MRI visualization of melanoma cells by targeting overexpressed sialic acid with a GdIII-dota-en-pba imaging reporter

Article abstract of DOI:10.1002/anie.201207131

By design: The novel MRI reporter Gd^III-dota-en-pba (see picture) selectively binds to sialic acid residues, which are overexpressed on tumors. This imaging reporter accumulates in a murine melanoma B16-F10 mice xenograft and from the MRI signal intensity it is possible to quantify the sialic acid concentration that is directly associated with tumor malignancy. Copyright

Full text of DOI:10.1002/anie.201207131

Angewandte
Chemie
DOI: 10.1002/anie.201207131
Imaging Agents
MRI Visualization of Melanoma Cells by Targeting Overexpressed
Sialic Acid with a GdIII-dota-en-pba Imaging Reporter**
Simonetta Geninatti Crich,* Diego Alberti, Ibolya Szabo, Silvio Aime, and Kristina Djanashvili*
Altered glycosylation of cell surface proteins and lipids is
frequently associated with tumorigenic and metastatic pro-
cesses.[1] These changes in glycosylation are mainly caused by
an increased expression of sialyltransferases, which glycosy-
late exposed glycans at their terminal positions with anionic
monosaccharide sialic acid residues (Sia). Tumor cells over-
Scheme 1. Formation of the five- and six-membered ring esters from
1,2- or 1,3 diols and phenylboronic acid at physiological pH.
expressing Sia appear protected against the immune defense
system, and as a result, malignancy is increased.[2] Numerous
studies have demonstrated that Sia are relevant biomarkers of
metastatic activity of tumors,[3] and that the amount of Sia
expression on cancer cells correlates with the prognosis of
patients.[4] Currently the determination of the sialylation
status of solid tumors is carried out on sections of tissue
obtained from biopsies. This procedure is not only invasive, it
also bears an intrinsic risk that the sample is not representa-
tive for the whole, potentially heterogeneous tumor. There-
fore, we have investigated probes that are able to non-
invasively visualize Sia in vivo.
of various saccharide sensors[7] and Sia determination
assays,[8] its application for in vivo targeting of tumors has
not yet been reported. Herein, for the first time, we
demonstrate in vivo visualization of tumors by the highly
selective dynamic covalent binding between the diol function
of overexpressed Sia and a pba-targeting vector conjugated
with an MRI reporter (Figure 1).
Macrocyclic or linear lanthanide complexes appear inter-
esting candidates for this intended application, because they
can be used either as contrast agents (CA) for magnetic
resonance imaging (MRI) or as radiopharmaceuticals for
imaging and treatment of tumors. For both tasks, accurate
delivery and retention of the complex at the site of disease is
mandatory. Previously, we have demonstrated the affinity of
160Tb-dtpa-bisamide (dtpa = diethylenetriamine pentaacetic
acid) conjugated to a phenylboronic acid (pba) moiety
towards human glioma cells.[5a] The recognition mechanism
is based on the reversible formation of five- and six-
membered cyclic boronate esters between pba and the
exocyclic polyol function of Sia[6] (Scheme 1). Although this
type of binding has extensively been exploited for the design
Figure 1. Schematic representation of sialic acid recognition by the
imaging reporter GdIII-dota-en-pba through diol–phenylboronic acid
interactions on the surface of tumor cells.
