**4. Discussion**

The uPA system is recognized to play a central role in the ability of breast cancer cells to escape the primary site of the tumor and colonize other parts of the body. Hence, targeting this system using novel approaches might be effective in the treatment of highly invasive or metastatic breast cancer. This work aimed to improve upon the solubility and delivery of the potent *N*-alkylisatin (*N*-AI) cytotoxin to uPA/uPAR positive breast cancer cells, through the conjugation of PAI-2 to the surface of PEGylated *N*-AI-loaded liposomes.

The successful encapsulation of a hydrophobic drug into liposomes can greatly enhance the aqueous solubility and bioavailability of the molecule, and therefore increase the suitability for its use in parenteral drug delivery applications. With a logarithmic octanol/water partition coefficient (LogP) of 3.39, the aqueous solubility of non-liposomal *N*-AI is negligible. In this work, the thin film hydration method was successfully used to load *N*-AI into the bilayer of soy PC PEGylated liposomes, to achieve a final *N*-AI concentration of 2.2 mM in aqueous solution. *N*-AI was substituted for cholesterol in the formulation, in order to increase the drug-loading capacity of *N*-AI in the bilayer. As the molecular weight of *N*-AI and cholesterol were similar, and both were hydrophobic molecules, we achieved a greater encapsulation of *N*-AI in the liposomes, in the absence of cholesterol, without affecting the zeta potential, liposome size, or stability (Supplementary Information Figure S5).

The zeta potential of liposomes is dependent on a number of factors [38], while PEG itself does not affect the surface charge of liposomes, PEG-DSPE introduces a negative surface potential due to the phosphate diester moiety [39]. In this study, liposomes displayed a slightly negative, but near-neutral, zeta potential, which did not vary greatly with *N*-AI encapsulation or PAI-2 conjugation to the liposome surface. Although it was noted that PAI-2-functionalized liposomes were slightly more negative than the non-functionalized liposomes, as PAI-2 has a predicted isoelectric point of 5.4 and therefore a negative charge at physiological pH [40]. It was reported that liposomes with mildly charged or near-neutral surfaces show a propensity to aggregate faster than liposomes with a strong surface charge, as the latter show a greater particle–particle repulsion and hence are more electrostatically stabilized in suspension [41]. However, our liposome formulations were found to be remarkably stable with no aggregation detected or drug leakage under the storage conditions for >4 weeks (data not shown).

The optimization of surface-functionalized liposomes through the covalent attachment of targeting moieties to the outer lipid leaflet is complex. According to a recent study by Herda et al., only 3.5% of proteins attached to the surface of SiO2−PEG8−Tf particles had appropriate orientation for receptor recognition [42]. Furthermore, increasing antibody density on the surface of nanoparticles was found to reduce receptor-dependent targeting [43]. In this study, the conjugation of PAI-2 to the surface of liposomes was confirmed by Western blotting and the maintenance of PAI-2 inhibitory activity post-conjugation was validated by uPA assay. PAI-2-liposomes were as effective at inhibiting the enzymatic activity of uPA as the unconjugated PAI-2, demonstrating that PAI-2 liposomes were fully active. While Western Blotting was successful in qualitatively confirming the conjugation of PAI-2 to liposomes and in confirming the absence of unconjugated PAI-2 in the purified liposome, the use of emerging methods such as single molecule fluorescent imaging to quantify ligand density on the liposome surface would enable a more precise characterization of targeted liposome formulations [44].

As reported previously, cell surface uPA and uPAR expression is low in MCF-7 cells relative to the MDA-MB-231 cells [12,45]. This difference in uPAR expression was associated with a significant increase in fluorescently labelled EMP PAI-2 liposome uptake, relative to the non-functionalized EMP liposomes by MDA-MB-231 cells, but not by MCF-7 cells. Liposomes can be taken into cells via several different mechanisms, including adsorption, lipid exchange, intracellular membrane fusion, and receptor-mediated endocytosis (RME) [46]. The presence of a fluorescent signal from cells treated with EMP fluorescently labelled liposomes indicates that these liposomes were taken up into cells by fusion or other non-specific mechanisms, rather than by RME. In contrast, the uptake of EMP PAI-2 liposomes by the MDA-MB-231 cells was greater than the uptake of EMP liposomes in MDA-MB-213 cells. As the average liposome diameters of the FITC-labelled EMP and EMP PAI-2 liposomes were equivalent, this difference in uptake was likely due to the presence of PAI-2 at the liposome surface and interaction with uPA/uPAR overexpressed on the surface of MDA-MB-231 cells. Competition binding studies using excess PAI-2 or uPAR antibody could be used to further confirm this mechanism [47,48], however, the uPA/uPAR-dependent uptake of PAI-2 has been extensively characterized by our group [18–20,49] and others [50]. Studies assessing the colocalization of liposome signal with lysosomes through confocal microscopy, further confirmed that liposomes were indeed internalized by cells and accumulated in lysosomes, in addition to being present elsewhere in the cell. This result was not unexpected, given that the liposomes were incubated with cells at a high phospholipid concentration, creating an environment where liposomes in solution would passively fuse with cell membranes over time [46]. Collectively, our results indicate that EMP and EMP PAI-2 liposomes are taken into cells via a number of mechanisms, including RME.

