Solid-phase peptide synthesis (SPPS) builds a peptide one amino acid at a time on tiny beads of insoluble resin. Bruce Merrifield introduced it in 1963 and received the 1984 Nobel Prize in Chemistry for it[1]. Today most synthetic peptides are made this way, from research reagents to peptide drugs produced by the tonne[2].
Making one peptide bond is not difficult. The hard part is making dozens of them in a row, in the right order, with every step close to perfect. Before 1963 each step also ended with the product being isolated and purified, which made long chains slow and wasteful to build. Most of this page follows from that problem: how protecting groups make each amino acid react only where it should, how the resin removed the purification step, and why the small losses that remain still limit how long a chain can be.
| Introduced | 1963 by R. Bruce Merrifield; Nobel Prize in Chemistry 1984 |
|---|---|
| Direction | From the C-terminus to the N-terminus, the opposite way to a ribosome |
| Main chemistries | Fmoc/tBu, the standard today; Boc/Bzl, kept for special cases |
| Routine length | Peptides of 30–50 amino acids; longer chains are joined from pieces |
| End product | A crude peptide that is then purified by HPLC |
Peptide synthesis before the solid phase
The first peptides were made more than a century ago. In 1882 Theodor Curtius prepared benzoylglycylglycine, the first dipeptide with a protected amino end, and in 1901 Emil Fischer and Ernest Fourneau made glycylglycine, the first free dipeptide. A year later Fischer proposed the name "peptide"[1]. Using acid chlorides he went on to make chains of up to 18 amino acids, but neither he nor Curtius had a protecting group for the amino end that could be taken off again without breaking the new peptide bond[3][1].
That came in 1932, when Max Bergmann, a former student of Fischer, and Leonidas Zervas introduced the benzyloxycarbonyl group, Cbz or Z. It is removed by catalytic hydrogenation, which leaves the peptide bond intact. The idea of a temporary protecting group, put on for one step and taken off for the next, opened a new era in peptide synthesis[1]. Further tools followed: dicyclohexylcarbodiimide (DCC) as a coupling reagent in 1955 and the acid-labile tert-butyloxycarbonyl (Boc) group in 1957[1]. With such methods Vincent du Vigneaud made oxytocin in 1953, the first synthetic peptide hormone[3].
Every new bond, however, meant another reaction in solution followed by isolation and purification before the next amino acid could be added, and the longer intermediates often became hard to dissolve. The work was slow, losses piled up at every stage, and longer peptides and proteins were very hard to make[3][4].
Merrifield's idea
Merrifield's answer was to anchor the growing chain by its C-terminal end to an insoluble support. The peptide then stays on the solid at every stage, and whatever is left over after a reaction, excess reagent or by-products, is removed simply by filtering and washing the beads. Washing replaces purification, all the reactions run in one vessel, and the reagents can be used in large excess to push each step close to completion[3]. The support turned out to help in another way too. In one test, a chain of isoleucines that stopped dissolving at the tetrapeptide stage in solution could be extended on the resin to at least twelve residues[3].
His first paper, in 1963, described the tetrapeptide Leu-Ala-Gly-Val, made on a chloromethylated polystyrene resin with Cbz-protected amino acids[5][1]. A year later he switched to Boc protection[1]. The support he settled on, polystyrene cross-linked with 1% divinylbenzene, is still widely used. The dry beads are about 50 µm across and swell five- to six-fold in solvents such as dichloromethane, so most of the chemistry happens inside the swollen bead rather than only on its surface[3].
Because every step came down to adding and draining liquids, the method suited machines. A hand-operated apparatus was followed by an automated synthesiser, in which a programmer pumped solvents and reagents in and out on schedule[3]. With such methods Merrifield and Bernd Gutte made the enzyme ribonuclease A, 124 amino acids long, in 123 cycles. Only about 3% of the starting material ended up as purified enzyme: most chains were lost from the resin during assembly, and more during cleavage and purification. The product nevertheless had about 80% of the activity of the natural enzyme[3], which showed that a working enzyme could be made entirely by chemistry.
Why protecting groups are needed
An amino acid has an amino group at one end and a carboxyl group at the other. If two different amino acids are simply mixed and activated, each reacts with itself as readily as with the other: glycine and alanine give Gly-Gly, Gly-Ala, Ala-Gly and Ala-Ala, and longer chains besides. An amino acid whose carboxyl group has been activated would even polymerise on its own[6]. To join amino acids in a chosen order, every group that should not react has to be blocked.
