Senescent Tumor Cells Are Frequently Present at the Invasion Front: Implications for Improving Disease Control in Patients with Locally Advanced Prostate Cancer

Malignant transformation and tumor progression are processes in which tumor cells acquire new properties that allow them to overcome various selection barriers. Early stages of malignant progression are characterized by evasion of apoptosis, enhanced survival, and increased proliferation. In later stages, tumor cells become increasingly able to invade the surrounding tissue and ultimately colonize distant metastatic niches [1].

In prostate cancer, the most common non-cutaneous malignancy in men [2], the molecular underpinnings leading to tumor progression, metastatic dissemination, and castration resistance are increasingly well understood [[3], [4]]. This is, e.g., highlighted by the finding that DNA damage repair gene defects play a more prominent role in metastatic prostate cancer than previously thought [[5]–[7]].

However, surprisingly little is known about the functional state in which prostate cancer cells engage in local tissue invasion. This information is of critical importance since the difference between cure and prostate cancer recurrence is in many patients defined by the extent and the anatomic site of local tumor invasion. The latter notion is highlighted by the fact that patients with locally advanced prostate cancer and seminal vesicle invasion (SVI) have been reported to have a more unfavorable prognosis than patients with extraprostatic extension (EPE) alone without SVI [[8]–[15]].

Based on a number of previous studies, it can be envisioned that tumor cell invasion involves a more aggressive subpopulation of cells with enhanced proliferative activity [[16]–[18]]. However, there are a number of notable exceptions from this idea [[19]–[23]]. In fact, it has been suggested that a functional switch from a proliferative to an invasive state may be required implying that invading tumor cells have temporarily or permanently withdrawn from the cell division cycle [24]. In the present proof-of-concept study, we report the surprising finding that prostate cancer cells in locally advanced tumors that invade the surrounding tissue are frequently withdrawn from the cell division cycle and show signs of cellular senescence, in particular in the area of SVI.

The prognostic impact of local tumor progression was analyzed in 1,003 men who underwent radical prostatectomy between 1993 and 2015 at the Department of Urology of the University Hospital Heidelberg. All patients were diagnosed with locally advanced prostate cancer and lacked lymph node or distant metastases (pT3, pN0, cM0; online suppl. Table 1; for all online suppl. material, see www.karger.com/doi/10.1159/000530430). Patients were subdivided into stage pT3a (EPE alone, n = 780) or pT3b (with SVI, n = 223; online suppl. Table 2). The time to biochemical recurrence was defined as the time from radical prostatectomy to the first confirmed rise of the serum prostate-specific antigen level >0.2 ng/mL. The time to clinical recurrence was defined as the time from radical prostatectomy to the first clinical signs of local recurrence or distant metastasis. All patients provided written informed consent for the use of their tissue and data for research and publication, and the analysis was approved by the Ethics Committee of the Medical Faculty of the University of Heidelberg School of Medicine (206/2005, 207/2005, S-864/2019, S-287/2022).

A double-immunohistochemical staining of the tumor invasion zone and perineural invasion (PNI) was performed in 21 prostate cancers stages pT3a (n = 6), pT3b (n = 12), or pT4 (n = 3). In addition, in 18 prostate cancers with SVI, the intraprostatic tumor periphery and tumor center as well as the area of SVI were analyzed in a detailed zonal fashion. All tissue samples were provided by the Tissue Bank of the National Center for Tumor Diseases Heidelberg in accordance with the regulations of the tissue bank and the approval of the Ethics Committee of the Medical Faculty of the University of Heidelberg School of Medicine (206/2005, 207/2005, S-864/2019). Written informed consent was provided from all patients for the use of their tissue for research and publication.

