ARTICLES




                                                                        Acute upregulation of an NKG2D ligand...
ARTICLES

                                                                                                                ...
ARTICLES


                                                                        a        SingleTg 72 h                 ...
ARTICLES

                                                                        epidermis of 15 bi-transgenic mice with
...
ARTICLES

                                                                        with the involvement of local Vg5Vd1+ DE...
ARTICLES


                                                                                                               ...
ARTICLES


                                                                         a                             WT
     ...
Acute upregulation of an NKG2D ligand promotes rapid ...
Acute upregulation of an NKG2D ligand promotes rapid ...
Upcoming SlideShare
Loading in...5
×

Acute upregulation of an NKG2D ligand promotes rapid ...

469
-1

Published on

Published in: Technology, Health & Medicine
0 Comments
0 Likes
Statistics
Notes
  • Be the first to comment

  • Be the first to like this

No Downloads
Views
Total Views
469
On Slideshare
0
From Embeds
0
Number of Embeds
0
Actions
Shares
0
Downloads
2
Comments
0
Likes
0
Embeds 0
No embeds

No notes for slide

Acute upregulation of an NKG2D ligand promotes rapid ...

  1. 1. ARTICLES Acute upregulation of an NKG2D ligand promotes © 2008 Nature Publishing Group http://www.nature.com/natureimmunology rapid reorganization of a local immune compartment with pleiotropic effects on carcinogenesis Jessica Strid1, Scott J Roberts2, Renata B Filler2, Julia M Lewis2, Bernice Y Kwong2, William Schpero2, Daniel H Kaplan3, Adrian C Hayday1,4 & Michael Girardi2,4 The self-encoded ligands MICA (human) and Rae-1 (mouse) for the cytotoxic lymphocyte activating receptor NKG2D are highly expressed in carcinomas and inflammatory lesions and have been linked to immunosurveillance and graft rejection. However, whether NKG2D ligands have an intrinsic ability to acutely regulate tissue-associated immune compartments is not known. Here we show that epidermis-specific upregulation of Rae-1 induced rapid, coincident and reversible changes in the organization of tissue-resident Vc5Vd1 TCRcd+ intraepithelial T cells and Langerhans cells, swiftly followed by epithelial infiltration by unconventional ab T cells. Whereas local Vc5Vd1+ T cells limited carcinogenesis, Langerhans cells unexpectedly promoted it. These results provide unique insight into the early phases of tissue immunosurveillance and indicate that acute changes in NKG2D ligands may alone initiate a rapid, multifaceted immunosurveillance response in vivo. The immunosurveillance theory of Burnet and Thomas proposed that gd T cells are much more susceptible to several regimens of skin adaptive lymphocytes could respond to and reduce tumor growth by carcinogenesis5,6. Indeed, adjuvants for gd T cells are now being used recognizing tumor antigens. Many approaches have provided experi- clinically as part of tumor immunotherapy regimens7,8. mental data in support of this theory. For example, B cell and T cell The carcinogenesis regimens to which gd T cells respond are reactivity to tumors is noted in tumor-bearing patients, whereas commonly characterized by overexpression of the major histocompat- greater susceptibility to spontaneous and chemically induced carcino- ibility complex (MHC) class Ib protein Rae-1. Mouse Rae-1, H60 and genesis has been reported for mice lacking various T cell subsets MULT and their human orthologs MICA and MICB9–12 are likely (ab T cells, gd T cells and natural killer T (NKT) cells), effector candidates for triggering immunosurveillance, because these proteins molecules (perforin, TRAIL and interferon-g (IFN-g)) and associated engage the activating receptor NKG2D, which is expressed by cells of signaling molecules (IFN-g receptor and STAT1)1. the innate (NK cells), transitional (gd and NKT cells) and adaptive Those facts notwithstanding, the generality and clinical applicability (CD8+ T cells) immune responses. Moreover, NKG2D ligand expres- of tumor-specific adaptive responses has remained conspicuously sion seems to be activated very early after cellular dysregulation. controversial, with some questioning whether immune systems can Transcripts encoding Rae-1 and H60 are upregulated in the skin of detect the very early signs of cellular dysregulation that precede mice within 24 h of carcinogen treatment, and this expression is tumorigenesis and that occur in the absence of non-self signals and/ sustained throughout papilloma and carcinoma formation5, reminis- or inflammation2. At the same time, it has become increasingly cent of the common expression of MICA and MICB by human apparent that antitumor potential exists in lymphoid cells such as carcinomas13. Although Rae-1 induction occurs within hours of NKT cells and gd T cells that may ‘sit’ between innate and adaptive DNA damage14, NKG2D ligands are also upregulated during viral immunity. Like adaptive cells, they bear receptors encoded by somat- infection15, in transplant