Partitioning of Poly(amidoamine)Dendrimers between n-Octanol andWaterJ Y O T S N E N D U G I R I , †

M A M A D O U S . D I A L L O , * , †

W I L L I A M A . G O D D A R D I I I , †

N A T H A N F . D A L L E S K A , ‡

X I A N G D O N G F A N G , § A N DY O N G C H U N T A N G §

Materials and Process Simulation Center, Division ofChemistry and Chemical Engineering, California Institute ofTechnology, Pasadena, CA, Environmental Science andEngineering, Environmental Analysis Center, Department ofCalifornia Institute of Technology, Pasadena, CA, and Power,Environmental, and Energy Research Center, Division ofChemistry and Chemical Engineering California Institute ofTechnology, Pasadena, CA

Received February 5, 2009. Revised manuscript receivedApril 27, 2009. Accepted April 29, 2009.

Dendritic nanomaterials are emerging as key building blocksfor a variety of nanoscale materials and technologies.Poly(amidoamine) (PAMAM) dendrimers were the first classofdendriticnanomaterialstobecommercialized.Despitenumerousinvestigations, the environmental fate, transport, and toxicityof PAMAM dendrimers is still not well understood. As a first steptoward the characterization of the environmental behavior ofdendrimers in aquatic systems, we measured the octanol-waterpartition coefficients (logKow) of a homologous series ofPAMAM dendrimers as a function of dendrimer generation(size), terminal group and core chemistry. We find that the logKow

of PAMAM dendrimers depend primarily on their size andterminal group chemistry. For G1-G5 PAMAM dendrimers withterminal NH2 groups, the negative values of their logKow

indicate that they prefer to remain in the water phase. Conversely,the formation of stable emulsions at the octanol-water (O/W) interface in the presence of G6-NH2 and G8-NH2 PAMAMdendrimers suggest they prefer to partition at the O/Winterface. In all cases, published studies of the cytotoxicity of Gx-NH2 PAMAM dendrimers show they strongly interact withthe lipid bilayers of cells. These results suggest that the logKow

of a PAMAM dendrimer may not be a good predictor of itsaffinity with natural organic media such as the lipid bilayers ofcell membranes.

IntroductionNanotechnology has great potential to provide new and moreeffective functional materials to a broad range of industriesincluding electronics, chemical, energy, water, food, andbiomedical (1). Dendritic nanomaterials (NM), which includehyperbranched polymers, dendigrafts, dendronized poly-

mers, dendrons and dendrimers, are emerging as key buildingblocks for a variety of nanoscale materials and technologies(2-4). Dendrimers are highly branched 3D globular nano-structures with controlled composition and architectureconsisting of three components: a core, interior branch cells,and terminal branch cells (3, 4). These are among the mostversatile classes of NM available to date (3, 4). These softnanostructures, with sizes in the range of 2-20 nm, can beused as hosts for cations, anions and organic/inorganicsolutes. Dendrimers can also be used as scaffolds andtemplates for the preparation of metal-encapsulated nano-particles with tunable electronic, optical, and catalyticproperties. They have also been successfully used as deliveryvehicles or scaffolds for bioactive compounds. These uniqueproperties of dendrimers are providing new opportunitiesfor developing more effective functional materials forchemical separations and catalysis, medical imaging, drugdelivery and water purification (2-5). As the United StatesEnvironmental Protection Agency (EPA) begins its assessmentof the impact of nanotechnology on human health and theenvironment, there is a critical need for data and quantitativetools for assessing the environmental fate, transport andtoxicity of NM such as dendrimers (6).

Poly(amidoamine) (PAMAM) dendrimers (Figure 1) werethe first class of dendritic NM to be commercialized. Thepotential use of PAMAM dendrimers in biomedical applica-tions such as gene therapy, drug delivery and magneticresonance imaging has led many investigators to probe theirbiodistribution and toxicity. In vitro and in vivo (mice) toxicitystudies show that dendrimer cytoxicity depends on con-centration, generation and terminal group chemistry (7-14).Fuchs et al. (9) measured the in vitro toxicity of amine-baseddendrimers (1.0-20.0 µM) using the human breast cancerMCF-7 cell line. The dendrimers evaluated consisted of G1and G2 polyamidoamine dendrimers with various triaminecores and terminal groups including (1) quaternized amines,(2) benzyloxycarbonyl (boc) protecting groups, (4) bocprotected and unprotected L-amino acids (phenylalanine,methionine, and aspartic acid), (5) diaminopropionic acidand (6) 5-dimethylamino-naphthlene-1-sulfonyl chloride(dansyl). They found that the dendrimers with neutral surfacegroups (i.e., dansyl groups) were nontoxic. Surprisingly, thedendrimers with positively charged diaminopropionic acidgroups were also nontoxic in sharp contrast to a commonlyaccepted view that electrostatic interactions between pro-tonated terminal groups of Gx-NH2 PAMAM dendrimers andnegatively charged head groups of lipid bilayers cause celldeath by disrupting the integrity of cell membranes (10-12).More recently, Tisha et al. (14) have assessed the develop-mental toxicity of dendrimers using the zebrafish embryo.They found that a G4 PAMAM dendrimer with primary amineterminal groups attenuate the growth and development ofzebrafish embryos at sublethal concentrations. In contrast,a G3.5 PAMAM dendrimer with carboxylic acid terminalgroups, had no effect on the growth and development of thezebrafish embryos.

