Synthesis of Dense and Chiral Dendritic Polyols Using Glyconanosynthon Scaffolds

Most classical dendrimers are frequently built-up from identical repeating units of low valency (usually AB2 monomers). This strategy necessitates several generations to achieve a large number of surface functionalities. In addition, these typical monomers are achiral. We propose herein the use of sugar derivatives consisting of several and varied functionalities with their own individual intrinsic chirality as both scaffolds/core as well as repeating units. This approach allows the construction of chiral, dense dendrimers with a large number of surface groups at low dendrimer generations. Perpropargylated β-d-glucopyranoside, serving as an A5 core, together with various derivatives, such as 2-azidoethyl tetra-O-allyl-β-d-glucopyranoside, serving as an AB4 repeating moiety, were utilized to construct chiral dendrimers using “click chemistry” (CuAAC reaction). These were further modified by thiol-ene and thiol-yne click reactions with alcohols to provide dendritic polyols. Molecular dynamic simulation supported the assumption that the resulting polyols have a dense structure.


Introduction
Dendrimers are hyperbranched, globular shaped, well-defined macromolecules possessing a wide range of applications in nanomedicine, novel dendrimer space concept in medicinal chemistry, as multivalent carriers for drug delivery, material sciences, catalysis, and gene therapy [1][2][3][4][5][6][7][8]. During the last three decades, there has been tremendous growth in this field motivating chemists and material scientists to conceive new ways to synthesize these molecules more efficiently, rapidly, and in a cost effective way. Most of the common synthetic methods start with low-valent scaffolds with reactive functionalities followed by the build-up of higher generation architectures using repetitive building blocks. The syntheses are accomplished either divergently or convergently. Unfortunately, the introductions of a large number of surface groups are usually achieved at the expense of high generation. To counterbalance this situation, only a handful of examples have been described to generate dendrimers with a high number of end groups at low generation, and the use of sugars (coined "glyconanosynthons") has been recently reviewed to achieve this goal [5]. Hence, the use of carbohydrates as either core or multivalent, high valency branching building blocks is clearly immerging as an alternative approach to achieve higher numbers of surface groups in a fewer steps. This approach [9][10][11] has not been systematically investigated, particularly in regard to their inherent diversified stereochemical identities, functional group diversities and the chemical orthogonalities these molecules can offer [5]. Propargyl 2,3,4,6-tetra-O-propargyl-β-D-glucopyranoside (5) was treated with excess 2-azidoethyl 2,3,4,6-tetra-O-acetyl-β-D-glucopyranoside (3) (Scheme 2) using copper mediated CuAAC, reaction. All propargyl groups of 5 reacted with the azido-functionalities of 3 forming five distinct triazole rings. Indeed, NMR spectrum of the resulting product 10 showed four singlets at δ 8.04 (1H), 7.93 (1H) 7.88 (2H), 7.71 (1H) and absence of any residual propargyl signal. Similarly, excess of 2-azido ethyl 2,3,4,6-tetra-O-allyl-β-D-glucopyranoside (7) was treated with propargyl 2,3,4,6-tetra-O-propargyl-β-D-glucopyranoside (5) under ''Click'' condition using CuI·P(OEt)3 [16] in dry toluene at 70 °C under microwave yielding perallylated product 12. The presence of five signals at δ 8.08, 8.03, 7.98, 7.93 and 7.80 ppm due to five triazole rings and absence of signal at 2.41-2.51 ppm due to propargyl groups in the NMR spectrum indicated complete conversion. Dendrimer (10) was de-O-acetylated under Zemplén condition using NaOMe in MeOH-water mixture (9:1) yielding perhydroxylated compound (11). This served as a common intermediate for the synthesis of both perallylated (12) and perpropargylated 14 derivatives. Perallylation [23] of (11) was accomplished by forming the anion from 11 using NaH in DMF first, then by reacting with excess allyl bromide at room temperature for 18 h. Perallylated 12 was obtained in good yield (77%) and was identical in TLC and NMR with the product obtained from the ''Click'' reaction involving 5 and 7. Moreover, it also showed a perfect match in its ESI-HRMS spectrum.
