Selectively functionalized glycerol/diacid dendrimers via click chemistry of azido fatty acids

Article abstract of DOI:10.1007/s11746-010-1675-x

Dendrimers consisting of glycerol and dicarboxylic acid units were synthesized using a convergent or "outside-in" strategy. The multimeric arms, or dendrons, are joined to a central core in the final step using the azide/alkyne click reaction. The combination of these approaches allows the preparation of dendrimers with variable and controlled degrees of substitution at their periphery. For example, a single protected amino acid has been situated at the periphery, with all other substituents being hydrophobic. In another example, all the peripheral substituents are protected amino acids. The identity of the dendrimers is unambiguous due to the purification and characterization of all the synthetic intermediates. Functional groups have also been selectively incorporated along the arms (dicarboxylic acids) and at the cores.

Full text of DOI:10.1007/s11746-010-1675-x

J Am Oil Chem Soc (2011) 88:403–413
DOI 10.1007/s11746-010-1675-x
ORIGINAL PAPER
Selectively Functionalized Glycerol/Diacid Dendrimers
via Click Chemistry of Azido Fatty Acids
•
•
Jonathan A. Zerkowski Alberto Nun˜ez
Daniel K. Y. Solaiman
Received: 26 March 2010 / Revised: 18 August 2010 / Accepted: 7 September 2010 / Published online: 26 September 2010
Ó US Government 2010
Abstract Dendrimers consisting of glycerol and dicar-
boxylic acid units were synthesized using a convergent or
‘‘outside-in’’ strategy. The multimeric arms, or dendrons,
are joined to a central core in the final step using the azide/
alkyne click reaction. The combination of these approaches
allows the preparation of dendrimers with variable and
controlled degrees of substitution at their periphery. For
example, a single protected amino acid has been situated at
the periphery, with all other substituents being hydropho-
bic. In another example, all the peripheral substituents are
protected amino acids. The identity of the dendrimers is
unambiguous due to the purification and characterization of
all the synthetic intermediates. Functional groups have also
been selectively incorporated along the arms (dicarboxylic
acids) and at the cores.
possess have been thoroughly reviewed elsewhere [1]; two
that deserve mention are the increased density of terminal
or surface or capping groups, which can be chosen to
impart a property of choice, and the possibility of entan-
glement or hosting of a guest molecule at the interior. This
combination of features has led to the use of hyperbranched
materials as drug delivery agents and matrices for tissue
engineering [2, 3]. As the partner reactant for glycerol,
dicarboxylic acids are also logical choices, considering the
ease of making ester linkages, and the likely biocompati-
bility of these building blocks. (Polyglycerol, which is also
highly branched, will not be considered here since as a
homopolyether it is chemically quite distinct, and it is not
necessarily prepared from glycerol itself but from other
three-carbon units [4]). There have been two general styles
of routes to glycerol/diacid hyperbranched materials. First
is the ‘‘all-at-once’’ route, where the two reactants are
combined in a selected ratio and esterified using a catalyst
(either inorganic or enzymatic) [3, 5–7]. Second is a
stepwise route where protected monomer units are added
one layer or shell at a time [8]. Both routes have their
strengths and drawbacks. The first is simple and suited to
larger scale, but there is little control over the extent of
reaction, and many glycerol OH groups may remain
unreacted. The second offers better control over the thor-
oughness of reaction, although once again at later stages
the increased number of reacting groups may make 100%
completion difficult to achieve, and there is the comple-
mentary problem that every single protecting group needs
to be removed cleanly. Furthermore, the completed mac-
romolecule still can only be derivatized globally and not
at selected sites. These two routes are illustrative of the
distinction between hyperbranched polymers (the former
route) where an incomplete reaction leads to incomplete
branching and ‘‘dangling’’ or free OH groups, and a true
Keywords Dendrimers Á Nanochemistry Á
Functionalized oligoesters Á Click reaction
Introduction
Glycerol, with its three reactive functional groups, is an
obvious choice as a building block for highly branched
oligomers or polymers. The practical advantages that a
highly branched structure, as opposed to a linear one, can
Mention of trade names or commercial products in this publication is
solely for the purpose of providing specific information and does not
imply recommendation or endorsement by the US Department of
Agriculture.
J. A. Zerkowski (&) Á A. Nun˜ez Á D. K. Y. Solaiman
USDA, ARS, Eastern Regional Research Center,
600 E. Mermaid Lane, Wyndmoor, PA 19038, USA
e-mail: jonathan.zerkowski@ars.usda.gov
123

404
J Am Oil Chem Soc (2011) 88:403–413
‘‘dendrimer’’ structure (the latter) where the degree of
branching is exactly 1 and all OH groups have been
esterified (except for any that are intentionally left at the
final surface). Another important distinction between the
two routes is that molecular characterization of the product
of the one-pot route, e.g., by NMR or mass spectrometry,
can be extremely challenging, since there are many alter-
native incomplete branching patterns, while the stepwise
nature of the other route allows characterization along the
way. In fact, the broad product distribution of the former
route makes clear-cut structure/function relationships
murky at best.
(APCI) MS data were recorded on a Waters/Micromass
(Milford, MA, USA) ZMD instrument, using direct injec-
tion without an LC column. Matrix-Assisted Laser
Desorption/Ionization mass spectra with automated tandem
time of flight fragmentation of selected ions (MALDI-TOF/
TOF) were acquired with an ABI 4700 Proteomics Ana-
lyzer mass spectrometer (Applied Biosystems, Framing-
ham, MA, USA) in the positive reflectron mode. Masses
were determined as sodiated adducts of the products,
[M ? Na]?, using 2,5-dihydroxybenzoic acid as matrix in
a concentration of 10 mg/mL in acetonitrile/water (1:1)
with 0.1% TFA. Approximately 0.7 lL of matrix was
spotted on the MALDI plate and allowed to dry. The
sample (0.5–1 lL) dissolved in chloroform or acetone at a
concentration of 1–2 mg/mL was spotted on the top of the
matrix crystals. Averages of 1,000–2,000 spectra were
acquired for optimal signal to noise ratio.
