Specific lipofuscin staining as a novel biomarker to detect replicative and stress-induced senescence. A method applicable in cryo-preserved and archival tissues

Introduction

Cellular senescence is the state of irreversible cellular growth arrest in which the cell remains metabolically active [1, 2]. Two types of cellular senescence have been described in mammalian cells [3]. Replicative Senescence (RS) that is triggered by the arrest of cellular proliferation after a certain number of divisions due to telomere attrition [1, 3] and Stress Induced Premature Senescence (SIPS) that is a more acute phenomenon in which the cells stop to proliferate under various stress conditions, regardless of telomere length [3]. The phenomenon of cellular senescence was originally described in vitro [4]. More recently, senescent cells were also identified in aged skin [2], benign tumors and premalignant lesions [5-9] as well as in age-related pathologies [10]. Also, the number of senescent fibroblasts reportedly increases exponentially in the skin of aging primates, reaching >15% of all cells in very old individuals [11]. The evidence so far from in vitro and in vivo studies suggests that cellular senescence acts as a tumor barrier, whereas it contributes to the processes of tissue aging and age-related diseases [12]. The significance of cellular senescence incarcinogenesis and age-related disorders, renders the detection of these phenomena essential. This urgent need of reliable biomarkers of senescence is even more apparent given the evidence for cellular senescence induced in response to anticancer therapy [13].

The most widely used biomarker of cellular senescence reported so far is the detection of Senescence-Associatedβ-Galactosidase activity (SA-β-gal) in sub-optimal pH [2, 14]. Nevertheless, a major disadvantage in designinglarge-scale studies of cellular senescence in humanlesionsis that SA-β-gal staining requires fresh tissue as it is based on an enzymatic reaction [14]. This fact seriously limits the exploitation of the widely available formalin-fixed paraffin-embedded (FFPE) archival tissues, including tissue microarrays [1]. In an effort to establish a biomarker of cellular senescence that could be applicable for FFPE archival tissue material, we focused on lipofuscin, also known as an "age-pigment" [15]. Lipofuscin is an aggregate of oxidized proteins that accumulates progressively mostly in aged post mitotic cells [16]. It is considered a hallmark of aging and is also involved in the pathogenesis of certain age related pathologies such as macular degeneration [16]. Sudan Black B (SBB) is a lipophilic histochemical stain that identifies lipofuscin and is applicable for in vitro and in situ studies [17-19]. Here we employed SBB in a series of experiments designed to demonstrate that lipofuscin accumulates in vitro in normal human cells during RS or SIPS, as well as in stressed human cancer cells. Furthermore, we sought to identify lipofuscin deposits in benign lesions already known to contain senescent cells. As a control marker of the cellular senescence state we used the SA-β-gal assay. Our results show that the SBB-stained lipofuscin is present in all the cells that express SA-β-gal activity and it is absent in SA-β-gal-negative cells. Hence, SBB positivity could be used as an additional cellular senescence biomarker. Moreover, SBB staining was applicable in FFPE tissue sections, providing evidence that this assay can provide a reliable biomarker for detection of senescent cells in archival clinical material that is stored in paraffin.

Results

To assess the value of lipofuscin as a potential biomarker of cellular senescence in vitro, five cellular systems of normal diploid cells and cancer cells were applied. In these experimental settings cellular senescence was triggered by means of proliferative exhaustion (Replicative Senescence, RS) or Stress Induced Premature Senescence (SIPS). Specifically, we used young proliferating primary human diploid lung fibroblasts (DLF) (at passage 6), along with replicatively senescent cells (at passage 42) and cells (at passage 7) in which SIPS was triggered by γ-irradiation [20] (Fig. 1). The effect of p53 and p21^WAF-1, two well established effectors of the senescence program, was studied in two inducible osteosarcoma cell lines, namely Saos-2 p21-Tet-ON and Saos-2 p53-Tet-ON [21-24]. After 8 days of p21^WAF-1 or p53 induction senescent cells were evident (Fig. 2A, 3A). Finally, the inducible osteosarcoma U2OS E2F1-ER cell line was also tested. Over-expression of E2F1 has been shown to trigger SIPS via activation of the DNA damage response (DDR) pathway [25]. The induction of E2F1 yielded senescent cells in 10 days (Fig. 4A) as concluded by positive SA-β-gal staining.

In all the above systems cells that acquired the morphological features of senescence (i.e. became enlarged and flattened) were positive for SA-β-gal (Fig. 1A, 2A, 3A, 4A). These preparations were also stained with SBB and as illustrated in Fig. 1B, 2B, 3B and 4B all contained clearly visible perinuclear and cytoplasmic aggregates of lipofuscin, stained by SBB as dark blue-black granules. In addition, as lipofuscin produces auto-fluorescence, we could verify that the blue-black granules, stained by SBB, represented lipofuscin aggregates, by masking the lipofuscin's auto-fluorescence with SBB staining (Fig. 1C, 2C, 3C, 4C) [26]. This unique property of theSBB wasthe reason it was selected amongother methodsthat detectlipofuscin [19, 27]. To assess the extent of agreement between the results from the two assays, we performed SA-β-gal and SBB co-staining in the same cells. Indeed, co-staining results showed a complete overlap (Fig. 1D, 2D, 3D, 4D). Likewise, concomitant staining of SA-β-gal, SBB and the proliferative marker Ki67 verified that cells positive for lipofuscin and SA-β-gal activity were not proliferating (Fig. 1E, 2E, 3E, 4E).

