Efficient solid-phase synthesis of peptide-based phosphine ligands: Towards combinatorial libraries of selective transition metal catalysts

Article abstract of DOI:10.1002/chem.200500105

A new methodology for the solid-phase synthesis of peptide-based phosphine ligands has been developed. Solid supported peptide scaffolds possessing either primary or secondary amines were synthesised using commercially available Fmoc-protected amino acids

Full text of DOI:10.1002/chem.200500105

FULL PAPER
Efficient Solid-Phase Synthesis of Peptide-Based Phosphine Ligands:
Towards Combinatorial Libraries of Selective Transition Metal Catalysts
Christian A. Christensen and Morten Meldal*[a]
Abstract: A new methodology for the
solid-phase synthesis of peptide-based
phosphine ligands has been developed.
Solid supported peptide scaffolds pos-
sessing either primary or secondary
amines were synthesised using com-
mercially available Fmoc-protected
amino acids and readily available
Fmoc-protected amino aldehydes for
reductive alkylation, in standard solid-
phase peptide synthesis (SPPS). Phos-
phine moieties were introduced by
phosphinomethylation of the free
amines as the final solid-phase synthet-
ic step, immediately prior to complexa-
tion with palladium(ii), thus avoiding
tedious protection/deprotection of the
phosphine moieties during the synthe-
sis of the ligands. The extensive use of
commercial building blocks and stan-
dard SPPS makes this methodology
well suited for the generation of solid-
phase combinatorial libraries of novel
ligands. Furthermore, it is possible to
generate several different phosphine
ligand libraries for every peptide scaf-
fold library synthesised, by functional-
ising the scaffold libraries with differ-
ent phosphine moieties. The synthes-
ised ligands were characterised on solid
support by conventional 31P NMR spec-
troscopy and, cleaved from the sup-
port, as their phosphine oxides by
HPLC, 1H NMR, 31P NMR and high
resolution ESMS. Palladium(ii) allyl
complexes were generated from the
resin bound ligands and to demonstrate
their catalytic properties, palladium
catalysed asymmetric allylic substitu-
tion reactions were performed. Good
yields and moderate enantioselectivity
was obtained for the selected combina-
tion of catalysts and substrate, but
most importantly the concept of this
new methodology was proven. Screen-
ing of ligand libraries should afford
more selective catalysts.
Keywords: asymmetric catalysis
·
allylic compounds · peptides · phos-
phine ligands · solid-phase synthesis
Introduction
catalysts for the Strecker reaction have been developed by
Jacobsen and co-workers.[13–15]
In Nature enzymes achieve their excellent efficiency and se-
lectivity in the catalysis of biological processes through a
combination of binding affinity and a delicate catalytic ma-
chinery. Hence, it is obvious to mimic Nature by designing
peptidic catalysts for asymmetric synthesis.[1,2] This strategy
has been explored by several groups. Peptides have been ap-
plied for asymmetric azidation and for regioselective func-
tionalisation of carbohydrates by Miller and co-workers.[3,4]
Functionalised peptides complexing transition metals, have
been utilised for a number of asymmetric addition and sub-
stitution reactions in the group of Hoveyda,[5–12] and peptidic
However, because of the expected high chemo-, regio-
and stereoselectivity of peptide-based catalysts, there is a
need for the development of a method for the identification
of the most efficient catalyst for any specific reaction. The
solution to this problem could be high-throughput screening
of combinatorial libraries of potential catalysts, thereby, for
each new reaction/synthetic challenge, being able to rapidly
pick the best catalyst.[16–19] Immobilisation of library mem-
bers on individual beads would greatly facilitate this screen-
ing. Thus it is important to be able to synthesise combinato-
rial libraries of peptide-based catalysts, preferentially on
solid support.
Resin bound catalysts based upon phosphine functional-
ised peptides combines the excellent catalytic properties
known from phosphine transition metal complexes with the
selectivity and possibly increased reactivity imposed by a
folded peptide scaffold, due to binding of the substrate. In
this context it is important for the phosphines to be situated
near the peptide backbone, so as to most efficiently transfer
[a] Dr. C. A. Christensen, Prof. M. Meldal
Carlsberg Laboratory, SPOCC-Centre
Gamle Carlsbergvej 10, 2500 Valby (Denmark)
Fax : (+45)3327-4708
E-mail: mpm@crc.dk
Supporting information for this article is available on the WWW
under http://www.chemeurj.org/ or from the author.
Chem. Eur. J. 2005, 11, 4121 – 4131
DOI: 10.1002/chem.200500105
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
4121

the chiral information to the transition metal.[20] Two types
of catalysts can be envisaged (Figure 1). For type I the che-
lating phosphines are attached to the chiral peptide back-
bone via substitution on a single side chain, which not neces-
sarily guaranties the transition metal to be embedded in a
chiral environment. On the other hand, when the chelating
phosphines are positioned at two different sites of the pep-
tide chain (type II), complexation can assist the folding of
the peptide, increasing the possibility for the active site to
be surrounded by a chiral architecture.
ised with phosphine moieties, several advantages could be
obtained. There would be no tedious protection/deprotec-
tion of the phosphine moieties and by using different phos-
phine reagents for functionalisation, several different ligands
and complexes thereof could be obtained from the same
peptide scaffold. Ideally, it should be possible to synthesise
such a scaffold from mainly commercially available building
blocks, combined with building blocks requiring a minimum
amount of simple solution-phase synthesis. Importantly, the
phosphine functionalisation of the peptide scaffolds should
position the phosphine moieties close to the chiral peptide
backbone and preferentially afford type II catalysts.
The new methodology presented is based upon this strat-
egy, with the ultimate goal of synthesising combinatorial li-
braries of resin bound highly selective catalysts. The meth-
odology is demonstrated by solid-phase synthesis of four dif-
ferent peptide-based phosphine ligands, the formation of
palladium(ii) allyl complexes on solid support and the
screening of these complexes in the palladium catalysed
asymmetric allylic substitution reaction.
Results and Discussion
Figure 1. Peptide-based phosphine-transition metal catalysts of type I and
II. In type II the folding of the peptide is assisted by the complexation.
M=transition metal.
Solid-phase synthesis of peptide-based bidentate phosphine
ligands: Initially, a reagent suitable for functionalising a
solid supported peptide scaffold with phosphine moieties
was investigated. Hydroxymethylphosphines can be easily
obtained in situ, by heating a neat mixture of paraformalde-
hyde and a secondary phosphine,[32] and have been shown to
react efficiently with both primary and secondary amines, af-
fording the phosphinomethylated species in a Mannich type
condensation. Furthermore, these very mild conditions
allows for the presence of several unprotected functional
groups like carboxylic acids, alcohols, phenols, esters and al-
kenes.[33] The most commonly used hydroxymethylphosphine
reagent has been the diphenyl derivative. In solution it has
been applied in the synthesis of spectacular dendritic struc-
tures,[34–37] as well as of more simple phosphine ligands made
from amino acids[38,39] and aminopyridines.[40,41] By phosphi-
nomethylation of primary amines, Alper, Arya and co-work-
ers[42–46] have illustrated the use of hydroxymethyldiphenyl-
phosphine as a reagent also suitable for solid-phase synthe-
sis. In this manner, dendrimers have been functionalised
with phosphine moieties and the derived palladium and rho-
dium complexes used for carbonylation reactions and selec-
tive hydroformylation, respectively. Two examples of non-
selective solid supported catalysts, made using hydroxyme-
thyldiphenylphosphine, were reported during the course of
this work.[47,48] These catalysts did not posses a peptide struc-
ture rather the primary amines of amino-resins were phos-
phinomethylated directly.
Recently Landis et al.[21] reported the incorporation of a
3,4-diazaphospholane moiety into a tetrapeptide, furnishing
a monodentate solid supported phosphine ligand, which was
applied in palladium catalysed allylic substitution. However,
the pioneering work of Gilbertson and co-workers,[22–31] re-
ported over the last ten years, constitutes the most impor-
tant examples of resin bound peptide-based phosphine tran-
sition metal catalysts. Gilbertsonꢁs diphenyl- and dicyclohex-
ylphosphinoserine building blocks contain an Fmoc-protect-
ed amine and a sulfide-protected phosphine.[22,23] They have
been used for the parallel synthesis of small libraries of
resin bound peptides containing sulfide-protected phosphine
moieties. Subsequent reduction[24] of the phosphine sulfides
and complexation with transition metals such as palladium
and rhodium afforded resin bound catalysts of type II
(Figure 1). Some of these catalysts were found to show good
selectivity in transformations such as hydrogenations,[25,26] al-
lylic substitutions[27,28] and desymmetrisation of meso-
diols.[29] However, although the synthesis of the diphenyl-
and dicyclohexylphosphinoserine building blocks was re-
cently improved,[30] resulting also in a series of new aryl
phosphinoserine building blocks,[28,31] it still involves several
steps and protection of the phosphine moiety. Furthermore,
incorporation of different phosphinoserine residues in the
peptide sequence, requires new building block synthesis for
each type of phosphine moiety.
