capillary electrophoresis of polycationic poly(amidoamine) dendrimers

Xiangyang Shi1

István Bányai2

Wojciech G. Lesniak1

Mohammad T. Islam1

István Országh2

Peter Balogh1

James R. Baker, Jr.1

Lajos P. Balogh1

1Center for BiologicNanotechnology, University ofMichigan Medical School,Ann Arbor, MI, USA

2Institute of Physical Chemistry,University of Debrecen,Debrecen, Hungary

Capillary electrophoresis of polycationicpoly(amidoamine) dendrimers

Generation 2 to generation 5 poly(amidoamine) (PAMAM) dendrimers having differentterminal functionalities were analyzed by capillary electrophoresis (CE). Polyacryl-amide gel electrophoresis was also used to assess the composition of the individualgenerations for comparison with the CE results. Separation of PAMAMs can beaccomplished by either using uncoated silica or silanized silica capillaries, althoughreproducibility is poor using the uncoated silica capillary. To improve run-to-run repro-ducibility, silanized capillary was used and various internal standards were also tested.Relative and normalized migration times of primary amine terminated PAMAM den-drimers were then determined using 2,3-diaminopyridine (2,3-DAP) as an internalstandard. Using silanized capillaries and internal standards, the relative and normal-ized migration times are fully reproducible and comparable between runs. Apparentdimensionless electrophoretic mobilities were determined and the results were com-pared to theoretical calculations. It is concluded that for PAMAMs a complex separa-tion mechanism has to be considered in CE, where the movement of the ions is due tothe electric field, but the separation is rather the consequence of the adsorption/de-sorption equilibria on the capillary wall (“electrokinetic capillary chromatography”). Thedescribed method may be used for quality control and may serve as an effectivetechnique to analyze polycationic PAMAM dendrimers and their derivatives with dif-ferent surface modifications.

Keywords: Capillary electrophoresis / Internal standards / Poly(amidoamine) dendrimersDOI 10.1002/elps.200500134

1 Introduction

Dendrimers are synthetic, highly branched, nearly sphe-rical and symmetrical macromolecules with well-definedsizes and compositions. Tailored dendrimers can be syn-thesized by appropriately selecting the cores, connectingunits, branching sites and terminal groups. Since the pio-neering works of the Tomalia, Newkome, and Vogtlegroups in the 1980s [1–4], a surge of interest has beenexperienced in the synthesis and applications of den-drimers [5–7]. Altering terminal functionalities of commer-cially available dendrimer materials, such as poly(amidoamine) dendrimers (Starburst™ PAMAMs) andpoly(propyleneimine) (Astramol™, PPI) dendrimers, afforda wide range of potential applications in catalysis [8], drugdelivery [9, 10], biosensor [11–13], optics [14], and elec-

tronics [15]. As the efforts to synthesize and manufacturevarious dendrimers intensify, the quality control and char-acterization of various dendritic macromolecular materialsstill offers challenges. Dendritic polymers are often talkedabout as close-to-perfect structures, but practical sam-ples should be handled as narrow size-distribution andcharge-distribution polymers [16]. In addition to the rou-tine chemical analysis methods (UV-vis, Fourier transforminfrared (FTIR), and NMR), numerous methods have beenemployed for the characterization of PAMAM dendrimersincluding potentiometric titration, scattering methods(small-angle Y-ray scattering (SAXS), small-angle neutronscattering (SANS), dynamic light scattering), SEC, HPLC,ESI-MS and MALDI-TOF-MS methods [5, 17, 18].

CE is a powerful chromatographic method in the analysisof biologic macromolecules, such as DNA [19], proteins[20], peptides [21], etc. CE has high efficiency, high sen-sitivity, short run time, high automation capability, and issuitable for the routine analysis of diverse dendrimerarchitectures, especially those that are carrying multiplecharges. Still, only a limited number of reports have beenpublished about the analysis of polyelectrolyte den-

Correspondence: Professor Lajos P. Balogh, Department ofRadiation Medicine, Roswell Park Cancer Institute, Elm and Carl-ton Streets, Buffalo, NY 14263, USAE-mail: [email protected]: 1716-845-8254

Abbreviations: 2,3-DAP, 2,3-diaminopyridine; 2,6-DAP, 2,6–dia-minopyridine; DAPM, 2,4-diaminopyrimidine; DMAP, 4-dimethy-laminopyridine; EDA, ethylenediamine; GPC, gel permeationchromatography; PAMAM, polyamidoamine

Electrophoresis 2005, 26, 2949–2959 2949

Supplementary material for this article is available on the WWWunder www.electrophoresis-journal.de.

© 2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

CE

and

CE

C

2950 X. Shi et al. Electrophoresis 2005, 26, 2949–2959

drimers using CE [22–26] and, in general, there have beenno reliable methods available for the CE analysis ofPAMAM dendrimers.

CE is a fast and reliable practical analytical technique thatcan also provide data of theoretical interest for PAMAMdendrimers, the main question being how much chargethe dendrimer molecules in fact carry in solution. Forexample, amine-terminated PAMAM dendrimers maycarry a large number of positive charges when protonatedand their high generations display particle-like character-istics. Using a simple model for CE, there is a direct wayto estimate the effective charge that various PAMAMdendrimers experience under run conditions.

Theoretically, when a particle moves with a stationary rate,the sum of all the forces on that particle must be zero. AtpH 2.5, when a protonated and positively charged EX.NH2

(E refers to ethylenediamine (EDA) core and X refers to thegeneration number) PAMAM dendrimer molecule moves ina capillary by means of an electric field towards the cath-ode, the following main effects should be considered: First,the electric force (Fe) consists of two components: an elec-trophoretic part and an electroosmotic part. The latter isnegligible under our conditions because at this pH the wallof the silica capillary has no negative charge. Second, thereis a hydrodynamic force, which comes from the differenttype of “frictions” of the medium (hydrodynamic friction orviscosity and the dynamics of counter ion cloud or relaxa-tion effect). The mobility of the dendrimers can be written inEq. (1) in a dimensionless form:

m0 ¼m03Ze2ee0kT

¼ ILVtm

3Ze2ee0kT

(1)

where l is the effective length, L is the total length, V is thevoltage, tm is the migration time, m0 is the electrophoreticmobility, e is the dielectric constant of water (80.5), e0 isthe dielectric permittivity of the vacuum, k is the Boltzmanconstant, T is the temperature, Z is the viscosity and e ischarge of the electron.

