Chromophoric and dendritic phosphoramidites enable construction of functional dendrimers with exceptional brightness and water solubility

Article abstract of DOI:10.1002/chem.201403445

The fluorescence brightness of a molecular probe determines whether it can be effectively measured and its water solubility dictates if it can be applied in real-world biological systems. However, molecules brighter than the most efficient fluorescent dye

Full text of DOI:10.1002/chem.201403445

DOI: 10.1002/chem.201403445
Full Paper
&
Fluorescent Probes
Chromophoric and Dendritic Phosphoramidites Enable
Construction of Functional Dendrimers with Exceptional
Brightness and Water Solubility
Andrew D. Shaller, Wei Wan, Baoming Zhao, and Alexander D. Q. Li*[a]
Abstract: The fluorescence brightness of a molecular probe
determines whether it can be effectively measured and its
water solubility dictates if it can be applied in real-world bio-
logical systems. However, molecules brighter than the most
efficient fluorescent dyes or particles brighter than quantum
dots are hard to come by, especially when they must also be
soluble in water. In this report, chromophoric phosphorami-
dites are used in a solid-state synthesis to construct func-
tional dendrimers. When highly twisted chromophores are
chosen and the proper spacers and dendrons are intro-
duced, the resultant dendrimers emit exceptionally bright
fluorescence. Chromophores, spacers, and dendrons are
stitched together by efficient phosphoramidite reagents,
which afford high-yield water-soluble phosphodiester linkag-
es after deprotection. The resulting water-soluble dendrim-
ers are exceptionally bright.
Introduction
ylene-based dendrimers were up to six times brighter than
a single fluorophore.[7] Although various terminal groups prom-
ise future aqueous compatibility, water solubility is currently
not reported in these rigid dendrimers without the use of de-
tergents.[8] The progress on water-soluble and non-water-solu-
ble perylene-based polymers has been summarized in recent
publications.[9] A major obstacle encountered in the tightly
spaced dendrimers is fluorescence self-quenching, resulting in
limited dendrimer brightness in aqueous systems.
Fluorescent imaging probes reveal a wealth of information
about molecular mechanisms and biological processes. The
key enabling factors are the brightness and water solubility of
the probes. Newly designed water-soluble, multi-chromophoric
dendrimers offer a promising approach to solve such daunting
challenges. Dendrimers are a class of spherical polymers that
fill the gap between small molecules and nanoparticles. As
such, dendrimers exhibit unique functions that cannot be
found in either small molecules or nanoparticles.[1] Because of
the tree-like structures, unlike those of polymeric random coils,
spherical dendrimers have been applied in many important
areas, including high contrast reagents for magnetic resonance
imaging (MRI),[2] carriers for DNA vectors in gene delivery,[3]
and versatile vehicles for drug delivery.[4] Indeed, the water-
soluble poly(amidoamine) has played an instrumental role in
the above nano-biotechnologies.[5]
We report the first successful use of solid-phase synthesis to
construct dendrimers by using the strategy of attenuating sur-
face reactive sites. The linchpin, which was not realized before,
is that dendritic growth is prohibited when there are too many
growing chains crowded on the solid phase.
Previously, phosphoramidites of fluorophores (F1) based on
planar perylene building blocks were synthesized
(Scheme 1).[10] These phosphoramidite reagents are too reac-
tive and suffer self-degradation because the phosphorus atom
is attached to the primary alcohol. Moreover, the resultant
multi-chromophoric polymers had low fluorescence quantum
yields owing to p-stacking. Herein, new phosphoramidites of
fluorophores, which use a secondary alcohol and have an
asymmetric structure (F2 and F3 in Scheme 1 and 2), are used
to solve these challenges. The net results are that both water-
soluble and exceptionally bright dendrimers are constructed.
Although dendrimers can function both as carrying vehicles
and molecular probes, they are not widely used in fluores-
cence imaging. The reasons include the fact that they are not,
thus far, much brighter than organic dyes at the single-mole-
cule level and syntheses of water-soluble, multi-chromophoric
dendrimers can be really difficult. A considerable amount of
pioneering work has been focused on constructing fluorescent
dendrimers (FD) utilizing either flexible or rigid architectures.[6]
Single-molecule fluorescence measurements revealed that per-
Results and Discussion
In the new chromophoric phosphoramidites, a rigid cyclohexa-
nol amide was used to replace one of the two flexible tetra-
ethylene glycol (TEG) chains for the following important rea-
sons (F2 and F3; also see the Supporting Information,
Scheme S1). First, the more sterically demanding cyclohexanol
group prevents efficient p-stacking, thus enhancing the bright-
[a] Dr. A. D. Shaller, W. Wan, B. Zhao, Prof. A. D. Q. Li
Department of Chemistry
Washington State University
Pullman, WA 99164 (USA)
E-mail: dequan@wsu.edu
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/chem.201403445.