The dota chelate (dota = 1,4,7,10-tetraazacyclododecane
1,4,7,10-tetraacetate) has been chosen for the design of the
targeting imaging reporter dota-en-pba because of the
superior kinetic stability of its complexes with LnIII ions
compared to previously reported dtpa-bisamide-pba-com-
plexes.[5,9] The pba group is conjugated via an ethylenedi-
amine (en) spacer. As the amino group of dota-en-pba is
positively charged at physiological pH it facilitates the
recognition mechanism through electrostatic affinity to the
negatively charged carboxylic groups of the Sia residues. The
role of this auxiliary interaction has been supported previ-
ously by a computer modeling study.[5a]
As Sia expression is directly correlated to melanogenesis,
the recognition ability of the designed complex were eval-
uated on murine melanoma B16-F10 cells. This cell line is
commonly used as a well-characterized, highly proliferative
tumor model. In melanoma cells, the activity of tyrosinase, the
[*] Dr. S. Geninatti Crich, D. Alberti, I. Szabo, Prof. S. Aime
Department of Chemistry, Molecular Imaging Center
Via Nizza 52, 10126 Turin (Italy)
E-mail: simonetta.geninatti@unito.it
Dr. K. Djanashvili
Department of Biotechnology, Delft University of Technology
Julianalaan 136, 2628 BL, Delft (The Netherlands)
E-mail: k.djanashvili@tudelft.nl
[**] This research was performed in the framework of the EU COST
Action TD1004, “Theranostics Imaging and Therapy: an Action to
Develop Novel Nanosized Systems for Imaging-Guided Drug
Delivery”, and supported by Regione Piemonte (PIIMDMT and
nano-IGT projects), MIUR (PRIN 2009235JB7). The authors are
grateful to Dr. Joop A. Peters for valuable scientific discussions.
dota=1,4,7,10-tetraazacyclododecane 1,4,7,10-tetraacetate, en=e-
thylenediamine, pba=phenylboronic acid.
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.201207131.
Angew. Chem. Int. Ed. 2013, 52, 1161 –1164
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1161

.
Angewandte
Communications
key enzyme for melanogenesis, shows a good correlation with
intra- and extracellular Sia levels.[10] This correlation indicates
that Sia have an essential role in enzyme expression and,
therefore, in the differentiation of melanoma cells. Sialyla-
tion/desialylation of some relevant molecules, such as mela-
nosomal sialoglycoconjugates and tyrosinase itself, partici-
pate in the regulation of the expression of melanogenic
functions. For these reasons, two clones of B16-F10 murine
melanoma each characterized by a different phenotype,
namely melanogenic (B16-F10m) and nonmelanogenic (B16-
F10non) cells, have been used to test the pba-based imaging
reporter in vitro. The two phenotypes were obtained by
growing cells at different pH values. In fact, it is known that
B16-F10 cells only form melanin when they are maintained in
culture at pH 7.2–7.4.[11] At lower pH melanogenesis is
inhibited, whereas other metabolic processes remain almost
unaffected. The level of Sia in the B16-F10m clone was
determined to be 3.6 times higher than that in B16-F10non; this
result is in agreement with previously reported values.[11] The
specific binding of GdIII-dota-en-pba to the tumor cells
overexpressing Sia was assessed by relaxation rate measure-
ments on the washed cell pellets after incubation with
a 0.6 mm solution of the complex for 4 h at 378C.
activity, B16-F10 cells were incubated for 24 h with a 0.9 mm
solution of theophylline, well-known pigmentation
a
enhancer in murine melanoma cells.[11] As an inhibitor of
phosphodiesterase, theophylline has been shown to exten-
sively stimulate pigment production and tyrosinase activity of
melanoma cells in vitro. As expected, expression of Sia by the
cells preincubated with theophylline increased significantly,
thus resulting in accumulation of larger quantities of the Gd
complex (Figure 3). This increased binding effect is especially
From the results reported in Table 1 it is evident that there
is a marked accumulation of the complex at the melanogenic
cells that yields high relaxation rates (R1 = 1/T1), as a result of
a higher Gd content, as confirmed by ICP-MS measurements.
The relaxation rates are in good correlation with the Sia
expression (Figure 2).
Figure 3. Gd content measured by ICP-MS after incubation of B16-
F10m (grey solid bar) and B16-F10non (white solid bar) cells with GdIII-
dota-en-pba, with or without preincubation with theophylline (dashed
bars). Error bars have been calculated on the basis of three replicate
experiments. Student’s t-test was used to compare the differences
between Gd content before and after theophylline addition. B16-F10m:
P=0.0015; B16-F10non-m: P<0.0001. A P value <0.05 is considered
statistically significant.
