PAPER

www.rsc.org/analyst | The Analyst

Molecular heterogeneity analysis of poly(amidoamine) dendrimer-based mono- and multifunctional nanodevices by capillary electrophoresis{
Xiangyang Shi, Istvan J. Majoros, Anil K. Patri, Xiangdong Bi, Mohammad T. Islam, Ankur Desai, T. Rose Ganser and James R. Baker, Jr.*
Received 7th November 2005, Accepted 21st December 2005 First published as an Advance Article on the web 18th January 2006 DOI: 10.1039/b515624f Poly(amidoamine) (PAMAM) dendrimer-based nanodevices are of recent interest in targeted cancer therapy. Characterization of mono- and multifunctional PAMAM-based nanodevices remains a great challenge because of their molecular complexity. In this work, various mono- and multifunctional nanodevices based on PAMAM G5 (generation 5) dendrimer were characterized by UV-Vis spectrometry, 1H NMR, size exclusion chromatography (SEC), and capillary electrophoresis (CE). CE was extensively utilized to measure the molecular heterogeneity of these PAMAM-based nanodevices. G5FA (FA denotes folic acid) conjugates (synthesized from amine-terminated G5.NH2 dendrimer, approach 1) with acetamide and amine termini exhibit bimodal or multi-modal distributions. In contrast, G5FA and bifunctional G5FAMTX (MTX denotes methotrexate) conjugates with hydroxyl termini display a single modal distribution. Multifunctional G5.AcnFIFA, G5.AcnFAOHMTX, and G5.AcnFIFAOHMTX (Ac denotes acetamide; FI denotes fluorescein) nanodevices (synthesized from partially acetylated G5 dendrimer, approach 2) exhibit a monodisperse distribution. It indicates that the molecular distribution of PAMAM conjugates largely depends on the homogeneity of starting materials, the synthetic approaches, and the final functionalization steps. Hydroxylation functionalization of dendrimers masks the dispersity of the final PAMAM nanodevices in both synthetic approaches. The applied CE analysis of mono- and multifunctional PAMAM-based nanodevices provides a powerful tool to evaluate the molecular heterogeneity of complex dendrimer conjugate nanodevices for targeted cancer therapeutics.

Introduction
Poly(amidoamine) (PAMAM) dendrimers are a novel class of advanced synthetic nanoparticles, possessing a core unit, branches, and multiple terminal groups.1,2 They are synthesized through a stepwise repetitive reaction sequence starting at the core molecules and giving rise to increasing generations, with increasing molecular weights and sizes. The interior cavities of PAMAMs are often used to encapsulate hydrophobic or hydrophilic drugs.3,4 In order to achieve better control of the release rate, imaging, and targeting properties of drugs, the terminal groups of PAMAMs have been functionalized with various moieties, such as drugs, biospecific ligands, and fluorescent tags.512 These dendrimer-based nanodevices hold great promise for various biomedical applications, especially for targeted cancer therapy. For instance, Baker and coworkers6,8 have developed a novel PAMAM dendrimer-based cancer therapeutic agent carrying biospecific targeting moieties (e.g. folic acid, denoted as FA), fluorescent dyes, as well as drugs. These dendrimer
Michigan Nanotechnology Institute for Medicine and Biological Sciences, University of Michigan, Ann Arbor, MI 48109, USA. E-mail: jbakerjr@umich.edu { Electronic supplementary information (ESI) available: A deconvoluted normalized electropherogram of G5.NH2FA conjugates and UV-Vis spectrum analysis of the CE peaks in the electropherogram. See DOI: 10.1039/b515624f

nanodevices are demonstrated to be able to effectively transverse vascular pores and directly infuse tumor cells. Wiener et al. have also developed dendrimerFA conjugates to target tumor cells expressing high-affinity FA receptors, and demonstrated that FA-conjugated magnetic resonance imaging contrast agents represent a promising new approach for tumor targeting and imaging.13,14 Although the research on medical applications of dendrimer-based nanodevices continues progressively, the characterization of these multifunctional dendrimer nanodevices still remains a great challenge. The comprehensive understanding and analytical characterization of the structural heterogeneity of these novel synthetic nanodevices have not been widely investigated. PAMAM dendrimers exhibit much narrower polydispersity than general synthetic polymers. The polydispersity of PAMAM dendrimers can be summarized as generational dispersity (i.e. trailing generations and dimers), skeletal dispersity (i.e. missing arms and intramolecular loops), and substitutional dispersity (differences between molecules due to the non-uniform surface functionalization).15 The former two dispersities are usually generated by the divergent synthetic technology,16,17 while the substitutional dispersity is always present in surface-functionalized PAMAM derivatives.15 For multifunctional dendrimer-based nanodevices, substitutional dispersity is often expected due to the multi-step surface modifications with different functional moieties.
This journal is The Royal Society of Chemistry 2006

