Cellular and soluble immune checkpoint signaling forms PD-L1 and PD-1 in renal tumor tissue and in blood

Immune checkpoint blockade therapy is a treatment option of various metastatic cancer diseases including renal cell carcinoma (RCC). Approved antibody drugs target the co-inhibitory signaling of Programmed Cell Death Ligand-1 (PD-L1) and its receptor Programmed Cell Death-1 (PD-1). The combined evaluation of PD-L1 and PD-1 at the mRNA and protein levels in tumor tissue with differentiation of tumor and immune cells as well as of soluble forms (sPD-L1) and (sPD-1) in blood is of basic interest in assessing biomarker surrogates. Here, we demonstrate that PD-L1 determined as fraction of stained tumor cells (TPS-score) correlates with PD-L1-mRNA in tumor tissue, reflecting the predominant expression of PD-L1 in tumor cells. Conversely, PD-1 in immune cells of tumor tissue (IC-score) correlated with PD-1-mRNA tissue levels reflecting the typical PD-1 expression in immune cells. Of note, sPD-L1 in blood did not correlate with either the TPS-score of PD-L1 or with PD-L1-mRNA in tumor tissue. sPD-L1 released into the supernatant of cultured RCC cells closely followed the cellular PD-L1 expression as tested by interferon γ (IFNG) induction and siRNA knockdown of PD-L1. Further analysis in patients revealed that sPD-L1 significantly increased in blood following renal tumor resection. In addition, sPD-L1 correlated significantly with inflammation marker C-reactive protein (CRP) and with PD-L1 mRNA level in whole blood. These results indicate that the major source of sPD-L1 in blood may be peripheral blood cells and not primarily tumor tissue PD-L1. Supplementary Information The online version contains supplementary material available at 10.1007/s00262-022-03166-9.


Introduction
Programmed Cell Death (PD-1 or CD279) and Programmed Cell Death Ligand (PD-L1 or CD274) are co-inhibitory signaling components that act between antigen presenting cells and T-cells. PD-L1/PD-1 signaling suppresses major histocompatibility complex (MHC) and T-cell receptor (TCR) mediated antigen recognition [1] and thereby weakens foreign antigen recognition including neo-antigens in tumor cells. Thus, active PD-L1/PD-1 signaling disturbs cancer immune surveillance that arises from tumor infiltrating immune cells in the tumor microenvironment [2,3]. The intercellular co-inhibitory action of PD-L1/PD-1 signaling can be successfully blocked by therapeutically administered antibodies that target PD-1 or PD-L1 [4][5][6]. Initially, these drug therapies were approved for melanoma and later for renal cell cancer [7,8], but analog treatment of further malignancies, particularly, in advanced stages has also being used [9]. For RCC, the therapy response exerted by these drugs is significant; however, these drugs fail in a Précis: Soluble PD-L1 in blood correlates with inflammation marker C-reactive protein and with PD-L1-mRNA determined in whole blood. Soluble PD-L1 is not associated with PD-L1 levels in renal tumors.
considerable portion of patients [10,11]. As a primary drug target, PD-L1/PD-1 expression is investigated as a prognostic and predictive biomarker for disease outcome and therapy response toward PD-L1/PD-1 blockade [12][13][14]. Different parameters of PD-L1/PD-1 expression in tumor and immune cells as well as in blood can be measured in patients. PD-L1 and PD-1 expressed in tumor or immune cells are commonly assessed by immunohistochemistry. PD-L1-mRNA and PD-1-mRNA levels in tumor tissue more closely reflect transcriptional gene regulation. Soluble PD-L1 (sPD-L1) and soluble PD-1 (sPD-1) are detectable in peripheral blood but their cellular origin has not yet been clearly defined. sPD-L1 can be released by the shedding of membranous PD-L1 with ADAM10 and ADAM17 as relevant cleavage proteases [15][16][17]. In addition, alternative mRNA splice variants of the genes for PD-L1 [18,19] and PD-1 [20,21], which can be translated to secreted isoforms, have been identified. So far, exploiting PD-L1/PD-1 signaling as a biomarker for corresponding immune checkpoint blockade has not been conclusive in RCC patients [12,17,22] and the relationship between PD-L1 tumor expression and sPD-L1 in blood is unclear. The function of sPD-L1 and its value as a prognostic or predictive marker have been diversely described in the literature [15,23].
In this study, the expression of PD-L1 and PD-1 in tumor tissue and their soluble forms sPD-L1 and sPD-1 in blood were evaluated in RCC patients. We aimed to verify possible associations between the parameters and to test for biomarker surrogates. In particular, PD-L1 and PD-1 were assessed by immunohistochemistry. PD-L1-mRNA and PD-1-mRNA, including markers of immune signaling, were measured in tissue homogenate. sPD-L1 and sPD-1 levels were measured before and after tumor resection. Significant correlations between some of the tissue parameters were found, but not between tissue and blood levels. Noteworthily, sPD-L1 in blood correlated with the inflammation marker C-reactive protein.

