Combination of Drug Delivery Properties of PAMAM Dendrimers and Cytotoxicity of Platinum(IV) Complexes—A More Selective Anticancer Treatment?

Based on their drug delivery properties and activity against tumors, we combined PAMAM dendrimers with various platinum(IV) complexes in order to provide an improved approach of anticancer treatment. Platinum(IV) complexes were linked to terminal NH2 moieties of PAMAM dendrimers of generation 2 (G2) and 4 (G4) via amide bonds. Conjugates were characterized by 1H and 195Pt NMR spectroscopy, ICP-MS and in representative cases by pseudo-2D diffusion-ordered NMR spectroscopy. Additionally, the reduction behavior of conjugates in comparison to corresponding platinum(IV) complexes was investigated, showing a faster reduction of conjugates. Cytotoxicity was evaluated via the MTT assay in human cell lines (A549, CH1/PA-1, SW480), revealing IC50 values in the low micromolar to high picomolar range. The synergistic combination of PAMAM dendrimers and platinum(IV) complexes resulted in up to 200 times increased cytotoxic activity of conjugates in consideration of the loaded platinum(IV) units compared to their platinum(IV) counterparts. The lowest IC50 value of 780 ± 260 pM in the CH1/PA-1 cancer cell line was detected for an oxaliplatin-based G4 PAMAM dendrimer conjugate. Finally, in vivo experiments of a cisplatin-based G4 PAMAM dendrimer conjugate were performed based on the best toxicological profile. A maximum tumor growth inhibition effect of 65.6% compared to 47.6% for cisplatin was observed as well as a trend of prolonged animal survival.


Introduction
Five decades after the discovery of the anticancer activity of cisplatin, platinum(II)based anticancer agents still play an essential role in modern cancer treatment. The breakthrough of platinum(II) complexes was further consolidated by the introduction of carboplatin and oxaliplatin in clinical practice [1][2][3]. Together, these three platinum(II) drugs are integrated into about 50% of all cancer chemotherapies worldwide and cisplatin belongs to the most lucrative anticancer agents [4,5]. Despite the enormous success, platinum(II)based cancer treatment lacks selectivity against tumor cells, leading to severe side effects, and is further affected by intrinsic and/or acquired resistance, diminishing its clinical efficacy [6].
The prodrug strategy, employing kinetically more inert platinum(IV) complexes, enabled a new approach to improve selectivity and reduce systemic toxicity. The reduction of the corresponding platinum(II) complexes, required to unleash their cytotoxic capacities, is facilitated by the characteristic oxygen-deficient milieu of tumor tissue [7,8]. Despite

Preparative RP-HPLC
An Agilent 1200 Series system controlled by ChemStation ® software was used for purifications by preparative RP-HPLC. A XBridge ® Prep C18 10 µm OBD TM Column (19 mm × 250 mm) from Waters served as the stationary phase, whereas different ratios of Milli-Q water, acetonitrile and methanol with the addition of 0.1% formic acid were used as eluents.

Elemental Analysis
The Microanalytical Laboratory of the Faculty of Chemistry at the University of Vienna performed the elemental analyses of the platinum(IV) compounds, using an Eurovector EA3000 elemental analyzer. All obtained values are in the range of ±0.4% of the calculated values, therefore confirming purity of at least 96%.  19 F chemical shifts are given relative to CCl 3 F, whereas NH 4 Cl and K 2 [PtCl 4 ] served as external reference for 15 N and 195 Pt NMR spectroscopy.

NMR Spectroscopy
DOSY experiments were measured with a standard Bruker DOSY pulse sequence without sample spinning in D 2 O at 298 K. The diffusion coefficient D, obtained from the spectra, was used for the estimation of the diameter of PAMAM dendrimers and conjugates by assuming a spherical size and using the Stokes-Einstein equation: R 0 is the (hydrodynamic) radius, k B the Boltzmann constant (JK −1 ), T the temperature [K], η the viscosity of the liquid (1.095 × 10 −3 Pa·s) and D the diffusion coefficient [m 2 s −1 ] [23].

Reduction Behavior
The reduction of platinum(IV) complexes and representatives of each series of conjugates by ascorbic acid was observed by 1 H NMR spectroscopic measurements at ambient temperature. An amount of 1 mM solutions of platinum(IV) complexes 1-3, 5-7 as well as of the conjugates C2, C11, C13, C22-C24 referred to their corresponding platinum units were prepared in a D 2 O phosphate-buffered solution (50 mM) at physiological pD = 7.4. Afterwards, ascorbic acid was added as a reducing agent (25 mM, 25 eq.) and 1 H NMR measurements were performed for several days. The reduction behavior for substances 1-3, 5, 6, C2, C11, C13, C22, C23 was monitored by the resulting decrease in intensity of the acetato signal (release of acetic acid). The percentage of reduced species was determined by using the integration ratios of these two signals. Contrarily, ratios of signals of the succinato ligands and free succinic acid were consulted for the reduction determination of substances 7 and C24. However, the reduction half-time of C24 could not be obtained due to the superimposition of the succinato peaks with other signals.

