Characterization of cyclic peptides containing disulfide bonds

Abstract

Unlike linear peptides, analysis of cyclic peptides containing disulfide bonds is not straightforward and demands indirect methods to achieve a rigorous proof of structure. Three peptides that belong to this category, p-Cl-Phe-DPDPE, DPDPE, and CTOP, were analyzed and the results are presented in this paper. The great potential of two dimensional NMR and ESI tandem mass spectrometry was harnessed during the course of peptide characterizations. A new RP-HPLC method for the analysis of trifluoroacetic acid is also presented. It is robust, simple, and efficient compared to the currently available methods.

Graphical Abstract

1. Introduction

Developments of opioid peptides as promising nonaddicting analgesics [1-3] began in 1770s when the discovery of opiate receptors led to the discovery of two endogenous opioid peptides Met⁵-Enkephalin (H-Tyr-Gly-Gly-Phe-Met-OH) and Leu⁵-Enkephalin (H-Tyr-Gly-Gly-Phe-Leu-OH). Given the multitude of reported side effects of currently available analgesics, a search for nonaddicting analgesics based on endogenous opioid peptide ligands is a timely endeavor. In addition to the nonaddicting properties, other features such as improved biological activity, bioavailability, and reduced toxicity drew attention during these studies.

With the above objectives, a strategy that has been employed in opioid peptide research includes replacing natural amino acids of endogenous peptides with modified amino acids and peptidomimetic ligands [4-10]. In comparison to those endogenous ligands that are linear, cyclization was designed in this class of synthetic peptides in order to reduce conformational freedom as it enhances the receptor selectivity along with other advantages [10]. Cyclic opioid peptides such as DPDPE [Tyr-c(D-Pen-Gly-Phe-D-Pen); D-Pen (2,5)-enkephalin] and p-Cl-Phe-DPDPE [Tyr-c(D-Pen-Gly-p-Cl-Phe-D-Pen)] in which the 2- and 5- amino acid residues of enkephalin have been replaced by a sterically constrained derivative D-Pen (penicillamine; β, β -dimethylcysteine) belong to this category. Both peptides contain a disulfide bond via the two D-Pen derivatives and were proven to be potent, stable and possessing δ receptor selectivity. CTOP [D-Phe-c(Cys-Tyr-DTrp-Orn-Thr-Pen)Thr-NH₂; [Cys(2),Tyr(3),Orn(5),Pen(7)amide] is a synthetic opioid peptide that contains a disulfide bond via Cysteine and D-Pen moieties, also proven to be potent, but showing μ receptor selectivity.

During the synthesis of a cyclic peptide that contains a disulfide bond, the linear form is usually assembled first. The oxidation of the two thiols to form the disulfide bridge is performed towards the end of the synthetic plan. Once such a cyclic peptide is synthesized, a variety of analytical challenges to prove its structure are presented. Unlike a linear peptide, achieving complete amino acid information for such cyclic compounds via conventional amino acid analysis will not be feasible and one may have to seek alternative means, such as mass spectrometry (MS).

Also, conventional Edman degradation is not amenable to the sequencing of cyclic peptides. Here again, MS is required to obtain amino acid sequencing information. But in this case, perhaps only a few sequencing fragments may be evident, though observing sufficient fragments to prove the sequence is not guaranteed. Under the circumstance of not observing sequencing fragments, the question arises on how the structure can be verified. A ready answer could be the advanced two-dimensional NMR techniques, such as correlation spectroscopy (COSY) and nuclear overhauser effect spectroscopy (NOESY) that may offer valuable structural information.

Furthermore, if a cyclic peptide such as DPDPE that contains a disulfide bond is prepared from its linear precursor, the two peptides differ in their molecular masses by only 2 mass units (daltons, Da). Under such circumstances, an important question arises: How would one ascertain whether there is any contamination by the linear precursor? Here, the subject peptide and the presumed impurity both contain the same amino acid skeleton. The only difference being that one has a disulfide bond whereas the other is a dithiol with the two additional protons.

Under a separation technique such as HPLC, the precursor may be represented as an impurity peak. However, observation of a single major peak, say in an extreme case with no observable impurities, is not an unambiguous answer because of the possibility that the linear precursor may not be retained under those particular separation conditions or it might get co-eluted with the major peak. Incorporation of other HPLC methods and /or other analytical methods such as CE with different separation principles in the analytical protocol seems to be logical and can be effective [11-12]. Also, a comparison with the linear precursor of the peptide is highly relevant. A simple and fast way to obtain this is by reducing the disulfide bond of the cyclic peptide. This reduction can be easily achieved by using a reducing agent such as dithiothreitol (DTT). This reagent has been used extensively, particularly on the studies of peptide and protein chemistry [13-17].

