Carboproteins: A 4-α-helix bundle protein model assembled on a D-galactopyranoside template

Article abstract of DOI:10.1016/S0960-894X(01)00035-X

We have recently introduced the concept of monosaccharides as templates for de novo design of protein models and described the synthesis of a model 'carbopeptide'. Here, we report the synthesis of a 64 amino acid (AA) 'carboprotein' by chemoselective ligation of a C-terminal hexadecapeptide aldehyde to a tetra-aminooxy functionalized methyl α-D-galactopyranoside (D-Galp) template. Biophysical characterizations by CD spectroscopy and NMR amide H-D exchange experiments indicated that the four-stranded carboprotein forms a 4-α-helix bundle structure.

Full text of DOI:10.1016/S0960-894X(01)00035-X

Bioorganic & Medicinal Chemistry Letters 11 (2001) 697±700
Carboproteins: A 4-ꢀ-Helix Bundle Protein Model Assembled
on a D-Galactopyranoside Template
Jesper Brask and Knud J. Jensen*
Department of Organic Chemistry, Technical University of Denmark, Building 201, Kemitorvet, DK-2800 Kgs. Lyngby, Denmark
Received 7 November 2000; revised 14 December 2000; accepted 10 January 2001
AbstractÐWe have recently introduced the concept of monosaccharides as templates for de novo design of protein models and
described the synthesis of a model `carbopeptide'. Here, we report the synthesis of a 64 amino acid (AA) `carboprotein' by che-
moselective ligation of a C-terminal hexadecapeptide aldehyde to a tetra-aminooxy functionalized methyl a-d-galactopyranoside
(d-Galp) template. Biophysical characterizations by CD spectroscopy and NMR amide H±D exchange experiments indicated that
the four-stranded carboprotein forms a 4-a-helix bundle structure. # 2001 Elsevier Science Ltd. All rights reserved.
Introduction
drug targeting,15 or to allow molecular sensing.16 How-
ever, using the primary and secondary hydroxyls of
mono- or disaccharides should provide a more ¯exible
control of the directionality and distances between
anchoring points for the peptide chains in de novo design
of protein models. We employed a methyl a-d-Galp
derivative in which the non-anomeric hydroxyls are in
or above a plane de®ned by C-2, C-3, C-5, and O-5, and
suggested the terms `carbopeptide' or `carboprotein' for
the chimeric constructs, depending on their size and
degree of folding. Aminooxyacetyl (Aoa) functionaliza-
tion of the template allowed a convergent strategy for
carbopeptide synthesis, in which C-terminal peptide
aldehydes were coupled by oxime17 formation.12 Here
we report an application of this strategy for the synthe-
sis of a single-stranded 16 AA carbopeptide (1), a four-
stranded 48 AA carbopeptide (2), and a four-stranded
64 AA carboprotein (3). We further present CD and
NMR biophysical studies of the three structures to
evaluate the stabilization provided by the carbohydrate
templates.
