The analytical case for a full-length labeled standard
Isotope dilution theory, spike-point selection, the published head-to-head evidence, and the questions an MS group asks before putting a new internal standard into a validated method.
What a stable isotope-labeled standard actually is
And why the word protein in that phrase does most of the work.
A stable isotope-labeled standard is a molecule chemically identical to your analyte, built from atoms of a heavier isotope. Ours are full-length proteins expressed cell-free from a reaction in which L-arginine and L-lysine carry six 13C atoms in place of 12C. The primary sequence is the same, and a correctly folded preparation presents closely matched surface chemistry and chromatographic behaviour.
The qualification matters and is stated here rather than buried: conformation, disulfide status, aggregation and post-translational modification can differ between any recombinant standard and its endogenous counterpart, and where they differ, behaviour can differ with them. This is a limitation shared by every recombinant standard format, including cell-expressed ones. It is a reason to match denaturation and reduction conditions deliberately, not a reason to prefer a peptide.
What changes unambiguously is mass. Trypsin cleaves after arginine and lysine, so in a tryptic digest every peptide except the protein's C-terminal one carries exactly one heavy residue — a uniform +6 Da shift. The mass spectrometer sees two populations of the same peptide, separated by six mass units and otherwise indistinguishable.
Because the heavy and light forms co-elute and ionise alike, their intensity ratio corrects for ion suppression, gradient drift and instrument response — the dominant sources of run-to-run variability in a targeted assay. The ratio is the measurement; the instrument is largely reading it.
This is isotope dilution — the reference principle underlying trace quantification in analytical chemistry for decades. The question that distinguishes a good protein assay from a poor one is where in your workflow the standard is allowed to join the sample. A labeled protein can join at the beginning. A labeled peptide cannot.
The published basis
The AQUA method established synthetic heavy peptides as absolute quantification standards, with the explicit limitation that the standard enters after proteolysis.1 PSAQ — protein standard absolute quantification — was introduced specifically to close that gap, using full-length labeled proteins spiked at the start of the workflow.3,8
Cell-free synthesis has been applied to producing labeled internal standards for targeted proteomics, and commercial cell-free labeling systems report greater than 90% stable-isotope incorporation with expression in under eight hours, against two to three days for cell-based expression.6,7
Avenyx is developing its standards against internal design targets for isotope incorporation, purity and quantification accuracy. These are design specifications — we have not published performance data, and nothing on this page should be read as evidence of validated product performance.
How a ratio becomes an absolute amount
Five steps. The standard has to be present for all of them — that is the entire argument.
-
Spike a known amount, as early as possible
A quantified amount of labeled protein — by amino acid analysis, or by A280 where its limitations are acceptable — is added to the sample before any handling step that can lose material: before capture, before depletion, before ultracentrifugation, before digestion.
-
Let both forms undergo the same workflow
To the extent the standard and the analyte behave alike, whatever fraction of the endogenous protein is lost on a resin, adsorbed to plastic, or left undigested, a comparable fraction of the standard is lost with it. Loss moves from an error term toward a shared denominator.
-
Digest and acquire
Trypsin releases light and heavy versions of every proteotypic peptide. In PRM, MRM or DIA the two are monitored as a co-eluting pair separated by +6 Da on the precursor and on every Arg- or Lys-terminated fragment.
-
Take the ratio
Light area divided by heavy area. Injection volume, ionisation efficiency, matrix suppression, column condition and day-to-day instrument response largely cancel, because both species experience them together.
-
Multiply by the amount you added
Ratio × known spike = amount of endogenous protein in the original sample. No external calibration curve in matrix, and no reliance on an unlabeled recombinant calibrator that may not behave like the endogenous molecule. A linearity assessment is still required — see the FAQ.
If enrichment happens at the protein level, only a protein standard is in the tube for it.
Immunocapture and AAV affinity capture
Anti-protein IP · Protein A/G · AVB Sepharose · POROS CaptureSelect
Affinity capture is often the step in a targeted assay with the largest and least predictable recovery. Binding depends on epitope accessibility, resin lot, incubation kinetics, wash stringency, elution chemistry, and the matrix the sample arrived in. Recovery is rarely complete and rarely identical between runs.
