3.1. In Silico Characterization of R-Cla Derivatives
It is important to verify if all the proposed R-Cla derivatives have an intrinsically available pathway for
T1 chemiexcitation during their chemiluminescent reaction, which can be enhanced by the heavy-atom effect, before their synthesis and characterization. Thus, we calculated the potential energy curves for both the
S0 and
T1 states during the thermolysis of OH-Cla (
Scheme 1 and
Figure 1), with a density functional theory (DFT) approach [
11,
17]. Similar calculations were previously performed for Br-Cla dioxetanone [
12] (
Figure 1). Therefore, OH-Cla dioxetanone was chosen to be studied as a reference, since OH-Cla and Br-Cla are on opposite sides regarding the expected enhancement of ISC rate due to the heavy-atom effect.
As consistent with previous reports for different dioxetanones (including Br-Cla) [
6,
9,
11,
12,
17], the
S0 thermolysis of OH-Cla dioxetanone proceeds via a stepwise biradical mechanism that involves the cleavage of two bonds of the peroxide ring: first by the breaking of the O-O bond, followed by C-C bond cleavage. The
S0 activation energy for the thermolysis reaction of OH-Cla dioxetanone is 24.7 kcal·mol
−1, only 0.8 kcal·mol
−1 lower than that of Br-Cla (25.5 kcal mol
−1) [
12]. The interplay between
S0 and
T1 states during the thermolysis reaction (
Figure 1) is more important. While, for both dioxetanones, the
S0 –
T1 energy gap is significant at the beginning of the reaction (~75 kcal·mol
−1), when reaching the TS and onward, both states become degenerated in a large and flat region of the potential energy surface (PES).
This indicates that both chemiluminescent reactions possess an intrinsic pathway for
T1 chemiexcitation, in line with previous studies for this type of system [
9,
11,
12,
18]. However, as ISC is a spin-forbidden process, the
S0 →
T1 chemiexcitation pathway for OH-Cla is not expected to be efficient [
19]. Given the identical energetic profiles found for OH-Cla and Br-Cla (
Figure 1), the addition of halogen atoms of increasing size in Cl-Cla and Br-Cla should only enhance the rate of the ISC process, due to the heavy-atom effect. Thus, this substitution appears to be ideal to assess solely the heavy-atom effect on the possible anticancer activity of these molecules.
It should be noted that an efficient ISC is not the only parameter determining the performance of a photosensitizer in singlet oxygen generation, as it should also have a
T1 state with an energy higher than 0.98 eV (the required energy to convert molecular oxygen into singlet oxygen) [
20]. We modeled the adiabatic
S0 –
T1 energy gap of OH-Cla coelenteramide and found a value of 2.78 eV. Furthermore, this value is identical to what was found by us previously for Br-Cla [
12]. Thus, both OH-Cla and Br-Cla compounds possess enough energy to generate singlet oxygen. It should be noted that this value was obtained with the M06-2X functional and the 6-31 + G (d,p) basis set, in implicit water, using the SMD model. This approach was used to ensure consistency between studies, as this was the approach used before to calculate the
S0 –
T1 energy gap of Br-Cla coelenteramide [
12].
3.3. Luminometric and Photophysical Characterization
After synthesizing the target derivatives and completing their structural characterization, it is essential to perform their luminometric and photophysical characterization. Specifically, before assessing the potential role of the heavy-atom effect in the anticancer properties of these compounds, it is required to find out if the introduction of heteroatoms of increasing size enhances the T1 chemiexcitation of these compounds. Thus, we subjected these compounds to a detailed luminometric and photophysical characterization.
First, we measured the chemiluminescent output of the three Cla derivatives in an aprotic solvent (DMF) with addition of buffer (acetate buffer, pH 5.14), conditions in which Clz and derivates are known to readily generate chemiluminescence by reacting with dissolved oxygen [
17]. All compounds emitted chemiluminescence (
Figure 2a and
Figure S18) with the typical kinetic profile, with a quick rise in light emission on the millisecond timescale and subsequent decay to basal levels (all within 600 ms). Interestingly, there is a clear halogen substitution effect in which the light emitted by Cl- and Br-Cla is significantly lower than that emitted by OH-Cla. Furthermore, among halogenated compounds, the light output decreases in the order of OH > Cl > Br. In solution, triplet states are generally more easily quenched than singlet excited states, thereby not leading to light emission in solution at room temperature. Thus, the quite lower light output of halogenated Cla compounds is indicative of the halogen’s ability to enhance ISC during dioxetanone’s thermolysis, by increasing the triplet-to-singlet product ratio of the studied chemiluminescent reactions [
19,
21].
One other possible explanation for these variations in intensity could be that the introduction of the halogen atoms decreased the energetic favorability of the
S0 reaction. However, in our previous study of Br-Cla [
12], we already found that bromination does not impede the reaction, as it is still highly exothermic (−97.1 kcal·mol
−1). Furthermore, the
S0 energetics for the thermolysis of OH-Cla and Br-Cla dioxetanones are identical (
Figure 1), which means that halogenation should not affect the efficiency of the
S0 chemiluminescent reaction.
