Flores Bueso, Y., Walker, S. P., & Tangney, M. (2020). Characterization of FFPE-induced bacterial DNA damage and development of a repair method. Biology Methods and Protocols, 5(1), bpaa015.

Flores Bueso et al. (2020) — FFPE-Induced Bacterial DNA Damage and Repair

Bibliographic Reference

Flores Bueso, Y., Walker, S. P., & Tangney, M. (2020). Characterization of FFPE-induced bacterial DNA damage and development of a repair method. Biology Methods and Protocols, 5(1), bpaa015. DOI: 10.1093/biomethods/bpaa015.

Core Argument

FFPE processing damages bacterial DNA in the same ways it damages human DNA — fragmentation, crosslinking, and sequence artifacts — but the dominant source of FFPE sequence artifacts is oxidative damage, not cytosine deamination. This finding challenges the deamination-centric narrative of the FFPE literature (e.g., Do & Dobrovic 2015) and has direct implications for microbiome studies from FFPE archives, where DNA template is often minimal and the sequences studied span multiple genomes. The authors characterize the damage spectrum in bacterial FFPE DNA and develop two repair strategies — (1) an optimized low-temperature de-crosslinking procedure that reduces heat-induced artifacts while maintaining DNA yield, and (2) an in vitro reconstitution of the base excision repair (BER) pathway that repairs oxidative lesions and abasic sites enzymatically. Both strategies significantly increased fragment length, reduced sequence artifacts, and improved sequencing readability for bacterial AND mammalian FFPE DNA.

Methods

• Model system: E. coli K-12 MG1655 genomic DNA subjected to standard FFPE processing (formalin fixation, paraffin embedding, 24h storage)
• Damage characterization: Fragment size analysis (Bioanalyzer), PCR amplification efficiency (qPCR with varying amplicon sizes), sequencing (WGS on Illumina MiSeq), crosslink quantification
• Repair Strategy 1 — Optimized de-crosslinking: Comparison of incubation temperatures (90°C standard vs. lower temperatures) with time-course analysis; measurement of DNA yield, fragment length, and sequence artifact frequency
• Repair Strategy 2 — In vitro BER reconstitution: Sequential treatment with DNA glycosylases (removing oxidized bases), AP endonuclease (cleaving abasic sites), DNA polymerase I (gap-filling), and DNA ligase (sealing nicks)
• Validation: WGS comparison of untreated vs. repaired FFPE DNA; applied to both bacterial (E. coli) and mammalian (mouse liver) FFPE specimens
• Metagenomic application: Demonstration on mock bacterial communities

Key Findings

• Oxidative damage dominates FFPE sequence artifacts. Contrary to the prevailing focus on cytosine deamination (C→T artifacts), the authors found that oxidative base modifications — producing abasic sites and a broader spectrum of base substitutions — were the predominant source of sequence artifacts in FFPE bacterial DNA. This is consistent with the “A-rule” biochemistry (Dahlmann et al., 2009; dahlmann2009-biochemical-dna-damage): polymerases insert adenine opposite abasic sites, producing characteristic mutation patterns distinct from the C→T transitions of deamination. The oxidative-damage dominance likely reflects the fact that formalin is readily oxidized to formic acid, creating an oxidizing environment during fixation and storage.

• FFPE bacterial DNA is highly fragmented and a poor PCR template. Fragment sizes were substantially reduced compared to fresh DNA, with corresponding decreases in PCR amplification efficiency — particularly for longer amplicons. Crosslinking further reduced the effective template count. These effects are quantitatively similar to those observed in mammalian FFPE DNA, confirming that FFPE damage is chemistry-driven, not genome-specific.

• Low-temperature de-crosslinking reduces heat-induced artifacts. The standard 90°C/1h de-crosslinking protocol introduces single-strand breaks and chimeric sequences as a side effect of thermal damage. A lower incubation temperature (combined with extended duration) achieved comparable de-crosslinking yields with significantly fewer heat-induced artifacts. This is a direct, immediately implementable protocol improvement for any lab working with FFPE DNA.

• In vitro BER reconstitution repairs oxidative lesions and improves sequencing. The authors reconstituted the core BER pathway in vitro: DNA glycosylases remove oxidized bases → AP endonuclease cleaves abasic sites → DNA polymerase I fills gaps → DNA ligase seals nicks. Treatment with this BER cocktail significantly increased DNA fragment length, reduced sequence artifacts, and improved whole-genome sequencing quality metrics (read mappability, coverage uniformity, SNP concordance with reference). Critically, BER treatment was effective on both bacterial and mammalian FFPE DNA — the repair biochemistry is universal.