Next we examined whether SBB staining of lipofuscin could serve as a reliable senescence bio-marker in vivo. To this end the same experimental procedure was followed in frozen tissue sections from mice lung adenomas. The lung adenomas were developed in a mouse model expressing conditionally K-rasV12in the lung [5], one of the first in vivo settings used to demonstrate the role of senescence as an anti-tumor barrier in premalignant lesions [5]. In line with the in vitro findings, the lung adenomas that demonstrated strong SA-β-gal activity stained positive for lipofuscin while normal lung tissues were negative (Fig. 5). Next, we examined frozen human samples from patients with benign prostatic hyperplasia (BPH) from enlarged prostates (>55gr). These lesions had been previously shown to feature senescence [7, 28]. As shown in Fig.6, SA-β-gal activity and SBB staining co-localized, whereas adjacent normal prostatic glands were negative.

Having verified specific staining of senescent cells by SBB we then asked whether this approach could be also applicable in archival tissues. Thus, SBB stain was performed in FFPE tissue samples prepared from the above in vivo settings. As demonstrated in Fig.7, SBB staining clearly demonstrated lipofuscin in the lung adenomas (Fig.7). Strikingly, adjacent adenocarcinomas that spontaneously developed in these mice [5] were negative (Fig. 7). This finding supports the reliability of SBB as a method for staining senescent cells in FFPE (Fig. 7). Also, it is in line with the finding of Collado et al., that spontaneously developing adenocarninomas in this model bypass the senescence anti-tumor barrier [5]. In addition, as representatively shown in Fig.8, SBB specifically stained blue-black lipofuscin granules in the FFPE material from BPH (see Fig.6) that were characterized as SA-β-gal positive. To further validate SA-β-gal and SBB staining tissue co-localization, we applied triple staining for SA-β-gal, SBB and NFR on BPH samples.

Specifically, the frozen samples were fixed in formaldehyde for 2h, stained with SA-β-gal; routinely processed for FFPE as previously described [6], sectioned and then stained with SBB (Fig.9). Although,SA-β-gal activity was detected mostly in the periphery of the sections, leaving the core of the tissue unstained (likely due to slow penetration of the SA-β-gal stain into the tissue) a clear co-localization of SA-β-gal activity and SBB staining was noted close to the tissue periphery marked by pathological features of BPH. Interestingly, SBB foci were also observed in the core of the tissue (Fig. 9).

It has been reported that false positive SA-β-gal staining may be detected in confluent cultures of quiescent cells [29, 30]. To investigate whether lipofuscin may be present in confluent cultures, we performed the SA-β-gal assay and SBB stain in confluent DLF cultures. We observed that as soon as the cells reached confluence they showed SA-β-gal activity while, in contrast, such cultures showed only negligible lipofuscin staining (Fig.10A). Extending the two assays for 72 hours we observed that all the cells clearly demonstrated SA-β- gal activity, whereas cells containing lipofuscin granules were fewer (Fig. 10B). Furthermore, the enzymatic activity of galactosidase [31] may produce positive results of SA-β-gal assay in non-senescent cells, when incubated for prolonged time. We incubated sub-confluent cultures of young DLFs in SA-β-gal for 72 hours (prolonged incubation) and then stained immediately with SBB. In these cells we observed SA-β -gal activity while there was no detectable SBB staining for lipofuscin (Fig. 10C).

Discussion

In this study, we examined whether lipofuscin, “a hallmark of aging”, [15] is also a cellular senescence biomarker and especially whether it is a marker of SIPS. Our working hypothesis was based on the notion that senescent cells accumulate in aged tissues [2]. We argued that the presence of lipofuscin in senescent cells could complement SA-β-gal by providing and additional marker of cellular senescence. To this end a series of experiments were designed in which various cellular systems, based on normal and cancer cells, were driven to either RS or SIPS. In line with our hypothesis all cells that displayed the senescence phenotype under either RS, or SIPS scenarios demonstrated co-localization of SA-β-gal activity and SBB-detected lipofuscin granules, both in vitro and in vivo, (Fig. 1D, 2D, 3D, 4D, Fig. 1E, 2E, 3E, 4E and Fig. 5C, 6C) whereas tissues negative for SA-β-gal activity were negative for lipofuscin, as well (Fig. 5E, 6D).

