Background
Limitless proliferative potential is thought to be one of the most characteristic features of cancer cells [
1], however, it is not fully reflected in vitro, as maintenance of primary cancer cells in culture is highly limited [
2]. Glioblastoma (GB) is one of the most aggressive tumors of the central nervous system, associated with poor prognosis and lack of effective treatment. Although extensively explored, stable GB cell lines, such as U87-MG or T98G, are rather more homogenous populations than an accurate representation of molecularly diverse glioma cells observed in vivo [
3]. The availability of cell lines characterized with oncogenes amplifications,
EGFRvIII or
IDH1R132H mutations, commonly observed in this tumor type is severely limited [
4,
5], while primary GB cultures tend to be difficult to establish.
Senescence is one of the mechanisms associated with culturing difficulties of primary cancer cells and it has already been described in various cancer cell types [
6,
7]. We previously reported that GB cells undergo senescence in vitro very early in culture (2
nd – 3
rd passage) and avoid stabilization attempts [
4]. Other accompanying phenomena include spontaneous or idiopathic apoptosis and cell death resulting from mitotic catastrophe [
4], but these have not been profoundly analyzed so far.
Recent analysis of culturing methods of primary GB cells indicates that there is plethora of published protocols, differing in culture medium, plate coating or culture type [
4,
8‐
14]. Therefore, there is a lack of standardized and unified method of establishment and maintenance of such cultures. Due to this fact it is difficult to compare establishment efficiency between different laboratories, as obtained results are often even contradictory [
15]. To further complicate this issue, it is worth to emphasize that glioblastoma is molecularly classified into four subtypes [
16], and each may require different culture conditions or establishment approach. Nevertheless, it remains debatable whether culturing inconsistencies actually depend on applied conditions, or rather on molecular profile of tumor cells. Without the precise molecular characterization it is not clear what type of cells (tumor or normal cells infiltrating tumor mass) actually preserves in long-term culture. Currently there is a tendency to limit the number of reported passages and restrict molecular identification of cells to tissue samples, with no molecular data of cultured cells available [
12,
16].
As the exact mechanism hindering stabilization of proliferating GB cells remains elusive, in this paper we analyzed three different approaches of glioblastoma cells culturing in an attempt to try to understand and circumvent senescence and cell death, hence, prolonging in vitro maintenance of cells with preserved phenotype and genotype. Determination of the most optimal approach will not only enable to employ primary GB cultures for complex in vitro analyses, but also possibly provide an insight into the mechanisms underlying culturing difficulties.
Discussion
Despite many years of extensive research glioblastoma, the most malignant brain tumor, remains incurable. This is not only associated with its location, severly limiting treatment options, but also high intratumoral heterogeneity [
3,
4,
12,
16]. There are very few established, indefinitely proliferating GB cell lines, characterized by molecular alterations that are commonly found in vivo [
4]. Therefore, there is a need to develop the reliable GB in vitro testing platform. Unfortunately, establishment of primary glioblastoma cultures, definately more heterogenous and better reflecting in vivo tumor state, is difficult.
In theory, cancer progression model, based on transition from preneoplastic to neoplastic phase, requires cancer cells bypassing cellular senescence and apoptosis [
1]. However, despite being one of the most aggressive tumors, GB cells tend to undergo idiopathic senescence and apoptosis, in some cases even quite extensive, in early passages during in vitro culturing [
4]. In very few cases of established GB cultures, notably these characterized by a specific combinations of molecular alterations that are almost never found in vivo, these phenomena are counterbalanced by proliferation enabling culture establishment [
4]. It remains unknown what determines stabilization success and whether is it associated with culture conditions, glioblastoma subtype or molecular characteristics of the tumor. In this paper, cells that failed stabilization attempts (8 out of 10 analyzed cases) were cultured not only according to NSC conditions-mimicking protocol (previously applied by Xie et al.), but also in standard monolayer culture conditions. This clearly indicates that optimization of culture conditions does not constitute the ultimate solution to stabilization problem of primary cancer cells. We have previously detected the atypical combination of
TP53 and
CDKN2A mutations in almost half cases of stabilized GB cultures [
4]. As these alterations are considered mutually exclusive in glioblastoma, it may suggest that they predispose cells to stabilization. On the other hand, their very rare in vivo occurrence indicates that when stabilized, primary GB cells tend to deviate from the original genotype. Indeed, during in vitro maintenance uncontrolled selection of cells with specific mutation or phenotype can be observed [
3,
12,
16] and due to glioblastoma genomic instability additional genetic alterations may commonly occur, especially late in the culture course. One of the cultures stabilized in the current study (GB7) was initially characterized by
TP53 mutation, while in both established cultures (GB7 and GB10) we detected deletion of one allele of
PTEN further in the culture course. Therefore, successful stabilization of these particular cases might have been associated with positive selection of cells characterized by the deletion of these two suppressor genes. Basing on the obtained results we cannot directly confirm our previous assumptions regarding the co-existence of tumor suppressor mutations. Still, it seems that lack of cell cycle control may facilitate cancer cells transfer and maintenance in vitro. Moreover, we have previously described that there is a higher frequency of homozygous mutations in tumor suppressors as well as mutations resulting in lack of tumor suppressor proteins in cell lines when compared to surgical samples. Hence, there seems to be a complete loss of function of tumor suppressor genes in time [
23]. On the other hand, mesenchymal GB subtype is suggested to be more prone to stabilization, even if patient was diagnosed with another GB subtype [
16], what seems to be reasonable, as cells in monolayer conditions attach to the culture vessel and compete with normal stromal cells [
24]. Indeed, some researchers even observed the transition from mesenchymal type (observed in vitro) to other phenotypes in xenografts [
16]. In general, all these data clearly indicate discrepancies between in vivo and in vitro analytical models what can be of utmost importance in terms of drugs testing perspective.