Chem. Eur. J. 2014, 20, 12165 – 12171
12165
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
Scheme 1. Chemical structures of chromophoric phosphoramidites, dendritic
phosphoramidites, spacers, and terminator phosphoramidites used to con-
struct dendrimers on solid supports.
Scheme 2. Chemical structures of the various dendrimer building blocks and
a schematic illustration of the dendrimer structure. See Table 1 for the den-
drimer sequences.
ness of the dendrimers constructed by using such moieties.
Similar approaches to the suppression of p–p interactions by
using extended spacers have been reported.[11] Second, the ri-
gidity and the negative charges of the phosphodiesters steer
the chromophores away from each other, thus greatly sup-
pressing self-quenching and improving fluorescent quantum
yields. This point is supported by the fact that perylene di-
imides functionalized with charged groups have improved
aqueous fluorescence quantum yields.[12] Third, the cyclohexa-
nol amide provides a secondary alcohol for phosphitylation, re-
sulting in a more stable phosphoramidite than that from a pri-
mary alcohol. Fourth, the new chromophores are highly twist-
ed, further frustrating p–p stacking and self-quenching. Phos-
phoramidites from primary alcohols are known for their high
reactivity and should be used immediately, within days for
better results. In contrast, phosphoramidites from secondary
alcohols are rather stable in the solid state and can be stored
for later use. The reason that F2 was chosen as the dendritic
building block is that it has a large extinction coefficient (e=
es S2–S5 for details). The hydroxyl groups on M1 coupled
quantitatively at the 5’-end, but this phosphoramidite multipli-
er did not couple well at the 3’-end, giving “stable” trityl
yields,[14] rather than exponential trityl yields. Adopted from
DNA nomenclature, the 5’-end is the dimethoxyl trityl (DMTr)
terminal; the 3’-end, the phosphoramidite terminal. Short tre-
bler M2 and doubler M4 gave higher 3’-coupling yields, but
only had 5’-coupling yields around 80%; the likely cause of
this lower yield is the close proximity of the terminal hydroxyl
groups. Amazingly, the triazole-based trebler (M3) shown in
Scheme 1 gave excellent tripling yields despite the short link-
age to the branching center. The resultant tripling yields as re-
ported by the absorbance of the DMTr group are 275%, or
ꢀ91% coupling efficiency, at the root of the phosphoramidite.
The terminal hydroxyl groups of the three dendrites are suffi-
ciently far away from each other, enabling nearly quantitative
coupling at the 5’-ends, comparable with the coupling efficien-
cy of the standard DNA bases.
4ꢁ104 mÀ1 cmÀ1)[13] and near unity quantum yield (f ).
fl
Complementary to the highly twisted fluorophores, dendritic
multipliers based on phosphoramidite chemistry are required
for solid-state dendrimer syntheses. Thus, several phosphora-
midite multipliers were synthesized (M1, M2, M3, and M4 in
Schemes 1 and 2): a pentaerythritol-based trebler, M1, with
four propyl branching arms; a short pentaerythritol trebler M2;
a pentaerythritol-triazole-based trebler M3; and a glycerol
based doubler M4 (see the Supporting Information Schem-
Multi-chromophoric dendrimer syntheses were carried out
on solid-state supports by alternating the coupling of chromo-
phoric phosphoramidite and trebler phosphoramidite to
achieve dendritic growth. Where needed, spacers (S1 and S2
in Schemes 1 and 2; also see the Supporting Information,
Scheme S6) were introduced to decouple unfavorable molecu-
lar interactions that cause fluorescence self-quenching. One
advantage of using solid-state synthesis is that it offers rapid
Chem. Eur. J. 2014, 20, 12165 – 12171
www.chemeurj.org
12166
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
purification. The solid phase can be separated from the com-
plex liquid mixture easily and the dendrimers anchored on the
solid supports are pure after a few rinses. Another advantage
of using the phosphoramidite approach is high-yield coupling
with a reporter trityl group that allows facile monitoring and
reporting of the reaction yield for each generation because re-
moving the protecting DMTr groups yields quantitative meas-
ures of the growing chains.
Such attenuation creates the necessary space needed for the
exponential growth of dendrimers on surfaces.