Despite these advances, the environmental fate andtoxicity of dendritic NM such as PAMAM dendrimers is stillnot well understood. The partition coefficient (logKow) of acompound between the isotropic solvent n-octanol and wateris considered a measure of membrane affinity or lipophilicity.In environmental assessment, logKow is widely employed topredict the tendency of organic molecules to (i) sorb ontosoil organic matter and (ii) bioaccumulate in the tissues of“living” media (e.g., zooplankton, benthic organisms, andfish) (15, 16). Note that this classical approach may be not

* Corresponding author phone: 626 395 2730; fax: 626 585 0918.† Materials and Process Simulation Center Division of Chemistry

and Chemical Engineering.‡ Environmental Science and Engineering Division of Engineering

and Applied Science.§ Power, Environmental, and Energy Research Center.

Environ. Sci. Technol. 2009, 43, 5123–5129

10.1021/es9003747 CCC: $40.75 2009 American Chemical Society VOL. 43, NO. 13, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 5123

Published on Web 05/19/2009

applicable to the assessment of the environmental fate andtoxicity of NM such as carbon nanotubes, fullerenes, andmetal oxide nanoparticles due to their tendency to aggregatein aqueous solutions; thus resulting in unpredictable bio-availability (17). Because dendrimer synthesis provides avariety of means for controlling molecular composition, size,shape, polydispersity, and solubility in aqueous solutions,dendrimers are ideal model systems for carrying out fun-damental investigations of the environmental fate and toxicityof monodisperse and nonaggregated NM. As a first steptoward the characterization of the environmental fate andbehavior of dendrimers in aquatic systems, we initially focuson quantifying their partitioning between water and naturalorganic media using n-octanol as model system. We mea-sured the logKow of a homologous series of 18 PAMAMdendrimers as a function of dendrimer generation (size),terminal group, and core chemistry. We find that the logKow

of PAMAM dendrimers depend primarily on their size andterminal group chemistry. For G1-G5 PAMAM dendrimerswith terminal NH2 groups, the negative values of their logKow

indicate that they prefer to remain in the water phase.Conversely, the formation of stable emulsions at theoctanol-water (O/W) interface in the presence of G6-NH2

and G8-NH2 PAMAM dendrimers suggest they prefer topartition at the O/W interface. In all cases, published studiesof the cytotoxicity of Gx-NH2 PAMAM dendrimers show theystrongly interact with the lipid bilayers of cells (10-12). Theseresults suggest that the logKow of a PAMAM dendrimer maynot be a good predictor of its affinity with natural organicmedia such as the lipid bilayers of cell membranes.

Materials and MethodsPAMAM dendrimers of different generations, core andterminal groups (Figure 1) were purchased (as methanol

solutions or solids) from Dendritech, Sigma-Aldrich andDendritic Nanotechnologies. Spectroscopic grade n-octanoland reagent grade trifluoroacetic acid were purchased fromSigma-Aldrich. OmniSolve acetonitrile was purchased fromEMD.

Measurements of Octanol-Water Partition Coefficient(logKow). The octanol-water coefficients (logKow) of thePAMAM dendrimers were measured at room temperatureusing the standard “slow-stirring” method (16). Stocksolutions of water-saturated octanol (WSO) and octanol-saturated water (OSW) were prepared by adding 450 mLof Milli-Q deionized water + 450 mL of octanol into a 1000mL glass bottle with a tap at the bottom. The content ofthe bottle was mixed for 7 days by magnetic stirring usinga Teflon-coated bar. The stirring speed was adjusted tocreate a vortex of ∼2 cm at the octanol-water interface.Following equilibration and phase separation, the OSWwas collected by decantation; whereas the WSO wascollected through the bottle tap. The logKow measurementswere carried out in 150 mL volumetric flasks. Theconcentration of each dendrimer was kept constant at 2.24mM of equivalent terminal groups in all experiments unlessotherwise specified. An aliquot of 25 mL of dendrimer +OSW was added to each flask. Following Diallo et al. (18),the pH of each dendrimer + OSW was adjusted to thedesired value by addition of drops of concentrated HCl orNaOH. After pH adjustment, an aliquot of 90 mL of WSOwas poured slowly against the upper wall of the flasks,which were then sealed with a threaded cap to minimizeevaporation. The sample solutions and reference solutions(OSW + dendrimers) were placed and mixed at 160-180rpm on a Corning magnetic stirrer. A picture of theexperimental setup is shown in Supporting Information(SI) Figure S1.

FIGURE 1. Core and terminal group chemistry of the PAMAM dendrimers evaluated in this study.