To demonstrate the versatility of structural and func-
tional control that the basic pattern of glycerol/diacid
oligomers can offer, but that has yet to be taken full
advantage of, we report here the modular synthesis of
glycerol/dicarboxylic oligomers that are tailored at their
periphery, arms, and core. Our method hinges on two key
strategies. First is a convergent approach that builds the
dendrimer from the outside in. Second is use of the azide/
alkyne ‘‘click’’ reaction for linking the bulky branched arms
to a central core, a reaction that has proven to be reliable in a
number of sterically demanding situations, e.g., at cell
surfaces [9] as well as in the construction of dendrimers
[10–12]. We recently prepared some azido fatty acids and
used the click reaction to ligate them to biomolecules [13],
so we were prompted to consider their use in the con-
struction of larger aggregates such as dendrimers. One
specific aim was to be able to choose the extent of labeling
at the exterior of the dendrimer, that is, to vary the coverage
from single-point derivatization to full coating with a
functional group of choice, which is an area of active
research in dendrimer chemistry [14]. A further aim was to
construct nanoscale molecules of a defined (monodisperse)
size and shape from bioderived, renewable building blocks.
Synthesis of Benzyl Glyceryl Adipate, 1b
To a solution of adipic acid (5.0 g, 34 mmol), benzyl
alcohol (3.7 g, 34 mmol), solketal (4.5 g, 34 mmol), and
triethylamine (20 mL, 0.14 mol) in 50 mL THF, cooled in
an ice bath and stirred magnetically, is added a solution of
benzoyl chloride (9.5 g, 68 mmol) in 10 mL THF dropwise
over the course of 5 min [15]. After about half of the
benzoyl chloride solution has been added, 4-dimethyl-
amino pyridine (DMAP, 200 mg, 1.6 mmol) is added, and
when the dropwise addition is complete (a thick precipitate
forms), another 200 mg of DMAP is added. The solution is
allowed to warm to room temperature overnight. Solvent is
removed, and the crude is taken up in 100 mL ethyl acetate
and washed with 50 mL each saturated citric acid, 1 M
NaOH, and water. Solvent is again removed, and the crude
is chromatographed on silica gel (Rf = 0.3 in 3:1 hexane/
ethyl acetate) to yield 1a (4.8 g, 81% based on an ideal
1:2:1 ratio of the mixed diester to the two symmetrical
diesters).
Materials and Methods
To remove the acetonide, the benzyl/solketal adipic
diester 1a is dissolved in approximately 10 mL of 80%
AcOH/H2O. The solution is heated at 45 °C for 3–4 h, then
stirred at RT overnight. The syrup is poured into a crys-
tallizing dish and evaporated under a stream of nitrogen. It
is then dissolved in 1:1 EtOAc/hexane and passed through
a short silica column to remove all traces of acetic acid.
The yield is quantitative. 1H NMR: 1.64–1.78 (m, 4H,
internal CH2), 2.33–2.47 (m, 4H, C(O)CH2), 3.55–3.76
(m, 2H, CH2OH), 3.83–4.01 (m, 1H, CH(OH)), 4.11–4.28
(m, 2H, C(O)OCH2), 5.14 (s, 2H, CH2Ar), 7.36 (br s, 5H,
Ar). 13C NMR: 24.3, 24.4, 33.9 C(O)CH2, 63.4 CH2OH,
65.4 CH2Ar, 66.4 CHOH, 70.3 esterified gly CH2, (128.3,
128.4, 128.7, and 136.0) all Ar, 173.5 and 173.7 C=O.
17-Azido stearic acid was synthesized as previously
described [13]. Chemicals and reagents were obtained
commercially from Sigma-Aldrich (St. Louis, MO, USA)
and Lancaster Synthesis (Alfa Aesar, Ward Hill, MA,
USA). All solvents and reagents were used as received.
˚
Silica gel (Grade 60 A, Mesh 230–400, particle size
40–63 lm), used for all column chromatography, was
obtained from Fisher Scientific (Fairlawn, NJ, USA). Sol-
vents were removed on a Yamato rotary evaporator. NMR
spectra were recorded in CDCl3 on either a Varian Asso-
ciates (Walnut Creek, CA, USA) Gemini 200 MHz or
Inova 400 MHz instrument and were referenced to
Si(CH3)4. Atmospheric pressure chemical ionization
123

J Am Oil Chem Soc (2011) 88:403–413
405
General Procedure for Pivalic Anhydride Couplings
overnight, then solvent was removed on the rotary evapo-
rator and the crude was purified by flash chromatography on
silica gel to afford 3 (0.53 g, 93%). In the synthetic steps
that follow, the ‘‘standard procedure’’ is to use the coupling
agent, pivalic anhydride, in slight excess (10–20%) relative
to the carboxylic acid component, along with 10–20 mg
DMAP as catalyst. This method routinely gave yields better
than 80% and frequently better than 90%. 1H NMR:
1.18–1.41 (m, 81H), 1.41–1.52 (m, 6H), 1.55–1.70 (m, 6H),
[16]: Synthesis of Core 3
To a solution of 17-azido stearic acid [13] (0.65 g,
2.0 mmol) and glycerol (52 mg, 0.56 mmol) in 10 mL
dioxane was added a solution of pivalic anhydride (0.39 g,
0.43 mL, 2.1 mmol) in 2 mL THF, followed by 4-dimeth-
ylamino pyridine (DMAP, 10 mg, 0.08 mmol). The mixture
was stirred vigorously with a magnetic stir bar at RT
Fig. 1 General outline for the
iterative construction of
clickable dendrons. In this
paper, n = 4 (adipic acid).
