Conformational change and epimerization of diketopiperazines containing proline residue in water

Article abstract of DOI:10.1248/cpb.c17-00164

In water, diketopiperazines cyclo(L-Pro-L-Xxx) and cyclo(L-Pro-D-Xxx) (Xxx-Phe, Tyr) formed an intramolecular hydrophobic interaction between the main skeleton part and their benzene ring, and both cyclo(L-Pro-L-Xxx) and cyclo(L-Pro-D-Xxx) took a folded conformation. The conformational changes from folded to extended conformation by addition of several deuterated organic solvents (acetone-d₆, metanold₄, dimethyl sulfoxide-d₆ (DMSO-d₆)) and the temperature rise were investigated using ¹H-NMR spectra. The results suggested that the intrarmolecular hydrophobic interaction of cyclo(L-Pro-D-Xxx) formed more strongtly than that of cyclo(L-Pro-L-Xxx). Under a basic condition of 1.0×10^-1 mol/L potassium deuteroxide, enolization of O₁-C₁-C₉-H₉ moiety of cyclo(L-Pro-L-Xxx) occurred, while that of the O₄-C₄-C₃-H₃ moiety did not. Cyclo(L-Pro-L-Xxx) epimerized to cyclo(D-Pro-L-Xxx), while cyclo(L-Pro-D-Xxx) did not change.

Full text of DOI:10.1248/cpb.c17-00164

598
Chem. Pharm. Bull. 65, 598–602 (2017)
Vol. 65, No. 6
Note
Conformational Change and Epimerization of Diketopiperazines
Containing Proline Residue in Water
Takashi Ishizu,* Hiroyuki Tsutsumi, Emi Yokoyama, Haruka Kawamoto, and Runa Yokota
Faculty of Pharmacy and Pharmaceutical Sciences, Fukuyama University; Sanzo Gakuen-cho 1, Fukuyama,
Hiroshima 729–0292, Japan.
Received February 22, 2017; accepted March 28, 2017
In water, diketopiperazines cyclo(L-Pro-L-Xxx) and cyclo(L-Pro-D-Xxx) (Xxx=Phe, Tyr) formed an
intramolecular hydrophobic interaction between the main skeleton part and their benzene ring, and both
cyclo(L-Pro-L-Xxx) and cyclo(L-Pro-D-Xxx) took a folded conformation. The conformational changes from
folded to extended conformation by addition of several deuterated organic solvents (acetone-d6, metanol-
1
d4, dimethyl sulfoxide-d6 (DMSO-d6)) and the temperature rise were investigated using H-NMR spectra.
The results suggested that the intrarmolecular hydrophobic interaction of cyclo(L-Pro-D-Xxx) formed more
strongtly than that of cyclo(L-Pro-L-Xxx). Under a basic condition of 1.0ꢀ10ꢁ1 mol/L potassium deuteroxide,
enolization of O1-C1-C9-H9 moiety of cyclo(L-Pro-L-Xxx) occurred, while that of the O4-C4-C3-H3 moiety did
not. Cyclo(L-Pro-L-Xxx) epimerized to cyclo(D-Pro-L-Xxx), while cyclo(L-Pro-D-Xxx) did not change.
1
Key words diketopiperazine; conformation; epimerization; intramolecular hydrophobic interaction; H-NMR
spectrum
The human immunodeficiency virus (HIV) protease selec- was inserted into 5mmϕNMR tube containing a diketopipera-
tively cleaves the peptide bonds L-Phe-L-Pro and L-Tyr-L-Pro. zine and 260µLH2O.
