Solution conformation of a potent cyclic analogue of tuftsin: Low temperature nuclear magnetic resonance study in a cryoprotective mixture

Article abstract of DOI:10.1021/jm980442+

Tuftsin, a linear tetrapeptide (Thr-Lys-Pro-Arg), corresponding to the sequence 289-292 of the heavy chain of leukokinin, has been the object of intensive SAR studies during the past 30 years, owing to its numerous biological activities and to the possibility of generating a novel anticancer drug. A cyclic tuftsin analogue, c-[T-K-P-R-G], has biological activity 50 times higher than that of the parent linear peptide. Here we present a conformational study of c-[T-K-P-R-G] based on NMR data in a cryoprotective DMSO/water mixture. The preferred conformation is a type via turn centered on the K-P residues. The orientation of the side chains of the two basic residues (K and R) may represent the essential feature of the bioactive conformation of tuftsin. A possible role of tuftsin as a DNA binding motif is suggested by the similarity of the bioactive conformation of c-[T-K-P-R-G] and of the β-turn conformation proposed by Suzuki for the [T,S]-P-K-R motif.

Full text of DOI:10.1021/jm980442+

J . Med. Chem. 1999, 42, 1705-1713
1705
Solu tion Con for m a tion of a P oten t Cyclic An a logu e of Tu ftsin : Low -
Tem p er a tu r e Nu clea r Ma gn etic Reson a n ce Stu d y in a Cr yop r otective Mixtu r e
Annamaria D′Ursi,*,† Stefania Albrizio,† Caterina Fattorusso,† Antonio Lavecchia,† Giancarlo Zanotti,‡ and
Piero A. Temussi§
Dipartimento di Scienze Farmaceutiche, Universita` di Salerno, piazza V. Emanuele 9, 84084 Penta di Fisciano, Salerno, Italy,
Centro di Chimica del Farmaco, CNR, Dipartimento di Studi Farmaceutici, Facolta` di Farmacia “La Sapienza”, Roma, Italy,
and Dipartimento di Chimica, Universita` di Napoli Federico II, via Mezzocannone 4, 80134 Napoli, Italy
Received J uly 29, 1998
Tuftsin, a linear tetrapeptide (Thr-Lys-Pro-Arg), corresponding to the sequence 289-292 of
the heavy chain of leukokinin, has been the object of intensive SAR studies during the past 30
years, owing to its numerous biological activities and to the possibility of generating a novel
anticancer drug. A cyclic tuftsin analogue, c-[T-K-P-R-G], has biological activity 50 times higher
than that of the parent linear peptide. Here we present a conformational study of c-[T-K-P-
R-G] based on NMR data in a cryoprotective DMSO/water mixture. The preferred conformation
is a type VIa turn centered on the K-P residues. The orientation of the side chains of the two
basic residues (K and R) may represent the essential feature of the bioactive conformation of
tuftsin. A possible role of tuftsin as a DNA binding motif is suggested by the similarity of the
bioactive conformation of c-[T-K-P-R-G] and of the â-turn conformation proposed by Suzuki
for the [T,S]-P-K-R motif.
In tr od u ction
life of several minutes,9,10 preparation of peptidase-
resistant tuftsin analogues would allow oral adminis-
tration, thus facilitating clinical use. In addition, an
ecto-enzyme on the membrane of PMNs, leucine ami-
nopeptidase, inactivates tuftsin by cleaving the N-
terminal threonine producing a competitive inhibitor,11
Lys-Pro-Arg. As a result tuftsin analogues that are
resistant to this peptidase attack would be expected to
possess enhanced therapeutic potential.
Tuftsin (Thr-Lys-Pro-Arg), a linear peptide corre-
sponding to the sequence 289-292 of the heavy chain
of leukokinin, a cytophilic γ-globulin,1 is endowed with
several remarkable biological activities. Complete re-
lease of this tetrapeptide from the parent protein occurs
in two stages. The cleavage of the Arg292-Glu293 peptide
bond by a splenic enzyme, tuftsin endocarboxypeptidase,
produces leukokinin-S containing the free carboxyl
terminus of tuftsin. Second, after leukokinin-S is bound
to the blood and tissue granulocyte, the Lys288-Thr289
peptide bond is cleaved by the membrane enzyme
leukokininase.
It is generally believed that tuftsin binds directly to
specific tuftsin receptors on polymorphonucleates (PMN),
monocyte-macrophages, and (natural killer) NK cells
and modulates their biological activities,2,3 but direct
interaction with DNA has also been invoked to explain
the biological activity of this peptide.4 Tuftsin deficiency
has been found in patients with some types of cancer,
myelofibrosis, idiopathic trombocytopenic purpura, sple-
nectomy, sickle cell disease, AIDS, and AIDS-related
complex, in addition to tuftsin congenital abnormali-
ties.2,3,5,6 In animal and clinical studies, tuftsin dis-
played antitumor, antiinfection, and anti-AIDS activi-
ties with no detectable toxicity.2,3,7
The design of more active and resistant tuftsin
analogues can greatly benefit from an accurate confor-
mational study of tuftsin itself. A previous study on
linear tuftsin in solution12 performed in our laboratory
identified two conformers (dubbed conformer II and
conformer III, respectively) as relevant components of
the mixture of solution conformations. Both structures,
characterized by a trans Lys-Pro bond, are very com-
pact, consistent with the observation that one NH
proton is strongly solvent-shielded. The features of both
conformations were strongly suggestive of possible cyclic
analogues, an obvious choice when looking for analogues
resistant to peptidase attack.
Cyclic analogues were in fact explicitly proposed by
a computational study that appeared nearly simulta-
neously to our solution study.13 These authors suggested
the presence of a type IV â-turn for linear tuftsin. This
conformer is not coincident with the absolute minimum
energy conformation found by us but is fairly similar
to the other less populated conformer of the quoted
solution work.12
Many attempts were made to produce active tuftsin
analogues,8 but these efforts were hampered by either
loss of activity or formation of competitive inhibitors.
Since tuftsin is degraded rapidly in serum with a half-
In addition to the simplest possible cycle, i.e., c-[T-
K-P-R] (dubbed ctuf), O’Connor et al.13 proposed three
other analogues characterized by the insertion of an
† Universita` di Salerno.
‡ CNR.
§ Universita` di Napoli Federico II.
10.1021/jm980442+ CCC: $18.00 © 1999 American Chemical Society
Published on Web 05/01/1999

1706 J ournal of Medicinal Chemistry, 1999, Vol. 42, No. 10
D′Ursi et al.
