Upregulation of CD26 expression in epithelial cells and ... - Nature

Oncogene (2012) 31, 992–1000

& 2012 Macmillan Publishers Limited All rights reserved 0950-9232/12 www.nature.com/onc

ORIGINAL ARTICLE

Upregulation of CD26 expression in epithelial cells and stromal cells during wound-induced skin tumour formation EN Arwert1,3, RA Mentink1,3, RR Driskell2, E Hoste1, SJ Goldie1, S Quist1 and FM Watt1,2 1 Cancer Research UK Cambridge Research Institute, Cambridge, UK and 2Wellcome Trust Centre for Stem Cell Research, Cambridge, UK

We have previously described InvEE transgenic mice in which non-dividing, differentiating epidermal cells express oncogenically activated MAPK kinase 1 (MEK1). Skin wounding triggers tumour formation in InvEE mice via a mechanism that involves epidermal release of IL-1a and attraction of a pro-tumorigenic inflammatory infiltrate. To look for potential effects on the underlying connective tissue, we screened InvEE and wild-type epidermis for differential expression of cytokines and immune modulators. We identified a single protein, CD26 (dipeptidyl peptidase-4). CD26 serum levels were not increased in InvEE mice. In contrast, CD26 was upregulated in keratinocytes expressing mutant MEK1 and in the epithelial compartment of InvEE tumours, where it accumulated at cell–cell borders. CD26 expression was increased in dermal fibroblasts following skin wounding but was downregulated in tumour stroma. CD26 activity was stimulated by calcium-induced intercellular adhesion in keratinocytes, suggesting that the upregulation of CD26 in InvEE epidermis is due to expansion of the differentiated cell layers. IL-1a treatment of dermal fibroblasts stimulated CD26 activity, and therefore epidermal IL-1a release may contribute to the upregulation of CD26 expression in wounded dermis. Pharmacological blockade of CD26, via Sitagliptin, reduced growth of InvEE tumours, while combined inhibition of IL-1a and CD26 delayed tumour onset and reduced tumour incidence. Our results demonstrate that inappropriate activation of MEK1 in the epidermis leads to changes in dermal fibroblasts that, like the skin inflammatory infiltrate, contribute to tumour formation. Oncogene (2012) 31, 992–1000; doi:10.1038/onc.2011.298; published online 18 July 2011 Keywords: epidermis; MEK1; IL-1a; CD26; dipeptidyl peptidase-4; Sitagliptin

Correspondence: Professor FM Watt, Epithelial Biology, Cancer Research UK Cambridge Research Institute, Li Ka Shing Centre, Robinson Way, Cambridge, Cambridgeshire CB2 0RE, UK. E-mail: [email protected] 3 These authors are joint first authors. Received 20 January 2011; revised and accepted 6 June 2011; published online 18 July 2011

Introduction There is considerable evidence to support the concept that tumours are organs, their function and properties depending not solely on the cancer cells bearing oncogenic mutations, but also on the neighbouring cells with which they communicate (Egeblad et al., 2010). Even within a single cell type, reciprocal signalling between different cell compartments can profoundly influence tumour initiation and progression. This is evident in the interfollicular epidermis, where suprabasal, differentiating cells can positively or negatively regulate proliferation of stem cells in the underlying basal layer (Owens and Watt, 2003; Janes and Watt, 2006). One way in which epidermal cell communication is disturbed is via suprabasal expression of integrin extracellular matrix receptors. Integrin expression is normally confined to the basal cell layer; however, in many squamous cell carcinomas (SCCs) integrins are also expressed by suprabasal cells (Owens and Watt, 2003). Transgenic mice in which integrins are expressed above the basal layer under the control of the involucrin promoter can show increased susceptibility to chemical carcinogenesis (Owens and Watt, 2003; Janes and Watt, 2006). Suprabasal b1 integrins signal through Erk MAPK (Owens and Watt, 2003). Transgenic mice that express constitutively active MEK1 in the suprabasal epidermal layers under the control of the involucrin promoter (InvEE transgenics) have hyperproliferative epidermis and a chronic inflammatory infiltrate. In addition, InvEE mice develop papillomas and keratoacanthomas on wounding (Hobbs et al., 2004; Arwert et al., 2010). We have recently shown that differentiating cells that express activated MEK1 stimulate cells in the underlying basal layer to divide and form the proliferative compartment of the tumour. In addition, they express high levels of IL-1a, which contributes to the inflammatory infiltrate (Arwert et al., 2010). In the present study, we have identified a new mechanism by which MEK1-expressing keratinocytes promote tumour growth, involving upregulation of CD26, also known as dipeptidyl peptidase-4 (DPPIV/DPP4). Results Elevated CD26 levels and activity in InvEE epidermis To identify cytokines and immune modulators that were differentially expressed in InvEE and wild-type