Following a few steps of attenuation, dendritic growth start-
ed by coupling the phosphoramidite to the spacers, chromo-
phores, or treblers (Scheme 2). Figure 1 depicts that coupling
efficiency is manifested by the stepwise change in the trityl re-
sponse. The coupling yields for linear growth, such as attach-
ing spacers or chromophores to the growing chains, were basi-
cally quantitative. The average response increased by ꢁ270–
280% (compared with 300% in ideal reactions) for each tre-
bling step, indicating ꢁ92% coupling efficiency for attaching
the trebler to its root (from S2 to S2M3 in Figure 1). Subse-
quent dendritic growth gave the same near-quantitative cou-
pling of a trebler to the growing chains; this quantitative cou-
pling again resulted in near 270–300% trityl yields after the
trebler reaction as shown in Figure 1.
After many trial-and-error attempts, an important prerequi-
site for solid-supported dendrimer synthesis was discovered—
the active sites on the commercial CPG (control porous glass)
must be first “attenuated” in order to create adequate spacing
for dendritic growth later. Without attenuation, trityl yields
from previous dendrimer synthesis remained “stable” and no
exponential or dendritic growth occurred.[14] Two attenuation
techniques were tested: the first method used a mixture of
phosphoramidite terminator Tr (Scheme 1, also see the Sup-
porting Information, Scheme S6) with a spacer phosphorami-
dite S1 or S2, whereas the second method limited the reaction
of the spacer S2 only. A 2:1 ratio of Tr/S1 was found to work
the best and trityl analysis reported that this attenuation con-
sistently killed ~75% of the growing chains in each reactive se-
quence (Figure 1). Furthermore, the monomethoxy terminator
(Tr) likely acts as a “comb” to keep the functional ethylene
glycol chains optimally extended for the next coupling step.
The attenuation and exponential growth strategy worked
well and we used it to synthesize dendrimers D1--D8, depicted
in Table 1. The first-generation dendrimer D1 has a dendritic
sequence of [A-S2-F2-M3ꢂ(F2-S2)3]; the second-generation
dendrimer D2 has a dendritic sequence of [C-S1-S1-S2-M3ꢂ
(F2-M3ꢂ(F2-S2)3)3)]; the fourth generation dendrimers D6 and
D7 have dendritic sequences of [T-S2-S2-S2-S2-S2-M3ꢂ(F2-
M3ꢂ(F2-M3ꢂ(F2-M3ꢂ(F2-ACGT)3)3)3)3] and [T-S2-S2-S2-S2-S2-
S2-M3ꢂ(F2-S1-M3ꢂ(F2-S1-M3ꢂ(F2-S1-M3ꢂ(F2-S1-
ACGT)3)3)3)3], respectively. Spacers, chromophores and multipli-
ers are defined in Scheme 2; A, T, C, and G stand for regular
DNA bases. The automated solid-state synthesizer generates
a dendrimer in a few hours, much superior to manual dendri-
mer syntheses. Once the dendrimer synthesis was completed,
concentrated NH4OH(aq.) simultaneously cleaved the dendrimers
from the solid support and converted the phosphotriester link-
ages resulting from the phosphoramidite coupling to negative-
ly charged phosphodiesters. Multiple-charged phosphodiester
linkages in the dendritic branches make the resultant dendri-
mer rather soluble in water.
Matrix-assisted laser desorption ionization (MALDI) mass
spectrometry was able to confirm the mass of dendrimer D1,
giving an observed mass of m/z=7597, which agrees with the
calculated mass of 7597 (C312H425Cl16N22O137P10Na2). Ensemble
optical properties of dendrimers were also collected before
single-molecule measurements. Absorption and fluorescence
spectra for dendrimers D2 and D8 are shown in Figure 2a and
Figure 1. On porous glass spheres, the growing chains were attenuated by
using a mixture of S1 and Tr at 1:2 ratio to create space for dendritic
growth later. After three attenuation steps, dendrimer synthesis started with
the S2 phosphoramidite to extend the chain and the M3 trebler to branch
the chain. Each tripling step yields the trityl responses on a log scale of
nearly ~300%, indicating the anticipated exponential growth of the den-
drites. Chemical structures of the dendrimer building blocks are given in
Schemes 1 and 2.
Table 1. Dendrimer sequences along with brightness and diameter.
