pH-Sensitive Dendrimersomes of Hybrid Triazine-Carbosilane Dendritic Amphiphiles-Smart Vehicles for Drug Delivery

Supramolecular constructions of amphiphilic dendritic molecules are promising vehicles for anti-cancer drug delivery due to the flexibility of their architecture, high drug loading capacity and avoiding off-target effects of a drug. Herein, we report a new class of amphiphilic dendritic species—triazine-carbosilane dendrons readily self-assembling into pH-sensitive dendrimersomes. The dendrimersomes efficiently encapsulate anticancer drugs doxorubicin and methotrexate. Chemodrug-loaded dendrimersomes have dose-related cytotoxic activity against leukaemia cell lines 1301 and K562. Our findings suggest that triazine-carbosilane dendrimersomes are prospective drug carriers for anti-cancer therapy.


Introduction
The use of engineered nanosized constructions as drug delivery systems has become an emerging area of attention in recent years due to their ability to improve pharmacological properties of cargo compounds, i.e., improving selectivity and efficiency of drug delivery into target cells/tissues, reducing off-target effects of drugs, and protecting drugs from the biological environment amid the delivery [1]. Furthermore, nanoconstructions are expected to release a cargo in a controllable manner upon the action of external stimuli (changes of ambient pH or temperature, enzymatic digestion etc.) [2]. In light of this, the use of supramolecular assemblies for drug delivery seems to be very promising [3].

Statistical Analysis
We used GraphPad Prism software and Statistica 7.0 software for data analysis and visualization. The Mann-Whitney criterion was used; the differences were considered to be significant if p<0,05.

Statistical Analysis
We used GraphPad Prism software and Statistica 7.0 software for data analysis and visualization. The Mann-Whitney criterion was used; the differences were considered to be significant if p<0,05.

Synthesis of Dendron Amphiphiles
Cyanuric chloride (970 mg, 5.26 mmol) and dodecylamine (2.0 g, 10.8 mmol) were mixed in 150 mL CHCl3. The mixture was cooled with an ice bath, then 10% aqueous KOH was added up to the pH~10. The reaction mixture was stirred overnight at room temperature, then organic phase was separated, the solvent was evaporated under vacuum, and the solid residue was recrystallized from the mixture CHCl3:CH3OH (5:1) to yield the desired product as white solid (2.28 g, 90%).  2.5.2. 2,4-dodecylamino-6-piperazino-1,3,5-triazine 2,4-didodecylamino-6-chloro-1,3,5-triazine (600 mg, 1.2 mmol) and piperazine (645 mg, 7.5 mmol) were mixed in 15 mL CHCl3; the reaction mixture was stirred at room temperature. When the starting triazine derivative was fully consumed, as shown by TLC (5% CH3OH in CH2Cl2), the solution was washed several times with 1M NaOH, then with water. The organic phase was separated, dried over Na2SO4, and the solvent was removed under vacuum to yield the desired product as yellowish solid (600 mg, 92%). 2,4-didodecylamino-6-piperazino-1,3,5-triazine (220 mg, 0.41 mmol) and vinyl-terminated carbosilane dendron BrG 2 V 4 (140 mg, 0.3 mmol), K 2 CO 3 (70 mg, 0.5 mmol) were mixed in 30 mL acetone in a sealed ampule with catalytic amounts of 18-crown-6 and KI added. The reaction mixture was stirred for 24 h at 90 • C. The reaction completion was monitored by TOCSY 1 H NMR following the disappearance of BrCH 2 protons. When the reaction was over, the solvent was removed under vacuum, the residue was dissolved in ethyl acetate and washed with brine. The organic phase was dried over Na 2 SO 4 , and the solvent was removed. The crude product was purified by silica gel column chromatography (eluent: ethyl acetate:hexane 1:1) to yield functionalized dendron as light-yellow oil (245 mg, 90%). 2,4-didodecylamino-6-piperazino-1,3,5-triazine (220 mg, 0.41 mmol) and vinyl-terminated carbosilane dendron BrG2V4 (140 mg, 0.3 mmol), K2CO3 (70 mg, 0.5 mmol) were mixed in 30 mL acetone in a sealed ampule with catalytic amounts of 18-crown-6 and KI added. The reaction mixture was stirred for 24 h at 90 °C. The reaction completion was monitored by TOCSY 1 H NMR following the disappearance of BrCH2 protons. When the reaction was over, the solvent was removed under vacuum, the residue was dissolved in ethyl acetate and washed with brine. The organic phase was dried over Na2SO4, and the solvent was removed. The crude product was purified by silica gel column chromatography (eluent: ethyl acetate:hexane 1:1) to yield functionalized dendron as light-yellow oil (245 mg, 90%).