374 | Analyst, 2006, 131, 374381

Various analytical techniques including, but not limited to, mass spectrometry (i.e. MALDI-TOF),18 NMR techniques,19 polyacrylamide gel electrophoresis,2024 UV-Vis and fluorescence spectroscopy,2528 high performance liquid chromatography (HPLC),2931 size exclusion chromatography (SEC),15,32,33 and capillary electrophoresis (CE),15,23,24,3337 have been used to analyze PAMAM dendrimers and derivatives. As opposed to other techniques, CE allows simultaneous evaluation of both purity and charge distribution of the analyzed materials. Previous studies15,23,24,33 have shown that CE can be effectively used to estimate molecular distribution of various functionalized PAMAM dendrimer derivatives. In this present study, we have synthesized a variety of generation 5 (G5) PAMAM dendrimer-based functional nanodevices using two synthetic approaches (See Scheme 1). Approach 1: synthesis from amine-terminated G5.NH2 dendrimers; the synthesized materials include monofunctional G5FA conjugates (terminated with amine, acetamide, and hydroxyl groups), and bifunctional G5.FAOHMTX (MTX denotes methotrexate) conjugates terminated with hydroxyl groups. Approach 2: synthesis from partially acetylated G5 dendrimers; the synthesized materials include a set of monoand multifunctional G5.AcnFA, G5.AcnFIFA, G5.Acn FAOHMTX, and G5.AcnFIFAOHMTX nanodevices (FI denotes fluorescein). SEC was used to evaluate the molecular weights of these dendrimer nanodevices, while UV-Vis spectrometry was employed to detect the functional moieties upon dendrimer surface modification. NMR techniques were also utilized to confirm the dendrimer surface

functionalization and estimate the average number of surface functional moieties. CE was extensively used to evaluate the molecular distribution of these dendrimer-based nanodevices, since the charge distribution and electrophoretic mobility often change upon dendrimer surface conjugation. The applied CE analysis provides a unique way to evaluate the molecular distribution and heterogeneity of the dendrimer-based nanodevices for targeted cancer therapy. To our best knowledge, this is the first extensive research report regarding the novel application of CE for the analysis of complex dendrimer-based mono- and multifunctional nanodevices.

Experimental
Materials Ethylenediamine (EDA)-cored PAMAM G5.NH2 dendrimer was synthesized in the Michigan Nanotechnology Institute for Medicine and Biological Sciences, University of Michigan (Ann Arbor, MI) and was completely characterized by various techniques.9,38 The number of terminal amine groups was determined to be 110. The molecular weight and polydispersity data are listed in Table 1. Acetic anhydride, glycidol, MTX, absolute methanol, triethylamine, dimethyl sulfoxide (DMSO), FA, 1-(3-dimethylaminopropyl)-3-ethylcarbodiimide (EDC), and all the other chemicals and solvents were obtained from Aldrich (Milwaukee, WI) and used as received. Phosphate buffer (pH = 2.5, 50 mM) and water used for CE analysis were obtained from Agilent Technologies (Waldbronn, Germany) and used as received. Water used in all the other experiments

Scheme 1 Schematic representation of the synthetic approaches used to synthesize mono- and multifunctional dendrimer nanodevices.

This journal is The Royal Society of Chemistry 2006

Analyst, 2006, 131, 374381 | 375

Table 1

Physicochemical parameters of dendrimer and dendrimer conjugates Conjugate Mn Mw/Mn Number of functional moieties

Dendrimer

G5.NH2 26 330 1.032 N/A 1 28 090 1.084 3.8(FA)a, 5.0(FA)b G5.NH2FA G5.FAAc 2 32 870 1.172 3.8(FA)a, 5.0(FA)b G5.FAOH 3 40 210 1.108 3.8(FA)a, 5.0(FA)b G5.FAOHMTX 4 44 270 1.120 3.8(FA)a,5.0(FA)b, 9.3(MTX)a,10.0 (MTX)b G5.Ac75 29 900 1.021 N/A 5 31 400 1.027 5.5 (FA)b, 5.5 (FA)c G5.Ac75FA G5.Ac75FIFA 6 34 100 1.003 4.7 (FI)c, 5.5 (FA)b, 5.5 (FA)c G5.Ac82FAOHMTX 7 36 730 1.006 5.7 (FA)a, 4.5 (FA)b, 4.8 (FA)c, 5.5 (MTX)a G5Ac82FIFAOHMTX 8 39 550 1.008 5.8 (FI)a, 5.7 (FA)a,4.5 (FA)b, 4.8 (FA)c, 5.5 (MTX)a a Based on the differences between the measured Mn by SEC, which is divided by the molecular mass of the incoming substituent. b Based on the differences between the integrals of 1H NMR signals associated with dendrimers and the surface moieties. c Based on UV-Vis spectroscopy analysis using the concentration calibration curve of free FA.

