A dendritic tetra(bisphosphonic acid) for improved targeting of proteins to bone

Article abstract of DOI:10.1002/anie.200500350

(Chemical Equation Presented) Close to the bone: A dendritic molecule that contains four bone-seeking bisphosphonic acid groups (see picture) has been prepared for targeting proteins to bone tissue. The tetra(bisphosphonic acid) molecule provides a high d

Full text of DOI:10.1002/anie.200500350

Communications
Bioorganic Chemistry
A Dendritic Tetra(bisphosphonic acid) for
Improved Targeting of Proteins to Bone**
Geeti Bansal, Jennifer E. I. Wright, Cezary Kucharski,
ˇ
and Hasan Uludag*
Protein-based therapeutic agents are being actively pursued
for the treatment of skeletal diseases.[1] As endogenous
regulators of cellular activity, proteins have the potential to
modulate cellular activity at skeletal sites to obtain a desired
tissue response. For example, protein-based cytokines that
can stimulate the formation of new bone or inhibit the cell
activity that is responsible for bone loss will be beneficial for
the treatment of diseases such as osteoporosis and arthritis.
Despite the demonstrated efficacy of several proteins in
preclinical disease models,[2] the clinical entry of most
proteins has been hampered as a result of undesired side
effects at extraskeletal sites. As systemically administered
proteins are not naturally targeted to skeletal tissue, the
desired efficacy is achieved only at exuberant doses of the
protein by nonspecific deposition at skeletal tissues. In this
case, pharmacological activity at extraskeletal sites becomes
unacceptable. If proteins could be engineered to exhibit a
strong affinity for bone, then systemic administration would
allow delivery of the proteins selectively to the skeletal sites.
Bone affinity can be imparted to a molecule by chemical
modification with bisphosphonic acids (BPs), a class of
compounds with exceptionally high affinity for the bone
mineral hydroxyapatite (HA). Bisphosphonic acids are chem-
ical analogues of endogenous pyrophosphate in which the
central, hydrolytically labile P-O-P linkage has been replaced
by the hydrolysis-resistant P-C-P bond. Intravenous admin-
istration of molecules that contain bisphosphonic acid groups
results in significant deposition (20–50% of injected dose) of
the molecules at bone tissue.[3,4] The exceptional affinity of
bisphosphonic acid derivatives for minerals has been utilized
to deliver radiopharmaceutical and imaging agents to skeletal
tissue.[5,6] To target proteins to bone, we devised conjugation
schemes by which 1-amino-1,1-diphosphonic acid (ami-
noBP)[7]
and
2-(3-mercapto-propylsulfanyl)-ethyl-1,1-
bisphosphonic acid (SH-BP)[8] were conjugated to proteins
ˇ+
[*] Dr. G. Bansal, J. E. I. Wright, C. Kucharski, Dr. H. Uludag
Chemical & Materials Engineering Department
526, Chemical & Materials Engineering Building
University of Alberta
Edmonton, Alberta T6G2G6 (Canada)
Fax: (+1)780-492-2881
E-mail: hasan.uludag@ualberta.ca
[+] Other addresses:
Department of Biomedical Engineering, Faculty of Medicine
and
Faculty of Pharmacy & Pharmaceutical Sciences
University of Alberta
Edmonton, Alberta T6G2G6 (Canada)
[**] Financial support was provided by the Canadian Institutes of Health
Research (CIHR) and the Whitaker Foundation.