5124 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 13, 2009

Following equilibration, the concentrations of dendrimersin the OSW phase of each sample solution (C d

w) and referencesolution C d

total were measured by high performance liquidchromatography (HPLC) (19). The HPLC system consistedof an Agilent 1100 vacuum degasser, a binary pump, an autosampler and a column compartment with a UV diode arraydetector (G1315A). A Jupiter C5 silica-based reverse phaseHPLC column (250 × 4.6 mm, 300 Å from Phenomenex,Torrance, CA) was used to separate the dendrimers. APhenomenex Widepore C5 guard column (4 × 3 mm) wasattached ahead of the analytical column. Milli-Q deonizedwater and acetonitrile (ACN) (50% v/v) were used as mobilephases. Trifluoracetic acid (0.14% v/v) was added to themobile phases as ion pairing reagent and thus enablingdendrimer separation with a strictly reverse phase system.Typically, after a 5 min hold in 100% water, a linear gradientwas employed from 0 to 50% ACN over 10 min followed byholding the system at 50% ACN water for 5 min. The injectionvolume was 25 µL with a solvent flow rate of 2 mL/minute.Dendrimer concentrations were determined by integrationof the 210 nm chromatograms using Agilent’s Chemstationsoftware. The dendrimer concentrations in the WSO phaseand reference solution were measured three times andaveraged to determine C d

o and C dw. The concentration of

dendrimer in the octanol phase C do ) (C d

total - C dw) was

determined by simple mass balance. logKow was expressedas follows:

ResultsTo our knowledge, only a few measured values of logKow fordendrimers have been reported in the literature (20, 21).Thus, for the PAMAM dendrimers evaluated in this study, itwas important to optimize the slow-stirring technique andHPLC assay to determine (1) the time required to reachequilibrium, (2) the optimum dendrimer concentration, and(3) the accuracy of the logKow measurements.

Measurements of Dendrimer Concentration by HPLC.In a previous article, Diallo et al. (22) showed that UV-visspectroscopy (λ ) 201 nm) can be used to measure theconcentration of PAMAM dendrimers in aqueous solutions.Because n-octanol absorbs UV light at λ ∼210 nm, our initialattempts to measure dendrimer concentration with UV inOSW solutions were unsuccessful. Thus, in this study, weused HPLC to measure dendrimer concentrations in thereference aqueous and OSW solutions. This techniqueexploits the ability of protonated PAMAM dendrimers to formion pairs with trifluoroacetic acid (TFA) that can be sorbedonto and eluted from a C5 HPLC column. Note that thehydrophobicity of a PAMAM-TFA ion pair will depend onthe size of the dendrimer and its number/type of terminalgroups. This suggests that the elusion times of the sorbedPAMAM-TFA ion pairs will vary with generation and terminalgroup chemistry. SI Figures S2 and S3 show the effects ofdendrimer generation and terminal group chemistry onelution time. For the Gx-NH2 PAMAM dendrimers withethylene diamine (EDA) core, we observed an increase inelution times with dendrimer generation (from 4.72 min forG1 to 7.3 min for G6). Similarly, the elution times of thenonamine terminated PAMAM dendrimers depend onterminal group chemistry: Tris- (t ) 5.6 min), carboxylate (t) 6.3 min), amine (t ) 6.3 min), succinamic acid (t ) 6.9min), amidoethanolamine (t ) 6.9 min), pyrrolidinone (t )8.7 min). Our results are consistent with those reported byIslam et al. (19). In all cases, the chromatograms are wellresolved thereby suggesting HPLC is effective at separatingPAMAM dendrimers. SI Figure S4 shows (as a typical example)

that HPLC with UV detection (λ ∼210 nm) can be used toquantify the concentration of PAMAM dendrimers in aqueoussolutions. A series of standard solutions (10-60 µM ofdendrimer) were used to generate the INSET calibration curveshown in SI Figure S4. Note that for all dendrimers evaluatedin this study, we were able to detect and quantify very smalldifferences (∼1%) between the areas of their HPCL chro-matograms in OSW solutions before and after equilibrationwith WSO solutions. SI Figure S5 shows a typical example ofthe integrated chromatographic areas of a G4 PAMAMdendrimer with succinamic acid groups in OSW before (area) 6051 ( 6 counts) and after equilibration [area ) 5985 (2 counts]. In all cases, we measured the HPLC chromatogramareas three times to ensure that they were consistent andreproducible. We also found that the HPLC chromatogramsbecome less resolved as dendrimer concentration decreases(SI Figure S4). Below a dendrimer concentration of 10 µM,the chromatograms of the PAMAM dendrimers were not wellresolved. This suggests that the HPLC method is only effectiveat quantifying dendrimer concentrations greater than 10 µMusing our experimental setup and measurement procedures.

Optimum Reaction Time, Dendrimer Concentration andAccuracy of LogkOw Measurements. Figure 2 illustrates theeffect of reaction time and dendrimer concentration on thelogKow of the G4-NH2 PAMAM dendrimer. The logKow of thisdendrimer reached constant value (-2.38) after 100 h of slow-stirring. De Bruijin et al. (20) have reported that the transferof highly hydrophobic compounds (with logKow> 5) fromn-octanol to water in slow-stirring experiments reachedequilibrium in 4-5 days. Interestingly, we also found thatthe transfer of highly hydrophilic PAMAM dendrimers fromwater to n-octanol also reached equilibrium in 4-5 days.Thus, we have allowed 120 h slow-stirring in all experiments.Figure 2 also shows that the logKow of the G4-NH2 PAMAMdendrimer remains constant as the concentration of den-drimer was increased from 30 to 52 µM. Thus, dendrimerconcentration was kept constant at 2.24 mM of equivalentterminal groups in all subsequent logKow measurements. Forexample, this corresponds to a concentration of 34.0 µM forall G4 PAMAM dendrimers with 64 terminal groups. Becausethe logKow measurements required significant amounts ofdendrimers, we only carried out three replicate measure-ments in selected cases to estimate an upper limit for theaccuracy of the logKow measurements. However, we used atleast two replicate in all cases. Note that the logKow

measurements were reproducible in all cases (Figures 3-6).We subsequently selected three generation (G4) with differentterminal groups (amine, succinamic acid and pyrrolidinone)and carried out three replicate logKow measurements for thesedendrimers. These replicate measurements and average

log Kow )Cd

o

CdW

(1)