Initially the diphenyl derivative was selected as the hy-
droxymethylphosphine reagent which should be used for the
phosphinomethylation of solid supported amino-functional-
ised peptide scaffolds, well aware that the use of other com-
mercially available secondary phosphines could be a source
Alternatively, if a peptide scaffold could be synthesised
on solid support and, as the final synthetic step, immediately
prior to complexation with transition metals, be functional-
4122
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2005, 11, 4121 – 4131

Phosphine Ligands
FULL PAPER
for randomisation in a future library synthesis. Reaction of a
primary amine with two equivalents or an excess of a hy-
droxymethylphosphine derivative affords a bidentate phos-
phine ligand, which upon complexation with a transition
metal furnishes a favoured six-membered chelate. Thus
solid-phase synthesis of peptide 6 (Scheme 1), possessing a
primary amine as a side-chain, provides a resin bound scaf-
fold suitable for the subsequent synthesis of peptide-based
bidentate phosphine ligands. Accordingly, complexation
with transition metals affords catalysts of type I (Figure 1).
As solid support, the PEGA1900 resin was selected due to
excellent swelling in organic solvents (acetonitrile, dichloro-
methane, N,N-dimethylformamide, dioxane, tetrahydrofuran
and toluene) as well as in water.[49] This would allow the use
of standard organic transformations during the synthesis of
the ligands and later the screening of the catalysts to be per-
formed in a range of solvents, including water, which is
timely in the light of the development of green chemistry.
The free primary amines of the PEGA1900 resin were func-
tionalised with Fmoc-glycine by N-[(1H-benzotriazol-1-yl)-
(dimethylamino)methylene]-N-methylmethanaminium tetra-
fluoroborate N-oxide (TBTU) activation in N,N-dimethyl-
formamide. Deprotection using 20% (v/v) piperidine in
N,N-dimethylformamide and subsequent TBTU coupling of
the hydroxymethylbenzoic acid (HMBA) linker furnished
the functionalised solid support 1 (Scheme 1), which was
used for the synthesis of all the ligands included in this
work. The HMBA linker is efficiently cleaved under mild
basic conditions and has proven to be ideal for the analysis
of split/mix combinatorial libraries.[50] Also the HMBA
linker is suitable for on-bead NMR analysis, since it possess-
es no stereocenter. In contrast, most chiral linkers (e.g. the
Rink linker) are only commercially available as racemic
mixtures, which would give rise to two sets of resonances in
the NMR spectrum.
Ligand precursor peptide 6 was synthesised following the
Fmoc protocol,[51] using commercially available Fmoc-amino
acid derivatives. Phenylalanine was selected as the main res-
idue, to increase the steric bulk around the primary amine
side chain, which in turn should increase the selectivity of
the derived catalyst. The first Fmoc-phenylalanine was at-
tached to the HMBA linker by 1-(mesitylene-2-sulfonyl)-3-
nitro-1,2,4-triazole (MSNT) activation in dichloromethane.
Removal of the Fmoc group and TBTU coupling of a
second Fmoc-phenylalanine yielded resin 3. To position the
primary amine and hence the reactive site of the catalyst as
close as possible to the chiral backbone of the peptide, a,b-
diaminopropionic acid (Dap) was incorporated as the b-ami-
nopropionic acid, affording a primary amine side chain.
Thus Boc-Dap(Fmoc)-OH was coupled using TBTU to give
resin 4 and elongation by another Fmoc-phenylalanine af-
forded resin 5, which was Fmoc deprotected and capped by
TBTU coupling of benzoic acid. Subsequently, the Boc pro-
tecting group of the side chain primary amine was removed
by trifluoroacetic acid (TFA) treatment, which yielded pep-
tide scaffold 6 in high purity.[52] The resin bound bidentate
phosphine ligand 7 was furnished by overnight treatment of
6 with a solution of hydroxymethyldiphenylphosphine. The
formation of 7 was indicated by the presence of only a
single resonance at ꢀ29.7 ppm in the 31P NMR spectrum of
the resin bound ligand. However, to investigate the purity of
the cleaved product in solution, the phosphines were oxi-
dised to their phosphine oxides before cleavage and HPLC
analysis. In this manner partial oxidation of the ligand
during analysis was avoided. The HPLC showed only a
single peak, proving a clean and quantitative conversion of
the amine 6, and the structure of the oxidised ligand 8 was
confirmed by HRMS and NMR spectroscopy.[52] Thus hy-
droxymethyldiphenylphosphine proved to be a reagent com-
patible with both peptide scaffolds and the PEGA resin.
Furthermore, excess hydroxymethyldiphenylphosphine was
easily washed out of the resin. This is a very important fea-
ture, since excess reagent trapped in the resin could other-
wise compete for catalysis of reactions and thus reduce the
selectivity of the immobilised peptide-based catalyst.
Scheme 1. Solid-phase synthesis and oxidation of peptide-based bidentate
phosphine ligand 7. a) Fmoc-Phe-OH, MSNT, methylimidazole, CH2Cl2;
b) piperidine/DMF 2:8; c) Fmoc-Phe-OH, TBTU, NEM, DMF; d) Boc-
Dap(Fmoc)-OH, TBTU, NEM, DMF; e) benzoic acid, TBTU, NEM,
DMF; f) TFA/CH2Cl2 1:1; g) Ph2PCH2OH, MeCN; h) 3% aqueous H2O2.
Chem. Eur. J. 2005, 11, 4121 – 4131
www.chemeurj.org
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
4123

M. Meldal and C. A. Christensen
With the methodology for the synthesis and analysis of bi-
dentate ligands such as resin 7 in hand, the more challenging
peptide scaffolds 18a–c could be pursued (Schemes 3 and
4). Scaffolds 18a–c contain two reduced amide bonds and
the secondary amines are able to accommodate one phos-
phinomethyl moiety each, upon reaction with hydroxyme-
thyldiphenylphosphine. Thus, not only do scaffolds 18a–c
bring the phosphine moieties closer to the chiral peptide
backbone, they also provide catalysts of the desired type II.
To prepare peptide scaffold 18a, a reduced bond dipep-
tide building block 14, which could be used in standard
Fmoc-based solid-phase peptide synthesis, was synthesised
in solution (Scheme 2). It was known that reduced peptide
bonds could be made by reductive alkylation of an amino
acid employing a Fmoc-protected amino aldehyde.[53] Thus
Fmoc-protected phenylalaninal 11 was synthesised from the
corresponding Fmoc-protected amino acid 9 via the Weinreb
amide[54] 10. The Weinreb amide was reduced to the alde-
hyde 11 using lithium aluminumhydride, following the pro-
cedure of Wen and Crews.[55] However, in our hands the
near quantitative yields could only be reproduced by paying
special attention to the quenching of the reaction.[56] The re-
action mixture was poured onto a large volume of stirred
ice water, to quickly dilute the generated base and in this
way prevent cleavage of the Fmoc group, a problem also de-
scribed by other groups[57] following the Wen and Crews
procedure. The tert-butyl ester of leucine 12 was alkylated,
generating a sterically hindered secondary amine 13 possess-
ing two stereocenters. To avoid dialkylation, the reductive
alkylation was carried out by slow addition of the aldehyde
to a small excess of the amine, in the presence of sodium cy-
anoborohydride and acetic acid. The amine 13 was purified
by column chromatography, however, evaporation to com-
plete dryness of the fractions containing pure 13 was avoid-
ed, since this caused self-cleavage of the Fmoc protecting
group by the free secondary amine. Instead the secondary
amine of 13 was immediately protected using Fmoc-chloride
in the presence of diisopropylethylamine, followed by de-
protection of the tert-butyl ester using TFA. This afforded
building block 14 in good yield after column chromatogra-
phy.
Scheme 2. Synthesis of dipeptide building block 14 possessing a reduced
amide bond.
and capped with benzoic acid to give 18a in high purity, ac-
cording to HPLC.[52] Benzoic acid proved to be an appropri-
ate capping reagent, since it did not react with the secon-
dary amines of 18a.