3710
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
DOI: 10.1002/anie.200500350
Angew. Chem. Int. Ed. 2005, 44, 3710 –3714

Angewandte
Chemie
with heterofunctional linkers in aqueous medium.[9–14] The
affinity of proteins for hydroxyapatite in vitro was shown to
be proportional to the extent of conjugation of the bisphos-
phonic acid groups.[9,12,14] Intraosseous injection of the
aminoBP conjugates confirmed the high affinity in vivo of
the conjugates for bone[10] as well as their capability to “seek”
bone tissue after systemic injection in rats.[11] The length of the
tether between the bisphosphonic acid derivative and the
protein was found to be an important factor for the affinity of
the conjugates to minerals: conjugates with shorter tethers
displayed a higher mineral affinity than those with longer
tethers.[13] A higher density of the bisphosphonic acid ligand at
the vicinity of proteins was postulated for the beneficial effect
of shorter tethers. On the basis of these observations, the ideal
approach to target bone would rely on creating a high density
of bisphosphonic acids on proteins through short tethers. To
this end, we have prepared a new dendritic molecule
composed of four bisphosphonic acid moieties linked through
a benzene ring (8, COOH-4BP; Scheme 1). This compound
should provide a higher density of bisphosphonic acid groups
on the proteins as four BP units are attached per protein site,
unlike the conventional approach that relies on one-to-one
linkage (Scheme 2). A carboxylic acid moiety was also
incorporated into 8 to allow its conjugation to proteins
through a short (zero carbon atom) tether using carbodiimide
chemistry (Scheme 2).
adding a catalytic amount of p-toluenesulfonic acid mono-
hydrate to a solution of 2 in toluene and then heating the
mixture at reflux with a Soxhlet apparatus (CaH2) to give 3,
which was purified by chromatography on silica gel (60–270
mesh; 4:1 CHCl3/hexane as eluent). Meanwhile, 3,5-bis(3,5-
dinitrobenzoylamino)benzoic acid (6) was prepared from a
mixture of 3,5-diaminobenzoic acid (4; 19.7 mmol) and 3,5-
dinitrobenzoyl chloride (5; 43.4 mmol) in N,N-dimethylacet-
amide and was then reduced to 3,5-bis(3,5-diaminobenzo-
ylamino)benzoic acid (7) with H2 and 10% Pd/C.[16] Then
compound 3 (1.6 mmol) was coupled to the amine moieties of
7 (0.4 mmol) by an anti-Markovnikov reaction in THF
(15 mL) at 608C for 5 hours to give 3,5-bis[3,5-di(ethyla-
mino-2,2-tetraethylbisphosphonate)benzoylamino]benzoic
acid. After removal of the solvent, the residue was dissolved
in CH2Cl2 and the phosphonate esters were hydrolyzed with
bromotrimethylsilane (BrSi(CH3)3)[15] for 48 hours to afford
the dendritic tetra(bisphosphonic acid) 8 (Scheme 1).[17]
Bovine serum albumin (BSA) served as the model protein
for conjugation of the bisphosphonic acid derivative. Acid 8
was activated by treating equimolar concentrations (0–
20 mm) of NHS and EDC in MES (morpholinoethanesulfonic
acid) buffer (0.1m, pH 4.5) for 45 minutes. The activated acid
8 (0.15–1.25 mm) was then incubated with BSA (5 mgmLÀ1
)
in MES buffer for 3 hours, after which time the unreacted
components were removed by dialysis against 0.2m carbonate
buffer (pH 10; ꢀ 4) and deionized water (ꢀ 2). The dialyzed
samples were further purified by gel permeation chromatog-
raphy (Bio-Rad P2 gel-fine; 18 ꢀ 1.2 cm2 column) with
deionized water as eluent. Samples of BSA mixed with 8 in
the absence of EDC/NHS as well as BSA conjugated to SH-
BP[14] served as controls. These control samples were purified
by dialysis only. All conjugates were analyzed for an average
The synthesis of COOH-4BP (8) is outlined in Scheme 1.
Tetraethyl methylenebisphosphonate (1; 4 mmol) was con-
verted into tetraethyl ethylidenebisphosphonate (3) by reac-
tion, first, with formaldehyde (20 mmol) in diethylamine
(0.42 mL) according to the method of Degenhardt and
Burdsell to give the intermediate tetraethyl 2-methoxyethy-
lene-bisphosphonate (2).[15] Methanol was removed from 2 by
Scheme 1. Synthesis of tetra(bisphosphonic acid) 8. TsOH=p-toluenesulfonic acid, DMAc=N,N-dimethylacetamide.
