Modulating the therapeutic activity of nanoparticle delivered paclitaxel by manipulating the hydrophobicity of prodrug conjugates

Article abstract of DOI:10.1021/jm800002y

A series of paclitaxel prodrugs designed for formulation in lipophilic nanoparticles are described. The hydrophobicity of paclitaxel was increased by conjugating a succession of increasingly hydrophobic lipid anchors to the drug using succinate or diglycolate cross-linkers. The prodrugs were formulated in well defined block copolymer-stabilized nanoparticles. These nanoparticles were shown to have an elimination half-life of approximately 24 h in vivo. The rate at which the prodrug was released from the nanoparticles could be controlled by adjusting the hydrophobicity of the lipid anchor, resulting in release half-lives ranging from 1 to 24 h. The diglycolate and succinate cross-linked prodrugs were 1-2 orders of magnitude less potent than paclitaxel in vitro. Nanoparticle formulations of the succinate prodrugs showed no evidence of efficacy in HT29 human colorectal tumor xenograph models. Efficacy of diglycolate prodrug nanoparticles increased as the anchor hydrophobicity increased. Long circulating diglycolate prodrug nanoparticles provided significantly enhanced therapeutic activity over commercially formulated paclitaxel at the maximum tolerated dose.

Full text of DOI:10.1021/jm800002y

3288
J. Med. Chem. 2008, 51, 3288–3296
Modulating the Therapeutic Activity of Nanoparticle Delivered Paclitaxel by Manipulating the
Hydrophobicity of Prodrug Conjugates
Steven M. Ansell,*,† Sharon A. Johnstone,† Paul G. Tardi,† Lily Lo,† Sherwin Xie,† Yu Shu,† Troy O. Harasym,†
Natashia L. Harasym,† Laura Williams,† David Bermudes,† Barry D. Liboiron,† Walid Saad,‡ Robert K. Prud’homme,‡ and
Lawrence D. Mayer†
Celator Pharmaceuticals Corporation, 1779 West 75th AVenue, VancouVer, B.C., V6P 6P2, Canada, and Department of Chemical Engineering,
Princeton UniVersity, Princeton, New Jersey
ReceiVed January 2, 2008
A series of paclitaxel prodrugs designed for formulation in lipophilic nanoparticles are described. The
hydrophobicity of paclitaxel was increased by conjugating a succession of increasingly hydrophobic lipid
anchors to the drug using succinate or diglycolate cross-linkers. The prodrugs were formulated in well
defined block copolymer-stabilized nanoparticles. These nanoparticles were shown to have an elimination
half-life of approximately 24 h in vivo. The rate at which the prodrug was released from the nanoparticles
could be controlled by adjusting the hydrophobicity of the lipid anchor, resulting in release half-lives ranging
from 1 to 24 h. The diglycolate and succinate cross-linked prodrugs were 1-2 orders of magnitude less
potent than paclitaxel in vitro. Nanoparticle formulations of the succinate prodrugs showed no evidence of
efficacy in HT29 human colorectal tumor xenograph models. Efficacy of diglycolate prodrug nanoparticles
increased as the anchor hydrophobicity increased. Long circulating diglycolate prodrug nanoparticles provided
significantly enhanced therapeutic activity over commercially formulated paclitaxel at the maximum tolerated
dose.
Introduction
compatible with lipid based delivery systems, it is possible to
adjust the properties of two disparate drugs such that their
effective release rates in vivo are matched. Micelles or lipophilic
nanoparticle carriers can be used to suspend these prodrugs in
aqueous environment because the individual drugs themselves
are otherwise insoluble.
In recent years there has been increasing interest in chemo-
therapeutic treatment regimes involving multiple drugs. Recent
evidence from our laboratory has revealed that many drug
combinations act synergistically when exposed at appropriate
ratios and antagonistically at other ratios in cell based studies.1–4
One of the problems encountered when translating these findings
to in vivo studies is that the exposed drug ratio changes over
time because of the different pharmacokinetic behavior of the
individual drugs when administered in a conventional aqueous
based cocktail. To address this issue, we have developed
liposome-based delivery systems that maintain drug ratios at
synergistic levels in vivo through coordinated release of the
drugs.1–4 Although liposomes are adequate delivery vehicles for
hydrophobic compounds, they are predominantly designed for
water soluble drugs. Hydrophobic drugs typically dissolve in
the liposome bilayer where they are subject to facile exchange
processes when exposed to hydrophobic sinks such as lipopro-
teins or cells in the blood compartment after in vivo administra-
tion.5 As a result, development of alternative delivery technolo-
gies is being pursued that are compatible with both hydrophobic
and hydrophilic drugs and allow the simultaneous control of
the rates at which drugs are made bioavailable.
Parameters that are likely to affect the in vivo availability of
a drug when optimizing the design of such systems include (1)
the plasma elimination of the carrier particle, (2) the partitioning
rate of the drug out of the particle, and (3) the hydrolysis rate
of the prodrug. In an ideal system, the particles should remain
intact upon iv administration and be cleared relatively slowly
from the central blood compartment. In addition, the prodrug
should undergo rapid hydrolysis upon partitioning out of the
delivery vehicle, preferably through enzymatic means rather than
chemical hydrolysis to enhance stability in the formulation. The
rate limiting process affecting drug availability is then the
partitioning rate of the prodrug from the particle to the plasma
compartment. A series of prodrugs based on paclitaxel as a
model agent were investigated here in order to validate this
general approach to achieving pharmacokinetic control in vivo.
Paclitaxel is a widely used chemotherapeutic agent for treating
a range of carcinomas. The clinical material is formulated in a
50:50 mixture of ethanol and Cremophor EL (polyoxyethyl-
eneglyceroltriricinoleate) and is diluted with buffer prior to
administration. There are many reports in the literature describ-
ing attempts to improve the formulation of paclitaxel using
micelles, liposomes, or emulsions.7 In almost all cases, however,
it is clear from the reported pharmacokinetic data that while
these carriers formulate paclitaxel, they do not act as true
delivery vehicles in vivo because the drug rapidly partitions
out of the carrier with half-lives on the order of minutes. One
reported exception appears to be a formulation of paclitaxel in
micelles comprising poly(ethylene glycol)-poly(ꢀ-(4′-phenylbu-
tanyl)aspartate).8
The drug delivery approach applied here was to combine two
well-known concepts, namely, the use of prodrugs and the
utilization of micellar or nanoparticle delivery vehicles to
facilitate pharmacokinetic control of hydrophobic drugs such
as paclitaxel. The goal of most prodrug technologies is typically
to make hydrophobic drugs more hydrophilic for increased
solubility in an aqueous environment.6 However, when such
drugs are made more hydrophobic and consequently more
* To whom correspondence should be addressed. Phone: (604) 675-2120.
Fax: (604) 708-5883. E-mail: sansell@celatorpharma.com.
†
Celator Pharmaceuticals Corp.
Princeton University.
‡
10.1021/jm800002y CCC: $40.75
2008 American Chemical Society
Published on Web 05/09/2008

Nanoparticle DeliVered Paclitaxel Prodrugs
Journal of Medicinal Chemistry, 2008, Vol. 51, No. 11 3289
Many attempts have been made to produce functional
lipophilic paclitaxel prodrugs to improve the performance of
paclitaxel or to address formulations issues associated with the
drug. These include conjugates with phospholipids,9,10 choles-
terol,11 R-bromo fatty acids,12,13 oleic acid,14,15 fullerene,16 and
docosanhexaneoic acid.17 These prodrugs were formulated in
lipid vehicles, such as liposomes,9,13,16 oil emulsions,14,15 or
micelles.11,12,17 Most of these reports claim improved efficacy
over paclitaxel in preclinical models; however, in most cases
they either provide no information on plasma drug elimination
or present data focusing on the terminal drug elimination phase
rather than the early distribution phase. Drug elimination
information during the first 24 h after administration is the period
of most significant interest from a tumor delivery perspective
because of the enhanced permeability and retention (EPR)
phenomenon observed with particulate carriers, including mi-
celles and nanoparticles.18–21
confined volume impinging jet (CVIJa) mixer. This method
involved rapid mixing of water and a miscible solvent containing
the formulation components in a confined space using carefully
controlled flow rates. Under conditions where the rate of mixing
exceeds the rate of precipitation, these devices have been shown
to achieve homogeneous particle formation kinetics, resulting
in minimum sized particles for a particular formulation
composition.23,24
Formulations were composed of a prodrug, a co-lipid, and
an amphipathic stabilizer in a weight ratio of 1:1:2 at a
concentration of 40 mg per mL of solvent. The use of higher
initial concentrations (>40 mg/mL) often led to the generation
of large particles that compromised downstream handling of
the preparation. The co-lipids used in this study were either
R-tocopherol succinate (VES) or POPC, which facilitated
processing. VES was replaced by POPC in later formulations
because VES precipitated on long-term storage at 4 °C.
