Challenges and Perspectives in Chemical Synthesis of Highly Hydrophobic Peptides

Solid phase peptide synthesis (SPPS) provides the possibility to chemically synthesize peptides and proteins. Applying the method on hydrophilic structures is usually without major drawbacks but faces extreme complications when it comes to “difficult sequences.” These includes the vitally important, ubiquitously present and structurally demanding membrane proteins and their functional parts, such as ion channels, G-protein receptors, and other pore-forming structures. Standard synthetic and ligation protocols are not enough for a successful synthesis of these challenging sequences. In this review we highlight, summarize and evaluate the possibilities for synthetic production of “difficult sequences” by SPPS, native chemical ligation (NCL) and follow-up protocols.

Focusing on the class of heavy synthetically accessible proteins such as membrane proteins and considering their structural and functional divergence we will summarize in following efforts undertaken for their synthetic production. It is worth to mention that membrane proteins are encoded by 20–30% of the human genome ( Wallin and von Heijne, 1998 ; Santoni et al., 2000 ; Krogh et al., 2001 ; Melnyk et al., 2003 ; Li et al., 2017 ). Thereupon, it is not surprising, that because of this relatively high abundance, membrane proteins are of great interest as drug targets. This can be seen in the number of available drugs that target these structures. Around 60% of all FDA-approved drugs address these structures, e.g., GPCRs or ion channels ( Yildirim et al., 2007 ) and the interest in therapeutic applications or for the design of nanopore-based bio-inspired sensors is rising. Surprisingly, until 2003 only about 60 high-resolution structures of transmembrane proteins were available despite their great importance for the cell function and drug design ( Melnyk et al., 2003 ). With the Nobel prize in 2017, awarded to J. Dubochet, J. Frank, and R. Henderson on their outstanding work in cryo-electron-microscopy, this situation changed rapidly ( Cressey and Callaway, 2017 ; Allen, 2019 ). High profile structures, dynamics and structural studies of various membrane proteins are now available, including works on calcium-selective ion channels ( Yoo et al., 2018 ), voltage-gated potassium channels ( Shigematsu et al., 2017 ; Shigematsu et al., 2019 ) and a membrane-embedded monomeric yeast ATP synthase ( Srivastava et al., 2018 ). However, there is still a challenge to provide access to enough material to determine structures and functions of membrane proteins. Two major ways are possible either to chemically synthesize or recombinantly express membrane proteins. Four major problems in expression of these structures led to the development of strategies using chemical synthesis: (1) protocols for the recombinant expression of hydrophilic structures are often not applicable to hydrophobic proteins, (2) over-expression of membrane proteins usually leads to membrane disruption and thus cell toxicity, (3) extraction and purification are not trivial, and (4) limitations in incorporation of post-translational modifications/isotopic labels ( Shen et al., 2011 ). Trying to overcome these problems, a bridge between biology and chemistry was required, resulting in an outstanding cooperation. A great advantage of chemical synthesis displays the possibility to custom design of the desired sequence. As soon as the successful synthetic protocol is elaborated, the integration of unnatural amino acids, mutations at arbitrary positions, post-translational modifications (PTMs) and site-specific labels for e.g., solid-state/solution NMR spectroscopy or fluorescence microscopy experiments is readily possible ( Sato, 2016 ). Furthermore, product in the multi-milligram range can be obtained making numerous analytical experiments possible, leading to a better structural and functional understanding. The establishment of a successful chemical synthesis strategy, though, faces some challenges. Due to the challenges in synthesis of “difficult sequences” especially membrane proteins or their functional parts, only few manuscripts report successful examples and are summarized in Table 1 identifying special features of the synthetic strategy. TABLE 1 Overview of successful chemical (semi-)synthesis protocols for transmembrane proteins (extended from Shen et al., 2011 ). Name Protein length Protocol, SPPS SPPS Facilitating NCL Special features References NS4A , cofactor protein of serine protease from Hepatitis C virus 1–66 Fragment 1: Boc-based Fragment 2: Fmoc-based Fragment 1: tri-lysine solubilizing tag Fragment 2: tetra-lysine solubilizing tag β-octyl-glucoside One of the first described synthesis routes Bianchi et al., 1999 BM2 proton channel, influenza A 1–97 Both fragments: Boc-based 30% TFE Kochendoerfer et al., 1999 Potassium Channel KcsA 1–125 Fragment 1-73 recombinant expression Fragment 74-125: Boc-based 50% TFE, 1% SDS Thioester Fragment 1-73, Great difficulties in solubilizing synthesized fragment, T74C mutation Valiyaveetil et al., 2002 Mechano-selective ion channels: Ec-MscL and Tb-MscL 1–136 All fragments: Boc-based Dodecyl-phospho-choline, DPC Ec-MscL: Q56C and N103C mutations Tb-MscL: E102C and S52C mutations, Acm protection group Clayton, 2005 Diacylglycerol Kinase from E. Coli, DAGK 1–121 Three fragments: Boc-based Polyethylene glycol-polyamide (PPO) tag and hexa-arginine tag DPC or OG Several solubilizing tags were tested Lahiri et al., 2011 Inward rectifier K+ channel protein Kir 5.1 64–179 Four fragments: Fmoc-based Fragment 3: tetra-arginine tag DPC Hydrazides for NCL, usage of pseudoprolines Zheng et al., 2014 Hepatitis C Virus cation-specific ion channel p7 1–63 Both fragments: Fmoc-based Both fragments: tetra-arginine tag

