Author Correction: Traceless native chemical ligation of lipid-modified peptide surfactants by mixed micelle formation

A Correction to this paper has been published: https://doi.org/10.1038/s41467-021-22508-2

T emplated chemistry is a hallmark of biological processes, bringing reactants into close proximity with one another, thus greatly increasing effective molarity and dramatically raising the rate of reaction [1][2][3] . Many reactions that would otherwise not occur in practical time periods are facilitated by templating. Nucleic acid [4][5][6][7][8] and protein templates [9][10][11] are widely appreciated in biology and biomimetic chemistry, and a number of seminal studies have exploited such templates for programming the synthesis of complex molecules such as oligopeptides [12][13][14] . All living cells possess lipid membranes, which in addition to acting as compartments and barriers, also act as scaffolding structures for guiding biological reactions. By restricting substrates in two-dimensions, local concentration is increased. Signaling from clusters of membrane bound proteins benefits from this phenomenon, as does the biosynthesis of membrane components such as the bacterial cell wall. Given that templated protein synthesis is a principal tenet of the central dogma of molecular biology, it would be of particular interest to understand if lipid moiety mediated self-assembly of mixed micelles 15 can drive the selective formation of discreet polypeptides. Here, the mixed micelles would be composed of two different lipid detergents, bearing polar head groups that would be reactive with one another. If successful, mixed micelle assistance could provide a straightforward mechanism to accelerate sluggish chemical reactions and could be potentially be a useful synthetic method if mechanisms existed to make the process traceless through removal of lipid auxiliary groups.
To address these issues, we focused on applying lipid modification to accelerate oligopeptide coupling, specifically through native chemical ligation (NCL) 16 . NCL is a chemoselective reaction that occurs at neutral pH between two unprotected peptides, one bearing an N-terminal cysteine residue and the other containing a C-terminal thioester. NCL is a robust method [17][18][19] for the synthesis and derivatization of large peptides, proteins [20][21][22][23][24] , and even nucleic acids 13,25,26 and phospholipids [27][28][29] . Despite its widespread use in chemical protein synthesis, NCL has several limitations. First, C-terminal peptide thioesters are commonly prepared as alkyl thioesters, which simplifies synthesis, purification and storage, but results in precursors with low reactivity 30 . Second, although some thiol-free NCL strategies have been developed recently [31][32][33] , addition of thiol catalysts is often required in conventional NCL method to promote the in situ formation of more reactive thioesters 30,34 . Thiol catalysts, such as MPAA 30 introduced by Kent and coworkers, have been widely applied in synthetic protein chemistry 35,36 . Despite their proven robustness, thiol catalysts may have the potential for unwanted side reactions 31,33 . Furthermore, thiol additives can interfere with the NCL-desulfurization synthetic strategy 37,38 by acting as radical scavengers 39 , which may necessitate a removal step prior to desulfurization 32,40 . Third, the modest reaction rates of NCL make it necessary to use millimolar concentrations of the substrate peptide fragments for practical application 13,34,41,42 .
Development of a lipid-facilitated NCL methodology that allows the rapid production of ligated polypeptides, using low concentrations of reactants 43 and in the absence of thiol additive catalysts, would be of considerable value. Herein, we describe a strategy for lipid-facilitated acceleration of NCL via micelle mixing (Fig. 1). Our approach takes advantage of a peptide bearing a C-terminal alkyl thioester (1a-d, 5a,b or 6a,b), which is capable of forming micelles in aqueous solution. Addition of an Cys-peptide containing a photolabile lipid 44,45 moiety near the Nterminus (2a,b) leads to rapid NCL. We hypothesize that the close proximity [46][47][48][49][50] of the alkyl thioester and cysteine within formed micelles increases the rate of transthioesterification substantially, even at low concentrations of reagents (1 mM as standard concentration) and in the absence of thiol catalysts.
Although there are several reported cases 51-53 of facilitated NCL in hydrophobic environments or in lipid bilayers, in our work, the surfactants are the reacting peptide segments, to which lipid moieties are covalently attached. Micelles are formed by the lipidmodified reactants themselves, which upon addition to the same solution, fuse to form mixed micelles. During NCL, the lipid moiety attached to the thioester peptide is eliminated in the initial thiol-thioester exchange. After NCL, the lipid moiety maintained on the product can be removed by a photochemical process. Thus, a traceless synthetic system can be established. We demonstrate the utility of this strategy through the successful one-pot synthesis of the natural antimicrobial peptide Magainin 2 and derivatization of the natural protein Ubiquitin (Ubi). Our studies suggest that micelle-mixing assembly may be a broadly applicable method for the acceleration of chemical reactions.