scenarios16 and in many nonmalignant ically rearranged genes, yet like innate cells, they generally lack inflammatory lesions such as type I diabetes (in the nonobese diabetic distinct potential for establishing antigen-specific clonal memory. mouse)17 and rheumatoid arthritis (in humans)18. Indeed, it has been The type of response mounted by such cells has been referred to suggested that antibodies generated against MICA have prognostic as ‘transitional immunity’, and its importance in tumor surveillance value in renal graft rejection16. Consistent with a substantive function is evident in the antitumor activity of the NKT cell agonist for Rae-1 and MICA in immune activation, it has been reported that a-galactosylceramide3,4 and the observation that mice deficient in methylcholanthrene-induced fibrosarcomas develop more readily in 1Peter Gorer Department of Immunobiology, King’s College London School of Medicine at Guy’s Hospital, London SE1 9RT, UK. 2Department of Dermatology and Skin Diseases Research Center, Yale University School of Medicine, New Haven, Connecticut 06511, USA. 3Department of Dermatology and Center for Immunology, University of Minnesota, Minneapolis, Minnesota 55455, USA. 4These authors contributed equally to this work. Correspondence should be addressed to A.C.H. (adrian.hayday@kcl.ac.uk). Received 12 October 2007; accepted 11 December 2007; published online 6 January 2008; doi:10.1038/ni1556 146 VOLUME 9 NUMBER 2 FEBRUARY 2008 NATURE IMMUNOLOGY
  2. 2. ARTICLES Figure 1 Inducible transgenic expression of Rae-1 in the epidermis, where a BiTg SingleTg c + TCRγδ DETC NKG2D is constitutively expressed by resident DETCs. (a) RT-PCR of RNA Rae-1 from tail tissue biopsies, analyzing the induction of Rae-1 transgene β-actin expression in the skin of bi-transgenic mice (BiTg) and single-transgenic mice (SingleTg) after 120 h on doxycycline. Results are compared with amplification of b-actin in the same sample (below). (b) Confocal b microscopy of an epidermal sheet isolated from the ear skin of a wild-type FVB mouse, showing highly dendritic MHC class II–positive Langerhans cells (green) in nonoverlapping association with TCRgd+ DETCs (red). MHCII+ LC Original magnification, Â63. (c) Flow cytometry of epidermal cells from body wall skin, stained with NKG2D-specific (red) or isotype control (blue) © 2008 Nature Publishing Group http://www.nature.com/natureimmunology antibodies, for gated CD3+ TCRgd+ DETCs or CD3– MHC class II–positive Langerhans cells (MHCII+ LC). Data are representative of 21 (a), 15 (b) or 24 (c) experiments. NKG2D lymph nodes27. Added to that, evidence that gd T cells compose a component of host resistance to carcinogenesis5 has fueled the wide- spread belief that the intraepithelial lymphocyte compartment, often mice in which antibodies are used to neutralize NKG2D in vivo19. rich in gd T cells, is itself immunoprotective. Nonetheless, the relative Not unexpectedly, then, viruses and tumors collectively have contributions to carcinogenesis of local versus systemic gd T cells (and several mechanisms for evading NKG2D-mediated recognition15,20, likewise of local dendritic cells, such as Langerhans cells) have never and tumors that develop in the absence of perforin-mediated been tested in vivo. Indeed, the established immunoregulatory capa- cytotoxicity have higher Rae-1 expression19. Such observations cities of T cell receptor gd–positive (TCRgd+) intraepithelial lympho- are consistent with the hypothesis that NKG2D-expressing cells cytes28,29 and of Langerhans cells30 have raised questions about their ‘immunoedit’ tumor phenotypes21. respective immunoprotective potentials. At the same time, one immunoevasion mechanism attributed to Here we present an experimental system that provides a new tumors is chronic expression of Rae-1 or MICA that downregulates framework with which to determine how the early events in tissue NKG2D expression on effector cells20. Modeling of this chronic immunosurveillance may unfold. By divorcing Rae-1 expression from NKG2D ligand expression in transgenic mice has established an any other aspect of tumorigenic or inflammatory dysregulation, our intrinsic capacity of NKG2D ligands to affect immune cells22,23. experiments show that tissue-associated immune cells constitute an Despite such findings, experiments assessing in vivo the intrinsic extremely dynamic compartment that is sensitive to upregulation of immunoregulatory capacity of acute alterations in NKG2D ligand self-encoded MHC-like stress molecules, in the absence of any expression, which would seem to be a prerequisite for its efficacy as a other overt molecular signals. The local immune reorganization generic sentinel of cellular dysregulation, have not been undertaken. induced by acute expression of these stress molecules was unexpect- Indeed, no existing experimental system provides a framework for edly multifaceted and featured an unanticipated infiltration by understanding early events in tissue immunosurveillance in vivo, unconventional ab T cells that proved predictive of the phenotype which may occur before the gross dysregulation that characterizes of tumor-infiltrating lymphocytes. Moreover, although local T cells tumorigenesis, infection and