Dendrimer
Dendritic sequence
Brightness[a]
N[b]
Diameter[c] [nm]
D1
D2
D3
D4
D5
D6
D7
D8
A S2 F2 M3 (F2 S2)3
C (S1)2 S2 M3 (F2 M3(F2 S2)3)3
T (S2)4 M4 (S2 F2 M4(S2 F2 M4(S2 F2 M4(S2 F2 S2)3)3)3)3
T (S2)3 M3 (F2 M3(F2 M3(F2 ACTG)3)3)3
N (S2)3 M3 (F2 S2 M3(F2 S2 M3(F2 S2)3)3)3
T (S2)5 M3 (F2 M3(F2 M3(F2 M3(F2 ACTG)3)3)3)3
T (S2)6 M3 (F2 S1 M3(F2 S1 M3(F2 S1 M3(F2 S1 ACTG)3)3)3)3
T (S2)5 M3 (F2 M3(F2 M3(F2 M3(F2 M3(F2 M3(F2 M3(F2 M3(F2 ACTG)3)3)3)3)3)3)3)3
2.1
3.9
6.8
7.5
20
18
74
114
0.53
0.33
0.23
0.19
0.51
0.15
0.62
0.012
–
12
–
9
10
12
16
12
[a] Modal brightness of dendrimers is determined by using the single-molecule modal brightness as the unit, i.e., the modal intensity of fluorescein or tet-
rachloro perylene diimide. [b] Normalized brightness per dye. [c] Diameters of the dendrimers in nanometers are measured using dynamic light scattering.
Chem. Eur. J. 2014, 20, 12165 – 12171
www.chemeurj.org
12167
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
Figure 3. CCD images reveal that dendrimers D1--D8 are much brighter
than single-molecules fluorescein dyes. Note the scale bar differences.
Figure 2. Absorption (left) and fluorescence (right) spectra of dendrimers: a)
D2 and b) D8 theoretically containing up to 12 and 9840 F2 fluorophores,
respectively. Dynamic light scattering results for c) D4 and d) D5 report hy-
drodynamic diameters that are much smaller than the fully elongated mo-
lecular length or even as predicted by the freely jointed chain model.
The brightness of these dendrimers were determined by
using a home-built single-molecule microscope equipped with
a liquid-N2 cooled CCD and a 488 nm Ar ion laser and two ava-
lanche photodiodes (APDs). The water-soluble dendrimers
were uniformally sprinkled on a coverslip so that a single den-
drimer could be imaged in the wide field by using either CCD
or APD raster scanning.[17] Figure 3 reveals the representative
10ꢁ10 mm CCD images with 2s integration. Fluorescein and
1,6,7,12-tetrachloro perylene diimide (1e: see the Supporting
Information, Scheme S1) were used to establish the brightness
of a single molecule; both had modal brightness of 125 cps
(counts per second) under the experimental conditions. The
scale bar at the top of each image calibrates the relative
brightness intensity, and the modal brightness of dendrimers
D1 and D2 is 250 and 500 cps, respectively. Thus, the first-gen-
b. The spectra are similar to the monomer spectra in CH2Cl2
except they tend to be broadened and there are varying de-
grees of intensity reversal between the A0À1/A0À0 and Fl0À1/Fl0À0
vibronic bands, an indication of hydrophobically driven p-
stacking, to some degree, between some of the twisted tetra-
chloro perylene diimides. The A0À0/A0À1 ratio changes from
dendrimer to dendrimer, and there is not enough data yet to
determine a clear trend. The experimental extinction coeffi-
cient per chromophore in the dendrimers is ꢀ7ꢁ103 mÀ1 cmÀ1
because interior and exterior dyes are likely to have different
absorption properties depending on generation number, G,
fluorophore spacing, and surface groups.
eration dendrimer D1 (f ꢀ24%) that contains four fluoro-
fl
phores is about twice as bright as the highly fluorescent fluo-
rescein as indicated by the scale bar (Figure 3). Progressively,
Dynamic light scattering (DLS) measures hydrodynamic di-
ameters of 9–16 nm for these water-soluble dendrimers as
shown in Table 1, and Figure 2c and d. These values are small-
er than the maximum fully elongated radius of the dendrimers
(sequential sum of the segment lengths determined by molec-
ular modelling, for example 10 nm for D1 to 47 nm for D8),
whereas the freely jointed chain (FJC) model predicts radii
from 6 to 38 nm.[15] Unlike quantum dots (QDs) or fluorescent
nanoparticles that have strong Tyndall effects,[16] these low-
density polyethylene glycol (PEG)-based dendrimers do not
scatter light very much, and thus enhancing fluorescence col-
lection efficiency.
