Transport of Anions across the Dialytic Membrane Induced by Complexation toward Dendritic Receptors.

A novel approach to inducing anion transport over the dialytic membrane was proposed and successfully tested using the dihydrogen phosphate anion. The anion receptor based on isophthalamide was anchored on a dendritic skeleton, resulting in a macromolecular structure with a limited possibility to cross the dialytic membrane. The dendritic receptor was placed in a compartment separated from a mother anion solution by a membrane. The resulting anion complexation reduced the actual concentration of the anion and induced the anion transfer across the membrane. The anion concentration in mother solution decreased, while it was found to be increased in the compartment with the dendritic receptor. This phenomenon was observed using dendritic receptors with four and eight complexation sites. A detailed analysis of a series of dialytic experiments by 1H NMR spectroscopy enabled an assessment of the complexation behavior of both receptors and an evaluation of the dendritic effect on the anion complexation.

Introduction

Anions play an important role in nature and are responsible for diverse phenomena; however, in excessive amounts, they can represent a serious environmental[1] or medical[2] issue. Therefore, supramolecular chemists pay significant attention to the design of receptors capable of recognizing and sensing anionic or electron-rich neutral substrates.[3−6] Numerous receptors were developed, and their complexation properties were investigated focusing on the selectivity, kinetics, and thermodynamics of anion complexation. The technological and engineering aspects of the actual extraction and possible isolation of the complexed substrate are rather rare. Besides the selectivity and effectivity issues, limitations arise mainly from the difficulty of separating receptors from a given solution.[7−9] To facilitate anion isolation, receptors can be detached from the mother solution by a certain barrier. The receptor has to be able to induce an anion transport through it in the initial step. One of the most employed barriers is represented by a water–organic interface utilized in the liquid–liquid extraction method. In this experimental setup, the receptor contained in an organic phase binds an anion and induces its transfer between the solvent layers.[7] In this context, Beletskiy and Kass described an extraction of phosphate salt from water into chloroform using a tripodal pentafluorophenyl thiourea receptor forming stoichiometric complexes with H2PO4–.[10] Moyer and co-workers developed the simple alkylated di(imino)-guanidinium receptor showing exceptional selectivity for SO42–.[11] Liquid–liquid extraction has severe drawbacks, especially the necessity of recycling large quantities of organic solvents. A variant of liquid–liquid extraction is represented by transport experiments using the U-tube technique, where the anions are transferred from an aqueous-source phase across bulk organic solvent containing a convenient receptor into another aqueous receiving phase.[12,13] This diffusion-driven process is highly time-consuming and requires all components to be well balanced.

Liquid–solid extraction is another option that utilizes predominantly ion exchange to induce an anion transport into a semisolid environment as the organic receptor is grafted on a membrane[14,15] or solid support such as MOFs,[16] polymers,[17] or hydrogels.[18,19] There were also attempts to utilize a discrete insoluble anion receptor.[20,21] Generally, liquid–solid extraction is a slow, diffusion-driven process employing only a fraction of the present complexation sites.

In this study, a new approach to induce anion transport across the dialytic membrane was tested. The isophthalamide receptors were covalently attached on the periphery of a carbosilane dendrimer,[22−25] a well-defined ordered symmetrical structure with an exact number of receptors on the periphery. Inspired by Gaab et al.’s tea-bag catalysis,[26] the macromolecular species were separated from the mother solution by a membrane. This experimental setup enabled continuous refill of anions depleted by complexation and a facile isolation of the complexes formed. Advantageously, the used dendrimer can be easily recycled by nanofiltration.[27,28] This approach was employed for extraction of phosphate anions.

Results and Discussion

It was found that receptors employing hydrogen bonding toward anionic species can possess high efficiency and selectivity.[6,29] Among others, the isophthalamide group acting as the hydrogen bond donor has been found to be very useful utilizing in syn–syn conformation a mutual cooperation of two amidic hydrogen atoms for anion cooperation. The chosen model receptor M1 was synthesized from 5-(benzyloxy)isophthalic acid via in situ conversion to the corresponding acid chloride and subsequent transformation to final amide by reaction with m-methoxyaniline (Scheme ).

Scheme 1

Synthesis of Receptor M1

The structure identification of receptor M1 was based on standard 1D and 2D NMR experiments and was irrefutably confirmed by X-ray crystallography (for details, see the Supporting Information). The conformational preferences of receptor M1 do not differ significantly from other isophthalamide derivatives possessing an anion complexation cavity between two NH groups, which have been frequently discussed in the literature.[30−34] According to X-ray data, the receptor M1 holds the syn–syn conformation of NH groups in the solid state. In the liquid state, 1H NMR steady-state NOE experiments proved free rotation around all the single bonds, enabling quick interconversion between all possible conformers including the complexation-favorable syn–syn conformation.

