Background
Pancreatic cancer remains one of the most lethal human cancers and causes > 30,000 deaths per year in the United States [
1]. Conventional treatments, such as surgery, radiation, and chemotherapy, or combination of these approaches, have had little impact on patient survival [
1]. Over the past decade, improved understanding and knowledge of the immune system have generated novel strategies for immunotherapy [
2]. While the success of tumor immunotherapy primarily relies on the identification of tumor antigens, the expression of transformation-associated stress genes commonly provokes innate immune reactions. These responses could be exploited to develop immunotherapeutic approaches to treat cancer [
3].
MICA is a glycosylated, polymorphic and membrane-anchored non-classical MHC class I molecule [
4,
5]. The structure of MICA resembles other MHC class I heavy chains. However, MICA is not associated with β2 microglobulin, lacks a CD8 binding site and does not present any antigens [
4,
5]. MICA is broadly expressed in a variety of malignancies, including melanoma, breast, colon and hepatocellular cancers [
6‐
8]. MICA can be cleaved by matrix metalloproteinases and ADAM proteinase, and released into the blood stream or tissue culture medium as a soluble molecule (sMICA) [
7,
9‐
11]. MICA functions as a ligand for NKG2D, an important immunoreceptor expressed on NK cells, CD8 and γδ T cells [
5]. The interaction of MICA and NKG2D plays an important role in immune surveillance by both innate immunity and adaptive immunity.
In vitro studies have provided strong evidence that MICA is critical for the susceptibility of target cells to NK cells, CD8 cytotoxic T cells, and γδ T cells [
5]. Antibodies that block the interaction of MICA and NKG2D can inhibit NK and T cell-mediated cytolysis [
5]. Tumor cells stably expressing NKG2D ligands at high level are rejected by CD8 T cells and/or NK cells [
12]. Mice immunized with NKG2/D ligand-transfected tumor cells develop adaptive immunity against re-challenge with the parental tumor cell lines [
13].
Gemcitabine is a first-line chemotherapy drug for pancreatic cancer [
14]. Gemcitabine alone or in combination with 5-fluorouracil (5-FU) or radiation treatment can prolong the survival of pancreatic cancer patients [
14]. We have recently characterized the change of immune cells in pancreatic cancer patients treated with gemcitabine [
15]. Our data suggest that gemcitabine therapy may decrease memory T-cells and promote naive T-cell activation, and that gemcitabine therapy is not immunosuppressive but rather may enhance antitumor immunity induced by tumor vaccine [
15]. Our present study aims at analyzing MICA/B expression in pancreatic tumor tissues and sera, and determining if gemcitabine can stimulate antitumor immunity by inducing MICA/B expression on pancreatic cancer cells.
Materials and methods
Reagents and cell lines
Anti-MICA/B mAbs (Clones 6D4 and SR99) have been previously described [
6,
16‐
18]. Both antibodies can be used in flow cytometric analysis to detect MICA/B cell surface expression. SR99 but not 6D4 can be used in immunohistochemical staining to monitor MICA/B expression in sections of paraffin-embedded tumor blocks. A polyclonal anti-MICA rabbit IgG was purchased from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). CAPAN-1, CAPAN-2, COLO-587, HPAF-II, MiaPaCa, PANC-1, and MPANC-96 cell lines were purchased from the American Type Culture Collection (Manassas, VA). MRO87, originally thought to be a thyroid cancer cell line but later identified as a colon adenocarcinoma cell line, was included as a positive control [
19]. CAPAN-1, CAPAN-2, and HPAF-II cells were grown in complete MEM medium containing 10% FBS, non-essential amino acids, sodium pyruvate, HEPES, penicillin and streptomycilin. COLO-587 and MRO87 cells were grown in complete RPMI 1640 medium containing 10% FBS. PANC-1 cells were grown in complete RPMI 1640 medium containing 10% FBS, non-essential amino acids, sodium pyruvate, HEPES, penicillin and streptomycilin. MPANC-96 and MiaPaCa were grown in complete DMEM medium containing 10% FBS. NK-92 cells were grown in MyeloCult H5100 medium (StemCell Technologies, Vancouver, British Columbia, Canada) supplemented with IL-2 (100 units/ml) (R&D Systems).