was purified by a Milli-Q Plus 185 water purification system (Millipore, Bedford, MA) with resistivity higher than 18 MV cm. Regenerated cellulose dialysis membranes (MWCO = 10,000) were acquired from Fisher (Fairlawn, NJ). Synthesis of G5FA conjugates (terminated with amine, acetamide, and hydroxyl groups) and G5.FAOHMTX conjugates terminated with hydroxyl groups using approach 1 The amine-terminated G5.NH2 dendrimer was conjugated with FA using our published procedures (conjugate 1).6 Based on the 1H NMR integration values and the comparison between the aromatic signals of FA and aliphatic region of dendrimer methylene protons, we estimated that an average of 5.0 FA moiety molecules were present per dendrimer. The remaining amines on the dendrimer surface were modified with acetic anhydride to give a complete dendrimer acetamide functionalized with the FA moiety (conjugate 2). The 1H NMR of this acetylated dendrimer shows merging methylene protons and an additional singlet at 1.8 ppm for the terminal COCH3 groups.32 Alternatively, the amine terminated G5FA conjugate was further hydroxylated with glycidol (conjugate 3), followed by esterification with methotrexate using EDC coupling chemistry (conjugate 4).9 All physicochemical parameters of the above dendrimer conjugates are shown in Table 1. Synthesis of G5.AcnFA, G5.AcnFIFA, G5.AcnFAOH MTX, and G5.AcnFIFAOHMTX conjugates using approach 2 This set of G5 dendrimer conjugates were synthesized from partially acetylated G5 dendrimers. A list of the synthesized conjugates (conjugate 5, 6, 7, 8) is shown in Table 1. Details can be found in a previous report.9 The numbers of the conjugated FA, FI, and MTX were determined using 1H NMR, UV-Vis spectrometry, and SEC (See Table 1). Characterization techniques SEC analysis. SEC was used to determine the absolute molecular weights of the as-prepared PAMAM dendrimer conjugates. SEC experiments were performed using an Alliance Waters 2690 separation module (Waters Corporation, Milford, MA) equipped with a Waters 2487 UV absorbance
376 | Analyst, 2006, 131, 374381

detector (Waters Corp.), a Wyatt Dawn DSP laser photometer (Wyatt Technology Corporation, Santa Barbara, CA), and an Optilab DSP interferometric refractometer (Wyatt Technology Corporation). Citric acid buffer (0.1 M) with 0.025% sodium azide in water was used as a mobile phase. The pH of the mobile phase was adjusted to 2.74 using NaOH, and the flow rate was maintained at 1 mL min21. Sample concentration was kept at approximately 2 mg mL21 and 100 mL of analytes was injected for all measurements. Molar mass moments of the PAMAM dendrimers were determined using Astra software (version 4.7) (Wyatt Technology Corp.). UV-Vis spectrometry. UV-Vis spectra of dendrimer conjugates were collected using a Perkin Elmer Lambda 20 UVVis spectrometer. All the dendrimer conjugates were dissolved in pH 2.5 phosphate buffer (50 mM) and their pHs were adjusted to 2.5 using 0.1 M phosphoric acid. The final concentrations of the dendrimer conjugates were 1 mg mL21. NMR measurements. 1H NMR spectra of PAMAM dendrimer conjugates were recorded on a Bruker DRX 500 nuclear magnetic resonance spectrometer. Dendrimer samples were dissolved in D2O at a concentration of 5 mg mL21 before NMR measurements. CE analysis. An Agilent Technologies (Waldbronn, Germany) CE instrument was used in this work. Unmodified quartz capillaries were purchased from Polymicro Technologies (Phoenix, AZ). Voltage was kept at 20 kV for all the separations. On-capillary UV diode-array detection was used, operating at four different wavelengths depending on the functional moieties conjugated on dendrimer surfaces. Samples were introduced by hydrodynamic injection at a pressure of 50 mbar. Silanized capillaries24 (id 100 mm) with total length of 78.5 cm and effective length of 70 cm were employed. Some of the experiments were performed using a short capillary (id 100 mm) with total length of 48.5 cm and effective length of 40 cm. The capillary temperature was maintained at 40 uC. Before use, the silanized capillary was initialized by rinsing with 0.2 M H3PO4 for 15 min, deionized water for 15 min, and the running buffer (50 mM, pH 2.5 phosphate buffer) for another 15 min. Prior to each injection, the capillary was rinsed with a sequence of 0.2 M H3PO4 (3 min), deionized water (3 min), and the running buffer (10 min). G5 PAMAM
This journal is The Royal Society of Chemistry 2006

conjugates were dissolved in the running buffer and the sample solutions were adjusted to pH 2.5 using 0.1 M phosphoric acid to give a final concentration of 1 mg mL21. All dendrimer conjugate samples contained 0.05 mg mL21 2,3-diaminopyridine (2,3-DAP) as an internal standard to normalize the migration time of analytes.24 Each sample was run 5 times to confirm the reproducibility. The CE electropherograms were deconvoluted using PEAKFIT program (SYSTAT Software Inc., Point Richmond, CA 94804) to perform compositional analysis of the dendrimer conjugate species.