Ta ble 1. Relevant Proton Chemical Shifts of the Cis Conformers of ctuf2 and ctufâa
ctuf2
ctufâ
NH
CRH
CâH
CγH
CδH
CꢀH
NHsc
Thr
7.28 (-2.5)
7.87 (-7.5)
8.03 (-2.0)
8.52 (-6.6)
3.67
4.07
4.08
4.13
4.44
4.52
4.32
4.23
3.99
3.83
1.03
1.10
Thr
Lys
Lys
Pro
Arg
Gly
1.59/1.63
1.59/1.62
2.02/2.21
1.98/2.20
1.56/1.61
1.60/1.77
1.11/1.28
1.20/1.40
1.67/1.86
1.53/1.87
1.34/1.38
1.35/1.40
1.46/1.48
1.49/1.52
3.38/3.62
3.38/3.43
3.06
2.72
2.77
7.68
7.77
Pro
Arg
â-Ala
7.51 (-6.6)
8.20 (-8.0)
9.05 (-3.3)
7.64 (-1.1)
7.48
7.49
3.05
3.20/4.13
3.47/3.23
2.14/2.46
a
All values were referenced to the residual DMSO resonance at 2.5 ppm. Temperature coefficients of NH resonances are reported in
brackets.
appropriate spacer: glycine in c-[T-K-P-R-G] (dubbed
ctuf2), lysine in c-[T-K-P-R-K] (dubbed ctuf3), and
aspartic acid in c-[T-K-P-R-D] (dubbed ctuf4). In their
calculations addition of either G or D in position 5
resulted in a backbone conformation similar to that they
proposed for linear tuftsin, i.e., a type IV â-turn at the
Lys-Pro position. Another computational study14 reex-
amined the conformations accessible to linear tuftsin
and ctuf2, treating explicitly the solvent (pure water and
NaCl-water). The results of this study stressed the
relevance of cis peptide bonds related to salt effects and
prevalent in proline-containing peptides.
Ctuf2 and ctuf4 were eventually synthesized and
tested for biological activity.15 One isomer of ctuf4
proved nearly as active as linear tuftsin, whereas ctuf2
was found to be 50 times more active than linear tuftsin.
Therefore we decided to examine the conformation of
ctuf2 in solution by means of a combination of NMR and
molecular modeling. A new cyclic analogue, c-[Thr-Lys-
Pro-Arg-â-Ala], dubbed ctufâ, was also studied for
comparison, in particular to check whether a more
flexible link might influence cis-trans isomerism.
Tuftsin analogues ctuf2 (c-[Thr-Lys-Pro-Arg-Gly]) and
ctufâ (c-[Thr-Lys-Pro-Arg-â-Ala]) were examined in a 80:
20 (v:v) DMSO-d6/water cryomixture at 275 K. Sequen-
tial assignment of all protons of ctuf2 and ctufâ in the
80:20 (v:v) DMSO-d6/water cryomixture was achieved
by standard methods20 via the usual systematic applica-
tion of DQF-COSY,21 TOCSY,22 and NOESY23 experi-
ments.
1D protonic spectra of the two analogues show that
several resonances are split in two signals corresponding
to the two families of conformations expected from the
cis-trans isomerism around the Lys-Pro bond. Trans
conformers of linear tuftsin are largely predominant in
water, neat DMSO, and the 80:20 (v:v) DMSO-d6/water
cryomixture. On the contrary, the cis conformers are
more abundant in cyclic analogues, with a cis/trans ratio
of 5:1 and 30:1 for ctuf2 and ctufâ, respectively. Since
ctufâ was designed primarily to check whether the
introduction of a more flexible (and longer) linker was
still consistent with the prevalence of a cis K-P bond,
it is interesting to note that the cis/trans ratio in ctufâ
is even higher than that in ctuf2. Another piece of
interesting information that can be inferred from the
1D spectra in the temperature range 270-310 K is that
some NH resonances have fairly low temperature coef-
ficients.
Resu lts a n d Discu ssion
The flexibility of small peptides, either linear or cyclic,
may lead to inextricable mixtures of isoenergetic con-
formers and makes NOEs very difficult to detect,
particularly in neat solvents of low viscosity such as
water. A proper choice of the solvent medium can
improve this situation. We have shown16,17 that NMR
problems linked to flexibility can be partially overcome
by running NMR spectra at low temperature in a
cryoprotective mixture. These experimental conditions
imply fairly high viscosities that favor the building up
of sizable NOEs.16 In addition, the use of a viscous
medium can affect the equilibrium among isoenergetic
conformers, selecting the more ordered conformers.18
Accordingly, we chose to rely on the best characterized
among the cryoprotective mixtures, i.e., DMSO/
water.16-18 Such a mixture at 275 K, i.e., the tempera-
ture used for all NOESY experiments, has a viscosity
of 6.7 cp.19
Relevant proton chemical shift data are summarized
in Table 1 along with temperature coefficients of NH
resonances. The observation of small coefficients, -2.5
ppb/°C for T and -2.0 ppb/°C for K in ctuf2 and -1.0
ppb/°C for â-Ala in ctufâ, hints at a high structural
rigidity, but it is still not sufficient to indicate precise
conformations nor even to guarantee that the corre-
sponding hydrogens are involved in intramolecular
hydrogen bonds. Amide hydrogens involved in intramo-
lecular H-bonding or otherwise shielded from solvent
can be identified by their relatively slow rate of H/D
exchange. This information is commonplace in proteins
but is generally very difficult to obtain in small peptides
whose amide hydrogens normally exchange very fast.
Nonetheless, owing to the peculiar behavior of the amide
protons of tuftsin in DMSO,12 we did perform these
measurements for the two cyclic analogues. Slowly
exchanging amide NH’s of ctuf2 were detected by
dissolving the peptide in the DMSO-d6/D2O (80:20, v:v)
cryomixture and monitoring signal intensities of re-
sidual amide NH resonances at 275 K. The NH’s of Gly
and Thr have half-lives respectively of 60 and 80 min,
whereas the corresponding NH’s of Arg and Lys have
lower exchange rates, with half-lives respectively of 330
To check whether this cryoprotective mixture is
compatible with the known conformational behavior of
tuftsin in DMSO,12 we examined also linear tuftsin in
a 80:20 (v:v) DMSO-d6/water cryomixture. A comparison
1
of the H NMR spectra of tuftsin in neat DMSO and in
the 80:20 (v:v) DMSO-d6/water cryomixture showed that
the NMR data in the two solvent media are completely
consistent, the only difference being that the cryomix-
ture favors the detection of a larger number of NOEs.

Solution Conformation of an Analogue of Tuftsin
J ournal of Medicinal Chemistry, 1999, Vol. 42, No. 10 1707
F igu r e 1. Low-field parts (a, b) of the 300-ms NOESY spectra of ctuf2 in the 80:20 (v:v) DMSO-d6/water cryomixture at 275 K.
P/K diagnostic cross-peaks are shown in the expansion (c).
F igu r e 2. Low-field parts (a, b) of the 300-ms NOESY spectra of ctufâ in the 80:20 (v:v) DMSO-d6/water cryomixture at 275 K.
