Peptides are made by bacteria, fungi, plants and animals, and they do very different jobs: hormones and neurotransmitters, antibiotics, plant growth signals, venom toxins. What is less obvious is that they are not all made in the same way. The ribosome, which builds every protein from a gene, is only part of the story.
Living things use three routes. The ribosome can make a long precursor protein from which enzymes cut the active peptide. It can make a precursor whose core is then rebuilt by enzymes into something no gene could spell out directly. Or the ribosome can be bypassed altogether, and the peptide is assembled amino acid by amino acid on a giant enzyme. Which route a peptide takes explains a good deal about it, including whether it is likely to contain D-amino acids, the mirror images of the L-amino acids found in proteins.
| Cut from a precursor | Ribosome, then rebuilt (RiPPs) | Without a ribosome (NRPS) | |
|---|---|---|---|
| Sequence set by | The gene | The gene, then enzymes | The enzyme assembly line |
| Building blocks | Standard amino acids; cleavage and limited later modification | Standard amino acids, then chemically changed | Standard and many non-standard amino acids |
| Made by | Animals, also plants | All three domains of life | Mostly bacteria and fungi |
| Examples | Enkephalins, ACTH, insulin | Nisin, conotoxins, amanitin, cyclotides | Cyclosporine, vancomycin, gramicidin S |
Sources: Hook et al. (2008)[1], Arnison et al. (2013)[2], Süssmuth and Mainz (2017)[3].
Three ways to make a peptide
The first two routes start in the same place, with a gene, its messenger RNA and the ribosome. The difference is what happens to the chain afterwards: in the first it is mainly cut, in the second it is chemically rebuilt. The third route has no gene for the peptide sequence at all; the sequence is set by the order of modules in an enzyme. A fourth, minor possibility is a single enzyme joining a few amino acids, which is how cells make glutathione[3], a tripeptide described on its own page.
Cut out of a larger protein
The peptide hormones and neurotransmitters of animals, together called neuropeptides, are typically 3 to 40 residues long, and more than a hundred are known. None is made directly in its final form. Each is a segment of a larger, inactive precursor protein, a prohormone, and has to be cut out[1].
The sequence of events is the same for most of them. The ribosome makes a preprohormone; its N-terminal signal peptide is removed as the chain enters the endoplasmic reticulum; the prohormone travels through the Golgi apparatus and is packed into secretory vesicles together with the proteases that will cut it. Processing happens as the vesicle matures, so the vesicle ends up holding the finished peptide, ready to be released on a signal[1].
The cuts are made at pairs of basic residues, most often Lys-Arg, sometimes Lys-Lys, Arg-Arg or a single Arg. Two families of proteases do the work. Prohormone convertases cut on the C-terminal side of the pair, and carboxypeptidase E then trims off the leftover basic residues; cathepsin L cuts on the N-terminal side, and an aminopeptidase removes them from the other end[1]. Knockout mice show how much depends on these enzymes: without prohormone convertase 2, proinsulin is processed incompletely and α-MSH almost disappears from the pituitary[1].
One precursor can hold several peptides. Proenkephalin contains four copies of Met-enkephalin, one of Leu-enkephalin and two related opioid peptides. Pro-opiomelanocortin (POMC) is cut differently in different tissues: into ACTH in the anterior pituitary, and into α-MSH and β-endorphin in the intermediate lobe[1]. After cutting, some peptides are modified further, for example by disulfide bonds or C-terminal amidation[1]. Insulin's precursor is described on the insulin page.
Plants use the same logic. Most plant signalling and antimicrobial peptides characterised so far are also cut from larger precursors, which in most cases have no other known function; a few are released from working proteins, as described below[4].
Made by the ribosome, then rebuilt
For a long time, research on unusual peptide antibiotics focused on the nonribosomal ones described below; only in 1988 was a peptide antibiotic with unusual modifications, epidermin, shown to be made on the ribosome[3]. Genome sequencing has since shown that such peptides form a major class of natural products. In a 2013 review, 65 researchers gave the class a common name, ribosomally synthesized and post-translationally modified peptides, or RiPPs, and noted that they are made in all three domains of life[2].
A RiPP begins as a precursor of typically 20 to 110 residues with two parts: a leader peptide, which the modifying enzymes recognise, and a core peptide, which becomes the product. The enzymes rework the core, and the leader is then cut off. Because the enzymes bind the leader and tolerate changes in the core, the core sequences can vary freely, which the authors describe as access to high chemical diversity at low genetic cost[2]. The modifications typically make the peptide more rigid, more stable and harder for proteases to degrade[2].
