A dicarboxylic fatty acid derivative of paclitaxel for albumin-assisted drug delivery

Article abstract of DOI:10.1002/jps.23213

Paclitaxel (PTX) is a potent chemotherapy for many cancers but it suffers from very poor solubility. Consequently, the TAXOL formulation uses copious amounts of the surfactant Cremophor EL to solubilize the drug for injection, resulting in severe hypersensitivity and neutropenia. In contrast to Cremophor EL, presented is a way to solubilize PTX by conjugation of a dicarboxylic fatty acid for specific binding to the ubiquitous protein, serum albumin. The conjugation chemistry was simplified to a single step using the activated anhydride form of 3-pentadecylglutaric (PDG) acid, which is reactive to a variety of nucleophiles. The PDG derivative is less cytotoxic than the parent compound and was found to slowly hydrolyze to PTX (～5% over 72h) in serum, tumor cytosol, and tumor tissue homogenate. When injected intravenously to tumor-bearing mice, [³H]-PTX in the TAXOL formulation was cleared rapidly with a half-life of 7 h. In the case of the PDG derivative of PTX, the drug is quickly distributed and approximately 20% of the injected dose remained in the vasculature experiencing a 23h half-life. These improvements from modifying PTX with the PDG fatty acid present the opportunity for PDG to become a generic modification for the improvement of many therapeutics.

Full text of DOI:10.1002/jps.23213

RESEARCH ARTICLE
A Dicarboxylic Fatty Acid Derivative of Paclitaxel for
Albumin-Assisted Drug Delivery
MICHAEL J. HACKETT, SHYAMSUNDER JOOLAKANTI, MEGAN E. HARTRANFT, PATRICK C. GULEY, MOO J. CHO
Division of Molecular Pharmaceutics, Eshelman School of Pharmacy, University of North Carolina at Chapel Hill,
Chapel Hill, North Carolina 27599-7571
Received 31 January 2012; revised 21 March 2012; accepted 4 May 2012
Published online in Wiley Online Library (wileyonlinelibrary.com). DOI 10.1002/jps.23213
ABSTRACT: Paclitaxel (PTX) is a potent chemotherapy for many cancers but it suffers from
very poor solubility. Consequently, the TAXOL formulation uses copious amounts of the sur-
factant Cremophor EL to solubilize the drug for injection, resulting in severe hypersensitiv-
ity and neutropenia. In contrast to Cremophor EL, presented is a way to solubilize PTX by
conjugation of a dicarboxylic fatty acid for specific binding to the ubiquitous protein, serum
albumin. The conjugation chemistry was simplified to a single step using the activated anhy-
dride form of 3-pentadecylglutaric (PDG) acid, which is reactive to a variety of nucleophiles.
The PDG derivative is less cytotoxic than the parent compound and was found to slowly hy-
drolyze to PTX (∼5% over 72 h) in serum, tumor cytosol, and tumor tissue homogenate. When
3
injected intravenously to tumor-bearing mice, [ H]-PTX in the TAXOL formulation was cleared
rapidly with a half-life of 7 h. In the case of the PDG derivative of PTX, the drug is quickly
distributed and approximately 20% of the injected dose remained in the vasculature expe-
riencing a 23 h half-life. These improvements from modifying PTX with the PDG fatty acid
present the opportunity for PDG to become a generic modification for the improvement of
many therapeutics. © 2012 Wiley Periodicals, Inc. and the American Pharmacists Association
J Pharm Sci
Keywords: albumin; cancer chemotherapy; endogenous drug delivery systems; pharmacoki-
netics; synthesis
proximately 130 nm diameter after conebulization.2
ABRAXANE showed itself superior to TAXOL in effi-
cacy, half-life, and toxicity and is currently indicated
as a second-line therapy for breast cancer.3
INTRODUCTION
Paclitaxel (PTX) is a mitotic spindle inhibitor pri-
marily indicated for ovarian cancer or in combina-
tion therapy for the treatment of non-small cell lung
cancer and breast cancer. With an intrinsic solubil-
,4
Albumin is a highly soluble protein with a vascu-
lature concentration of 40–50 mg/mL (0.6–0.75 mM),
1
ity of 10 :g/mL, PTX requires dissolution in ethanol
5
making it the most abundant serum protein. The
and the surfactant Cremophor EL in the commonly
known product, TAXOL. The large quantities of Cre-
mophor EL lead to neutropenia and hypersensitiv-
ity in patients, thus requiring a prophylactic regimen
high vasculature concentrations of albumin are main-
tained because of recycling via the protective Bram-
bell receptor (FcRn), which grants albumin and IgG
5
1
9 and 21 day half lives in humans, respectively. The
2
of dexamethasone. A second-generation formulation
use of albumin as a drug carrier for antineoplastics
devoid of the surfactant uses the endogenous serum
protein albumin as a vehicle. ABI-007, also known
as ABRAXANE, utilizes the intrinsic binding of PTX
to albumin to entrap the drug in particles of ap-
6
is not a novel concept as albumin has been shown
to be a major nutritional staple for a variety of grow-
7
ing tumors. The high vascular concentrations beget
high tumoral concentrations, making albumin abun-
8
,9
dantly available for catabolism.
This tumoral ac-
Additional Supporting Information may be found in the online
version of this article. Supporting Information
cretion is believed to proceed via passive filtration
and sequestration in the tumor microenvironment of
leaky endothelia. This observation was the founda-
tion for Maeda’s now universal “enhanced permeabil-
Correspondence to: Moo J. Cho (Telephone: +919-966-1345; Fax:
+
919-966-0197; E-mail: m j cho@unc.edu)
Journal of Pharmaceutical Sciences
2012 Wiley Periodicals, Inc. and the American Pharmacists Association
6
,10
©
ity and retention” hypothesis.
It has recently been
JOURNAL OF PHARMACEUTICAL SCIENCES
1