In a more direct approach the reductive alkylation was
performed on solid support, as previously reported for the
synthesis of mouse melanocortin receptor agonists.[59] How-
ever, in that work racemisation of the aldehydes during the
For the solid-phase peptide synthesis of the diamine pep-
tide scaffold 18a (Scheme 3), the HMBA functionalised
PEGA1900 resin 1 was utilised, and again the first amino acid
in the sequence was phenylalanine. Building block 14 was
coupled to the free amine of resin 2 using TBTU activation,
and after 3 h a Kaiser test[58] showed complete reaction. The
subsequent Fmoc deprotection also liberated the secondary
amine of 15, however, as expected, due to its sterically hin-
dered nature, the secondary amine did not react in the fol-
lowing coupling reactions. Before the second incorporation
of building block 14 into the peptide sequence, a proline
was introduced to give 16. A proline between the two re-
duced bonds in the scaffold was expected to induce a turn
that together with the flexibility of the reduced bonds would
facilitate chelation of the metal. After coupling of the
second building block 14, the peptide 17 was deprotected
Scheme 3. Solid-phase synthesis of diamine 18a using building block 14.
a) Piperidine/DMF 2:8; b) 14, TBTU, NEM, DMF; c) Fmoc-Pro-OPfp,
DhbtOH, DMF; d) benzoic acid, TBTU, NEM, DMF.
4124
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2005, 11, 4121 – 4131

Phosphine Ligands
FULL PAPER
synthesis resulted in some formation of diastereomers,
which could only be separated after cleavage from the sup-
port. This is detrimental to the synthesis of the phosphine li-
gands, since they should remain bound to the solid support.
Thus the problem of racemisation was investigated.
It was already known from the literature that silica gel
chromatography caused racemisation of a-amino alde-
hydes.[60] However, for the synthesis of building block 14 the
aldehydes could be obtained in an optically pure form and
be used for further reaction without chromatography. The
other critical factor for racemisation was the actual reduc-
tive alkylation. It was found that prolonged reaction times,
sometimes needed to complete the reductive alkylations on
solid support, but not in solution, lead to increased racemi-
sation. If however the reaction time was restricted to 3 h,
only negligible (<5% by HPLC) racemisation was ob-
served, even when Fmoc-phenylalaninal, prone to racemisa-
tion, was used. To drive the reductive alkylation to comple-
tion within 3 h, the number of equivalents of Fmoc-amino
aldehyde was increased (typically from three to six equiva-
lents).
The synthesis of the two peptide scaffolds 18b and 18c
was carried out starting from resin 2 (Scheme 4). Fmoc-ala-
nine was coupled to the phenylalanine residue of 2 by
TBTU activation to give 19. Subsequent reductive alkylation
using aldehyde 11 afforded compound 20 possessing the first
of the two reduced amide bonds. The variation was intro-
duced before the next reductive alkylation. For series b and
c a proline and an alanine was incorporated, respectively, to
investigate the effect of the proline for catalyst stability and
selectivity. Coupling of another Fmoc-alanine to give 22b,c,
this time using the pentafluorophenyl (Pfp) ester with 1
equivalent of 3-hydroxy-3,4-dihydro-4-oxo-1,2,3-benzotria-
zine (DhbtOH), followed by reductive alkylation to give
23b,c, and capping with benzoic acid, afforded the two dia-
mine peptide scaffolds 18b,c in high purity. Only a single
Scheme 4. Synthesis of diamines 18b,c by reductive alkylation on solid
support. a) Piperidine/DMF 2:8; b) Fmoc-Ala-OPfp, DhbtOH, DMF;
c) 11, NaCNBH3, AcOH, DMF; d) Fmoc-Pro-OPfp (for 18b) or Fmoc-
Ala-OPfp (for 18c), DhbtOH, DMF; e) benzoic acid, TBTU, NEM,
DMF.
1
peak was observed in HPLC and from the H NMR spec-
trum it was clear that racemisation during the reductive al-
kylations had not been a problem with a reaction time less
than 3 h.[52]
Batches of resin bound diamines 18a–c were treated over-
night with hydroxymethyldiphenylphosphine, followed by
hydrogen peroxide oxidation to give the oxidised phosphine
ligands 25a–c (Scheme 5).[52] HPLC showed full and clean
conversion of the diamines 18a–c, in spite of their limited
reactivity in peptide couplings, due to steric hindrance.
Again 31P NMR spectra of batches of resin bound ligands
24a–c, recorded prior to oxidation, showed that no oxida-
tion of the phosphines took place during the synthesis, as
evidenced by only two resonances in the range ꢀ26 to
ꢀ29 ppm, typical for phosphines (Figure 2a). For compari-
son, the solution 31P NMR spectra of the oxidised ligands
25a–c showed resonances in the range from 32 ppm to
35 ppm (Figure 2b).
Scheme 5. Phosphinomethylation and oxidation of the diamines 18a–c on
solid support. a) Ph2PCH2OH, MeCN; b) 3% aqueous H2O2.
ty of the peptide-based phosphine ligands 7 and 24a–c to
form complexes with transition metals for asymmetric syn-
thesis, palladium(ii) complexes were synthesised on solid
Formation of resin bound palladium(ii) complexes and their
application in asymmetric synthesis: To investigate the abili-
Chem. Eur. J. 2005, 11, 4121 – 4131
www.chemeurj.org
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
4125

M. Meldal and C. A. Christensen
yellow to reddish-brown colours, and after 3 h the resins
were drained and washed, affording resin bound allyl palla-
dium(ii) complexes 26–27a–c.
In Figure 3 molecular dynamics modelling of complex 26
and complex 27c are presented. It is clear that the active
metal in type II catalyst 27c is much closer to and more em-
bedded in the chiral environment of the peptide backbone
than that of type I catalyst 26. However, it was not con-
firmed by spectroscopy that all the ligand molecules in the
resin form the complex represented by the model, but the
low loading and the excellent swelling properties of the
PEGA1900 resin favour 1:1 complexes rather than larger
cross linked aggregates of ligands and palladium ions.
To study the catalytic properties of the resin bound com-
plexes, the palladium catalysed asymmetric allylic substitu-
tion reaction[61] was performed. Not only is this a common
model reaction for evaluation of new ligands for palladium
catalysed asymmetric synthesis, also used to assay the pep-
tide-based phosphine catalysts synthesised by Gilbertson
and co-workers,[27,28,31] it has also found widespread use in
the synthesis of a vast array of bioactive targets.[62] Thus sup-
ported catalysts 26 and 27a–c were used in 5 mol% (relative
to the initial resin loading) for the reaction of 1,3-diphenyl-
propenyl acetate 28 with dimethylmalonate 29 under basic
conditions[63] (Scheme 7). After 3 h the reactions were
worked up and 30 isolated as a mixture of enantiomers. The
Figure 2. 31P NMR of a) resin bound ligand 24a in CDCl3 and b) cleaved
oxidised ligand 25a in [D6]DMSO. The spectra were referenced against
H3PO4.
support. Batches of the phosphine ligands were freshly pre-
pared and immediately after removal of excess hydroxyme-
thyldiphenylphosphine, a degassed solution of allyl palladi-
um chloride dimer (1 equivalent palladium) was added
(Scheme 6). Within minutes the resin batches attained
1
enantiomeric excess was determined by H NMR, using the
chiral shift reagent [Eu(hfc)3].[64] Isolated yields after
column chromatography, and ee values are listed in Table 1.
Type I catalyst 26 afforded 30 in excellent yield, which
could be expected due to its stable six-membered chelate.
The enantioselectivity was modest, affording the S enantio-
mer in 15% excess, which could be a consequence of the
metal being only partially embedded in a chiral environment
in the open type I structure. Type II catalysts 27a and 27b,
both possessing a proline residue between the phosphine
moieties, showed a disappointingly low selectivity, yielding
the S enantiomer in 9 and 8% ee, respectively. Furthermore,
the sterically hindered leucine containing catalyst 27a only
converted 22% of substrate 28 compared to a conversion of
73% for the less hindered 27b. Presumably this is due to
steric congestion around the metal, preventing a geometry
appropriate for the formation of a stable palladium(ii) com-
plex. The exchange of proline for alanine in 27c did not
seem to lower the stability compared to 27b, in fact 27c
converted the substrate quantitatively within 3 h. Further-
more, the selectivity was better, unexpectedly affording the
S enantiomer in 21% ee. Thus, prediction of the most selec-
tive catalyst was not possible, and even the relative stability
of the complexes 26 and 27a–c was not clear prior to synthe-
sis, in spite of computer modelling, which underlines the im-
portance of being able to synthesise the ligands in a combi-
natorial manner. Only screening of combinatorial catalyst li-
braries will allow for the selection of both stable and highly
selective catalysts. This is in agreement with the unpredicta-
ble observations made by Gilbertson and co-workers[28] for
their peptide-based phosphine palladium catalysts. Com-
Scheme 6. Formation of palladium(ii) allyl complexes on solid support.
a) [PdCl(h3-C3H5)]2, MeCN.
