Increasing αvβ3 selectivity of the anti-angiogenic drug cilengitide by N-methylation

Article abstract of DOI:10.1002/anie.201102971

A subtle change: Structural changes upon amide bond methylation improve the selectivity of the anti-angiogenic drug Cilengitide, which after N-methylation at distinct positions discriminates between the closely related pro-angiogenic integrins αvβ3 and αv

Full text of DOI:10.1002/anie.201102971

Communications
DOI: 10.1002/anie.201102971
Drug Selectivity
Increasing avb3 Selectivity of the Anti-Angiogenic Drug Cilengitide by
N-Methylation**
Carlos Mas-Moruno, Johannes G. Beck, Lucas Doedens, Andreas O. Frank, Luciana Marinelli,
Sandro Cosconati, Ettore Novellino, and Horst Kessler*
The drug Cilengitide, c(RGDf(NMe)V), is a cyclic RGD
pentapeptide (R = arginine, D = aspartic acid, G = glycine)
currently in clinical phase III for the treatment of brain
tumors and in phase II for other cancer types.[1] The anti-
tumoral properties of this peptide are based on its antago-
nistic activity for pro-angiogenic integrins, such as avb3, avb5,
or a5b1. However, the specific roles of these integrin subtypes
in angiogenesis and cancer are not yet clear and fully
understood. In this work, we present di-N-methylated ana-
logues of the stem peptide c(RGDfV) which retain an avb3-
binding activity in the nanomolar range but have lost most of
the activity for integrins avb5 and/or a5b1. Highly active and
selective peptides for avb3 are important tools to study the
specific role of this integrin in angiogenesis and cancer.
Integrins are heterodimeric receptors that govern cell–cell
and cell–extracellular matrix (ECM) interactions, and play
crucial roles in a plethora of cellular functions.[2] The fact that
many integrins are involved in pathological processes, such as
tumor angiogenesis, has stimulated their study as therapeutic
targets.[3] A number of integrin receptors recognize and bind
the tripeptide sequence RGD, which is a prominent cell-
adhesion motif present in ECM proteins.[4] Mimicking this
tripeptide sequence with RGD-peptides or peptidomimetics
is hence a promising approach to target integrins involved in
angiogenesis and to develop anti-cancer agents.[1,3b,5]
growth factor-a is avb5-dependent. These two integrins are
also described to be important mediators in the regulation of
hypoxia in glioblastomas.[7] However, mice lacking either av
or b3 and b5 integrins showed extensive angiogenesis.[8] These
intriguing results were a matter of debate and challenged our
understanding about the role of these two integrins in
angiogenesis.[9] The integrin a5b1 is also highly expressed in
angiogenic vasculature by several angiogenic stimuli, such as
bFGF but not by VEGF.[10] Since avb3, avb5 and a5b1 have
partially overlapping ligand affinities,[4b] it is plausible that
a5b1 might substitute the pro-angiogenic activity of the other
integrins. Paradoxically, another recent study showed that low
concentrations of Cilengitide stimulates VEGF-mediated
angiogenesis.[11] Although the doses used in this study are
far lower than therapeutic concentrations[12] and hence such a
“pro-angiogenic” effect is not likely to be observed in the
clinical studies, it becomes evident that a better understand-
ing of anti-angiogenic agents is necessary.[13]
It has been shown by us and others that N-methylation can
increase the selectivity towards specific receptor subtypes.[14]
These biological effects are often caused by the induction of
conformational constraints in the peptide backbone, which
lead to preferred single conformers essential for biological
activity.[14a,d,h,15] Thus, we envisioned that further N-methyl-
ation of Cilengitide could result in enhanced selectivity
profiles. For this reason we designed a library containing all
the di-N-methylated analogues of c(RGDfV) (Figure 1).
