Understanding protease catalysed solid phase peptide synthesis

Article abstract of DOI:10.1039/b211890d

A protease (thermolysin) was used to directly synthesise a number of dipeptides from soluble Fmoc-amino acids onto a solid support (PEGA₁₉₀₀) in bulk aqueous media, often in very good yields. This shift in equilibrium toward synthesis is remarkable because for soluble dipeptides in aqueous solution hydrolysis rather than synthesis is observed. Three possible reasons for the equilibrium shift were considered: (i) using a solid support makes it easy to use an excess of reagents, so mass action contributes towards synthesis; (ii) reduction in the unfavourable hydrophobic hydration of the Fmoc group within the solid support compared with the free amino acid in solution and (iii) suppression of the ionization of amino groups linked to the solid phase due to mutual electrostatic repulsion. It was found that under the conditions studied the second effect was most important.

Full text of DOI:10.1039/b211890d

View Article Online / Journal Homepage / Table of Contents for this issue
Understanding protease catalysed solid phase peptide synthesis
Rein V. Ulijn,a Nicola Bisek,a Peter J. Hallingb and Sabine L. Flitsch*a
a Department of Chemistry, The University of Edinburgh, King’s Buildings, West Mains Road,
Edinburgh, Scotland UK EH9 3JJ; Fax: ϩ44(0)131 650 4737; Tel: ϩ44(0)131 650 4737
b Department of Chemistry, The University of Strathclyde, Thomas Graham Building,
Cathedral Street, Glasgow, Scotland UK G1 1XL; Fax: ϩ44(0)141 552 4822;
Tel: ϩ44(0)141 548 2683
Received 5th December 2002, Accepted 4th March 2003
First published as an Advance Article on the web 19th March 2003
A protease (thermolysin) was used to directly synthesise a number of dipeptides from soluble Fmoc-amino acids
onto a solid support (PEGA1900) in bulk aqueous media, often in very good yields. This shift in equilibrium toward
synthesis is remarkable because for soluble dipeptides in aqueous solution hydrolysis rather than synthesis is
observed. Three possible reasons for the equilibrium shift were considered: (i) using a solid support makes it easy
to use an excess of reagents, so mass action contributes towards synthesis; (ii) reduction in the unfavourable
hydrophobic hydration of the Fmoc group within the solid support compared with the free amino acid in solution
and (iii) suppression of the ionization of amino groups linked to the solid phase due to mutual electrostatic repulsion.
It was found that under the conditions studied the second effect was most important.
Introduction
Applications of enzyme catalysis on substrates that are
immobilised onto insoluble supports are on the increase.
Examples are chemo-enzymatic synthesis on solid supports,1
‘on-bead’ screening for enzyme substrates or inhibitors in com-
binatorial libraries,2 enzyme cleavable linkers for release of
polymer bound libraries3 and enzyme cleavable protecting
groups in solid phase chemistry.4 In the future, increased appli-
cations are expected in the area of small molecule micro-
arrays,5 and solid-phase combinatorial biocatalysis.6
It may seem rather surprising that the fundamentals of
enzyme reactions on immobilised substrates (kinetics, thermo-
dynamics) have not been studied in detail to date, even though
one might predict that these could be significantly different
from those of reactions in dilute aqueous solutions. Thus, we
have recently reported an example where reaction thermo-
dynamics were altered when the enzymatic peptide synthesis
was carried out on solid PEGA1900 support (poly(acrylamide)–
Scheme 1 Phenylalanine was coupled via the Wang linker to PEGA1900
and treated with excess N-protected amino acid in the presence of
ethyleneglycol). We found that the amide synthesis/hydrolysis
equilibrium was significantly shifted toward synthesis when
compared to the reaction in dilute aqueous solution where the
hydrolytic reaction predominates.7 This shift from preferred
hydrolysis with dissolved amine substrate toward preferred syn-
thesis with immobilised amine substrate allowed for efficient
protease catalysed synthesis of a variety of peptides on solid
support (Scheme 1).
