N-Methylation of Peptides: A New Perspective in

Medicinal Chemistry
JAYANTA CHATTERJEE,† CHAIM GILON,‡ AMNON HOFFMAN,§
AND HORST KESSLER*,†
†
Center for Integrated Protein Science at the Department Chemie, Technische
Universität München, Lichtenbergstrasse 4, Garching 85747, Germany,
‡
Institute of Chemistry, The Hebrew University of Jerusalem,
Jerusalem 91904, Israel, §Department of Pharmaceutics, The Hebrew University
of Jerusalem, Jerusalem 91120, Israel
RECEIVED ON MARCH 5, 2008

CON SPECTUS

T he potential of peptides as drug candidates is limited by their poor pharmacokinetic properties. Many peptides have a
short half-life in vivo and a lack of oral availability. Inspired by the excellent pharmacokinetic profile of cyclosporine, a
natural, multiply N-methylated cyclic peptide, we envisioned multiple N-methylation as a promising way to rationally improve
key pharmacokinetic characteristics. In this Account, we summarize our efforts toward modulating the properties of pep-
tides by multiple N-methylation.
As a first step, we simplified the synthesis of N-methylated amino acids in solution, by employing very mild conditions
that could be tolerated by the diverse protecting groups required when working with naturally occurring amino acids. We
also report the rapid and inexpensive syntheses of N-methylated peptides on a solid support; this facilitated the N-methyl
scanning of bioactive peptides. Because of a lack of information regarding the conformational behavior of multiply N-me-
thylated peptides, a complete library of N-methylated cyclic alanine pentapeptides was synthesized. The library provided
valuable insight into the conformational modulation of cyclic peptides by N-methylation. This information is extremely valu-
able for the design of bioactive peptides and spatial screening of cyclic N-methylated peptides.
To demonstrate the applicability of N-methylation to highly active but poorly bioavailable peptides, we performed a full
N-methyl scan of the cyclopeptidic somatostatin analog cyclo(-PFwKTF-), known as the Veber-Hirschmann peptide. We show
here for the first time that the simple approach of multiple N-methylation can drastically improve the metabolic stability
and intestinal permeability of peptides, for example, resulting in 10% oral bioavailability for a tri-N-methylated
Veber-Hirschmann peptide analog. In addition, we also describe a designed approach to N-methylated peptide library syn-
thesis, which can accelerate the screening of N-methylated bioactive peptides. Finally, we find that multiple N-methylation
of a cyclic hexapeptide integrin antagonist of GPIIb-IIIa (RIIbβ3 integrin), cyclo(-GRGDfL-), increases the selectivity of this
peptide toward different integrin subtypes. This result demonstrates the utility of multiple N-methylation in elucidating the
bioactive conformation of peptides.

Published on Web 07/18/2008

www.pubs.acs.org/acr Vol. xxx, No. xx Month XXXX 000 ACCOUNTS OF CHEMICAL RESEARCH A
10.1021/ar8000603 CCC: $40.75 © XXXX American Chemical Society
N-Methylation Chatterjee et al.

Introduction This search of matching is done by “spatial screening”.5

Peptides have found tremendous attention in diverse aspects Another major problem in developing peptidic drugs is their
of science ranging from rational drug design1 to nanomateri- enzymatic degradation in vivo, which eventually results in the
als.2 These diverse applications are due to their distinctive lowering of the pharmacokinetic profile (half-life, bioavailabil-
properties, such as ease of synthesis and characterization, ity, etc.). Medicinal chemists have developed an array of strat-
introduction of chemical diversity by simple amino acid sub- egies over the years to confront this problem, such as
stitution, and modulation of 3D structure by chemical modi- incorporating peptide bond isosters,6 peptoids,7 retro-inverso
fication. The application of peptides as drugs stems from their peptides,8 and peptidomimetics.9 Although these strategies
key role in many signal transduction pathways, which makes have elegant properties of their own, they demand careful
them an attractive avenue to target diseases. Despite the high design with challenging syntheses.
activity and receptor selectivity of naturally occurring bioac- Here we envisage the minimalist approach of N-methy-
tive peptides (or active protein fragments), they have distinct lation to overcome various obstacles of peptides as a “ratio-
disadvantages for practical application in medicine, such as nal” way toward drug development. Mono-N-methylation
short half-life in vivo and lack of oral availability. The initial has been used for years to change phamacological proper-
step in drug research of peptides is usually simplification (e.g., ties of peptides.10 However, due to difficulties in synthe-
reduction in size), followed by peptidomimetic approaches to sis 11 and the expectation of losing activity, multiple
ensure metabolic stability, with the final goal of an orally avail-
N-methylation has seldom been used.12 Although, there are
able, highly active, and selective drug. Whereas the prelimi-
prominent examples of multiply N-methylated natural cyclic
nary steps can be done in a rational way with relatively high
peptides,13 cyclosporine, omphalotin, etc. (Figure 1), with
probability of success, the final, crucial step of conversion from
remarkable biological and pharmacological profile.
peptide into a drug is often more problematic.
Cyclosporine A, with its seven N-methylated peptide bonds,
Not all of the amino acids in a peptide sequence are essen-
violates all the Lipinski rules for oral availability14 but is
tial to achieve the biological effect. The initial identification of
marketed as an orally available (oral bioavailability 28% (
the “bioactive sequence”, the minimal sequence3 required to
18%) immunosuppressive drug. Thus, in recent years, we
achieve the biological activity, is often done by alanine scan-
have introduced multiple N-methylation into cyclic peptides
ning. This is the systematic substitution of each amino acid by
alanine to identify the key residues, that is, those whose sub- to characterize the versatile properties of this modification.
stitution results in reduced activity. The next important factor We have developed an efficient and practical synthesis of
is the conformation of the peptide. In the majority of such pep- N-methylated amino acids in solution 15 and on solid
tides, a major obstacle in the study of the “bioactive sequence” phase16 and showed that multiple N-methylation not only
is intrinsic flexibility. Thus, the active sequence must be rigid- can dramatically improve the receptor subtype selectivi-
ified in a defined conformation in order to achieve the desired ty 17 but also can confer oral bioavailability. 18 Here, we
activity and selectivity. Reduction of conformational space can summarize the synthesis, conformational behavior, 19,20
be achieved by cyclization, resulting in highly active and selec- pharmacokinetic properties and modulation of bioactivity
tive derivatives when the bioactive conformation is matched.4 by multiple N-methylation of cyclic peptides.