Temporary internal (structural) modifications of the peptide sequence have revolutionized the use of the Fmoc-based SPPS protocol on “difficult sequences.” Today several different backbone modifications exist mainly focused to prevent side-reaction such as diketopiperazine or aspartimide formation ( Paradis-Bas et al., 2016 ). Many of them offer also the advantage to reduce peptide aggregation on solid support thus facilitate the accessibility of the N-terminus during SPPS ( Tickler and Wade, 2007 ). Depending on their structure, most backbone modifications can be classified in ortho-hydroxybenzyl groups [Hmb ( Hyde et al., 1994 ), Hmsb ( Huang and Liu, 2015 ), Hbz ( Johnson and Quibell, 1994 ), Hnb ( Meutermans et al., 1999 )] ortho-mercaptobenzyl groups [Dmmb ( Kawakami et al., 2001 ), Tmb ( Offer et al., 2002 )] and methoxybenzyl groups [Dmb ( Zahariev et al., 2006 ), Mmsb ( Thennarasu and Liu, 2010 )]. Additionally, several other protection groups were developed which cannot be assigned to one of the aforementioned backbone modifications such as EDOTn ( Isidro-Llobet et al., 2008 ), Dcpm ( Carpino et al., 2009 ), Etom ( Fernandez-Llamazares et al., 2014 ), 2-furfury and 2-thlenylmethyl ( Johnson et al., 1995 ) groups. In the 1990ies, one of the first backbone amide protection groups, which remarkably improved handling of aggregation-prone peptides, was developed, N,O-bis(Fmoc) derivates of Nα-2-hydroxy-4-methoxybenzylamino acids (Hmb) ( Zeng et al., 1997 ). This group demonstrated the ability to facilitate synthesis of an acyl carrier protein, a 65–74 decapeptide that showed strong inter-chain association ( Weygand et al., 1966b ; Quibell, 1999 ). Thus, the Hmb moiety and its derivates were developed, which were frequently used in Fmoc-based SPPS ( Simmonds, 1996 ; Table 2 ). TABLE 2 Backbone amide protecting groups that can be removed during cleavage. Structure Advantages Limitations Hmb ( Johnson et al., 1993 , 1995 ) (Fmoc SPPS) • Inhibition of aspartimide formation • Commercial availability of dipeptides and amino acids containing Hmb group • Poor O-N -acyl transfer Dmb ( Weygand et al., 1966a ; Blaakmeer et al., 1991 ) (Fmoc/Boc SPPS) • Suppression of aspartimide formation • Rapid removal at high concentrations of TFA • Commercial availability of dipeptides and amino acids containing Dmb group • Bulkiness • Dipeptides are restricted to Substitution sites containing Ser, Thr and Gly (and in some cases Cys) Tmob ( Johnson et al., 1995 ) (Fmoc SPPS) • Commercial availability of Tmob-protected amino acids • Faster acylation compared to Dmb. • High acid lability, can be removed with 5% TFA in DCM • Bulkiness MIM ( Isidro-Llobet et al., 2008 ) (Fmoc SPPS) • Lower steric hindrance and faster acylation compared to Dmb • Bulkiness EDOTn ( Isidro-Llobet et al., 2008 ) (Fmoc SPPS) 2-furfury/2-thienylmethyl ( Johnson et al., 1995 ) (Fmoc SPPS) • Higher acid lability compared to Hmb • Inefficiency of the coupling of the incoming amino acid onto the N-(furfury/thienyl)peptidyl-resin Further strategies aiming to decrease on-resin aggregation during SPPS include the use of pseudoprolines ( Mutter et al., 1995 ; Wohr and Mutter, 1995 ) or O-acyl isopeptides ( Horikawa et al., 1998 ) (depsipeptides) ( Figure 4 ). Mutter’s lab introduced pseudoprolines incorporating them into a sequence with numerous building blocks being commercially available ( Mutter et al., 1995 ; Wohr and Mutter, 1995 ). The cyclic oxazolidine (Ser, Thr) thiazolidine (Cys) ring system shows structural similarities with Pro, resulting in a “kink” conformation within the growing peptide chain preventing aggregation, self-association and a β-sheet formation ( Wohr et al., 1996 ). An alternative strategy to interrupt unfavorable secondary structures is the synthesis with O-acyl isopeptides which were developed by Sohma et al. (2011) . This strategy relies on the introduction of oxo-esters over Ser or Thr residues within the primary sequence. The strategy was successfully applied for SPPS of several lipophilic peptides ( Kawashima et al., 2014 ). FIGURE 4 Example of pseudoproline (A) and isopeptide (B) strategies as temporary structural modifications. Apart from their structure, backbone modifications can be classified into two categories: (1) modifications, removable during cleavage from the resin ( Table 2 ) and (2) modifications, removable after cleavage from the resin ( Table 3 ). TABLE 3 Backbone amide protecting groups that remain on the sequence after cleavage from solid support and require special conditions to be detached. Structure Advantages Limitations Hmsb ( Howe et al., 2000 ) (Fmoc SPPS) • “Safety-catch” protecting group • Bulkiness 1,3-Benzoxathiole-3-oxide derivative ( Offer, 1997 ) (Fmoc/Boc SPPS) • Efficient acylation, suppression of epimerization • “Safety-catch” protecting group • Bulkiness 4-Methoxy-2-Nb ( Johnson and Kent, 2006 ) (Fmoc/Boc SPPS) • “Safety-catch” protecting group • Slightly slower coupling of incoming amino acid

