Design and application of basic amino acids displaying enhanced hydrophobicity

Article abstract of DOI:10.1021/ja029892o

Three noncoding basic amino acids, mono-, di-, and trimethyldiaminopropionic acid (mmdap, dmdap, and tmdap), have been synthesized for use in protein design. Covalent modification of a diaminopropionic acid (dap) side chain with an increasing number of me

Full text of DOI:10.1021/ja029892o

Published on Web 06/04/2003
Design and Application of Basic Amino Acids Displaying
Enhanced Hydrophobicity
Juliana K. Kretsinger and Joel P. Schneider*
Contribution from the Department of Chemistry and Biochemistry, UniVersity of Delaware,
Newark, Delaware 19716-2522
Received December 23, 2002; E-mail: schneijp@udel.edu
Abstract: Three noncoding basic amino acids, mono-, di-, and trimethyldiaminopropionic acid (mmdap,
dmdap, and tmdap), have been synthesized for use in protein design. Covalent modification of a
diaminopropionic acid (dap) side chain with an increasing number of methyl moieties results in a family of
residues displaying short basic side chains with varying degrees of enhanced hydrophobic character. These
residues may be used to introduce charged/polar interactions into the confining hydrophobic interior or
interfacial spaces of proteins. As a demonstration of their utility, the ability of these residues to promote
interior salt bridge formation at the helix/helix interface of GCN4-p1, a dimeric two-stranded coiled coil,
was assessed. Heterodimerization mediated by buried salt bridge formation between a GCN4-based peptide
containing either mmdap, dmdap, or tmdap at position 16 and an analogous peptide containing aspartic
acid at the same position was studied. Mmdap-derived heterodimers are 0.5 kcal/mol more stable than the
corresponding dap-derived heterodimers. This result indicates that the addition of one methyl group to the
dap side chain can stabilize the heterodimeric fold. The stabilization can most likely be attributed to a
decrease in the desolvation penalty incurred upon folding as well as enhanced van der Waals contacts in
the folded state. The addition of three methyl groups to the dap side chain results in heterodimers that are
significantly less stable than the corresponding dap-derived heterodimers, suggesting that increased steric
bulk is not well accommodated in the interior of this protein. Unexpectedly, the addition of two methyl
groups leads to homotrimerization of the dmdap-peptide. The resulting trimer is relatively stable (∆G37°C
°
) 11.8 kcal/mol) and undergoes cooperative thermal unfolding. The GCN4-p1 system exemplifies how
small incremental changes in size and hydrophobicity can alter the folding preferences of a protein. Generally,
this versatile suite of residues can be utilized in any protein and offer new options to the protein chemist.
Ddap, this antibiotic becomes active against gram-negative
bacteria. Gramicidin S is typically active against gram-positive
Introduction
Basic amino acid side chains contain hydrophobic regions
which seem to contrast their polar/charged regions. Lysine, for
example, contains four methylene groups capped by an am-
monium ion at neutral pH. Although it is sometimes convenient
to consider lysine as a hydrophilic residue, the apolar methylene
groups of the residue often participate in favorable hydrophobic
interactions within folded peptides and proteins.1,2 In fact, the
basic residues found in proteins are some of the largest residues,
mostly due to their hydrophobic portions.
Nature has also found use for more compact basic residues
with shortened side chains. For instance, the antituberculosis
compound capreomycin (isolated from Streptomyces capreolus)
and its active analogues contain two diaminopropionic acid (dap)
residues, one of which is suggested to be important for biological
activity.3 Dap is also found in other antibiotics including edeine,
viomycin, and the antitumor antibiotic, bleomycin.4 Interestingly,
when the two Dphe residues of Gramicidin S are mutated to
bacteria, and the change in selectivity is attributed to the
incorporation of these residues into the hydrophilic face of this
amphiphilic peptide.5 Shortened basic residues have also found
use in biophysical studies. For example, dap has been used to
interrogate the snorkeling behavior of positively charged
residues in transmembrane helices,6 and both dap and di-
aminobutyric acid (dab) have been incorporated into solvent
exposed regions of peptides to study helix propensity.2
In protein design, dap and dab residues can be used to
introduce specific polar interactions into the protein matrix. For
example, a buried salt bridge, formed from the association of a
basic and acidic residue, may be incorporated into the hydro-
phobic interior of designed proteins to help specify a desired
fold.7 For this application, the compact dap and dab residues
would seem to be ideal because they are small in size and meet
the charge requirements of participating in the buried polar
interaction. However, these compact residues, lacking the large
(1) Stites, W. E.; Gittis, A. G.; Lattman, E. E.; Shortle, D. J. Mol. Biol.1991,
221, 7-14.
(2) Groebke, K.; Renold, P.; Tsang, K. Y.; Allen, T. J.; McClure, K. F.; Kemp,
D. S. Proc. Natl. Acad. Sci. U.S.A. 1996, 93, 4025-4029.
(3) Nomoto, S.; Shiba, T. Bull. Chem. Soc. Jpn. 1979, 52, 1709-15.
(4) Andruszkiewicz, R. Pol. J. Chem. 1995, 69, 1615-29.
(5) Ando, S.; Aoyagi, H.; Shinagawa, S.; Nishino, N.; Waki, M.; Kato, T.;
Izumiya, N. FEBS Lett. 1983, 161, 89-92.
(6) Liu, F.; Lewis, R. N. A. H.; Hodges, R. S.; McElhaney, R. N. Biochemistry
2002, 41, 9197-9207.
(7) Schneider, J. P.; Lear, J. D.; DeGrado, W. F. J. Am. Chem. Soc. 1997,
119, 5742-5743.