The use of specimens from human subjects was approved by the Institutional Review Board of Rush University Medical Center. Paraffin-embedded tissue blocks derived from patients with either primary pancreatic adenocarcinomas and/or metastases were obtained from the Department of Pathology. The stage of the pancreatic adenocarcinomas was classified according to the TNM scheme used by the AJCC. Blood samples were collected from 61 patients with pancreatic ductal adenocarcinoma. Ten of them were treated with gemcitabine (Gemzar®; Eli Lilly, Indianapolis, IN) reconstituted from lyophilized powder in 200-mg and 1000-mg aliquots. Patients were treated at a dose of 1000 mg/m2 weekly by an intravenous infusion in 250 mL over 30 minutes. Patients were treated for 3 weeks. Blood samples were drawn before (day 0) and on day 3, 7, 14, 21 after gemcitabine treatment. Serum samples were prepared, aliquoted, and stored at -80°C until analysis of sMICA using a sandwich ELISA as described below.
Flow cytometric analysis
The monolayers of MRO87 and 7 pancreatic cancer cell lines were harvested when they reached approximately 80% confluence. To determine whether DNA damage stimulated MICA/B expression in pancreatic cancer cells, PANC-1 cells grown in 6-well plates were preincubated with allopurinol (250 μg/ml) for 1 hr, followed by treatment with 5-FU (10 μM), gemcitabine (2 μM) or radiation (40 Gy). PANC-1 cells were also treated with uric acid crystals (200 μg/ml) and analyzed for MICA/B expression. Cell surface MICA/B was stained with an anti-MICA/B mAb (Clone 6D4) and analyzed on live cells, which were gated on the forward scattering, in a FACScan flow cytometer (Becton Dickinson, Palo Alto, CA) as previous described [
20].
Chromium release assay
Pancreatic cancer cell lines were grown in T-25 flasks. Upon 90% confluence, the cell monolayers were washed and treated with Cell Dissociation Solution (Sigma, St. Louis, MO). Single cell suspensions were labeled with 51Cr ( 50 μ Ci per 1 × 106 cells) at 37°C for 1 hr. The cells (5000 per well) were aliquoted in triplicate in a 96-well U-bottom plate in the absence or presence of NK-92 cells with a ratio of effector to target at 25:1, 5:1, or 1:1. Cells were incubated at 37°C in a humidified CO2 incubator for 4 hr. The supernatants were collected, transferred to Ready-Caps, and the radioactivity was measured in a scintillation counter (Becton Dickinson, Palo Alto, CA). To determine whether genotoxic drugs and radiation can sensitize pancreatic cancer cells for NK-92 cell killing, PANC-1 cells grown in 6-well plates were preincubated with 5-FU (10 μM), gemcitabine (2 μM) or radiation (40 Gy) in the absence or presence of allopurinol (250 μg/ml). The experiments were conducted in triplicate and repeated at least twice with similar results.
Immunohistochemical (IHC) analysis
Tissue sections were de-paraffinized with xylene and rehydrated. MICA/B were stained with an anti-MICA/B mAb (SR99) as previously described [
20]. Two investigators (X.X. and P. G.) graded MICA/B expression in a blinded fashion. MICA/B expression was confirmed with a rabbit polyclonal anti-MICA/B IgG with a similar immunohistochemical staining procedure as previously published [
20]. Negative MICA/B expression was graded as no MICA/B signal (-)with weak signal (+) in less than 20% of tumor cells. MICA/B-positive tumors were defined as weak (+), moderate (++) or strong (+++) MICA/B signal in more than 20% of tumor cells.
ELISA
A MICA ELISA kit (IMMATICS Biotechnologies, Tubigen, Germany) was used to detect sMICA in sera following a protocol described previously [
21]. Plates were coated with the AMO1 capture mAb against MICA in PBS. After incubation overnight at 4°C, the plates were blocked with PBS containing 5% bovine serum albumin for 1 hr at 37°C and washed. Standard recombinant sMICA and serum samples (diluted 1:3) were added to the plates and incubated for 2 hrs at 37°C. After 3x wash, detection mAb BAMO3 (IgG2a specific for MICA/B) was added and incubated at 37°C for 2 hrs. The plates were washed and incubated at 37°C with HRP- conjugated anti mouse IgG2a. Color was developed using ABTS (2,2-Azino-bis-(3-ethylbenzenthiazoline-6-sulfnoic acid). Absorbance was measured at 450 nm. A standard curve of the logarithimic relationship between concentration and absorbance was used to calculate the sMICA concentration in serum samples.
Quantification of uric acid
PANC-1 cells were left untreated or treated with 5-FU (10 μM), gemcitabine (2 μM) or irradiated with 40 G in the absence or presence of allopurinol (250 μg/ml) for 24 hr. Cells were collected by trypsinization and counted with Trypan Blue staining in a hemacytometer. The cells were then washed 3 times with PBS and lysed at 5 × 105 cells/100 μl in the buffer containing Tris-HCl, 50 mM, pH 8.0; 2 mM EDTA, 1% Triton X-100 and followed by a 10-second homogenization. The lysates were incubated on ice for 30 min then centrifuged at 2000 g for 15 min. The supernatants were analyzed for uric acid concentration by using a uric acid kit (BioAssay Systems, Hayward, CA).