4126
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2005, 11, 4121 – 4131

Phosphine Ligands
FULL PAPER
the two most effective type II
catalysts 27b and 27c (entries 5
and 6). However, changing the
solvent did not affect the yields,
nor did it alter the enantioselec-
tivity significantly. Two control
experiments were also per-
formed.
Unsubstituted
PEGA1900 resin 31 was treated
with allyl palladium chloride
dimer and used in catalysis
under the same conditions as
for the catalysts 26–27a–c. As
expected no background reac-
tion was observed. Palladium(ii)
Figure 3. Molecular dynamics modelling of a) palladium(ii) complex 26 and b) palladium(ii) complex 27c, gen-
erated using the Insight-Discover package from MSI. The chloride counter ions have been omitted for clarity.
The structures were constructed and subjected to several rounds of MD-calculation at 650 and 350 K using
CVFF forcefield and 1 fs steps to finally yield a structure that was stable for 100000 steps.
complex
32,
made
from
PEGA1900 functionalised direct-
ly with phosphinomethyl moieties, was used for the second
control experiment. Again the stable six-membered chelate
afforded 30 in near quantitative yield, but this time as a rac-
emic mixture, due to lack of ligand chirality.
Conclusion
Scheme 7. Application of the complexes 26 and 27a–c in palladium cata-
lysed asymmetric allylic substitution.
This work paves the way for the synthesis of peptide-based
phosphine transition metal catalyst libraries. An efficient
and direct route from commercially available Fmoc-protect-
ed amino acids to elaborate peptide-based phosphine li-
gands has been developed. Solution-phase synthesis is limit-
ed to that of Fmoc-protected amino aldehydes needed for
formation of backbone secondary amines as attachment
points for phosphine moieties. In that context the choice of
different phosphinomethylation reagents, made from differ-
ent commercially available secondary phosphines, can afford
sublibraries of phosphine ligands from the same library of
ligand precursor scaffolds. Furthermore, protection/depro-
tection of phosphine moieties can be completely avoided.
The building block approach (Schemes 2 and 3) may be con-
sidered for large-scale synthesis of identified catalysts, par-
ticularly when bulky amino acids more prone to racemisa-
tion are used in the reductive alkylation.
Table 1. Palladium catalysed asymmetric allylic substitution.[a]
Entry Catalyst
Solvent Yield 30 [%][b] ee 30 [%][c]
1
2
3
4
5
6
26
THF
THF
THF
THF
MeCN
MeCN
93
22
73
97
95
90
15 (S)
9 (S)
8 (S)
21 (S)
10 (S)
18 (S)
27a
27b
27c
27b
27c
7
THF
0
-
8
THF
96
0
It was shown that ligands 7 and 24b,c could form palla-
dium(ii) allyl complexes suitable for the palladium catalysed
asymmetric allylic substitution reaction, whereas ligand 24a
seemed too sterically hindered. Although the enantioselec-
tivities obtained using substrate 28 were only moderate, the
concept of this new methodology was successfully proven. It
is expected that screening of catalyst libraries will afford
more selective catalysts. Furthermore, screening of other
substrates could reveal catalysts 26–27a–c to be more selec-
tive. New high-throughput screening methods are currently
being developed for the screening of single-bead split/mix
catalyst libraries, with the aim of discovering highly selective
catalysts for several different reactions.
[a] All reactions were performed at room temperature using 5 mol% Pd,
relative to the initial resin loading. [b] Isolated yield. [c] Determined by
1H NMR using a chiral shift reagent.
plexes based upon b-turn motifs catalysed the allylic substi-
tution of cyclic allyl acetates with high selectivity, but could
not catalyse substitution of linear substrates such as 28 at
all. Furthermore, in the desymmetrisation of meso diols the
selectivities of the b-turn based catalysts were inferior to
catalysts possessing no particular secondary structure, illus-
trating the need for a catalyst screening for every new sub-
strate or type of reaction considered.[29]
The effect of the solvent was investigated by performing
the catalytic allylic substitution of 28 in acetonitrile, using
Chem. Eur. J. 2005, 11, 4121 – 4131
www.chemeurj.org
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
4127

M. Meldal and C. A. Christensen
peroxide solution was added to cover the resin for 1 h. The resin was
washed with methanol (ꢃ2), N,N-dimethylformamide (ꢃ6), methanol
(ꢃ2) and with dichloromethane (ꢃ2), whereupon it was dried in vacuo.
General procedure for recording 31P NMR spectre of resin bound li-
gands: The freshly prepared resin bound ligand was washed with de-
gassed dichloromethane (ꢃ3) and with degassed 1% diisopropylethyla-
mine in CDCl3 (ꢃ1) under argon. The resin was swollen in a second
volume of degassed 1% diisopropylethylamine in CDCl3 and transferred
to a standard 5 mm NMR tube under argon, whereupon the spectrum
was recorded.
Experimental Section
General methods: ESMS spectra were recorded on a Micromass QTOF
Global Ultima instrument and high-resolution MS determined using an
appropriate internal reference. 1H NMR and 31P NMR spectra were re-
corded on a Bruker DRX250 250 MHz instrument. Chemical shifts for
1H spectra are reported in ppm relative to the internal solvent peak
(2.50 ppm for [D6]DMSO and 7.26 for CDCl3) and for 31P spectra refer-
enced against H3PO4. NMR spectra of reduced bond ligand precursors
18a–c were recorded under acidic conditions to allow full protonation.
Solid-phase reactions were performed in flat-bottomed polyethylene sy-
ringes equipped with sintered Teflon filters (50 mm pores), Teflon tubing,
Teflon valves for flow control, and suction to drain the syringes from
below. For solid-phase reactions carried out under argon, the syringes
were equipped with a rubber septum and an argon inlet. TLC plates used
were Merck silica gel 60 F254 on aluminium. Column chromatography was
performed on silica gel 60H (230–400 mesh). Analytical reverse-phase
HPLC was performed on a Waters system (490E detector at 215 nm, two
510 pumps with gradient controller and a Zorbax RP-18 column, 300 ꢂ,
0.45ꢃ50 mm) with a flow rate of 1 mLminꢀ1. Eluents A (0.1% TFA in
water) and B (0.1% TFA in acetonitrile/water 9:1 v/v) were used in a
linear gradient (0% B ! 100% B in 25 min) and retention times refer to
this solvent system. PEGA1900 was purchased from VersaMatrix. All sol-
vents were HPLC grade. Anhydrous solvents were obtained by storing
over 3 ꢂ activated molecular sieves. Degassed solutions were obtained
by bubbling with argon for 30 min. All other starting materials were pur-
chased from commercial suppliers and used without further purification.
Procedure for determining enantioselectivity: The chiral shift reagent
tris[3-(heptafluoropropylhydroxymethylene)-d-camphorato]europium(iii)
(0.4 equiv) was added to the NMR sample of 30 and the spectrum re-
corded. The two CO2Me singlets at about 4 ppm were now split into two
singlets for each of the two antipodes (see Supporting Information). One
CO2Me singlet was well-resolved into the R- and S-singlet, whereas the
other was seen as an unresolved doublet. Using the d-antipode of the
shift reagent, the singlet with the highest ppm value corresponds to the
R-enantiomer.[64]
[49]
Preparation of functionalised resin 1: PEGA1900
0.23 mmolgꢀ1
(2.00 g, loading
,
0.46 mmol) was washed with N,N-dimethylformamide
(ꢃ6), methanol (x 6) and dichloromethane (x 2) and dried in vacuo. Cou-
pling of Fmoc-Gly-OH using TBTU activation according to the general
procedure was performed twice, whereupon the Fmoc groups were re-
moved. The HMBA linker was attached by TBTU coupling.
