*2.9. Hemolysis*

To determine the hemolytic activity of the complexes, whole blood from healthy mice was diluted in Ringer's solution and erythrocytes were purified by several centrifugation and washing steps at 5000 rpm for 5 min, until the supernatant became clear. After the final washing step, the cells were taken up into physiological NaCl solution. Then, 1 <sup>×</sup> 106 cells in 50 <sup>μ</sup>L were mixed with 50 <sup>μ</sup>L complexes containing different siRNA amounts and incubated for 1 h at 37 ◦C. For absorption measurements, 25 μL of each sample were transferred into a 96 well plate in triplicates. Erythrocytes incubated with buffer served as negative control and erythrocytes treated with Triton X-100 to a final concentration of 1% served as positive control (= 100%). The absorbance of the samples was measured at 550 nm using a plate reader.

To evaluate erythrocyte aggregation, 50 <sup>μ</sup>L of the purified cells (~1 <sup>×</sup> 10<sup>6</sup> cells/mL) were mixed with 50 μL aqueous solution containing 7.5 μg P5Y/3 μg siCtrl, or 750 kDa branched PEI (1 and 3 μg) as positive control. After incubation for 2 h at 37 ◦C, the samples were mounted onto glass slides and examined microscopically.

#### *2.10. In Vivo Tumor Therapy*

Athymic nude mice (Foxn1nu, Charles River Laboratories, Sulzfeld, Germany) were kept at 23 ◦C in a humidified atmosphere, 12 h light/dark cycle, with standard rodent chow and water ad libitum. Animal studies were performed according to the national regulations and approved by the local authorities (Landesdirektion Sachsen, approval NO. TVV 38/16, date: 20 January 2017). Tumor xenografts were established by injecting 3 <sup>×</sup> <sup>10</sup><sup>6</sup> PC3 cells in 150 <sup>μ</sup>L PBS subcutaneously (s.c.) into both flanks of mice. When tumors reached a size of ~100 mm3, the mice were randomized into different treatment groups (6–9 tumors/group). The polyplexes were prepared as described above and amounts corresponding to 10 μg siRNA were i.p. injected every 2–3 days over 14 days.

For in vivo luciferase knockdown experiments, 3 <sup>×</sup> 106 HCT116-Luc cells were s.c. injected for the induction of tumor xenografts described above. Mice were treated five times every 2–3 days with P5Y/siRNA complexes (siLuc2 = siCtrl and siLuc3 = Luciferase specific siRNA). Tumors were excised and lysed with Luciferase Cell Culture Lysis Reagent (app. 100 mg tumor tissue/1 mL buffer), prior to homogenization using an ULTRA-TURRAX®. Homogenates were repeatedly centrifuged at 13,000 rpm for 5 min and the supernatants were transferred into fresh tubes until the lysate became clear. Luminescence was determined as described above. Relative light units (RLU) were normalized for total protein concentration of the lysates using the BCA assay (Pierce Thermo Fisher, Schwerte, Germany) according to the manufacturer's protocol.

For the determination of blood serum markers, healthy nude mice were i.p. injected with P5Y/siLuc3 complexes containing 10 μg siRNA every 2–3 days, with a total of four times over 8 days. Three hours after the last injection, the blood was collected. Untreated mice served as negative control. The serum was diluted 1:20 with water and serum levels of various analytes were determined using an AU480 (Beckman Coulter, Krefeld, Germany).

For the analysis of the immunostimulatory cytokines TNF-α and INF-γ, P5Y/siLuc3 complexes (10 μg siRNA) were i.v. injected twice within 24 h into immunocompetent C57BL/6 mice. Four hours after the last injection, the blood was collected. Mice treated with lipopolysaccharides (LPS) 50 μg in 150 μL (single injection) served as positive control and untreated mice served as negative control. The serum levels of TNF-α and INF-γ were measured using ELISA kits (PreproTech, Hamburg, Germany) following the manufacturer's instructions.

#### *2.11. Statistics*

Statistical analyses were performed by Student's *t*-test or One-way ANOVA, and significance levels are \* = *p* < 0.05, \*\* = *p* < 0.01, \*\*\* = *p* < 0.001, and # = not significant, with at least *n* = 3.