FIGURE 1. Naturally occurring multiply N-methylated cyclic peptides: (A) cyclosporine A and (B) omphalotine.

B ACCOUNTS OF CHEMICAL RESEARCH 000 Month XXXX Vol. xxx, No. xx

 N-Methylation Chatterjee et al.

Synthesis of N-Methylated Peptides: letter code denotes D-amino acid) exhibits a βII′-turn around
8
Problems and Solutions D-Trp and Lys9 and a βVI-turn around Phe11 and Pro6, and
Over the years, multiply N-methylated peptides have failed to we could achieve efficient cyclization by using MeLys9 as the
attract the attention of medicinal chemists due to various dis- C-terminus and Thr10 as the N-terminus. On the other hand,
advantages encountered in their synthesis. First, a general when Pro6 and Phe7 were taken as the C- and N-termini,
approach to the synthesis of the N-methylated amino acids is respectively, the peptide completely failed to cyclize despite
very high yield of the linear peptide (Figure 2). Thus, termini
a challenging task, and this is followed by the difficult cou-
for efficient peptide cyclizations should be chosen in a way
pling of the preceding amino acids to the sterically hindered
that results in the closure of a turn; preferably a βII/II′-turn.
N-methylated site. Synthesis of N-methylated peptides and
cyclic peptides was revolutionized after the groundbreaking
Synthesis of N-Methylated Amino Acids
total synthesis of cyclosporin by Wenger.21 The synthesis was
and Peptides
carried out in solution using Boc chemistry, and owing to the
Solution Synthesis of N-Methylated Amino Acids. When
fortunate lack of diversely functionalized amino acids in
the goal is the synthesis of libraries of N-methylated peptides,
cyclosporine, the difficult couplings on the N-methylated termi-
it is preferable to use N-methylated amino acids directly as
nus could be carried out by the formation of reactive acid chlo-
building blocks. Despite a plethora of available methods,25 the
ride. In solid phase peptide synthesis, easy and fast coupling of
commercially available N-methylated amino acids are still very
N-methylated amino acids is achieved by using bis(2-oxo-3-ox-
expensive. Whereas amino acids with aliphatic side chains
azolidinyl)phosphonic chloride (BOP-Cl),22 2-(7-aza-1H-benzotria-
and without functional groups are best synthesized by the
zole-1-yl)-1,1,3,3-tetramethyluronium hexafluorophosphate
Freidinger method (via the reductive ring opening of the 5-ox-
(HATU)/1-hydroxy-7-azabenzotriazole (HOAt),23 and bis-(trichlo-
azolidinone using TFA and triethylsilane),26 the side chains of
romethyl)carbonate (BTC).24 We observed that, although expen-
functionalized amino acids may react under these conditions.
sive, difficult couplings with HATU/HOAt using double coupling
Hence, we developed an improved method yielding enan-
are more efficient than using BTC and BOP-Cl. Although BTC is
tiopure N-methylated amino acids quantitatively to be used
an effective reagent, it demands strict anhydrous conditions
directly in Fmoc solid phase peptide synthesis.
(alteration results in cleavage from the resin) and higher equiv-
The most efficient method for the site-selective N-methy-
alents of base, which might lead to racemization.
lation of peptides to date was developed by Fukuyama27 and
We prefer solid phase synthesis for all linear precursor pep- Miller.28 This is a three-step procedure involving amine acti-
tides. The sequence of amino acids during peptide synthesis vation by protection with an o-nitrobenzenesulfonyl group
and subsequent cyclization is crucial both for averting side (o-NBS), followed by alkylation and deprotection of the o-NBS
reactions and for efficient cyclization. A common problem group on a solid support (Scheme 1A). However, the major
encountered in the solid phase synthesis of N-methylated pep- drawback of the procedure is that only small quantities of
tides is the cleavage from the resin by the formation of dike- N-methylated amino acids can be prepared on the solid phase.
topiperazine due to the higher population of cis peptide bond Therefore, for the large-scale synthesis of N-methylated amino
(of the tertiary amide bond) when an N-methylated amino acid acids in solution, our strategy was to modify this procedure as
or proline are loaded to the resin.11 The cyclization yield of follows:
N-methylated peptides is strongly dependent on the linear • substitution of the expensive base MTBD (7-methyl-1,5,7-
sequence. We always perform the cyclizations in solution triazabicyclo[4.4.0]dec-5-ene) by a structurally similar inex-
under high dilution conditions (0.1-0.3 mM), because cycliza- pensive base DBU (1,8-diazabicyclo[5.4.0]undec-7-ene) and
tion on solid phase is often accompanied by cyclodimeriza- dimethyl sulfate as the methylating agent (Scheme 1B).
tions due to the relatively high concentration of peptide on the • masking the carboxyl group of the amino acid by a methyl
resin. Linear peptides exhibit a large number of conforma- ester for the synthesis of N-methyl-o-NBS-Xaa methyl
tions in solution; hence, any preferred conformation that esters.
brings the C- and N-termini in close proximity enhances the The racemization-free saponification of the N-methyl-o-
cyclization yield. We encountered several instances where the NBS-Xaa methyl esters to N-methyl-o-NBS-Xaa was done by
linear peptide completely failed to cyclize owing to the lack of a SN2 mechanism, because N-methylated amino acids are
a preferred conformation in solution: for example, a soma- infamous for undergoing racemization by base-mediated
tostatin analog cyclo(-P6F7Mew8MeK9T10MeF11-) (small single- saponification due to the absence of an amide proton.29 How-

Vol. xxx, No. xx Month XXXX 000 ACCOUNTS OF CHEMICAL RESEARCH C

N-Methylation Chatterjee et al.

FIGURE 2. Choice of the preferred cyclization site in a linear peptide, TMeFPFMewMeK. The yield of the cyclized peptide is given.