FIGURE 2. Effects of stirring time and dendrimer concentrationon the logKow of G4-NH2 PAMAM dendrimer at roomtemperature and pH 7.4.

VOL. 43, NO. 13, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 5125

logKow are given in Table 2. The corresponding standarddeviations are, respectively, equal to 0.02, 0.06, and 0.08 forthe G4 PAMAM dendrimers with amine, succinamic acid,and pyrrolidinone terminal groups. Based on the results ofthese replicate experiments, we estimate the accuracy of thelogKow of the PAMAM dendrimers evaluated in this study tobe ∼ (0.10.

Effects of pH on logKow. We assessed the effect of solutionpH on the logKow of PAMAM dendrimers using the G4-NH2

dendrimer as model system. Figure 3 shows the logKow valuesof the G4-NH2 dendrimer in acidic (pH 4.0), neutral (pH 7.4)and basic (ph 12.0) OSW solutions. Note that this dendrimer

has higher (less negative) logKow (-2.15) at acidic pH (4)than at neutral (7.4) and basic pH (12). There is no significantchange in the dendrimer logKow (-2.38) when pH changesfrom neutral to basic.

Effects of Dendrimer Generation, Terminal Group andCore Chemistry on logKow at pH 7.4. Tables 1, 2, and 3 listthe logKow values at physiological pH 7.4 of the 18 PAMAMdendrimers (Figure 1) evaluated in this study. Note that thelogKow of the dendrimers are all negative thereby suggestingthese dendrimers are hydrophilic. Figure 4 shows that thelogKow of Gx-NH2 PAMAM dendrimers gradually decreasewith generation/size (i.e., become more negative) fromgeneration 1 (G1) to generation 4 (G4) followed by a slightincrease at generation 5 (G5). To our knowledge, only a limitednumber of measured logKow values for PAMAM dendrimershave published. Najlah et al. (20) reported that G1-NH2

PAMAM dendrimers functionalized with two different ter-minal hydrophobic groups (terfenadine-succinyl and ter-fenadine-succinyl-diethylene glycol) have negative logKow

values (-0.16). These measurements corroborate our findingof negative logKow values for the 18 PAMAM dendrimersevaluated in this study. Figure 5 and Table 2 illustrate theeffect of dendrimer terminal chemistry on the logKow ofPAMAM dendrimers. We tested seven different dendrimerterminal groups: cationic (amine, amidoethylethanolamine),anionic (carboxylate, succinamic acid, and pyrrolidinone),and neutral (amidoethanol and tris(hydroxymethyl)ami-domethane (Tris)). Note that the G4 PAMAM dendrimer withTris terminal groups has the highest (less negative) logKow

(-1.39). Figure 6 and Table 3 show the variation of logKow

values of G4-NH2 dendrimer with different core. There is aslight increase of the logKow values of the G4-NH2 PAMAMdendrimers as the number of their core carbon atoms increasefrom 2 (EDA) to 6 (DAH).

FIGURE 3. Effects of solution pH on the logKow of G4-NH2PAMAM dendrimer at room temperature and pH 7.4.

FIGURE 4. Effects of generation on the logKow of Gx-NH2PAMAM dendrimers at room temperature and pH 7.4.

FIGURE 5. Effects of dendrimer terminal group chemistry on thelogKow of G4 PAMAM dendrimers (G3.5 for Na-Carboxylate) atroom temperature and pH 7.4. All the dendrimers have an EDAcore except the G4 dendrimer with amidoehylethanolamineterminal groups which has a DAB core.

FIGURE 6. Effects of dendrimer core chemistry on the logKow ofG4 PAMAM dendrimer at room temperature and pH 7.4.

TABLE 1. Octanol-Water Partition Coefficients (logKow) ofPAMAM Dendrimers with Ethylene Diamine (EDA) Core andTerminal NH2 Groups at Room Temperature and pH 7.4

generation logKow1 logKow2 logKow3 logKow (reported)

G0 -1.48 -1.53 -1.53G1 -1.81 -1.81 -1.81G2 -1.91 -1.99 -1.99G3 -2.24 -2.24 -2.24G4 -2.39 -2.36 -2.39 -2.38 ( 0.02G5 -2.26 -2.16 -2.16G6 aNDG8 aND

a ND: not determined. This dendrimer caused theformation of an emulsion at octanol-water interface (seeSI Figure S6).