#### **3. Results**

#### *3.1. Identification of Optimal Complexes for In Vitro Transfection*

Based on the superior siRNA complexation and transfection efficacy of 10 kDa PEI upon its tyrosine-modification, we analyzed the performance of even smaller PEIs, which had been shown previously to be inactive [51], upon tyrosine engraftment. We hypothesized that even in the case of 5 kDa or 2 kDa PEIs, the tyrosine modification may result in polymers capable of efficient siRNA complexation and delivery, thus leading to target gene knockdown and RNAi-mediated tumor cell-inhibitory effects in various cellular assays. The tyrosine was covalently coupled onto branched PEIs by using standard NHS/EDC coupling chemistry as outlined in Figure 1A. The degree of tyrosine modification was ~30% compared to ethylenimine monomers, determined by 1H-NMR analysis as described ([50,51]; see Figure S1). Transfection experiments with complexes based on the tyrosine-modified 5 kDa PEI (P5Y) revealed >90% knockdown of serine/threonine-protein kinase PLK1 (polo-like kinase 1; PLK-1) when transfecting a specific siPLK1 as compared to a negative control siRNA (siCtrl; Figure 1B, left). Complexes based on P10Y or P5Y were even more efficient than their higher molecular weight 25 kDa counterpart (P25Y; data not shown), while P2Y/siRNA complexes comprising the very low molecular weight 2 kDa PEI were less efficient. Similar results were obtained after transfection of an siRNA targeting the anti-apoptotic protein survivin, with P5Y/siRNA and P10Y/siRNA complexes again resulting in a 80–90% knockdown (Figure 1B, right) and thus performing better than their P25Y/siRNA counterparts (not shown). P2Y-based nanoparticles showed somewhat reduced activity, even when using higher N/P ratios. The knockdown of the oncogenes/proto-oncogenes PLK-1 and survivin also exerted profound tumor cell-inhibitory effects in anchorage-dependent proliferation assays (Figure 1C). When using P5Y or P10Y, the transfection of 6 pmol/well siRNA was sufficient to largely abolish cell proliferation. In contrast, in the case of P2Y/siRNA complexes, even double amounts (12 pmol) were incapable of inhibiting cell growth. Only the transfection of a very potent ubiquitin siRNA, which essentially abolished tumor cell viabilities in the P5Y and P10Y transfection groups, was strong enough to partially inhibit cell growth when complexed with P2Y. For the very efficient P10Y and P5Y, the siRNA concentration could be further reduced to 3 pmol siRNA with still showing strong inhibitory effects (Figure S2A). In addition, this reduction revealed differences in the potencies of the siRNAs against different target genes.

**Figure 1.** (**A**) Reaction scheme of tyrosine-coupling onto branched polyethylenimines (PEIs). Abbreviations: TFA, Trifluoroacetic acid; NHS, *N*-Hydroxysuccinimid; EDC, 1-Ethyl-3-(3 dimethylaminopropyl) carbodiimid (**B**) Knockdown efficacies of various tyrosine-modified PEI/siRNA complexes in PC3 cells targeting the oncogenes polo-like kinase 1 (PLK1) (left) or survivin (right), quantitated by RT-qPCR. (**C**) Determination of anchorage-dependent proliferation of PC3 cells upon transfection with the tyrosine-modified PEI/siRNA complexes using different anti-proliferative siRNAs at the amounts indicated in the figures. SiRNA-mediated cell inhibition is compared to negative-control transfected or untreated cells. (**D**) Knockdown upon transfecting Saos-2 cells with siRNAs against PLK1 or survivin, as determined by RT-qPCR. Increasing mass ratios from 2.5 to 3.75 does not further improve the knockdown efficacy. (**E**) Determination of anchorage-dependent proliferation of Saos-2 cells upon transfection with P5Y/siRNA complexes targeting PLK1 or survivin. The statistical significance indicates the difference to negative-control transfected cells. Left: quantitation based on WST-8 measurements; right: original pictures.

