A single band on a gel and a single intact mass still leave one question open: which residue is actually first, and what follows it. Edman degradation answers by chemistry rather than by inference. One residue is removed from the amino terminus per cycle, every bond further down the chain is left intact, and the removed residue is identified on its own. The method has been largely neglected since mass-spectrometry-based protein sequencing became established [1], but what it does has no direct substitute: it reads the terminus, in order, on the species you actually purified.
Three reactions, deliberately kept apart
Coupling comes first. Phenyl isothiocyanate (PITC) thiocarbamylates the free α-amino group of a polypeptide immobilised on a support — in one standard configuration, a Polybrene-coated glass-fibre filter sitting in the sequencer cartridge [2]. The reaction needs base. In gas-phase instruments the coupling reagent delivered in the carrier gas stream is 5% aqueous triethylamine and the cleavage reagent that follows it is anhydrous trifluoroacetic acid; conversion of the released anilinothiazolinone (ATZ) to the phenylthiohydantoin (PTH) is then carried out separately in 25% aqueous trifluoroacetic acid, and the PTH amino acids are analysed by reversed-phase HPLC [3].
Keeping the two acid steps apart is the point. Cleavage has to remove residue 1 and leave the rest of the backbone untouched, so it is run in anhydrous acid [3]. Acid exposure is still not free: random acid hydrolysis of the sample accumulates as background stepwise through a run, and that accumulation is what limits how far extended sequencing data can be pushed [4].
Cleavage also does not hand you a single compound. The post-cleavage products of the classical degradation include the ATZ, the phenylthiocarbamyl (PTC) amino acid and the PTH, and an alternative conversion chemistry that drives all of them to a homogeneous ATZ preparation was developed precisely so that one species reaches the detector [5]. In the classical route that one species is the PTH, produced in the separate aqueous-acid step [3]. The ATZ is what comes off the chain; in the classical route it is an intermediate rather than the species quantified.
Identifying the residue that came off
Identification is by retention time against PTH standards on reversed-phase HPLC [3], and that readout sets the sensitivity floor of the classical method. A fluorescent alternative — isolating the ATZ from the solid-phase membranes and reacting it with a primary amine, 4-(1'-cyanoisoindolyl)aniline, at an average yield of 85% under neutral pH conditions — gave limits of detection of 0.16 to 0.52 pmol on column at a signal-to-noise ratio of 3, approximately ten times higher sensitivity than ultraviolet absorbance detection of PTH products in the conventional degradation. It was demonstrated by manual sequencing of insulin chain B, a 30-residue peptide [6].
Retention time is only as good as the gradient behind it. Some widely used standard protocols for separating PTH derivatives do not resolve PTH-Trp from diphenylurea, a by-product generated in variable amounts during the degradation, and PTH-Trp is in any case usually recovered in low yield under typical automated conditions. A modified reversed-phase gradient made tryptophan assignment consistent and reproducible on Applied Biosystems 477A sequencers with on-line PTH identification, even at 5 pmol [7].
Chromatography also buys something mass alone does not. Using 4-(3-pyridinylmethylaminocarboxypropyl)phenyl isothiocyanate in place of PITC, microbore reversed-phase HPLC gave baseline separation of the isobaric leucine and isoleucine derivatives; adding on-line electrospray ionisation mass analysis to the same separation dramatically improved signal-to-noise and the ability to interpret sequence data conclusively [8].
Repetitive yield, and how it sets the read length
Two numbers describe a run. Initial yield is the fraction of loaded material that gives callable signal in cycle 1; repetitive yield is the fraction carried forward per cycle. Manual gas-phase chemistry in the 100–200 pmol range gave initial yields of 30–60% and repetitive yields of 90–96%, with protein samples sequenced through 15–20 cycles and peptide samples usually run to the C-terminus, at a cycle time of 15–20 min for six to ten samples in parallel [3].
Repetitive yield compounds, and that compounding is the whole story of read length. Arithmetically, 95% per cycle leaves about 21% of the starting signal after 30 cycles; 90% leaves about 4%. A run ends not because the chemistry stops working but because the residue-specific peak sinks into the background. Instrument design aimed at low repetitive loss pushes the boundary out: the gas-liquid solid-phase sequenator gave useful sequence data on as little as 5 pmol of protein and extended runs of more than 30 residues on subnanomole quantities of proteins purified by SDS-PAGE, with the sample embedded in a Polybrene matrix on a porous glass-fibre disc — not covalently attached, but essentially immobile through the cycle because only relatively apolar liquid-phase solvents pass through the cell. The same instrument could sequence hydrophobic peptides to completion, at a degradative cycle time of 50 min using a complete double cleavage program [9]. That cycle time is the throughput cost of the complete double cleavage program on that instrument: close to an hour per position, so a 30-cycle run occupies it for more than a day.