The work presented here covers the methods and results we obtained during analysis and structural verification of three cyclic opioid peptides p-Cl-Phe-DPDPE, DPDPE, and CTOP. Each peptide contains a disulfide bond, the first two via two D-Pen molecules and the latter via a cysteine and a D-Pen molecule. They were all fully characterized, identity verified, and the purity and potency were determined. As discussed below, all peptides needed indirect methods for their characterization. Our experimental protocols included DTT reduction of the cyclic peptide followed by LC-MS. For p-Cl-Phe-DPDPE and DPDPE, LC-MS was coupled with fraction collection and electrospray ionization tandem MS (MS-MS) as unequivocal proof of structure.

Trifluoroacetic acid (TFA) is a reagent that has many applications, particularly in peptide/protein chemistry, including their synthesis, purification and analysis. Sometimes, in addition to TFA, the purified peptide samples contain acetic acid (AA) as well and therefore, simultaneous determination of these acids is desirable and has been achieved by RP-HPLC [18] and CE with indirect UV detection [19].

Quantification of TFA contained in each sample was performed using an efficient and robust HPLC method developed during our studies. Development of this RPHPLC method is also reported in this paper. The method is efficient, each HPLC run taking 5 min. It is simple and can be easily integrated into any analytical laboratory that has an HPLC system. As illustrated here, the method is applicable for the analysis of both TFA and AA.

Additional information is provided as supplementary data.

2. Experimental

2.1. Reagents and Materials

The reagents used in our experiments were acetonitrile, HPLC-grade and tris from Mallinckrodt, (Paris, Kentucky, USA), trifluoroacetic acid, >99.0% (GC), DL-dithiothreitol, DMSO-d₆, and D₂O from Sigma-Aldrich (St. Louis, Missouri, USA). The mobile phases were prepared in Milli-Q water.

All peptide samples were supplied by the PolyPeptide Laboratories, San Diego, CA.

2.2. Instrumentation and Methods

2.2.1. RP-HPLC for peptide purity determination

Analysis was performed on an Agilent 1100 HPLC system with diode-array detection (DAD). 10 μL of each sample (1 mg/mL in water) was injected on a Phenomenex Gemini-NX C18 column, 5 μm, 110Å, 250 × 4.6 mm. This column was purchased from Phenomenex (Torrance, California, USA). The detection wavelength was set at 215 nm. Mobile phase A was 0.05% TFA in water and mobile phase B was 0.05% TFA in acetonitrile. The separation was obtained at a flow rate of 1 mL/min with a gradient program that allowed for a 20 min step that raised eluent B from 20-55%. Equilibration at 20% B was then performed for 10 min for a total analysis time of 30 min.

2.2.2. RP-HPLC for TFA and AA analysis

Analysis was performed on an Agilent 1100 HPLC system with DAD. 20 μL of each sample (1 mg/mL in water) was injected on a Phenomenex Synergi Hydro-RP C18 column, 4 μm, 80Å, 150 × 4.6 mm. This column was purchased from Phenomenex (Torrance, California, USA). The detection wavelength was set at 215 nm. Mobile phase A was acetonitrile and mobile phase B was 20 mM KH₂PO₄ adjusted to pH 2.5. The separation was obtained at a flow rate of 1 mL/min with a run time of 5 min.

2.2.3. LC-MS

Analysis was performed on a Thermo-Scientific LCQ Fleet MS and Finnigan Surveyor HPLC system. 5 μL of each sample (1 mg/mL in water) was injected on a Phenomenex Gemini-NX C18 column, 5 μm, 110Å, 250 × 4.6 mm. This column was purchased from Phenomenex (Torrance, California, USA). The detection wavelength was set at 215 nm. Mobile phase A was 0.1% TFA in water and mobile phase B was 0.1% TFA in acetonitrile. The separation was obtained at a flow rate of 0.5 mL/min with a gradient program that allowed for a 7 min step that raised eluent B from 35-50%. Eluent B was held at 50% for 6 minutes. Equilibration at 35% B was then performed for 10 min for a total analysis time of 23 min.

2.2.4. NMR

The instrument used to perform NMR analysis was a Varian Mercury 400. 11.7 mg sample was dissolved in 0.5 mL DMSO-d₆ with two drops of D₂O.