9
10.1021/ja029892o CCC: $25.00 © 2003 American Chemical Society
J. AM. CHEM. SOC. 2003, 125, 7907-7913
7907

A R T I C L E S
Kretsinger and Schneider
Figure 1. Structures of mono-, di-, and trimethylated analogues of
diaminopropionic acid and the calculated transfer free energies (kcal/mol)
for the protonated side chains of each residue for octanol/watera and vacuum/
c
waterb systems. Calculated solvent-accessible area for the protonated Nγ
atom (Å2). Protected analogues 4, 6, and 7 were synthesized for use in
Fmoc-based solid-phase peptide synthesis.
Figure 2. Helical wheel diagram of a parallel heterodimeric coiled coil
showing the positions of a-g residues comprising one heptad repeat.
Intended heterodimer formation is mediated via the buried salt bridge and
may occur between the Asp-p1 peptide and any one of four peptides
containing dap or the designed short basic residues.
number of side-chain methylene moieties of their larger cousins,
are relatively polar. As a result, their incorporation into the
hydrophobic interior of a protein is expected to be energetically
unfavorable.
incorporating a buried salt bridge usually leads to thermo-
dynamic destabilization due to unfavorable desolvation contri-
butions to the free energy incurred upon protein folding. The
charged side chains forming a buried salt bridge must be
desolvated for folding to occur.13,14 Although buried salt bridges
are associated with much higher desolvation costs than are
solvent exposed bridges, buried bridges can compensate via
stronger electrostatic interactions due to reduced solvent screen-
ing.15 In addition to this compensatory effect, the unnatural
residues presented here are designed to minimize the desolvation
penalty due to their increased hydrophobic character.
GCN4-p1 is the 33 residue peptide spanning the leucine
zipper portion of the yeast transcriptional activator protein,
GCN4, and is known to fold as a parallel homodimeric coiled
coil.16 A single polar asparagine (asn) residue at position 16 of
each helix comprising the homodimer specifies this fold by
forming an interhelical hydrogen bond. Mutation of this partially
buried asn16 to valine,17 aminobutyric acid,18 or glutamine19
results in the formation of coiled coils of varying aggregation
number, demonstrating the importance of this buried asparagine
H-bond in specifying a single fold. In addition, it has been
shown that a heterodimer can be specified by incorporating an
interhelical salt bridge, where one helix contains an aspartic
acid residue at position 16 and the other helix contains a dap
residue at the same position, Figure 2. Although this buried salt
bridge specifies a single fold, the resulting heterodimer is about
2 kcal/mol less stable than the wild-type homodimer.7 The
charged asp and dap residues forming the buried salt bridge of
the heterodimer must become largely dehydrated upon folding.
In comparison, wild-type GCN4-p1 only needs to dehydrate
neutral asn residues to fold - an energetically less costly
process.20 Increasing the hydrophobicity of the basic residue
involved in salt bridge formation should decrease this desol-
In this paper, we report the synthesis and application of a
family of noncoding basic residues that have shortened side
chains and display enhanced hydrophobic character, Figure 1.
These residues are analogues of dap in which an increasing
number of methyl moieties are covalently incorporated at the
side-chain nitrogen. Figure 1 shows calculated free energies for
the transfer of each side chain from both octanol8 and vacuum9,10
into water. As the number the of Nγ-methyl groups increases,
the ∆Gt’s also increase, indicating an increase in hydrophobic
character. Because the magnitude of ∆Gt is related to the
solvent-accessible area (SAA) of each atom comprising a given
side chain, the ability to decrease the SAA of the most
hydrophilic atoms in the chain should effect the greatest change
in ∆Gt. Calculated values of SAA for the positively charged
side-chain nitrogen of each residue show that each methyl
substituent is effective in incrementally decreasing the SAA of
the protonated nitrogen atom and that two methyl groups are
sufficient to effectively shield this atom from solvent, Figure
1. In regards to protein design, these residues fill a void in the
existing palette of coding and noncoding residues as they are
designed to fit well into protein interiors where a combination
of charge and hydrophobicity is desired. Dap analogues 4, 6,
and 7 were synthesized in a suitably protected form for facile
incorporation into synthetic peptides via Fmoc-based solid-phase
peptide synthesis (SPPS).
The utility of these residues is demonstrated herein by their
incorporation into the hydrophobic interior of a small protein,
GCN4-p1, to study their ability to support buried salt bridge
formation. Buried salt bridges have been found to be conserved
among members of several protein families, and the degree of
conservation increases with decreasing solvent accessibility.11
The ability of buried polar residues to hydrogen bond and
participate in electrostatic interactions is often important in
defining structural attributes.12 In de novo protein design, the
incorporation of a buried salt bridge can be used to drive the
formation of a desired fold over alternate topologies. However,
(13) Hendsh, Z. S.; Tidor, B. Protein Sci. 1994, 3, 211-226.
(14) Hill, R. B.; Raleigh, D. P.; Lombardi, A.; DeGrado, W. F. Acc. Chem.
Res. 2000, 33, 745-754.
(15) Kumar, S.; Nussinov, R. J. Mol. Biol. 1999, 293, 1241-1255.
(16) Oshea, E. K.; Klemm, J. D.; Kim, P. S.; Alber, T. Science 1991, 254, 539-
544.
(8) Eisenberg, D.; McLachlan, A. D. Nature 1986, 319, 199-203.
(9) Sharp, K. A.; Nicholls, A.; Friedman, R.; Honig, B. Biochemistry 1991,
30, 9686-9697.
(10) Wesson, L.; Eisenberg, D. Protein Sci. 1992, 1, 227-235.
(11) Schueler, O.; Margalit, H. J. Mol. Biol. 1995, 248, 125-135.
(12) McClain, D. L.; Binfet, J. P.; Oakley, M. G. J. Mol. Biol. 2001, 313, 371-
383.