2
HACKETT ET AL.
shown that albumin and albumin-based therapies
such as ABRAXANE may be more specifically sought
by tumors. One mechanism of selection is through
caveolae-mediated transcytosis via selective overex-
pression of gp60 (albondin) in the tumor endothelium.
Inhibition of this protein had profound effects on the
tumor accumulation of ABRAXANE.11,12 Another al-
bumin binding protein, the secreted protein acidic and
rich in cysteines (SPARC), is also believed to facilitate
ity, serum half-life, or potentially increased tumoral
delivery.
MATERIALS AND METHODS
Chemicals and Reagents
All chemicals involved in the synthesis were pur-
chased from Sigma–Aldrich (St. Louis, Missouri) at
9
8% or more purity and used without further purifi-
the accumulation of albumin via increased paracellu-
lar transit due to endothelial remodeling.13 Increased
cation. Delipidated human serum albumin was also
purchased from Sigma–Aldrich at 96% purity. All sol-
vents were purchased from BDH (Radnor, Pennsyl-
vania) and were ACS grade or better. PTX was pur-
chased from AK Scientific (Union City, California)
and used without further purification. CT26 murine
colon carcinoma cells were obtained from American
Type Culture Collection (Manassas, Virginia) and
expression of SPARC directly correlated with the ef-
ficacy of ABRAXANE, showing it to be more effective
than surfactant-based PTX or docetaxel, even in po-
1
4
tentially PTX-resistant HER2-positive xenografts.
SPARC has also been linked to the upregulation of
matrix metalloproteases,15 which is partially respon-
sible for the high proteolytic activity observed in tu-
◦
1
6,17
grown at 37 C at 5% CO2 in Roswell Park Memo-
mor tissues.
This selectivity for albumin by tu-
rial Institute-1640 media supplemented to 10% fetal
bovine serum, and 1% penicillin and streptomycin.
Cells were passaged twice before implantation to
mice. Tritium-labeled PP and PTX were synthesized
by Moravek Biochemicals (Brea, California) at a spe-
cific activity of 11 and 36 Ci/mmol, respectively, with
mors makes albumin excellently suited as a natural
drug carrier for cancer therapy.
The primary function of albumin is plasma expan-
sion accomplished by maintaining an oncotic pres-
sure across the vasculature with high (0.60–0.75 mM)
intravasculature and lower (0.30 mM) extravascula-
3
ture concentrations.5 Ancillary to expansion, albu-
H-substitutions at the ortho positions on the benza-
mide ring. Female BALB/c 6–8-week-old mice weigh-
ing 18–20 g were acquired from National Cancer In-
stitute (NCI). Mice were allowed 1 week acclimation
prior to beginning studies. All animal studies were
performed under approval of the University of North
Carolina Office of Animal Care and Use in accordance
with the standards set forth by the Institutional Ani-
mal Care and Use Committee.
min acts as a sponge for various metals, hydrophobic
1
8
xenobiotics, and fatty acids. Albumin was shown to
have six to seven binding sites for long-chain fatty
acids with dissociation constants (Kd) in the nanomo-
1
9,20
lar range.
These highly specific binding sites are
due to hydrophobic channels in the protein, which
cradle the fatty acid tails. The apex of these chan-
nels, at the protein–water interface, provides lysine
or arginine cations, which anchor the anionic car-
Spectroscopy
2
1,22
boxylate in place via an electrostatic interaction.
Mass spectra were obtained on an Agilent (Santa
Clara, California) 1100-DAD-FLD Ion Trap electro-
spray ionization (ESI) mass spectrometer in MeOH or
Historically, drug–fatty acid conjugation was typi-
cally accomplished through amides or esters, con-
sequently removing the electrostatic interaction
1
13
2
3
MeCN. H NMR and C NMR spectra were recorded
on a Varian (Palo Alto, California) Inova 400 MHz
between conjugate and protein. Such examples in-
2
4
25
clude taxoprexin and squalenoyl–PTX. The affini-
ties of these conjugates for the protein are thus
expected to diminish when compared with the re-
spective unconjugated fatty acids. One exception is
the myristoylated insulin LEVEMIR where the C-
terminal anion of the B-chain can fortuitously bring
about the gain in binding energy from the electro-
◦
NMR at 30 C in CDCl3 or CD3OD; chemical shifts
are referenced to tetramethylsilane (* 0.00 ppm) for
1
H NMR and CDCl3 (* 77.23 ppm) or CD3OD (* 49.15
1
3
ppm) for C NMR. Ultraviolet (UV) spectra were ob-
tained with a Shimadzu (Kyoto, Japan) UV-2401PC
spectrophotometer. Scintillation counting was con-
ducted on a Packard (Waltham, Massachusetts) Tri-
Carb 2900TL. Dynamic light scattering (DLS) was
acquired on a Malvern (Malvern, Worcestershire,
2
6
static interaction. In the present study, a new fatty
acid, 3-pentadecylglutaric (PDG) acid, is presented
that can be easily conjugated to nucleophiles. The
product of the reaction is a fatty acid derivative that
maintains an anionic carboxylate for the purpose of
generating long-circulating, soluble, albumin-bound
conjugates. The technology is demonstrated through
a PDG–PTX conjugate (PP), but could potentially be
applied to any chemical entity with a nucleophilic
functionality that requires an increase in solubil-
◦
United Kingdom) Zetasizer Nano ZS detecting 173
noninvasive backscatter.
Ethyl-(E)-Octadec-2-Enoate (1)
The intermediate was prepared following a litera-
2
7
1
ture procedure (85% yield over two steps). H NMR
(400 MHz, CDCl ) * 6.96 (m, 1H, HC = CHCH ), 5.81
3
2
JOURNAL OF PHARMACEUTICAL SCIENCES
DOI 10.1002/jps

PACLITAXEL DERIVATIVE FOR ALBUMIN-ASSISTED DRUG DELIVERY
3
(
2
(
d, 1H, HC = CHVCH2), 4.18 (m, 2H, O CH2CH3),
MHz, CDCl3) *167.1, 36.3, 34.7, 32.1, 29.8, 28.9, 26.5,
+
.19 (q, 2H, (CH2), 1.54 (s, 1H, CH2CHCH2), 1.45
22.9, 14.3. ESI–MS (pos, MeCN) m/z 325.4 [M + H] ,
+
t, 2H, *CH2), 1.30 to 1.26 (m, 29H, CH2CH2,
347.6 [M + Na] .
OCH2CH3), 0.88 (t, 3H, CH3). ESI–mass spectrom-
ꢀ
Paclitaxel-2 -O-3-Pentadecylhemiglutarate (5; PP)
+
etry (MS) (pos, MeOH): 333.5 [M + Na] .
3
Intermediate 4 (21 mg, 64.5 :mol), PTX (50 mg, 58.6
-Pentadecyldiethylglutarate (2)
:
mol), and N,N-dimethylaminopyridine (DMAP) (0.7
To a 10 mL solution of anhydrous EtOH containing
sodium metal (111 mg, 4.83 mmol) was added di-
mg, 5.86 :mol) were dissolved in 0.8 mL dry pyridine
and the reaction mixture stirred overnight at room
temperature. After evaporation, the solid was taken
up in dichloromethane (DCM) and washed with 0.1 N
HCl. The DCM was dried, filtered, and concentrated.
The solid was adsorbed onto silica and eluted with
(1:1) EtOAc in petroleum ether supplemented with
1% AcOH. The product was recovered as a white solid
ethylmalonate (386 mg/367 :L, 2.42 mmol) and (1)
◦
(
500 mg, 1.61 mmol) in 5 mL dry EtOH at 0 C. The
solution was stirred overnight at room temperature.
The solution was concentrated, diluted with water,
and extracted with EtOAc. The EtOAc layer was dried
with magnesium sulfate and evaporated to a colorless
oil. To the oil in DMSO was added 1 eq of H2O and
1
(58.4 mg, 85%). H NMR (400 MHz, CDCl3) * 8.12 (d,
2
eq of NaCl at a concentration of 6 g/mL and the
2H, ArH), 7.75 (d, 2H, ArH), 7.62 to 7.24 (m, 11H,
ArH), 6.29 (s, 1H, C10), 6.24 (t, 1H, C 13), 5.99 (dd,
1H, C 3’), 5.68 (d, 1H, C 2), 5.50 (d, 1H, C 2’), 4.97
(d, 1H, C 5), 4.40 (dd, 1H, C 7), 4.29 (d, 1H, C 20),
4.20 (d, 1H, C 20), 3.80 (d, 1H, C 3), 2.56 2.50 (m,
1H, CH2 CH CH2 ), 2.45 (s, 3H, OAc), 2.40–2.28
(m, 3H, CH2 CH CH2 , C 6"), 2.24–2.13 (m, 7H,
OAc, C 14, CH2 CH CH2 ), 2.00–1.85 (m, 4H,
C 18, C 6$), 1.66 (s, 3H, C 19), 1.26–1.19 (m, 31H,
CH2CH3, C 17), 1.13 (s, 3H, C 16), 0.88 (t, 3H,
solution refluxed for 5 h. Dilution with water and ex-
traction into EtOAc yielded a colorless oil. The prod-
uct was purified on silica gel eluting with 5% EtOAc
in petroleum ether (603 mg, 94% over two steps). H
NMR (400 MHz, CDCl3) * 4.13 (q, 4H, OCH2CH3),
1
2.34 (m, 5H, CH2CHCH2), 1.33 to 1.24 (m, 34H,
CH2CH2, OCH2CH3), 0.88 (t, 3H, CH3). 13C NMR
(
2
100 MHz, CDCl3) * 172.8, 60.4, 38.9, 34.2, 32.4, 32.1,
9.8, 26.8, 22.9, 14.4, 14.3. ESI–MS (pos, MeOH) m/z
+
+
+
13
399.7 [M + H] , 421.8 [M + Na] , 437.8 [M + K] .
CH2CH3). C NMR (100 MHz, CDCl3) * 204.0,
76.7, 171.9, 171.5, 170.2, 168.3, 167.7, 167.2, 142.9,
37.1, 133.8, 133.0, 132.2, 130.5, 129.0, 127.1, 84.7,
1
PDG Acid (3)
1
To 15 mL of MeOH containing 2 (1.06 g, 2.66 mmol)
was added 30 mL 3 N KOH. The mixture was heated
to a reflux with stirring for 3 h causing the suspen-
sion to become clear. The solution was cooled to am-
bient temperature, acidified to pH 1 with HCl, and
extracted with methyl t-butyl ether. The organic layer
was dried and concentrated and the product collected
as a white solid (900 mg, 99%). The compound was
81.3, 79.2, 75.8, 75.3, 74.4, 72.2, 58.7, 53.1, 45.8,
43.4, 38.0, 35.7, 34.1, 32.2, 29.9, 29.7, 27.1, 22.9, 22.3,
21.0, 15.0, 14.3, 9.8. ESI–MS (pos, MeOH) m/z 1201.1
[M+Na] .
+
Cytotoxicity of PP
The cytotoxicity of PP and PTX was determined by
the NCI as part of the NCI-60 DTP Human Tumor
1
28
pure by TLC 10% Me2CO in petroleum ether. H NMR
Cell Line Screen. Briefly, various human cancer
(
400 MHz, CD3OD) * 2.33–2.27 (m, 5H, CH2CHCH2),
cell lines are plated at 5000–40,000 cells/well and
treated with five concentrations of PP or PTX from
10 nM to 100 :M. The effect on cell growth is com-
pared with cell lines fixed at time 0 and the untreated
control via the sulforhodamine B test. Dose–response
curves are then generated across 60 cell lines and the
1
.37 to 1.29 (m, 28H, CH2CH2 ), 0.90 (t, 3H, CH3).
C NMR (100 MHz, CD3OD) *176.7, 39.6, 35.2, 33.5,
3.3, 30.9, 27.8, 23.9, 14.6. ESI–MS (neg, MeOH) m/z
1
3
3
3
−
41.7 [M − H] .
PDG Anhydride (4)
50% growth inhibition (GI50), total growth inhibition
Intermediate 3 (900 mg, 2.63 mmol) was suspended in
(TGI), and 50% lethal concentration (LC ) values are
determined for each cell line.
5
0
3
mL EtOAc and heated gently until dissolved. To the
solution was added trifluoroacetic anhydride (773 mg/
Biochemical Stability of PP
0
.513 mL, 3.68 mmol) and the solution was heated to
◦
37 C with shaking for 2 h. After evaporation, the
Murine-derived CT26 cells were grown in a T-75
6
solid was recrystallized from pentane. Filtration of
the solid and drying under vacuum produces the prod-
flask to confluency (∼4 × 10 cells). The cells were
then washed twice with Hank’s balanced salt solu-
tion (HBSS), scraped off the flask with a cell scraper,
and suspended in 5 mL HBSS. The suspension was
uct as a fluffy white powder in quantitative yield
1
(
853 mg, 100%). H NMR (400 MHz, CDCl3) * 2.89
◦
(dd, 2H, CH2CH CH2), 2.43 (dd, 2H, CH2CHVCH2),
centrifuged at 6000g for 5 min at 4 C. The pellet was
2
2
.19 to 2.12 (m, 1H, CH2CHCH2), 1.40 to 1.26 (m,
homogenized in 0.5 mL fresh HBSS on ice using a
Potter–Elvehjem homogenizer attached to a Hitachi
8H, CH2CH2V), 0.88 (t, 3H, CH3). 13C NMR (100
DOI 10.1002/jps
JOURNAL OF PHARMACEUTICAL SCIENCES