Amphiphilic Dendron G2
Functionalized vinyl-terminated dendron (400 mg, 0.44 mmol), 2-(dimethylamino)ethanethiol hydrochloride (275 mg, 1.76 mmol) and dimethoxyphenylacetophenone (DMPA) (12 mg, 0.044 mmol) were dissolved in 5 mL of mixture THF:CH3OH (1:2). The reaction mixture was deoxygenated by bubbling argon and irradiated by UV for 2 h (365 nm, 120W). Then, another 12 mg (0.044 mmol) of DMPA was added, and the reaction mixture was irradiated for another 2 h. The reaction completion was monitored by 1 H NMR. After the reaction was completed, the solvents were removed under vacuum and then the residue was dissolved in methanol. Afterward, it was precipitated in diethyl ether, and after the solvent was separated, the solid was dried under vacuum to afford the desired dendron as light-yellow solid (495 mg, 75%). For characterization, the dendron was deprotonated with K2CO3.

Amphiphilic Dendron G2
Functionalized vinyl-terminated dendron (400 mg, 0.44 mmol), 2-(dimethylamino)ethanethiol hydrochloride (275 mg, 1.76 mmol) and dimethoxyphenylacetophenone (DMPA) (12 mg, 0.044 mmol) were dissolved in 5 mL of mixture THF:CH 3 OH (1:2). The reaction mixture was deoxygenated by bubbling argon and irradiated by UV for 2 h (365 nm, 120W). Then, another 12 mg (0.044 mmol) of DMPA was added, and the reaction mixture was irradiated for another 2 h. The reaction completion was monitored by 1 H NMR. After the reaction was completed, the solvents were removed under vacuum and then the residue was dissolved in methanol. Afterward, it was precipitated in diethyl ether, and after the solvent was separated, the solid was dried under vacuum to afford the desired dendron as light-yellow solid (495 mg, 75%). For characterization, the dendron was deprotonated with K 2 CO 3 . 2,4-didodecylamino-6-piperazino-1,3,5-triazine (220 mg, 0.41 mmol) and vinyl-terminated carbosilane dendron BrG2V4 (140 mg, 0.3 mmol), K2CO3 (70 mg, 0.5 mmol) were mixed in 30 mL acetone in a sealed ampule with catalytic amounts of 18-crown-6 and KI added. The reaction mixture was stirred for 24 h at 90 °C. The reaction completion was monitored by TOCSY 1 H NMR following the disappearance of BrCH2 protons. When the reaction was over, the solvent was removed under vacuum, the residue was dissolved in ethyl acetate and washed with brine. The organic phase was dried over Na2SO4, and the solvent was removed. The crude product was purified by silica gel column chromatography (eluent: ethyl acetate:hexane 1:1) to yield functionalized dendron as light-yellow oil (245 mg, 90%).

Amphiphilic Dendron G2
Functionalized vinyl-terminated dendron (400 mg, 0.44 mmol), 2-(dimethylamino)ethanethiol hydrochloride (275 mg, 1.76 mmol) and dimethoxyphenylacetophenone (DMPA) (12 mg, 0.044 mmol) were dissolved in 5 mL of mixture THF:CH3OH (1:2). The reaction mixture was deoxygenated by bubbling argon and irradiated by UV for 2 h (365 nm, 120W). Then, another 12 mg (0.044 mmol) of DMPA was added, and the reaction mixture was irradiated for another 2 h. The reaction completion was monitored by 1 H NMR. After the reaction was completed, the solvents were removed under vacuum and then the residue was dissolved in methanol. Afterward, it was precipitated in diethyl ether, and after the solvent was separated, the solid was dried under vacuum to afford the desired dendron as light-yellow solid (495 mg, 75%). For characterization, the dendron was deprotonated with K2CO3.

CMC Measurements
Association of dendron molecules into supramolecular constructions was studied by fluorescence spectroscopy using pyrene as a marker [32]. Aliquots of pyrene solution in acetone (5 µL, 20 µM) were dispensed in 200 µL glass vials and the solvent was removed under vacuum. Then, solutions of the amphiphilic dendron (100 µL, 0.3 to 100 µM) in 10 mM sodium phosphate buffer, pH 7.0 (PB) or 10 mM phosphate-buffered saline, pH 7.0-7.4 (PBS) were added. Samples were sonicated for 30 min and then incubated for 2 days at room temperature. After the incubation, the samples were transferred into the wells of 96-well black microplate COSTAR 96 half-area, and pyrene fluorescence spectra were recorded in the range 360-500 nm upon excitation at 345 ± 5 nm. The ratio of fluorescence intensities I 373 /I 383 was plotted against −lgC, fitted with Boltzmann sigmoidal curve (r 2 > 0.95), and the −lg(CMC) values were estimated from the fitting data.