P/K diagnostic cross-peaks are shown in the expansion (c).
and 450 min, thus confirming that the NH of Lys is
indeed involved in hydrogen bonding.
isomers is crucial for the subsequent structure inter-
pretation; it is well-known that the two isomers are best
discriminated on the basis of R-R X-P or R-δ X-P
cross-peaks.
Figure 1 shows the low-field parts of the NOESY
spectra of ctuf2 (Figure 1a,b) in the 80:20 (v:v) DMSO-
d6/water cryomixture at 275 K and a mixing time of 300
ms, together with the expansion of the R-CH region that
shows the P/K correlations crucial for the sequential
assignment of the resonances to the cis isomer (Figure
1c). The corresponding spectra for ctufâ at the same
experimental conditions are shown in Figure 2a-c,
respectively. The resonances can be grouped in two sets
of different intensity corresponding to molecules con-
taining a cis or trans Lys-Pro bond, respectively.
Assignment of the two components to cis and trans
The quoted conformational analysis of linear tuftsin
in DMSO12 has shown that the most populated confor-
mational family for tuftsin is characterized by a trans
Lys-Pro bond. The two preferred conformations (con-
former II and conformer III in the original paper12),
consistent with all NOE-derived constraints, are rigid,
folded structures, characterized by an approximate
inverse γ-turn centered on Pro. The findings that the
prevailing family of conformers in the cyclic analogues
contains a cis Lys-Pro bond and that ctuf2 is 50 times

1708 J ournal of Medicinal Chemistry, 1999, Vol. 42, No. 10
D′Ursi et al.
Ta ble 2. Comparison of Experimental Interproton Distances
Derived from NOEs (dexp) and Calculated Distances (dcalc) of
Models of ctuf2 Calculated from Canonical Turns (G1, G2) or
the Conformation of the Corresponding Tuftsin Sequence
(TKPR) in the Parent Protein (G3, G4)
more active than the linear parent peptide imply that
the conformers previously identified in the DMSO
solution of tuftsin may not be representative of the
bioactive conformation. Accordingly, it is not possible
to use them as starting conformers for the interpretation
of the NMR data of the cyclic analogues.
dexp, Å
dcalc, Å
G2 G3
residue
NOE
NMR
G1
G4
It is possible, in principle, to run an exhaustive
conformational search on the cyclic analogues, similar
to those previously performed on tuftsin12,24 since the
size of both peptides is rather small. However, the main
conformational preferences of tuftsin and ctuf2 are well-
known from the quoted conformational studies,12-14,24
and an improvement over these analyses would require
a much finer mesh in the search with a substantial
scaling up of computing times. Accordingly, we did not
attempt to map out the conformational space of the two
analogues but chose to rely on a combination of known
tuftsin structures and NMR data to find the prevailing
conformers of ctuf2 in solution. The data of ctufâ are so
similar to those of ctuf2 that the structure determina-
tion of this analogue can be tailored after that of ctuf2.
The features of the prevailing conformer of ctuf2 were
used as an internal check, i.e., to examine whether the
stability of conformers, of ctuf2 and ctufâ, containing a
K-P cis bond is influenced by the flexibility of the whole
cycle.
Thr-1 NH
NH Gly-5
NH Lys-2
CHR1 Gly-5
CHR2 Gly-5
3.64
3.22
2.74
3.79
3.14 3.79 3.57 4.28
3.45 3.55 3.79 3.76
2.27 2.28 3.70 3.75
3.43 3.40 4.07 4.11
Lys-2 NH
HN Thr-1
HN Arg-4
CHR Pro-3
CHR Thr-1
CHγ THr-1
3.22
4.27
4.49
3.08
4.41
3.55 3.08 2.5
3.11
5.64 4.80 3.24 2.99
5.12 5.25 2.54 3.16
2.16 2.15 2.20 2.24
3.81 3.66 4.62 5.34
Pro-3 CHR
CHR Lys-2
*CHγ1Lys-2
CHâ Lys-2
3.34
3.70
4.05
2.44 2.58 4.03 4.01
3.15 4.04 5.34 5.52
4.11 4.22 4.08 4.97
Arg-4 NH
HN Lys-2
4.27
3.89
3.43
3.67
3.41
5.64 4.80 3.24 2.99
3.57 3.52 2.25 3.45
3.70 3.11 5.71 4.28
3.32 3.57 3.42 3.81
2.68 3.22 5.66 3.54
CHR Pro-3
CHR Lys-2
*CHâ Pro-3
*CHδ Pro-3
Gly-5 NH
HN Arg-4
HN Thr-1
CHR Arg-4
*CHâ Arg-4
*CHγ Arg-4
4.39
3.64
2.80
3.33
4.51
2.89 3.73 3.80 3.48
3.14 3.79 3.57 4.28
3.56 3.33 3.61 3.64
3.85 2.19 3.06 2.92
3.30 4.52 3.64 4.65
The strategy adopted to find the prevailing conform-
ers of ctuf2 in solution was based on two methods: (a)
a simple check of canonical conformations, containing
a cis Lys-Pro peptide bond, to test their consistency
with experimental NMR data and (b) the use of the
structure of the parent protein25 to generate starting
conformations in procedures of EM and MD.
Ta ble 3. Significant Internal Rotation Angles of Models of
ctuf2 Calculated from Canonical Turns (G1, G2) or the
Conformation of the Corresponding Tuftsin Sequence (TKPR) in
the Parent Protein (G3, G4) Calculations
dihedral
G1
G2
32
G3
34
G4
Alternatively one might resort to more sophisticated
approaches, e.g., ensemble calculations,26 but the num-
ber of the constraints that can be derived from the NMR
data in our case does not warrant this kind of approach.
E, kcal mol-1
Thr-1
29
94.24
32
φ
ψ
φ
ψ
φ
ψ
φ
ψ
φ
98.94 -103.9
111.70 88.10
-51.32 -115.02
-104.44
108.41
-58.05
127.81
-74.22
-45.06
78.56
Lys-2
Pro-3
Arg-4
Gly-5
-92.05
-35.15
-72.57
-9.50
-50.90
-58.72
126.23
-67.64
-26.53
-46.38
-75.76
159.56
114.81
-80.29
A qualitative interpretation of the NMR data suggests
a well-defined family of structures characterized by a
cis Lys-Pro bond. In particular the two adjacent short
NH-NH distances (K-T and T-G), concomitant low-
temperature coefficients, and slow H/D exchange rates
are consistent with the features of canonical (cis) type
VI â-turns.27 Using families of type VI â-turns (VIa and
VIb) as starting conformers in unrestrained EM runs,
it was clear that the best agreement with the NMR data
can be obtained with turns of the VIa type. Most
conformational features of the backbone of this structure
(G1) are in fair agreement with the NOE-derived
distances reported in Table 2. The relative arrangement
of the side chains was optimized by MD runs in which
the backbone conformation was kept fixed at the values
reported in Table 3. Subsequent restrained EM runs
yielded a structure (G2) that is in good agreement with
the experimental data. In particular, there are small
violations of observed NOE-derived constraints but also
no significant violation of absent distance constraints
(ADC in the terminology of Bru¨schweiler et al.28).