In SPPS the C-terminal carboxyl group is held by the resin, which acts as its protecting group[1]. Each incoming amino acid carries a temporary protecting group on its amino group. Once the amino acid has been coupled to the chain, this group is removed to free the amine for the next coupling. Reactive side chains, such as the carboxyl groups of aspartic and glutamic acid, the amine of lysine, the guanidino group of arginine, the hydroxyl groups of serine, threonine and tyrosine and the thiol of cysteine, carry permanent protecting groups that stay on until the end[6][1]. Alanine, glycine, leucine, isoleucine, valine, phenylalanine, methionine and proline need none[1]. A good protecting group goes on easily, survives everything the synthesis puts it through and comes off cleanly when it is no longer needed[6].
The temporary group must come off under conditions that leave the permanent groups untouched, and it must do so dozens of times. The cleanest way to achieve this is orthogonal protection, a concept introduced by Barany and co-workers: the groups belong to different chemical classes and are removed by different mechanisms, for example one by a base and the other by an acid, so each can be taken off in the presence of the other[6]. Orthogonal schemes are usually also milder, because selectivity depends on the type of reaction rather than on a difference in reaction rates[6].
Protection also explains why the chain is built from the C-terminus towards the N-terminus. The carboxyl group activated at each step belongs to a single amino acid, and urethane-type groups such as Fmoc, Boc and Cbz on its amino group suppress loss of its configuration during activation[2][7]. Activating the C-terminus of a peptide instead brings a real risk of epimerisation[1].
Two strategies: Boc/Bzl and Fmoc/tBu
Boc/Bzl
Merrifield's classical scheme protects the amino group with Boc and the side chains with benzyl-based groups, and holds the chain on the resin through a benzyl ester. Both kinds of group are removed by acid, but by acids of very different strength: their sensitivity differs by more than 1000 to 1. Each cycle removes Boc with 50% trifluoroacetic acid (TFA) in dichloromethane, and the resulting ammonium salt is neutralised with a tertiary amine such as diisopropylethylamine. At the end, anhydrous hydrogen fluoride (HF) releases the peptide and removes the side-chain groups[3].
The scheme works, but it is not orthogonal. Each round of TFA can remove small amounts of side-chain protection and loosen the anchor to the resin[2]; in the ribonuclease synthesis, 83% of the chains were lost during assembly because the anchoring bond was not stable enough[3]. HF is dangerous and needs special apparatus made of Teflon, which many laboratories lack and some safety departments do not allow[1]. It also generates reactive carbocations that can attack tyrosine, tryptophan, methionine and cysteine in the peptide[3].
Fmoc/tBu
The 9-fluorenylmethoxycarbonyl (Fmoc) group was introduced by Louis Carpino and Grace Han in 1970. It is removed by a mild base, usually piperidine, and is stable to acid[1][6]. In solution it proved disappointing, because its by-product, dibenzofulvene, is reactive and can add back to the freed amine. On a solid support the dibenzofulvene is simply washed away[2]. In the late 1970s Atherton and Sheppard in Cambridge, and independently Chang and Meienhofer, turned Fmoc into a complete solid-phase method[1][2]. Side chains are protected with acid-labile groups based mainly on tert-butyl (tBu) and trityl (Trt), and the final cleavage uses TFA instead of HF.
Because the temporary group is removed by base and the permanent groups by acid, Fmoc/tBu is truly orthogonal[2]. It spread first among biologists, who could make peptides for raising antibodies on inexpensive machines without handling HF. It was easy to automate, since no corrosive TFA was needed during the cycles, and the removal of each Fmoc group releases a fluorene compound with strong UV absorption, which gives a running check on the synthesis. Its mild conditions also made it possible to build phosphorylated and glycosylated peptides[2]. Today most synthetic peptides are made by Fmoc SPPS, and multi-tonne production of peptide drugs has made very pure Fmoc amino acids cheap. Boc chemistry is now used mainly for specialist applications[2].