Upregulation of CD26 expression EN Arwert et al

993

(WT; transgene-negative InvEE littermates) epidermis, we screened a protein-based array, consisting of 96 cytokines and immune response-related antibodies spotted on a membrane, with pooled WT (n ¼ 3) and InvEE (n ¼ 3) epidermal lysates. One protein was significantly upregulated in InvEE epidermis: CD26 (Figure 1a). CD26/DPPIV is a dipeptidyl peptidase expressed on the surface of several cell types, including epithelial cells, fibroblasts, endothelial cells and activated immune cells (Ohnuma et al., 2008; Takasawa et al., 2010). CD26 is primarily associated with immune regulation, via enzymatic cleavage of cytokines, and with T cell activation (Ohnuma et al., 2008). Western blotting of epidermal lysates confirmed that CD26 was more abundant in InvEE than WT skin (Figure 1b). WT epidermal CD26 ran with an apparent molecular mass of 105 kDa, similar to that described in other cell types (Vivier et al., 1991). In contrast, the major CD26 band in InvEE epidermis ran at B90 kDa (Figure 1b). As CD26 is N-glycosylated (Fan et al., 1997), we investigated whether the difference in mobility of WT and InvEE epidermal CD26 was due to differences in N-glycosylation. Protein lysates were digested with PNGaseF and EndoH. PNGaseF cleaves all N-glycans, while EndoH specifically cleaves those N-glycans that have not been modified in the Golgi (Maley et al., 1989) (Figure 1c). Treatment with EndoH did not change CD26 mobility, indicating that the lower molecular mass of InvEE CD26 did not reflect incomplete maturation. However, treatment with PNGaseF reduced the mobility of both the 105- and 90-kDa bands to B80 kDa. Therefore, the difference in molecular mass between WT and InvEE epidermal CD26 reflected differential glycosylation of the mature protein. WT

To explore whether there were differences in overall glycosylation in InvEE and WT epidermis, we probed western blots of epidermal lysates with wheat germ agglutinin (WGA), which recognises sialic acid and N-acetylglucosaminyl sugar residues. InvEE and WT epidermal lysates had similar profiles of glycosylated proteins, but the levels were higher in InvEE lysates (Figure 1d). When skin sections were stained with WGA–fluorescein isothiocyanate (FITC), we noted that binding was predominantly to the suprabasal, differentiating epidermal layers (Figure 1e). As there are more suprabasal layers in InvEE than WT epidermis, this could account for the increased level of glycosylated proteins. We next used a colorimetric assay to determine whether CD26 activity was upregulated in InvEE epidermis (Figure 1f). The specificity of the assay for CD26 was established by treating lysates with the CD26 inhibitors Sitagliptin (SG) or Ile–Pro–Ile at 10 IC50 and subtracting residual activity from the values presented. Our results indicate that although CD26 was differentially glycosylated in InvEE and WT epidermis, both forms were active and there was more CD26 activity in InvEE than WT epidermis (Figure 1f). Therefore, both CD26 activity and CD26 protein were elevated in InvEE epidermis. Our results are consistent with a previous report that differential glycosylation does not affect the enzyme activity of CD26 (Aertgeerts et al., 2004). Biological fluids, such as serum, contain relatively high levels of soluble CD26 (sCD26), but the physiological role of sCD26 and its mode of release from the cell surface are incompletely characterised (Cordero et al., 2009). There was no difference in CD26 serum activity

InvEE

WT 150 kDa 102 kDa 76 kDa

InvEE

150 kDa

WT InvEE CD26

102 kDa 76 kDa

CD26

GAPDH GAPDH

-

+

-

-

+

-

-

-

+

-

-

+

InvEE

WT

WGA-FITC Actin

CD26 activity - serum (nmol/hour/μl serum)

WGA lectin/ DAPI

(nmol/hour/mg protein)

CD26 activity - Lysate

CD26 levels - serum x1000 pg/ml serum)

WT

EndoH PNGaseF

WGA lectin/ DAPI InvEE

Figure 1 CD26 is upregulated in InvEE epidermis. (a) Cytokine protein array (Ray Bioscience) incubated with WT (left) and InvEE (right) epidermal protein lysates in NP-40 buffer. Each lysate was pooled from skin of three male mice. Cytokines were spotted in duplicate. Arrows: CD26 spots. Positive controls (green box), negative controls (red box) and blank spots (black box) are indicated. (b, c) Western blots of NP-40 protein lysates probed for CD26 or GAPDH. In (c), lysates were prepared±endoglycosidase EndoH or PNGaseF. (d) Epidermal protein lysates probed with WGA–FITC or anti-b-actin (loading control). Molecular mass markers: 52, 76, 102, 150 kDa. (e) Skin sections labelled with WGA–FITC (green) and DAPI (blue). Dashed lines indicate epidermal-dermal boundary. Scale bars: 50 mm. (f, g) Chromogenic activity assays for CD26 in epidermal lysates (f) and serum (g). Aminopeptidase activity in the presence of 100 nM SG (10 IC50) was subtracted from baseline so that activity presented is specific for CD26. (h) CD26 protein levels in serum. (f–h) Data are means±s.e. Oncogene