Compare to Scheme 1, where
the ‘‘nested’’ character of
specific intermediates is
diagrammed explicitly in the
building up of dendron 8
123

406
J Am Oil Chem Soc (2011) 88:403–413
Scheme 1
2.33 (t, 7.2 Hz, 6H, C(O)CH2), 3.42 (hex, 6.5 Hz, 3H,
C(17)H-N3), 4.08–4.21 and 4.25–4.38 (m, 4H, gly CH2),
5.21–5.37 (m, 1H, gly CH). 13C NMR: 19.6 CH3, 25.0, 26.2,
29.2–29.8, 34.2, 36.3 C(O)CH2, 58.1 C(17)-N3, 62.2 gly
CH2, 69.0 gly CH, 173.0 and 173.4 C=O. APCI: calculated
for C57H107N9O6ÁNa?: 1,036.8 Da, found 1,037.0 Da.
sequentially to a new unit of diol 1b. First the dicaproic
monomer 1d (0.42 g, 1.0 mmol) was coupled to 1b (0.65 g,
2 mmol). The resulting mono-hydroxy compound (0.60 g,
0.85 mmol) was then reacted with the Boc-b-Ala/caproic
monomer 1c (0.49 g, 1.0 mmol). This trimer is shown in
condensed fashion in the open rectangle at the top of Fig. 5
and explicitly as 1e in Scheme 1. The benzyl ester is
removed hydrogenolytically, and the resulting free acid
(0.82 g, 0.75 mmol) is coupled to the monoacylglycerol of
17-azido stearic acid, H3C–CH(N3)(CH2)15CO–OCH2-
CH(OH)CH2(OH) (1.0 g, 2.5 mmol) to afford a tetramer
unit 1f that still possesses one free OH. An excess of the
mono-17-azido stearoyl glycerol is used so that di-acylation
is minimized; unreacted mono-17-azido stearoyl glycerol is
readily recovered with column chromatography. After
purification, this molecule 1f (0.85 g, 0.58 mmol) is coupled
with a second trimer 1g (0.68 g, 0.67 mmol) possessing four
caproic units (Scheme 1 and shaded rectangle at the top of
Fig. 5), to afford the heptameric dendron 8 (see Fig. 5 and
Scheme 1), 1.2 g (0.49 mmol, 84% for this final acylation).
1H NMR: 0.88 (t, 6.6 Hz, 21H, caproic CH3), 1.20–1.41 m,
1.44 (s, 9H, Boc tBu), 1.53–1.76 m, 2.22–2.45 (m, 40H,
C(O)CH2), 2.56 (t, 6.0 Hz, 2H, b-Ala C(O)CH2), 3.41
(m, 3H, b-Ala CH2NH and C(17)H-N3), 4.09–4.23 and
4.25–4.50 (m, 28H, gly CH2’s), 5.04–5.16 (m, 1H, Boc-
NH), 5.21–5.36 (m, 14H, gly CH). 13C: 14.0, 21.1, 22.4,
24.2–24.9, 27.2, 28.5, 29.2–29.9, 31.2, 33.6–34.6, 36.1, 41.1,
58.2 C(17)–N3, 62.1–62.6, 68.8–69.1, 79.5, 155.6,
172.3–173.3. MALDI: calculated for C125H208N4O44ÁNa?:
2,492.41 Da, found 2,492.46 Da.
Representative Synthesis of a Dendron (Molecule 8)
(Refer to Figs. 1, 5; Scheme 1)
To a solution of diol 1b (1.15 g, 3.70 mmol) and Boc-
b-alanine (0.53 g, 2.8 mmol) in 10 mL THF is added a
solution of pivalic anhydride (0.52 g, 2.8 mmol) in 2 mL
THF dropwise, followed by DMAP (15 mg, 0.1 mmol). A
substoichiometric amount of the alanine is used to mini-
mize formation of its di-adduct onto 1b. The solution is
stirred magnetically overnight. Solvent is removed and the
diacylglycerol product—C6H5CH2OC(O)(CH2)4C(O)OCH2
CH(OH)CH2OC(O)CH2CH2NHCO2tBu—is isolated by
column chromatography, which is also used to purify all
subsequent synthetic intermediates (0.99 g, 74%). This
diacylglycerol (0.6 g, 1.3 mmol) is then reacted with caproic
acid (0.2 g, 1.7 mmol) using the standard pivalic anhydride/
DMAP protocol (260 mg, 1.4 mmol pivalic anhydride,
10 mg DMAP) to yield adipoyl triacylglycerol 1c, sche-
matically represented by the AB monomer in Fig. 1 (0.65 g,
86%). After it is purified, it is dissolved in THF and the
benzyl ester is removed hydrogenolytically over 5% Pd/C to
yield 0.49 g (1.0 mmol). A second monomer 1d, having two
identical acyl units attached to the glycerol unit, (i.e. C=D in
Fig. 1) was prepared similarly from 1b (1.2 g, 3.9 mmol)
and two equivalents of caproic acid (1.0 g, 8.6 mmol) and
likewise debenzylated. The next step is constructing a trimer
(see Fig. 1), which is done by coupling these two monomers
Dendrons 2, 6, and 9 were built up using similar iterative
steps. Other Dendron MALDI data:
Dendron 2: calculated for C154H254N8O60ÁNa? 3,198.70
Da, found 3,198.75 Da.
123

J Am Oil Chem Soc (2011) 88:403–413
407
Dendron 6: calculated for C137H229N3O46ÁNa? 2,675.57
Da, found 2,675.61 Da.
THF, K2CO3 (1.4 g, 10 mmol) was added, and the solution
was warmed to approximately 45 °C under N2. A solution
of diphenylphosphoryl azide (2.6 g, 9.5 mmol) in THF was
added and the heat was increased to reflux the solution for
7 h, after which time it was stirred at RT overnight. Solvent
was removed and the crude was taken up in EtOAc and
washed with water. Column chromatography yielded C6H5
CH2OCO(CH2)7NHCO-solketal, which was then depro-
tected using H2 over Pd/C to give the free acid, 1.67 g,
57% from benzyl azelate. This material was coupled with
propargyl alcohol using the pivalic anhydride/DMAP
method to give the clickable ester, which was subsequently
deprotected at the diol end with aq. HOAc as described for
1b above to give HCCCH2OC(O)(CH2)7NHCO2CH2CH
(OH)CH2(OH), 1.41 g, 79%. This diol is coupled with
Dendron 9: calculated for C117H189NO44ÁNa? 2,335.25
Da, found 2,335.26 Da.
