Growing Capan-2 spheroids display a proliferation gradient
Our first aim was to examine how regionalization of proliferation parameters occurs in Capan-2 multicellular tumor spheroids. To address this issue we selected two representative spheroid growth stages that will subsequently be referred to in this work as small and large spheroids. These stages correspond to spheroids of an average diameter of 300-350 μm and 500-600 μm, respectively. The experiments reported below were performed and strictly restricted to spheroid sections that were identified as closely passing through the spheroid center.
We first assessed proliferation using Ki-67 immunofluorescence staining procedure. The Ki-67 antigen is a marker of cell proliferation widely used to evaluate the proliferation index in tissue biopsies [18]. As shown in Figure 1A small spheroids display a homogeneous Ki-67 positive cells distribution pattern, while in contrast on large spheroids Ki-67 positive cells staining is restricted to the outmost cell layers. This first result illustrates the notion of an existing proliferation gradient that is initially absent and develops as multicellular tumor spheroids grow.
To refine this analysis, we sought to visualize the population of cells that are engaged in cell cycle progression and therefore undergoing S-phase replication. To this aim we performed EdU (5-ethynyl-2´-deoxyuridine) incorporation for 24 h. EdU is a nucleoside analog to thymidine that is incorporated in DNA during replication and is therefore a functional marker of actively proliferating cells. As shown in Figure 1B, while in small spheroids staining was observed in the whole volume, a clear regionalization was observed in large spheroids with EdU incorporation restricted to the outmost layers. In such large spheroid quantification of the percentage of EdU positive cells (Figure 1C) as a function of the distance to the spheroid surface displays a progressive decrease of the proportion of proliferative cells from the surface to the center of the spheroid. This experiment shows that approximately 60% of cells have undergone S-phase over the last 24 h and can be considered as actively proliferating in the outer 40 μm of the spheroid. Strikingly, in the same conditions one can also observe a rapid decrease with less than 20% of cells that had undergone DNA replication 150 μm away from the spheroid surface and the absence of any EdU positive cell in the center of the spheroid.
Finally, we also examined whether the observed regionalization of Ki67 and EdU staining matched the hypoxia gradient that progressively takes place in growing spheroids. Indeed, as shown in Figure 1D, while no hypoxia is detected in small spheroids, a strong Pimonidazole staining is observed in large spheroids and its regionalization matches those of the proliferation parameters.
Cell cycle regionalization within growing Capan-2 spheroids
The observed regionalization of Ki-67 staining and EdU incorporation in large spheroids indicates that the ability of cells to progress in the cell cycle is dependent on their position to the spheroid surface and suggests that expression of cell cycle regulatory proteins might be dependent on that parameter. We therefore explored systematically cyclins expression, which is representative of the distinct phases of cell cycle, and of phospho-histone H3 to detect mitotic cells on small and large spheroids sections. In Figure 2, from top to bottom are presented staining observed for cyclin D1, cyclin E, cyclin A, cyclin B and phospho-Histone H3.
Cyclin levels are known to mirror the progression of a cell within the cell cycle. As expected, in small spheroids mainly constituted of actively proliferating cells, all four cyclins were highly expressed, with level variations reflecting their position in the cell cycle. In contrast, large and regionalized spheroids displayed an expression pattern of all cyclins that mirrored the proliferation gradient reported above. Finally, detection of phosphorylated histone H3 allowed the detection of mitotic cells that are evenly spread in small spheroids and again restricted to the outer layers in large spheroids.