994

between WT (n ¼ 6) and tumour-free InvEE (n ¼ 5) mice, but activity was significantly lower in tumour-bearing InvEE animals (n ¼ 5, Po0.05; Figure 1g). The serum concentration of sCD26 was lower in both tumour-free and tumour-bearing InvEE mice compared with WT controls (Po0.05; Figure 1h). We conclude that the elevated levels of CD26 in InvEE epidermis (Figures 1a–c) do not result in increased levels of serum CD26. CD26 expression by differentiated epidermal cells To determine where CD26 was expressed, we performed immunolocalisation studies (Figure 2). Compared with WT epidermis, CD26 protein was upregulated in the differentiated layers of InvEE epidermis, at cell–cell borders (Figures 2a and b). Consistent with this, in stratified cultures of InvEE keratinocytes, CD26 was primarily expressed by involucrin-positive, suprabasal cells (Figure 2c). Whereas CD26 activity was higher in InvEE than in WT epidermis (Figure 1f), CD26 activity was similar in WT and InvEE cultured keratinocytes (Figure 2d). This led us to hypothesise that the increased level of CD26 in InvEE epidermis reflected the expansion of the suprabasal layers (Figures 2a and b), as in culture InvEE and WT keratinocytes stratify to the same extent (Figure 2c and data not shown). To test this, we increased the level of calcium ions in the culture medium, which induces stratification and assembly of intercellular adhesive junctions (Watt, 1984). As shown in Figure 2d, increasing the level of calcium ions increased CD26 activity in both InvEE and WT keratinocytes. Although IL-1a is known to have a role in InvEE tumour formation and is reported to increase CD26 activity in some cell types

CD26

(Nemoto et al., 1999; Arwert et al., 2010), IL-1a had no effect on CD26 activity in keratinocytes (Figure 2d). In addition, we did not detect sCD26 in keratinocyte culture medium (data not shown). CD26 expression by dermal fibroblasts CD26 was highly expressed in the dermis of WT and InvEE skin (Figures 2a and b). When InvEE and WT animals were wounded, CD26 expression in the epidermis was unaffected, but there was a marked accumulation of CD26 in the wound bed, particularly in InvEE wounds (day 10 after wounding, n ¼ 5 WT, 5 InvEE, Figures 3a and b). On the other hand, CD26 was undetectable in the stroma of wound-induced InvEE papillomas, except for a few small clusters of cells (n ¼ 5 tumours; Figure 3c). CD26 was also absent from the stroma of chemically induced papillomas and SCCs in WT mice (Figure 3d and e; n ¼ 5). The change of CD26 expression between normal dermis and tumour stroma was also observed when we compared human skin and human SCCs (Figures 3f and g; n ¼ 4). In all types of tumour, CD26 accumulated at cell–cell borders in the epithelial compartment (Figures 3c–e, g). To identify the dermal cell type that expressed CD26, we performed flow cytometry on isolated dermal cells using anti-CD26 and PDGFRa, a cell-surface marker of dermal fibroblasts (Karlsson et al., 1999). Nearly all CD26-positive dermal cells expressed PDGFRa (Figure 4a), and approximately one-third of PDGFRapositive cells were CD26 positive. When murine dermal fibroblasts were treated with IL-1a, CD26 activity was stimulated (Figure 4b), consistent with an earlier study

CD26 DAPI

e d

CD26

CD26 DAPI

e d

CD26 DAPI Involucrin

WT

InvEE

InvEE

7 Protease activity (nmol/hour/50.000 cells)

WT

6 5 4 3 2 1 0

-

+ WT

+

+ +

-

+ -

+

+ +

Ca IL-1a

InvEE

Figure 2 CD26 is upregulated in differentiating keratinocytes. (a, b) WT (a) and InvEE skin (b) labelled with anti-CD26 (red) and DAPI nuclear counterstain (blue). (e) epidermis; (d) dermis; dashed lines: epidermal–dermal boundary. (c) CD26 and involucrin expression in stratified keratinocytes. Extended focus of Z-stack in XY plane showing DAPI-stained nuclei (white), CD26 (green) and involucrin (magenta). Red line (left hand panel) indicates plane of XZ projection (right hand panels). Arrows: basal and suprabasal cell nuclei. (d) CD26 enzymatic activity of cultured WT and InvEE mouse keratinocytes±1.8 mM calcium ions and 10 ng/ml IL-1a (mean±s.d.; n ¼ 3). Keratinocytes were transferred from low-calcium medium to 1.8 mM of CaCl2 and/or treated with 10 ng/ml IL-1a for 3–5 days. Scale bars: 50 mm. Oncogene