Patients
We prospectively analyzed renal tumor tissues and blood of patients undergoing renal tumor resection (radical nephrectomy or nephron sparing surgery). The patients were admitted from the commuting area of the hospital. The study was approved by the Philipps-University Marburg Ethics Committee (AZ 125/16), and written informed consent to participate in the study was obtained from all patients. An overview of the numbers of analyzed parameters along with statistics is summarized in a schematic diagram (Fig. 1). Characteristics of patients with clinicopathological evaluation of renal tumors are displayed in Table S1. The numbers of parameters vary since complete data sets were not retrieved from all patients' samples. The actual number of parameters are indicated in Fig. 3 and supplementary Tables S2-S5.

Culturing and processing of cell lines
The cell lines were authenticated from the Leibniz Institute DSMZ-German Collection of Microorganisms and Cell Cultures, Germany, and are listed with the DSMZ-No. (CaKi1, ACC731; HepG2, ACC180; Jurkat, ACC282; THP-1, ACC-16). Cells were cultured in accordance with the recommended conditions in RPMI full growth medium supplemented with fetal bovine serum. The experiments were performed with tested mycoplasma-free cells (Minerva Biolabs GmbH, Berlin, Germany). Interleukin 2 (IL-2) (10 ng/ml), interleukin 6 (IL-6) (25 ng/ml), or interferon γ (IFNG) (10 ng/ml) from R&D Systems were added to the cells (24 h). RNA was extracted using the TriFast procedure (Peqlab, Erlangen, Germany). Protein was isolated with RIPA buffer (Cell signaling Technology Europe, Frankfurt a.M., Germany) supplemented with protease inhibitor cocktail (Sigma-Aldrich Chemie, GmbH, Munich, Germany) and phosphatase blocker (Phos-STOP, Roche Diagnostics GmbH, Mannheim, Germany). Extraction procedure was performed according to the manufacturers' protocols. Cell supernatant was centrifuged for 10 min at 500 g and frozen at −20 °C until measurement of sPD-L1 and sPD-1 (see above). Transfection of double-stranded siRNA was performed with synthetic siRNAs sequences (Biomers GmbH, Germany) employing Lipofectamine RNAiMAX (Thermo Scientific Fisher). si-PD-L1 was compared to si-control sequence (supplementary Table S6). Cells were incubated with the respective siRNAs for 48 h and with IFNG in the last 24 h as indicated.

Western blot
Protein samples (40 µg) were analyzed by sodium dodecyl sulfate polyacrylamide gel electrophoresis with subsequent electrotransfer to nitrocellulose membrane (Bio-Rad-System, Germany). The membranes were blocked in TRISbuffered saline with 0.1% tween containing 5% dry milk and the primary antibodies were added and incubated at 4 °C for 24 h. The antibodies were as follows: PD-L1, host rabbit (#13,684, Cell signaling); PD-1, host goat (AF1086, R&D Systems); cytoplasmic β-actin, host mouse (#MAK6019 Linaris GmbH). Then, the respective secondary anti-host antibodies coupled with horseradish peroxidase were added for band detection with enhanced chemiluminescent luciferase kit (Thermo Scientific, Rockford, USA) by an imager system (Fluorchem IS-8900, Alpha Innotech, San Leandro, USA).  Table S7). For patients' samples, the "PD-L1(s)" primer sets and for the cell line samples the "PD-L1" primer sets served for qPCR. The RT-PCR amplicons were characterized by DNA length and melting temperature profile and contaminations were excluded by appropriate controls (data not shown).

Statistical analysis
Correlation analyses between TPS-score, IC-score, different mRNA target levels in tissue and whole blood as well as sPD-L1 and sPD-1 forms were performed by nonparametric Spearman test and linear regression (Pearson) analysis.