Angew. Chem. Int. Ed. 2005, 44, 3710 –3714 ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
3711

Communications
Scheme 2. Conjugation of 8 (COOH-4BP) and SH-BP to model protein BSA. EDC=1-ethyl-3-(3-dimethylaminopropyl)carbodiimide,
NHS=N-hydroxysuccinimide, SPDP=N-succinimidyl 3-(2-pyridyldithio)propionate.
number of bisphosphonic acid substituents (i.e. the number of
BP groups attached per BSA) by using the Bradford assay for
protein concentrations[18] and the Ames assay[19] for phos-
phate concentrations.
8–BSA showed an increasing efficiency of conjugation (up to
3.6 molecules of 8 per BSA) as the concentration of EDC and
NHS was increased (Figure 1a). The binding of the conju-
gates to hydroxyapatite was assessed in phosphate buffer
(100 mm, pH 7.4), which was previously optimized for inves-
tigating such protein–HA binding.[7] Figure 1b summarizes
the relationship between the binding of the conjugates to
hydroxyapatite and the conjugation efficiency (that is, the
number of bisphosphonic acid groups per BSA) for SH-BP–
BSA and 8–BSA. Both conjugates exhibited an increased
level of binding to hydroxyapatite as a function of conjugation
efficiency. Whereas a plateau in the percentage binding to
hydroxyapatite was evident with SH-BP–BSA, no such
plateau was seen for 8–BSA, partly as a result of our inability
to obtain more than 3.6 bisphosphonic acid groups per BSA
(two other independent reactions did not yield higher
numbers of bisphosphonic acid substituents). This excellent
correlation between the conjugation efficiency and the
affinity for hydroxyapatite was in accord with the results
obtained previously for aminoBP and SH-BP conjugations
with hydroxyapatite.[9,14]
SDS-PAGE (sodium dodecylsulfate–polyacrylamide-gel
electrophoresis) analysis of the obtained proteins indicated
the migration of protein as single bands, which confirmed a
lack of protein–protein linking during the conjugation
reactions (data not shown). The number of bisphosphonic
acid substituents for a representative set of 8–BSA conjugates
([EDC] = [NHS] = 0–20 mm; [8] = 1.25 mm) are summarized
in Figure 1a. The control sample of BSA ([EDC] = [NHS] =
0 mm) revealed 0.8 molecules of 8 per BSA protein which
presumably represents the extent of free COOH-4BP (8) not
removed by dialysis (or gel chromatography). The conjugates
The conjugates were then analyzed for binding to
hydroxyapatite and bone in the presence of serum proteins.
Fresh bone matrix was obtained from Sprague–Dawley rats,[9]
and the serum proteins originated from tissue-culture grade
adult bovine serum. The use of bone matrix and serum
proteins in the binding medium provides more stringent
conditions for an assessment of binding and better represents
the situation in vivo. The BSA conjugates for this analysis had
similar levels of substitution by bisphosphonic acids: 3.4 and
3.6 for SH-BP–BSA and 8–BSA, respectively. Owing to the
presence of serum, 125I-labeled conjugates were used for this
analysis and were added to unlabeled (cold) proteins to give
Figure 1. a) Number of tetra(bisphosphonic acid) substituents 8 per
BSA conjugate upon coupling 8 (1.25 mm) to BSA in the presence of
equimolar quantities of EDC and NHS ([EDC]=[NHS]=0–20 mm).
b) Correlation between the number of bisphosphonic acid (BP) sub-
stituents per BSA protein conjugate (meanÆSD) and the percentage
binding to hydroxyapatite (HA; meanÆSD) in phosphate buffer
(100 mm). Note the linear increase in the percentage binding of the
*
conjugates at low numbers of BP groups for both SH-BP–BSA ( ) and
*
8–BSA ( ). See Ref. [9] for details about the methods.