Excess of TMS-protected propargyl derivative 9 was treated with perpropargylated glucopyranoside 5 under "Click" condition using soluble copper reagent (CuI¨P(OEt) 3 ) in the presence of Hunig's base [22] at 70˝C for 20 h affording per-TMS-propargylated dendrimer 13 in 90% yield. The use of Hunig's base was necessary to avoid loss of TMS groups during the reaction and to improve the yield. TMS-protected propargyl ethers were removed by treatment with tetrabutylammonium fluoride (TBAF) [22] buffered with acetic acid in THF, yielding perpropargylated dendrimer (14) in 85% yield. The same perpropargylated product was obtained by direct propargylation [24] of the anion derived from 11 by treating with NaH in DMF in presence of excess propargyl bromide in acceptable yield (70%). Both the products were identical by TLC and by NMR spectroscopy (Figure 1) and showed once again a perfect ESI-HRMS spectrum.

Molecular Dynamics Simulations
In order to gain additional insight into these interesting dense nanostructures, we created an atomistic model for dendrimer 16 and have simulated it in a box filled of explicit water molecules by means of all-atom molecular dynamics (MD) simulation. The initially extended conformation of 16 has been equilibrated during 200 ns of MD simulation. During this time, the dendrimer folded assuming a quasi-globular shape (Figures 2 and 3). The radius of gyration of 16 was found to be Rg = 9.8 ± 0.5 Å, while the strong folding was found to be stable during the MD simulation (see Supplementary Materials).  All monomeric building blocks were properly characterized by NMR ( 1 H, 13 C and COSY), IR and mass spectrometry. Dendrimers 10, 11, 12 and 14 showed correct mass corresponding to their molecular formula in HRMS and exhibited good 1 H-NMR and 13 C-NMR in agreement with the assigned structures. All dendrimers 15, 16, 17, 18, 19, 20, and 21 showed molecular ions in MALDI-TOF, though in some cases M + signals were broad signals, likely due to the facile loss of water molecules or the presence of varied cations (Na + , K + , NH 4 + ) in the polyols. Dendrimers 20 and 21 exhibited good GPC curves (single peaks) indicating their mono-dispersity (see Supplementary Materials).

Molecular Dynamics Simulations
In order to gain additional insight into these interesting dense nanostructures, we created an atomistic model for dendrimer 16 and have simulated it in a box filled of explicit water molecules by means of all-atom molecular dynamics (MD) simulation. The initially extended conformation of 16 has been equilibrated during 200 ns of MD simulation. During this time, the dendrimer folded assuming a quasi-globular shape (Figures 2 and 3). The radius of gyration of 16 was found to be R g = 9.8˘0.5 Å, while the strong folding was found to be stable during the MD simulation (see Supplementary Materials). All monomeric building blocks were properly characterized by NMR ( 1 H, 13 C and COSY), IR and mass spectrometry. Dendrimers 10, 11, 12 and 14 showed correct mass corresponding to their molecular formula in HRMS and exhibited good 1 H-NMR and 13 C-NMR in agreement with the assigned structures. All dendrimers 15, 16, 17, 18, 19, 20, and 21 showed molecular ions in MALDI-TOF, though in some cases M + signals were broad signals, likely due to the facile loss of water molecules or the presence of varied cations (Na + , K + , NH4 + ) in the polyols. Dendrimers 20 and 21 exhibited good GPC curves (single peaks) indicating their mono-dispersity (see Supplementary Materials).

Molecular Dynamics Simulations
In order to gain additional insight into these interesting dense nanostructures, we created an atomistic model for dendrimer 16 and have simulated it in a box filled of explicit water molecules by means of all-atom molecular dynamics (MD) simulation. The initially extended conformation of 16 has been equilibrated during 200 ns of MD simulation. During this time, the dendrimer folded assuming a quasi-globular shape (Figures 2 and 3). The radius of gyration of 16 was found to be Rg = 9.8 ± 0.5 Å, while the strong folding was found to be stable during the MD simulation (see Supplementary Materials).   From the equilibrated phase MD trajectories, we obtained the radial distribution functions (g(r)) of the atoms of 16, of the surface groups, and of the water molecules calculated using the dendrimer's center of mass ( Figure 4). The g(r) curves are indicative of the probability density for finding certain groups of the dendrimer in space and are useful to understand the level of density in the various regions of the dendrimer structure [26], of hydration and water penetration [27,28] into the scaffold and also of the displacement of the surface groups at the dendrimer's surface both in terms of surface groups crowding and directionality [29,30].