Significant internal rotation angles of G1 and G2 are
reported in Table 3. The differences between overall
geometrical features are small: such a behavior is
mirrored by the fact that the “restrained structure”
relaxes back, to some extent, to the “unrestrained
-78.33 -113.69
-50.42 -112.22
-164.42
-82.58 -161.30 -162.86
ψ
74.69
78.54
85.43
75.47
structure” when the NMR-derived restraints are re-
lieved. The molecular model of G2 is shown in Figure
3.
Alternatively, we tried to use the structure of the
tuftsin fragment in the parent protein as a starting
point. We are aware, of course, of the fact that the K-P
peptide bond in the protein is trans and that the
conformation of the tuftsin fragment in the protein may
bear no relationship with the bioactive conformation of
tuftsin or even with the solution conformation of tuftsin.
However, it seemed interesting to verify whether using
the “protein conformation” as a starting point we might
“naturally” end up with a correct cyclic tuftsin confor-
mation. In human Igg tuftsin residues are part of a
â-sheet strand in the Fc fragment. Tuftsin is actually
produced, in vivo, as a cleavage fragment of this
leukokinin (residues 289-292). We ‘extracted’ the four
residues with the conformation they have in the pro-
tein25 and tried to use this ‘protein conformer’ to produce

Solution Conformation of an Analogue of Tuftsin
J ournal of Medicinal Chemistry, 1999, Vol. 42, No. 10 1709
F igu r e 3. Stereoviews of the molecular models of the conformers of ctuf2 derived from a starting canonical VIa turn centered on
the K-P bond (G2, top) and from the conformation of the sequence T-K-P-R in the parent protein (G4, bottom).
stable structures of the two cyclic analogues. If one uses
this ‘protein conformer’ in EM runs that do not over-
emphasize the role of electrostatic interactions, i.e., by
using neutral end groups and/or employing a distance-
dependent dielectric constant, the predictable result is
a structure still characterized by a trans Lys-Pro bond.
However, when one performs an EM calculation with
fully charged N- and C-terminal groups and a low value
of the dielectric constant, as already noted by Valdeavel-
la et al.,14 overestimation of electrostatic interactions
leads to a “spontaneous cyclization”, i.e., to a structure
rearrangement characterized by a cis Lys-Pro bond and
a quasi-cyclic backbone shape.
A comparison of calculated distances of the two final
conformers (G4 and G2) derived from the two methods
described above with experimental data (Table 2) shows
that, although the agreement of G2 is fairly good,
neither structure is fully consistent with the NOE-
derived distances. However, it can be seen that nearly
in all cases where there is a violation, the calculated
distances can be interpolated between the two struc-
tures to give a satisfactory agreement with experimental
data. An interesting feature of these structures is the
fact that, despite differences in the backbone torsion
angles, they have similar energies and a very similar
topology of the side chains, which implies a very similar
surface of interaction. It is interesting to note that while
refinement of the starting VIa conformer does not alter
the backbone topology too much, i.e., conformer G2 is
still a VIa turn, restrained minimization alters the
backbone conformation of G3, as indicated by the rmsd
value between G3 and G4 (0.96), leading to significant
transitions of ψ3 and φ4 torsion angles.
A crucial point in this structure determination is the
stability of conformations containing a cis K-P peptide
bond. It is surprising that in going from linear tuftsin
to ctuf2 there is a drastic change in the relative stability
of conformers containing a trans or cis K-P peptide
bond. Although it can be expected that cyclization
imposes some sterical strain on the K-P peptide bond,
it must be taken into account that a cyclic pentapeptide
has considerable conformational flexibility. An indirect
way to check the intrinsic stability of the conformers of
ctuf2 containing a cis K-P peptide bond is to modify
the constitution of the peptide by inserting a residue
The gap between N- and C-terminal regions can be
easily filled by a glycine residue. After cyclization with
glycine another unrestrained EM run yielded a stable
structure of ctuf2 containing a cis Lys-Pro amide bond
(G3). Its relevant distances are reported in Table 2. As
already described for conformer G1, the relative ar-
rangement of the side chains of G3 was optimized by
MD runs in which the backbone conformation was kept
fixed at the values reported in Table 3. After constrained
EM calculations using NOE-derived distances, the
resulting structure (G4) shows a better agreement with
experimental NOEs with respect to the starting con-
former (G3). Relevant conformational parameters are
reported in Table 3. A final EM calculation was per-
formed to allow relaxation of the structure derived by
the constrained application of NOE distances and to
check whether it represents a true energy minimum
conformational state. The molecular model of G4 is
shown in Figure 3.

1710 J ournal of Medicinal Chemistry, 1999, Vol. 42, No. 10
D′Ursi et al.
F igu r e 4. Comparison of the molecular models of the minimum energy conformer of ctufâ, B2 (red), and of the corresponding
ctuf2, G2 (blue), displayed using capped sticks renderings. Heteroatoms are labeled.
even more flexible than Gly. We have examined the
solution conformation of ctufâ and found that both the
NMR data and energy calculations similar to those
performed on ctuf2 lead to the same results, i.e., to
isomers containing a cis K-P peptide bond and char-
acterized by backbone torsion angles very similar to
those of ctuf2. Even in this case, using families of type
VI â-turns (VIa and VIb) as starting conformers in
unrestrained EM runs, it was clear that the best
agreement with the NMR data can be obtained with
turns of the VIa type. The relative arrangement of the
side chains was optimized by MD runs in which the
backbone conformation was kept fixed at the values of
the starting VIa conformation. Subsequent restrained
EM runs yielded a final structure that is in good
agreement with the experimental data. Figure 4 shows
a comparison of this model of ctufâ with the correspond-
ing model of ctuf2 (G2).