These findings prompted us to investigate the reasons for the
shift in equilibrium towards peptide synthesis when the amine
substrate is immobilised. We considered three possible factors
that could play a role.7 These were i) the effect of using an
excess of soluble amino acid in solid phase synthesis. Secondly
(ii), the synthesis reaction involves transfer of the protected
amino acid 1 from aqueous solution into the polyethylene
glycol micro-environment of the solid phase resin, where
unfavourable hydrophobic hydration is expected to be reduced.
This should increase the synthesis yield of solid phase dipeptide
3, especially when hydrophobic amino acid substrates are used.
And finally (iii), ionisation of PEGA1900 immobilised amino
groups (in this case the phenylalanine-amine) could be
suppressed in comparison to a dissolved amino acid, due to the
proximity of neighbouring amines. Suppression of ionisation is
well known to shift peptide synthesis/hydrolysis equilibria
thermolysin and a phosphate buffer (i). After 16 hours, reaction
products were released from the resin by TFA cleavage of the Wang
linker (ii) and quantified by reverse phase HPLC. X is either Fmoc or
Z protecting group. R1 is the amino acid side chain.
as observed when organic (co-) solvents are used in peptide
synthesis.
In this article we aim to quantify the relative importance of
each of these three factors in order to understand what governs
the equilibrium shift that leads to the high yielding peptide
synthesis that was observed.
Results and discussion
What causes the equilibrium shift toward peptide synthesis when
the substrate is immobilized?
All reactions were performed as shown in Scheme 1. Phenyl-
alanine was chosen as the immobilised amino component
of the reaction because of the known substrate preference of
thermolysin at an amine site for hydrophobic and aromatic
amino acid residues. PEGA1900 immobilised phenylalanine was
reacted overnight in the presence of thermolysin with a range
of N-protected amino acids with unprotected side chains in an
aqueous phosphate buffer.
T h i s j o u r n a l i s © T h e R o y a l S o c i e t y o f C h e m i s t r y 2 0 0 3
O r g . B i o m o l . C h e m . , 2 0 0 3 , 1, 1 2 7 7 – 1 2 8 1
1277

View Article Online
total
i. The effect of the excess soluble amino acids on the equi-
librium position. A well-known method of shifting reaction
equilibria is to apply larger concentrations of starting materials
to push the equilibrium toward synthesis of the end-product. In
particular in solid phase chemistry this method is commonly
used because unreacted excess starting material can simply be
washed from the solid supported product. In order to deter-
mine if excess of amino acid 1 was responsible for the high
yields of synthesis observed, the solid phase reaction was com-
pared to an equivalent reaction in aqueous solution. In this case
the acylation equilibrium is compared for PEGA1900-Phe (2A)
and free phenylalanine methyl ester (Phe-OMe) (2B, Scheme 2).
ments, the total initial concentration [Phe-OMe]
was chosen
initial
as equal to the amount of immobilized phenylalanine amine
per unit volume present in the solid phase experiments (2 µmol
per 2 ml = 1mM). The pKa of the amine group of Phe-OMe
equals 7.0.8 The pKa values of the Fmoc amino acids 1a–1g
were estimated as 3.7. For 1h they were taken as 3.0 and 8.0 for
the macroscopic ionization constants and 3.6 and 7.4 for the
microscopic ionization constants. For 1i they were 3.2 for the α-
acid and 4.8 for the side chain. (For methods of pKa estimation,
see ref. 8). The pH of the reaction was 7.4 and the total satur-
ated concentrations of 1a–1i were measured (Table 1). Using
eqn. (6) then resulted in the predicted conversions in aqueous
solution given in Table 1. These results could be directly com-
pared to those experimentally observed on immobilised Phe
(Scheme 2A).
From Table 1 it is clear that the conversions on immobilised
Phe-PEGA are in all cases much higher than those expected
for the soluble Phe-OMe. These observations clearly show that
the high yields of peptide synthesis obtained on immobilised
Phe cannot exclusively be explained in terms of the excess of
1 present.
Scheme 2 A: Amide synthesis/hydrolysis on solid supported substrate;
B: Amide synthesis in aqueous solution.
As expected, the amino acids that are present in smaller con-
centrations (poorly soluble) are predicted to give rise to the
lowest yields in solution. By contrast, poor aqueous solubility
of 1 leads to generally higher conversion on immobilized Phe.
As seen in particular for 3c and 3e as compared to e.g. 3g–3i.