Dendrimers are believed to be the first experimentalexample of moving charged spheres for which theHuckel-Smoluchowski theory [27, 28] works rather well.The electrostatics for the motion of dendrimer moleculesin the presence of a 1:1 electrolyte (z1 and z2 is equally 1)is described by the Poisson-Boltzman equation:

1r2

dd

r2 dcdr

� �¼ 2000INA

ee0sinh

eckT

� �(2)

The appropriate boundary conditions are: (i) when r = a,then dc/dr = 2s/e0e, wheres is the surface charge density;(ii) (r??) = 0.The remainingsymbols are as follows:c is thesurface potential, r is the distance from the center of the

sphere, a is the radius of the sphere, NA is Avogadro’s con-stant, and I is the ionic strength. Equation (2) can be solvednumerically, and some of the values are tabulated in the lit-erature [29, 30]. In these articles, the charge distributionfunction I0 is given, which correlates to s surface densityaccording to the following empirical formula:s = 5.8718(I)1/

2I0 6 1022 C?m22. The tables in Loeb’s work [29] show thevalues of reduced distances q0 = ka (where k is the Debye–Hückel parameter), and the reduced surface potentialsy0 = e0/kT as functions of each other and also I0, whichdepends on both. Huang et al. [26] found that in the case ofcarboxylic acid-terminated cascade polymers, the largenegative charge of the macromolecule is partially compen-sated by counterion binding to the spherical molecules,analogously to the Manning hypothesis [31] about counter-ion condensation on polyelectrolytes. Recently, Welch andHoagland [32] have reported that poly(propyleneimine) (PPI)dendrimers can be considered as moving charged spheresand follow the predictions of the standard electrokineticmodel described by Eq. (2). Three factors are dominant inthe motion of these spherical macromolecules towards theanode: the electric field, the hydrodynamic friction and therelaxation effect of the counterion cloud. Welch and Hoag-land could not uncover nonspecific ion binding. PAMAMdendrimerscarrying thesame numberofnitrogenatoms aremuch larger than the corresponding PPI dendrimers.Therefore, if there is a counterion effect, it should be morepronounced for PAMAMs than for PPI dendrimers.

Brothers et al. [22] and Ebber et al. [23] recently reportedthe application of CZE to separate and characterizePAMAM dendrimers using bare fused-silica capillary.According to our experience, the measured migrationtimes of polycationic PAMAM dendrimers were some-times irreproducible, and day-to-day data differed fromeach other even when the same sample, instrument,capillary and run conditions were applied. It is known thatpolycationic PAMAM dendrimers have a strong tendencyto be adsorbed onto negatively charged silica or glasssurfaces [33–35]. It has been reported that pKa of the ter-minal primary amine groups and the tertiary amine groupsof PAMAM dendrimers is between 9 and 10 and 4 and 5,respectively [2, 36a]. At low pH conditions (pH , 3.0),when all amine groups in a PAMAM molecule are proto-nated [36b], adsorption of dendrimers onto the capillarywall surface [37] makes the separation irreproducible.

In the present study, we investigated the separation ofpolycationic PAMAM dendrimers including derivatives ofdifferent generations and different termini both by PAGE[6] and CE. Knowing that silanization reduces theadsorption of PAMAM dendrimers on a silica substrate[38], we also compared the reproducibility of the separa-tions of primary amine-terminated (“full generation”)


Electrophoresis 2005, 26, 2949–2959 CE of polycationic PAMAM dendrimers 2951

PAMAMs using native silica or silanized capillary. We haveexpected that the chemical derivatization of the surfacesilanol groups of the capillary with an alkyl silane will de-crease the adsorption of protonated PAMAM dendrimersdue to the increased hydrophobic nature of the wall, thusit will increase the reproducibility. To reduce the experi-mental error, we also tested potential internal standardsand determined electrophoretic mobilities based onmeasured (relative and normalized) migration times ofvarious PAMAM dendrimers and their derivatives.

2 Materials and methods

2.1 Instrumentation

2.1.1 PAGE

Analysis of PAMAM dendrimers by PAGE was performedon a Micrograd vertical electrophoresis system (Gradi-pore, Sydney, Australia) with a commercial power supply(Model 500/200; BioRad, Hercules, CA, USA). Electro-phoresis experiments were performed on 10 6 8 cm gelsin a vertical electrophoresis unit (Model Protean I;BioRad). Precast 4–20% gradient express gels for PAGEwere obtained from ISC BioExpress (Kaysville, UT, USA).

2.1.2 CE

The CE instrument used in this work was purchased fromAgilent Technologies (Waldbronn, Germany). Unmodifiedsilica capillaries (100 mm ID 6 78.5 cm) were obtained fromPolymicro Technologies (Phoenix, AZ, USA). These capil-laries had an effective length of 70 cm. Some of thesecapillaries have been silanized as described in Section 2.3.All the capillaries had a total length of 78.5 cm and aneffective length 70 cm, unless specified otherwise. Thecapillary temperature was maintained at 407C (in order todecrease the adsorption of PAMAM dendrimers onto thecapillary surface) and separation voltage was kept at 20 kVfor all the separations. On-capillary UV diode-array detec-tion was used, operated at wavelengths of 200, 210, 250and 300 nm. Samples were introduced by hydrodynamicinjection at a pressure of 50 mbar.