SCHEME 1. The Site Selective N-Methylation (A) on solid support and (B) Synthesis of N-Methyl-o-NBS-Amino Acids15

ever, LiI in refluxing pyridine30 yielded a mixture of products The optimization of the o-NBS protection step revealed
(including cleavage of N-methyl groups due to nonselective completion of the reaction without any racemization in the
attack of iodine), but the reaction worked well by using reflux- surprisingly short time of 15 min in the polar NMP (N-meth-
ing ethyl acetate for 16 h, yielding the corresponding enan- ylpyrrolidone) using 4 and 10 equiv each of o-NBS-Cl and col-
tiopure N-methyl-o-NBS-Xaa in quantitative yields to be used lidine, respectively, whereas reactions in THF (>2 h) or DCM
directly for solid phase peptide synthesis (SPPS). (∼1 h) required much longer reaction times.
Optimized Solid Phase Synthesis. When small amounts The subsequent alkylation step in the original procedure,
of amino acids are needed, for example, during N-methyl using MTBD in DMF is completed in 30 min. However, when
scanning of bioactive peptides, the simplest strategy is the 3 equiv of the less expensive DBU and 10 equiv of dimeth-
direct N-methylation of the desired amino acid on the solid ylsulfate in NMP were used, the reaction was complete in only
support during the peptide synthesis. The Miller and Scanlan 5 min. This method was efficient for all the amino acids inves-
procedure28 was chosen, and optimization was done using tigated, yielding products with >99% purity. The only excep-
DBU as base and the more polar NMP (N-methylpyrrolidone) tion was His(Trt), which showed N-methylation of the side
as a solvent to reduce the total time (Scheme 2). chain with the unexpected loss of the trityl protecting group.
SCHEME 2. On-Resin Synthesis of N-Methyl-Xaa Dipeptides16
To overcome this problem, the Mitsunobu procedure was
used, although it requires the change of solvent from NMP to
THF.31 N-Methylation of the resin bound NR-o-NBS-dipeptide
is performed with 5 equiv of triphenylphosphine, 10 equiv of
methanol, and 5 equiv of diisopropyl azodicarboxylate (DIAD)
in THF. Monitoring the reaction over time revealed that only
10 min is required for the completion of the reaction, and
hence it can be employed as a very fast and efficient method
for N-methylation of peptides on a solid support. It is worth
mentioning that this optimized method of N-methylation is

D ACCOUNTS OF CHEMICAL RESEARCH 000 Month XXXX Vol. xxx, No. xx

 N-Methylation Chatterjee et al.

also efficient for the introduction of functionalized larger alkyl NMR time scale.19 To serve as templates for rational drug
groups into peptides.32 design, we were interested only in those peptides that exhib-
The removal of the o-NBS protection is achieved by the for- ited a preferred conformation (>80% population of a single
mation of a Meisenheimer complex by reaction of the resin conformer). It should be noted that in these small cyclic pep-
bound dipeptide with 5 equiv of DBU and 10 equiv of mer- tides, the conformation is dictated primarily by the steric inter-
captoethanol and was optimized to be complete in 5 min, actions of the bulky -CH3 groups (R-methyl or N-methyl),
compared with 30 min in the original procedure. Thus, we are rather than the internal hydrogen bonds, which have been
now equipped with fast and highly efficient methods to pre- overemphasized over the years in stabilizing cyclic peptide
pare large amounts N-methylated amino acids and peptides conformation.
to create libraries of multiply N-methylated peptides.

Conformational Impact of Multiple

N-Methylation
Spatial Screening. The conformation of cyclic peptides of
smaller ring size is mainly dictated by the pattern of chirality
(D- or L-) of the amino acids in the peptide sequence.20 Hence,
peptides that consist of only alanine with a fixed pattern of
chirality can be used as template structures for designing bio-
active peptides, where alanine is replaced with appropriate
amino acids (pharmacophores), barring glycine and proline.
N-Methylation introduces another dimension to this “spatial
screening”5 (Figure 3) owing to the remarkable property of
conformational modulation. N-Methylation facilitates the
occurrence of a cis peptide bond and blocks potential hydro-
gen bonds, resulting in a long-range impact, especially on the
backbone conformation of cyclic peptides.33

FIGURE 4. The library of N-methylated cyclic alanine peptides.

Numbers in parentheses describe the relative populations of
detectable conformers by NMR (yellow squares denote the
conformationally homogeneous peptides (>98%), and the gray
ones denote the peptides showing a preferred conformation
(>80%) on the NMR time scale).
FIGURE 3. A peptide with pharmacophoric groups A, B, C, D, and E
can be screened for the spatial orientation in the bioactive
conformation by the synthesis of the five position-shifted cyclic Template Structures. Out of the 16 peptides selected, 15
isomers. In the absence of N-methylation, the five isomers would
have identical constitution but present pharmacophores differently.
are grouped into five different classes by virtue of the site of
In this example, however, the five di-N-methylated peptides with their cis peptide bond (10 could not be characterized due to
shifted N-methylated peptide bonds are constitutional isomers. spectral overlap). It should be noted that cyclic pentapeptides
To elucidate the impact of N-methylation on the backbone still have considerable conformational flexibility.34 Flipping of
of cyclic peptides, we synthesized a library of 30 N-methy- the plane of the peptide bond by synchronous rotation of
lated peptides, based on cyclo(-D-Ala-L-Ala4-) (Figure 4). NMR adjacent Φ and Ψ angles is often observed and can be fast on
analysis of these peptides displayed various populations of the NMR time scale. Hence, the conformations shown in the
major and minor conformers slowly interconverting on the figures are preferred structures.

Vol. xxx, No. xx Month XXXX 000 ACCOUNTS OF CHEMICAL RESEARCH E

N-Methylation Chatterjee et al.

FIGURE 5. (A) Classes of N-methylated cyclo(-D-Ala-L-Ala4-) (wavy lines indicate cis peptide bond; a indicates D-Ala1) and (B) superimposed
backbone conformation of the members in each class. Peptides 18 and 26 are in cyan, highlighting differences in peptide bond orientation
caused by the N-methyl group.