Some examples show how far the rebuilding can go:
- Nisin, first reported in 1928, contains lanthionine rings: serine and threonine are first dehydrated, then cysteine side chains add across them to form thioether bridges. Its ribosomal origin was proposed in 1970 and proved in 1988. Nisin binds lipid II, a building block of the bacterial cell wall[2], and is used as a food preservative; see nisin.
- Amanitin and phalloidin, toxins of Amanita and some other mushrooms, are cut from precursors as 8- and 7-residue rings and cross-linked between a cysteine and a tryptophan. Amanitin blocks RNA polymerase II; phalloidin binds actin[2].
- Cyclotides are plant peptides of 28 to 37 residues with a head-to-tail backbone and three disulfide bonds forming a knot. A single plant can make dozens to hundreds of them, and they act as insecticides by damaging the membranes of the insect gut[2].
- Conotoxins, the venom peptides of cone snails, are mostly 10 to 30 residues with a high density of disulfide bonds and a wide range of further changes: hydroxyproline, γ-carboxyglutamate, bromotryptophan, sulfotyrosine, sugars and D-amino acids[2]. One of them is an approved painkiller (Prialt); see conotoxins.
- Polytheonamides, 48-residue peptides from a marine sponge, were long thought to be nonribosomal because they are so heavily modified. The genes turned out to belong to an uncultivated bacterial symbiont of the sponge, and the peptide is ribosomal: its precursor undergoes 48 modification steps, including the conversion of 18 L-amino acids into D-amino acids[2].
One cone snail enzyme shows how old some of these tools are. The enzyme that converts glutamate into γ-carboxyglutamate in conotoxins is closely related to the vitamin K-dependent enzyme that does the same job in human blood-clotting proteins, down to the positions of its ten introns[2].
Made without a ribosome
The first hint of a third route came in the early 1960s. Edward Tatum's group found that bacteria went on making tyrocidine, a cyclic peptide containing D-amino acids, even when protein synthesis on ribosomes was blocked. Fritz Lipmann's and Søren Laland's groups then showed how: amino acids are activated with ATP and attached by thioester bonds to the enzyme itself[3].
These enzymes, nonribosomal peptide synthetases (NRPSs), are built like assembly lines. In a typical line each module adds one amino acid, so a heptapeptide needs seven modules; some synthetases reuse modules, as described below. Within a module, an adenylation (A) domain selects and activates the amino acid, a thiolation (T) domain holds it on a flexible arm about 18 Å long, and a condensation (C) domain joins it to the growing chain. At the end, a thioesterase (Te) domain releases the product, either as a linear peptide or by closing it into a ring[3]. The choice of amino acid is fixed by up to about ten residues lining the A domain's binding pocket, a pattern known as the nonribosomal code[3].
Freed from the genetic code, these assembly lines can use building blocks that proteins never contain: ornithine, β-amino acids, hydroxylated and halogenated amino acids, α-hydroxy acids[3]. Extra domains add N-methyl groups to the backbone; cyclosporine, a ring of 11 residues, has seven of them[3]. Nonribosomal peptides range from tripeptides upwards; in 2017 the largest known was the 25-residue syringopeptin 25A[3]. Some lines reuse their modules: gramicidin S synthetase builds the same short unit more than once and closes the copies into one symmetric ring[3].
The producers are mostly bacteria and fungi; an NRPS in the fruit fly and one in a nematode appear to be rare examples in animals[3]. Their products include more than 20 marketed drugs, among them penicillin, vancomycin, daptomycin, bleomycin and cyclosporine, as well as toxins such as the ergot alkaloids, which caused the medieval poisoning known as St Anthony's Fire, and the liver toxins microcystins from cyanobacterial blooms[3]. More on the cyclic ones is on the cyclic peptides page.
What the peptides are for
The same few jobs come up again and again, across very distant organisms.
Signals. In animals, neuropeptides control pain, feeding, stress, water balance and glucose metabolism[1]. Plants rely on peptide signals too. CLAVATA3, a peptide of 12 to 13 residues, maintains the pool of stem cells at the growing tip of the shoot[4]. Systemin, an 18-residue peptide from tomato, is released where insects attack and triggers defences, including protease inhibitors and volatiles that attract the insects' enemies[4]. Bacteria use small cyclic peptides to sense their own numbers; in Staphylococcus aureus such a signal switches on virulence genes[2].
Defence. Many peptides kill microbes or deter predators. Nisin and other lantibiotics are made by bacteria and act against other bacteria[2]. Plants carry large families of antimicrobial peptides: a motif search found 317 defensin-like genes in the small weed Arabidopsis thaliana[4]. Some plant peptides are hidden inside ordinary proteins. Inceptins are fragments of a chloroplast enzyme that are released when a caterpillar digests leaf tissue, and they then signal the plant to defend itself[4]. Antimicrobial peptides are covered on their own page.