2.2 Chemicals

Amine-terminated PAMAM dendrimers of generation 2through 4 with EDA core were purchased from Dendritech(Midland, MI, USA). E5.NH2 was synthesized in our center.All amine-terminated PAMAM dendrimers of different gen-eration were characterized with gel permeation chromatog-

raphy (GPC), HPLC, acid-base titration, and NMR. Chloro-trimethylsilane (1.0 M in THF), pyridine, 2,3-diaminopyridine(2,3-DAP), 2,6-diaminopyridine (2,6-DAP), 4-dimethylami-nopyridine (DMAP), 2,4-diaminopyrimidine (DAPM), andphosphoric acid were obtained from Aldrich and used asreceived. Acetyl (-COCH3) and glycidyl (-CH2CH(OH)-CH2OH)-terminated PAMAM derivatives of generation 4 and5, which are denoted as E4.NHAc, E5.NHAc and E5.(Gly)OH,were synthesized in our lab and were characterized by NMR,ESI-MS, and MALDI-TOF-MS.

2.3 Silanization of the capillary

The bare fused-silica capillary was pretreated by rinsingwith a 1.0 M NaOH solution [39, 40] for 30 min to activatethe hydroxyl groups on the silica surface, followed by anaqueous wash with water for 30 min, then with methanolfor 30 min and subsequently with THF for 20 min. Theinternal wall of the capillary was silanized by filling up thecapillary with the mixture of chlorotrimethylsilane andpyridine (10:1 v/v) for 24 h. The capillary was extensivelyrinsed using THF (for at least 20 min) then washed withwater and equilibrated with the running buffer. Rinsingand flushing were performed at room temperature undersyringe-induced vacuum.

2.4 Procedures

2.4.1 PAGE

Tris-glycine (TG) buffer (pH 8.3) was purchased fromInvitrogen (Carlsbad, CA, USA). It was diluted by a factorof ten to prepare the running buffer. PAGE separationstypically required 50 min at 200 V. Reverse polarity wasused for the analysis of the polycationic PAMAM den-drimers. Into every sample well 6 mL of a sample solutioncomposed of 3 mL 1 mg/mL PAMAM dendrimer and3 mL methylene blue sucrose dye solutions (50% sucrose,1% methylene blue) was injected. Developed gels werestained with 0.025% Coomassie Blue R-250 in 40%methanol and 7% acetic acid aqueous solution overnight.The gels were destained with 7% v/v acetic acid and5% v/v methanol in water. Image analysis was performedon a Macintosh computer using the public domain NIHImage program (developed at the U.S. National Institutesof Health and available on the Internet at http://rsb.info.-nih.gov/nih-image/).

2.4.2 CE

Before initial use, the uncoated, fused-silica capillarieswere rinsed with 0.1 M NaOH for at least 15 min, thenwashed with deionized water (resistivity of 18 MO?cm;



Milli-Q Plus 185 water purification system, Millipore,Bedford, MA, USA) for 15 min, and then conditioned withthe running buffer for an additional 15 min. Before eachinjection, the capillary was rinsed with a sequence of0.2 M H3PO4 (5 min) solution, followed by deionized water(5 min), then with the running buffer (5 min). For a silanizedcapillary, 0.2 M H3PO4 was used for both the daily initi-alization of the capillary and before each injection;50 mM phosphate buffer (pH 2.5, Agilent Technologies)was used as running buffer. PAMAM dendrimers possessa large number of amine groups in a relatively small vol-ume of the macromolecule thus they can be consideredas a powerful buffer system. When PAMAM dendrimersare directly dissolved in pH 2.5 (50 mM) phosphate buffer,an undesirable pH-change may occur depending on thesample concentration due to the high concentration ofnitrogen ligands. For example, 1 mg of PAMAM den-drimer consumes all the H3PO4 contained in 1 mL ofpH 2.5 (50 mM) buffer and the pH increases to 6–7.Hence, in this study, first we always adjusted the pH of theaqueous dendrimer solution to pH 2.5 by 0.2 M pho-sphoric acid to form a salt. Dendrimer salt solutions werethen freeze-dried for three days using a Labconco sys-tem. The dry PAMAM dendrimer phosphate salts werethen dissolved in a calculated amount of pH 2.5 phos-phate buffer to give a solution equivalent to c = 1 mg/mLPAMAM at the required pH. As a result of this procedure,every nitrogen atom in the dendrimer molecule is proto-nated [36b] (please note that even at pH 2, there is stillone unprotonated tertiary amine group for PAMAMs) andno aggregation takes place between individual den-drimers. We have also tested the influence of sampleinjection mode on the electropherograms of PAMAMdendrimers and compared hydrodynamic pressure injec-tion mode with electrokinetic injection. We found thatPAMAM separations were unaffected by the two modesof injection. Thus, in all of the subsequent studies, hydro-dynamic (pressure) injection was used.

3 Results and discussion

3.1 General characterization of PAMAMdendrimers by PAGE

Figure 1 shows the electropherograms of primary amine-terminated PAMAMs from generation 2 to generation 5(E2.NH2, E3.NH2, E4.NH2, and E5.NH2) separated on agradient polyacrylamide gel. Under the applied PAGEconditions (running buffer pH 8.3) primary amine terminiof the primary amine terminated PAMAM dendrimers areprotonated while their tertiary amine groups are not. It isalso reported [22] that when the buffer pH value wasdecreased to pH 3, the bands of dendrimers were rela-

Figure 1. (a) Electropherogram of several generations ofEDA-core amine-terminated PAMAM dendrimers ana-lyzed on a 4–20% express gel. Tris-glycine buffer (pH 8.3)was used as the running buffer. The left lane correspondsto a mixture of standard PAMAMs: E2.NH2, E3.NH2,E4.NH2, and E5.NH2. (b) Histogram of the dendrimer dis-tributions in which the heights of columns are proportionalto the density values of the bound dye (the density valuesshown are of the rest of the lanes) determined by thesoftware NIH Image.

tively tighter. However, the use of acidic pH values posesthe problem of high instrument heating, thus the commonTris-Glycine buffer (pH 8.3) was used for PAGE analysis.