Study of the pH-Dependence of the Particle Size
The water solution of amphiphilic dendron in a glass vial (1 mL, 100 µM) was sonicated for 30 min and then incubated overnight at room temperature. 100 µL aliquots of the solution were taken into plastic Eppendorf-type test tubes, and 5 µL of 200 mM sodium phosphate buffer (pH 7.0; 6.5; 6.0; 5.5; 5.0) was added. The samples were incubated at room temperature for 2 h, then DLS profiles of the samples were recorded.

Drug Encapsulation
A drop of water solution of amphiphilic dendron (10 µL, 10 mM) and a drop of water solution (10 µL, 10 mM) of a doxorubicin hydrochloride (DOX), sodium methotrexate (MTX) or 5-fluorouracil (5FU) were rapidly mixed in 1 mL of deionized water; an UV-vis spectrum of the sample [As-mixed] was recorded. The solution was sonicated for 30 min and then incubated for 2 days at room temperature; an UV-vis spectrum of the sample [After treatment] was recorded. Non-encapsulated chemodrug was removed by dialysis against deionized water (4 × 200 mL, 6 h); an UV-vis spectrum of the sample [After dialysis] was recorded. The degree of chemodrug encapsulation was estimated from UV-vis spectra as a ratio of a chemodrug absorption at λ max before and after dialysis. Drug loading capacity was estimated as a ratio of the weight of encapsulated drug to the total weight of nanoparticles neglecting the weight of the dendron washed off during the dialysis. The real loading capacity values are thus higher than estimated.
The UV-vis spectra of chemodrugs at different stages of encapsulation into dendrimersomes are shown in the Figure 1. DLS and zeta potential distribution profiles of drug-loaded dendrimersomes were recorded. Nanomaterials 2020, 10, x FOR PEER REVIEW 6 of 15

Drug Release from Dendrimersomes
Doxorubicin-loaded dendrimersomes solution (65 µM DOX) was placed in a 200 µL Eppendorf-type plastic tube, sealed with a dialysis membrane and incubated in 5 mL of 10 mM Na phosphate buffer pH 7.0; 6.5; 6.0 for 24 h upon gentle stirring. At given time points, 50 µL aliquots of dialysis buffer were taken, and their fluorescence was read at 590 nm. Fluorescence intensity was plotted vs. incubation time and fitted with a first order kinetic curve (r 2 > 0.95). After cultivation, 10 µL of the WST-1 reagent (Takara Bio Inc, Kusatsu, Japan) per well was added, and the plates were incubated for 4 h. To evaluate the cell viability, optical absorbance at 450 nm was directly read against the background control, the reference was read at 620 nm (TriStar LB 941 Multimode Microplate Reader, Berthold Technologies GmbH&Co., Bad Wildbad, Germany). Cell viability was calculated as a ratio of absorbance of treated cells samples to that of non-treated control, then converted into percentage. Cell viability values were plotted against lgC, fitted with Boltzmann sigmoidal curve (r 2 > 0.95), and the -lg(IC50) values were estimated from the fitting data.

Drug Internalization Studies
0.5 × 10 6 1301 cells were cultivated in total volume 100 µL in flat-bottomed 48-well cell culture plates (TPP, Switzerland) for 4 h. Then, water solutions of amphiphilic dendron (5 µM), doxorubicin (3 µM) or doxorubicin-loaded dendrimersomes (doxorubicin content 3 µM) were added. The equal volume of PBS was added to control cells. After cultivation, cells were collected, washed once with PBS, then 50 µL of acidic glycine solution were added, gently pipetted for 30 s and washed again with PBS. Flow cytometric analysis was performed on a FACSCanto II cytometer (BD Biosciences, San Jose, CA). The analysis was made in FITC and PE channels. 15,000 to 50,000 events per sample were acquired and analysed using FACSDiva 6.1.2 software. 1301 cells were incubated with blank dendrimersomes, doxorubicin and doxorubicin-loaded dendrimersomes (6.6 µM DOX) in complete cell culture media for 4 h in humidified atmosphere containing 5% CO2 at 37 °C, then fixed in ethanol and glacial acetic acid. Samples were stained with DAPI solution (1.5 µg/mL) with the addition of antifade solution (10 µL) for 20 min, protected from

