KIF7 (human, Q2M1P5) — curation notes

Working journal for the GO annotation review. Appended to as work proceeded.

1. What the protein is

KIF7 is a 1343-residue kinesin-family protein with an N-terminal kinesin motor
domain (UniProt feature DOMAIN 15..349), a long coiled-coil/α-helical stalk
(COILED 480..542 and further coiled segments), and an extensive C-terminal
region carrying the GLI-binding and NPHP1-binding determinants
(REGION 513..775 /note="Sufficient for interaction with NPHP1"). UniProt places
it in the "TRAFAC class myosin-kinesin ATPase superfamily. Kinesin family. KIF27
subfamily."

PANTHER classification (authoritative source interpro/panther/panther-members.tsv):

Q2M1P5  PTHR24115:SF445

PTHR24115 is KINESIN-RELATED; PTHR24115:SF445 is KINESIN-LIKE PROTEIN KIF7,
and its sister subfamily PTHR24115:SF949 is KINESIN-LIKE PROTEIN COSTA
(Drosophila Costal-2). The KIF7/Cos2 orthology is therefore supported directly by
the family classification, not only by pathway analogy. Note that the cached
UniProt flat file still prints the older PTHR47969 / PTHR47969:SF8 mapping in
its DR PANTHER lines; panther-members.tsv is the authority and the
modules/hedgehog_signaling.yaml grounding (PTHR24115:SF445) matches it.

Human disease: KIF7 loss-of-function causes acrocallosal syndrome, hydrolethalus
syndrome 2, Joubert syndrome 12 and Al-Gazali-Bakalinova syndrome, and KIF7
missense alleles act as modifiers across the ciliopathy spectrum
PMID:21552264, PMID:21633164.

2. The decisive biochemistry: KIF7 is an immotile kinesin

This is the single most important fact for grading the molecular-function
annotations, and two independent purified-protein studies establish it.

He et al. 2014 (Nat Cell Biol) purified recombinant KIF7 and assayed it on dynamic
microtubules by TIRF:

Yue et al. 2018 (J Cell Biol) did the kinetic dissection across the whole
kinesin-4 family and reached the same conclusion, with a mechanism:

Consequences for the review:

  1. GO:0003777 microtubule motor activity ("A motor activity that generates
    movement along a microtubule, driven by ATP hydrolysis") is contradicted for
    KIF7. The defensible residue of those rows is microtubule binding, and
    specifically plus-end binding (GO:0051010).
  2. GO:0007018 microtubule-based movement and GO:0030705 cytoskeleton-dependent intracellular transport inherit the same problem. The
    real microtubule-directed activity is negative regulation of microtubule
    polymerization (GO:0031115) at the axonemal plus end.
  3. ATP binding and ATP hydrolysis are not contradicted. KIF7 hydrolyses ATP —
    weakly, with only modest microtubule stimulation — and the catastrophe-promoting
    activity is ATP-hydrolysis dependent: PMID:29351996 So
    GO:0005524 and GO:0016887 stand; what fails is the coupling of hydrolysis to
    stepping, not hydrolysis itself.

The motor domain is still functionally required — it is what targets KIF7 to the
ciliary tip PMID:19666503 This is why the older literature
called KIF7 a "putative ciliary motor protein" PMID:21633164 and
"may also act as a ciliary motor" PMID:19666503 — a hypothesis
that the 2014 and 2018 biochemistry then falsified. Reactome's own text has
already absorbed the correction: the cached entry
reactome/R-HSA-5610733.md says KIF7 "binds directly to the plus ends of axonemal
microtubules and inhibits their growth in an ATP-dependent manner (He et al, 2014)".

3. What KIF7 actually does: organize the ciliary tip compartment

He et al. 2014 is also the paper that gives KIF7 a positive functional
description rather than a motor description:

Two negative results in the same paper matter for the transport annotations:

Targeting to the tip does not depend on the pathway components it regulates:
PMID:24952464.

4. Position in the Hedgehog pathway: both signs, downstream of SMO

Three 2009 papers established KIF7 as the mammalian Cos2 counterpart, all with
the same bidirectional result.

The negative and positive arms were separated genetically in two mouse tissues,
and in both the positive arm runs through SUFU:

Human genetics agrees on the GLI read-out: PMID:21552264

Zebrafish adds the cross-species substitution experiment that seals the Cos2
equivalence, and the cytoplasmic pool:

This is exactly the compartment axis the pan-metazoan module records
(modules/hedgehog_signaling.yaml, gli_activation_compartment_axis): the same
kinesin scaffold step runs at the ciliary tip in vertebrates and on a cytoplasmic
complex in Drosophila. Nothing found here contradicts the module; the zebrafish
cytoplasmic puncta show the two implementations are not mutually exclusive even
within one vertebrate.

5. The interaction set, and what it means for GO:0005515

GOA carries eight bare GO:0005515 rows for KIF7. They fall into three groups.

(a) Core pathway partners — GLI3 (P10071), Gli1 (P47806), SMO (Q99835) and
Sufu (Q9Z0P7), all from PMID:19592253, matching the UniProt SUBUNIT line
"Interacts with GLI1, GLI2, GLI3, SMO and SUFU (PubMed:19592253)". These are not
generic interactions: they are the tethering activity that the functional papers
describe, and the informative molecular function is adaptor/scaffold activity
(GO:0030674), the term the module already assigns to the KIF7 annoton.

(b) P4HA2 (O15460), three rows from three separate high-throughput interactome
papers (PMID:33961781 BioPlex, PMID:35271311 OpenCell, PMID:40205054 multimodal
cell maps). Ordinarily three screen hits would still leave a bare
protein-binding row uninformative — but this interaction has since been validated
and given a mechanism, and the validating paper explicitly credits those three
screens: PMID:38909089 The mechanism is scaffolding by KIF7:

So these three rows also resolve to GO:0030674 rather than to removal.

(c) NPHP1 (O15259) from PMID:21633164 — PMID:21633164 A single co-precipitation with no
established functional consequence. Nothing more informative is supported, so the
bare row should go.

6. Localisation summary

7. Decisions taken, and the ones deliberately not taken

Actions assigned: GO:0003777 (×3) MODIFY to plus-end/microtubule binding;
GO:0007018 (×2) MODIFY to GO:0031115; GO:0030705 REMOVE; of the eight
GO:0005515 rows, six MODIFY to GO:0030674 and two (SMO, NPHP1) REMOVE; everything cilium-,
microtubule-, ATP- and smoothened-related ACCEPTed, with GO:0005737 cytoplasm
and GO:0005871 kinesin complex kept as non-core.

Three NEW annotations were considered and rejected:

Terms verified against QuickGO before use (labels and non-obsolete status):
GO:0051010 microtubule plus-end binding, GO:0031115 negative regulation of microtubule polymerization, GO:0030674 protein-macromolecule adaptor activity.