Immunohistochemical staining was performed using the CINtec^® PLUS Kit (Roche mtm Laboratories, Heidelberg, Germany) for the simultaneous detection of p16^INK4A and Ki-67, according to the manufacturer’s instructions with some minor modifications as described previously [25]. Briefly, specimens were deparaffinized using xylene and afterward rehydrated by a graded ethanol series. After heat-induced (95–99°C) epitope retrieval in the supplied solution, incubation with the primary antibodies was performed for 30 min. Two primary antibodies were used for this procedure, a monoclonal mouse antibody against p16^INK4a (clone E6H4™) and a monoclonal rabbit antibody against Ki-67 (clone 274-11 AC3). Secondary antibodies were conjugated with horseradish peroxidase or alkaline phosphatase, and the chromogens 3,3-diaminobenzidine (for p16^INK4A) or Fast-Red (for Ki-67) were used. Slides were then counterstained with hematoxylin (Dako, Glostrup, Denmark) and mounted with the CINtec^® PLUS Mount. An immunohistochemical staining for β-galactosidase (anti-GLB1, 1:20 dilution; Abcam, Cambridge, UK) was performed as described previously [26]. The stained tissue specimens were examined using a Leica (Wetzlar, Germany) DM5000B microscope equipped with a Leica K3C camera. For image processing, the Leica Application Suite (version 4.7) was used.

LNCaP cells were obtained from LGC (Teddington, UK) and were maintained in RPMI 1640 (Life Technologies, Darmstadt, Germany) supplemented with 0.2% amphotericin B (Life Technologies), 0.5% streptomycin/penicillin (Sigma-Aldrich, Taufkirchen, Germany), and 10% fetal bovine serum (Life Technologies). Cells were cultured at 37°C and 5% CO₂.

For RT-PCR, cDNA was synthesized with the Maxima First Strand cDNA Synthesis Kit (Life Technologies) from total RNA extracted using the RNeasy Mini Kit (Qiagen, Hilden, Germany). RT-PCR was performed with the SsoFast EvaGreen Supermix (Bio-Rad, Feldkirchen, Germany) on a CFX96 Real-Time PCR system (Bio-Rad) with 30-s activation at 95°C, 5-s denaturation at 95°C, and 5-s annealing at 60°C (GAPDH) or 58°C (PAI-1) for 40 cycles. Primer sequences were PAI-1 forward 5′-CTC​ATC​AGC​CAC​TGG​AAA​GGC​A-3′; reverse 5′-GAC​TCG​TGA​AGT​CAG​CCT​GAA​AC-3′ and GAPDH forward 5′-ACA​ACT​TTG​GTA​TCG​TGG​AAG​G-3′; reverse 5′-GCC​ATC​ACG​CCA​CAG​TTT​C-3′. The relative expression of PAI-1 was calculated according to the ΔΔCT method. PAI-1 primers were adapted from [27].

To induce senescence, LNCaP cells were treated with doxorubicin (Selleck, Planegg, Germany) at a 0.01 μm concentration for 72 h. Controls were treated with DMSO (Sigma-Aldrich).

A nearly confluent cell layer of senescent LNCaP prostate cancer cells was scratched using a 10 μL pipette tip. Images of the scratches were obtained after 0 h, 72 h, and 120 h. Open surface areas were evaluated with the TScratch (https://github.com/cselab/TScratch) software.

LNCaP cells were fixed in 4% paraformaldehyde (Sigma-Aldrich) for 10 min at RT and permeabilized in 0.01% Triton-X100 (Sigma-Aldrich) for 10 min at 4°C. Nonspecific staining was blocked with 10% normal donkey serum (Jackson ImmunoResearch, Ely, UK) for 30 min at 4°C. The primary antibody used was directed against p16^INK4a (E6N8P; Cell Signaling, Leiden, Netherlands) at a 1:1,000 dilution followed by an anti-rabbit Alexa-488 secondary antibody at a 1:750 dilution. Cells were mounted with VECTASHIELD (Vector Laboratories, Burlingame, USA) containing DAPI.

Pearson’s χ² test was used for comparisons of categorical variables such as the event of biochemical and clinical recurrence, as appropriate. Numerical variables were analyzed using the Mann-Whitney U test or Student’s t test for independent samples, two-tailed. The biochemical and clinical recurrence-free survival was analyzed using the Kaplan-Meier estimator and the log-rank test. Both univariate and multivariate Cox regression models were calculated. Results were considered statistically significant with p values lower than 0.05. Statistical analyses and calculations were performed using IBM SPSS v. 17.0 (SPSS Inc., Chicago, IL, 2008).

We first sought to confirm the impact of local tumor invasion on patient prognosis in a cohort of 1,003 patients treated with a radical prostatectomy at the Department of Urology of the University Hospital Heidelberg with a mean postoperative follow-up of 62.6 months. All patients lacked nodal involvement and distant metastasis at the time of diagnosis (online suppl. Table 1).