To get further insights into the relationship between GdIII-
dota-en-pba uptake and the sialylation level/melanogenic
Table 1: Response of melanogenic and nonmelanogenic cells to the
incubation with GdIII-dota-en-pba (0.6 mm, 4 h, 378C).
remarkable for the nonmelanogenic cells, as it is 7 times
higher than that for the melanogenic cells, thus resulting in
a similar level of Sia expression for both cell types. The
differences in the Gd levels measured in this experiment
compared to the one reported in Table 1 can be explained by
the variability of the Sia expression by the B16-F10 cells, as
this expression depends on the conditions, that is, pH,
composition of medium, number of passages after cell
defreezing, etc.
Cell type
R1 [sÀ1
obs
]
[Gd] [molmgÀ1 pro-
tein][a]
[Sia] [molmgÀ1 pro-
tein][b]
B16-F10m 0.74Æ0.03 1.50 (Æ0.5)ꢀ10À9
7.3 (Æ0.6)ꢀ10À9
B16-
0.51Æ0.04 0.20 (Æ0.5)ꢀ10À9
1.5 (Æ0.4)ꢀ10À9
F10non-m
[a] Determined by ICP-MS. [b] Determined by AbCAM assay. The data
represent the results obtained on the basis of five independent
experiments.
It is reasonable to assume that successful in vivo tumor
targeting, based on the recognition of Sia residues by a pba
vector, could be hampered by its competitive binding to
glucose in blood (typically present in a concentration of 3.5–
5.8 mm). The formation of cyclic esters of pba and glucose is
well-documented,[12] and therefore, the interaction of GdIII-
dota-en-pba with B16-F10m cells was evaluated at different
concentrations in the presence (5.8 mm) or in the absence of
glucose. In both cases, the cells were incubated for 4 h with
the complex, after which time the incubation medium was
removed, the cells were washed, detached by using EDTA,
and the water proton relaxation rates of cell suspensions were
measured. As expected, a moderate decrease in the relaxation
rates as a result of the competitive binding between the
complex and glucose in the medium was observed. However,
when the incubation is performed using greater than 1 mm of
GdIII-dota-en-pba, the competition effect is negligible. At
Figure 2. Correlation between sialic acid expression and relaxation
rates (R2 =0.9329) measured in triplicate on cells incubated with GdIII-
dota-en-pba.
1162
www.angewandte.org
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2013, 52, 1161 –1164

Angewandte
Chemie
these concentrations the amount of the free (not bound to
glucose) complex is high enough to saturate all Sia residues
available on the cell surface, as demonstrated by the saturat-
ing profile of the curve. This observation can be rationalized
by the lower affinity of pba to glucose than to Sia.[13] This
lower affinity is due to the fact that d(+)glucose is present in
solution predominantly (> 99%) in its pyranose anomeric
form. In this configuration, the diol groups are in an
unfavorable position for the formation of pba esters com-
pared to those of Sia, that are positioned on a flexible glycerol
tail.[6] The analysis of the relaxation rate behavior in the
absence of glucose as a function of the concentration of GdIII-
dota-en-pba and the equilibria involved is discussed in the
Supporting Information. The probe uptake by cells is also
only slightly influenced by the presence of glucose, as
demonstrated by MRI images of both melanoma clones
(Figure 4B). Melanogenic cells (capillaries 2 and 3) express-
ing high levels of Sia on their surfaces appear hyperintense on
T1-weighted images compared to nonmelanogenic cells
(capillaries 5 and 6). When cells were incubated with the
probe in the presence (capillaries 3 and 6) or in the absence
(capillaries 2 and 4) of glucose, suspensions containing the
same type of cells give images of almost the same intensity,
thus demonstrating that glucose has only a minor influence on
the binding of GdIII-dota-en-pba to the cells.
Figure 4. A) Relaxation rates measured for B16-F10 murine melanoma
cells, incubated for 4 h at 378C, with various concentrations of GdIII-
dota-en-pba in the presence (circles) or the absence (squares) of
glucose (5.8 mm). Detachment of cells by EDTA was applied. The
continuous line is the result of the fitting of the data by the equations
described in the Supporting Information. B) T1-weighted spin-echo MR
images of agar phantoms containing unlabeled cells (1 and 4) and
cells incubated with 0.6 mm GdIII-dota-en-pba in the absence (2 and 4)
or in the presence (3 and 6) of 5.5 mm glucose.