Poly(ethylene glycol)-b-polystyrene (2kPS3k or 2.5kPS3k) was
used as the amphipathic stabilizer. The application of a number
of other stabilizers has been examined in formulations of this
type and will be reported elsewhere. By use of the flash
precipitation procedure, the mean particle diameter ranged from
10 to 20 nm for the prodrug/VES/2kPS3k nanoparticles and
from 20 to 30 nm for the prodrug/POPC/2kPS3k nanoparticles.
In this work, the development of a series of lipophilic
paclitaxel prodrugs and associated micellar/nanoparticle for-
mulations is reported. Particulate delivery vehicles with pro-
longed circulation half-lives are described where the release of
drug is modulated by manipulating the degree of lipid anchor
hydrophobicity and the lability of the prodrug cross-linkers. The
efficacy of the prodrugs in vivo is shown to be dependent on
the nature of the linkage and the relative partitioning rate of
the lipid anchor.
Chemical and Physical Stabilitsy of Nanoparticle
Formulations. The stability of nanoparticle formulations was
monitored by HPLC after long-term storage at 4 °C. Analysis
of 4-6 formulated with POPC/2kPS3k (1:1:2) in unbuffered
300 mM sucrose showed less than 5% free paclitaxel present
after 11 weeks of storage. Free drug at that level was shown to
have no antitumor activity in control efficacy experiments (data
not shown). Accordingly, stability was considered adequate for
the in vivo experiments carried out in these investigations.
Nanoparticle formulations stored in either sucrose solution
or water at 4 °C were found to be physically stable for 3 months
by visual inspection, with the exception of 8 which formed
precipitates in some formulation compositions after 1-2 weeks
of storage. Formulations using VES as the colipid also led to
the slow formation of colloidal or crystalline precipitates, likely
due to VES partitioning to the aqueous phase followed by
subsequent precipitation outside the nanoparticle. VES formula-
tions were therefore only suitable for short-term studies and
not for longer term experiments such as efficacy evaluations.
The physical stability of formulations was investigated by
examining changes in particle size after vortexing prodrug/co-
lipid/2.5kPS3k (1:1:2) nanoparticles in water, 150 mM saline,
or 300 mM sucrose (Table 1). These formulations consisted
primarily of nanoparticles with diameters between 20 and 30
nm. Particles that were sensitive to mechanical stress underwent
aggregation or phase separation when subjected to these
conditions, resulting in the formation of larger particles in the
sample. The increase in particle size was monitored by using
dynamic light scattering to measure the Z average (the intensity
weighted mean size of the particle distribution) value for samples
before and after vigorous vortexing. The change in the Zave value
was used as surrogate readout for the formation of aggregates
or precipitates resulting from mechanical instability, where a
higher value for ∆Zave indicated a higher level of aggregation
Results
Prodrug Synthesis. The prodrugs used in this study were
based on the generalized structure paclitaxel-linker-lipid. The
lipid anchors used were lipophilic alcohols, and these anchors
were conjugated to paclitaxel using di-acid cross-linkers. This
approach allowed both the anchor and the cross-linker to be
varied independently in order to investigate the effects of
changes in hydrolysis rates and partitioning behavior on the
pharmacokinetics and efficacy for nanoparticle formulations of
these prodrugs.
A range of commonly available lipid alcohols were used as
the anchor component (Scheme 1), including R-tocopherol (1
and 3), oleyl alcohol (2 and 4), octadecanol (5), cosanol (6),
docosanol (7), cholesterol (8), and 1,2-dimyristoyl-sn-glycerol
(9). The lipids were conjugated to the cross-linker by treatment
with the corresponding cyclic anhydride, followed by condensa-
tion of the anchor-linker product with paclitaxel using diiso-
propylcarbodiimide (DIPC) in the presence of N,N-4-dimeth-
ylaminopyridine (DMAP). Succinic acid (1, 2) and diglycolic
acid (3-9) were used as the cross-linkers. The 3-oxa moiety of
the latter was intended to increase the susceptibility of the 2′-
acyl group of the cross-linker to hydrolysis relative to the
succinate analogues.
The 2′-acyl paclitaxel derivatives were prepared selectively
by exploiting the difference in reaction rates between the
paclitaxel 2′- and 7-hydroxyl groups.22 Under the reaction
conditions used here, the majority of paclitaxel was consumed
before significant levels of the 2′,7-diacyl product were gener-
ated, as monitored by TLC. Column chromatography was used
to remove unreacted paclitaxel, the 2′,7-diacyl product, and other
impurities in the crude reaction mixture. Purity and identity of
the final products were confirmed by HPLC and NMR analysis,
respectively.
a Abbreviations: VES, vitamin E succinate; POPC, 1-palmitoyl-2-oleoyl-
sn-glycero-3-phosphocholine; CVIJ, confined volume impinging jets;
2kPS3k, poly(ethylene glycol)2000-b-polystyrene3000; 2.5kPS3k, poly(ethylene
glycol)2500-b-polystyrene3000; 3H-CHE, tritiated cholesterol hexadecyl ether;
TLC, thin layer chromatography; DIPC, diisopropylcarbodiimide; DMAP,
N,N-4-dimethylaminopyridine.
Nanoparticle Formulation. The prodrugs were formulated
in lipid-based nanoparticles by flash precipitation using a

3290 Journal of Medicinal Chemistry, 2008, Vol. 51, No. 11
Scheme 1. Synthesis of Lipophilic Paclitaxel Prodrugs
Ansell et al.
or precipitation (Table 1). Nanoparticles with POPC as the co-
lipid were found to be significantly more stable in the presence
of salt than those prepared with VES or in the absence of a
co-lipid (∆Zave ) 24, 372, and 774 nm, respectively, for
formulations of 5). In most cases mechanical stability was
significantly improved in the presence of 300 mM sucrose
compared to water. Sucrose samples where ∆Zave was in the
range 0-40 nm had very little or no micrometer-sized material
(<2% of the sample). Some prodrug formulations (3 and 9)
that generated higher levels of aggregation in water and saline
under these conditions (∆Zave ) 100-200 nm) were stable in
sucrose (∆Zave ) 0-40 nm). The conclusion of the study was
that the best stability was obtained using POPC co-lipids in
nanoparticles formulated in 300 mM sucrose.