The complexation properties of M1 toward selected anions were tested by 1H NMR titrations monitoring the chemical shift change of the NH signal induced by addition of aliquots of the anion in the form of tetrabutylammonium (TBA+) salt. In CDCl3, the model compound M1 showed superior complexation behavior with the stability constants that are out of range that can be accurately determined by NMR. To find the preferences of M1 toward selected anions, the titration experiments were repeated using DMSO-d6 as a solvent. Under these conditions, the receptor M1 binds the molecules of DMSO itself via hydrogen bonding, as proven by titration experiments with DMSO in CDCl3 (Ka = 9.3 ± 0.2 M–1). Despite the competitive environment, compound M1 exhibits satisfactory complexation behavior in DMSO with preference toward dihydrogen phosphate and carboxylates over halides. The obtained complexation constants are shown in Table . With respect to complexation affinity and selectivity, M1 behaves similarly to other receptors utilizing the isophthalamide moiety.

Table 1

Anion Association Constants of Receptor M1

	Ka [M–1]
	AcO–	BzO–	H2PO4–	Cl–	Br–
CDCl3	>104	>104	>104	>104	4700(500)
DMSO-d6	100(6)	160(30)	140(10)	33(7)	3(1)

Subsequently, the idea of receptor M1 utilization in dialytic experiments was explored. For this purpose, it was essential to anchor the receptor M1 molecule on a bulky substrate that cannot pass through the dialytic membrane. In this respect, a dendritic scaffold seems to be an ideal choice, which additionally enables multiplication of complexation sites within one molecule. The hydroxy group of M1 can be easily deprotected by hydrogenation, and the resulting 5-hydroxy-N,N′-bis(3-methoxyphenyl)isophthalamide can be subsequently reacted under Williamson’s conditions with reactive iodopropyl groups of carbosilane dendrimers (Scheme ). For this purpose, the first- and second-generation dendrimer moieties (G1PrI4 and G2PrI8)[25] were used, providing multivalent dendritic receptors Dm1 and Dm2 with four and eight complexation sites, respectively. Molecular dynamics simulations were performed to estimate the size characteristics of the receptors Dm1 and Dm2 in DMSO. The average largest dimensions of receptors Dm1 and Dm2, calculated from the last 2 ns of the 6 ns simulation, were found to be 3.0 and 3.6 nm, respectively (for details, see the Experimental Section and the Supporting Information), and corresponding estimates of radii of gyration were 1.0 and 1.3 nm, respectively (Figure S45). This information, together with the corresponding molecular masses of the receptors, 2163 Da for Dm1 and 4639 Da for Dm2, predetermined the selection of the dialytic membrane. The membrane with a molecular weight cutoff of 3.5–5 kDa seemed to be the most convenient.

Scheme 2

Synthesis of Dendritic Receptors Dm1 and Dm2

The mobility of individual parts of the dendritic molecules was investigated again by steady-state NOE NMR experiments. Generally, the tumbling of the complexation sites decreased due to bonding to the large dendritic molecule; however, the rotation of all the important parts remains free enabling isophthalamide groups at the periphery to adopt a convenient geometry for anion complexation. The complexation properties of receptors Dm1 and Dm2 were examined by NMR titrations with dihydrogen phosphate in CDCl3 and DMSO-d6.

CDCl3 represents a noncompetitive solvent for hydrogen bonding, where the complexation abilities of anion complexing groups can fully develop. According to the obtained results, the four possible complexation sites of Dm1 in CDCl3 do not mutually compete, and strong 1:4 complexes are formed. In the case of Dm2, the equilibrium is shifted toward a 1:8 receptor/anion ratio indicating participation of all complexation sites. It was not possible to determine the complexation constants due to the high stoichiometry of the complexes formed in both the aforementioned cases; however, the complexation is almost quantitative, as evidenced by binding isotherms (Figure ). In DMSO, a considerable competitive complexation of the solvent causes a decrease of receptor complexation effectivity. For complexation of dihydrogen phosphate with Dm1 in DMSO, the complexes with a prevailing stoichiometry of 1:2 were observed. For this model, an overall complexation constant can be calculated with an approximate value of 2000 M–2 (20% error). The stoichiometry of formed complexes was found to be even more complicated for Dm2, which possesses eight complexation sites. Obviously, not all binding sites are occupied even in the presence of an excess of the phosphate salt. Due to high error caused by unclear stoichiometry, the calculation of complexation constants in this case was not considered meaningful.

Figure 1

1H NMR binding isotherms for complexation of dihydrogen phosphate by Dm1 and Dm2; 5 mmol/L receptor concentration in CDCl3 (left) and 10 mmol/L in DMSO-d6 (right).