3. Results and Discussion

3.1. Analysis of p-Cl-Phe-DPDPE

3.1.1. RP-HPLC experiments

When we analyzed the peptide sample by RP-HPLC, the result indicated a high degree of purity, 99.5 ± 0.1%, n=3 (Figure 1), based on the assumption that all components have equal UV absorptions at the detection wavelength. Furthermore, it has to be recognized that the level of purity is dependent on the specific analytical methods used. The HPLC purity result was in agreement with the manufacturer's data.

FIG. 1.

RP-HPLC chromatogram of p-Cl-Phe-DPDPE. Inset is the expanded baseline showing the minor/trace impurities.

In addition to HPLC, CE experiments were also performed. They indicated a purity result of 99.8 ± 0.1% and are in close agreement with the HPLC result.

3.1.2. Determination of TFA and Acetic Acid

TFA was present in the sample as the counter ion and also needed quantification. This was achieved using a polar endcapped C18 column that provides extreme hydrophobic and slight polar selectivity.

Five TFA standards ranging from 10μg/mL to 500μg/mL were prepared and chromatographed to construct a standard curve. The equation obtained from the standard curve was y = 0.3387x – 1.3203, with R² = 1.

Based on the results of three test samples, the peptide contains 10.18 ± 0.12% TFA.

Although the three particular peptides, p-Cl-Phe-DPDPE, DPDPE, and CTOP that are discussed in this paper do not contain any acetic acid, we have occasionally encountered some peptides containing both TFA and AA. Therefore, we extended the method for the detection of AA as well. As claimed by Phenomenex, the Synergi Hydro-RP column manufacturers, this method is capable of resolving AA from eight other acids present in a mixture. Our observation is that the method is capable of detecting AA and is well separated from TFA as is illustrated in Figure 2.

FIG. 2.

HPLC chromatogram of mixture of TFA (250μL/mL water) and AA (250 μL/mL water).

Elemental nitrogen data indicated a peptide content of 80.3 ± 0.1%, n=2 and Karl Fischer water determination yielded a result of 5.5 ±0.2%, n=3 for the water content.

On the material balance, considering the data gathered so far, a peptide content of 80.3%, water content of 5.5%, and TFA content of 10.2% yield a total 96%. Impurities may account for the remainder. Underlying method errors in each technique also become a contributing factor in the final purity result. However, our efforts were next directed towards proving the structure of the peptide and verifying whether there is any contamination by the linear precursor peptide.

3.1.4. MS, AAA and AASA

Mass spectrometry revealed an (M+H)⁺ ion at 679.99 and an alkali ion, the sodium adduct (M+Na)⁺ at 702.05 revealing the expected average molecular mass 680.2 of the peptide, but the MS spectrum was devoid of any sequencing fragments to prove the structure.

Usually, the information on the amino acid composition and their sequence are obtained by conventional amino acid analysis (AAA) and Edman degradation methods. But for p-Cl-Phe-DPDPE, AAA information was incomplete and amino acid sequence analysis (AASA) was not amenable because of the cyclic structure.

Thus, as evidenced above, despite the fact that the subject peptide can be considered as a small molecule, its cyclic nature prevented the use of direct conventional techniques MS, AAA, and AASA for its analysis.

3.1.5. NMR of p-Cl-Phe-DPDPE

In the absence of sequencing fragments in MS, and inability to obtain any AAA and AASA information, we were prompted to use NMR for structural scrutiny.

The ¹H-NMR spectrum of p-Cl-Phe-DPDPE is shown in Figure 3.

FIG. 3.

¹H-NMR spectrum of p-Cl-Phe-DPDPE.

The NMR chemical shifts and the respective proton assignments are presented in Table 1.

Table 1.

The ¹H-NMR chemical shifts and the respective proton assignments.

Chemical Shift (ppm)	Pattern J(Hz)	Relative Intensity	Proton Assignment
7.26	d (8.4)	2	11, 13
7.20	d (8.8)	2	10, 14
7.03	d (8.8)	2	26, 30
6.66	d (8.8)	2	27, 29
4.32	s	1	2
4.26	dd (4.4, 10 .4)	1	7
4.25	s	1	18
4.23	d (14.8)	1	16b
4.20	t (7.6)	1	23
3.21	d (14.8)	1	16a
3.02	dd (4.4, 14.0)	1	8b
2.89	m	2	24
2.78	dd (10.4, 14.0)	1	8a
1.27, 1.22	s	6	20, 21
1.26, 0.86	s	6	4, 5