• UDG alone is insufficient. The authors explicitly note that commercial repair kits (e.g., NEB FFPE Repair Mix) and UDG pretreatment address only a subset of FFPE lesions — primarily uracil-derived C→T artifacts. Oxidative lesions, abasic sites, and strand breaks require the full BER pathway for repair. This provides empirical support for the Steiert et al. (2023) observation that the FFPE artifact spectrum extends well beyond C→T, and that UDG-only approaches leave the majority of damage types unaddressed.

Concepts Introduced or Used

FFPE bacterial DNA damage, oxidative DNA damage, base excision repair (BER), de-crosslinking optimization, DNA glycosylase, AP endonuclease, DNA polymerase I, DNA ligase, heat-induced artifacts, chimeric sequences, microbiome from FFPE, metagenomics

Entities Referenced

• E. coli K-12 MG1655 — model system for FFPE damage characterization
• NEB FFPE Repair Mix — commercial kit; addresses only partial damage spectrum
• UDG (uracil-DNA glycosylase) — addresses C→T deamination artifacts only
• DNA glycosylases (Fpg, Endo III, Endo VIII) — remove oxidized bases in BER
• MiSeq (Illumina) — sequencing platform for WGS validation

Limitations (as stated by authors)

• Bacterial model (E. coli) may not fully represent the complex mixed bacterial communities in clinical FFPE microbiome samples
• FFPE processing was performed under controlled laboratory conditions (24h storage); archival clinical specimens with years-to-decades of storage may show different damage profiles
• The BER reconstitution uses E. coli enzymes; mammalian BER enzymes may produce different repair outcomes at specific lesion types
• Metagenomic application was demonstrated on a simple mock community, not complex clinical samples
• The repair methods improve but do not eliminate FFPE artifacts — residual damage remains even after optimized de-crosslinking + BER treatment

Relevance to Clonal Evolution

This paper provides a critical empirical correction to the deamination-centric FFPE narrative and a practical repair methodology applicable to FFPE DNA from any source. Its relevance to clonal evolution is threefold:

1. Oxidative damage, not deamination, is the dominant FFPE artifact source. The wiki’s FFPE knowledge layer — built from Do & Dobrovic (C→T deamination as the primary mechanism), Greytak (scope), and Steiert (complete damage spectrum) — needs to incorporate this finding: oxidative base modifications producing abasic sites are the numerically dominant lesion type in FFPE DNA. UDG pretreatment, while valuable, addresses only the minority (deamination-derived) fraction. This explains why Steiert et al. observed non-C→T artifacts at frequencies comparable to C→T: they are the majority damage type, not a secondary contributor.

2. The BER repair strategy is a general solution for FFPE DNA. Unlike UDG (lesion-specific), the BER reconstitution approach repairs the full spectrum of oxidative and abasic lesions. For clonal evolution studies using FFPE archives, BER pretreatment before library preparation could substantially reduce the three-confounder problem identified in the FFPE series synthesis: true subclonal mutations + CNA-induced miscalibration + FFPE artifacts. By reducing the artifact component, BER pretreatment shrinks the SMF floor — the minimum measurable subclonal mutation fraction that can be distinguished from fixation artifact.

3. The principle extends to human FFPE DNA. The authors demonstrated BER repair efficacy on both bacterial and mammalian FFPE specimens. For human cancer FFPE archives — the source material for most clonal evolution studies with clinical outcome data — BER pretreatment represents a practical, enzymatically principled approach to improving variant-calling accuracy. This is not yet standard practice in clinical NGS pipelines (which typically use UDG alone or no repair), representing a translational gap between what is biochemically possible and what is operationally implemented.

Integration with the FFPE series:

Paper	Artifact mechanism emphasized	Countermeasure
Do & Dobrovic 2015	C→T deamination	UDG, molecular tagging
Buesco et al. 2020	Oxidative damage (dominant)	BER reconstitution + optimized de-crosslinking
Steiert et al. 2023	Full damage spectrum	ERROR-FFPE-DNA checklist; notes UDG insufficiency

The series now documents a clear progression: deamination is the best-characterized artifact type but NOT the most prevalent — oxidative damage is. The repair strategy needs to match the damage spectrum: UDG alone is insufficient; BER reconstitution (or commercial kits implementing the full BER pathway) is the biochemically appropriate response.