5126 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 13, 2009

DiscussionDue to the wide utilization of logKow in drug design, medicinalchemistry, and environmental assessment, a database of over60 000 measured values of logKow for organic molecules hasbeen developed by the Biobyte Corporation (24). Severalmethods of estimations of logKow from “structure” have alsobeen developed during the last 20 years. Quantitative-structure activity relationships (QSAR) of estimation of logKow

include the fragment-based methods (CLOGP) of Hanschand Leo (25), the extended group contribution method(KLOGP) of Klopman et al. (26), and the atomic constantapproach (ALOGP) of Goshe et al. (27). Although these QSARhave been shown to reproduce very well the logKow valuesof their training data sets, there are several classes of solutessuch as peptides and nucleosides which often cause them“difficulties”. To quote Leo “when CLOGP fails badly, thereare strong indications that it is conformational informationthat is lacking” (25).

The G4-NH2 PAMAM dendrimer provides a good modelsystem for explaining the effect of solution pH and macro-molecular conformation on the octanol-water partitioningbehavior of PAMAM dendrimers. First, a G4-NH2 PAMAMdendrimer exhibits three distinct charged states in water asa function of solution pH (18). At acidic pH (<5), all theprimary and tertiary amine groups of a G4-NH2 PAMAMdendrimer are protonated. At neutral pH (∼7.0), only theprimary amine groups of this dendrimer are protonated.Conversely, all the amine groups of the G4-NH2 PAMAMbecome unprotonated and neutral at basic pH (>10.0). Figure7 shows the 3D structures of a G4-NH2 PAMAM in water withCl- counterions at these three distinct protonation levels asdetermined from atomistic molecular dynamics (MD) simu-lations by Yi et al. (28). By using a modified Dreiding forcefield with interaction potentials that were fitted to high leveldensity functional theory (DFT) calculations, Yi et al. (28)obtained calculated radii of gyration (Rg) (∼2.11-2.20 nm)that were in excellent agreement with estimated Rg from smallangle neutron scattering (SANS) experiments (29). Note thatthe G4 PAMAM dendrimer undergoes drastic conformationalchanges as solution pH decreases (Figure 7). Although theRg of the G4-NH2 PAMAM dendrimer does not vary with pH,

Yi et al. (28) found that a decrease in solution pH causes aredistribution of mass within the G4 PAMAM dendrimer asits conformation changes from a “dense core” at high pH toa “dense shell” at low pH. They also found that the backfoldingof the terminal NH2 groups of the G4 PAMAM is mainlylocalized at the dendrimer periphery. Thus, the conforma-tional flexibility of the G4 PAMAM dendrimer in aqueoussolutions of different pH strongly suggest that QSAR basedon group contributions (e.g., CLOGP, KLOGP, and ALOGP)cannot provide reliable estimates of logKow for NM such asdendrimers.

At the present time, we do not have a quantitativeatomistic model for explaining the effect of solution pH onthe logKow of the G4-NH2 PAMAM dendrimer. However, thiseffect could be attributed to a competition between theelectrostatic and cavity contributions to the free energy oftransfer of a dendrimer from water to n-octanol. The logKow

(i) of a nonassociating solute “i” can be expressed as follows(30):

Where ∆Gwo (i) is the free energy of transfer of solute “i” from

water to n-octanol; ∆Go(i) and ∆Gw(i) are, respectively, thesolvation free energies of “i” in n-octanol and water; T is thetemperature and R is the ideal gas constant. Macroscopicsolvent models can provide simple, fast, and reliable meansof calculating the solvation energies of biological macro-molecules such as proteins (31). These models treat thesolvent as a dielectric continuum and the solute as “lowdielectric” molecular cavity with point charges placed onthe atomic nuclei of its van der Waals surface. The basicpremise of a macroscopic solvent model is that the elec-trostatic and nonpolar contributions to the solvation freeenergy of solute can be evaluated separately. Following Honiget al. (30), the solvation free energy of solute “i” in solvent“s” [∆Go(i)] can be expressed as follows:

where ∆G sel(i) and ∆G s

c(i) are, respectively, the electrostaticand cavity (i.e., nonpolar) contributions to the free energyof solvation of solute in solvent s; γs and As are, respectively,the microscopic surface tension and solvent-accessiblesurface area of solute “i” in solvent “s”. Substituting eqs 3,4, and 5 into eq 2 yields

TABLE 2. Octanol-Water Partition Coefficients (logKow) of PAMAM Dendrimers with Different Terminal Groups at RoomTemperature and pH 7.4

generation terminal group logKow1 logKow2 logKow3 logKow (reported)

4 amine -2.39 -2.36 -2.39 -2.38 ( 0.024 amidoethanol -2.47 -2.49 -2.474 succinamic acid -2.53 -2.59 -2.48 -2.53 ( 0.063.5 sodium carboxylate -2.30 -2.33 -2.334 tris(hydroxymethyl) aminomethane -1.39 -1.39 -1.394 pyrrolidinone -2.55 -2.42 -2.39 -2.45 ( 0.084 amidoethylethanolamine -2.54 -2.54 -2.54

TABLE 3. Octanol-Water Partition Coefficients (logKow) of G4PAMAM Dendrimers with Different Cores at RoomTemperature and pH 7.4

corea terminal group logKow1 logKow2 logKow3 logKow (reported)

