Sequence-specific Ni(II)-dependent peptide bond hydrolysis for protein engineering. combinatorial library determination of optimal sequences

Article abstract of DOI:10.1021/ja907567r

Previously we demonstrated for several examples that peptides having a general internal sequence R_N-Yaa-Ser/Thr-Xaa-His-Zaa-R_c (Yaa = Glu or Ala, Xaa = Ala or His, Zaa = Lys, R_N and R_c = any N- and C-terminal amino acid sequence) were hydrolyzed specifically at the Yaa-Ser/Thr peptide bond in the presence of Ni(II) ions at alkaline pH (Krezel, A.; Mylonas, M.; Kopera, E.; Bal, E. Acta Biochim. Polon. 2006, 53, 721-727 and references therein). Hereby we report the synthesis of a combinatorial library of CH₃CO-Gly-Ala-(Ser/Thr)-Xaa-His-Zaa-Lys-Phe- Leu-NH₂ peptides, where Xaa residues included 17 common a-amino acids (except Asp, Glu, and Cys) and Zaa residues included 19 common a-amino acids (except Cys). The Ni(II)-dependent hydrolysis at 37 and 45 °C of batches of combinatorial peptide mixtures randomized at Zaa was monitored by MALDI-TOF mass spectrometry. The correctness of librarybased predictions was confirmed by accurate measurements of hydrolysis rates of seven selected peptides using HPLC. The hydrolysis was strictly limited to the Ala-Ser/Thr bond in all library and individual peptide experiments. The effects of individual residues on hydrolysis rates were quantified and correlated with physical properties of their side chains according to a model of independent contributions of Xaa and Zaa residues. The principal component analysis calculations demonstrated partial molar side chain volume and the free energy of amino acid vaporization for both Xaa and Zaa residues and the amine pK_a for Zaa residues to be the most significant empirical parameters influencing the hydrolysis rate. Therefore, efficient hydrolysis required bulky and hydrophobic residues at both variable positions Xaa and Zaa, which contributed independently to the hydrolysis rate. This relationship between the peptide sequence and the hydrolysis rate provides a basis for further research, aimed at the elucidation of the reaction mechanism and biotechnological applications of Ni(II)-dependent peptide bond hydrolysis.

Full text of DOI:10.1021/ja907567r

Published on Web 02/18/2010
Sequence-Specific Ni(II)-Dependent Peptide Bond Hydrolysis
for Protein Engineering. Combinatorial Library Determination
of Optimal Sequences
Artur Kre¸z˙el,† Edyta Kopera,‡ Anna Maria Protas,‡ Jarosław Poznan´ski,‡,§
Aleksandra Wysłouch-Cieszyn´ska,‡ and Wojciech Bal*,‡,|
Laboratory of Protein Engineering, Faculty of Biotechnology, UniVersity of Wrocław, Tamka 2,
50-137 Wrocław, Poland, Institute of Biochemistry and Biophysics, Polish Academy of Sciences,
Pawin´skiego 5a, 02-106 Warsaw, Poland, Institute of Physical Chemistry, Polish Academy of
Sciences, Kasprzaka 44/52, 01-224 Warsaw, Poland, and Central Institute for Labour
ProtectionsNational Research Institute, Czerniakowska 16, 00-701 Warsaw, Poland
Received September 7, 2009; E-mail: wbal@ibb.waw.pl
Abstract: Previously we demonstrated for several examples that peptides having a general internal
sequence RN-Yaa-Ser/Thr-Xaa-His-Zaa-RC (Yaa ) Glu or Ala, Xaa ) Ala or His, Zaa ) Lys, RN and RC
)
any N- and C-terminal amino acid sequence) were hydrolyzed specifically at the Yaa-Ser/Thr peptide bond
in the presence of Ni(II) ions at alkaline pH (Kre¸z˙el, A.; Mylonas, M.; Kopera, E.; Bal, E. Acta Biochim.
Polon. 2006, 53, 721-727 and references therein). Hereby we report the synthesis of a combinatorial
library of CH3CO-Gly-Ala-(Ser/Thr)-Xaa-His-Zaa-Lys-Phe-Leu-NH2 peptides, where Xaa residues included
17 common R-amino acids (except Asp, Glu, and Cys) and Zaa residues included 19 common R-amino
acids (except Cys). The Ni(II)-dependent hydrolysis at 37 and 45 °C of batches of combinatorial peptide
mixtures randomized at Zaa was monitored by MALDI-TOF mass spectrometry. The correctness of library-
based predictions was confirmed by accurate measurements of hydrolysis rates of seven selected peptides
using HPLC. The hydrolysis was strictly limited to the Ala-Ser/Thr bond in all library and individual peptide
experiments. The effects of individual residues on hydrolysis rates were quantified and correlated with
physical properties of their side chains according to a model of independent contributions of Xaa and Zaa
residues. The principal component analysis calculations demonstrated partial molar side chain volume
and the free energy of amino acid vaporization for both Xaa and Zaa residues and the amine pKa for Zaa
residues to be the most significant empirical parameters influencing the hydrolysis rate. Therefore, efficient
hydrolysis required bulky and hydrophobic residues at both variable positions Xaa and Zaa, which contributed
independently to the hydrolysis rate. This relationship between the peptide sequence and the hydrolysis
rate provides a basis for further research, aimed at the elucidation of the reaction mechanism and
biotechnological applications of Ni(II)-dependent peptide bond hydrolysis.
Introduction
sequence specificities, limitations due to specific reaction
conditions, or a requirement for tedious procedures of protease
The sequence-specific cleavage of the peptide bond is a
crucial procedure in protein engineering and purification. The
extreme stability of this bond, with half-life for spontaneous
hydrolysis estimated as 350-600 years at neutral pH and room
temperature, limits the range of appropriate cleavage reagents.1
Those used currently include natural or engineered proteolytic
enzymes,2 self-cleaving intein sequences,3 and chemical agents,
such as cyanogen bromide.4 All these methodologies have
specific disadvantages, resulting from side reactions, insufficient
removal.
The search for alternative agents for sequence-specific cleav-
age of peptides and proteins among metal ions and their
complexes included redox-active metal ions. They can cleave
peptide bonds, especially in the presence of reactive oxygen
species (ROS) precursors, e.g., H2O2/ascorbate.5-7 The redox
cleavage, while rapid, is area- rather than site-specific and
therefore is not suitable for applications requiring homogeneous
protein products.
Hydrolytic chemistry of peptide bond has been used with
more success in terms of sequence specificity. Many metal ions
and metal complexes were found to promote or catalyze the
† Faculty of Biotechnology, University of Wrocław.
‡ Institute of Biochemistry and Biophysics, Polish Academy of Sciences.
§ Institute of Physical Chemistry, Polish Academy of Sciences.
| Central Institute for Labour ProtectionsNational Research Institute.
(1) Radzicka, A.; Wolfenden, R. J. Am. Chem. Soc. 1996, 110, 6105–
6109.
(5) Rana, T. M.; Meares, C. F. J. Am. Chem. Soc. 1990, 112, 2457–2458.
(6) Rana, T. M.; Meares, C. F. J. Am. Chem. Soc. 1991, 113, 1859–1861.
(7) Andberg, M.; Jantti, J.; Heilimo, S.; Pihkala, P.; Paananen, A.;
Koskinen, A. M. P.; Soderlund, P.; Linder, M. B. Protein Sci. 2007,
16, 1751–1761.
(2) Arnau, J.; Lauritzen, C.; Petersen, G. E.; Pedersen, J. Protein
Expression Purif. 2006, 48, 1–13.
(3) Muir, T. W. Annu. ReV. Biochem. 2003, 72, 249–289.
(4) Gross, E. Methods Enzymol. 1967, 11, 238–255.
9
10.1021/ja907567r  2010 American Chemical Society
J. AM. CHEM. SOC. 2010, 132, 3355–3366 3355

