Anthracycline Therapy Modifies Immune Checkpoint Signaling in the Heart

Cancer survival rates have increased significantly because of improvements in therapy regimes and novel immunomodulatory drugs. Recently, combination therapies of anthracyclines and immune checkpoint inhibitors (ICIs) have been proposed to maximize neoplastic cell removal. However, it has been speculated that a priori anthracycline exposure may prone the heart vulnerable to increased toxicity from subsequent ICI therapy, such as an anti-programmed cell death protein 1 (PD1) inhibitor. Here, we used a high-dose anthracycline mouse model to characterize the role of the PD1 immune checkpoint signaling pathway in cardiac tissue using flow cytometry and immunostaining. Anthracycline treatment led to decreased heart function, increased concentration of markers of cell death after six days and a change in heart cell population composition with fewer cardiomyocytes. At the same time point, the number of PD1 ligand (PDL1)-positive immune cells and endothelial cells in the heart decreased significantly. The results suggest that PD1/PDL1 signaling is affected after anthracycline treatment, which may contribute to an increased susceptibility to immune-related adverse events of subsequent anti-PD1/PDL1 cancer therapy.


Introduction
New antineoplastic therapies have dramatically improved the long-term survival rates of cancer patients in recent years [1]. Conventional chemotherapeutics target rapidly dividing cells via the inhibition of cell division, e.g., by intercalation to DNA. In contrast, targeted and immune therapies affect inter-and intracellular signaling pathways and immune cells, resulting in lower tumor burden [2]. In addition to higher survival rates, many adverse effects of these therapies have been found, of which cardiotoxic side effects are among the most lethal in this group [3,4].
Conventional anthracycline cancer therapeutics, e.g., Doxorubicin (DOX), belong to a family of antibiotics used to treat solid tumors and hematologic malignancies. They are well known for their acute and chronic cardiotoxic side effects that finally result in cardiotoxic cardiomyopathy and congestive heart failure, potentially limiting their clinical use [5,6]. Mechanisms of anthracycline cardiotoxicity are considered multifactorial, and the global understanding of the underlying mechanisms is still incomplete. Pathomechanisms include the binding of topoisomerase IIβ in cardiomyocytes, the generation of reactive oxygen species due to direct and indirect stress damage to cardiomyocyte mitochondria, impaired ion handling, and disrupted mitochondrial quality control [4,7,8]. In some cases, anthracyclines are applied in combination regimens with novel cancer therapies including ICI. Interestingly, anthracyclines have been found to affect immune checkpoint signaling pathways targeted by these subsequently applied therapeutics [9][10][11][12][13].
The integration of inhibiting and activating signals at the level of T cells determines the effectivity of the cellular immune response against tumor cells [14]. T cells recognize tumorspecific antigens via their T cell receptors, initiating their anti-tumor cell activity [15]. PD1, an inhibitory checkpoint molecule, is responsible for limiting a permanently activated T cell response, thus protecting healthy as well as tumor cells from an overshooting immune response. ICI against PD1 belongs to the family of immune therapies and promotes a distinct lasting antitumor activity in T cells in solid tumors and hematologic malignancies via releasing the brake of the immune system [16,17]. However, ICI therapy inherits the risk of immune-related adverse events, of which myocarditis is the most severe therapylimiting or in many cases fatal form [16,18,19]. Recently, PDL1 on cardiac endothelial cells has been identified as the main mediator of immune crosstalk in a mouse model of anti-PD1 therapy, which promoted myocardial infiltration with inflammatory cells preceding apparent cardiotoxicity [20].
Synergistic therapeutic regimens are used in the treatment of advanced cancer. Here, the sequential combination of conventional and immune therapies in advanced cancer patients can improve patient outcomes, e.g., as a treatment for triple-negative breast cancer [21][22][23][24][25][26]. Due to the still infrequent but expected increasing use of combination therapies, information on clinical outcomes is limited. Nevertheless, PD1-PDL1 interaction plays a crucial role in maintaining cardiac immunological integrity and is involved in various forms of cardiac injury. However, its role in patients pre-treated with anthracycline has not yet been characterized.
The amplifying effect of the combination therapy in tumor tissue may trigger increased susceptibility for ICI-related cardiotoxicity, further increasing anthracycline-induced cardiotoxicity with considerable consequences on cardiovascular morbidity and mortality. Therefore, this study aims to elucidate the response of PD1/PDL1 signaling to anthracyclines exposure.