Results
Design and synthesis of model peptide fragments. To explore the potential of lipid-facilitated NCL, solid-phase peptide synthesis (SPPS) was used to construct the peptide LYRMG-OH, which was subsequently modified with linear aliphatic alkyl chains of different lengths to form the corresponding thioesters ( Fig. 1)-C8 (1a), C16 (1b), and C18 (1c). We note that LYRMG peptide thioesters with Cn alkyl chain will henceforth be denoted as LYRMG-αCOSCn; e.g. LYRMG-αCOSC8 (1a). As a control, we prepared LYRMG-αCOSC2 (1d), a short-chain peptide thioester containing an ethyl moiety on its C-terminus, which lacks the necessary hydrophobicity for micelle formation. This behavior was confirmed by dynamic light scattering (DLS) measurements, showing that at 1 mM concentration, no micelle formation was observed for 1d. Additionally, we synthesized Cys-peptides (2a,b) using a standard Fmoc-SPPS protocol. These peptide segments were adapted from the peptide model sequence CRANK-OH 18 . In our case, we substituted the arginine for the unnatural amino acid Dap(PhCn), which bears a photolabile lipid group 44,45 at the N β position of 2,3-diaminopropionic acid (Dap). The photolabile group possesses a linear aliphatic alkyl chain of either C8 (2a) or C16 (2b). Peptide 2c, where the photolabile group is absent, was also prepared as a control.
Kinetic analysis of lipid-facilitated model peptide ligations. We tested the NCL reactions between the oligopeptide thioesters (1a-d) and the cysteine-functionalized peptide substrates (2a-c) in a thiol-additive-free aqueous phosphate buffered system at pH 7.0, containing tris(2-carboxyethyl) phosphine hydrochloride (TCEP·HCl) to maintain a reducing environment.
Cysteine-based oligopeptide CDap(PhC16)ANK (2b) was treated with various oligopeptide LYRMG-αCOSZ thioesters (1a-d) (Fig. 2a). The ligation of 1a and 2b provided product 3b in quantitative yield within 180 min. Half-maximal product formation was achieved in 10 min, and 95% ligation yield in 70 min. The reaction between 1b and 2b was slower, with quantitative yield within 960 min (16 h), but only 60% ligation yield in 180 min. The reaction of 1c and 2b was slow; less than 50% ligation yield could be achieved after 960 min. Use of thioester 1d, which lacks a long aliphatic chain, presented less than 20% ligation yield with 2b after 960 min, indicating the importance of mixed micelle self-assembly in accelerating the reaction.
Of the tested oligopeptide thioesters, 1a (C8) promoted the greatest acceleration of the NCL. Therefore, we studied its reactivity against various cysteine-based peptide analogs (2a-c) (Fig. 2b). The ligation rate of 2b and 1a, with~60% ligation yield in 70 min and quantitative conversion after 420 min, was slower when compared to 2a and 1a. As expected, the control reaction of 2c with 1a did not generate ligated product.
When keeping the Cys-peptide substrate constant (Fig. 2a), we observed that the ligation rate decreases with increasing alkyl thioester chain length [LYRMG-αCOSC18 (1c) < LYRMG-αCOSC16 (1b) < LYRMG-αCOSC8 (1a)]. Micelles composed of shorter alkyl thioester chains are predicted to have a higher rate of dissociation 54,55 , resulting in facilitated micelle mixing, potentially explaining this result. However, when 1a was reacted with various Cys-peptides differing in lipid chain length (Fig. 2b), we found that C16 reacted slightly faster than C8. At this time, the relationship between reaction rate and lipid chain length is unclear. Our system is complicated by the reaction between two different amphiphiles introduced as micelles, which form mixed micelles, as indicated by the observed reaction. Mixing two amphiphiles with different critical micelle concentrations (CMCs) can often lead to complex micelle formation behavior [56][57][58][59] . Although beyond the scope of the current work, future studies using advanced characterization techniques could shed light on the kinetics of mixed micelle formation and their overall composition as a function of lipopeptide precursor.