A R T I C L E S
Kre¸z˙el et al.
hydrolysis of peptide bonds in short peptides.8-10 In particular,
many metal ions promote the hydrolysis of Xaa-Ser peptide
sequences (Xaa ) any amino acid) by polarizing the peptide
carbonyl group by coordination and assisting in its migration
to the intramolecular hydroxyl group.11,12 These reactions,
however, have not been demonstrated for downstream peptide
bonds in longer peptides or proteins. In contrast, the Cu(II) ions
were reported to specifically cleave the Lys226-Thr227 peptide
bond in the hinge region of human IgG1.13 In follow-up studies,
the specific Cu(II)-related hydrolysis was demonstrated for
peptide bonds preceding Ser-His and Thr-His sequences at
elevated temperatures an pH values.14-16 These studies indicated
that the sequence specificity for metal ion-promoted peptide
bond hydrolysis can be based on more than one amino acid
residue.
Coordinatively unsaturated Pd(II) complexes, such as cis-
[Pd(en)(H2O)2]2+, can exchange water for a sulfur atom present
in the side chain of Met or Cys or for a His imidazole nitrogen
atom. The cleavage of the second peptide bond upstream results
from these interactions.17,18 Pt(II) aqua ion and its complexes,
such as cis-[Pt(en)(H2O)2]2+, bind to Cys and Met sulfurs,
causing the cleavage of the first peptide bond downstream.19,20
These reactions proceed at low pH, around 2. Their sequential
specificity is limited to single Cys/Met/His residues. Interest-
ingly, cisplatin (cis-[Pt(NH3)2Cl2]) was found to share the
reactivity with Pd(II) complexes, rather than Pt(II) complexes
mentioned above, but in a wider pH range.21 The Trp nitrogen
atom was also found to provide an anchor for hydrolytic Pd(II)
and Pt(II) complexes.22,23 The cleavage of peptide bonds directly
preceding Ser/Thr-His/Met sequences in human serum albumin
(HSA) was accomplished with the use of a Pd(II) complex, in
line with peptide studies.24
fragmentation, rather than specific cleavage sites, in both
cases.26,27 Altogether, the reactions of metal ions presented
above share a disadvantage of low sequence specificity, based
on one or two adjacent amino acid residues. Therefore, they
are suited better for protein fragmentation, e.g., for mass
spectrometry, than for selective cleavage of dedicated sequences
for protein engineering.
We found that the CH3CO-Thr-Glu-Ser-His-His-Lys-NH2
hexapeptide underwent a slow hydrolysis in the presence of
Ni(II) ions in a phosphate buffer, at pH 7.4 and 37 °C.28
A
Ni(II) complex of the C-terminal tetrapeptide amide Ser-His-
His-Lys-NH2 was found to be the product of this reaction, with
a yield between 3% and 9% after 140 h of incubation, depending
on the concentration of Ni(II) ions. The cleavage occurred solely
between the Glu and Ser residues. Subsequent studies revealed
that a 34-residue peptide and histone H2A, both comprising the
above sequence, were hydrolyzed with an identical sequence
specificity by Ni(II) ions under analogous conditions, but ca. 7
times faster.29 Cu(II) ions hydrolyzed this 34-residue peptide
with the same specificity as Ni(II) ions, but 3 times slower,
while Co(II) and Zn(II) ions were inactive. In further studies
on Ala-substituted analogues of CH3CO-Thr-Glu-Ser-His-His-
Lys-NH2, we found that the Ser residue and the C-terminal His
residue were necessary for hydrolysis to occur.30 Subsequently,
we showed that substitution of the Ser residue with a Thr residue
maintained the reactivity toward Ni(II) ions.31 We demonstrated
that the reaction proceeded above pH 7, with an acceleration at
pH 9-10. Relatively large differences of hydrolysis rates were
found for individual sequences studied.
In brief, our prior studies demonstrated that peptides of a
general sequence Yaa-(Ser/Thr)-Xaa-His-Zaa, where Yaa ) Glu
or Ala, Xaa ) Ala or His, and Zaa ) Lys, could be hydrolyzed
specifically at the Yaa-(Ser/Thr) peptide bond in the presence
of Ni(II) ions at alkaline pH. On this basis, we hypothesized
that some amino acid substitutions in positions Xaa and Zaa
could accelerate hydrolysis further, thus providing a specific
hydrolysis site comprising a tri- or even tetrapeptide sequence.
The addition of Ni(II) ions to such a site would yield an artificial
endopeptidase system with a specificity comparable to those of
enzymes used in biotechnological practice. We therefore decided
to systematically review the effects of substitutions in positions
Xaa and Zaa on the rate of hydrolysis of the Yaa-Ser/Thr bond
in the Yaa-(Ser/Thr)-Xaa-His-Zaa-sequence. For this purpose,
we synthesized a combinatorial library of R1-(Ser/Thr)-Xaa-
His-Zaa-R2 peptides (R1 ) CH3CO-Gly-Ala, R2 ) Lys-Phe-
Leu-NH2), studied their Ni(II)-related hydrolysis using MALDI-
TOF mass spectrometry, and performed a thorough statistical
analysis of relations between the reaction rate and the peptide
sequence. We verified the library screening results for selected
R1-Ser/Thr-Xaa-His-Zaa-R2 peptides using HPLC. The results
The hydrolytic cleavage of proteins was achieved only in a
few other cases. The scission of myoglobin at Gln91-Ser92 and
Ala94-Thr95 peptide bonds was found for a series of Cu(II)
compounds.25 In two separate studies Cu(II) complexes were
reacted with bovine serum albumin (BSA), yielding protein
(8) Grant, K. B.; Kassai, M. Curr. Org. Chem. 2006, 10, 1035–1049.
(9) Fujii, Y.; Kiss, T.; Gajda, T.; Tan, X. S.; Sato, T.; Nakano, Y.; Hayashi,
Y.; Yashiro, M. J. Biol. Inorg. Chem. 2002, 7, 843–851.
(10) Yashiro, M.; Sonobe, Y.; Yamamura, A.; Takarada, T.; Komiyama,
M.; Fujii, Y. Org. Biomol. Chem. 2003, 1, 629–632.
(11) Yashiro, M.; Yamamura, A.; Takarada, T.; Komiyama, M. J. Inorg.
Biochem. 1997, 67, 225.
(12) Kassai, M.; Ravi, R. G.; Shealy, S. J.; Grant, K. B. Inorg. Chem. 2004,
43, 6130–6132.
(13) Smith, M. A.; Easton, M.; Everet, P.; Lewis, G.; Payne, M.; Riveros-
Moreno, V.; Allen, G. Int. J. Pept. Protein Res. 1996, 48, 48–55.
(14) Allen, G.; Campbell, R. O. Int. J. Pept. Protein Res. 1996, 48, 265–
273.
(15) Humphreys, D. P.; Smith, B. J.; King, L. M.; West, S. M.; Reeks,
D. G.; Stephens, P. E. Protein Eng. 1999, 12, 179–184.
(16) Humphreys, D. P.; King, L. M.; West, S. M.; Chapman, A. P.; Sehdev,
M.; Redden, M. W.; Glover, D. J.; Smith, B. J.; Stephens, P. E. Protein
Eng. 2000, 13, 201–206.
(25) Zhang, L.; Mei, Y.; Li, S.; Sun, X.; Zhu, L. Inorg. Chem. 2003, 42,
492–498.
(17) Milovic´, N. M.; Kostic´, N. M. J. Am. Chem. Soc. 2002, 124, 4759–
4769.
(26) Hegg, E. L.; Burstyn, J. N. J. Am. Chem. Soc. 1995, 117, 7015–7016.
(27) de Oliviera, M. C. B.; Scarpellini, M.; Neves, A.; Terenzi, H.;
Bortoluzzi, A. J.; Szponganics, B.; Greatti, A.; Mangrich, A. S.; de
Souza, E. M.; Fernandez, P. M.; Soares, M. R. Inorg. Chem. 2005,
44, 921–929.
(18) Milovic´, N. M.; Kostic´, N. M. J. Am. Chem. Soc. 2003, 125, 781–
788.
(19) Milovic´, N. M.; Dutcaˇ, L.-M.; Kostic´, N. M. Inorg. Chem. 2003, 42,
4036–4045.
(20) Milovic´, N. M.; Dutcaˇ, L.-M.; Kostic´, N. M. Chem.sEur. J. 2003, 9,
5097–5106.
(28) Bal, W.; Lukszo, J.; Bialkowski, K.; Kasprzak, K. S. Chem. Res.
Toxicol. 1998, 11, 1014–1023.
(21) Manka, S.; Becker, F.; Hohage, O.; Sheldrick, W. S. J. Inorg. Biochem.
2004, 98, 1947–1952.
(29) Bal, W.; Liang, R.; Lukszo, J.; Lee, S.-H.; Dizdaroglu, M.; Kasprzak,
K. S. Chem. Res. Toxicol. 2000, 13, 616–624.
(22) Zhu, L.; Kostic´, N. M. Inorg. Chim. Acta 2002, 339, 104–110.
(23) Kaminskaia, N. V.; Johnson, T. W.; Kostic´, N. M. J. Am. Chem. Soc.
1999, 121, 8663–8664.
(30) Mylonas, M.; Kre¸z˙el, A.; Plakatouras, J. C.; Hadjiliadis, N.; Bal, W.
J. Chem. Soc., Dalton Trans. 2002, 4296–4306.
(31) Kre¸z˙el, A.; Mylonas, M.; Kopera, E.; Bal, W. Acta Biochim. Polon.
2006, 53, 721–727.
(24) Kaminskaia, N. V.; Kostic´, N. M. Inorg. Chem. 2001, 40, 2368–2377.
9
3356 J. AM. CHEM. SOC. VOL. 132, NO. 10, 2010