ICP-MS
Digestion of all conjugates (0.5-1.5 mg) was performed in 2 mL of HNO 3 (20%) and 0.1 mL of H 2 O 2 (30%) with a temperature-controlled heating plate of graphite from Labter. After dilution (1:10,000) of digested samples with HNO 3 (3%), the total platinum amount was determined with an Agilent 7800 ICP-MS instrument. For each sample, 10 measurements were performed and the obtained data were analyzed with the Agilent MassHunter software package (Workstation Software, Version C.01.04, 2018, Agilent, Santa Clara, CA, USA).

Cytotoxicity Tests
Culture of and 96 h cytotoxicity tests in the human cell lines CH1/PA-1 (ovarian teratocarcinoma), SW480 (colon carcinoma) and A549 (non-small cell lung cancer) were performed as described previously [24], with the exception that test compounds were dissolved either in sterile water or supplemented MEM and then serially diluted in the latter medium.
The MTT assay in the murine cell line CT26 was performed as described in [25].

In Vivo Experiments and Organ Distribution
Female and male BALB/c mice, bread in-house (originally Janvier), were kept in a controlled environment under pathogen-free conditions. By the toxicity tests, non-tumorbearing female BALB/c mice were treated intravenously at a dose: G4 PAMAM dendrimer at 0.058 mg/20 g, C11 at 0.17 mg/20 g, C12 at 0.075 mg/20 g and 0.15 mg/20 g, C13 at 0.91 mg/20 g and C22 at 0.52 mg/20 g three times in the first week. All drugs were dissolved in 0.9% NaCl. For the efficacy study, CT26 (5 × 10 5 in 50 µL serum-free RMPI medium) murine colon cells were injected into the right flank (subcutaneously) of female BALB/c mice. When all tumors were measurable (at day 5), animals were treated intravenously with solvent (0.9% NaCl, 100 µL/20 g), C12 (7.5 mg/kg in 0.9% NaCl, 100 µL/20 g) and cisplatin (3 mg/kg in 0.9% NaCl, 100 µL/20 g) twice a week for two weeks. Tumor volume and size (measured by caliper) and body weight were evaluated every working day; the animals were sacrificed by cervical dislocation upon a loss in body weight, tumor size or other indications of deteriorated health. In vivo experiments were done according to the regulations of the Ethics Committee for the Care and Use of Laboratory Animals at the Medical University Vienna (proposal number BMBWF-V/3b 2020-0.380.502).
Organs and tumors for platinum content/distribution were harvested 24 h after the second application of C12 (7.5 mg/kg in 0.9% NaCl, 100 µL/20 g) and cisplatin (3 mg/kg in 0.9% NaCl, 100 µL/20 g) in male CT26-allograft-bearing BALB/c mice. Blood was taken after 30 min and 24 h after the first application from the facial vein and terminally by cardiac puncture 24 h after the second application. To separate serum from blood pellet, blood was twice spun down at 900 g for 10 min at 4 • C. The platinum amount of all samples was measured by ICP-MS.
General procedure 1: carboxylation of unsymmetrically oxidized platinum(IV) complexes (1-6): The corresponding precursor platinum(IV) complex and succinic anhydride were stirred overnight in absolute DMF under argon atmosphere at 50 • C. The solvent was removed under reduced pressure. Purification was performed by using preparative RP-HPLC. Finally, the product was freeze-dried.

Synthesis
Acetatohydroxidoplatinum(IV) precursor complexes were synthesized according to standard procedures via unsymmetric oxidation with hydrogen peroxide in acetic acid [27]. The trifluoromethyl oxaliplatin analog was synthesized as reported recently [26]. Further reaction with succinic anhydride in absolute DMF resulted in complexes 1-6 after purification via RP-HPLC [23]. The synthesis of compound 7 followed a method previously published (Scheme 1) [28].
The conjugation of platinum(IV) complexes to G2 and G4 PAMAM dendrimers was performed in two steps adapted from a previously published procedure [23]. At first, COOH groups of compounds 1-7 were activated with the coupling reagents N-(3-dimethylaminopropyl)-N -ethylcarbodiimide hydrochloride (EDC·HCl) and Nhydroxysuccinimide (NHS) forming an NHS-ester [29]. The addition of G2 and G4 PAMAM dendrimers containing primary amines as terminal groups resulted in amide bond formations leading to conjugates C1-C24 (Scheme 2). Purification was performed via dialysis against distilled water using dialysis tubings with a molecular weight cut-off (MWCO) of 1 kDa for conjugates of G2 and MWCO of 3.5 kDa for G4, respectively. The final conjugates, C1-C24, were obtained via lyophilization. Additionally, conjugates with platinum(IV) complex 6 were synthesized. However, analysis by NMR spectroscopy revealed an additional signal in 195 Pt NMR spectra, probably caused by hydrolysis during the dialysis process. Consequently, these conjugates could not be included in this study. Furthermore, numerous couplings of the G4 PAMAM dendrimer with platinum(IV) complexes 1 and 3 were conducted with different reaction times in order to vary the loading with platinum(IV) complexes (Table 1).