EDA amine -2.39 -2.36 -2.39 -2.38 ( 0.02DAB amine -2.43 -2.35 -2.35DAH amine -2.25 -2.25 -2.25DAD amine -2.26 -2.33 -2.33Cyst amine -2.62 -2.73 -2.62

a EDA (ethylene diamine); DAB (diaminobutane); DAH(diaminohexane); DAD (diaminododecane), and Cyst(cystamine).

logKow )-∆GW

o (i)

2.30RT(2)

∆Gwo (i) ) ∆Go(i) - ∆Gw(i) (3)

∆Gs(i) ) ∆Gsel(i) + ∆Gs

c(i) (4)

∆Gsc(i) ) γsAs(i) (5)

logKow )[∆Gw

el(i) + γwAw(i)] - [∆Goel(i) + γoAo(i)]

2.30RT(6)

VOL. 43, NO. 13, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 5127

Equation 6 provides a thermodynamic framework forexplaining the effect of solution pH on logKow of the G4-NH2

PAMAM dendrimer. First, the negative values of the logKow

of PAMAM dendrimers (Tables 1-3) indicate that dendrimerhydration free energy [∆G w

el(i) + γwAw(i)] is very favorable(i.e., smaller than dendrimer solvation free energy in theoctanol phase). Recall that at pH 4.0, all the primary andtertiary amine groups of a G4-NH2 PAMAM dendrimer areprotonated (18). However, SANS experiments by Chen et al.(29) and atomistic MD simulations by Liu et al. (28) (Figure7) have established that a significant fraction (>66%) of theprotonated amine groups of a G4 PAMAM dendrimer becomeneutralized by bound Cl- counterions at acid pH (<5).Moreover Liu et al. (28) have found that the solvent (water)accessible of the G4 PAMAM dendrimer is lower in magnitude(by ∼15%) than at neutral pH. Thus, we hypothesize that thesolvation free energy of the G4 PAMAM dendrimer in water[∆Gw(i)] to become less favorable as its cavity term ∆G s

c(i)decrease in magnitude and its electrostatic component ∆G w

el(i)becomes less negative due the neutralization of its protonatedamine groups by bound Cl- anions at pH 4.0.

Figure 3 and Table 1 show a slight increase of logKow forthe G5 PAMAM dendrimer (-2.16) compared to that of theG4 PAMAM dendrimer (-2.38). SI Figure S6 shows that theG6-NH2 PAMAM dendrimer causes the formation of emul-sions as illustrated by the cloudiness of the octanol-water(O/W) interface. We hypothesized this increased affinity forthe O/W interface of the G5-NH2, G6-NH2, and G8-NH2

PAMAM dendrimers may be the result of changes in theirhydrophilic-lipophilic balance due to global and localconformational events within the macromolecules. First, Gx-NH2 PAMAM dendrimers undergo a gradual transition inoverall shape (from a more extended conformation for“earlier” generation dendrimers (Ge3.0) to a more compact/globular shape for “later” generation dendrimers (Gg 5.0)])(31). Second, atomistic MD simulations with explicit watermolecules by Yi et al. (28) (Figure 7) have shown that thebackfolding of the terminal NH2 groups of PAMAM den-drimers at neutral pH is mainly localized at the dendrimerperiphery. Thus, it is reasonable to postulate that global andlocal conformational changes within the more compact andsterically crowded G6-NH2 and G8-NH2 PAMAM dendrimerscould cause these macromolecules to act as amphiphiles atthe O/W interface. This hypothesis is consistent with recentexperiments showing that homologues of p-methyl benzyl

alkylamines also undergo conformational changes andaccumulate at the octanol-water interface (32). However,independent experimental measurements and atomistic MDsimulations will be needed to corroborate this hypothesis.

Except for the PAMAM dendrimer with sodium carboxy-late terminal group (G3.5), all dendrimers were fourthgeneration (G4) and thus have similar size and same numberof tertiary amine and amide groups. Note that the highest(i.e., less negative) logKow value (-1.39) for the G4 PAMAMdendrimer with Tris terminal groups may be attributed dueto its large number of terminal OH groups (64 × 3 ) 192).Thus, its increased affinity for the octanol phase could beattributed to increased H-bonding between the G4-Trisdendrimer macromolecules and n-octanol molecules. ThelogKow values of all the other G4 PAMAM dendrimers arecomparable in magnitude ranging from -2.54 to -2.33. Notsurprisingly, the logKow of the G4-NH2 PAMAM with EDA,DAB, DAH, and DAD cores are also comparable ranging from-2.38 to-2.25. However, the G4-NH2 PAMAM with cystaminecore has the lowest logKow value (-2.62). At the present time,we do not have a quantitative explanation for this observation.