Library Selection for Peptide Bond Hydrolysis
A R T I C L E S
Scheme 1. Design of Synthesis of Peptide Libraries
anhydrate. The resulting N-terminally blocked peptides were cleaved
from the resin with a mixture of trifluoroacetic acid (TFA) and a
cleavage scavenger reagent, such as triisopropylsilane (TIPS), and
water (95:2.5:2.5). The resins were then filtered away, and diethyl
ether was added next in order to precipitate the peptide. Each
precipitated peptide was filtered through a sintered glass funnel and
redissolved in water with acetic acid. The peptide solution was then
frozen and lyophilized. Peptides were purified by reverse-phase high
performance liquid chromatography (HPLC, Breese Waters) on C18
preparative columns (Vydac). The eluting solvent A was 0.1% TFA/
water and solvent B was 0.1% TFA/90% acetonitrile/water. The
linear gradient from 10% to 40% in 30 min at a flow rate of 2
mL/min, with detection at 220 nm (common to all peptides) and
280 nm (aromatic residues), was applied routinely. The identities
and purities of peptides were confirmed using mass spectrometry,
on a Q-Tof1 ESI-MS spectrometer (Waters).
Mass Spectrometry Screening of Hydrolysis Rates. Each of
34 sublibraries randomized in position Zaa was divided into two
portions for separate reactions at 37 and 45 °C. The hydrolysis
reactions were performed in a 10 mM Tris buffer at pH 8.2. The
total concentrations of peptide mixtures and Ni(II) ions were 1 and
2 mM, respectively. The detection of reaction substrates and
products was accomplished with the use of a MALDI-TOF
(Micromass) mass spectrometer. The aliquots of reaction mixtures
were withdrawn and measured at measurement times, tm, of 0, 2,
4, 6, 8, and 24 h of incubation. The progress of reaction was
evaluated by visual detection in mass spectra of the appearance of
signals corresponding to expected products of hydrolysis, Ser/Thr-
Xaa-His-Zaa-Lys-R2, at given tm. The formula used for the
semiquantitative evaluation of reaction progress had the form of
eq 1.
are described below. In a forthcoming paper (J. Am. Chem. Soc.,
submitted), we will present a detailed analysis of reaction
kinetics and mechanism for these peptides, together with the
optimization of reaction conditions for its practical applications.
Experimental Section
Synthesis of Peptide Libraries. The synthesis of the CH3CO-
Gly-Ala-Ser/Thr-Xaa-His-Zaa-Lys-Phe-Leu-NH2 library was ac-
complished on an AM-RINK amide resin (NovaBiochem, Merck)
on a 800 mg scale, according to a standard manual Fmoc protocol.32
Residues in position Xaa included the canonical set of standard
DNA-coded amino acids, except for Asp, Glu, and Cys, and those
in position Zaa included the standard amino acids except for Cys.
The peptides were assembled on the resin, starting with the Fmoc-
Leu-OH. The Fmoc group was used for temporary protection of
the amino group and it was removed before each coupling step
with 20% (v/v) piperidine solution in N′,N′-dimethylformamide
(DMF). 1-Hydroxybenzotriazole (HOBt) and benzotriazol-1-yl-
oxytripyrrolidinophosphonium hexafluorophosphate (pyBOP) were
used as coupling reagents in the presence of DIEA (N,N-diisopro-
pylethylamine). The resin-attached Leu-Phe-Lys peptide was
synthesized first. In the next step, the isokinetic mixture of 19 amino
acids selected for position Zaa was coupled,33 followed by the His
residue. The resin carrying the library of pentapeptides was divided
into Ser and Thr halves at this stage. Each of these halves was
subdivided into 17 portions, which were subsequently coupled with
individual amino acids selected for position Xaa. The coupling with
Fmoc-Ser(tBu)-OH or Fmoc-Thr(tBu)-OH, as appropriate, and
additions of Fmoc-Ala-OH and Fmoc-Gly-OH residues were
accomplished subsequently. The acetylation of the N-terminus was
performed using acetic anhydrate (Ac2O) in the presence of DIEA.
N-Terminal acetylated resin-attached peptides were cleaved from
the resin with the mixture of TFA, AcOH, and DCM 1:2:7 (v/v)
followed by evaporation. The library design is presented in Scheme
1.
Synthesis of Selected Peptides. The peptides were synthesized
in the solid state according to a standard manual Fmoc protocol
(usually 0.5 mg per each peptide with the loading capacity of 0.5
mmol/g) with the 2.5-fold excess of amino acids and coupling
reagents diluted in DMF/DCM/NMP (1:1:1) and in the presence
of (DIEA) using the AM-RINK (RapPolymere) resin. The pair of
HOBt and O-(benzotriazol-1-yl)-N,N,N′,N′-tetramethyluronium tet-
rafluoroborate (TBTU) were used as coupling reagents. The
coupling reactions were performed in the glass vessels with filters.
The Fmoc protection group was removed from growing peptide
chains with 20% (v/v) piperidine in DMF followed by a thorough
wash with DMF (4×), DCM/DMF (4×), and DCM (4×) in order
to remove the excess of piperidine. The progress of amino acid
couplings was monitored by the p-chloranil test. At the end of chain
elongation, the N-terminal amine was acetylated with acetic
Sc ) 24/tm
(1)
According to eq 1, a given peptide was assigned a score (Sc) of
12, if a peak corresponding to its predicted hydrolysis product was
detected at 2 h, and scores of 6, 4, and 3 were assigned for such
detection at 4, 6, and 8 h, respectively. A score of 0 was assigned
if no trace of the product of hydrolysis was seen after 8 h of
incubation. The data collected at 24 h were not used for the analysis,
because the crowding of the spectra made assignments problematic.
Equation 1 reflects in a simplified form the pseudo-first-order of
the hydrolysis reaction, seen previously for the CH3CO-Thr-Glu-
Thr-His-His-Lys-NH2 peptide.31 For the most active peptides, for
which the formation of Ser/Thr-Xaa-His-Zaa-R2 products was seen
already at 2 h, the progress of reaction was also controlled at longer
incubation times. The decrease of signals of substrates was
controlled in these cases, as well.
HPLC Measurements of Hydrolysis Rates. The hydrolysis
rates were measured individually for seven selected peptides, R1-
SRHW-R2, R1-SKHW-R2, R1-SAHW-R2, R1-SRHA-R2, R1-SGHA-
R2, R1-TRHW-R2, and R1-THHW-R2, at 37 and 45 °C. The samples
contained 1 mM peptide and 1 mM Ni(II) in 20 mM Tris buffer,
pH 8.2. The 10 µL aliquots were withdrawn at variable incubation
intervals, adjusted individually to reaction times. These aliquots
(32) Guy, C. A.; Fields, G. B. Methods Enzymol. 1997, 289, 67–83.
(33) Ostresh, J. M.; Winkle, J. H.; Hamashin, V. T.; Houghten, R. A.
Biopolymers 1994, 34, 1681–1689.
9
J. AM. CHEM. SOC. VOL. 132, NO. 10, 2010 3357