The apparent inverse correlation between substrate solubility
and yield will be discussed in the next section.
Given the finding that a large excess of amino acid was not
required for successful synthesis of peptide on solid support,
a systematic study relating yield of synthesis to excess of
substrate was carried out.
Thus, the synthesis of peptide Fmoc-Nle-Phe (3c) was
carried out with increasing ratio of equivalents of protected
amino acid over supported amino acid. Fig. 1 shows that a
10-fold excess was sufficient to allow for complete conversion to
peptide 3c. This compares favourably with conventional chem-
ical peptide synthesis, where 5–10 fold excess of reagants are
usually employed.
The equilibrium position of reaction for the solution phase
reaction can be calculated reliably from the initial reactant
concentrations, reactant ionisation constants, and the pH of
the reaction which are known. This calculation is based on the
single reference equilibrium constant that was recently identi-
fied for amide hydrolysis/synthesis reactions in aqueous
medium. When all reactant concentrations in the amide equi-
librium were expressed in their uncharged form, a reference
value of Kr0ef = 103.6 MϪ1 was consistently found, regardless
of the actual molecular form of the reactants.8 Hence, at
equilibrium we have for reaction 2B:
(1)
The concentrations of uncharged [Phe-OMe]0 and [X-aa2]0
follow from the simple ionisation relationships:
(2)
(3)
For the zwtterionic molecule Fmoc-His and the di-acid
Fmoc-Asp with two pKa values each the equations are slightly
more complicated.8 The peptide concentration at equilibrium
follows from combining eqn. (1)–(3).
Fig. 1 Stoichiometry of thermolysin catalysed synthesis of peptide 3c
on PEGA immobilized Phe at pH 8 in 0.1 M K-phosphate buffer. Note
that 1 equivalent of Fmoc-Nle corresponds to a 2 mM solution of
Fmoc-Nle.
(4)
After adding the mass balance equation
(5)
ii. Removal of hydrophobic groups from aqueous solution.
The micro-environment within the polymeric PEGA1900
support is more hydrophobic than that of the bulk aqueous
solution. Thus one might expect a transfer of hydrophobic
enzyme substrates (such as Fmoc protected amino acids used
here) from the aqueous bulk solution to the polymer support
to be favourable. Hence, the synthesis reaction is expected to be
more favoured with more hydrophobic enzyme substrates. We
have used log P values to quantify the hydrophobicity of the
we obtain for the fraction equilibrium conversion eqn. 6
below.
This equation shows that the fractional conversion of Phe-
OMe to peptide is independent of its initial concentration.
However, for the best direct comparison to solid phase experi-
(6)
O r g . B i o m o l . C h e m . , 2 0 0 3 , 1, 1 2 7 7 – 1 2 8 1
1278

View Article Online
Table 1 Thermolysin catalysed peptide synthesis yields from protected amino acids in the solid phase reaction on PEGA1900-Phe compared to the
calculated yields of the equivalent reaction in aqueous solution on free phenylalanine methyl ester (Phe-OMe). The aqueous solubility and log P of
the protected amino acids are also given
Saturated conc.
acyl donor/mM
Observed conversion on
solid phase (%)a, b
Calculated conversion
in solution (%)c
a
Peptide (3)
R1
Acyl donor (1)
Log P
a
b
c
d
e
f
g
h
i
H
Fmoc-Gly
Fmoc-Leu
Fmoc-Nle
Z-Phe
Fmoc-Phe
Fmoc-Gln
Fmoc-Ser
Fmoc-His
Fmoc-Asp
40
16
5
63
3
99
99
99
72
99
84
10
77
70
2.3
0.9
0.3
2.8
0.1
0.4
2.0
0.3
0.4
1.7
3.5
3.6
1.6
3.5
0.3
1.0
1.2
0.7
CH2-CH2-(CH3)2
(CH2)3-CH3
CH2-C6H5
CH2-C6H5
CH2-CH2-CONH2
CH2-OH
7
36
24
25
CH2-(imidazole)ϩ
CH2-COOϪ
a Conversions (%) determined by HPLC after acid cleavage of the Wang linker. b The conversions were always measured after at least 16 h reaction
time. In several cases results were compared to 80 hours and no significant differences were found. Hence, it is reasonable to assume that reactions
had approached thermodynamic equilibrium. c If this value exceeds the peptide solubility limit of the product, precipitation of the peptide would be
observed leading to overall higher conversions. d Calculated using the c log P calculator on www. Daylight.com
protected amino acids used. From Table 1 it is clear that sub-
strates with the highest log P gave rise to the best conversions
(1a, 1b, 1c, 1e and to a lesser extent 1d), while less hydrophobic
amino acids lead to lower conversion yields (1f–1i). These
observations indicate that the hydrophobic effect is an impor-
tant factor in the equilibrium shift observed when comparing
solid phase reactions to those in aqueous solution. However,
there is no absolute correlation between log P and conversion.