995

e

e

d

d

WT DAPI

d e

e InvEE papilloma

InvEE CD26

E-cad

DAPI

CD26

E-cad

DMBA/TPA pap DAPI

e

DMBA/TPA SCC

CD26

E-cad

DAPI

CD26

E-cad d

e

d e

d Normal human skin

e

e Human SCC

Figure 3 CD26 expression in wounded skin and tumours. (a–c) CD26 (green) expression in the back skin 10 days after wounding (a, b) and in InvEE wound-induced papilloma (c). (e) epidermis; (d) dermis; dashed lines: epidermal–dermal boundary. Arrows: wound edge. (d, e) CD26 (green) and E-cadherin (red) immunolabelling of chemically induced papilloma (d) and SCC (e) in WT mouse with DAPI nuclear counterstain (blue). (f, g) CD26 (green) and E-cadherin (red) immunolabelling of human skin (f) and SCC (g) with DAPI nuclear counterstain (blue). (d–g) Individual staining is shown in grey scale; merge in colour. Dashed lines: boundary between epithelium (e) and dermis (d). Scale bars: 75 mm.

(Nemoto et al., 1999) and in contrast to the lack of an effect of the cytokine on epidermal CD26 activity (Figure 2d). Although in some contexts CD26 is expressed by endothelial cells (Takasawa et al., 2010), CD26 was not expressed by CD31-positive blood vessels in WT and InvEE skin or InvEE tumour stroma (Figure 4c). In WT and InvEE skin, most bone marrowderived (CD45-positive) cells lacked CD26; however, in InvEE tumours the few CD26-positive cells in the stroma were CD45 positive (Figure 4d). CD26 inhibition reduces tumour growth To test the role of CD26 in wound-induced InvEE tumour formation, we inhibited CD26 activity with SG,

a drug that is used to treat diabetes (Deacon, 2007). In a pilot study, SG was administered orally to WT mice and SG levels in the skin were measured by liquid chromatography–mass spectrometry (Jackson and Mi, 2008). SG reached concentrations well above the IC50 of 18 nM (n ¼ 5, mean 38.62 nM, s.e.m. 13.29) (data not shown; Kim et al., 2005). The dose-dependent inhibitory effect of SG in epidermal lysates is shown in Figure 5a. Directly after wounding, one group of InvEE mice was fed mash containing SG (n ¼ 30) and the other mash without SG (n ¼ 30). The mash was made fresh every day and given for 30 days. There was no effect on the number of tumours that formed, nor on the kinetics of induction (Figure 5b). Although treatment with SG did not delay tumour formation, it did affect tumour Oncogene


996 104 1.95

25.9

45.3 101

0 PDGFRα

102

0 II1 α

100

2

103

N o

CD26

101

* 4

IL 1α

102

CD26 activity on dermal fibroblasts 6

5d

103

protein activity (nmol/hr/40000 cells)

26.9

CD26CD31DAPI

CD26CD31DAPI

CD26CD31DAPI

CD45CD26DAPI

CD45CD26DAPI

CD45CD26DAPI

*

Figure 4 CD26 expression and activity in fibroblasts. (a) Flow cytometry of dermal cells. Cells were isolated from postnatal day 2 skin and stained using DAPI (live/dead marker), CD26-PerCP-Cy5.5 and PDGFRa-PE. (b) CD26 activity of confluent dermal fibroblasts treated for 5 days with IL-1a or vehicle (means±s.e.; *Po0.05). (c, d) Immunolabelling for CD26 and CD31 (c) or CD45 (d) with DAPI nuclear counterstain. Dashed lines: boundary between epidermis and dermis. Asterisk indicates hair follicle. Arrowheads in (c) indicate CD26-negative blood vessels. Arrowheads in (d) indicate co-expression of CD26 and CD45. Scale bars: 75 mm.

growth. When mice received SG, a reduction in tumour size was observed 30 days after SG treatment had stopped (Figure 5c). We have previously shown that treatment with the IL-1 inhibitor Kineret delays the onset of woundinduced tumours (Arwert et al., 2010). We therefore investigated the effects of combined treatment with SG and Kineret. The combination of SG and Kineret both delayed tumour onset and reduced tumour incidence (Figure 5d). The effect of SG on tumour growth (Figure 5c) was not correlated with a decrease in tumour angiogenesis or proliferation. When wound-induced tumours were either left untreated or treated with SG for 21 days, there was no difference in CD31 (Figure 5e) or Ki67 labelling (Figure 5f). We also saw no differences in CD31 and Ki67 labelling between the treated and untreated tumours in the experiment shown in Figure 5d (data not shown). Oncogene

Discussion Studies of CD26 have demonstrated or suggested functions in areas as diverse as tumour biology and malignancy, T cell activation, cell migration, HIV infection and diabetes (Iwata and Morimoto, 1999; Ohnuma et al., 2008). We now report that CD26 is upregulated in the epidermis of InvEE mice and has a role in the growth of wound-induced tumours. Unwounded InvEE epidermis shares characteristics with psoriatic human skin, including epidermal hyperproliferation, perturbed epidermal differentiation and a chronic inflammatory infiltrate (Hobbs et al., 2004). It is thus striking that CD26 is upregulated in human psoriatic plaques and is linked to inflammatory infiltrates in neoplastic skin (Novelli et al., 1996). CD26 is proposed to function as a keratinocyte activation antigen (Novelli et al., 1996). Moreover, CD26 inhibitors, currently used to manage type-2 diabetes, are being


