Real-time monitoring of photocytotoxicity in nanoparticles-based photodynamic therapy: a model-based approach

Abstract

Nanoparticles are widely suggested as targeted drug-delivery systems. In photodynamic therapy (PDT), the use of multifunctional nanoparticles as photoactivatable drug carriers is a promising approach for improving treatment efficiency and selectivity. However, the conventional cytotoxicity assays are not well adapted to characterize nanoparticles cytotoxic effects and to discriminate early and late cell responses. In this work, we evaluated a real-time label-free cell analysis system as a tool to investigate in vitro cyto- and photocyto-toxicity of nanoparticles-based photosensitizers compared with classical metabolic assays. To do so, we introduced a dynamic approach based on real-time cell impedance monitoring and a mathematical model-based analysis to characterize the measured dynamic cell response. Analysis of real-time cell responses requires indeed new modeling approaches able to describe suited use of dynamic models. In a first step, a multivariate analysis of variance associated with a canonical analysis of the obtained normalized cell index (NCI) values allowed us to identify different relevant time periods following nanoparticles exposure. After light irradiation, we evidenced discriminant profiles of cell index (CI) kinetics in a concentration- and light dose-dependent manner. In a second step, we proposed a full factorial design of experiments associated with a mixed effect kinetic model of the CI time responses. The estimated model parameters led to a new characterization of the dynamic cell responses such as the magnitude and the time constant of the transient phase in response to the photo-induced dynamic effects. These parameters allowed us to characterize totally the in vitro photodynamic response according to nanoparticle-grafted photosensitizer concentration and light dose. They also let us estimate the strength of the synergic photodynamic effect. This dynamic approach based on statistical modeling furnishes new insights for in vitro characterization of nanoparticles-mediated effects on cell proliferation with or without light irradiation.

Figure 1. Dark cytotoxicity without light exposure of nanoparticles-grafted photosensitizer (NP-PS) or control nanoparticles (NP) using MTT assay.

MDA-MB-231 cells were exposed to NP-PS (dark grey) or NP (clear grey) at the mentioned concentrations for 24 h. Cell viability was evaluated by MTT assay (data points show the mean ± S.D., n = 6).

Figure 2. Normalized cell index (NCI) kinetics of the MDA-MB-231 cells exposed to nanoparticles without light irradiation.

The cells were exposed to the indicated concentrations of (A) nanoparticles-grafted photosensitizers (NP-PS), or (B) photosensitizer-free nanoparticles (NP) for 24 h before washing. Cell Index (CI) was monitored during 143 h after nanoparticles exposure. Reported data are the means of six replicates.

Figure 3. Canonical representation of a multivariate ANOVA with respect to the measurement time instants.

The two canonical axes contain more than 95% of the total information contained in the time profiles over the study range [25–120] h presented in Figures 4A–B. According to this synthetic representation, there are two significant time regions of interest, described by blues and red ellipses. These two regions correspond to the most distant time points (with respect to the initial time instant t₀) for the two axes. These two time regions of interest are reported in the time plots of Figures 4A–B with the same color code. Data used in this statistical analysis have been previously normalized by fixing the cell index (CI) values at 1 at initial time t = 0 for all the cell cultures.

Figure 4. Study area [25–120] h of the normalized cell index (NCI) kinetics for different concentrations of nanoparticles.

MDA-MB-231 cells were exposed to various concentrations of A) nanoparticles-grafted photosensitizer (NP-PS) or B) photosensitizer-free nanoparticles (NP) as described in Figure 2. Two time regions of interest, colored in blue and red, were identified by a statistical analysis (results in Figures 3A–B) as the most informative ones: around 45 h and 120 h after nanoparticles exposure. At t₀ = 25 h (beginning of the study area) the NCI values are normalized at one.

Figure 5. Analysis of variance of the normalized cell index (NCI) values at times T1 = 45 h and T2 = 120 h.

This analysis was performed with respect to the indicated six concentration groups of photosensitizer into NP-PS or the six concentration groups of nanoparticles without photosensitizer (NP) at times T1 = 45 h (left panel) and T2 = 120 h (right panel).

Figure 6. Kinetics of photo-induced cytotoxicity of nanoparticles-grafted photosensitizer (NP-PS) according to real-time impedance analysis.

MDA-MB-231 cells were monitored for 24 h during interaction with NP-PS at the indicated concentrations of photosensitizer (left panel) before washing and light irradiation at 1 J/cm² (A), 5 J/cm² (B) or 10 J/cm² (C) (right panel). Presented cell index (CI) values are the mean of 6 replicates.

Figure 7. Photo-induced cytotoxicity of nanoparticles-grafted photosensitizer using impedance analysis and WST-1 test at 24 h post-irradiation.

For impedance-based analysis (A), the MDA-MB-231 cells were exposed to various concentrations of nanoparticles-grafted photosensitizer (NP-PS) (from 0.05 to 10 µM) for 24 h followed by a washing step before exposition to the indicated doses of light. At 24 h post-irradiation, WST-1 test (B) was carried out on the same E-Plate. Data are presented as the mean ± SE of the mean, (n = 6).

Figure 8. Exponential-Linear model structure of the transformed cell index (TCI) profile.

The TCI is decomposed into two parts: a transient and a steady-state periods. This model is composed of three parameters (T,k,r) used as quantitative indicators in the therapeutic efficiency analysis. T, time constant of the transient phase; K, magnitude of the transient decrease; r, steady-state growth rate.

Figure 9. Model quality assessment.

Comparison of measured (+) and predicted (−) responses for 16 cases among 144 time profiles of the transformed cell index (TCI). These 16 profiles sum up the variability of the cell response profiles observed between all the cell cultures.

Figure 10. Synergic effects between the photosensitizer concentration, C, and light fluence, F, on the model parameters.

The three model parameters (T, r, k) were analyzed with respect to C.F values. A) Any reduction of the time constant T suggests a faster decrease of the transient phase (positive therapeutic effect). Almost all the values of T below a reference threshold fixed to log(T) = 0 correspond to a synergistic condition defined by C.F ≥5. B) Any reduction of the transformed cell index (TCI) growth rate r suggests a slow down of the steady-state growth rate during its post-transient phase (positive therapeutic effect). Conversely, any increase of it leads to locally degrading the therapeutic response. Almost all the values of r below a reference threshold fixed to r = 0 (no steady-state growth) correspond to a synergistic condition defined by C.F ≥5. C) Any positive enhancement of the magnitude of the transient decrease k suggests a deeper decrease of the TCI during its transient decrease (positive therapeutic effect). Conversely, any decrease of it leads to locally degrading the therapeutic response. The largest values of k all correspond to large values of the interaction between the concentration and the fluence.

Figure 11. Profile of the photodynamic effects based on the time constant (T) and the growth rate (r).

This synthetic representation describes both transient and steady-state effects. Corners correspond to four distinct scenarios of the therapeutic responses. The bottom left-hand corner (C) is the ideal case: a rapid decrease followed by a null steady-sate growth. All the estimates of T and r are projected in this map. Two distinct groups are described by red and blue regions. The most efficient group (red group) corresponds to the values of the concentration-fluence interaction C.F ≥2.5.