1H n class="Chemical">NMR binding isotherms for complexation of dihydrogen phosphate by Dm1 and Dm2; 5 mmol/L receptor concentration in CDCl3 (left) and 10 mmol/L in DMSO-d6 (right).

Dialytic Experiments

The ability of prepared dendritic receptors to induce the transport of anions through the dialytic membrane was tested in a series of experiments. For dialytic experiments, the dihydrogen phosphate anion was again chosen as a biologically relevant anion with a strong interaction toward dendritic receptors. A cellulose-based membrane was used for detachment of two solutions containing different concentrations of the chosen ionic pair. According to preliminary experiments, the ionic compounds dissolved in chloroform interact strongly with the pores of polar membranes during the transport from one environment into another. Their transport across the membrane in chloroform is therefore much slower and difficult to quantify. The dialytic experiments were thus performed in DMSO, where the movement of ionic compounds across the membrane is not restricted. All the experiments were carried out in 5 mL vials containing the starting solution of a given concentration of phosphate salt (compartment I). Another solution was injected into dialytic tubing (compartment II) and inserted into the vial. In all cases, compartment II, containing 0.5 mL of solution, was surrounded by 3 mL of the stock solution at the beginning of the experiment. Preliminary experiments are summarized in Figure 4 showing that phosphate and its counter ion can pass through the membrane and equilibrate concentrations in both compartments (Figure a). Preliminary experiments with pure DMSO also confirmed that receptor leakage through the chosen membrane is negligible even for the smaller multivalent receptor Dm1 (Figure b). An interesting phenomenon was observed when both previous experiments were combined and compartment II was filled with the solution of the dendritic receptor. The concentration of the ionic pair was found to be significantly higher in compartment II, containing the dendritic receptor. Moreover, the volume in compartment II increased significantly (Figure c). Based on these experiments, one can assume that the dendritic receptor binds the anion, and the concentration of free anions in compartment II lowers and induces the transport of the ionic pair through the membrane. The increasing concentration of ions in compartment II consequently induces the transport of solvent molecules through the membrane to equilibrate the ionic strength on both sides. Obviously, the contribution of osmosis with respect to different concentrations of the dendritic receptor in both compartments is negligible as no volume changes were observed during the experiment conducted with the receptor and pure DMSO (Figure b). The experimental data of the preliminary experiments can be found in the Supporting Information.

Figure 2

Experimental settings of preliminary experiments; membrane permeability for the ion pair (a), membrane permeability for the dendritic receptor (b), and ion transport enhanced by the dendritic receptor (c).

Additional experiments were performed to explore the behavior of the system in detail, focusing mainly on the kinetics of the process, mass transport through the membrane, and induced volume changes. In order to suppress the influence of the initial concentration gradient of the dialytic experiment, further experiments were performed using the same concentration of the ionic pair in both compartments. Using this experimental setup, the whole observed mass transport could be attributed to the presence of a receptor.

The kinetics of the dialytic experiment for both dendritic receptors was investigated using the receptor dissolved in 0.5 mL of dihydrogen phosphate stock solution. The dendrimer concentration chosen was about one receptor molecule per 6 mol of the salt. This solution was loaded into compartment II, while the respective stock solutions were added into the corresponding compartment I of the experimental vial. The concentration of the ionic pair in compartment I was tested regularly by 1H NMR spectroscopy. The obtained curves reveal the time dependency of anion transport across the membrane (Figure ). The initial steep changes of concentration reflect the driving force of the process. After 2 h, the driving force decreased significantly, while after 6 h, the experiment was close to equilibrium and further concentration changes were found to be negligible. The time required for the equilibration of the dialytic experiment was estimated at 24 h. The signals of both dendrimers were detected via 1H NMR spectroscopy in compartment I, indicating a certain amount of dendrimer leakage. This leakage was subsequently quantified to be around 10% of the initial dendrimer mass in compartment II. The experiments also suggested that the receptor with more complexation sites can induce a larger decrease of the salt concentration in compartment I.

Figure 3

Kinetics of TBA+ H2PO4– transport through the membrane in DMSO monitored by the concentration decrease in compartment I.

Kinetics of TBA+ n class="Chemical">H2PO4– transport through the membrane in DMSO monitored by the concentration decrease in compartment I.

