The use of caspase inhibitors in pulsed-field gel electrophoresis may improve the estimation of radiation-induced DNA repair and apoptosis

The use of pulsed-field gel electrophoresis (PFGE) is widespread in the evaluation of DNA fragmentation caused by double-strand breaks (DSBs) following ionizing radiation [1‐4]. The DNA-DSBs may result in the formation of small (often acentric) chromosomal fragments. Following this initial damage, cells activate DNA repair mechanisms to prevent catastrophic mitosis and cell death due to the loss of acentric DNA fragments [5]. The principle of PFGE methodology is that the release of DNA from cells correlates adequately with the intensity of DNA fragmentation [6]. The estimation of DNA repair by PFGE is based on the diminution of DNA released from cells as the length of the DNA fragments increases through the process of rejoining. Thus, a decrease in the ratio of DNA extracted from the cells over a period of time can be used as an evaluation of DNA repair [7].

In the PFGE technique, cells are encapsulated in agarose to form cell-plugs, thus preventing physic damage of the cells while facilitating their manipulation and placement into agarose gels where electrophoresis will take place. Usually in laboratory routine, cells are encapsulated after a period of repair which is allowed to occur in physiological conditions such as either cell cultures or xenografts. Thus, extraction ratios depend exclusively on induced and repaired DNA damage. While the desired strategy is to encapsulate cells after the period of repair has finalized, in a clinical setting, where the availability of cells is limited by the small size of tumour biopsies, it is crucial to concentrate cells in agarose plugs-- to obtain enough cell number for PFGE analysis -- before irradiation [8]. However, this tactic means that DNA breakage and repair occurs in a non-physiological environment. Whitaker and McMillan reported in 1992 that encapsulating cells before irradiation impair the estimation of DNA repair due to the interference of spontaneous DNA fragmentation [8]. This pioneering observation has been confirmed by other studies [9] leading to the belief that results from these sorts of studies (conditions) are not robust enough to properly estimate DNA repair. Nevertheless, the underlying mechanisms affecting cells embedded in agarose during DNA repair in PFGE methodology are still not well understood.

To better understand spontaneous DNA breakage--whether cells are irradiated or not--while cells are encapsulated, we decided to examine PFGE outcomes in cell-plugs over a period of incubation. Our main findings were that 1) incubation of agarose-encapsulated cells induces DNA fragmentation, 2) spontaneous DNA breakage is caused by apoptosis (which can be inhibited by caspase inhibitors), and 3) reducing the interference of spontaneous breakage improves our ability to estimate DNA repair and to determine apoptosis intensity.

Irrespectively whether cells were irradiated or not, we found that the pattern of DNA smears consisted of a compression zone (CZ) just below the wells, which was more patent in irradiated cells, followed with an area of distribution of DNA fragments ending in a sudden edge (Figure 1). The GelDoc settings for fluorescence acquisition were adjusted (exposure time and iris aperture) in each gel to obtain smears below saturation levels but sufficiently high to be measured. Thus, the range of fluorescence could vary depending on whether the gel contained unirradiated or irradiated cell plugs. In unirradiated cell-plugs, we observed an initial low extraction of DNA which progressively increased over time. In the A431 cell line, this spontaneous DNA fragmentation increased significantly over the incubation time (Figure 1), and similar figures were seen in the NP18 cell line (data not shown). In sharp contrast, in irradiated cell-plugs the initial DNA extraction was higher than in unirradiated cell-plugs indicating that radiation-induced DNA breakage occurred in a manner quantifiable by PFGE. Moreover, in this irradiated plugs we found a time-dependent decrease in DNA fragmentation compatible with the process of rejoining (Figure 1). Although some rejoining might have taken place in this experimental setting, proper estimation of DNA repair was clearly interfered by spontaneous DNA fragmentation, logically occurring in both unirradiated and irradiated cell-plugs.