A synthetic peptide cannot correct for any of it. The epitope the antibody or capture ligand recognises does not exist in a tryptic peptide, and by the time the peptide exists the capture step is already over.
In immunoaffinity LC-MS/MS practice, labeled peptides in a protein-IA method are "typically introduced after IA capture but before digestion" — structurally unable to report on capture losses. Labeled full-length proteins "have been shown to deliver superior assay performance, particularly in the presence of patient-specific matrix effects," and are recommended where higher validation rigour is required.4
A ¹³C-labeled full-length protein presents the same epitope and, when correctly folded, a closely matched conformation and charge distribution to the capture ligand. It binds with the analyte, washes with it, and elutes with it. Whatever the capture step did, it did to both.
For AAV, the same logic applies to an assembled labeled VLP on AVB or CaptureSelect resin. A labeled capsid experiences capsid-level chromatography; a labeled VP peptide experiences none of it. Capture recovery becomes a measured quantity rather than an assumed constant.
What the published comparisons actually show
Most of the case for full-length standards is mechanistic. A small number of studies have compared the formats directly, in the same workflow, on the same samples. Those are worth reading closely — including where they go against the argument.
The L19 biodistribution comparison
Ravazza et al., Advanced Science, 2026. Three internal-standard strategies, one antibody, one workflow.
An anti-fibronectin-EDB antibody (L19) was quantified in mouse tissue by Protein A capture and nanoLC-HRMS, with three internal standards run against it: a matched full-length SIL L19, an unlabeled homologous antibody used as a surrogate, and a post-digestion SIL peptide. Analyte and standards were tested at an equimolar 10 pmol amount. Both antibody standards entered before homogenisation; the peptide entered after digestion.
| Internal standard | PBS | Liver |
|---|---|---|
| Matched full-length SIL L19 | 1.03 ± 0.03 | 0.98 ± 0.00 |
| Unlabeled surrogate antibody (KSF) | 0.92 ± 0.40 | 0.34 ± 0.08 |
| SIL peptide ALPAPIEK, post-digestion | 0.15 ± 0.12 | 3.59 ± 0.24 |
The matched full-length standard held near the expected ratio in both matrices. The peptide standard moved from 0.15 in buffer to 3.59 in liver — a roughly twenty-fold swing between matrices in the same assay. The matched SIL L19 was reported at >99.5% isotope incorporation, with 11 peptides covering variable and constant regions of both chains.
The cross-matrix swing is the persuasive part, not the distance from 1. A matrix-matched calibration curve can absorb a constant response-factor offset. It cannot absorb a correction factor that changes with the tissue. That distinction is what makes this study relevant to anyone transferring a method between matrices.
The same study's supplementary data qualifies the headline: across the full tissue panel the matched standard gave 1.08 in heart, 1.10 in intestine, 1.04 in lung and 1.09 in tumour, but 1.40 ± 0.52 in kidney and 0.46 ± 1.02 in spleen. Broad matrix compatibility was demonstrated unevenly. Universal matrix-independent correction is not supported by all the reported values.
Whole antibody against peptide formats
Li et al., Analytical Chemistry, 2012.
A whole SIL antibody, an extended SIL peptide and a conventional SIL peptide were compared as internal standards for three therapeutic mAbs. The whole-antibody approach gave more consistent QC accuracy and precision. Internal-standard peak-area CVs were 15.5%, 28.1% and 27.7% respectively (n = 32).
Two qualifications from the authors themselves. Those are internal-standard signal CVs, not final concentration CVs. And the peptides entered after immunocapture, so the comparison confounds standard format with addition timing — which is arguably the point, but it is not a format-only result. The authors noted that optimised preparation can make peptide standards adequate.
Low recovery need not mean poor accuracy
Heudi et al., Analytical Chemistry, 2008.
An isotope-labeled version of the analyte mAb was added before digestion in marmoset serum. Overall method recovery was 14%. The assay still reported a 5–1000 µg/mL range with accuracy and precision meeting the stated limits.
This is the shared-denominator argument in its clearest form: if 86% of the analyte is lost and 86% of the standard is lost with it, the ratio survives. Low absolute recovery is a robustness concern, not automatically an accuracy one — provided the standard was in the tube when the loss happened.
The counterweight
AZD7442 clinical bioanalysis. Huang et al., Bioanalysis, 2024.