Next, in vivo MRI experiments were carried out on mice
models bearing a tumor xenograft obtained by subcutaneous
injection of approximately 1 ꢀ 106 B16-F10m cells on the right
flank. After 8–10 days, the tumor volume was between 20 and
40 mm3, that is, easily detectable by MRI. Five animals were
then intravenously administered with 0.1 mmolkgÀ1 of GdIII-
dota-en-pba. A second group of mice (n = 5) received the
same dose of Gd-HPDO3A (a commercial MRI agent with
unspecific vascular and extravas-
and without any significant difference between muscle and
tumor regions. Next, to demonstrate the specificity of the
binding of GdIII-dota-en-pba to the tumor cells, a competition
experiment was performed with EuIII-dota-en-pba, which
does not influence the SI in T1-weighted MR images. A group
of four mice were treated with 0.15 mmolkgÀ1 of EuIII-dota-
cular distribution) as control. Fat-
suppressed T1-weighted multislice
multiecho MR images were
recorded before, 4 h after, and
24 h after contrast agent adminis-
tration (Figure 5). The analysis of
signal intensity (%SI) enhance-
ment measured on the region of
interest, which is drawn on the
whole
tumor
regions
(see
Figure 5), was of 25% and 10%
at 4 h and 24 h after the injection
of the targeting probe, respec-
tively, whereas no persistent SI
was observed for animals injected
with the control probe (see the
Supporting Information, Fig-
ure S1). From Figure 5B, it is ap-
parent that the probe can visualize
the heterogeneity of tumor. The
%SI enhancement recorded on
mice treated with Gd-HPDO3A
at the same Gd dose (Figure 5D
and Figure S1 in the Supporting
Information) was markedly lower
Figure 5. Fat-suppressed T1-weighted MR spin-echo images of C57BL/6 mice grafted subcutaneously
with B16-F10 melanoma cells recorded at 7T before (A) and 4 h after (B) the administration of GdIII-
dota-en-pba. For comparison, analogous measurements were performed before (C) and 4 h after (D)
the administration of Gd-HPDO3A. The corresponding MRI %SI enhancements (E) were measured in
melanoma tumors and muscle 4 h and 24 h after the injection of GdIII-dota-en-pba; the data obtained
after injection of GdIII-HPDO3A is reported in the Supporting Information (Figure S1). F) A plot of MRI
%SI enhancements measured in melanoma tumor and muscle 4 h after administration of GdIII-dota-
en-pba with or without the preinjection with a high dose (0.15 mmolkgÀ1) of EuIII-dota-en-pba.
Angew. Chem. Int. Ed. 2013, 52, 1161 –1164
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
1163

.
Angewandte
Communications
en-pba, with subsequent administration of 0.1 mmolkgÀ1 of
GdIII-dota-en-pba after 3 h. Figure 5F shows that %SI
enhancements measured 4 h after the injection of Gd complex
in the region indicated on the whole tumor regions were
significantly lower than those measured in the absence of the
competitor, thus supporting the view of GdIII-dota-en-pba
binding specificity.
[2] a) R. Schauer, Zoology 2004, 107, 49 – 64; b) R. Kannagi, M.
Izawa, T. Koike, K. Miyazaki, N. Kimura, Cancer Sci. 2004, 95,
377 – 384.
[3] I. M. Duncker, R. Salinas-Marin, C. Martinez-Duncker, Int. J.
Mol. Imaging 2011, 283497.
[4] a) A. Cazet, S. Julien, M. Bobowski, M.-A. Krzewinski-Recchi,
A. Harduin-Lepers, S. Groux-Degroote, P. Delannoy, Carbo-
hydr. Res. 2010, 345, 1377 – 1383; b) A. Fernꢁndez-Briera, I.
Garzia-Parcerio, E. Cuevas, E. Gili-Martin, Oncology 2010, 78,
196 – 204.
[5] a) K. Djanashvili, G. A. Koning, A. J. G. M. van der Meer, H. T.