Statistical analysis
Chi-square tests were performed to analyze the difference in the clinicopathologic parameters of MICA/B-positive and MICA/B-negative pancreatic ductal adenocarcinomas for significant association. Student t test was used to analyze the difference in NK-92 cell-mediated cytotoxicity. Mann-Whitney U test was used to compare serum MICA levels between the control group and pancreatic cancer patients. Paired student t test was used to determine whether serum MICA/B levels were significantly changed before (day 0) and on day 3 after gemcitabine treatment. The survival of pancreatic cancer patients with serum MICA/B activity at a cutoff of 200 pg/ml was determined according to the Kaplan-Meier method, and the difference was evaluated by the log-rank test. A p value less than 0.05 was considered statistically significant. All statistics was conducted by using SigmaStat 3 software (Richmond, CA).
Discussion
It has been well documented that MICA/B expression is increased in several types of malignancy. A very recent study showed that MICA/B expression was detected in 92 of 103 pancreatic ductal adenocarcinomas from a cohort of Chinese patients [
22]. Marten et al. reported that MICA/B was expressed in three pancreatic cancer cell lines, including PANC-1, DNA-G, and PatSci [
21]. Our study confirms the expression of MICA/B in pancreatic adenocarcinomas, showing that 17 of 25 pancreatic adenocarcinomas (68%) were positive for MICA/B expression; and that MICA/B expression was detected in 4 of 7 pancreatic cancer cell lines. The molecular mechanisms of increased MICA/B expression in pancreatic cancer are unknown. Our prior study suggests that MAP kinase activation due to BRAF gene mutation in thyroid cancer may contribute to increased MICA/B expression [
20]. KRAS is mutated in more than 95% of pancreatic cancers [
14]. We speculate that MAP kinase activation due to RAS gene mutation may in part contribute to increased MICA/B expression. Further epigenetic changes may be required for MICA/B overexpression since MICA/B promoter methylation regulates MICA/B gene expression in hepatomas [
23].
Ectopic expression of NKG2D ligands Rae1β or H60 in several murine tumor cell lines leads to the sensitization of these cells to immune cell-mediated cytolysis and tumor rejection [
13]. In contrast to murine NKG2D ligands, tumor-derived soluble MICA/B molecules can induce endocytosis and degradation of NKG2D on both tumor-infiltrating and peripheral-blood lymphocytes from patients with cancer [
7,
24]. Thus, serum MICA/B molecules released from tumor cells act as a negative force to counteract the effect of membrane-bound MICA/B in immune surveillance and sensitization for immune cell killing. Cell surface levels of NKG2D in NK and T cells from cancer patients are decreased, subsequently leading to the loss of cytotoxic activity [
22,
25]. Therefore, MICA/B expression in pancreatic cancer and subsequent release as soluble molecules may have a very intriguing impact in clinical outcome. In this study, we found that elevated serum MICA levels were not associated with a prolonged or shorter survival. In contrast, Duan et al. [
22] recently reported that the mean survival of pancreatic cancer patients with low serum sMICA levels are significantly longer than those with high serum MICA levels, whereas the mean survival of patients with low MICA/B expression in pancreatic tumor tissues are significantly shorter than those with high MICA/B expression in tumor tissues. We have no explanation for this discrepancy. We found that serum sMICA levels in patients with stage IV pancreatic cancer were higher than those with stage I to III cancer, but this was not statistically significant. An earlier study showed that sMICA levels in patients with stage IV pancreatic cancer are significantly higher than those with stage III cancer [
21].