In gradient PAGE, due to the gel filtration effect, lower gen-eration PAMAM macromolecules are more mobile thanhigher generation dendrimers and mixtures of E2.NH2,E3.NH2 E4.NH2 and E5.NH2 separate from each otheraccording to their molecular mass. PAGE electro-pherogram of amine-terminated PAMAMs (Fig. 1a) showsthat a given generation separates well from its dimer, whichhas practically the same molecular mass as the next highergeneration molecules do (molecular mass approximatelydoubles in each successive generation). Similarly, trailinggenerations (that originate from traces of EDA in the reac-tion mixture) [22, 41] appear in positions corresponding toprevious generations. The image may be evaluated forsemiquantitative information using image-analysis soft-ware although one has to remember that the measurableinformation is generated by the chromophores present inthe given dye bound by the dendrimer molecules, whichmay or may not be in a linear relation with the number (ormass) of dendrimer molecules (Fig. 1b).

3.2 Separation of polycationic PAMAMdendrimers of different generations by CE

The electrophoretic mobilitym0 for cations and polycationicdendrimers can be calculated from a simple equation [42]when EOF is negligible or absent EOF at pH 2.5.

m0 ¼IL

Vtm(3)

where l is the effective length of the capillary (from inlet todetector window), L is the total length of the capillary (bothin cm), V is the applied voltage, and tm is the migrationtime. The electrophoretic mobility of the internal standardsin the absence of dendrimers was found always constant(e.g., for 2,3-DAP tIS = 11.80 6 0.26 min, average of sixmeasurements). We assume that the internal standardsexperience the same varying CE conditions as PAMAMdendrimer samples do, consequently the relative PAMAMmigration times (trel = tm/tIS) should be constant. (Alter-natively, this number can be multiplied by the independ-ently determined tIS value and further expressed as a nor-malized migration time, which is also useful for compar-isons.) Thus, the use of relative or normalized mobilityvalues should permit direct comparison between runs ofvarying generations and functional groups.

The electrophoretic mobility of positively chargedPAMAM dendrimers under the described conditionspoints to the same direction as the EOF would, i.e., to-wards the cathode. However, at pH 2.5 the EOF is negli-gible [43]. When all amine groups (both primary and ter-tiary) in a dendrimer structure are protonated, the PAMAMdendrimers should have approximately similar electro-

phoretic mobility due to the nearly constant value of theircharge/mass ratios based on both theoretical calculationand practical measurements (see Table 1) [44]*. In spite ofthis anticipation, successful separations of differentPAMAM generations have been reported by CE using abare fused-silica capillary [22, 23].

Table 1. Molecular characteristics of amine-terminatedEDA core PAMAM dendrimers and their measuredrelative migration times (RMTs) and RSD relative toE2.NH2 as internal standard (the analysis was per-formed using a silanized silica capillary)

Dendrimer E2.NH2 E3.NH2 E4.NH2 E5.NH2

Mra) 3256 6909 14215 28826

Diametera) (nm) 2.9 3.6 4.5 5.4

No. of NH2 groupsa) 16 32 64 128

Total no. of N ligandsa) 30 62 126 254

Charge/Mr ratiob)(C/g) 0.009214 0.008973 0.008864 0.008811

Mrc) 3185 6648 12910 27250d)

Average no. of -NH2

groupsc)11.9 26.3 46.3 120d)

Total average no.of –N= ligandsc)

21.1 54 98.3 238d)

Charge/Mr ratioc) (C/g) 0.00662 0.00812 0.00761 0.00873

RMT/RSD (n = 8) 1.00(0%)

1.06(0.04%)

1.12(1.2%)

1.18(1.6%)

a) Theoretical valuesb) Assuming theoretical values and full protonation [36b]c) Practical values determined from titration and GPC

measurementsd) See [44]

We have compared the reproducibility of the separation ofa mixture of E2.NH2, E3.NH2, E4.NH2, and E5.NH2

PAMAM dendrimers using uncoated silica (Fig. 2) andsilanized capillary (Fig. 3).

Migration times of dendrimers are clearly decreasing inthe consecutive runs approaching a plateau. The longmigration times (30–35 min) (see Fig. 2a) of the first tworuns indicate strong binding of PAMAM dendrimers to thesilica wall. After an initial “conditioning” of the capillary, adynamic equilibrium develops and the PAMAM migrationtimes become very similar to the ones in the previousruns. Figure 2b shows the achievable separation of theprimary amine-terminated PAMAM generations E2.NH2,

* Regarding the value of charge/mass ratios in all the Tables inthis paper, molecular mass increase of PAMAMs due to proto-nation at pH 2.5 is not considered in the both theoretical andpractical calculations as compared to the PAMAM molecularweight, the increase is negligible.



Figure 2. (a) Relationship between the migration times ofE2.NH2, E3.NH2, E4.NH2, and E5.NH2 separated using abare fused-silica capillary as the function of consecutiverun numbers. (b) A typical capillary electropherogram ofthe mixture of EDA core amine-terminated PAMAM den-drimers of generation 2 to 5 using a bare fused-silicacapillary (ID 100 mm, total length 78.5 cm and effectivelength 70 cm). Injection time: 3 s. Original data is the realelectropherogram, and generated data was obtainedfrom Peakfit program. They are overlapped. Peaks 1, 2, 3,4, 5 correspond to E1.NH2 trace (in E2.NH2 dendrimer),E2.NH2, E3.NH2, E4.NH2, E5.NH2.

E3.NH2, E4.NH2, and E5.NH2 using a bare fused-silicacapillary in the fifth run. Peak shapes show signs ofstrongly non-linear adsorption, and generation 5 den-drimer is preferably retained from the mixture, which isindicated by the much smaller area under the E5.NH2

peak.

Figure 3. (a) Relationship between the migration timesand the consecutive run numbers for E2.NH2, E3.NH2,E4.NH2, and E5.NH2 dendrimers separated on silanizedsilica capillary. (b) A typical capillary electropherogram ofthe mixture of EDA core amine-terminated PAMAM den-drimers of generation 2 to 5 using a silanized silica capil-lary (ID 100 mm, total length 78.5 cm and effective length70 cm). Injection time: 3 s. Original data is the real elec-tropherogram, and generated data was obtained fromPeakfit program. They are overlapped. Peaks 1, 2, 3, 4, 5correspond to E1.NH2 trace (in E2.NH2 dendrimer),E2.NH2, E3.NH2, E4.NH2, E5.NH2.