DOX Treated Mice Show a Profound Effect on Cardiac Function
To study the effects of anthracycline therapy on the PD1/PDL1 pathway, mice were injected with a single high-dose of Doxorubicin (20 mg/kg; 55 mg/m 2 ) after baseline echocardiographic assessment. After 6 days, heart function was measured and cardiac PD1/PDL1 signaling was evaluated ( Figure 1) and compared to baseline.

PDL1 Expression following Anthracycline Treatment
We next investigated cardiac tissue composition and the expression of PD1 and PDL1 on these different cell populations using flow cytometry. We found a decreased number of cardiomyocytes relative to the total number of cells in DOX-treated mice (49.03 ± 4.60% vs. 58.79 ± 6.66%, p = 0.021, Figure 3A). However, the number of immune cells and endothelial cells was consistent in proportion in both groups (immune cells: 4.87 ± 0.57% vs. 4.50 ± 0.79%, p = 0.417; endothelial cells: 7.01 ± 1.16% vs. 6.19 ± 1.01%, p = 0.272, Figure 3A). When quantifying the number of PDL1-positive cells in the same experimental approach, there was only a small, albeit significant, difference between DOX-treated and control animals in terms of fibroblasts (2.81 ± 1.20% vs. 9.61 ± 1.11%, p = 0.003, Figure 3B) and cardiomyocytes (0.88 ± 0.5% vs. 1.73 ± 0.3%, p = 0.031, Figure 3B), mainly due to the rather low expression of PDL1 in these populations. In contrast, immune cells and endothelial cells in the control group showed high expression levels of PDL1, which was lower in DOX-treated animals in both cell types (immune cells: 41.83 ± 10.42% vs. 74.06 ± 6.59%, p = 0.003; endothelial cells: 72.83 ± 4.38% vs. 84.37 ± 2.36%, p = 0.003, Figure 3B). The number of PD1-positive cells in both groups did not change significantly for immune cells, fibroblasts and cardiomyocytes and showed only a small increase for endothelial cells (mean ± SD: 2.52 ± 0.58% vs. 1.05 ± 0.41%, p = 0.003, Figure 3B).

Discussion
In our present study, we found that substantial cell damage is detectable in a high-dose model of DOX treatment after 6 days. This damage is associated with decreased heart function and decreased numbers of cardiomyocytes in the hearts of DOX-treated mice. Furthermore, DOX acts on the PD1/PDL1 pathway by decreasing PDL1 expression in several cardiac cell populations, which is not restricted to a specific region in the myocardium but is diffusely distributed throughout the tissue.
The effects of DOX treatment on the PD1/PDL1 pathway in various cancer types remain elusive. Some reports have found increased expression of PDL1 in cancer cells, while others have reported less PDL1 expression [9,27,28]. There are even fewer sources of DOX affecting the PD1/PDL1 pathway in other tissue cells. Given the recent preclinical and clinical studies, focusing on combination therapies of DOX and PD1-PDL1 inhibitors, especially in triple-negative breast cancer, reliable data for adverse effect risk assessment are urgently needed [26,29,30].
We specifically used a high-dose model of DOX injection to take a first look at PD1 and PDL1 expression on cardiac cells to induce cell damage, as previously shown [31,32]. Decreased heart function in DOX-injected animals, as evidenced by reduced FS and EF, was detectable at day 6 after exposure, similar to other recent studies [33,34]. Acute tissue damage was also apparent by the high amounts of CK and cTnI in the blood of DOX-treated animals compared with the control group. Both factors have been described earlier as markers of DOX-induced cardiac injury [35]. Similar, dramatically reduced levels of circulating blood leucocytes were found in the DOX group [36]. Using flow cytometry, we detected a difference in the relative quantities of fibroblasts and cardiomyocytes, whereas there was no such difference in endothelial and immune cells. The decline in cardiomyocytes is consistent with our findings of cell damage markers CK and cTnI in the blood, which indicate cellular damage to cardiomyocytes, and has been shown extensively by other groups [37][38][39][40]. Of note, in one study, DOX was found to cause endothelial cell death in immune-incompetent mice but not decreased cardiomyocyte numbers, which is reversed in our model, hinting at a substantial involvement of the immune system, possibly due to altered PD1/PDL1 signaling [41]. Therefore, we believe that the established model adequately resembles DOX-induced myocardial injury to further study the effects on the PD1-PDL1 pathway.
Following the model analysis, we checked the number of cells expressing either PD1 or PDL1 distributed in these populations. In the control group animals, similar levels of PD1 and PDL1 were measured in all four cell populations as previously published for the baseline mice [20]. In DOX-treated animals, the number of PD1-expressing endothelial cells changed significantly, although it is debatable whether this is also a biologically relevant change. PD1 expression in previous studies of cancer cell types varied following doxorubicin treatment but was increased in a cardiac injury model [12,42,43]. Therefore, PD1 expression in our model may be mediated by cardiac damage rather than doxorubicin signaling. Of note, no cell population expressed a high amount of PD1, not even immune cells. The sharp decrease in PDL1-positive immune cells and endothelial cells requires further research, as it is of high importance when considering combination therapy with PD1/PDL1 pathway inhibitors, at least for considering possible cardiac side effects. PDL1 expression was already studied following DOX treatment, albeit mostly in cancer cells, with heterogeneous results regarding increased or decreased expression [9,28,44].
A limiting factor in this study is the difference between flow cytometry and Western blot on the one hand and immuno-staining on the other with respect to PDL1 expression. To this point, determining PDL1 expression levels on different cardiac cell populations by Western blotting from sorted populations would be an interesting option for further research. Furthermore, this study shows only the cardiotoxic effects of anthracycline therapy, but not actual worsening by sequential or combined ICI therapy to further substantiate our hypothesis.
To conclude, our study is the first to provide a detailed look at the PD1 and PDL1 expression on cardiac cells following DOX treatment. Whereas PD1 expression was scarcely present on any cell in all animals, PDL1 was significantly less expressed by immune cells and endothelial cells after DOX treatment. Especially, the decreased expression on endothelial cells could lead to an increased immune response when anthracycline therapy is combined with or precedes PD1/PDL1 ICI therapy. However, whether this holds true needs to be tested by such combination therapy in an animal model.