Further experiments were focused on determination of the amount of dihydrogen phosphate salt passing through the membrane and its dependence on dendrimer and dihydrogen phosphate concentrations. For this purpose, a correct quantification of volume changes accompanying the course of the experiment is necessary. Unfortunately, the only information that can be obtained precisely is the final volume in compartment I, which can be determined, e.g., by weighing. The quantification of the volume changes in compartment II by syringe was significantly affected by an error induced by membrane swelling. Obviously, an unknown amount of both the dendrimer and dihydrogen phosphate salt remained absorbed in the membrane. Due to the dendrimer leakage mentioned above, one can assume that even in the membrane the mutual ratio of DMSO, dendrimer, and dihydrogen phosphate salt is similar to the ratio in compartment II. Therefore, the missing volume in compartment I was attributed as a whole to the final volume of compartment II. In the following, this volume will be labeled as the volume transferred. Using this approximation, an indicative mass balance corresponding well with the starting conditions can be achieved. The concentrations of dendrimer and dihydrogen phosphate salt (namely, the TBA+ signal) in respective compartments were determined by integration of corresponding signals in 1H NMR spectra. Additionally, a calibration-based comparison of both signals to a residual solvent signal affords an indicative estimation of the volume of a given compartment. The mass balance calculated from this data was fully in agreement with the balance obtained previously just by weighing compartment I. The calculation based on NMR data was subsequently employed in all dialytic experiments.

The experiments using both dendritic receptors were performed, varying the concentration of dihydrogen phosphate salt while keeping the receptor concentration the same. The concentration of H2PO4– in the experiments with receptor Dm1 covered molar ratios from 1:1 to 1:12. A summary of the experimental conditions and obtained results is found in Table (for details, see the Supporting Information). Each experiment began by an immersion of compartment II, containing the dendritic receptor, into compartment I. After 24 h, the experiment was terminated by removal of compartment II from the experimental vial as further volume changes were imperceptible. In all cases, significant volume changes were observed, and compartment II almost doubled its volume. The concentrations in both compartments also changed. The H2PO4– concentration slightly decreased in compartment I, while it increased in compartment II. These concentration changes undoubtedly exceeded the error of the integration of the 1H NMR spectrum, which oscillates around 2%. Almost half of the concentration of the dendrimer receptor was found in compartment II at the end of the experiment, due to dilution in double solvent volume.

Table 2

Brief Summary of the Experimental Conditions of Dialytic Experimentsa

entry	c(Dm1)0h [mol/L]	Dm1:H2PO4–	ΔV24h [mL]	Δc(H2PO4–)24h [mol/L]	n(H2PO4–)trans [mol]
1	0.0084	1:1	0.35	0.0011	0.0041
2	0.0091	1:3	0.55	0.0081	0.0169
3	0.0091	1:5.5	0.42	0.0073	0.0197
4	0.0071	1:11.5	0.49	0.0037	0.0344
5	0.0049	1:10	0.37	0.0026	0.0143

c(Dm)0h - starting concentration; Dm:H2PO4– - molar ratio; ΔV24h - a volume change in comp. II after 24 h; Δc(H2PO4–)24h - a concentration difference between comp. I and II after 24 h; n(H2PO4–)trans - transferred molar amount.

c(Dm)n class="Chemical">0h - starting concentration; Dm:H2PO4– - molar ratio; ΔV24h - a volume change in comp. II after 24 h; Δc(H2PO4–)24h - a concentration difference between comp. I and II after 24 h; n(H2PO4–)trans - transferred molar amount.

Taking into account all possible experimental errors, the induced volume changes were approximately 0.4 mL in all cases. This change should correspond to the molar mass of the dendritic receptor or specifically to the total number of functional groups on the dendrimer. Keeping the dendrimer concentration at the same level, the amount of complexed anion depends solely on its initial concentration. Low H2PO4– concentration leads to an occupation of less than one functional group of the dendritic molecule on average, while in higher concentrations, H2PO4– can occupy up to two or four functional groups per Dm1 or Dm2 molecule, respectively. The amount of complexed H2PO4– was calculated from the concentration difference. According to the degree of complexation, the seemingly decreased anion concentration in compartment II induces a flux from compartment I to equilibrate concentration in both compartments. Close to equilibrium, the actual anion concentration in compartment II is always higher than in compartment I due to the complexed amount of H2PO4– on the dendritic receptor. A higher concentration requires an extra solvent molecule to balance the ionic strength on both sides, raising the volume in compartment II significantly. On the other hand, the raised volume in compartment II dilutes the receptor, which slightly decreases its complexation ability. The mass of H2PO4– transferred through the membrane not only corresponds to the complexed matter but is also proportional to the raised volume. This amount of H2PO4– steadily rises with the increasing anion concentration (entries 1–4). Increasing the starting H2PO4– concentration leads to a larger amount of H2PO4– transferred through the membrane; however, the final concentration difference is not that pronounced for obvious reasons. Keeping the mutual dendrimer/H2PO4– ratio constant while diluting the overall concentration results in a decreased level of complexation (Table , entries 4 and 5). This observation is fully in agreement with common knowledge and also with the former screening of the complexation behavior of M1.