Silanization of the internal wall decreases the number ofactive centers and not only leads to faster equilibration,but clearly makes the separation more reproducible.When silanized capillary was used for the separation ofthe same mixture of PAMAM dendrimers (Fig. 3a), themigration times rapidly decreased in subsequent injec-



tionsand stabilized at steady values. A typical separation atrun #6 is shown in Fig. 3b, for E2.NH2, E3.NH2, E4.NH2, andE5.NH2 separated from their mixture. Silanization of thecapillary minimizes but does noteliminate the adsorption ofPAMAM onto the capillary surface. Decreasing the numberof active sites results in faster stabilization (see Figs. 2a, b).It is believed that in both bare capillary and silanized capil-lary are equilibrated by unremoved PAMAM dendrimermolecules from previous runs. Larger peak areas for G5 arenot expected as the same w/w sample concentrationresults in the same numberof UV chromophores present. Inaddition, in both Figs. 2b and 3b, Gaussian-shaped elec-tropherograms can be observed because the electro-pherograms shown in Figs. 2b and 3b are recorded afterseveral runs. This means that the migration and separationof PAMAMs are stabilized in both cases (bare and silanizedcapillaries) when run by capillaries equilibrated by unre-moved PAMAM dendrimers on the capillary surface in pre-vious runs. However, we do observe the non-Gaussian CEpeaks at the first run in both cases (data not shown); thenon-Gaussian shape of CE peaks run by bare silica capil-lary is more prominent than those run by silanized capillary.

Migration times of dendrimer samples are well reproduc-ible on the same day RSD of the migration times is lessthan 4.0%, see Table 2), which was also confirmed by theanalysis of PAMAM derivatives. Peak shapes are regularand the peak positions are nearly equally spaced in theelectropherograms (Figs. 2b and 3b). However, we foundthat the day-to day reproducibility is inconsistent andpeak position values measured on different days were

Table 2. Charge/mass ratios and migration times ofPAMAM dendrimers and derivatives of genera-tion 4 and 5 analyzed using a silanized silicacapillary (ID 100 mm, total length = 68.5 cm andeffective length = 60 cm)

PAMAMdendrimers

E4.NH2 E4.NHAc E5.NH2 E5.NHAc E5.GlyOH

Ideal Mr 14215 16903 28826 34202 38810

Charge/massratio (C/g)a)

0.008864 0.003668 0.008811 0.003684 0.003246

Migrationtime tm (min)

9.79 10.35 10.72 13.44 13.07

RSD of tmb) 3.37% 0.25% 0.76% 1.71% 3.31%

a) Ideal charge/mass ratiosb) The comparison of migration times of primary amine,

acetyl, and hydroxyl-terminated PAMAM dendrimerswere run on the same day applying the same CE con-ditions. Each sample was analyzed three times andaveraged numbers are shown

incomparable. A similar tendency was observed withother PAMAM derivatives. Thus, day-to-day measure-ments still needed improvement.

3.3 Use of internal standards in the analysis ofindividual dendrimers

We selected and studied potential internal standards (ISs)based on the following requirements: (i) similar charge buta greater mobility, i.e., fair separation from PAMAM den-drimers, i.e., higher charge/mass ratio (derivatization ofPAMAM dendrimers usually increases the molecularmass and simultaneously decreases the global charge ofthe macromolecules, so it is advisable to select a com-pound with an electrophoretic mobility higher than thedendrimer samples.); (ii) easy detection between 200 and600 nm by UV-visible absorption; (iii) high stability underCE conditions; (iv) no chemical interaction with PAMAMdendrimers; (v) high purity; and (vi) easy availability. Con-ceptually, lower generation PAMAM dendrimers (i.e., G0and G1) could also be used as ISs for higher generationdendrimers due to their similar properties. However, CEelectropherograms of practical G0 and G1 dendrimersalways contain side products [23] that appear in the formof several small peaks scattered around the main peak(electropherograms are not shown), hence they are notapplicable for this purpose. We have tested various smallnitrogen-containing organic molecules with differentcharge/mass ratios and pKa values [45–47], such as 2,4-diaminopyrimidine (2,4-DAPM, pKa = 8.73), 2,3-DAP,(pKa = 6.48), 2,6-diaminopyridine (2,6-DAP, pKa = 6.48),and 4-dimethylaminopyridine (4-DMAP, pKa = 6.09). Thecharge/mass ratio order for several different ISs (atpH 2.5) are as follows: 2,4-DAPM . 2,3-DAP = 2,6-DAP. DMAP. Table 3

Table 3. Relative migration times and their correspond-ing SDs of PAMAM dendrimers using differentISs such as 2,3-DAP, 2,6-DAP, DMAP, 2,4-DAPM, and melamine

PAMAMdendrimers

Migration time ratios between dendrimers and internalstandards and their RSDs (n = 3)a)

E2.NH2 E2. NH2/2,3-DAP0.929 (0%)

E2. NH2/2,4-DAPM0.92 (1.54%)

–

E4.NH2 E4. NH2/2,3-DAP1.072 (2.02%)

E4. NH2/2,6-DAP1.055 (0.670%)

E4. NH2/DMAP1.027 (0.562%)

E5.NH2 E5. NH2/2,3-DAP1.107 (1.04%)

E5. NH2/melamine0.873 (2.88%)

E5. NH2/DMAP1.04 (0%)

a) Average of three day-to-day measurements using thesame capillary



lists the relative migration times of PAMAM dendrimersusing these various ISs. (Relative migration times of 2,3-DAP are very similar to those of 2,6-DAP, but its molar A isconsiderably higher at 210 nm therefore an internalstandard peak can easily be identified by comparing sig-nals detected at 250 and 210 nm.)

Since most of our dendrimer-related applications usegeneration 4 and generation 5 PAMAM dendrimers asbuilding blocks, in this study we focused on thesePAMAMs. The relative migration times (tD/tIS = trel, migra-tion time ratios) between various dendrimers and all IScandidates were essentially constant RSDs were lessthan 3%, see Table 3). Based on the above listedrequirements, 2,3-DAP was selected and used as an ISfor polycationic PAMAMs.