997 Sitagliptin (CD26 inhibitor) 60% 225

Mice with tumors

Protein activity (nmol/hour/mg protein)

275

175 125

50%

Sitagliptin

40%

Control

30% 20% 10% 0%

75 0

50

0 6 11 14 18 20 29 34 42 48 53 60 treatment phase Days after wounding

100

Concentration sitagliptin (nM) Tumor size + Sitagliptin

Sitagliptin + Kineret 60%

GrpA-Sitagliptin GrpB-Control

60

Mice with tumors

Size tumor mm^2

80

40 20 0 19 treatment phase

29

69

Days after wounding

Control CD31

SG CD31

50%

Sitagliptin + Kineret

40%

Control

30% 20% 10% 0% 0

6 11 14 20 25 35 39 42 49 54 60

treatment phase Days after wounding

Control Ki67

SG Ki67

Figure 5 Inhibition of CD26 activity with SG. (a) Epidermal lysates of WT mice were treated with the SG concentrations shown and CD26 activity was measured. Data are from triplicate measurements. (b, c) Effect of SG on wound-induced InvEE tumour formation (b) and tumour size (c). Treatment phase is shown in green. (c) Data are means±s.e. (d) Mice were untreated or treated with SG (treatment phase: green) in combination with Kineret (treatment phase: blue). (e, f) Wound-induced tumours that were either untreated (control) or SG treated for 21 days, labelled with anti-CD31 (e) or Ki67 (f) and counterstained with haematoxylin. Scale bars: 50 mm.

evaluated as treatments for psoriasis (Ansorge et al., 2009). CD26 is expressed in the differentiating epidermal layers and thus the upregulation in InvEE transgenic mice is correlated with the increased number of differentiated cell layers. The observation that raising the level of calcium ions in the culture medium leads to increased CD26 activity in cultured keratinocytes supports this conclusion, as calcium is known to stimulate intercellular adhesion and promote epidermal stratification (Watt, 1984). Elevated levels of CD26 protein correlated with elevated CD26 activity in InvEE epidermal lysates, even though CD26 was differently glycosylated in WT and InvEE epidermis. This is consistent with a previous report that glycosylation does not affect CD26 peptidase activity or dimerisation (Aertgeerts et al., 2004). Altered CD26 glycosylation was correlated with an overall increase in protein glycosylation in InvEE epidermis, again associated with the increased number of differentiated cell layers. CD26 was expressed not only in the epidermis but also in the dermis of both WT and InvEE mice. In the dermis, CD26 was predominantly expressed by fibroblasts and not by blood vessel endothelial cells (Takasawa et al., 2010) or CD45-positive bone marrow-derived cells. CD26 activity in cultured fibroblasts was stimulated by treatment with IL-1a (Nemoto et al.,

1999). This suggests that a likely mechanism of CD26 activation in InvEE dermis is via wound-induced epidermal IL-1a release, as InvEE keratinocytes store high levels of IL-1a (Hobbs et al., 2004; Arwert et al., 2010). Wounding increased CD26 expression in the underlying dermis. However, CD26 expression levels declined in the stroma of InvEE papillomas. The low levels of stromal CD26 were also a feature of chemically induced tumours in WT mice and in human SCCs of the oral cavity. In tumour stroma, CD26 was not expressed by blood vessels, but could be detected in CD45-positive cells, albeit at low frequency. The contrast between CD26 levels in wounded dermis and tumour stroma could potentially reflect a switch from a pro-inflammatory Th1 environment in wounds to an anti-inflammatory/pro-tumor Th2 cytokine profile in papillomas (Willheim et al., 1997). In contrast to the observed downregulation of CD26, FAP-1, a CD26 family member, is expressed exclusively in the tumour stroma (Kraman et al., 2010). When comparing CD26 activity in serum, we found no significant differences between WT and InvEE animals. However, CD26 activity was lowered in InvEE animals bearing tumours. This is consistent with studies in humans with SCCs, rheumatoid arthritis and lupus erythematosus (Cordero et al., 2009). Changes in sCD26 Oncogene


998

levels are often associated with immune status changes: low concentrations of sCD26 are correlated with failure of the immune response (for example, in solid tumours), whereas high concentrations are found in inflammatory and infectious diseases (Cordero et al., 2009). Inhibition of CD26 with SG did not affect the onset of wound-induced tumour formation, but did result in reduced tumour size. We have previously observed that inhibition of IL-1a delays the onset of tumour development (Arwert et al., 2010). We now show that combined treatment with SG and Kineret both delays tumour formation and reduces tumour incidence. We did not obtain evidence that inhibiting CD26 activity decreased keratinocyte proliferation (Reinhold et al., 1998) or inhibited neo-angiogenesis (Kohl et al., 1991). We speculate that by regulating blood glucose levels the drug may affect cell turnover within the tumour mass. Alternatively, SG may affect the tumour stroma by modulating production of cytokines and extracellular matrix remodelling enzymes (Ta et al., 2010). Our findings are likely to be relevant to human tumours because of recent reports that patients being treated for type 2 diabetes with Metformin, another CD26 inhibitor, show a reduced incidence of certain cancers (Lindsay et al., 2005; Landman et al., 2010).