Additional experiments with n class="Chemical">Dm2 covered the ratios of dendrimer:H2PO4– from 1:3 to 1:30 and were mainly focused on the evaluation of the dendritic effect. Experiments performed in similar ratios and concentration ranges indicated that in diluted solutions a higher number of complexation sites on the dendrimer do not mean a higher amount of complexed H2PO4– or its transferred mass (entries 2 and 6 for the 1:3 ratio). This feature starts to prevail in higher concentrations of H2PO4– (entries 5 and 8 for the 1:10 ratio).

A similar effect was observed in the second part of the experiments, which were performed with half the concentration of the Dm2 receptor, keeping the number of complexation sites at the same level as in experiments with Dm1. For lower concentrations, Dm1 was found to be more efficient than Dm2 (Table , entries 2 and 7 for the 4:3 ratio of complexation sites/anion). At 2:3 ratios, the complexation efficiency seems to be similar for both dendritic receptors (Table , entries 3 and 8), while at 1:2.5 ratios Dm2 induces greater complexation and transfer of H2PO4– (Table , entries 5 and 9). Both series of experiments indicate a negative dendritic effect at low H2PO4– concentrations and a positive one at higher concentrations. At low anion concentrations, a homogeneous distribution of complexation sites is advantageous, and therefore, the complexation ability should decrease in the row of free isophthalamide receptor (M1) and dendrimer with four (Dm1) and eight (Dm2) complexation sites. At high anion concentrations, a migration of the anion from one complexation site to another in its proximity can be expected and results in a positive dendritic effect.

After the dialytic experiments, the complexes formed were decomposed by appropriate solvents, and free dendritic receptors were recycled in a nanofiltration cell (for details, see the Experimental Section). Using this approach, up to 85% of the initial amount of each receptor was recovered.

Conclusions

A novel approach to induce anion transport across the dialytic membrane was proposed and successfully tested. For this purpose, the isophthalamide groups were anchored on dendritic structures providing macromolecules with complexation abilities toward anions and with a limited possibility to cross the dialytic membrane. Placing such a receptor in a compartment, which is separated from a mother anion solution by a dialytic membrane, reduces the actual anion concentration and induces anion flux across the membrane. When the equilibrium is reached, the anion concentration in the mother solution is decreased, while that in the compartment with the dendritic receptor is increased. The excess of the anion can be attributed to the complexation with the receptor. The increasing anion concentration is accompanied by osmosis; solvent transfers through the membrane to equalize the actual anion concentrations on both sides of the membrane. The complexation efficiency depends on concentrations of both the anion and receptor. For diluted anion solutions, the use of the dendrimer with fewer complexation sites is advantageous, while the dendrimer with more complexation sites is more efficient in higher anion concentrations. The dendritic receptor can be subsequently regenerated by nanofiltration without a significant loss. The particular isophthalamide group used in this work was of relatively low complexation ability. However, the results presented can serve as a base for studies utilizing receptors with higher anion affinity that should induce higher anion transfer through the membrane.

Experimental Section

General Procedures

The reagents were purchased from commercial sources and used without further purification. The starting carbosilane dendrimers were prepared by the standard iterative procedure.[35] The preparation of the starting dendritic scaffold was accomplished by the previously reported procedure.[25] The solvents used for synthesis and chromatography were purchased from commercial sources and distilled before use. Anhydrous solvents were dried by standard procedures. DCM was dried over calcium hydride, triethyl amine was stored above NaOH(s), and DMF was stored above molecular sieves.

The 1H (400.1 MHz), 13C (100.6 MHz), and 29Si (79.5 MHz) NMR spectra were recorded using a Bruker Avance 400 spectrometer at 25 °C. Used solvents (DMSO-d6, CDCl3) were stored above molecular sieves. The 1H and 13C NMR spectra were referenced to the line of the solvent (δ/ppm; δH/δC: DMSO-d6, 2.50/39.52, CDCl3, 7.26/77.16). The 29Si spectra were referenced to the line of external standard hexamethyldisilane (δ/ppm; −19.79). The HRMS spectra were measured on a MicroTOF III spectrometer (Bruker) with an ESI ionization source in positive mode. For calibration of accurate masses, an ESI-APCI Low Concentration Tuning Mix (Agilent) was used. The samples were delivered into the ion source in methanol solution; the ionization in Dm2 had to be enhanced by addition of ammonium formate buffer. FTIR spectra of the samples were obtained using an FTIR spectrometer (Nicolet Avatar 360, Thermo Nicolet). The samples were dissolved in acetone, dropped on a germanium window (thickness 1 mm), and measured after evaporation of the solvent. The transmission spectra with a resolution of 4 cm–1 and 200 scans per spectrum were collected at ambient temperature in the range of 4000 to 400 cm–1. Diffraction data were collected on a Bruker D8 VENTURE Kappa Duo PHOTON 100 CMOS with monochromatic Cu Kα radiation. The structure was solved by the direct methods (SHELXT)[36] and refined by full-matrix least-squares on F2 values (CRYSTALS).[37] The water molecule trapped in the structure was found with 0.11 occupancy. All heavy atoms were refined anisotropically. Hydrogen atoms were localized from the expected geometry. NH and water molecule hydrogen atoms were located from difference electron density maps. Only hydrogen atoms with full occupancy were refined isotropically. ORTEP-3 was used for structure presentation.[38] The crystallographic data have been deposited in the Cambridge Crystallographic Data Centre as a supplementary publication. These data are provided free of charge by the joint Cambridge Crystallographic Data Centre and Fachinformationszentrum Karlsruhe Access Structures service www.ccdc.cam.ac.uk/structures.