Further support for this conclusion was obtained by analyzing the chemiluminescent kinetic profiles of these compounds (
Figure S18). There were no significant differences between compounds, with the kinetic profiles showing identical rises and subsequent decay of light emission in the same timeframe, showing that halogenation does not affect the kinetics of the reaction. The steady-state chemiluminescent spectra (measured during the chemiluminescent reaction in a spectrofluorometer, without the use of an excitation source) were identical for all R-Cla compounds (
Figure 2c), with emission maxima at ~480 nm. Lastly, we also measured the 2D excitation-emission matrices (EEMs) for the spent reaction mixtures (30 min after addition to aprotic solvents) of the three compounds (
Figure 3). The resulting EEMs were quite similar to each other, with just one emissive center with an excitation wavelength maximum at ~280 nm. The main difference is only a small blue-shift of ~25 nm in the emission maxima between OH-Cla (~425 nm) and Cl-/Br-Cla (~400 nm). Given this high similarity between EEMs for the spent reaction mixtures, we can conclude that halogenation does not affect the outcome of the chemiluminescent reaction in terms of obtained products.
Lastly, a possible explanation for the lower light output, other than the increase in
T1 chemiexcitation, could be that halogenation decreases the singlet emission efficiency of the resulting chemiluminophore (coelenteramide). To verify this hypothesis, we also measured the fluorescence intensity of the spent reaction mixtures (after 30 min of reaction) in an aprotic solvent (
Figure 2d and
Figure 3). There is indeed a small decrease in emission intensity of the reaction product that correlates with halogen substitution. However, for one, this decrease is significantly lower than that found for the chemiluminescent output (
Figure 2a). Furthermore, given that the decrease in emission intensity occurs in the order of OH > Cl > Br, this also indicates a heavy-atom effect. More specifically, this observation is consistent with fluorescence quenching due to ISC enhancement, during the photo-excitation process. Thus, these results also support the conclusion that the decrease in chemiluminescent light output is due to the heavy-atom effect, which enhances the
T1 chemiexcitation and increases the resulting triplet-to-singlet product ratio.
Having demonstrated that the addition of halogens enhances the production of triplet states, it is also important to assess the ability of the superoxide anion to trigger the chemiluminescent reaction of the studied molecules. This type of ROS is overexpressed in cancer cells [
22,
23]; thus, it could be thought of as a cancer marker inducing chemiluminescence and a tumor-selectivity profile to our molecules. It should be noted that we previously demonstrated our compounds as having selectivity for breast cancer, as Br-Cla induced significant toxicity toward breast cancer cell lines without affecting non-cancer cell lines in the same concentration range [
12].
The ability of superoxide anions to trigger the chemiluminescent reaction of these compounds was demonstrated by measuring their chemiluminescent output, after KO
2 addition (a known superoxide anion source) in methanol. Interestingly, all compounds responded to the addition of KO
2 with immediate emission of light and subsequent decay to basal levels (
Figure S19). The kinetics of this reaction is significantly quicker than in aprotic solvents, but this difference can be attributed to the instability of KO
2 in solution. The more relevant result is that all compounds responded similarly to superoxide anion, meaning that they can be activated by a cancer marker. It should also be pointed out that the measured chemiluminescent output showed a substitution-induced effect, in the order OH > Cl > Br (
Figure 2b). Once again, this points to halogen substitution increasing the triplet-to-singlet ratio.
3.4. In Vitro Cytotoxic Activity of R-Cla
The next step of this study was the evaluation of the in vitro anticancer activity of these compounds. Both the in silico modeling and the luminometric/photophysical characterization showed that the introduction of halogens enhances the superoxide anion-triggered generation of triplet states (with enough energy to sensitize singlet oxygen), without affecting the chemiluminescent reaction. Thus, if the anticancer activity found before for this type of molecules [
12,
13] is indeed related to a light-free and self-activating photodynamic effect, we should see a halogen-dependent effect on their anticancer activity.
Given this, the potential anticancer activity of the three R-Cla compounds was evaluated in both colon cancer (HT-29) and neuroblastoma (SH-SY5Y) cell lines with a standard MTT assay for exposure times of 48h, along with changes in cell morphology analyzed by microscopy.
OH-Cla did not have any effect on the viability of colon cancer cells (
Figure 4) and did not cause any changes in their morphology (
Figure 5b). This indicates that OH-Cla has no anticancer activity toward these cells, which is expected since, without the heavy-atom effect introduced by the halogens, there is no reason for efficient
T1 chemiexcitation. Interestingly, while Cl-Cla induced virtually no anticancer effect in these cells, there was a statistically significant difference at a concentration of 100 µM (
Figure 4). In addition, microscopic analysis at the two highest concentrations indicated that the cells were rounder and in lesser number, without the formation of aggregates (
Figure 5c). As for Br-Cla, there was a relevant decrease in cellular viability at the highest concentration (100 µM) (
Figure 4), with the morphological analysis showing a clear decrease in the number of cells at the highest concentration, as well as a change in the size and shape of the cells, which were smaller and rounder (
Figure 5d).
