Amino Acid Analysis: The Reference Method for Peptide Content

A certificate of analysis usually leads with a purity figure from a chromatogram. That figure answers a question about composition: of the material that eluted from the column and produced a signal at the detector, what fraction was the intended peptide? It does not answer the question that determines how much peptide is in the vial. Counterions, bound water, residual salts and other non-peptidic mass either do not elute in the analytical window or do not absorb where the peptide absorbs. They are invisible to the number, and they can be a large fraction of the weighed solid.

Amino acid analysis (AAA) answers the mass question directly. The peptide is hydrolyzed to its constituent amino acids, those amino acids are separated and quantified against calibrants of known purity, and the peptide amount is back-calculated from the sequence. Because the calibration chain ends at pure amino acid standards rather than at the peptide itself, AAA does not require a pure sample of the very thing being measured. It is conventionally employed as a reference method when purity values are assigned to peptide reference standards ^[1].

The scale of the gap between purity and content

The gap is not a rounding error. In work developing a candidate certified reference material for the peptide angiotensin II, the trifluoroacetic acid counterion was measured at nearly 25% of the material by mass and treated as an impurity for purity assignment; the final assigned value was 691 ± 9 mg/g (k = 2) ^[1]. Expressed as a percentage, that is 69.1% ± 0.9% peptide by mass. A chromatographic purity figure for the same material would have said nothing about the remaining 30.9% of the mass.

This is the distinction the site glossary draws between purity and peptide content, and AAA is the measurement that populates the second term.

How the measurement is made

The method dates to Moore, Stein and coworkers in the 1950s, and despite substantial subsequent work the basic approach — hydrolyze, separate, detect, quantify — has remained relatively unchanged ^[2]. For synthetic peptides the standard hydrolysis is 6 M HCl at 110 °C for 20–24 h, preferably carried out in the gas phase ^[3]. The reasons usually given for the gas phase are that the sample never contacts liquid acid and that the acid is easier to remove afterward; that rationale is convention rather than the result of a published comparison.

The liberated amino acids are then separated by ion-exchange or reversed-phase chromatography and quantified against calibrants. The peptide amount follows from dividing each amino acid's molar yield by the number of times that residue appears in the sequence. Several residues are measured, and agreement among them is the internal check on whether the hydrolysis actually went to completion.

What acid hydrolysis destroys

Hydrolysis is not a neutral operation. Under standard conditions tryptophan is destroyed, cysteine is destroyed unless it has been derivatized beforehand, and the amides glutamine and asparagine are deamidated to glutamic acid and aspartic acid respectively ^[3]. Deamidation under prolonged acid exposure is a general phenomenon rather than an artifact of one protocol: in an unrelated acid-catalyzed labeling workflow, deamidation of asparagine, glutamine and carbamidomethylated cysteine during extended exposure to acid was identified as the cause of peptide-dependent inconsistency in quantitation ^[4].

The practical consequence is that a standard AAA report cannot distinguish Asn from Asp or Gln from Glu, and conventionally reports them combined. It also means tryptophan and cysteine values from a single standard hydrolysis should not be trusted as content-bearing residues.

There are established ways to recover tryptophan, and their published recoveries are informative about how much is lost without them. Adding 3% (w/v) phenol to 6 M HCl largely prevented tryptophan destruction during rapid hydrolysis at 166 °C for 25 min or at 145 °C for 4 h, with recovery from proteins of at least 80%; the same addition improved recovery of methionine and carboxymethylcysteine ^[5]. Thioglycolic acid in a trifluoroacetic acid/HCl mixture at 166 °C gave a maximum tryptophan recovery of 88.3% for a 25-min lysozyme hydrolysate, but recovery varied by protein — 88% for lysozyme, 73% for α-chymotrypsinogen A, 85% for apomyoglobin — and the authors noted that extrapolating to zero hydrolysis time brought those values much closer together, at 94%, 87% and 88% ^[6].

Two things follow from that spread. First, a single correction factor for tryptophan does not transfer safely between materials: in that study the recovery measured at a fixed hydrolysis time depended on which of the three proteins was hydrolyzed, though the authors themselves described the variation as slight. Second, extrapolation to zero time is a way of separating what was released from what was subsequently destroyed, but it requires a hydrolysis time course rather than a single time point.