Results
Size exclusion chromatography SEC allows for the detection of the absolute molecular weights of the as-synthesized dendrimer conjugates. The polydispersities of all dendrimer conjugates are very close to 1.000 (See Table 1), indicating the narrowly dispersed nature of PAMAM dendrimers. Based on the molar mass difference between G5 dendrimer conjugates and the starting materials, one can calculate the number of functionalized groups or moieties. For instance, the number of MTX moiety molecules attached on conjugate 4 was calculated to be 9.3 based on its molar mass difference from conjugate 3, which is in agreement with that calculated based on NMR integration. The molar masses of G5FA conjugates are in the order of G5.FAOH (conjugate 3) > G5.FAAc (conjugate 2) > G5.NH2FA (conjugate 1). We also found that an incomplete hydroxylation reaction took place when comparing the molar mass difference between 1 and 3. Based on the calculation, the number of terminal glycidol moiety of 3 was 164, which was less than 210 according to the ideal stoichiometry. Namely, there are 46 secondary amines remaining intact under the experimental conditions. This observation is consistent with our previous results showing that incomplete hydroxylation always appears regardless of the number of dendrimer generations.15,33 The incomplete hydroxylation may contribute to the structural heterogeneity of the dendrimer conjugates (vide infra). The same SEC measurements were carried out for a set of monoand multifunctional G5 dendrimer nanodevices synthesized using approach 2 (conjugates 5, 6, 7, 8, see Table 1). From SEC results, G5 dendrimer nanodevices synthesized using both approach 1 and approach 2 exhibited similar polydispersity, mainly due to the limitation of SEC technique that functions based on the sizes of the macromolecules. UV-Vis spectrometry UV-Vis spectrometry was used to detect the functional moieties attached on dendrimer molecules. Fig. 1 shows the UV-Vis spectra of G5FA conjugates terminated with amine (1), acetamide (2), and hydroxyl (3) groups and a bifunctional G5.FAOHMTX conjugate (4). It is clear that curves 1, 2, and 3 exhibit the specific maximum absorption peak at 287 nm, which is ascribed to the contribution of the FA moiety. This suggests the successful conjugation of FA with dendrimers (UV-Vis spectrum of free FA not shown). The UV-Vis absorption profiles of the G5FA conjugates are almost similar regardless of the type of the terminal groups.
This journal is The Royal Society of Chemistry 2006
Fig. 1 UV-Vis spectra of the G5FA conjugates with different terminal groups synthesized using approach 1. Curves 1, 2, 3, and 4 represent the spectra of G5.NH2FA (1), G5.FAAc (2), G5.FAOH (3), and G5.FAOHMTX (4), respectively.

For the bifunctional G5.FAOHMTX nanodevice, due to the structural similarity of FA and MTX,39 their absorptions overlap. Therefore, it is difficult to differentiate the maximum absorption of MTX and FA moieties. The maximum absorption at 303 nm may result from the contribution of both FA and MTX moieties. The shoulder present at 243 nm is virtually related to the contribution of the MTX moiety as compared with the UV-Vis spectrum of free MTX (not shown). The UV-Vis spectra of dendrimer conjugates demonstrates the successful conjugation of the functional moieties. The absorption features also direct the optimal selection of the oncapillary UV detection wavelengths for CE measurements in order to achieve maximum sensitivity. The same UV-Vis spectroscopy studies were performed on the mono- and multifunctional G5 dendrimer nanodevices synthesized using approach 2. Details can be found in the previous literature.9 Overall, by using UV-Vis spectrometry, one can confirm the successful conjugation of imaging, targeting, and therapeutic agents onto the G5 dendrimer surfaces. Capillary electrophoresis Analysis of G5FA conjugates synthesized using approach 1. For the analysis of G5FA conjugates, four different UV wavelengths (200 nm, 210 nm, 250 nm, and 280 nm) were selected based on the UV-Vis spectroscopic studies. The selection of the 280 nm wavelength is expected to increase the detection sensitivity of the FA moiety. Fig. 2 shows the normalized electropherograms of G5FA conjugates terminated with amino (1), acetamide (2), and hydroxyl (3) groups. It is apparent that conjugate 3 exhibits a relatively homogeneous migration peak, while 1 and 2 display at least bimodal distribution. Estimation of the FA moiety distribution in dendrimer conjugates can be accomplished by performing offline UV-Vis spectrum analysis of individual migration peaks using Agilent software. Inset of Fig. 3a shows the UV-Vis spectrum assignments of the corresponding migration peaks shown in Fig. 3a and b. A dendrimer material associated with peak 2 displays much higher absorption at 280 nm (characteristic maximum absorption of the FA moiety) than that corresponding to peak 1, indicating that peak 2-related
Analyst, 2006, 131, 374381 | 377