Class I consists of six peptides with all trans peptide bonds, a Φ angle close to 120°.19 A similar N-methyl scan of a cyclic
having N-methylation at D-Ala1, Ala2, and Ala5 or a combina- hexapeptide cyclo(-D-Ala-L-Ala5-) revealed that the N-methy-
tion of these three sites (Figure 5). Class II contains peptides lation of D-alanine results in conformational homogeneity, as
having a cis peptide bond between Ala2 and Ala3. Class III con- has been recently confirmed in designing N-methylated cyclo-
tains the three peptides with a cis peptide bond between Ala4 peptidic scaffolds against colon cancer.35 Thus, the turn-in-
and Ala5. One notable difference in this class is the orienta- ducing property of N-methylated D-alanine or any
tion of the D-Ala1 N-methyl group in 18, which undergoes a N-methylated D-amino acid (except glycine) is at least equal to
flip of about 180° from its preferred orientation (projecting that of D-proline. This will open a new dimension for the
above the plane of the ring in Figure 5) as a consequence of design of β-hairpin conformations36 in cyclic protein epitope
strong steric clash between the N-methyl group and Ala5 mimetics by using NMe-D-Xaa-L-Pro, NMe-D-Xaa-L-MeXaa, and
methyl group. Class IV peptides show the characteristics of NMe-D-Xaa-L-Xaa as templates to induce a βII′-turn instead of
both classes II and III, exhibiting two cis peptide bonds the conventionally used D-Pro-L-Pro36 (Figure 6). The clear
between Ala2 and Ala3 and Ala4 and Ala5. Class V contains a advantage of this method is the ability to incorporate differ-
unique peptide with a cis peptide bond between Ala3 and ent N-methylated amino acids, providing much greater flexi-
Ala4. All the other peptides with an Ala3-Ala4 N-methylated bility in functionalization of the turn-inducing region.
peptide bond, that is, 13, 16, 18, 21, and 26, exhibit a trans Systematic Modulation. A clear picture of the conforma-
peptide bond. This pattern probably arises from the fact that tional modulation by successive N-methylation can be
the parent peptide (4, Figure 4) with the N-methylated obtained by classifying these peptides based on their N-me-
Ala3-Ala4 peptide bond exists in a 1:1 equilibrium between thylation site (Figure 7). Starting with 5, N-methylation on
cis and trans conformers by NMR. Thus, N-methylation at Ala2 either side of the N-methylated peptide bond results in 15 and
shifts the equilibrium strongly toward a cis orientation whereas 13. N-methylation of D-Ala1 is tolerated in 15 without intro-
N-methylation at any other site shifts toward trans. ducing any cis peptide bond, whereas N-methylation of Ala4
The Ala5-Ala3 region of the cyclic pentapeptide is con- introduces a cis peptide bond in 13, and this pattern is fol-
served, and N-methylation does not introduce cis peptide lowed upon further N-methylation.
bonds. Of the conformationally homogeneous peptides Thus, conformational modulation by N-methylation on
(>98%), six out of seven have N-methylated D-Ala1 and show cyclic peptide backbone is defined and not irregular. These

F ACCOUNTS OF CHEMICAL RESEARCH 000 Month XXXX Vol. xxx, No. xx

 N-Methylation Chatterjee et al.

FIGURE 6. Conventional β-hairpin turn inducer, D-Pro-L-Pro, which could be replaced by NMe-D-Xaa-L-Pro, NMe-D-Xaa-L-MeXaa, and NMe-D-
Xaa-L-Xaa. R and R′ represent amino acid side chains.

development of metabolically stable analogues. In the

approach taken by Sandoz, higher metabolic stability of soma-
tostatin analogs was achieved by the following modifications,
resulting in the compound dubbed “Sandostatin” (Figure 8):38
(i) reduction in the size from 14 amino acids to 8; (ii) exchange
of Trp8 with D-Trp8; (iii) shift of the disulfide bridge closer to
the “active loop” (amino acids 6-11); (iv) change of the N-ter-
minal phenylalanine into D-phenylalanine and the C-termi-
nal threonine into reduced threoninol to avoid enzymatic
cleavage (Figure 8). This resulted in a dramatically longer half-
life in vivo. However, Sandostatin (octreotide) is not orally
available and must be administered by i.v. injection. Out of a
plethora of somatostatin receptor agonists, the earliest was
FIGURE 7. Modulation of conformation by successive N- cyclo(-PFwKTF-), discovered rationally by the group of
methylation.
Ralph Hirschmann at Merck Inc. and known as the Veber-
described templates can be now used as scaffolds in drug
Hirschmann peptide (Figure 8). This peptide was reported to
design by substituting alanine with other amino acids repre-
be selective toward the somatostatin receptor subtypes sst2
senting the pharmacophores of interest. The knowledge of the
and sst5 and showed excellent activity toward the inhibition
impact of N-methylation allows the design of N-methylated
of insulin, glucagon, and growth hormone secretion, surpass-
biologically active peptides without distorting the (bioactive)
ing the activity of somatostatin itself.39 However, the com-
conformation, leading to modification of pharmacokinetic
pound was administered by subcutaneous injection. Although
parameters without loss of biological activity.
end-to-end cyclization improved metabolic stability of the pep-
Multiple N-Methylation Imparts Oral tide in serum, to confer oral bioavailability, the improvement
Bioavailability to Somatostatin Analogs of intestinal permeability and stability against gut enzymes
To investigate possible improvements in the pharmacokinetic and enhanced uptake from the gut to systemic circulation is
properties of bioactive peptides by N-methylation, we chose necessary. We envisioned that multiple N-methylation of
the well-studied somatostatin system. Somatostatin is a major cyclo(-PFwKTF-) might convey improved pharmacokinetic
endocrine hormone and physiological inhibitor of pancreatic properties, making it orally bioavailable.
and gastrointestinal secretion, of growth hormone, glucagons, Library Approach. A library of all possible N-methylated
and insulin.37 However, somatostatin has a very short plasma analogs of cyclo(-PFwKTF-) was synthesized (except the penta-
half-life, <3 min, and therefore, there was a need for the N-methylated analog), resulting in 30 analogs. Out of these

Vol. xxx, No. xx Month XXXX 000 ACCOUNTS OF CHEMICAL RESEARCH G

N-Methylation Chatterjee et al.

FIGURE 8. Truncation of somatostatin into Sandostatin and the Veber-Hirschman peptide. The active analogs resulted from the N-
methylation of the amides shown by arrows.