inflammation (such as altered cell showed strong immunoprotective potential, local Langerhans cells growth, perturbations in tissue organization, increased oxidative stress promoted carcinogenesis. and the potential expression of neoantigens). Thus, the key issue of whether NKG2D ligands might initiate immunosurveillance and/or RESULTS inflammation by acting as primary activators of an immune compart- Local immunoregulation by Rae-1 ment remains unresolved. Indeed, the hypothesis has been proposed Although the intrinsic immunoregulatory potential of microbial that full stimulation de novo of lymphocytes requires cell necrosis and/ ligands for host receptors such as Toll-like receptors has been or activation of antigen-presenting cells by microbial sensors such as established24,31, there has been no such assessment of acute exposure Toll-like receptors24. Furthermore, were NKG2D ligands sufficient to to the self-encoded ligands for the activating NKG2D receptor. To promote responses among immune cells, there is as yet no apprecia- assess the intrinsic capacity of acute Rae-1 upregulation to regulate tion of the extent of those responses. For example, are early responses immune responses in a tissue, we designed a ‘bi-transgenic’ mouse exclusively demonstrated by NKG2D+ cells, or can cells respond (Supplementary Fig. 1 online) in which Rae-1 mRNA expression indirectly? These are critical issues relevant not only to the intrinsic could be routinely induced specifically in the epidermis simply capacity of the immune system to respond to nonmicrobial dysregula- through the administration of doxycycline (Fig. 1a). In normal tion but also to the potential of manipulating NKG2D ligands for conditions, the resident immune cells in the epidermis compose an clinical effect. interdigitating network of Langerhans cells and TCRgd+ intraepithelial Much attention has been paid to the fact that stromal immune cells, lymphocytes, known as dendritic epidermal T cells (DETCs). Both cell including macrophages25 and T cells6,26, may in fact promote rather types were highly dendritic in morphology and seemed sufficient in than inhibit tumor growth. Such observations notwithstanding, the number and area to monitor most (and possibly all) basal keratino- initial responses of local, tissue-associated immune cells to acute cytes (Fig. 1b and Supplementary Video 1 online). DETCs, but not activating signals are generally assumed to be protective. In part, Langerhans cells, constitutively expressed NKG2D (Fig. 1c). this assumption is due to the widely accepted hypothesis that during At 3 d after the addition of doxycycline to control (single- infection and tumor growth, tissue-resident dendritic cells such as transgenic) mice, there was no change in the characteristic pattern Langerhans cells may promote the expansion of host-protective of immune cells in the epidermis (Fig. 2a). Because doxycycline was antigen-specific T cell populations after their migration to draining administered with food provided ad libitum, specific time points NATURE IMMUNOLOGY VOLUME 9 NUMBER 2 FEBRUARY 2008 147
  3. 3. ARTICLES a SingleTg 72 h BiTg 72 h b * 350 CD69 (mean MFI) 300 Isotype SingleTg 250 BiTg 200 150 100 Isotype SingleTg BiTg CD69 c ** © 2008 Nature Publishing Group http://www.nature.com/natureimmunology 900 SingleTg BiTg TCRγδ (mean MFI) BiTg 120 h BiTg 120 h 800 700 600 FSH 500 400 300 200 100 Isotype SingleTg BiTg TCRγδ d 450 ** 400 CD86 (mean MFI) Isotype 350 SingleTg 300 BiTg 250 Figure 2 Acute upregulation of Rae-1 in the epidermis induces morphological 200 and activational changes in both Langerhans cell and DETC compartments in vivo. 150 (a) Confocal microscopy of epidermal sheets freshly isolated from mice treated 100 with doxycycline for defined time periods. Arrows in top right image indicate Isotype SingleTg BiTg CD86 juxtaposition of MHC class II–positive Langerhans cells (green) and TCRgd + DETCs (red). Original magnification, Â63 (main images) with Â2.5 ‘zoom’ (enlargement of boxed area). At least ten bi-transgenic and six single-transgenic mice were analyzed per condition. (b–d) Flow cytometry of epidermal cell suspensions from ear skin of mice after 120 h on doxycycline (n ¼ 6 mice). Left (graphs): each dot represents an individual mouse; small horizontal lines indicate the mean. Right (histograms and dot plots): data from one representative mouse. (b) CD69 expression on CD3+ TCRgd+ DETCs. (c) TCR expression on CD3+ TCRgd+ DETCs. (d) CD86 expression on CD3– MHC class II–positive Langerhans cells. *, P ¼ 0.002; **, P o 0.001. Data are representative of three (a) or four (b–d) experiments. reflect the maximum time of induction. There was also no change in expression, but not in single-transgenic mice or in bi-transgenic mice the immune compartment of doxycycline-treated bi-transgenic mice with inducible corneodesmosin expression, all Langerhans and DETC in which the expression of the human epidermal protein corneodes- cell bodies seemed fully rounded, and the residual dendrites of the mosin, instead of Rae-1, was doxycycline inducible (data not shown). Langerhans cells were very narrow (Fig. 2a, inset, and data not These controls collectively establish that neither doxycycline nor the shown). In