arnF (ArnF / PmrM / YfbJ) - Research Notes

Gene Identity

• UniProt: P76474 (ARNF_ECOLI)
• Gene names: arnF (preferred), pmrM (old), yfbJ (old)
• Ordered locus names: b2258, JW5373
• Organism: E. coli K12
• Size: 128 AA, 14 kDa
• Structure: 4 transmembrane helices, inner membrane protein

Core Function

ArnF is one subunit of an undecaprenyl phosphate-alpha-L-Ara4N flippase heterodimer (ArnE/ArnF). This flippase translocates 4-amino-4-deoxy-L-arabinose-phosphoundecaprenol (alpha-L-Ara4N-phosphoundecaprenol) from the cytoplasmic to the periplasmic side of the inner membrane PMID:17928292.

This flipping step is essential for the L-Ara4N modification of lipid A by ArnT (whose active site faces the periplasm), which in turn is required for polymyxin resistance PMID:17928292.

Pathway Context

ArnF operates in the L-Ara4N lipid A modification pathway:
1. UDP-glucose → UDP-glucuronic acid (Ugd)
2. UDP-glucuronic acid → UDP-4-ketopentose (ArnA C-terminal domain)
3. UDP-4-ketopentose → UDP-beta-L-Ara4N (ArnB, transamination)
4. UDP-beta-L-Ara4N → N-formylated form (ArnA N-terminal domain)
5. N-formyl-L-Ara4N → undecaprenyl-phosphate-L-Ara4N (ArnC, transfer to carrier lipid)
6. Deformylation (ArnD)
7. Flipping of undecaprenyl-phosphate-L-Ara4N across inner membrane (ArnE/ArnF) ← arnF acts here
8. Transfer of L-Ara4N to lipid A on periplasmic face (ArnT)

Key Evidence (PMID:17928292 - Yan, Guan, Raetz 2007)

Deletion phenotype

• arnF deletion (pmrM deletion) in pmrAc background: polymyxin-sensitive PMID:17928292
• Lipid A lacks L-Ara4N modification (>95% reduction) PMID:17928292
• Undecaprenyl phosphate-alpha-L-Ara4N levels NOT reduced - the substrate accumulates but can't be flipped PMID:17928292

Flippase evidence (sulfo-NHS-biotin assay)

• Membrane-impermeable sulfo-NHS-biotin labeling of undecaprenyl-phosphate-L-Ara4N reduced 4-5 fold in arnF mutant vs parent PMID:17928292
• Membrane-permeable NHS-biotin labeling unaffected - ruling out sequestration or degradation
• arnT mutant shows NO reduction in sulfo-NHS-biotin labeling, proving ArnT is not the flippase

Complementation

• Polymyxin resistance restored by pWSK29-PmrM (arnF) but NOT by pWSK29-PmrL (arnE), demonstrating non-redundant, indispensable roles PMID:17928292

Subunit

• Forms heterodimer with ArnE (PmrL) PMID:17928292

Protein Family

• Related to small multidrug resistance (SMR) transporters / EmrE-like
• DMT (Drug/Metabolite Transporter) superfamily (TCDB 2.A.7.22.1)
• Distantly related to EmrE, but functionally distinct (lipid flippase vs drug efflux)
• PANTHER: PTHR30561 (SMR family)

Regulation

• Induced by BasR (PmrA homolog in E. coli) PMID:15569938
• Part of the arnBCADTEF operon (previously yfbE operon / pmrHFIJKLM in Salmonella)
• The BasS-BasR two-component system responds to iron PMID:15322361

Response to Iron

• The yfbE operon (now arn operon) is induced in response to external iron PMID:15322361
• BasS-BasR system is involved in iron responses PMID:15322361
• Fe(III) toxicity studies show PmrA/PmrB system important for resistance to Fe(III) PMID:12139617
• Note: arnF's role in iron response is INDIRECT - iron induces BasR which induces the arn operon, leading to lipid A modification with L-Ara4N

Subcellular Localization

• Inner membrane, multi-pass membrane protein (4 TM helices)
• IDA from global topology study [PMID:15919996 - C-terminal GFP/PhoA fusion]
• Labeled "plasma membrane" in GO (GO:0005886) - this is the E. coli inner membrane
• GO:0005887 (integral component of plasma membrane) is obsolete — cannot be used

GO Hierarchy Analysis: Intramembrane Lipid Transporter Activity

The GO:0140303 branch reveals an important ontology gap for ArnF:

GO:0140303 intramembrane lipid transporter activity
  "Enables the transport of a lipid from a region of a membrane
   to a different region on the same membrane."
  ├── GO:0017128 phospholipid scramblase activity
  │     (ATP-independent, non-selective, bidirectional)
  └── GO:0140326 ATPase-coupled intramembrane lipid transporter activity
        "Catalysis of the movement of lipids from one membrane leaflet
         to the other, driven by ATP hydrolysis."
        ├── GO:0140327 flippase activity
        │     (exoplasmic → cytosolic, ATP-dependent)
        │     ├── GO:0015247 aminophospholipid flippase activity
        │     ├── GO:0140333 glycerophospholipid flippase activity
        │     ├── GO:0140345 phosphatidylcholine flippase activity
        │     ├── GO:0140351 glycosylceramide flippase activity
        │     └── GO:0140347 N-retinylidene-PE flippase activity
        └── GO:0140328 floppase activity
              (cytosolic → exoplasmic, ATP-dependent)
              ├── GO:0015161 lipid III floppase activity (ECA assembly)
              ├── GO:0015437 lipopolysaccharide floppase activity
              ├── GO:0034202 glycolipid floppase activity
              ├── GO:0046623 sphingolipid floppase activity
              └── GO:0090554 phosphatidylcholine floppase activity

Why GO:0140303 is the correct (and only correct) MF term for ArnF

ArnE/ArnF flips undecaprenyl phosphate-α-L-Ara4N from cytosolic → periplasmic leaflet.
This is the floppase direction (cytosolic → exoplasmic). However:

• GO:0140328 (floppase activity): requires ATP hydrolysis — ArnE/ArnF is a small 4-TM
  protein with NO ATPase domain. Not applicable.
• GO:0140327 (flippase activity): wrong direction AND requires ATP. Not applicable.
• GO:0017128 (scramblase activity): ATP-independent but defined as non-selective and
  bidirectional — ArnE/ArnF is directional and substrate-specific. Not applicable.
• GO:0015161 (lipid III floppase activity): closest analog (also undecaprenyl-linked
  substrate, also bacterial inner membrane), but defined as ATP-dependent. Not applicable.

Conclusion: GO:0140303 is the only term that correctly captures ArnF's function without
making false claims about energy coupling. All children either require ATP or don't match
the directionality/selectivity.

Ontology gap

There is no GO term for "ATP-independent, directional intramembrane lipid transporter
activity" — the space between scramblase (non-selective, bidirectional) and
flippase/floppase (ATP-dependent, directional). ArnE/ArnF falls in this gap: it is
directional and substrate-specific, but not ATP-coupled.

This could be flagged as a potential new term request: something like
"energy-independent intramembrane lipid transporter activity" or
"non-ATPase floppase activity" under GO:0140303.

The closest existing term by analogy is GO:0015161 (lipid III floppase activity) which
also handles undecaprenyl-linked substrates flipped across the bacterial inner membrane
during cell surface polysaccharide assembly — but for ECA rather than L-Ara4N lipid A
modification, and classified as ATP-dependent.

IBA Analysis: PANTHER PTHR30561

WITH/FROM proteins in the IBA annotations

The IBA annotation to GO:0022857 (transmembrane transporter activity) was inferred from:

All non-ArnE/ArnF members are genuine solute exporters — they move substrates from
cytoplasm across the membrane to the periplasm/exterior. ArnE/ArnF does something
fundamentally different: it flips a lipid-linked substrate between membrane leaflets.

The PANTHER family PTHR30561 groups these by shared SMR/DMT fold (4-TM helices), but the
functional annotation "transmembrane transporter activity" correctly describes only the
exporters, not the flippase pair.

PANTHER subfamily structure

From PTHR30561-entries.csv, the family contains at least these subfamilies:
- SF9: ArnF-related (4-amino-4-deoxy-L-arabinose-phosphoundecaprenol flippase subunit)
- SF23: ArnE-related
- SF6: MdtI (spermidine export)
- SF2: MdtJ (spermidine export)
- Family-level (not assigned to SF): EmrE, Gdx/SugE

This means the IBA could in principle be refined at the subfamily level — SF9/SF23
should get intramembrane lipid transporter annotations, while SF2/SF6 and family-level
members keep transmembrane transporter annotations.

Note on SF9 subfamily mapping inconsistency: One entry in PTHR30561-entries.csv
(A8FRQ9, Shewanella sediminis ArnE) is named as ArnE but assigned to SF9
(ARNF-RELATED) rather than SF23 (ARNE-RELATED). This is likely a genuine PANTHER
classification based on sequence similarity — the Shewanella ArnE ortholog may be
more similar to ArnF sequences. This does not affect the E. coli annotations or the
IBA analysis, since E. coli ArnE (Q47377) and ArnF (P76474) are correctly assigned
to SF23 and SF9 respectively.

EcoCyc Annotation Quality Assessment

EcoCyc contributed 5 of the 18 total GO annotations for arnF:

The iron response annotations are a textbook case of over-annotation: iron activates
BasS-BasR → induces arn operon → ArnF expressed. But ArnF doesn't sense, bind, or
detoxify iron. By this logic, every gene in an iron-responsive regulon would be annotated
to "response to iron(III) ion," which is clearly inappropriate.

The EcoCyc IMP annotations are well-supported and accurate. The IDA for plasma membrane
is the strongest CC evidence available.