4
HACKETT ET AL.
(Tokyo, Japan) 12-Volt 3/8” cordless peak lithium ion
The precipitate was pelleted at 800g for 2 min and the
micro driver drill at 1300 rpm while continuously
plunging for 15 min. After centrifugation at 6000g,
the supernatant was diluted to 1 mL and stored in
supernatant removed for analysis on an Aquasil C18
˚
column (2.1 × 50 mm) with 5 :m particles and 80 A
pore size using 20 :L injections.
◦
1
00-:L aliquots at −18 C until use. Prior to use, the
The eluent was a binary gradient containing 0.1%
formic acid in water (A) and 0.1% formic acid in
MeOH (B). The gradient began with 1 min of 30%
B, followed by a linear increase over 2 min to 95%
B. The eluent was then held constant at 95% B for
15 s, followed by a 30 s linear decrease down to 30%
B and 30 s re-equilibration at 30% B. The flow rate
and sample volume were 0.75 mL/min and 100 :L,
respectively. Analysis was by an API 4000 Triple
Quadrupole LC–MS–MS mass spec (AB Sciex Instru-
ments, Framingham, Massachusetts) scanning for
parent to daughter transitions of 854 to 525 m/z (PTX)
and 854 to 525 m/z (PP) in negative mode. The 1178
to 854 transition for PP was inconsistent and thus the
daughter PTX and its subsequent daughter fragment
were used for analysis for both the standard curve
and stability study. Concentrations were determined
using peak area per the standard curves.
aliquots were diluted another twofold with HBSS.
Tumor tissue homogenate was prepared by inoc-
5
ulating 5 × 10 CT26 cells on the right flank of
an 8-week-old female BALB/c mouse, and the tu-
3
mor allowed to grow to approximately 1000 mm . At
this point, the mouse was anesthetized with 0.1 mL
1
00 mg/mL intraperitoneal (i.p.) ketamine with sub-
sequent cervical dislocation. The tumor mass was ex-
cised, washed with phosphate-buffered saline (PBS)
and blotted dry. The tumor was weighed and di-
luted with two volumes of PBS. The suspension was
coarsely homogenized on ice using a POWERGEN
7
00 tissue homogenizer (Fisher Scientific, Waltham,
Massachusetts) for 5 min until no lumps of tissue
remained. The tissue homogenate was mixed thor-
oughly, separated into 1 mL aliquots, and stored at
◦
−
18 C until use.
For serum preparation, eight female 8-week-old
Formulation of PP with Albumin and PTX as TAXOL
BALB/c mice were anesthetized with ketamine un-
til unresponsive to toe pinching. The mice were then
exsanguinated by cardiac puncture with subsequent
cervical dislocation. The whole blood was pooled and
The concentration of PP in a stock solution in t-
BuOH was confirmed via UV absorption in EtOH
−1
−1
(ε228 = 30,593 M cm , EtOH). The PP stock was
then added to a 4% or 5% (w/v) delipidated HSA
in water at a 1:10 final volume ratio of EtOH–HSA
◦
stored at 4 C overnight. The fibrin clot was pelleted
◦
by centrifuging the suspension at 6000g at 4 C and
−1
−1
29
the supernatant removed and separated into 500 :L
aliquots.
(ε278 = 39,800 M cm , water), which immediately
caused turbidity due to precipitation of PP. The sam-
ple was shell frozen and lyophilized over 24 h. The
lyophilized powder was then reconstituted in 0.1 M
phosphate buffer at pH 7.4 and filtered through a
0.22 :m polyvinylidene fluoride (PVDF) filter (Milli-
pore, Billerica, Massachusetts). If the stock contained
The extraction efficiencies of both PP and PTX
were determined in duplicate using the individual mi-
lieu that was spiked with radiolabeled stock solution
3
3
of either [ H]-PP or [ H]-PTX. Aliquots of the stock
were weighed out and analyzed by liquid scintilla-
tion counting (LSC) to determine the concentration.
Next, the matrices were extracted with one volume
of 1 N HCl and eight volumes of MeCN. The samples
were centrifuged at 16.1 × 1000 g for 5 min and the
supernatants quantitated by LSC. Efficiencies were
calculated as a ratio of concentration found from the
extraction over the initial concentration in the matrix.
The disappearance of PP and concomitant appear-
ance of PTX in the test milieus were monitored by a
liquid chromatography (LC)–MS procedure. Standard
curves were generated by combining PP (160 to1 :M)
and PTX (1–160 :M) stock solutions in EtOH with
aliquots of the test milieu (serum, cell, or tissue ho-
mogenate) at a 1:10 (v/v) ratio, respectively. The mix-
tures were then diluted with equal volumes of 1 N HCl
and precipitated with eight volumes of MeCN contain-
ing 500 nM docetaxel as internal standard. The sta-
bility of the PP was carried out at 16 :M in duplicate
3
[ H]-PP, it was quantified via LSC. If the stock is not
radiolabeled, it is extracted by diluting with one vol-
ume of 1 N HCl and eight volumes of MeCN, then
added a quantity sufficient of EtOH to make the vol-
ume 5 mL in a volumetric flask for UV–visible anal-
ysis. The loading efficiency of PP varied depending
on such processing parameters as injection rate, stir-
ring dynamics, concentration of albumin, and ionic
strength, with a maximum loading of about 50% for
higher doses. TAXOL was prepared by dissolving PTX
at 7.5 mg/mL in a 1:1 (v/v) mixture of EtOH–Cre-
mophor EL. The drug concentration was confirmed by
diluting with EtOH and analyzing by UV absorption
−1
−1
(ε228 = 29,800 M cm , EtOH). The organic stock
was diluted fivefold with a phosphate buffer to pro-
duce a concentration of about 1.5 mg/mL immediately
prior to injection.
Dialysis of PP–HSA Complex
◦
at 37 C while shaken continuously. At given times,
2
0 :L aliquots were mixed with 20 :L 1N HCl fol-
To determine whether the PP–HSA complex would
dissociate over time, the complex was dialyzed against
lowed by 160 :L MeCN containing 500 nM docetaxel.
JOURNAL OF PHARMACEUTICAL SCIENCES
DOI 10.1002/jps