It can be noted that all models of Figures 3 and 4,
despite differences in constitution and/or backbone
conformation, are characterized by similar orientations
of the two basic side chains. It is difficult to say whether
this feature is characteristic of a bioactive conformation
interacting with a specific membrane receptor, but it is
interesting that this orientation is shared also by one
of the conformers (III) of linear tuftsin in DMSO
solution,12 by the native structure of the sequence in
the parent protein,25 and even by the type I turn
proposed by Suzuki for the S-P-X-X peptide correspond-
ing to the [S,T]-P-X-X motif that the same author
proposed as a general motif in gene regulatory pro-
teins,29 if one substitutes K and R for the two X. This
observation, in turn, prompted a more in-depth check
of the hypothesis that the main biological action of
tuftsin is a direct interaction with DNA.4
[T,S]-[K,R]-P-[K,R], corresponding to tuftsin and its
closest analogues, is comparable to that of the Suzuki
motif, [T,S]-P-[K,R]-[K,R]. For instance, in the Swis-
sProt data bank, the pattern [T,S]-[K,R]-P-[K,R] was
found in 1 737 instances in 1 664 proteins out of a total
of 70 423 entries, whereas that of the Suzuki motif was
found 2 290 times in 2 012 proteins. Thus the sequence
permutation showed by tuftsin and its analogues with
respect to the [S,T]-P-[K,R]-[K,R] motif, i.e., T-[K,R]-P-
[K,R], may not prevent interaction with DNA, particu-
larly if tuftsin acts as an inhibitor with respect to DNA
binding proteins. As noted above it is interesting that
the two basic side chains of ctuf2 in the G2 and G4
models are oriented in way very similar to that of the
turn proposed by Suzuki.29 Figure 5 shows the com-
parison of all mentioned models: i.e., G2 (yellow) and
G4 (pink) of ctuf2, ctufâ (orange), model III of linear
tuftsin (green), type I â-turn of the Suzuki peptide
(gray), and parent protein TKPR sequence (blue). The
similarity of the orientation of the K and R side chains
in all structures is remarkable, especially if one takes
into account the differences in constitution.
We can conclude by saying that the prevailing con-
former of ctuf2 is characterized by a cis K-P bond and
has a type VIa turn conformation. This structure
appears a likely candidate for the ‘bioactive conforma-
tion’ of tuftsin and has an outer shape strongly sugges-
tive of a possible interaction with DNA.
Exp er im en ta l Section
P ep tid es. Tuftsin was purchased from Bachem and used
without further purification. c-[Thr-Lys-Pro-Arg-Gly] and
c-[Thr-Lys-Pro-Arg-â-Ala] were synthesized by standard solu-
tion methods according to the following procedure.
Boc(ter t-b u t yloxyca r b on yl)-Lys(2-Cl-Z)(2-ch lor ob en -
zyloxyca r bon yl)-P r o-OBg(N-ben zh yd r yl-glycola m id e es-
ter ) (1). To a stirred solution of 3.56 g of Boc-Lys(2-Cl-Z)-OH
(8.6 mmol) in 100 mL of CH2Cl2 were added at -10 °C 1.16 g
of iBCCl (isobutyl chloroformate) (8.6 mmol) and 1.16 g of
NMM (N-methylmorpholine) (11.4 mmol). After 10 min a
precooled solution of 2.9 g of H-Pro-OBg (8.6 mmol) in 30 mL
of CH2Cl2 was added; the reaction mixture was stirred at room
temperature for 2 h, extracted with aqueous 0.5 M KHSO4, a
saturated NaHCO3 solution, and water, dried with Na2SO4,
and evaporated in vacuo, to give 5.9 g (95% yield) of title
In fact the sequence of tuftsin is reminiscent of the
[S,T]-P-X-X motif that is involved in specific binding of
AT-rich sequences along the minor groove of DNA.29 The
similarity is even higher with other peptides that bind
AT-rich DNA such as the so-called BD peptide, a
consensus peptide from HMGI30 whose sequence con-
tains a proline flanked by two basic residues: TP-
KRPRGRPKK. It is interesting to note that the fre-
quency of occurrence in the data banks of the pattern

Solution Conformation of an Analogue of Tuftsin
J ournal of Medicinal Chemistry, 1999, Vol. 42, No. 10 1711
in 20 mL of TFA for 1 h at room temperature. The solution
was evaporated to dryness and repeatedly evaporated after
addition of ether. The TFA salt was dissolved in 40 mL of
acetonitrile and added to a solution of 0.394 g of Boc-Thr-OH
(1.8 mmol), 0.795 g of BOP (benzotriazol-1-yloxytris(dimethyl-
amino)phosphonium hexafluorophosphate) (1.8 mmol), and
0.528 mL of TEA (triethylamine) (3.76 mmol) in 20 mL of
acetonitrile. After stirring overnight at room temperature, the
reaction mixture was evaporated to dryness, and the residue,
kept in ethyl acetate, was washed with aqueous 0.5 M KHSO4,
saturated NaHCO3 solution, and water. The organic phase,
dried with Na2SO4 and evaporated in vacuo, gave 1.9 g of a
residue which was chromatographed on a silica gel column
(100 cm × 2.5 cm) in chloroform-methanol (85:15) as eluent.
Fractions (30 mL) 20-45 afforded 1.52 g (85% yield) of TLC
pure 6 (Rf ) 0.6 in chloroform-methanol, 85:15).
Boc-Lys(2-Cl-Z)-P r o-Ar g(N02)-â-Ala -OBg (7). The pro-
tected tetrapeptide 7 was obtained by reacting a MA prepared
from 2.26 g of 4 (3.10 mmol) with 0.96 g of H-â-Ala-OBg (3.10
mmol) following the procedures described in the previous
paragraphs. The crude reaction residue was purified on a silica
gel column (100 cm × 2.5 cm) in chloroform-methanol (9:1)
as eluent obtaining 1.84 g (59% yield) of 7 as aTLC pure foam
(Rf ) 0.55 in chloroform-methanol, 9:1).
Boc-Th r -Lys(2-Cl-Z)-P r o-Ar g(NO2)-â-Ala -OBg (8). A to-
tal of 1.5 g of Boc-tetrapeptide 7 (1.5 mmol) was deprotected
from Boc by TFA (trifluoroacetic acid) treatment as previously
reported. The obtained TFA salt, dissolved in 20 mL of
acetonitrile, was added at room temperature to a solution of
0.463 g of Boc-Thr-OH (1.5 mmol), 0.660 g of BOP (1.5 mmol),
and 0.41 mL of TEA (5.6 mmol) in 20 mL of acetonitrile. The
reaction mixture was left under stirring overnight and worked
up as previously described obtaining a residue which was
chromatographed on a silica gel column (100 cm × 2.5 cm) in
chloroform-methanol (85:15) as eluent. Fractions (30 mL) 29-
35 afforded 1.35 g (81% yield) of the title compound (Rf ) 0.6
in chloroform-methanol, 85:15).
TF A‚Th r -Lys(2-Cl-Z)-P r o-Ar g(NO2)-Gly-OH (9). For OBg
deprotection, 1.56 g of protected pentapeptide 8 (1.42 mmol)
was dissolved in 20 mL of DMF and treated with a solution of
0.39 g of K2CO3 (2.84 mmol) in 20 mL of water. After stirring
1 h at room temperature, the reaction mixture, diluted with
water, was extracted with 100 mL of ether. The aqueous phase
was acidified with KHSO4 and extracted with ethyl acetate.
The organic layer, washed with water and dried with Na2SO4,
was evaporated in vacuo, and the obtained residue was kept
for 1 h at room temperature in 10 mL of TFA for Boc removal.