Note that the synthesis of NMe peptides (especially if they
are cyclic) is not without challenges that need to be carefully
considered.[14a,16] In the first place, although many N-methyl
amino acids are commercially available, most of them are still
expensive. Therefore, we synthesized, in solution, the NMe
residues of Gly, Val, and d-Phe by reduction of the
corresponding oxazolidinone using Freidinger conditions.[17]
Alternatively, Arg and Asp were methylated on resin using
the Miller and Scanlan method,[18] later optimized by Biron
et al.,[19] which is compatible with acid-sensitive side-chain
It is known that avb3 and avb5 are involved in two
different angiogenic pathways.[6] Whereas angiogenesis
induced by basic fibroblast growth factor (bFGF) or tumor
necrosis factor-a depends on avb3, angiogenesis triggered by
vascular endothelial growth factor (VEGF) or transforming
[*] Dr. C. Mas-Moruno, J. G. Beck, Dr. L. Doedens, Dr. A. O. Frank,
Prof. Dr. H. Kessler
Institute for Advanced Study and Center of Integrated Protein
Science, Department Chemie, Technische Universitꢀt Mꢁnchen
85747 Garching (Germany)
E-mail: kessler@tum.de
Prof. Dr. H. Kessler
Chemistry Department, Faculty of Science,
King Abdulaziz University, 21589 Jeddah (Saudi Arabia)
Prof. Dr. L. Marinelli, Dr. S. Cosconati, Prof. Dr. E. Novellino
Dipartimento di Chimica Farmaceutica e Tossicologica, Universitꢂ
di Napoli “Federico II”, 80131 Napoli (Italy)
[**] This work was partially supported by the International Graduate
School of Science and Engineering. We thank B. Cordes for technical
assistance with mass spectroscopy. C.M.M. thanks the Generalitat
de Catalunya for a Beatriu de Pinꢀs postdoctoral fellowship. J.G.B
thanks the TUM Graduate School for support.
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.201102971.
Figure 1. Schematic representation of our library of di-N-methylated
analogues of c(RGDfV).
9496
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2011, 50, 9496 –9500

protecting groups. Moreover, other limitations were encoun-
tered. For instance, the presence of NMeGly at the C-
terminus resulted in diketopiperazine formation. Also, to
improve coupling efficiency, powerful coupling reagents, such
as HATU, were required. Another critical point was the
cyclization step: to favor cyclization NMe amino acids were
avoided at the N-terminus and whenever possible Gly was
fixed at the C-terminus to prevent racemization. Finally, both
reaction time and TFA concentration were optimized to avoid
peptide fragmentation during side chain deprotection. The
synthesis of peptide 4, which summarizes all these consid-
erations, is shown in Scheme 1 (see the Supporting Informa-
tion for a detailed description of the synthesis of all the
analogues).
The impact of the extra N-methylation on Cilengitide in
terms of integrin binding activity and selectivity was eval-
uated using a solid-phase binding assay for the pro-angiogenic
integrins avb3, avb5, and a5b1 as well as for the platelet
receptor aIIbb3 (Table 1). The analogues in which Val was
non-methylated (5–10) showed a dramatic decrease in avb3-
binding activity, regardless of the position of the N-methy-
lated residues. This effect was particularly observed for
peptides 9 and 10, in which the antagonistic activity for the
vitronectin receptor was totally lost. In contrast, when Val was
N-methylated, the resulting analogues (1–4) displayed low
nanomolar activity for the avb3 integrin receptor. These
results indicate that NMeVal is a crucial residue to retain the
activity for this receptor, probably by inducing a preferred
bioactive avb3-binding conformation.[5a,20] Analogues 9 and
10, which are totally inactive, are both N-methylated at d-
Phe. It could be hypothesized that this biological effect was
due to the loss of a hydrogen-bond donor at this position;[21]
however, peptide 2, which also has NMe-d-Phe unit, exhibits
a remarkable nanomolar antagonistic activity. In this regard,
the effect of NMe at an Arg residue is also interesting. In a
previous study, a peptide with a single N-methylation of this
residue showed an IC50 of 5.5 nm.[5a] Herein, the presence of
NMeArg is found in peptides with activities ranging from
superpotent (1.9 nm, 4), moderate (142 nm, 5), low
(> 1000 nm, 8) and very low (> 10000 nm, 10). These data
clearly indicate that the biological activity of these peptides
more strongly depends on their overall conformation rather
than on the local effects of a single
N-methylation.[22]
Noteworthy,
most members of the library are
inactive for the integrins avb5 and
a5b1. If we focus on peptides 1 to 4
(highly active for avb3) only 4
shows nanomolar activity for
these receptors, with selectivity
ratios very similar to Cilengitide.