the second-generation dendrimer D2 (f ꢀ8%), which contains
fl
up to 12 dyes theoretically, is ~4 times brighter than the stan-
dard dye fluorescein (Figure 3). Although fluorescence self-
quenching still exists in these dendrimers, they are much
brighter than previous multi-chromophoric oligomers using
planar chromophore F1 owing to the reasons stated above.[12]
The fourth-generation dendrimer D6 (f ꢀ4%), which con-
fl
tains up to 120 fluorophores and has a modal brightness of
2250 cps, is ~18 times brighter than the modal brightness of
highly fluorescent fluorescein (125 cps) (Figures 3 and 4). Such
dendrimers are as bright as QDs, which have been character-
ized as 10–20 times as bright as the most efficient fluorescent
dyes.[18] However, the brightness per fluorophore within the
dendrimer for this fourth-generation dendrimer is apparently
not as high as those of the first- or second-generation den-
drimers (D1 or D2). Thus, further potential brightness gain can
be realized in the fourth-generation dendrimer if its structure
is optimized to further reduce self-quenching and improve
The fluorescent quantum yields, f , for dendrimers D1–D8
fl
range between 2% and 24%. Similar to single-molecule bright-
ness, dendrimers with greater intramolecular spacing tend to
have higher f values. Also, for generations 2–4, f appears to
fl
fl
decrease with increasing G. Interestingly, for the 8th genera-
tion dendrimer D8, where hydrophobic encapsulation is likely,
f rebounds to 20%.
fl
Chem. Eur. J. 2014, 20, 12165 – 12171
www.chemeurj.org
12168
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
Figure 5. Time traces for dense dendrimers D1, D2, D4, D6, and D8 are com-
pared with a single dye—tetrachloro perylene diimide (1e, see the Support-
ing Information, Scheme S1). Although a single molecule typically shows
one-step on–off blinking before photobleaching, D1 typically shows four or
less discreet steps. Higher generation dendrimers show increasing numbers
of smaller bleaching steps and increased overall lifetime. As a result, the tra-
jectories change from stepwise for lower generation dendrimers to continu-
ous for higher generation of dendrimers.
Figure 4. Intensity histograms reveal that fluorescein peaks at 125 cps,
whereas the dendrimer D7 reaches a maximum at 10000 cps, a near 80
times enhancement in brightness. Other dendrimers are also brighter than
the high-quantum-yield dye fluorescein.
that individual dyes are less bright in the larger dendrimers,
consistent with relatively lower quantum yields discussed
above. Similar to fluorescent nanoparticles and in contrast to
QDs, fluorescent dendrimers do not completely blink on and
off. Dendrimer brightness is affected by individual dye photo-
bleaching, photoblinking, and typically exhibits exponential
decay where higher generation dendrimers exhibit longer
decay lifetimes.
quantum yields. Promisingly, increasing the fluorophore spac-
ing has been shown to augment fluorescent intensity.[19] When
a TEG spacer was introduced between F2 chromophores, the
single-molecule fluorescence of the derivatized fourth-genera-
tion dendrimer D7 (f ꢀ7%), with a 16 nm hydrodynamic di-
fl
ameter, has a modal brightness of 104 cps, 74 times brighter
than a single fluorophore (Figures 3 and 4). Individual QDs
seem to have reached a fundamental ceiling at 20 times
brighter than single-molecule fluorophores, but dendrimers,
which have the molar mass equivalent to 5 nm QDs, can deliv-
er much brighter intensity than the ceiling brightness of QDs.
Time-dependent fluorescence trajectories from single den-
drimers were monitored by using diffraction-limited excitation,
and representative traces for each generation are shown in
Figure 5. Upon shutter opening, a single molecule having the
tetrachloro perylene diimide skeleton typically exhibits discreet
on–off photoblinking, with eventual photobleaching to the
background noise level before shutter closing (Figure 5: 1e). In
contrast, first-generation D1, with four dye units, exhibits at
least four discreet photo-blinking or photobleaching steps.
With increasing generation number, the number of apparent
steps increases until D8 (G=8), which approaches a smooth
decay. Interestingly, the size of the step decreases and initial
intensity increases sub-linearly with increasing G, suggesting
The big picture is that these fluorescent dendrimers have
hydrodynamic diameters, as determined by DLS, similar to
functionalized QDs and smaller than fluorescent nanoparticles,
with similar or higher fluorescent intensity.[20] Fluorescent den-
drimers are smaller than fluorescent polymers or silica nano-
particles but can achieve similar brightness at smaller sizes.
This is because dendrimers allow packing of the maximum
density of fluorophores into the smallest space with controlled
intramolecular fluorophore spacing.
Conclusion
Exceptionally bright, fluorescent dendrimers have been synthe-
sized by using automated solid-state synthesis and phosphora-
midite chemistry, thus opening potential alternatives for highly
fluorescent probes for biological applications.[21] Thinning of
the reactive sites on the solid support was found to be crucial
Chem. Eur. J. 2014, 20, 12165 – 12171
www.chemeurj.org
12169
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
ever, jumps nearly 300% with the addition of the first trebler, fol-
lowed by stable response for addition of another PEG spacer. This
trend continues for four generations, so that the final trityl re-
sponse is higher than the initial response from the dC CPG
column. The response increases by ~275% (300% ideal) for each
trebling step, and remains near 100% for nondendritic coupling.