All together these observations led us next to investigate whether and how cell cycle distribution was a parameter subjected to changes upon spheroid growth and in various layers of a large spheroid. To address that question in a more systematic and quantitative manner than immunostaining detection, we developed spheroid models expressing Fucci (Fluorescence Ubiquitination Cell Cycle Indicators) tools that allow monitoring cell cycle position of a cell [19]. Fucci-red (Cdt1-mKO2) is expressed and stable in G1 cells, while Fucci-green (Geminin-mAG) marks S- and G2-phases. We cloned these constructs and prepared lentiviruses particles that were used to transduce and establish stable Capan-2 cell lines expressing these fluorescent fusion proteins [17]. To validate the cell cycle regulation of these reporters in 3D MCTS, dynamics of Fucci-green and -red expression was examined by co-immunostaining with cyclin antibodies on small and large spheroids. No Fucci-green positive cells (i.e. S-G2-M cells) were found to express the G1-phase cyclins D or E, and no Fucci-red positive cells (i.e. G1 cells) were found to express the G2-phase cyclins A or B (see Additional file 1).
Figure 3A and 3C present typical wide field fluorescence microscopy images of spheroids sections illustrating the distribution of Fucci-red and Fucci-green in small and large spheroids. As shown, an even distribution of both markers is observed in small spheroids, while a clear gradient of both Fucci-red and Fucci-green is detected in large spheroids. These results were quantified by determining the percentage of Fucci-red and Fucci-green positive cells as a function of the distance to the spheroid surface. These graphs drawn at the same scale for small and large spheroids clearly illustrate the fact that G1-phase cells are evenly spread in small spheroids (average 50% of positive cells), while in large spheroids a gradient is observed with a noticeable decay approximately 150 μm away from the cell surface (Figure 3B). The percentage of Fucci green cells (i.e. S, G2 and mitosis) is less homogenously distributed in small spheroids with a progressive decrease, while in large spheroids the percentage of positive cells is reduced and no detectable Fucci-green positive cells are observed approximately 120-150 μm away from the surface (Figure 3D).
Altogether, these data indicate that the proliferative zone of a spheroid is restricted to the outer 150 μm where Ki-67 staining is positive and EdU incorporation occurs. Indeed cells are apparently able to efficient cell cycle commitment and to proceed in S-phase in this area only, confirming the above results. These experiments further underline the importance of the growth regionalization aspect occurring upon MCTS growth when investigating cell cycle checkpoint in 3D using a drug that impairs cell cycle progression.
Lovastatin-induced cell cycle arrest in G1
Lovastatin inhibition of mevalonate synthesis results in farnesyl- and geranylgeranyl-pyrophosphate deprivation that ultimately causes an inactivation of Ras and Rho. It has been demonstrated that up-regulation of p27 mediated by the reduction of Rho geranylgeranylation is responsible for the inhibition of CDK2 activity and consequently for the arrest of the cell cycle at the G1-phase [20, 21]. Accordingly, we observed that treatment of a 2D monolayer culture of Capan-2 cells expressing Fucci with lovastatin (60 μM) results in an efficient cell cycle arrest in G1 (85%) with about 90% of Fucci-red and 10% of Fucci-green cells (see Additional file 2). In order to validate that Fucci-red or Fucci-green expressing Capan-2 MCTS could allow evaluating an arrest of the cell cycle at the G1-phase, we analyzed the variation of Fucci reporters’ expression induced by a Lovastatin treatment.
When applied on spheroids, we find that lovastatin treatment impairs growth in a dose-dependent manner with an IC50 of 5 μM and induces cell death with spheroid collapse at high concentration. We therefore examined the effect of 5 μM (data not shown) and 10 μM lovastatin after 24 h and 48 h treatment of Fucci-red or -green spheroids. As illustrated with the micrographs shown in Figure 4A and the quantification shown in Figure 4B, the percentage of Fucci-green cells decreases over time while an increase in Fucci-red positive cells is detected. At 48 h, we observe a nearly 40% reduction of Fucci-green positive in the outmost region of the spheroid and in parallel in the same region the percentage of Fucci-red positive cells increases from 57% to 68%. Thus, albeit about 15% of Fucci-green cells are still detected after 48 h, a high level of Fucci-red positive cells indicative of an arrest in G1 is observed in the outmost proliferative layers.