short, the resident local immune compartment is highly epidermal upregulation of a transgenic protein substantially affects dynamic in response to acute changes in the expression of a single, Langerhans cells or DETCs. self-encoded, ‘stress-associated’ gene product. Conversely, by the same time point, bi-transgenic mice showed considerable changes in response to doxycycline-induced epidermal Unconventional ab T cells expression of Rae-1. First, their DETCs became rounded (Fig. 2a and By 72 h after doxycycline treatment, we detected occasional ab T cells Supplementary Video 2 online), consistent with the established in the epidermis of bi-transgenic mice with inducible Rae-1 expres- guanine nucleotide–exchange factor Vav1–mediated regulation of sion, whereas they were very rare in single-transgenic mice or cytoskeleton induced by NKG2D signaling32. We also detected other bi-transgenic mice with inducible corneodesmosin expression phenotypic changes, such as the consistent, albeit modest, upregula- (Fig. 3a,b and data not shown). By 120 h, we detected large tion of CD69 on DETCs, and the appearance of cells with much lower clusters of ab T cells in bi-transgenic mice with inducible TCRgd expression (Fig. 2b,c). Conversely, the surface expression of Rae-1 expression (Fig. 3c). These ab T cells did not intersperse with NKG2D on DETCs did not decrease during the time of induction DETCs; instead, these two cell populations occupied almost mutually studied here (Supplementary Fig. 2 online). exclusive regions (Fig. 3d). In focal areas, the ab T cells assumed Overt and coordinate changes also characterized the Langerhans cell the dendritic morphology normally associated with DETCs (Fig. 3e), compartment, which was perhaps unexpected, given that Langerhans indicative of intimate molecular associations with keratinocytes cells did not express NKG2D (Fig. 1c). These cells too became and akin to those of ab T cells that fill the epidermal space rounded and had fewer and longer dendrites, consistent with the vacated by DETCs in Tcrd–/– mice34 (Fig. 3f). These findings demon- morphologic changes reported after more complex epidermal strate that such morphological changes to ab T cells occur rapidly stimuli33 (Fig. 2a). Langerhans cells also showed modest but consis- and may constitute a normal aspect of acute immunosurveillance tent CD86 upregulation (Fig. 2d). It was common for DETC and in epithelia. Langerhans cell bodies to become juxtaposed (Fig. 2a); as a result of Notably, despite the complete reconfiguration of the epidermal these collective changes, large areas of the epidermal sheets were no immune compartment evident by 120 h, all changes were fully longer proximally contacted by DETCs or Langerhans cells. These reversible within 3 d of doxycycline withdrawal (Fig. 3g,h), which changes were particularly accentuated by 120 h after doxycycline demonstrated a rapid ‘resetting’ of the local immune compartment administration, when in the bi-transgenic mice with inducible Rae-1 after removal of the epithelial stress ligand. Flow cytometry of the 148 VOLUME 9 NUMBER 2 FEBRUARY 2008 NATURE IMMUNOLOGY
  4. 4. ARTICLES epidermis of 15 bi-transgenic mice with inducible Rae-1 expression permitted us to a SingleTg 72 h b BiTg 72 h c BiTg 120 h d BiTg 120 h establish that the greater representation of ab T cells in the induced mice was statistically significant relative to that of either single- transgenic mice (P ¼ 0.003) or induced bi- transgenic mice from which doxycycline had been withdrawn (P ¼ 0.024; Supplementary Fig. 3 online). Because overt epidermal infiltration by e f g DETC and LC h DETC and αβ T cell © 2008 Nature Publishing Group http://www.nature.com/natureimmunology BiTg 120 h Tcrd –/– ab T cells after Rae-1 upregulation occurred rapidly and in the absence of obvious antigen, it seemed unlikely that it was attributable to conventional ab T cells recruited from the local lymph nodes. Infiltrating ab T cells were uniformly NKG2D+NK1.1+CD4–CD8– (Fig. 4a), thus refuting the hypothesis that these cells may have been conventional NKG2D+CD8+ memory T cells ‘percolating’ Figure 3 TCRab+ cells rapidly infiltrate the epidermis after upregulation of Rae-1. (a–f) Confocal through the skin. Consistent with an effector microscopy of TCRgd+ DETCs (red) and TCRab+ cells (green) in epidermal sheets freshly isolated phenotype, infiltrating ab T cells expressed from mice treated with doxycycline for defined time periods. At least ten bi-transgenic and six single- surface CD44 but not CD62L (Fig. 4a). Stain- transgenic mice were analyzed per condition; images of representative fields were obtained after ing with a set of antibodies recognizing spe- analysis of whole epidermal sheets. (g,h) Confocal microscopy of the reversibility of changes induced cific TCRb variable (TCR Vb) regions showed by Rae-1 expression in the epidermis of bi-transgenic mice given doxycycline (dox) for 120 h, then taken off doxycycline for 72 h. (g) TCRgd+ DETCs are red; MHC class II–positive Langerhans cells that the TCR repertoire of infiltrating ab are green. (h) TCRgd+ DETCs are red; TCRab+ cells are green. Original magnification, Â63. Data are T cells was distinct from that of splenic ab representative of four (a–f) or three (g,h) experiments. T cells (Fig. 4b). For example, although we consistently detected Vb6+ cells in both skin and spleen, there was negligible representation in the skin of other TCR Vg5+ T cells juxtaposed with upregulated Rae-1 protein expression Vb regions (such as Vb14 and Vb7) that were invariably present among mainly in basal and follicular areas from which squamous cell splenic T cells of wild-type mice and among the epidermal ab T cells of tumors most often develop37,38 (Supplementary Fig. 4 online). Tcrd –/– mice. Conversely, many of the skin cells in the bi-transgenic Those findings prompted us to determine whether the immune cell mice with inducible Rae-1 expression expressed Vb2, which is rare populations coordinately responsive to Rae-1 upregulation were uni- among systemic cells of wild-type mice but which frequently pairs with versally host protective. Although it is established that Tcrd –/– mice Va14 on NKT cells35,36. The mean fluorescence intensity of TCR are more susceptible to two-stage chemical carcinogenesis5, the relative staining on infiltrating ab T cells was consistently lower than that of contributions of systemic versus intraepithelial T cells have not been systemic cells. However, because the diverse repertoire of ab T cells that established in this or any other tumor surveillance system. fills the epidermal space in Tcrd –/– mice commonly had large amounts We exploited the fact that over 90% of the natural DETC of surface TCR complexes (Fig. 4b), we could exclude the possibility population expresses Vg5Vd1 TCRs that bind to the clonotypic that low TCR expression in bi-transgenic mice with inducible Rae-1 TCR-specific antibody 17D1 (ref. 39). Such cells develop in the fetal expression was an artifact of the isolation procedure. Instead, it more thymus and in the adult are found only in the skin. By intercrossing likely reflected the cells’ activated state and/or a characteristic feature of Tcrg-V5 –/– and Tcrd-V1–/– mice of the FVB strain, we generated Tcrg- the cells, as has been reported for NKT cells36. Consistent with that, V5 –/–Tcrd-V1–/– double-knockout mice as well as Tcrd-V1–/– and most T cells infiltrating the epidermis of bi-transgenic mice with Tcrg-V5 –/– single-knockout mice. Tcrd-V1–/– but not Tcrg-V5 –/– inducible Rae-1 expression bound specifically to a-galactosylceramide– mice had Vg5+ DETCs (about 60% of epidermal gd T cells; CD1d tetramers, in contrast to most splenic T cells, DETCs or resident Fig. 5a), whereas, as noted before39, some DETCs in Tcrg-V5 –/– epidermal ab T cells in Tcrd–/– mice (Fig. 4c). Infiltration of mice expressed TCRs that bound to 17D1 (about 40% of epidermal CD4–CD8– ab T cells may be a previously unrecognized aspect of gd T cells). As expected, neither Vg5+ nor 17D1+ cells were present in tissue immunosurveillance. Indeed, by individual analysis of four mice, the Tcrg-V5 –/–Tcrd-V1 –/– mice, although such mice had many ‘repla- we consistently found that tumor-infiltrating lymphocyte populations cement gd TCR+ DETCs’ (Fig. 5a). Notably, the Tcrg-V5 –/–Tcrd-V1 –/– from papillomas and carcinomas induced by two-stage chemical mice resembled Tcrd–/– mice in their susceptibility to tumors induced carcinogenesis were enriched approximately tenfold for CD4–CD8– ab by a two-stage carcinogenesis regimen, thereby establishing a key host- T cells, relative to the enrichment of systemic lymphocytes for protective function for local T cells (Fig. 5b). Nonetheless, given that such cells (7.68% ± 2.50% for tumor-infiltrating lymphocytes, versus Tcrg-V5 –/–Tcrd-V1–/– mice had replacement DETCs, local T cell 0.70% ± 0.31% for peripheral blood lymphocytes; P ¼ 0.03). compartments themselves are insufficient to provide protection. Instead, intraepidermal T cells with appropriate TCR specificities Protective function of skin intraepithelial lymphocytes and/or responses are required, as were retained in the Tcrd-V1–/– Because Rae-1 RNA is rapidly upregulated during chemical carcino- and Tcrg-V5 –/– mice, which showed no significant increase in tumor genesis5, the coordinated response to acute local Rae-1 expression may susceptibility relative to that of wild-type mice (Fig. 5b). be considered a model for the early phases of tumor immunosurveil- Our results established a reliance of tumor resistance on certain lance. Indeed, in carcinogen-treated mice, we detected rounded-up types and/or TCR specificities of intraepidermal T cells. Consistent NATURE IMMUNOLOGY VOLUME 9 NUMBER 2 FEBRUARY 2008 149
