RCA Protocol: Direct RNA Strategy

This is the direct RNA library construction protocol shared by all sample types (FF, FFPE) and readout technologies (PRISM, SPRINTseq). Padlock probes hybridize directly to mRNA, and SplintR ligase ligates the DNA nick on the RNA template. Start here after completing sample pre-treatment (Fresh Frozen or FFPE).

cDNA strategy

For the cDNA ligation strategy (reverse transcription before hybridization, with T4 DNA Ligase for SNP specificity), see RCA Protocol: cDNA.

• Applicable samples: All sample types (FF, FFPE); all readouts (PRISM, SPRINTseq)
• Version: v7.0
• Last modified: 2026-08-18

Materials

Ordering: Bill of Materials → RCA (Library Construction). Prepared ahead of time: PDMS Chambers (forms the 100 μl on-slide reaction well) · Pepsin Solution (pre-treatment). Strategy choice: Library Construction Strategies. Buffer chemistry: Reaction Buffers & Stringency.

1. Hybridization (~2.5 h, or overnight)

Goal: Probe-specific binding.

There is no separate blocking step — the blockers are in this mix

Up to v5.7 this protocol opened with a 30 min room-temperature "blocking" incubation, inherited from SCRINSHOT.¹ It was deleted in v6.0: its composition was identical to the hybridization mix below apart from an oligo-dT that the 55 °C denaturation melts off before the probes ever anneal. The blockers that do the work — recombinant albumin and tRNA — are in the mix that carries the probe, which is where the only controlled evidence puts them.⁵ See the Changelog for the full reasoning.

Probe concentration and hybridization time -- two working points

Preferred: 10 nM per probe, overnight (~16 h) at 45°C. The low probe concentration substantially reduces nonspecific probe deposition, and the long incubation recovers on-target yield. Use this whenever the schedule allows.

Same-day: 50 nM per probe, 120 min at 45°C. Use when the run has to finish in one day; expect more nonspecific background than the overnight condition.

Probe stock concentration and pooled-panel volume are panel-dependent -- compute the volume that reaches the target per-probe final concentration and fill H₂O to 100 μl total.

Reagent	Stock	Final	100 μl
RNase Free H₂O			to 100 μl
Ampligase Buffer	10X	1X	10 μl
KCl	1 M	0.05 M	5 μl
Formamide	100%	20%	20 μl
Probes	(panel-dependent)	10 nM (overnight) / 50 nM (2 h)	(variable)
Recombinant Albumin	20 μg/μl	0.2 μg/μl	1 μl
tRNA	10 μg/μl	0.2 μg/μl	2 μl
RiboLock	40 U/μl	1 U/μl	2.5 μl

• Incubate: 55°C (15--20 min denaturation) → 45°C (~16 h overnight at 10 nM, or 120 min at 50 nM). Seal to prevent evaporation -- for the overnight condition use a humid chamber and check the seal before walking away.
• Stringent Wash: 10% formamide in 2X SSC, 3 x 10 min, at room temperature. Washing buffer: 100 µL 20X SSC + 100 µL formamide + 800 µL NFW.
• Wash: PBS-Tween 0.05% wash 3 x 2 min. Two independent reasons for the third change, and the salt one is the stronger: 10% formamide is ~2.5 M, and 2X SSC is 390 mM Na⁺ while SplintR is strongly inhibited above 100 mM NaCl. See Reaction Buffers & Stringency.

Pin the temperature — it is worth more than the formamide

Room temperature, not 45 °C. The two are ~20 °C apart in effective stringency, against ~5 °C for the 10%-vs-20% formamide difference. A page that offers both is not specifying a condition.

This matters here more than anywhere else in the protocol because the wash lands while the padlock is still linear. Once ligated it is a closed circle catenated around the target and washing is nearly free — but at this point base pairing is the only thing holding it on.

20% formamide is the high-stringency arm, not the default

Available as a deliberate comparison arm — washing buffer: 100 µL 20X SSC + 200 µL formamide + 700 µL NFW. Record it in the run notes if used: yield and background both move, so data from the two conditions are not directly comparable.