Figure 4 compares uncorrected (Fig. 4a) and normalized(Fig. 4b) electropherograms of E2.NH2, E4.NH2 andE5.NH2 PAMAMs in the presence of 2,3-DAP. It is worth tonote that the peak intensity of the IS is not fully consistentin the separation of different generation PAMAMs. This isprobably due to the partial overlap of IS peaks andPAMAM dendrimer peaks, which display relativelybroader peaks due to its polymeric nature. However,dendrimer peak identification is not influenced by theoccasional partial overlap. The normalized curves clearlyreflect the change in the mobility of the investigated den-drimers and the relative mobility of E5.NH2 is somewhatsmaller than that of E4.NH2 (Table 2). According to thecalculation based on Eq. (3) and the known migration timeof 2,3-DAP (tM = 11.80 min, the average of six differentruns), the electrophoretic mobilities of E4.NH2 andE5.NH2 were calculated as 3.63 6 1025 cm2/V?s and3.50 6 1025 cm2/V?s, respectively. Apparent electropho-retic mobilities of dendrimer samples calculated in thismanner are constant in different runs and dendrimersamples with varying generations or surface modifica-tions may be compared even though the data have beenacquired in different runs. E2.NH2 and E4.NH2 have aslightly higher apparent electrophoretic mobility thanE5.NH2 does (this trend is illustrated in Figs. 2b and 3b).The normalized electropherograms may be further ana-lyzed using a commercial software (e.g., PEAKFIT fromSYSTAT) to deconvolute the curves (see Fig. 5) into majorcomponents thus generating even more quantitative data(Table 4).

3.4 Evaluation of CE data

Protonated polycationic PAMAM dendrimers can hardlybe considered as simple ions as it is demonstrated in thecalculations summarized in the supplementary Table 3S.Three basic facts can be obtained from the CE experi-

Figure 4. Comparison of uncorrected (a) and normal-ized (b) electropherograms of E2.NH2, E4.NH2 andE5.NH2 PAMAM dendrimers in the presence of the 2,3-DAP internal standard using a silanized capillary(id: 100 mm, effective length 70 cm, and total length78.5 cm). Peak 1 indicates the internal standard 2,3-DAP.Injection time: 3 s.

mental data, which should be discussed in detail. The firstis that the PAMAM dendrimers have different migrationtime in CE experiments although their theoretical charge/mass ratios are very close to each other. The second isthat the migration time for higher generation dendrimersare longer as it is shown in the supplementary Table 3S.The third is that the adsorption of the PAMAM dendrimerson the capillary surface affects the measured migrationtime.

The mobility of PAMAM dendrimers was found to de-crease with increasing generation number. Although itmay appear to be in agreement with the decreasing ten-



Figure 5. CE traces of E4.NH2 (a) and E5.NH2 (b) decon-voluted by PEAKFIT. The dotted line shows the compo-nents found (see areas in Table 4).

dency of the formal charge/mass ratio, the difference (seeTable 1) is too small to explain a successful CE separa-tion. Mass and size of dendrimers are not in a linear cor-relation because density of the dendrimers increases withincreasing number of generations. Thus, although thecharge/mass ratio values (Table 1) are almost constantbased on both theoretical calculation and practicalmeasurements (from acid-base titration and GPC data,their actual charge/mass ratios are ranged from 0.0066 to0.0087), the formal charge/surface area ratio (supple-mentary Table 3S) increases with increasing number ofgeneration. Considering that the electrophoretic mobilitydepends rather on the charge/surface ratio than thecharge/mass ratio (for small ions the two values are di-

rectly proportional), the smaller mobility of larger den-drimers cannot be explained simply on electrophoreticgrounds. The first possible explanation is that anions maypartially shield the charge or “neutralize” the protonatedamine groups. In this case the H2PO4

2 ions could be themost probable partners. In order to approach the behav-ior of PAMAM dendrimers from this point of view, wemade similar calculations for the effective charge/surfacearea ratio from the electrophoretic mobility as Huang et al.[26] and Welch and Hoagland [32] made for other den-drimers. We assumed that a standard electrokineticmodel is valid for PAMAM dendrimers and we used thenormalized CE migration times (determined in silanizedcapillaries), from which data Table 3S was constructed(for details, see supplementary material).

The apparent charge/surface area ratios calculated fromthe experimental mobility data show an opposite trendand are much lower than the geometric surface charge.That means that behavior of PAMAM ions are more similarto the cascade polymers investigated by Huang et al. [26]than to the PPI dendrimers studied by Welch and Hoag-land [32] who found no interaction between the den-drimers and anions. It is very important to point out here,that the calculated apparent charges are valid only ifexclusively electrophoretic forces govern the motion ofthe dendrimer polyions in the capillary. This is clearly notthe case and consequently the standard electrokineticmodel is invalid here.

It is known that the silica surface strongly binds PAMAMdendrimers, which are positively charged at applied pH[33]. Long migration times are indicative of the shift of theadsorption/desorption equilibria on the capillary wall sur-face towards increased adsorption. The surface-boundPAMAMs provide electrostatic shielding for the silanolgroups on the silica, and decrease the migration times.Initially, the injected sample adheres strongly to the sur-face of capillary (Fig. 2) and it can be removed only veryslowly and sometimes incompletely [48]. In consecutiveinjections and runs, the migration time levels off due to theensuing dynamic equilibrium between the wall and thepolycationic dendrimer molecules. Then a stationarystate arises, and a new competitive binding develops withevery change, such as the injection of a new sample and/or change of solvent. Unfortunately, this mechanism notonly means that unknown part of the injected sample maygo undetected (due to the excessive broadening of thepeaks and the stronger retention of the higher generationdendrimers), but the same phenomenon results in dyna-mically changing separation conditions on the wall, andthe position of every newly injected PAMAM peak willdepend both on the history of the sample and on the his-tory of the column. Injecting the same samples repeatedly