To further analyze possible reasons for these observed differences in biological activity, we determined intracellular levels of fluorophore-labeled siRNAs upon transfection of PC3 cells complexed

with the different tyrosine-modified PEIs. The comparison of P5Y/atto488-siRNA complexes with their P10Y counterparts used for transfection in the presence of fetal calf serum (FCS) revealed a ~30% decrease in siRNA fluorescence when switching from P5Y to P10Y, while siRNA delivery in P2Y complexes was poor (Figure S3A). This was also confirmed in confocal microscopy, demonstrating atto488-siRNA signals to be largely absent upon transfection with P2Y (Figure S3B, upper panel). In contrast, profound fluorescence was observed in the case of P5Y and P10Y, with somewhat stronger signals in the case of P5Y/siRNA complexes (Figure S3B, center and lower panels). Beyond complex uptake, these differences may be based on variations in complex stabilities. Interestingly, a heparin displacement assay revealed even enhanced P5Y/siRNA complex stability in the presence vs. in the absence of serum, as indicated by the shift towards higher heparin concentrations required for siRNA release from the complex upon its pre-incubation in FCS (Figure S3C, upper panels). Under the same conditions, siRNA release from P2Y/siRNA complexes was already observed at lower heparin concentrations, indicating poor complex stability. In stark contrast, siRNA bands from heparin displacement of P10Y/siRNA complexes were weaker, independent of the heparin concentration, and thus indicating incomplete siRNA release even under very stringent conditions. Taken together, this suggests differences in siRNA delivery, complex stability, and siRNA release as underlying reasons for molecular weight-dependent differences between the tyrosine-modified PEIs. Very high efficacies of tyrosine-modified branched PEIs were observed up to as low as 5 kDa (P5Y), while only an even further reduction of the molecular weight (P2Y) eventually led to reduced transfection efficacies. This prompted us to focus in particular on P5Y as the tyrosine-modified PEI with the lowest molecular weight and highest activity, which was even slightly above P10Y.

The very profound siRNA transfection efficacies of P5Y were also confirmed in other cell lines like Saos-2 osteosarcoma cells (Figure 1D). A 1.5-fold increase of the polymer/siRNA mass ratio did not lead to enhanced knockdown, indicating that the very low polymer/siRNA mass ratio of 2.5 was already sufficient for maximum complex activity (Figure 1D). In contrast, knockdown efficacies were found to be dependent on the selection of the siRNA and the target gene, as indicated by the particularly profound >90% knockdown in the case of an siRNA targeting glyceraldehyde-3-phosphate dehydrogenase (GAPDH). As before in PC3 cells, an almost complete abolishment of cell proliferation was observed upon transfecting Saos-2 cells with P5Y/siPLK-1 or P5Y/siSurv complexes (Figure 1E). The microscopic evaluation of the wells also revealed a complete (siSurv) or almost complete (siPLK-1) loss of viable cells after 72 h (Figure 1E, right panel). In all experiments, P5Y-complexed negative control siRNA (siCtrl) was transfected in parallel, with the comparison to untransfected cells indicating non-specific transfection effects to be largely absent. As seen before in PC3 cells, an increase in the polymer/siRNA ratio by 1.5-fold did not further enhance biological efficacies. However, the P5Y/siCtrl curve indicated some non-specific inhibitory effects at this ratio due to excess polymer (Figure S2B). Specific growth inhibition upon P5Y/siRNA-mediated survivin knockdown was also observed in MV3 melanoma cells (Figure S2C), and GAPDH target gene reduction similar to the results shown above were also found in other cell lines (H441 lung adenocarcinoma cells, PC3 prostate carcinoma cells, and G55T2 glioblastoma cells; Figure S2D).

#### *3.2. Characterization of P5Y*/*siRNA Complex Properties*

For physicochemical characterization, complexes with different polymer/siRNA mass ratios were subjected to Zetasizer and Nanosight measurements. The hydrodynamic diameters determined by dynamic light scattering (DLS) were around 200–400 nm, with a slight trend towards smaller complexes with lower P5Y/siRNA mass ratios (Figure 2A).