Synthesis of the Triazine Core (5, Fig. 4)
(A) Propylamino-dichloro-triazine was prepared by react-
ing propylamine (0.16 g, 2.7 mmol) with cyanuric chloride
(0.5 g, 2.7 mmol) at RT in 15 mL THF for 3 h. Removal
of solvent and column chromatography afforded the
product (0.24 g, 86%), CH3CH2CH2NH(C3N3)Cl2. (B)
N-phenoxyacetamidoyl c-tert-butyl glutamic piperazine
amide: Z-piperazine (Z = C6H5CH2OCO, 0.77 g, 3.5 mmol)
was coupled to Fmoc-Glu(OtBu)-OH (1.5 g, 3.5 mmol),
with diisopropylcarbodiimide (DIC, 0.44 g, 3.5 mmol) in
20 mL CH2Cl2 for 2 h. Solvent was removed, then 10 mL
N,N-diisopropylethylamine and 1,8-diazabicycloundecene
(0.76 g, 5 mmol) were added to remove the Fmoc group.
Reaction proceeded for 18 h, then solvent was removed,
15 mL dioxane was added, and phenoxy acetic acid
(0.61 g, 4 mmol) and DIC (0.44 g, 3.5 mmol) were added.
The mixture was stirred magnetically overnight. Solvent
was removed, and column chromatography gave 1.15 g, 61
% from Fmoc-Glu(OtBu)OH. Finally the Z group was
removed from the piperazine with H2 on 5% Pd/C in
25 mL EtOH to give N-phenoxyacetamidoyl c-tert-butyl
glutamic piperazine amide, C6H5OCH2C(O)NHCH(CH2
CH2COOtBu)C(O)(NC4H8NH) (0.80 g, 95%). (C) To a
solution of propylamino-dichloro-triazine (0.17 g, 0.80 mmol)
in 25 mL dioxane was added N-phenoxyacetamidoyl
c-tert-butyl glutamic piperazine amide (0.73 g, 1.8 mmol)
and the solution was refluxed under N2 for 6 h. Over the
course of this time, N,N-diisopropylethylamine was added
periodically (0.23 g, 1.8 mmol) to neutralize the HCl that
is formed. After removal of solvent, the product was
purified by column chromatography to give 636 mg prod-
uct (84%). The t-Bu protecting groups were removed with
CF3CO2H, which was then removed under a stream of N2.
To prepare the di-propargyl ester core 5, the free diacid
(0.44 g, 0.53 mmol) and propargyl alcohol (280 mg,
5 mmol) were reacted using the pivalic anhydride/DMAP
protocol (0.37 g, 2.0 mmol, and 20 mg, respectively).
After purification (Rf = 0.4 in ethyl acetate), 415 mg
(87%) of 5 was obtained. Calculated for C46H56N10O10ÁH?
909.4, found 909.8 Da by APCI. The NMR spectrum
is incorporated within that for dendrimers 7a and 7b
below.
1
2 equiv of trimer 1 g for the preparation of dendron 9. H
NMR: 1.24–1.42 (m, 6H), 1.42–1.56 (m, 4H), 1.56–1.75
(m, 4H), 2.37 (t, 7.4 Hz, 2H, CH2C(O)), 2.49 (t, 2.6 Hz,
alkyne CH), 3.15 (q, 6.7 Hz, 2H, CH2NH), 3.54–3.74
(m, 2H, CH2OH), 3.82–3.96 (m, 1H, CH(OH)), 4.11–4.27
(m, 2H, C(O)OCH2), 4.69 (d, 2.4 Hz, alkyne–CH2–O),
5.05–5.20 (m, 1H, NH). 13C NMR: 24.7, 26.5, 28.9, 29.0,
29.8, 34.0 C(O)CH2, 41.2 CH2NH, 51.9 OCH2–alkyne,
63.3 gly CH2OH, 65.9 gly CHOH, 70.8 esterified gly
OCH2, 74.9 HC (alkyne), 77.9 quaternary alkyne, 157.2
carbamate C=O, 173.0 ester C=O. APCI: calculated for
C15H25NO6ÁNa? 338.2 Da, found 338.3 Da.
Representative Conditions for the Click Reaction
Triacylglycerol core 3 (Fig. 2) (0.11 g, 0.11 mmol) and
dendron 2 (1.25 g, 0.39 mmol) were dissolved in 10 mL
2:1 H2O/t-BuOH and 3 mL THF. The solution was stirred
vigorously with a magnetic stir bar. An aqueous solution
of CuSO4Á5 H2O (10 mol%, 10 mg in 0.5 mL) was added,
followed immediately by sodium ascorbate (20 mol%,
15 mg in 0.5 mL) and reaction was continued overnight
[17]. TLC of the crude showed no remaining 3. Solvent
was removed and the crude was purified by column
1
chromatography to afford dendrimer 4 (1.0 g, 90%). H
NMR: 0.94 (d, 6.5 Hz, 144 H, Leu CH3), 1.09–1.35, 1.45
(s, 216 H, Boc tBu), 1.50–1.78, 2.22–2.40 (m, 90H,
(C(O)–CH2), 4.08–4.41 (m, 112H, gly CH2’s and Leu
C(a)H), 4.63 (h, 6.8 Hz, 3H, C(17)H–N), 4.92–5.19
(m, 24H, Boc NH), 5.21 (s, 6H, OCH2–triazole),
5.20–5.38 (m, 22H, gly CH), 7.58 (s, 3H, triazole). 13C
NMR (When a range of ppm values is listed, there are
several overlapping peaks).: 21.4 C(18)H3, 21.8 Leu CH3,
22.9, 24.2, 24.9, 26.1, 28.4, 29.2–29.7, 33.6, 33.7, 34.1,
37.3, 41.4–41.6 Leu b-CH2, 52.2 Leu a-CH, 57.6 C(17)–
triazole, 62.0–62.7 gly CH2 and OCH2–triazole, 68.9–69.8
gly CH, 79.9 Boc quaternary, 121.6 triazole C, 142.6
triazole CH, 155.5 Boc C=O, 172.4–173.4 carbonyls. MS:
Synthesis of the Azelaic/Glycerol Carbamate Unit (Fig. 5)
The monobenzyl ester of azelaic acid (2.6 g, 9.3 mmol)
and solketal (20 mL, 0.16 mol) were dissolved in 20 mL
123

408
J Am Oil Chem Soc (2011) 88:403–413
Fig. 2 Construction of a dendrimer with 24 external units. Ad adipic, G glycerol, Leu-Boc C(O)CH(CH2CH(CH3)2)NHC(O)OC(CH3)3
calculated for C519H869N33O186ÁNa?, 10,563.9 Da, found
10,566.7 Da.