For example, amino acid 1f has the lowest log P value but
reaches a good conversion of 84%. Thus there are other specific
interactions between the polymer and amino acids that con-
tribute to the conversion yields obtained.
In order to obtain more accurate data on the contribution of
substrate hydrophobicity to the equilibrium of the reaction, the
conversions obtained for the same amino acid with different
protecting groups under otherwise identical conditions were
investigated. Fmoc-Phe and Z-Phe have the same pKa of 3.7
and only differ in the hydrophobicity due to their different
protecting groups. Fig. 2 shows the synthesis of peptides 3d and
3e as a function of pH. For both these two acyl donors the
shape of the curve is the same and its shape can be explained in
term of ionization effects (see next section). The difference
in conversions observed when comparing Z-Phe and Fmoc-Phe
therefore directly reflects the contribution of substrate hydro-
phobicity.
group of the immobilized phenylalanine may well be signifi-
cantly suppressed as compared to the amino group of a soluble
phenylalanine derivative. Peptide hydrolysis is favored in dilute
aqueous solution largely because of favourable ionization
of both the amine and acid components. Any suppression of
ionization is therefore expected to result in a shift toward
peptide synthesis.
When in solution, amino acid amines will protonate
independently of each other. By contrast, solid supported
amino groups are restricted in their movement, and when close
together they are likely to influence each other. When the pH of
the bulk solution is lowered, polymer bound amines will start
to become protonated leading to local positive charges within
the polymer. As the pH drops further, protonation of polymer
bound amino groups will become increasingly more difficult
due to electrostatic repulsion between the charged amines.
Hence, ionization will be more and more suppressed at low
pH values when compared to free solution. And suppressed
ionisation will result in an equilibrium shift towards peptide
synthesis as is also observed when organic (co-) solvents are
used in solution phase biocatalysis.
To analyze the behavior quantitatively we have considered
the case where excess solid X-aa2 is present, so that [X-aa2]0
remains constant at its solubility in the reaction mixture.
Eqn. (6) may then be re-written as:
(7)
Note, that Kr0ef is now replaced by K0 because hydrophobic
effects may cause a shift from the aqueous solution reference
value, as explained above. Eqn. (7) can be rewritten as:
(8)
If one assumes that there is no electrostatic repulsion
Fig. 2 Effect of substrate hydrophobicity on the synthesis yields of the
thermolysin catalysed peptide synthesis on solid phase PEGA1900 as a
function of pH: Z-Phe-Phe (squares) and Fmoc-Phe-Phe (diamonds).
between amines, all amines ionise according to a single pKamine
value. In that case, eqn. (8) suggests that a plot of 1/f peptide
eq
versus [Hϩ] will be linear (K0 and [X-aa2]0 are constants).
Unfortunately such a plot gives rise to bunched data points
even over the relatively small pH range (2 units) in our experi-
Thus, it appears that the use of more hydrophobic protecting
groups will allow for a further decrease of the number of equiv-
alents of soluble amino acid needed. It is expected that conver-
sion yields could be improved further by using tailor-made
highly hydrophobic protecting groups.