Table 2 Compositional estimation of the studied dendrimer conjugates based on the deconvolution of CE electropherograms at 210 nm using PEAKFIT program Dendrimer Peak # Migration time/min Peak area Percentage (%) G5.NH2FA (1) 1 17.12 48.2 55.8 2 17.88 20.6 23.8 3 20.51 17.6 20.4 G5.FAAc (2) 1 19.05 61.2 68.6 2 22.60 18.0 20.2 3 24.50 10.0 11.2

Fig. 2 Normalized electropherograms of G5FA conjugates terminated with amino (1, curve 1), acetamide (2, curve 2), and hydroxyl (3, curve 3) groups detected at 210 nm. The first peak is related to the internal standard 2,3-DAP, which exhibit an average migration time of 12.14 min (7 runs). The CE was run using a silanized silica capillary (id 100 mm) with effective length of 70 cm and total length of 78.5 cm.

material possesses more FA moiety molecules than that related to peak 1. This phenomenon can also be easily differentiated by comparing the absorbances of peak 1 and peak 2 at different detection wavelengths. At 210 nm, peak 1-related dendrimer conjugate displays higher absorbance than that related to peak 2, while at 280 nm this is reversed. The deconvoluted electropherograms of G5.FAAc (2) at 210 nm (Fig. 3a) and 280 nm (Fig. 3b) are also shown, which were used for quantitative analysis of peak compositions (Table 2). The same analysis was applied to G5.NH2FA conjugate (1). It appears that peak 1-related dendrimer conjugate possesses more FA moiety molecules than peak 2-associated dendrimer material (See supporting information{). Analysis of G5FAMTX conjugate (4) synthesized using approach 1. G5.FAOHMTX bifunctional nanodevice (4) was also analyzed using CE. To determine different moieties attached onto dendrimer surfaces, the suitable wavelengths (210 nm, 240 nm, 300 nm, 350 nm) of on-capillary UV-Vis diode array detector were selected. The selection was based on the UV-Vis spectrum of 4 collected in the CE running buffer (pH 2.5 phosphate buffer, 50 mM) (Fig. 1). 210 nm was used to detect the dendrimer aliphatic backbone, while 240 nm and

Fig. 4 A typical electropherogram of G5.FAOHMTX conjugate (4) detected at 210 nm using a silanized silica capillary (id 100 mm) with effective length of 40 cm and total length of 48.5 cm. Peaks 1 and 2 correspond to the internal standard 2,3-DAP and conjugate 4, respectively. An off-line UV-Vis spectrum of peak 2 is also shown in the inset.

300 nm were used for MTX and FA moiety analysis, respectively. Fig. 4 shows a typical CE electropherogram of 4. One can clearly observe that conjugate 4 exhibits a monomodal symmetrical peak. The material was confirmed to be pure, as no impurity peaks were present in the electropherogram even at a long migration time up to 50 min. To determine the real composition of the conjugate, an off-line UV-Vis spectrometry analysis was performed using the Agilent software (Fig. 4, inset). The UV-Vis spectrum of peak 2-related conjugate species (Fig. 4, inset) is consistent with the UV-Vis spectrum of conjugate 4 measured by UV-Vis spectrometer (See Fig. 1), further indicating that peak 2 corresponds to the studied bifunctional G5 conjugate.

Fig. 3 Deconvoluted normalized electropherograms of G5.FAAc conjugate (2) at 210 nm (a) and 280 nm (b). The off-line UV-Vis spectra of individual migration peaks are shown in the inset of (a).

378 | Analyst, 2006, 131, 374381

This journal is The Royal Society of Chemistry 2006

Fig. 5 CE electropherograms of G5 dendrimer conjugates synthesized using approach 2 detected at 210 nm. The first peak is related to the internal standard 2,3-DAP. The CE was run using a silanized silica capillary (id 100 mm) with effective length of 40 cm and total length of 48.5 cm. Curve 1: G5.Ac75; curve 2: G5.Ac75FA (5); curve 3: G5.Ac75FIFA (6); curve 4: G5.Ac82FAOHMTX (7); curve 5: G5Ac82FIFAOHMTX (8).