Separate hydrolyses are the usual answer for the labile residues: performic acid oxidation before hydrolysis for methionine and cysteine, and base hydrolysis for tryptophan, are described as standard companion procedures ^[2]. Alternative acids have also been compared directly; in one study across ubiquitin, bovine serum albumin and a soy protein reference material, methane sulfonic acid hydrolysis gave the best recovery of the labile residues cysteine, methionine and tryptophan, and the authors state plainly that a single hydrolysis method with sufficient recovery of all amino acids does not exist ^[7].

What acid hydrolysis fails to finish

The opposite failure is incomplete cleavage. A review of hydrolysis conditions frames the whole problem as a compromise: variation in the ease of cleaving different peptide bonds, differences in amino acid stability, and matrix effects from non-proteinaceous components together argue against any single set of conditions that both hydrolyzes every bond quantitatively and destroys nothing ^[8].

Analysts handle this by choosing which residues to base the calculation on. In one isotope-dilution AAA method the four amino acids used — proline, isoleucine, valine and phenylalanine — were selected both for abundance in the sequences and deliberately to include the harder-to-cleave peptide bonds; the authors reported that peptide amounts calculated from each of the four agreed, and took that consistency as pointing to complete hydrolysis ^[9].

That is the check worth asking for. A content value derived from one amino acid carries no information about whether hydrolysis finished. A value derived from four, with the four in agreement, does.

Derivatization chemistries

Underivatized amino acids absorb weakly in the usual UV range and are retained poorly on reversed-phase columns, so most AAA workflows attach something to them. That is general chemistry rather than a finding from any one study. The chemistries divide into two families.

Post-column

The classical configuration separates underivatized amino acids by ion exchange and reacts the eluate with ninhydrin after the column. Because derivatization happens downstream of the separation, reagent byproducts cannot create false peaks and the separation itself is unaffected by derivative stability. The tradeoff is sensitivity: ninhydrin post-column detection, as implemented in automatic amino acid analyzers, is limited by low sensitivity ^[10].

Pre-column

Pre-column reagents convert amino acids into derivatives that are retained on reversed-phase columns and detected by fluorescence or UV. Derivatization with the fluorescent reagent 6-aminoquinolyl-N-hydroxysuccinimidyl carbamate (AQC) followed by ion-pair reversed-phase separation gave an amino acid composition for bovine serum albumin below 100 ng — about 1.5 pmol — that matched the theoretical composition with low standard deviations, and the authors suggest the method is potentially applicable, as a complement to mass-spectrometry-based proteomics ^[10]. The costs of the pre-column family are real and were summarized in one method paper as derivatives that are unstable, toxic and can be costly, together with incomplete reactions and inadequate chromatographic separations ^[7].

Mass spectrometric detection

Coupling the separation to a mass spectrometer with isotopically labeled amino acids as internal standards changes the error structure rather than just the sensitivity. Because a labeled version of each analyte is present, losses occurring after the standard is added are corrected by ratio. One method development paper sets out the comparison specifically: absorbance- or fluorescence-detected AAA, in their account, covers only a limited set of amino acids, frequently lacks traceability to high-quality reference standards, and returns coefficients of variation above 10% often enough to be worth noting; the isobaric-tagged isotope dilution method reported there averaged 3.57% CV (range 2.72–4.20%) across three peptides measured on seven different days ^[11]. That precision figure belongs to those peptides and that method, not to isotope dilution AAA in general.

Hydrophilic interaction chromatography with tandem mass spectrometry has been used to separate all proteinogenic amino acids including the isobaric pair leucine and isoleucine, with labeled amino acids as internal standards ^[7].

A worked example

Take a peptide of molar mass 1,000 g/mol containing two phenylalanine residues and one valine residue. Weigh 10.0 mg of the lyophilized solid and bring it to 10.0 mL. Hydrolyze an aliquot and analyze.

• Phenylalanine is found at 1.40 µmol/mL. Divide by 2 residues per molecule: 0.700 µmol of peptide per mL.
• Over 10.0 mL that is 7.00 µmol of peptide.
• At 1,000 g/mol — equivalently 1,000 µg/µmol — that is 7,000 µg, or 7.00 mg of peptide.
• Peptide content = 7.00 mg peptide / 10.0 mg weighed solid = 70.0%.