ments. When plotting instead Ϫlog f peptide limiting behaviour
eq
can be recognised at high and low pH: above the pKamine the
[Hϩ] term in eqn. (8) becomes negligible and peptide yield will
be independent of pH resulting in a straight line of slope 0. At
low pH the [Hϩ] term dominates, and the equation predicts a
iii. The effect of suppressed amine ionization on the equi-
librium position. As a third possible contribution to the equi-
librium shift it was suggested that protonation of the amino
linear relationship between Ϫlog f peptide and pH, with a gradient
eq
of Ϫ1. By contrast, if the amine charges did affect each other
O r g . B i o m o l . C h e m . , 2 0 0 3 , 1, 1 2 7 7 – 1 2 8 1
1279

View Article Online
(i.e. no single pKamine), the protonation of the amino groups
becomes increasingly more difficult when the pH is lowered. In
this case the amino groups would behave as if they had a range
of pKamine values. For the plot this would result in a curved line
Solid phase peptide synthesis
Solid phase substrates were linked to amino functionalised
PEGA1900–NH2 (Polymer Laboratories) beads via a Wang-type
linker (hydroxymethylphenoxyacetic acid, HMPA). Attach-
ment of the linker was achieved in DMF with 3 equivalents of
HMPA, 4 equivalents diisopropylcarbodimide (DIC), and 6
equivalents hydroxybenzotriazole (HOBt). After overnight
rotation on a blood rotator at 20 ЊC, the resin was washed
(5 × 50 : 50 acetonitrile : water, 5 × MeOH, 5 × DMF, 5 ×
acetonitrile, 5 × DCM, 5 × DMF) and the first amino acid was
coupled to the resin. This step used 3 equivalents Fmoc-amino
acid, 4 equivalents DIC and 0.1 equivalents of dimethylamino-
pyridine (DMAP) in DMF. The mixture was incubated on the
blood rotator overnight at 20 ЊC. This was again followed by a
washing sequence as above.
Unreacted OH groups were capped by addition of 5 equiv-
alents acetic anhydride in DMF and left on the blood rotator
at 20 ЊC overnight. This was followed by a washing sequence
(as above) and removal of the Fmoc group by addition of a
solution of 20% piperidine in DMF, and left 30 minutes on
blood rotator.
instead of a straight line at pH values below the of pKamine
.
Fig. 3 shows the pH dependence of the reactions producing
solid supported peptides 3d and 3e. Both show the two straight
line regions as expected for a system where amines protonate
independently. The break between the two occurs at bulk pH
around 7.0, which is the pKa of the amino group of Phe-OMe
in free solution. Hence it can be concluded that under the con-
ditions used suppressed amine ionisation is a minor factor
in the observed equilibrium shift. The shift of the line at all
pH values observed when comparing the two peptides reflects
the difference in K0 through the hydrophobic contribution as
discussed in the previous section.
Determining loading of Phe on PEGA1900
A weighed amount of PEGA1900-Wang-Phe-Fmoc resin was
cleaved with 95 : 5 (v/v) TFA–water during 2 hours. The
cleavage mixture was washed with 10 ml of a mixture of
50 :50 (v/v) water : acetonitrile. The solvent was evaporated
in a vacuum centrifuge (Christ, Germany). The residue was
redissolved in 1 ml 50 : 50 (v/v) water : acetonitrile containing
0.1% TFA. The loading was then determined by reverse phase
HPLC (for conditions, see below).
Fig. 3 Effect of solid supported amine ionisation on the synthesis
yields of the thermolysin catalysed peptide synthesis on solid phase
PEGA1900 as a function of pH: Z-Phe-Phe (squares) and Fmoc-Phe-Phe
(diamonds).
In addition, a known amount of PEGA1900-Phe-Fmoc was
weighed, then dried in a vacuum oven at RT to constant weight.
This gave a value for the swelling of the resin and allowed for the
dry weight to be calculated. The loading of the resin obtained
varied from batch to batch between 70 and 110 µmol gϪ1.
In these experiments a relatively high ionic strength buffer
was used (0.1 M K-phosphate buffer). Electrostatic interaction
between the charged groups will be greater at lower ionic
strength, and hence ionization of solid supported amines will
be suppressed more. It is therefore expected that synthesis will
be more favoured at lower ionic strength. We are currently
studying the effect of ionic strength on the peptide synthesis/
hydrolysis equilibrium in more detail.