Similar results were obtained when samples were subjected
to normal processing steps required for large sample prepara-
tions. Nanoparticles formulated in water or with either VES or
no co-lipid were not stable when exposed to mechanical stress
and exhibited significant precipitate formation during concentra-
tion steps using diafiltration. Stability while concentrating
formulations could be improved by diluting the nanoparticle
solution with sucrose for a final concentration of 300 mM prior
to concentration. Use of POPC as the co-lipid provided further
improvement of physical stability during storage and postfor-
mation processing, consistent with the results of the earlier
stability studies. All of the concentrated samples used in this
work subsequently were based on POPC as a co-lipid and were

Nanoparticle DeliVered Paclitaxel Prodrugs
Journal of Medicinal Chemistry, 2008, Vol. 51, No. 11 3291
Table 1. Physical Stability of Prodrug Formulationsa
Table 2. Inhibition of Human Tumor Cell Lines Using Prodrug/POPC/
2kPS3k (1:1:2) Formulationsa
∆Zave (nm)
IC50 ( SD (nM)
diameter
(nm)
150 mM
saline
300 mM
sucrose
prodrug
co-lipid
water
compd
A2780
MCF-7
2
3
4
5
6
7
8
9
5
5
POPC
POPC
POPC
POPC
POPC
POPC
POPC
POPC
VES
26
24
22
21
23
23
23
23
7
20
125
15
20
21
18
47
167
3
109
43
55
25
24
26
24
42
171
372
774
10
42
5
7
36
0
41
0
10
71
paclitaxel
2.4 ( 0.3
192 ( 18
3.8 ( 0.6
158 ( 72
1
2
3
4
5
6
7
8
9
75.1 ( 3.9
25.3 ( 1.3
28.4 ( 1.4
20.9 ( 10.9
29.0 ( 1.3
29.5 ( 5.2
27.0 ( 1.1
34.7 ( 1.3
67.0 ( 5.9
13.4 ( 2.4
15.9 ( 0.7
15.1 ( 1.5
16.6 ( 4.5
23.5 ( 1.3
24.1 ( 0.6
32.2 ( 1.3
28
a All formulations were prepared as prodrug/co-lipid/2.5kPS3k (1:1:2)
nanoparticles in water. Particle diameter is the volume weighted average
as determined with a Malvern Nano-ZS particle sizer. Stability was assessed
by measuring the change in the Zave (∆Zave) of samples diluted with
equivolumes of water, 300 mM saline, or 600 mM sucrose after vortexing
for 30 s.
a The ovarian A2780 and breast MCF-7 human cell lines were evaluated
for viability 72 h after drug addition using an MTT assay as outlined in the
Experimental Section. IC50 values were determined for each of the three
replicate studies with standard deviations (SD) presented.
prodrugs. For example, in the A2780 cell line the IC50 values
for the succinate prodrug 1 and diglycolate prodrug 3 were 192
and 25.3 nM, respectively, compared to 2.4 nM for paclitaxel.
Likewise, the succinate 2 and diglycolate 4 had IC50 values of
75.1 and 28.4 nM, respectively. These results are consistent with
literature reports of related compounds, where IC50 values in
the MCF-7 breast carcinoma cell line for paclitaxel, 2′-O-
hexadecanoylpaclitaxel, and 2′-(2′′-bromohexadecanoyl)pacli-
taxel were reported as <1, 8600, and 70 nM, respectively.13
In Vivo Nanoparticle Plasma Elimination. Partitioning rates
for lipophilic drugs from lipid based particle formulations are
commonly determined by dialysis of the drug carrier system
against a bulk aqueous phase. The low aqueous solubility of
the drug under these circumstances results in a high probability
of the drug partitioning back into the carrier system rather than
eluting through the dialysis membrane, resulting in artificially
low apparent partitioning rates. Such an experiment does not
accurately reflect behavior observed for hydrophobic drugs in
vivo, where free drug can rapidly partition into the large
hydrophobic reservoir present (for example, cell membranes or
lipophilic proteins) rather than back into the delivery vehicle.5
Dialysis experiments therefore misrepresent the true delivery
capability of these formulations. The most accurate means of
determining true partitioning rates in vivo is to track both the
drug and the particle using a nonexchangeable marker. Conse-
quently it was necessary to identify such a marker and to confirm
that it reflected the particle pharmacokinetics.
Elimination studies tracking both the prodrug and the particle
were conducted to obtain a more accurate description of the
partitioning behavior of the prodrugs in vivo (Figure 1). Prodrug
elimination was followed by HPLC, while particle elimination
rates were determined by labeling nanoparticles with tritiated
cholesterylhexadecyl ether (3H-CHE). This label is widely used
to track liposomes in vivo because it is regarded as a
nonexchangeable, nonmetabolizable lipid marker in those
systems. It was expected that the hydrophobicity and low level
of hydration of this molecule would result in low partitioning
rates from the nanoparticles used in this study and therefore
would provide an assessment of particle fate, provided that
particle integrity was maintained.
prepared in 300 mM sucrose. Under these conditions precipitate
formation was negligible up to drug concentrations of 8 mg/
mL.
The diglycolate linkage was found to be more susceptible to
hydrolysis over the succinate linkage, as may be predicted on
the basis of the relative lability of the ester linkages. The
increased reactivity of the diglycolate linkage posed an ad-
ditional difficulty for accurate analysis of prodrug in biological
samples. Methanol/acetonitrile 1:4 (v/v) was effective in near-
complete precipitation of plasma proteins and liberation of the
prodrug from the nanoparticle. However, a steady increase of
free paclitaxel over time was observed after processing plasma
samples, indicating hydrolysis during matrix workup and HPLC
analysis. The diglycolate prodrug 6 was completely hydrolyzed
after 9 h, while the succinate prodrug 2 was reduced to
approximately 25% after 68 h. Hydrolysis was not observed
when the samples were buffered with acetate at pH 4. Unbuf-
fered samples of 6/POPC/2kPS3k in the absence of plasma did
not exhibit this hydrolysis, indicating that the time-dependent
hydrolysis of prodrug in these formulations was most likely due
to the presence of a nonprecipitated plasma component. This
result suggests that the prodrug is stable when it is in the particle
and that hydrolysis in biological media takes place after release
of the prodrug.
In Vitro Activity. Evaluation of prodrug cytotoxicity was
carried out using A2780 ovarian and MCF-7 breast human tumor
cells treated with prodrug/POPC/2.5kPS3k (1:1:2) formulations
prepared in distilled water (Table 2). Control formulations
containing no prodrug confirmed that other nanoparticle com-
ponents did not contribute to the growth inhibition of the cells.
In addition, 1 was evaluated using a number of different
nanoparticle formulations (1/VES/2kPS3k, 1/2kPS3k, and 1/cre-
mophor) to show that the observed activity was independent of
the type of formulation used.
Succinate-linked prodrugs (1 and 2) were found to be
approximately 30- to 100-fold less potent than free paclitaxel
against A2780 and MCF-7 cells. The decrease in observed
activity was likely the result of resistance of the succinate group
to hydrolysis. Subsequent prodrugs (3-9) were prepared with
the diglycolate linker moiety, which was expected to be more
susceptible to hydrolysis because of the presence of the 3-oxa
group. The diglycolate prodrugs all showed similar potency in
growth inhibition assays (50% growth inhibition concentration
values, IC50, of ∼20 nM) and were an order of magnitude less
potent than paclitaxel positive controls. The succinate analogues
were always less potent than the corresponding diglycolate
A number of experiments were carried out to determine the
validity of the above assumptions and to approximate the particle
plasma elimination rate. Formulations comprising 1/VES/
2kPS3k (1:1:2) and 2/VES/2kPS3k (1:1:2) were labeled with
3
trace amounts of H-CHE, and plasma levels of both the label
and prodrug were monitored in Foxn1nu mice (Figure 1A). The
3H-CHE label cleared at the same rate in both groups, with a

3292 Journal of Medicinal Chemistry, 2008, Vol. 51, No. 11
Ansell et al.
Figure 2. Elimination of prodrug formulated as prodrug/POPC/2kPS3k
(1:1:2; w/w) formulations and administered intravenously to athymic
nude Foxn1nu mice at a dose of 7 mg/kg (n ) 3/time point). Drug
concentrations were determined by HPLC analysis of plasma isolated
at various time points. The prodrugs used were 1 (b), 2 (O), 3 (]), 4
(9), 5 (0), 6 (2), 7 (4), 8 (1), and 9 (3). Error bars represent standard
deviation (n ) 3).
stabilizer poly(ethylene glycol)/polystyrene ratio had little effect
on the elimination of 1 relative to that seen in Figure 1 (data
not shown).