(17) Harbury, P. B.; Zhang, T.; Kim, P. S.; Alber, T. Science 1993, 262, 1401-
1407.
(18) Gonzalez, L.; Brown, R. A.; Richardson, D.; Alber, T. Nat. Struct. Biol.
1996, 3, 1002-1010.
(19) Lino, G.; Woolfson, D. N.; Alber, T. Nat. Struct. Biol. 1996, 3, 1011-
1018.
(20) Hendsch, Z. S.; Tidor, B. Protein Sci. 1999, 8, 1381-1392.
9
7908 J. AM. CHEM. SOC. VOL. 125, NO. 26, 2003

Basic Amino Acids Displaying Enhanced Hydrophobicity
A R T I C L E S
Scheme 1
GCN4-p1 peptide analogues containing substitutions at posi-
tion 16 were prepared via automated SPPS, Figure 2. Residues
4 and 6 behaved ideally during automated SPPS and were
incorporated using standard HBTU/HOBt activation chemistry
affording peptides Mmdap-p1 and Dmdap-p1, respectively.
However, the incorporation of residue 7 via SPPS proved
problematic, and an alternate strategy was employed for the
preparation of the corresponding peptide, Tmdap-p1. This
peptide was most easily prepared by treating resin-bound
Dmdap-p1 with methyl iodide under basic conditions. Because
all of the side chains of Dmdap-p1 were protected, selective
methylation of the dmdap residue was realized. Subsequent
TFA-mediated cleavage from the resin and side-chain depro-
tection afforded Tmdap-p1. Care was taken in handling this
peptide. Under neutral aqueous solution conditions, the unnatural
side chain of Tmdap-p1 was prone to Hofmann elimination,
affording dehydroalanine (half-life ) 17 h at pH 7.0, 20 °C).22
Under mildly acidic conditions (pH 6), elimination occurred at
room temperature with a half-life greater than 3.5 days, and
folding studies could be performed as long as they were
accomplished in a timely manner. Although methylated variants
of amino acids having longer side chains such as ornithine or
lysine have been used in protein studies33,34 and trimethylated
derivatives would not be prone to â-elimination, their larger
size precludes incorporation into most sterically demanding
protein interiors.
Scheme 2
The ability to perform elimination of the tmdap side chain
quickly and cleanly with either base (pH > 8) or elevated
temperatures may be of further interest. Dehydroalanine residues
are found naturally in the active sites of some enzymes,23,24 as
well as in antibiotics of bacterial origin, including nisin25 and
subtilin.26 These residues are additionally useful as electrophiles
in the formation of nonnatural amino acids,27 cyclic peptides,28
and prenylated peptides.29 In principle, tmdap is a synthon for
dehydroalanine and can be used to introduce dehydroalanine
into peptide sequences with exact temporal resolution (e.g.,
initiate elimination by raising the pH or temperature). Further
study is underway to establish the potential advantages of this
methodology.
Folding Attributes of Peptides. Homooligomerization. CD
spectroscopy was used to probe the folding behavior of
individual peptides to establish optimal solution conditions for
studying heterodimerization. Figure 3a shows that Asp-p1,
Mmdap-p1, and Dmdap-p1 adopted helical structure in a pH-
dependent manner and that helical content was maximal at pH
vation penalty. Here, we incorporate each of the dap analogues
shown in Figure 1 into GCN4-p1 at position 16 and study their
ability to form buried salt bridges upon heterodimer formation
with the asp16 containing peptide, asp-p1 (Figure 2).
Results and Discussion
Synthesis. Fmoc-protected mono-, di-, and trimethyldiamino-
propionic acid residues (Figure 1) were all prepared from
â-lactone 1 as outlined in Schemes 1 and 2. Synthesis of the
monomethyl analogue 4 is shown in Scheme 1. Ring opening
of lactone 1 at the â-carbon with N-methyltrimethylsilylamine21
in acetonitrile followed by acidic workup yielded the Z-protected
monomethyl compound 2 in 72% yield. Treatment of 2 with
di-tert-butyl dicarbonate followed by hydrogenation afforded
amine 3 in 81% yield. Near quantitative incorporation of the
Fmoc group yielded NR-Fmoc-Nâ-Boc-protected unnatural
residue 4. The synthesis of residue 6, outlined in Scheme 2,
began with the ring opening of lactone 1 with N,N-dimethyl-
trimethylsilylamine followed by acidic workup, yielding the
Z-protected hydrochloride 5. Fully characterized 5 was normally
hydrogenated directly and Fmoc-protected, affording dimethyl
analogue 6 in 60% yield from 1. Last, the trimethyl analogue 7
was easily obtained via alkylation of 6 with methyl iodide.
(22) Nomoto, S.; Sano, A.; Shiba, T. Tetrahedron Lett. 1979, 20, 521-522.
(23) Schwede, T. F.; Retey, J.; Schulz, G. E. Biochemistry 1999, 38, 5355-
5359.
(24) Langer, B.; Roether, D.; Retey, J. Biochemistry 1997, 36, 10867-10871.
(25) Fukase, K.; Kitazawa, M.; Sano, A.; Shimbo, K.; Fujita, H.; Horimoto, S.;
Wakamiya, T.; Shiba, T. Tetrahedron Lett. 1988, 29, 795-8.
(26) Gross, E.; Kiltz, H. H.; Craig, L. C. Hoppe-Seyler’s Z. Physiol. Chem.
1973, 354, 799-801.
(27) Burkett, B. A.; Chai, C. L. L. Tetrahedron Lett. 1999, 40, 7035-7038.
(28) Okeley, N. M.; Zhu, Y.; van der Donk, W. A. Org. Lett. 2000, 2, 3603-
3606.