PACLITAXEL DERIVATIVE FOR ALBUMIN-ASSISTED DRUG DELIVERY
5
delipidated human serum albumin in phosphate
buffer to create sink conditions. Dialysis was con-
ducted across a Spectra Por 12–14 kDa molecular
weight cutoff (MWCO) regenerated cellulose dialysis
membrane with 500 :L fiberglass compartments in
a custom-made dialysis block.0 A 0.1 M phosphate
buffer pH 7.4 containing 0.01% NaN3 was diluted
traacetic acid disodium, followed by 300 :L 30% H2O2
in 100 :L aliquots. The reaction can be quite vigor-
ous and was allowed to continue at room temperature
until the foaming subsided. The samples were then
◦
heated to 55 C for an additional hour, yielding clear
to slightly yellow solutions. The samples were cooled
to ambient and 15 mL Ultima Gold scintillation cock-
tail was added and mixed thoroughly. The samples
were rested in the liquid scintillation counter for 1 h
to adapt to light and temperature prior to counting.
1
0:1 (v/v) with EtOH. This EtOH-containing buffer
3
was saturated by shaking with solid [ H]-PP for 24 h
at 37 C, then cooled to ambient temperature, and fil-
◦
ꢀ
R
tered through a 0.22 :m PVDF filter. Albumin was
dissolved in phosphate buffer without EtOH at con-
centrations of 40 and 10 mg/mL as confirmed by UV.
To the organs was added 1 mL of Solvable fol-
◦
lowed by heating to 55 C for 2 h. Once dissolved (yel-
low to slightly yellow), the samples were cooled to am-
3
Finally, a stock solution of [ H]-PP in EtOH was
bient and 200 :L of 30% H2O2 was added in 100 :L
7
◦
made to a concentration of 15 mg/8.63 × 10 dpm/mL.
aliquots. The samples were then heated to 55 C again
The PP stock was added to the albumin stocks in a
for 30 min. The samples were cooled to ambient and
15 mL Ultima Gold scintillation cocktail was added
and thoroughly mixed. The samples were rested in
the LSC for 1 h to adapt to light and temperature
prior to counting.
1
:10 (v/v) ratio and the stocks were filtered through a
0.22 :m PVDF filter. These stocks were analyzed by
LSC for PP content and UV for HSA content.
3
The saturated solution of [ H]-PP in 0.1 M phos-
phate buffer pH 7.4 was dialyzed against blank phos-
phate buffer for 5 h with sampling of both compart-
ments every hour to determine how quickly free PP
could establish equilibrium. The albumin solutions
loaded with PP were sampled after shaking for 24 h
Aliquots of PP stock were weighed in triplicate
and analyzed via LSC. A tumor-bearing mouse was
then anesthetized with 0.1 mL 100 mg/mL ketamine
i.p. and exsanguinated via cardiac puncture. The
mouse was sacrificed and the target organs har-
vested. Aliquots of 175, 200, and 225 :L of blood were
weighed and spiked with a weighed amount of PP
◦
at 37 C and the compartment replenished with fresh
aliquots of the initial contents. After another 24 h of
◦
4
shaking at 37 C, all contents were analyzed. Analy-
stock (∼1 × 10 dpm). In addition, approximately 75,
sis involved weighing 200 :L aliquots of the sample
into scintillation vials and adding 5 mL Ultima Gold
scintillation cocktail followed by LSC.
100, and 125 mg samples of organs were washed, blot-
ted dry, and weighed, then spiked with a weighed
4
amount of PP stock (∼1 × 10 dpm). The samples were
then processed in the same fashion as the unknown
samples. The counting efficiency was calculated as the
ratio of dpm in the organ sample/calculated PP stock
dpm × 100. The efficiencies were plotted as a func-
tion of organ or blood mass and fitted with a linear
regression. The calculated efficiency was then used to
correct the counts for the unknown samples based on
blood volume or organ mass.
3
Pharmacokinetics/Biodistribution of [ H]-PP–HSA and
3
[
H]-TAXOL
Female, 6–8-week- old BALB/c mice were adminis-
5
tered 5 × 10 CT26 murine colon carcinoma cells in
0
.1 mL HBSS on the right flank. When the tumors
3
were approximately 1000 mm , the mice received the
radiolabeled dose as formulated in 4% HSA in 0.1 M
phosphate buffer (0.08 mg PP eq to 0.058 mg PTX/
7
1
2
.75 × 10 dpm/0.1 mL) or as TAXOL (0.147 mg PTX/
6
.22 × 10 dpm/0.1 mL) via the tail vein. Approxi-
RESULTS
mately 5 min prior to the timepoint, the mice were
injected i.p. with 0.1 mL 100 mg/mL ketamine. Once
unresponsive to toe pinching, the mice were exsan-
guinated via cardiac puncture with subsequent cervi-
cal dislocation. The whole blood was quickly weighed
as 200 :L aliquots in triplicate. The target organs
were harvested, washed with PBS, blotted dry and
small (∼100 mg) pieces excised, weighed, and stored
in scintillation vials in duplicate. The target organs
include the liver, lungs, heart, kidneys, spleen, tumor,
and injection site on the tail.
Chemistry
The synthesis of 3-alkylglutaric anhydrides has alrea-
3
0
dy been demonstrated. This synthetic method was
shortened and modified (Fig. 1). Hexadecanol was
oxidized to palmitaldehyde using pyridiniumchloro-
chromate (PCC) and subsequently subjected to
Wittig olefination using the ylide (carboethox-
ymethylene)triphenylphosphorane. 1 The ",$-uns-
aturated ethyl ester can then be used as a
Michael acceptor for diethyl sodiomalonate at room
2
7
3
1
To the whole blood aliquots was added 1 mL of
temperature. The resultant triester is decarboxy-
lated and saponified to the 3-alkylglutaric acid using
KOH; NaOH resulted in incomplete saponification.
ꢀ
R
Solvable tissue solubilizer. The sample was incu-
◦
31
bated for 1 h at 55 C turning brownish- green. To
this was added 100 :L of 0.1 M ethylenediaminete-
Dehydration of the diacid can efficiently be affected
DOI 10.1002/jps
JOURNAL OF PHARMACEUTICAL SCIENCES

6
HACKETT ET AL.
Figure 1. Synthesis of PDG anhydride. (i) pyridiniumchlorochromate, NaOAc, (car-
o
boethoxymethylene)triphenylphosphorane, DCM RT, 85%; (ii) Na , EtOH, diethylmalonate RT
overnight, 1 eq H2O, 2 eq NaCl, DMSO reflux 5 h, 94%; (iii) 3 N KOH/MeOH reflux 3 h, 99%;
◦
(
iv) trifluoroacetic anhydride, EtOAc,37 C,2h,100%.
with trifluoroacetic anhydride in EtOAc at 37 C.32
PDG anhydride itself is reactive to amines and alco-
hols, thus it could react with PTX at three positions.
PTX is known to be reactive to glutaric anhydride
◦
particular concentration of cell homogenate attaining
under 1% conversion to PTX after 36 h. The PP is
similarly stable in serum attaining about 1.5% con-
version to PTX within 72 h. Under the experimental
conditions, the PP was also stable in the tumor tissue
homogenate only attaining 5.5% conversion to PTX
after 72 h.
ꢀ
at the 2 -OH position at room temperature. The 7-
OH is only reactive under heated conditions and the
1
-OH is unreactive.33,34 Mixing PDG anhydride and
PTX in pyridine with a catalytic amount of DMAP al-
lows complete reaction at room temperature in 24 h
Pharmaceutics
(
Fig. 1).
Paclitaxel is very poorly soluble in water (∼10 :g/
1
mL) and adding a fatty acid to this molecule only
Cytotoxicity In Vitro
exacerbates the issue. The solubility of PP was cal-
The IG50, TGI and LC50 were determined for PP and
PTX via the NCI-60 DTP Human Tumor Cell Line
Screen. The results for all three values consistently
showed a superior cytotoxicity of PTX to PP by at least
an order of magnitude and up to several orders; 0.2-
3
culated using [ H]-PP to be 0.36 :g/mL (0.26 :g/
mL PTX) in 0.1 M phosphate buffer at pH 7.4, room
temperature. In the presence of HSA, the solubility
3
:M for PP and 0.02-1 :M for PTX. For simplicity,
representative data from the TGI data is presented
Table 1). Because of the large number of cell lines,
Table 1. Cytotoxicity of PTX and PP in Various Cancer Cell
Lines
(
TGI
:M
:M
PP
only two were selected from each cancer type which
demonstrated the greatest toxicity for PTX and PP.
Cancer Type
Leukemia
Cell Line
HL-60(TB)
RPMI-8226
NCI-H522
NCI-H460
HC-2998
KM12
PTX
0.03
0.32
0.08
0.79
0.13
0.16
0.06
0.08
0.02
1.00
0.16
0.79
0.50
1.00
0.16
0.63
0.79
1.00
1.78
2.75
2.40
1.35
1.95
1.26
1.78
1.20
0.20
1.78
0.60
2.63
2.29
2.69
1.23
3.24
2.24
0.27
Biochemical Stability
NSCLC
Colon
3
Prior to the experiment, stock solutions of [ H]-PP
and [ H]-PTX in EtOH were introduced to various
3
milieu at a 1:10 (v/v) ratio. Subsequent to incubation
for a given period of time, samples were taken and ex-
tracted with one volume of 1 N HCl and eight volumes
of MeCN, vortexed and pelleted by centrifugation. For
undiluted murine serum, PP and PTX were quan-
titatively extracted. For CT26 cytosolic fraction, PP
was extracted at 97% efficiency and PTX at 91%. For
CT26 tissue homogenate, PP was extracted at 90% ef-
ficiency and PTX at 94% efficiency. Results of the sta-
bility study are presented as an average of duplicates
and results from LC–MS were corrected for extrac-
tion efficiency. The stability of PP in serum, tumor cy-
tosol, and tumor tissue homogenate was determined
as a function of time (Fig. 3). The PP is stable in this
CNS
SF-539
SNB-75
Melanoma
Ovarian
Renal
MDA-MB-435
LOX IMVI
OVCAR-3
OVCAR-5
RXF 393
SN12C
DU-145
PC-3
BT-549
MDA-MB-468
Prostate
Breast
Total growth inhibition of PP and PTX from the NCI-60 DTP human
tumor cell line screen. For simplicity, only the TGI data are presented for
two cell lines representative of each type of cancer. The cell lines were
selected as having the highest toxicity for PTX and PP.
JOURNAL OF PHARMACEUTICAL SCIENCES
DOI 10.1002/jps