Elimination of 1 in these formulations was found to be
independent of co-lipid and stabilizer composition and cor-
responded to the loss of the H-CHE marker over time. These
3
Figure 1. (A) Elimination of 1 (O) and 2 (b) in prodrug/VES/2kPS3k
(1:1:2; w/w) formulations labeled with 3H-CHE (0 and 9, respectively)
in athymic nude Foxn1nu mice dosed at 35 mg drug/kg doses (n )
3/time point). Drug concentrations were determined by HPLC analysis
of plasma recovered at various time points. (B) Relative drug retention
was determined using the ratio of drug/3H-CHE in individual mice for
1 and 2 systems (F and π, respectively). Error bars represent standard
deviation (n ) 3).
results suggest that the 3H-CHE label remains with the particle
throughout the pharmacokinetic experiment, independent of
changes in the composition of the particle. The circulation half-
lives for these particles were notably long (approximately 24 h),
an order of magnitude longer than has been reported for micelles
prepared using other polymers at these concentrations25 and
comparable to that observed with long-circulating liposomes.26
half-life of approximately 24 h. Prodrug 1 cleared at ap-
proximately the same rate as H-CHE in this system, while 2
Effect of Prodrug Partitioning on Plasma Drug Eli-
mination. The prodrug panel was screened using prodrug/
POPC/2kPS3k (1:1:2) nanoparticles in Foxn1nu mice to deter-
mine elimination rates for the different prodrug conjugates
(Figure 2). Prodrug circulation levels appeared to be dependent
primarily on the prodrug partitioning rate and consequently
indirectly reflect that rate, since the particles are sufficiently
stable to remain intact over the course of the experiment.
Conjugates were formulated at a concentration of 0.7 mg/mL.
Variations in concentration did not appear to affect the particle
elimination rate, since the relative rates of plasma elimination
of 1/VES/2.5kPS1.6k (1:1:2) were independent of dose over
the range 3-25 mg/kg (data not shown).
3
was eliminated significantly faster. Comparison of prodrug/3H-
CHE ratios (Figure 1B) demonstrated that the ratio of 1 to
particle label was virtually unchanged over the course of the
experiment, suggesting that the partitioning half-life was
significantly longer than the duration of the experiment. Prodrug
2 decreased relative to 3H-CHE at a rate consistent with a
partitioning half-life of approximately 4 h. The data in the
experiment could be fitted to a four-parameter double-
exponential decay implying that the partitioning rate constant
changed over time, presumably in response to local environ-
mental changes as drug partitioned out of individual particles.
In order to estimate the partitioning half-life for each prodrug,
it was assumed that 1 tracked the particle elimination rate based
on the results discussed above. The elimination of all other
prodrugs was then fitted to four-parameter double-exponential
decay curves and normalized against 1. The presence or absence
of prodrug in the formulation did not significantly affect the
elimination behavior of the particles, implying that changes in
composition over time as drug partitioned out did not affect
particle elimination rates. Prodrug elimination normalized to 1
therefore yielded a reasonable approximation of the relative
partitioning half-lives of the prodrugs in these formulations.
Prodrug 1 had a negligible partitioning rate over 24 h, whereas
2 in this composition partitioned out with a half-life of
approximately 1.7 h. Both of these prodrugs used succinate
cross-linkers and were significantly less active than paclitaxel
The effects of formulation composition on elimination were
3
determined by comparing the plasma elimination of 1 and H
3
CHE using 1/2kPS3k, 1/POPC/2kPS3k, and H-CHE labeled
1/VES/2kPS3k (1:1:2) formulations in vivo (data not shown).
For all three formulations both the label and prodrug elimination
rates were found to be similar to that observed for 1 in Figure
1, indicating little to no role of co-lipid on the rate of elimination
of 1 in these formulations. Additional elimination studies were
performed using 3H-CHE labeled 2kPS3k micelles and POPC/
2kPS3k (1:2) nanoparticles, neither of which contained prodrug
3
(data not shown). In both cases the H-CHE label was cleared
at approximately the same rate as observed for 1 in Figure 1.
Finally, studies using 1/VES/3kPS3k and 1/VES/2.5kPS1.6k
formulations demonstrated that moderate variations in the

Nanoparticle DeliVered Paclitaxel Prodrugs
Journal of Medicinal Chemistry, 2008, Vol. 51, No. 11 3293
Figure 4. Effect of dose on the efficacy of 7/POPC/2kPS3k nanopar-
ticles in the HT29 human colon carcinoma model. Athymic nude mice
were injected intravenously with formulations 14 days after tumor cell
inoculation. 7/POPC/2kPS3k (1:1:2) was administered at 69 µmol/kg
(3 paclitaxel equiv, 0), 46 µmol/kg (2 paclitaxel equiv, 4), 34 µmol/
kg (1.5 paclitaxel equiv, 3), 23 µmol/kg (1 paclitaxel equiv, ]), and
5.7 µmol/kg (0.25 paclitaxel equiv, 1) using a Q2D×5 schedule. Saline
(Q2Dx5, 9) was used as a negative control. Paclitaxel was administered
at 23 µmol/kg using a Q2D×5 (b) dosing schedule. Error bars represent
standard error of the mean (n ) 6). All treatments were initiated on
day 12 after tumor cell inoculation.
Figure 3. Relative efficacy of prodrugs formulated as prodrug/POPC/
2kPS3k (1:1:2; w/w) nanoparticles when administered to athymic nude
mice bearing HT29 human colon carcinoma xenographs. Drug was
administered at a dose of 36 µmol/kg using a Q4D×6 schedule
beginning on day 14. The prodrugs and controls used were saline (b),
1 (0), 2 (2), 4 (3), 5 ([), 6 (O), 7 (9). Error bars represent standard
error of the mean (n ) 6).
in vitro. The diglycolate versions of these compounds (3 and
4) showed accelerated elimination relative to their succinate
parents, with partitioning half-lives of approximately 13 and
1 h, respectively. This is consistent with an increased level of
hydration on partitioning into the aqueous phase due to the
presence of the 3-oxa group of the diglycolate linkage.
The two succinate linked prodrugs, 1 and 2, showed no
evidence of any therapeutic activity at the 34 µmol/kg dose.
The results correlated with the observation that these prodrugs
were 2 orders of magnitude less potent than paclitaxel in in
vitro cell growth inhibition assays (Table 2). The partitioning
half-lives of 2 and 1 differed widely, ranging from 1.7 h with
2 to effectively nonexchangeable on experimental time frames
with 1. These results indicate that even though the succinate-
linked species 2 is being made available to cells, the lower
hydrolysis rate and concomitant release of paclitaxel are
insufficient to provide a meaningful therapeutic response.
The four diglycolate linked species, 4-7, all resulted in
varying degrees of inhibition of tumor growth. In vitro growth
inhibition studies of these prodrugs had demonstrated that the
cytotoxicity of diglycolate-linked prodrugs was greater than their
succinate analogues (Table 2). Elimination data for these
formulations indicated that aliphatic chain length was related
to partitioning half-lives for the prodrugs, which were 1, 1.7,
6.5, and 10 h, respectively. This correlates well with the
observed efficacy trends, where the greatest antitumor response
is seen with the prodrug having the longest partitioning half-
life, namely, the docosanyl conjugate 7. Relative efficacy was
determined using the treatment induced delay in reaching a mean
tumor size of 500 mm3, yielding values of 3, 4, 15, and 17
days, respectively. On the basis of these results, further
investigations to optimize dose and scheduling using the
nanoparticle formulation of 7 were undertaken.
Effect of Dose on the Efficacy of 7. The activity of 7 in
POPC/2.5kPS3k nanoparticle formulations relative to paclitaxel
was determined by comparing the efficacy of doses ranging from
0.25 to 3 mol equiv of paclitaxel (Figure 4) in the HT29 tumor
model. The Q2D dosing regime was chosen because the greatest
efficacy for 7 and paclitaxel was observed on this schedule.
Sporadic deaths were observed with paclitaxel during the sixth
dose when administered at 20 mg/kg (23 µmol/kg), consistent
with reported literature values.12,17,27 The Q2D×5 schedule was
chosen for the study to avoid this apparent Cremophor related
toxicity. The maximum tolerated dose of 7 on the Q2D×5
schedule was approximately 3 paclitaxel equiv (69 µmol/kg).