Drug Release from Dendrimersomes
Doxorubicin-loaded dendrimersomes solution (65 µM DOX) was placed in a 200 µL Eppendorf-type plastic tube, sealed with a dialysis membrane and incubated in 5 mL of 10 mM Na phosphate buffer pH 7.0; 6.5; 6.0 for 24 h upon gentle stirring. At given time points, 50 µL aliquots of dialysis buffer were taken, and their fluorescence was read at 590 nm. Fluorescence intensity was plotted vs. incubation time and fitted with a first order kinetic curve (r 2 > 0.95). After cultivation, 10 µL of the WST-1 reagent (Takara Bio Inc, Kusatsu, Japan) per well was added, and the plates were incubated for 4 h. To evaluate the cell viability, optical absorbance at 450 nm was directly read against the background control, the reference was read at 620 nm (TriStar LB 941 Multimode Microplate Reader, Berthold Technologies GmbH&Co., Bad Wildbad, Germany). Cell viability was calculated as a ratio of absorbance of treated cells samples to that of non-treated control, then converted into percentage. Cell viability values were plotted against lgC, fitted with Boltzmann sigmoidal curve (r 2 > 0.95), and the −lg(IC 50 ) values were estimated from the fitting data.

Synthesis of the Amphiphilic Dendron
The novel dendritic amphiphile designed herein consists of a branched hydrophobic moiety (substituted triazine) and a cationic carbosilane dendron conjugated through a piperazine moiety, a convenient linker allowing further dendron grafting in relatively mild conditions.
We have synthesized the amphiphile in a convergent way (Scheme 1). The hydrophobic triazine unit has been obtained by the substitution of two chlorides in cyanuric chloride with dodecylamino-residues followed by the subsequent grafting of piperazine. The vinyl-terminated carbosilane dendron G2 having bromine in the focal point has been obtained starting from 4-bromobutylmethyldichlorosilane by iterative hydrosilylation and Grignard reaction steps as described earlier [33]. Two units, namely triazine and dendron ones, were conjugated by simple nucleophilic substitution, and vinyl residues at the periphery of the dendron have been modified via thiol-ene reaction. The resulting amphiphilic molecule contains two hydrophobic tails long enough to form a bilayer, a stimuli-sensitive triazine unit, and a branched polycationic dendron unit.

Synthesis of the Amphiphilic Dendron
The novel dendritic amphiphile designed herein consists of a branched hydrophobic moiety (substituted triazine) and a cationic carbosilane dendron conjugated through a piperazine moiety, a convenient linker allowing further dendron grafting in relatively mild conditions.
We have synthesized the amphiphile in a convergent way (Scheme 1). The hydrophobic triazine unit has been obtained by the substitution of two chlorides in cyanuric chloride with dodecylamino-residues followed by the subsequent grafting of piperazine. The vinyl-terminated carbosilane dendron G2 having bromine in the focal point has been obtained starting from 4-bromobutylmethyldichlorosilane by iterative hydrosilylation and Grignard reaction steps as described earlier [33]. Two units, namely triazine and dendron ones, were conjugated by simple nucleophilic substitution, and vinyl residues at the periphery of the dendron have been modified via thiol-ene reaction. The resulting amphiphilic molecule contains two hydrophobic tails long enough to form a bilayer, a stimuli-sensitive triazine unit, and a branched polycationic dendron unit.

Self-Assembly of Dendrimersomes
Dendritic amphiphile molecules are efficiently self-organised into supramolecular associates in water medium. The CMC determination plots in PB and PBS are given in the Figure 2. Comparing two plots, a clear difference is observed. In a low-salt solution (PB), there is only one association stage, quite broad in terms of concentration range, with CMC of 9.5 ± 1.2 µM. In contrast, at physiological salt concentration (PBS), there are two association stages with CMC of 2.1 ± 1.0 µM and 21 ± 1.1 µM. Apparently, the high salt concentration fosters the association of dendron molecules at low concentrations, with these aggregates further rearranging into vesicle-like dendrimersomes at higher concentrations. Nanomaterials 2020, 10, x FOR PEER REVIEW 8 of 15