¹H- NMR spectrum of p-Cl-Phe-DPDPE showed twelve methines (CH) including eight olefinic protons at δ_H = 6.66, 7.03, 7.20 and 7.26 ppm, and four methyl (CH₃) singlets at δ_H = 0.86, 1.22, 1.26 and 1.27 ppm. Several multiplets observed between δ_H 2.75 ppm and 4.23ppm are attributed to three methylenes (CH₂). The ¹H-¹H COSY spectrum, shown in Figure 4, revealed six proton-bearing spin coupling units and are illustrated with bold bonds as shown in Figure 4a. These six units are H-7/H₂-8, H-10/H-11, H-13/H-14, H-23/H₂-24, H-26/ H-27, and H-29/H-30. The presence of two disubstituted benzene rings(AA′BB′ system) was easily recognized by the ¹H resonances{ [δ_H 7.26, (2H, d J = 8.4 Hz, H-11 and H-13), δ_H 7.20, (2H, d J = 8.8 Hz, H-10 and H-14)] and [δ_H 7.03, (2H, d J = 8.8 Hz, H-26 and H-30), δ_H 6.66, (2H, d J = 8.8 Hz, H-27 and H-29)] }. ¹H-¹H COSY data also exhibit two doublets ( δ_H 3.21/ 4.23) from AB spin system with large germinal coupling constants ( J = 14.8 Hz), which assigned to the isolated methylenes H₂-16. The connection of C-24/C-25 and C-8/C-9 was achieved by the NOE correlations from H₂-24 to H-26/H-30 and from H₂-8 to H-10/H-14. The NOE correlations from H-18 to CH₃-20/CH₃-21 and from H-2 to CH₃-4/CH₃-5 established the connection of C-2/C-3 and C-18/C-19, and also placed CH₃-20/CH₃-2 and CH₃-4/CH₃-5 on C-19 and C-3, respectively. In conclusion, the NOE correlations of CH₃-4 and CH₃-20 indicated that p-Cl-Phe-DPDPE was a cyclic peptide containing a disulfide bond between C-3 and C-19. The NOESY spectrum is shown in Figure 5.

FIG. 4.

¹H-¹H COSY spectrum of p-Cl-Phe-DPDPE.

FIG. 5.

NOESY spectrum of p-Cl-Phe-DPDPE.

3.1.6. DTT reduction

As shown above, the peptide structure was proven by NMR techniques ¹H-NMR, ¹H-¹H COSY and NOESY.

However, the question of whether there is any contamination by the linear precursor still remains. If the sample is contaminated with its linear precursor, we needed to address the following three potential scenarios:

• 1)Co-elution with the major peak in HPLC
• 2)The linear precursor's presence as a minor peak
• 3)The linear precursor not eluting under the HPLC conditions being used.

When the structures of the two compounds are considered, the precursor molecule is a linear peptide that contains two thiol groups whereas the subject compound is a cyclic peptide with a disulfide bond. With such a structural difference between the two compounds, one would expect them to resolve during HPLC. On an RPHPLC column, the linear peptide should be more retained due to the polarity imparted by the two thiol groups. It was apparent that additional experiments including the linear precursor were needed, so we attempted to obtain this by reducing the cyclic peptide. Here, the cyclic peptide was opened up at the disulfide bond using DTT in the presence of tris buffer. The peptide (2 mg) was dissolved in 1 mL 50 mM tris buffer. DTT was added to the peptide solution (in excess, 1.5 mg) and kept stirring. The LC-MS experiments discussed below were performed after the solutions were stirred overnight.

Figure 6 shows an HPLC profile of the DTT-treated sample solution. There were two early eluting peaks which were shown to be DTT artifacts, a major peak that eluted around the similar time (10.7min) as the untreated sample (Figure 1) and a new minor peak at 13.8 min, presumably the linear form. We endeavored to prove that this peak was the linear form of the peptide using LC-MS.

FIG. 6.

HPLC chromatogram of p-Cl-Phe-DPDPE treated with DTT over 30 min.

3.1.7. LC-MS

In addition to HPLC experiments discussed above, treatment of the sample with DTT was followed by LC-MS analysis. During the method transfer, the conditions used for HPLC experiments needed modification to suit the LC-MS format.

The HPLC chromatogram after DTT treatment under LC-MS conditions is shown in Figure 7. The first peak at 6.48 min was shown to be a DTT artifact and due to the adjustment of the method, the other two peaks eluted a little earlier than what was observed under the HPLC conditions.

FIG. 7.

HPLC chromatogram of p-Cl-Phe-DPDPE upon treatment with DTT under LC-MS conditions.