NK-92 cells have a superior effect over lymphokine activated killer cells and other NK cell lines such as the YT cell line [
26], largely because of an abundant expression of perforin and granzyme B [
27]. Romanski et al. [
28] showed that NK-92 cells can selectively kill a panel of leukemia cell lines in a MICA/B-dependent manner. NK-92 cells moderately express the NKG2D receptor and two other NK activation receptors, NKp30 and NKp46 [
27]. NK-92 cells preferentially kill a panel of MICA/B-positive tumor cell lines [
20]. In this study, we demonstrated that NK-92 cells were able to kill two MICA/B positive pancreatic cancer cell lines (PANC-1 & CAPAN-1) but had only minimal effect on other two MICA/B-positive cell lines (HPAF-II and MPANC-96). Marten et al. [
21] found that NK-92 cells are able to kill HLA-ABC-positive PANC-1 cells with similar potency. The resistance of MICA/B-positive pancreatic cancer to NK-92 cell killing could be due to the inhibitory effect of co-expressed HLA-G antigen that suppresses NK cell function [
29‐
31]. Three MICA/B-negative cell lines, CAPAN-2, COLO-587, and MiaPaCa, displayed very low to moderate sensitivity to NK-92 cells. It is likely that other NK activation receptors or other members of MICA/B such as ULBP1-4 and Letal, a recently identified NKG2D ligand [
32,
33], may also participate in NK-92 cell-mediated cytotoxic activity against MICA/B-negative tumor cell lines [
21].
Activation of the DNA damage pathway leads to increased expression of NKG2D ligands in several murine tumor cell lines [
34]. Other non-genotoxic anticancer drugs can also increase MICA/B and other NKG2D ligand expression in various tumor cell lines and in patients [
35‐
37]. Mechanistic studies suggest that increased cell surface MICA/B levels by anticancer drugs could be due to transcriptional up-regulation of
MICA/B gene expression or due to the suppression of proteinases that cleave MICA/B [
35‐
37]. Our prior study showed that MAP kinase activation due to RAS or BRAF oncogene activation contributed to increased MICA/B expression [
20]. Additional study showed that DNA damage induced by genotoxic drugs and radiation leads to uric acid accumulation and MAP kinase activation, subsequently resulting in increased MICA/B expression in MRO87 and HeLa cell lines (Manuscript under review). Our in vitro study showed that uric acid crystals were able to induce MICA/B expression in PANC-1 cells, and that the blockade of uric acid production by allopurinol abrogated DNA damage-induced uric acid production and MICA/B expression. Our clinical study showed that gemcitabine treatment led to a transient increase of serum sMICA levels in 6 of 10 pancreatic cancer patients. These observations collectively suggest that uric acid accumulation in DNA-damaged pancreatic cancer cells plays an important role in mediating genotoxic drug- and radiation-induced MICA/B expression. Consistent with this notion, we were able to detect the expression of xanthine oxidoreductase, a metabolic enzyme that generates uric acid, in PANC-1 cell line and in pancreatic adenocarcinomas (X. Xu, unpublished observations).
Since MICA/B expression in PANC-1 cells plays a critical role in NK-92 cell-mediated cytotoxicity [
21], it is highly likely that increased MICA/B expression contributes significantly to DNA damage-enhanced sensitivity of PANC-1 cells to NK-92 cell killing. Increased MICA/B expression in HeLa and MRO87 cells by uric acid crystals also leads to increased sensitivity to NK-92 cell killing. It should be noted that other NKG2D ligands or Fas can be up-regulated by DNA damage too and may also sensitize tumor cells to NK cell killing. Our clinical study showed that gemcitabine treatment led to a transient increase of serum MICA levels in pancreatic cancer patients. The highest levels of serum sMICA were on day 3 after gemcitabine administration. Serum sMICA declined slightly thereafter, which could be due to tumoricidal effect of gemcitabine or due to the suppression of proteinase gene expression. Since soluble MICA/B can antagonize membrane-bound MICA/B-mediated antitumor immunity, the impact of genotoxic drug- or radiation-induced MICA/B expression in pancreatic tumor cells in a clinical setting remains unknown. Recent clinical trials revealed that pancreatic cancer patients treated with chemotherapy followed by immunization with a GM-CSF-transfected pancreatic cancer cell line developed a strong antitumor immunity and prolonged patient survival [
38,
39]. Combinational use of proteinase inhibitors that block MICA/B cleavage or use of chemotherapeutic drugs that can also suppress proteinase expression may further improve the therapeutic outcome of immunotherapy for pancreactic cancer.
Acknowledgements
The authors would like to thank Dr. Sophie Caillat-Zucman (INSERM U561, Equipe Avenir, Hôpital Saint-Vincent de Paul, Paris) for kindly providing the anti-MICA/B mAb (Clone SR99), Dr. Hans-G. Klingemann (Tufts-New England Medical Center, Boston) for kindly providing the NK-92 cell line. This work was supported in part by a grant from American Cancer Society, IL Division (X. Xu) and by the Eli Lilly Research Laboratories (J. Plate).
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
All authors read and approved the final manuscript.
XX, GR, PG, J, & RAP: Conducted basic, pathological and clinical studies.
XX, VG, TS, HLK, JP, & RAP, conceived and executed the project, writing and/or critically reading the manuscript