Self-Assembly of Dendrimersomes
Dendritic amphiphile molecules are efficiently self-organised into supramolecular associates in water medium. The CMC determination plots in PB and PBS are given in the Figure 2. Comparing two plots, a clear difference is observed. In a low-salt solution (PB), there is only one association stage, quite broad in terms of concentration range, with CMC of 9.5 ± 1.2 µM. In contrast, at physiological salt concentration (PBS), there are two association stages with CMC of 2.1 ± 1.0 µM and 21 ± 1.1 µM. Apparently, the high salt concentration fosters the association of dendron molecules at low concentrations, with these aggregates further rearranging into vesicle-like dendrimersomes at higher concentrations.  Vesicle-like nanoconstructions, dendrimersomes, are readily assembled upon quite simple treatment (short sonication and relaxation). High uniformity of nanoconstructions' topology and size, likely defined by the molecular structure of the dendron precursor, is achieved without specific post-treatment (e.g., extrusion). Having been formed at neutral pH, dendrimersomes have mean diameter of 20-30 nm; however, when put into acidic medium (pH below 6.5), they reorganize into larger particles 100-150 nm diameter (Figure 3). TEM shows that the vesicle-like structure of dendrimersomes is maintained upon reorganization.
Based on the findings above, we suggest a provisional scheme of the bilayer behaviour ( Figure  3D). The main driving force of self-assembly is the interaction of hydrophobic tails of amphiphilic dendron molecules additionally stabilized by stacking interactions of triazine moieties [34]. On the other hand, the bilayer structure is destabilized by the steric and electrostatic repulsion of dendron fragments. Such a balance of factors favouring and unfavouring the bilayer stability likely gives the supramolecular flexibility to the whole dendrimersome. In acidic medium, triazine moieties get protonated, and the stacking is distorted. Thus, the stabilizing factor turns into the destabilizing one; this causes loosening of the bilayer followed by fusion with other vesicles. Vesicle-like nanoconstructions, dendrimersomes, are readily assembled upon quite simple treatment (short sonication and relaxation). High uniformity of nanoconstructions' topology and size, likely defined by the molecular structure of the dendron precursor, is achieved without specific post-treatment (e.g., extrusion). Having been formed at neutral pH, dendrimersomes have mean diameter of 20-30 nm; however, when put into acidic medium (pH below 6.5), they reorganize into larger particles 100-150 nm diameter (Figure 3). TEM shows that the vesicle-like structure of dendrimersomes is maintained upon reorganization.
Based on the findings above, we suggest a provisional scheme of the bilayer behaviour ( Figure 3D). The main driving force of self-assembly is the interaction of hydrophobic tails of amphiphilic dendron molecules additionally stabilized by stacking interactions of triazine moieties [34]. On the other hand, the bilayer structure is destabilized by the steric and electrostatic repulsion of dendron fragments. Such a balance of factors favouring and unfavouring the bilayer stability likely gives the supramolecular flexibility to the whole dendrimersome. In acidic medium, triazine moieties get protonated, and the stacking is distorted. Thus, the stabilizing factor turns into the destabilizing one; this causes loosening of the bilayer followed by fusion with other vesicles.
It is worth noting that the effects of pH-sensitivity are observed at slightly acidic pH (below 6.5), which corresponds to early stages of endosome maturation. Thus, once delivered into a cell by endocytosis, an encapsulated biologically active cargo can be released into the cytosol in the most favourable moment to escape endosomes. Fusing dendrimersomes can contribute to the endosomal escape release by assisting the endosome membrane disruption.  It is worth noting that the effects of pH-sensitivity are observed at slightly acidic pH (below 6.5), which corresponds to early stages of endosome maturation. Thus, once delivered into a cell by endocytosis, an encapsulated biologically active cargo can be released into the cytosol in the most favourable moment to escape endosomes. Fusing dendrimersomes can contribute to the endosomal escape release by assisting the endosome membrane disruption.