Analysis
Characterization of platinum(IV) complexes 1-7 was performed by using multinuclear one-and two-dimensional NMR spectroscopy ( 1 H, 13 C, 15 N, 19 F, 195 Pt) (Supporting Information, Figures S1-S10) and their purity was validated by elemental analysis (>95%). The platinum(IV)-PAMAM conjugates C1-C24 were analyzed by 1 H and 195 Pt NMR spectroscopy (Supporting Information, Figures S11-S20). 195 Pt resonances between 2611 and 3507 ppm are indicative of the presence of platinum(IV) units. As an example, the 1 H NMR spectra of G4 PAMAM conjugate C14 and platinum(IV) complex 3 are shown for comparison ( Figure S21). 1 H signals of conjugates in the region above 2.2 ppm result in part from signals of the bound platinum(IV) moiety as well as from signals of the inner and outer part of the dendrimer with peripheral amine groups which are free or bound to the platinum complex. Therefore, complete signal assignment in proton NMR spectra was unfortunately not possible.
Furthermore, ICP-MS was used for the determination of the platinum amount of conjugates C1-C24. The different platinum(IV) units per dendrimer and the corresponding loading rates are shown in Table 1. The molecular weight of conjugates C1-C24 was calculated based on the molecular weight of the PAMAM dendrimer (according to Sigma-Aldrich: PAMAM G2 = 3256 g/mol, PAMAM G4 = 14214 g/mol) and the addition of the molecular weight of attached platinum(IV) units, while also considering the release of water molecules during the conjugation process (Table 1).
Additionally, pseudo-2D diffusion-ordered spectroscopy (DOSY) spectra were measured for selected compounds, confirming the conjugation of platinum(IV) complexes to PAMAM dendrimers. As an example, an overlay of DOSY spectra of the conjugate C14, the unloaded G4 PAMAM dendrimer and platinum(IV) complex 3 is shown in Figure S22 (Supporting Information). The derived diffusion coefficient was used to estimate the average diameter of unloaded G2 and G4 PAMAM dendrimers as well as of conjugates C1-C3 and C12-C14 using the Stokes-Einstein equation and the assumption that the molecules are spherical ( Table 2). As expected, the conjugation of platinum(IV) complexes to dendrimers significantly increased the diameter compared to unloaded PAMAM dendrimers. The calculated diameters are consistent with previously published data [23].

