AP5S1 (sigma-5): what the AP-5 structure says the subunit does

Regenerate with uv run python sigma5_structure.py in this directory. Structures are
downloaded from RCSB and sequences from UniProt at run time; expected sequence lengths
are asserted, and no result below is hardcoded.

Inputs and chain verification

Every chain indexes its UniProt sequence directly except zeta, whose deposited
numbering is shifted by one (an extra N-terminal residue in the construct); the offset
is found by scanning rather than assumed, and all zeta residue numbers below are
reported in UniProt coordinates.

Sequence lengths verified: {'Q9NUS5': 200, 'Q3U829': 807, 'Q2VPB7': 878, 'Q96JI7': 2443, 'Q8BJ63': 490, 'P62743': 142}

Q1 -- sigma-5 sits on the zeta trunk, and touches nothing else

Free solvent-accessible surface of the isolated sigma-5 chain: 11230.0 A^2.

• sigma-5 residues at the zeta5 (AP5Z1, mouse) interface: M1, V2, L24, F29, G30, A31, H41, A43, E44, D46, R47, L48, R50, K51, M64, L67, Q68, Q70, A71, S72, G73, R74, P75, R95, G96, A97, F98, R99, T111, L120, H129, N131, L132, L133, L134, E136, G137, L152, L153, A154, S156, T157, L160, L161, R162, A163, D164, E167, T171, P175, H176, G177, Q178, L179, L180, F181, N183, Q185, F186, W198
• sigma-5 residues at the SPG11 WD40-hairpin (spatacsin, human) interface: L20, R22, S34, R45, L48, L49, K51, E52, L55, A56, R59, Q60, S63, M64, L67, L100, A101, E103, N104, P105

The zeta N-terminal segment tested is UniProt residues 1-40. It contacts 9 sigma-5 residues (L24, L120, L152, L153, A154, S156, T157, L160, L161); the zeta residues involved are [1, 2, 3, 5, 6, 9, 10].

Q2 -- the acidic dileucine site

The site is defined from 2JKR: the sigma2 residues within 4.5 A of the bound CD4 dileucine peptide (modelled sequence MSQIKRLLS). That gives 19 residues: N9, R10, A11, R15, Y62, A63, G64, L65, F67, H85, V88, E89, N92, N97, V98, C99, E100, L101, L103.

Superposition method: Bio.PDB.cealign.CEAligner (Combinatorial Extension, CA atoms, sequence-independent), reference 2JKR chain S (sigma2).

Read the RMSD column, not the coverage column: beta5 is a 551-residue solenoid, so its
CA atoms are dense enough to fall within 4 A of most reference positions whatever the
fold. RMSD separates the three cases cleanly, and sigma-5 sits between the identical
control and the unrelated fold, which is what a divergent homologue should do.

Summary: of 19 sigma2 residues that contact the dileucine peptide, 18 have a structurally equivalent position in sigma-5 (CA within 4 A after superposition), 3 of those carry the identical residue, and 3 of them are in contact with the zeta N-terminus in the assembled AP-5 core.

Does the mouse anchor's numbering transfer to the human protein?

The dileucine site is defined on P62743 (mouse Ap2s1, the sigma2 chain of 2JKR), while the functional
weight of these positions comes from work on the human protein, P53680 (AP2S1).
Same length: True. Identical over the full chain: True. All peptide-contacting positions identical between the two: True.

Sequence control -- can this comparison be made from sequence alone?

Every pairwise alignment of sigma-5 against a human sigma paralog scores negative and
sits within a few standard deviations of shuffled sequence, so the correspondence used
in Q2 could not have been obtained from pairwise sequence alignment. That is consistent
with Hirst et al. 2011 (PMID:22022230), who needed profile methods (HHpred, probability
96.8% against sigma2) rather than BLAST to recognise this protein as a sigma subunit --
and it is why the mapping above is done on structures.

Conclusion

1. Sigma-5 is the zeta-adaptin partner: it buries 2621.8 A^2 against the zeta trunk over 60 residues, and touches neither beta5 nor mu5. Structurally it occupies the position that sigma1/sigma2/sigma3 occupy in AP-1/AP-2/AP-3.

2. Sigma-5 is also part of the SPG11 binding surface, burying 755.4 A^2 against the SPG11 WD40-hairpin over 20 residues. This is a role the other AP sigma subunits do not have, and it is the structural counterpart of the pull-down result that SPG11 associates with the zeta/sigma-5 subcomplex and is needed to assemble the AP-5 heterotetramer.

3. The fold that carries the acidic dileucine site is retained (18 of 19 peptide-contacting positions have a structural equivalent) but the chemistry is not: only 3 of those positions carry the same residue. The hydrophobic pocket linings and the basic patch are the positions that change, and in the assembled AP-5 core 3 of the equivalent positions are covered by the zeta N-terminus. Retention of the sigma fold is therefore not evidence that sigma-5 reads dileucine sorting signals, and the structure gives a positive reason to think it does not.

4. This is a structural argument, not a binding assay. No one has tested AP-5 against a dileucine motif, so the claim defended here is the negative one: nothing in the structure supports annotating cargo-signal recognition to sigma-5.