To verify the proposed micelle formation-mixing mechanism, we carried out NCL experiments between 1a and 2b at multiple concentrations (Fig. 2c). When the concentrations of both peptide oligomers (1 mM, or 500 µM) were above their respective CMCs (~380 µM for 1a, and~130 µM for 2b) ( Supplementary  Fig. 23), conditions under which mixed micelles consisting of both reactants are expected to form, ligations were fast (completed within 70 min). The curves for 1 mM and 500 µM were similar, which suggests that once the concentration is above the CMC, there is a limited increase in ligation rate with increasing reactant concentration, as one would expect if micelle formation is the critical parameter for achieving rate acceleration. When the concentration of reactants (250 µM) is above the CMC of 2b but below the CMC of 1a [conditions in which monomers of 1a can fuse on micelles of 2b], the ligation reaction proceeded b NCL reaction between aliphatic alkyl peptide thioesters (1a-c, 5a or 6a) and photocaged cysteine-based peptides (2a,b), which yields the desired polypeptides (3a,b, 7 and 8, respectively). Control reactions were performed with peptides containing short alkyl chains (1d, 2c, 5b, and 6b). Subsequent photocleavage reaction of the ligated products (3a,b, 7 or 8) leads to the formation of the uncaged polypeptides (4, 9, and 10, respectively).
with an intermediate kinetic velocity (~70% ligation yield within 70 min). When the concentrations of the reactants (100 µM, or 50 µM) are below the CMCs of both reactants [conditions in which both 1a and 2b do not form micelles], ligations once again become slow (below 35% ligation yield for 100 µM, and below 17% ligation yield for 50 µM within 180 min). The significant difference in NCL kinetics at various reactant concentrations provides further evidence for the role of mixed micelle selfassembly in facilitating the reaction. In comparison, MPAAcatalyzed NCL reactions between the unlipidated precursors LYRMG-αCOSC2 (1d) and CDapANK (2c) were also performed ( Supplementary Fig. 55). At 500 µM, the MPAA-catalyzed ligation was considerably slower than the lipid-facilitated ligation ( Supplementary Fig. 36), which highlights the importance of mixed micelle self-assembly at low reactant concentrations.
Scope of the lipid-facilitated peptide ligations. To test the robustness and widen the applicability of our methodology, we generated more complex peptide ligation sites such as Asn-Cys and Leu-Cys (Fig. 3a). We synthesized peptide thioesters LYRMN-αCOSC8 (5a) and LYRML-αCOSC8 (6a) and treated them at room temperature with the N-terminal cysteine peptide CDap(PhC16)ANK (2b). The desired ligated polypeptides (7 and 8, respectively) were afforded in good yields (>80%) after 900 min. Control reactions of LYRMN-αCOSC2 (5b) and LYRML-αCOSC2 (6b) with 2b did not generate the corresponding ligation products, supporting the role of lipid modification in acceleration of these ligations.
We next studied the effect of the distance between the lipid chains and the reactive centers. We prepared the peptide CADap (PhC16)NK (S10) ( Supplementary Fig. 34), an analogue of 2b where the unnatural amino acid containing the photolabile lipid group was located at the third position, increasing the distance between peptide reactive sites. As expected, kinetics experiments showed that NCL reaction between 1a and S10 was slower than the ligation with 2b, demonstrating the importance of lipid chain positioning on reaction rate (Supplementary Figs. 35 and 54).
In classical NCL conditions, thiol catalysts, such as MPAA, are able to reverse nonproductive transthioesterification with the thiol moieties of non N-terminal cysteine residues 30 . NCL reaction between CK(PhC16)ANC (S12) (Supplementary Fig. 37) Photoliberation of the lipid-free peptides. Without purification, the nitrobenzyl photocaging group was removed using a standard photocleavage protocol 45 , liberating the lipid-free peptide. This photochemical process allowed the generation of the uncaged peptide product 4 in nearly quantitative yield (Fig. 3b, Supplementary Figs. 17 and 51). Decapeptides 9 and 10 were obtained using analogous conditions (Supplementary Figs. 18 and 19, respectively). In this manner, a 'traceless' synthetic system can be established, in which both lipid modifications (the C-terminal thioester and the photocaging auxiliary group) on the starting materials (1a, 5a or 6a, and 2b) are removed during the reaction process.