Raise stringency only against a background you have shown to be sequence-based. The nonspecific deposit characterized in-house is not — it survives 20% formamide and is removed by SDS, so it is adsorptive, and more formamide costs signal without touching it.

2. Ligation (~2 h)

Goal: Ligation of hybridized padlock probes into circles; SplintR ligase (PBCV-1 DNA Ligase) works on DNA:RNA hybrids.

Reagent	Stock	Final	100 μl
RNase Free H₂O			76.5 μl
SplintR Buffer	10X	1X	10 μl
Recombinant Albumin	20 μg/μl	0.2 μg/μl	1 μl
SplintR Ligase	25 U/μl	2.5 U/μl	10 μl
RiboLock	40 U/μl	1 U/μl	2.5 μl

• Incubate: 37°C for 2 hours (sealed).
• Wash: PBS-Tween 0.05% wash 2 x 2 min.

3. Rolling Circle Amplification (~12--16 h)

Goal: Amplify circularized probes into DNA nanoballs (rolonies).

Reagent	Stock	Final	100 μl
RNase Free H₂O			70.5 μl
Phi29 Buffer	10X	1X	10 μl
Glycerol	100%	~10%	10 μl
dNTPs	10 mM	0.25 mM	2.5 μl
Amino-dUTP	2 mM	50 μM	2.5 μl
Recombinant Albumin	20 μg/μl	0.2 μg/μl	1 μl
RCA Primer	10 μM	0.1 μM	1 μl
Phi29 Polymerase	10 U/μl	0.25 U/μl	2.5 μl

• Incubate: 30°C for 12--16 h overnight (sealed / humid chamber).
• Wash: PBS-Tween 0.05% wash 2 x 2 min.

Glycerol -- reaction speed and rolony uniformity

Glycerol raises the viscosity and lowers the water activity of the mix, which slows phi29 elongation and restricts how far the growing concatemer can spread. In practice this keeps rolonies compact and more even in size, which is what the readout needs: signal calling thresholds and counts individual rolonies, so oversized products that sprawl and merge with their neighbours are lost to segmentation. Glycerol also helps stabilize phi29 across the long incubation.

If rolonies come out too small or too dim, lower the glycerol (10% → ~5%) or leave it out entirely. The trade is a wider size distribution.

Fresh frozen commonly runs ~5% glycerol

FF sections have historically used ~5% glycerol (5 μl) rather than 10%, taking the extra 5 μl as water (70.5 → 75.5 μl). FF tissue gives more circles than FFPE, so the extra confinement is less necessary and the larger rolonies are welcome. Either value is a valid starting point — judge it on rolony size in the first run.

Nucleotide load is sample-dependent

Some tissues give under-sized rolonies at the standard nucleotide load; TNBC is the known case in-house. For these, double both nucleotide components -- dNTPs to 0.5 mM (5 μl) and amino-dUTP to 100 μM (5 μl) -- and take the extra 5 μl out of the water (70.5 → 65.5 μl). Scale the two together so the amino-dUTP : dTTP ratio, and therefore the amine density available for labelling, stays constant.

No RNase inhibitor in the RCA mix

RiboLock is not used at this step. By the time RCA starts, the template is the ligated DNA circle -- the RNA has already served its purpose in hybridization and ligation -- so an RNase inhibitor adds nothing.

4. Post-amplification Fix & Strip

Goal: Fix RCA products and strip unbound probes.

• Fix: Add 4% PFA, mix by gentle pipetting (3--5 times), incubate 30 min at RT.
• Wash: PBS-Tween 0.05% wash 3 x 2 min.
• Strip: 65% formamide wash 3 x 2--10 min at 30°C (e.g. on a PCR block or Eppendorf Thermostat). Washing buffer: 650 µL formamide + 350 µL NFW.
• Wash: PBS-Tween 0.05% wash 2 x 2 min.