The effect of small changes on the size of the lipid anchor
was determined using the series 4, 5, 6, and 7, which have oleyl
(∆C18), stearyl (C18), cosanyl (C20), and docosanyl (C22)
moieties, respectively. On the basis of the results in Figure 2,
we estimated the partitioning half-lives for these prodrugs to
be approximately 1, 1.7, 6.5, and 10 h, respectively. These
results indicate that prodrug release from the nanoparticles can
be regulated by adjusting the length of the alkyl anchor used.
The effect of other common lipid types on partitioning
kinetics was also examined. A cholesteryl diglycolate conjugate
(8) was retained considerably longer than the straight chain
aliphatic species, with an estimated partitioning half-life of
approximately 21 h. Unfortunately 8 was susceptible to phase
separation from nanoparticles on storage and was not fully
evaluated in later efficacy studies in this work. A 1,2-
dimyrsistoyl-sn-glycero-3-succinate derivative (9) was evaluated
in anticipation that it would be better retained than 7 since it
had two C14 anchors; however, it partitioned out more quickly,
with a half-life of 8.5 h. The higher partitioning rate was likely
due to increased levels of hydration on the glycerol backbone
during exchange out of the nanoparticle.
Efficacy of Paclitaxel Prodrug Conjugates in Nano-
particle Formulations. The relative efficacies of nanoparticle
formulations of 1-7 were determined in the HT29 human
colorectal xenograft solid tumor model using a 36 µmol/kg dose
on a Q4D×6 schedule (Figure 3). Saline was used as a negative
control. Earlier experiments conducted with submaximal doses
suggested that this dose would result in a range of responses
bracketing the behaviors of positive and negative control groups
and in this way would allow discernment of the relative
therapeutic potencies of the individual prodrug species. Prodrug
8 was not included in the study because of precipitation of drug
from formulations on storage. Prodrugs 3 and 9 were omitted
as well because their partitioning half-lives were similar to those
of 6 and 7.

3294 Journal of Medicinal Chemistry, 2008, Vol. 51, No. 11
Ansell et al.
Dosing at 4 paclitaxel equiv (92 µmol/kg) resulted in weight
loss in excess of 15% over the course of treatment.
paclitaxel prodrugs in nanoparticles. Paclitaxel partitions rapidly
out of Cremophor micelles on injection and distributes broadly
to lipophilic sinks it encounters in the blood compartment. Over
time the drug redistributes back into the blood compartment
before being cleared.29,30 Prodrug nanoparticles do not exhibit
this early distribution phase but rather are continually releasing
drug over time, the rate of which depends on the partitioning
rate of the prodrug component. The prodrug partitioning rate
was shown to be directly correlated to efficacy (Figures 2 and
3), and this rate was readily tunable by varying the lipid anchor
(and hence hydrophobicity) of the prodrug.
The tumor growth data were analyzed by two methods to
determine what dose of 7 would be equivalent to the paclitaxel
positive control. In the first method only the treatment groups
that resulted in tumor regression were considered. The delay in
reaching a tumor size of 125 mg was measured from the start
of treatment and then normalized against the delay observed
for the paclitaxel group. The relative delay was plotted against
paclitaxel dose equivalents of 7, yielding a data set that showed
a linear dose response up to the MTD of 7. The fitted curve
indicated that 7 achieved parity with paclitaxel at a dose of 1.75
dose equiv of paclitaxel. In the second analysis the delay in
reaching a tumor size of 300 mg was measured relative the saline
negative control and then normalized against the delay observed
for the paclitaxel group. Plotting the relative delay against
paclitaxel dose equivalents of 7 again yielded a linear dose
response that showed that parity with the paclitaxel group was
achieved with 1.65 paclitaxel equiv of 7. Importantly, when the
nanoparticle formulation of 7 was dosed at 3 molar paclitaxel
equiv (69 µmol/kg), reflecting its MTD, the antitumor activity
was significantly greater than that achieved with Cremophor
formulated paclitaxel administered at its MTD. Specifically,
tumors treated with the nanoparticle formulation of 7 at MTD
remained below 50 mg at day 60 in contrast to tumors treated
with paclitaxel at MTD, which had grown to nearly 300 mg
(Figure 4).
The dependence of efficacy on partitioning rate can be
rationalized in terms of accumulation of nanoparticles in tumors
followed by gradual release of prodrug. Nanoparticles that
release their payload rapidly or are cleared rapidly from
circulation will deliver less drug to the tumor site and
consequently may be expected to exhibit reduced efficacy.
Preliminary in vivo particle distribution studies indicated that
tumor accumulation levels plateau at ∼3% of the injected dose
per gram of tumor tissue, with the majority of the accumulation
occurring within the first 8 h. Similar results have been reported
in the literature.18–21 In the case of prodrugs such as 4 and 5
where partitioning half-lives are approximately 1 and 1.7 h,
respectively, the level of drug delivered to tumors is expected
to be significantly lower, likely accounting for the reduced
efficacy observed for these species. Prodrugs 6 and 7 release
drug at slower rates, with partitioning half-lives of approximately
6.5 and 10 h, respectively. Since the particle accumulation in
tumor is expected to be similar in all cases, formulations with
these prodrugs are capable of delivering more drug to the tumor.
The effects of dose on the efficacy of 7 nanoparticles were
evaluated relative to commercially formulated paclitaxel. The
optimal therapeutic response was observed with a Q2D dosing
in the HT29 solid tumor model. The nanoparticles used to
formulate 7 have an elimination half-life of approximately
24-36 h, while 7 itself has a partitioning half-life of about 10 h.
These values are consistent with an effective release of prodrug
at the tumor site over a period of approximately 2 days
postadministration, coinciding with the Q2D schedule. Titration
of drug dose through 0.25-3 paclitaxel equiv compared to
paclitaxel at MTD showed that 7 achieved equivalent efficacy
at approximately 1.7 paclitaxel equiv. Since the dose response
for 7 is linear up to the MTD, it is possible to significantly
improve efficacy over the maximum tolerated dose of paclitaxel
using higher doses of 7. We expect that prodrugs that are better
retained than 7 in nanoparticles would achieve parity at lower
doses than that seen for 7, offering the potential for further
improvement of this nanoparticle system.
Discussion
The nanoparticle delivery system employed in this work
required the synthesis of paclitaxel prodrugs to increase the
hydrophobicity of the drug. While paclitaxel itself is nearly
insoluble in water, it possesses sufficient aqueous solubility that
it can rapidly partition out of lipid-based delivery systems,
typically with half-lives on the order of a few minutes.28–30 The
objective of this work was to control the pharmacokinetics of
the formulated drug in vivo while maximizing efficacy. A
variety of prodrugs were synthesized, characterized, and for-
mulated into nanoparticles to characterize the effects of iterative
changes to the anchor-linker-drug structural motif on partition
kinetics and efficacy. A spectrum of in vitro and in vivo
antitumor activity was observed, dictated by the physical and
chemical properties of the prodrug rather than the delivery
vehicle itself.
Nanoparticle formulations of the type reported here are most
likely cleared as a result of single polymer chains partitioning
out of the particle over time in vivo. Gradual loss of the stabilizer
component destabilizes the particle and would result in the
elimination of any payload that had not already partitioned out.
We expect that particles with a high drug/stabilizer ratio would
be cleared rapidly because the loss of relatively small amounts
of polymer exposes the particle core to potential protein
interactions. Particles with low drug/stabilizer ratios have greater
tolerance to loss of polymer and may survive long enough to
be cleared through other mechanisms in vivo. This implies that
an ideal delivery system should approach the size of a micelle
with a relatively low drug/polymer ratio, preferably with a
stabilizer that has a low partitioning rate.
A major driving force in the development of the prodrug
nanoparticles described here is the generation of delivery
systems in which two drugs are released at matched rates to
maintain a synergistic ratio. The purpose of this study was to
demonstrate that it is possible to control drug release using a
paclitaxel as a model agent. Improved performance over the
free drug in vivo could be achieved through adjustment of
prodrug properties and complementary design of the delivery
system. This technology provides a delivery platform from
which synergistic combinations of different prodrug conjugates
can be coformulated in nanoparticles that are able to maintain
the desired drug/drug ratio in vivo, thereby optimizing synergy
and avoiding antagonism.
