Permeabilization Optimization (Pilot)

• Applicable samples: Any new tissue type, fixation route, block, or pepsin lot
• Version: v0.3
• Last modified: 2026-08-18

v0.x — designed, not yet run here

This page is a design, assembled from published tissue-optimization workflows. It has not yet been executed in this lab. It moves to v1.0 the first time it is run end to end and the result is confirmed. Treat the specific ladder values as starting points, not as validated settings.

Why this exists

Permeabilization is the one step in the workflow with a genuine optimum rather than a monotonic dose-response: too little and probes and enzymes cannot reach the RNA; too much and the RNA is no longer anchored and is lost across the ~16 h hybridization, the formamide stringency washes and the overnight RCA.

Two consequences make it worth a dedicated pilot:

• "It produced data" is a floor, not an optimum. In the only published pepsin time-course read out on an in situ platform, transcript density varied 7.6- to 22-fold across the first minute of digestion alone, and every condition tested still produced data.¹ A working condition can be far from the best one and give no sign of it.
• The optimum moves. It differs by tissue, by fixation history, by section thickness, by pepsin catalogue number and lot, and — within one section — by compartment. It is not a constant to be looked up.

When to run it

• A tissue type not run before (required)
• A new pepsin catalogue number or lot (required — specific activity spans ~13× across products; see Pepsin Solution Preparation)
• A change of fixation route, section thickness, or embedding
• Signal loss on a sample type that previously worked
• Before committing an irreplaceable clinical block

1. Design: one slide, serial sections, one variable

The whole value of the design is that every step after permeabilization is shared, so section-to-section and batch-to-batch variance drops out of the comparison. Adapted from the Visium Tissue Optimization gasket workflow³ and its use on an in-situ platform.¹

• Cut 4–6 serial sections from the same block onto one slide.
• Mount one PDMS chamber over each section, so each gets its own ~100 µl reaction well.
• Apply a different permeabilization condition to each chamber. Change one variable across the ladder — time, or concentration, or acid — never two.
• Run hybridization, ligation, RCA and imaging on all chambers together.

Use a control panel, not the real panel

Read out with a 1–3 plex positive-control padlock set against high-abundance housekeepers (ACTB, GAPDH, POLR2A) plus one negative padlock matching no human transcript. Counting rolonies is the measurement; a multiplex panel only adds decoding as a confound.

For a tumour panel, add one marker of the compartment you actually care about — e.g. PTPRC/CD3E for immune infiltrate. Housekeepers report the optimum for high-abundance transcripts in large cells, which is not the optimum for a low-abundance transcript in a lymphocyte.

2. Ladders

2a. Fresh frozen — is the protease needed at all?

Starts at zero protease deliberately: four published padlock/RCA systems permeabilize FF tissue with none.⁴⁵

Chamber	Condition
1	0.1 M HCl, 3 min, RT
2	0.1 M HCl, 5 min, RT — current FF protocol v6.0
3	0.1 M HCl, 10 min, RT
4	0.1 mg/mL pepsin in 0.1 M HCl, 2 min, 37 °C
5	0.4 mg/mL pepsin in 0.1 M HCl, 2 min, 37 °C — the pre-v6.0 condition
6	1% SDS 2 min RT + cold 70% methanol 60 min

2b. FFPE — how much pepsin, on top of the retrieval

Keep the citrate retrieval; an in-house run without it gave almost no signal.

Chamber	Condition (after the standard FFPE retrieval and re-fixation)
1	no pepsin
2	0.1 mg/mL pepsin, 10 min, 37 °C — current protocol (v3.0)
3	0.1 mg/mL pepsin, 20 min, 37 °C
4	0.1 mg/mL pepsin, 30 min, 37 °C — the published FFPE condition
5	0.1 mg/mL pepsin, 45 min, 37 °C
6	0.4 mg/mL pepsin, 10 min, 37 °C — the pre-v3.0 condition

The ladder deliberately walks upward from the current condition: at 10 min we sit at roughly a third of the exposure the published protocols use, so under-digestion is the failure mode to look for, and chambers 3–5 are the ones most likely to win.

Chamber 6 is the pre-v3.0 condition at 4× the exposure of chamber 2, kept as the reference point. Because it differs from chamber 2 only in concentration, a 2-versus-6 difference is a clean dose result; a 2-versus-4 difference (same total mass × time, different concentration and time) would be a uniformity result.

The acid was decided, but keep it on the list if a sample regresses

All pepsin solutions moved to 0.1 M HCl in Pepsin Solution Preparation v3.0, matching the padlock/RCA literature. Both 0.01 M and 0.1 M have been run here historically, and there is no published head-to-head of the two for in-situ work — the P6887 datasheet's optima are substrate-dependent and do not settle it. So if a sample type that used to work regresses after that change, add 0.4 mg/mL pepsin in 0.01 M HCl as a seventh chamber before concluding anything else.