It is interesting to note that the high level of branching in the scaffold of 16 produces high-density in the dendrimer's interior and reduced surface group backfolding. Seen in Figure 4b, the g(r) of the surface OH-groups of the dendrimer (red) are most probably found at greater distance from the dendrimer's center than all other atoms of 16. This indicates that the surface groups well surround the dendrimer scaffold. This picture of high interior density is well consistent with the g(r) of the water molecules (Figure 4c), demonstrating that water penetration is present only close to the surface, identified by Rg (no water penetration for R < 0.5 Rg).  Table S1 reports other interesting structural data for 16. In particular, comparison between dendrimer's Rg and solvent accessible surface area (SASA) provides an interesting cue on the level of porosity of this dendrimer. Namely, SASA is the surface in contact with the solvent. From this parameter, it is possible to obtain RSASA, which is the radius of a sphere having surface equivalent with that of the dendrimer in contact with the solution as: RSASA = sqrt(SASA/4π). RSASA is larger than Rg, because the dendrimer is not a rigid sphere but a porous soft macromolecule with a non-perfect surface. Comparison between the volume of the voids in the dendrimer and the full one leads to a quantification of 16's porosity (0.78, see Table S1), which is slightly higher than that of From the equilibrated phase MD trajectories, we obtained the radial distribution functions (g(r)) of the atoms of 16, of the surface groups, and of the water molecules calculated using the dendrimer's center of mass ( Figure 4). The g(r) curves are indicative of the probability density for finding certain groups of the dendrimer in space and are useful to understand the level of density in the various regions of the dendrimer structure [26], of hydration and water penetration [27,28] into the scaffold and also of the displacement of the surface groups at the dendrimer's surface both in terms of surface groups crowding and directionality [29,30]. From the equilibrated phase MD trajectories, we obtained the radial distribution functions (g(r)) of the atoms of 16, of the surface groups, and of the water molecules calculated using the dendrimer's center of mass ( Figure 4). The g(r) curves are indicative of the probability density for finding certain groups of the dendrimer in space and are useful to understand the level of density in the various regions of the dendrimer structure [26], of hydration and water penetration [27,28] into the scaffold and also of the displacement of the surface groups at the dendrimer's surface both in terms of surface groups crowding and directionality [29,30].
It is interesting to note that the high level of branching in the scaffold of 16 produces high-density in the dendrimer's interior and reduced surface group backfolding. Seen in Figure 4b, the g(r) of the surface OH-groups of the dendrimer (red) are most probably found at greater distance from the dendrimer's center than all other atoms of 16. This indicates that the surface groups well surround the dendrimer scaffold. This picture of high interior density is well consistent with the g(r) of the water molecules (Figure 4c), demonstrating that water penetration is present only close to the surface, identified by Rg (no water penetration for R < 0.5 Rg).  Table S1 reports other interesting structural data for 16. In particular, comparison between dendrimer's Rg and solvent accessible surface area (SASA) provides an interesting cue on the level of porosity of this dendrimer. Namely, SASA is the surface in contact with the solvent. From this parameter, it is possible to obtain RSASA, which is the radius of a sphere having surface equivalent with that of the dendrimer in contact with the solution as: RSASA = sqrt(SASA/4π). RSASA is larger than Rg, because the dendrimer is not a rigid sphere but a porous soft macromolecule with a non-perfect surface. Comparison between the volume of the voids in the dendrimer and the full one leads to a quantification of 16's porosity (0.78, see Table S1), which is slightly higher than that of scaffold, red: OH-surface groups; (b) radial distribution functions g(r) of the atoms of 16 (black) and of the surface OH-groups (red) as a function of the distance from the dendrimer's center of mass. The hydrodynamic radius of 16 is calculated according to the model of a rigid sphere as: R h « 1.29 R g [28]; (c) radial distribution functions, g(r), of the water molecules (black), and number of water molecules (blue) as a function of the distance from 16's center of mass.
It is interesting to note that the high level of branching in the scaffold of 16 produces high-density in the dendrimer's interior and reduced surface group backfolding. Seen in Figure 4b, the g(r) of the surface OH-groups of the dendrimer (red) are most probably found at greater distance from the dendrimer's center than all other atoms of 16. This indicates that the surface groups well surround the dendrimer scaffold. This picture of high interior density is well consistent with the g(r) of the water molecules (Figure 4c), demonstrating that water penetration is present only close to the surface, identified by R g (no water penetration for R < 0.5 R g ). Table S1 reports other interesting structural data for 16. In particular, comparison between dendrimer's R g and solvent accessible surface area (SASA) provides an interesting cue on the level of porosity of this dendrimer. Namely, SASA is the surface in contact with the solvent. From this parameter, it is possible to obtain R SASA , which is the radius of a sphere having surface equivalent with that of the dendrimer in contact with the solution as: R SASA = sqrt(SASA/4π). R SASA is larger than R g , because the dendrimer is not a rigid sphere but a porous soft macromolecule with a non-perfect surface. Comparison between the volume of the voids in the dendrimer and the full one leads to a quantification of 16's porosity (0.78, see Table S1), which is slightly higher than that of Polyamidoamine (PAMAM) G2 dendrimer (0.75), having SASA « 3.5ˆ10 3 Å 2 and R g = 10.4 Å 3 . Compared to G2 PAMAM, the data from the MD simulation suggest that 16 is more heavy and dense in the interior even if having a larger surface. This is consistent with the fact that 16 is slightly smaller than a G2 PAMAM (slightly smaller R g ) while the MW is higher (4581 Da for 16 vs. 3256 Da for G2 PAMAM) and it has a larger number of surface groups: 20 vs. 16.