Animals and Ethical Statement
C57BL/6JRJ mice (12 ± 3 weeks of age) were housed in the central animal facility of the University Hospital Essen at a 12 h/12 h day-night-cycle and access to food and water ad libitum before and during the experiment. The experiments were approved beforehand by the ethical committee of the Landesamt für Natur-, Umwelt-und Verbraucherschutz (LANUV).

DOX Treatment Model and Experiment Setup
The cardiac function of mice was measured at baseline (day 0) before the beginning of the experiment. Animals were then randomly assigned to DOX or vehicle treatment group. After baseline cardiac assessment, mice in the DOX treatment group received a single intraperitoneal injection of DOX (20 mg/kg; MedacGmbH, Wedel, Germany) in max 150 µL NaCl (B. Braun Melsungen AG, Melsungen, Germany), whereas mice in the vehicle treatment group received a single injection of 150 µL NaCl only. Following published conversions, this DOX dose equals~55 mg/m 2 [45,46]. Mice were weighed and visually investigated every day to check for worsening overall fitness. On day 6 after injection, mice were again measured for cardiac function. Afterward, animals were killed by exsanguination in deep isoflurane (Baxter GmbH, Unterschleißheim, Germany) narcosis after injection of Ketamine/Xylazine (100/20 mg/kg; Hameln Pharma, Hameln, Germany/Bela Pharm, Vechta, Germany). The heart was extracted after perfusing the animal free of blood with PBS without calcium or magnesium (PBS −Ca/−Mg ; ThermoFisher Scientific, Waltham, MA, USA) including 20 IE/mL of heparin (Leo Pharma GmbH, Neu-Isenburg, Germany). From the extracted heart, a single midventricular slice (~2 mm thickness) was obtained for histology and fixated in 4% PFA at RT overnight, while the rest of the heart was used for flow cytometry.