Analysis
Characterization of platinum(IV) complexes 1-7 was performed by using multinuclear one-and two-dimensional NMR spectroscopy ( 1 H, 13 C, 15 N, 19 F, 195 Pt) (Supporting Information, Figures S1-S10) and their purity was validated by elemental analysis (>95%). The platinum(IV)-PAMAM conjugates C1-C24 were analyzed by 1 H and 195 Pt NMR spectroscopy ( Supporting Information, Figures S11-S20). 195 Pt resonances between 2611 and 3507 ppm are indicative of the presence of platinum(IV) units. As an example, the 1 H NMR spectra of G4 PAMAM conjugate C14 and platinum(IV) complex 3 are shown for comparison ( Figure S21). 1 H signals of conjugates in the region above 2.2 ppm result in part from signals of the bound platinum(IV) moiety as well as from signals of the inner and outer part of the dendrimer with peripheral amine groups which are free or bound to the platinum complex. Therefore, complete signal assignment in proton NMR spectra was unfortunately not possible.
Furthermore, ICP-MS was used for the determination of the platinum amount of conjugates C1-C24. The different platinum(IV) units per dendrimer and the corresponding loading rates are shown in Table 1. The molecular weight of conjugates C1-C24 was calculated based on the molecular weight of the PAMAM dendrimer (according to Sigma-Aldrich: PAMAM G2 = 3256 g/mol, PAMAM G4 = 14214 g/mol) and the addition of the molecular weight of attached platinum(IV) units, while also considering the release of water molecules during the conjugation process (Table 1).
Additionally, pseudo-2D diffusion-ordered spectroscopy (DOSY) spectra were measured for selected compounds, confirming the conjugation of platinum(IV) complexes to PAMAM dendrimers. As an example, an overlay of DOSY spectra of the conjugate C14, the unloaded G4 PAMAM dendrimer and platinum(IV) complex 3 is shown in Figure S22 (Supporting Information). The derived diffusion coefficient was used to estimate the average diameter of unloaded G2 and G4 PAMAM dendrimers as well as of conjugates C1-C3 and C12-C14 using the Stokes-Einstein equation and the assumption that the molecules are spherical ( Table 2). As expected, the conjugation of platinum(IV) complexes to dendrimers significantly increased the diameter compared to unloaded PAMAM dendrimers. The calculated diameters are consistent with previously published data [23]. G2 PAMAM -G2~27.9 -G4 PAMAM -G4~44.4 - Moreover, the reduction behavior of conjugates in comparison to unattached platinum(IV) complexes was investigated by time-dependent 1 H NMR spectroscopy (Table 3). In general, the reduction behavior is influenced by the nature of the ligands coordinated to the platinum(IV) core. In addition to the axial ligands, the equatorial coordination sphere plays a crucial part as shown by the significantly different rates of reduction of platinum(IV) complexes 1-3, 5, 6 featuring the same carboxylato ligands in axial position. In accordance with previous studies [30,31], the cisplatin core of substance 1 led to a faster reduction with a reduction half-time of 6 h, whereas only 6% of the carboplatin(IV) analog 2 was reduced after 95 h. Similar to complex 1, the two chlorido ligands of compound 6 allow fast electron transfer, thereby supporting rapid reduction [32]. The additional trifluoromethyl group in position 4 of the DACH ligand of complex 5 seems to cause a faster reduction in comparison to substance 3, comparable with the discovered relationship of electron-withdrawing power and rate of reduction for axial ligands [33,34]. Furthermore, the strong effect of axial ligands on the reduction behavior is displayed by the comparison of the two oxaliplatin(IV) analogs 3 and 7. The amine and carboxylato ligands of complex 3 cannot form bridges with the reducing agent and thus do not support the electron transfer from ascorbate to the platinum(IV) atom in the center. Consequently, only 14% of compound 3 was reduced after 165 h. In contrast, the hydroxido ligand of complex 7 facilitates electron transfer and results in a significantly lower reduction half-time of 29 h, consistent with previously published results [35,36]. The same order of rates of reduction was observed for the conjugates of the respective platinum(IV) complexes, except for C23. Generally, all conjugates underwent faster reduction compared to their corresponding platinum(IV) complexes, possibly caused by the increased bulkiness of the axial ligand. A potential connection between bulkiness and facilitated reduction has been reported in the literature [37]. As further expected, the influence between G2 and G4 dendrimers on the reduction behavior is marginal due to the huge distance to the platinum(IV) core. Therefore, no significant difference in the rate of reduction between conjugates of G2 and G4 dendrimers was observed based on the measurements of C2 and C13, respectively. Table 3. Overview of the reduction half-times of platinum(IV) complexes and representative conjugates at ambient temperature (ratio complex: ascorbic acid = 1:25). Due to long reduction half-times, some measurements were stopped before reaching the reduction half-time, and the percentage of the reduced species at this point is mentioned in brackets.

