure 1C-D), CD207 (Figure 1E-F), and CD1a (Figure 1G-H). For confirmation, CD207-expressing cells from a representative lesion (case 9) with VE1 positivity were sampled by laser capture microdissection. Consecutive sequencing revealed a BRAFV600E mutation in this fraction (data not shown). Of note, in some cases, the number of CD207-expressing cells (Figure 1H) was lower than the number of VE1-positive cells (Figure 1B). This heterogeneity could also be verified by immunofluorescence (Figure 2) performed on 3 independent LCH lesions (cases 1, 9, and 19). Although the majority of cells expressed both mutant BRAFV600E protein and CD207 (Figure 2A-I), single cells only yielded a signal for BRAFV600E (Figure 2D-E). All BRAFV600E-positive cells expressed the antigen CD36, which is frequently found in myeloid dendritic cells^18-21  (Figure 3A-C) and the monocyte marker CD14 (Figure 3D-F) but lacked expression of the T-cell lineage marker CD3 (Figure 3G-I). CD33 typically strongly expressed in mature Langerhans cells within the skin (Figure 3L)^21,22  was rarely found in LCH lesions (Figure 3J-K). Moreover, BRAFV600E-positive cells also expressed CD80 (Figure 3L-O) and some also CD86 (supplemental Figure 1A-C, see the Supplemental Materials link at the top of the article), both up-regulated during LC maturation,^6,22-24  but rarely the hematopoietic progenitor antigen CD34,²²  which was mainly limited to vessels (supplemental Figure 1D-E).

In the majority of the 34 cases with BRAFV600E, mutant protein was primarily detected in granuloma-like structures; however, single dispersed cells were also observed (Figure 4A). The ratio of BRAFV600E-positive cells to cells not harboring mutant protein varied from 3% to 90%. Accumulation of BRAFV600E-positive cells was observed in the regions with highest Ki-67 index and mitotic figures (Figure 4A). As determined on consecutive sections stained for S-100, CD1a, and Ki-67 (illustrated in supplemental Figure 1F-H), there was, however, no difference in proliferation rates of the respective cells in lesions with and without BRAFV600E mutations (Figure 4B, P > .61, Student t test). Similarly, the activation of the RAS/RAF/MEK/ERK signal cascade, also determined with regard to areas of CD1a⁺ and S-100⁺ cells, did not differ between lesions with and without BRAFV600E mutations (Figure 4E; P > .24, Student t test). This was assessed in at least 8 LCH lesions with and in 8 LCH lesions without BRAFV600E mutation. Because TP53 mutations and subsequent up-regulation of mutant p53 protein have also been discussed as a pathomechanism underlying LCH,^25,26  we assessed p53 expression in the lesions by immunohistochemistry. Strong expression of p53 was seen in LCH independent of BRAFV600E mutational status (Figure 4F; P > .19, Student t test). However, p53 was absent in Langerhans cells from the skin (data not shown).

Analysis of LCH is particularly challenging because it is composed of different cell types that are heterogeneously distributed within the lesions. Precise evaluation, therefore, is greatly assisted by tools successfully identifying individual cell populations. Several studies dissected LCH lesions with specific cellular lineage or differentiation markers, such as CD1a, CD207, CD34, CD3, and others.^2,6,27,28  However, none of these markers could unequivocally substantiate the neoplastic cell component in these heterogeneous lesions. Furthermore, the classification of LCH as either reactive or neoplastic disease had not been resolved until the recent detection of activating BRAF mutations in approximately half of LCH lesions. This observation favors the classification of LCH as a neoplastic disease.^9,11,12,29  So far, detection of BRAF mutations had to rely on DNA analysis of the heterogeneous cell populations in LCH. In the present study, we characterize the neoplastic component of LCH with BRAFV600E mutations using the mutation-specific monoclonal antibody VE1.

Mutant BRAF protein was detected by VE1 in 34 of 89 (38%) LCH lesions. This is lower than the frequencies of BRAF mutations in LCH previously published^9,12  and is possibly the result of less multifocal cases being included which tended to have higher mutation rates. We could perform both VE1 immunohistochemistry and direct sequencing in 46 of the 89 LCH lesions. Of these 46 lesions, 20 were identified harboring a BRAF mutation. In 18 lesions, the mutation was detected by both methods. One V600K mutation was missed by VE1 not able to detect this alteration, and one V600E mutation was missed by sequencing because of the low number of tumor cells (20, Figure 5). Sanger sequencing requires ∼ 10% mutant alleles for detection in our experience and according to reports from other laboratories.^30,31  Indeed, 1 lesion (case 19) was initially also missed by direct sequencing but could be identified on immunohistochemistry-guided selection of tissue for DNA extraction. Thus, there was no significant difference in the sensitivity for detection of BRAF mutations between direct sequencing and immunohistochemistry. This receives further support by the observation of the majority of BRAF mutations in LCH being of the V600E type with only a single mutation detected in position 599.^9,12 