Given that agarose is a non-physiological environment, we hypothesized that spontaneous DNA degradation observed in PFGE could be a consequence of apoptosis triggered by agarose associated with the loss of normal cell-matrix interactions. To evaluate this possibility we examined apoptosis activation in cells embedded in agarose. Thus, we decided to determine cleavaged caspase-3 levels, an apoptosis effector, in cell-plugs (Figure 2A). Cleavaged caspase-3 was gradually activated regardless of whether the cells were irradiated or not, suggesting that apoptosis was triggered by agarose environment (Figure 2B). To further confirm that the origin of apoptosis was the encapsulation in agarose, we investigated the activation of caspase-3 in cells growing as monolayer (in a physiological environment). We observed only a minimal activation of apoptosis after 45 Gy (of 174 cells evaluated, 6 were caspase positive after 4 hours of incubation time: 3.44% of cell population), a finding that also precludes a substantial radiation-induced origin of apoptosis and corroborates the participation of agarose in spontaneous DNA fragmentation (Figure 3). To study whether apoptosis and repair could coexist we determined simultaneously the activation of caspase-3 and histone γH2AX by a double staining technique. Immediately following irradiation we observed high levels of γH2AX induction and low levels of cleavaged caspase-3 (Figure 2A). However, over incubation time we found that a fraction of apoptotic cells was simultaneously visible with a relevant fraction of cells in which the γH2AX was present (Figure 2A). As we were able to exclude caspase-positive cells from γH2AX fluorescence measurements we could specifically determine the evolution of DNA-DSBs. We found that initial radiation-induced γH2AX fluorescence decreased significantly over the incubation time, suggesting that in those cells in which apoptosis was not triggered radiation-induced damage could be repaired (Figure 2C). Interestingly, while the intensity of γH2AX decreased in irradiated cell plugs, the γH2AX levels in unirradiated plugs did not vary, indicating that γH2AX levels in those cells were not significantly influenced by agarose (Figure 2C).

To further verify that the spontaneous DNA fragmentation observed in the PGFE experiments was induced by apoptosis, we treated cell-plugs with the pancaspase inhibitor z-vad-fmk. First, we corroborated that merely incubating cell-plugs was sufficient to induce DNA fragmentation in a time-dependent manner (Figure 4). We next found that treatment of these cell-plugs with the anti-apoptotic agent inhibited progressive spontaneous DNA degradation, confirming the crucial role of apoptosis and corroborating the mechanistic explanation of the caspase-3 activation we demonstrated by immunofluorescence method in cell-plugs (Figure 4).

Since encapsulating cells before irradiation is particularly useful in a clinical setting with tumour biopsies, our aim was to investigate whether the inhibition of apoptosis would facilitate the evaluation DNA repair in this context. To do this, we removed tumours (< 1000 mm³) derived from NP18 cells from the subcutaneous tissue of nude mice. In plugs containing cells from xenografts we reproduced the DNA fragmentation pattern previously observed in PFGE of cultured cells (shown in Figure 1). In irradiated cell-plugs, radiation-induced DNA-DSB decreased progressively, whereas spontaneous fragmentation increased gradually (Figure 5). However, when cell-plugs were treated with zvd-fmk, a specific caspase-3 inhibitor, spontaneous DNA degradation was inhibited over the entire time-course (P > 0.3). Importantly, radiation induced DNA breakage, which is attributable to rejoining, continued diminishing (Figure 5). The blockade of apoptosis thus allowed us to efficiently determine rejoining without the inference of the apoptosis. Figure 5 illustrates the effect of abrogating apoptosis in NP18 cells from a xenograft, showing clear differences between initial and residual radiation-induced damage depending on the absence or presence of zvd-fmk. By inhibiting apoptosis, we were able to reduce spontaneous DNA breakage and estimate rejoining as the initial DNA released minus residual DNA released without the interference of apoptosis. On the other hand, the intensity of apoptosis could be easily estimated by subtracting the total DNA released in unirradiated zvd-treated cell-plugs from the unirradiated non-zvd-treated cell-plugs.