Cytotoxicity
As expected due to their higher inertness, platinum(IV) complexes 1-3 are by one to two orders of magnitude less potent than the corresponding platinum(II) drugs cisplatin, carboplatin and oxaliplatin. Of the structural modifications imposed on 3, only the exchange of oxalate for two chlorido ligands had conspicuous consequences for biological activity: IC50 values of the dichlorido analog 6 are 9-26 times lower than those of 3, depending on the cell line, whereas addition of a CF3 substituent to the DACH ligand (in 5) or replacement of the axial acetato ligand with a hydroxido group (in 7) yielded minor, if any, changes in cytotoxic potency.
Except for the rather inefficient conjugate C2 (where, moreover, a comparison with the unconjugated complex 2 is only partially possible, as some IC50 values were not even reached), loading of the platinum(IV) complexes onto G2 PAMAM dendrimers (C1, C3-C5) resulted in products with 4-30 times increased cytotoxic potency in absolute numbers.  Table 4. This pattern of sensitivity also reflects throughout the data compiled here.
As expected due to their higher inertness, platinum(IV) complexes 1-3 are by one to two orders of magnitude less potent than the corresponding platinum(II) drugs cisplatin, carboplatin and oxaliplatin. Of the structural modifications imposed on 3, only the exchange of oxalate for two chlorido ligands had conspicuous consequences for biological activity: IC 50 values of the dichlorido analog 6 are 9-26 times lower than those of 3, depending on the cell line, whereas addition of a CF 3 substituent to the DACH ligand (in 5) or replacement of the axial acetato ligand with a hydroxido group (in 7) yielded minor, if any, changes in cytotoxic potency.
Except for the rather inefficient conjugate C2 (where, moreover, a comparison with the unconjugated complex 2 is only partially possible, as some IC 50 values were not even reached), loading of the platinum(IV) complexes onto G2 PAMAM dendrimers (C1, C3-C5) resulted in products with 4-30 times increased cytotoxic potency in absolute numbers. This implies that the products mostly exert at least the effect that could roughly be expected from their degree of platinum(IV) loading (4-9 platinum(IV) units per dendrimer), but even higher effects were observed in some cases (e.g., compare C4 with 5 in CH1/PA-1 cells). Apart from a singular exception (conjugate C23 containing complex 5), the effects of loaded G4 PAMAM dendrimers are not less remarkable: the majority of C6-C12 (loaded with cisplatin analog 1) and, even more so, C14-C22 (loaded with oxaliplatin analogs 3 or 4) is much more to tremendously more potent than could be explained by the mere ratios of platinum(IV) loading (especially compare C15, C20, C21 or also C22 with 3). Reasons are likely to be multifactorial. It has been reported that dendrimers can enhance membrane permeability and therefore increase the cellular uptake of drugs. The conjugation of anticancer agent paclitaxel to lauryl-modified G3 PAMAM dendrimers led to up to 12-times higher permeability than free paclitaxel in monolayers of the human colon adenocarcinoma cell line Caco-2, as well as in porcine brain endothelial cells [40]. Furthermore, enhanced cellular uptake by a factor of up to 11 was detected with G4 PAMAM dendrimers combined with cisplatin in A2780 ovarian cancer cells compared to free cisplatin [19]. Additionally, a relationship of fast reduction leading to increased cytotoxicity is widely accepted [41]. According to Ref. [19], the accelerated activation of conjugates by reduction based on faster reduction half-times (Table 2) as compared to their corresponding platinum(IV) complexes may play a significant part in increased cytotoxicity. In addition to enhanced permeability, cellular uptake and faster reduction half-times, it is conceivable that synergies between the individual effects of platinum(IV) complex and G4 PAMAM dendrimer additionally contribute to the extraordinary enhancement of cytotoxicity, since even the unloaded G4 PAMAM dendrimer (in contrast to G2) exerts antiproliferative activity in the low micromolar concentration range in all three cell lines. However, detailed investigations are needed to fully understand the mechanism of significantly increased cytotoxicity of platinum(IV)-based PAMAM dendrimer conjugates. Furthermore, it is intriguing that higher loading than the applied minimum of about 10 platinum(IV) units per dendrimer did not necessarily (or, in fact, rather occasionally and under proportionally) lead to further increase in potency, neither in series C6-C12 nor C14-C21. It is conceivable that reasons are similar to a previously conducted study of half-generation PAMAM dendrimers loaded with cisplatin, in which an incomplete drug release was detected probably caused by intramolecular interactions of platinum complex and dendrimer branches [42].