PACLITAXEL DERIVATIVE FOR ALBUMIN-ASSISTED DRUG DELIVERY
7
Figure 2. Synthesis of PP. (i) 3-Pentadecylglutaric anhydride, DMAP,pyridine,RT 24 h, 85%.
3
H corresponds to the sites replaced with tritium in the radiolabeled compounds.
Figure 3. Biochemical stability of PP in undiluted murine serum, CT26 murine tumor cell
homogenate, and CT26 tumor tissue homogenate. The data are presented as a percentage of
total mass quantified from LC–MS. Error bars represent the standard deviation of matrices
performed in duplicate with aliquots quantified in triplicate.
DOI 10.1002/jps
JOURNAL OF PHARMACEUTICAL SCIENCES

8
HACKETT ET AL.
of PP should increase dramatically to a theoretical
maximum of the number of fatty acid binding sites
multiplied by the concentration of HSA. This would be
a solubility of about 3.6 mM (3.07 mg/mL with respect
to PTX) conservatively assuming a 40 mg/mL physio-
logical solution of albumin and six fatty acid binding
sites. After 2 weeks of equilibration, the concentration
of PP with respect to PTX (54 :g/mL) was substan-
tially less than the theoretical maximum (data not
shown).
To increase dissolution, PP was dissolved in a
water-miscible cosolvent, t-BuOH, and introduced to
aqueous albumin at a 1:10 (v/v) ratio. Albumin was
found to be structurally stable to t-BuOH at this
concentration via circular dichroism analysis (data
not shown) and no aggregation was observed by DLS
carboxylate anion (Supporting Information 2). The
PDG–FLA–miniPEG and SA–FLA–miniPEG conju-
gated fatty acids as well as the corresponding uncon-
jugated fatty acids were then introduced to albumin
and analyzed by DLS with no observable aggregation
(Supporting Information 1).
The extremely poor aqueous solubility of PP pre-
cluded determination of the dissociation constant via
conventional methods such as isothermal titration
calorimetry or equilibrium dialysis. Instead, the sta-
bility of the PP–HSA complex was examined. Dialy-
sis of a PP solution without HSA across a 12–14 kDa
MWCO membrane over 5 h showed that a rapid equi-
librium could be attained within this time period.
In contrast, dialysis of the PP–HSA complex against
comparable concentrations of delipidated HSA, which
should act as a sink for free PP, showed only about 3%
of PP was released from the complex over 2 days at
(
Supporting Information 1). This is not surprising as
other small alcohols at this concentration do not per-
turb its secondary or tertiary structure.35 Distilled
water was chosen as a solvent as electrolytes tend
to salt-out PP, leading to precipitation. After loading,
the suspension was shell frozen and lyophilized to
a cake, which was reconstituted in an isotonic solu-
tion. Filtration of the suspension yields a clear blue
solution, indicative of Tyndall scattering. Analysis by
DLS showed aggregates of 120 nm diameter (Support-
ing Information 1). Formulation of unmodified PTX
via this same procedure also led to a blue solution of
PTX–HSA aggregates. To test if this phenomenon was
PTX specific, two water-soluble conjugates were syn-
thesized using fluoresceinamine (FLA) attached to a
short tetraethyleneglycol (miniPEG) linker as a drug
surrogate. The FLA–miniPEG was attached to either
PDG or stearic acid (SA) such that the only difference
between the two conjugates was the presence of the
◦
37 C for both 4% and 1% concentrations of HSA (Ta-
ble 2). To determine the effect of the anionic charge
from the PDG conjugates, the water-soluble FLA con-
jugates were introduced to albumin and analyzed by
isothermal titration calorimetry. Initial results sug-
gest the carboxylate from PDG does bring about a
3.5-fold increase in binding affinity, twofold decrease
in enthalpy, and about a twofold decrease in entropy
(Supporting Information 3).
Pharmacokinetics/Biodistribution
The PP formulation shows a rapid distribution (Fig. 4)
from the vasculature over 6 h accounting for a 10-fold
dilution in serum concentrations. The concentration
then rebounds to about 20% of the initial concentra-
tion with a secondary peak that is within the quan-
tified error. This concentration is then cleared very
3
Table 2. Dynamic Dialysis of [ H]-PP–HSA Complexes
A
B
Time (h)
HSA (:M)
EtOH [% (v/v)]
PP (:M)
HSA (:M)
EtOH [% (v/v)]
PP (:M)
0
1
2
3
4
5
0
0
0
0
0
0
0
0
10
0
10
10
10
10
10
10
1.3
10
10
0
2.8
2.1
1.8
1.7
1.5
1.5
0
0
0
0
0
10
10
10
10
10
0
0.7
1.0
1.1
1.3
2.8
1.2
467
455
131
131
609
609
10
10
0
0
1.6
0
2
0
2
4
0
0
2
4
4
0
722
722
722
454
179
179
179
4
8
0
0
12
609
0
0
0
14
127
123
115
146
146
146
0
0
0
0
4
4
4
8
0
[3
H]-PP–HSA in compartment A was dialyzed against delipidated HSA in compartment B across a
regenerated cellulose membrane (MWCO 12–14 kDa) to assess the stability of the aggregates. After dialysis, the
membrane showed minimal PP binding.
JOURNAL OF PHARMACEUTICAL SCIENCES
DOI 10.1002/jps

PACLITAXEL DERIVATIVE FOR ALBUMIN-ASSISTED DRUG DELIVERY
9
3
3
Figure 4. Serum pharmacokinetics for [ H]-PP-HSA and [ H]-TAXOL in CT26 tumor-bearing
mice. The insert magnifies the terminal phase of each treatment which was used to calculate
the half-life, t1/2 (Table 3). Error bars represent the standard deviation calculated from n = 3
mice, with blood samples analyzed in triplicate.
slowly with a half-life approaching 24 h. Biodistri-
bution analysis shows accumulation primarily in the
liver and spleen (Fig. 5). The concentration in these
organs also decreases slowly, mirroring the clearance
from the serum allowing quantitation over 72 h. The
tumor rapidly accumulates 1%–2% of the dose per
gram tissue, which remains constant through 72 h.
The PTX from the TAXOL formulation shows a
similarly rapid distribution (Fig. 4). No rebound is
observed and the terminal half-life of the formula-
tion was estimated to be 7 h. Biodistribution anal-
ysis shows accumulation primarily in the liver and
kidneys (Fig. 5), which eliminate PTX quickly from
the system such that the lower limit of detection is
reached after 12 h. The tumor shows a comparable up-
take of about 3% of the dose per gram tissue; however,
concentrations approach the lower limit of detection
by 6 h.
3
3
Figure 5. Biodistribution analysis of [ H]-PP-HSA (a) and [ H]-TAXOL (b) in CT-26 tumor-
bearing mice. Error bars represent the standard deviation calculated from n = 3 mice, with
organ samples analyzed in duplicate.
DOI 10.1002/jps
JOURNAL OF PHARMACEUTICAL SCIENCES