  5. 5. ARTICLES with the involvement of local Vg5Vd1+ DETCs in the early stages of Whereas all wild-type mice subjected to low doses of chemical immunosurveillance, Tcrg-V5 –/–Tcrd-V1–/– mice (like Tcrd –/– mice) carcinogens had tumors by 14 weeks after initiation, over 50% of were about 2.2-fold more susceptible than wild-type mice to papil- Langerhans cell–deficient mice remained tumor free by 16 weeks loma formation (Table 1). Conversely, tumor progression, as (Fig. 6a,b). Even when high doses of chemicals were applied to the measured by the ratio of carcinomas to total tumors, was similar in highly susceptible FVB strain, most Langerhans cell–deficient mice Tcrg-V5 –/–Tcrd-V1–/– mice (average, 0.634), wild-type mice (average, remained tumor free at 7 weeks after initiation, by which time point 0.665) and single-knockout mice (average, 0.619). all wild-type mice had many tumors (Fig. 6c). Moreover, Langerhans cell–deficient mice rarely had more than one or two tumors per Tumor-promoting function of Langerhans cells mouse, whereas wild-type mice often had more than 20 tumors per We undertook analogous studies with mice rendered deficient in mouse (Fig. 6a–c). This degree of resistance far exceeded that reported © 2008 Nature Publishing Group http://www.nature.com/natureimmunology Langerhans cells by the use of a diphtheria toxin transgene regulated before for Cd8–/– mice26, which suggested that the resistance could not by the Langerin promoter30. Of note, comprehensive ablation of be explained solely by a failure to prime tumor-promoting T cells, Langerhans cells in these mice is obtained without exogenous addition such as CD8+ tumor infiltrates in chemically induced carcinomas, of diphtheria toxin, and it is highly selective, with both dermal which express IFN-g, tumor necrosis factor and cyclooxygenase-2 but dendritic cell and splenic and lymph node Langerin-positive dendritic lack cytolytic effector molecules. Indeed, the resistance to carcino- cell populations remaining intact30. In addition, the number and genesis of Langerhans cell–deficient mice was independent of the morphology of gdTCR+ DETCs was normal30 and the epidermis was presence or absence of either some or all ab T cells (Fig. 6d,e). completely devoid of ab T cells (Supplementary Fig. 5 online). In sum, the two tissue-resident immune cell types that rapidly and a 6.8 93.2 5.8 5.5 98.8 4.5 95.5 TCRαβ TCRαβ CD8α CD8α CD44 0 0 1.2 NKG2D CD4 CD8β CD62L NK1.1 b WT spleen 0 5.5 10.8 10.3 7.0 1.5 10.1 BiTg CD3 epidermis 10.4 0.2 0.3 4.2 1.6 0.4 0.2 –/– Tcrd epidermis 0.8 0.1 2.9 15.4 34.1 12.1 13.7 TCR Vβ2 TCR Vβ3 TCR Vβ4 TCR Vβ6 TCR Vβ7 TCR Vβ10b TCR Vβ14 + c WT spleen Tcrd –/– epidermis BiTg epidermis TCRγδ DETC Unloaded + α-GalCer Unloaded + α-GalCer + α-GalCer + α-GalCer CD3 0 3.2 0.2 0.3 58.3 0.2 CD1d tetramer Figure 4 Epidermis-infiltrating TCRab+ cells differ from conventional circulating ab T cells and all express NK cell markers. (a,b) Flow cytometry of wild-type spleen cell suspensions and of epidermal cell suspensions prepared from the ears of Tcrd –/– and of bi-transgenic mice after 120 h on doxycycline. Stringent isolation ensured no dermal contamination, and no TCRab+ cells were found in single-transgenic or wild-type controls. Plots are of individual mice; at least ten bi-transgenic mice were analyzed per condition. (a) Plots gated on CD3+TCRab+ epidermal cells. (b) TCRb repertoire in bi-transgenic and Tcrd–/– epidermis and wild-type (WT) spleen; plots gated on CD3+TCRgd– cells. (c) Binding of a-galactosylceramide–CD1d tetramers to epidermal T cells; plots gated on CD3+TCRgd– splenic or epidermal cells. Seven bi-transgenic, three wild-type and three Tcrd –/– mice were analyzed. Numbers in quadrants or outlined areas indicate percent cells in each. Data are representative of four (a,b) or two (c) experiments. 150 VOLUME 9 NUMBER 2 FEBRUARY 2008 NATURE IMMUNOLOGY
  6. 6. ARTICLES Tcrg-V5 –/– Tcrd-V1 –/– Tcrg-V5 –/–Tcrd-V1–/– Tcrd –/– a WT b Tcrd –/– 0.7 9 7 0 8 12 10 0 0 0 Tcrg-V5 –/–Tcrd-V1 –/– Tcrg-V5 –/– TCRγδ (GL3) ** x Tcrd-V1 –/– 12 WT ** 10 0 0 0 0 0 * Tumors per mouse 8 Vγ 5 WT Tcrg-V5 –/– Tcrd-V1 –/– Tcrg-V5 –/–Tcrd-V1 –/– 6 x 0 91 4 0 0 0 0 0 * x © 2008 Nature Publishing Group http://www.nature.com/natureimmunology 4 Vγ 5Vδ1(17D1) x 2 x 0 6 0 39 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 Time after DMBA initiation (weeks) Vγ 5 Figure 5 Mice selectively deficient in the prototypic Vg5Vd1+ DETCs are more susceptible to tumor development. (a) Flow cytometry of intraepidermal T cell populations. Top row, gating on all epidermal cells; stained with anti-TCRgd (GL3) and anti-Vg5 (536); bottom row, gating on TCRgd+ cells; stained with anti-Vg5Vd1 (17D1) and anti-Vg5 (536). Numbers in quadrants indicate percent cells in each. (b) Tumor development in mice subjected to low-dose two- stage chemical carcinogenesis (200 nmol DMBA initiation; 10 nmol weekly TPA promotion). Statistical analysis, Table 1. Data are representative of three experiments (a) or four experiments with 10–14 mice in each (b; error bars, s.e.m.). coordinately responded to acute Rae-1 upregulation had, on aggre- reorganization of tissue layout and cellular composition, including gate, diametrically opposite effects on tumor incidence, thus establish- bulk coincident changes in DETCs and Langerhans cells. The changes ing the pleiotropy of local immunosurveillance. in the structure and the apparent migration of DETCs can be predicted by the known regulation of Vav-1 ‘downstream’ of DISCUSSION NKG2D32. However, the bulk changes in Langerhans cells are notable Tissue immunosurveillance, with particular reference to tumor given that Langerhans cells do not express NKG2D; these findings immunology and inflammatory diseases, has been the subject of indicate a potential for local T cells to ‘dictate’ the activity of tissue- much speculation. Interest in determining whether immune cells can resident myeloid cells. The mechanism supporting