Similar exponential growth continues even up to the 8th genera-
tion in the synthesis of D8.
for efficient dendrimer growth, and multiplier phosphorami-
dites with long, flexible arms give the highest coupling effi-
ciency. It was also confirmed that increased intramolecular flu-
orophore spacing results in brighter dendrimers.
The family of dendrimers discussed in this report demon-
strates that a single dendrimer can achieve intensities ꢀ70
times brighter than that of high-quantum-yield fluorescein.
Furthermore, dendrimers promise single-functionalization
points at the root, a unique property that QDs and fluorescent
nanoparticles do not have. Therefore, this class of fluorescent
dendrimers will be continuously optimized, aiming for small
hydrodynamic radius, narrow polydispersity, and limited scat-
tering profiles for ultra-bright molecular probes.
CPG size, coupling time, and scale
All dendrimers (summarized in Table 1) were synthesized at the
0.2 mmol scale by using CPG pore sizes of 500, 1000, or 2000 ꢂ.
Dendrimer growth efficiency did not seem to correlate with CPG
pore size for low generation dendrimers (G=1–3), whereas the
highest generation dendrimers (G=4 or 8) were only synthesized
using 2000 ꢂ pores. Repeated, extended coupling times were nec-
essary to maximize coupling yields of nonstandard phosphorami-
dites. Thus, sub-100% coupling is attributed mostly to steric hin-
drance of crowded reactive groups.
Experimental Section
Dendrimer syntheses
Dendrimers D1–D8 were synthesized by using an 8909 Expedite
Nucleic Acid Synthesis System. All reagents were freshly prepared
except for tetrazole activator solution, which was purchased from
Fisher Scientific. The solid supports or controlled pore glass (CPG)
columns were purchased from Glen Research at the 0.2 mmoles
scale with 500 ꢂ, 1000 ꢂ, or 2000 ꢂ pore sizes. Chromophoric phos-
phoramidites (F2, F3), spacer phosphoramidite (S1, S2), and multi-
plier phosphoramidite (M1, M2, M3, M4) with DMTr-protected hy-
droxyl groups for the chain growth were dissolved in CH2Cl2 at
60 mm. A custom coupling protocol for the phosphoramidite re-
agents was used in order to obtain the reported high yields.[10b,22]
Dendrimer growth on the CPG was either starting from a standard
3’-DNA base (A, T, C, or G) or a 3’-amino modifier (no DNA bases
attached to the final dendrimers). When required, oligo DNA se-
quences were attached to either the dendrimer root or its den-
drites or both. Once a spacer or dye was incorporated into the
growing chain, deprotection removed the trityl group, which re-
vealed the coupling yield from the previous step. Typically, the
coupling yields range from 95–100%. Similarly, the incorporation
of the dendritic unit reported an average yield of 92%. The dura-
tion of the solid-phase synthesis ranged from 160 s to 500 s de-
pending on the activity of the phosphoramidites. Repeating such
phosphoramidite coupling cycles produces 1st, 2nd, 3rd, 4th, and
8th generation dendrimers (D1–D8).
Dendrimer purification
All dendrimers were deprotected by using concentrated NH4OH(aq)
,
which simultaneously cleaved the dendrimer from the CPG sup-
port, removed the cyanoethoxy and 3’-amino FMOC (9-fluorenyl-
methoxycarbonyl) protecting groups, and removed any DNA base
protecting groups. Dendrimers were synthesized in either DMTr-On
or DMTr-Off mode. The DMTr-Off dendrimers were desalted by
using standard reverse phase C8 oligomer purification columns
(OPCs), easily removing unwanted salts and protecting group resi-
dues, whereas DMTr-On dendrimers were purified by using the
standard OPC purification protocol, which additionally removes
capped PEG chains owing to thinning. Even if coupling yields of
95–98% are achieved, the resulting hyperbranched polymer mix-
ture will still exhibit relatively low polydispersity and uniformly
bright single-molecule fluorescence.
Single-molecule spectroscopy
Single-molecule microscopy and spectroscopy were used to mea-
sure dendrimer brightness and photostability. The aqueous den-
drimers were first spin-coated onto a cover slip in order to allow
consistent CCD bright-field measurement of immobilized dendrim-
ers as well as APD raster-scan images and time-trace analysis of dif-
fraction-limited spots. Representative 10ꢁ10 mm CCD images were
collected with 2-second integration time for dendrimers D1–D8
and compared with fluorescein and monomer 1e (see the Support-
ing Information, Scheme S1). Statistical fluorescence histograms are
provided in Figure 4. The monomers and the first-generation den-
drimer, D1, are not much brighter than inorganic fluorescent im-
purities within the cover slip or other organic surface impurities.