Wolterbeek, J. A. Peters, Contrast Media Mol. Imaging 2007, 2,
35 – 41; b) K. Djanashvili, T. L. M. ten Hagen, R. Blangꢂ, D.
Schipper, J. A. Peters, G. A. Koning, Bioorg. Med. Chem. 2011,
19, 1123 – 1130.
In conclusion, we have demonstrated for the first time the
feasibility of in vivo tumor targeting based on the recognition
of overexpressed sialic acid by a pba-based imaging reporter.
The reported results for the murine melanoma tumor could
possibly be extended to other tumor types that are known to
exhibit high levels of Sia[1] as a function of their metastatic
potential, such as breast cancer. The high expression of Sia on
the cell surface (approximately 1 ꢀ 109/cell) permits their MRI
visualization both in vitro and in vivo by using a low molec-
ular weight Gd complex. The use of small-sized targeting
contrast agents has evident advantages compared to the use of
targeting nanoparticles in terms of their stability, access to the
cellular targets, and general toxicity issues. This proof of
concept opens new avenues for the design of tumor-specific
imaging probes. The prolonged retention of GdIII-dota-en-pba
at the tumor site after injection suggests the possibility of
using this probe for therapy, provided that a suitable radio-
nuclide is used for labeling the dota-en-pba ligand. Research
along these lines is currently under way.
[6] K. Djanashvili, L. Frullano, J. A. Peters, Chem. Eur. J. 2005, 11,
4010 – 4018.
[7] a) J. Ren, R. Trokowski, S. Zhang, C. R. Malloy, A. D. Sherry,
Magn. Reson. Med. 2008, 60, 1047 – 1055; b) S. Aime, M. Botta,
W. Dastrffl, M. Fasano, M. Panero, Inorg. Chem. 1993, 32, 2068 –
2071; c) S. Saito, T. L. Massie, T. Maeda, H. Nakazumi, C. L.
Colyer, Sensors 2012, 12, 5420 – 5431; d) Z. Guo, I. Shin, J. Yoon,
Chem. Commun. 2012, 48, 5956 – 5967 (and references therein).
[8] a) A. Matsumoto, H. Cabral, N. Sato, K. Kataoka, Y. Miyahara,
Angew. Chem. 2010, 122, 5626 – 5629; Angew. Chem. Int. Ed.
2010, 49, 5494 – 5497; b) S. M. Levonis, M. J. Kiefel, T. A.
Houston, Chem. Commun. 2009, 2278 – 2280.
[9] C. Cabella, S. G. Crich, D. Corpillo, A. Barge, C. Ghirelli, E.
Bruno, V. Lorusso, F. Uggeri, S. Aime, Contrast Media Mol.
Imaging 2006, 1, 23 – 29.
[10] Y. Kinoshita, S. Sato, T. Takeuchi, Cell Struct. Funct. 1989, 14,
35 – 43.
[11] a) D. C. Bennett, Environ. Health Perspect. 1989, 80, 49 – 59; b) I.
Szabó, S. G. Crich, D. Alberti, F. K. Kꢁlmꢁn, S. Aime, Chem.
Commun. 2012, 48, 2436 – 2438.
Received: September 3, 2012
Revised: September 26, 2012
Published online: December 6, 2012
[12] E. Battistini, A. Mortillaro, S. Aime, J. A. Peters, Contrast Media
Mol. Imaging 2007, 2, 163 – 171.
[13] a) G. Springsteen, B. Wang, Tetrahedron 2002, 58, 5291 – 5300;
b) M. Regueiro-Figueroa, K. Djanashvili, D. Esteban-Gómez, T.
Chauvin, É. Tóth, A. de Blas, T. Rodríguez-Blas, C. Platas-
Iglesias, Inorg. Chem. 2010, 49, 4212 – 4223.
Keywords: imaging agents · molecular recognition ·
phenylboronates · sialic acids · tumor targeting
.
[1] a) P.-H. Wang, J. Cancer Mol. 2005, 1, 73 – 81; b) H. Cui, Y. Lin,
L. Yue, X. Zhao, J. Liu, Oncol. Rep. 2011, 25, 1365 – 1371.
1164
www.angewandte.org
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2013, 52, 1161 –1164