Glioblastomas are malignant tumours that commonly occur in adult humans [1]. The incidence of these tumours increases with the age of the patients, and the survival rate is still low after diagnosis [2,3]. The main proposed treatment is surgery supported by radiotherapy and chemotherapy [4].

Advanced computed tomography, magnetic resonance imaging, and ultrasound methods are useful for neuronavigation to localize the tumour during surgery [5], [6], [7]. However, photodiagnostics, which makes use of fluorescence of light-sensitive molecules targeting to glioblastoma cells, provides another way to examine small infiltrating areas, which is not possible with magnetic resonance imaging [8]. In addition, photodiagnostics can be performed directly during resection [9,10]. Indeed, 5-aminolevulinic acid (5-ALA) is used in practise for fluorescence-guided surgery of glioblastomas [11]. This strategy can be used more than 4 h after the application of 5-ALA, which provides a large time window for intraoperative surgery [12]. However, not only 5-ALA and its derivatives are used in clinical trials for tumour identification, but also other fluorescent molecules such as fluorescein, indocyanine green and hypericin in various polymer and nano formulations have been proposed [13], [14], [15]. The applicability of these molecules is determined by the limitations arising from their origin. While fluorescein is mainly found in the extracellular areas of the tumour, protoporphyrin IX is produced in the cells when 5-ALA is applied. Spectral properties of these molecules are also different and due to the blue excitation of fluorescein and the detection in the range of 540–690 nm, it is preferable. Hypericin also has similar properties. In addition, the specificity of hypericin identification of tumour tissue is higher than that of fluorescein [16,17]. Due to its safer profile, easier administration and for financial reasons, sodium fluorescein has recently become the favourite contrast agent in preclinical studies. However, hypericin can be developed as a water-soluble formulation. In addition, Ritz and his group compared 5-ALA and hypericin and showed that hypericin has higher photostability and better penetration depth (10 mm) compared to 5- ALA (2–3 mm) compared to protoporphyrin IX [14,18,19].

While photodiagnostics uses light with a short wavelength (usually blue) [20], the same molecules can be used in photodynamic therapy, which is performed with light of longer wavelengths (in the red range) [21] and allows deeper penetration into the tissue.

The photodynamic reaction also leads to the formation of reactive oxygen species (ROS) of type I (superoxide, hydroxyl radical and hydrogen peroxide) or of type II (singlet oxygen) [22]. Photodamaged tumour cells, often targeted by high selectivity of the photosensitive molecule, undergo autophagy and apoptosis due to the formation of ROS [23,24].

Cells grown in monolayers have been replaced by 3D models in recent decades because of their greater similarity to tumours, and the models in which cells grow in an aggregate to form spherical structures have been well accepted by the scientific community [25], [26], [27]. We have recently shown that U87MG cells can be formed into 3D structures by the hanging drop method [28]. This formation allows the production of spheres of different sizes that can be readily analysed by fluorescence techniques to the limits of detection defined by the light transmission of the object. Tight junctions between cells restrict the growth of neighbouring cells. For this reason, cells grown in spheroids suffer from oxygen and nutrient starvation [29,30]. It is very likely that the composition and functionality of the central part of the spheroid differs from that of the peripheral part. These differences not only affect nutrition, but may also influence the uptake of hydrophobic molecules.

One of the model molecules that has fluorescence and is active in photodynamic therapy is hypericin [31,32]. Hypericin is very special molecules that is not active in aqueous solutions, in which it creates nonfluorescent aggregates [33]. This aggregation was shown to have impact on hypericin fluorescence and its fluorescence decay in biomacromolecules and subcellular structures that subsequently defines photoactivity of hypericin in cells [31,34]. Due to the relatively high quantum yield of hypericin fluorescence and singlet oxygen production, hypericin is attractive molecule for photodiagnosis and photodynamic therapy of selected cancers [14,35]. In addition to the extensive screening of hypericin in 2D models, there are several studies that have investigated hypericin penetration behaviour and interaction with cells in spheroids [28,36,37].

In the present study, we prepared U87MG glioblastoma spheroids, which were examined by various imaging techniques to visualize heterogeneity in the spheroids. Fluorescence microscopy was used to study morphological changes and subcellular differences in the cells. Three main populations of cells were identified in the spheroid by flow cytometry. The main objective of the study was to identify autophagic and apoptotic cells in different layers of the spheroids before and after photodynamic therapy induced by hypericin.