Prodrugs with lipid anchors of different hydrophobicity
demonstrated different partition rates from the same particle
composition. In these nanoparticles, drug physicochemical
properties largely dictated the release of drug from the carrier
rather than degradation of the carrier itself. This behavior leads
to a significant difference between paclitaxel in Cremophor and
Experimental Section
Materials. All reagents unless otherwise specified were pur-
chased from Sigma-Aldrich Canada Ltd., Oakville, Ontario. Sol-

Nanoparticle DeliVered Paclitaxel Prodrugs
Journal of Medicinal Chemistry, 2008, Vol. 51, No. 11 3295
vents were obtained from VWR International, Mississauga, Ontario.
Paclitaxel was purchased from Indena S.p.A., Milan, Italy. 3H-CHE
was obtained from Perkin-Elmer Life and Analytical Sciences, Inc.,
Waltham, MA. POPC was obtained from Northern Lipids, Burnaby,
d, J ) 2.96 Hz), 5.70 (1H, d, J ) 7.25 Hz), 6.05 (1H, dd, J ) 9.27
Hz, J′ ) 2.82), 6.29 (1H, t, J ) 8.3 Hz), 7.05 (1H, d, J ) 9.13
Hz).
2′-O-(5′′-O-Docosanyldiglycoloyl)paclitaxel 7. 7 was synthe-
sized from paclitaxel and docosanyl diglycolate. 1H NMR (400
MHz, CDCl3) δ 3.83 (1H, d, J ) 6.98 Hz), 4.0-4.4 (5H, m), 4.46
(1H, dd, J ) 10.75 Hz, J′ ) 6.72 Hz), 4.99 (1H, dd, J ) 9.40 Hz,
J′ ) 1.61 Hz), 5.60 (1H, d, J ) 2.96 Hz), 5.70 (1H, d, J ) 6.98
Hz), 6.05 (1H, dd, J ) 9.40 Hz, J′ ) 2.95 Hz), 6.28 (1H, t, J )
8.3 Hz), 7.05 (1H, d, J ) 9.40 Hz).
1
British Columbia. H NMR spectra were recorded in CDCl3 on a
Bruker Avance 400. HT-29 human colorectal adenocarcinoma cells
were obtained from ATCC, Manassas, VA. Foxn1nu mice were
obtained from Harlan, Indianapolis, IN. The confined volume
impinging jets mixer was custom-built at Princeton University. All
animal experiments were conducted according to protocols approved
by the University of British Columbia’s Animal Care Committee
and in accordance with the current guidelines established by the
Canadian Council of Animal Care.
Synthesis of Lipid Anchors. A lipid alcohol in pyridine was
treated with 3 equiv of succinic anhydride or diglycolic anhydride
at room temperature overnight. The solvent was removed on a
rotovap and the residue extracted from dilute hydrochloric acid with
methylene chloride. The organic fractions were dried over anhy-
drous magnesium sulfate and filtered, and the solvent was removed.
Conversion to the appropriate acid was monitored by TLC, which
in most cases was 100%. The resultant product was either dried
under vacuum or lyophilized from benzene. The lipid acids were
used in the following steps without further purification.
Synthesis of Paclitaxel Conjugates. Paclitaxel (1 equiv), a lipid
acid (2 equiv), and 4-N,N-dimethylaminopyridine (3 equiv) were
dissolved in alcohol-free chloroform. Diisopropylcarbodiimide (1.3
equiv) was then added and the solution stirred at room temperature.
The reaction was monitored by TLC until most of the paclitaxel
had been consumed (typically 2-4 h). The reaction mixture was
then washed with dilute hydrochloric acid and dried over anhydrous
magnesium sulfate. After removal of solvent the crude product was
passed down a silica gel column using a methanol/methylene
chloride gradient. The purified prodrug was lyophilized from
benzene and stored at room temperature.
2′-O-(4′′-O-Tocopherylsuccinoyl)paclitaxel 1. 1 was synthe-
sized from paclitaxel and tocopherol succinate. 1H NMR (400 MHz,
CDCl3) δ 3.83 (1H, d, J ) 7.25 Hz), 4.22 (1H, d, J ) 8.60 Hz),
4.33 (1H, d, J ) 8.60 Hz), 4.46 (1H, dd, J ) 10.88 Hz, J′ ) 6.58
Hz), 4.99 (1H, d, J ) 9.54 Hz), 5.52 (1H, d, J ) 3.22 Hz), 5.70
(1H, d, J ) 6.98 Hz), 5.98 (1H, dd, J ) 9.13 Hz, J′ ) 3.22 Hz),
6.27 (1H, t, J ) 8.6 Hz), 6.97 (1H, d, J ) 8.87 Hz).
2′-O-(5′′-O-Cholesteryldiglycoloyl)paclitaxel 8. 8 was synthe-
1
sized from paclitaxel and cholesteryl diglycolate. H NMR (400
MHz, CDCl3) δ 3.83 (1H, d, J ) x Hz), 4.0-4.4 (5H, m), 4.47
(1H, dd, J ) 10.6 Hz, J′ ) 6.8 Hz), 4.63 (1H, m), 4.99 (1H, d, J
) 7.79 Hz), 5.38 (1H, d, J ) 3.76 Hz), 5.61 (1H, d, J ) 2.69 Hz),
5.70 (1H, d, J ) 7.25 Hz), 6.06 (1H, dd, J ) 9.13 Hz, J′ ) 2.69
Hz), 6.28 (1H, t, J ) 8.3 Hz), 7.09 (1H, d, J ) 9.40 Hz).
2′-O-(5′′-O-(1′′′,2′′′-Dimyristoyl-sn-glycero)diglycoloyl)paclitax-
el 9. 9 was synthesized from paclitaxel and 3-(1,2-dimyristoyl-sn-
glycerol) diglycolate. 1H NMR (400 MHz, CDCl3) δ 3.84 (1H, d,
J ) 6.98 Hz), 4.47 (1H, dd, J ) 10.6 Hz, J′ ) 6.8 Hz), 4.99 (1H,
dd, J ) 9.54 Hz, J′ ) 1.75 Hz), 5.20 (1H, m), 5.61 (1H, d, J )
2.96 Hz), 5.71 (1H, d, J ) 7.25 Hz), 6.07 (1H, d, J ) 9.27 Hz, J′
) 2.82 Hz), 6.31 (1H, t), 7.11 (1H, d, J ) 9.40 Hz).
Nanoparticle Formulation. The prodrug, co-lipid, and stabilizer
polymers (typically on a 1:1:2 w/w basis) were dissolved in ethanol/
THF (4:1) at a concentration of 40 mg/mL. The solvent was rapidly
diluted with water using a four-port CVIJ mixer23,24 with flow rates
set at 12/12/53/53 mL/min (solvent/water/water/water). Flow rates
were controlled using Harvard apparatus PHD2000 syringe pumps.
The resultant solution was then dialyzed against water to remove
residual solvent. The final drug concentration was typically about
0.7 mg/mL. When higher concentrations were required, the dialyzed
solution was diluted with equivolumes of 600 mM sucrose and then
concentrated using a 100 kD, 0.5 mm lumen, 60 cm path length
MidGee hoop cartridge (GE Healthcare Life Sciences, Piscataway,
NJ) with a peristaltic pump. Particle size was determined using a
Malvern Zetasizer Nano-ZS particle sizer and reported as volume
weighted data.
In Vitro Activity. The MCF-7 human tumor cell line was
purchased from American Type Culture Collection (MEM media
supplemented with 2 mM L-glutamine and 10% FBS). The A2780
human tumor cell line was purchased from the European Collection
of Cell Culture (RPMI 1640 media supplemented with 2 mM
L-glutamine and 10% FBS). Cultures were incubated at 37 °C and
5% CO2.
On day 0, A2780 cells were plated at 2000 cells/well and MCF-7
cells at 1500 cells/well and allowed to adhere for 24 h. On day 1
cells were exposed in triplicate to a 2-fold serial dilution of prodrug/
POPC/2.5kPS3k (1:1:2) nanoparticle formulations for 72 h at 16
concentrations. Viable cells were quantified using standard 3-(4,5-
dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) de-
tection at 570 nm after DMSO addition.31 Survival rate after
treatment is expressed as a mean percentage relative to untreated
control wells.