There is no mammalian protease that cleaves such peptide
X-Ray Crystal Structure Analysis of Cyclo(L-Pro-
bonds, and it is a feature of retroviral protease. It is possible L-Tyr) Cyclo(L-Pro-L-Tyr) was dissolved in ethyl acetate.
to design anti-HIV drugs with high selectivity and few side Colorless needle crystals were obtained at room temperature
effects if this feature is utilized. Based on this concept, vari- after 1d. A crystal of the cyclo(L-Pro-L-Tyr) was determined
ous HIV protease inhibitors incorporating substrate transition by X-ray crystallographic analysis at 213K. The X-ray inten-
state analogues that mimicked the L-Phe-L-Pro of the HIV sity data of 50502 reflections (of which 9783 were unique)
protease substrate peptide sequence have been developed.1,2)
were collected on a Rigaku RAXIS RAPID II imaging plate
We have studied the basic properties of the L-Phe-L-Pro area detector with graphite monochromated Cu-Kα radia-
and L-Tyr-L-Pro moieties using a cyclic octapeptide cyclo tion (λ=1.54187Å). The data were corrected for Lorentz and
(L-Pro-L-Phe)4. As a result, it was found that cyclo (L-Pro- polarization effects. The structure was solved by direct
L-Phe)4 took C2-symmetric comformation containing two cis methods using SIR20115) and expanded using Fourier tech-
peptide bonds in CDCl3 and metanol-d4.3,4) However, since niques.6) The non-hydrogen atoms were refined anisotropi-
cyclo (L-Pro-L-Phe)4 was insoluble in water, we decided to cally. Hydrogen atoms were refined using the riding model.
investigate the properties of water soluble diketopiperazines The final cycle of full-matrix least-squares refinement on F2
cyclo(L-Pro-L-Phe) and cyclo(L-Pro-L-Tyr) in water (Fig. 1). For was based on 9741 observed reflections and 689 variable pa-
comparison, the property of diketopiperazines cyclo(L-Pro- rameters and converged with unweighted and weighted agree-
D-Phe) and cyclo(L-Pro-D-Tyr) was also examined (Fig. 1).
ment factors of R=Σ||Fo|−|Fc||/Σ|Fo|=0.0673 (I>2.00σ(I)),
In this paper, the conformations, conformational changes, Rw=[Σ(w(Fo2−Fc2)2)/Σ w(Fo2)2]1/2=0.2155. The goodness
and epimerization of cyclo(L-Pro-Xxx) (Xxx=Phe, Tyr) in of fit was 1.02. Unit weights were used. The maximum and
water are reported.
minimum peaks on the final difference Fourier map corre-
sponded to 0.54 and −0.22 e/Å3, respectively. All calculations
were performed using the CrystalStructure7) crystallographic
Experimental
NMR Experiments 1H-NMR spectra were recorded at software package except for refinement, which was performed
room temperature on a JEOL JMN-LA500 (Tokyo, Japan) using SHELXL97.8) Crystallographic data reported in this
operating at 500MHz. D2O was used as a solvent (99.9 atom% manuscript have been deposited with the Cambridge Crys-
D; Wako Pure Chemical Industries, Ltd., Osaka, Japan).
Chemical shift values are expressed in ppm downfield using
sodium 2,2-dimethyl-2-silapentane-5-sulfonate (DSS) as an
internal standard and all measurement temperature are 35°C.
Also the NMR measurerment under basic condition using
potassium deuteroxide (KOD) was performed at 35°C. The
nuclear Overhauser effect (NOE) difference experiments were
typically conducted with 32K data points covering a spectral
width of 10000Hz and with ca. 5s presaturation time. In mea-
surement using H2O, 3mmϕNMR tube containing 180µL D2O
Fig. 1. Cyclo(Pro-Phe) and Cyclo(Pro-Tyr)
*To whom correspondence should be addressed. e-mail: ishizu@fupharm.fukuyama-u.ac.jp
© 2017 The Pharmaceutical Society of Japan

Vol. 65, No. 6 (2017)
Chem. Pharm. Bull.