These results are interesting because the observed anticancer activity toward HT-29 cell lines was low, while we could observe a distinct halogen-dependent effect. Specifically, OH-Cla did not present any activity, while Cl-Cla and Br-Cla presented increasing activity. This is a first demonstration of the heavy-atom effect, which indicates that this anticancer activity is indeed related to a light-free and self-activating photodynamic effect.
It should also be noted that a positive control test was performed by exposing these cells to an antineoplastic drug commonly used in colon cancer therapy, 5-fluorouracil (5-FU), in the same concentration range as R-Cla (
Figure S20). This drug induced a significant decrease in cell viability for all concentrations, while microscopic evaluation revealed that all cells appeared to be morphologically changed (
Figure S21a–c).
In neuroblastoma cells (
Figure 6 and
Figure 7), the results were similar and neither OH-Cla nor Cl-Cla altered cell viability in a relevant manner. On the contrary, Br-Cla (
Figure 6 and
Figure 7d) induced a significant decrease in cell viability, although not as marked as for 5-FU (
Figures S20b and S22), which was also used as a positive control for neuroblastoma cells. This compound induced a sharp decrease in cell viability. The cell viability results were consistent with microscopic evaluation (
Figure 7 and
Figure S22).
Given this, it is clear that there was a halogen-dependent effect on the anticancer activity of these compounds for both cancer cell lines. This is in line with our proposition that the addition of halogens to Clzs will provide them with anticancer activity, by enhancing
T1 chemiexcitation and the subsequent intracellular generation of triplet states capable of sensitizing singlet oxygen [
12,
13]. In conclusion, our results indicate that the anticancer activity of our Clz derivatives [
12,
13] is indeed related to the heavy-atom-induced triplet state generation during their chemiluminescent reaction. Therefore, it appears that we can consider our compounds as prototypical single-molecule-photosensitizers capable of an intracellular and self-activating photodynamic effect, which is triggered by a cancer marker.
Having said that, it is also important to note that the intrinsic anticancer activity of the studied compounds (including Br-Cla) might not be as high as desired toward neuroblastoma and colon cancer cell lines, especially when compared to 5-FU. Given this, it is important to determine whether the magnitude of the obtained anticancer activity is similar across different cancer types or if it is dependent on the studied cell line.
To clarify this topic, we calculated for the first time the IC
50 of Br-Cla for different cell lines. Br-Cla was chosen for its consistent toxicity in this study, contrary to OH-Cla and Cl-Cla. IC
50 values were determined for neuroblastoma SH-SY5Y, prostate PC-3, and breast MCF-7 cancer cell lines (
Table 1). We did not include assays with the HT-29 cancer cell line, due to the limited toxicity of Br-Cla toward it (
Figure 4). We included assays with PC-3 and MCF-7 cell lines due to previous promising results of Br-Cla and other Clz derivatives toward them [
12,
13]. For the assays with these two cell lines, we maintained the conditions employed before for consistency purposes [
13].
The impact of treatment with Br-Cla on the cellular viability of PC-3 and MCF-7 cell lines can be found in
Figure 8. The anticancer activity of Br-Cla was improved in these two cell lines; in breast MCF-7 cancer cell lines, there was a noticeable decrease in cell viability from 10 μM onward, whereas, in prostate PC-3 cancer cells, Br-Cla decreased the cell viability from 25 μM onward, achieving toxicity values higher than 50% at a concentration of 75 μM. This improved efficiency was confirmed by the determined IC
50 values (
Table 1). More specifically, for SH-SY5Y, the obtained IC
50 was more than double the values found for the other cell lines. Interestingly, the IC
50 values found for PC-3 and MCF-7 cells were similar, although slightly lower for the latter cell line. Furthermore, to discard the hypothesis that these differences could be related to the different duration of treatment with Br-Cla, we also determined the IC
50 for the MCF-7 cell line with 24 h treatment (
Table 1). This value (33.84 μM) was found to be significantly higher than that found for 72 h treatment (21.56 μM), which indicates that increasing the incubation time increased the obtained IC
50. More important is that the IC
50 found for SH-SY5Y (50.92 μM) was still significantly higher than the IC
50 found for MCF-7, irrespective of the duration of treatment with Br-Cla.
Thus, the results indicate that it is safe to state that the magnitude of the anticancer activity of Br-Cla is dependent on the cancer cell type, being higher for prostate and breast cancer than for neuroblastoma and colon cancer. Further studies are required to assess if these differences arise from (among other possibilities) higher resistance of HT-29 and SH-SY5Y cell lines to the photodynamic effect, lower efficiency of internalization of R-Cla compounds into these cell lines, or a lower generation of superoxide anion in these cells.