The TFA was evaporated in vacuo, and the residue, triturated
with ether, gave a white solid which was filtered and repeat-
edly washed with ether obtaining 1.08 g (77% yield) of 9 (Rf )
0.3 in n-buthanol-acetic acid-water, 4:1:1).
F igu r e 5. Comparison of the molecular models G2 (yellow),
G4 (pink) of ctuf2 with the models of ctufâ (orange), linear
tuftsin (green), type I â-turn of the Suzuki peptide (gray), and
parent protein TKPR sequence (blue). The K and R side chains
are represented as capped stick (thick sticks) models, whereas
all other atoms of the peptides and of the parent protein are
represented as lines.
compound as a TLC (thin-layer chromatography) pure oil (Rf
) 0.7 in ethyl acetate-diethyl ether, 8:2).
Boc-Lys(2-Cl-Z)-P r o-OH (2). To a solution of 5.9 g of 1 (8
mmol) in 20 mL of DMF (dimethylformamide) was added a
solution of 2.26 g of K2CO3 (16 mmol) in 20 mL of water. After
stirring overnight at room temperature, the reaction solution
was diluted with water and extracted with 200 mL of diethyl
ether. The aqueous phase, acidified with 0.5 M KHSO4, was
extracted with ethyl acetate. The organic phase, washed with
water, dried with Na2SO4, and evaporated in vacuo, afforded
4 g (100% yield) of 2 as a TLC pure foam (Rf ) 0.3 in ethyl
acetate-acetic acid, 95:5).
Boc-Lys(2-Cl-Z)-P r o-Ar g(NO2)-OBg (3). To a solution of
4 g of 2 (8 mmol) in 60 mL of CH2Cl2 were added at -10 °C
under stirring 1.08 g of iBCCl (8 mmol) and 1.08 g of NMM
(10.68 mmol). After 10 min a solution of 3.54 g of H-Arg(NO2)-
OBg (8 mmol) in
a mixture of 60 mL of CH2Cl2-THF
(tetrahydrofuran) (1:1) was added. After 3 h stirring at room
temperature, the reaction mixture, worked up as previously
described, gave 7.2 g of a residue which was chromatographed
on a silica gel column (150 cm × 2.5 cm) in chloroform-
methanol (92:8) as eluent, obtaining 5.6 g (75% yield) of 3 (Rf
) 0.4 in chloroform-methanol 9:1).
TF A‚Th r -Lys(2-Cl-Z)-P r o-Ar g(NO2)-â-Ala -OH (10). In an
analogous way, as described in the preceding paragraph, 1 g
of title compound (1.1 mmol, 88% yield) was obtained from
1.35 g (1.22 mmol) of protected pentapeptide 8.
Cyclo[Th r -Lys(2-Cl-Z)-P r o-Ar g(NO2)-Gly] (11). To the
stirred solution of 0.89 g (1 mmol) of 9 in 120 mL DMF and
80 mL of THF was added at -15 °C 0.136 g of iBCCl (1 mmol).
After 10 min stirring, the precooled solution of 0.2 mL of NMM
in 340 mL of DMF and 160 mL of THF was added. After
stirring 1 h at 0 °C and overnight at room temperature, the
reaction mixture was evaporated in vacuo and the residue
chromatographed on a Sephadex LH 20 column (250 cm × 2.5
cm) in methanol as eluent. The residue obtained from fractions
(27 mL) 54-59 was further chromatographed on a silica gel
column (30 cm × 2.5 cm) in chloroform-methanol-water (65:
25:4) as eluent. Fractions (20 mL) 6-7 gave 276 mg (37% yield)
of title compound (Rf ) 0.4 in chloroform-methanol-water,
65:25:4).
Boc-Lys(2-Cl-Z)-P r o-Ar g(NO2)-OH (4). To a stirred solu-
tion of 5.6 g of 3 (6 mmol) in 30 mL of DMF was added a
solution of 1.65 g of K2CO3 (12 mmol) in 30 mL of water. After
stirring 1.5 h at room temperature, the reaction mixture was
worked up as described for the analogous deprotection reac-
tion, obtaining 4.8 g (100% yield) of 4 (Rf ) 0.3 in ethyl
acetate-acetic acid, 9:1).
Boc-Lys(2-Cl-Z)-P r o-Ar g(NO2)-Gly-OBg (5). To a MA
(mixed anhydride solution), prepared as already described,
from 2.26 g of 4 (3.10 mmol), 0.42 g of iBCCl (3.10 mmol), and
0.42 g of NMM (4.14 mmol) in CH2Cl2 was added 0.92 g of
H-Gly-OBg (3.10 mmol) in 20 mL of CH2Cl2. After the usual
workup, 3.2 g of the obtained residue was purified by chro-
matography on a silica gel column (100 cm × 2.5 cm) in
chloroform-methanol (92:8) as eluent obtaining 1.8 g (79%
yield) of 5 as a TLC pure foam (Rf ) 0.4 in chloroform-
methanol, 9:1).
Cyclo(Th r -Lys-P r o-Ar g-Gly) (12). A total of 276 mg of 11
dissolved in 40 mL of acetic acid-water (1:1) was hydrogenated
over 40 mg of Pd black. After 1 day further 40 mg of catalyst
was added; total deprotection occurred in 3 days. After removal
of catalyst by filtration, the filtrate was evaporated in vacuo
Boc-Th r -Lys(2-Cl-Z)-P r o-Ar g(NO2)-Gly-OBg (6). For re-
moval of Boc, 1.8 g of Boc-tetrapeptide 5 (1.8 mmol) was kept

1712 J ournal of Medicinal Chemistry, 1999, Vol. 42, No. 10
D′Ursi et al.
and the residue chromatographed on a Sephadex LH 20
column (100 cm × 2.5 cm) in water as eluent. Fractions (20
mL) 49-50 afforded 90 mg of cyclic title compound which was
further purified by preparative HPLC using a C18 RP column
eluted by linear gradient H2O/0.1% TFA and CH3CN/0.1% TFA
obtaining 70 mg (36% yield) of pure 12. Fast atom bombard-
ment mass value for (M + H)+ was 539 as expected.
Cyclo[Th r -Lys(2-Cl-Z)-P r o-Ar g(NO2)-â-Ala ] (13). A total
of 0.9 g (1 mmol) of 10 was reacted with 0.136 g of iBCCl
adopting the identical cyclization conditions used for the
synthesis of cyclopentapeptide 11 as previously described.
After purification of the crude reaction material by subsequent
chromatographic procedures on Sephadex LH 20 in methanol
as eluent and on silica gel eluting with chloroform-methanol-
water (65:25:4), 195 mg (25% yield) of cyclic title compound
was recovered (Rf ) 0.45 in chloroform-methanol-water, 65:
25:4).
Cyclo(Th r -Lys-P r o-Ar g-â-Ala ) (14). A total of 195 mg of
cyclopentapeptide 13 in 40 mL of acetic acid-water (1:1) were
hydrogenated over 20 mg of Pd black following the same
procedure used for obtaining cyclopeptide 12. The residue was
purified by chromatography on Sephadex LH 20 in water as
eluent followed by a preparative HPLC purification on a C18
RP column eluted by a linear gradient H2O/0.1% TFA and CH3-
CN/0.1% TFA obtaining 55 mg (40% yield) of pure 14. FAB
(M + H)+ was 553 as expected.
starting structures of cyclic analogues, was modeled starting
from the conformation adopted by the corresponding residues
(T-K-P-R) in human IgG. Peptide coordinates were transferred
directly from the protein X-ray structure using InsightII
Homology module. The obtained molecule was subjected to an
energy minimization, in its zwitterionic form. To allow a
gradual structural relaxation, first a steepest descents mini-
mization using a larger convergence criterion (until maximum
rms derivative was less than 0.5 kcal mol-1 Å-1) was applied;
second the geometry optimization proceeded using conjugate
gradient until the maximum rms derivative was less than
0.001 kcal mol-1 Å-1. This protocol was adopted in each EM
run.
Using InsightII Biopolymer module, tuftsin has been cy-
clized with glycine by automatic addition of the new amino
acid to tuftsin NH terminus. In the case of â-Ala, it proved
necessary to introduce this nonstandard residue manually. The
cyclic models, obtained by these two methods, were subjected
to an unrestrained EM calculation to let the peptide conforma-
tion reach its local energy minimum and then to a restrained
EM calculation by using distance restraints from NOESY
spectra. All restrained calculations were made using a range
of distances according to the percent error calculated for NOE-
derived distances.
Restraints were applied by a “flat-bottomed” potential
function with a force constant value of 1000 kcal mol-1 Å-1
.
NMR. NMR samples were prepared by dissolving appropri-
ate amounts of tuftsin, ctuf2, and ctufâ in a DMSO-d6/water
(80:20, v:v) cryomixture to obtain 1 mM solutions. NMR
spectra were run at 600 MHz on a Bruker DRX-600 instru-
ment. The samples for amide proton exchange were prepared
dissolving the peptides in a DMSO-d6/D2O water (80:20, v:v)
cryomixture to obtain 1 mM solutions.
This function applies a further energy penalty on the atoms
which exceeds the limits of the distance restraints ranges.
Regarding this latter parameter, two different energy penalty
values were assigned, to force more and less strongly the
conformation to fit NOE distances. These values were set to 1
and 100 kcal mol-1 Å-1, respectively, leading to two different
sets of structures defined as ‘less restrained’ and ‘more
restrained’. The obtained structures were relaxed by steepest
descents minimization until maximum rms derivative was less
than 0.5 kcal mol-1 Å-1 to let them reach the nearest local
minimum. The observed and absent NOEs were fully checked.
The conformation with better respected NOEs distances has
been chosen as starting point for restrained molecular dynam-
ics calculations.
1D NMR spectra were recorded in the Fourier mode, with
quadrature detection, and the water signal was suppressed
by a low-power selective irradiation in the homogated mode.
DQF-COSY,21 TOCSY,22 and NOESY23 experiments were run
in the phase-sensitive mode using quadrature detection in ω1
by time-proportional phase incrementation of initial pulse.30
Data block sizes were 2048 addresses in t2 and 512 equidistant
t1 values. Before Fourier transformation, the time domain data
matrices were multiplied by shifted sin2 functions in both
dimensions. A mixing time of 70 ms was employed for the
TOCSY experiment. NOESY experiments were run at 275 K
with mixing times in the range 100-300 ms. Cross-peak
volumes were quantitated in NOESY spectra with 75, 150, 225,
and 300 ms mixing times, using the AURELIA software
(Bruker). Distances were calibrated using d(δ,δ′) Pro3 distances
of 0.180 nm and calculating other distances according to the
r-6 relationship, fitting the build-up curve to a second-order
polynomial function. The necessary pseudoatom corrections
were applied for nonstereospecifically assigned protons at
prochiral centers and for Me group of Thr. The resulting values
were used as input upper distance restraints, considering a
percent error limit of 10%.
A molecular dynamics procedure was performed on the
peptide side chains while keeping the backbone geometry fixed.
Molecular dynamics simulations were run at high temperature
for 300 000 ps so that thermal energy sufficient to cross energy
barriers was added. Local minima conformations were then
equilibrated by running dynamics at 310 K for 600 000 ps to
allow the system to escape from high-energy states.
An average conformation was extracted from molecular
dynamics runs. The whole molecule was then relaxed using a
steepest descents minimization until maximum rms derivative
was less than 0.5 kcal mol-1 Å-1 to let it reach the nearest
local minimum. Observed and absent NOEs were fully checked.
The conformation of “Suzuki’s peptide”, T-P-K-R, was built
using canonical values of the type I â-turn for the internal
rotation angles as suggested in the original paper29 and
subsequently performing a molecular dynamics procedure on
the peptide side chains while keeping the backbone geometry
fixed.
En er gy Ca lcu la tion s. Energy calculations were based on
the all-atom parametrization of the CVFF force field32 as
provided by the MSI InsightII package. Every atom parameter
was assigned explicitly since the CVFF force field is especially
parametrized to reproduce peptide and protein properties. EM
calculations have been performed in vacuo with no distance
cutoff for nonbonded interactions; charges were computed as
provided by the CVFF force field.
Refer en ces
(1) Najjar, V. A.; Nishioka, K. A natural phagocytosis stimulating
peptide. Nature 1970, 288, 672-673.
X-ray coordinates of human immunoglobulin G25 were
extracted from the Brookhaven Protein Data Bank. Model
building was performed according to two different methods
(vide infra). In method (a) starting models for ctuf2 and ctufâ
were built from standard geometries based on canonical
peptide structures containing a cis K-P peptide bond.27 All
molecules were subjected to extensive preliminary EM and
then compared to NMR data. Introduction of NMR-based
restraints led to a good agreement only in the case of the VIa
model, which was thus chosen for subsequent calculations. In
method (b) linear tuftsin, used as a guideline to generate
(2) Nishioka, K.; Amoscato, A. A.; Babcock, G. F.; Banks, R. A.;
Philip, J . H. An immunomodulating peptide hormone and its
clinical potential as a natural biological response modifier.
Cancer Invest. 1984, 2, 39-49.
(3) Fridkin, M.; Najjar, V. A. Tuftsin: Its chemistry, biology and
physiological function. Crit. Rev. Biochem. Mol. Biol. 1989, 24,
1-40.
(4) Wagle, J . R.; Anservin, A. T.; Dessens, S. E.; Nishioka, K. Specific
translocation of Tuftsin (Thr-Lys-Pro-Arg), a natural immuno-
modulating peptide, into the nuclei of human monocytes. Bio-
chem. Biophys. Res. Commun. 1989, 159, 1147-1153.
(5) Constantopoulos, A. Congenital tuftsin deficiency. Ann. N.Y.
Acad. Sci. 1983, 419, 214-219.

Solution Conformation of an Analogue of Tuftsin
J ournal of Medicinal Chemistry, 1999, Vol. 42, No. 10 1713
(6) Corazza, G. R.; Zoli, G.; Ginaldi, L.; Cancelier, C.; Profeta, V.;
Gasparrini, G.; Quaglino, D. Tuftsin deficiency in AIDS. Lancet
1991, 337, 12-13.
(7) Marilus, R.; Spirer, Z.; Michaeli, D.; Zoferman, G.; Blum, I. First
case of AIDS in a homosexual in Israel. Results of different
therapeutic regimens. Isr. J . Med. Sci. 1984, 20, 249-251.
(8) Fridkin, M.; Gottlieb, P. Tuftsin, Thr-Lys-Pro-Arg, anatomy of
an immunologically active peptide. Mol. Cell Biochem. 1981, 41,
73-97.
(9) Surkirs, R.; Rubinraut, S.; Dagan, S.; Tzehoval, E.; Fridkin, M.;
Ben Yoseph, R.; Catane, R. Polytuftsin: A potential precursor
for slow release of phagocytosis stimulating peptide tuftsin. Int.
J . Biochem. 1990, 22, 193-195.
(10) Verdini, A. S.; Silvestri, S.; Becherucci, C.; Longobardi, M. G.;
Parente, L.; Peppoloni, S.; Perretti, M.; Pileri, P.; Pinori, M.;
Viscomi, G. C.; Nencioni, L. Immunostimulation by a partially
modified retro-inverso-tuftsin analogue containing Thr1Y-
[NHCO] (R,S)Lys2 modification. J . Med. Chem. 1991, 34, 3372-
3379.
(20) Wuthrich, K. NMR of Proteins and nucleic acids; J ohn Wiley &
Sons: New York, 1986.
(21) Piantini, U.; Soerensen, O. W.; Ernst, R. R. Multiple quantum
filters for elucidating NMR coupling networks. J . Am. Chem.
Soc. 1982, 104, 6800-6801.
(22) Bax, A.; Davis, D. G. MLEV-17-based two-dimensional homo-
nuclear magnetization transfer spectroscopy. J . Magn. Reson.
1985, 65, 335-360.
(23) J eener, J .; Meyer, B. H.; Bachman, P.; Ernst, R. R. Investigation
of exchange processes by two-dimensional NMR spectroscopy.
J . Chem. Phys. 1979, 71, 4546-4553.
(24) Fitzwater, S.; Hodes, Z. I.; Scheraga, H. A. Conformational
energy study of tuftsin. Macromolecules 1978, 11, 805-811.
(25) Deisenhofer, J . Cristallographic refinement and atomic models
of a human Fc fragment and its complex with fragment B of
protein A from Staphylococcus aureus at 2.9 and 2.8 Å resolu-
tion. Biochemistry 1981, 30, 2361-2370.
(26) Crescenzi, O.; Amodeo, P.; Cavicchioni, G.; Guerrini, R.; Picone,
D.; Salvadori, S., Tancredi, T.; Temussi, P. A. δ Selective Opioid
Peptides Containing a Single Aromatic Residue in the Message
Domain. A NMR Conformational Analysis. J . Pept. Sci. 1996,
2, 290-308.
(27) Richardson, J . S. The anatomy and taxonomy of protein struc-
ture. Adv. Protein Chem. 1981, 34, 167-339.
(28) Bru¨schweiler, R.; Blackledge, M. J .; Ernst, R. R. Multi-confor-
mational peptide dynamics derived from NMR data: A new
search algorithm and its application to antamanide. J . Biomol.
NMR 1991, 1, 3-11.
(29) Suzuki, M. SPXX, a frequent sequence motif in gene regulatory
proteins. J . Mol. Biol. 1989, 207, 61-84.
(30) Reeves, R.; Nissen, M. S. The AT-DNA-binding domain of
mammalian high mobility group I chromosomal proteins. A novel
peptide motif for recognizing DNA structure. J . Biol. Chem.
1990, 265, 8573-8582.
(31) Marion, D.; Wu¨thrich, K. Biochem. Application of phase sensitive
two-dimensional correlated spectroscopy (COSY) for measure-
ments of 1H-1H spin-spin coupling constants in proteins.
Biophys. Res. Commun. 1983, 113, 967-971.
(11) Nagoaka, I.; Yamashita, T. Inactivation of phagocytosis stimu-
lating activity of tuftsin by polymorfonuclear neutrophils. Bio-
chim. Biophys. Acta 1981, 675, 85-93.
(12) D′Ursi, A.; Pegna, M.; Amodeo, P.; Molinari, H.; Verdini, A.;
Zetta, L.; Temussi, P. A. Solution Conformation of Tuftsin.
Biochemistry 1992, 31, 9581-9586.
(13) O’Connor, S. D.; Smith, P. E.; Al-Obeidi, F.; Pettitt, B. M.
Quenched molecular dynamics simulations of tuftsin and pro-
posed cyclic analogues. J . Med. Chem. 1992, 35, 270-2881.
(14) Valdeavella, C. V.; Blatt, H. D.; Pettitt, B. M. Simulations of
conformation of tuftsin and cyclic tuftsin analogues. Int. J . Pept.
Protein Res. 1995, 46, 372-380.
(15) Nishioka, K.; Obeyesekere, N. U.; McMurray, J . S. Enhanced
phagocytosis activity of cyclic analogues of tuftsin. Biochem.
Pharmacol. 1995, 49, 735-738.
(16) Motta, A.; Picone, D.; Tancredi, T.; Temussi, P. A. NOE
Measurements on Linear Peptides in Cryoprotective Solvents.
J . Magn. Reson. 1987, 75, 364-370.
(17) Motta, A.; Picone, D.; Tancredi, T.; Temussi, P. A. Low-
Temperature NMR Studies of Leu-enkephalins in Cryoprotective
Solvents. Tetrahedron 1988, 44, 975-990.
(32) Dauber-Osguthorpe, P.; Roberts, V. A.; Osguthorpe, D. J .; Wolff,
J .; Genest, M.; Hagler, A. T. Stucture and energetics of ligand
binding to proteins: Escherichia coli dihydrofolate reductase-
trimethoprim, a drug-receptor system. Proteins: Struct. Funct.
Genet. 1988, 4, 31-47.
(18) Amodeo, P.; Motta, A.; Picone, D.; Saviano, G.; Tancredi, T.;
Temussi, P. A. Viscosity as a Conformational Sieve. NOE of
Linear Peptides in Cryoprotective Mixtures. J . Magn. Res. 1991,
95, 201-207.
(19) Schichman, S. A.; Amey, R. L. Viscosity and local liquid structure
in dimethyl sulfoxide-water mixtures. J . Phys. Chem. 1971, 75,
98-102.
J M980442+