Ligand precursor 6: Resin 1 (0.50 g, 0.11 mmol) was dried in vacuo in the
presence of phosphorous pentaoxide overnight. A solution of Fmoc-Phe-
OH (3 equiv) and methylimidazole (2.25 equiv) in dry dichloromethane
was added to MSNT (3 equiv), and the resulting solution was added to
the dry resin and reacted for 1 h. The resin was washed with dry dichloro-
methane (ꢃ3) and the MSNT coupling repeated to give 2. Resin 2 was
washed with dichloromethane (ꢃ6) and N,N-dimethylformamide (ꢃ6),
whereupon the Fmoc groups were removed. Fmoc-Phe-OH, Boc-Dap-
(Fmoc)-OH and Fmoc-Phe-OH were coupled successively to the resin by
using TBTU activation followed by Fmoc cleavage, to obtain resin bound
peptide 5. The Fmoc groups were removed and benzoic acid was coupled
by TBTU activation. The resin was washed with N,N-dimethylformamide
(ꢃ6) and dichloromethane (ꢃ6) and the Boc group was removed by
treating the resin with TFA-dichloromethane 1:1 (v/v) for 15 min, which
yielded ligand precursor 6. The resin was washed with dichloromethane
(ꢃ6), N,N-dimethylformamide (ꢃ6), methanol (ꢃ2) and dichloromethane
(ꢃ2), whereupon it was dried in vacuo overnight. A sample of 6 (100 mg
resin, 0.019 mmol) was cleaved for analysis, affording an amorphous col-
ourless solid (11 mg, 89%). Analytical HPLC: tR =14.7 min, >95%
purity; 1H NMR ([D6]DMSO): d=8.76 (d, 3J(H,H)=8 Hz, 1H; amide),
8.51 (brs, 1H; amide), 8.26 (d, 3J(H,H)=8 Hz, 1H; amide), 8.20 (d, 3J-
(H,H)=9 Hz, 1H; amide), 7.84 (d, 3J(H,H)=7 Hz, 2H; arom. H), 7.55–
7.12 (m, 18H; arom. H), 4.70 (brs, 3H; NH3+), 4.69–4.60 (m, 1H; CHa),
4.56–4.48 (m, 1H; CHa), 4.45–4.36 (m, 1H; CHa), 3.44–3.38 (m, 1H;
CHa), 3.22–2.76 (m, 8H; PhCH2, Dap CH2); HRMS (ES): m/z: calcd for
C37H40N5O6: 650.2973, found: 650.2947 [M+H]+.
General procedure for cleavage of resin supported peptides: Cleavage of
peptides was achieved with 0.1m aqueous NaOH for 2 h followed by neu-
tralisation with 0.1m aqueous HCl.
General procedures for peptide couplings: TBTU couplings were per-
formed by dissolving the acid (3 equiv) in N,N-dimethylformamide with
NEM (4 equiv), followed by addition of TBTU (2.9 equiv). The resulting
solution was left for preactivation for 5 min before being added to the
resin (reaction time 2–3 h). Coupling of Pfp esters were performed by
dissolving the Pfp ester (3 equiv) and DhbtOH (1 equiv) in N,N-dimeth-
ylformamide, whereupon the solution was added to the resin (reaction
time 3 h). Peptide couplings were generally run in an amount of solvent
just enough to cover the resin. After reaction, the resin was washed with
N,N-dimethylformamide (ꢃ6) and methanol (ꢃ2) and finally checked
using the Kaiser test.[58]
General procedure for Fmoc deprotection: Fmoc deprotection was ach-
ieved with 20% piperidine in N,N-dimethylformamide (v/v) for 2
+
18 min, followed by washing of the resin with N,N-dimethylformamide
(ꢃ6) and methanol (ꢃ2). The cleavage was checked using the Kaiser test.[58]
General procedure for reductive alkylation: Two solutions of the same
volume were prepared, so that the total volume was just enough to cover
the resin. Solution 1: Sodium cyanoborohydride (10 equiv) and 2% (v/v)
glacial acetic acid were dissolved in N,N-dimethylformamide. Solution 2:
Fmoc-phenylalaninal (3–6 equiv) were dissolved in N,N-dimethylforma-
mide. Solution 1 was added to the resin bound free amine, swollen in
N,N-dimethylformamide. The resin was stirred for 1 min, whereupon so-
lution 2 was added whilst stirring. The resin was left to react for 3 h,
drained, and washed with N,N-dimethylformamide (ꢃ6) and methanol
(ꢃ2). The reaction was checked using the Kaiser test.[58]
Phosphine ligand 7: Phosphinomethylation of resin 6 was carried out ac-
cording to the general procedure, to give resin bound phosphine ligand 7.
31P NMR (CDCl3): d=ꢀ29.7.
Oxidised phosphine ligand 8: Freshly prepared resin 7 was oxidised fol-
lowing the general procedure to give the resin bound oxidised ligand 8.
Oxidised ligand 8 (100 mg resin, 0.018 mmol) was cleaved, affording an
amorphous colourless solid (17 mg, 88%). Analytical HPLC: tR =
19.5 min, >95% purity; 1H NMR ([D6]DMSO): d=8.61 (d, 3J(H,H)=
9 Hz, 1H; amide), 8.50 (d, 3J(H,H)=9 Hz, 1H; amide), 8.42 (d, 3J-
(H,H)=8 Hz, 1H; amide), 8.07 (brs, 1H; amide), 7.84 (d, 3J(H,H)=
7 Hz, 2H; arom. H), 7.65–6.86 (m, 38H; arom. H), 4.82–4.73 (m, 1H;
CHa), 4.53–4.40 (m, 2H; CHa), 3.90–3.71 (m, 3H; CHa, CH2P), 3.49–3.19
(m, 2H; Dap CH2), 3.12–2.62 (m, 8H; CH2P, PhCH2); 31P NMR
([D6]DMSO): d=32.8; HRMS (ES): m/z: calcd for C63H62N5O8P2:
1078.4068, found: 1078.4037 [M+H]+.
General procedure for phosphinomethylation: Neat paraformaldehyde
(1 equiv) and diphenylphosphine (1 equiv) were heated at 1108C for
1.5 h under argon, affording hydroxymethyldiphenylphosphine. The resin
was dried in vacuo overnight and flushed with argon. To the resin was
added a 0.15m solution of hydroxymethyldiphenylphosphine in degassed
acetonitrile (two times the volume needed to swell the resin, ca.
20 equiv). The resin was left to react under argon for 12 h at room tem-
perature, drained and washed with degassed acetonitrile (ꢃ3) under
argon.
General procedure for preparation of oxidised phosphine ligands: The
freshly prepared resin bound phosphine ligand was washed with a 3%
aqueous solution of hydrogen peroxide (ꢃ1), whereupon the hydrogen
Fmoc-Phe-y[CH2N]-Leu-OtBu (13): The hydrochloride salt of the tert-
butylester of leucine 12 (839 mg, 3.75 mmol) and sodium cyanoborohy-
4128
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2005, 11, 4121 – 4131

Phosphine Ligands
FULL PAPER
dride (471 mg, 7.50 mmol) were dissolved in dry N,N-dimethylformamide
(10 mL) and glacial acetic acid (200 mL) was added. The reaction mixture
was stirred at room temperature whilst a solution of Fmoc-phenylalani-
nal[55,56] 11 (929 mg, 2.50 mmol) in N,N-dimethylformamide (10 mL) was
added drop-wise over 1 h. Stirring was continued for 2 h, whereupon the
reaction mixture was poured onto water. The mixture was extracted with
dichloromethane (ꢃ3) and the combined organic phases washed with
water (ꢃ2) and brine (ꢃ1), dried (MgSO4) and concentrated to a pale
yellow oil. The oil was purified by column chromatography (silica gel,
acetone/dichloromethane 5:95) and the fractions containing pure 13 were
combined and concentrated to a viscous colourless oil, but not to com-
plete dryness. Since loss of Fmoc by self-cleavage was observed upon
concentration and storage, 13 was not characterised any further, but used
immediately in the next reaction.