---

Next step: Signal readout -- PRISM Imaging or SPRINTseq Sequencing

Changelog

• v7.0 -- 2026-08-18 -- The stringent wash returns to 10% formamide in 2X SSC at room temperature, and the wash temperature is now pinned. MAJOR: detection efficiency and background both move, so data either side of this version are not directly comparable.

  v5.4 (2026-07-29) had raised the default from 10% to 20% and, in the same commit, added a "45 °C optimal, or RT" clause. Neither half carried a recorded reason — the change was filed only as "current bench practice". Reviewing it turned up three things, in ascending order of weight:

  • Every run this lab has published data from used 10%. The wash appears verbatim in eight lab records from 2023-06 to 2024-12 as "washing buffer (10% formamide in 2X SSC)" with no temperature stated, i.e. RT. The 20% default was four weeks old and covered no completed dataset.
  • Our hybridization mix is SCRINSHOT's, component for component — same Ampligase buffer, 0.05 M KCl, 20% formamide, 0.2 μg/μl albumin, 0.2 μg/μl tRNA, 1 U/μl RiboLock, same 55 → 45 °C ramp.¹ The wash was the one place we departed from the source protocol, and we departed in the harsher direction: SCRINSHOT washes at 10%, as do HybISS and the Nilsson open-source direct-RNA ISS.³ Xenium washes 13 °C below its hybridization temperature.⁴ Every published method washes below its own hybridization stringency; at 20% and 45 °C we were washing slightly above ours — 2X SSC carries 390 mM Na⁺ and no Mg²⁺ against ~454 mM Na⁺-equivalent in the hybridization.
  • A wash is not a second hybridization. During the anneal there is 10 nM probe in solution and a duplex that breathes open re-anneals; the moment the mix is replaced the probe concentration is zero and the same ΔG runs one way for 30 min. Applied at 45 °C, to a padlock that is still linear because ligation has not happened yet, this preferentially removes probes engaged over part of their 40 nt — which in fixed tissue is the normal case, and which SplintR would still have ligated.

  The temperature clause was the larger error. 10% versus 20% formamide is worth ~5 °C on a DNA:RNA duplex; RT versus 45 °C is worth ~20 °C. The two options together spanned roughly 25 °C of effective stringency inside one SOP, so two operators following this page were running different experiments.

  20% is retained as an explicitly labelled high-stringency arm, not a fallback, with the instruction to record its use. It is worth reaching for only against a background shown to be sequence-based — and the deposit characterized in-house is not, since it survives 20% formamide and is removed by SDS.

  Also corrected: the PBS-T rationale, which cited "20% formamide is ~5 M". At 10% that is ~2.5 M, and the stronger reason for the third change is salt — 2X SSC is 390 mM Na⁺ and SplintR is strongly inhibited above 100 mM NaCl.⁷

  Applied across the family in the same revision: RCA: cDNA v3.0, lab records 20260129_rca-ff-ilock v6.0 and 20260702_rca-cdna-snp v0.4. Full evidence: the internal evidence review Stringency, Nonspecific Deposition and FFPE Background (lab site → Research).

• v6.0 -- 2026-08-18 -- The 30 min Blocking step is deleted, and with it the oligo-dT. Sections renumbered 1--4. MAJOR: no reagent that touches the reaction changed, but every step number moved, and step numbers are how this protocol is called out at the bench.

  The step was a pre-hybridization incubation in the hybridization buffer plus 0.1 μM oligo-dT30VN, copied component-for-component from SCRINSHOT,¹ which introduced it with a one-sentence rationale and no supporting experiment. Three findings retired it:

  • It was the hybridization mix minus probes. Apart from the oligo-dT, every component was already in Section 1 at the same concentration — buffer, 50 mM KCl, 20% formamide, 0.2 μg/μl albumin, 0.2 μg/μl tRNA, 1 U/μl RiboLock. All it bought was a 30 min head start for blockers that then sit in the hybridization for 2--16 h at 5--25x the total probe mass.
  • The oligo-dT could not act where it was placed. It was applied at RT, then replaced by a hybridization mix containing no oligo-dT and taken to 55 °C for 15--20 min. An all-A:T 30-mer RNA:DNA heteroduplex in 20% formamide (Tm depressed ~0.6--0.7 °C per percent formamide⁶) does not survive that step, and nothing remains in solution to re-anneal at 45 °C. The poly(A) block was gone before the padlock probes began competing for sites.
  • The field does not run one. BOLORAMIS² and the Nilsson lab's open-source ISS³ run the same direct-RNA SplintR chemistry with no blocker in the hybridization at all; Xenium⁴ has no blocking step in its workflow. The one controlled in-situ result on blocking oligos adds them to the pre-hybridization and hybridization together, which supports blocker in the probe-carrying mix — where ours already is — not a separate pre-incubation.⁵

  The oligo-dT is dropped outright, not relocated into the hybridization mix: an oligo-dT held on the poly(A) tail can occlude padlocks targeting the extreme 3′ end, which is the likely reason SCRINSHOT omitted it from its own hybridization, and no panel in use here carries the T-rich sequence a poly(A) sink would compete for.

  What is unchanged: recombinant albumin, tRNA and RiboLock stay in the hybridization mix; the stringent formamide wash and the preferred 10 nM overnight hybridization — the two levers that actually control nonspecific deposition — are untouched. The FFPE RNase H pointer that lived in the old Section 1 is gone; see RCA: FFPE v3.0 for where RNase H now belongs.

• v5.7 -- 2026-08-18 -- Added a pointer in Blocking to the FFPE-specific RNase H addition. No change to this protocol's own steps: FFPE previously carried its own full copy of the workflow in RCA: FFPE, which has been reduced to the delta list, so the one recipe difference needs to be visible from here.

• v5.6 -- 2026-08-18 -- Post-stringency wash tightened from "2--3 x 2 min" to a firm 3 x 2 min. MINOR: same method, narrower spec. 20% formamide is ~5 M, and two changes of a ~100 μl chamber leave low-mM formamide at the ligase, which costs activity. Everywhere else the 2 x 2 min default stands — the reasoning, and the three cases that genuinely warrant a third change, are in Reaction Buffers & Stringency. Also: an Applicable samples line was added to the metadata block so this page matches the house format (see docs/CONTRIBUTING.md).

• v5.5 -- 2026-08-10 -- Post-amplification fix (Section 5) pinned to 30 min, from the previous 15--30 min. The range was a real ambiguity rather than a tolerance: formaldehyde's methylene-bridge crosslinking takes hours to complete, so 15 and 30 min both sit on the rising part of that curve and anchor measurably differently. This fix has to hold the rolonies through the 65% formamide strip that immediately follows, so err long. Applied across the whole RCA family -- see rca-cdna v2.0, which carries the same change plus the reverse-transcription and post-RT fixation corrections.

• v5.4 -- 2026-07-29 -- Hybridization: 10 nM per probe overnight is now the preferred condition (lower nonspecific background), with 50 nM / 2 h kept as the same-day option. Stringent wash: 20% formamide in 2X SSC is the new default, 10% retained as the milder fallback. RCA: primer lowered 0.3 → 0.1 μM, RiboLock removed (the template is DNA by this point), glycerol rationale documented with guidance for undersized rolonies, and a sample-dependent option to double dNTPs + amino-dUTP (e.g. TNBC) added. All PBS-T washes standardised to 2 min per change (previously 1 min). Corrected the RCA table, which previously summed to 97.5 μl instead of 100 μl. Added a Materials cross-reference block and linked the new Reaction Buffers & Stringency background page.
• v5.3 -- 2026-01-29 -- Direct-RNA library construction consolidated into one protocol shared across sample types and readouts.

