CD1c⁺ blood dendritic cells have Langerhans cell potential

Langerhans cells (LCs) are antigen-presenting cells of the epidermis that are self-renewing in the steady state¹  but recruited from blood-borne precursors after inflammation.^2-4  Observations in humans also confirm that LCs can be self-maintained^5-7  or replaced by bone marrow–derived cells in the context of transplantation and inflammation.^7-10 

The nature of bone marrow–derived LC precursors that repopulate the epidermis after inflammation is incompletely defined. Experiments in mice with clodronate depletion and bead-labeling suggest a monocyte origin but do not completely exclude other precursors.²  More recent observations in mice lacking LCs as a result of Id2 knockout or conditional ablation show that LC repopulation occurs in 2 waves.^3,4  The epidermis is initially infiltrated by a short-term precursor with low langerin expression and features in common with monocytes, followed by a long-term precursor that acquires the full phenotype of LCs, including self-renewal capacity.^3,4 

In humans, langerin⁺ cells can be made in vitro from monocytes.^11-13  Experiments with CD34⁺ progenitors also demonstrate the existence of an LC-differentiation pathway that appears independent of CD14⁺ monocytes.^14,15  A previous report suggested the existence of CD3/7/14/16/19-negative LC precursors in human blood and, although noted to express CD1c, these were incorrectly described as expressing CD1a.^16,17  Here we show that CD1c⁺ blood dendritic cells (DCs) are alternative LC precursors that achieve higher levels of langerin, CD1a, and Birbeck granules than CD14⁺ monocytes when exposed to soluble factors known to drive LC differentiation.

Blood mononuclear cells were obtained from healthy volunteers under local ethical approval. CD14⁺ monocytes, CD16⁺ monocytes, CD1c⁺ DCs, and CD123⁺ plasmacytoid DCs were sorted from peripheral blood mononuclear cells (PBMCs). LCs were isolated from epidermal sheets separated from whole-skin keratome sections with dispase (Invitrogen) 1 mg/mL incubated at 37°C for 90 minutes in RPMI 1640 and subsequently digested with collagenase (Worthington Type IV) 1.6 mg/mL incubated for 12 hours at 37°C in RPMI 1640 with 10% fetal bovine serum. Sorting was performed with an ARIA Fusion (Becton Dickinson) using previously described protocols (supplemental Figure 1, available on the Blood Web site).¹⁸  Ten-thousand cells were cultured in RPMI 1640 with 10% fetal bovine serum or X-Vivo in 100 μL. Supplements were added at the following concentrations: GM-CSF 50 ng/mL, TGFβ, 10 ng/mL, and BMP7 200 ng/mL. Cultures were maintained for 3 to 7 days and supplemented with fresh cytokines on day 4.

Flow cytometry analysis was performed with an FACS Canto (Becton Dickinson) using appropriate isotype controls. Antibodies were from Becton Dickinson unless stated otherwise (antigen fluorochrome clone): CD1a BV421 HI149, CD1c PeCy7 L161, CD3 FITC SK7, CD11b APC ICRF44, CD11c A700 B-ly6; CD11c APCCy7 Bu15, and CD14 BV650 M5E2 (all from Biolegend); CD14 ECD RMO52 (Beckman Coulter); CD16 APCCy7 3G8; CD19 FITC 4G7; CD34 FITC 8G12; CD20 FITC L27; CD56 FITC NCAM16.2; CD83 FITC HB15e; CD123 PerCPCY5.5 7G3; CD207 PE DCGM4 (Beckman Coulter); E-Cadherin APC 67A4 (Biolegend); EpCAM APC EBA-1; and HLA-DR V500 G46-6.

Cells were fixed for electron microscopy (EM) according to standard protocols in 2% glutaraldehyde, and then pelleted, dehydrated, and fixed in resin (reagents from TAAB Laboratory, Aldermarston, UK). Ultrathin sections were cut with a diamond knife RMC MT-XL ultramicrotome and examined with a Philips CM 100 Compustage (FEI) Transmission Electron Microscope. Images were collected with an AMT CCD camera (Deben).

DC and monocyte fractions of human PBMCs were cultured in conditions that induce langerin in progenitor cells, including BMP7.^19-23  This induced a rapid upregulation of langerin and CD1a dual expression by CD1c⁺ DCs, peaking within 3 days (Figure 1A-B). Langerin⁺ cells also appeared in CD14⁺ and CD16⁺ monocytes, but the level of expression was lower and did not increase further by day 7. Cells did not expand in culture, and approximately 60% to 70% of monocytes and 20% to 30% of CD1c⁺ DCs were recovered after 7 days (Figure 1C). Induction of a CD1a⁺ langerin^high population was restricted to CD1c DCs treated with TGFβ, BMP7, or both (Figure 1D). No significant synergy was observed between TGFβ and BMP7, in contrast to recent results obtained with TGFβ and thymic stromal lymphopoietin (TSLP).²⁴  No langerin induction was seen in medium with serum alone, but serum-free medium with supplements also failed to induce any expression, in contrast to results obtained with CD34⁺ progenitor cells.^20,23  Exhaustive testing of different serum-free media was not conducted.