| Fmoc/tBu | Boc/Bzl | |
|---|---|---|
| Temporary group | Fmoc | Boc |
| Removed each cycle by | Base: 20% piperidine in DMF | Acid: 25–50% TFA in dichloromethane |
| Side-chain protection | tBu, Trt, Boc, Pbf | Benzyl-based |
| Final cleavage | TFA with scavengers | Anhydrous HF |
| Orthogonal | Yes | No; relies on graded acid strength |
| Main weak points | Base-driven side reactions, above all aspartimide formation | HF; gradual losses in repeated acid |
From Isidro-Llobet et al.[6], Behrendt et al.[2] and Coin et al.[4]
By 2016 the standard side-chain protecting groups had hardly changed for more than fifteen years, although for arginine, asparagine, aspartic acid, histidine and cysteine they were known to be far from ideal[2]:
| Amino acid | Fmoc strategy | Boc strategy |
|---|---|---|
| Arg | Pbf | Tos |
| Asp, Glu | OtBu | OBzl, OcHex |
| Asn, Gln | Trt | none |
| Cys | Trt | 4-MeBzl |
| His | Trt | Dnp, Bom |
| Lys | Boc | 2-Cl-Z |
| Ser, Thr | tBu | Bzl |
| Tyr | tBu | 2-Br-Z |
| Trp | Boc | For |
From Behrendt et al.[2] and Jaradat[1]. Pbf, pentamethyldihydrobenzofuransulfonyl; Trt, trityl; Tos, tosyl; cHex, cyclohexyl; Bzl, benzyl; MeBzl, methylbenzyl; Dnp, dinitrophenyl; Bom, benzyloxymethyl; Cl-Z and Br-Z, chloro- and bromobenzyloxycarbonyl; For, formyl.
Boc chemistry has not disappeared. In aggregation-prone sequences, the TFA used in every Boc cycle helps break up aggregates, while the piperidine used in Fmoc chemistry does not, so some of these peptides are easier to make by the older method[4].
One cycle, step by step
In Fmoc SPPS every amino acid is added by the same short cycle[4]:
- Deprotection. A solution of 20% piperidine in dimethylformamide (DMF) removes the Fmoc group from the end of the chain, typically in two treatments of about 1 and 10 minutes.
- Washing. DMF rinses out the piperidine and the by-products.
- Coupling. The next Fmoc-amino acid is activated at its carboxyl group and added in excess, six equivalents in one standard protocol. It reacts with the free amine and forms the new peptide bond.
- Washing. The excess amino acid and reagents are rinsed away, and the cycle starts again.
The cycle can be checked as it runs. The compound formed from dibenzofulvene and piperidine absorbs at 301 nm, so measuring the used deprotection solution shows how much Fmoc came off and whether the amount of growing chain stays steady. The Kaiser test, a colour reaction of free primary amines, shows whether a coupling is complete; if amines remain, the coupling is repeated. Chains that still fail to react can be capped with acetic anhydride, which acetylates their free ends so that they cannot grow further[4].
When the sequence is complete, the last Fmoc group is removed and the resin is treated with a TFA mixture, for example TFA, water, phenol and triisopropylsilane in a ratio of 85:5:5:5. This releases the peptide from the resin and strips the side-chain protecting groups in a single step[4]. The extra components are scavengers that trap the reactive carbocations released from the protecting groups before they can attack the peptide[3]. The filtrate is dropped into cold diethyl ether, where the crude peptide precipitates[4], ready for purification by HPLC and a check by mass spectrometry.
Choosing the resin
The resin is a polymer bead with a linker, a small chemical handle that holds the first amino acid. The linker decides what the C-terminus will look like after cleavage[1]:
| Resin | C-terminus after cleavage | Introduced |
|---|---|---|
| Merrifield resin | Free acid (COOH) | 1963 |
| Benzhydrylamine (BHA) resin | Amide (CONH2) | 1970 |
| Wang resin | Free acid | 1973 |
| Rink amide resin | Amide | 1987 |
From Jaradat[1].
The choice matters for analysis: a C-terminal amide is 0.984 Da lighter than the free acid and has one acidic group fewer, which changes both the mass and the isoelectric point given by the peptide calculator. Besides polystyrene, polyacrylamide supports and polystyrene modified with polyethylene glycol (PEG) are widely used; and the more polar resins often give better results[3][4].
Joining the amino acids
To form a peptide bond, the amino group of one amino acid attacks the carbonyl carbon of the other, and that carbon first has to be activated by attaching a group that pulls electrons away from it and later leaves[7]. The reagents that do this are called coupling reagents, and they must work fast, completely and without scrambling the configuration of the amino acid.
The classic reagent is DCC, first used for peptides by John Sheehan and George Hess in 1955[1]. A carbodiimide reacts with the carboxylic acid to form an O-acylisourea, one of the most reactive species in peptide chemistry, which the amine then attacks to give the peptide bond[7]. On the solid phase DCC has a practical drawback. Its by-product, dicyclohexylurea, precipitates within minutes and dissolves only in TFA. That is tolerable in Boc chemistry but not in Fmoc chemistry, so Fmoc syntheses usually use diisopropylcarbodiimide (DIC), whose urea stays in solution[7].