However, for the second-generation dendrimer, D2, and higher,
there is a measurable increase in molecular brightness and signal-
to-noise ratio (SNR), with the fourth-generation dendrimer, D7, and
eighth-generation dendrimer, D8, being exceptionally bright. The
zoomed-in 1ꢁ1 mm CCD images of individual emitters from the
same samples demonstrate that these emitters have diffraction-
limited spots.
Trityl monitoring
One advantage of the phosphoramidite approach is that trityl
monitoring allows facile reporting of the reaction yield for each
generation. On the 8909 synthesizer, the built-in flow-through trityl
monitor provides qualitative monitoring of step-wise synthesis.
More importantly, the flow-through detector provides a non-linear
response owing to detector saturation from order-of-magnitude
concentration changes during dendrimer growth, as shown for
dendrimer D9 in Figure 1. Additionally, some phosphoramidites de-
tritylate react faster than others, resulting in concentration spikes
and thus response fluctuations of up to 20% for identical trityl
concentrations. However, manually collecting and measuring the
absorbance from the trityl byproduct after each step allows quanti-
tative yield determination, as depicted in Figure 1 for dendrimer
D9. Upon calibration to the manual trityl yields, it is apparent the
flow-through detector displays a quasi-logarithmic response. The
trityl yield decreases for the first three “thinning” steps, whereas
a PEG spacer provides a stable trityl yield. The trityl response, how-
Chem. Eur. J. 2014, 20, 12165 – 12171
www.chemeurj.org
12170
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
[11] a) S. K. Yang, X. H. Shi, S. Park, S. Doganay, T. Ha, S. C. Zimmerman, J.
Am. Chem. Soc. 2011, 133, 9964–9967; b) T. Heek, C. Fasting, C. Rest, X.
Zhang, F. Wurthner, R. Haag, Chem. Commun. 2010, 46, 1884–1886;
c) B. X. Gao, H. X. Li, H. M. Liu, L. C. Zhang, Q. Q. Bai, X. W. Ba, Chem.
Commun. 2011, 47, 3894–3896.
Acknowledgements
We acknowledge the support of the National Science Founda-
tion (CHE-1212429).
[12] a) C. Kohl, T. Weil, J. Q. Qu, K. Mullen, Chem. Eur. J. 2004, 10, 5297–
5310; b) J. Q. Qu, C. Kohl, M. Pottek, K. Mullen, Angew. Chem. 2004, 116,
1554–1557; Angew. Chem. Int. Ed. 2004, 43, 1528–1531; c) R. O.
Marcon, J. G. dos Santos, K. M. Figueiredo, S. Brochsztain, Langmuir
2006, 22, 1680–1687; d) F. Yukruk, A. L. Dogan, H. Canpinar, D. Guc,
E. U. Akkaya, Org. Lett. 2005, 7, 2885–2887.
[13] D. Aigner, S. M. Borisov, I. Klimant, Anal. Bioanal. Chem. 2011, 400,
2475–2485.
[14] M. S. Shchepinov, I. A. Udalova, A. J. Bridgman, E. M. Southern, Nucleic
Acids Res. 1997, 25, 4447–4454.
[15] a) J. S. Klos, J. U. Sommer, Macromolecules 2009, 42, 4878–4886; b) S.
Zou, H. Schonherr, G. J. Vancso, Angew. Chem. 2005, 117, 978–981;
Angew. Chem. Int. Ed. 2005, 44, 956–959.
[16] a) X. Feng, D. Taton, R. Borsali, E. L. Chaikof, Y. Gnanou, J. Am. Chem.
Soc. 2006, 128, 11551–11562; b) I. Willerich, H. Ritter, F. Grohn, J. Phys.
Chem. B 2009, 113, 3339–3354.
[17] a) J. J. Han, A. D. Shaller, W. Wang, A. D. Q. Li, J. Am. Chem. Soc. 2008,
130, 6974–6982; b) J. J. Han, W. Wang, A. D. Q. Li, J. Am. Chem. Soc.
2006, 128, 672–673; c) D. H. Hu, Z. Y. Tian, W. W. Wu, W. Wan, A. D. Q.
Li, J. Am. Chem. Soc. 2008, 130, 15279.
Keywords: dendrimers
phosphoramidite · solid-phase synthesis
·
fluorescence
·
nanoprobes
·
[1] V. J. Pugh, Q. S. Hu, L. Pu, Angew. Chem. 2000, 112, 3784–3787; Angew.
Chem. Int. Ed. 2000, 39, 3638–3641.
[2] C. Kojima, B. Turkbey, M. Ogawa, M. Bernardo, C. A. S. Regino, L. H.
Bryant, P. L. Choyke, K. Kono, H. Kobayashi, Nanomedicine 2011, 7,
1001–1008.