2′-O-(4′′-O-Oleylsuccinoyl)paclitaxel 2. 2 was synthesized from
paclitaxel and oleyl succinate. 1H NMR (400 MHz, CDCl3) δ 3.82
(1H, d, J ) 7.0 Hz), 3.96 (2H, t, J ) 6.81 Hz), 4.21 (1H, d, J )
8.38 Hz), 4.33 (1H, d, J ) 8.38 Hz), 4.46 (1H, br. s), 4.98 (1H, d,
J ) 8.45 Hz), 5.36 (2H, m), 5.50 (1H, d, J ) 2.89 Hz), 5.69 (1H,
d, J ) 7.00 Hz), 6.00 (1H, dd, J ) 9.04 Hz, J′ ) 2.74 Hz), 6.26
(1H, t, J ) 8.91 Hz), 7.08 (1H, d, J ) 9.06 Hz).
2′-O-(5′′-O-Tocopheryldiglycoloyl)paclitaxel 3. 3 was synthe-
1
sized from paclitaxel and tocopherol diglycolate. H NMR (400
MHz, CDCl3) δ 3.84 (1H, d, J ) 6.98 Hz), 4.22 (1H, d, J ) 8.3
Hz), 4.34 (1H, d, J ) 8.6 Hz), 4.99 (1H, d, J ) 7.79 Hz), 5.64
(1H, d, J ) 2.96 Hz), 5.70 (1H, d, J ) 6.98 Hz), 6.06 (1H, dd, J
) 9.27 Hz, J′ ) 2.82 Hz), 6.30 (1H, t), 6.97 (1H, d, J ) 9.13 Hz).
2′-O-(5′′-O-Oleyldiglycoloyl)paclitaxel 4. 4 was synthesized
from paclitaxel and oleyl diglycolate. 1H NMR (400 MHz, CDCl3)
δ 3.83 (1H, d, J ) 7.0 Hz), 4.46 (1H, t, J ) 7.96 Hz), 4.99 (1H,
d, J ) 8.45 Hz), 5.35 (2H, m), 5.60 (1H, d, J ) 2.74 Hz), 5.70
(1H, d, J ) 7.08 Hz), 6.05 (1H, dd, J ) 9.25 Hz, J′ ) 2.55 Hz),
6.28 (1H, t, J ) 8.9 Hz), 7.04 (1H, d, J ) 9.29 Hz).
2′-O-(5′′-O-Octadecyldiglycoloyl)paclitaxel 5. 5 was synthe-
sized from paclitaxel and octadecyl diglycolate. 1H NMR (400
MHz, CDCl3) δ 3.83 (1H, d, J ) 7.0 Hz), 4.46 (1H, dd, J ) 10.74
Hz, J′ ) 6.70 Hz), 4.99 (1H, d, J ) 8.45 Hz), 5.60 (1H, d, J )
2.82 Hz), 5.70 (1H, d, J ) 7.00 Hz), 6.05 (1H, dd, J ) 9.25 Hz,
J′ ) 2.55 Hz), 6.28 (1H, t, J ) 8.9 Hz), 7.05 (1H, d, J ) 9.29 Hz).
2′-O-(5′′-O-Cosanyldiglycoloyl)paclitaxel 6. 6 was synthesized
from paclitaxel and cosanyl diglycolate. 1H NMR (400 MHz,
CDCl3) δ 3.83 (1H, d, J ) 6.98 Hz), 4.0-4.4 (5H, m), 4.46 (1H,
dd, J ) 10.75 Hz, J′ ) 6.72), 4.99 (1H, d, J ) 9.40 Hz), 5.60 (1H,
In Vivo Plasma Elimination. Athymic nude mice (n ) 3/time
point) were injected iv with test samples (10 µL/g body weight, to
a maximum of 250 µL). Blood was collected at the designated time
point by cardiac puncture and placed into EDTA coated microtainer
tubes (BD Biosciences). The tubes were centrifuged at 2800 rpm
for 10 min. Plasma was recovered, and 50 µL aliquots were
analyzed by HPLC for drug content. Samples containing 3H-CHE
(50 µl) were diluted with Picofluor40 scintillation fluid (Packard)
and counted on a Beckman-Coulter LS 6500 multipurpose scintil-
lation counter. The total drug (counts) per mouse in the blood
compartment was calculated assuming a plasma volume of 0.041 25
mL per gram of body weight.
In Vivo Efficacy. Female athymic nude mice (7-8 weeks old)
were inoculated subcutaneously with 100 µL of HT29 human colon
carcinoma cells (2 × 106 cells) using a 26 gauge needle. Mice were
randomly grouped (n ) 6/group) with a mean tumor size between

3296 Journal of Medicinal Chemistry, 2008, Vol. 51, No. 11
Ansell et al.
(11) Stevens, P. J.; Sekido, M.; Lee, R. J. A folate receptor-targeted lipid
nanoparticle formulation for a lipophilic paclitaxel prodrug. Pharm.
Res. 2004, 21, 2153–2157.
(12) Perkins, W. R.; Ahmad, I.; Li, X.; Hirsh, D. J.; Masters, G. R.; Fecko,
C. J.; Lee, J. K.; Ali, S.; Nguyen, J.; Schupsky, J.; Herbert, C.; Janoff,
A. S.; Mayhew, E. Novel therapeutic nano-particles lipocores): trapping
poorly water soluble compounds. Int. J. Pharm. 2000, 200, 27–39.
(13) Ali, S.; Ahmad, I.; Peters, A.; Masters, G.; Minchey, S.; Janoff, A. S.;
Mayhew, E. Hydrolyzable hydrophobic taxanes: synthesis and anti-
cancer activities. Anti-Cancer Drugs 2001, 12, 117–128.
(14) Lundberg, B. B.; Risovic, V.; Ramaswamy, M.; Wasan, K. M. A
lipophilic pacltaxel derivative incorporated in a lipid emulsion for
parenteral administration. J. Controlled Release 1003, 86, 93–100.
(15) Rodrigues, D. C.; Maria, D. A.; Fernandes, D. C.; Valduga, C. J.;
Couto, R. D.; Ibanez, O. C. M.; Maranhao, R. C. Improvement of
paclitaxel therapeutic index by derivatization and association to a
cholesterol-rich microemulsion: in vitro and in vivo studies. Cancer
Chemother. Pharmacol. 2005, 55, 565–576.
(16) Zakharian, T. Y.; Seryshev, A.; Sitharaman, B.; Gilbert, B. E.; Knight,
V.; Wilson, L. J. A fullerene-paclitaxel chemotherapeutic: synthesis,
characterization and study of biological activity in tissue culture. J. Am.
Chem. Soc. 2005, 127, 12508–12509.
(17) Bradley, M. O.; Webb, N. L.; Anthony, F. H.; Devanesan, P.; Witman,
P. A.; Hemamalini, S.; Chander, M. C.; Baker, S. D.; He, L.; Horwits,
S. B.; Swindell, C. S. Tumor targeting by covalent conjugation of a
natural fatty acid to paclitaxel. Clin. Cancer Res. 2001, 7, 3229–3238.
(18) Torchilin, V. P. Lipid-core micelles for targeted drug delivery. Curr.
Drug DeliVery 2005, 2, 319–327.
(19) Kwon, G.; Suwa, S.; Yokoyama, M.; Okano, T.; Sakurai, Y.; Kataoka,
K. Enhanced tumor accumulation and prolonged circulation times of
micelle-forming poly (ethylene oxide-aspartate) block copolymer-
adriamycin conjugates. J. Controlled Release 1994, 29, 17–23.
(20) Weissig, V.; Whiteman, K. R.; Torchilin, V. P. Accumulation of
protein-loaded long-circulating micelles and liposomes in subcutaneous
Lewis lung carcinoma in mice. Pharm. Res. 1998, 15, 1552–1556.