599
tallographic Data Center as supplementary publication No. bic interactions were formed between the main skeleton of
1530756.
cyclo(L-Pro-L-Xxx) and cyclo(L-Pro-D-Xxx) and their benzene
ring, and both cyclo(L-Pro-L-Xxx) and cyclo(L-Pro-D-Xxx)
took a folded conformation in D2O (Fig. 2). Sano et al. report-
Results and Discussion
Conformation of Cyclo(L-Pro-Phe) and Cyclo(L-Pro-Tyr) ed the intramolecular CH–π interaction of diketopiperazines
1
in Water In the H-NMR spectrum in D2O, the proton sig- bearing a benzyl group in dimethyl sulfoxide-d6 (DMSO-d6)
nal for H8α of cyclo(L-Pro-L-Xxx) (Xxx=Phe, Tyr) at 0.764, and CDCl3.9–11)
0.768ppm was observed in a higher magnetic field than that
of cyclo(L-Pro-D-Xxx) (Xxx=Phe, Tyr) at 1.639, 1.646ppm,
and the proton signal for H9 of cyclo(L-Pro-D-Xxx) at 2.471,
2.478ppm was observed in a higher magnetic field than that of
cyclo(L-Pro-L-Xxx) at 4.056, 4.034ppm, respectively.
The crystal structures of cyclo(L-Pro-L-Phe),12) cyclo(L-Pro-
NOEs were observed between the proton signal for H8α
and the proton signals for the benzene ring in of cyclo(L-Pro-
L-Xxx), and between the proton signal for H9, H8β and the
proton signals for the benzene ring of cyclo(L-Pro-D-Xxx),
suggesting that the upfield shift in the proton signals for H8α
of cyclo(L-Pro-L-Xxx) and H9 of cyclo(L-Pro-D-Xxx) resulted
from the magnetic anisotropic shielding by the ring current
from the benzene ring.
Fig. 2. Stereochemical
Cyclo(L-Pro-D-Phe) Black Both Arrows Indicate NOE
Structures
of
Cyclo(L-Pro-L-Phe)
and
Judging from the above findings, intramolecular hydropho-
Fig. 3. Crystal Structures of Cyclo(Pro-Phe) and Cyclo(Pro-Tyr)
Fig. 4. Conformational Changes in Cyclo(L-Pro-L-Xxx) and Cyclo(L-Pro-D-Xxx) (Xxx=Phe, Tyr)

600
Chem. Pharm. Bull.
Vol. 65, No. 6 (2017)
Fig. 6. Ratio of the Extended Conformation of Cyclo(L-Pro-L-Xxx) and
Cyclo(L-Pro-D-Xxx) (Xxx=Phe, Tyr) with the Temperature Rise
Fig. 7. Chemical Shift Changes in the 1H-NMR Spectrum of
Cyclo(L-Pro-L-Xxx) and Cyclo(L-Pro-D-Xxx) (Xxx=Phe, Tyr) with the
Increase in the KOD Concentration
H8α of cyclo(L-Pro-D-Xxx) was used as that of the cyclo(L-Pro-
L-Xxx) (Xxx=Phe, Tyr) when they took the extented confor-
mation in D2O, and the chemical shift value (4.056, 4.034ppm)
of the proton signal for H9 of cyclo(L-Pro-L-Xxx) was used as
that of the cyclo(L-Pro-D-Xxx) when they takes extented con-
formation in D2O, respectively.
Figure 5 shows the ratio of the extended conformation of
cyclo(L-Pro-L-Xxx) and cyclo(L-Pro-D-Xxx) (Xxx=Phe, Tyr)
versus the concentration of the deuterated organic solvent
(acetone-d6, metanol-d4, DMSO-d6). The ratio of the extended
conformation of cyclo(L-Pro-L-Xxx) and cyclo(L-Pro-D-Xxx)
Fig. 5. Ratio of the Extended Conformation of Cyclo(L-Pro-L-Xxx) and
Cyclo(L-Pro-D-Xxx) (Xxx=Phe, Tyr) with the Increase in (a) Acetone-d6,
(b) Metanol-d4, and (c) DMSO-d6
D-Phe),13) and cyclo(L-Pro-L-Tyr) were investigated by X-ray was estimated by a shift value of the chemical shifts of
crystallographic analysis. As shown in Fig. 3, cyclo(L-Pro- the proton signal for H8α of cyclo(L-Pro-L-Xxx) and H9 of
L-Phe) took extended conformation, on the other hand cyclo(L-Pro-D-Xxx), respectively.
cyclo(L-Pro-D-Phe) and cyclo(L-Pro-L-Tyr) took folded confor-
mation.