The O-acylisourea is so reactive that it can also rearrange into an inactive N-acylurea or close into an oxazolone, through which the amino acid can lose its configuration. Additives such as 1-hydroxybenzotriazole (HOBt), introduced in 1970, turn it into a less reactive active ester and reduce both problems[1][7]. HOBt and its relative HOAt are benzotriazoles, and all benzotriazole derivatives have explosive properties; HOBt and HOAt can decompose rapidly when heated above 180 °C. A newer additive, Oxyma, suppresses racemisation as well as HOAt does, outperforms HOBt and carries a lower risk of explosion[7].
Stand-alone reagents of the aminium/uronium and phosphonium types, such as HBTU (1978), HATU (1993) and COMU (2009), combine activation and additive in one compound[1]. They are widely used, especially in automated synthesis, and HATU serves for demanding steps such as coupling bulky building blocks[4]. Used in excess, though, an aminium/uronium reagent can react directly with the free amine of the chain and cap it as a guanidine[7].
Why small losses add up
Merrifield stressed that the repeated steps must be fast, high-yielding and free of side reactions, because by-products accumulate from cycle to cycle[3]. If a coupling is incomplete, the chains that missed it go on growing in later cycles and end up lacking one residue somewhere in the middle. Such deletion peptides are almost identical to the target. Merrifield grouped the main by-products of SPPS into termination, deletion and modification peptides[3]. If every coupling has the same efficiency, the share of complete chains falls with each step:
| Residues | 97% | 99% | 99.5% | 99.9% |
|---|---|---|---|---|
| 10 | 76% | 91% | 96% | 99% |
| 20 | 56% | 83% | 91% | 98% |
| 30 | 41% | 75% | 86% | 97% |
| 50 | 22% | 61% | 78% | 95% |
| 70 | 12% | 50% | 71% | 93% |
| 100 | 5% | 37% | 61% | 91% |
Calculated as efficiencyn−1 for n residues, assuming every coupling has the same efficiency and ignoring all other side reactions, which lower the real figures further.
The quantitative ninhydrin reaction can detect as little as 0.1% of unreacted chains, which corresponds to a coupling 99.9% complete[3]. In practice, peptides of 30–50 residues are now made routinely; the 41-residue hormone corticotropin-releasing factor takes about 80 working hours by hand, purification included[4]. The often-quoted limit of about 50 residues is only a rough guide. Behrendt and colleagues call it practically meaningless, because some much shorter sequences are extremely difficult to make and success is never guaranteed[2].
Longer chains and small proteins are built from pieces. Protected fragments made on the resin can be joined in solution; a precursor of the green fluorescent protein, 238 residues long, was assembled from 26 of them. Protected fragments, however, are often poorly soluble and can epimerise at the activated end[1]. Chemical ligation avoids both problems by joining unprotected fragments in water. In native chemical ligation, introduced by Philip Dawson, Stephen Kent and colleagues in 1994, a peptide ending in a thioester reacts with a peptide that begins with cysteine, and the two end up joined by a normal peptide bond[8]. This is how whole enzymes, including mirror-image ones, have been made by chemistry.
Difficult sequences
Some sequences are hard to make for reasons that have little to do with the coupling chemistry. As the chain grows, it can fold back or pair with neighbouring chains on the same bead, forming β-sheet-like structures held together by hydrogen bonds between N–H and C=O groups[1]. The peptide–resin matrix collapses, reagents diffuse in slowly, deprotection and coupling become slow and incomplete, and even the Kaiser test can give false negative results[4]. Behrendt and colleagues regard this aggregation on the resin as the main obstacle to peptide synthesis today[2].
Most remedies stop the chains from lining up[1][2][4]:
- Pseudoproline dipeptides. Serine or threonine is temporarily locked into a ring that mimics proline and kinks the chain, which breaks up β-sheets. The ring opens in the final TFA treatment and restores the normal residue. They can be used only where a serine or threonine sits at a suitable position.
- Backbone protection. A removable Hmb or Dmb group on an amide nitrogen stops that N–H from forming hydrogen bonds. Placed before the point where aggregation starts and then every six residues, it keeps the chain solvated, although coupling onto the substituted nitrogen is slow.
