Headline: of the 12 residues that contact bound sodium in the human
NTCP/SLC10A1 cryo-EM structure 9QZQ, only 3 are conserved in SLC10A7, with
four non-conservative substitutions — including replacement of the NTCP
carboxylate E257 by phenylalanine and of N106 by alanine. The C-terminal
register is the weak point and the residue identities, not the superposition
distances, carry the argument there: the equivalents of 257 and 261 sit 19-23 Å
from their counterparts, which is alignment uncertainty rather than a measured
displacement. Separately,
all three polar anchors of the NTCP bile-salt pocket are lost (N103→V118,
N262→T268, Q264→A270). The SLC10 fold is still recognisable (TM-score 0.766
against the experimental NTCP chain), but neither the ion-binding chemistry nor
the substrate recognition chemistry is. This is independent structural support for the
review's position that SLC10A7's molecular function must stay at the
substrate-agnostic parent, and against reading it as an NTCP-type sodium-coupled
bile-salt carrier.
This is the SLC10A7 slice of a family-wide analysis; the pipeline, the method
control and the full limits section live with it, in
genes/human/SLC10A4/SLC10A4-bioinformatics/ (RESULTS.md, fetch_structures.py,
pocket_conservation.py, pocket_conservation.json). Nothing is duplicated
here beyond the SLC10A7 numbers.
Sites were defined empirically, as every NTCP residue with a heavy atom within
4.5 Å of a bound ligand: sodium in PDB 7ZYI (2.88 Å) and 9QZQ (3.11 Å),
glycochenodeoxycholate — a conjugated bile salt and genuine NTCP substrate — in
7ZYI, and N-tetradecanoylglycine (the myristoyl-glycine of the HBV preS1 anchor)
in 8RQF (3.41 Å). Each contact is additionally recorded as side-chain or
main-chain, since a substitution at a backbone-only contact is
sequence-independent. Author numbering in all three chains indexes exactly into UniProt
Q14973 (293/293, 286/286, 301/301). SLC10A7 is represented by its AlphaFold DB
v6 model (Q0GE19); it has no experimental structure.
Sodium site, 9QZQ, 12 positions, NTCP → SLC10A7:
Q68→T80* G101→P116~ S105→S120 N106→A121* S119→A135~ I120→I136
T123→S139~ E257→F263* T258→T264 G259→P265~ C260→A266* Q261→D267~
* non-conservative, ~ conservative. Totals: 3 identical, 5 conservative,
4 non-conservative. TM-score 0.766, RMSD 3.66 Å.
Family context from the same run (sodium site, identical / conservative /
non-conservative out of 12):
| Protein | id | cons | non-cons | TM | note |
|---|---|---|---|---|---|
| SLC10A1 (NTCP) | 12 | 0 | 0 | 0.973 | method control (AFDB vs its own structure) |
| SLC10A2 (ASBT) | 10 | 1 | 1 | 0.877 | positive control, Na⁺-coupled |
| SLC10A6 (SOAT) | 11 | 0 | 1 | 0.879 | positive control, Na⁺-coupled |
| SLC10A4 | 11 | 0 | 1 | 0.876 | orphan, ligands retained (its one change is a main-chain contact) |
| SLC10A7 | 3 | 5 | 4 | 0.766 | site degenerate |
SLC10A7 is the only human family member whose sodium site is not conserved. The
two paralogues with demonstrated sodium-coupled transport keep it, and even the
orphan SLC10A4 keeps it — which is what makes the SLC10A7 result interpretable
rather than merely a low number.
Bile-salt pocket (7ZYI, glycochenodeoxycholate, 21 positions): SLC10A7 scores 5
identical, 5 conservative, 9 non-conservative, and loses all three polar anchors
(N103→V118, N262→T268, Q264→A270). SLC10A4 keeps all three; ASBT, which does
transport bile salts, keeps two. So SLC10A7 is the only family member to have
lost both the ion site and the substrate recognition triad.
In the myristoyl/preS1 pocket (8RQF, 6 positions) SLC10A7 scores 2 identical,
2 conservative, 2 non-conservative; no human paralogue retains that pocket,
which is NTCP-specific.
An OpenScientist hypothesis job was run on the same question without being given
this analysis (SLC10A7-hypotheses/structure-translocation-pathway/openscientist.md).
It returned "partially supported", splitting the fold claim (retained — the
panel/core architecture and a traversable cavity are present) from the
NTCP-type-carrier claim (refuted — the sodium sphere is degraded), which matches
the result above. It differs on one detail, left open here: it reads SLC10A7's
Q68 position as deleted into an indel rather than substituted to T80. Both
readings agree the Na1 primary ligand is gone.
It supports two statements already made in the SLC10A7 review on other grounds:
that the GO:0022857 parent should not be deepened to a substrate-specific
child, and that the unidentified-substrate gap is real rather than a gap in the
literature search. It also makes the Golgi localization and the negative
bile-acid/steroid-sulfate transport assay structurally coherent.
It is not experimental evidence and no review action was changed because of
it. A degenerate site is not proof that no ion is bound — coordination could be
rebuilt by residues that do not align to NTCP's own ligand contacts, which is
all this analysis inspects — and the SLC10A7 coordinates are predicted, not
observed. The decisive experiment remains the one in the review's
suggested_experiments: identify the transported solute directly.