Another important aspect for cancer cell culturing is the lack of unified approach for primary cells establishment and conditions for their further maintenance. Currently, over 20 different protocols are in use [
14], hence obtained results cannot constitute a subject of reliable comparison. Indeed, there are a lot of discrepancies between laboratories working in the field of GB primary cultures. Research group headed by Pollard [
12] indicated inconsistencies between NSC-like adherent conditions and neurosphere 3D cultures, what is in line with our results and those obtained by other research teams [
4,
12,
16,
25,
26]. This report was immediately criticized by Reynolds and Vescovi [
13] as this group claimed that their processing of 150 GB specimens resulted in 100% efficiency in cell culture establishment using neurosphere-based approach. Intriguingly, this research group claimed that they were able to passage these cells approximately 50 times a year (up to 150 passages), which can be considered quite an achievement. Still, it is very hard to comprehend, since neither our team nor any other laboratory reported complete efficiency of primary GB cells stabilization, especially when neurospheres formation was involved. Usually, establishment rate is estimated to be 33–47%, while the number of passages reached by cultures in NSC-like conditions do not generally exceed 10 (when proper molecular analyses during the culture course are included) [
12,
16]. Actually, it was demonstrated that the subtype of GB cells matched tumor of origin only in 45% of cases, mostly due to high in vivo tumor heterogeneity [
16]. These examples clearly indicate that results of primary GB cells establishment and analyses conducted using such models should be analyzed more critically.
As mentioned, discrepancies concerning culture establishment may result from the fact that the term ‘stable cell line’ is not uniformly defined. In our laboratory, we consider cancer cell line stabilized when it reaches at least 50 passages (approx. twice as many as normal cells do) and shows any molecular changes at DNA level. The percentage of cultures that stabilize in our laboratory has been constant for many years, despite the application of new culturing approaches, culture media, supplements or coatings and is not directly associated with the technical abilities of the person performing the procedure. It seems that not all researchers tend to follow such strict criteria, as the percentage of stabilized cell lines reported in the literature is often overestimated. Cell line cannot be considered stable when it reached only several passages (e.g. only up to 10) or molecular status of these allegedly stabilized cells is not verified [
24]. Unfortunately, established cultures are often not molecularly analyzed at various time points. Nevertheless, it can be considered quite encouraging that articles analyzing the aspect of primary cells culturing from a more complex point of view begin to be published [
27]. One can criticize that stabilization success rate in this study is less than average, however, strict criteria used in our laboratory may ensure reliability of results.
Since appearance of senescence and apoptosis in primary neoplastic cells is considered spontaneous (idiopathic), recognizing and describing the mechanism of these phenomena, which tend to occur at early passages of GB culture, remain crucial to circumvent culturing difficulties. We are aware of the fact that senescence and apoptosis, which can be very extensive in some cases, are somehow associated with stabilization failure and may constitute a result of mitotic catastrophe. It is not the first report associating senescence-like phenotype (SLP) with mitotic catastrophe. Indeed, Eom et al. pointed out that cells may firstly undergo mitotic catastrophe and then enter a temporary senescence-like arrest prior to cell death [
28]. Moreover, Wang et al. demonstrated that GB cells with normal TP53 protein tend to exhibit SLP in response to inhibition of topoisomerase I, while cells with mutated TP53 undergo apoptosis [
29]. On the other hand, TP53 inhibition in fibrosarcoma cells was found to be associated with decreased drug-induced SLP and increased mitotic catastrophe [
30]. Although in mentioned studies these phenomena were not spontaneous, in this article we made an attempt not only to test various methods of GB culture establishment, but, most importantly, to try to provide a greater insight into the mechanisms underlying inability to stabilize these cells in vitro. Therefore, lentiviral vectors and in vitro
/in vivo cultivation approaches were mostly applied in the article not as a method of culture establishment, but rather as an attempt to understand the causes of the difficulties in stabilization of primary cancer cells.