1
0
HACKETT ET AL.
Table 3. Pharmacokinetic Parameters of PP–HSA and TAXOL
Treatment
Dose (:g)
C0 (:g/mL)
Vd (mL)
t1/2 (h)
Cl (mL/h)
AUC∞ [h/(:g mL)]
MRT∞ (h)
PP
TAXOL
PP/TAXOL
80
150
0.53
12.05
6.21
1.94
36.56
116.19
0.31
23.13
6.61
3.50
1.10
12.18
0.09
72.59
12.07
6.01
29.82
6.53
4.57
3
3
Summary of important pharmacokinetic parameters for [ H]-PP–HSA and [ H]-TAXOL. A noncompartmental model was
fit to the data using WinNonlin.
AUC, area under the curve; C0, estimated initial serum concentration; Cl, systemic clearance; Vd, volume of distribution;
MRT, mean residence time.
The serum pharmacokinetics were analyzed using
addition of EtOH or MeCN then precipitates the pro-
tein but dissolves the PP.
noncompartmental analysis. The major pharmacoki-
netic parameters were extracted and presented in Ta-
ble 3. The PP formulation shows a 3.5-fold increased
serum half-life due to a lower systemic clearance and
volume of distribution. Consequently, while at only
half the dose of TAXOL, PP shows a higher C0, area
under the curve (AUC), and mean residence time
The structural details of the particles are not
known. On the basis of published docking studies
from PTX and HSA, it is hypothesized that the PTX
moiety of the PP on the surface of albumin can inter-
act with other PTX moieties via van der Waals inter-
actions or bind other molecules of albumin through
3
7
(
MRT) in the system.
PTX-binding sites. Formulation of PTX without
the PDG moiety in this manner saturates at 0.5 eq
PTX–HSA, with formation of 100 nm particles as
confirmed by DLS (Supporting Information 1). This
finding suggests that particle formation is a PTX-
specific phenomenon, as also found with ABRAXANE.
This is further supported by the lack of aggrega-
tion of albumin upon binding PDG–FLA–miniPEG,
SA–FLA–miniPEG, PDG acid, or SA. Interestingly,
ABRAXANE allows solubilization of 10 equivalents of
PTX–HSA compared with a maximum of 0.5 by sim-
ple mixing. Formulation as ABRAXANE would thus
allow higher concentrations of PTX to be solubilized
compared with PP, which only has six potential bind-
ing sites for the fatty acid.
DISCUSSION
Formulation of PP–HSA Aggregates
The purpose of synthesizing the PP conjugate was to
improve the solubility and pharmacokinetics of PTX
by utilizing the endogenous serum protein, albumin.
Loading of albumin with PP as a buffered suspension
was exceedingly slow taking 2 weeks to reach only
5
4 :g/mL with respect to PTX. This is in contrast
to the expected 3 mg/mL at saturation of six binding
sites of 4% HSA. The slow binding could be a result
of rate-limited dissolution. To facilitate the kinetics
of binding, PP was dissolved in t-BuOH prior to mix-
ing with a solution of 4% HSA in water. Much of the
PP precipitates but this process can result in much
higher concentrations of bound PP; adding 12 eq of
PP allowed full saturation of six binding sites of HSA.
After loading, the solvents are subsequently sublimed
by lyophilization and the cake reconstituted in an iso-
tonic solution suitable for injection.
Cytotoxicity of PP
The PP conjugate is less cytotoxic than PTX toward a
variety of human cancer cell lines when tested in vitro
(Table 1). The mechanism of action of PTX is to bind
and prevent microtubulin assembly; however, numer-
ous PTX derivatives have been shown to relinquish
this binding capacity yet retain varying degrees of
2
5,38
The PP–HSA complex spontaneously formed
20 nm particles, which remain after lyophilization
cytotoxicity.
This appears to be regiospecific as
ꢀ
1
succinate esters on the 2 -OH or 7-OH also abolish
microtubule binding, whereas the 1-OH succinate
3
3
and reconstitution. Once the PP forms a complex
with HSA, it is exceedingly difficult to dissociate;
some failed attempts include surfactants with acyl
chains (tweens and SDS), high concentrations of urea,
high concentrations of water-miscible and water-
immiscible organic solvents, TCA precipitation. The
only reliable way to dissociate the complex was to
acidify the complex to pH 1 and precipitate with EtOH
or MeCN. The acidification step causes albumin to
unfold from the normal heart shaped N state to the
fully expanded E state, potentially exposing all the
fatty acid binding pockets.36 The acidification also
quenches the anion from the PP conjugate, making it
very water insoluble but highly organic soluble. The
3
9
ester retains it. The PP conjugate was shown to be
highly stable over 72 h in a variety of milieu including
undiluted murine serum, CT26 cellular cytosol, and
tumor tissue homogenate. The short time scale of 4 h
for the in vitro cytotoxicity assay precludes the more
active PTX from forming, suggesting the PDG moiety
likely affects microtubule bundling.
Pharmacokinetics/Biodistribution
Ideally, albumin-bound drugs will remain invisible to
the liver and reticuloendothelial system, as is the
case with IgG, due to the expression of FcRn on
4
0
monocytes. The release mechanism for bound drugs
JOURNAL OF PHARMACEUTICAL SCIENCES
DOI 10.1002/jps

PACLITAXEL DERIVATIVE FOR ALBUMIN-ASSISTED DRUG DELIVERY
11
would then be dilution-driven dissociation or release
due to digestion of the carrier albumin. Dissociation
from the protein is expected to be slowed due to the
increased binding affinity imparted by the anionic
charge of the PDG fatty acid. A dialysis experiment
aimed at assessing the off-rate of PP from albumin
shows it is very slow, even when sink conditions are
created by dialyzing against similar concentrations
of delipidated albumin (Table 2). This is not diffu-
sion rate limited as free PP was shown to equili-
brate across the membrane within hours with min-
imal membrane binding. Thus, spontaneous dissocia-
tion from the intact protein is unlikely to be a major
release mechanism for PP in vivo.
As the PP conjugate is utilizing albumin’s natural
utility as a fatty acid carrier but is not subject to rapid
dissociation, albumin is expected to impart its physi-
ological profile to the conjugate. Upon bolus intraven-
ous injection of unmodified albumin, a characteristic
two-phase profile is observed.41 The initial phase
shows a 60% removal of the dosed albumin representing
rapid distribution throughout all the extravascular
tissues and fluids. The slope of this phase is used to
calculate the transcapillary escape rate for albumin,
which is estimated to be approximately 4%–5% per
hour in humans. This is not irreversible clearance
from the body as the re-equilibration of the albumin
back into the vasculature via the lymphatic system
through the thoracic duct gives rise to the protracted
half-life of albumin. The second phase observed in the
albumin profile is metabolic clearance and the slope
is used to estimate the fractional degradation rate
of albumin, approximately 4% per day in humans.
This profile was also observed with albumin-based
therapies such as ABRAXANE42 as well as covalent
linkages to albumin such as through the single free
cysteine using a maleimide linker.43
trolling the affinity of the conjugate for albumin sim-
ply by reducing the alkyl length of the initial alcohol
(Fig. 1).
The volume of distribution, Vd, estimates the
amount of drug in the tissues as compared with the
amount of drug in the blood. Larger Vd values are
common for lipophilic drugs suggesting they tend to
permeate through the tissues of the body readily with
little remaining in the blood. Lower volumes of distri-
bution reflect hydrophilic or macromolecular drugs
that do not readily permeate the endothelium or
lipophilic drugs that are highly protein bound, caus-
ing retention in the vasculature. The apparent Vd for
PTX from the TAXOL formulation (120 mL) was much
higher than from the PP formulation (36 mL), sug-
gesting PTX from TAXOL permeates the tissues more
readily. Although both molecules are highly lipophilic
and bind albumin, the substantial decrease in Vd from
TAXOL to PP is indicative of a much stronger binding
of PP with albumin than TAXOL. Thus, PTX is ex-
pected to dissociate from albumin more readily than
PP in various tissues, particularly the liver, facilitat-
ing rapid clearance. In fact, the PTX showed a 12-
fold increase in clearance over PP. The PP formula-
tion shows a twofold higher initial concentration and
greater AUC compared with the TAXOL formulation
despite being half the dose; again due to the large Vd
for TAXOL. The longer retention within the vascu-
lature creates the potential for more drug to be ex-
posed to the tumor due to multiple passes through
the tumor vasculature with about a 4.5-fold increase
in MRT (Table 3).
Biodistribution analysis of the PP conjugate is sim-
ilar to unmodified albumin demonstrating uptake
throughout all major viscera but primarily the liver.
In the case of unmodified albumin, the liver is a pri-
mary destination due to the large pool of blood, but
4
1
This too is what was observed with the PP–HSA
complex, although the secondary peak has not been
seen with unmodified albumin before. This may not
be a real secondary peak as the pharmacokinetic data
has been repeated twice using a TAXOTERE-like for-
mulation of PP in which the calculated pharmacoki-
netic parameters are very similar but no secondary
peak is observed (Supporting Information 4). The ob-
served half-life of 23 h for the PP conjugate agrees
with recent data suggesting human serum albumin
is not protected by murine FcRn,4 leading to a half-
life approaching a day in mice. This is also comparable
to the half-life of ABRAXANE, which was found to be
it is not a major site of degradation. The biodistri-
bution data (Fig. 5) shows that the PP is not rapidly
cleared from any of the major organs and in fact the
concentrations mimic the concentration in serum, in-
cluding the secondary peak. This suggests that the PP
is not being deposited or rapidly degraded in any ma-
jor tissue, similar to native albumin. No specific anal-
ysis was conducted in vivo to determine if the radio-
label in the PP experiment is from intact PP or PTX;
however, it is likely intact PP because the pharma-
cokinetics are much more similar to that of albumin-
bound formulations (e.g. maleimide-based drugs) sug-
gesting the PP–HSA complex is very tight. In ad-
dition, when PP was formulated like TAXOTERE,
the docetaxel equivalent of TAXOL, the pharmacoki-
netics (Supporting Information 4) and biodistribu-
tion (Supporting Information 5) were very similar to
PP–HSA. If the PP is hydrolyzing to PTX, which binds
albumin weakly, then the kinetics should be more
similar to that of the TAXOL formulation. The
4
1
7.1 h; in the same study TAXOL was found to have
4
5
a half-life of 4.0 h. Although this may seem incon-
sequential, it is important to mention the half-life of
albumin increases proportionally with size.4 If the
complex mimics the half-life of albumin in humans,
as it does in mice, a 19-day half-life would be pre-
dicted. The half-life could then be modulated by con-
6
DOI 10.1002/jps
JOURNAL OF PHARMACEUTICAL SCIENCES

