Case Study 32.2 — What Is Actually in the Vial: A Worked Deletion-Sequence Problem

Chapter 32 · How Peptides Are Made

This is a [constructed teaching example]. The peptide below is invented for the purpose. No real product, vendor, manufacturer, or certificate is described, and no number here is a report about any actual batch of anything. The arithmetic is real arithmetic; everything it is applied to is hypothetical. The point of constructing it rather than citing one is that the arithmetic is general — it applies to every synthetic peptide ever made — and a constructed case lets us follow every gram of material without arguing about whose data it is.

The setup

[constructed teaching example]

A hypothetical peptide, which we will call Compound P, is 25 residues long. It contains only standard amino acids and no disulfide bonds. It is made by solid-phase synthesis: 24 couplings after the first residue is loaded onto resin. We will simplify to 25 couplings, as §32.3 does, because it makes the arithmetic legible and changes nothing about the conclusions.

Assume — generously — that the process is competent and couples at 99% efficiency at every step. Assume no capping is performed. Assume the cleavage from resin is quantitative and clean.

Start with 1,000,000 chains on the resin. That is a small number for a real synthesis, but it makes every subsequent number readable.

Question: at the end of the synthesis, what is on the resin?

Step 1 — How much is correct?

0.99²⁵ = 0.7778.

So approximately 778,000 chains are correct, full-length Compound P.

And approximately 222,000 chains — more than one in five — are something else.

This is the number the chapter wants you to feel. Not a trace contaminant. Not a rounding error. More than a fifth of the material, produced by a process performing at a level most people would call excellent.

Step 2 — What is the "something else"?

Without capping, a chain that fails a coupling is not removed from the process. It is deprotected on the next cycle and receives the following residue. It continues through every remaining cycle, grows to within one residue of full length, and is cleaved off at the end along with everything else.

COMPOUND P — where the 222,000 chains went
`[constructed teaching example]`

  Correct full-length (25 residues)          ~778,000 chains    77.8%

  SINGLE deletions (24 residues)             ~196,000 chains    19.6%
    · one missing residue, position unknown
    · distributed across all 25 positions
    · ~7,800 chains PER POSITION on average

  DOUBLE deletions (23 residues)              ~24,000 chains     2.4%
    · two missing residues, in 300 possible combinations

  TRIPLE and higher                            ~2,000 chains     0.2%

  ─────────────────────────────────────────────────────────────────
  Every one of these is a peptide.
  Every one has the same N-terminus and the same C-terminus.
  Every one is between 92% and 100% identical to Compound P by sequence.

That middle block is the whole problem. Nearly 200,000 chains are 24-residue peptides that differ from Compound P by exactly one residue, and they are not one impurity — they are up to twenty-five different impurities, one for each position where a coupling could have failed, each present in small amounts.

A single impurity present at 19.6% is a purification problem. Twenty-five impurities each present at roughly 0.8%, all of them near-twins of the product, is a different and much worse kind of problem.

Step 3 — Why purification struggles

Preparative HPLC separates molecules by how strongly they interact with the column packing, which is governed by their overall chemical character — hydrophobicity, charge, and shape.

Now consider Compound P and one of its deletion sequences, missing a single alanine from the middle of the chain.

The deletion does not elute somewhere else on the chromatogram. It elutes on the shoulder of the main peak, or underneath it.

This has two consequences that pull in opposite directions and cannot both be satisfied:

If you cut the collection window narrowly, you reject the shoulders and get high purity — and you throw away a large fraction of your correct product along with the deletions, because the correct product's own peak has shoulders too. Recovery collapses.

If you cut widely, you recover most of your product and carry the deletions with it.

Every real purification sits somewhere on that curve, and where it sits is a decision made against a specification. This is one of the concrete places where a pharmaceutical process and an unspecified one genuinely differ, and it happens before any question of good faith arises: a manufacturer working to a specification that limits individual impurities to a stated level has to cut where that specification requires. A manufacturer working to no specification cuts wherever they like.

Step 4 — What the certificate would say

[constructed teaching example]

Suppose our hypothetical Compound P is purified to a real and honest 98.0% by HPLC area, and a certificate is issued.

What that certificate truthfully establishes: in the analytical run performed, on the method used, the main peak accounted for 98% of integrated UV-absorbing peak area.