In contrast, in peptides 1 and 2 the
activity for avb5 is strongly or fully
suppressed, with selectivity ratios
much higher than those found for
Cilengitide (> 500-fold for 1 and
> 250-fold for 2). Compound 3
does not show an improved selec-
tivity towards avb5 but towards
a5b1. For all these compounds the
selectivity of Cilengitide against
aIIbb3 was either maintained or
improved. A strong reduction in
binding activity for avb5 and a5b1
was also observed for analogues 5–
10. However, these peptides are of
lower biological interest due to
their low (or absent) affinity for
avb3.
To rationalize these findings
and to explain the selectivities
obtained in a better way, three
peptides were chosen for structural
Scheme 1. Solid-phase synthesis of analogue 4. a) Fmoc-Gly-OH, DIEA, DCM; b) piperidine-NMP
(1:4); c) Fmoc-Arg(Pbf)-OH, TBTU, HOBt, DIEA, NMP; d) NBS-Cl, collidine, NMP; e) Ph3P, MeOH,
DIAD, THF; f) HS-CH2-CH2-OH, DBU, NMP; g) Fmoc-NMeVal-OH, HATU, HOAt, DIEA, NMP;
h) Fmoc-d-Phe-OH, HATU, HOAt, DIEA, NMP; i) Fmoc-Asp(OtBu)-OH, TBTU, HOBt, DIEA, NMP;
j) HFIP-DCM (2:8); k) DPPA, NaHCO3, DMF; l) TFA-DCM-H2O-TIS (60:35:2.5:2.5). Fmoc=9-fluore-
nylmethoxycarbonyl, DIEA=ethyldiisopropylamine, DCM=dichloromethane, NMP=N-methylpyrroli-
dine, TBTU=O-(benzotriazol-1-yl)-N,N,N’,N’-tetramethyluronium-tetrafluoroborate, NBS-Cl=nitroben-
zylsulfonylchloride, DIAD=diisopropylazodicarboxylate, DBU=1,5-diazabicyclo[5.4.0]undec-5-en,
HATU=O-(7-azabenzotriazol-1-yl)-tetramethyluronium hexafluorophosphate, HFIP=hexafluoroisopro-
pylalcohol, DPPA=diphenylphosphorylazide, TFA=trifluoro acetic acid, TIS=Triisopropylsilane.
studies: peptide 1,
a selective
ligand that shows nanomolar activ-
ity for avb3 and low activity for
avb5; peptide 4, which is active for
both receptors in the nanomolar
range and therefore not selective;
and peptide 10, totally inactive for
Angew. Chem. Int. Ed. 2011, 50, 9496 –9500
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org 9497

Communications
Table 1: The ten di-N-methylated analogues of Cilengitide and their binding activity (IC50 in nm) towards
the peptide occurs (see Figure 3b).
This result explains why 1 has
approximately tenfold lower affin-
ity to avb3 than Cilengitide (5.9 nm
and 0.65 nm, respectively). For
docking to the avb5 receptor,
100 avb5 homology models, differ-
ing in the specificity determining
loop (SDL) conformation, were
generated. Prior to docking calcu-
lations, all 100 models were tested
for their capability to host the
unselective Cilengitide and only
those able to bind were further
considered for docking of 1, 4, and
10. Predictably, in these models
avb3, avb5, a5b1, and aIIbb3.[a]
Peptide
avb3
avb5
a5b1
aIIbb3
avb5/ a5b1/
avb3
avb3
Cil
1
2
3
4
5
6
7
8
c(-R-G-D-f-V-) 0.65(Æ0.07) 11.7(Æ1.5) 13.2(Æ0.6) 815(Æ60) 18
20
c(-R-G-D-f-V-) 5.9(Æ2.5)
c(-R-G-D-f-V-) 36.2(Æ8.1)
c(-R-G-D-f-V-) 13.2(Æ1.8)
c(-R-G-D-f-V-) 1.9(Æ0.3)
c(-R-G-D-f-V-) 142(Æ33)
c(-R-G-D-f-V-) 173(Æ12)
c(-R-G-D-f-V-) 965(Æ96)
c(-R-G-D-f-V-) >1000
c(-R-G-D-f-V-) >10000
c(-R-G-D-f-V-) >10000
>3000
270(Æ95)
>2000
>1000
>10000
>2000
>500 46
>250 55
>10000
313(Æ122) >1000
24
22
>70
>58
–
–
–
–
>75
40.9(Æ3.2) 39.5(Æ1.3) >1000
21
>14
>29
–
–
–
>10000
>10000
>10000
>10000
>10000
>10000
>2000
>5000
>1000
>10000
>10000
>10000
>10000
>10000
>10000
>2000
>2000
>10000
9
10
–
[a] Residues in bold and italics are N-methylated.