Yield is not the only clock. Lag, the reappearance of a previously called residue in later cycles, is evaluated separately. The Edman Sequencing Research Group of the ABRF tested it directly by distributing one protein in three configurations — with an N-terminal polyhistidine tag, with a polyalanine tag, or with no tag — and asking core facilities to return uncorrected picomole yields for the first 17 cycles. The majority of participating laboratories successfully called all 17 cycles for all three constructs, but the polyhistidine construct was harder: lag was observed earlier and more consistently than with the polyalanine tag, histidine yields within the tag were significantly lower than the alanine yields, and the terminal histidines were demonstrated to be responsible for the high lag [10].
Polar residues are also the ones that extract badly out of the reaction cell. Covalently immobilising peptides through their terminal α-carboxyl and side-chain carboxyl groups on aminophenyl-derivatised glass-fibre paper allowed polar extraction solvents to be used, which increased the yield of PTH-His and PTH-Arg, and gave extraordinarily low backgrounds; peptides prepared this way could be routinely sequenced through to and including the C-terminal residue. Covalent attachment also loosened the sequencing conditions generally, including the use of alternative Edman-type reagents with enhanced detectability such as the chromophoric 4-(N,N'-dimethylamino)azobenzene-4'-isothiocyanate [11]. The trade is the extra chemistry and the extra handling step before cycle 1, which is exactly what you cannot afford at the low end of the loading range.
Getting the sample onto the sequencer
PVDF membrane is the usual presentation for gel-purified material. Between 7 and 250 pmol of myoglobin, β-lactoglobulin and other proteins and peptides can be spotted or electroblotted onto PVDF, stained with Coomassie Blue and sequenced directly, with no chemical activation of the membrane and no Polybrene pretreatment; average repetitive yields of 84–98% and initial couplings of 30–108% were comparable with the yields measured for the same proteins spotted onto Polybrene-coated glass-fibre discs [12].
Purity is not a nicety. Every cycle releases one derivative per amino terminus present, so a second chain in the preparation writes its own ladder over the first. That is not hypothetical: in recombinant humanised IgG preparations whose heavy chain was blocked, the light chain supplied its own first-cycle signal and was deliberately used as an internal standard, with the extent of heavy-chain deblocking read as the ratio of deblocked heavy chain to light chain in cycle 1 [4]. The same arithmetic works against you when the second species is a contaminant rather than a known partner, and two overlapping ladders offset by a residue or two are hard to distinguish from lag. PVDF fixes the handling problem, not the chemistry problem: a band that transfers cleanly and stains well can still be blocked, and the blot gives no warning of that before cycle 1.
The blocked amino terminus
The commonest failure mode is that the run never starts. N-terminal acetylation is a common post-translational modification and it blocks Edman degradation during sequencer analysis [13]. It is also widespread: COFRADIC analysis of 742 human and 379 yeast protein N-termini, described at the time as the largest eukaryotic dataset of N-terminal acetylation, found a higher proportion of N-terminally acetylated proteins in humans (84%) than in yeast (57%), with the major N-terminal acetyltransferase NatA acting on subclasses of proteins carrying Ser-, Ala-, Thr-, Gly-, Cys- and Val- N-termini [14].
Pyroglutamate is the other routine block, and it needs no enzyme at all. N-terminal glutaminyl and glutamyl residues tend to form pyroglutamic acid by intramolecular cyclisation, with spontaneous rate constants in aqueous solution at pH 6.5 of about 10-6 s-1 for glutamine and 10-9 s-1 for glutamic acid, roughly three orders of magnitude apart [15]. A glutamine at position 1 is therefore a standing liability in any preparation held in solution.