Materials and methods Mice InvEE founder line 3376A mice (Hobbs et al., 2004) were kept as heterozygotes on an F1 (CBA C57BL/6) background. Transgene-negative littermates were used as controls. Animal procedures were subject to Cancer Research UK ethical approval and performed under a UK Government Home Office licence. In order to dose mice with SG, the drug was extracted from tablets in water as previously described (Jackson and Mi, 2008). A solution of 5 mg/ml SG in water was combined with mash and fed to mice ad libitum. Control mice received mash without the drug. Cytokine array Epidermal protein lysates were made by scraping the epidermis from the dermis of frozen mouse skin on dry ice and extracting in NP-40 buffer (50 mM Tris–HCl, 150 mM NaCl, 1% NP40, pH ¼ 8) with complete protease inhibitor cocktail (Roche, Burgess Hill, UK). The cytokine protein array (RayBiotech, Peterborough, UK) was probed with epidermal lysates pooled from skin of three different mice. Cell culture Mouse keratinocytes were isolated as described previously (Romero et al., 1999; Jensen et al., 2010) and grown in lowcalcium complete FAD medium (3 parts of calcium-free Dulbecco’s modified Eagle’s medium, 1 part of calcium-free Ham’s F12 medium (F12), 1.8 104 M adenine), 10% fetal calf serum, 0.5 mg/ml hydrocortisone, 5 mg/ml insulin, 1010 M cholera toxin and 10 ng/ml epidermal growth factor) at 32 1C in 8% CO2. Keratinocytes were maintained on J2-3T3 feeder layers in Collagen-I-coated (0.0392 mg/ml) culture vessels. In some experiments, keratinocytes were transferred to standard Oncogene

complete FAD medium containing 1.8 mM CaCl2 and/or treated with 10 ng/ml IL-1a for 3–5 days. Immunostaining Human tissue was acquired with appropriate ethical approval. Frozen sections of mouse and human tissue embedded in optimal cutting temperature (Tissue-Tek, Sakura Finetek UK Ltd, Thatcham, UK) were air-dried for 30 min and fixed in 4% paraformaldehyde before staining. Cultured keratinocytes were fixed in 4% paraformaldehyde in phosphate-buffered saline and permeabilised in 0.2% Triton X-100 for 10 min at room temperature (RT). Subsequently, cells or slides were incubated for 1 h in blocking buffer (5% donkey serum and 0.5% fish gelatin in PBS) and incubated with primary antibodies for 1 h at RT, washed, incubated with secondary antibodies containing 0.5 mg/ml 4,6-diamidino-2-phenyl indole (DAPI) and mounted in Mowiol (Scientific Laboratory Supplies, Hessle, UK). Antibodies were as follows: goat antiCD26 (R&D Systems, Abingdon, UK; 1:50); rabbit anti-Ki67 (Covance, Crawley, UK; 1:400); rat anti-CD31 (1:500); anti-CD45-FITC (1:200) (both eBioscience, Hatfield, UK); streptavidin-Alexa555 donkey anti-rat-Alexa488; donkey anti-rabbit-Alexa555 and goat anti-mouse-Alexa555 (all Invitrogen, Paisley, UK; 1:300). WGA–FITC was purchased from Sigma-Aldrich (Dorset, UK) and diluted 1:200 for tissue labelling. Western blotting Protein lysates were made in NP-40 buffer. To remove N-glycans, protein extracts were supplemented with 10 protein denaturing buffer (5% SDS, 0.4 M dithiothreitol) and incubated at 100 1C for 10 min. 10 G7 reaction buffer (0.5 M Na2HPO4/NaH2PO4, pH 7.5, 10% NP-40) was added for PNGaseF or 10 G5 reaction buffer (0.5 M sodium citrate, pH 5.5) for EndoH digestion. In all, 500 U of either PNGaseF or EndoH (New England Biolabs, Hitchin, UK) was added and the mixture was incubated at 37 1C for 1 h. Proteins were run on 4–12% pre-cast polyacrylamide gels (Invitrogen), transferred to nitrocellulose membranes (GE Healthcare, Chalfont St Giles, UK), and blocked for 1 h at RT in 5% skimmed milk and 0.1% Tween-20 in PBS. Membranes were incubated with primary and secondary antibodies for 1 h at RT and washed in phosphate-buffered saline containing 0.1% Tween-20. Protein bands were visualised using Lumilight-Plus substrate solution (Roche) and photo-sensitive film (Kodak, Hemel Hempstead, UK). WGA–FITC was visualised using a Typhoon Laser Scanner (GE Healthcare). The following antibodies and lectin were used: goat anti-CD26 (R&D systems; 1:500); rabbit anti-GAPDH (Abcam, Cambridge, UK; 1:2500); donkey anti-rabbit-HRP (1:8000); anti-goat-HRP (1:8000); and WGA–FITC (Sigma-Aldrich, 1:300). CD26 concentration and activity assays To measure cell membrane-associated CD26 activity, 5 104 keratinocytes were seeded per well in 24-well plates and cultured to confluence in complete FAD medium. After removal of feeders, keratinocytes were washed with buffer-1 (0.1 M Tris–HCl, pH 7.4). Buffer-1 containing 1 mM Glycine– Proline–p–Nitroanilide (Gly–Pro–pNA) substrate was added to the samples and incubated at 37 1C for 2 h (Pereira et al., 2003). 1 M acetate buffer (pH 4.4) stopped the reaction. Samples were centrifuged for 10 min and supernatant absorbance was read at 405 nm. The specific CD26 inhibitors SG and Diprotin A (Ile–Pro–Ile, Tocris, Bristol, UK) were used at a concentration of 10 IC50 value to inhibit CD26 activity.