In Vivo Studies
In order to validate the in vivo efficacy of the platinum(IV)-loaded dendrimer strategy, one representative conjugate of the platinum(IV) series of cisplatin (C11), carboplatin (C13), and oxaliplatin (C22) was tested in the G4 PAMAM dendrimer background. The toxicity tests were performed by tail vein injection (every second day for 3 injections in total) into non-tumor-bearing mice at a dose equimolar to the released platinum(II) species: C11, 0.17 mg/20 g, equimolar to 3 mg/kg cisplatin; C13, 0.91 mg/20 g, equimolar to 17 mg/kg carboplatin; C22, 0.52 mg/20 g; equimolar to 9 mg/kg oxaliplatin. Out of the three tested substances, the cisplatin-based conjugate C11 showed by far the best tolerability without significant weight loss or profound changes in behavior and showed only temporal mild hair loss. In contrast, oxaliplatin-based conjugate C22 led in both treated mice to tail necrosis, forcing the termination of this experimental group based on ethical guidelines. The carboplatin-based conjugate C13 induced moderate weight and strong hair loss. Thus, the cisplatin-based conjugate was further analyzed in more depth. An additional toxicity assay with C12, a higher platinum(IV)-loaded conjugate compared to C11, was performed at concentrations equimolar to 1.5 and 3 mg/kg cisplatin and the same application scheme as before. Again, no signs of toxicity were observed.
Consequently, C12 was chosen for the therapy experiment. As the anticancer activity of platinum drugs includes also immunological mechanisms [43], the colon cancer allograft model was used. As a first step, the impact of cisplatin was tested compared to C12 in CT26 murine colon cancer cells in vitro. Compound C12 exerted a more than four-fold lower IC 50 value as compared to cisplatin (C12: 0.43 µM; cisplatin: 1.85 µM, Supporting Information, Figure S30), thus resembling data in the human cancer cell models (compare Table 4).
Based on this higher cytotoxic activity and clearly better tolerability in non-tumorbearing animals, the efficacy of C12 (7.5 mg/kg: equimolar to 3 mg/kg cisplatin) was compared with the respective dose of free cisplatin (3 mg/kg) and unloaded G4 PAMAM dendrimer in CT26-allograft-bearing mice. Drugs were given intravenously for two weeks twice a week. The impact of the different treatment groups as compared to solvent control on tumor volume and body weight until day 14 (loss of the first mouse in the solvent group due to big tumor size) are given in Figures 2 and 3, respectively. All treatments significantly reduced tumor volume as compared to the solvent control with a maximum tumor growth inhibition effect of 37% for unloaded G4 PAMAM dendrimer, 47.6% for cisplatin, as well as 65.6% for C12. In addition, concerning the tumor growth curves, the strongest activity was exerted by C12, although the difference between free cisplatin and C12 did not reach statistical significance in the multiple comparison tests (p = 0.075). Concerning toxicity, neither free G4 PAMAM dendrimer nor C12 led to any loss of body weight. In contrast, free cisplatin at the maximal tolerated dose of 3 mg/kg significantly reduced body weight as compared to all other experimental groups with a maximal loss of body weight of around 20% at day 14 of treatment. This data strongly suggests an improved therapeutic window for cisplatin when given as a PAMAM-dendrimer-based nano-formulation. on tumor volume and body weight until day 14 (loss of the first mouse in the solvent group due to big tumor size) are given in Figures 2 and 3, respectively. All treatments significantly reduced tumor volume as compared to the solvent control with a maximum tumor growth inhibition effect of 37% for unloaded G4 PAMAM dendrimer, 47.6% for cisplatin, as well as 65.6% for C12. In addition, concerning the tumor growth curves, the strongest activity was exerted by C12, although the difference between free cisplatin and C12 did not reach statistical significance in the multiple comparison tests (p = 0.075). Concerning toxicity, neither free G4 PAMAM dendrimer nor C12 led to any loss of body weight. In contrast, free cisplatin at the maximal tolerated dose of 3 mg/kg significantly reduced body weight as compared to all other experimental groups with a maximal loss of body weight of around 20% at day 14 of treatment. This data strongly suggests an improved therapeutic window for cisplatin when given as a PAMAM-dendrimer-based nano-formulation.  The promising effects further translate into a prolongation trend of animal survival (Figure 4). While in the cisplatin group, weight loss was critical in addition to tumor necrosis, even smaller tumors tended to get necrotic and broke up in C12-treated animals, making the sacrifice of the animals necessary due to ethical guidelines. In contrast to the on tumor volume and body weight until day 14 (loss of the first mouse in the solvent group due to big tumor size) are given in Figures 2 and 3, respectively. All treatments significantly reduced tumor volume as compared to the solvent control with a maximum tumor growth inhibition effect of 37% for unloaded G4 PAMAM dendrimer, 47.6% for cisplatin, as well as 65.6% for C12. In addition, concerning the tumor growth curves, the strongest activity was exerted by C12, although the difference between free cisplatin and C12 did not reach statistical significance in the multiple comparison tests (p = 0.075). Concerning toxicity, neither free G4 PAMAM dendrimer nor C12 led to any loss of body weight. In contrast, free cisplatin at the maximal tolerated dose of 3 mg/kg significantly reduced body weight as compared to all other experimental groups with a maximal loss of body weight of around 20% at day 14 of treatment. This data strongly suggests an improved therapeutic window for cisplatin when given as a PAMAM-dendrimer-based nano-formulation.  The promising effects further translate into a prolongation trend of animal survival (Figure 4). While in the cisplatin group, weight loss was critical in addition to tumor necrosis, even smaller tumors tended to get necrotic and broke up in C12-treated animals, making the sacrifice of the animals necessary due to ethical guidelines. In contrast to the The promising effects further translate into a prolongation trend of animal survival ( Figure 4). While in the cisplatin group, weight loss was critical in addition to tumor necrosis, even smaller tumors tended to get necrotic and broke up in C12-treated animals, making the sacrifice of the animals necessary due to ethical guidelines. In contrast to the cisplatin and G4 PAMAM dendrimer treatment arms, only the anticancer activity of C12 showed a trend towards longer survival (p-value of 0.0549 in log-rank and 0.05 in Gehan-Breslow-Wilcoxon test) in the direct comparison with the solvent control arm. A more extended analysis of different doses and schedules is needed to optimize the therapeutic effects of C12, but limit the massive necrotizing effects leading to the termination of the experiment due to tumor ulceration. cisplatin and G4 PAMAM dendrimer treatment arms, only the anticancer activity of C12 showed a trend towards longer survival (p-value of 0.0549 in log-rank and 0.05 in Gehan-Breslow-Wilcoxon test) in the direct comparison with the solvent control arm. A more extended analysis of different doses and schedules is needed to optimize the therapeutic effects of C12, but limit the massive necrotizing effects leading to the termination of the experiment due to tumor ulceration. In addition to tumor growth experiments, the tumor and organ distribution of platinum in mice treated with cisplatin or C12 were determined by ICP-MS ( Figure 5). Unexpectedly, the platinum levels in the tumor did not differ significantly 24 h after the second dosing. In contrast, C12 led to high platinum levels in the serum at all three time points, while blood cells contained enhanced platinum contents in the case of cisplatin treatment (Supporting Information, Figure S31). This suggests a lower clearance and reduced local interaction with blood cells of C12, as compared to cisplatin. In organ distribution ( Figure  5), C12 treatment led to a massive platinum accumulation in the kidney as compared to all other organs, while in the case of cisplatin, higher levels were detected in lung tissue, however, with great inter-individual differences. These data strongly suggest that the PA-MAM dendrimer formulation leads to trapping of the nano-formation probably based on filtration in the glomerulus and reabsorbed in the lumen of the proximal tubule. Renal excretion and glomerular filtration are typical for ≤G4 PAMAM dendrimers and in general for nanoparticles smaller than 5 nm [44,45]. The conjugation of complex 1 to G4 PAMAM dendrimers considerably increased the polymer diameter of C14 from about 4.5 to 5.4 nm. Additionally, the influence of the molecular weight of nanoparticles on the biodistribution behavior was reported previously. Nanoparticles <20 kDa primarily undergo renal clearance, whereas bigger molecules show longer blood circulation times and a shift towards clearance by the reticuloendothelial system [46]. Based on a particle size of over 5 nm and a molecular weight of 36.5 kDa of conjugate C12, the renal accumulation is rather unexpected and needs to be investigated in more detail. However, it is obvious that, despite these unwanted conditions, C12 exerted a comparable tumor accumulation to cisplatin, a tendency to enhance anticancer activity, and an improved therapeutic window. This strongly suggests further modifications of the novel cisplatin dendrimeric remedy to ameliorate kidney accumulation and, in parallel, enhance tumor response. On the contrary, In addition to tumor growth experiments, the tumor and organ distribution of platinum in mice treated with cisplatin or C12 were determined by ICP-MS ( Figure 5). Unexpectedly, the platinum levels in the tumor did not differ significantly 24 h after the second dosing. In contrast, C12 led to high platinum levels in the serum at all three time points, while blood cells contained enhanced platinum contents in the case of cisplatin treatment (Supporting Information, Figure S31). This suggests a lower clearance and reduced local interaction with blood cells of C12, as compared to cisplatin. In organ distribution ( Figure 5), C12 treatment led to a massive platinum accumulation in the kidney as compared to all other organs, while in the case of cisplatin, higher levels were detected in lung tissue, however, with great inter-individual differences. These data strongly suggest that the PAMAM dendrimer formulation leads to trapping of the nano-formation probably based on filtration in the glomerulus and reabsorbed in the lumen of the proximal tubule. Renal excretion and glomerular filtration are typical for ≤G4 PAMAM dendrimers and in general for nanoparticles smaller than 5 nm [44,45]. The conjugation of complex 1 to G4 PAMAM dendrimers considerably increased the polymer diameter of C14 from about 4.5 to 5.4 nm. Additionally, the influence of the molecular weight of nanoparticles on the biodistribution behavior was reported previously. Nanoparticles <20 kDa primarily undergo renal clearance, whereas bigger molecules show longer blood circulation times and a shift towards clearance by the reticuloendothelial system [46]. Based on a particle size of over 5 nm and a molecular weight of 36.5 kDa of conjugate C12, the renal accumulation is rather unexpected and needs to be investigated in more detail. However, it is obvious that, despite these unwanted conditions, C12 exerted a comparable tumor accumulation to cisplatin, a tendency to enhance anticancer activity, and an improved therapeutic window. This strongly suggests further modifications of the novel cisplatin dendrimeric remedy to ameliorate kidney accumulation and, in parallel, enhance tumor response. On the contrary, the efficient kidney accumulation of this dendrimer preparation might also be considered in kidney-specific drug delivery approaches [47][48][49]. the efficient kidney accumulation of this dendrimer preparation might also be considered in kidney-specific drug delivery approaches [47][48][49].