Immunohistochemical and mRNA analyses of PD-L1 and PD-1 in renal tumors
At first, we studied PD-L1 and PD-1 expression in renal tumor tissues. Immunohistochemically derived scores indicating the respective protein levels of PD-L1 and PD-1 in the tumor tissue served for correlation analyses with corresponding mRNA tissue levels and blood values of sPD-L1 and sPD-1 (see schematic diagram in Fig. 1). PD-L1 and PD-1 scores were determined along with markers of lymphocytes (CD3) and macrophages (CD68). We assessed the pathological scores that indicate the respective protein signals in immune and tumor cells. For markers expressed in immune cells such as PD-1, CD3, CD68 and PD-L1, the respective IC-scores were evaluated in tumor area (IC-tumor) and in stroma area (IC-stroma) of the tumor tissues. For PD-L1 that is predominantly expressed in tumor cells, we also determined the score in tumor cells (PD-L1-TPS) and a combined score for PD-L1 in tumor and immune cells (PD-L1-CPS). Exemplary figures of immunohistochemistry are displayed (Fig. 2). In the adjacent tumor tissue homogenates, we determined PD-L1-mRNA, PD-1-mRNA, CD3-mRNA and CD68-mRNA levels. Subsequent correlation analyses of the scores and the mRNA levels revealed certain traits of the targets (Table S2) For delineation of PD-L1 expression with cytokine signaling, we also determined levels of JAK2-mRNA, CXCL10-mRNA and CXCR3-mRNA (Table S2 and S3). JAK2 is both mediator and target of IFNG signaling. CXCL10 and CXCR3 represent cytokine ligand/receptor end-targets of IFNG signaling [27][28][29]. Strikingly, PD-L1-mRNA correlated with JAK2-mRNA (r = 0.37; p = 0.011) possibly reflecting its dependence on IFNG signaling. For PD-1-mRNA, we observed several correlations with components enriched in immune cells (CD3-mRNA) (r = 0.37; p = 0.02) or indicating active immune response (JAK2-mRNA) (r = 0.73; Comparing the immunohistochemical scores, significant correlations were observed between PD-L1-TPS and CD3-IC of the tumor (r = 0.37; p = 0.015) and between PD-1-IC and CD3-IC (r = 0.43; p = 0.0032) (Table S4). PD-L1-IC-tumor correlated with PD-L1-IC-stroma (r = 0.54; p = 0.00018) and PD-1-IC-tumor correlated with PD-1-ICstroma (r = 0.53; p = 0.00029) exhibiting some similarities between both compartments.

sPD-L1 and sPD-1 in blood of patients
sPD-L1 and sPD-1 were measured in patients' blood samples, and both sPD-L1 and sPD-1 were detectable at heterogeneous levels in all blood samples. Importantly, sPD-L1 did not correlate with PD-L1-mRNA levels in tumor tissue, nor with TPS-and IC-scores of PD-L1. In addition, sPD-1 did not correlate with PD-1-mRNA in tumor tissue nor with IC-scores of PD-1 and no correlation was found between sPD-1 and sPD-L1 (Table S5).

PD-L1 and PD-1 expression in cell lines
To characterize the regulation of PD-L1 and PD-1 along with sPD-L1 and sPD-1, we investigated cancer (CaKi-1, HEP-G2) and immune cell lines (Jurkat, THP-1). CaKi-1 and HEP-G2 cells displayed typical induction of PD-L1 by IFNG paralleled by release of sPD-L1 into the cell supernatant. In contrast to IFNG, IL-6 did not induce PD-L1 neither in CaKi-1 cells nor in HEP-G2 cells that were employed as principal IL-6 responsive cell line [31] (Fig. 4A-C). Silencing of sPD-L1 by siRNA in CaKi-1 cells reduced sPD-L1 level in supernatant confirming a direct link between cellular PD-L1 expression and the release of sPD-L1 into the supernatant (Fig. 4D). Considering PD-1 expression, it turned out that Jurkat cells displayed detectable levels of PD-1-mRNA whereas THP-1 did not. Accordingly, Jurkat cells revealed immune-reactive PD-1 bands in Western blot that were not detectable in THP-1 cells. IL-2, a possible inducer of PD-1 [32], had no considerable effect on PD-1 levels (Fig. 4E, F). In supernatants of Jurkat cells, sPD-1 levels were slightly above detection threshold (Fig. 4G).

Discussion
We determined PD-L1/PD-1 in resected renal tumor tissues along with the soluble forms sPD-L1/sPD-1 in patient's blood. Subsequent data analyses tested for related biomarker surrogates. The parameters comprised (i) immunohistochemical scores for PD-L1 and PD-1 in immune and tumor cells, (ii) PD-L1-mRNA and PD-1-mRNA concentrations in the adjacent tissue slides, (iii) sPD-L1 and sPD-1 in blood serum, (iv) PD-L1-mRNA and PD-1-mRNA in whole blood RNA and (v) comparison of sPD-L1 and sPD-1 in blood before and after tumor resection.