Analysis of mono- and multifunctional G5 conjugates synthesized using approach 2. In another experiment, a set of mono- and multifunctional G5 nanodevices were analyzed using CE. These materials were synthesized from partially acetylated G5 dendrimer. Fig. 5 shows the CE electropherograms of G5.Ac75, G5.Ac75FA (5), G5.Ac75FIFA (6), G5.Ac82FAOHMTX (7), and G5Ac82FIFAOHMTX (8) conjugates. It is clear that all dendrimer conjugates display a single peak, indicating their relatively good monodispersity. Conjugate 6 shows a longer migration time than both G5.Ac75 and conjugate 5, which is probably due to its lower charge/ mass ratio and structural complexity after conjugation with the FI moiety. G5.Ac82FAOHMTX (7) and G5Ac82FI FAOHMTX (8) conjugates were synthesized from a partially acetylated G5.Ac82 dendrimer. After conjugation with FA and/or FI, the intermediate products were hydroxylated by reacting with glycidol, followed by conjugation with MTX to give the final MTX-conjugated nanodevices. Both conjugates (7 and 8) show narrower peaks than those of G5.Ac75, 5, and 6 (please note that the comparison is reasonable because the difference between G5.Ac75 and

G5.Ac82 is small enough and can be neglected). The narrow peaks of 7 and 8 might result from the hydroxylation reaction, which has been shown to mask the polydispersity of dendrimers in our previous work.15,33 In a sense of molecular heterogeneity, hydroxylation actually improves the monodispersity of the multifunctional dendrimer nanodevices. The CE system holds a diode array UV detector and an offline UV-Vis spectral analysis feature, which allows identifying the individual CE peaks. By comparing the UV-Vis spectra of analyzed species with those recorded using conventional UVVis spectrometer, one can identify the composition of each CE peak related species. Fig. 6a shows the CE electropherograms of a tri-functional PAMAM nanodevice (8) detected at 210, 300, and 495 nm. The material displays a single peak at all wavelengths indicating that FI, FA, and MTX moieties are homogeneously present. A magnified electropherogram (Fig. 6b) shows the distribution of the FI moiety. Off-line UV spectra of peak 1 and peak 2 (Fig. 6a, inset) further confirmed that the peak compositions are related to the internal standard 2, 3-DAP and the analyzed nanodevice (UV spectrum of the tri-functional nanodevice peak is consistent with the spectrum measured using UV-Vis spectrometer).9 All the mono- and multifunctional G5 dendrimer nanodevices were analyzed in the same way. They all displayed a monodisperse molecular distribution.

Discussion
Molecular distribution of dendrimer-based nanodevices regarding polymer polydispersity does not increase significantly after surface multi-step conjugation, as verified by SEC measurements (Table 1). This is because SEC is only based on polymer size distribution that does not change significantly upon dendrimer surface conjugation. As opposed to other analytical techniques (i.e. SEC, UV-Vis spectrometry, and NMR) that can only detect the average molecular physicochemical parameters, CE provides a unique way to detect the distribution of molecular species based on the differences in their electrophoretic mobilities resulting from the change of their charge/mass ratios. Possible factors affecting the molecular distribution of functional dendrimer conjugates are briefly discussed as follows.

Fig. 6 CE electropherograms of a tri-functional PAMAM nanodevice (G5Ac82FIFAOHMTX, 8) detected at 210, 300, and 495 nm (a). A magnified electropherogram detected at 495 nm is shown in (b). Off-line UV-Vis spectra of peak 1 (internal standard) and peak 2 (nanodevice) are shown in the inset of (a). The CE was run using a silanized silica capillary (id 100 mm) with effective length of 40 cm and total length of 48.5 cm.