30, seven analogs showed binding affinity in the nanomolar

range toward sst2 and sst5 receptor subtypes (Table 1). Pre-
liminary tests of S1-S8 by oral administration into rats
showed only S1 and S8 to be significantly taken up into the
blood.

TABLE 1. pKd Values of S1-S8 toward hsst2 and hsst5 Receptorsa

peptide N-methylated amino acid hsst 2 (pKd) hsst 5(pKd)
octreotide none 9.18 7.71
S1 none 8.01 7.82
S2 Lys9 8.60 8.19
S3 Phe11 7.93 8.28
8
S4 D-Trp 7.61 7.87 FIGURE 9. Peptide permeability across Caco-2 monolayer.
S5 Lys9, Phe11 7.96 7.39
8 9
S6 D-Trp , Lys 7.60 7.19
8 11
S7 D-Trp , Phe 7.16 7.47 meability of S8 was observed, which exceeded even the per-
8 9 11
S8 D-Trp , Lys , Phe 7.21 7.22
a
meability of mannitol (paracellular marker).
Higher pKd corresponds to higher affinity.
Conformational Details. Whereas a βII′, βVI conforma-
Hence, detailed pharmacokinetic experiments were per- tion was established in one early NMR structure,45 the pres-
formed. These eight peptides showed a stable profile in the rat ence of a “flat” or “bent” conformation has been an issue of
serum; however, a significant difference was found between debate regarding the real bioactive conformation of cyclo-
S1 and S8 in their stability against gut enzymes, revealing the (-PFwKTF-);41 Goodman et al. suggested that the peptide
stability effect conferred by multiple N-methylation. exhibits a “bent” conformation,42 with a kink in the backbone
The transport mode of these peptides through the intes- about Phe7 and Thr10, stabilized by the two additional hydro-
tine revealed interesting facts. We expected multiple N-me- gen bonds between Pro6CO-D-Trp8HN and Lys9CO-Phe11HN,
thylation to confer sufficient lipophilicity to enable the forming two closed γ-turns. On the other hand, Veber et al.
peptides to cross the membrane via the transcellular mecha- suggested the “flat” conformation without the two γ-turns to
nism (where the peptides interact with the lipophilic mem- be the bioactive conformation.43 Interesting evidence was
brane of the enterocytes).40 However, no such trend was found while screening the interactions of these eight analogs
observed (Figure 9). All the N-methylated peptides permeate with the liposomal model of the cell membrane (Figure 10),
the membrane by a paracellular mechanism through the tight where there was a sudden increase in the liposomal interac-
junctions (aqueous extracellular route across the epithelia) with tion of S2 (the most active analog) despite being mono-N-
low permeability. Surprisingly a significant increase in the per- methylated.

H ACCOUNTS OF CHEMICAL RESEARCH 000 Month XXXX Vol. xxx, No. xx

 N-Methylation Chatterjee et al.

Multiple N-Methylation Imparts Subtype

Selectivity to Integrin Ligands:
Design Approach. Instead of the commonly used library
approach, we envisioned an approach using designed multi-
ple N-methylation, where only the externally oriented amide
bonds were targeted.17 This was based on our experience of
the somatostatin analogs, where N-methylation of only the
externally oriented amide bonds results in the bioactive ana-
logs, whereas targeting the internally oriented ones distorts
the bioactive conformation. To test our hypothesis, we chose
FIGURE 10. The effect of N-methyl position on interaction with the to N-methylate a cyclic hexapeptide RIIbβ3 integrin receptor
liposomal model of the cell membrane (higher %CR shows higher
antagonist, cyclo(-G1R2G3D4f5L6-)44 (Figure 12, R1) in an
interaction).
This is due to the enhanced lipophilicity of S2 compared attempt to confer oral bioavailability. RIIbβ3 is the most abun-
with S3 and S4, where a “bent” conformation of S2 is dant integrin on the surface of platelets and mediates forma-
observed in solution compared with the comparatively “flat” tion of thrombi by platelet aggregation. In the final step of
conformations of S3 and S4 (Figure 11). In the “bent” confor- blood clot formation, the activated RIIbβ3 binds to the blood
mation, the externally oriented amide hydrogens of D-Trp8 glycoprotein fibrinogen to cross-link platelets in a growing
and Phe11 are involved in a γ-turn with Pro7 and Lys9 carbo- thrombus. Thus compounds that compete with fibrinogen in
nyls, resulting in the solvent shielding of all amide HNs, mak- binding to RIIbβ3 can act as potent antithrombotic agents.45
ing the molecule more lipophilic. This “bent” conformation is The preliminary design criterion in this approach is a prior
observed in cyclo(-PFwKTF-) and S2 with N-methylated lysine; knowledge of the conformation of the lead peptide. An added
N-methylation of any other site results in partial to complete advantage would be knowledge of its active sequence.
loss of this “bent” conformation. Thus, the highest pKd of S2 Selectivity. The activity and selectivity of the lead com-
and the gradual decrease in activity with increasing N-methy- pound R1 and seven N-methylated analogs (R2-R7) are
lations suggests that the “bent” conformation is more active shown in Table 2. The lead structure was unselective; how-
than the “flat” one, and a successive loss in activity is seen in ever, satisfactory activity and selectivity toward RIIbβ3 was
moving from “bent” to the “flat” conformer. first obtained in R4, where the Arg2 was N-methylated. The
A 5-fold difference in the elimination half-life between S1 preferred RIIbβ3 selectivity is due to the hexapeptidic ligands,
and S8 (15.5 ( 2 and 74 ( 6 min, respectively) suggested where the -RGD- recognition sequence is part of the βII-turn
reduced proteolytic digestion or low hepatic or renal clear- (Figure 12B), and particularly by an extended conformation in
ance, which is preferable for a good drug. A 10-fold differ- this β-turn, such that the guanidine and the aspartic acid
ence in the volume of distribution at steady state for S1 and groups are farthest apart (length of binding pocket: RIIβ3 >
S8 (0.3 ( 0.1 and 3.7 ( 1.3 L/Kg, respectively) revealed that R5β1 > Rvβ3).45
distribution of S8 is not only limited to interstitial fluid and Absence of selectivity in R1 is due to the conformation of
blood but also to biological membranes. The absolute oral bio- cyclic hexapeptides of the family cyclo(-D-Xaa-L-Xaa5-) (Xaa )
availability of S8 was found to be 9.9%, which is remarkable all amino acids except proline and glycine) exhibiting a “pre-
for a peptidic drug obtained by the relatively simple modifi- ferred” βII′ and flexible βII/βI-turn. This flexibility about the rec-
cation of N-methylation. ognition motif RGD results in lower selectivity. However, when

FIGURE 11. Front and side view of the solution conformations of (A) S2 (note the “bend” in the backbone) and (B) S8 (complete loss in the
“bend” by N-methylation).