A R T I C L E S
Kre¸z˙el et al.
Table 1. Hydrolysis Scores and Xaa and Zaa Residue Rankings, Obtained by MALDI-TOF Screening of the Library of 1 mM
R1-Ser-Xaa-His-Zaa-Lys-R2 Peptides Incubated with 2 mM Ni(II) at 45 °C for 8 h in 10 mM Tris Buffer, pH 8.2
were mixed with 50 µL of 2% TFA at room temperature to quench
hydrolysis and stored at 4 °C until the HPLC analysis. The samples
were analyzed on a Breeze HPLC system equipped with an
analytical C-18 column (ACE, 250 × 4.6 mm). The substrates and
products were separated with elution buffers A (0.1% TFA in water)
and B (90% acetonitrile, 0.1% TFA in water), using the gradient
of 10-40% buffer B in 15 min, at flow rate of 1 mL/min, and
detecting at 220 nm. The peaks were identified on a Q-Tof1 ESI-
MS instrument. Pseudo-first-order kinetic constants were calculated
using Microcal Origin v. 8.0 (OriginLab).
Statistical Calculations. The library kinetic data were catego-
rized according to the Sc values, which roughly describe the rates
of peptide bond hydrolysis for individual library members. Initially,
the relationship between Xaa and Zaa contributions into the overall
hydrolysis rate was tested separately for each of Tables 1-4 of Sc
values (Ser and Thr sets, each at 37 and 45 °C). The zero hypothesis
claimed independence of contributions of Xaa and Zaa, which
would mean an absence of spatial and electronic interactions
between these two residues. In order to test this hypothesis,
semiempirical ranks were constructed as products of individual
contributions of Xaa and Zaa residues. A positive verification of
the hypothesis was obtained using a standard approach based on
the Pearson’s ꢀ2-test.34 The natural logarithms of rate constants
obtained in this way for all individual peptides within a given library
were then used as smoothed estimates of the free energy of reaction
barrier. These kinetic data were subjected to a standard principal
components analysis (PCA) procedure, performed iteratively.35
Various physical and chemical parameters determined for amino
acid residues were included in the initial trial set of parameters
characterizing the contributions of individual residues (see Results
for details).36,37 In order to balance statistical contributions of
individual parameters, all these data were normalized in the PCA
analysis according to eq 2.
xi - µ(x)
σ(x)
˜
x¯i )
(2)
where µ(x) and σ2(x) are the mean and variance of parameter x.
After each cycle of calculations the parameter contributing the least
to two main eigenvectors corresponding to the largest eigenvalues
was purged from the calculations. This was performed to identify
various sets of X and/or Z residues that differ in their contribution
to the estimated values of effective reaction rates. The PCA analysis
was continued until the major parameter, the previously estimated
reaction rate, was pointed as the less significant one among the
parameters used to distinguish between the peptide clusters.
Otherwise, the next steps of parameter reduction would yield a
simple clustering of individual residues according to their literature
properties, which would not pertain to the analysis of reaction rates.
Additionally, after each PCA iteration, the scatterplots built with
(35) Jolliffe, I. T. In Principal Component Analysis; Springer Series in
Statistics; Springer: Heidelberg 2002.
(36) Carugo, O. In Silico Biol. 2003, 3, 417–428.
(37) IUPAC Stability Constants Database, V. 4.74, IUPAC and Academic
Software, 2009.
(34) Pearson, K. Mathematical Contributions to the Theory of EVolution.
XIII. On the Theory of Contingency and Its Relation to Association
and Normal Correlation; Draper’s Company Research Memoirs,
Biometric Series; Cambridge University Press: London, 1904.
9
3358 J. AM. CHEM. SOC. VOL. 132, NO. 10, 2010