Patients with SVI (Fig. 1a) had a higher Gleason score (p < 0.001) and worse performance status (p = 0.003) and were more likely to develop a biochemical and/or clinical recurrence (p < 0.001; online suppl. Table 2). Patients with SVI also received adjuvant treatment more frequently (p < 0.001; online suppl. Table 2). The median time to biochemical recurrence was significantly shorter in patients with SVI (2.9 years) when compared to patients with EPE alone (pT3a, 10.7 years; p < 0.001; Fig. 1b). Similarly, the mean time to clinical recurrence was 15.7 years in patients with SVI and 19.1 years in patients with stage pT3a prostate cancer (median not reached for both groups; p < 0.001; Fig. 1c).

SVI was an independent risk factor for biochemical recurrence in univariate (online suppl. Table 3) and multivariate (HR 1.847 [CI: 1.406–2.427], p < 0.001; online suppl. Table 4) Cox regression analyses. SVI was likewise an independent risk factor for clinical recurrence in univariate (online suppl. Table 3) and multivariate Cox regression models (HR 2.504 [CI: 1.409–4.451], p = 0.002; online suppl. Table 4). These results underscore that SVI has a significant negative impact on both biochemical and clinical recurrence-free survival of men with prostate cancer.

Having confirmed that SVI is associated with a poorer patient prognosis, we next tested the hypothesis that SVI creates a niche with more aggressive tumor cell behavior, i.e., enhanced proliferation. In order to characterize invading prostate cancer cells, we first surveyed 21 locally advanced prostate cancers (pT3a, n = 6; pT3b, n = 12; pT4, n = 3) using a double-immunohistochemical staining procedure for the proliferation marker Ki-67 and the senescence marker p16^INK44.

Perplexingly, we found that the invasive margins of all 21 tumors commonly harbored p16^INK4A-positive and Ki-67-negative tumor cells indicating cellular senescence (Fig. 2). In addition, senescent prostate cancer cells were present in areas of PNI in six of six pT3a tumors (100%), 10 of 10 pT3b tumors with PNI (100%), and one of two pT4 prostate cancers with PNI (50%). There were also proliferating p16^INK4A-negative and Ki-67-positive tumor cells present at the invasion zone (Fig. 2a). In contrast to malignancies associated with infection by human papillomaviruses [28], we very rarely detected p16^INK4A and Ki-67 double-positive tumor cells.

The notion that p16^INK4A-positive prostate cancer cells are in fact senescent was confirmed by showing expression of β-galactosidase in invading prostate cancer cells (Fig. 3). Expression of β-galactosidase has been reported to correlate with its senescence-associated enzymatic activity [29]. Taken together, these results indicate that the local invasion of prostate cancer involves tumor cells with features of cellular senescence.

To further corroborate the notion that local tumor invasion is associated with senescent prostate cancer cells, we selected 18 prostate cancers with SVI from our original cohort (online suppl. Table 1) and stained the tissue specimens for both Ki-67 and p16^INK4A. We performed a detailed spatial analysis for each of the 18 tumors that included the tumor center, the intraprostatic tumor periphery, and the area of SVI. For each of the three regions, four 40x HPFs containing the highest number of Ki-67- or p16^INK4A-positive tumor cells in a 1,000-μm zone were selected, the percentage of positive cells was calculated, and results were averaged for the four microscopic fields (Fig. 4).

Most areas contained both, proliferating and senescent cells, albeit with different proportions. We detected the highest proliferation rate in the intraprostatic tumor periphery in 10 of 18 tumors (55.6%), whereas five of 18 tumors (27.8%) showed the highest proliferation in the tumor center and only three of 18 tumors (16.7%) in the area of SVI (Fig. 5).