Environmental ImplicationsAs previously stated, the logarithm of the partition coefficientof an organic compound between n-octanol and water(logKow) is one of the most widely used measures of itstendency to accumulate in natural biomedia such as thelipid membranes of animals, plants, and bacteria (15). Severalinvestigators have shown that the lipid-water coefficients(logKlipw) of many polar and nonpolar organic pollutants arepositively and often linearly correlated with their logKow (15).Because of these correlations, logKow is often employed asdescriptor in QSAR used to evaluate the baseline toxicity ofxenobiotic organic compounds as they accumulate in naturalbiomedia such as lipid bilayers (15). However, to ourknowledge, correlations between the logKow of organicnanomaterials such as dendrimers and their accumulationin natural biomedia have not been established. Note thatthe G1-G5 PAMAM dendrimers with terminal NH2 groupsevaluated in this study (Figure 1) have negative logKow valuesthereby suggesting they prefer to remain in water (Table 1).Conversely, the formation of stable emulsions at theoctanol-water (O/W) interface in the presence of G6-NH2

and G8-NH2 PAMAM dendrimers suggest they prefer to

FIGURE 7. Three-dimensional structures of a G4-NH2 PAMAM dendrimer in water (at low, neutral, and high pH) from atomisticmolecular dynamics (MD) simulations (see ref 28). The green atoms are bound Cl- counterions that neutralize the protonated primaryand tertiary amine groups of the dendrimer.

5128 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 13, 2009

partition at the O/W interface (SI Figure S6). However,published atomic force microscopy (AFM) imaging andcytotoxicity studies show that Gx-NH2 PAMAM dendrimersstrongly interact with the lipid bilayers of cells (10-12). Thissuggests that logKow of a PAMAM dendrimer may not be agood predictor of its affinity with natural organic media suchas the lipid bilayers of cell membranes.

AcknowledgmentsThis work was carried out in the Materials Process SimulationCenter and Power, Environmental and Energy ResearchCenter of the Division of Chemistry and Chemical Engineer-ing at the California Institute of Technology. Funding wasprovided by the Environmental Protection Agency (EPA STARGrant RD832525) and the National Science Foundation (NIRTCBET Award No. 050695).

Supporting Information AvailableFurther details are shown in three tables and six figures. Thismaterial is available free of charge via the Internet at http://pubs.acs.org.

Literature Cited(1) U.S. NNI. http://www.nano.gov/html/facts/nanoapplication-

sandproducts.html.(2) Dendrimers and Other Dendritic Polymers; Frechet, J. M. J.,

Tomalia, D. A., Eds.; Wiley and Sons: New York, 2001.(3) Helms, B.; Meijer, E. W. Dendrimers at work. Science. 2006, 313,

929–930.(4) Tomalia, D. A., Henderson, S. A.; Diallo, M. S. DendrimerssAn

enabling synthetic science to controlled organic nanostructures.In Handbook of Nanoscience, Engineering and Technology,Chapter 24, 2nd ed.; Goddard, W. A. III.; Brenner, D. W.;Lyshevski, S. E. and Iafrate, G. J.; Eds.; CRC Press: Boca Raton,FL, 2007.

(5) Diallo M. S. Water treatment by dendrimer-enhanced filtration:Principles and applications. In Nanotechnology Applicationsfor Clean Water; Savage, N., Diallo, M. S, Duncan, J., Street,S. R., Eds.; William Andrew Applied Science Publishers: Norwich,2009.

(6) Tinkle, S. S. Nanotechnology: collaborative opportunities forecotoxicology and environmental health. Environ. Toxicol.Chem. 2008, 27, 1823–1824.

(7) Roberts, J. C.; Bhalgat, M. K.; Zera, R. T. Preliminary biologicalevaluation of polyamidoamine (PAMAM) Starburst dendrimers.J. Biomed. Mater. Res. 1996, 30, 53–65.

(8) Malik, N.; Wiwattanapatapee, R.; Klopsch, R.; Lorenz, K.; Frey,H.; Weener, J. W.; Meijer, E. W.; Paulus, W.; Duncan, R.Dendrimers: Relationship between structure and biocompat-ibility in vitro, and preliminary studies in the biodistribution of125I-labelled polyamidoamine dendrimers in vivo. J. ControlledRelease 2000, 65, 133–148.

(9) Fuchs, S.; Kapp, T.; Otto, H.; Schoneberg, T.; Franke, P.; Gust,R.; Schluter, A. D. A Surface-modified dendrimer set for potentialapplication as drug delivery vehicles: synthesis, in vitro toxicityand intracellular localization. Chem.sEur. J. 2004, 10, 1167–1192.

(10) Hong, S. P.; Bielinska, A. U.; Mecke, A.; Keszler, B.; Beals, J. L.;Shi, X. Y.; Balogh, L.; Orr, B. G.; Baker, J. R.; Holl, M. M. B.Interaction of poly(amidoamine) dendrimers with supportedlipid bilayers and cells: hole formation and the relation totransport. Bioconjugate Chem. 2004, 15, 774–782.

(11) Leroueil, P. R.; Hong, S. Y.; Mecke, A.; Baker, J. R.; Orr, B. G.;Holl, M. M. B. Nanoparticle interaction with biological mem-branes: Does nanotechnology present a janus face? Acc. Chem.Res. 2007, 40, 335–342.

(12) Leroueil, P. R.; Berry, S. A.; Duthie, K.; Han, G.; Rotello, V. M.;McNerny, D. Q.; Baker, J. R.; Orr, B. G.; Holl, M. M. B. Widevarieties of cationic nanoparticles induce defects in supportedlipid bilayers. Nano Lett. 2008, 8, 420–424.