Drug Loading Into Dendrimersomes
To explore the potential of dendrimersomes as drug carriers, we encapsulated three anti-tumor drugs-doxorubicin, methotrexate and 5-fluorouracil into dendrimersomes. Concentrated solutions of a drug and a dendron were rapidly mixed in water, sonicated and incubated for 48 h, followed by removal of non-encapsulated drug by dialysis. This procedure yields drug-loaded dendrimersomes with low size dispersion (PDI < 0.3) with a high degree of drug encapsulation (ca. 65% for doxorubicin and 75% for methotrexate) ( Table 1). On the contrary, 5-fluorouracil encapsulated much less efficiently: for ca. 25%. The differences in the drug encapsulation efficiency resulted in 10-fold variation of the drug loading capacity: ~20% for doxorubicin and methotrexate vs. ~2% for 5-fluorouracil (Table 1). Such a dissimilarity is likely explained by the hydrophobicity of the encapsulated compound: doxorubicin and methotrexate can not only be entrapped into the inner cavity of a dendrimersomes but also incorporated into a bilayer (that can be observed in the TEM images ( Figure 4) by the thickening of dendrimersomes' membranes), whereas for 5-fluorouracil, the entrapment likely prevails. Thus, the efficiency of 5-fluorouracil encapsulation can give an idea of the efficiency of the physical entrapment of compounds into the hydrophilic interior of dendrimersomes. It should be noted that methotrexate appeared to be additionally retained on the dendrimersomes' surface due to electrostatic interactions, as suggested by zeta potential values and TEM images. Drug-loaded dendrimersomes have mean diameter less than 100 nm that makes them suitable for drug delivery studies (Table 1).

Figure 3.
Self-assembly of dendritic amphiphiles. DLS profiles for supramolecular associates of dendron (100 µM) exposed to different pH in 10 mM Na-phosphate buffer (A); representative TEM images of dendrons' supramolecular associates at pH 7.0 (B) and pH 5.0 (C); tentative scheme of the pH-dependent behaviour of a dendron bilayer (D).

Drug Loading into Dendrimersomes
To explore the potential of dendrimersomes as drug carriers, we encapsulated three anti-tumor drugs-doxorubicin, methotrexate and 5-fluorouracil into dendrimersomes. Concentrated solutions of a drug and a dendron were rapidly mixed in water, sonicated and incubated for 48 h, followed by removal of non-encapsulated drug by dialysis. This procedure yields drug-loaded dendrimersomes with low size dispersion (PDI < 0.3) with a high degree of drug encapsulation (ca. 65% for doxorubicin and 75% for methotrexate) ( Table 1). On the contrary, 5-fluorouracil encapsulated much less efficiently: for ca. 25%. The differences in the drug encapsulation efficiency resulted in 10-fold variation of the drug loading capacity:~20% for doxorubicin and methotrexate vs.~2% for 5-fluorouracil (Table 1). Such a dissimilarity is likely explained by the hydrophobicity of the encapsulated compound: doxorubicin and methotrexate can not only be entrapped into the inner cavity of a dendrimersomes but also incorporated into a bilayer (that can be observed in the TEM images ( Figure 4) by the thickening of dendrimersomes' membranes), whereas for 5-fluorouracil, the entrapment likely prevails. Thus, the efficiency of 5-fluorouracil encapsulation can give an idea of the efficiency of the physical entrapment of compounds into the hydrophilic interior of dendrimersomes. It should be noted that methotrexate appeared to be additionally retained on the dendrimersomes' surface due to electrostatic interactions, as suggested by zeta potential values and TEM images. Drug-loaded dendrimersomes have mean diameter less than 100 nm that makes them suitable for drug delivery studies (Table 1).  We have studied the pH dependence of the drug release from dendrimersomes using doxorubicin-loaded constructions as a model. The release was followed by measuring fluorescence intensity of dialysate (see Section 2.9). The rate of the drug release from dendrimersomes has not been found to increase significantly in slightly acidic medium ( Figure S1): The half-release time being 6.3 ± 1.4 h at the pH 7.0 decreased to 6.0 ± 1.5 h (pH 6.5) and 5.4 ± 1.4 h (pH 6.0). However, overall fluorescence intensity of dialysate was higher with the decrease of the pH. The completeness of the doxorubicin release from dendrimersomes after 24 h was >80%. Thus, the pH-dependent behaviour shown for blank dendrimersomes (Figure 3) likely takes place in drug-loaded constructions as well, triggering the release of a drug in slightly acidic media frequently occurring in endocytic vesicles at early stages of endosome maturation [35].