References

---

1. Sountoulidis A, Liontos A, Nguyen HP, Firsova AB, Fysikopoulos A, Qian X, Seeger W, Sundström E, Nilsson M, Samakovlis C. 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. The source of the blocking step this protocol carried until v5.7 — same padlock / SplintR / phi29 chemistry, 30 min room-temperature preconditioning in the hybridization mixture with 0.1 μM Oligo-dT30VN "to block unspecific binding of DNA", omitted again from the padlock hybridization. No block/no-block comparison is reported. ↩↩↩

2. Liu S, Punthambaker S, Iyer EPR, Ferrante T, Goodwin D, Fürth D, et al. (Church GM). Barcoded oligonucleotides ligated on RNA amplified for multiplexed and parallel in situ analyses. Nucleic Acids Res 2021;49(10):e58. DOI: 10.1093/nar/gkab120. Direct-RNA padlock + SplintR + phi29 with a hybridization buffer of 10% formamide in 6X SSC and nothing else — no blocker, no carrier, no pre-hybridization step. ↩

3. Lee H, Langseth CM, Marco Salas S, Sariyar S, Metousis A, Rueda-Alaña E, et al. (Nilsson M). Open-source, high-throughput targeted in situ transcriptomics for developmental and tissue biology. Development 2024;151(16):dev202448. DOI: 10.1242/dev.202448. Padlock probe solution is 2X SSC + 10% formamide + 10 nM of each probe; no blocking agents anywhere in the hybridization. Also the wash reference for v7.0: "we washed the unhybridized excess probes by two washes of 10% formamide in 2× SSC, followed by two washes in 2× SSC", after an overnight 37 °C hybridization — the mildest condition in the direct-RNA comparison. ↩↩

4. 10x Genomics. CG000582 — Xenium In Situ Gene Expression: Probe Hybridization, Ligation & Amplification, Rev H. PDF. Workflow is Probe Hybridization (16--24 h) → Post Hybridization Wash → Ligation → Amplification → Autofluorescence Quenching. There is no blocking step; background is controlled by low probe concentration, long hybridization and a stringent wash. ↩↩

5. Kuboe Y, Nakanishi K, Tazawa I, Nakajima K. Universal improvement of in situ hybridization chain reaction by reducing background signals caused by single probes. Dev Growth Differ 2025;67(6):336--343. DOI: 10.1111/dgd.70018. Random oligonucleotides added to both the pre-hybridization and the hybridization reduced in-situ HCR background ~3--90x — the closest published controlled test of blocking oligos in situ, and one that does not separate the two placements. ↩↩

6. Blake RD, Delcourt SG. Thermodynamic effects of formamide on DNA stability. Nucleic Acids Res 1996;24(11):2095--103. DOI: 10.1093/nar/24.11.2095. Basis for the linear Tm depression by formamide. The two duplex classes differ and the distinction matters here: 0.6--0.72 °C per percent is a DNA:DNA figure, while at hybridization-relevant formamide concentrations the stability ordering is RNA:RNA > RNA:DNA > DNA:DNA (Casey J, Davidson N, Nucleic Acids Res 1977;4(5):1539--52), so a DNA-probe:mRNA duplex loses less per percent. The direct-RNA route therefore uses the conservative in-house 0.5 °C/% — which is where the "~5 °C for 10% formamide" in the v7.0 entry comes from. See Reaction Buffers & Stringency. ↩

7. Lohman GJS, Zhang Y, Zhelkovsky AM, Cantor EJ, Evans TC Jr. Efficient DNA ligation in DNA–RNA hybrid helices by Chlorella virus DNA ligase. Nucleic Acids Res 2014;42(3):1831--1844. DOI: 10.1093/nar/gkt1032. The primary characterisation of SplintR, cited here for the salt limit: ligation is "strongly inhibited by concentrations of NaCl > 100 mM", with the inhibited fraction diverted to dead-end 5′-adenylylated probe. 2X SSC carries 390 mM Na⁺, which is why the PBS-T chase before ligation is 3 x rather than 2 x. The same paper is the reason the 37 °C / 2 h ligation needs no change — rate rises monotonically up to 37 °C. ↩

Comments

Ask / Feedback