**4. Discussion**

HepG2 cells cultured in 2D- and 3D-forms are used as experimental model in various studies related with hepatic functions and pathology, drug delivery and safety, toxicology and others [11,12,14,15]. Such a wide range of applications is due to the unique properties of this cell line, which reproduces structural and functional features of the liver, in particular, in 3D-form (spheroids), which is reviewed in [7,9,11]. Liver hepatocytes and HepG2 cells possess unique structural and functional polarity, it was interesting to compare their features with another cell line, representing "standard columnar" polarity, and we used HEK293 cell line for that. Using the TEM, we compared not only morphology of HepG2 and HEK293 cells in monolayers and spheroids, but also their interaction with three types of gold NPs differing by a coating nature, hydrodynamic size, and net charge.

In monolayer, HepG2 cells formed bile capillaries with typical microvilli and tight junctions, thereby showing liver-specific features. In contrast, HEK293 cells did not show any tissue-specific signs. In spheroids, both HepG2 and HEK293 cells formed outer surface by basal plasma membrane, which contacts with exterior environment (culture medium), and this finding can be important for understanding of the mechanisms of experimental influences. Our data show that spheroids consist of structural blocks, formation of which is specified by bile capillaries in case of HepG2 cells, and conglomerates of cell apical parts in case of HEK293. In HepG2 spheroids, blocks are separated by pseudosinusoids and intercellular spaces formed by lateral membranes, but more research is needed to find out whether pseudosinusoids form a common network within the spheroid. More research is also needed to understand how the blocks interact with each other; nevertheless, at present we can definitely say that the cells do not form shaped layers in HepG2 and HEK293 spheroids.

AuNPs are considered a promising platform for development of targeted nanomedicines and are extensively explored; however, there are many unknown details in their interaction with a cell [1–6]. We examined internalization of positively charged AuPEI-NPs, and negatively charged AuNPs and AuBSA-NPs with HepG2 and HEK293 cells in monolayers and spheroids to learn more about the mechanisms of their internalization. To our surprise, AuNPs did not penetrate HepG2 cells either in the monolayer or in the spheroids. Published TEM studies of AuNPs penetration into HepG2 cells were conducted in presence of serum [23], which contains proteins forming corona and so change a pattern of AuNPs interaction with a cell. However, our data corresponded to published TEM data that AuNPs (10 nm) did not penetrate hepatocytes in mouse liver, their uptake was detected in Kuppfer and endothelial cells (in contrast to AgNPs which penetrated the hepatocytes [4]. More research is needed to explain why hepatocytes and HepG2 cells "ignore" AuNPs, but an important point following from this observation is that cells can possess a selectivity for at least one type of the NPs. In connection with the selectivity of cells, it is pertinent to note the known effect of the size of AuNPs on penetration into cells. Thus, size-dependent accumulation of AuNPs (2–15 nm) coated with tiopronin was shown for MCF-7 breast cancer cells and their multicellular spheroids, and tumors in mice [61]. Different penetration rates were also reported for AuNPs (50 and 100 nm) coated with thiopronine in the same experimental MCF-7 cell models [62].

Interest to penetration of a drug into spheroid tissue is related with known block of the diffusion in tissue of solid tumors in vivo influencing drug effect [63]. Published studies show that the penetration of various NPs into HepG2 and other spheroids is limited to a depth of 20–50 μm, despite a fairly long incubation (for 24–72 h) [10,15,64]. The same values of penetration into HepG2 spheroids were reported for sorafenib [14]. Examination of MCF-7 spheroids treated with doxorubicin, revealed a dependence of penetration on spheroid sizes and ability of the drug to completely penetrate into the small-size spheroids [65]. Ability of doxorubicin for complete penetration into C3-HepG2 spheroid during 24 h was shown using confocal microscopy [45]. Our study showed that all studied NPs penetrated to a depth of 30–40 μm into HepG2 and HEK293 spheroids during 4 h of incubation. Limited penetration of NPs into spheroids is a "good" feature when a research is devoted to antitumor drugs, however, it may influence the results of toxicological and other studies.

Obviously, the use of spheroids in NPs research will expand and the methods of their cultivation and study of experimental effects will be increasingly improved. Presently, direct visualization of a NP in a cell and identification of cell structures is possible only by TEM of ultrathin sections. Currently, electron microscopy is overshadowed by other methods, faster and sometimes simpler, however, do not allowing to see directly nanoparticles in cells. Application of TEM allowed us to obtain a new data about features of HepG2 and HEK293 morphology in monolayer and spheroids, and clarify details of AuNPs, AuBSA-NPs and AuPEI-NPs uptake by these cells.