The ratio of the extended conformation of cyclo(L-Pro-
L-Xxx) increased more predominantly than that of cyclo(L-Pro-
Conformational Change by Addition of an Organic Sol- D-Xxx) with increasing concentration of deuterated organic
vents and a Temperature Rise The conformational change solvents. In acetone-d6, cyclo(L-Pro-L-Xxx) (Xxx=Phe, Tyr)
of cyclo(L-Pro-L-Xxx) and cyclo(L-Pro-D-Xxx) (Xxx=Phe, Tyr) reached extended conformation at a ratio of 99.6, 100%, on
from folded to extended conformation was investigated by ad- the other hand, cyclo(L-Pro-D-Xxx) (Xxx=Phe, Tyr) reached
dition of deuterated organic solvent into D2O (Fig. 4).
extended conformation at a ratio of only 26.9, 25.5%, as
Changes in chemical shift values of the proton signal for shown in Fig. 5a.
H8α of cyclo(L-Pro-L-Xxx) and the proton signal for H9 of
cyclo(L-Pro-D-Xxx) in the H-NMR spectrum were measured cyclo(L-Pro-L-Xxx) and cyclo(L-Pro-D-Xxx) to temperature
Nextly, the stability of the hydrophobic interaction of
1
while increasing the ratio of deuterated organic solvents rises was investigated. Figure 6 shows a ratio of the extended
(acetone-d6, metanol-d4, DMSO-d6) against D2O. Here, the conformation of cyclo(L-Pro-L-Xxx) and cyclo(L-Pro-D-Xxx)
chemical shift value (1.639, 1.646ppm) of the proton signal for (Xxx=Phe, Tyr) versus temperature change from 25°C to

Vol. 65, No. 6 (2017)
Chem. Pharm. Bull.
601
Fig. 8. 1H-NMR Spectra
1
All H-NMR measurement temperature are 35°C. (a) Cyclo(L-Pro-L-Phe) in D2O. (b) Cyclo(L-Pro-D-Phe) in D2O. (c) Cyclo(L-Pro-L-Phe) 1.0×10−1 mol/L KOH in H2O. (d)
Cyclo(L-Pro-L-Phe) 1.0×10−1 mol/L KOD in D2O. (e) Cyclo(L-Pro-D-Phe) 1.0×10−1 mol/L KOD in D2O.
Fig. 9. Enolization and Epimerization of Cyclo(L-Pro-L-Xxx) (Xxx=Phe, Tyr)
80°C. The ratio of the extended conformation of cyclo(L-Pro- was investigated in the KOD concentration range from
L-Xxx) increased more predominantly than that of cyclo(L-Pro- 1.0×10−6 to 1.0×10−1 mol/L using the proton signal for H8α in
D-Xxx) with rising temperature. No significant chemical shift 1H-NMR spectrum.
1
value was observed other than the shift values of the chemical
No change in the H-NMR spectrum of cyclo(L-Pro-D-Xxx)
shifts of proton signals for H8α of cyclo(L-Pro-L-Xxx) and H9 was observed in the range from 1.0×10−6 to 1.0×10−1 mol / L,
1
of cyclo(L-Pro-D-Xxx), suggesting that the conformational and no change in the H-NMR spectrum of cyclo(L-Pro-L-Xxx)
change from folded to extended conformation occurred due to was observed in the range from 1.0×10−6 to 1.0×10−2 mol / L.
1
the temperature rise.
At 1.0×10−1 mol / L H-NMR, the signal of cyclo(L-Pro-L-Xxx)
Therefore, the results of the conformational change by ad- changed to that of cyclo(L-Pro-D-Xxx) or cyclo(D-Pro-L-Xxx)
dition of organic solvents and a temperature rise suggested (Figs. 7, 8).