A validated RBD-capture assay using post-digestion SIL peptides supported three clinical trials over more than 17 months and at least 720 analytical runs, covering roughly 30,000 samples and around 3,000 incurred sample reanalyses per analyte, at a reported range of 0.300–30.0 µg/mL per antibody.
Reproducible sample preparation can carry demanding clinical work even when the internal standard corrects neither capture nor digestion. Any honest account of this field has to hold both facts at once: the matched standard corrects more, and peptide standards have nonetheless delivered at scale. Which one your assay needs is an empirical question about your workflow, not a general truth.
Limits of the inference
Worth stating explicitly, because the numbers above are easy to over-read:
- The L19 figures are analyte/IS response ratios, not independently calibrated concentration recoveries. A reported SD of 0.00 reflects rounding, not zero variability.
- The surrogate-antibody comparison averaged signals from different peptide sequences. Equal molar amounts of different peptides do not necessarily give equal MS signal.
- No universal fold-improvement, cost reduction, sample-volume saving or sensitivity gain can be assigned to full-length standards from these studies. They differ in capture reagent, matrix, instrument, peptide, calibration and sample volume.
- Matched standards compensate shared matrix effects. They do not remove all of them, do not correct loss that occurred before the standard was added, and an exogenous spike need not equilibrate with endogenous drug–target complexes.
- Bottom-up digestion showing several agreeing peptides supports regional consistency; it cannot prove every detected region came from the same intact, functional molecule.
The strongest claim the evidence supports is a specific one: a matched, characterised full-length standard normalises variability from its addition point through capture, digestion and detection. Claims of universally better sensitivity, guaranteed accuracy, or mandatory regulatory use go beyond it.
A second use: the standard as a functional probe
El Amrani et al., J Translational Autoimmunity, 2019.
A SIL infliximab was used not as a normaliser but as a reagent. Neutralising antibodies in patient plasma prevented its capture by TNF-α, so the labeled-peptide signal fell in proportion to neutralising capacity. The study reported 94% sensitivity and 100% specificity for detecting neutralising antibodies in its evaluated sample set.
This only works with a full-length, correctly folded standard — a peptide has no target to be blocked from. It also raises the bar on what the standard must be: authentic binding behaviour becomes a requirement rather than a convenience.
What this asks of a cell-free product
Production route alone does not establish commutability. A cell-free-derived antibody standard has to be shown to behave sufficiently like the therapeutic molecule in the intended procedure. Sequence matching is necessary but not sufficient — assembly, disulfides, aggregation, glycosylation and binding can all affect capture or digestion, and a workflow that relies on native target binding demands more structural comparability than one that immediately denatures the material.
We would rather say this plainly than let a customer discover it in validation. The qualification evidence a buyer should ask us for is: identity and assembly (peptide mapping, intact and subunit masses, chain integrity, monomer and aggregate assessment); isotope quality (peptide-resolved incorporation, residual light contamination, interference at the assay LLOQ); quantity and stability (documented concentration assignment with uncertainty, formulation, storage and lot comparability); workflow matching (parallel capture recovery and digestion behaviour against the therapeutic, in your matrices); and functional similarity where target or anti-idiotype binding is part of the assay.
The most informative pilot compares a matched full-length standard, a plausible unlabeled surrogate, a conventional SIL peptide and an extended SIL peptide against independently assigned calibrators — evaluating final concentration bias, precision and robustness across matrices and realistic preparation variation, with internal-standard signal CV reported separately. That is a study design we propose, not a result we have.
Why VP stoichiometry needs measuring
Native MS of intact capsids has shown AAV assembly to be stochastic rather than fixed, with modelled VP1:VP2:VP3 ratios ranging from roughly 4:1:95 in AAV1 to 6:14:80 in AAV9 — far from the textbook 1:1:10, and dependent on serotype and expression platform.5 The same work estimates the probability of any given capsid having one exact VP composition and configuration at around 10−14.
An assumed ratio is therefore not a measurement, and a single labeled VLP is an awkward reference for one: it carries its own VP ratio, which itself requires characterisation.
This is why individually expressed VP1, VP2 and VP3 exist as a separate modality. Three independently quantified standards, spiked at three known amounts, anchor the measured VP ratio to three separate calibrations rather than to one particle's composition.