(29) Zhu, Y.; van der Donk, W. A. Org. Lett. 2001, 3, 1189-1192.
(30) Boice, J. A.; Dieckmann, G. R.; DeGrado, W. F.; Fairman, R. Biochemistry
1996, 35, 14480-14485.
(31) Pansare, S. V.; Huyer, G.; Arnold, L. D.; Vederas, J. C. Org. Synth. 1992,
70, 1-9.
(32) Kim, J.; Bott, S. G.; Hoffman, D. M. Inorg. Chem. 1998, 37, 3835-3841.
(33) Demmers, J. A. A.; Rijkers, D. T. S.; Haverkamp, J.; Killian, J. A.; Heck,
A. J. R. J. Am. Chem. Soc. 2002, 124, 11191-11198.
(34) Ashfield, J. T.; Meyers, T.; Lowne, D.; Varley, P. G.; Arnold, J. R. P.;
Tan, P.; Yang, J. C.; Czaplewski, L. G.; Dudgeon, T.; Fisher, J. Protein
Sci. 2000, 9, 2047-2053.
(21) Ratemi, E. S.; Vederas, J. C. Tetrahedron Lett. 1994, 35, 7605-7608.
9
J. AM. CHEM. SOC. VOL. 125, NO. 26, 2003 7909

A R T I C L E S
Kretsinger and Schneider
Figure 3. (a) pH-dependent folding as monitored by changes in θ222 for Mmdap-p1, Dmdap-p1, and Asp-p1. Peptide samples are 50 µM in 50 mM buffer,
150 mM NaCl. Buffers are glycine at pH 2 and 3, acetate at pH 4 and 5, MES at pH 6, BTP at pH 7, 8, and 9, and CAPS at pH 10 and 11. (b) CD spectra
of 100 µM solution of Tmdap-p1 at pH 6.0 and 50 µM solutions of Mmdap-p1 at pH 2.0 and 9.0. Buffers as defined in (a).
values in which the side chains of the residues at position 16
were in their nonionized, neutral states. This pH-dependent
behavior is expected as electrostatic repulsion will disfavor
homooligomerization when the side chains at position 16 are
ionized. Under folding conditions, sedimentation equilibrium
analytical ultracentrifugation (SEAU) experiments indicated that
Asp-p1 and Mmdap-p1 formed homodimers and Dmdap-p1
formed a homotrimer (see Supporting Information). Tmdap-p1
is not included in panel a because â-elimination of the tmdap
side chain occurs under basic solution conditions (vide supra).
Knowing that tmdap contains a quaternary amine, it must be
positively charged at all pH’s, and consequently Tmdap-p1
should be monomeric and unstructured. In fact, the CD spectrum
of Tmdap-p1 at pH 6.0 was indicative of random coil, closely
resembling unfolded Mmdap-p1, Figure 3b.
Heterooligomerization. Solution conditions that promote
heterodimerization between Asp-p1 and the base-containing
peptides are those in which the side chains at position 16 are
fully ionized and homooligomerization is disfavored. The data
in Figure 3a suggest that a compromise must be met in choosing
the optimal pH to drive heterodimer formation as no pH value
exists where both the asp and the amine-containing side chain
will be fully ionized. Heterodimerization of Mmdap-p1 with
Asp-p1 was studied at pH 6.0, where the maximal amounts of
each peptide bear charged side chains at position 16. For
consistency, heterodimerization of Tmdap-p1 with Asp-p1 was
also studied at pH 6.
Assessing the ability of the dmdap residue to support buried
salt bridge formation in the context of GCN4-based hetero-
dimerization was not possible. It is apparent from Figure 3a
that the intrinsic pKa of the dmdap side chain was significantly
depressed in comparison to the mmdap side chain. Therefore,
no pH value exists in which homooligomerization was not
favored for either the Asp-p1 or the Dmdap-p1 peptides, and
Figure 4. Mole fraction titration of (a) Mmdap-p1 and (b) Tmdap-p1 with
Asp-p1. Total peptide concentration is 50 µM in 50 mM MES, 150 mM
all attempts to coax heterodimerization failed. An explanation
for the inability of Dmdap-p1 to form heterodimers can be
derived from thermal denaturation experiments (described
below), which indicated that homotrimerization of Dmdap-p1
was a much more favorable process than salt bridge-mediated
heterodimerization. However, this observation does not mean
that the dmdap residue is limited in its ability to form
energetically favorable buried salt bridges. It only means that
GCN4-p1 may not be the optimal model system to study dmdap-
based salt bridge formation; this residue may be ideal for
NaCl, pH 6.0. In panel b, a line is included to show the expected [θ]222
values if no interaction occurred between the two peptides.
incorporation into monomeric proteins or proteins whose
olimerization state is not as sensitive to structural changes as
GCN4-p1.17
Heterodimer formation of Asp-p1 with Mmdap-p1 and
Tmdap-p1 was initially indicated by mole fraction titration
experiments performed at pH 6.0 (Figure 4). In Figure 4a, a
distinct minimum is seen at a 1:1 molar ratio of Mmdap-p1 to
9
7910 J. AM. CHEM. SOC. VOL. 125, NO. 26, 2003

Basic Amino Acids Displaying Enhanced Hydrophobicity
A R T I C L E S
Table 1. Thermodynamic Assessment of Oligomerization
b
NaCl ∆G° (37 °C)a ∆∆G
°
spec
Tm (1 M)c Tm (100 µM)d
peptide
Asp-p1
Dap-p1
Mmdap-p1
Dap-p1:Asp-p1
Mmdap-p1:Asp-1 0.15
Mmdap-p1:Asp-1 0.025
Mmdap-p1:Asp-1 0.5
(M)
(kcal/mol)
(kcal/mol)
(°C)
(°C)
0.15
0.15
0.15
0.15
5.4
3.8
5.1
6.7
7.2
7.1
7.2
11.5
74.9
66.6
75.6
86.6
94.0
93.5
95.9
99.4
31.7
-0.7
28.5
44.2
48.8
49.0
49.7
73.0
2.1
2.0
1.9
2.0
Dmdap-p1
0.15
a ∆G° was determined from thermal denaturation of indicated peptides
(100 µM, pH 6.0, 50 mM MES) monitored by CD at 222 nm (standard
state ) 1 M); the uncertainties in ∆G° and Tm are (0.1 kcal mol-1 and 1
°C, estimated by sensitivity analysis of the theoretical curves. b ∆∆Gspec
°
) ∆G°ab - 1/2[∆G°aa + ∆G°bb]. Tm extrapolated to the standard state (1
M)c and at 100 µM total peptide concentration,d where Tm is the temperature
at which ∆G ) 0.