De novo protein design has emerged as a valuable tool
to critically test our understanding of protein folding
and structure.1 A design approach relying on the
anchoring of peptide strands to a molecular template
has been proposed by Mutter and co-workers2 and later
explored also by others, for example, the groups of
Sasaki,3 DeGrado,4 Ghadiri,5 and Goodman.6 The
main advantage of this totally synthetic strategy is a
bypassing of the problematic folding step of linear
polypeptides, thereby allowing more freedom in the
sequence design for the construction of small structures
with native-like folding. The 4-a-helix bundle, consist-
ing of four amphiphilic helices forming a hydrophobic
core, has been a target for a number of de novo
designed proteins. In recent contributions, templated 4-
a-helix bundle constructs have been synthesized on
peptide templates7,8 or aromatic derivatives.9,10
In previous reports, we introduced carbohydrates as
templates for de novo design of protein models based
on their multifunctionality, the relative rigidity of pyr-
anose rings, the availability of epimeric forms, and their
well-described chemistry.11,12 Several groups have used
the anchoring of peptide chains through the modi®ed
O-6 of cyclodextrin sca€olds as a means to improve the
pharmacological properties of peptides,13,14 to aid in
Synthesis
The previously reported12 methyl 2,3,4,6-tetra-O-Aoa-
a-d-Galp (4) was used as a template for the four-stran-
ded carbopeptide 2 and carboprotein 3. A new template,
methyl 6-O-Aoa-2,3,4-tri-O-acetyl-a-d-Galp (5), was
required for the single-stranded carbopeptide 1 (Scheme
1). Methyl a-d-Galp was tritylated at O-6 and acety-
lated at O-2, O-3, and O-4 in a two-step±one-pot pro-
*Corresponding author. Tel.: +45-4525-2147; fax: +45-4593-3968;
e-mail: okkj@pop.dtu.dk
0960-894X/01/$ - see front matter # 2001 Elsevier Science Ltd. All rights reserved.
PII: S0960-894X(01)00035-X

698
J. Brask, K. J. Jensen / Bioorg. Med. Chem. Lett. 11 (2001) 697±700
cedure. A potential problem in the synthesis is acetyl
migration to O-6 under conditions of acidic trityl
deprotection. However, deprotection with in situ
formed TMS-I18 prevented this problem as less than 2%
migration was observed by NMR. The desired template
5 was then obtained after DIPCDI/DMAP acylation
with Boc2-Aoa-OH, followed by Boc removal with
TFA±CH2Cl2.
The sequence of the peptide aldehyde building blocks
was based on an amphiphilic peptide used by Mutter et
al.19 in the synthesis of a 4-a-helix bundle template-
assembled synthetic protein (TASP), T4-(4a15-Ac). This
should allow a comparison between Mutter's cyclic
peptide and d-Galp derivatives as templates. Both
structures 1 and 3 were assembled from the amphiphilic C-
terminal hexadecapeptide aldehyde Ac-Glu-Ala-Leu-Glu-
Lys-Ala-Leu-Lys-Glu-Ala-Leu-Ala-Lys-Leu-Gly-Gly-H (6).
For compound 2, the truncated dodecapeptide aldehyde
Ac-Lys-Ala-Leu-Lys-Glu-Ala-Leu-Ala-Lys-Leu-Gly-Gly-
H (7) was used. The peptide aldehydes were synthesized
by a BAL strategy, starting with reductive amination of
o-PALdehyde on 0.34 mmol/g Champion I (PEG-
AMPS) resin with aminoacetaldehyde dimethyl ace-
tal.20À22 After coupling of the second residue as (Fmoc-
Gly)2O, the remaining synthesis was performed on an
automated continuous-¯ow peptide synthesizer with 4
equiv of HBTU/HOBt activated Fmoc amino acid in 30
min couplings. Release of the peptide, dimethyl acetal
hydrolysis, and deprotection of side-chain functional-
ities occurred by treatment with TFA/H2O (19:1) for 2 h
(Scheme 2). Puri®cation of the two peptide aldehydes by
prep. HPLC and lyophilization provided 6 in 21% yield
and 7 in 15% yield based on an Fmoc quanti®cation
of the resin-bound dipeptide acetal.23 In both cases
ES±MS analysis showed dominating signals corre-
sponding to [M+3H]3+ and [M±H2O+2H]2+. While
the former signal most likely originated from protona-
tion of the three Lys residues in the sequences, we spec-
ulate that the latter signal resulted from imine
formation between one of these Lys residues and the
aldehyde functionality.
Scheme 1. Synthesis of methyl 6-O-Aoa-2,3,4-tri-O-acetyl-a-d-Galp
(5).