1
2
HACKETT ET AL.
TAXOL formulation shows rapid equilibration into
the liver and rapid elimination from the liver. The tu-
mor uptake is low for both formulations at 1%–3% of
the injected dose per gram tissue, but like the serum
concentrations of PP the tumor concentrations are
also sustained over 3 days. Conversely, for the TAXOL
formulation, the serum and tumor concentrations of
PTX decrease quickly.
Submicron particles, including albumin, do not dif-
fuse far into high-pressure tumors.47 In order for the
PP–HSA complex to be cytotoxic, the albumin must
be degraded in the tumor environment releasing PP,
which can then diffuse to the vicinal tumor cells. PP
can then be activated by hydrolysis to PTX. From the
literature, albumin digestion in the tumor vicinity is
expected to be rampant.17 Conversely, the hydroly-
sis of PP to PTX (Fig. 3) appears to be slow in both
CT26 tumor cell cytosol as well as the tumor tissue
homogenate. Because of the decreased cytotoxicity of
the conjugate, greater doses of PP will be required in
the tumor environment, which could be possible due
to the protracted half-life of the carrier albumin.
is currently licensed to NovoLipid, Inc. of which they
are both founders.
REFERENCES
1
2
3
. Trissel LA. 1997. Pharmaceutical properties of paclitaxel and
their effects on preparation and administration. Pharma-
cotherapy 17:133S–139S.
. Hawkins MJ, Soon-Shiong P, Desai N. 2008. Protein nanopar-
ticles as drug carriers in clinical medicine. Adv Drug Deliv Rev
60:876–885.
. Sparreboom A, Scripture CD, Trieu V, Williams PJ, De T, Yang
A, Beals B, Figg WD, Hawkins M, Desai N. 2005. Compara-
tive preclinical and clinical pharmacokinetics of a cremophor-
free, nanoparticle albumin-bound paclitaxel (ABI-007) and pa-
clitaxel formulated in Cremophor (Taxol). Clin Cancer Res
11:4136–4143.
4
. Gradishar WJ, Tjulandin S, Davidson N, Shaw H, Desai
N, Bhar P, Hawkins M, O’Shaughnessy J. 2005. Phase III
trial of nanoparticle albumin-bound paclitaxel compared with
polyethylated castor oil-based paclitaxel in women with breast
cancer. J Clin Oncol 23:7794–7803.
5
. Andersen JT, Sandlie I. 2009. The versatile MHC class I-
related FcRn protects IgG and albumin from degradation: Im-
plications for development of new diagnostics and therapeu-
tics. Drug Metab Pharmacokinet 24:318–332.
6. Kratz F. 2008. Albumin as a drug carrier: Design of pro-
drugs, drug conjugates and nanoparticles. J Control Release
CONCLUSIONS
132:171–183.
The utter insolubility of PP requires formulation in
a solution containing albumin prior to injection. Sev-
eral issues arose from this including PP precipita-
tion during formulation, protein aggregation, and a
decrease in activity with little hydrolysis of the fatty
acid. Admittedly, PTX was not an ideal choice for PDG
but nevertheless the PP-containing formulation is ca-
pable in providing: (1) a solution devoid of any un-
desirable solvents or detergents; (2) a cost-effective
method for high affinity albumin binding via the sim-
ple reaction with PDG’s one-step chemistry; (3) sim-
ilar characteristics to ABRAXANE; and (4) superior
pharmacokinetics. As is the case with ABRAXANE,
albumin may also facilitate increased tumor accumu-
lation, which can make PDG a very useful tool in
improving cancer drug therapies. This could be true
not only in the cancer model, but the success of LEV-
EMIR demonstrates albumin is a useful carrier for
therapies of many disease states.
7. Stehle G, Sinn H, Wunder A, Schrenk HH, Stewart JC,
Hartung G, Maier-Borst W, Heene DL. 1997. Plasma protein
(albumin) catabolism by the tumor itself—Implications for tu-
mor metabolism and the genesis of cachexia. Crit Rev Oncol
Hematol 26:77–100.
8
9
. Sinn H, Schrenk HH, Friedrich EA, Schilling U, Maier-Borst
W. 1990. Design of compounds having an enhanced tumour
uptake, using serum albumin as a carrier. Part I. Int J Rad
Appl Instrum B 17:819–827.
. Schilling U, Friedrich EA, Sinn H, Schrenk HH, Clorius
JH, Maier-Borst W. 1992. Design of compounds having en-
hanced tumour uptake, using serum albumin as a carrier—-
Part II. In vivo studies. Int J Rad Appl Instrum B 19:685–
695.
1
1
1
0. Maeda H. 2010. Tumor-selective delivery of macromolecular
drugs via the EPR effect: Background and future prospects.
Bioconjug Chem 21:797–802.
1. Schnitzer JE. 1992. gp60 is an albumin-binding glycoprotein
expressed by continuous endothelium involved in albumin
transcytosis. Am J Physiol 262:H246–H254.
2. Desai N, Trieu V, Yao Z, Louie L, Ci S, Yang A, De T,
Beals B, Dykes D, Noker P, Yao R, Labao E, Hawkins M,
Soon-Shiong P. 2006. Increased antitumor activity, intra-
tumor paclitaxel concentrations, and endothelial cell trans-
port of cremophor-free, albumin-bound paclitaxel, ABI-007,
compared with cremophor-based paclitaxel. Clin Cancer Res
ACKNOWLEDGMENTS
1
2:1317–1324.
We would like to acknowledge Dr. Arlene Bridges
of the UNC ADME Mass Spectrometry Center for
preparing a protocol for separation and quantita-
tion of our compounds by LC–MS and the NCI/NIH
Developmental Therapeutics Program for providing
cytotoxicity analysis. Funds were provided by NIH
R01 CA126825 and the North Carolina Biotechnol-
ogy Center Grant #2011-CFG-8003 including funds
from NovoLipid, Inc. Drs. Cho and Hackett are in-
ventors on a patent involving PDG anhydride, which
1
3. Young BA, Wang P, Goldblum SE. 1998. The counteradhesive
protein SPARC regulates an endothelial paracellular pathway
through protein tyrosine phosphorylation. Biochem Biophys
Res Commun 251:320–327.
14. Desai NP, Trieu V, Hwang LY, Wu R, Soon-Shiong P,
Gradishar WJ. 2008. Improved effectiveness of nanoparticle
albumin-bound (nab) paclitaxel versus polysorbate-based do-
cetaxel in multiple xenografts as a function of HER2 and
SPARC status. Anticancer Drugs 19:899–909.
1
5. Gilles C, Bassuk JA, Pulyaeva H, Sage EH, Foidart JM,
Thompson EW. 1998. SPARC/osteonectin induces matrix
JOURNAL OF PHARMACEUTICAL SCIENCES
DOI 10.1002/jps