Figure 5. Thermal denaturations monitored by CD for homo- and
heterogeneous solutions of peptides. Total peptide concentration is 100 µM
in 50 mM Mes, 150 mM NaCl, pH 6.0.
Asp-p1, suggesting heterodimerization. Figure 4b suggests
heterodimer formation for the Tmdap-p1:Asp-p1 pair as well,
although the heterodimer exhibited approximately the same
mean residue ellipticity as the Asp-p1 homodimer. Importantly,
SEAU of 1:1 mixtures of Mmdap-p1:Asp-p1 and Tmdap-p1:
Asp-p1 showed the expected apparent molecular weights for
heterodimer formation (see Supporting Information).
Thermodynamic Assessment. By design, increasing the side-
chain hydrophobicity of the amine containing peptide should
result in the formation of heterodimers exhibiting increased
stability. Figure 5 compares thermal unfolding curves for
equimolar mixtures of the Dap-p1:Asp-p1 and the Mmdap-p1:
Asp-p1 pairs. As compared to Dap-p1:Asp-p1, the unfolding
curve for the Mmdap-p1:Asp-p1 solution shifted to higher
temperatures and exhibited greater helicity at low temperatures.
Qualitatively, this suggests that the Mmdap-p1:Asp-p1 hetero-
dimer enjoyed enhanced thermal stability as a result of
incorporating the additional methyl moiety. Figure 5 also
demonstrates that at pH 6 homogeneous solutions of Mmdap-
p1 and Asp-p1 formed marginally stable homodimers. The data
in Figure 5 as well as concentration-dependent data (see
Supporting Information) were globally fit to the Gibbs-
Helmholtz equation. Equilibrium constants were first determined
for each competing homodimer. Once these values were known,
thermal denaturation data resulting from appropriate heteroge-
neous mixtures were fit to competing equilibria, and heterodimer
equilibrium constants were determined. The results provided
in Table 1 show that, for both the Dap-p1:Asp-p1 and the
Mmdap-p1:Asp-p1 solutions, heterodimerization was pre-
ferred over competing homodimerization by 2.0 kcal/mol
(∆∆Gspecificity°). The degree of specificity was insensitive to
electrolyte concentration, suggesting that the Mmdap-p1:Asp-
p1 salt bridge was sufficiently shielded from solvent, Table 1.
Importantly, at 37 °C and 150 mM NaCl, the Mmdap-p1:Asp-
p1 heterodimer was 0.5 kcal/mol more stable than the corre-
sponding Dap-p1:Asp-p1 dimer (∆G° of 7.2 versus 6.7 kcal/
mol). The incorporation of one methyl group into the amine
side chain resulted in a more stable heterodimer presumably
due to a decrease in the desolvation penalty experienced upon
folding.
Figure 6. Themal denaturations monitored by CD for 100 µM homoge-
neous solutions of Mmdap-p1 and Dmdap-p1 in 50 mM MES, 150 mM
NaCl, pH 6.0.
corresponding heterodimers should exhibit increased stability.
However, this linear extrapolation in thought negates additional
factors important for proper folding such as sterics. In fact, the
incorporation of a dmdap residue, whose side chain bears two
methyl moieties, into the GCN4-p1 scaffold afforded a peptide
(Dmdap-p1) incapable of forming heterodimers with Asp-p1.
Instead, Dmdap-p1 formed homotrimers as evident by SEAU.
This aggregation state can better accommodate the steric bulk
of the additional methyl group. Dmdap-p1-derived homotrimers
underwent a cooperative, reversible two-state thermal unfolding/
folding transition reminiscent of nativelike proteins, Figure 6.
A fit of the unfolding curve invoking a trimer-monomer
equilibrium afforded a ∆G°(37 °C) of 11.8 kcal/mol (1 M
standard state), indicating that the trimer was of moderate
stability.30 Although there are no reports of a GCN4-p1 single
mutant containing the isostructural residue leucine at position
16, one would predict that such a mutant would favor the
formation of trimers based on the observed aggregation state
of Dmdap-p1.