**Figure 2.** (**A**) Determination of the hydrodynamic diameters and zeta potentials of P5Y/siRNA complexes at different mass ratios as indicated in the figure. (**B**) Size measurement of P5Y/siRNA complexes by nanoparticle tracking (NTA) at the optimal mass ratio of 2.5. (**C**) Luciferase knockdown efficacies in H441-Luc cells for P5Y/siRNA complexes pre-incubated with increasing amounts of FCS (*v*/*v*) prior to transfection for 1 h (left) or upon storage for 4 weeks at 4 ◦C.

The diameter for the optimal mass ratio of 2.5 was additionally characterized by nanoparticle tracking (NTA) of which the main peak was at 200 nm and a second small peak at 400 nm (Figure 2B). The measurement of the zeta potentials revealed a slightly negative value for the lowest mass ratio of 1.25 indicating non complexed siRNA, and strongly increased to +20 mV for the mass ratio 2.5. The zeta potential further increased to ~30 mV at mass ratio 5 and reached a plateau of ~25 mV for mass ratios above 10. In many nanoparticle formulations, serum stability has been identified as one major issue limiting biological activity. Our transfection experiments, however, were exclusively performed in serum-containing media (10% FCS). Moreover, the presence of serum even proved beneficial: While PEI-based complexes tend to readily aggregate in aqueous solutions [42], our P5Y/siRNA complexes could be stored over weeks in the presence of 10% or even 50% FCS without losing activity, while 5% FCS was insufficient (Figure 2C).

#### *3.3. High Biocompatibility*/*Absence of Toxicity of P5Y*/*siRNA Complexes*

High biological activity may well be associated with increased cytotoxicity or other adverse effects, which were therefore assessed next. Lactate dehydrogenase (LDH) release assays, however, revealed no increase of LDH levels over background (untransfected cells). This was true for both, P5Y/siRNA and P10Y/siRNA complexes and was found in two reporter cell lines (PC3-Luc-EGFP and G55T2-Luc-EGFP cells; Figure 3A,D).

Again, the transfection led to profound target gene knockdown, indicated by 80–90% decreased EGFP activities upon siEGFP transfection as determined in flow cytometry (Figure 3B,E). Concomitantly, cell viabilities remained unaffected by P5Y/siRNA or P10Y/siRNA transfection, even after 48 h and independent of siRNA amounts, which could be increased from 4 pmol to 12 pmol without appreciable decrease in metabolic activity (Figure 3C,F). In line with the absence of adverse effects on cell membrane integrity (LDH release assay), a hemoglobin release assay in erythrocytes revealed no adverse effects. Within an almost 10-fold range of different complex amounts, no hemoglobin release over background was observed (Figure 3G). Similarly, the P5Y/siRNA complexes were tested for their erythrocyte aggregation potential. Identical complex amounts were incubated with red blood cells and analyzed under the microscope (Figure 3H). While the positive control a 750 kDa PEI led to strong aggregates at 1 μg, the P5Y complexes did not damage the red blood cells up to 3 μg complexed siRNA.

**Figure 3.** (**A**,**D**) Absence of early cytotoxic effects, as determined in lactate dehydrogenase (LDH) release assays for P10Y/siCtrl and P5Y/siCtrl complexes in PC3 (**A**) and G55T2 (**D**) cells. LDH release was quantitated 24 h after transfection of 30 pmol siCtrl/polymer per 24 well. (**B**,**E**) Knockdown of the Enhanced Green Fluorescent Protein (EGFP) reporter gene in stably expressing PC3-EGFP/Luc (**B**) and G55T2-EGFP/Luc (**E**) cells. Cells were transfected with P10Y and P5Y siRNA complexes and analyzed after 72 h by flow cytometry. (**C**,**F**) Cell viabilities of PC3 cells (**C**) and G55T2 cells (**F**) 48 h after transfection of P10Y/siCtrl and P5Y/siCtrl complexes at mass ratio 2.5 with increasing siRNA-concentrations. (**G**) Hemolysis assay for P5Y/siRNA complexes, demonstrating the absence of erythrocyte damage over a wider range of different concentrations. (**H**) Erythrocyte aggregation assay, further confirming the biocompatibility of P5Y/siRNA complexes as shown for 3 μg complexed siRNA (upper left; all lens magnifications: 4×). A total of 750 kDa bPEI complexes served as positive control, which led to a dose-dependent, profound aggregation of the red blood cells.