this indirect respond to dysregulated self, in addition to foreign pathogen- NKG2D-mediated regulation remains to be investigated and may be associated molecular patterns, is compounded by the clinical potential related to increasing evidence that T cells can profoundly regulate of enhancing antitumor responses or inhibiting excessive inflammatory monocytes and dendritic cells, as well as vice versa. tissue surveillance. In this context, the prospect that immunosurveil- The aggregate consequence of our study is extension of the primary lance may be focused on the recognition of self-encoded MHC-related responsiveness of immune compartments from engagement of microb- stress molecules such as Rae-1 and MICA, which are upregulated in a ial pattern-recognition receptors on myeloid and epithelial cells to spectrum of tissue inflammation events associated with both malignant the recognition of dysregulated self-encoded molecules by lymphoid and nonmalignant lesions, is of great interest. Nonetheless, few cells. Because it has been established in transgenic mice that human experimental data have demonstrated either the events that constitute MICA and mouse Rae-1 have mostly equivalent capacities to engage tissue immunosurveillance or their temporal sequence. At least part of mouse NKG2D22, it is reasonable to extrapolate from our data to the the problem is the complexity of the myriad components presented by effects of MICA upregulation in humans. Further refinements of our full-fledged malignant or inflammatory lesions, coupled with the system to allow upregulation of other molecules in the skin and in difficulty in identifying and examining early lesions. various other tissues will permit the classification of other self- In contrast, here we exploited a transgenic system to establish that encoded MHC-related molecules according to their capacity to acutely acute upregulation of Rae-1 is itself an axis of immunoregulation. regulate local immune compartments. Our data have shown the dynamic nature of a tissue-associated We have not established whether NKG2D-ligand upregulation immune compartment (in this case, the skin) in response to acute alone is sufficient to promote a full-fledged adaptive immune alterations in only a single self-encoded molecule. We noted complete response. However, the changes described here are overt and make it difficult to accept the proposal that the immune system cannot ‘see’ cellular dysregulation and hence cannot naturally mount tumor Table 1 Tumor development in wild-type and DETC-deficient mice immunosurveillance2,40. Indeed, the response of local T cells to Rae-1 upregulation and their contribution to protection at a Mouse strain Tumors P value Carcinomas P value seemingly early stage in carcinogenesis jointly add weight to other Tcrd –/– 11.08 ± 1.74 r 0.002 5.46 ± 1.24 r 0.05 experimental evidence of the existence of a natural tumor immuno- Tcrg-V5–/– 5.83 ± 1.09 NS 3.58 ± 0.99 NS surveillance potential41. Tcrd-V1–/– 6.07 ± 0.95 NS 3.79 ± 0.73 NS The primary responsiveness of the epidermal immune compart- Tcrg-V5–/–Tcrd-V1–/– 11.17 ± 1.36 r 0.0004 7.08 ± 1.05 r 0.01 ment to acute upregulation of NKG2D ligand described here is a WT 5.08 ± 0.84 – 3.38 ± 0.75 – logical model for the early phases of immunosurveillance, because upregulation of Rae-1 and MICA in vivo can occur early after infection Data are mean values (± s.e.m.) per mouse at week 17 after DMBA initiation. P values are relative to wild type; NS, not significant. Data are representative of two experiments or carcinogen exposure; in the latter case, it is upregulated before the with 10–14 mice in each. formation of a palpable tumor5. Indeed, upregulated Rae-1 protein NATURE IMMUNOLOGY VOLUME 9 NUMBER 2 FEBRUARY 2008 151
  7. 7. ARTICLES a WT b c WT d Tcrb –/– 8 Langerin-DTA 25 Langerin-DTA Tcrb –/– Langerin-DTA 8 Tumors per mouse Tumors per mouse Tumors per mouse WT 6 20 6 15 4 4 10 2 2 Langerin-DTA 5 0 0 0 © 2008 Nature Publishing Group http://www.nature.com/natureimmunology 8 9 10 11 12 13 14 15 16 4 5 6 7 8 9 10 11 12 8 9 10 11 12 13 14 15 16 Time after DMBA initiation (weeks) Time after DMBA initiation (weeks) Time after DMBA initiation (weeks) –/– Figure 6 Mice deficient in Langerhans cells are protected from tumor development. (a,b) Induction of tumor formation e Cd4 by a low-dose two-stage chemical carcinogenesis protocol (200 nmol DMBA initiation, 10 nmol weekly TPA promotion) Cd4 –/– Langerin-DTA in wild-type mice and in mice rendered deficient in Langerhans cells via diphtheria toxin transgene regulated by the Langerin promoter (Lagerin-DTA): 5.67 ± 1.63 versus 1.00 ± 0.47 tumors per mouse at week 16, respectively 8 Tumors per mouse (P o 0.005). (b) Wild-type and Langerhans cell–deficient mice from the low-dose experiment in a. (c) Tumor induction by a high-dose two-stage chemical carcinogenesis protocol (400 nmol DMBA; 40 nmol TPA weekly) in wild-type and 6 Langerhans cell–deficient mice: 20.20 ± 0.59 versus 2.39 ± 0.59 tumors per mouse at week 16, respectively (P o 0.0000001). (d,e) Induction of tumor formation by the low-dose two-stage chemical carcinogenesis protocol in a 4 for mice deficient in either all ab T cells (TcrbÀ/À) or both ab T cells and Langerhans cells (Tcrb–/–Langerin-DTA; d) and 2 for mice deficient in either all CD4+ T cells (Cd4–/–) or both CD4+ T cells and Langerhans cells (Cd4–/–Langerin-DTA; e): 9.20 ± 1.80 tumors per mouse for Tcrb–/– versus 0.36 ± 0.28 tumors per mouse for Tcrb–/– Langerhans cell–deficient 0 (P o 0.0005); 6.14 ± 1.48 tumors per mouse for Cd4–/– versus 0.36 ± 0.20 tumors per mouse for Cd4–/– Langerhans 8 9 10 11 12 13 14 15 16 cell–deficient (P o 0.001). Data are representative of three (a,b) or two (c–e) experiments with 10–15 mice (a), Time after DMBA initiation 10 mice (c,d) or 10–14 mice (e) in each. (weeks) ‘decorated’ the basal and follicular areas from which tumors most immunosuppressive functions44. In yet another system, different frequently emanate37,38. Thereafter it was generally sustained through- subsets of NKT cells enhance and suppress tumor surveillance36,45,46. out the formation of papillomas and carcinomas, reminiscent of Obviously, the main functions of the skin-infiltrating CD4–CD8– the sustained expression of MICA by many human cancers. Because T cells cells detected here merit further study that may clarify the sustained chronic expression of Rae-1 and/or MICA ultimately cells’ aggregate contributions in human tissue surveillance. However, downregulates T cell and NK cell responses20,22,23, there is consider- in practical terms, this may require specific deletion of the CD4–CD8– able weight to the idea that the response to acute expression of Rae-1 NKT cell population, as CD4–CD8– NKT cells and CD4+ NKT cells and/or MICA is a key component of immunosurveillance and exerts a have been proposed to have different functions36. substantial pressure for immunoevasion. The second insight provided by our study relates to the functional Our work has provided much new insight into the events occurring potential of local T cells. Whereas DETCs can have aggregate im- during early tumor immunosurveillance. First, we have documented a munosuppressive functions29, our study has unequivocally established hitherto unrecognized enrichment of CD4–CD8– TCRab+ NKT cells in their immunoprotective function in terms of carcinogenesis. Given tumor infiltrates. It will be useful to resolve the origin of these cells that DETCs act early (on papilloma formation), our data seem to (dermis or blood). Furthermore, such cells may represent a common uphold the long-held hypothesis that intraepithelial lymphocytes feature of tissue immunosurveillance distinct from the antigen-driven directly monitor surrounding tissue dysregulation to the host’s benefit, infiltration of tissues by cognate conventional ab T cells after their a hypothesis that has nonetheless not been tested experimentally clonal expansion in local lymph nodes. Of note, independent reports before. And yet our data refute the related hypothesis that ‘any have indicated that NKT cells are greatly enriched among intraepithelial intraepithelial lymphocyte will do the job’. Tcrg-V5–/–Tcrd-V1–/– lymphocytes in human psoriatic epidermis42,43, where they are asso- double-knockout mice were more susceptible to carcinogenesis yet ciated with cells expressing CD1d that is likewise upregulated within had many ‘replacement’ skin intraepithelial lymphocytes that were 24 h of the induction of contact dermatitis42. The NKT cells in such insufficient for protection. In contrast, the intraepithelial lymphocytes lesions are not obviously of the classical, invariant type; future studies of Tcrg-V5–/– and Tcrd-V1–/– single-knockout mice did provide should clarify the composition and heterogeneity of the infiltrating protection. Examination of the genes differentially expressed in these CD4–CD8– ab T cell repertoire in the system described here. Notably, different subsets may identify key mechanisms of local T cell protec- the possibility exists that CD4–CD8– T cells and/or NKT cells may form tion. Moreover, were the results to extend to other tissues, such as the an important commonality in the epidermal immunology of mice and gastrointestinal tract, our data might hold promising clinical implica- humans that has been obscured by the lack of an obvious DETC tions for targeted immunotherapy. compartment in humans. Rather than accepting the idea that the two The third insight provided by our work is that Langerhans cells, on species have radically different mechanisms, we now hypothesize that aggregate, promoted carcinogenesis. This result was unexpected for infiltration of CD4–CD8– T cells and/or NKT cells (noted here in mice several reasons. First, Langerhans cells are ordinarily viewed as and reported before in humans42,43) may be a critical conserved feature contributing to the early activation of protective immune responses. of epidermal surveillance. Second, deficiency in Langerhans cells reverted tumor susceptibility CD4–CD8– T cells and NKT cells can show considerable pleiotropy, even to high-dose carcinogen in the FVB strain, which is particularly and whereas human skin-derived NKT cells can produce potentially susceptible to carcinogenesis. The detailed mechanisms forming the pathogenic IFN-g42, NKT cells from irradiated mice seem to exert basis of Langerhans cell–mediated tumor promotion remain to be 152 VOLUME 9 NUMBER 2 FEBRUARY 2008 NATURE IMMUNOLOGY

×