The GCN4-p1 scaffold seems to be rather unyielding in
accommodating steric challenges to its interior. Although
Tmdap-p1 was capable of forming heterodimers with Asp-p1,
the helicity of the resultant heterodimers is only 75% that of
the Mmdap-p1-derived heterodimers, Figure 4. Chemical de-
naturation studies employing guanidine hydrochloride (see
Conceptually, amine side chains bearing multiple methyl
groups should further decrease the desolvation penalty, and their
9
J. AM. CHEM. SOC. VOL. 125, NO. 26, 2003 7911

A R T I C L E S
Kretsinger and Schneider
room temperature, the reaction solution was transferred to a separatory
funnel, washed with 0.1 M citric acid (3 × 30 mL), water (1 × 30
mL), dried (Na2SO4), and concentrated under reduced pressure to yield
an oil. The oil was bathed in hexanes to remove residual (Boc)2O and
dried under vacuum for 1 h. The resulting oil was then dissolved in
150 mL of methanol, and 10% Pd/C (0.3 g) was added. This mixture
was stirred under a balloon of H2 for 1.5 h after which time the reaction
mixture was filtered, concentrated, and the resulting oil triturated with
acetonitrile to afford a white precipitate which was collected by filtration
and dried under vacuum yielding 2.88 g of monomethylamine 3 (13.2
mmol, 81%). 1H NMR (D2O, 400 MHz): δ 3.90 (dd, J ) 7.5 and 4.3
Hz, 1H), 3.79 (m, 1H), 3.65 (dd, J ) 15 and 7.6 Hz, 1H), 2.87 (br d,
3H, cis/trans isomers), 1.44 (s, 9H). 13C NMR (D2O, 100 MHz): δ
172.57, 157.73, 83.12, 54.18, 49.97, 35.03, 27.95. HRMS (ESI) m/z
241.1161 [(M + Na)+, calcd for C9H18N2O4 Na, 241.1164].
Supporting Information) demonstrated that the Tmdap-p1:Asp-
p1 heterodimer is relatively unstable, suggesting that the
trimethylammonium side chain was not easily accommodated
into the protein interior.
Conclusion
Three Fmoc-protected amino acids (4, 6, and 7) have been
prepared, and their potential utility in protein design has been
explored. These mono-, di-, and trimethylated diaminopropionic
acid derivatives were designed such that each additional methyl
moiety covalently incorporated at the side-chain nitrogen serves
to incrementally increase the hydrophobicity of the residue. The
result is a suite of residues whose side chains offer ammonium
ions capable of engaging in polar/charged interactions, but
contradictorily express enhanced hydrophobic character. These
residues may be used to introduce polar functionality into the
interior of designed proteins while simultaneously offering
nonpolar surface area for easier accommodation into hydro-
phobic environments. The utility of these residues has been
explored here via the construction of buried salt bridges into
the coiled coil scaffold of GCN4-p1.
Synthesis of Fmoc-Protected Monomethylamine 4. A 100 mL
round-bottomed flask was sequentially charged with monomethylamine
3 (1.0 g, 4.59 mmol), 12 mL of acetonitrile, 12 mL of water, Fmoc-
OSu (1.56 g, 4.63 mmol), and 2.38 mL of DIEA (13.8 mmol). The
reaction mixture was stirred at room temperature for 5 h, after which
time the solvent was removed by evaporation. The resulting oil was
partitioned between 30 mL of ethyl acetate and 30 mL of 0.1 M citric
acid. The aqueous and organic phases were separated, and the organic
phase was subsequently washed with 0.1 M citric acid (2 × 30 mL),
water (1 × 30 mL), dried (Na2SO4), and concentrated under reduced
pressure to yield 2.0 g of 4 as an oil after drying on high vacuum
Experimental Section
General Methods and Materials. (Z)-Serine â-lactone 1 was
synthesized from (Z)-serine according to the procedure of Pansare et
al.31 N-Methyltrimethylsilylamine was synthesized according to the
procedure of Kim et al.32 Ultrapure Gdn HCl was purchased from ICN.
Palladium on carbon and N-(9H-fluoren-2-ylmethoxycarbonyloxy)-
succinimide (Fmoc-OSu) were purchased from Acros Organics. Fmoc-
protected amino acids, 2-(1H-benzotriazole-1-yl)-1,1,3,3-tetramethyl-
uronium hexafluorophosphate (HBTU), N-hydroxybenzotriazole (HOBT),
piperidine, N-methylpyrrolidone, and amide resin were purchased from
Applied Biosystems. Methyl iodide was filtered through alumina prior
to use. Solvents were purchased from Fisher Scientific and used without
further purification except the acetonitrile and dichloromethane, which
were distilled from calcium hydride. N,N-Dimethyltrimethylsilylamine
and the remaining chemicals were purchased from Aldrich and used
as received.
1
(4.54 mmol, 99%). H NMR ((CD3)2SO, 400 MHz): δ 7.89 (d, J )
7.5, 2H), 7.71 (d, J ) 7.2, 2H), 7.41 (t, J ) 7.4, 2H), 7.32 (t, J ) 7.4,
2H), 4.30-4.20 (m, 4H), 3.65 (m, 1H), 3.37 (bs, 1H), 3.29 (m, 1H),
2.78 (bd, 3H, cis/trans isomers). 13C NMR ((CD3)2SO, 100 MHz): δ
172.24, 156.00, 154.53, 143.77, 140.76, 127.69, 127.11, 125.26, 120.18,
78.85, 65.70, 52.54, 49.41, 46.65, 34.96, 27.96. HRMS (ESI) m/z
463.1832 [(M + Na+), calcd for C24H28NO4Na+ 463.1845].
Synthesis of (Z)-Dimethyl Diaminopropionic Acid 5. Under
nitrogen atmosphere, a dry 100 mL round-bottomed flask was charged
with (Z)-serine â-lactone 1 (1.0 g, 4.5 mmol) and 60 mL of dry
acetonitrile, followed by N,N-dimethyltrimethylsilylamine (0.762 mL,
4.7 mmol) via syringe. The reaction was stirred at room temperature
for 2 h, monitoring the disappearance of 1 (Rf ) 0.68) by TLC (85:15;
CHCl3:MeOH). After completion, the reaction was cooled in an ice
bath, and 100 mL of 0.1 N HCl was added. The resulting solution was
stirred and allowed to warm to room temperature over 30 min,
transferred to a separatory funnel, and washed with CH2Cl2 (3 × 100
mL). The aqueous layer was lyophilized to afford ∼1.2 g of 5 which
normally contains a moderate amount of dimethylamine hydrochloride
salt byproduct. Although the salt was generally taken directly to the
next step, product free from salt can be obtained using a Waters Sep-
Synthesis of (Z)-Monomethyl Diaminopropionic Acid‚HCl 2.