Chemoselective oxime ligation to form single-stranded
carbopeptide 1 was achieved simply by dissolving hexa-
decapeptide aldehyde 6 and 2 equiv of mono-functio-
nalized template 5 in 0.1 M aqueous acetate bu€er at
pH 4.76 (Scheme 3). After 2 h, 1 was isolated by prep.
HPLC in 67% yield after lyophilization.23 By a similar
Scheme 2. Synthesis of C-terminal peptide aldehydes by a BAL strategy.
Scheme 3. Synthesis of templated structures by chemoselective oxime ligation of C-terminal peptide aldehydes to Aoa functionalized methyl a-d-
Galp templates.

J. Brask, K. J. Jensen / Bioorg. Med. Chem. Lett. 11 (2001) 697±700
699
procedure, but now using 1.5 equiv of 6 for each ami-
nooxy functionality in tetra-functionalized template 4,
carboprotein 3 was prepared in 76% yield, and with the
same excess of dodecapeptide aldehyde 7 relative to
template 4, carbopeptide 2 was obtained in 67% yield.
Although analytical HPLC of the puri®ed products
showed somewhat broad peaks, ES±MS indicated pure
compounds with the multiple protonated species
[M+xH]x+ as the only signi®cant signals. The putative
derivatives of 2 and 3 missing one or two peptide
four-stranded carboprotein 3. A pronounced e€ect of
inter-helical stabilization was observed, as 3 in contrast
to 1 displayed a concerted denaturation curve with a
midpoint at 5.1 M GuHCl (Fig. 1, right). From the data
for carboprotein 3 the free energy of folding in water
was calculated according to the linear extrapolation
method, ÁG=ÁGH O+m[GuHCl].27,28 The folding free
2
energy was extrapolated to ÁGH O=À6.9 kcal molÀ1
,
while the slope, re¯ecting the cooperativity of folding,
2
was m=1.4 kcal molÀ1 MÀ1. These results are in accor-
1
strands were not observed. H NMR (500 MHz, D2O)
dance with data obtained for other templated 4-a-helix
showed for all three structures an approximate 1:1 ratio
of Z and E oxime isomers according to the CHˆN tri-
plet signal from the Z isomer at 6.8 ppm and the E iso-
mer at 7.5 ppm. The presence of E/Z oxime isomers
could, in principle, give two4 isomers of each of 2 and 3.
bundles by Sasaki and Kaiser3 (À4.4 kcal molÀ1
;
0.84 kcal molÀ1 MÀ1) and Ghadiri and co-workers5
(À5.6 kcal molÀ1; 1.0 kcal molÀ1 MÀ1), but are not as
high as the extraordinary result recently obtained by
Mezo and Sherman10 with a thiophenyl cavitand tem-
plate (À22.9 kcal molÀ1; 3.1 kcal molÀ1 MÀ1). Mutter
and co-workers19 reported a GuHCl denaturation
curve for T4-(4a15) with a midpoint of 3 M, considerably
less than we observe for the N-terminally acetylated
carboprotein 3.26
Biophysical Characterization
The CD spectra of the three templated structures 1, 2,
and 3 were recorded at 22 ꢀC in 10 mM phosphate bu€er
at pH 7.0.24 All three compounds showed some degree
of a-helicity (Fig. 1, left). Based on [y]222 and the for-
mula proposed by Chen and co-workers,25 the content
of a-helix was calculated to be 45% in single-stranded
carbopeptide 1, 29% in the short-chained four-stranded
2, and 67% in the 64 AA four-stranded carboprotein 3.
Dilution studies, in which the CD spectrum of single-
stranded 1 was recorded at 0.91±91 mM concentrations
indicated weak concentration dependence, with increas-
ing concentration leading to increasing helicity. In
contrast, the CD spectra of four-stranded 3 were con-
centration independent in the concentration range 2.7±
27 mM, i.e., no secondary structural changes were
observed. The CD spectrum of carboprotein 3 resembles
the reported spectrum for Mutter's TASP T4-(4a15-
Ac).19 However, for the shorter TASP T4-(4a11), with a
peptide sequence corresponding26 to our 2, Mutter
and co-workers19 reported a CD spectrum characteristic
of a random coil, with a single minimum around 198
nm.