- O-acyl isopeptides. The chain is temporarily attached through the side-chain oxygen of a serine or threonine instead of its amino group, which prevents the secondary structure from forming.
- Conditions. Resins with a low loading or a PEG component, salts such as lithium chloride or potassium thiocyanate, and heat, including microwave heating.
Heat speeds up the synthesis but brings its own risks: loss of chains from the resin, racemisation of cysteine and histidine, and more aspartimide formation. In a direct comparison, backbone protection prevented aggregation better than heating did[2]. Such tools can rescue peptides that the standard protocol cannot deliver: a 37-residue WW domain, an analogue of FBP28, could not be made by the standard procedure but was obtained once pseudoproline units were built in[4].
Common side reactions
| Side reaction | Where and why | Consequence |
|---|---|---|
| Aspartimide formation | Asp followed by Gly (the most prone), or by Asp, Asn, Arg, Thr or Cys; caused by the base used to remove Fmoc | Up to nine by-products; some, such as the epimerised α-aspartyl peptide, are very hard or impossible to separate[2] |
| Racemisation of His | During coupling, especially with base, long preactivation or heat | D-His in the product: with His(Trt), 1% without preactivation, 7.8% after 5 minutes and 16.6% with microwave heating at 80 °C[2] |
| Racemisation of Cys | Activation with base, worst with heating and preactivation | D-Cys in the product; avoided by carbodiimide activation[2] |
| Diketopiperazine formation | When Fmoc is removed from the second residue, mostly after Pro or an N-alkylated residue | The first two residues close into a ring and are lost[4][7] |
| Alkylation during cleavage | Tyr, Trp, Met, Cys | Adducts with carbocations from the protecting groups; reduced by scavengers[3] |
| Guanidinylation | Excess aminium/uronium coupling reagent | The free N-terminus is capped and stops growing[7] |
Aspartimide formation is regarded as the most serious side reaction of Fmoc chemistry. Adding Oxyma to the piperidine solution reduces it: in one test peptide containing Asp-Gly, the related impurities fell from 44% to 15%[2].
Because SPPS adds one residue at a time and carries every error forward, it is not expected ever to match the accuracy of protein synthesis in a cell, which copies a template. What it offers instead is complete control over every atom: any residue, natural or not, and any modification can be placed exactly where the chemist wants it[2].
In short
- SPPS builds a peptide from the C-terminus on insoluble resin beads. Washing replaces purification after each step, which made long chains practical.
- Protecting groups decide which groups react: a temporary group on the amine, removed every cycle, and permanent groups on the side chains, removed at the end.
- Fmoc/tBu, deprotected by base and cleaved by TFA, is today's standard; Boc/Bzl, which needs HF, is kept for special cases.
- Every incomplete coupling leaves deletion peptides. At 99% per step, only about 61% of the chains of a 50-residue peptide are complete.
- Aggregation on the resin is the main obstacle. Pseudoprolines and backbone protection help, and longer proteins are joined from pieces by ligation.
References
- Jaradat DMM (2018). Thirteen decades of peptide synthesis: key developments in solid phase peptide synthesis and amide bond formation utilized in peptide ligation. Amino Acids 50:39–68. DOI: 10.1007/s00726-017-2516-0
- Behrendt R, White P, Offer J (2016). Advances in Fmoc solid-phase peptide synthesis. Journal of Peptide Science 22:4–27. DOI: 10.1002/psc.2836
- Merrifield RB (1985). Solid phase synthesis (Nobel lecture). Angewandte Chemie International Edition in English 24:799–810. DOI: 10.1002/anie.198507993
- Coin I, Beyermann M, Bienert M (2007). Solid-phase peptide synthesis: from standard procedures to the synthesis of difficult sequences. Nature Protocols 2:3247–3256. DOI: 10.1038/nprot.2007.454
- Merrifield RB (1963). Solid phase peptide synthesis. I. The synthesis of a tetrapeptide. Journal of the American Chemical Society 85:2149–2154. DOI: 10.1021/ja00897a025
- Isidro-Llobet A, Álvarez M, Albericio F (2009). Amino acid-protecting groups. Chemical Reviews 109:2455–2504. DOI: 10.1021/cr800323s
- El-Faham A, Albericio F (2011). Peptide coupling reagents, more than a letter soup. Chemical Reviews 111:6557–6602. DOI: 10.1021/cr100048w
- Kent SBH (2009). Total chemical synthesis of proteins. Chemical Society Reviews 38:338–351. DOI: 10.1039/b700141j