However, as seen in Figures 6 and 7, disregarding the DTT artifacts, the proportions of the other two peaks are different. This is because of the time dependence of the DTT reduction. We performed additional experiments to demonstrate this effect using DPDPE and another cyclic opioid peptide CTOP. The first peak at 6.48 min of Figure 7 had neither a specific UV profile nor appeared on the total ion chromatogram (TIC) spectrum during MS studies. Mass spectrometry proved that the peaks at 8.46 (Figure 8) and 10.84 min (Figure 9) were the cyclic form of the peptide (MW 680) and the reduced form (MW 682), respectively. Thus, even though the molecular mass difference is only 2 Da, each form can be identified under the LC-MS conditions.

FIG. 8.

MS spectrum of DTT-treated sample at retention time 8.46 min.

FIG. 9.

MS spectrum of DTT-treated sample at retention time 10.84 min.

Fraction collection was performed on the peak at 10.84 min, which is presumably the reduced form of the peptide. Mass spectrometry and MS-MS were performed on the collected fraction. The fragmentation pattern of this peak is shown in Figure 10.

FIG. 10.

MS fragmentation pattern of the linear form, the reduced p-Cl-Phe-DPDPE.

The MS-MS spectrum of the peak eluted at 10.84 shows sufficient sequencing information of the linear peptide yielding direct evidence on the proof of the correct structure for the cyclic peptide.

Thus we have verified that the later eluting peak in the DTT treated sample (Figure 6) is indeed the linear precursor peptide. Therefore the HPLC method we used is capable of resolving the cyclic peptide and its linear precursor peptide.

3.2. DPDPE

Here again, we have performed a full analysis of the peptide but only LC-MS data are presented. Although the MS did not show sufficient fragments to confirm the sequence, the reduced form yielded all the expected fragments to prove the structure.

DTT was used to perform the reduction of DPDPE. Similar to p-Cl-Phe-DPDPE, this reaction was also found to be time dependent. Both the linear and cyclic forms were observable in the HPLC profile Figure 11. The corresponding MS spectra indicated the correct molecular weights 646 and 648 for DPDPE and the reduced linear form, respectively. Fraction collection was performed on the linear form and analyzed by MS and MS-MS. The fragmentation pattern of the reduced peptide, the linear form is shown in Figure 12.

FIG. 11.

LC and TIC chromatograms of DTT-treated DPDPE.

FIG. 12.

MS fragmentation pattern of the linear form, the reduced DPDPE.

3.3. CTOP

Similar to the analyses of the two peptides p-Cl-Phe-DPDPE and DPDPE described above, we have also successfully analyzed the cyclic peptide CTOP (data not presented here).

While recognizing the careful planning and the effort expended to synthesize a well purified peptide, it is to be noted that analysis also demands a similar well-planned effort to verify its identity and integrity. The work presented here clearly demonstrates that.

The reagent we used to reduce the disulfide bond in our experiments, DTT, is quite reliable and we have extended the same method to analyze other disulfide containing cyclic peptides in addition to those mentioned here. However, other methods have been reported that are equally applicable, including reagents such as performic acid [20], sodium sulfite [21], and tris(2-carboxyethyl)phosphine (TCEP) [16]. More recent advances include electrochemical methods [22-23].

Of course, the utility of each method depends upon the application. Though observing the molecular mass difference of 2 Da between the cyclic and linear peptide for small peptides, as reported here, poses no problems, our methods may be of limited use on larger peptides and proteins containing a multitude of disulfide bonds, because of the broad isotopic distributions observed in MS. In such circumstances, β-mercaptoethanol has been shown to be a useful reagent [24]. Using this reagent, an amplification of the molecular mass difference by 78 Da is obtained for each disulfide bond reduced.

4. Conclusions

Reduction via DTT, followed by LC-MS, is a convenient way to elucidate the structure of cyclic peptides containing disulfide bonds and to prove the presence and/or absence of any linear precursor peptide. The MS fragmentation pattern of the linear form implies the correct sequence of the cyclic form of the peptide. NMR methods such as ¹H-NMR, ¹H-¹H COSY, and NOESY are also extremely useful tools that can be employed for structural elucidations of cyclic peptides when the standard methods are not amenable.

The RP-HPLC method reported in this paper for the analysis of TFA is very efficient, highly reliable, accurate and robust. It can be easily extended for analyzing other acids, e.g. AA.

Highlights.

• Challenges in the characterization of cyclic peptides containing disulfide bonds

• Incorporation of HPLC with ESI MS-MS, LC-MS, and NMR to analyze cyclic peptides

• A novel RP-HPLC method for the analysis of trifluoroacetic acid

• Use of dithiothreitol for reducing disulfide bonds in cyclic peptides