Computer Modeling – Methods

3D computer models of dendrimers Dm1 and Dm2 were created using a dendrimer builder, as implemented in the Materials Studio software package (MS) from BIOVIA (formerly Accelrys). Similarly, the DMSO molecule was created using the sketch tool, which is available in MS for the purpose of creating molecules from scratch. For the parameterization of the molecular structures, the COMPASS II force field[39] was used (bond, nonbond terms, dendrimer partial charges). DMSO partial charges were adopted from the JM4 model[40] to ensure good agreement of simulated and experimental DMSO density. Dendrimers were solvated in DMSO (3000 DMSO molecules were used for each dendrimer structure) using the Amorphous Cell MS module (see Figure S44). Lengths of the cubic simulation box oscillated around 73 Å. The used simulation method was Molecular Dynamics, NPT ensemble (T = 298 K, P = 105 Pa). These simulations were performed using the MS simulation module Forcite. Temperature was regulated using a Nosé–Hoover thermostat (Q ratio = 0.01), and pressure was controlled by a Berendsen barostat (decay constant = 0.1 ps). Simulations were carried out using 3D periodic boundary conditions. The electrostatic and van der Waals interactions were calculated using group-based and the atom-based summation approaches, respectively, with a cutoff distance of 18.5 Å in both cases. The total length of simulations was 6 ns with a time step of 1 fs (i.e., 2 × 106 integration steps for each simulation). The first 4 ns was used for equilibration of the simulated systems; the last 2 ns was used for analysis. The time evolution of equilibration of dendrimer structures was monitored using the time evolution of their radius of gyration.

Determination of Complexation Constants

Complexation constants were measured in CDCl3 and DMSO-d6 by the standard 1H NMR titration methodology.[41−43] A solution of tetrabutylammonium salt of a selected anion (purchased from commercial sources, stored in a dry box) was gradually added in aliquots into a solution of a given receptor to reach at least a 1:5 ratio of binding group to anion. Concentrations of respective receptors were about 10–2 mol/L for DMSO-d6 titrations and 5 mmol/L for CDCl3 titrations and were kept constant during the titrations to avoid the effects of dilution. The corresponding complexation constants were calculated based on the analysis of binding isotherms obtained from the complexation induced shift (CIS) of NH protons. For nonlinear curve fitting of experimental data, the freely available software Bindfit was utilized.[44]

Phosphate Extraction by Dialysis

For dialytic experiments, the cellulose-based n class="Disease">dialytic tubing Spectra/Por with MWCO 3.5–5 kDa, 16 mm flat width, dry, treated with glycerin was used (Spectrum Laboratories).

A specified amount of TBA+ salt of a target anion was dissolved in 5 mL of DMSO-d6 to form a stock solution. The dendritic receptor was loaded in a 1 mL vial, 0.5 mL of the stock solution was added, and the vial was sonicated until the dendrimer was completely dissolved. A 5 cm piece of the tubing was closed at one end using a coated wire, and the whole content of the vial was transferred into the tubing. The filled tubing was placed into a 5 mL vial and surrounded by 3 mL of the stock solution. The vial was closed tight, with the upper end of the dialysis tubing fixed between the vial rim and screw cap, and left at room temperature for a given time period. Then the dialysis tubing was removed from the vial, the outer volume was determined by weighing, and both solutions and the stock solution were analyzed by 1H NMR spectroscopy.

Receptor Recycling and Nanofiltration

For recycling of dendritic receptors, a solvent resistant stirred cell (Merck Millipore) equipped with 1 kDa MWCO regenerated cellulose ultrafiltration membrane disc (Merck Millipore) and sealed with FEP-coated O-rings (Eriks) was used. The filtration was driven by nitrogen under a pressure of 4.5 bar. Solutions of complexes were diluted by a 2:1 mixture of dichloromethane and methanol to 20 mL total volume and filtered to a target volume of 1 mL of retentate; this run was repeated five times. Evaporation of the retentate gave the dendritic receptor, while salts accumulated in the permeate.