Cilengitide assumed
a
binding
all integrins. Based on NMR spectroscopic assignments
(Supporting Information, Table S2), ROEs, homo- and het-
eronuclear scalar coupling constants, HN temperature gra-
dients, and on distance geometry calculations, distinct pre-
ferred structures could be derived for the three peptide
backbones (Figure 2). The structures of 1, 4, and 10 possess
pronounced differences that are described in detail in the
Supporting Information along with additional information
about their dynamics. Further, molecular-docking studies of
these peptides were attained into the avb3 X-ray structure[20]
as well as in the newly constructed avb5 homology models.
Docking of 1 on avb3 showed that this peptide is able to
interact with this receptor similarly to Cilengitide (Figure 3a).
Nevertheless, note that the substitution of the Asp residue in
Cilengitide with NMeAsp in 1 does affect to a certain extent
the binding mode to avb3. In particular, this modification
causes the loss of a hydrogen bond with (b3)-D216 CO but
more importantly an evident relocation of the lower part of
pose similar to the experimentally determined bound state
in avb3.[20] Interestingly, analysis of the multiple docking
simulations performed on the avb5 selected models demon-
strated that in the case of 1, a well-defined binding mode
could not be easily identified. Therefore, in this case the
ligand Asp N-methylation causes a pronounced effect on the
binding to avb5. In an attempt to rationalize such a behavior,
the predicted 1/avb3 complex was superimposed to the
modeled avb5 receptor structure. As represented in Fig-
ure 3c, it is clear that the (b3)-A252/(b5)-D279 mutation
results in a remarkable restriction of the available space.
Therefore, the methyl group of the NMeAsp would be hardly
adapted in the same binding fashion as in the 1/avb3 complex.
This, in turn, seems to strongly affect the RGD binding to
avb5.
Docking studies were also helpful in suggesting why the
N-methylation of Arg residue (4) is ineffective in producing
the avb3/avb5 selectivity (Table 1). In fact, such a modifica-
tion, while inducing a different peptide conformation with
respect to Cilengitide, does not influence the binding of 4
which is still assured by the conserved RGD sequence
(Supporting information, Figure S1). Conversely, docking of
10 revealed that this peptide is unable to efficiently bind to
the metal-ion-dependent adhesion site (MIDAS) and the av
subunit b propeller at the same time. Indeed, a comparison
between the NMR solution structure of 10 and the X-ray
bound conformation of Cilengitide showed that the double
methylation of Arg and d-Phe residues (10) induces marked
differences in distance between the Arg and Asp Ca atoms
(5.0 ꢀ and 6.4 ꢀ for 10 and Cilengitide, respectively) as well
À
as in the orientation of the Ca Cb bond vectors of the same
residues (Supporting Information, Figure S2). Both features
are well known to be critical for integrin binding and
selectivity.[1a,23] Hence, unlike in 1 and 4, the double N-
methylation in 10 seems to induce a non-productive peptide
conformation that prevents binding to avb3 and avb5
integrins.
In conclusion, we have demonstrated that double N-
methylation of the peptide backbone allows fine tuning of the
peptideꢁs biological activity by inducing preferred bioactive
conformations. Certain members of our library of di-N-
methylated c(RGDfV) retained nanomolar affinity for avb3
Figure 2. Stereoviews of 1, 4, and 10, as determined from NMR-based
distance geometry calculation and subsequent minimization (see the
Supporting Information for details).
9498 www.angewandte.org
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2011, 50, 9496 –9500

[1] a) C. Mas-Moruno, F. Rechenmacher, H. Kessler, Anti-Cancer
Agents Med. Chem. 2010, 10, 753 – 768; b) G. Tabatabai, M.
Weller, B. Nabors, M. Picard, D. Reardon, T. Mikkelsen, C.
Ruegg, R. Stupp, Target. Oncol. 2010, 5, 175 – 181.