The great potential and importance of membrane proteins in elucidation of their structures and function as well as development of novel drug leads targeting membrane proteins causes research in this area to flourish. Moreover, need in being able to incorporate post-translational modifications, isotopic labels or peptide-mimetics rely on robust approaches of chemical synthesis ( Zheng et al., 2016 ). Schmidt et al. (2017) recently summarized the possibilities of ligases to perform amide bond formation during ligation reactions in a review opening a platform for biological engineering. In a very recent review, Nuijens et al. (2019) discuss advantages and disadvantages of enzymes used for ligation and cyclization suggesting a sortase-mediated ligation strategy to be straightforward. They also elucidate the possibility to use enzymes for cyclization and labeling, showing the versatile applications of engineered and naturally occurring enzymes. With great interest to solve solubility problems during NCL, but also handling of peptides, ionic liquids seem to have enormous potential ( Tietze et al., 2012 ; Baumruck et al., 2017 ). Besides SPPS and NCL, purification of these “difficult sequences” is not trivial. Apart from the usage of the already mentioned alternative organic solvents application at RP-HPLC to purify fragments and products, just recently an alternative approach has been introduced by using catch-and-release purification method which is based on base-labile cleavable linkers using oxime-based and hydrazine-based ligation chemistry ( Reimann et al., 2019 ). Recently, a new and rapid ligation method was presented: the additive-free diselenide-selenoester ligation (DSL) ( Mitchell et al., 2015 ). Making use of a peptide selenoester and a peptide diselenide bearing an N-terminal (Sec) 2 unit, the thiol-free ligation was completed within minutes. Giving an example of the time-saving advantage this method provides, Mitchell et al. were able to completely synthesize an early secretory antigenic protein-6 (ESAT-6) within only 16 h including a deselenization of the Sec unit to Ala ( Mitchell et al., 2015 ). The deselenization reaction is further employable to other amino acids, such as aspartate and glutamate auxiliaries ( Conibear et al., 2018 ) and is even employable to poorly soluble compounds. This is demonstrated by synthesis of the poorly soluble therapeutic lipopeptide tesamorelin and variants of the transmembrane lipoprotein phospholemman FXYD1 using this method and nanomolar concentrations circumventing the integration of solubilizing units ( Chisholm et al., 2019 ). Broadening the ligation toolkit, the α-ketoacid-hydroxylamine (KAHA) ligation route is also employable to poorly soluble and highly hydrophobic proteins such as IFITM3 (see Table 1 ) or the antibacterial cyclic AS-48 protein ( Rohrbacher et al., 2017 ) making use of 5-oxaproline within acidic conditions. The KAHA ligation is applicable to synthesized fragments from Fmoc-synthesis, integrated solubilizing tags and based on ligation conditions in organic solvents and thus presents an alternative to fragments facing solubility problems. All presented methods and protocols are based on traditional “batch” chemistry but a novel approach toward continuous-flow peptides synthesis ( Mijalis et al., 2017 ) or ligation and desulfurization ( Chisholm et al., 2018 ). They presented an in-line flow-based ligation and desulfurization protocol and presented synthesis of enfuvirtide (HIV drug) and the diagnostic agent somatorelin. This procedure could be interesting in the future especially considering scale-up of ligation reactions. Having given an overview of synthetic strategies and follow- up protocols available to date, it is obvious that there is no one-fits-all approach. Membrane proteins are of great interest and thus robust synthesis routes will help to investigate structural behavior leading to a better understanding of the diagnostic points of action and possibilities. Limitations in a straight-forward design are the extreme hydrophobic regions within the protein sequences that lead to aggregation on resin making synthesis challenging. Overcoming this limitation, NCL enables segmentation of the sequence into two or more fragments that can be successively condensed. Especially transmembrane regions are challenging to be synthesized even out of smaller fragments, incorporation of removable solubilizing tags represent a method to facilitate handling, synthesis and purification. However, the general protocols and a variety of choice can be used in order to meet a right choice for the synthesis of any particular “difficult sequences” especially membrane proteins of functional parts of them ( Figure 3 ).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.