Conclusions
In the present study, an alternative antitumor strategy was presented by conjugating various platinum(IV)-analogs of cisplatin, carboplatin and oxaliplatin to the surface of G2 and G4 PAMAM dendrimers. Twenty-four novel conjugates were synthesized and characterized by 1 H and 195 Pt NMR spectroscopy, as well as DOSY measurements substantiating the successful conjugation. Reduction behavior analysis revealed a significantly faster reduction of all conjugates in comparison to their corresponding platinum(IV) complexes, probably caused by the increased bulkiness of the axial ligands of the conjugates. The accelerated reduction of the conjugates may also, amongst others, be responsible for improved cytotoxicity of the conjugates. Specifically, the conjugation of platinum(IV) complexes to G4 PAMAM dendrimers resulted in IC50 values in the low micro-to the nanomolar range, tremendously lower compared to the corresponding platinum(IV) or even platinum(II) complexes. Remarkably, an oxaliplatin(IV)-based conjugate even reached an IC50 value of 780 ± 260 pM in the CH1/PA-1 cancer cell line.
Furthermore, the cisplatin(IV)-based conjugate C12 was investigated in vivo in CT26allograft-bearing mice due to its best toxicological profile. Concentrations of the conjugate equimolar to 3 mg/kg cisplatin (maximum tolerated dosage [50]) were very well tolerated