Vol. xxx, No. xx Month XXXX 000 ACCOUNTS OF CHEMICAL RESEARCH I

N-Methylation Chatterjee et al.

FIGURE 12. (A) R1 with two β-turns (solvent exposed amides (red) targeted for N-methylation) and (B) stereoview of cyclo(-GRGDfL-) (R1).
Note the stabilizing γ-turns about Asp4 and Gly1 (similar as that discussed above for the somatostatin hexapeptides).

TABLE 2. IC50 (nM) of the N-Methylated Analogs and Cyclo(- The solution conformations reveal the reduced flexibility in
GRGDfL-) toward Different Integrinsa the β-turn about Arg2 and Gly3 in R4; in addition, we observe
no. analogue R5β1 Rvβ3 RIIbβ3 Rvβ3/RIIbβ3 a kink in the backbone about Gly1 and Asp4 in R1 (Figure
R1 c(-GRGDfL-) 740 100 195 0.5 12B), a pattern that was also observed in the somatostatin
R2 c(-GRGDfL-) 3900 103 560 0.2
R3 c(-GRGDfL-) 4300 490 2000 0.2 analogs and is probably typical of the cyclo(-D-Xaa-L-Xaa5-)
R4 c(-GRGDfL-) 1200 770 12 64
R5 c(-GRGDfL-) >20000 1200 620 2
class of peptides. This kink helps in the formation of a γ-turn
R6 c(-GRGDfL-) ∼20000 1300 15 86 about Gly3, preventing an extended conformation in the
R7 c(-GRGDfL-) >20000 2730 165 16
R8 c(-GRGDfL-) >20000 12,200 30 406 β-turn and bringing the side chains of Arg2 and Asp4 in close
a
N-methylated residues are in bold. proximity, which is favored for binding to Rvβ3. This kink is
lost partially by the N-methylation of Arg2 resulting in the loss
Arg2 is N-methylated, the flexibility is reduced and the βII/βI- in binding to Rvβ3 in R4 (Figure 13A). A total loss in the kink
turn is presented in an extended orientation, affecting both the is observed by N-methylation of D-Phe5 in R7 and in R8 (Fig-
activity and selectivity of the ligand. ure 13B), presenting the peptide in a “flat” conformation.
IC50 values suggest that N-methylation at D-Phe5 is not tol- Docking (Figure 13C) revealed differences in the upper part
erated; whereas N-methylation at Leu6 is favored and results of the peptides, where N-methylation of D-Phe5 in R8 com-
in enhanced selectivity of R6. The most surprising result was
pared with R4 lowered the π-π interaction between the
the activity and selectivity profile of R8, where all the three
phenyl rings with β3-Tyr122 and a change in the preferred ori-
sites are N-methylated, in contrast to R5, where N-methyla-
entation of Leu6 carbonyl group to form hydrogen bond with
tion of D-Phe5 results in the loss of activity. Thus, one would
β3-Arg214 side chain. Unfortunately, docking could not give a
also expect a further loss in the activity in R8; instead a tre-
clear distinction between the binding of the ligands to Rvβ3
mendous enhancement in the selectivity was observed with a
slight loss in activity. and RIIbβ3.

FIGURE 13. Conformation of (A) R4 and (B) R8 and (C) docked R4 (yellow) and R8 (pink) in the RIIbβ3 integrin. RIIb subunit is represented
by the green surface, β3 by the violet, and metal ion by the magenta sphere. Reproduced with permission from ref 17. Copyright 2007
American Chemical Society.

J ACCOUNTS OF CHEMICAL RESEARCH 000 Month XXXX Vol. xxx, No. xx

 N-Methylation Chatterjee et al.

We have shown that a designed approach to multiple versity of Jerusalem. His main interest is the pharmacokinetics
N-methylation based on the conformation of the stem pep- and pharmacodynamics of drugs and drug interactions.
tide leads to the development of very potent and receptor sub- Horst Kessler is Professor at the Department Chemie of the Tech-
type selective ligands. Unfortunately, these peptides showed nische Universität München. His main interest is drug develop-
ment from peptides and peptidomimetics, as well as the
low permeability in the Caco-2 test. Although these peptides
development and application of multidimensional NMR experi-
might show enhanced duration of action due to proteolytic ments to proteins, small molecules, and their interactions.
stability, we discontinued the project due to lack of enhanced
permeability. FOOTNOTES
* To whom correspondence should be addressed. E-mail: [email protected]. Tel: +49-
Summary 89-28913300. Fax: +49-89-28913210.