Construction of native peptides. Having achieved traceless peptide ligation by lipid-facilitated NCL, we sought to extend this approach to the synthesis of native peptides rather than Dapcontaining products. By utilizing the amino acid Lys(PhCn), which bears the photolabile lipid group at the N ε position of the lysine, we successfully synthesized the decapeptide LYRMGCK- Lipid-facilitated total synthesis of Magainin 2. Encouraged by these results, we next applied our methodology to the total synthesis of a natural peptide product. Magainin 2 is a 23aa cationic and amphipathic polypeptide that acts as a potent antibiotic in a variety of organisms ( Fig. 4) 60,61 . This natural antimicrobial peptide has been reported to be a difficult sequence to synthesize by Fmoc chemistry SPPS 62,63 . We devised a straightforward one-pot synthesis of Magainin 2 based on our lipidfacilitated NCL methodology (Fig. 4). The peptide thioester GIGKFLHS-αCOSC8 (11)  We note that sodium dodecyl sulfate (SDS) was required to completely solubilize the peptide fragment 12. Addition of this anionic detergent provides a satisfying solution for the issues related with low solubility of certain lipopeptides. Additionally, a control reaction between GIGKFLHS-αCOSC2 (S8) (Supplementary Fig. 28) and peptide 12 under the same conditions generated no ligated product ( Supplementary Fig. 33). By eliminating the need for thiol additives, our lipid-facilitated NCL strategy enables successive peptide ligation, photocleavage, and desulfurization in one-pot to afford the natural peptide Magainin 2 (15, 19% isolated yield for 3-step one-pot reaction).
Lipid-facilitated expressed protein ligation of Ubiquitin. Having demonstrated the utility of this approach in the synthesis of a natural peptide, we sought to apply our methodology to the Cterminal derivatization of proteins. Using expressed protein ligation, recombinant proteins bearing C-terminal thioesters can be generated and reacted with N-terminal cysteine peptides through NCL. Expressed protein ligation [64][65][66][67] has been extensively used for the site-specific introduction of probes, unnatural amino acids, and complex post-translational modifications. The recombinant protein-derived thioesters are usually prepared from a modified Cys-intein fusion protein, which can be cleaved with thiol to provide the corresponding thioester moiety. Thus, we obtained the Ubiquitin-derived thioester Ubi-αCOSC8 (16, Supplementary Fig. 58) through thiol cleavage from a recombinant Cys-intein fusion protein and transthioesterfication. The protein thioester Ubi-αCOSC8 (16, 0.5 mM) was subsequently treated with the Cys-peptide CK(PhC16)ANK (S4, 2 mM) to afford the ligation product, lipidated protein Ubi-CK(PhC16)ANK (17, 71% conversion), within 5 h at 37°C without thiol additives ( Fig. 5 and Supplementary Fig. 59). In contrast, the control reaction between Ubi-αCOSC8 (16) and peptide CKANK (S13, Supplementary  Fig. 60) under the same conditions generated only 6% ligation product ( Supplementary Figs. 61 and 62). These results demonstrate that our lipid-facilitated NCL methodology can be used in conjunction with expressed protein ligation to efficiently derivatize proteins. Indeed, our method could be particularly useful for the addition of lipid modifications onto full-length proteins.

Discussion
In summary, we have developed a mixed micelle-assisted NCL methodology that demonstrates the ability of lipid self-assembly to facilitate the synthesis of larger oligopeptides through accelerating NCL. This approach takes advantage of the presence of cleavable lipid chains in the reactants, leading to self-assembly of reactants into micelles, and acceleration of the ligation reaction by mixed micelle formation. Peptide ligation occurs quickly, at moderate concentration ranges, and in the absence of catalyst additives. Moreover, ligation studies showed that a variety of amino acids were compatible with our methodology. The lipidfacilitated approach was successfully employed to synthesize the natural antimicrobial peptide Magainin 2 in one-pot. The hydrophobic effects conferred by the alkyl chains (C8 on the thioester segment, C16 on the Cys-peptide segment) were enough to efficiently facilitate the reactions with all the peptides studied. We believe that the hydrophobic character of the peptide fragments may also play a key role in the acceleration of the NCL. Therefore, incorporation of hydrophobic residues (especially close to the reaction centers) could considerably increase the rate of the peptide ligations 53 . Further studies using our approach will be focused on the construction of optimized hydrophobic lipopeptides that can drive fast NCL reactions. Since the cysteinecontaining segments incorporate a relevant lipid modification, we foresee potential applications of our lipid-facilitated NCL strategy in the synthesis of structurally complex peptides, especially lipoproteins, which has been demonstrated by derivatization of Ubiquitin. Furthermore, lipid-facilitated peptide ligation could be applied in the construction and selective modification of proteins and other biopolymers such as nucleic acids. More broadly, lipidfacilitated ligation could become a general method to accelerate reactions that would otherwise be too slow to be practical.