999 CD26 activity was performed on NP-40 lysates of epidermis using the same procedure. To measure CD26 activity in serum, blood was collected via cardiac puncture under general anaesthesia. The blood was left to clot for 2 h at RT, then centrifuged at 5000 g so that the serum could be collected. The CD26 activity assay used 10 ml serum with 90ml of buffer-1, with or without CD26 inhibitor at 10 IC50. In all, 100 ml buffer-1 containing 1 mM Gly–Pro–pNA substrate was added and incubated at 37 1C for 2 h before analysis. To quantify CD26 levels in serum and protein lysates of epidermis and papillomas, an ELISA assay was used (Duoset, R&D systems).

Vigorous pipetting resulted in a single dermal cell suspension. Cells were stained using DAPI (live/dead marker), CD26-PerCPCy5.5 and PDGFRa-PE and then analysed using flow cytometry.

Conflict of interest The authors declare no conflict of interest.

Acknowledgements Flow cytometry A dermal single cell suspension was prepared as described previously (Jensen et al., 2010). Total skin from postnatal day 2 mice was treated with dispase/trypsin for 1 h at 37 1C. Subsequently, the epidermis was removed and discarded, and the dermis was minced and incubated in collagenase-I for at least 1 h at 37 1C.

We are grateful to Donna Michelle Smith for performing SG chromatography. We acknowledge support from Cancer Research UK, the Dutch Cancer Society (to RM), the Wellcome Trust, the MRC, the European Union, Hutchison Whampoa and Cambridge University.

References Aertgeerts K, Ye S, Shi L, Prasad SG, Witmer D, Chi E et al. (2004). N-linked glycosylation of dipeptidyl peptidase IV (CD26): effects on enzyme activity, homodimer formation, and adenosine deaminase binding. Protein Sci 13: 145–154. Ansorge S, Bank U, Heimburg A, Helmuth M, Koch G, Tadje J et al. (2009). Recent insights into the role of dipeptidyl aminopeptidase IV (DPIV) and aminopeptidase N (APN) families in immune functions. Clin Chem Lab Med 47: 253–261. Arwert EN, Lal R, Quist S, Rosewell I, van Rooijen N, Watt FM. (2010). Tumor formation initiated by nondividing epidermal cells via an inflammatory infiltrate. Proc Natl Acad Sci USA 107: 19903–19908. Cordero OJ, Salgado FJ, Nogueira M. (2009). On the origin of serum CD26 and its altered concentration in cancer patients. Cancer Immunol Immunother 58: 1723–1747. Deacon CF. (2007). Dipeptidyl peptidase 4 inhibition with sitagliptin: a new therapy for type 2 diabetes. Expert Opin Investig Drugs 16: 533–545. Egeblad M, Nakasone ES, Werb Z. (2010). Tumors as organs: complex tissues that interface with the entire organism. Dev Cell 18: 884–901. Fan H, Meng W, Kilian C, Grams S, Reutter W. (1997). Domainspecific N-glycosylation of the membrane glycoprotein dipeptidylpeptidase IV (CD26) influences its subcellular trafficking, biological stability, enzyme activity and protein folding. Eur J Biochem 246: 243–251. Hobbs RM, Silva-Vargas V, Groves R, Watt FM. (2004). Expression of activated MEK1 in differentiating epidermal cells is sufficient to generate hyperproliferative and inflammatory skin lesions. J Invest Dermatol 123: 503–515. Iwata S, Morimoto C. (1999). CD26/dipeptidyl peptidase IV in context. The different roles of a multifunctional ectoenzyme in malignant transformation. J Exp Med 190: 301–306. Jackson EK, Mi Z. (2008). Sitagliptin augments sympathetic enhancement of the renovascular effects of angiotensin II in genetic hypertension. Hypertension 51: 1637–1642. Janes SM, Watt FM. (2006). New roles for integrins in squamous-cell carcinoma. Nat Rev Cancer 6: 175–183. Jensen KB, Driskell RR, Watt FM. (2010). Assaying proliferation and differentiation capacity of stem cells using disaggregated adult mouse epidermis. Nat Protoc 5: 898–911. Karlsson L, Bondjers C, Betsholtz C. (1999). Roles for PDGF-A and sonic hedgehog in development of mesenchymal components of the hair follicle. Development 126: 2611–2621. Kim D, Wang L, Beconi M, Eiermann GJ, Fisher MH, He H et al. (2005). (2R)-4-oxo-4-[3-(trifluoromethyl)-5,6-dihydro[1,2,4]triazolo [4,3-a]pyrazin -7(8H)-yl]-1-(2,4,5-trifluorophenyl)butan-2-amine: a