Materials and Methods
All reactions in organic medium were performed in standard oven dried glassware under an inert atmosphere of nitrogen using freshly distilled solvents stored over molecular sieves. Solvents were deoxygenated when necessary by bubbling nitrogen through the solution. All reagents were used as supplied without prior purification and obtained from Sigma-Aldrich Chemical Co. (Toronto, ON, Canada) Reactions were monitored by analytical thin-layer chromatography (TLC) using silica gel 60 F254 precoated plates (E. Merck, Darmstadt, Germany) and compounds were visualized by 254 nm light and/or by dipping into a mixture of sulfuric acid and methanol in water or into a mixture of KMnO 4 and K 2 CO 3 in water followed by gentle warming with a heat-gun. Purifications were performed by flash column chromatography using silica gel from Canadian Life Science (60 Å, 40-63 µm) (Peterborough, ON, Canada) with the indicated eluent. 1 H-NMR and 13 C-NMR spectra were recorded at 300 and/or 600 MHz and 75 and/or 150 MHz, respectively, on a Bruker spectrometer (300 MHz and 600 MHz) (Milton, ON, Canada) and Varian spectrometer (600 MHz) (Milton, ON, Canada). All NMR spectra were measured at 25˝C in indicated deuterated solvents. Proton and carbon chemical shifts (δ) are reported in ppm and coupling constants (J) are reported in Hertz (Hz). The resonance multiplicity in the 1 H-NMR spectra are described as "s" (singlet), "d" (doublet), "t" (triplet), and "m" (multiplet) and broad resonances are indicated by "broad". Residual protic solvent of CDCl 3 ( 1 H, δ 7.27 ppm; 13  2-Azidoethyl 2,3,4,6-tetra-O-propargyl-β-D-glucopyranoside (8). 2-azidoethyl 2,3,4,6-tetra-O-acetyl-β-Dglucopyranoside (300 mg, 0.72 mmol) was dissolved in dry methanol (6 mL) and NaOMe (1.1 M) was added slowly to bring pH to 9. The mixture was stirred at room temperature for 2.5 h. Acidic resin (IR-120) was added to neutralize the base. The mixture was filtered and evaporated to obtain fully deprotected glucoside derivative. Before being used in the next step, it was placed under vacuum for 1 h. It was dissolved in DMF (3 mL) and the solution was cooled to 0˝C. Sodium hydride (60% in oil, 240 mg, 6.0 mmol) was added. The mixture was stirred at 0˝C for 15 min. Propargyl bromide (80% in toluene, 1.5 mL, 10.1 mmol) was added dropwise. The mixture was stirred at 0˝C for 1 h. Saturated NH 4 Cl solution was added to quench the reaction. It was extracted with ethyl acetate. The extract was washed with brine, dried over MgSO 4  2-Azidoethyl 2,3,4,6-tetra-O-trimethylsilylpropargyl-β-D-glucopyranoside (9) A mixture of 2-azidoethyl 2,3,4,6-tetra-O-propargyl-β-D-glucopyranoside (8) (700 mg, 1.74 mmol), AgCl (300 mg, 2.1 mmol) and DBU (3.5 g, 23 mmol) in DCM (25 mL) was brought to 40˝C. Chlorotrimethyl silane (3.1 mL, 24.4 mmol) was added dropwise. It was stirred at 45˝C for 18 h. The reaction mixture was diluted with DCM/H 2 O (300 mL, 2:1 v/v) and extracted with DCM (200 mL). The extract was washed with water (50 mL) and with brine (50 mL), dried and evaporated. The crude material was passed through a column of silica gel (hexane-ethyl acetate (0-10%) as eluent) to obtain 2-azidoethyl 2,3,4,6-tetra-O-trimethylsilylpropargyl-β-D-glucopyranoside (9) as colorless oil (1.0 g, 1.45 mmol, 83%).   ) and CuI¨P(OEt) 3 (7 mg, 0.02 mmol) in dry toluene (2 mL) was held at 70˝C with stirring for 20 h. It was cooled and diluted with ethyl acetate (100 mL). The solution was washed with EDTA (2ˆ5 mL) and with a mixture of aqueous NaHCO 3 and brine (10 mL), dried and evaporated. The crude material was passed through a column of silica gel with hexane-ethyl acetate (1:1) and 0-6% MeOH in DCM as eluents yielding 10 as colorless oil (73 mg, 0.03 mmol, 85%). 1  Computational Methods. The computational approach used for this study is the same recently adopted for the simulation of a variety of dendrimers in aqueous solution [26][27][28][29][30]. The simulation work was conducted using the AMBER 12 software (version 12, University of California, San Francisco, CA, USA, 2012) [31]. The molecular model for 16 has been parameterized according to the "general AMBER force field (GAFF)" [32].