[2] R. O. Hynes, Cell 2002, 110, 673 – 687.
[3] a) D. Cox, M. Brennan, N. Moran, Nat. Rev. Drug Discovery
2010, 9, 804 – 820; b) T. Arndt, U. Arndt, U. Reuning, H. Kessler
in Cancer Therapy: Molecular Targets in Tumor Host Interac-
tions (Ed.: G. F. Weber), Horizon Bioscience, Norfolk, UK,
2005, pp. 93 – 141; c) C. J. Avraamides, B. Garmy-Susini, J. A.
Varner, Nat. Rev. Cancer 2008, 8, 604 – 617; d) J. S. Desgrosellier,
D. A. Cheresh, Nat. Rev. Cancer 2010, 10, 9 – 22.
[4] a) E. Ruoslahti, M. D. Pierschbacher, Science 1987, 238, 491 –
497; b) E. F. Plow, T. A. Haas, L. Zhang, J. Loftus, J. W. Smith, J.
Biol. Chem. 2000, 275, 21785 – 21788.
[5] a) M. A. Dechantsreiter, E. Planker, B. Mathꢂ, E. Lohof, G.
Hçlzemann, A. Jonczyk, S. L. Goodman, H. Kessler, J. Med.
Chem. 1999, 42, 3033 – 3040; b) R. Haubner, D. Finsinger, H.
Kessler, Angew. Chem. 1997, 109, 1440 – 1456; Angew. Chem.
Int. Ed. Engl. 1997, 36, 1374 – 1389; c) A. Meyer, J. Auernheimer,
A. Modlinger, H. Kessler, Curr. Pharm. Des. 2006, 12, 2723 –
2747.
[6] M. Friedlander, P. C. Brooks, R. W. Shaffer, C. M. Kincaid, J. A.
Varner, D. A. Cheresh, Science 1995, 270, 1500 – 1502.
[7] N. Skuli, S. Monferran, C. Delmas, G. Favre, J. Bonnet, C. Toulas,
E. C. J. Moyal, Cancer Res. 2009, 69, 3308 – 3316.
[8] a) B. L. Bader, H. Rayburn, D. Crowley, R. O. Hynes, Cell 1998,
95, 507 – 519; b) L. E. Reynolds, L. Wyder, J. C. Lively, D.
Taverna, S. D. Robinson, X. Z. Huang, D. Sheppard, R. O.
Hynes, K. M. Hodivala-Dilke, Nat. Med. 2002, 8, 27 – 34.
[9] a) P. Carmeliet, Nat. Med. 2002, 8, 14 – 16; b) R. O. Hynes, Nat.
Med. 2002, 8, 918 – 921; c) D. A. Cheresh, D. G. Stupack, Nat.
Med. 2002, 8, 193 – 194.
[10] a) S. Kim, K. Bell, S. A. Mousa, J. A. Varner, Am. J. Pathol. 2000,
156, 1345 – 1362; b) N. J. Boudreau, J. A. Varner, J. Biol. Chem.
2004, 279, 4862 – 4868.
[11] A. R. Reynolds et al., Nat. Med. 2009, 15, 392 – 400.
[12] a) M. Weller, D. Reardon, B. Nabors, R. Stupp, Nat. Med. 2009,
15, 726; b) R. Stupp, C. J. Ruegg, J. Clin. Oncol. 2007, 25, 1637 –
1638.
[13] A. R. Reynolds, K. M. Hodivala-Dilke, Nat. Med. 2009, 15, 727.
[14] a) J. Chatterjee, C. Gilon, A. Hoffman, H. Kessler, Acc. Chem.
Res. 2008, 41, 1331 – 1342; b) P. Pratim Bose, U. Chatterjee, C.
Nerelius, T. Govender, T. Norstrçm, A. Gogoll, A. Sandegren, E.
Gçthelid, J. Johansson, P. I. Arvidsson, J. Med. Chem. 2009, 52,
8002 – 8009; c) K. S. Harris et al., J. Biol. Chem. 2009, 284, 9361 –
9371; d) L. Doedens, F. Opperer, M. Cai, J. G. Beck, M. Dedek,
E. Palmer, V. J. Hruby, H. Kessler, J. Am. Chem. Soc. 2010, 132,
8115 – 8128; e) H. Qu, P. Magotti, D. Ricklin, E. L. Wu, I.