The highest number of tumor cells expressing p16^INK4A was also found in the intraprostatic tumor periphery (12 of 18 tumors, 66.7%) followed by SVI (four of 18 tumors, 22.2%) and the tumor center (two of 18 tumors, 11.1%; Fig. 5). When only tumor cells with higher nuclear than cytoplasmic p16^INK4A expression (p16^INK4A nuclear>cytoplasmic) were considered, the highest percentage was found in eight of 17 tumors (47.1%) in the area of SVI, in six of 17 tumors (35.3%) in the intraprostatic tumor periphery and in three of 17 tumors (17.6%) in the tumor center (Fig. 5). One tumor (#9) was excluded since no cells with nuclear p16^INK4A expression were detected in any of the three areas. Taken together, these results indicate that the tumor periphery and the area of SVI are hotspots for prostate cancer cells expressing p16^INK4A and p16^INK4A nuclear>cytoplasmic, respectively.

To further substantiate our findings, we next performed a series of in vitro experiments in which LNCaP prostate cancer cells were treated with doxorubicin to induce senescence (Fig. 6a). A wound healing assay using senescent LNCaP cells and DMSO-treated control cells showed a significantly enhanced wound closure in the senescent cell populations (Fig. 6b, c). Moreover, senescent LNCaP cells that have migrated into the wound area were found to express p16^INK4 (Fig. 6d), which lends further support to the notion that senescence promotes prostate cancer cell migration.

The extent of local tumor invasion is a limiting factor for the curability of prostate cancer. It is therefore critical to understand the molecular basis and the functional state of invading prostate cancer cells in order to develop effective therapeutic strategies to suppress this process.

In the present proof-of-concept study, we made the surprising discovery that prostate cancer cells that invade the surrounding tissue show frequently signs of cellular senescence, i.e., an upregulation of p16^INK4A and a withdrawal from proliferation. Increased expression of p16^INK4A has previously been reported in human premalignant and cancerous prostate lesions [31]. Our finding that senescent cells are present at the leading edge of human prostate cancer invasion has, to the best of our knowledge, not been reported before.

Cellular senescence is considered a stable cell cycle arrest that can be triggered by a number of stimuli including telomere shortening, oncogene activation, DNA damage, or other forms of cellular stress [32]. Senescent tumor cells show important additional features such as a senescence-associated secretory phenotype [33]. There are also changes in the cellular morphology such as enhanced cell spreading and enlarged nuclei [32].

In our study, a predominantly nuclear expression of p16^INK4A was mostly found in areas of SVI and in the intraprostatic tumor periphery. At the same time, tumor cell proliferation was found to peak frequently in the intraprostatic tumor periphery. Whether the increased proliferation in this zone leads to senescent tumor cells via enhanced replication stress-induced DNA damage remains to be determined. The question whether the increase of nuclear p16^INK4A expression in the area of SVI, which at the same time showed less frequently the peak proliferative activity, may involve other senescence-inducing factors likewise needs to be addressed. While it is conceivable that nuclear expression of p16^INK4A is necessary for its function as a CDK inhibitor and to mediate cell cycle exit, there are also reports that cytoplasmic expression of p16^INK4A may be associated with invasion and poor prognosis [34]. Unrelated to prostate cancer, it has been shown that in high-risk HPV-infected cells, p16^INK4A can become an essential factor for cell survival [35].

Despite the expression of p16^INK4A, β-galactosidase, and the reduction of proliferation at the invasive margin of locally advanced prostate cancer, we did not observe other morphological features of cellular senescence such as cell spreading und nuclear enlargement. Nonetheless, our data clearly show that invading prostate cancer cells are withdrawn from proliferation, which was unexpected. This observation stands in striking contrast to kidney and other cancers, in which the invasive margin is a zone of high proliferative activity [[16]–[18]].

Senescent tumor cells at the invasion front have been detected in a number of other malignancies including breast, pancreatic, colon, and thyroid cancer as well as basal cell carcinoma of the skin [[19]–[23]]. In fact, it has been suggested that tumor cells are required to switch from a proliferative to a non-proliferative state in order to become invasive [24]. Previous studies have suggested that senescent tumor cells at the invasive margin can create a senescence-associated secretory phenotype-related chemokine gradient to enhance chemotaxis, proliferation, and survival of non-senescent tumor cell [[19], [33]]. In our study, non-senescent, proliferating tumor cells were commonly present at the invasion zone (Fig. 2).