Day 7 cultures of CD1c⁺ DCs and CD14⁺ monocytes treated with GM-CSF and BMP7 were harvested for EM (all cells). Birbeck granules were observed in 12 of 20 CD1c⁺ ultrathin sections of CD1c⁺ DCs, with many sections containing >10 granules. In contrast, Birbeck granules were rarely found in cultures of CD14⁺ monocytes (a single granule in 20 cell sections) (Figure 2A-B). CD1c⁺ DCs developed similar cytologic appearance to freshly isolated LCs (supplemental Figure 2). Extended phenotyping showed induction of EpCAM and E-cadherin, and low expression of CD11b, similar to primary LCs. Langerin⁺ CD1c⁺ DCs remained immature as assessed by the expression of HLA-DR and CD83 (Figure 2C). The expression of EpCAM was highest on langerin^high cells derived from CD1c⁺ DCs; in contrast, langerin⁺ cells derived from CD14⁺ monocytes had low EpCAM and higher CD11b expression (Figure 2D).

While this paper was under review, LC formation by CD1c⁺ DCs treated with TGFβ and TSLP was reported.²⁴  Together, these results indicate that there are several pathways to the generation of langerin⁺ cells in humans and that, under the conditions tested, CD1c⁺ DCs but not CD14⁺ monocytes, generate langerin^high CD1a⁺ EpCAM⁺ cells with Birbeck granules. Notably, monocytes prepared by negative selection contain CD1c⁺ DCs (supplemental Figure 3). It is therefore likely that CD1c⁺ DCs contribute significantly to LC development when these preparations are induced to differentiate.¹¹ 

There are 2 limitations to this study. First, other conditions not tested here, such as Notch ligation, may be capable of inducing higher langerin in CD14⁺ monocytes.^12,13  Birbeck granule formation was reported in monocytes exposed to Delta-1,¹²  although with Jagged, langerin expression did not exceed that observed with GM-CSF and TGFβ.¹³  Second, although the Birbeck granule is the ultrastructural hallmark of LC phenotype in vivo, the derivation of Birbeck granule–containing cells in vitro does not prove a precursor-progeny relationship. The expression of EpCAM and E-cadherin and low level of CD11b are consistent with an LC phenotype, but further expression profiling and functional studies would be required to evaluate fully the proximity of derived cells to primary LCs. It would be of interest to determine whether a similarly high capacity to crosspresent antigen is found in CD1c⁺ DC–derived LC-like cells, as shown by those obtained from CD34⁺ progenitors.^21,22  However, it has been reported that crosspresentation is not as marked in freshly isolated primary LCs¹⁸  and may not be discriminatory for LC differentiation, because it is easily induced in CD1c⁺ DCs under other conditions.²⁵ 

Although the cultures contained serum, as in other recent publications describing DC differentiation,^24,26  the factors GM-CSF, TGFβ, and BMP7 are expressed by epithelial tissues under physiologic conditions.²³  Previous experiments on CD34⁺ progenitors have highlighted the difference between a TGFβ-dependent pathway of LC generation from CD1a⁺ intermediates vs the TGFβ-independent generation of monocyte-derived DCs that are unable to express langerin.^14,15  It is likely that this dichotomy reflects the differential ability of CD1c⁺ DCs and monocytes to generate LCs with high langerin, CD1a, EpCAM, and Birbeck granules that we now observe.

Recent data in mice describing 2 waves of repopulation of the LC compartment show an early monocytelike EpCAM^– langerin^low wave succeeded by a myeloid precursor of unknown origin. In situ studies also show that LC repopulation after inflammation occurs with similar kinetics in humans.²⁷  CD1c⁺ DCs have properties that make them candidates for the second-wave LC precursor. Potential homologs of the CD1c⁺ blood DC in mice such as the circulating pre-cDC may contain the long-term precursor of murine LCs.

Finally, these data explain the puzzling claim that there are CD1a⁺ “LC precursors” in human blood.¹⁶  Blood DCs were reported to express CD1a as a result of incorrect assignment of clone BB5 as a CD1a-specific antibody. It is now known that BB5 recognizes CD1b/c and not CD1a.¹⁷  The nonmonocyte cells in human blood that give rise to LCs in response to TGFβ and BMP7 in vitro are now clearly shown to be CD1c⁺ DCs.

The authors thank Tracey Davey and Newcastle University Electron Microscopy Research Services for assistance with electron microscopy.

This work was supported by Wellcome Trust grants WT088555MA (M.H.) and WT101155/Z/13/Z (V.B.), the Histiocytosis Research Trust, the Histiocytosis Association, and Bright Red (registered charity 1105891).

Contribution: P.M. performed experiments and analyzed data; V.B. designed and performed experiments and analyzed data; M.G. performed experiments; M.H. designed experiments and analyzed data; and M.C. designed experiments and wrote the paper.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Matthew Collin, Institute of Cellular Medicine, Newcastle University, Framlington Pl, Newcastle upon Tyne NE2 4HH, UK; e-mail: matthew.collin@ncl.ac.uk.