Library Selection for Peptide Bond Hydrolysis
A R T I C L E S
Table 2. Hydrolysis Scores and Xaa and Zaa Residue Rankings, Obtained by MALDI-TOF Screening of the Library of 1 mM
R1-Ser-Xaa-His-Zaa-Lys-R2 Peptides Incubated with 2 mM Ni(II) at 37 °C for 8 h in 10 mM Tris Buffer, pH 8.2
the two main eigenvectors were subjected to a cluster analysis
procedure performed with the aid of the K-means algorithm.38
Molecular Modeling. The structural calculations were performed
for the model 4N square-planar Ni(II) complex of the R1-SRHW-
R2 peptide using X-PLOR.39,40 Original X-PLOR “allhdg” param-
etrization was modified to model the coordination environment of
the Ni(II) ion. The lengths of Ni(II)-nitrogen bonds were set to
1.85 and 1.90 Å for backbone amide and histidine side-chain
nitrogen, respectively. The complex was kept planar using ap-
propriate improper constraints both for Ni(II) and nitrogen atoms:
four nitrogen atoms were kept in the sp2 hybridization in square
planar arrangement around the Ni(II).
The initial complex structure was subjected the simulated
annealing protocol (SA), consisting of 6 ps high-temperature
evolution and 3 ps cooling down to 300 K, followed by 10 000
steps of energy minimization. This was repeated 999 times, and
the 30 lowest energy conformers were chosen as a representative
sample of the conformational space. The ensemble of conformers
was analyzed with the aid of MolMol.42
amino acids (except Asp, Glu, and Cys), and Zaa included 19
common R-amino acids (except Cys). For the clarity of
description we shall use the “Ser library” and “Thr library” terms
for the whole sets of R1-Ser-Xaa-His-Zaa-R2 and R1-Thr-Xaa-
His-Zaa-Lys-R2 peptides, respectively. The peptides were
synthesized in 34 batches, each containing a mixture (sublibrary)
of 19 peptides randomized at Zaa, with known Ser/Thr and Xaa
residues. Peptide sequences R1 and R2 were designed so that
the masses of substrates and their expected variable C-terminal
hydrolysis products, (Ser/Thr)-Xaa-His-Zaa-R2, were in the
range of 700-1200, optimal for the detection by the MALDI-
TOF spectrometer as +1 ions. The molecular mass of the lightest
expected C-terminal product of the Ser library was 744 Da, and
that of the heaviest substrate of the Thr library was 1187 Da.
The spread of masses within clusters of substrates and C-
terminal products was 129 Da, equal to the difference between
the lightest and the heaviest Zaa substituents, Gly and Trp.
Therefore, the constant mass separation between the substrates
and these products, equal to 170 Da (CH3CO-Gly-Ala-OH),
assured a good separation and, consequently, easy identification
of signals of reaction products. The presence of the Lys residue
in the invariable R2 tail of the C-terminal hydrolysis products
assured the sufficient positive ionization of all library members.
Results
Principles of Library Design and Screening. The peptide
libraries designed for this study had a general sequence R1-
(Ser/Thr)-Xaa-His-Zaa-Lys-R2. Xaa included 17 common Ra-
(38) Hartigan, J. A.; Wong, M. A. Appl. Stat. 1979, 28, 100–108.
(39) Nilges, M.; Clore, G. M.; Gronenborn, A. M. FEBS Lett. 1988, 239,
129–136.
(40) Nilges, M.; Kuszewski, J.; Brunger, A. T. In Computational Aspects
of the Study of Biological Macromolecules by NMR; Hoch, J. C., Ed.;
Plenum Press: New York, 1991.
(41) Bal, W.; Djuran, M. I.; Margerum, D. W.; Gray, E. T., Jr.; Mazid,
M. A.; Tom, R. T.; Nieboer, E.; Sadler, P. J. J. Chem. Soc., Chem.
Commun. 1994, 1889–1890.
(42) Koradi, R.; Billeter, M.; Wu¨thrich, K. J. Mol. Graphics 1996, 14, 51–
55.
9
J. AM. CHEM. SOC. VOL. 132, NO. 10, 2010 3359

A R T I C L E S
Kre¸z˙el et al.
Table 3. Hydrolysis Scores and Xaa and Zaa Residue Rankings, Obtained by MALDI-TOF Screening of the Library of 1 mM
R1-Thr-Xaa-His-Zaa-Lys-R2 Peptides Incubated with 2 mM Ni(II) at 45 °C for 8 h in 10 mM Tris Buffer, pH 8.2
The omission of Asp and Glu residues at position Xaa was
dictated by the fact that they contain carboxylates in their side
chains. Such carboxylates may bind the Ni(II) ion in a way
which would quench or slow down the hydrolysis reaction.43
The Cys residue was eliminated because it was likely to provide
alternative Ni(II) binding in either of variable positions, resulting
in unwanted coordination heterogeneity. Cysteine peptides are
also prone to free-radical redox processes in the presence of
Ni(II) ions, which might result in oxidative side reactions.44-46
The temperatures of 37 and 45 °C and pH of the reaction of
8.2 were chosen on the basis of our previous studies as relatively
mild and at the same time potentially selective for highly active
sequences.28-31 In particular, our previous study of the CH3CO-
Thr-Glu-Thr-His-His-Lys-NH2 peptide indicated that its Ni(II)-
dependent hydrolysis was effective at pH values higher than
8.5 and temperatures higher than 50 °C.31 Therefore, the
detection of substantial amounts of reaction products during
library screening would prove the presence of peptides much
more susceptible to Ni(II)-dependent hydrolysis than those
known to date.
observed. Tables 1 and 2 present the Sc values (defined in the
Experimental Section), obtained for the Ser library at 45 and
37 °C, respectively. Tables 3 and 4 show the respective values
for the Thr library. The residues in these tables were provision-
ally sorted in rows (Xaa, in short X) and columns (Zaa, in short
Z) according to the van der Waals volumes of their side chains
and octanol/water partition coefficients, respectively. One can
see that these parameters provided a good, but not a perfect,
ordering of hydrolytic properties.
On the a priori assumption of the additivity of Xaa and Zaa
contributions, the Sc values were summed up in lines and
columns of Tables 1-4, thereby creating cumulative scores for
substituents in these positions, XSc and ZSc, respectively.
Individual peptide scores, ISc, were calculated by adding XSc
and ZSc of its substituents. The Ser library provided a higher
number of reactive peptides than the Thr library, and the
hydrolysis was faster at 45 °C than at 37 °C for both cases.
The sums of all Sc values within these individual experiments
are presented in Table 5 to illustrate these tendencies. Figure 2
presents correlations of XSc and ZSc values between Ser and
Thr libraries. The correlation of XSc values was weak, with R
) 0.62, while that of ZSc values was strong, with R ) 0.93.
Measurements of Hydrolysis Rates for Individual Pep-
tides. Seven peptides were chosen for detailed measurements
of rates of Ni(II)-related hydrolysis and their temperature
dependence. This selection included five Ser library peptides:
two of them with top-scoring Xaa and Zaa substituents, R1-
SRHW-R2 and R1-SKHW-R2; two with one top-scoring and one
bottom-scoring substituent, R1-SAHW-R2 and R1-SRHA-R2; and
one peptide without top-scoring substituents, R1-SGHA-R2. This
Library Screening and Evaluation. Figure 1 shows a MALDI-
TOF mass spectra of a representative sublibrary (R1-SRHZ-
R2) at 0 and 24 h of incubation at 45 °C. Only +1 ions were
(43) Kozłowski, H.; Lebkiri, A.; Onindo, C. O.; Pettit, L. D.; Galey, J.-F.
Polyhedron 1995, 14, 211–218.
(44) Cherifi, K.; Decock-Le Reverend, B.; Varnagy, K.; Kiss, T.; Sovago,
I.; Loucheux, C.; Kozłowski, H. J. Inorg. Biochem. 1990, 38, 69–80.
(45) Van Horn, J. D.; Bulaj, G.; Goldenberg, D. P.; Burrows, C. J. J. Biol.
Inorg. Chem. 2003, 8, 601–610.
(46) Kre¸z˙el, A.; Szczepanik, W.; Sokołowska, M.; Jez˙owska-Bojczuk, M.;
Bal, W. Chem. Res. Toxicol. 2003, 16, 855–864.
9
3360 J. AM. CHEM. SOC. VOL. 132, NO. 10, 2010