What it does not establish, and this is the list worth memorizing:

That the main peak is Compound P. Area percent is silent on identity. A different peptide, beautifully purified, gives the same number.

That the 2% is one thing. It is more likely twenty-five things, individually small. Some pharmacopeial specifications limit any single impurity as well as the total, precisely because "2% total" and "2% of one unknown compound" are different risk profiles — and because an impurity present at 0.3% can be pharmacologically or immunologically relevant while contributing almost nothing to a total.

That the 2% is even visible. Anything co-eluting under the main peak is counted as product. The method has to be demonstrated to resolve this particular process's impurities. An unnamed method carries no such demonstration.

That 98% of the vial's mass is peptide. From §32.4: counterion and residual water are not in the denominator of an area-percent calculation and can be a double-digit percentage of the powder.

Anything about sterility, endotoxin, particulates, or stability. None of these appear on a chromatogram.

Step 5 — What capping would have changed

Suppose the same synthesis had included a capping step: after each coupling, any chain that failed is permanently blocked.

The yield of correct product does not change. It is still 0.99²⁵ ≈ 77.8%. Capping does not rescue a failed coupling; nothing does.

What changes is the character of the remaining 22%. Instead of ~196,000 near-twin 24-residue deletion sequences, you get a population of truncated chains that stopped at every possible length from 1 to 24 residues. A chain that stopped at residue 11 has a different mass, a different terminus, a different hydrophobicity, and a different retention time. It is a straightforward separation.

THE CAPPING TRADE
`[constructed teaching example]`

  WITHOUT CAPPING              WITH CAPPING
  ───────────────              ────────────
  yield:      77.8%            yield:      77.8%   ← unchanged
  impurities: 24-residue       impurities: chains of every length
              near-twins                   from 1 to 24
  separation: very hard        separation: comparatively easy
  what ends
  up in the                    what ends
  purified                     up in the
  fraction:   deletions        purified
                               fraction:   much less

  Capping buys separability, not yield. Whether it is worth the extra
  cycle time and reagent is a real decision with a real answer that
  depends on the sequence and the specification.

Step 6 — Scaling the intuition

[constructed teaching example]

Run Compound P's arithmetic at three lengths, at 99% per step, to see the shape of the problem:

At 60 residues, nearly half the crude material is not the product — at a coupling efficiency almost nobody would criticize. This is the arithmetic behind everything in §32.6: it is not that chemists lack ambition, it is that the exponential wins.

And it is the arithmetic behind a point worth carrying into Chapter 34. The impurity you should expect in a synthetic peptide preparation is not an exotic contaminant. It is a peptide that is almost the one you wanted. That is a harder thing to detect, a harder thing to remove, and — because it is chemically unremarkable — a much easier thing to overlook.

Discussion Questions

1. The example assumes 99% coupling efficiency at every step. §32.2 explains why that assumption is known to be false. Redo the reasoning qualitatively: if three of the twenty-five couplings ran at 95% because of an aggregating stretch, how would the distribution of deletion sequences change — not just the total? What would that imply for someone trying to design an analytical method?

2. Capping does not improve yield. Construct the strongest argument for capping and the strongest argument against it. Then name the single piece of information that would settle the decision for a particular product.

3. A specification that limits total impurities to 2% and one that limits any individual impurity to 0.2% are very different requirements. Using the Compound P numbers, explain why a preparation could pass the first and fail the second. Which limit do you think protects a patient better, and why?

4. The table in Step 3 shows a deletion sequence differing from the product by about 2.6% in mass. Mass spectrometry can resolve that easily; HPLC retention time often cannot. Why, then, is HPLC purity the number that appears on certificates? Give the practical answer, not the cynical one.

5. Everything in this case study concerns a process operating in good faith and performing well. Nothing here involves fraud, incompetence, or contamination. Restate, in one sentence, what that implies about how much of the uncertainty around unregulated peptide products would remain even if every seller were honest and every synthesis competent.

6. Return to your Evidence Dossier. Pick the compound on your list that is longest, and estimate — using the table in Step 6 — what fraction of a crude synthesis of it would not be the target molecule at 99% per step. Then write down what you would need a seller to tell you before that number stopped mattering to you. Be specific enough that your list could actually be answered.