Venoms. Venom has evolved many times, in animals as different as sea anemones, cone snails, spiders, scorpions, fish, lizards, snakes, the platypus and shrews[5]. Most venom toxins started as ordinary body proteins: a gene was duplicated, the copy was switched on in the venom gland, and its surface changed while the overall fold was kept. The same protein families, including defensins and natriuretic peptides, have been recruited into venoms again and again[5]. Recruited toxins tend to have stable scaffolds held together by many disulfide bonds[5]. Cone snail venoms are an extreme case: each species makes 50 to 200 conotoxins, and the genes evolve so fast that even closely related species share almost none[2].
Where D-amino acids come from
The ribosome builds proteins only from L-amino acids, yet D-amino acids turn up in natural peptides. The three routes explain where.
On nonribosomal assembly lines they are common. In the words of Süssmuth and Mainz, “the majority of NRPs are equipped with D-amino acids”, which can fix the active shape of the peptide and make it harder for proteases to break down. Most are made on the line itself: an epimerization (E) domain turns the L-amino acid held by the enzyme into its D form, and the next condensation domain accepts only the D form[3]. Tyrocidine and gramicidin S, the peptides that revealed this route, both contain them.
Ribosomal peptides can acquire them only after translation, by enzymes or chemistry acting on the finished chain. In lanthipeptides, a dehydrated serine can be reduced to D-alanine; in polytheonamides, one enzyme converts 18 residues; phalloidin contains one D-amino acid; conotoxins contain D-valine, D-leucine, D-phenylalanine or D-tryptophan; and in cyanobactins, residues next to certain rings switch to the D form spontaneously[2]. Some animal peptides carry D-residues too, made by epimerases that are not yet well characterized[2]. The enkephalins and insulin, which are only cut and trimmed, contain none. The D-amino acids page explains what the D form changes.
The same handful of features, rings, disulfide bonds, N-methyl groups and D-amino acids, turns up on routes that evolved independently, and the sources describe the same effect for each: a more rigid peptide that proteases attack less easily[2][3]. Why that matters for a peptide's lifetime in the body is explained on the peptide half-life page.
In short
- Living things make peptides in three main ways: by cutting them out of larger precursor proteins, by rebuilding a ribosomal precursor with enzymes (RiPPs), or by assembling them on enzyme assembly lines without a ribosome (NRPSs).
- Animal peptide hormones and neuropeptides are cut from prohormones at pairs of basic residues such as Lys-Arg; one precursor can yield several peptides, and tissues can cut it differently.
- RiPPs are made in all three domains of life. Enzymes guided by a leader peptide add rings, cross-links and other changes; nisin, amanitin, cyclotides and conotoxins belong here.
- Nonribosomal peptides, mostly from bacteria and fungi, can contain non-standard building blocks, N-methyl groups and D-amino acids; they include penicillin, vancomycin and cyclosporine.
- D-amino acids are common in nonribosomal peptides and appear in ribosomal peptides only after enzymes or chemistry act on the chain.
- The same jobs recur everywhere: signalling, defence against microbes and predators, and venom, where ordinary body proteins have been recruited as toxins many times.
References
- Hook V, Funkelstein L, Lu D, Bark S, Wegrzyn J, Hwang SR (2008). Proteases for processing proneuropeptides into peptide neurotransmitters and hormones. Annual Review of Pharmacology and Toxicology 48:393–423. DOI: 10.1146/annurev.pharmtox.48.113006.094812
- Arnison PG, Bibb MJ, Bierbaum G, Bowers AA, Bugni TS, Bulaj G, et al. (2013). Ribosomally synthesized and post-translationally modified peptide natural products: overview and recommendations for a universal nomenclature. Natural Product Reports 30(1):108–160. DOI: 10.1039/C2NP20085F
- Süssmuth RD, Mainz A (2017). Nonribosomal peptide synthesis—principles and prospects. Angewandte Chemie International Edition 56:3770–3823. DOI: 10.1002/anie.201609079
- Tavormina P, De Coninck B, Nikonorova N, De Smet I, Cammue BPA (2015). The plant peptidome: an expanding repertoire of structural features and biological functions. The Plant Cell 27(8):2095–2118. DOI: 10.1105/tpc.15.00440
- Fry BG, Roelants K, Champagne DE, Scheib H, Tyndall JDA, King GF, et al. (2009). The toxicogenomic multiverse: convergent recruitment of proteins into animal venoms. Annual Review of Genomics and Human Genetics 10:483–511. DOI: 10.1146/annurev.genom.9.081307.164356