Our finding that invading prostate cancer cells show features of senescence has important translational consequences. In general, senescence has a somewhat paradoxical role in tumor development and progression. While acute induction of senescence can provide a barrier against the propagation of premalignant or cancer cells, persistent senescent cells have the potential to promote tumor progression and metastatic dissemination [[36], [37]]. This notion is supported by findings in prostate cancer, where PTEN loss has been found to lead to tumor cell senescence, thereby impeding tumor progression [38]. Remarkably, the matrix metalloproteinase inhibitor TIMP1 has recently been identified to suppress metastatic dissemination of senescent prostate cancer cells [39]. Metastasis was found to be enhanced in a Pten/Timp1 double knock-out mouse model and further accelerated by treatment-induced senescence [39]. Strikingly, this study showed that pharmacological elimination of senescent cells abolishes metastasis [39]. There are a number of possibilities to target and eliminate senescent tumor cells (“senolysis”). Since these cells rely on anti-apoptotic pathways to ensure survival, e.g., through upregulation of BCL-2 family members, senolytics that inhibit this pathway such as the BCL-2 inhibitor navitoclax (ABT-263) are particularly promising [37]. Similarly, a complete elimination of senescent prostate cancer cells and metastasis was achieved with a MCL-1 inhibitor [40]. Immunological approaches to clear senescent cells are likewise promising strategies [36]. The robust upregulation of p16^INK4A in senescent cells makes this protein also a promising candidate for a therapeutic vaccine [[41], [42]].

What are the necessary steps to introduce “senolysis” as novel therapeutic approach in prostate cancer patients? Many advances in establishing senescent cells as targets are based on mouse models. Hence, the presence and number of senescent cells should be assessed in larger cohorts and different disease stages in human prostate cancer. This should involve a combination of markers as previously suggested [43]. Second, prospective and randomized clinical trials need to be initiated to test whether senolytic drugs provide a benefit for prostate cancer patients. There are several scenarios in which senolytic drugs could be further explored. These include combination therapies with antineoplastic compounds known to induce senescence such as docetaxel or olaparib [[39], [44]]. In light of the findings presented herein, senolytic therapy may also become relevant in the adjuvant or neoadjuvant treatment of locally advanced primary prostate cancer.

Our clinical results confirm previous studies showing a more unfavorable prognosis of prostate cancer patients with SVI [[8]–[13], [15]]. However, we hasten to add that this does not prove that the increased frequency of senescent cells with enhanced nuclear expression of p16^INK4 in areas of SVI compared to other tumor regions is involved in this finding. Additional studies are required to test this notion.

In summary, this proof-of-concept study shows that prostate cancer cells at the invasion front, including areas of SVI, show features of cellular senescence and are frequently withdrawn from proliferation. Our findings may lead to novel treatment concepts in locally advanced prostate cancer that focus on eliminating or suppressing senescent tumor cells.

We are grateful to the tissue bank of the National Center for Tumor Diseases Heidelberg for tissue procurement.

All patients provided written informed consent to participate and for the use of their data and tissue for research and publication. All tissue samples were provided by the Tissue Bank of the National Center for Tumor Diseases Heidelberg in accordance with the regulations of the Tissue Bank. This study protocol was reviewed and approved by the Ethics Committee of the Medical Faculty Heidelberg of the University of Heidelberg, approval numbers 206/2005, 207/2005, S-864/2019, S-287/2022.

The authors declare no competing interests. E.-S.P. has received lecture fees from MSD Sharp & Dohme GmbH and Frauengesundheit bei Tumorerkrankungen e.V. and has been involved in a research project funded by MSD Sharp & Dohme GmbH outside the submitted work. She is Managing Director and shareholder of the company ViMREX GmbH.

This study was supported by the Medical Faculty Heidelberg and the Wilhelm Sander-Stiftung.

Sebastian Schwarz performed experiments, analyzed data, and wrote a draft of the manuscript. Cathleen Nientiedt helped with data interpretation and study design. Elena-Sophie Prigge, Adam Kaczorowski, Christine Geisler, and Carlota Lucena Porcel performed experiments and provided clinical data or pathological information. Magnus von Knebel Doeberitz, Markus Hohenfellner, and Stefan Duensing supervised the study. Stefan Duensing conceptualized the study and wrote the final manuscript.

All data generated or analyzed during this study are included in this article. Further inquiries can be directed to the corresponding author.