This journal is The Royal Society of Chemistry 2006

Analyst, 2006, 131, 374381 | 379

(1) The classification of dendrimer terminal groups influences the surface moiety distribution of dendrimer conjugates. We previously demonstrated that the dispersities of dendrimers could be magnified by acetylation and masked by hydroxylation for amine-terminated PAMAM dendrimers.15 For all materials synthesized using approach 1 (starting from amine-terminated G5.NH2), more prominent FA moiety distribution can be observed in G5.FAAc (2) than that of G5.NH2FA (1) (Fig. 2), while G5.FAOH (3) and G5.FA OHMTX (4) (Fig. 2 and 4) only show a mono-modal distribution. In the case of hydroxyl-terminated dendrimer conjugates, because of the incomplete hydroxylation and the existence of the resulting secondary terminal amines, the very close number of terminal hydroxyl groups masks the molecular differences between individual dendrimer conjugate molecules. This is expected to result in a mono-modal distribution of dendrimer conjugate species. These results are consistent with our previous report.15 Similarly, multifunctional G5.Ac82FA OHMTX (7) and G5Ac82FIFAOHMTX (8) nanodevices synthesized using approach 2 display narrower CE peaks than those of G5.Ac75, 5, and 6 that do not possess hydroxyl terminal groups (Fig. 5). This further suggests the dispersity masking effect of hydroxylation reaction for dendrimer conjugation. Please note that peak 2-related dendrimer conjugate in the electropherogram of G5.FAAc (2) display a higher concentration of FA moiety molecules than peak 1-related material, which is opposed to peak composition comparison of G5.NH2FA (1) conjugate (Fig. 3 and supporting information{). This may be reasoned as follows: in the separation of 2, the hydrophobic interaction between the silanized capillary surface (methyl groups) and 2 may be stronger than the non-specific electrostatic interaction, which makes conjugate 2 containing more hydrophobic FA moiety molecules migrate slower than those containing less FA moiety molecules. (2) The synthetic approaches. Two synthetic approaches were employed to synthesize mono- and multifunctional G5 dendrimer nanodevices. Compared with approach 1 which the conjugation reaction started from amine-terminated G5.NH2 dendrimer, in approach 2 partially acetylated G5 dendrimer was instead used as a starting material. The high local concentration of amine groups of G5.NH2 makes it highly reactive, thereby significantly distributing the conjugated moiety molecules. Whereas, after partial acetylation reaction, the small portion of amine groups remaining on the G5 dendrimer surface can preserve the homodispersity of the dendrimer for subsequent bioconjugation.9 Consequently, mono- and multifunctional dendrimer nanodevices synthesized using approach 2 always display a monomodal distribution, which is opposed to those dendrimer conjugates synthesized from an amine-terminated G5 dendrimer (approach 1). Please note that the partial acetylation reaction is crucial for one to produce targeted monodisperse dendrimer conjugates. The acetylation reaction is very fast. One has to add dilute acetic anhydride solution slowly to a dendrimer solution under vigorous stirring in order to achieve highly monodispersed partially acetylated G5 dendrimer. If the preparation procedure is not appropriate, highly dispersed G5.Acn can be obtained. This definitely influences the polydispersity of the final G5 PAMAM nanodevices.
380 | Analyst, 2006, 131, 374381

Conclusions
In summary, various mono- and multifunctional dendrimer nanodevices were synthesized either from amine-terminated or from partially acetylated G5 dendrimers and were characterized using SEC, UV-Vis spectrometry, 1H NMR, and CE. CE was extensively used to evaluate the molecular distribution of the as-synthesized dendrimer conjugates. The CE results show that the molecular distribution of the as-synthesized dendrimer nanodevices largely depends on the classification of the final functionalization step and the synthetic approaches. For dendrimer conjugates synthesized from amine-terminated G5 dendrimers, hydroxyl-terminated monofunctional and multifunctional nanodevices display a monomodal distribution, while amino and acetamide-terminated G5FA conjugates show multimodal distributions. Mono- and multifunctional G5 dendrimer conjugates synthesized from partial acetylated G5 dendrimers always display a monomodal distribution. It indicates that partial acetylation is critical to preserve the homodispersity of dendrimers for subsequent bioconjugation reactions. The applied CE analysis of monoand multifunctional PAMAM-based nanodevices provides an alternative approach to evaluate the molecular heterogeneity of complex dendrimer conjugate nanodevices, which are being used for targeted cancer therapy.

Acknowledgements
We thank Chunyan Chen for her valuable remarks and suggestions. This work is financially supported by the National Cancer Institute (NCI), National Institute of Health (NIH), under the contract # NOI-CO-97111.

References
1 D. A. Tomalia, A. M. Naylor and W. A. Goddard, III, Angew. Chem., Int. Ed. Engl., 1990, 29, 138. 2 D. A. Tomalia and J. M. J. Frechet, Dendrimers and Other Dendritic Polymers, John Wiley & Sons Ltd, New York, 2001. 3 R. Esfand and D. A. Tomalia, Drug Discovery Today, 2001, 6, 427. 4 A. E. Beezer, A. S. H. King, I. K. Martin, J. C. Mitchel, L. J. Twyman and C. F. Wain, Tetrahedron, 2003, 59, 3873. 5 R. Abu-Rmaileh, D. Attwood and A. DEmanuele, Drug Delivery Syst. Sci., 2003, 3, 65. 6 A. Quintana, E. Raczka, L. Piehler, I. Lee, A. Myc, I. Majoros, A. K. Patri, T. Thomas, J. Mule and J. R. Baker, Jr., Pharm. Res., 2002, 19, 1310. 7 A. K. Patri, I. Majoros and J. R. Baker, Jr., Curr. Opin. Chem. Biol., 2002, 6, 466. 8 J. R. Baker, Jr., A. Quintana, L. T. Piehler, M. Banazak-Holl, D. Tomalia and E. Raczka, Biomed. Microdevices, 2001, 3, 61. 9 I. J. Majoros, T. P. Thomas, C. B. Mehta and J. R. Baker, Jr., J. Med. Chem., 2005, 48, 5892. 10 I. J. Majoros, T. P. Thomas and J. R. Baker, Jr., in Handbook of Theoretical and Computational Nanotechnology, ed. M. Reilly and W. Schommers, American Scientific Publishers, Stevenson Ranch, 2006, in press. 11 T. P. Thomas, A. K. Patri, A. Myc, M. T. Myaing, J. Y. Ye, T. B. Norris and J. R. Baker, Jr., Biomacromolecules, 2004, 5, 2269. 12 T. P. Thomas, I. J. Majoros, A. Kotlyar, J. F. Kukowska-Latallo, A. Bielinska, A. Myc and J. R. Baker, Jr., J. Med. Chem., 2005, 48, 3729. 13 S. D. Konda, M. Aref, S. Wang, M. Brechbiel and E. C. Wiener, Magma, 2001, 12, 104. 14 E. C. Wiener, S. Konda, A. Shadron, M. Brechbiel and O. Gansow, Invest. Radiol., 1997, 32, 748.