Synthesis of Receptors

5-Benzyloxy-N,N′-bis(3-methoxyphenyl)isophthalamide (M1)

2.72 g (10 mmol) of 5-(benzyloxy)isophthalic acid, 17 mL (0.2 mol) of oxalyl chloride, and 40 mL of anhydrous DCM were placed into a round-bottom flask, and the mixture was heated for 4 h to 80 °C. Then, the liquid part was distilled, and 5 mL of DCM was added to the crude product and distilled again. This procedure was repeated three times, and the crude product was dried under vacuum. The residue was dissolved in 40 mL of THF, and 10.6 mL (94 mmol) of m-anisidine and 5.5 mL (40 mmol) of anhydrous triethyl amine were added. The mixture was heated to 50 °C overnight, then evaporated to dryness, dissolved in 50 mL of ethyl acetate, and washed with 3 × 20 mL of 1 M HCl and once with 20 mL of water. The organic phase was dried over MgSO4. The crude mixture was evaporated to dryness, and the product was isolated by silica column chromatography, using a THF–DCM 1:1 (v/v) eluent. The resulting diamide M1 was obtained as a pale brown solid in 44% yield (2.1 g).

Data for M1: 1H NMR (DMSO-d6, 400 MHz, H-H COSY) δ (ppm): 10.36 (s, 2H), 8.15 (t, J = 1.5 Hz, 1H), 7.78 (d, J = 1.5 Hz, 2H), 7.53–7.51 (m, 2H), 7.48 (dd, J = 2.0, 2.6 Hz, 2H), 7.43 (tt, J = 6.5, 1.0 Hz, 2H), 7.40 (ddd, J = 8.0, 2.0, 0.9 Hz, 2H), 7.38–7.34 (m, 1H), 7.27 (dd, J = 8.2, 8.0 Hz, 2H), 6.71 (ddd, J = 8.2, 2.6, 0.9 Hz, 2H), 5.29 (s, 2H), 3.76 (s, 6H). 13C {1H} NMR (DMSO-d6, 100 MHz, HSQC, HMBC) δ (ppm): 164.7 (2C), 159.4 (2C), 158.3, 140.2 (2C), 136.6, 136.5 (2C), 129.5 (2C), 128.5 (2C), 128.1, 127.8 (2C), 119.5, 116.9 (2C), 112.6 (2C), 109.4 (2C), 106.1 (2C), 69.9, 55.0 (2C). HRMS (ESI+) [C29H27N2O5]+ calc. 483.1914 [M + H]+, found 483.1919. IR: 1650, 1602, 1548, 1460, 1426, 1338, 1263, 1158, 1048, 853, 685. X-ray data: triclinic system, space group P-1, a = 8.4290(4), b = 11.2479(6), c = 13.2101(6) Å, α = 81.940(2), β = 77.217(2), γ = 78.149(2)°, Z = 2, V = 1189.45(10) Å3, Dc = 1.36 g/cm3, μ(Cu Kα) = 0.761 mm–1, T = 150 K, crystal dimensions of 0.04 × 0.11 × 0.24 mm, colorless prism. The structure converged to the final R = 0.0390 and Rw = 0.0981 using 4235 independent reflections for 438 refined parameters (θmax = 72.31°). CCDC registration number 2054265.

5-Hydroxy-N,N′-bis(3-methoxyphenyl)isophthalamide

2 g of diamide 1 (4.14 mmol) was n class="Disease">dissolved in a mixture of solvents THF–EtOH 1/1 (50 mL), and the Pd/C catalyst (0.2 g) was added. The mixture was stirred under a hydrogen atmosphere at room temperature for 6 h. Then the mixture was filtered through Celite, and the filtrate was concentrated to obtain a white solid in 94% yield (1.5 g, 3.9 mmol).

Data for 5-hydroxy-N,n class="Chemical">N′-bis(3-methoxyphenyl)isophthalamide: 1H NMR (DMSO-d6, 400 MHz, H-H COSY) δ (ppm): 10.39 (s, 2H), 10.16 (s, 1H), 8.05 (t, J = 1.1 Hz, 1H), 7.52 (dd, J = 2.0, 1.0 Hz, 2H), 7.50 (d, J = 1.1 Hz, 2H), 7.43 (dd, J = 8.1, 1.0 Hz, 2H), 7.25 (t, J = 8.1 Hz, 2H), 6.69 (dd, J = 8.1, 2.0 Hz, 2H), 3.76 (s, 6H). 13C {1H} NMR (DMSO-d6, 100 MHz, HSQC, HMBC) δ (ppm): 165.1 (2C), 159.4 (2C), 157.5, 140.3 (2C), 136.5 (2C), 129.4 (2C), 117.6 (2C), 117.5, 112.5 (2C), 109.2 (2C), 106.0 (2C), 55.0 (2C). HRMS (ESI+) [C22H20N2O5Na]+ calc. 415.1264 [M + H]+, found 415.1267. IR: 1657, 1604, 1540, 1489, 1426, 1334, 1263, 1153, 1041, 774, 685.