0.019 mmol) was cleaved from the resin to give an amorphous colourless
solid (12 mg, 82%). Analytical HPLC: tR =13.6 min, >95% purity;
1H NMR ([D6]DMSO): d=9.15 (brs, 2H; R2NH2+), 8.98 (brs, 2H;
R2NH2+), 8.87 (d, 3J(H,H)=8 Hz, 1H; amide), 8.43 (d, 3J(H,H)=8 Hz,
1H; amide), 8.30 (d, 3J(H,H)=8 Hz, 1H; amide), 7.78 (d, 3J(H,H)=
8 Hz, 2H; arom. H), 7.55–7.41 (m, 3H; arom. H), 7.32–7.15 (m, 15H;
arom. H), 4.57–4.46 (m, 2H; CHa), 4.33–4.05 (m, 3H; CHa), 3.91–3.77
(m, 1H; CHa), 3.66–2.64 (m, 12H; PhCH2, NCH2Ca, Pro CH2N), 2.04–
1.76 (m, 4H; Pro CaCH2CH2), 1.40 (d, 3J(H,H)=7 Hz, 3H; Ala CH3),
1.36 (d, 3J(H,H)=7 Hz, 3H; Ala CH3); HRMS (ES): m/z: calcd for
C45H55N6O6: 775.4178, found: 775.4182 [M+H]+.
Ligand precursor 18c: The synthesis of 18c from resin
1 (0.50 g,
0.11 mmol) was similar to the synthesis of 18b, except that resin 20 was
Fmoc deprotected and reacted with Fmoc-Ala-OPfp to give 21c, instead
of reaction with Fmoc-Pro-OPfp to give series b. A sample of 18c
(100 mg resin, 0.019 mmol) was cleaved from the resin, affording an
amorphous colourless solid (13 mg, 91%). Analytical HPLC: tR =
13.4 min, >95% purity; 1H NMR ([D6]DMSO): d=8.94 (brs, 4H;
R2NH2+), 8.87 (d, 3J(H,H)=8 Hz, 1H; amide), 8.66 (d, 3J(H,H)=8 Hz,
1H; amide), 8.42 (d, 3J(H,H)=8 Hz, 1H; amide), 8.09 (d, 3J(H,H)=
8 Hz, 1H; amide), 7.80 (d, 3J(H,H)=8 Hz, 2H; arom. H), 7.55–7.41 (m,
3H; arom. H), 7.31–7.14 (m, 15H; arom. H), 4.57–4.47 (m, 2H; CHa),
4.25–4.07 (m, 2H; CHa), 3.94–3.80 (m, 2H; CHa), 3.17–2.63 (m, 10H;
PhCH2, NCH2), 1.37 (d, 3J(H,H)=7 Hz, 3H; Ala CH3), 1.33 (d, 3J-
(H,H)=7 Hz, 3H; Ala CH3), 1.17 (d, 3J(H,H)=7 Hz, 3H; Ala CH3);
HRMS (ES): m/z: calcd for C43H53N6O6: 749.4021, found: 749.4069
[M+H]+.
Building block 14: Amine 13 (from 929 mg, 2.50 mmol 11) was dissolved
in dichloromethane (20 mL) and diisopropylethylamine (642 mL,
3.75 mmol) was added. The solution was cooled to 08C and stirred under
argon. Fmoc-chloride (745 mg, 2.88 mmol) was added as a solid in one
portion and the reaction mixture was left to stir for 2 h, whereupon it
was allowed to attain room temperature. The reaction mixture was
poured onto water and extracted with dichloromethane (ꢃ3). The com-
bined organic phases were washed with water (ꢃ3), dried (MgSO4) and
concentrated. The resulting oil was redissolved in a 1:1 (v/v) mixture of
TFA and dichloromethane (10 mL) and left to stir at room temperature
for 12 h. Toluene (10 mL) was added and the reaction was concentrated
to a viscous oil in vacuo. Purification by column chromatography (silica
gel, methanol/dichloromethane 5:95) afforded 14 as an amorphous col-
ourless solid (1.38 g, 78% from 11). Analytical HPLC: tR =23.0 min,
>95% purity; 1H NMR ([D6]DMSO): d=7.85 (d, 3J(H,H)=7 Hz, 4H;
arom. H), 7.63–7.55 (m, 4H; arom. H), 7.38–7.17 (m, 14H; arom. H,
NH), 4.41–4.10 (m, 7H; Fmoc CH2CH, CHa), 3.89–3.85 (m, 1H; CHa),
3.43–3.06 (m, 2H; NCH2), 2.87–2.59 (m, 2H; PhCH2), 1.78–1.46 (m, 3H;
Leu CHCH2), 0.82 (brs, 6H; Leu CH3); 13C NMR ([D6]DMSO): d=
173.9, 156.8, 155.8, 155.5, 143.71, 143.68, 140.67, 140.57, 139.0, 138.8,
129.0, 127.9, 127.5, 126.9, 125.8, 125.1, 124.9, 120.0, 66.7, 66.5, 65.2, 59.3,
58.5, 52.3, 52.1, 50.1, 49.4, 46.65, 46.58, 38.2, 37.7, 24.7, 24.6, 22.9, 22.7,
21.73, 21.67;[65] HRMS (ES): m/z: calcd for C45H45N2O6: 709.3272, found:
709.3253 [M+H]+.
Phosphine ligands 24a–c: Phosphinomethylation of resins 18a–c was car-
ried out according to the general procedure, to give resin bound phos-
phine ligands 24a–c.
Compound 24a: 31P NMR (CDCl3): d=ꢀ27.2, ꢀ28.7.
Compound 24b: 31P NMR (CDCl3): d=ꢀ27.2, ꢀ28.9.
Compound 24c: 31P NMR (CDCl3): d=ꢀ26.8, ꢀ28.2.
Oxidised phosphine ligands 25a–c: Freshly prepared resins 24a–c were
oxidised following the general procedure to give the resin bound oxidised
ligands 25a–c. The oxidised ligands 25a–c (100 mg resin, 0.017 mmol)
were cleaved to give amorphous colourless solids.
Ligand precursor 18a: Fmoc-Phe-OH was attached to the HMBA linker
of resin 1 (0.50 g, 0.11 mmol) to give resin 2, in a MSNT coupling per-
formed as described in the procedure for the synthesis of 6. The Fmoc
groups of 2 were removed and the resin reacted with building block 14
using TBTU activation, to give 15. The Fmoc groups of 15 were removed
before elongation by reaction with Fmoc-Pro-OPfp affording 16. Remov-
al of Fmoc groups and successive TBTU couplings of building block 14
Compound 25a: (19 mg, 87%). Analytical HPLC: tR =22.5 min, >95%
purity; 1H NMR ([D6]DMSO): d=9.25 (d, 3J(H,H)=5 Hz, 1H; amide),
8.03 (brs, 1H; amide), 7.94–7.71 (m, 6H; amide, arom. H), 7.66–6.91 (m,
35H; arom. H), 4.54–4.43 (m, 1H; CHa), 4.21–4.08 (m, 1H; CHa), 3.95–
3.83 (m, 1H; CHa), 3.70–2.33 (m, 18H; CHa, PhCH2, PCH2NCH2, Pro
CH2N), 1.93–1.69 (m, 2H; Pro CH2), 1.38–0.25 (m, 9H; Pro CH2, Leu
CaCH2CH), 0.65 (d, 3J(H,H)=6 Hz, 3H; Leu CH3), 0.54 (d, 3J(H,H)=
6 Hz, 3H; Leu CH3), 0.52 (d, 3J(H,H)=6 Hz, 3H; Leu CH3), 0.42 (d, 3J-
(H,H)=6 Hz, 3H; Leu CH3); 31P NMR ([D6]DMSO): d=35.0, 33.8;
HRMS (ES): m/z: calcd for C77H90N6O8P2: 644.3142, found: 644.3143
and benzoic acid afforded ligand precursor 18a.
A sample of 18a
(100 mg resin, 0.019 mmol) was cleaved for analysis, yielding an amor-
phous colourless solid (14 mg, 86%). Analytical HPLC: tR =15.6 min,
>95% purity; 1H NMR ([D6]DMSO): d=9.03 (brs, 4H; R2NH2+), 9.01
(d, 3J(H,H)=8 Hz, 1H; amide), 8.46 (d, 3J(H,H)=8 Hz, 1H; amide),
8.23 (d, 3J(H,H)=7 Hz, 1H; amide), 7.77 (d, 3J(H,H)=7 Hz, 2H; arom.
H), 7.55–7.41 (m, 3H; arom. H), 7.30–7.16 (m, 15H; arom. H), 4.63–4.46
(m, 2H; CHa), 4.35–4.19 (m, 2H; CHa), 4.10–3.99 (m, 1H; CHa), 3.83–
3.64 (m, 2H; CHa, Pro CH2N), 3.42–2.54 (m, 11H; PhCH2, NCH2Ca, Pro
CH2N), 2.06–1.49 (m, 10H; Pro CaCH2CH2, Leu CaCH2CH), 0.94–0.81
(m, 12H; Leu CH3); HRMS (ES): m/z: calcd for C51H67N6O6: 859.5117,
found: 859.5119 [M+H]+.