This journal is The Royal Society of Chemistry 2006

15 X. Shi, I. Banyai, M. T. Islam, W. Lesniak, D. Z. Davis, J. R. Baker, Jr. and L. Balogh, Polymer, 2005, 46, 3022. 16 D. A. Tomalia, H. Baker, J. R. Dewald, M. Hall, G. Kallos, S. Martin, J. Roeck, J. Ryder and P. Smith, Polym. J., 1985, 17, 117. 17 D. A. Tomalia, H. Baker, J. R. Dewald, M. Hall, G. Kallos, S. Martin, J. Roeck, J. Ryder and P. Smith, Macromolecules, 1986, 19, 2466. 18 B. L. Schwartz, A. L. Rockwood, R. D. Smith, D. A. Tomalia and R. Spindler, Rapid Commun. Mass Spectrom., 1995, 9, 1552. 19 D. I. Malyarenko, R. L. Vold and G. L. Hoatson, Macromolecules, 2001, 34, 7911. 20 C. Zhang and D. A. Tomalia, in Dendrimers and other dendritic polymers, ed. J. M. J. Frechet and D. A. Tomalia, John Wiley and Sons, Ltd, New York, 2001, pp. 239253. 21 A. Sharma, D. K. Mohanty, A. Desai and R. Ali, Electrophoresis, 2003, 24, 2733. 22 A. Sharma, A. Desai, R. Ali and D. Tomalia, J. Chromatogr., A, 2005, 1081, 238. 23 X. Shi, A. K. Patri, W. Lesniak, M. T. Islam, C. Zhang, J. R. Baker, Jr. and L. Balogh, Electrophoresis, 2005, 26, 2960. 24 X. Shi, I. Banyai, W. Lesniak, M. T. Islam, I. Orszagh, P. Balogh, J. R. Baker, Jr. and L. Balogh, Electrophoresis, 2005, 26, 2949. 25 J. Zheng and R. M. Dickson, J. Am. Chem. Soc., 2002, 124, 13982. 26 W. I. Lee, Y. Bae and A. J. Bard, J. Am. Chem. Soc., 2004, 126, 8358.

27 D. Wang and T. Imae, J. Am. Chem. Soc., 2004, 126, 13204. 28 A. Sharma, M. Rao, R. Miller and A. Desai, Anal. Biochem., 2005, 344, 70. 29 M. T. Islam, I. J. Majoros and J. R. Baker, Jr., J. Chromatogr., B: Biomed. Appl., 2005, 822, 21. 30 M. T. Islam, X. Shi, L. Balogh and J. R. Baker, Jr., Anal. Chem., 2005, 77, 2063. 31 P. E. Froehling and H. A. J. Linssen, Macromol. Chem. Phys., 1998, 199, 1691. 32 I. J. Majoros, B. Keszler, S. Woehler, T. Bull and J. R. Baker, Jr., Macromolecules, 2003, 36, 5526. 33 X. Shi, W. Lesniak, M. T. Islam, M. C. MuNiz, L. Balogh and J. R. Baker, Jr., Colloids Surf., A, 2006, 272, 139. 34 E. Seyrek, P. L. Dubin and G. R. Newkome, J. Phys. Chem. B, 2004, 108, 10168. 35 Q. R. Huang, P. L. Dubin, C. N. Moorefield and G. R. Newkome, J. Phys. Chem. B, 2000, 104, 898. 36 H. M. Brothers, L. T. Piehler and D. A. Tomalia, J. Chromatogr., A, 1998, 814, 233. 37 A. Ebber, M. Vaher, J. Peterson and M. Lopp, J. Chromatogr., A, 2002, 949, 351. 38 M. S. Diallo, S. Christie, P. Swaminathan, L. Balogh, X. Shi, W. Um, C. Papelis, W. A. Goddard, III and J. H. Johnson, Jr., Langmuir, 2004, 20, 2640. 39 S. J. Duthie, Nutrition, 2001, 17, 736.

This journal is The Royal Society of Chemistry 2006

Analyst, 2006, 131, 374381 | 381