Dendrimer 16 was immersed in a simulation box filled with explicit TIP3P water molecules [33]. After preliminary energy minimization, the system has been thermalized through a short 100 ps MD simulation in NVT (constant N: number of atoms, V: volume and T: temperature) periodic boundary conditions to reach the temperature of 37˝C (310 K). Then, dendrimer 16 has been equilibrated through 200 ns of MD simulations NPT in periodic boundary condition (constant N: number of atoms, P: pressure and T: temperature during the run) at 37˝C of temperature and 1 atm of pressure. During this time, dendrimer 16 reached the equilibrium with good stability, as demonstrated by the root mean square displacement (RMSD), radius of gyration (R g ) and enthalpy (H) data reported in the SI. For the MD run, a 2 femtoseconds time step was used, the Langevin thermostat, a 8 Å cutoff, the particle mesh Ewald [34] (PME) for long-range electrostatics, and the SHAKE algorithm to treat bonds involving Hydrogen atoms [35] Enthalpy H was calculated through the MMPBSA approach [36,37] according to the same procedure followed previously [26][27][28][29][30]. Namely, H is the sum of the total gas-phase in vacuo non-bonded energy (Egas) and a solvation term (Gsol = G PB + G NP ) [38] The polar component of G PB was evaluated using the Poisson-Boltzmann [39] (PB) term while the non-polar contribution to the solvation energy was calculated as G NP = a (SASA) + b, in which a = 0.00542 kcal/Å 2 , b = 0.92 kcal/mol, and the SASA (solvent accessible surface area) was estimated with the MOLSURF program included in AMBER 12 [40]. The last 100 ns of the MD trajectories were considered as representative of the equilibrium condition for dendrimer 16 and used for the analysis of the structural parameters reported in Table S1.

Conclusions
An efficient synthesis of dense and chiral dendritic polyols using sugars as multivalent nanosynthons (glyconanosynthons) at the core and at the repeating units was established. The click reactions worked best with a catalytic amount of the copper (I) organic catalyst CuI¨P(OEt) 3 . The high yielding synthetic strategy happens to be very versatile, as it permitted access to the desired functionalized materials by many different alternatives, depending on the surface group functionality of the building blocks chosen. In addition, the alkenyl and alkynyl surface groups could be conveniently transformed into dendritic polyols using thiol-ene or thyol-yne reactions using either thioethanol or thioglycerol, respectively. Hence, the number of OH-groups that could be added can be readily doubled. Furthermore, in principle, the allyl end groups of 21 can be further transformed by an additional layer of protected sugar derivative to reach the G 3 level via a thiol-ene reaction to afford dendrimers with 160 hydroxyl groups using thioglycerol and with 320 hydroxyl groups using 2-thiolethyl β-D-glucopyranoside, as shown previously [12,13]. It is also possible to add cysteamine to the double bond of 21 to construct cationic dendrimers, which may find application as gene transfection agents. Moreover, the synthesis is able to generate dendrimers with defined stereochemistry and with more rigidity compared to the existing ones. For instance, polyol 16 had a diameter of 2 nm, as calculated using molecular dynamic simulations. The strategy described herein can thus provide dendrimers having a large number of surface functionalities at very low generation efficiently and rapidly, hence simplifying the entire process of making complex architectures. Work is also in progress with the cationic equivalents of these structures as gene transfecting agents [13] together with their potential as chiral catalysts.