Synthesis of LYRMG-αCOSC8 (1a). Here, we describe the representative procedure for the preparation of LYRMX-αCOSZ derivatives (General procedure II). Peptide 1a was prepared manually by standard Fmoc chemistry solid phase peptide synthesis (SPPS) protocols ( Supplementary Fig. 49). 2-Chlorotrityl chloride resin (500 mg; loading: 1.5 mmol/g) was soaked in anhydrous DCM (4 mL) for 30 min. The solvent was filtered off, and a solution of Fmoc-Gly-OH (268 mg, 0.9 mmol) and DIEA (522 μL, 3 mmol) in anhydrous DCM (4 mL) was added to the resin. After 1 h, the solvent was filtered off and the resin was washed with DCM (4 mL). A mixture of DCM:MeOH:DIEA (8.5:1:0.5; 4 mL) was added and the resin was shaken for 30 min, then washed with DCM (3 × 4 mL) and Et 2 O (4 mL). The resin was dried under high vacuum and the loading was determined by quantification of the Fmoc group. For this, a small portion of the resin (~2 mg) was treated with a solution of 20% piperidine in DMF (1 mL) for 30 min. To an aliquot of this solution (100 μL) was added 900 μL of DMF and the absorbance was read at 301 nm. The concentration of the dibenzofulvene-piperidine adduct was obtained by using the extinction coefficients (ε) tabulated in the literature. Thus, the resin loading was estimated to be 0.70 mmol/g. A portion of resin (0.1 mmol loaded Fmoc-Gly-OH) was used for the synthesis of the desired peptide. After the coupling of the last amino acid, the protected peptide was released from the resin by treatment with freshly prepared mixture of 20% 1,1,1,3,3,3hexafluoro-2-propanol (HFIP) in DCM (4 mL) for 1 h and then filtered. The resin was washed with DCM (2 × 1 mL) and the combined fractions were evaporated in vacuo, affording Boc-Leu-Tyr( t Bu)-Arg(Pbf)-Met-Gly-OH as a white solid (58.6 mg, 56%). Analytical HPLC: t R = 2.9 min (5 to 95% Phase B over 1 min, then 95% Phase B for 9 min, Eclipse Plus C8 analytical column) (Supplementary Fig. 4) (Supplementary Fig. 4).
Synthesis of LYRMGCDap(PhC16)ANK (3b). Here, we describe the representative procedure for the preparation of LYRMGCDap(PhC16)ANK derivatives by NCL reaction (General procedure III). Peptide 3b was synthesized according to Supplementary Fig. 50. The lyophilized model peptides 1a (0.4 µmol, 1.0 equiv) and 2b (0.4 µmol, 1.0 equiv) were dissolved in separate vials using freshly prepared ligation buffer [200 µL for each vial respectively. In all, 200 mM NaH 2 PO 4 pH 7.0 containing 10 mM tris(2-carboxyethyl)phosphine (TCEP)]. After sonication of both mixtures for 10 minutes, the solutions from both vials were combined (final concentration of each peptide: 1 mM). The resulting solution was gently shaken at rt. Aliquots were taken at various time intervals and analyzed by analytical HPLC-MS. The yield was calculated based on the peak areas of [2b + 3b] versus the desired ligation product 3b at 210 nm. Analytical HPLC: t R = 13.43 min (0% Phase B for 1 min, then 0 to 95% Phase B over 14 min, then 95% Phase B for 2 min, Eclipse Plus C8 analytical column) (Supplementary Fig. 13 Fig. 13). The ligation product was purified by semipreparative HPLC using a C18 column [gradient of H 2 O with 0.1% formic acid and MeOH with 0.1% formic acid 50:50 (0 min) to 5:95 (10 min to 15 min)], giving 3b (0.5 mg, 81%) as a white solid.

Synthesis of LYRMGCDapANK (4).