Cardiac Function and Blood Analysis
Mouse echocardiography was conducted using a Visualsonics Vevo 3100 Imaging system (FUJIFILM Visualsonics, Toronto, ON, Canada). After initial anesthesia with 4 Vol.-% isoflurane in 1 L/min O 2 , mice were placed on a heated plate and anesthesia was maintained with 1.5 Vol.-% isoflurane. A rectal probe was introduced, and heart rate, respiratory rate and body temperature were monitored continuously. After hair removal, images of long and short parasternal axis were acquired in M-mode and B-mode. Measurements were analyzed offline by a blinded investigator using Visualsonics VevoLab 3.2.6 software (FUJIFILM Visualsonics, Toronto, ON, Canada). Fractional shortening was determined by comparison of midventricular short-axis enddiastolic and endsystolic volumes. The ejection fraction was calculated using LV tracing on the parasternal short axis [47,48]. CK and cTnI levels were measured from serum samples using high-sensitive ELISA assays (both Siemens, Erlangen, Germany).

Immuno-Histology
After overnight fixation, obtained midventricular slices were dehydrated in an ascending ethanol series (50%, 70%, 100% v/v in ddH 2 O, Roth, Karlsruhe, Germany) at RT for 1 h each, followed by two changes of isopropyl alcohol (Roth, Karlsruhe, Germany) and one change of xylene (Roth, Karlsruhe, Germany) at RT for 1 h each, before overnight incubation in one additional change of xylene (Roth, Karlsruhe, Germany). Afterward, samples were embedded in paraffin and 5 µm histological slices were obtained using a microtome. Three histological slices per sample were used for staining and an additional three slices from one sample were used as a secondary antibody control. Samples were deparaffinized in an oven at 60 • C for 20 min and rehydrated in two changes of xylene (Roth, Karlsruhe, Germany), and a descending ethanol series (100%, 100%, 96%, 96%, 70%, 0% v/v in ddH 2 O, Roth, Karlsruhe, Germany) for 3 min each, followed by an antigen retrieval procedure in citrate buffer (Sigma-Aldrich, St. Louis, MO, USA) at 95 • C for 30 min. Samples were allowed to cool to RT for 20 min in the citrate buffer. Samples were washed in three changes of tap water, two changes of ddH 2 O and one change of tris-buffered saline, containing 10 mM TRIS (Roth, Karlsruhe, Germany) and 100 mM NaCl (Roth, Karlsruhe, Germany) in ddH 2 O, with 0.1% (v/v) Tween-20 (Roth, Karlsruhe, Germany; TBS-T) for 5 min each. Blocking was conducted in TBS-T + 5% normal goat serum (Invitrogen, Waltham, MA, USA; NGS) at RT for 1 h. Samples were briefly rinsed with TBS-T and then incubated with primary antibodies for CD31 (rat, 1:20, Dianova, Hamburg, Germany) and PDL1 (rabbit, 1:50, Invitrogen, Waltham, MA, USA) in TBS-T + 5% NGS at 4 • C overnight. The secondary antibody control was incubated in TBS-T + 5% NGS only. Samples were again briefly rinsed with TBS-T and then washed thrice in TBS-T for 5 min each. Staining with secondary antibodies for rat (goat, 1:200, Cy3, Invitrogen, Waltham, MA, USA) and rabbit (goat, 1:200, AlexaFluor 680, Invitrogen, Waltham, MA, USA) was conducted in TBS-T + 5% NGS at RT for 1 h. Afterward, samples were rinsed with TBS-T and then washed thrice in TBS-T for 5 min each. Nuclear staining was conducted with DAPI (Invitrogen, Waltham, MA, USA) 1:5000 in TBS-T at RT for 5 min, followed by washing samples in TBS-T thrice for 5 min each. Samples were preserved using Prolong Gold Antifade (Invitrogen, Waltham, MA, USA) before and after imaging. Two images from brightfield (BF), DAPI, CD31-Cy3 and PDL1-AlexaFluor 680 of each slice (three slices per animal, resulting in six images per animal) were obtained using an EVOS microscope (ThermoFisher Scientific, Waltham, MA, USA) using the transmitted light setting and the DAPI, RFP and Cy5 EVOS light cubes at 20× magnification. Tissue outlines were traced and fluorescence intensity of CD31-Cy3 and PDL1-AlexaFluor 680 were measured using ImageJ 1.53 software (NIH, New York City, NY, USA) [49]. Similarly, DAPI and CD31-Cy3 high signals were traced and PDL1-AlexaFluor 680 intensity was measured in these areas using ImageJ software [49]. Signal intensities were normalized to the mean of CTRL group values. Data Availability Statement: The raw data for the findings provided in the publication are available from the corresponding author upon reasonable request.