3712
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
Angew. Chem. Int. Ed. 2005, 44, 3710 –3714

Angewandte
Chemie
approximately 106 cpm (counts per minute) at a concentration
of 0.1 mgmLÀ1 protein (hot/cold = 1:100).[9] In 30 and 60%
serum media, both conjugates exhibited a higher level of
binding to hydroxyapatite than unmodified BSA (Figure 2a).
conjugate 8–BSA, with a similar extent of bisphosphonic acid
substituents, gave 4.1- and 4.7-fold higher delivery (relative to
control BSA) at the femora and tibiae, respectively
(Figure 3). A similar result was obtained in a repeat study
in which the control BSA and 8–BSA were injected subcuta-
neously: delivery increased by 3.7- (femora) and 3.4-fold
(tibiae) as a result of conjugation of 8 (data not shown).
In conclusion, we have reported the synthesis of a novel,
dendritic tetra(bisphosphonic acid) 8. The synthesized mol-
ecule is introduced through a minimal tether length at the
attachment site and provides a high density of bisphosphonic
acid groups per protein site modified. Dendritic 8 gave higher
numbers of total bisphosphonic acid groups attached per
protein relative to SH-BP conjugates and exhibited its
intended effect, namely bone-targeting, at a lower extent of
protein modification which is an important consideration for
maintaining the pharmacological activity of proteins when
they are derivatized with bone-seeking ligands. The studies
reported here should further stimulate the efforts to design
“bone-seeking” proteins with minimal degree of modifica-
tion.
Figure 2. Percentage binding of BSA protein (&) and its conjugates
&
&
(SH-BP–BSA , 8–BSA ) to a) HA and b) bone in serum-containing
media (200 mm phosphate buffer). For both binding matrices, 8–BSA
gave the highest extent of binding, followed by SH-BP—BSA, while
only a low level of binding was observed with the control (BSA protein)
under all conditions. See Ref. [9] for details about the methods.
Conjugates prepared with 8 exhibited a higher level of
binding to hydroxyapatite relative to SH-BP–BSA, despite
the same number of bisphosphonic acid substituents. A
similar result was obtained when hydroxyapatite was replaced
by bone as the binding matrix (Figure 2b), with the 8–BSA
conjugate exhibiting superior levels of binding than the SH-
BP–BSA conjugate.
Received: January 29, 2005
Published online: May 6, 2005
Keywords: bioorganic chemistry · bone-targeting ·
.
drug delivery · protein engineering
Finally, the ability of these two conjugates to target bone
was evaluated in rats. The 125I-labeled conjugates (as above)
were injected intravenously through the tail vein and bone
(femora and tibiae), and deposition of the proteins was
assessed by explanting the bones and determining the counts
in these tissues. The depositions of the control BSA and the
SH-BP–BSA conjugate at the bone were similar at both
femora and tibia, which indicates a lack of targeting by the
SH-BP–BSA conjugate (Figure 3). The presence of approx-
imately 3.4 bisphosphonic acid substituents per BSA protein
was not expected to impart a strong affinity to bone, as our
previous studies required around 10 bisphosphonic acid
groups per BSA for successful bone-targeting.[11] However,
[1] D. Puleo, BioDrugs 2003, 17, 301 – 314.
ˇ
[2] S. A. Gittens, H. Uludag, J. Drug Targeting 2001, 6, 407 – 429.
[3] A. G. Porras, S. D. Holland, B. J. Gertz, Clin. Pharm. 1999, 36,
315 – 328.
[4] G. M. Blake, S. J. Park-Holohan, G. J. Cook, I. Fogelman, Semin.
Nucl. Med. 2001, 31, 28 – 49.
[5] C. L. De Ligny, W. J. Gelsema, T. G. Tji, Y. M. Huigen, H. A.
Vink, Nucl. Med. Biol. 1990, 17, 161 – 179.
ˇ
[6] H. Uludag, Current Pharm. Design. 2002, 8, 1929 – 1944.
[7] D. Kontaci, J.K. Denike, W. J. Wechter, Synth. Commun. 1996,
26, 2037 – 2043.
[8] I. S. Alferiev, N. R. Vyavahare, C. V. Song, R. J. Levy, J. Polym.
Sci. Part A 2001, 39, 105 – 116.
ˇ
[9] H. Uludag, N. Kousinioris, T. Gao, D. Kantoci, Biotechnol. Prog.
2000, 16, 258 – 267.