Two enzymatic deblocking routes are in general use. Pyroglutamate aminopeptidase acts on Nα-pyrrolidone carboxyl proteins in solution or blotted onto a membrane. Acylaminoacyl-peptide hydrolase handles other Nα-acyl blocks, but sequencing with it requires fragmenting the protein before unblocking, with the newly generated peptides chemically blocked using succinic anhydride or phenyl isothiocyanate/performic acid so that the acylated N-terminal peptide is the only substrate the hydrolase should see; the whole peptide mixture then goes to sequence analysis [16]. An optimised pyroglutamate aminopeptidase procedure — reduction and carboxymethylation in guanidine buffer, buffer exchange by gel-permeation chromatography, then overnight digestion at 37 °C — deblocked five different recombinant IgGs, including one blocked on both heavy and light chains, in nearly quantitative yields. Quantitative matters, because any portion that remains blocked contributes to the undesirable background rather than simply being silent [4].
For acetyl blocks, chemical deacetylation is the alternative to an enzyme. Incubation in trifluoroacetic acid/methanol (1:1 by volume) at approximately 47 °C for two to three days gave efficient deacetylation with initial yields up to approximately 50% of the amount applied for deblocking. Deacetylation relative to internal peptide bond cleavage, evaluated as recovery of residues from the deblocked sequence over those from the background, ran up to the order of 10:1 for peptides and greater than 2:1 for proteins. Yields may still vary and some sequences are only partly susceptible, but in many cases the chemistry allows unambiguous interpretation of an N-terminally acetyl-blocked sequence [13].
Residues that need handling before cycle 1
Cysteine is the residue that will silently cost you a position. Cysteine residues in polypeptides are not easily identified during automated N-terminal sequence analysis. Reaction of the side chain with 4-vinylpyridine and identification as the pyridylethylated PTH derivative (PE-PTH-Cys) is the standard answer, but after that reaction a desalting step is necessary. On-sequencer pyridylethylation avoids the desalting when sample amounts do not allow it, at the cost of preview of the consecutive amino acid, which is usually observed; derivatising the peptide with PITC before adding the 4-vinylpyridine suppressed that preview and gave PE-PTH-Cys in much higher yields, with fewer side products and no losses from extra sample handling [2].
The derivative chemistry has since been put on a formal analytical footing. PTH-S-methylcysteine and PTH-S-carbamidomethylcysteine were characterised and validated as bioanalytical reference standards for cysteine detection and quantification, then tested as standards on a series of model peptides, and applied to the native disulfide-bonded peptides CCAP-vil, μ-conotoxin KIIIA and human insulin as case studies to determine their disulfide connectivity completely independently of mass spectrometry. Partial or complete reduction and chemical cysteine modification remain required first steps [1].
Serine causes trouble at the detection end rather than the chemistry end. Work locating the single phosphorylation site in recombinant rat and murine immune interferon preparations reported inherent problems in measuring both PTH-serine and PTH-phosphoserine with an automated sequenator, and worked around them by subjecting aliquots of the peptide to partial Edman degradation and analysing the products by HPLC rather than reading the sequencer trace straight through [17]. Tryptophan, as above, is a gradient problem rather than a chemistry problem — a different kind of failure with the same consequence for the call. The procedural lesson covers both: derivative behaviour is residue-specific, so a position occupied by a residue whose derivative is unstable, poorly extracted or poorly resolved warrants an orthogonal check rather than a confident assignment from retention time alone.
What an Edman call licenses that a spectrum does not
De novo sequencing reconstructs sequence from mass differences between fragment ions, and peptide mapping identifies a protein from the masses and fragmentation of its proteolytic fragments. Both infer the terminus rather than reading it, and the inference is weakest exactly at the terminus: the b1 ion is typically not observed in the fragmentation of protonated peptides, and determining the identity and order of the two N-terminal amino acids is named as one of the major limitations of traditional mass-spectrometric peptide sequencing [18]. Disulfide connectivity in multiply bridged peptides is a second gap — it is still not detectable directly through a simple and sufficiently validated protein sequencing or peptide mapping method [1].
That is the residual niche, and it is narrow but real. On a single purified species, an Edman run returns an ordered list of residues, each identified as a discrete chemical entity by retention time, with no database and no fragmentation model in the loop. It is also why interfacing Edman chemistry to a mass spectrometer is complementary rather than redundant, as the leucine/isoleucine separation described earlier shows [8].
Be explicit about what a result does and does not support. A clean twelve-cycle call licenses the statement that those residues are the first twelve of the chain you loaded, in that order. It says nothing about the rest of the sequence, nothing about species removed during purification, and nothing about heterogeneity present below the initial-yield threshold. A blank first cycle is not a sequence result at all: a blocked terminus [13, 15], an unalkylated cysteine at position 1 [2] and simply too little material [3] all produce the same uninformative trace, and only orthogonal evidence separates them.