Under nitrogen atmosphere, a dry round-bottomed flask was charged
with N-methyltrimethylsilylamine (3.0 g, 29 mmol) followed by 10
mL of dry acetonitrile. A solution of (Z)-serine â-lactone 1 (5 g, 22.6
mmol) in 9 mL of dry acetonitrile was added via syringe. The resulting
solution was stirred at room temperature for 12 h, subsequently cooled
in an ice bath, and 100 mL of cold 0.1 M HCl was then added to
reduce the pH to 2-3. The resulting solution was stirred and allowed
to warm to room temperature over 30 min, transferred to a separatory
funnel, and washed with CH2Cl2 (4 × 100 mL) and diethyl ether (2 ×
100 mL) to remove undesired carbonyl ring-opened amide. The aqueous
layer was then rotary evaporated, yielding 4.7 g of (Z)-monomethyl
diaminopropionic acid‚HCl 2 (16.3 mmol, 72%) as a light yellow foam.
1H NMR (D2O, 400 MHz): δ 7.43 (m, 5H), 5.12 (s, 2H), 4.30 (dd, J
) 5.6 and 8.2, 1H), 3.42 (dd, J ) 5.4 and 12.9, 1H), 3.24 (dd, J ) 8.5
and 12.8, 1H), 2.73 (s, 3H). 13C NMR (D2O, 100 MHz): δ 171.73,
157.81, 138.05, 129.85, 129.38, 128.27, 68.08, 52.16, 50.67, 45.44,
1
Pak Vac 35 cm3 (10 g) C18 cartridge. H NMR (CD3CN, 400 MHz):
δ 7.3 (m, 5H), 6.89 (d, J ) 6.6 Hz, 1H), 5.08 (s, 2H), 4.39 (m, 1H),
3.34 (d, J ) 7.1 Hz, 2H), 2.77 (s, 6H). 13C NMR (CD3CN, 100 MHz):
δ 173.24, 157.31, 138.09, 129.54, 129.00, 128.84, 67.35, 57.79, 51.21,
43.97. HRMS (ESI) m/z 267.1342 [(M + H+), calcd for C13H19N2O4
+
267.1345].
Synthesis of Fmoc-Dimethyl Diaminopropionic Acid 6. A 250
mL round-bottomed flask containing 1.2 g of crude (Z)-dimethyl
diaminopropionic acid 5 was charged with 225 mL of MeOH and 0.14
g of 10% Pd/C. Hydrogen was bubbled through the solution while it
was stirred for 1 h. The Pd/C was removed by gravity filtration, and
rotary evaporation of the filtrate afforded an oily solid. This oil was
dissolved in 40 mL of 50% acetonitrile/water, and DIEA was added
until a constant pH of 9 was obtained. Next, Fmoc-OSu (1.5 g, 4.4
mmol) was added at once, and the reaction was stirred at room
temperature for 5 h. The resulting solution was washed with diethyl
ether (3 × 40 mL), and the aqueous layer was lyophilized to afford
crude 6 which was purified using a Waters Sep-Pak Vac 35 cm3 (10 g)
C18 cartridge. Crude 6 was loaded onto the Sep-Pak in water as a
34.75. HRMS (ESI) m/z 253.1186 [(M + H+), calcd for C12H17N2O4
+
253.1188].
Synthesis of Monomethylamine 3. Under a nitrogen atmosphere,
a 200 mL round-bottomed flask was charged with Z-protected mono-
methylamine hydrochloride 2 (4.7 g, 16.3 mmol) and 82 mL of dry
methylene chloride. The mixture was stirred, cooled via an ice bath,
and diisopropyl ethylamine (DIEA) (17.0 mL, 97.8 mmol) was added
slowly via pipet. The ice bath was removed, and (Boc)2O (3.73 g, 17.1
mmol) was added at once as a solid. After being stirred for 5 h at
9
7912 J. AM. CHEM. SOC. VOL. 125, NO. 26, 2003

Basic Amino Acids Displaying Enhanced Hydrophobicity
A R T I C L E S
suspension and washed with 15 mL of water. The product was then
eluted with 50% acetonitrile/water and lyophilized to afford 0.96 g (2.7
mmol, 60%) of 6 as a white solid. H NMR (CD3CN, 400 MHz): δ
Synthesis of Tmdap-p1. A dry 5 mL pear-shaped flask equipped
with a stirbar was charged with resin-bound Dmdap-p1 (130 mg; Ld
) 0.127 mmol/g) and 1.5 mL of anhydrous DMF. 1,3,4,6,7,8-
Hexahydo-1-methyl-2H-pyrimido[1,2-a]-pyrimidine (MTBD) (2.86 µL,
19.9 µmol) was added to the swollen resin-bound peptide followed by
methyl iodide (23.6 mg, 166 µmol). The resin slurry was stirred for 4
h under an atmosphere of nitrogen, after which time the resin-bound
peptide was filtered and washed with DMF (3 × 10 mL), DCM (3 ×
10 mL), and dried under vacuum. Treatment of resin bound peptide
with 90% TFA, 5% thioanisole, 3% ethanedithiol, 2% anisole for 2 h
under nitrogen atmosphere affords pure Tmdap-p1 after RP-HPLC
purification (yields determined by HPLC ranged from 50% to 90%
depending on starting material purity). ESI (M)+ calcd 4051.8, obsd
4052.0.