Mono- and multiple N-methylations of cyclic peptides were

REFERENCES
investigated to elucidate their remarkable conformational 1 Marx, V. Watching peptide drugs grow up. Chem. Eng. News 2005, 83, 17–24.
modulation ability by imparting steric constraints in the pep- 2 Teixido, M.; Giralt, E. The role of peptides in blood-brain barrier nanotechnology. J.
Pept. Sci. 2008, 14, 163–173.
tidic backbone and to improve the pharmacokinetic profile of 3 Gurrath, M. Peptide-binding G protein-coupled receptors: New opportunities for drug
the peptides in order to be used as drug leads. The develop- design. Curr. Med. Chem. 2001, 8, 1605–1648.
4 Kessler, H. Peptide conformations. 19. Conformation and biological-activity of cyclic-
ment of simplified and easily scalable synthesis of N-methy- peptides. Angew. Chem., Int. Ed. 1982, 21, 512–523.
lated amino acids in solution and of peptides in solid phase 5 Kessler, H.; Gratias, R.; Hessler, G.; Gurrath, M.; Müller, G. Conformation of cyclic
peptides. Principle concepts and the design of selectivity and superactivity in
accelerated the N-methyl scanning technique of peptides. The bioactive sequences by “spatial screening”. Pure Appl. Chem. 1996, 68, 1201–
elucidation of the conformational impact of N-methylation on 1205.
6 Houben-Weyl Methods of Organic Chemistry; Goodman, M., Felix, A., Moroder, L.,
cyclic peptides highly facilitates the design of bioactive pep- Tonolio, C., Eds.; Georg Thieme Verlag: Stuttgart, Germany, 2002; Vol. E22c,
tides by “spatial screening”, wherein the side chains in the Chapter 10, pp 213-633.
7 Kessler, H. Peptoids-a new approach to the development of pharmaceuticals.
template structures are functionalized by appropriate pharma- Angew. Chem., Int. Ed. 1993, 32, 543–544.
cophores. The improvement of oral bioavailability by multi- 8 Fletcher, M. D.; Campbell, M. M. Partially modified retro-inverso peptides:
Development, synthesis, and conformational behavior. Chem. Rev. 1998, 98, 763–
ple N-methylation is a significant advance toward the 795.
development of peptide-based therapeutics, which has ham- 9 Giannis, A. Peptidomimetics for receptor ligands discovery, development, and
medical perspectives. Angew. Chem., Int. Ed. 1993, 32, 1244–1267.
pered over the years due to poor pharmacokinetic proper- 10 Gilon, C.; Dechantsreiter, M. A.; Burkhart, F.; Friedler, A.; Kessler, H. Synthesis of
ties. Multiple N-methylation resulted in enhancement in the peptides and peptidomimetics. Houben-Weyl Methods of Organic Chemistry;
Goodman, M., Felix, A., Moroder, L., Tonolio, C., Eds.; Georg Thieme Verlag:
activity and selectivity of receptor subtypes using either library Stuttgart, Germany, 2002; Vol. E22c, pp 215-291.
or designed approaches and helps in understanding finer 11 Teixido, M.; Albericio, F.; Giralt, E. Solid-phase synthesis and characterization of
N-methyl-rich peptides. J. Pept. Res. 2005, 65, 153–166.
details of the bioactive conformation. Thus, with these diverse 12 Holladay, M. W.; Kopecka, H.; Miller, T. R.; Bednarz, L.; Nikkel, A. L.; Bianchi, B. R.;
properties, we foresee a bright future for peptide chemistry by Witte, D. G.; Shiosaki, K.; Lin, C. W.; Asin, K. E.; Nadzan, A. M. Tetrapeptide CCK-a
agonists - Effect of backbone N-methylations on in-vitro and in-vivo CCK activity.
multiple N-methylation toward their development as a thera- J. Med. Chem. 1994, 37, 630–635.
peutic prototypes. 13 Hamada, Y.; Shioiri, T. Recent progress of the synthetic studies of biologically active
marine cyclic peptides and depsipeptides. Chem. Rev. 2005, 105, 4441–4482.
14 Lipinski, C. A.; Lombardo, F.; Dominy, B. W.; Feeney, P. J. Experimental and
We thank the Humboldt Foundation for Max-Planck-Fors- computational approaches to estimate solubility and permeability in drug discovery
and development settings. Adv. Drug Delivery Rev. 1997, 23, 3–25.
chungspreis, German-Israel Foundation for their funding sup- 15 Biron, E.; Kessler, H. Convenient synthesis of N-methylamino acids compatible with
Fmoc solid-phase peptide synthesis. J. Org. Chem. 2005, 70, 5183–5189.
port, and our collaborators at Novartis, and Jerini A.G. We also
16 Biron, E.; Chatterjee, J.; Kessler, H. Optimized selective N-methylation of peptides
thank Dr. Luciana Marinelli for docking studies in the RGD field. on solid support. J. Pept. Sci. 2006, 12, 213–219.
17 Chatterjee, J.; Ovadia, O.; Zahn, G.; Marinelli, L.; Hoffman, A.; Gilon, C.; Kessler, H.
Multiple N-methylation by a designed approach enhances receptor selectivity.
J. Med. Chem. 2007, 50, 5878–5881.
BIOGRAPHICAL INFORMATION
18 Biron, E.; Chatterjee, J.; Ovadia, O.; Langenegger, D.; Brueggen, J.; Hoyer, D.;
Jayanta Chatterjee is a graduate student pursuing his Ph.D. in Schmid, H. A.; Jelinek, R.; Gilon, C.; Hoffman, A.; Kessler, H. Improving oral
bioavailability of peptides by multiple N-methylation: somatostatin analogs. Angew.
the group of Prof. Horst Kessler. Chem., Int. Ed. 2008, 47, 2595–2599.
Chaim Gilon is Professor of Chemistry at the Institute of Chem- 19 Chatterjee, J.; Mierke, D.; Kessler, H. N-methylated cyclic pentaalanine peptides as
istry, The Hebrew University of Jerusalem. His main interest is the template structures. J. Am. Chem. Soc. 2006, 128, 15164–15172.
20 Chatterjee, J.; Mierke, D. F.; Kessler, H. Conformational preference and potential
development of technologies to convert peptides and active templates of N-methylated cyclic pentaalanine peptides. Chem.sEur. J. 2008, 14,
regions in proteins into drugs. 1508–517.
21 Wenger, D. M. Total syntheses of “Cyclosporin A” and “Cyclosporin H”, Two fungal
Amnon Hoffman is Associate professor of Pharmaceutics and metabolites isolated from the species Tolypocladium Inflaturn1 GAMS. Helv. Chim.
Clinical Pharmacy at the School of Pharmacy, The Hebrew Uni- Acta 1984, 67, 502–525.

Vol. xxx, No. xx Month XXXX 000 ACCOUNTS OF CHEMICAL RESEARCH K

N-Methylation Chatterjee et al.