Library Selection for Peptide Bond Hydrolysis
A R T I C L E S
Table 4. Hydrolysis Scores and Xaa and Zaa Residue Rankings, Obtained by MALDI-TOF Screening of the Library of 1 mM
R1-Thr-Xaa-His-Zaa-Lys-R2 Peptides Incubated with 2 mM Ni(II) at 37 °C for 8 h in 10 mM Tris Buffer, pH 8.2
set was complemented with two top-scoring Thr library peptides,
R1-THHW-R2 and R1-TRHW-R2. The Ni(II)-dependent hy-
drolysis of these peptides was studied in a detailed, quantitative
fashion under the conditions analogous to those used in library
screening. HPLC was used to separate and quantitate reaction
substrates and products. Examples of chromatograms are
Figure 1. Example of MALDI-TOF spectra of peptide libraries. (Top) MALDI-TOF spectrum of the R1-SRHZ-R2 sublibrary, incubated at 45 °C at 0 h
(immediately after Ni(II) addition). (Bottom) MALDI-TOF spectrum of the same sublibrary, incubated at 45 °C for 24 h. Red and green boxes mark the
spectral regions of substrates and products, respectively. Blue dotted lines mark the positions of R1-SRHW-R2 (substrate) and SRHW-R2 (product) peptides.
Asterisks mark major impurities.
9
J. AM. CHEM. SOC. VOL. 132, NO. 10, 2010 3361

A R T I C L E S
Kre¸z˙el et al.
Scestimate values are also provided in Table 6 to show that
correct Sc values were reproduced in most cases. This agreement
validated the library scoring methodology and prompted a
conclusive statistical analysis of the library data.
Statistical Analysis of Library Scores. The analysis of the
categorized set of individual scores determined for the Ser
library at 45 °C was performed first, because this experiment
provided the highest number of nonzero Sc entries. This analysis
was based on an assumption that the contributions of Xaa and
Zaa residues are mutually independent. This assumption was
justified by the Pearson ꢀ2-test. This approach enabled us to
generate a set of semiexperimental scores. These scores were
compared to experimental rate constants. Figure 6 demonstrates
an excellent linear correlation between these parameters for all
five Ser library peptides. The slope of the correlation, 2.5 (
0.3, is due to a 2-fold excess of Ni(II) over the mixture of
peptides used in library experiments, while only a 20% excess
of Ni(II) over the peptide was applied in detailed studies.
Semiempirical scores validated in this way were then
subjected to the PCA procedure. Partial molar volumes, mo-
lecular volumes, and various polarity coefficients of amino acid
residues were included in the initial trial set of parameters
characterizing the contributions of individual residues.36 Step-
wise reductions of this initial set in foregoing cycles of PCA
analysis yielded models that included partial molar volumes of
residues Xaa and Zaa (XV2O and ZV2O, respectively), and free
energies of the amino acid side-chain vaporization of residues
Xaa and Zaa (Xvap and Zvap, respectively).
Figure 2. Correlations of cumulative scores at positions X (XSc) and Z
(ZSc) between Ser and Thr peptide libraries. Trend lines (red), 95%
confidence bands (blue), and linear correlation coefficients R are indicated
in the plots.
Table 5. Sums of Library Score (Sc) Values for Individual Library
Experiments
library
45 °C
37 °C
Ser
Thr
850
367
505
160
presented in Figure 3. Individual peaks were identified by ESI-
MS. A substantial amount of a transient reaction product was
observed in some, but not all, cases. Its retention time was
always between those of the substrate and the final product,
while its molecular mass and MS isotopic profile were always
identical with those of its substrate. The final reaction product
corresponded to the removal of the N-terminal CH3CO-Gly-
Ala-OH dipeptide in all cases, as expected. This dipeptide was
not detected in HPLC as a separate peak. Due to its low size it
eluted within the dead volume of the column. No peptide bond
hydrolysis was observed in control samples, incubated in the
absence of Ni(II) ions.
The reaction rates were obtained by fitting the normalized
integrals of substrate and C-terminal product peaks to the first-
order rate equation. The intermediate products was automatically
summed up with their substrates during the MALDI-TOF
scoring of peptide libraries, since their masses were identical.
That approach was reproduced in these calculations, as il-
lustrated in Figure 4. As mentioned above, the CH3CO-Gly-
Ala-OH product was not observed in HPLC analysis; thus, this
hydrolysis product was not included in the calculations. Despite
that, the substrate and product integrals were sufficiently similar
for all peptides studied. This can be explained by the fact that
the CH3CO-Gly-Ala-OH molecule absorbs light only marginally
at the detection wavelength of 220 nm.
The peptides were then bifurcated into two groups according
to the electronic properties of the Zaa residue. At this stage,
the pKa value of the R-amino group of the free amino acid (ZpK
)
a
was included as an additional parameter.37 This parameter
significantly improved the prediction of the peptide hydrolytic
activity. The detailed analysis of the PCA data of the Ser library
proved that the clustering of the peptides was related to the
type of Zaa residue. The peptides carrying a Trp, Asp, or Glu
residue at Zaa belonged to the minor cluster of highly active
peptides, while all other peptides built the major cluster. The
peptides from the major cluster were then finally subjected to
the generalized linear regression of the PCA-derived parameters
against the estimated rate constants. In the Thr library, another
subset of highly active peptides could be discerned according
to the parametrization obtained for the Ser library. It contained
peptides with His, Lys, and Arg residues as Xaa. The separation
of the Zaa ) Trp set from the major cluster was much smaller
(note that Zaa ) Asp and Glu yielded null ZSc values for this
experiment, and therefore, they were excluded from the
analysis). Table 7 presents regression results, which demonstrate
a good correlation between the predicted and the semiexperi-
mental values. Minor clusters could not be analyzed in the same
way, because their low sizes did not satisfy formal requirements
of the linear regression algorithm for the minimal number of
independent variables.
The rate constants obtained from the above calculations are
presented in Table 6. Figure 5 compares them with ISc values
for respective peptides of Ser library. A positive correlation
between these two measures of susceptibility to Ni(II)-related
hydrolysis is obvious. These constants can also be used to re-
estimate Sc values, on an assumption, based on inspection of
noise levels in MALDI-TOF spectra, that the threshold of
reaction product detection in these spectra corresponded to ca.
20% of product formation. This condition relates Scestimate values
with t1/5 of first-order reactions according to eq 3:
Figure 7 shows the regression plots for both Ser and Thr
libraries, arranged according to the major clusters (circles). The
Z ) Trp (and Glu, Asp for the Ser library) peptides are marked
with squares, and the X ) His, Lys, Arg peptides are marked
with triangles. These subsets stay out for the Ser and Thr
libraries, respectively, as explained above. Within the major
clusters, the largest deviations were observed for the peptides
with Z ) Thr, Ser, but this seems to be an effect of specific
interactions of the hydroxyl group, rather than a systematic
deviation. The analysis of the parameters selected by PCA
Scestimate ) 24/t1/5 [h]
(3)
9
3362 J. AM. CHEM. SOC. VOL. 132, NO. 10, 2010