Dendrimer 7b (see text and Fig. 3): 0.88 (t, 6.6 Hz,
36H, caproic CH3), 0.93 (t, 7.3 Hz, 3H, propyl CH3),
1.07–1.44 m, 1.49–1.77 m, 1.77–1.93 m, 2.22–2.45 (m,
92H, C(O)CH2), 2.49 (t, 2.6 Hz, 4H, alkyne CH), 3.36 (q,
6.9 Hz, 2H, propyl NHCH2), 3.55–3.72 (m, 8H, piperazine
CH2), 3.72–3.94 (m, 8H, piperazine CH2), 4.05–4.22 (m,
28H, gly CH2), 4.22–4.38 (m, 28H, gly CH2), 4.49 (m, 4H,
CH2OAr), 4.63 (q, 6.6 Hz, 2H, C(17)H–triazole), 4.66
(d, 2.5 Hz, 8H, OCH2–alkyne), 4.94 (t, 6.5 Hz, 1H, propyl
NH), 5.08–5.19 (m, 2H, Glu C(a)H), 5.19–5.36 (m, 18H,
gly CH and OCH2–triazole), 6.90–7.07 (m, 6H, Ar),
7.23–7.39 (m, 4H, Ar), 7.60 (d, 8.5 Hz, 2H, NH Glu), 7.67
(s, 2H, triazole). 13C NMR: 11.6 propyl CH3, 14.0 caproic
CH3, 21.4 C(18)H3, 22.4, 23.1, 24.2, 24.6–24.9, 26.1, 27.2,
28.3, 28.9–29.8, 31.2, 33.6, 34.2, 37.3, (42.3–43.3, 45.4,
and 47.6 piperazine and propyl NHCH2), 51.8 OCH2–
alkyne, 57.6 C(17)–triazole, 58.2 glu C(a), 62.1–62.4 gly
CH2 and OCH2–triazole, 67.2 CH2O–Ar, 68.9 and 69.1 gly
CH, 74.9 alkyne CH, 77.9 quaternary alkynyl C, 114.8 Ar,
121.5 triazole CH, 122.2 Ar, 129.8 Ar, 142.6 triazole C,
157.2 O–C(Ar) quaternary, 165.3 and 166.4 triazine C,
168.2 and 169.5 amide C=O, 172.3–173.3 carbonyl.
MALDI: calculated for C316H490N16O102ÁH? 6,142.37 Da,
found 6,142.54 Da.
Dendrimer 7a: 0.85 (t, 6.5 Hz, 36H, caproic CH3), 0.93
(t, 7.3 Hz, 3H, propyl CH3), 1.04–1.37 m, 1.39 (s, 36H,
tBu), 1.46–1.73 m, 1.75–1.91 m, 2.18 (t, 7.2 Hz, 8H,
tBuOC(O)CH2), 2.21–2.47 (m, 84H, other C(O)CH2), 3.32
(q, 6.9 Hz, 2H, propyl NHCH2), 3.50–3.70 (m, 8H, piper-
azine), 3.70–3.91 (m, 8H, piperazine), 4.04–4.19 (m, 28H,
gly CH2), 4.19–4.36 (m, 28H, gly CH2), 4.47 (m, 4H,
CH2OAr), 4.62 (h, 6.6 Hz, 2H, C(17)H–triazole), 4.86
(t, 6.5 Hz, 1H, propyl NH), 5.05–5.14 (m, 2H, Glu C(a)H),
5.14–5.34 (m, 18H, gly CH and OCH2–triazole), 6.86–7.03
(m, 6H, Ar), 7.20–7.31 (m, 4H, Ar), 7.57 (d, 8.7 Hz, 2H,
NH Glu), 7.64 (s, 2H, triazole). 13C NMR: 11.6 propyl
CH3, 14.0 caproic CH3, 21.4 C(18)H3, 22.3, 23.1, 24.2,
24.6–25.0, 26.0, 27.1, 28.2, 28.9–29.7, 31.2, 33.6, 34.1,
35.5, 37.2, (42.2–43.3, 45.4, and 47.6 piperazine and pro-
pyl NHCH2), 57.6 C(17)-triazole, 58.2 Glu C(a)H,
62.1–62.3 gly CH2 and OCH2–triazole, 67.2 CH2OAr, 68.9
gly CH, 79.9 tBu quaternary, 114.8 Ar, 121.5 triazole CH,
122.2 Ar, 129.8 Ar, 142.6 triazole C, 157.1 O–C(Ar)
quaternary, 165.3 and 166.4 triazine C, 168.2 and 169.5
amide C=O, 172.3–173.3 carbonyls. MALDI: calculated
for C320H514N16O102ÁH? 6,218.63 Da, found 6,220.51 Da.
123

J Am Oil Chem Soc (2011) 88:403–413
409
1
Fig. 3 Comparison of the H
NMR traces for core 3 (bottom)
and dendrimer 4 (top). Note the
complete conversion of the
C(17)H–N3 protons around
3.4 ppm in core 3 to the
‘‘clicked’’ triazole product
around 4.6 ppm
Dendrimer 10: 0.85 (t, 6.0 Hz, 69H, caproic CH3),
1.07–1.40 m, 1.42 (s, 9H, Boc tBu), 1.48–1.78 m,
1.78–2.02 m, 2.20–2.45 (m, 128H, C(O)CH2), 2.54 (t,
6.8 Hz, 2H, b-Ala C(O)CH2), 3.13 (q, 6.6 Hz, 4H, carba-
mate CH2NH), 3.38 (q, 6.5 Hz, 2H, b-Ala CH2NH),
4.03–4.21 (m, 44H, gly CH2), 4.21–4.39 (m, 44H, gly
CH2), 4.64 (h, 7.1 Hz, 3H, C(17)H–triazole), 4.96–5.11 (m,
1 Boc NH and 2H azelaic carbamate NH), 5.18 (s, 6H,
OCH2–triazole), 5.21–5.35 (m, 22H, gly CH), 7.58 (s, 3H,
triazole). 13C NMR: 13.9 caproic CH3, 21.4 C(18)H3, 22.3,
24.2, 24.6–24.8, 26.0, 26.5, 27.1, 28.4, 28.8–29.8, 31.2,
33.5, 34.1, 37.2 carbamate CH2NH, 41.1 b-Ala CH2NH,
57.6 C(17)–triazole, 62.1–62.3 gly CH2 and OCH2–tria-
zole, 68.7–69.5 gly CH, 79.3 Boc quaternary C, 121.5
triazole CH, 142.7 triazole C, 155.8 Boc C=O, 172.3–173.6
carbonyl. MALDI: calculated for C410H676N12O140ÁNa?