22 Tung, R. D.; Rich, D. H. Bis(2-Oxo-3-Oxazolidinyl)Phosphinic Chloride (1) as a 36 Robinson, J. A., β-Hairpin peptidomimetics: Design, structures and biological
coupling reagent for N-alkyl amino-acids. J. Am. Chem. Soc. 1985, 107, 4342– activities. Acc. Chem. Res., published online April 16, 2008, http://dx.doi.org/
4343. 10.1021/ar700259k.
23 Carpino, L. A.; Elfaham, A.; Albericio, F. Efficiency in peptide coupling - 1-Hydroxy- 37 Gerich, J. E.; Lovinger, R.; Grodsky, G. M. Inhibition by somatostatin of glucagon
7-azabenzotriazole vs 3,4-dihydro-3-hydroxy-4-oxo-1,2,3-benzotriazine. J. Org. and insulin release from perfused rat pancreas in response to arginine, isoproterenol
Chem. 1995, 60, 3561–3564.
and theophylline - Evidence for a preferential effect on glucagon-secretion.
24 Falb, E.; Yechezkel, T.; Salitra, Y.; Gilon, C. In situ generation of Fmoc-amino acid Endocrinology 1975, 96, 749–754.
chlorides using bis(trichloromethyl) carbonate and its utilization for difficult couplings
in solid-phase peptide synthesis. J. Pept. Res. 1999, 53, 507–517. 38 Bauer, W.; Briner, U.; Doepfner, W.; Haller, R.; Huguenin, R.; Marbach, P.; Petcher,
25 McDermott, J. R.; Benoiton, N. L. N-Methylamino acids in peptide synthesis 0.2. T. J.; Pless, J. SMS 201-995 - a very potent and selective octapeptide analog of
New synthesis of N-benzyloxycarbonyl,N-methylamino Acids. Can. J. Chem. 1973, somatostatin with prolonged action. Life Sci. 1982, 31, 1133–1140.
51, 1915–1919. 39 Veber, D. F.; Freidinger, R. M.; Perlow, D. S.; Paleveda, W. J.; Holly, F. W.;
26 Freidinger, R. M.; Hinkle, J. S.; Perlow, D. S.; Arison, B. H. Synthesis of 9- Strachan, R. G.; Nutt, R. F.; Arison, B. H.; Homnick, C.; Randall, W. C.; Glitzer,
fluorenylmethyloxycarbonyl-protected N-alkyl amino-acids by reduction of M. S.; Saperstein, R.; Hirschmann, R. A potent cyclic hexapeptide analog of
oxazolidinones. J. Org. Chem. 1983, 48, 77–81. somatostatin. Nature 1981, 292, 55–58.
27 Fukuyama, T.; Jow, C. K.; Cheung, M. 2-Nitrobenzenesulfonamides and 4- 40 Knipp, G. T.; Velde, D. G. V.; Siahaan, T. J.; Borchardt, R. T. The effect of beta-turn
Nitrobenzenesulfonamides - Exceptionally versatile means for preparation of
structure on the passive diffusion of peptides across Caco-2 cell monolayers.
secondary-amines and protection of amines. Tetrahedron Lett. 1995, 36, 6373–
6374. Pharm. Res. 1997, 14, 1332–1340.
28 Miller, S. C.; Scanlan, T. S. Site-selective N-methylation of peptides on solid 41 Kessler, H.; Bernd, M.; Kogler, H.; Zarbock, J.; Sørensen, O. W.; Bodenhausen, G.;
support. J. Am. Chem. Soc. 1997, 119, 2301–2302. Ernst, R. R. Relayed heteronuclear correlation spectroscopy and conformational
29 Cheung, S. T.; Benoiton, N. L. N-Methylamino acids in peptide-synthesis 0.5. analysis of cyclic hexapeptides containing the active sequence of somatostatin.
Synthesis of N-tert-Butyloxycarbonyl, N-methylamino acids by N-methylation. Can. J. Am. Chem. Soc. 1983, 105, 6944–6952.
J. Chem. 1977, 55, 906–910. 42 He, Y. B.; Huang, Z. W.; Raynor, K.; Reisine, T.; Goodman, M. Syntheses and
30 Müller, P.; Siegfried, B. SN2 Reactions with carboxylic esters - Selective cleavage of conformations of somatostatin-related cyclic hexapeptides incorporating specific
methyl-esters. Helv. Chim. Acta 1974, 57, 987–994. alpha-methylated and beta-methylated residues. J. Am. Chem. Soc. 1993, 115,
31 Reichwein, J. F.; Liskamp, R. M. J. Site-specific N-alkylation of peptides on the solid 8066–8072.
phase. Tetrahedron Lett. 1998, 39, 1243–1246. 43 Veber, D. F. Design and discovery in the development of peptide analogs. In
32 Demmer, O.; Dijkgraaf, I.; Schottelius, M.; Wester, H.-J.; Kessler, H. Introduction of Proceedings of the 12th American Peptide Symposium; Smith, J. A., Rivier, J. E.,
functional groups into peptides via N-alkylation. Org. Lett. 2008, 10, 2015–2018. Eds.; ESCOM, Leiden: Cambridge, MA, 1991; pp 3-14.
33 Kessler, H. Detection of hindered rotation and inversion by NMR spectroscopy.
Angew. Chem., Int. Ed. 1970, 9, 219–235. 44 Pfaff, M.; Tangemann, K.; Müller, B.; Gurrath, M.; Müller, G.; Kessler, H.; Timpl, R.;
Engel, J. Selective recognition of cyclic RGD peptides of NMR defined conformation
34 Heller, M.; Sukopp, M.; Tsomaia, N.; John, M.; Mierke, D. F.; Reif, B.; Kessler, H.
The conformation of cyclo(-D-Pro-Ala(4)-) as a model for cyclic pentapeptides of the by alpha-II-beta-3, alpha-v-beta-3, and alpha-5-beta-1 integrins. J. Biol. Chem.
DL4 type. J. Am. Chem. Soc. 2006, 128, 13806–13814. 1994, 269, 20233–20238.
35 Otrubova, K.; McGuire, K. L.; McAlpine, S. R. Scaffold targeting drug-resistant colon 45 Heckmann, D.; Kessler, H. Design and chemical synthesis of integrin ligands.
cancers. J. Med. Chem. 2007, 50, 1999–2002. Integrins 2007, 426, 463–503.

L ACCOUNTS OF CHEMICAL RESEARCH 000 Month XXXX Vol. xxx, No. xx