It is believed that specific tumour radiosensitivity is highly dependent on DNA repair and, reasonably, tumour response is also influenced by this biologic variable. Initially, PFGE was devised as a test for predicting cancer response to radiation therapy. However, soon thereafter the few studies that examined the utility of the PFGE method in tumour biopsies concluded that PFGE was insufficiently robust to be a predictive test, mainly due to spontaneous DNA fragmentation that occurs during sample processing. In the present study, we revisited the PFGE method to demonstrate that apoptosis induced by agarose is a component of total DNA breakage, and that the inhibition of apoptosis allows for more proper estimation of radiation-induced DNA DSB repair and apoptosis itself. First, we found spontaneous cleavage of caspase-3 in the cell-plugs, suggesting that the agarose environment triggered apoptosis activation, and second we inhibited spontaneous DNA fragmentation in PFGE by means of selective and specific caspase inhibitors. This is the first study to describe that the spontaneously released DNA in cell-plugs reflects an ongoing homeless-induced apoptosis, referred to as anoikis [17], when cells are embedded in agarose. DNA cleavage into large fragments is an early event observed in the apoptotic cascade before the typical endonuclease cleavage into 180 to 200 bp can be detected as a DNA ladder [18]. This DNA breakage is exactly what we observed in our experiments, and further supports the apoptotic aetiology of spontaneous degradation. Furthermore, the fact that we did not detect caspase-3 activation in irradiated cells growing as a monolayer, and caspase-3 activation was seen only in cells embedded in agarose (whether irradiated or not), allows us to rule out a radiation-induced origin of apoptosis.

After elucidating the mechanism behind DNA degradation, it seemed reasonable to conclude that apoptosis initiation was occurring in both unirradiated and irradiated cell-plugs. However, our findings show that γH2AX fluorescence, a surrogate marker of DNA-DSB [19], continued to decrease progressively, a result compatible with radiation-induced DNA-DSB repair. We should keep in mind that during incubation, apoptosis activation was triggered in a percentage of cells, whereas in non-apoptotic cells repair took place, as we were able to show by eliminating apoptotic cells from the quantification of fluorescence in cell-plugs. Thus, in our PFGE we obtained smears that represent the final stage of two independent processes--DNA repair and anoikis--at each time point.

This type of apoptosis is physiologically induced to prevent proliferation of cells at inappropriate locations. Tumour cells that lose this homeostatic control can manage to thrive in an anchorage-independent manner, an aggressive characteristic of different types of human malignancies. The clinical significance of resistance to anoikis is increasingly associated with malignant phenotypes, therapy resistance, and poor prognosis [20].

Finally, by incubating cell plugs in absence or presence of caspase inhibitors we were able to determine two traits associated with tumour aggressiveness, DNA repair and apoptosis. Reasonably, in a short time-course scheme, a relatively low DNA release rate will indicate an aggressive phenotype because the majority of radiation-induced DNA fragments are quickly rejoined (resistance to radiation) and apoptosis is not triggered (resistance to apoptosis). In this scenario, cancer cells would survive in an adverse environment (i.e., loss of extracellular matrix due to radiation cell killing) and would rapidly recover their proliferative potential. On the other hand, a significant smear would indicate a less aggressive behaviour due to low radiation repair and high sensitivity to anoikis.

Our study confirms previous findings that apoptosis occurring in agarose-encapsulated cells interferes with PFGE evaluation of radiation-induced DNA repair analysis. However, we found that it is possible to reduce this interference by using caspase inhibitors, thereby greatly improving the estimation of DNA repair in tumour cells. Simultaneously, apoptosis induced by agarose can also be determined. The ability to determine together two traits--repair and apoptosis--involved in cell fate opens new possibilities for PFGE as functional assay.