ˇ
[10] H. Uludag, T. Gao, G. R. Wohl, D. Kantoci, R. F. Zernicke,
Biotechnol. Prog. 2000, 16, 1115 – 1118.
ˇ
[11] H. Uludag, J. Yang, Biotechnol. Prog. 2002, 18, 604 – 611.
ˇ
[12] S. A. Gittens, J. R. Matyas, R. F. Zernicke, H. Uludag, Pharm.
Res. 2003, 20, 978 – 987.
ˇ
[13] S. A. Gittens, P. I. Kitov, J. R. Matyas, R. Lꢁbenberg, H. Uludag,
Pharm. Res. 2004, 21, 608 – 616.
ˇ
[14] G. Bansal, J. E. I. Wright, S. Zhang, R. F. Zernicke, H. Uludag, J.
Biomed. Mater. Res. 2005, in press.
[15] C. R. Degenhardt, D. C. Burdsall, J. Org. Chem. 1986, 51, 3488 –
3490.
[16] Y. Ishida, M. Jikei, M. Kakimoto, Macromolecules 2000, 33,
3202 – 3211.
&
&
Figure 3. Delivery of BSA (control; &), SH-BP–BSA ( ), and 8–BSA ( )
to femora and tibiae after intravenous injection in Sprague–Dawley
rats (n=3 per group). Bone-targeting was assessed one day after injec-
tion. Note the lack of bone-targeting by the SH-BP–BSA conjugate (i.e.
similar to BSA) and the significantly improved bone-targeting by
8–BSA. See Ref. [11] for details about the methods.
[17] 2: 1H NMR (CDCl3, J [Hz]): d = 4.19 (m, 8H), 3.90 (td, 2H, J =
5.4, 16.2), 3.36 (s, 3H), 2.68 (tt, 1H, J = 10.8, 24), 1.31 ppm (t,
12H, J = 9.3); The hydrogen atom on the carbon atom in P-CA-P
resonates as a triplet of triplets, through coupling to two
phosphorus and to two hydrogen atoms on the adjacent CB
atom, whereas the two hydrogen atoms on CB couple with two
Angew. Chem. Int. Ed. 2005, 44, 3710 –3714
www.angewandte.org
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
3713

Communications
phosphorus nuclei to give a triplet that splits into a doublet by
the hydrogen atom on CA. 3: 1H NMR (CDCl3, J [Hz]): d = 7.00
(dd, 2H, J = 34.2, 33.9), 4.15 (m, 8H), 1.32 ppm (t, 12H, J = 6.9);
13C NMR (CDCl3): d = 149.01, 132.03, 63.12, 16.2 ppm.[15]
6: 1H NMR (DMSO): d = 11.08 (s, 2H), 9.23 (s, 4H), 9.02 (s,
2H), 8.76 (s, 1H), 8.19 ppm (s, 2H); MS: m/z = 539 [MÀH]À.
7: 1H NMR (DMSO): d = 10.11 (s, 2H), 8.42 (s, 1H), 8.04 (s,
2H), 6.31 (s, 4H), 5.99 ppm (s, 2H), in agreement with those
reported.[17] The addition of four ethylidenebisphosphonic acid
groups to the four amine (NH2) groups of 7 was confirmed by
1H NMR spectroscopy and mass spectrometry. 8: 1H NMR (D2O
and NaOD, J [Hz]): d = 7.94 (m, 1H), 7.81 (m, 2H), 6.74 (m,
4H), 6.52 (m, 2H), 3.45 (m, 8H), 2.05 ppm (tt, 4H, J = 3.3, 7.5);
13C NMR: d = 147.52, 138.96, 121.31, 88.44, 47.66, 40.51 ppm;
31P NMR: d = 18.1 ppm; MS: m/z = 1171 [MÀH]À, 1173
[M+H]+; IR: n˜ = 1202 cmÀ1 (P O).
=
[18] M. M. Bradford, Anal. Biochem. 1976, 72, 248 – 254.
[19] B. N. Ames, Methods Enzymol. 1966, 8, 115 – 117.
3714
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
Angew. Chem. Int. Ed. 2005, 44, 3710 –3714