Increased PD-L1 Expression in Human Skin Acutely and Chronically Exposed to UV Irradiation

Abstract

Overexpression of PD-L1 (CD274) on tumor cells may represent a hallmark of immune evasion, and overexpression has been documented in several tumors including cutaneous squamous cell carcinoma (cSCC). While PD-L1/PD-1 activity in the skin has been primarily described in inflammatory models, our goal was to examine PD-L1 expression in human keratinocytes exposed to UV irradiation. We assessed PD-L1 expression in human sun-protected (SP) and sun-damaged (SD) skin, actinic keratosis (AK), and cSCC using IHC and protein microarray. Both methods found low baseline levels of PD-L1 in SP and SD skin and significantly increased expression in cSCC. Next, we examined PD-L1 expression in acute models of UV exposure. In human SP skin exposed to 2-3 MED of UV (n = 20), epidermal PD-L1 was induced in 70% of subjects after 24 h (P = 0.0001). SKH-1 mice exposed to acute UV also showed significant epidermal PD-L1 induction at 16, 24 and 48 h. A time- and dose-dependent induction of PD-L1 was confirmed in cultured human keratinocytes after UV, which was markedly reduced in the presence of MEK/ERK, JNK or STAT3 inhibitors. These findings suggest that UV induces upregulation of PD-L1 through established, pharmacologically targetable stress-signaling pathways in keratinocytes.

Figure 1.

Increased PD-L1 in cSCC compared to normal skin and AK. (A, B) IHC staining for PD-L1 in sun-protected skin (n = 20), sun-damaged skin (n = 20), actinic keratosis (AK, n = 25), low risk (n = 40) and high risk (n = 31) cSCC showed minor staining in AK and significantly increased staining in tumor samples (> 5% total tumor section, p < 0.001). (C) Reverse Phase Protein Microarray analysis (RPPA) was performed on matched human Normal Sun Protected Skin (NSPS), Normal Sun Exposed Skin (NSES), AK, or cSCC. RPPA showed significant induction of PD-L1 in cSCC compared to all other groups [p < 0.001; anti-PD-L1 (E1L3N), Cell Signaling, Inc.].

Figure 2.

Acute SSL exposure induces PD-L1 expression in human epidermis. Twenty subjects were exposed to either 2, 2.5 or 3 MED of SSL and biopsies were taken at baseline, 5hr and 24hr post SSL. (A) Representative PD-L1 IHC images after 2 MED SSL from one subject, 20X. (B) Quantification of epidermal PD-L1 IHC staining from 20 subjects (p < 0.0001).

Figure 3.

Acute SSL exposure induces PD-L1 expression in mouse epidermis. (A) IHC staining of mouse cSCC. (B) SKH-1 mice were exposed to an acute dose of SSL (105 kJ/m² UVA/6.4 kJ/m² UVB) and harvested at the indicated times after exposure stopped. FFPE samples were subjected to IHC. (C) Mouse negative control samples for both UV-treated skin and cSCC were run at the same time as those in A and B, but subjected to only the secondary antibody. (D) The acute SSL-induced epidermal PD-L1 staining was quantified using ImagePro Plus software. Error bars shown depict n = 3, standard error of the mean.

Figure 4.

Acute SSL exposure induces PD-L1 expression in human keratinocytes in culture. (A) HaCaT cells were exposed to low (20 kJ/m² UVA, 1.3 kJ/m² UVB), medium (30 kJ/m² UVA, 2 kJ/m² UVB) or high (40 kJ/m² UVA, 2.7 kJ/m² UVB) doses of SSL or mock treated, harvested along a timecourse and used for immunoblotting of PD-L1 (GAPDH loading control, band density quantified using ImageJ). (B) HEKa cells were exposed to high dose SSL or mock treated and harvested at the indicated timepoints. (C) HaCaT cells were pretreated with vehicle, U0126 (MEK inhibitor, 10μM), SP600125 (JNK inhibitor, 5μM) or SH5–07 (STAT3 inhibitor, 3μM) 1 hour prior and immediately after high dose SSL. Cells were harvested at 24 hr for immunoblotting as in (A). Results representative of n ≥ 3.