Previous analyses of cells from LCH lesions sorted for CD207 underlined these cells as the disease-determining population.²  Our data clearly confirm this concept, demonstrating coexpression of mutant BRAF protein and CD207 (Figure 2), thus proving the CD207 cells being clonal and neoplastic. However, our data also extend previous observations by supporting the existence and illustrating the extent of neoplastic cells not expressing CD207. First, we can demonstrate a significant difference in the abundance of VE1 and CD207-positive cells in selected cases (Figure 1B,H). Second, applying double-immunofluorescence analysis, we could detect cells positive for VE1 but not for CD207 (Figure 2D-F). Thus, BRAFV600E mutations appear not to be exclusively restricted to pathologic LCs but may rather occur in a population of cells of different maturation, with pathologic LCs constituting the most advanced differentiation status.

Two origins for pathologic LCs have been proposed: one is development from immature myeloid cells in the bone marrow; the other is development from peripheral LCs.^2,9,11,23,32  Our findings are compatible with development from immature myeloid cells because of detection of BRAFV600E mutations in both CD207-positive pathologic LCs and in cells expressing CD14 and CD36 but not CD207. The initial BRAF mutation may occur in a less mature myeloid cell with most, but not all, of them differentiating to pathologic LCs. However, these results do not exclude the possibility that mature LCs or their precursors alter their expression profile after acquiring a BRAF mutation or another respective alteration. Indeed, the finding of CD80/CD86 expression on pathologic LCs is well compatible with that hypothesis and in line with previous investigations.^6,24  Moreover, in 2 patients (cases 2 and 17) available for analysis, we did not detect BRAFV600E in bone marrow, although the mutation was present in their LCH lesions. This supports previous observations of not detecting BRAFV600E in peripheral blood of LCH patients with mutant lesions.¹²  Thus, although we do succeed in further characterizing the pathologic cells in LCH, we cannot resolve the debate on their origin.

Investigations on the existence of BRAF mutations in non-LCH histiocytosis with different neoplastic cells might provide further insight into these mechanisms. Similar to LCH, there have been reports on Erdheim-Chester disease establishing clonality by X-chromosome inactivation analysis³³  and cytogenetics,³⁴  making this disease also a candidate for neoplastic rather than reactive inflammatory disease. Very recent findings demonstrate BRAFV600E mutations in Erdheim-Chester disease.³⁵ 

BRAFV600E mutations constitutively activate the RAS/RAF/MEK/ERK signal transduction cascade. Consequently, we detected phosphorylated ERK in these lesions. However, phosphorylated ERK was also present in comparable amounts in LCH lesions without BRAF mutations (Figure 4). Our observation confirms previous findings⁹  and suggests activation of the MAPK/ERK/AKT cascade and perturbation of its regulation by mechanisms other than BRAF point mutations. Similarly, the proliferation activity did not differ in LCH lesions with and without BRAF mutations.

In conclusion, we demonstrate the ability of monoclonal antibody VE1 to detect BRAF mutation in LCH with high sensitivity and the potential of VE1 to assist in determination of the origin of pathologic LCs in these lesions. Immunohistochemical detection of BRAF mutations may greatly assist the diagnosis of LCH, especially in cases with a low proportion of pathologic LCs. Furthermore, it may help direct BRAF mutation-targeting therapy to these patients. The BRAF inhibitors vemurafenib and dabrafenib have shown evidence of therapeutic activity in several BRAFV600E-mutated cancer types, including metastatic melanoma, refractory hairy cell leukemia, and others,^16,36,37  and might also be a treatment option for patients with LCH harboring BRAFV600E mutations. It remains to be seen by analysis of large patient cohorts whether the BRAF mutation characterizes a distinct subgroup among LCH.

The authors thank Kerstin Lindenberg and Jutta Scheuerer for skillful technical assistance.

Contribution: F.S. and A.v.D. designed the study, analyzed data, and wrote the manuscript; F.S., J.M., A.H., and F.L. performed experiments; D.C., M.P., A.S., A.-S.B., M.S., A.K., I.A., L.M., and G.M. provided tissue samples and patient information and analyzed the data; and all authors were involved in manuscript preparation.

Conflict-of-interest disclosure: A.v.D. and D.C. declare shared inventorship of BRAF antibody clone VE1. A patent for diagnostic application of VE1 has been applied for. All terms are being managed by the German Cancer Research Center in accordance with its conflict-of-interest policies. The remaining authors declare no competing financial interests.

Correspondence: Andreas von Deimling, Department of Neuropathology, Institute of Pathology, Ruprecht-Karls-Universität Heidelberg, D-69120 Heidelberg, Germany; e-mail: andreas.vondeimling@med.uni-heidelberg.de.