PACLITAXEL DERIVATIVE FOR ALBUMIN-ASSISTED DRUG DELIVERY
13
metalloproteinase 2 activation in human breast cancer cell
lines. Cancer Res 58:5529–5536.
6. Søreide JA, Lea OA, Kvinnsland S. 1991. Cytosol protein con-
tent and prognosis in operable breast cancer. Correlations with
steroid hormone receptors and other prognostic factors. Breast
Cancer Res Treat 20:25–32.
7. Andersson C, Iresj o¨ BM, Lundholm K. 1991. Identification of
tissue sites for increased albumin degradation in sarcoma-
bearing mice. J Surg Res 50:156–162.
8. van der Vusse GJ. 2009. Albumin as fatty acid transporter.
Drug Metab Pharmacokinet 24:300–307.
32. Buron F, Deguest G, Bischoff L, Fruit C, Marsais F. 2007. On
the racemisation of aspartic anhydride during its preparation.
Tetrahedron: Asymmetry 18:1625–1627.
1
33. Deutsch HM, Glinski JA, Hernandez M, Haugwitz RD,
Narayanan VL, Suffness M, Zalkow LH. 1989. Synthesis of
congeners and prodrugs. 3. Water-soluble prodrugs of taxol
with potent antitumor activity. J Med Chem 32:788–792.
34. Lin S, Fang K, Hashimoto M, Nakanishi K, Ojima I. 2000. De-
sign and synthesis of a novel photoaffinity taxoid as a potential
probe for the study of paclitaxel–microtubules interactions.
Tetrahedron Lett 41:4287–4290.
1
1
1
9. Simard JR, Zunszain PA, Hamilton JA, Curry S. 2006. Loca-
tion of high and low affinity fatty acid binding sites on human
serum albumin revealed by NMR drug-competition analysis.
J Mol Biol 361:336–351.
35. Kumar Y, Muzammil S, Tayyab S. 2005. Influence of fluoro,
chloro and alkyl alcohols on the folding pathway of human
serum albumin. J Biochem 138:335–341.
36. Rosenoer VM. 1977. Albumin structure, function and uses.
Pergamon Press. Oxford, United Kingdom.
37. Bertucci C, Cimitan S, Riva A, Morazzoni P. 2006. Binding
studies of taxanes to human serum albumin by bioaffinity
chromatography and circular dichroism. J Pharm Biomed Anal
42:81–87.
38. Bradley MO, Webb NL, Anthony FH, Devanesan P, Witman
PA, Hemamalini S, Chander MC, Baker SD, He L, Horwitz
SB, Swindell CS. 2001. Tumor targeting by covalent conju-
gation of a natural fatty acid to paclitaxel. Clin Cancer Res
7:3229–3238.
20. Spector AA. 1975. Fatty acid binding to plasma albumin. J
Lipid Res 16:165–179.
2
1. Cistola DP, Small DM, Hamilton JA. 1987. Carbon 13 NMR
studies of saturated fatty acids bound to bovine serum albu-
min. II. Electrostatic interactions in individual fatty acid bind-
ing sites. J Biol Chem 262:10980–10985.
2
2. Simard JR, Zunszain JR, Ha C-E, Yang JS, Bhagavan NV, Pe-
titpas I, Curry S, Hamilton JA. 2005. Locating high-affinity
fatty acid-binding sites on albumin by x-ray crystallogra-
phy and NMR spectroscopy. Proc Natl Acad Sci U S A
1
02:17958–17963.
39. Rice A, Michaelis ML, Georg G, Liu Y, Turunen B, Audus KL.
2003. Overcoming the blood–brain barrier to taxane delivery
for neurodegenerative diseases and brain tumors. J Mol Neu-
rosci 20:339–344.
2
3. Lambert DM. 2000. Rationale and applications of lipids
as prodrug carriers. Eur J Pharm Sci 11(Suppl 2):S15–
S27.
2
4. Sparreboom A, Wolff AC, Verweij J, Zabelina Y, van Zomeren
DM, McIntire GC, Swindell CS, Donehower RC, Baker SD.
40. Zhu X, Meng G, Dickinson BL, Li X, Mizoguchi E, Miao L,
Wang Y, Robert C, Wu B, Smith PD, Lencer WI, Blumberg
RS. 2001. MHC class I-related neonatal Fc receptor for IgG is
functionally expressed in monocytes, intestinal macrophages,
and dendritic cells. J Immunol 166:3266–3276.
2003. Disposition of docosahexaenoic acid-paclitaxel, a novel
taxane, in blood: In vitro and clinical pharmacokinetic studies.
Clin Cancer Res 9:151–159.
2
5. Dosio F, Reddy LH, Ferrero A, Stella B, Cattel L, Cou-
vreur P. 2010. Novel nanoassemblies composed of squalenoyl-
paclitaxel derivatives: Synthesis, characterization, and biolog-
ical evaluation. Bioconjug Chem 21:1349–1361.
6. Kurtzhals P, Havelund S, Jonassen I, Kiehr B, Larsen UD,
Ribel U, Markussen J. 1995. Albumin binding of insulins acy-
lated with fatty acids: Characterization of the ligand–protein
interaction and correlation between binding affinity and tim-
ing of the insulin effect in vivo. Biochem J 312(Pt 3):725–
41. Peters T.
37:161–245.
1985. Serum albumin. Adv Protein Chem
42. Nyman DW, Campbell KJ, Hersh E, Richardson K, Trieu V,
Desai N, Hawkins MJ, Von Hoff DD. 2005. Phase I and phar-
macokinetics Trial of ABI-007, a novel nanoparticle formula-
tion of paclitaxel in patients with advanced nonhematologic
malignancies. J Clin Oncol 23:7785–7793.
43. Xie D, Yao C, Wang L, Min W, Xu J, Xiao J, Huang M, Chen
B, Liu B, Li X, Jiang H. 2010. An albumin-conjugated peptide
exhibits potent anti-HIV activity and long in vivo half-life.
Antimicrob Agents Chemother 54:191–196.
2
731.
27. Shet J, Desai V, Tilve S. 2004. Domino primary alcohol
oxidation—Wittig reaction: Total synthesis of ABT-418 and
44. Andersen JT, Daba MB, Berntzen G, Michaelsen TE, San-
dlie I. 2010. Cross-species binding analyses of mouse and
human neonatal Fc receptor show dramatic differences
in immunoglobulin G and albumin binding. J Biol Chem
285:4826–4836.
45. Desai N, Yao Z, Trieu V, Soon-Shiong P, Dykes D, Noker P.
2003. Evidence of greater tumor and red cell partitioning and
superior antitumor activity of cremophor free nanoparticle pa-
clitaxel (ABI-007) compared to taxol. Proc Am Soc Clin Oncol
22. (abstr 625)
46. Nguyen A, Reyes AE, Zhang M, McDonald P, Wang WLT, Dam-
ico LA, Dennis MS. 2006. The pharmacokinetics of an albumin-
binding Fab (AB.Fab) can be modulated as a function of affinity
for albumin. Protein Eng Des Sel 19:291–297.
47. Lichtenbeld HC, Yuan F, Michel CC, Jain RK. 1996. Perfusion
of single tumor microvessels: Application to vascular perme-
ability measurement. Microcirculation 3:349–357.
(E)-4-oxonon-2-enoic acid. Synthesis 35:1859–1863.
2
8. Shoemaker RH. 2006. The NCI60 human tumour cell line an-
ticancer drug screen. Nat Rev Cancer 6:813–823.
9. Fang Y, Tong GC, Means GE. 2006. Structural changes
accompanying human serum albumin’s binding of fatty
acids are concerted. Biochim Biophys Acta 1764:285–
2
291.
3
0. Poldy J, Peakall R, Barrow RA. 2008. Pheromones and analogs
from Neozeleboria wasps and the orchids that seduce them: A
versatile synthesis of 2,5-dialkylated 1,3-cyclohexanediones.
Tetrahedron Lett 49:2446–2449.
1. Alvarez E, Cuvigny T, du Penhoat CH, Julia M. 1988. Syn-
theses with sulfones XLIX: Stereo- and enantioselective syn-
thesis of (s)-(-)-3,9-dimethyl 6-(1-methylethyl) (e)-5,8-decadien
3
1
4
-ol acetate, sexual pheromone of yellow scale. Tetrahedron
4:119–126.
DOI 10.1002/jps
JOURNAL OF PHARMACEUTICAL SCIENCES