Lipofuscin is a non degradable aggregate of oxidized proteins, lipids and metals which accumulates inside the lysosomes of cells that do not replicate [16]. Such accumulation possibly reflects the inability of nonproliferating cells to dispose of lipofuscin by cellular division, a process that naturally results in dilution of lipofuscin [32]. Lipofuscin accumulation in aged tissues and age- related pathologies is considered a progressive phenomenon, probably as a consequence of the decline of the cellular clearance systems of misfolded (lipo)proteins and possibly other ‘aberrant’ metabolites [16, 33, 34]. We observed the lipofuscin granules in cells shortly after they became senescent. This fact implies that lipofuscin formation is possibly related to the senescent state, rather than just arandomtime-coincident event. Our findings reinforce the observation that senescent cells are hyper-metabolic and full of lipids in the cytoplasm. [35]. As no specific antibody for lipofuscin exists, we have used the SBB histochemical stain to demonstrate the presence of lipofuscin [16]. The dye is diluted in ethanol and due to its lipophilic nature, when in contact with lipids; it assembles on the lipid surface, as it is more soluble in lipids than in ethanol [26]. Accordingly, SBB stains the lipid component of lipofuscin [26]. Several methods were reported to detect lipofuscin, including fluorescence microscopy due to the natural auto-fluorescence of lipofuscin [36], as well as histochemical dyes like SBB, Berlin Blue, Nile Blue, Ziehl-Neelsen and Periodic acid Schiff [37]. As SBB is shown to specifically mask lipofuscin's fluorescence, it was considered the most reliable histochemichal stain to apply [19, 26, 27, 38]. SBB is suitable for use both in frozen and FFPE material [27]. We also verified that SBB staining for lipofuscin may depict senescent cells in FFPE sections from precancerous lesions already shown to contain senescent cells [5, 7, 28] (Fig. 7, 8A-C and Fig.9), while it is negative in the adjacent normal tissue (Fig. 8D) and the related carcinomas (Fig. 7) that have over-come the senescence anti-tumor barrier [5]. These characteristics support the candidacy of SBB for a highly desirable tool to study senescence in archival material.

There is a broadrange of potential senescence markers whose reliability, however,varies [39, 40]. None of the candidate markers proposed to date, however, is considered entirelyspecificfor cellular senescence, especially for in vivo applications[3]. The most commonly used senescence biomarker is SA-β-gal activity [14]. On the other hand, even the SA-β-gal assay apparently produces false positive results, under certain culture conditions such as confluence and serum starvation [29, 30>]. Also, cells that do not express the galactosidase gene, show no SA-β-gal activity, but fully execute the senescence program [31]. SA-β-gal, can only be used in fresh-frozen tissue as it is based on the enzymatic activity of galactosidase [14]. This limitation prevents to use SA-β-gal for studies of senescence in FFPE- archival material [1]. Furthermore, if cells or tissues are left in SA-β-gal solution for a prolonged time, all cells eventually will acquire the characteristic turquoise stain, due to their normal enzymatic galactosidase activity (Fig. 10C) [31]. To perform the technique a control sample is always required and the process is usually stopped when the sample under investigation starts to stain. As different tissues and cells require different times to stain and the desired stain is selected subjectively by the observer, the density and intensity of the turquoise color considered as positive varies between studies. On the other hand, SBB is a fast dye and it takes only few minutes to stain lipofuscin in tissues and cells, while the results are homogeneous and reproducible. This technique for identifying senescent cells is therefore easier, more rapid, likely more reproducible, and especially more suited for a wider spectrum of applications in diagnostic pathology laboratories. Moreover, SBB is a technique employable in frozen tissue [27], so it is also ideal for use in parallel with SA-β-gal as an additional biomarker of the senescent state. Of note, like all the other senescence biomarkers, lipofuscin is not 100% specific for senescence as it aggregates in degenerative conditions such as macular degeneration [16]. As shown (Fig. 10A, 10B) in confluent cultures, cells containing lipofuscin may be detected. However, in contrast to SA-β-gal activity which is immediately demonstrated in confluence and is present in all the cells (Fig. 10A), lipofuscin granules appear later and to a significantly lesser extent (Fig. 10B).

The majority of pathologies associated with lipofuscin are age-related diseases [16]. The role of cellular senescence in age-related pathologies and cancer, conditions that are both considered integral components of “aging”, is a widelygrowing fieldof biomedical research [39, 40]. Future studies will focus on the role of cellular senescence in cancer, especially considering the increasing evidence thatsenescence is one of major outcomes and a determinant of treatment response in oncology, broadly analogous to significance of apoptosis [13, 41]. Furthermore, there is evidence that presence of senescent cells in certain malignant tumors may be a sign of better prognosis [42]. There is little doubt that a convenient bio-marker applicable in archival tissue materialwould greatly facilitate research on cellularsenescence in cancer. Thedetectionoflipofuscin with SBB could be applied in studies that evaluate the effects of chemotherapy and otherantineoplastic treatments.

Taken together, our present study showed that lipofuscin can provide a senescence biomarker comparable to the SA-β-gal activity. Detection of lipofuscin with SBB stain can be applied as a stand-alone or auxiliary to SA-β-gal technique. Furthermore, SBB may visualize senescent cells in FFPE tissues, the most common form of archival clinical material, thereby extending the applicability of currently available candidate senescence biomarkers to a much wider selection of research topics related to diverse diseases and aging.

Methods