CD Spectroscopy. All experiments were performed on an Aviv
model 62DS circular dichroism spectropolarimeter. Sample concentra-
tions were determined by UV absorbance of the tyrosine residue (∈275
) 1450 cm-1 M-1). Wavelength scans were recorded using a 2 nm
step size and a 15 s average time. Thermal denaturation experiments
were monitored at 222 nm from 2 to 94 °C in 2 °C steps with a 4 min
equilibration time and data averaging for 60 s at each temperature.
Thermodynamic parameters were determined via global fits of thermal
denaturation curves as previously described.7 A Microlab 500 series
automated titrator was used for the guanidine hydrochloride denaturation
experiments; peptide concentration was held constant, while the Gdn
HCl concentration was increased from 0 to 3.5 M in 0.1 M increments,
and then from 3.5 to 7 M in 0.25 M increments. Data were collected
and averaged for 10 s after 1 min stirring and 20 s settling.
1
7.82 (d, J ) 7.6 Hz, 2H), 7.66 (m, 2H), 7.41 (t, J ) 7.4 Hz, 2H), 7.33
(m, 2H), 6.19 (m, 1H), 4.35 (d, J ) 6.7, 2H), 4.23 (t, J ) 6.7, 1H),
4.08 (m, 1H), 3.17 (dd, J ) 5.1 and 12.04, 1H), 2.98 (m, 1H), 2.72 (s,
5H), 2.35 (bs, 1H). 13C NMR (CD3CN, 100 MHz): δ 173.30, 157.54,
145.43, 142.43, 129.06, 128.49, 126.58, 121.33, 67.79, 59.13, 51.05,
48.27, 44.28. HRMS (ESI) m/z 355.1662 [(M + H+), calcd for
+
C20H23N2O4 355.1658].
Synthesis of Fmoc-Trimethyl Diaminopropionic Acid 7. A dry
round-bottomed flask was sequentially charged with 6 (0.79 g, 2.2
mmol), 22 mL of methanol, KHCO3 (267 mg, 2.7 mmol), and methyl
iodide (1.58 g, 11.1 mmol). The reaction mixture was stirred under an
atmosphere of nitrogen at room temperature for 26 h and subsequently
transferred to a separatory funnel where 20 mL of water was added.
The aqueous solution was washed with CH2Cl2 (1 × 20 mL), and the
aqueous fraction was lyophilized to afford 1.23 g of crude 7. The
product was purified by reverse phase HPLC (C18 Vydac peptide/
protein column) employing a linear gradient of solvent B (where solvent
A was composed of water and 0.1% TFA, and solvent B was composed
of 90% acetonitrile, 10% water, and 0.1% TFA) to afford 0.64 g (1.3
mmol) of 7 as a white solid. 1H NMR (CD3CN, 400 MHz): δ 7.80 (d,
J ) 7.5 Hz, 2H), 7.67 (d, J ) 7.5 Hz, 2H), 7.40 (t, J ) 7.4 Hz, 2H),
7.32 (dt, J ) 7.4, 2H), 4.73 (d, J ) 7.2, 2H), 4.61 (dd, J ) 6.2 and
10.76, 1H), 4.53 (dd, J ) 6.0 and 10.75, 1H), 4.25 (t, J ) 6.1, 1H),
4.01 (dd, J ) 2.3 and 13.9, 1H), 3.58 (dd, J ) 8.8 and 13.9, 1H), 3.31
and 3.13 (two singlets are observed due to side-chain conformational
isomerization; total integration 9H). 13C NMR (CD3CN, 100 MHz): δ
170.48, 158.82, 143.76, 140.83, 127.73, 127.13, 125.17, 120.22, 65.70,
65.60, 52.86, 49.47, 46.73. HRMS (ESI) m/z 369.1827 [(M+), calcd
Sedimentation Equilibrium Analytical Ultracentrifugation. SEAU
data were collected at 20 °C on a Beckman XL-I Analytical ultra-
centrifuge with an An-60 Ti rotor. Sample concentrations were
determined by tyrosine absorbance (∈275 ) 1450 cm-1 M-1). Cell
length-dependent absorbance measurements at 275 nm were made at
three rotor speeds (32 000, 37 000, and 42 000 rpm). The resulting data
were fit globally as previously described7 using the software package
Igor Pro (WaveMetrics, Inc.).
+
for C21H25N2O4 369.1814].
Peptide Synthesis. Peptides were synthesized on amide resin using
an ABI 433A peptide synthesizer employing HBTU/HOBT activation.
Peptide N-termini were acetylated using acetic anhydride and DIEA.
Resin-bound peptides were cleaved and deprotected via treatment with
90% TFA, 5% thioanisole, 3% ethanedithiol, 2% anisole for 4 h under
nitrogen atmosphere. Peptides were purified by RP-HPLC (Vydac C4
or C18 peptide/protein column) employing linear gradients of solvent
B. Mass spectroscopy data can be found in the Supporting Information.
The molecular weights reported here are calculated using the multiple
charge states observed from electrospray ionization. Mmdap-p1, ESI
(M) calcd 4023.7, obsd 4023.6; Dmdap-p1, ESI (M) calcd 4037.8, obsd
4038.0; Dap-p1, ESI (M) calcd 4009.7, obsd 4009.5; Asp-p1, ESI (M)
calcd 4038.7, obsd 4038.0.
Acknowledgment. We would like to thank Dan Camac and
Jim Lear for helpful discussions. This work was supported in
part by ACS PRF-G: 36376-G4 (J.P.S.).
Supporting Information Available: Index, mass spectros-
copy, denaturation, and SEAU data for all peptides (PDF). This
material is available free of charge via the Internet at
http://pubs.acs.org.
JA029892O
9
J. AM. CHEM. SOC. VOL. 125, NO. 26, 2003 7913