Based on the approaches that were actually developed for normal cells, we put forward the hypothesis that in vitro immortalization may enable longer propagation of primary GB cells. Several immortalizing factors were selected considering genetic pathways involved in cancer cells senescence. Transduction of primary cell cultures with BMI-1 (inactivating CDKN2A/RB pathway), SV40 (binding and inactivating tumor suppressor protein TP53) and hEST2 (generally up-regulated in tumor cells and during immortalization, preventing senescence resulting from telomere shortening) was insufficient to provide infinite proliferation. Since it is well established that both pathways, TP53/p14 and p16, are essential to induce replicative senescence [
31] we assume that this mechanism is not associated with the observed phenomena. Neoplastic cells are thought to maintain high proliferation capacity mainly due to their telomerase activity – large ribonucleoprotein complex (dependent on hEST2), regulating cell replicative lifespan through stabilization and extension of telomeres [
32]. Repression of telomerase or lack of its activity finally leads to cellular senescence, apoptosis or premature aging [
33]. Various reports have demonstrated link between neoplastic progression and telomerase activity emphasizing the fact that only minority (10–15%) of malignant cells does not possess upregulated hEST2 expression [
34]. However, recently it was proven that elongation of telomeres does not eliminate occurrence of senescence [
35]. Moreover, inactivation of proteins involved in cell cycle control is possibly not enough to circumvent senescence, despite the fact that this was suggested by the results of our previous analyses, indicating that proper balance between senescent/apoptotic and proliferating cells may be achieved only when specific genetic alterations ensuring lack of cell cycle control (e.g. a combination of
TP53 mutation and
CDKN2A homozygous deletion) co-exist in the cell [
4].
Considering the fact that all analyzed in vitro conditions might have an impact on triggering senescence and apoptosis, we decided to test sequential in vitro
/in vivo model of culturing glioblastoma-derived cells, as it was reported that the majority of glioblastoma subtypes can efficiently give rise to tumors when implanted in vivo, with two thirds retaining characteristics of original primary tumor tissue [
16]. Despite the fact that this approach did not enable to stabilize analyzed GB cases, so far it seems to be the most promising option for long-term culturing of these cells. Above all, it facilitates fast propagation of native, unaltered cancer cells for the purpose of e.g. screening of molecules with antineoplastic potential in both, in vivo and in vitro conditions. It is especially important as in the majority of cases the amount of material directly derived from patients is not sufficient for standardized analyses and further statistical assessment. Importantly, despite the fact that our model was based on subcutaneous GB cells injection, it still enabled to impede growth arrest and phenotypic changes of tumor cells longer than all tested in vitro approaches.
One of the most important questions is whether the inability to evade senescence/apoptosis by the majority of primary GB cultures observed in the present study, is a feature inherent in the behavior of these cells. Despite our many years of research on this topic, we are still unable to unequivocally determine whether senescence constitutes the major reason of stabilization failure or is it a feature characteristic to GB cells. It has to be mentioned that senescent glioblastoma cells have been reported to be present in vivo [
36], however, as we demonstrated in this paper, the percentage of senescent cells tends to increase with in vitro passages. To make it even more complicated, our team and others demonstrated that a small, but rather constant, population of senescent, and even apoptotic cells is also present in stable cancer cell lines [
6,
37,
38]. Hence, we decided to evaluate the percentage of senescent cells in several stable glioblastoma cell lines. Our analyses indicated a considerable diversity both, in the presence and percentage of senescent cells – cells positive for SA-β-Gal activity were detected within T98G, U87-MG and DK-MG lines, with the last line characterized by the highest percentage of senescent cells. These results may support the theory of dual – anti- and pro-neoplastic role of senescence [
39], complicating manipulations of this phenomenon in anticancer research as well as unequivocal understanding of the mechanism of this process and its implications. Finally, not only idiopathic senescence, but also idiopathic/spontaneous apoptosis was observed in primary and stabilized GB cell lines. These data suggest that tumor is highly heterogeneous not only from a molecular, but also a functional point of view, meaning that not all cancer cells are immortal, at least under in vitro conditions. Such intratumoral heterogeneity is associated with the fact that not single cells, but rather cell subpopulations tend to exhibit specific functions (e.g. with part of cells being immortal, while another part being associated with secretory/senescent phenotype). Therefore, not only cancer stem cells should be considered in such a context, as these interdependencies may be more complicated, as demonstrated on an example of stable cells lines (“fixed” percentage of senescent/apoptotic cells).
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.