Now the consequence. A researcher who weighs out the solid as though it were pure peptide and prepares a nominal 1.00 mg/mL solution has in fact prepared 0.700 mg/mL. Relative to the nominal value, that is low by 0.300/1.00 = 30.0%. Stated the other way — with the true value as the denominator — the nominal figure is high by 0.300/0.700 = 42.9%. Both statements describe the same solution; which denominator is in use has to be said out loud, because 30% and 43% are not the same claim.

The internal consistency check works the same way. Suppose valine, at one residue per molecule, reads 0.665 µmol/mL, implying 6.65 µmol of peptide against phenylalanine's 7.00 µmol. The disagreement is 0.35/7.00 = 5.0% taking the phenylalanine result as the denominator, or 0.35/6.65 = 5.3% taking valine. A spread of that size is a prompt to look at hydrolysis time, not a result to average silently.

Where the error actually comes from

The calibrants. An AAA result is a comparison against amino acid standards, so it inherits their uncertainty. In one metrological implementation, traceability to SI was established by characterizing the amino acid calibrants by nuclear magnetic resonance, neutron activation analysis and Karl Fischer titration ^[9]. A content figure quoted without any statement of what the amino acids were calibrated against is a ratio to an unknown.

Related peptide impurities. This is the error that a chromatogram and an AAA result share. A deletion sequence or a degradation fragment releases many of the same amino acids on hydrolysis and is counted as parent peptide. For a candidate angiotensin I reference material, high-resolution tandem mass spectrometry identified degradation products including shorter fragments and estimated the major impurities at 10.4 mg/g — 1.04% by mass — and the authors concluded that failing to correct for them would produce a 1% error in the peptide concentration determined by amino acid analysis ^[12].

Racemization. Acid hydrolysis itself generates D-amino acids from L residues. One study addressed it by hydrolyzing model peptides in deuterated hydrochloric acid, so that only D-amino acids newly generated during hydrolysis carried a deuterium at the α-carbon, which lets innate D residues be told apart from artifacts; in a peptide containing D-Phe the same approach identified hydrolysis-induced conversion to L-Phe ^[13]. For a content measurement on an achiral separation this does not change the total, since both enantiomers are counted. It matters when the question is the stereochemical composition of the material, where hydrolysis artifacts and genuine D residues are not separable without an experiment designed to distinguish them.

Blanks. Amino acids are ubiquitous contaminants. Sensitive analysis of low-abundance samples requires strict prevention of environmental contamination ^[10]; at the picomole scale a blank is not a formality.

What AAA does not answer

AAA measures composition, not sequence. Two peptides with the same residue composition and different order are indistinguishable to it. It also cannot see impurities that share the parent's amino acids, which is why the metrological literature does not treat it as sufficient alone: purity assignment for a peptide reference material has been done by combining isotope-dilution AAA with quantitative NMR, which quantifies the intact peptide and so eliminates potential bias from hydrolysis, and with a mass balance approach in which each impurity class is measured separately — in that case including a validated fluorine-19 qNMR measurement of the trifluoroacetic acid counterion ^[1]. Both AAA and qNMR were noted there as susceptible to interference from related peptide impurities ^[1].

Under favorable conditions, and taking the destroyed and deamidated residues into account, quantitation better than 5% can usually be obtained by AAA on synthetic peptides ^[3]. That is a statement about a well-run analysis with the limitations acknowledged, not a floor that every reported number meets.

Reading an AAA number on a document

Four questions make the difference between a content value that means something and a number on a page.

• Which residues was it calculated from, and did they agree? One residue gives no check on hydrolysis completeness; three or four in agreement do.
• What were the hydrolysis conditions? Acid, temperature, time, phase, and any protective additive. Without them, the recoveries for the labile residues are unassessable.
• Are tryptophan and cysteine quoted, and from what? If they appear at all, they should come from a separate hydrolysis, not from the standard HCl run.
• What were the amino acid calibrants? Their characterization is the root of the traceability chain.

A great deal of AAA practice is convention rather than measurement: the 24-hour hydrolysis, the internal standard added before hydrolysis, the combined reporting of Asx and Glx, the exclusion of air from the hydrolysis tube. These conventions are defensible and widely shared, but they are not each backed by a primary determination, and a document that reports a content figure without its conditions is asking to be taken on faith about all of them at once.