IHC-scores of PD-L1 and PD-1 versus mRNAs in tumor tissues
Comparing immunohistochemistry and mRNA data (Table S2), we found correlations between PD-L1-TPS and PD-L1-mRNA and between PD-1-IC-score and PD-1-mRNA in the respective adjacent tissue layers. This observation reflects predominant congruent expression of PD-L1-mRNA and PD-L1 protein in tumor cells and between PD-1-mRNA and PD-1 protein in immune cells [33]. For PD-1, we determined correlation of mRNA level with several immune cell specific markers (CD3, CD68) including the cytokine signaling markers CXCL10 and CXCR3 that are engaged in active immune cell responses [27][28][29]34].

PD-L1 expression in tumor tissue and cells in relation to sPD-L1 in blood
We observed that PD-L1 protein and mRNA expression in tumor tissue was not associated with sPD-L1 levels in circulating blood (Table S5). Likewise, in a study of hepatocellular carcinoma (HCC), no link between immunohistochemical staining and sPD-L1 was detected [35]. These observations are not implicitly expected since both RCC and HCC cell lines release sPD-L1 into cell supernatant (Fig. 4B, C). Of note, it is speculated that the amounts of sPD-L1 in blood versus PD-L1 in tumor tissue may be reciprocally interconnected [15]. In particular, active metalloproteases ADAM10/17 can release sPD-L1 from tumor Our experimental tools cannot directly track this process. We found neither positive nor negative correlations of sPD-L1 with PD-L1 tissue expression when considering protein and mRNA levels, or their ratios.

sPD-L1 in relation to inflammation marker CRP and PD-L1-mRNA in blood
The increased levels of sPD-L1 after tumor resection suggest a primarily non-tumor derived origin of sPD-L1 in blood (Fig. 3A). In parallel, CRP was elevated in the post-versus pre-tumor resection samples (Fig. 3C), an anticipated effect due to tissue damage and wound healing processes after the surgical insult [30]. Corresponding analysis revealed that sPD-L1 also correlated with CRP in the pre-tumor resection samples (Fig. 3D). This trait of sPD-L1 has been described in pancreatic cancer patients [35] and HCC patients [36] where dichotomized patients groups with high and low CRP values were compared. Additionally, in patients with pancreatic cancer, an association between sPD-L1 and sPD-1 was revealed [35] that was not noticed in HCC patients [36] and not in our study. Of note, the tumor volume may particularly add to the pre-surgery levels of sPD-L1 levels in blood and may overlap with the inflammatory conditions regulating sPD-L1.
Another point is that we noticed correlation of sPD-L1 in serum with PD-L1 mRNA in whole blood (Fig. 3E). This observation supports peripheral blood cells as a relevant source for sPD-L1. In line, activated mature dendritic cells isolated from peripheral blood mononuclear cells (PBMNs) were shown to release sPD-L1 [16]. Furthermore, transcriptomic RNAseq analysis identified PD-L1 transcripts in sorted peripheral blood cells from healthy persons that were accessible from the GTEx database [19]. In particular, polymorphonuclear neutrophil granulocytes (PMNs), T-lymphocytes, B-lymphocytes and monocytes exhibited considerable levels of PD-L1-mRNAs expressed as full length along with truncated splice variants. IL-6 signaling represents a major pathway for induction of acute phase proteins such as CRP in hepatocytes during inflammation [31,37]. Since sPD-L1 correlated with CRP, we tested for induction of PD-L1 and sPD-L1 through IL-6. Analysis of cultured RCC and HCC cell lines could not demonstrate any effects of IL-6 on PD-L1 and sPD-L1 (Fig. 4). Speculatively, IL-6 may affect sPD-L1 levels by course of events in blood. Possibly, IL-6 triggered inhibition of apoptosis of neutrophil granulocytes [38] may favor sPD-L1 enrichment in blood.
For sPD-1, we confirmed PD-1 mRNA expression in the T-cell derived Jurkat cell line [39,40] and additionally detected soluble sPD-1 in the Jurkat cell supernatant. In contrast to sPD-L1, we did not observe alterations of sPD-1 in response to renal tumor resection.
In summary, soluble PD-L1 in blood correlated with inflammation marker C-reactive protein and with PD-L1-mRNA level in whole blood. Of note, sPD-L1 was not associated with PD-L1 expression in renal tumors.
Funding Open Access funding enabled and organized by Projekt DEAL.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.