Table 4. Comparison of standardized and deconvoluted CE and PAGE results for E4.NH2 andE5.NH2 dendrimer

PAMAM Peak PAMAMcomponents

Normalizedmigrationtime (min)

Relativemigrationtime

Area%by CE

Area%by PAGE

E4.NH2 2 E3.NH2 11.42 1.021 4.14 5.43 E4.NH2 11.75 1.052 90.02 90.64 (E4NH2)2 12.41 1.110 5.84 4.0

E5.NH2 2 E4.NH2 12.67 1.076 6.53 13.43 E5.NH2 13.02 1.105 76.67 74.94 (E5.NH2)2 13.61 1.155 16.80 11.7

will create this stationary state and a virtual equilibriumarises, although injecting different samples leads to vary-ing peak positions. Consequently, incomplete capillarypreconditioning and post-conditioning, which changesthe adsorption and desorption of dendrimers in differentruns, also may contribute to the irreproducibility. Silani-zation of the capillary internal surface reduces the numberof binding centers therefore decreases the absorption ofPAMAM dendrimers [38]. This results in a much betterreproducibility of the measurement, while essentially thesame mechanism applies.

We conclude that although the movement of ions is gen-erated by the electric field and is due to the charge/massratio or rather the effective charge/surface area ratio, butthe separation is due to the adsorption/desorption ofdendrimers on the capillary surface. Absorption is alsosuggested by the increase of the width/height ratio and bythe increasingly nonlinear shape of the peaks, which is atypical LC feature as opposed to the Gaussian CZE peak-shapes (as shown in Fig. 2b). Although adsorption is thedominant factor that significantly changes the migrationtime between runs, there are also secondary factors toconsider: e.g., smaller generation PAMAM dendrimersmay have a smaller contact area but higher flexibility topromote adsorption onto the capillary surface, whilehigher generations have a larger contact area, but theyare more rigid [48].

To summarize, the degree of separation and the order ofCE peaks of polycationic PAMAM dendrimers can be theconsequence of both the adsorption and the partial neu-tralization of the dendrimers resulting in an opposite trendin the apparent charge/surface area ratio and decreasingapparent mobility of PAMAM dendrimers under CE con-ditions. The two phenomena take place simultaneouslyand their effect cannot be separated. In contrast with theseparation mechanism of polycationic PAMAM den-drimers, we have also investigated the CE separation ofpolyanionic carboxyl-terminated PAMAM succinamic

acid of different generations at pH 8.3, which have noelectrostatic interaction with the uncoated silica capillarysurface (see subsequent manuscript ‘Capillary electro-phoresis of polycationic poly(amidoamine) dendrimers’).All generations show almost identical electrophoreticmobilities. The results of polyanionic PAMAM separationby CE will be reported separately.

4 Concluding remarks

A chromatographic method was developed for the analy-sis and characterization of polycationic PAMAM den-drimers of different generations and termini using CE. Themethod is demonstrated on PAMAM dendrimers betweengenerations 2 to 5 using either an uncoated bare fused-silica capillary or a silanized capillary. Separation andcharacterization is more reproducible on a silanized cap-illary than on uncoated capillary. To improve reproducibil-ity, various small aromatic molecules with differentcharge/mass ratios were selected and tested. Based onthese test results, 2,3-DAP was chosen as an internalstandard for polycationic PAMAMs analysis. Both mix-tures of various PAMAM generations and PAMAM den-drimers of the same generation with different terminalgroups were analyzed. Ratios of migration times tostandards proved to be very reproducible. Various com-parisons were made using normalized migration times(i.e., extrapolating to normal conditions). Apparent sur-face charges were also calculated based on theoreticalconsiderations and were compared to literature data. Ourmeasurements show that although the electric fieldmoves the polycationic dendrimers in the capillary, theretention differences are dominantly due to the interac-tions of the PAMAM dendrimers with the capillary wall. Itis concluded that for polycationic PAMAMs a complexseparation mechanism has to be considered, which is incharacter similar to electrokinetic capillary chromatogra-phy.



We thank Dr. Anil K. Patri for his kind assistance withPAGE. This work is financially supported in part by theEnvironmental Protection Agency, EPA NanotechnologyAward R829626 as part of the STAR program and by theNational Cancer Institute (NCI) of the National Institute ofHealth (NIH), under the contract # NOI-CO-97111. TheHungarian Science Foundation (OTKA T-035127) is alsoacknowledged for partial support of IB.

Received February 16, 2005

5 References

[1] Tomalia, D. A., Baker, H., Dewald, J. A., Hall, M., Kallos, G.,Martin, S., Roeck, J., Ryder, J., Smith, P., Macromolecules1986, 19, 2466.

[2] Tomalia, D. A., Baker, H., Dewald, J. R., Hall, M., Kallos, G.,Martin, S., Roeck, J., Ryder, J., Smith, P., Polym. J. 1985, 17,117.

[3] Newkome, G. R., Yao, Z. Q., Baker, G. R., Gupta, K., J. Org.Chem. 1985, 50, 2003.

[4] Buhleier, E., Wehner, W., Vogtle, F., Synthesis 1978, 155–158.

[5] Newkome, G. R., Moorefield, C. N., Vögtle, F., DendriticMacromolecules: Concepts, Syntheses, Perspectives,Wiley-VCH, New York 1997.

[6] Tomalia, D. A., Fréchet, J. M. J., Dendrimers and OtherDendritic Polymers, John Wiley & Sons Ltd, New York 2001.

[7] Rutt, J. S., Dendrimers and Nanotechnology: A PatentExplosion; National Nanotechnology Initiative Conference,Washington D.C., April 29, 2002.

[8] Van Manen, H. J., van Veggel, F. C. J. M., Reinhoudt, D. N.,in: Vögtle, F., Schalley, C. A. (Eds.), Dendrimers IV: MetalCoordination, Self-assembly, Catalysis, Springer-Verlag,Berlin 2001, Vol. 217, pp. 163–199.

[9] Esfand, R., Tomalia, D. A., Drug Discovery Today 2001, 6,427–436.