8,036.74 Da, found 8,035.92 Da.
synthesis is obtaining 100% reaction at each site, with no
sites ‘‘skipped’’, the odds of attaining a complete reaction
decreases at each stage. Even if acylation goes in 99%
yield at each site, an octa-hydroxy compound would afford
(0.99)8 or 92% yield. Purifying the intermediates, at least
using silica gel chromatography, is also difficult, since a
highly polar multiple-hydroxy compound can be hard to
elute, and there would be practically no chance of sepa-
rating out incompletely reacted side products that have
slightly different degrees of acylation. There is also the
matter of requiring 100% removal of the protecting groups,
as well as any reagents used for deprotection that could
interfere with future reactions. As an aside, these problems
are reminiscent of those faced by peptide chemistry that
were essentially solved by the introduction of solid-phase
protocols.
A better strategy is the ‘‘outside-in’’ or convergent one,
which facilitates unambiguous characterization and purifi-
cation at every stage of reaction. The external units are
constructed first, and protecting groups are removed when
needed. Using AB2 monomers, each reaction stage occurs
at two sites at most, for example between a diol-acid
protected at its carboxy end and two equivalents of free
carboxylic acids (see Fig. 1). Differences in polarity are
such that separation of unreacted starting materials from
product, including partially reacted variants where one OH
of the diol-acid is left unreacted on purpose, is straight-
forward. In an iterative way, the multimeric arms, referred
to as dendrons, are built up after removal of the carboxy
protecting group and subsequent rounds of coupling
Results and Discussion
Synthetic Strategies
There are two routes that can be followed in the con-
struction of dendrimers. The more obvious, at first glance,
is by starting at a core and adding successive layers,
moving ‘‘inside-out’’. But this divergent procedure has the
drawback that the number of reacting groups at each stage
increases, for example from four to eight to sixteen
hydroxy groups. Since the critical aspect of dendrimer
123

410
J Am Oil Chem Soc (2011) 88:403–413
(Fig. 1). The dendrons we employed are heptamers, since
they contain seven glycerol and seven dicarboxylic or fatty
acid units, not counting the external capping groups. While
we speculate that our iterative methodology would have
been successful in preparing dendrons of a larger size as
well (the next generation would be a pentadecamer), since
the ester-forming reactions did not appear to get slower as
the dendrons got larger—and if they did, the temperature
could be raised—we chose to stop at the heptamer stage
due to anticipated analytical limitations, that is, we were
concerned that dendrimers larger than about 10 kDa might
be difficult to ionize and characterize by mass spectrome-
try. The final step, then, is linkage of dendrons to the core,
and again only a small number of reactions occur at this
stage, three at most in our examples. Depending on the size
of the dendrons, steric interference when bringing together
several units of kilodalton size toward the central core
could be severe. Since we need these reactions to go with a
high degree of confidence in virtually 100% yield, we
chose the azide/alkyne ‘‘click’’ reaction for the final step,
which has been shown to meet this criterion [9–12, 17]. An
additional benefit is that it is compatible with a wide range
of functional groups.
glycerol side is ‘‘outward’’ and the benzyl group is the
inward end. It is true that this convergent strategy requires
the use of protecting groups, but the benefits of specificity,
that is, being able to perform a reaction at only selected sites
when desired, far outweigh any drawbacks or tedium arising
from a few extra steps. Furthermore, the particular pro-
tecting groups we used are easy to remove mildly, and just
as with the bond-forming reactions in the synthetic direc-
tion, only a small number of groups need to be removed at
each step, which is again a major contrast with the divergent
route described above. The acetonide on solketal only needs
to be removed in an initial batch step, not repeatedly during
preparation of dendrons. It is removed with 80% acetic acid/
water at room temperature or slightly above (40–50 °C).
These mild conditions do not destroy the ester linkages. All
traces of acetic acid do, however, need to be removed,
otherwise they would compete with the incoming carbox-
ylic component as a capping/acylating agent. The benzyl
ester protecting group is removed by hydrogenation over
palladium on carbon; again, ester bonds in the rest of the
molecule are undisturbed. The only limitation of course is
that olefins cannot be used elsewhere in the molecule.
Synthetically, ester bonds are the connections used to
build up the dendrons, and for these we have employed two
variants of mixed anhydride routes that are very mild and
go in high yields. For construction of all the dendron ester
bonds, we use pivalic anhydride [16], but for construction
of the diester starting materials, e.g., 1, we found that the
benzoyl chloride route works better [15]. As would be
expected, three different diester products are obtained—the
dibenzyl, di-solketal, and desired mixed ester 1a—but they
are easily separated on silica gel, having DRf’s of at least
0.1. It may be that the polarity of the solketal acetonide unit
contributes to the ease of separation, however, a similar
effect was noted with the three products obtained in pre-
paring the benzyl/tert-butyl diester of azelaic acid, used for
the synthesis of dendron 6. We propose that the benzoyl
chloride mixed anhydride method may be a general way for
preparing non-symmetrical diesters of dicarboxylic acids.
Interestingly, this route does not work for adding a tert-
butyl ester on azelaic acid, giving only dark decomposition
products, presumably due to acidity of benzoyl chloride
and accompanying HCl, but the pivalic method does work,
although the reaction is sluggish and needs to be refluxed in
THF. The benzyl/tert-butyl azelaic diester had to be pre-
pared by first making the mono benzyl ester through the
benzoyl chloride route, then attaching the tert-butyl group
with the pivalic route. The benzoyl chloride route is also
less satisfactory for preparation of the dendron units, since
a minor side product (approximately 5–10%) is benzoyla-
tion of free hydroxy groups in 1b or its congeners. We have
found that the pivalic route does not afford a similar side
product (the trimethylacetylated derivative).