2c. Fix the rest

Across the whole ladder, hold constant and record: pepsin catalogue number and lot, stock preparation date, time since dilution, acid molarity, and the measured bath temperature rather than the setpoint. Section thickness must be consistent — a published time course had to discard two points as outliers on thickness alone.¹

3. In-process check: stop on morphology, not on the clock

Digestion adequacy is visible under an ordinary light microscope while it is happening, which is the only chance to catch it before the readout. Criteria from a study that validated them on 400 samples, breast cancer and lymphoma.²

State	Under the light microscope	Downstream
Under-digested	unclear nuclei, glassy perinuclear space	high background, uninterpretable
Adequate	clear bare nuclei, intact nuclear membrane, translucent perinuclear space	bright signal, low background
Over-digested	disrupted, unclear nuclear envelope; soft bare nuclei; enlarged perinuclear space	weak signal, damaged morphology

• Check a chamber at the microscope before quenching the digestion.
• If a section is already at the over-digested appearance, record it and continue — that chamber is the upper bound of the ladder and is informative.
• The general operating rule from the FISH literature is the strongest digestion that does not cause the section to lift; the morphology criteria above make that judgeable before the section is lost.

4. Read-out metrics

Consult in this order. The first two disagree characteristically at the over-digested end, and that disagreement is the signal.

Metric	How	What it tells you
Rolony density — per 100 µm² and per nucleus	spot counts from the standard detection pipeline, normalized to segmented area and to DAPI nuclei	Primary yield curve: rises, peaks, falls. Report both — per-nucleus inflates once digestion has enlarged apparent cell size
Nuclei per mm², nuclear morphology	DAPI segmentation count; visual check for ghosting and loss of boundary	Tissue integrity. Falls monotonically and turns before yield does — earliest over-digestion alarm
Negative-probe rate	non-targeting padlock rolonies ÷ total	Non-specific deposition; rises past the optimum. Also the check that a high count is real
Section retention	brightfield image before digestion and after the last wash; compare area	Detachment and loss; explains outliers that otherwise look like chemistry failures

Decision rule — marginal gain, not absolute maximum

Do not simply take the chamber with the highest count. Compute the geometric mean of per-gene density fold-change for each step of the ladder, and stop at the step where it drops to ≈ 1. Beyond that you are paying tissue integrity for nothing. On a 3-gene control panel this is a three-number calculation, and it gave a consistent answer across three human samples where per-cell and per-area metrics disagreed.¹

For a tumour panel, compute every metric inside the annotated compartment the panel is for — tumour nests, or immune infiltrate — not over the whole section. Stroma, adipose and necrosis otherwise dominate the average and pull the apparent optimum away from the compartment you care about.

5. Recording the result

Write the chosen condition and the ladder it came from into the run's lab record, including the metric values for every chamber — not only the winner. A ladder whose losing arms are unrecorded cannot be re-analysed when the next tissue behaves differently.

References

Changelog

• v0.3 — 2026-08-18 — FFPE ladder recentred on the new default (0.1 mg/mL / 10 min, pretreatment-ffpe v3.0): the arms walk time upward at a fixed 0.1 mg/mL, since the current condition sits below the published range, with the old 0.4 mg/mL / 10 min kept as the dose reference. Still not run.

• v0.2 — 2026-08-18 — Ladder run-out step follows RCA Protocol v6.0: blocking is no longer part of the shared workflow, so it is no longer listed among the steps run across all chambers together. MINOR — the design is unchanged.

• v0.1 — 2026-08-18 — First draft. Assembled from the Visium/Xenium tissue-optimization workflow, the Teng morphology criteria, and the permeabilization evidence review in the internal evidence review Permeabilization / Pepsin Digestion (lab site → Research). Consolidates the ladder that was previously duplicated inline in pretreatment-ff.md and in that note. Not yet run.

---

1. Roberts K, Bassett AR. Optimisation of Xenium automated in situ sequencing for PAXgene-fixed tissue samples. bioRxiv 2025.02.11.637091. DOI: 10.1101/2025.02.11.637091. ↩↩↩↩

2. Teng X, Zhang S, Liu W, Bi K, Zhang L. A new method for real-time evaluation of pepsin digestion of paraffin-embedded tissue sections, prior to fluorescence in situ hybridisation. Virchows Arch 2017;470(5):567–573. DOI: 10.1007/s00428-017-2097-z. ↩

3. 10x Genomics. CG000238 — Visium Spatial Gene Expression Reagent Kits: Tissue Optimization User Guide, Rev E. ↩

4. Sountoulidis A, et al. SCRINSHOT enables spatial mapping of cell states in tissue sections with single-cell resolution. PLoS Biol 2020;18(11):e3000675. DOI: 10.1371/journal.pbio.3000675. ↩

5. Gyllborg D, et al. Hybridization-based in situ sequencing (HybISS) for spatially resolved transcriptomics in human and mouse brain tissue. Nucleic Acids Res 2020;48(19):e112. DOI: 10.1093/nar/gkaa792. ↩

Comments

Ask / Feedback