Arabidopsis Single-Cell Response to Fungal Infection

Bottom line: this project reviews the GO annotations of the Arabidopsis
genes that a single-cell atlas of fungal infection singled out
(PMID:37741284, Tang et al. 2023,
Cell Host & Microbe 31:1732). The paper profiled 95,040 leaf cells infected
with the hemibiotrophic fungus Colletotrichum higginsianum. The first batch
covers six genes: MYB122, the one gene the paper tested genetically, plus its
indole-glucosinolate regulator paralogs MYB51 and MYB34, the camalexin
cytochromes CYP71A12 and CYP71A13, and the immune marker FRK1. Results are in
the review status table. All six reviews are complete: 96
existing annotations were assessed, with 49 accepted, 20 kept as non-core, 8
modified, 7 marked over-annotated, 11 removed and 1 left undecided, plus 7 new
annotations proposed.

We picked this paper because it separates two kinds of evidence that GO
annotation usually blurs together. Most of its findings say where a gene is
expressed: which cell type, and how close the cell is to the fungus. Expression
like that supports at most a response to term, not a function. Only the
myb122 mutant phenotype is functional evidence. The review question for each
gene is whether its existing annotations hold up, and whether the paper adds
anything that meets the bar for a new annotation.

Project start date: 2026-10-02
Organism: Arabidopsis thaliana (ARATH)
Source paper: Tang B, Feng L, Hulin MT, Ding P, Ma W (2023). Cell-type-specific
responses to fungal infection in plants revealed by single-cell transcriptomics.
Cell Host Microbe 31:1732–1747. doi:10.1016/j.chom.2023.08.019 (open access,
CC BY-NC-ND). The cached copy (publications/PMID_37741284.md) holds only the
abstract, so review quotes from this paper are limited to the abstract.

What the paper found

Curation approach

• Cell-type expression is not function. That a gene is induced in procambium or
  in guard cells near the fungus is a fact about the gene's regulation. GO has no
  "expressed in cell type" relation, so these findings do not become annotations.
  At most they support a response to fungus row with IEP evidence, and only when
  that row is not already there.
• MYB122 is the only gene with new functional evidence from this paper. Even
  there, the phenotype shows that MYB122 is needed for resistance. It does not show
  which step of defense MYB122 carries out. Glucosinolate regulation is the likely
  route, but the paper did not measure glucosinolates in the mutant. Any new process
  term has to meet the NEW bar in CLAUDE.md.
• FRK1 is a marker. It is the standard reporter gene for PTI, and its own
  biochemical role is barely characterized. Being induced by flg22 is not evidence
  that it takes part in defense.

Review status

The first batch covers the genes on the paper's functional thread (MYB122) and
its glucosinolate and camalexin context. Action counts are taken from the review
files.

Genes already reviewed in other projects

SUR1 (core glucosinolate C-S lyase), ABI1 (negative regulator of ABA
signaling), and EDS1 and PAD4 (TNL and RNL signaling hubs) already have
reviews in the repo.

Candidate backlog

These genes are named in the paper but not yet reviewed. UniProt accessions were
taken from UniProt's reviewed entries (October 2026). The four helper-NLR paralogs
(ADR1-L1, ADR1-L2, NRG1.1, NRG1.2) do not resolve by name in UniProt and need
their accessions looked up before they are fetched.

The NLRs are the natural next batch. Many of them carry EC 3.2.2.6 (NAD+
hydrolase) in UniProt, so the reviews should check that TIR-domain NADase
activity is annotated consistently across the TNLs.

Findings

• Regulators were annotated as enzymes. MYB34 and MYB51 carried GO:0009759
  indole glucosinolate biosynthetic process. Both were changed to GO:0010439
  regulation of glucosinolate biosynthetic process, which MYB122 also received as a
  new annotation, because the transcription factors catalyse no step of the pathway.
  GO has no positive or indole-specific child of GO:0010439; the MYB34 review
  proposes one.
• MYB122's fungal-defense annotation already exists. GOA carries
  GO:0050832 defense response to fungus for MYB122 from this paper. It was
  accepted, with a note that the evidence is a mutant phenotype (IMP) rather than
  IDA. The paper's key-resources table lists both myb122 lines under the wrong
  locus (AT3G25800, not At1g74080) and calls one a CRISPR line while the text says
  T-DNA. These look clerical and are recorded on the reference.
• The three MYBs divide by cell type and by signal. Several independent
  studies agree that MYB34, unlike MYB51 and MYB122, is not induced by pathogens or
  wounding, and Tang et al. found the same at infection sites. MYB34's broad
  response to other organism row is still kept as non-core, because it is an
  ancestor of the defense response to insect term MYB34 retains.
• The CYP71A12 and CYP71A13 paralogs were kept distinct. CYP71A12 gained its
  enzyme activity (GO:0047720, IDA) and defense response to fungus (IMP), as
  the main enzyme for pathogen-induced indole-3-carboxylic acid derivatives;
  CYP71A13 remains the main leaf camalexin enzyme. Both carry an ER-lumen IDA from
  PMID:33831160. PubMed's redirection notice shows that this is a deleted duplicate of
  Mucha et al. 2019 (PMID:31511315). The evidence there is confocal colocalization
  with the ER lumenal marker RFP-HDEL, which places the enzymes in the ER but cannot
  tell lumen from membrane, and the paper itself describes these P450s as
  membrane-anchored with the catalytic domain facing the cytosol. Both rows were
  changed to GO:0005789 endoplasmic reticulum membrane. The same TAIR curation also
  put ER lumen on CYP71B15 (PAD3), which is not reviewed here.
• FRK1 is a marker, not a demonstrated defense component. The expression-only
  defense response to bacterium row was changed to response to bacterium. The
  TAIR IDA row for the same term (He et al. 2006) was left undecided, because that
  paper's full text is not available and the curator may have seen an FRK1-specific
  experiment. One flg22-only row was changed to response to molecule of bacterial origin. A motif
  scan shows FRK1 has an intact kinase domain, but no activity, substrate or mutant
  phenotype has been published.
• Induced systemic resistance does not fit its source. MYB51, CYP71A12 and
  CYP71A13 carry GO:0009682 from van de Mortel et al. 2012 (IMP), and the two
  cytochromes also inherit it by IBA from a PAINT node grounded only in those rows.
  That paper states the resistance is salicylic-acid-dependent, while GO:0009682
  is defined as salicylic-acid-independent. All five rows are marked over-annotated,
  and the mutant phenotypes are covered by defense response to bacterium for the
  two cytochromes and by response to bacterium for MYB51. This is
  worth raising with PAINT and the TAIR curators.
• Open hypotheses for single-cell reanalysis. Whether MYB122 levels track its
  indole glucosinolate targets cell by cell in epidermis, and whether CYP71A12-high
  and CYP71A13-high cells switch on different downstream branches. Only raw reads
  are deposited (ENA PRJEB61052); there is no processed count matrix.