Conclusions
In the present study, an alternative antitumor strategy was presented by conjugating various platinum(IV)-analogs of cisplatin, carboplatin and oxaliplatin to the surface of G2 and G4 PAMAM dendrimers. Twenty-four novel conjugates were synthesized and characterized by 1 H and 195 Pt NMR spectroscopy, as well as DOSY measurements substantiating the successful conjugation. Reduction behavior analysis revealed a significantly faster reduction of all conjugates in comparison to their corresponding platinum(IV) complexes, probably caused by the increased bulkiness of the axial ligands of the conjugates. The accelerated reduction of the conjugates may also, amongst others, be responsible for improved cytotoxicity of the conjugates. Specifically, the conjugation of platinum(IV) complexes to G4 PAMAM dendrimers resulted in IC 50 values in the low micro-to the nanomolar range, tremendously lower compared to the corresponding platinum(IV) or even platinum(II) complexes. Remarkably, an oxaliplatin(IV)-based conjugate even reached an IC 50 value of 780 ± 260 pM in the CH1/PA-1 cancer cell line.
Furthermore, the cisplatin(IV)-based conjugate C12 was investigated in vivo in CT26allograft-bearing mice due to its best toxicological profile. Concentrations of the conjugate equimolar to 3 mg/kg cisplatin (maximum tolerated dosage [50]) were very well tolerated and even higher doses could be considered in further investigations. Additionally, biodistribution was analyzed in tumor-bearing mice 24 h after the second application. Unexpectedly, increased accumulation in the kidney was detected despite a higher cut-off molecular weight and particle size of C12. It needs to be investigated in more detail, whether C12 behaves like nanoparticles with a hydrodynamic diameter between 5 nm and 100 nm efficiently crossing the endothelial layer, but blocked by the glomerular basement membrane [51]. Due to the connection of preferred renal excretion with small molecular weights and particle sizes as well as cationic surface charges, supporting attraction to the negatively charged endothelial and podocyte glycocalyx, the use of amine-terminated PAMAM dendrimers >G4 could be considered to reduce renal accumulation. Nevertheless, besides increased molecular weight and size, higher generations of cationic PAMAM dendrimers (>G4) are accompanied by a sharp increase in cytotoxicity based on their increased positive charge density [52,53]. However, additional surface modifications (e.g., PEGylation) of the terminal positively charged amines under physiological conditions could reduce undesired toxicities and further decrease the preference for renal excretion [54,55].
The following anticancer activity experiments revealed a maximum tumor growth inhibition effect of 65.6% for C12 compared to 47.6% for cisplatin. Additionally, the treatment with C12 had no negative influence on body weight, whereas cisplatin application led to a maximal loss of body weight of around 20%, enabling an improved therapeutic window for C12 compared to cisplatin. Furthermore, a trend of extended animal survival with the treatment of C12 could be observed compared to the solvent control and cisplatin group.
Finally, it can be concluded that the combination of platinum(IV) complexes coupled with PAMAM dendrimers enables a promising approach to further improve existing anticancer therapy. The full potential could be exploited by further investigations of the therapeutic window as well as adjustments on the surface of PAMAM dendrimers to further optimize pharmacological properties.