Library Selection for Peptide Bond Hydrolysis
A R T I C L E S
Figure 3. HPLC chromatograms of 1 mM R1-SRHW-R2 peptide incubated in 20 mM Tris buffer, pH 8.2 at 45 °C, in the presence (right) or absence (left)
of 1.2 mM Ni(II) ions, recorded at incubation times t0 ) 0 min (before incubation), t1 ) 93 min, t2 ) 221, and t3 ) 384 min. Peak labels: S, R1-SRHW-R2
reaction substrate; IP, intermediate product; P, SRHW-R2 reaction product.
Table 6. Comparison of Experimental Rate Constants (k1) of
Hydrolysis of Seven Selected Peptides with Their Library Score
(Sc) Valuesa
37 °C
45 °C
k1
k1
peptide
(s-1 × 10-5
)
Sc
Scestimate
(s-1 × 10-5
)
Sc
Scestimate
R1-SRHW-R2
R1-SKHW-R2
R1-SAHW-R2
R1-SRHA-R2
R1-SGHA-R2
R1-TRHW-R2
R1-THHW-R2
4.7(2)
3.7(3)
2.0(5)
1.1(1)
0.35(7)
1.5(5)
2.1(3)
12 18.0 [12]
12 14.3 [12]
8.9(6)
4.5(5)
2.4(3)
1.5(1)
0.87(8)
1.3(3)
3.9(8)
12 34.4 [12]
12 23.2 [12]
6
3
0
7.6 [6]
4.2 [3-4]
1.4 [0]
6
6
0
9.4 [6]
5.9 [4-6]
3.4 [3-0]
12 5.7 [4-6]
12 7.9 [6]
12 5.0 [4-6]
12 15.2 [12]
a The values in parentheses denote standard errors (SE) on the last
significant digit of k1 values. Scestimate values were obtained by
application of eq 3 to these values. The values in brackets in Scestimate
columns result from the rounding of estimates to the nearest
experimentally possible Sc values (12, 6, 4, 3, or 0).
Figure 4. Procedure of calculations of rate constants from HPLC data,
presented for the R1-SKHW-R2 peptide incubated at 45 °C. (A) relative
peak area of the substrate (down triangles), the intermediate product (up
triangles), and the SKHW-R2 reaction product (hexagons). (B) The result
of summing up the substrate with the intermediate product (squares). (C)
The first-order rate law fit (line) of the summed up data of panel B (circles).
with the trend line. This parametrization overestimates the Thr
major cluster peptides’ activities approximately 5-fold, thus
indicating that the Thr methyl group counteracts the reaction
for Xaa * His, Lys, Arg. This parametrization also overestimates
the Zaa ) Trp (Asp, Glu) contributions in both libraries, by a
factor of ca. 3 for the most active peptides of this group, which
are located in the top right corner of the plots, and up to nearly
40-fold for the least active ones. Therefore, Trp (Asp, Glu)
residues may be involved in cooperative interactions that
destabilize the active complex for the majority of Xaa
substitutions.
proved that bulky and hydrophobic residues improve the
efficiency of hydrolysis at both Xaa and Zaa positions (Table
7). This effect is ca. 3 times stronger for Zaa than for Xaa.
Additionally, the high pKa of the Zaa residue contributed
positively to the reaction rate, while the electrostatic effect of
the Xaa residue was negligible.
The parametrization optimized for the major cluster of the
Thr library was close to that obtained for the Ser library. The
only significant difference concerned the effect of the size of
the Xaa residue, found to be 3 times stronger than that observed
in the Ser library. This is certainly due to a possibility of direct
interactions between Xaa side chains and the Thr methyl group.
However, looking at the original scores of the Thr sublibrary,
one can immediately see that the dominant contribution to the
activity of these peptides is confined to the Xaa ) His, Lys,
Arg subset (cumulative Sc of 264 per total 367 and all Sc ) 12
entries). For these three peptides, the Ser major cluster
parametrization is appropriate. This is illustrated in Figure 7
Discussion
Validation of Study Concept. As shown above, certain specific
combinations of substitutions in positions Xaa and Zaa of the
general sequence R1-Ser/Thr-Xaa-His-Zaa-R2 resulted in a
dramatic increase of the rate of hydrolysis of the Ala-Ser/Thr
peptide bond in the presence of Ni(II) ions, while preserving
the sequence specificity of the reaction. Among 646 peptides
tested by library screening at 45 °C, only 32 were highly active
(Sc ) 12), while 413 were inactive (Sc ) 0). This proportion
was even more striking at more selective 37 °C: 6 highly active
9
J. AM. CHEM. SOC. VOL. 132, NO. 10, 2010 3363

A R T I C L E S
Kre¸z˙el et al.
Figure 7. The regression plots of natural logarithms of predicted (Table
7) versus semiempirical reaction rate constants for libraries at 45 °C.
Symbols and lines are explained in the graph. Ser library trend lines are
projected onto the Thr library to demonstrate special properties of X ) H,
K, R peptides of the latter. The circle marks the X ) H, K, R; Z ) W
peptides.
Figure 5. Linear correlations between ISc values and accurate experimental
rate constants, determined at 37 and 45 °C for selected R1-SXHZ-R2 peptides
(labeled using their SXHZ portions). Trend lines (red), 95% confidence
bands (blue), and linear correlation coefficients R are indicated in the plots.
Figure 8. Structural sketch of the proposed hydrolytically reactive NiH-2
L
complex of a library peptide L. Green and violet clouds indicate areas of
influence of Xaa (X) and Zaa (Z) residues, respectively.
donors (4N structure).47 This structure, presented schematically
in Figure 8, is analogous to those well-known for serum
albumins and related peptides, including products of Ni(II)-
related hydrolysis characterized by us before.28-31,41,48-51 The
only difference is that it includes an amide rather than amine
nitrogen, provided by Ser/Thr residue.
The structure in Figure 8 is consistent with the following
results of statistical analysis and empirical observations: (1) The
4N structure separates Xaa and Zaa residues spatially, which
accounts for the independence and additivity of Xaa and Zaa
contributions. (2) The 4N structure cannot be formed with Xaa
) Pro.47-52 In fact, all Sc values were 0 for Xaa ) Pro in all
library experiments (Tables 1-4). (3) The Xaa side chain in
the 4N structure is adjacent to the Ser/Thr side chain.48 This
explains the much higher contribution of Xaa residue to the
linear regression over the Thr library, compared with the Ser
library (Table 7); the same effect manifested itself in a poor
correlation of XSc parameters between Ser and Thr libraries
(Figure 2). (4) The side chain of the Zaa residue has a spatial
freedom to interact at the bottom of the Ni(II) chelate structure
(the top is defined by Ser/Thr and Xaa side chains).48 This
Figure 6. A linear correlation between accurate experimental rate constants
and semiempirical library rate constants at 45 °C for selected R1-SXHZ-R2
peptides (labeled using their SXHZ portions). Trend lines (red), 95%
confidence bands (blue), linear correlation coefficient R, and slope of the
fitted line are indicated in the plot.
vs 481 inactive (Tables 1-4). Among peptides tested individu-
ally, the enhancement of the hydrolysis rate for R1-SRHW-R2,
compared to R1-SGHA-R2, was 10-fold at 45 °C and higher
than 13-fold at 37 °C (Table 6). Therefore, the cornerstone
hypothesis of the study was tested positively. Moreover, the
results of this study contribute significantly to a prospective
mechanism of the reaction studied and pave the way for further
optimization of reaction conditions.
Structure-Reactivity Relationships. All peptides that under-
went hydrolysis in the presence of Ni(II) ions reacted according
to the same pattern: only the peptide bond preceding the Ser/
Thr residue was hydrolyzed. This fact indicates that residues
Xaa and Zaa do not participate directly in the reaction, instead,
their participation has a sterical character. Our previous study
on the CH3CO-Thr-Glu-Thr-His-His-Lys-NH2 peptide31 indi-
cated that the hydrolysis was enabled by the formation of a
characteristic, square-planar complex with the hydrolysis sub-
strate, which contains the Ni(II) ion coordinated to four nitrogen
(47) Kozłowski, H.; Bal, W.; Dyba, M.; Kowalik-Jankowska, T. Coord.
Chem. ReV. 1999, 184, 319–346.
(48) Bal, W.; Chmurny, G. N.; Hilton, B. D.; Sadler, P. J.; Tucker, A.
J. Am. Chem. Soc. 1996, 118, 4727–4728.
(49) Bal, W.; Wo´jcik, J.; Maciejczyk, M.; Grochowski, P.; Kasprzak, K. S.
Chem. Res. Toxicol. 2000, 13, 823–830.
(50) Bal, W.; Christodoulou, J.; Sadler, P. J.; Tucker, A. J. Inorg. Biochem.
1998, 70, 33–39.
(51) Sokołowska, M.; Kre¸z˙el, A.; Dyba, M.; Szewczuk, Z.; Bal, W. Eur.
J. Biochem. 2002, 269, 1323–1331.
(52) Bal, W.; Kozłowski, H.; Pettit, L. D.; Robbins, R. Inorg. Chim. Acta
1995, 231, 7–12.
9
3364 J. AM. CHEM. SOC. VOL. 132, NO. 10, 2010