Cytotoxic Activity of Drug-Loaded Dendrimersomes Towards Cancer Cells
We have tested the performance of triazine-carbosilane dendrimersomes as drug carriers by doing a comparative study of cytotoxic effects of nanoconstructions loaded with doxorubicin and methotrexate towards two leukaemia cell lines, 1301 and K562. 5-Fluorouracil-loaded dendrimersomes were excluded from the study due to the insufficient encapsulation efficiency resulting in the low concentration of the drug in the nanoparticle preparations.
Cell lines 1301 and K562 have been chosen as targets, for they represent leukaemia variants related to different types of progenitor cells: 1301 is the acute T-cell leukaemia cell line; K562 is chronic myelogenous leukaemia cell line derived from the cells of a patient having blast crises [36]. Thus, the use of these cell lines permits to estimate the efficiency of drug-loaded nanoconstructions in the haemoblastosis therapy. We have studied the pH dependence of the drug release from dendrimersomes using doxorubicin-loaded constructions as a model. The release was followed by measuring fluorescence intensity of dialysate (see Section 2.9). The rate of the drug release from dendrimersomes has not been found to increase significantly in slightly acidic medium ( Figure S1): The half-release time being 6.3 ± 1.4 h at the pH 7.0 decreased to 6.0 ± 1.5 h (pH 6.5) and 5.4 ± 1.4 h (pH 6.0). However, overall fluorescence intensity of dialysate was higher with the decrease of the pH. The completeness of the doxorubicin release from dendrimersomes after 24 h was >80%. Thus, the pH-dependent behaviour shown for blank dendrimersomes (Figure 3) likely takes place in drug-loaded constructions as well, triggering the release of a drug in slightly acidic media frequently occurring in endocytic vesicles at early stages of endosome maturation [35].

Cytotoxic Activity of Drug-Loaded Dendrimersomes Towards Cancer Cells
We have tested the performance of triazine-carbosilane dendrimersomes as drug carriers by doing a comparative study of cytotoxic effects of nanoconstructions loaded with doxorubicin and methotrexate towards two leukaemia cell lines, 1301 and K562. 5-Fluorouracil-loaded dendrimersomes were excluded from the study due to the insufficient encapsulation efficiency resulting in the low concentration of the drug in the nanoparticle preparations.
Cell lines 1301 and K562 have been chosen as targets, for they represent leukaemia variants related to different types of progenitor cells: 1301 is the acute T-cell leukaemia cell line; K562 is chronic myelogenous leukaemia cell line derived from the cells of a patient having blast crises [36]. Thus, the use of these cell lines permits to estimate the efficiency of drug-loaded nanoconstructions in the haemoblastosis therapy.
The ability of drug-loaded dendrimersomes to suppress the viability of target cell lines was studied using WST-1 assay. Dose-response cytotoxic effects of nanoconstructions were observed. Doxorubicin-loaded dendrimersomes have been found to be less cytotoxic in comparison with free drug at low concentrations (below 1 µM) and exhibit similar activity at higher concentrations ( Figure 5A,B). These differences are likely connected with the slow release of doxorubicin from dendrimersomes, so, being taken at low concentrations, they do not cause a commeasurable cytotoxic effect, which results in higher cell viability after treatment. In the case of methotrexate, free drug and drug-loaded dendrimersomes possess similar cytotoxicity ( Figure 5C,D): In most cases, there were no statistically significant differences found between two groups. In general, the cytotoxic effects have been observed at dendron concentrations below the IC 50 (Table 2, Figure S2), which permits to assign them to the encapsulated chemodrugs.
there were no statistically significant differences found between two groups. In general, the cytotoxic effects have been observed at dendron concentrations below the IC50 (Table 2, Figure S2), which permits to assign them to the encapsulated chemodrugs.
The differences observed between effects of doxorubicin-and methotrexate-loaded dendrimersomes on the viability of target cells are likely explained by the distribution of drugs in dendrimersomes. Since doxorubicin is mostly retained in the hydrophobic part of a vesicle, it is released as a result of reorganization (or even dissociation) of the carrier nanoparticle. Meanwhile, methotrexate is retained mostly at the surface of dendrimersomes, so it can be released relatively easier.    The differences observed between effects of doxorubicin-and methotrexate-loaded dendrimersomes on the viability of target cells are likely explained by the distribution of drugs in dendrimersomes. Since doxorubicin is mostly retained in the hydrophobic part of a vesicle, it is released as a result of reorganization (or even dissociation) of the carrier nanoparticle. Meanwhile, methotrexate is retained mostly at the surface of dendrimersomes, so it can be released relatively easier.
To understand the effects of drug-loaded dendrimersomes better, we have quantified cell subpopulations at different stages of cell death after the treatment with blank dendrimersomes, free drug and drug-loaded nanoconstructions. As a model system, we have taken doxorubicin-loaded dendrimersomes and 1301 cells as a cell model.
Both free doxorubicin and doxorubicin-loaded dendrimersomes have been shown to be efficiently accumulated in 1301 cells (Figures S3 and S4). However, treating cells with free doxorubicin (3 µM) resulted in slight increasing of early and late apoptosis cell fractions in comparison with control (ca. 10% each), whereas doxorubicin-loaded dendrimersomes (3 µM doxorubicin) caused a sharp increase of late apoptosis and necrosis cell fractions (50% and 20%, respectively) ( Figure 6 and Figure  S5). Blank dendrimersomes do not cause any significant effect on the 1301 cells in comparison to control at the concentration used (5 µM dendron). Certain differences with WST data likely occurred due to methodological features of the techniques. dendrimersomes and 1301 cells as a cell model.
Both free doxorubicin and doxorubicin-loaded dendrimersomes have been shown to be efficiently accumulated in 1301 cells (Figures S3 and S4). However, treating cells with free doxorubicin (3 µM) resulted in slight increasing of early and late apoptosis cell fractions in comparison with control (ca. 10% each), whereas doxorubicin-loaded dendrimersomes (3 µM doxorubicin) caused a sharp increase of late apoptosis and necrosis cell fractions (50% and 20%, respectively) (Figures 6 and S5). Blank dendrimersomes do not cause any significant effect on the 1301 cells in comparison to control at the concentration used (5 µM dendron). Certain differences with WST data likely occurred due to methodological features of the techniques. Our findings show that the encapsulation does not improve the rate of accumulation of doxorubicin into tumour cells, however, the cytotoxic effect of the doxorubicin-containing nanoconstruction is visibly higher. This likely occurs due to the switch of the mechanism of drug penetration into a cell. Remarkably, the presence of serum, an important component of the cell culture medium, does not suppress the dendrimersomes' penetration into cells nor their cytostatic activity. The interaction of nanoconstructions with serum components is known to imminently take place during co-incubation with cells. We suppose that the encapsulation of anti-cancer chemodrugs into dendrimersomes can prolongate their persistence in the organism as well as reduce side effects. Such a phenomenon has been reported while using supramolecular associates [37] and macromolecules, in particular, dendrimers [38,39], as carriers for chemodrugs. In respect of novel amphiphilic triazine-carbosilane dendrimersomes reported herein, this hypothesis deserves further validation in animal experiments.