1
that the intramolecular hydrophobic interaction of cyclo(L-Pro-
In the H-NMR spectrum of both cyclo(L-Pro-L-Xxx) and
D-Xxx) bound more strongly than that of cyclo(L-Pro-L-Xxx). cyclo(L-Pro-D-Xxx) in D2O the proton signal of H9 was dis-
It might be because in aqueous solution structure, the part appered at 1.0×10−1 mol/L KOD and the proton signal of H3
that the main skeleton and benzene ring of cyclo(L-Pro-D-Xxx) remained, indicating that the enolization of the O1-C1-C9-H9
(Xxx=Phe, Tyr) overlapped each other was larger than the moiety of both cyclo(L-Pro-L-Xxx) and cyclo(L-Pro-D-Xxx) oc-
part that those of cyclo(L-Pro-L-Xxx) (Xxx=Phe, Tyr) over- curred, while that of the O4-C4-C3-H3 moiety of them did not
lapped.
Epimerization of Cyclo(L-Pro-L-Phe) and Cyclo(L-Pro- L-Xxx), while cyclo(L-Pro-D-Xxx) did not change.
L-Tyr) under Basic Conditions The isomerization of Generally speaking, racemization occurs as a result of
cyclo(L-Pro-L-Xxx) and cyclo(L-Pro-D-Xxx) (Xxx=Phe, Tyr) enolization. Actually, epimerization from cyclo(L-Pro-L-Xxx)
occur (Fig. 8). Cyclo(L-Pro-L-Xxx) changed to cyclo(D-Pro-

602
Chem. Pharm. Bull.
Vol. 65, No. 6 (2017)
K., Chem. Pharm. Bull., 41, 2029–2031 (1993).
to cyclo(D-Pro-L-Xxx) occurred due to the enolization of the
O1-C1-C9-H9 moiety (Fig. 9), while cyclo(L-Pro-D-Xxx) was not
epimerized. Since cyclo(L-Pro-D-Xxx) and cyclo(D-Pro-L-Xxx)
are enantiomer each other, it was thought that cyclo(D-Pro-
L-Xxx) formed a strong intramolecular hydrophobic interac-
tion between the main skeleton part and the benzene ring
similar to cyclo(L-Pro-D-Xxx). Therefore, it is considered that
cyclo(L-Pro-L-Xxx) may be epimerized to cyclo(D-Pro-L-Xxx)
in order to obtain a stronger intramolecular hydrophobic in-
teraction.
4) Ishizu T., Noguchi S., Chem. Pharm. Bull., 46, 1303–1307 (1998).
5) SIR2011: Burla M. C., Caliandro R., Camalli M., Carrozzini B.,
Cascarano G. L., Giacovazzo C., Mallamo M., Mazzone A., Polidori
G., Spagna R., J. Appl. Cryst., 45, 357–361 (2012).
6) DIRDIF99: Beurskens P. T., Admiraal G., Beurskens G., Bosman
W. P., De Gelder R., Israel R., Smits J. M. M., “The DIRDIF-99
program system, Technical Report of the Crystallography Labora-
tory,” University of Nijmegen, The Netherlands, 1999.
7) CrystalStructure 4.1: Crystal Structure Analysis Package, Rigaku
Corporation 2000–2014, Tokyo 196–8666, Japan.
8) SHELX97:Sheldrick G. M., Acta Crystallogr. A, 64, 112–122
(2008).
9) Sano S., Nakao M., Takeyasu M., Kitaike S., Yoshioka Y., Nagao
Y., Heterocycles, 79, 781–789 (2009).
Conflict of Interest The authors declare no conflict of
interest.
10) Nakao M., Toriuchi Y., Fukayama S., Sano S., Chem. Lett., 43,
340–342 (2014).
11) Sano S., Nakao M., Heterocycles, 91, 1349–1375 (2015).
12) Ramani R., Venkatesan K., Marsh R. E., Hu Kung W. J., Acta Crys-
tallogr. B, 32, 1051–1056 (1976).
Supplementary Materials The online version of this ar-
ticle contains supplementary materials.
References
1) Kiso Y., J. Synth. Org. Chem. Jpn., 52, 403–412 (1994).
2) Hamada Y., Kiso Y., Kagaku to Seibutsu, 41, 796–803 (2003).
3) Ishizu T., Hirayama J., Noguchi S., Iwamoto H., Hirose J., Hiromi
13) Mazza F., Lucente G., Pinnen F., Zanotti G., Acta Crystallogr. C,
40, 1974–1976 (1984).