[10] Patri, A. K., Majoros, I. J., Baker J. r., J. R., Curr. Opin. Chem.Biol. 2002, 6, 466–471.

[11] Kim, C., Park, E., Song, C. K., Koo, B. W., Synth. Met. 2001,123, 493–496.

[12] Koo, B. W., Song, C. K., Kim, C., Sens. Actuators B 2001, 77,432–436.

[13] Miller, L. L., Kunugi, Y., Canavesi, A., Rigaut, S., Moorefield,C. N., Newkome, G. R., Chem. Mater. 1998, 10, 1751.

[14] Busson, P., Ortegren, J., Ihre, H., Gedde, U. W., Hult, A.,Andersson, G., Eriksson, A., Lindgren, M., Macromolecules2002, 35, 1663–1671.

[15] Ma, H., Jen, A. K. Y., Adv. Mater. 2001, 13, 1201.

[16] Zhou, L., Russell, D. H., Zhao, M., Crooks, R. M., Macro-molecules 2001, 34, 3567–3573.

[17] Schutz, B. L., Rockwood, A. L., Smith, R. D., Tomalia, D. A.,Spindler, R., Rapid Commun. Mass Spectrom. 1995, 9,1552.

[18] Kallos, G. J., Tomalia, D. A., Hedstrand, D. M., Lewis, S.,Zhou, J., Rapid Commun. Mass Spectrom. 1991, 5, 383.

[19] Mitnik, L., Novotny, M., Felten, C., Buonocore, S., Koutny,L., Schmalzing, D., Electrophoresis 2001, 22, 4104–4117.

[20] Patrick, J. S., Lagu, A. L., Electrophoresis 2001, 22, 4179–4196.

[21] Righetti, P. G., Biopharmaceutics & Drug Disposition 2001,22, 337–351.

[22] Brothers II, H. M., Piehler, L. T., Tomalia, D. A., J. Chroma-togr. A 1998, 814, 233–246.

[23] Ebber, A., Vaher, M., Peterson, J., Lopp, M., J. Chromato-gr. A 2002, 949, 351–358.

[24] (a) Balogh, L., Leuze-Jallouli, A., Dvornic, P., Kunugi, Y.,Blumstein, A., Tomalia, D. A., Macromolecules 1999, 32,1036–1042. (b) Tomalia, D. A., Huang, B., Swanson, D. R.,Brothers, H. M., Klimash, J. W., Tetrahedron 2003, 59, 3799–3813.

[25] (a) Pesak, D. J., Moore, J. S., Wheat, T. E., Macromolecules1997, 30, 6467–6482. (b) Castagnola, M., Zuppi, C., Ros-setti, D. V., Vincenzoni, F., Lupi, A., Vitali, A., Meucci, E.,Messana, I., Electrophoresis 2002, 23, 1769–1778.

[26] Huang, Q. R., Dubin, P. L., Moorefield, C. N., Newkome, G.R., J. Phys. Chem. 2000, 104, 898–904.

[27] Hückel, E., Phys. Z. 1924, 25, 24.

[28] Smoluchowski, M., Z. Phys. Chem. 1918, 92, 129.

[29] Loeb, A. L., Overbeek, J. Th. G., Wiersema, P. H., The Elec-trical Double Layer Around a Spherical Colloid Particle, MITPress, Cambridge, MA 1961.

[30] Wiersema, P. H., Loeb, A. L., Overbeek, J. Th. G., J. ColloidInterface Sci. 1966, 22, 78–99.

[31] Manning, G. S., J. Chem. Phys. 1969, 51, 924.

[32] Welch, C. F., Hoagland, D. A., Langmuir 2003, 19, 1082.

[33] (a) Esumi, K., Goino, M., Langmuir 1998, 14, 4466–4470. (b)Ottaviani, M. F., Turro, N. J., Jockusch, S., Tomalia, D. A., J.Phys. Chem. 2003, 107, 2046–2053.

[34] Tsukruk, V. V., Rinderspacher, F., Bliznyuk, V. N., Langmuir1997, 13, 2171–2176.

[35] Baker, L. A., Zamborini, F. P., Sun, L., Crooks, R. M., Anal.Chem. 1999, 71, 4403–4406.

[36] (a) Tomalia, D. A., Naylor, A. M., Goddard III, W. A., Angew.Chem. Int. Ed. Engl. 1990, 29, 138. (b) Cakara, D., Kleimann,J., Borkovec, M., Macromolecules 2003, 36, 4201–4207.

[37] Weinberger, R., Am. Lab. 2002, 34, 32–40.

[38] Betley, T. A., Holl, M. M. B., Orr, B. G., Swanson, D. R.,Tomalia, D. A., Baker Jr., J. R., Langmuir 2001, 17, 2768–2773.

[39] Jung, H. J., Bae, Y. C., J. Polym. Sci. Polym. Chem. 2002,40, 1405–1420.

[40] Bruggemann, O., Freitag, R., Whitcombe, M. J., Vulfson, E.N., J. Chromatogr. A 1997, 781, 43–53.

[41] Sharma, A., Mohanty, D. K., Desai, A., Ali, R., Electrophore-sis 2003, 24, 2733–2739.

[42] Li, S. F. Y., Capillary Electrophoresis Theory & Practice.Academic Press, San Diego, CA 1992.

[43] Linhardt, R. J., Toida, T., Science 2002, 298, 1441–1442.

[44] Majoros, I. J., Keszler, B., Woehler, S., Bull, T., Baker Jr., J.R., Macromolecules 2003, 36, 5526–5529.

[45] Mazik, A., Zielinski, W., Monatsh. Chem. 1996, 127, 587–591.

[46] Lide, D. R., Handbook of Chemistry and Physics 80th ed.,CRC Press, New York 1999–2000.

[47] Dean, J. A., Lange’s Handbook of Chemistry 13th ed.,McGraw-Hill Book Company, New York 1985.

[48] Rahman, K. M. A., Durning, C. J., Turro, N. J., Tomalia, D. A.,Langmuir 2000, 16, 10154–10160.


capillary electrophoresis of polycationic poly(amidoamine) dendrimers

Documents