Monitoring G2/M checkpoint activation in MCTS treated with etoposide
Etoposide (VP-16) is a topoisomerase inhibitor that can be used to induce DNA damage and activate cell cycle checkpoint. As this is the case with many other cell lines, a one hour treatment with 40 μM etoposide results after a 24 h release in the massive accumulation of Capan-2 cells at the G2/M checkpoint that can be detected either by flow cytometry or by the observation of the accumulation of 90% of Fucci-green positive cells with a very low residual percentage of Fucci-red (see Additional file 2).
Since it is not technically possible to wash away etoposide from a treated spheroid, we used continuous treatment with a 1 μM and 5 μM concentration for 24 h and 48 h. The micrographs of spheroids treated for 48 h (Figure 5A) illustrate the decrease in Fucci-red cells with a concomitant increase in Fucci-green cells percentage. As presented in the quantitation of these experiments shown in Figure 5B, while modestly affected at 24 h, after a 48 h treatment with 5 μM the percentage of Fucci-green positive cells increases by approximately two-fold (25% to 45%) and in parallel the percentage of Fucci-red positive cells drops from 60% to 30%. Strikingly, this effect is observed up to 120 μm deep in the spheroid. Phospho H2AX staining was used to detect DNA damage and confirmed the presence of numerous positive cells in the whole volume of the spheroid, thus indicating that etoposide has penetrated in the spheroid (see Additional file 3). One can therefore conclude that etoposide treatment of 3D MCTS efficiently activates a DNA damage checkpoint resulting in a global change in cell cycle distribution of the outmost layers of the spheroid.
Cell cycle arrest in G1 and G2 in response to EGF removal
In order to investigate growth factor starvation effect on cell cycle parameters in MCTS we referred to our recent observation of the dependence of Capan-2 spheroid growth on the presence of EGF [17]. In that context, EGF deprivation effect on cell cycle progression can only be explored within 3D MCTS as Capan-2 cells proliferate in the absence of EGF when grown as 2D monolayer. Capan-2 Fucci-red or -green expressing spheroids grown in DMEM/F12 media were deprived of EGF in the presence of 10% serum and monitored over 6 days. EdU incorporation was performed every day for a 24 h duration to assess the ability of the cells to enter S-phase during that time.
As shown in Figure 6A (left panels) EdU incorporation is totally abolished after 6 days indicating that most cells are likely in a quiescent stage i.e. in a prolonged cell cycle arrest. Quantification of the evolution of EdU incorporation overtime as a function of the distance to the cell surface is presented in Figure 6B (top). As shown, there is virtually no EdU incorporation after 5 days and a major decrease is already detected in EGF-deprived spheroids at day 2 in the deepest layers and at day 3 on the outer layers.
According to the classical view of cell cycle regulation such result suggests that upon growth factor removal cycling cells are unable to progress in G1, do not pass the restriction point and are arrested in G1 prior to enter a quiescent state. We examined this hypothesis by looking at the distribution of the percentage of Fucci-red and Fucci-green cells in the proliferative zone of the spheroid during this 6 days experiment. As presented in the micrographs shown in Figure 6A (middle and right panels) both the percentage of Fucci-red and Fucci-green positive cells are diminished after 6 days. The quantification presented in Figure 6B shows that percentage of Fucci-green (i.e. S and G2 cells) progressively dropped and is decreased by a two-fold factor after 6 days. However, unexpectedly about 15% of positive Fucci-green cells are still detected at day 6, while as indicated above EdU incorporation is totally abolished. The evolution of the percentage of Fucci-red positive cells is different. During the first three days, this percentage increases as compared to control spheroid, indicating that cells are accumulating in G1-phase. However, at days 5 and 6 we observe a major decrease in the percentage of Fucci-red positive cells, illustrating the entry of the cells in a quiescent stage accompanied by the loss of Fucci-red expression, that reproduces the decrease of the level of the endogenous Cdt1 protein in the same conditions (data not shown).
Our results strongly suggest that epidermal growth factor removal slows down cell cycle progression not only in G1 but also in G2. The progressive changes in Fucci-green and -red cells percentage thus would likely reflect a combined effect and its progressive consequences on the cell cycle distribution.