The structures of four-stranded 3 and the single-stran-
ded reference 1 were further compared by NMR H±D
exchange experiments to evaluate exposure of amide
protons to solvent. The lyophilized samples were dis-
solved in D2O, and the time course of exchange was
followed by 1H NMR spectroscopy. For carboprotein 3,
containing 64 amides, the ®rst spectrum recorded after 5
min showed 35 exchangeable protons.29 After 7 h, a
total of four NH remained, while the last trace of
exchangeable protons disappeared only after 42 h. For
carbopeptide 1, containing 16 amides, the spectrum
after 5 min showed six exchangeable protons, and none
after only 30 min. These results indicate a signi®cant
shielding of amide protons from the solvent in carbo-
protein 3.
In conclusion, carboprotein 3 shows the characteristics
of a 4-a-helix bundle, according to CD spectroscopy
and H±D exchange experiments. Denaturation experi-
ments indicate a higher stability for 3 than reported for
TASP T4-(4a15), with which it shares the peptide
sequence, and the shorter carbopeptide 2 shows a higher
degree of a-helicity than reported for the corresponding
TASP T4-(4a11).19,26 Thus, on the basis of these initial
Denaturation experiments with guanidinium chloride
(GuHCl) were carried out to compare the stabilities of
single-stranded carbopeptide 1 and the corresponding
Figure 1. Left: CD spectra of 1, 2, and 3. Recorded with 91 mM of 1, 25 mM of 2, and 27 mM of 3. Right: GuHCl titration of 1 and 3. Recorded with
91 mM of 1, and 27 mM of 3. Compound numbers are written on the spectra.

700
J. Brask, K. J. Jensen / Bioorg. Med. Chem. Lett. 11 (2001) 697±700
results it appears that the d-Galp template and/or its
linker region might be more apt in inducing 4-a-helix
bundle formation than a cyclic peptide. Further investi-
gations are required to establish whether the structure
resembles that of a molten globule or in fact is native-like.
13. Hristova-Kazmierski, M. K.; Horan, P.; Davis, P.;
Yamamura, H. I.; Kramer, T.; Horvath, R.; Kazmierski,
W. M.; Porreca, F.; Hruby, V. J. Bioorg. Med. Chem. Lett.
1993, 3, 831.
14. Schaschke, N.; Musiol, H.-J.; Assfalg-Machleidt, I.;
Machleidt, W.; Moroder, L. Bioorg. Med. Chem. Lett. 1997, 7,
2507.
15. Pean, C.; Creminon, C.; Wijkhuisen, A.; Grassi, J.; Gue-
È
not, P.; Jehan, P.; Dalbiez, J.-P.; Perly, B.; Djedaõni-Pilard, F.
Acknowledgements
J. Chem. Soc., Perkin Trans. 2 2000, 853.
16. Matsumura, S.; Sakamoto, S.; Ueno, A.; Mihara, H.
Chem. Eur. J. 2000, 6, 1781.
17. Rose, K. J. Am. Chem. Soc. 1994, 116, 30.
18. Klemer, A.; Bieber, M.; Wilbers, H. Liebigs Ann. Chem.
1983, 1416
19. Mutter, M.; Tuchscherer, G. G.; Miller, C.; Altmann, K.-
H.; Carey, R. I.; Wyss, D. F.; Labhardt, A. M.; Rivier, J. E. J.
Am. Chem. Soc. 1992, 114, 1463.
20. Jensen, K. J.; Alsina, J.; Songster, M. F.; Vagner, J.;
Albericio, F.; Barany, G. J. Am. Chem. Soc. 1998, 120, 5441.
21. Alsina, J.; Jensen, K. J.; Albericio, F.; Barany, G. Chem.
Eur. J. 1999, 5, 2787.