Here, we describe the representative procedure for the preparation of lipid-free peptides by photouncaging reaction (General procedure IV). Peptide 4 was synthesized according to Supplementary  Fig. 51. When the native chemical ligation was completed, the reaction mixture containing the product LYRMGCDap(PhC16)ANK (3b) was directly transferred to a quartz tube. Then, the solution was irradiated with UV light (365 nm) for 5 min 45 . Afterwards, an aliquot was taken and analyzed by analytical HPLC-MS and the formation of the peptide 4 was observed. Analytical HPLC: t R = 5.0 min (0% Phase B for 1 min, then 0% to 95% Phase B over 19 min, then 95% Phase B for 2 min, Eclipse Plus C8 analytical column) (Fig. 3b) (Supplementary Fig. 17).
Determination of critical micelle concentrations. Each lipid-modified peptide (1a, 1b, 2a, 2b) subjected to CMC determination was initially lyophilized, and then dissolved in separate vials by addition of freshly prepared ligation buffer. The resulting aqueous solutions (100 µL) of LYRMG-αCOSC8 (1a, 5 µM, 25 µM, 100 µM, 250 µM, 500 µM, 750 µM, 1 mM, 1.5 mM, 2 mM) and CDap(PhC16)ANK (2b, 10 µM, 25 µM, 50 µM, 100 µM, 150 µM, 250 µM, 375 µM, 500 µM, 750 µM) were analyzed by DLS in order to determine the CMCs 68 . This technique allowed us to measure the CMCs at 25°C, since the scattered light intensity (measured at 90°) is dependent on the molecular weight and size of the particle, it increases when monomers start to aggregate in solution. Therefore, in our particular case, light scattering intensity values were plotted versus the concentration of the lipidmodified peptide acting as a surfactant . Lightscattering intensity was very low and constant when the concentration was low, denoting the presence of monomers in solution. After sequential increases in concentration, light-scattering intensity increased, and the CMC was determined by intersecting the two extrapolated lines corresponding to monomer and micelle regions.
Preparation The peptide was next unprotected and released from the resin (S7c) by treatment with freshly prepared cocktail K (TFA:phenol:thioanisole:H 2 O:EDT; 82.5: 5: 5: 5: 2.5; 3 mL per 100 mg of resin) for 2 h, and then filtered. The resin was washed with TFA (0.5 mL) and the combined fractions were evaporated to 1-2 mL by bubbling argon. The concentrated solution was added dropwise to cold Et 2 O (10 mL of Et 2 O per ml of TFA). The resulting precipitate was centrifuged for 10 min at 1008 x g. The supernatant was discarded, then fresh Et 2 O was added, and the suspension was sonicated and centrifuged.  (Supplementary Fig. 26).
Photocleavage: Without purification, the previous NCL reaction mixture was directly transferred to a quartz tube. Then, the photocleavage reaction was performed following the general procedure IV. An aliquot was taken and analyzed by HPLC-MS (Fig. 4). GIGKFLHSCKKFGKAFVGEIMNS (14) was verified to be the peak with t R = 9.66 min (5% Phase B for 1 min, then 5% to 95% Phase B over 14 (Supplementary Fig. 30).
Desulfurization: After photocleavage, to the vial containing 400 µL of the reaction mixture was added 40 µL of 0.5 M Bond-breaker ® TCEP solution (neutral pH), 40 µL of 6 M guanidinium chloride (GuHCl) aqueous solution, 8 μL of 2-methyl-2-propanethiol and 8 μL of radical initiator 2,2′-azobis[2-(2-imidazolin-2-yl)propane] dihydrochloride (VA-044; 0.1 M in degassed water). Then, the reaction mixture was stirred at 37°C. After 120 min, an aliquot was taken and analyzed by HPLC-MS (Fig. 4), confirming that the desulfurization was completed by MS. The desired product (main peak) was purified and verified to be GIGKFLHSAKKFGKAFVGEIMNS (Magainin 2, 15). Analytical HPLC: t R = 9.58 min (5% Phase B for 1 min, then 5 to 95% Phase B over 14 69,70 , SDS [usually 1% (w/v), 35 mM] is difficult to be removed by HPLC. In our case, the concentration of SDS is 2 mM, which is far below the SDS concentration used in conventional methods. Exhaustive analysis of the obtained HPLC-MS traces showed that the retention time of SDS is 15.5 min, and the final product was SDS free.
The expected mass of the purified product (Magainin 2, 15) was also corroborated by high resolution mass spectrometry (ESI-TOF) analysis ( Supplementary Fig. 32).