A brief comment on the isomeric nature of these den-
drons and dendrimers is in order. When differential sub-
stitution is used on the external units—that is, when
A = B in the monomer molecules shown in Fig. 1—the
acylations can occur at either the free primary or the sec-
ondary OH of the glycerol. We have found that the 1,2 and
1,3 diacylglycerol isomers, e.g., 1b acylated with one
equivalent of Boc-b-alanine as described in the synthesis of
dendron 8 in the methods section, can be separated using
silica gel chromatography, but we have chosen not to
separate them. Our hypothesis is that any local effects
imposed on an acyl unit by attachment at a primary versus
a secondary position of glycerol will be minuscule com-
pared to the overall sterics and flexibility of the dendrimer.
Considering the structural features of a quasi-spherical
dendrimer, there are three elements that can be varied to
‘‘tune’’ the properties of the completed oligomer, for
example in terms of size, polarity, or functional groups.
These elements are the peripheral units (A–D in Fig. 1), the
arms or dicarboxylic acids, and finally the core. As our basic
AB2 arm unit we chose the mono-glycerol adipic ester 1b.
We have no reason to believe that longer diacids would not
work (see for example reference [3], where sebacic acid was
used), but the adipate NMR pattern is simple, and the two
carboxylic units should be far enough apart to not interfere
with each other, either sterically or reactively. Linear, non-
hyperbranched glycerol/adipic polyesters have also found
use in drug delivery [18]. Succinic acid, by contrast, could
conceivably participate in side-reactions where one free
carboxylate cyclizes with an OH at the other end. The
123

J Am Oil Chem Soc (2011) 88:403–413
Design and Preparation of Targets
411
N-terminus of glutamic acid, were simply chosen for
convenience; in a dendrimer geared toward some specific
application they could have been some other aromatic or
acyl group, perhaps a fluorescent one, or a binding point for
a specific guest. The attachment to the triazine nucleus is
via the piperazine amide at the C-terminus, and the side
chain COOH of Glu provides the attachment point for the
clickable dendrons, a propargyl ester. The third substituent
of the triazine ring, the propylamino group, was again
chosen for convenience and its clear NMR signal; as with
the phenoxy groups above, a more ‘‘useful’’ functionalized
unit could have been in its place. One interesting advantage
of the triazine nucleus is that it can be substituted in a
sequential fashion, that is, three different Glu units, each
with its own unique substituent at the N-terminus, could
have been attached. The dendrons used in this target are
substituted with functional groups to a medium extent,
namely two reactive groups each (the tert-butyl protected
carboxyl groups of azelaic acid) with the other caps being
hydrophobic caproic acid units. The t-Bu groups are an
excellent choice for pursuing further derivatization once
the dendrimer is constructed: we removed them using tri-
fluoroacetic acid, then installed new propargyl esters in
their place (molecule 7b). In this way, the dendrimer is ‘‘re-
primed’’ for further rounds of click chemistry, although
other amide or ester derivatives of the azelaic units could
have been added. The final dendrimer has a bola shape.
The three dendrimers we have prepared using this meth-
odology were designed to exhibit varying degrees of
structural complexity. The first and most straightforward
combines dendrons that are homogeneously capped with a
symmetrical core (molecule 4, Fig. 2). The 24 external
units of the dendrons are protected amino acids, the arms
are all adipic acid, and the connection point to the core is a
propargyl ester. The core is the triacylglycerol prepared
using 17-azido stearic acid. As expected, the threefold click
reaction went in high yield. There was no ‘‘partial-click’’
product observed, where only one or two of the azide units
of core 3 reacted, as shown by the complete disappearance
of the C(17)H–N3 signal at 3.4 ppm (Fig. 3). One potential
application of this macromolecule is that it would serve as
a polycation after deprotection of the amino groups.
Structures of this sort have been referred to as ‘‘unimo-
lecular micelles’’ and are actively being studied for their
self-assembling properties [19].
The second target is composed of a polyfunctional core,
namely the triazine shown in Fig. 4. This core was
designed to demonstrate one way in which functional
groups and/or hydrogen-bonding moieties could be ‘‘hid-
den’’ at the center of the dendrimer. To accomplish this, a
multifunctional linker, glutamic acid, is used to build out
the core. The phenoxyacetic amide groups, attached to the
Fig. 4 Construction of a bola-shaped dendrimer with a functional-
ized triazine core. G glycerol, Ad adipic, Cap caproic, Az azelaic.
Removal of the t-Bu esters with CF3CO2H, then coupling with
propargyl alcohol, affords the alkynyl dendrimer 7b that could
undergo further extension via click chemistry
123

412
J Am Oil Chem Soc (2011) 88:403–413
The third dendrimer (Fig. 5) shows the kind of structure
the other end of the dicarboxylic acid with a clickable
propargyl group. The first dendron added is singly substi-
tuted, that is, it possesses at its periphery only a single group
(the NH-Boc) that can be further reacted with, for example,
that can be assembled by performing sequential click
reactions with different dendrons. The core begins as a
monoacylglycerol built from azelaic acid, monoesterified at
Fig. 5 Construction of a dendrimer with a single reactive group (NH-Boc) at its periphery. Abbreviations as in Fig. 3
123

J Am Oil Chem Soc (2011) 88:403–413
413
2. Svenson S (2009) Dendrimers as versatile platform in drug
delivery applications. Eur J Pharm Biopharm 71:445–462
3. Wang Y, Ameer GA, Sheppard BJ, Langer R (2002) A tough
biodegradable elastomer. Nat Biotechnol 20:602–606
4. Wilms D, Stiriba S-E, Frey H (2010) Hyperbranched polygly-
cerols: from the controlled synthesis of biocompatible polyether
polyols to multipurpose applications. Acc Chem Res 43:129–
141
5. Wyatt VT, Nun˜ez A, Foglia TA, Marmer WN (2006) Synthesis of
hyperbranched poly(glycerol-diacid) oligomers. J Am Oil Chem
Soc 83:1033–1039
6. Stumbe JF, Bruchmann B (2004) Hyperbranched polyesters
based on adipic acid and glycerol. Macromol Rapid Commun
25:921–924
7. Kulshrestha AS, Gao W, Gross RA (2005) Glycerol copolyesters:
control of branching and molecular weight using a lipase catalyst.