Kourtzelis, Y. Q. Wu, Y. N. Kaznessis, J. D. Lambris, Mol.
Immunol. 2011, 48, 481 – 489; f) A. C. Bach II, C. J. Eyermann,
J. D. Gross, M. J. Bower, R. L. Harlow, P. C. Weber, W. F.
DeGrado, J. Am. Chem. Soc. 1994, 116, 3207 – 3219; g) A. C.
Bach II, J. R. Espina, S. A. Jackson, P. F. W. Stouten, J. L. Duke,
S. A. Mousa, W. F. DeGrado, J. Am. Chem. Soc. 1996, 118, 293 –
294; h) J. Chatterjee, O. Ovadia, G. Zahn, L. Marinelli, A.
Hoffman, C. Gilon, H. Kessler, J. Med. Chem. 2007, 50, 5878 –
5881.
Figure 3. a) Structure of 1 (yellow) docked in the avb3 integrin binding
pocket. The av and b3 subunits are represented by the pink and cyan
surfaces, respectively. In both subunits the amino acid side chains
important for the ligand binding are represented as sticks. The metal
ion in the MIDAS region is represented by a magenta sphere. For
comparison, the X-ray structure of b) Cilengitide (white sticks) as well
as c) the (b5)-D279 residue (blue spheres) are shown. Red circles
highlight N-methyl groups in (a) and (c).
but were totally inactive for the integrin subtypes avb5 and
a5b1, thus improving the selectivity of Cilengitide. Com-
pounds displaying such selectivity profiles represent new
promising tools to study the role of closely related integrins in
essential biological processes.
[15] G. Mꢃller, Angew. Chem. 1996, 108, 2941 – 2943; Angew. Chem.
Int. Ed. Engl. 1996, 35, 2767 – 2769.
Received: April 29, 2011
Revised: May 27, 2011
[16] a) M. Teixidꢄ, F. Albericio, E. Giralt, J. Pept. Res. 2005, 65, 153 –
166; b) J. Tulla-Puche, N. Bayꢄ-Puxan, J. A. Moreno, A. M.
Francesch, C. Cuevas, M. ꢅlvarez, F. Albericio, J. Am. Chem.
Soc. 2007, 129, 5322 – 5323.
[17] R. M. Freidinger, J. S. Hinkle, D. S. Perlow, B. H. Arison, J. Org.
Chem. 1983, 48, 77 – 81.
Published online: August 25, 2011
Keywords: conformational studies · cyclic peptides · integrin ·
.
N-methylation · receptor selectivity
Angew. Chem. Int. Ed. 2011, 50, 9496 –9500
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org 9499

Communications
[18] S. C. Miller, T. S. Scanlan, J. Am. Chem. Soc. 1997, 119, 2301 –
[22] a) J. Chatterjee, D. Mierke, H. Kessler, J. Am. Chem. Soc. 2006,
128, 15164 – 15172; b) J. Chatterjee, D. Mierke, H. Kessler,
Chem. Eur. J. 2008, 14, 1508 – 1517; c) B. Laufer, A. O. Frank, J.
Chatterjee, T. Neubauer, C. Mas-Moruno, G. Kummerlçwe, H.
Kessler, Chem. Eur. J. 2010, 16, 5385 – 5390.
[23] a) M. Pfaff, K. Tangemann, B. Mꢃller, M. Gurrath, G. Mꢃller, H.
Kessler, R. Timpl, J. Engel, J. Biol. Chem. 1994, 269, 20233 –
20238; b) G. Mꢃller, M. Gurrath, H. Kessler, J. Comput.-Aided
Mol. Des. 1994, 8, 709 – 730.
2302.
[19] E. Biron, J. Chatterjee, H. Kessler, J. Pept. Sci. 2006, 12, 213 –
219.
[20] J. P. Xiong, T. Stehle, R. Zhang, A. Joachimiak, M. Frech, S. L.
Goodman, M. A. Arnaout, Science 2002, 296, 151 – 155.
[21] R. Haubner, R. Gratias, B. Diefenbach, S. L. Goodman, A.
Jonczyk, H. Kessler, J. Am. Chem. Soc. 1996, 118, 7461 – 7472.
9500 www.angewandte.org
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2011, 50, 9496 –9500