potent, orally active dipeptidyl peptidase IV inhibitor for the treatment of type 2 diabetes. J Med Chem 48: 141–151. Kohl A, Volk HD, Buntrock P, Kohl G, Diamantstein T, von Baehr R. (1991). The role of dipeptidylpeptidase IV positive T cells in wound healing and angiogenesis. Agents Actions 32: 125–127. Kraman M, Bambrough PJ, Arnold JN, Roberts EW, Magiera L, Jones JO et al. (2010). Suppression of antitumor immunity by stromal cells expressing fibroblast activation protein-alpha. Science 330: 827–830. Landman GW, Kleefstra N, van Hateren KJ, Groenier KH, Gans RO, Bilo HJ. (2010). Metformin associated with lower cancer mortality in type 2 diabetes (ZODIAC-16). Diabet Care 33: 322–326. Lindsay JR, Duffy NA, McKillop AM, Ardill J, O’Harte FP, Flatt PR et al. (2005). Inhibition of dipeptidyl peptidase IV activity by oral metformin in Type 2 diabetes. Diabet Med 22: 654–657. Maley F, Trimble RB, Tarentino AL, Plummer Jr TH. (1989). Characterization of glycoproteins and their associated oligosaccharides through the use of endoglycosidases. Anal Biochem 180: 195–204. Nemoto E, Sugawara S, Takada H, Shoji S, Horiuch H. (1999). Increase of CD26/dipeptidyl peptidase IV expression on human gingival fibroblasts upon stimulation with cytokines and bacterial components. Infect Immun 67: 6225–6233. Novelli M, Savoia P, Fierro MT, Verrone A, Quaglino P, Bernengo MG. (1996). Keratinocytes express dipeptidyl-peptidase IV (CD26) in benign and malignant skin diseases. Br J Dermatol 134: 1052–1056. Ohnuma K, Dang NH, Morimoto C. (2008). Revisiting an old acquaintance: CD26 and its molecular mechanisms in T cell function. Trends Immunol 29: 295–301. Owens DM, Watt FM. (2003). Contribution of stem cells and differentiated cells to epidermal tumours. Nat Rev Cancer 3: 444–451. Pereira DA, Gomes L, El-Cheikh MC, Borojevic R. (2003). Dipeptidyl peptidase IV (CD26) activity in the hematopoietic system: differences between the membrane-anchored and the released enzyme activity. Braz J Med Biol Res 36: 567–578. Reinhold D, Vetter RW, Mnich K, Buhling F, Lendeckel U, Born I et al. (1998). Dipeptidyl peptidase IV (DP IV, CD26) is involved in regulation of DNA synthesis in human keratinocytes. FEBS Lett 428: 100–104. Romero MR, Carroll JM, Watt FM. (1999). Analysis of cultured keratinocytes from a transgenic mouse model of psoriasis: effects of suprabasal integrin expression on keratinocyte adhesion, proliferation and terminal differentiation. Exp Dermatol 8: 53–67. Ta NN, Li Y, Schuyler CA, Lopes-Virella MF, Huang Y. (2010). DPP-4 (CD26) inhibitor alogliptin inhibits TLR4-mediated ERK activation and ERK-dependent MMP-1 expression by U937 histiocytes. Atherosclerosis 213: 429–435. Oncogene


1000 Takasawa W, Ohnuma K, Hatano R, Endo Y, Dang NH, Morimoto C. (2010). Inhibition of dipeptidyl peptidase 4 regulates microvascular endothelial growth induced by inflammatory cytokines. Biochem Biophys Res Commun 401: 7–12. Vivier I, Marguet D, Naquet P, Bonicel J, Black D, Li CX et al. (1991). Evidence that thymocyte-activating molecule is mouse CD26 (dipeptidyl peptidase IV). J Immunol 147: 447–454.

Oncogene

Watt FM. (1984). Selective migration of terminally differentiating cells from the basal layer of cultured human epidermis. J Cell Biol 98: 16–21. Willheim M, Ebner C, Baier K, Kern W, Schrattbauer K, Thien R et al. (1997). Cell surface characterization of T lymphocytes and allergen-specific T cell clones: correlation of CD26 expression with T(H1) subsets. J Allergy Clin Immunol 100: 348–355.