Conclusions
In summary, we have designed a new class of dendritic amphiphiles self-assembling into vesicle-like nanosized supramolecular associates (dendrimersomes). Rationally designed molecular topology of dendrons permits to use simple procedures to yield monodisperse nanoparticles. Because the reorganization is driven by subtle changes of ambient pH, triazine-carbosilane dendrimersomes can be useful as carriers for the delivery of sensitive, poorly soluble or highly toxic compounds into cells. As a proof-of-concept study, we attempted to encapsulate anti-cancer drugs doxorubicin and methotrexate and to deliver them into human leukaemia cells. The encapsulation, though efficient, does not change the vesicular topology of nanoconstructions. Drug-loaded dendrimersomes efficiently penetrate into cells and induce cell death. Our findings suggest that Our findings show that the encapsulation does not improve the rate of accumulation of doxorubicin into tumour cells, however, the cytotoxic effect of the doxorubicin-containing nanoconstruction is visibly higher. This likely occurs due to the switch of the mechanism of drug penetration into a cell. Remarkably, the presence of serum, an important component of the cell culture medium, does not suppress the dendrimersomes' penetration into cells nor their cytostatic activity. The interaction of nanoconstructions with serum components is known to imminently take place during co-incubation with cells. We suppose that the encapsulation of anti-cancer chemodrugs into dendrimersomes can prolongate their persistence in the organism as well as reduce side effects. Such a phenomenon has been reported while using supramolecular associates [37] and macromolecules, in particular, dendrimers [38,39], as carriers for chemodrugs. In respect of novel amphiphilic triazine-carbosilane dendrimersomes reported herein, this hypothesis deserves further validation in animal experiments.

Conclusions
In summary, we have designed a new class of dendritic amphiphiles self-assembling into vesicle-like nanosized supramolecular associates (dendrimersomes). Rationally designed molecular topology of dendrons permits to use simple procedures to yield monodisperse nanoparticles. Because the reorganization is driven by subtle changes of ambient pH, triazine-carbosilane dendrimersomes can be useful as carriers for the delivery of sensitive, poorly soluble or highly toxic compounds into cells. As a proof-of-concept study, we attempted to encapsulate anti-cancer drugs doxorubicin and methotrexate and to deliver them into human leukaemia cells. The encapsulation, though efficient, does not change the vesicular topology of nanoconstructions. Drug-loaded dendrimersomes efficiently penetrate into cells and induce cell death. Our findings suggest that triazine-carbosilane dendrimersomes hold considerable potential for nanomedicine as stimuli-sensitive drug carriers.