[M+2H]2+
.
Compound 25b: (17 mg, 83%). Analytical HPLC: tR =19.7 min, >95%
purity; 1H NMR ([D6]DMSO): d=9.18 (d, 3J(H,H)=6 Hz, 1H; amide),
8.09 (d, 3J(H,H)=8 Hz, 1H; amide), 7.92–6.99 (m, 41H; arom. H,
amide), 4.54–4.45 (m, 1H; CHa), 4.24–4.11 (m, 1H; CHa), 3.93–3.80 (m,
1H; CHa), 3.67–2.13 (m, 19H; CHa, PhCH2, PCH2NCH2, Pro CH2N),
1.08–0.44 (m, 4H; Pro CaCH2CH2), 0.87 (d, 3J(H,H)=6 Hz, 3H; Ala
CH3), 0.81 (d, 3J(H,H)=6 Hz, 3H; Ala CH3); 31P NMR ([D6]DMSO):
d=34.9, 33.2; HRMS (ES): m/z: calcd for C71H77N6O8P2: 1203.5278,
found: 1203.5186 [M+H]+.
Ligand precursor 18b: Fmoc-Phe-OH was attached to the HMBA linker
of resin 1 (0.50 g, 0.11 mmol) to give resin 2, in a MSNT coupling per-
formed as described in the procedure for the synthesis of 6. The Fmoc
groups of 2 were removed and the resin reacted with Fmoc-Ala-OPfp to
give 19. Cleavage of the Fmoc groups followed by reductive alkylation
using Fmoc-phenylalaninal[55, 56] 11 (6 equiv) afforded 20. Removal of the
Fmoc groups of 20 and coupling with Fmoc-Pro-OPfp and Fmoc-Ala-
OPfp successively, afforded 22b. The Fmoc groups of 22b were removed
and reductive alkylation using Fmoc-phenylalaninal[55,56] 11 (3 equiv) af-
forded 23b. Cleavage of the Fmoc groups of 23b followed by TBTU cou-
pling of benzoic acid, afforded 18b. A sample of 18b (100 mg resin,
Compound 25c: (16 mg, 80%). Analytical HPLC: tR =19.7 min, >95%
purity; 1H NMR ([D6]DMSO): d=8.49 (d, 3J(H,H)=8 Hz, 1H; amide),
8.14 (d, 3J(H,H)=8 Hz, 1H; amide), 7.86–7.02 (m, 42H; arom. H,
amide), 4.57–4.48 (m, 1H; CHa), 4.34–4.26 (m, 1H; CHa), 3.99–3.83 (m,
2H; CHa), 3.74–2.18 (m, 16H; CHa, PhCH2, PCH2NCH2), 0.98 (d, 3J-
(H,H)=7 Hz, 3H; Ala CH3), 0.83 (d, 3J(H,H)=7 Hz, 3H; Ala CH3),
0.75 (d, 3J(H,H)=7 Hz, 3H; Ala CH3); 31P NMR ([D6]DMSO): d=33.9,
33.2; HRMS (ES): m/z: calcd for C69H75N6O8P2: 1177.5122, found:
1177.5029 [M+H]+.
Chem. Eur. J. 2005, 11, 4121 – 4131
www.chemeurj.org
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
4129

M. Meldal and C. A. Christensen
General procedure for formation of resin bound palladium(ii) allyl com-
plexes: The freshly prepared resin bound phosphine ligand, still under
argon and swollen in acetonitrile, was washed further with dry degassed
acetonitrile (ꢃ3). To the drained resin was added a solution of allyl palla-
dium chloride dimer (1 equiv) in dry degassed acetonitrile, whereupon it
was left under argon for 3 h, in which time the resin attained a yellow
(for 26) or reddish-brown (for 27a–c) colour. The resin was drained,
washed with dry degassed acetonitrile (ꢃ3) and dried in vacuo overnight.
The dry resin was used immediately for palladium catalysed asymmetric
allylic substitution.
[9] C. A. Luchaco-Cullis, H. Mizutani, K. E. Murphy, A. H. Hoveyda,
Angew. Chem. 2001, 113, 1504–1508; Angew. Chem. Int. Ed. 2001,
40, 1456–1460.
[10] J. F. Traverse, A. H. Hoveyda, M. L. Snapper, Org. Lett. 2003, 5,
3273–3275.
[11] L. C. Akullian, M. L. Snapper, A. H. Hoveyda, Angew. Chem. 2003,
115, 4376–4379; Angew. Chem. Int. Ed. 2003, 42, 4244–4247.
[12] D. M. Mampreian, A. H. Hoveyda, Org. Lett. 2004, 6, 2829–2832.
[13] M. S. Sigman, E. N. Jacobsen, J. Am. Chem. Soc. 1998, 120, 4901–
4902.
[14] M. S. Sigman, P. Vachal, E. N. Jacobsen, Angew. Chem. 2000, 112,
1336–1338; Angew. Chem. Int. Ed. 2000, 39, 1279–1281.
[15] P. Vachal, E. N. Jacobsen, J. Am. Chem. Soc. 2002, 124, 10012–
10014.
[16] A. H. Hoveyda, Chem. Biol. 1999, 6, R305–R308.
[17] M. T. Reetz, Angew. Chem. 2001, 113, 292–320; Angew. Chem. Int.
Ed. 2001, 40, 284–310.
[18] J. F. Traverse, M. L. Snapper, Drug Discovery Today 2002, 7, 1002–
1012.
[19] C. Gennari, U. Piarulli, Chem. Rev. 2003, 103, 3071–3100.
[20] C. W. Edwards, M. R. Shipton, M. Wills, Tetrahedron Lett. 2000, 41,
8615–8619.
[21] C. R. Landis, T. P. Clark, Proc. Natl. Acad. Sci. USA 2004, 101,
5428–5432.
[22] S. R. Gilbertson, G. Chen, M. McLoughlin, J. Am. Chem. Soc. 1994,
116, 4481–4482.
[23] S. R. Gilbertson, X. Wang, J. Org. Chem. 1996, 61, 434–435.
[24] S. R. Gilbertson, X. Wang, G. S. Hoge, C. A. Klug, J. Schaefer, Orga-
nometallics 1996, 15, 4678–4680.
General procedure for palladium catalysed asymmetric allylic substitu-
tion—Synthesis of 30: Dimethyl malonate 29 (3 equiv) was dissolved in
dry degassed tetrahydrofuran (4.0 mLmmolꢀ1) under argon. To the solu-
tion was added tetrabutylammonium fluoride (3 equiv of a 1.0m solution
in tetrahydrofuran) and N,O-bis(trimethylsilyl)acetamide (3 equiv)
whereupon the solution was stirred for 15 min at room temperature. To
the freshly prepared dry catalyst resin 26 or 27a–c (5 mol% Pd, relative
to the initial resin loading), now under argon, was added 1,3-diphenylpro-
penyl acetate 28 (1 equiv, typically 38 mg, 0.15 mmol) and then the malo-
nate solution. The resin mixture was shaken under argon for 3 h, where-
upon it was drained and washed with ethyl acetate (ꢃ3). The combined
organic phases were concentrated in vacuo and purified by column chro-
matography (silica gel, pentane/diethyl ether 3:1) affording compound 30
as a colourless oil which solidified upon standing. 1H NMR (CDCl3): d=
7.17–7.33 (m, 10H; Ph), 6.46 (d, 3J(H,H)=16 Hz, 1H; C1 CH), 6.31 (dd,
3J(H,H)=16 Hz, 3J(H,H)=8 Hz, 1H; C2 CH), 4.25 (dd, 3J(H,H)=11 Hz,
3J(H,H)=8 Hz, 1H; C3 CH), 3.93 (d, 3J(H,H)=11 Hz, 1H; C4 CH), 3.69
(s, 3H; CO2CH3), 3.50 (s, 3H; CO2CH3). The 1H NMR spectrum was
identical to that described in the literature.[64]
[25] S. R. Gilbertson, X. Wang, Tetrahedron Lett. 1996, 37, 6475–6478.
[26] S. R. Gilbertson, X. Wang, Tetrahedron 1999, 55, 11609–11618.
[27] S. R. Gilbertson, S. E. Collibee, A. Agarkov, J. Am. Chem. Soc.
2000, 122, 6522–6523.