Why we make them cell-free
Labeling a protein inside a living cell means competing with the cell's own metabolism. An open reaction has no such competition.
No endogenous amino acid pool
In a cell, unlabeled arginine and lysine are continuously synthesised and recycled, diluting the label. In a cell-free reaction the amino acid supply is what you put in it, so incorporation is governed by the feed rather than by competition with metabolism.
No metabolic conversion
Arginine-to-proline conversion is a recognised artefact of cell-based labeling, moving label onto residues you did not intend to label and complicating every peptide containing proline. A cell-free reaction has no active arginase pathway to perform it.
Only the target is labeled
In a labeled cell line, every protein in the cell carries the isotope — expensive amino acids spent on material that is discarded. Cell-free synthesis directs the label into the one protein being made.
Which is also the rest of what we do
Avenyx Bio is building a cell-free protein synthesis platform for viral vector manufacturing. Labeled standards are produced on the same platform: the same open reaction that lets us control what a vector is made of is what lets us control which atoms a standard is made of. We describe the platform on the main site; the reaction chemistry itself is proprietary.
Analytical questions
The questions an MS group asks before it will put a new internal standard into a validated method.
Is +6 Da enough separation from the light peptide's natural isotope envelope?
For most tryptic peptides, yes. A peptide of 1,000–2,500 Da has a natural isotope envelope whose M+6 contribution is small relative to the monoisotopic peak, and PRM or MRM further isolates the pair by fragment. The M+6 overlap grows with peptide mass and with the number of sulfur atoms, so it matters for long or Cys/Met-rich peptides.
Where it matters, the correction is deterministic: the natural isotope distribution of a known sequence is calculable, so the M+6 contribution can be subtracted rather than estimated. In practice the more robust route is peptide selection — choose proteotypic peptides short enough that the overlap is negligible.
For assays where residual overlap is unacceptable, ¹³C₆¹⁵N₄-Arg / ¹³C₆¹⁵N₂-Lys (+10 / +8 Da) removes the question entirely. That option can be specified when you submit a request.
What isotope incorporation do you achieve, and what happens if it is incomplete?
Avenyx works to an internal incorporation specification and reports the measured value for each lot. We are in early access and have not published performance data; the current target and the verification method are shared as part of a pilot discussion.
Incomplete incorporation matters because unlabeled standard is indistinguishable from analyte. If a fraction f of the standard is light, it adds to the light channel and inflates the apparent endogenous amount. This is a systematic, one-directional bias.
It is also directly measurable: acquire the standard alone and read the residual light signal. Once measured, the correction is arithmetic. What is not acceptable is an unmeasured incorporation figure — which is why a per-lot value belongs on the certificate rather than in a specification range.
How is the standard itself quantified? A tracer is not a calibrant.
Correct, and this is the step that determines the accuracy ceiling of the whole assay. A labeled protein spiked at an uncertain amount gives a precise ratio multiplied by an uncertain number.
Amino acid analysis is the reference approach — acid hydrolysis followed by quantification of released amino acids against certified standards — because it is traceable and independent of the protein's own spectroscopic properties. UV A280 with a sequence-derived extinction coefficient is faster but inherits any error in aggregation state or in the coefficient itself.
Our quantification method, its uncertainty, and how it appears on the certificate are defined per modality. If your assay must be traceable to a reference material, say so in the request — it changes what we supply.
Does the labeled protein really digest identically to the endogenous protein?
It has the same primary sequence, so it presents trypsin with the same cleavage sites in the same order. This is the property no peptide standard and no concatamer has, and it is why digestion kinetics — including missed cleavages and slow-cleaving sites — tend to affect both species together.
The honest qualification is conformation and modification. If the endogenous protein is disulfide-bonded, aggregated, glycosylated or bound to a partner in a way the standard is not, accessibility can differ. This is a real limitation shared by every recombinant standard, including cell-expressed ones, and it is why denaturation and reduction conditions should be matched deliberately rather than inherited from a legacy SOP.
Where exactly should I spike it?
As early as the sample allows. Every step between the spike point and the measurement is corrected; every step before it is not.
For a protein-level immunocapture assay: into the neat matrix, before capture. For an AAV workflow with affinity capture: before the capture load. For a simple lysate digest: into the lysate before denaturation. If the sample must be diluted or buffer-exchanged first, spike into the first tube it occupies.