The authors thank Ms. Karen Jùrgensen, University of
Copenhagen, for providing the CD spectra, Dr. John
Nielsen, Technical University of Denmark, for the ES±
MS spectra, Professor Arne Holm, Royal Veterinary
and Agricultural University, for donating a MilliGen
9050 PepSynthesizer, and Dr. Derek Hudson, Solid
Phase Sciences, for a gift of Champion resins. Further-
more, we are grateful to the Alfred Benzon Foundation,
the Lundbeck Foundation, the Torkil Holm Founda-
tion, and the Danish Natural Science Research Council
for ®nancial support to KJJ.
22. Guillaumie, F.; Kappel, J. C.; Kelly, N. M.; Barany, G.;
Jensen, K. J. Tetrahedron Lett. 2000, 41, 6131.
23. Yields were determined gravimetrically based on the
assumption that all peptide Lys residues had formed TFA
salts.
24. The mean residue ellipticity was calculated from the for-
mula [y]=y/(10ÂlÂcÂn), where y [mdeg] is the measured
ellipticity, 1 [cm] is the path length, and n is the number of
residues in each helix, not counting the C-terminal glycinal
oxime. The concentration c [mol LÀ1] was calculated from the
mass of lyophilized carbopeptide or -protein used in prepara-
tion of a stock solution, assuming TFA salts as in ref 23.
25. Chen, Y.-H.; Yang, J. T.; Chau, K. H. Biochemistry 1974,
13, 3350.
References
1. DeGrado, W. F.; Summa, C. M.; Pavone, V.; Nastri, F.;
Lombardi, A. Annu. Rev. Biochem. 1999, 68, 779.
2. Mutter, M.; Vuilleumier, S. Angew. Chem., Int. Ed. Engl.
1989, 28, 535.
3. Sasaki, T.; Kaiser, E. T. Biopolymers 1990, 29, 79.
4. Akerfeldt, K. S.; Kim, R. M.; Camac, D.; Groves, J. T.;
Lear, J. D.; DeGrado, W. F. J. Am. Chem. Soc. 1992, 114, 9656.
5. Ghadiri, M. R.; Soares, C.; Choi, C. J. Am. Chem. Soc.
1992, 114, 4000.
6. Goodman, M.; Feng, Y.; Melacini, G.; Taulane, J. P. J.
Am. Chem. Soc. 1996, 118, 5156.
7. Tuchscherer, G.; Grell, D.; Mathieu, M.; Mutter, M. J.
Peptide Res. 1999, 54, 185.
26. TASPs T4-(4a11) and T4-(4a15) were not N-terminally
acetylated. However, it was reported (ref 19) that N-acetyla-
tion, at least for T4-(4a15), only caused minor changes in the
CD spectrum.
8. Rau, H. K.; DeJonge, N.; Haehnel, W. Angew. Chem., Int.
Ed. 2000, 39, 250.
9. Wong, A. K.; Jacobsen, M. P.; Winzor, D. J.; Fairlie, D. P.
J. Am. Chem. Soc. 1998, 120, 3836.
10. Mezo, A. R.; Sherman, J. C. J. Am. Chem. Soc. 1999, 121,
8983.
11. Jensen, K. J.; Barany, G. J. Peptide Res. 2000, 56, 3.
12. Brask, J.; Jensen, K. J. J. Peptide Sci. 2000, 6, 290.
27. Pace, C. N. Methods Enzymol. 1986, 131, 266.
28. The calculations in the transition region 3.2 to 6.1 M
GuHCl were based on the values [y]222=À20,000 deg cm2
dmolÀ1 for folded protein and [y]222=0 deg cm2 dmolÀ1 for
denatured protein.
29. The number of remaining exchangeable protons was cal-
culated from the amide integral, 7.6±8.6 ppm in the case of 3,
normalized to the template methoxy signal at 3.4 ppm.