(21) Lukyanov, A. N.; Gao, Z.; Mazzola, L.; Torchilin, V. P. Polyethylene
glycol-diacyllipid micelles demonstrate increased accumulation in
subcutaneous tumors in mice. Pharm. Res. 2002, 19, 1424–1429.
(22) Deutsch, H. M.; Glinski, J. A.; Hernandez, M.; Haugwitz, R. D.;
Narayanan, V. L.; Suffness, M.; Zalkow, L. H. Synthesis of congeners
and prodrugs. 3. Water-soluble prodrugs of Taxol with potent antitumot
activity. J. Med. Chem. 1989, 32, 788–792.
80 and 120 mg prior to the first treatment. (Tumors greater than
200 mg and smaller than 30 mg were excluded.) Test samples were
injected with a volume of 10 µL drug/g body weight, to a maximum
of 250 µL. Subsequent injections were made according to the
designated schedule. Tumor length and width measurements, body
weight, and in life observations were recorded 2-3 times per week.
Tumor weight was calculated according to the formula (L)(W2)/2.
The paclitaxel positive controls were administered using the
commercial clinical formulation Taxol. Taxol comprises a 6 mg/
mL solution of paclitaxel dissolved in 527 mg of Cremophor EL/
49.7% (v/v) ethanol and was reconstituted immediately prior to
administration by dilution with 0.9% saline in a 1:2 ratio (v/v) for
a final paclitaxel concentration of 2 mg/mL.
HPLC Analysis. Plasma samples (50 µL) or nanoparticle
suspensions (50 µL) were mixed with 150 µL of diluents (methanol/
acetonitrile, 2:1 v/v) by vigorous vortexing followed by centrifuga-
tion at 10000g for 10 min. Supernatant (20 µL) was injected into
a Waters HPLC for quantitation using a Phenomenex SynergiFusion
reverse phase analytical column monitored by UV detection at 227
nm. Chromatography was conducted with a 1 mL/min gradient
elution of methanol and 10 mM sodium acetate buffer (pH 5.6)
mobile phases from an initial solvent mix of 70:30 to 100:0,
respectively. Column temperature was set at 30 °C. Samples were
kept in the autosampler compartment at 4 °C prior to HPLC
analysis.
Acknowledgment. We thank Suzen Lines, Lindsey Knight,
Fiona Ross, and Sarah Kelly for assistance in the vivarium.
Supporting Information Available: NMR and HPLC data,
elimination curves for control particles, modeled partition data,
relative weight loss data for Figure 4. This material is available
free of charge via the Internet at http://pubs.acs.org.
References
(23) Johnson, B. K.; Prud’homme, R. K. Flash nanoprecipitation of organic
actives and block copolymers using a confined impinging jets mixer.
Aust. J. Chem. 2003, 56, 1021–1024.
(24) Johnson, B. K.; Prud’homme, R. K. Mechanism for rapid self-assembly
of block copolymer nanoparticles. Phys. ReV. Lett. 2003, 91, 118302.
(25) Liu, J.; Zeng, F.; Allen, C. In vivo fate of unimers and micelles of a
poly(ethylene glycol)-block-poly(caprolactone) copolymer in mice
following intravenous administration. Eur. J. Pharm. Biopharm. 2007,
65, 309–319.
(26) Allen, T. M.; Hansen, C.; Martin, F.; Redemann, C.; Yau-Young, A.
Liposomes containing synthetic lipid derivatives of poly(ethylene
glycol) show prolonged circulation half lives in vivo. Biochim. Biophys.
Acta 1991, 1066, 29–36.
(27) Desai, N.; Trieu, V.; Yao, Z.; Louie, L.; Ci, S.; Yang, A.; Tao, C.;
De, T.; Beals, B.; Dykes, P.; Noker, P.; Yao, R.; Labao, E.; Hawkins,
M.; Soon-Shiong, P. Increased antitumor activity, intratumor paclitaxel
concentrations and endothelial cell transport of cremophor-free,
albumin bound paclitaxel, ABI-007, compared with Cremophor-based
paclitaxel. Clin. Cancer Res. 2006, 12, 1317–1324.
(28) Fetterly, G. J.; Straubinger, R. M. Pharmacokinetics of paclitaxel-
containing liposomes in rats. AAPS PharmSci 2003, 5, 1–11.
(29) Sparreboom, A.; Scripture, C. D.; Trieu, V.; Williams, P. J.; De, T.;
Yang, A.; Beals, B.; Figg, M.; Hawkins, W. D.; Desai, N. Comparative
preclinical and clinical pharmacokinetics of a Cremophor free,
nanoparticle albumin-bound paclitaxel (ABI-007) and paclitaxel
formulated in Cremophor (Taxol). Clin. Cancer Res. 2005, 11, 4136–
4143.
(30) Sparreboom, A.;.; van Zuylen, L.; Brouwer, E.; Loos, W. J.; de Bruijn,
P.; Gelderblom, H.; Pillay, M.; Nooter, K.; Stoter, G.; Verweij, J.
Cremophor EL-mediated alteration of paclitaxel distribution in human
blood: clinical pharmacokinetic implications. Cancer Res. 1999, 59,
1454–1457.
(1) Harasym, T. O.; Tardi, P. G.; Harasym, N. L.; Harvie, P.; Johnstone,
S. A.; Mayer, L. D. Increased precilinical efficacy of irinotecan and
floxuridine coencapsulated inside liposomes is aassociated with tumor
delivery of synergistic drug ratios. Oncol. Res. 2007, 16, 361–374.
(2) Mayer, L. D.; Janoff, A. S. Optimizing combination chemotherapy
by controlling drug ratios. Mol. InterVentions 2007, 7, 216–223.
(3) Harasym, T. O.; Tardi, P. G.; Johnstone, S. A.; Bally, M. B.; Janoff,
A. S.; Mayer, L. D. Fixed Drug Ratio Liposome Formulations of
Combination Cancer Therapeutics. In Liposome Technology, 3rd ed.;
Gregoriadis, G. Ed.; Informa Healthcare: New York, 2007; Volume
III, pp 25-48.
(4) Mayer, L. D.; Harasym, T. O.; Tardi, P. G.; Harasym, N. L.; Shew,
C. R.; Johnstone, S. A.; Ramsay, E. C.; Bally, M. B.; Janoff, A. S.
Ratiometric dosing of anticancer drug combinations: controlling drug
ratios after systemic administration regulates therapeutic activity in
tumor-bearing mice. Mol. Cancer Ther. 2006, 5, 1854–1863.
(5) Fahr, A.; van Hoogevest, P.; May, S.; Bergstand, N.; Leigh, M. L. S.
Transfer of lipophilic drugs between liposomal membranes and
biological interfaces: consequences for drug delivery. Eur. J. Pharm.
Sci. 2005, 26, 251–265.
(6) Skwarczynski, M.; Hayashi, Y.; Kiso, Y. Paclitaxel prodrugs: toward
smarter delivery of anticancer agents. J. Med. Chem. 2006, 49, 7253–
7269.
(7) Hennenfent, K. L.; Govindan, R. Novel formulations of taxanes: a
review. Old wine in a new bottle. Ann. Oncol. 2006, 17, 734–749.
(8) Hamaguchi, T.; Matsumura, Y.; Suzuki, M.; Shimizu, K.; Goda, R.;
Nakamura, I.; Yokoyama, M.; Kataoka, K.; Kakizoe, T. NK105, a
paclitaxel-incorporating micellar nanoparticle formulation, can extend
in vivo antitumor activity and reduce the neurotoxicity of palitaxel.
Br. J. Cancer 2005, 92, 1240–1246.
(9) Ansell, S. Lipophilic Drug Derivatives for Use in Liposomes. U.S.
Patent 5,534, 499, 1996.
(10) Hostetler, K. Y.; Sridhar, N. C. Prodrugs for Oral Administration
Containing Taxol Covalently Bound to a Phospholipid. U.S. Patent
5,484,809, 1996.
(31) Mosmann, T. Rapid colorimetric assay for cellular growth and survival:
application to proliferation and cytotoxicity assays. J. Immunol.
Methods 1983, 65, 55–63.
JM800002Y