A useful diagnostic: spike the same standard at two points — before and after the step you are questioning — in parallel aliquots. The difference between the two ratios is that step's recovery, measured rather than assumed.
Do I still need a calibration curve?
Not in the classical matrix-matched sense, which is the point of isotope dilution. You do need a linearity assessment: a series of light-to-heavy ratios spanning your expected range, confirming the response is linear and identifying where it is not — typically at extreme ratios where the minor channel approaches the noise floor or the major channel saturates the detector.
Practically, that means choosing a spike level that keeps the light:heavy ratio within roughly an order of magnitude of unity across your samples. A standard spiked far above or far below the analyte gives worse precision than one spiked near it.
How do you avoid arginine-to-proline conversion and other label scrambling?
Arg→Pro conversion happens because living cells run arginase and the downstream proline biosynthesis pathway, converting supplied heavy arginine into heavy proline. The result is proline-containing peptides carrying unexpected mass shifts and split signal. A cell-free reaction is not running that metabolism.
Cell-free reactions are lysate-based and so are not literally metabolically inert; residual enzymatic activity and any unlabeled amino acid carried in with the lysate are real considerations, and they are exactly what our process controls address. The verified answer for any given lot is empirical — the observed mass spectrum of the standard, which is what the certificate reports.
What about missed cleavages and semi-tryptic peptides?
They affect both channels together, which is the advantage. A peptide with a consistently missed cleavage still gives a valid light:heavy ratio, because the standard missed the same cleavage.
The practical guidance is unchanged: prefer fully cleaved, non-ragged proteotypic peptides for the primary quantifier, and monitor a missed-cleavage form as a qualifier. If the two disagree, digestion is incomplete in a way that varies between light and heavy — usually a sign of a conformational difference worth investigating.
Can one labeled VLP quantify VP1, VP2 and VP3 at once?
It can quantify total capsid protein robustly, because the whole particle is a single known amount and every VP-derived peptide reports against it.
VP stoichiometry is a harder question. A labeled VLP carries its own VP ratio, which — per the native MS work on stochastic capsid assembly — is not a fixed constant.5 Using it to measure your capsid's VP ratio means comparing against a ratio that itself needs characterising.
This is why we offer individually expressed VP1, VP2 and VP3 as a separate modality. Three independently quantified standards, spiked at three known amounts, anchor the measured VP ratio to three separate calibrations rather than to one particle's composition.
Will a labeled VLP behave like my capsid on AVB or CaptureSelect resin?
It should, to the extent that it presents the same capsid surface — the binding determinants for these ligands are conformational epitopes on the assembled particle, which is precisely why an assembled standard is the right format and a peptide is not.
The qualification worth stating plainly: our VLPs are genome-free. If your capture step or your matrix distinguishes full from empty capsids, the standard tracks the empty population's behaviour. For most affinity capture chemistries this is not a difference; for density- or charge-based steps it can be. This is a good thing to discuss for your specific workflow rather than to assume either way.
My target is glycosylated or otherwise post-translationally modified. Now what?
Choose proteotypic peptides that do not carry the modification, and the standard works normally — this is standard practice and usually sufficient.
Where the modification itself is the analyte, or where it is dense enough that no clean proteotypic peptide exists, a cell-free standard is not automatically the right tool: the PTM profile of a cell-free product will not match a mammalian-expressed glycoprotein. We will tell you when that is the case rather than sell you a standard that answers a different question.
Does this work with PRM, MRM and DIA equally?
Yes. The standard is agnostic to acquisition scheme — it produces a co-eluting heavy partner for every proteotypic peptide, and any method capable of measuring a peptide pair can use it.
PRM and MRM give the cleanest ratios because the pair is explicitly targeted. In DIA the heavy channel serves both as quantifier and as a per-peptide retention time and identity anchor, which is often as valuable as the quantification itself.
How stable is it, and how should I store and handle it?
Format, buffer, concentration, carrier protein where appropriate, freeze-thaw guidance and storage conditions are defined per modality. Stability data is generated as part of each pilot engagement.