(13) Metullio, L.; Ferrone, M.; Coslanich, A.; Fuchs, S.; Fermeglia,M.; Paneni, M. S.; Pricl, S. Polyamidoamine (Yet not PAMAM)dendrimers as bioinspired materials for drug delivery: sStruc-ture-activity relationships by molecular simulations. Biomac-romolecules. 2004, 5, 1371–1378.

(14) Heiden, T. C. K.; Dengler, E.; Kao, W. J.; Heideman, W.; Peterson,R. E. Developmental toxicity of low generation PAMAM den-drimers in zebrafish. Toxicol. Appl. Pharmacol. 2007, 225, 70–79.

(15) Schwarzenbach, R. P.; Gschwend, P.; Imboden, D. M. Envi-ronmental Organic Chemistry, 2nd.; John Wiley & Sons, Inc:New York, 2003.

(16) Sangster, J. Octanol-Water Partition Coefficients: Fundamentalsand Physical Chemistry, Wiley Series in Solution Chemistry;John Wiley & Sons: New York, 1997; Vol. 2.

(17) Klaine, S. J.; Alvarez, P. J. J.; Batley, G. E.; Fernandes, T. F.; Handy,R. D.; Lyon, D. Y.; Mahendra, S.; Mclaughlin, M. J.; Lead, J. R.Nanomaterials in the environment: Behavior, fate, bioavailabilityand effects. Environ. Toxicol. Chem. 2008, 27, 1825–1851.

(18) Diallo, M. S.; Chritie, S.; Swaminathan, P.; Balogh, L.; Shi, X.;Um, W.; Papelis, L.; Goddard, W. A. III.; Johnson, J. H. Jr.Dendritic chelating agents 1. Cu(II) binding to ethylene diaminecore poly(amidoamine) dendrimers in aqueous solutions.Langmuir. 2004, 20, 2640–2651.

(19) Islam, M. T.; Shi, X. Y.; Balogh, L.; Baker, J. R. HPLC separationof different generations of poly(amidoamine) dendrimersmodified with various terminal groups. Anal. Chem. 2005, 77,2063–2070.

(20) Najlah, M.; Freeman, S.; Attwood, D.; D’Emanuele, A. Synthesisand assessment of first-generation polyamidoamine dendrim-erprodrugs to enhance the cellular permeability of P-gpsubstrates. Bioconjugate Chem. 2007, 18, 937–946.

(21) Florence, A. T.; Sakthivel, T.; Toth, I. Oral uptake and trans-location of a polylysine dendrimer with a lipid surface. J.Controlled Release 2000, 65, 253–259.

(22) Diallo, M. S.; Chritie, S.; Swaminathan, P.; Johnson, J. H., Jr;Goddard, W. A. III. Dendrimer enhanced ultrafiltration. 1.Recovery of Cu(II) from aqueous solutions using Gx-NH2

PAMAM dendrimers with ethylene diamine core. Environ. Sci.Technol. 2005, 39, 1366–1377.

(23) De Bruijn, J.; Busser, F.; Seinen, W.; Hermens, J. Determinationof octanol/water partition coefficients for hydrophobic organicchemicals with the ‘slow-stirring’ method. Environ. Sci. Technol.1989, 8, 499–512.

(24) Bio-Loom. Available at http://www.biobyte.com/bb/prod/bioloom.html

(25) Leo, A. J. Calculating log Poct from structures. Chem. Rev. 1993,83, 1281–1306.

(26) Sedykh, A. Y.; Klopman, G. Structural analogue approach to theprediction of the octanol-water partition coefficient. J. Chem.Inf. Model. 2006, 46, 1598–1603.

(27) Goshe, A. K.; Wishwanadhan, V. N.; Wendoloski, J. J. Predictionof hydrophobic (Lipophilic) properties of small organic mol-ecules using fragment methods: an analysis of ALOGP andCLOGP methods. J. Phys. Chem. A. 1998, 102, 3762–3772.

(28) Liu, Yi.; Bryantsev, V. S.; Diallo, M. S.; Goddard, W. A. III. PAMAMdendrimers undergo pH responsive conformational changeswithout swelling. J. Am. Chem. Soc. 2009, 131 (8), 2798–2799.

(29) Chen, W. R.; Porcar, L.; Liu, Y.; Butler, P. D.; Magid, L. J. Smallangle neutron scattering studies of the counterion effects onthe molecular conformation and structure of charged G4PAMAM dendrimers in aqueous solutions. Macromolecules.2007, 40, 5887–5898.

(30) Honig, B.; Sharp, K.; Yang, A. S. Macroscopic models of aqueous-solutions - biological and chemical applications. J. Phys. Chem.1993, 97, 1101–1109.

(31) Bosman, A. W.; Janssen, H. M.; Meijer, E. W. About Dendrimers:Structure, physical properties and applications. Chem. Rev. 1999,99, 1665–1668.

(32) Fruttero, R.; Caron, G.; Fornatto, E.; Boschi, D.; Ermondi, G.;Gasco, A.; Carrupt, P.; Testa, B. Mechanisms of liposomes/waterpartitioning of (p-methylbenzyl) alkylamine. Pharm. Res. 1998,15, 1407–1413.

ES9003747

VOL. 43, NO. 13, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 5129