**4. Discussion**

Tyrosine-modified polymers based on linear or branched PEIs have been identified as highly efficient for siRNA delivery in this and previous studies [50–52,54,56–58]. In this paper, we introduce for the first time the extension of this approach towards PPI dendrimers. In contrast to their non-modified counterparts, the tyrosine-grafted polymers show substantially increased complexation efficacy and complex stability, which is particularly critical for small oligonucleotides like siRNAs and also allows for using smaller polymers at lower mass ratios. The latter aspect may also provide a difference with regard to previous studies on unmodified PPI, where, when using higher siRNA concentrations than

in our experiments, considerable knockdown was observed [29,30]. The positive effect of tyrosine grafting may rely on the contribution of others than electrostatic interactions, including π−π and cation−π interactions as suggested previously [59–61]. Notably, however, even the markedly enhanced complex stabilities still allow for efficient intracellular siRNA release, as seen from the high knockdown efficacies. In this context, it should be noted that the analysis of complex decomposition by heparin displacement has repeatedly proven to be insufficient since its results, falsely, suggested no siRNA release at all. Rather, studies in the presence of biological media like protein-containing cell or tissue lysates seem to be more appropriate in this regard [54].

Another major bottleneck for nanoparticle activity is their efficient cellular delivery and uptake. Notably, despite their large sizes, especially in the case of the PPI-G4-Y/siRNA complexes, and their very low (if at all) surface charge, cellular internalization of tyrosine-modified PEI- or PPI-based nanoparticles proved to be high. This indicates that surface charge is not a major determinant of nanoparticle activity. Still, the larger sizes may come with issues in vivo, e.g., regarding the penetration into intact tissue. In the case of P10Y/siRNA [50] or LP10Y/siRNA complexes [54], however, high biological activities in xenograft tumors were observed, arguing against major problems related to size. While it remains to be seen if this is also true for the even larger PPI-G4-Y/ siRNA complexes, it can already be noted that even sizes of almost 600 nm did not lead to issues regarding biocompatibility after i.v. injection. Also, limited tissue penetration and thus altered biodistribution may prove beneficial with regard to specificities for certain compartments and cell types. In this context, it could be of interest that we found PPI-G4-Y/siRNA complexes to be particularly efficient for macrophage transfection in vitro (data not shown). Thus, they may well represent candidates for preferentially transfecting these cells also in vivo, without the introduction of targeting moieties for targeted delivery. Taken together, these aspects provide the basis and clearly call for future in vivo experiments, covering biodistribution upon different modes of administration, gene knockdown in various target cells including tumor cells, stroma cells as well as hematopoietic cells, and therapeutic effects in relevant tumor models. While beyond the scope of this paper, we have already selected a relevant siRNA dosage (10 μg) for our toxicity studies presented here, based on previous results on PEI/siRNA or P10Y/siRNA complexes [50,55]. Thus, the combination of these and previous data, i.e., (i) the absence of toxicity at (ii) dosages previously identified in PEI-based systems as relevant for therapy, and (iii) in the light of the even enhanced transfection efficacy seen here, provides a strong basis for extensive in vivo and therapy studies.

The very high biocompatibilities in vitro and in vivo described here for PPI-G4-Y/siRNA complexes as well as in our previous studies for tyrosine-modified PEIs may be readily explained by the low nanoparticle zeta potentials, considering that positive surface charges have been associated with cytotoxicity [62], and the comparably little polymer amounts required for siRNA complexation. Still, more detailed studies will be required for comprehensively analyzing their toxicological profile. As reported previously, G4 PPI induced DNA damages in a COMET assay but no effects were observed in the case of a maltose-modified G4 PPI derivative. Likewise, G4 PPI, but not the maltose-modified G4 PPI, showed toxic effects in vivo, e.g., changes in serum parameters, body weight and behavior. These studies indicate that PPI dendrimers, when chemically modified, are interesting starting materials for synthesizing new effective and biocompatible polymers for nucleic acid delivery. In this regard, the approach of combining tyrosine-grafting with the introduction of biodegradable linkers at defined branching points in the polymer will be of particular interest. It should be noted, however, that the systems reviewed and described here rely on a comparably minor chemical modification, thus increasing the likeliness of their possible translation into the clinics. It must be kept in mind that successful systems for therapeutic siRNA delivery will have to combine maximum efficacy and biocompatibility with favorable properties with regard to manufacturing, upscaling, charge variability and GMP production.