Enzymatic peptide synthesis on solid support
For the enzymatic reactions 5 mg of thermolysin (protease type
X from Sigma) was added to a suspension of 10 mg PEGA1900
resin, 0.2 mmol protected amino acid, and 2 mL 0.1 M
potassium phosphate buffer of the appropriate pH. Reactions
were briefly mixed, and then incubated overnight at 20 ЊC on a
blood rotator. The next day the resin was washed extensively
using 5 mL volumes in the following sequence: 5 × 50 : 50
acetonitrile : water, 5 × MeOH, 5 × DMF, 5 × acetonitrile,
5 × DCM, 5 × DMF. The products were cleaved from the resin
with 2ml TFA : water 95 : 5 for 2 hours. Resin was then washed
with 10 mL of a mixture of 50 : 50 acetonitrile in water, solvent
was removed by evaporation and the residue redissolved in
1 mL 50 : 50 mixture of acetonitrile and water.
Conclusion
There can be a substantial equilibrium shift in favour of amide
synthesis when amines are immobilized on a solid support,
compared with preferred hydrolysis in aqueous solution. This
allows efficient protease catalysed solid-phase synthesis of a
variety of peptides. By comparison with the reaction in aque-
ous solution it was shown that the shift in equilibrium is not
caused just by the use of excess amino acids. The most
important factor appears to be substrate hydrophobicity and
consequently the highest conversions are observed with amino
acid substrates carrying hydrophobic protecting groups.
Analysis
The samples were analysed by HPLC on a Waters 2690 LC
system equipped with a Waters 468 UV detector and a reverse
phase column (0.46 × 25cm Hichrom HIRPB-250A). Mobile
phases were acetonitrile and water with 0.1% TFA added to
each, a gradient was used that started with 20% acetonitrile
increasing to 50% acetonitrile in 30 minutes at a flow rate of
1 ml minϪ1. LCMS was done on a Waters 2790 LC system
coupled with a Micromass Platform II mass spectrometer using
Electrospray ionisation mode.
Experimental
Enzymes, reagents and solvents
Thermolysin was obtained from Sigma (UK) as protease type
X. Fmoc-Phe,
Z-Phe, Fmoc-Nle, and Fmoc-Gly were all from Sigma (UK).
Fmoc-Asp, Fmoc-His, Fmoc-Ser, Fmoc-Gln were from
Bachem (UK). Fmoc-Leu, hydroxymethylphenoxyacetic acid
(HMPA), and dimethylaminopyridine (DMAP) were from
Novabiochem. PEGA1900 was a kind gift provided by Polymer
Laboratories (UK). Diisopropylcarbodimide (DIC), hydroxy-
benzotriazole (HOBt) and acetic anhydride were from Aldrich.
All solvents were of the highest purity available and obtained
from Aldrich.
Acknowledgements
The authors gratefully acknowledge financial support from the
BBSRC, EC and the Wellcome Trust. We would also like to
thank Polymer Laboratories (UK) for the generous supply of
PEGA1900
O r g . B i o m o l . C h e m . , 2 0 0 3 , 1, 1 2 7 7 – 1 2 8 1
1280

View Article Online
5 F. G. Kuruvilla, A. F. Shamji, S. M. Sternson, P. J. Hergenrother and
S. L. Schreiber, Nature, 2002, 416, 653–657.
References
1 N. Bezay, G. Dudziak, A. Liese and H. Kunz, Angew. Chem., Int. Ed.,
2001, 40, 2292–2295.
6 P. C. Michels, Y. L. Khmelnitsky, J. S. Dordick and D. S. Clark,
Trends Biotechnol., 1998, 16, 210–215.
7 R. V. Ulijn, B. Baragana, P. J. Halling and S. L. Flitsch, J. Am. Chem.
Soc., 2002, 124, 10988–10989.
8 R. V. Ulijn, A. E. M. Janssen, B. D. Moore and P. J. Halling, J. Chem.
Soc., Perkin Trans. 2, 2002, 5, 1024–1028.
2 M. Meldal, Biopolymers, 2002, 66, 93.
3 R. Reents, D. A. Jeyaraj and H. Waldmann, Drug Disc. Today, 2002,
7, 71–76.
4 D. W. Kadereit and H. Waldmann, Chem. Rev., 2001, 101, 3367–3396.
O r g . B i o m o l . C h e m . , 2 0 0 3 , 1, 1 2 7 7 – 1 2 8 1
1281