Library Selection for Peptide Bond Hydrolysis
A R T I C L E S
Table 7. Linear Regression Parameters for Ser and Thr Libraries at 45 °Ca
Ser library
Zaa * Trp, Glu, Asp
Thr library
Z * Trpb; Xaa * His, Lys, Argc
parameter
value
95% CI limits
t-stat
value
95% CI limits
t-stat
free
ZV2O
-39(2)
0.039(1)
3.8(2)
-0.100(10)
0.006(1)
-0.055(7)
[-43; -36]
[0.036; 0.042]
[3.4; 4.2]
[-0.121; -0.080]
[0.004; 0.009]
[-0.068; -0.041]
-21.7
27.3
20.1
-9.6
4.9
-31(2)
0.035(2)
2.7(2)
-0.084(13)
0.021(2)
-0.035(9)
[-36; -27]
[0.031; 0.038]
[2.2; 3.1]
[-0.110; -0.059]
[0.017; 0.025]
[-0.053; -0.017]
-14.0
19.5
11.4
-6.5
10.4
ZpK
a
Zvap
XV2O
Xvap
-7.9
-3.8
a Values in parentheses denote standard errors (SE) on the last significant digit. b Glu and Asp were excluded from the analysis because they yielded
null cumulative scores. See the text for more details. c Hydrolytic properties of these peptides agree with the parametrization estimated for the major
cluster in the Ser library.
Figure 9. Orthographic projections of the ensemble of 30 lowest-energy structures of the proposed hydrolytically reactive NiH-2L complex of the R1-
SRHW-R2 peptide. The Ser residue is in red, Arg in green, His in blue, and Trp in violet, while the Ni(II) is in gold and the four N-Ni(II) bonds are in cyan.
Other residues are invisible for the sake of clarity.
property agrees well with a very high correlation of ZSc
parameters between Ser and Thr libraries (Figure 2). (5) The
Pro residue in position Zaa cannot bend back toward the Ni(II)
chelate plane due to its rigidity. Accordingly, all Sc values were
0 for this substitution in all library experiments (Tables 1-4).
The above concepts are further corroborated by results of
molecular modeling of the 4N (NiH-2L) complex of the R1-
SRHW-R2 peptide, presented in Figure 9. The planar arrange-
ment of nitrogen ligands, enforced by the diamagnetic Ni(II)
ion, separates in space the side chains of residues X (here Arg,
green) and Z (here Trp, violet). The X residue has a freedom
of rotation above the complex plane, while the less restricted Z
residue moves freely aside and below the plane.
volumes and low free energies of the side-chain vaporization.
These parameters define quantitatively the terms of bulkiness
and hydrophobicity, respectively, so the hydrolysis is enhanced
by molecular crowding around the Ni(II) ion. Molecular
crowding enhances the stability of the 4N complex, according
to the mechanism of shielding of Ni(II)-N bonds.48-53 There-
fore, the rate enhancement at pH 8.2 may be at least partially
due to the increase of 4N complex stability and therefore its
ability to facilitate hydrolysis.
The pKa of the amino acid is an indirect measure of electronic,
rather than sterical properties of the side chain. Trp was found
to be the most effective Zaa substitution. The calculations
indicated, however, that it underperforms, compared to linear
regression predictions, as well as Asp and Glu do. Perhaps, these
The PCA and linear regression calculations (Table 7)
demonstrated that hydrolysis rates were enhanced for Xaa and
Zaa substituents characterized with both high partial molar
(53) Raycheba, J. M.; Margerum, D. Inorg. Chem. 1980, 19, 837–843.
9
J. AM. CHEM. SOC. VOL. 132, NO. 10, 2010 3365

A R T I C L E S
Kre¸z˙el et al.
residues can participate in some additional electrostatic interac-
tions that destabilize the active form of the peptide.
in the general sequence R1-(Ser/Thr)-Xaa-His-Zaa-R2 provide
peptides that are particularly susceptible to specific hydrolysis
of the R1-(Ser/Thr) peptide bond in the presence of Ni(II)
ions. We also found that this susceptibility is related to the
bulkiness and hydrophobicity of Xaa and Zaa substituents.
These results provide a firm basis for subsequent studies
aimed at elucidating the molecular mechanism of Ni(II)-
dependent hydrolysis and developing this reaction for bio-
technological applications.
Perspectives of Further Studies. Above, we presented the
structural concept which rationalizes the relationships between
sequences of peptides and their susceptibility to Ni(II) hydroly-
sis. In a future paper (J. Am. Chem. Soc., submitted), we will
verify it by investigating the relationship between the Ni(II)
complexation properties of seven selected peptides (Table 6)
and their hydrolytic reactivity. We will also research the optimal
conditions of the reaction in the perspective of its application
for protein engineering.
Acknowledgment. This study was financed by Polish Ministry
of Science and Higher Education, Grant No. PBZ-MIN-007/P04/
01.
Conclusions
We proved the validity of the study concept, by demon-
strating that some combinations of Xaa and Zaa substitutions
JA907567R
9
3366 J. AM. CHEM. SOC. VOL. 132, NO. 10, 2010