[3] a) D. Luo, K. Haverstick, N. Belcheva, E. Han, W. M. Saltzman, Macromole-
cules 2002, 35, 3456–3462; b) Y. J. Tsai, C. C. Hu, C. C. Chu, I. Toyoko,
Biomacromolecules 2011, 12, 4283–4290.
[4] a) Y. E. Kurtoglu, R. S. Navath, B. Wang, S. Kannan, R. Romero, R. M.
Kannan, Biomaterials 2009, 30, 2112–2121; b) L. Albertazzi, B. Storti, L.
Marchetti, F. Beltram, J. Am. Chem. Soc. 2010, 132, 18158–18167.
[5] A. Arakaki, K. Shibata, T. Mogi, M. Hosokawa, K. Hatakeyama, H. Gomyo,
T. Taguchi, H. Wake, T. Tanaami, T. Matsunaga, T. Tanaka, Polym. J. 2012,
44, 672–677.
[6] a) D. J. Liu, S. De Feyter, M. Cotlet, A. Stefan, U. M. Wiesler, A. Herrmann,
D. Grebel-Koehler, J. Q. Qu, K. Mullen, F. C. De Schryver, Macromolecules
2003, 36, 5918–5925; b) T. Weil, U. M. Wiesler, A. Herrmann, R. Bauer, J.
Hofkens, F. C. De Schryver, K. Mullen, J. Am. Chem. Soc. 2001, 123,
8101–8108; c) D. J. Liu, S. De Feyter, M. Cotlet, U. M. Wiesler, T. Weil, A.
Herrmann, K. Mullen, F. C. De Schryver, Macromolecules 2003, 36, 8489–
8498; d) J. M. J. Frechet, Proc. Natl. Acad. Sci. USA 2002, 99, 4782–4787.
[7] a) I. Oesterling, K. Mullen, J. Am. Chem. Soc. 2007, 129, 4595–4605;
b) M. Cotlet, T. Vosch, S. Habuchi, T. Weil, K. Mullen, J. Hofkens, F. De
Schryver, J. Am. Chem. Soc. 2005, 127, 9760–9768.
[18] a) W. C. W. Chan, S. M. Nie, Science 1998, 281, 2016–2018; b) M. Bru-
chez Jr., M. Moronne, P. Gin, S. Weiss, A. P. Alivisatos, Science 1998, 281,
2013–2016.
[19] a) S. Masuo, T. Vosch, M. Cotlet, P. Tinnefeld, S. Habuchi, T. D. M. Bell, I.
Oesterling, D. Beljonne, B. Champagne, K. Mullen, M. Sauer, J. Hofkens,
F. C. De Schryver, J. Phys. Chem. B 2004, 108, 16686–16696; b) F. C. De
Schryver, T. Vosch, M. Cotlet, M. Van der Auweraer, K. Mullen, J. Hofkens,
Acc. Chem. Res. 2005, 38, 514–522; c) P. Tinnefeld, K. D. Weston, T.
Vosch, M. Cotlet, T. Weil, J. Hofkens, K. Mullen, F. C. De Schryver, M.
Sauer, J. Am. Chem. Soc. 2002, 124, 14310–14311.
[8] C. Minard-Basquin, T. Weil, A. Hohner, J. O. Radler, K. Mullen, J. Am.
Chem. Soc. 2003, 125, 5832–5838.
[9] D. Gçrl, X. Zhang, F. Wurthner, Angew. Chem. 2012, 124, 6434–6455;
Angew. Chem. Int. Ed. 2012, 51, 6328–6348.
[20] Z. Tian, A. D. Shaller, A. D. Q. Li, Chem. Commun. 2009, 180–182.
[21] Z. Y. Tian, A. D. Q. Li, Acc. Chem. Res. 2013, 46, 269–279.
[22] W. Wang, A. D. Q. Li, Bioconjugate Chem. 2007, 18, 1036–1052.
[10] a) A. D. Shaller, W. Wang, H. Y. Gan, A. D. Q. Li, Angew. Chem. 2008, 120,
7819–7823; Angew. Chem. Int. Ed. 2008, 47, 7705–7709; b) W. Wang, W.
Wan, H. H. Zhou, S. Q. Niu, A. D. Q. Li, J. Am. Chem. Soc. 2003, 125,
5248–5249; c) W. Wang, L. S. Li, G. Helms, H. H. Zhou, A. D. Q. Li, J. Am.
Chem. Soc. 2003, 125, 1120–1121.
Received: May 7, 2014
Published online on August 8, 2014
Chem. Eur. J. 2014, 20, 12165 – 12171
www.chemeurj.org
12171
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim