The NMR assignment of both peptides was accomplished following the traditional methodology based on 2D 1H1H NOESY and TOCSY. Also, 13C edited HSQC data at natural abundance were used to assign carbon resonances and to solve ambiguities in the 1H spectra to derive 98% 1H and 95% 13C assignments for both isomers. The chemical shifts of CIGB-300 and CIGB-300iso (Table S1) were compared to identify possible differences in peptide conformations. Overall, these differences were found to be rather small. The N-terminal amino acid chain corresponding to the TAT region G1-Q13 does not show considerable variations because chemical shift differences were less than 0.1 ppm for protons and 0.3 ppm for carbons. In contrast, residues in the cyclic part of the peptides (C15 – C25) showed notable differences (Figure S3). Inspection of the backbone proton chemical shifts revealed the largest differences for the Hα protons of M17 and H21 and the HN protons of S18 and R20. This was the first indication that structural variations between the two CIGB-300 and its isomer are located in the cyclic peptide region. Due to a specific amino acid’s L to D configuration change, the side chain orientations undergo a more pronounced alteration. Consequently, we also examined the side chain protons resonances, where we found that only H21 demonstrated an average difference of 0.15ppm, while for the other amino acids, it consistently remained below 0.05ppm. The carbon chemical shift comparison for the side chains yielded small changes only for C15, M17, H21, L22, and C25. At least one minor conformation was identifiable in the NMR spectra of both CIGB-300 isomers by some very weak extra signals, especially in the cyclic part in the corresponding TOCSY and NOESY spectra. These additional cross-peaks were assigned (Table S2). Because of the presence of three prolines, P11, P12, and P19, in CIGB-300, we consider that the additional observed conformers could be due to cis-trans equilibria, but in none of the spectra, we found additional signals for prolines. We also calculated δ(Cβ) - δ(Cγ), confirming that all prolines of the major conformer of both CIGB-300 isomers are in the trans conformation.

It has been reported that cysteine and mostly histidine residues are prone to racemization during peptide synthesis (Kovacs et al. 1985; Han et al. 1997; Yang et al. 2022). Based on that fact and the results from the chemical shift differences between CIGB-300 and CIGB-300iso, seven diastereomers with all combinations of D and L residues at positions C15, H21, and C25 were synthesized. Figure 1 shows the 1D 1H-NMR spectra of these seven isomers compared to CIGB-300 and CIGB-300iso. An eye inspection revealed that the L-C15/D-H21/L-C25 isomer corresponded best to the spectrum of CIGB-300iso. A quantification by calculating the variance of the entire 1D spectrum further corroborates that this isomer is the only one where all the sidechain signals and the extra signals corresponding to minor conformations match CIGB-300iso.

Comparison of the 1D 1H NMR spectra of CIGB-300 (blue), and CIGB-300iso (red) with the spectra of synthesized diastereomers (black): DLL, LDL, LLD, DDL, DLD, LDD, DDD, where each letter corresponds to the L or D configuration of C15, H21, and C25 respectively. The variance of the corresponding synthesized isomers in parenthesis is reported in respect to the CIGB-300iso spectrum

In the next step, we compared the 2D NOESY spectra of CIGB-300 and CIGB-300iso to find possible structural rearrangements due to D-His21. Despite minimal changes in the distances of protons in most residues, NOE contacts for H21 revealed notable differences (Fig. 2). The NOESY spectra of CIGB-300iso contained, for example, NOEs between the sidechains of H21 and L22, not present in the CIGB-300 (red cross peak assignment in the top NOESY section of Fig. 2). Meanwhile, the β-protons of H21 showed contacts to the amide proton of S18 in CIGB-300 (blue assignment in the second NOESY section from the top in Fig. 2) which are not present in CIGB-300iso. The same holds for one β-proton of S18, which shows only in CIGB-300iso a contact to the amide proton of H21 (red assignment in the second spectrum from the bottom in Fig. 2) but not in CIGB-300 (missing cross peak in the bottom spectrum in Fig. 2). Another important difference was found for the orientation of the W16 side chain, which showed contacts to the amide proton of M17 in CIGB-300, but to the α-proton of C15 in CIGB-300iso. This different orientation could explain why M17 shows the biggest difference in the Hα/Cα chemical shift analysis.

Characteristics NOE contacts that are only present either in CIGB-300 (blue) or CIGB-300iso (red). The corresponding proton positions are indicated in the structural representation by blue and red arrows for the cyclic part of the CIGB-300 isomers. In the blue box, the absent peak 18Hβ-21HN for CIGB300 is indicated by dotted lines at the corresponding chemical shift

For a very detailed comparison of CIGB-300 and CIGB-300iso and to further study the influence of D- instead of L-His21, their three-dimensional structures based on the distance constraints derived from NOEs and dihedral angles were determined by simulated annealing with the Xplor program package. A superposition of the 20 structures of lowest energies for both peptides of the region corresponding to the linear part (residues Gly1 – βAla14) showed an extended but otherwise featureless conformation for both peptides and no restricted orientation towards the cyclic parts (Fig. S4). On the other hand, the superposition of the cyclic region showed a much more well-defined structure in both peptides (Fig. 3A and C). The cyclic part of the CIGB-300 revealed three loops (Fig. 3B), with the first loop formed by C15 to S18, the second loop by S18 to H21, and a slightly less defined third loop from L22 to C25. The three loops show distances between Cα of residue i and i + 3 shorter than 7Å, classifying them as β-turns (Lewis et al. 1973; Chou 2000). The first and third loops are open-turns (Chou 2000) with an average distance between the Cα carbons of C15 and S18 of 5.47 ± 0.56 Å (cyan in Fig. 3B) and L22 and C25 of 5.79 ± 0.91 Å (magenta in Fig. 3B). The second loop is stabilized by a hydrogen bond between the carbonyl group of S18 and the amide proton of H21 (black dotted line in Fig. 3B) leading to an average distance between Cα of the mentioned residues of 5.20 ± 0.27 Å. The HN temperature coefficient also supports the hydrogen bond formation (Fig. S8).

NMR-derived structures of CIGB-300 and CIGB-300iso. (A) Superposition of the 10 NMR structures with the lowest energies of peptide CIGB-300. The TAT region from G1 to βA14 is depicted with narrow lines and the cyclic part from C15 to C25 in a licorice representation. (B) The backbone of the different β-turns present in the structure of CIGB-300 are highlighted, the first β-turn from C15 to S18 in light blue, the second β-turn from S18 to H21 in dark blue, and the third β-turn from L22 to C25 in magenta. (C) Corresponding superposition of the 10 lowest energy NMR structures of peptide CIGB-300iso. (D) Representation of the small β−hairpin in the CIGB300iso. Black dotted lines in (B) and (D) represent hydrogen bonds

The high-resolution NMR structures of CIGB-300 allowed a classification of the turn conformations. Multiple and continuous turn motives have been described before, and 58% of β-turns occur as multiple turns (Hutchinson and Thornton 1994; Guruprasad et al. 2000, 2001). Even if double turns are the most common, there are reports of triple and higher-order multiple turns (Hutchinson and Thornton 1994). β-turns can be grouped into types I, II, I´, II´, VIa1, VIa2, VIba, and VIII based on the backbone dihedral angles of residue i + 1 and i + 2 (Figure S6). β-turns that do not fit in any of those groups are classified as type IV (Venkatachalam 1968; Richardson 1981; Hutchinson and Thornton 1996). 1/3 of the β-turns in protein correspond to type IV, subdivided into types IV, IV1, IV2, IV3, and IV4 (de Brevern 2016).

The Φ and Ψ dihedral backbone angles of the first and third turn of CIGB-300 do not match with any of the classical groups (Fig. S6), so they were assigned to the miscellaneous category type IV (Fig. S5A and Fig. S5B). The analysis of P19 and R20 dihedral angles allows the classification of a β-turn type IV3 (Fig. 4) (de Brevern 2016). It should be noted that it has been reported that proline in position i + 1 and a trans conformation generally is involved in β-turns of type I and II (Weißhoff et al. 1995). The β-turn type IV3 is similar to type I in its dihedral angle of residue i + 1 but different in its dihedral angle of residue i + 2. Additionally, type IV3 β-turns found in proteins are not associated with either α-helices or β-sheets (de Brevern 2016), a behavior that we also observed in CIGB-300. Also, notice that amino acid i + 3 from the first loop is shared with amino acid i from the second loop, and the third loop continues without common residues otherwise found in triple ββ(i, i + 3)β turns(Hutchinson and Thornton 1994; Guruprasad et al. 2000, 2001).

Ramachandran plots of CIGB-300 (left) and CIGB-300iso (right). The Θ and Ψ values are given for P19 and R20 in the ensemble of twenty lowest energy NMR structures. The green circles represent the average angles for each residue and the shift from residue i + 1 to i + 2 for β-turn classification is represented with an arrow, based on the Thorthon methodology (Hutchinson and Thornton 1994). CIGB-300 (blue, left) represents a distribution characteristic of β-turn type IV3, and the CIGB-300iso (red) a characteristic of β-turn type I

The structure of CIGB-300iso shows remarkable differences from CIGB-300. The dihedral angles for W16, M17, and L22 correspond to the β-sheet region of the Ramachandran plot (Fig. S5C), and thus an extended conformation. It gets stabilized by hydrogen bonds between the HN of S18 and CO of D-H21, HN of D-H21 and CO of S18, and HN of G23 and CO of W16 CO (black dotted lines in Fig. 3D). The HN temperature coefficients for CIGB-300iso were different from that of CIGB300 (Fig. S8) and support the hydrogen bond pattern found from these structural elucidations. Interestingly, the presence of D-H21 reduces the number of β-turns to just one between residues S18 to H21. The hydrogen bound between the carbonyl of S18 and the amide proton of H21 restricts the average distance between the Cα carbons of the mentioned amino acids to 5.02 ± 0.26 Å.

The dihedral angle analysis of amino acids P19 and R20 reveals a β-turn type I instead of the type IV3 observed in the CIGB-300 (Fig. 4). D-amino acids can promote the stabilization of the β-turns, where all systematic studies found the D-amino acids in positions i + 1 and i + 2 (Imperiali et al. 1992; Weißhoff et al. 1999; Mitchell and Smith 2003), but best to our knowledge we did not find any report for position i + 3. Anyhow, literature reports show D-amino acids in β-turn type I but in position i, and type VI in position i + 3 (Mitchell and Smith 2003). The β-turn type I is not well known to promote β-hairpin in natural proteins, but there are some reports, especially in synthetic peptides (Sibanda and Thornton 1985; Alba et al. 1999). β-turns are generally considered the initiation site for β-sheet formation, and their sequence plays an important role in the stabilization and classification of β-hairpins(Alba et al. 1999).

As we have discussed, the D- or L-form of H21 determines the super secondary structure of the CIGB-300 peptide. The main difference at the i + 3 turn position is not only a classification matter; the change in the super secondary structure leads to different orientations of the local side chains and, therefore, probably differences in the activation of the molecular targets. In a previous alanine scanning, W16 and L22 were identified as important amino acids for anticancer activity (data not shown). The two CIGB-300 diastereomers show various sidechain orientations. In CIGB-300 residue, W16, M17, and H21 face each other, the S18 is oriented to the center of loop one, and the side chains of L22 and T24 are more isolated and exposed. On the other hand, in CIGB-300iso, the side chains of M17, D-H21, and L22 are on the same side of the cyclic peptide part, whereas for W16, S18, and T24, the side chains are oriented towards the opposite side. R20 shows a very similar orientation in both peptides. Notably, the deprotonation of histidine due to pH change in both peptides observed in their NMR spectra promote very small changes in the side chains according to the calculated variance of the aliphatic region in the 1D 1H NMR spectra recorded at pH 3 and 7 (s2 = 0.0152 for CIGB-300 and s2 = 0.0243 for CIGB-300iso). This indicates that the presented structure determined at low pH is very similar to the structure at a more physiological pH.