In the multicellular tumor spheroid experiments, imaging of spheroids over 96 h indicated a concentration- and time-dependent destruction of both MCF-7 and MDA-MB-231 spheroids, when treated with *N*-AI PAI-2 liposomes. However, MDA-MB-231 spheroids treated with *N*-AI PAI-2 liposomes showed a greater destruction of the spheroid architecture than MCF-7 spheroids at lower *N*-AI PAI-2 liposome concentrations and at earlier time-points. We posit that this effect was receptor-dependent, due to the uPA/uPAR targeting of PAI-2 functionalized liposomes, but acknowledge that it might also

be the result of differences in cellular adhesion molecules involved in spheroid formation and cell–cell interactions between the two cell lines [51], thereby influencing liposome perfusion. While spheroid models are increasingly used to study anti-tumor drug effects, a major limitation is that they mimic only the avascular region of in vivo tumors, and exclude important aspects of tumorigenesis, such as the vasculature, immune system, and fluid dynamics. This is especially important in the context of evaluating the efficacy of nanomedicines, where the enhanced permeability and retention (EPR) effect is likely at play and vascular permeability is a relevant factor in their accumulation at the tumor site.

Determining the in vivo properties of novel nanotherapies is important for evaluating how a new nanoparticle formulation can be expected to perform in humans. In this work, the pharmacokinetics and tissue distribution of *N*-AI and *N*-AI PAI-2 liposomes were evaluated in an orthotropic MDA-MB-231 breast tumor xenograft mouse model. The addition of PAI-2 to the surface of *N*-AI-loaded liposomes did not significantly alter the in vivo blood clearance properties of the formulation, but did increase accumulation of liposomes at the primary tumor site relative to non-functionalized liposomes. The PEGylated liposome formulation contained 10 mol% PEG-DSPE and ranged between 130 nm and 150 nm in diameter. This liposome size range and PEG density has been reported to avoid rapid clearance by the mononuclear phagocyte system (MPS) in circulation, and to utilize the EPR effect to extravasate and accumulate at the site of tumors for drug delivery [42]. The accumulation of nanoparticles in tumors via EPR is dependent on a number of factors, including interstitial fluid pressure, vascularity of the tumor, and the in vivo circulation time of the nanoparticle formulation [52]. As the plasma half-lives of the two liposomes were largely equivalent, the presence of PAI-2 at the liposome surface might have affected liposome extravasation and uptake at the tumor site, with PAI-2 liposomes binding to uPAR expressed by tumor cells, as was observed in the in vitro experiments. The difference in uptake might also be due to slight differences in surface charge. Surface charge was previously shown to affect tumor uptake of nanoparticles [48], whereby histological analysis showed that negatively charged and neutral liposomes are able to extravasate at the site of the tumor, while positively charged liposomes remain associated with the vascular endothelium, limiting their suitability for tumor-targeting applications [53].

Finally, the maximum tumor accumulation of the liposomes reported in this study was 0.5% of the ID at the 10 min time-point for *N*-AI PAI-2 liposomes and 0.02% for *N*-AI liposomes. These values were comparable to other PEGylated nanoparticles, which typically showed 1% or less of the total ID reaching the site of the primary tumor. Notably, the high tumor accumulation of Doxil® in humans (reported as high as 10% of the ID) was due in large part to the very long circulation half-life (up to 45 h) of the formulation [54]. Given that the in vivo half-life of *N*-AI PAI-2 liposomes was 5.82 h, modifications to the 'stealth' coating of liposomes to promote increased blood circulation and reduce MPS clearance, might further improve liposome uptake into tumors. Following systemic administration, the surface of PEGylated liposomes is modified by protein adsorption, forming a protein corona. Future studies should also aim to characterize the composition of the protein corona, before conducting in vivo efficacy studies, in order to further understand the mechanism of tumor uptake.