[28] A. Agarkov, S. J. Greenfield, T. Ohishi, S. E. Collibee, S. R. Gilbert-
son, J. Org. Chem. 2004, 69, 8077–8085.
[29] A. Agarkov, E. W. Uffman, S. R. Gilbertson, Org. Lett. 2003, 5,
2091–2094.
[30] S. J. Greenfield, S. R. Gilbertson, Synthesis 2001, 2337–2340.
[31] S. J. Greenfield, A. Agarkov, S. R. Gilbertson, Org. Lett. 2003, 5,
3069–3072.
[32] H. Hellmann, J. Bader, H. Birkner, O. Schumacher, Liebigs Ann.
Chem. 1962, 659, 49–63.
[33] S. E. Durran, M. R. J. Elsegood, N. Hawkins, M. B. Smith, S. Talib,
Tetrahedron Lett. 2003, 44, 5255–5257.
[34] M. Slany, A.-M. Caminade, J.-P. Majoral, Tetrahedron Lett. 1996, 37,
9053–9056.
[35] M.-L. Lartigue, A.-M. Caminade, J.-P. Majoral, Tetrahedron: Asym-
metry 1997, 8, 2697–2708.
For catalyst 26: Using 28 (43 mg, 0.17 mmol) and 26 (0.0085 mmol Pd),
compound 30 was isolated in 93% yield, 51 mg, and 15% ee (S).
For catalyst 27a: Using 28 (43 mg, 0.17 mmol) and 27a (0.0085 mmol
Pd), compound 30 was isolated in 22% yield, 12 mg, and 9% ee (S).
For catalyst 27b: Using 28 (43 mg, 0.17 mmol) and 27b (0.0085 mmol
Pd), compound 30 was isolated in 73% yield, 40 mg, and 8% ee (S).
For catalyst 27c: Using 28 (48 mg, 0.19 mmol) and 27c (0.0076 mmol Pd),
compound 30 was isolated in 97% yield, 60 mg, and 21% ee (S).
For catalyst 27b (acetonitrile as solvent): Using 28 (33 mg, 0.13 mmol)
and 27b (0.0065 mmol Pd), compound 30 was isolated in 95% yield,
40 mg, and 10% ee (S).
For catalyst 27c (acetonitrile as solvent): Using 28 (33 mg, 0.13 mmol)
and 27c (0.0065 mmol Pd), compound 30 was isolated in 90% yield,
38 mg, and 18% ee (S).
For catalyst 32: Using 28 (44 mg, 0.175 mmol) and 32 (0.0088 mmol Pd),
compound 30 was isolated in 96% yield, 55 mg, and 0% ee.
[36] V. Maraval, R. Laurent, B. Donnadieu, M. Mauzac, A.-M. Cami-
nade, J.-P. Majoral, J. Am. Chem. Soc. 2000, 122, 2499–2511.
[37] M. T. Reetz, G. Lohmer, R. Schwickardi, Angew. Chem. 1997, 109,
1559–1562; Angew. Chem. Int. Ed. Engl. 1997, 36, 1526–1529.
[38] G. Mꢄrkl, G. Yu Jin, Tetrahedron Lett. 1981, 22, 223–226.
[39] M. B. Smith, M. R. J. Elsegood, Tetrahedron Lett. 2002, 43, 1299–
1301.
Acknowledgement
This research was supported by the Danish National Research Founda-
tion.
[40] S. E. Durran, M. B. Smith, A. M. Z. Slawin, J. W. Steed, J. Chem.
Soc. Dalton Trans. 2000, 2771–2778.
[41] M. L. Clarke, A. M. Z. Slawin, M. V. Wheatley, J. D. Woollins, J.
Chem. Soc. Dalton Trans. 2001, 3421–3429.
[1] E. R. Jarvo, S. J. Miller, Tetrahedron 2002, 58, 2481–2495.
[2] B. R. Sculimbrene, A. J. Morgan, S. J. Miller, Chem. Commun. 2003,
1781–1785.
[42] S. C. Bourque, F. Maltais, W.-J. Xiao, O. Tardif, H. Alper, P. Arya,
L. E. Manzer, J. Am. Chem. Soc. 1999, 121, 3035–3038.
[43] S. C. Bourque, H. Alper, L. E. Manzer, P. Arya, J. Am. Chem. Soc.
2000, 122, 956–957.
[44] P. Arya, N. V. Rao, J. Singkhonrat, H. Alper, S. C. Bourque, L. E.
Manzer, J. Org. Chem. 2000, 65, 1881–1885.
[3] D. J. Guerin, S. J. Miller, J. Am. Chem. Soc. 2002, 124, 2134–2136.
[4] K. S. Griswold, S. J. Miller, Tetrahedron 2003, 59, 8869–8875.
[5] J. R. Porter, W. G. Wirschun, K. W. Kuntz, M. L. Snapper, A. H.
Hoveyda, J. Am. Chem. Soc. 2000, 122, 2657–2658.
[6] N. S. Josephsohn, K. W. Kuntz, M. L. Snapper, A. H. Hoveyda, J.
Am. Chem. Soc. 2001, 123, 11594–11599.
[45] P. Arya, G. Panda, N. V. Rao, H. Alper, S. C. Bourque, L. E.
Manzer, J. Am. Chem. Soc. 2001, 123, 2889–2890.
[46] S. Antebi, P. Arya, L. E. Manzer, H. Alper, J. Org. Chem. 2002, 67,
6623–6631.
[7] S. J. Degrado, H. Mizutani, A. H. Hoveyda, J. Am. Chem. Soc. 2001,
123, 755–756.
[8] J. R. Porter, J. F. Traverse, A. H. Hoveyda, M. L. Snapper, J. Am.
Chem. Soc. 2001, 123, 984–985.
4130
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2005, 11, 4121 – 4131

Phosphine Ligands
FULL PAPER
[47] Y. Uozumi, Y. Nakai, Org. Lett. 2002, 4, 2997–3000.
[48] E. Gonthier, R. Breinbauer, Synlett 2003, 1049–1051.
[49] M. Meldal, Tetrahedron Lett. 1992, 33, 3077–3080.
[50] M. Willert, J. M. Benito, M. Meldal, J. Comb. Chem. 2003, 5, 91–
101.
[59] J. Bondebjerg, Z. Xiang, R. M. Bauzo, C. Haskell-Luevano, M.
Meldal, J. Am. Chem. Soc. 2002, 124, 11046–11055.
[60] W. D. Lubell, H. Rapoport, J. Am. Chem. Soc. 1987, 109, 236–239.
[61] B. M. Trost, D. L. Van Vranken, Chem. Rev. 1996, 96, 395–422.
[62] B. M. Trost, J. Org. Chem. 2004, 69, 5813–5837.
[51] R. Sheppard, J. Pept. Sci. 2003, 9, 545–552.
[63] The base-system used was a mixture of N,O-bis(trimethylsilyl)aceta-
mide (BSA) and tetrabutylammonium fluoride (TBAF), conditions
developed by Gilbertson and co-workers, see: S. R. Gilbertson, C.-
W. T. Chang, J. Org. Chem. 1998, 63, 8424–8431.
[52] See Supporting Information for HPLC and 1H NMR data of all
ligand precursor peptide scaffolds and oxidised peptide-based phos-
phine ligands.
[53] J.-P. Salvi, N. Walchshofer, J. Paris, Tetrahedron Lett. 1994, 35, 1181–
1184.
[64] G. Brenchley, M. Fedouloff, M. F. Mahon, K. C. Molloy, M. Wills,
Tetrahedron 1995, 51, 10581–10592.
[54] S. Nahm, S. M. Weinreb, Tetrahedron Lett. 1981, 22, 3815–3818.
[55] J. J. Wen, C. M. Crews, Tetrahedron: Asymmetry 1998, 9, 1855–1858.
[56] F. Diness, J. Beyer, M. Meldal, QSAR Comb. Sci. 2004, 23, 130–144.
[57] F. Guillaumie, J. C. Kappel, N. M. Kelly, G. Barany, K. J. Jensen, Tet-
rahedron Lett. 2000, 41, 6131–6135.
[65] At the NMR time-scale a plurality of conformers of building block
14 were present, resulting in broadening and splitting of peaks, due
to the very bulky substituents.
[58] E. Kaiser, R. L. Colescott, C. D. Bossinger, P. I. Cook, Anal. Bio-
chem. 1970, 34, 595–598.
Received: January 28, 2005
Published online: April 28, 2005
Chem. Eur. J. 2005, 11, 4121 – 4131
www.chemeurj.org
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
4131