The general handling concern with any low-concentration protein standard is adsorptive loss to container surfaces — which, notably, is itself corrected if the standard is already in the tube with the analyte, and is not corrected at all if the loss happens to the standard stock before spiking. Aliquoting and a suitable carrier matter more than they do for peptide standards.
Which peptides should I monitor — do you supply candidate transitions?
Peptide and transition selection is target- and matrix-specific, and we would rather do it with you than hand over a list generated in isolation. Selection criteria we work to: fully tryptic, no Met or Cys where avoidable, no known modification sites, unique to the target across the relevant background proteome, and empirically well-behaved in your matrix rather than merely predicted to be.
Candidate peptide lists and method-setup support are part of a pilot engagement.
References
Cited claims on this page are drawn from the published literature and are the work of their authors, not of Avenyx Bio.
- Gerber SA, Rush J, Stemman O, Kirschner MW, Gygi SP. Absolute quantification of proteins and phosphoproteins from cell lysates by tandem MS. PNAS 2003;100(12):6940–6945. pnas.org
- Beynon RJ, Doherty MK, Pratt JM, Gaskell SJ. Multiplexed absolute quantification in proteomics using artificial QCAT proteins of concatenated signature peptides. Nat Methods 2005;2(8):587–589. Nature Methods
- Brun V, Dupuis A, Adrait A, et al. Isotope-labeled protein standards: toward absolute quantitative proteomics. Mol Cell Proteomics 2007;6(12):2139–2149. PubMed
- Protein biomarker quantification by immunoaffinity liquid chromatography–tandem mass spectrometry: current state and future vision. Clin Chem 2020;66(2):282–301. Clinical Chemistry
- Wörner TP, Bennett A, Habka S, et al. Adeno-associated virus capsid assembly is divergent and stochastic. Nat Commun 2021;12:1642. Nature Communications
- Cell-free synthesis of stable isotope-labeled internal standards for targeted quantitative proteomics. Synth Syst Biotechnol 2018;3(2):97–104. PMC
- Thermo Fisher Scientific. Cell-free protein expression for generating stable isotope-labeled proteins — application note. thermofisher.com
- Picard G, Lebert D, Louwagie M, et al. PSAQ standards for accurate MS-based quantification of proteins: from the concept to biomedical applications. J Mass Spectrom 2012;47(10):1353–1363. Wiley
- Ravazza D, et al. Quantitative mass spectrometry-based biodistribution of monoclonal antibodies: an alternative to radio-biodistribution. Advanced Science 2026. Wiley
- Li H, et al. General LC-MS/MS method approach to quantify therapeutic monoclonal antibodies using a common whole antibody internal standard, with application to preclinical studies. Anal Chem 2012;84:1267–1273. PubMed
- Heudi O, et al. Towards absolute quantification of therapeutic monoclonal antibody in serum by LC–MS/MS using isotope-labeled antibody standard and protein cleavage isotope dilution mass spectrometry. Anal Chem 2008;80(11):4200–4207. ACS Publications
- Huang Y, et al. Comprehensive performance evaluation of a ligand-binding assay–LC–MS/MS method for co-dosed monoclonal anti-SARS-CoV-2 antibodies (AZD7442). Bioanalysis 2024. Taylor & Francis
- El Amrani M, et al. Quantification of neutralizing anti-drug antibodies and their neutralizing capacity using competitive displacement and tandem mass spectrometry: infliximab as proof of principle. J Transl Autoimmun 2019;1:100004. ScienceDirect
- Faria M, et al. Comparison of a stable isotope labeled (SIL) peptide and an extended SIL peptide as internal standards to track digestion variability of an unstable signature peptide. J Chromatogr B 2015. ScienceDirect
- Prasad B, Unadkat JD. Comparison of heavy labeled (SIL) peptide versus SILAC protein internal standards for LC-MS/MS quantification of hepatic drug transporters. Int J Proteomics 2014;2014:451510. PubMed
- Matuszewski BK, Constanzer ML, Chavez-Eng CM. Strategies for the assessment of matrix effect in quantitative bioanalytical methods based on HPLC-MS/MS. Anal Chem 2003. PubMed
- Assessment of matrix effect in quantitative LC-MS bioanalysis. Bioanalysis 2024. PubMed
Tell us what you want quantified
We are currently evaluating a limited number of pilot projects. If another internal-standard format is better suited to your assay, we'll tell you.