Macromolecules 38:3193–3204
8. Degoricija L, Carnahan MA, Johnson CS, Kim T, Grinstaff MW
(2006) Synthesis and characterization of bola-type amphiphilic
dendritic macromolecules. Macromolecules 39:8952–8958
9. Moses JE, Moorhouse AD (2007) The growing applications of
click chemistry. Chem Soc Rev 36:1249–1262
a peptide or a fluorescent group. It was gratifying that the
click reaction could be performed with the free dihydroxy
glycerol group. The next step was to introduce two new
clickable units, namely 17-azido stearates, converting the
monoacylglycerol to a triacylglycerol. The final two click
reactions are then performed using dendrons that are
homogeneously substituted with hydrophobic (caproic)
exteriors. In addition, we explored the use of modified
dicarboxylic arms by preparing the monocarbamate deriv-
ative of azelaic acid (see Dendron 9 in Fig. 5). This group is
interesting in its own right as a way to increase hydrogen
bonding along the arms (relative to ester units) but it also
presents a way to introduce ‘‘side-chain’’ functionality to
the otherwise featureless dicarboxylic acids, e.g. by alkyl-
ating the amine. In general, the diphenyl phosphoryl azide-
modified Curtius rearrangement is a convenient way to
prepare a carbamate AB2 monomer that should be more
resistant to hydrolysis than the corresponding ester 1b.
The three dendrimers described here show how the extent
of functional coverage at the periphery of a quasi-spherical
polyester can be monotonically varied, from single-point
substitution to tetra-substituted to 24 substituents. Func-
tionalized triacylglycerols and fatty acids obviously have a
contribution to make to this important area of nanotech-
nology. It may be possible to engineer the selective targeting
or delivery of an encapsulated drug to particular biological
sites by surface modification of the dendrimer, either by
attaching a signaling peptide or by controlling the pattern of
charges. Some fundamental issues in macromolecular sci-
ence that this methodology opens for study is how ‘‘sticky’’
these oligomers could be toward each other. To what extent
for example would dendrimer X, coated with 12 positively
charged groups, associate with its congener Y, coated with
12 negatively charged groups, as a function of solvent
polarity? Alternatively, dendrimers with positive charges on
one face, negative charges on another, and hydrophobic
groups on a third might self-assemble to novel structures,
such as fibrils that lead to gelation [20] or ‘‘artificial lipid
clusters’’ [21]. Finally, there is the issue of their size. Crude
molecular modeling (Chem3D and CPK models) suggests
that these molecules would be approximately 10 nm in
diameter. Since they contain flexible alkyl chains, that
number can only be an average subject to molecular oscil-
lations. The variability of the dicarboxylic arms would allow
construction of a family of dendrimers of increasing size,
e.g., from adipic to octadecanedioic acid, and a comparison
of physical properties across such a family.
10. Lee JW, Kim JH, Kim B-K, Shin WS, Jin S-H (2006) Synthesis
´
of Frechet type dendritic benzyl propargyl ether and Frechet type
triazole dendrimer. Tetrahedron 62:894–900
´
11. Wu P, Feldman AK, Nugent AK, Hawker CJ, Scheel A, Voit B,
Pyun J, Frechet JMJ, Sharpless KB, Fokin VV (2004) Efficiency
and fidelity in a click-chemistry route to triazole dendrimers by
the copper(I)-catalyzed ligation of azides and alkynes. Angew
Chem Int Ed 43:3928–3932
12. Joralemon MJ, O’Reilly RK, Matson JB, Nugent AK, Hawker CJ,
Wooley KL (2005) Dendrimers clicked together divergently.
Macromolecules 38:5436–5443
13. Zerkowski JA, Nun˜ez A, Strahan GD, Solaiman DKY (2009)
Clickable lipids: azido and alkynyl fatty acids and triacylglyce-
rols. J Am Oil Chem Soc 86:1115–1121
14. Morgan JR, Cloninger MJ (2002) Heterogeneously functionalized
dendrimers. Curr Opin Drug Discov Dev 5:966–973
15. Dhimitruka I, SantaLucia J Jr (2006) Investigation of the Yam-
aguchi esterification mechanism. Synthesis of a Lux-S enzyme
inhibitor using an improved esterification method. Org Lett
8:47–50
16. Sakakura A, Kawajiri K, Ohkubo T, Kosugi Y, Ishihara K (2007)
Widely useful DMAP-catalyzed esterification under auxiliary
base- and solvent-free conditions. J Am Chem Soc 129:14775–
14779
17. Rostovtsev VV, Green LG, Fokin VV, Sharpless KB (2002) A
stepwise Huisgen cycloaddition process: copper(I)-catalyzed
regioselective ‘‘ligation’’ of azides and terminal alkynes. Angew
Chem Int Ed 41:2596–2599
18. Kallinteri P, Higgins S, Hutcheon GA, St. Pourc¸ain CB, Garnett
MC (2005) Novel functionalized biodegradable polymers for
nanoparticle drug delivery systems. Biomacromolecules 6:1885–
1894
19. Kainthan RK, Brooks DE (2008) Unimolecular micelles based on
hydrophobically derivatized hyperbranched polyglycerols: bio-
distribution studies. Bioconj Chem 19:2231–2238
20. Seo M, Kim JH, Kim J, Park N, Park J, Kim SY (2010) Self-
association of bis-dendritic organogelators: the effect of dendritic
architecture on multivalent cooperative interactions. Chem Eur J
16:2427–2441
21. Ariga K, Urakawa T, Michiue A, Kikuchi J (2004) Spider-web
amphiphiles as artificial lipid clusters: design, synthesis, and
accommodation of lipid components at the air-water interface.
Langmuir 20:6762–6769
References
1. Grayson SM, Frechet JMJ (2001) Convergent dendrons and den-
drimers: from synthesis to applications. Chem Rev 101:3819–3867
123