Dendrimer Dm1

A mixture of 5-hydroxy-N,n class="Chemical">N′-bis(3-methoxyphenyl)isophthalamide (300 mg, 0.77 mmol) and K2CO3 (106 mg, 0.77 mmol) in 20 mL of DMF was stirred for 30 min at ambient temperature. The polyiodide dendrimer G1PrI4 (188 mg, 0.17 mmol) was added, and the reaction mixture was stirred at 80 °C for 48 h. The solvent was removed under vacuum, the residue was extracted with Et2O and filtered through Celite, and the filtrate was concentrated to obtain Dm1 as a yellow waxy solid in 91% yield (335 mg, 0.155 mmol).

Data for Dm1: n class="Chemical">1H NMR (DMSO-d6, 400 MHz, H-H COSY), δ (ppm): 10.30 (s, 8H), 8.11 (t, J = 1.5 Hz, 4H), 7.62 (d, J = 1.5 Hz, 8H), 7.47 (dd, J = 2.5, 1.4 Hz, 8H), 7.38 (ddd, J = 8.0, 1.4, 0.9 Hz, 8H), 7.24 (dd, J = 8.2, 8.0 Hz, 8H), 6.68 (ddd, J = 8.2, 2.5, 0.9 Hz, 8H), 4.02 (t, J = 6.8 Hz, 8H), 3.73 (s, 24H), 1.77–1.69 (m, 8H), 1.36–1.29 (m, 8H), 0.61–0.54 (m, 24H), −0.05 (s, 24H). 13C {1H} NMR (DMSO-d6, 100 MHz, HSQC, HMBC) δ (ppm): 164.7 (8C), 159.4 (8C), 158.6 (4C), 140.2 (8C), 136.5 (8C), 129.4 (8C), 119.1 (4C), 116.5 (8C), 112.6 (8C), 109.3 (8C), 106.1 (8C), 70.6 (4C), 55.0 (8C), 23.3 (4C), 19.5 (4C), 18.2 (4C), 16.9 (4C), 10.9 (4C), −3.4 (8C). 29Si {1H} NMR (DMSO-d6, 80 MHz) δ (ppm): 2.16 (4Si), 0.82. HRMS [C120H148N8O20Si5Na]+ calc. 2184.9575 [M + Na]+ 100%, found 2184.9570. IR: 1650, 1605, 1544, 1496, 1338, 1247, 1158, 1044, 841, 684.

Dendrimer Dm2

The compound was synthesized according to the procedure for n class="Chemical">dendrimer Dm1 from 5-hydroxy-N,N′-bis(3-methoxyphenyl)isophthalamide (140 mg, 0.36 mmol) and polyiodide dendrimer G2PrI8 (100 mg, 0.04 mmol) to obtain Dm2 as a yellow waxy solid in 84% yield (140 mg, 0.033 mmol).

Data for Dm2: n class="Chemical">1H NMR (DMSO-d6, 400 MHz, H-H COSY) δ (ppm): 10.29 (s, 16H), 8.12 (bs, 8H), 7.62 (d, J = 2.2 Hz, 16H), 7.47 (bs, 16H), 7.38 (bd, J = 8.2 Hz, 16H), 7.22 (dd, J = 8.1, 8.0 Hz, 16H), 6.66 (dd, J = 8.1, 2.5 Hz, 16H), 4.00 (bs, 16H), 3.75–3.71 (m, 48H), 1.69 (m, 16H), 1.27 (m, 24H), 0.58–0.49 (m, 64H), −0.07 (m, 48H), −0.15 (m, 12H). 13C {1H} NMR (DMSO-d6, 100 MHz, HSQC, HMBC) δ (ppm): 164.7 (16C), 159.4 (16C), 158.6 (8C), 140.2 (16C),136.4 (16C), 129.4 (16C), 119.1 (8C), 116.5 (16C), 112.6 (16C), 109.3 (16C), 106.1 (16C), 70.6 (8C), 55.0 (16C), 23.3 (8C), 19.3 (8C), 18.4 (4C), 18.3 (4C), 18.2 (8C), 18.1 (8C), 17.1 (4C), 10.9 (8C), −3.4 (16C), −4.9 (4C). 29Si {1H} NMR (DMSO-d6, 80 MHz) δ (ppm): 2.10 (8Si), 1.02 (4Si), 0.75. HRMS [C256H336N17O40Si13]+ calc. 4656.1854 [M + NH4]+ 100%, found 4656.1872. IR: 1650, 1602, 1546, 1491, 1247, 1158, 1047, 839, 689.