ADPRHL1 (ARH2, Q8NDY3): catalytic-site census against its active paralogues

Generated by catalytic_site_census.py. Re-running reproduces this file byte-for-byte
(python catalytic_site_census.py && git diff --exit-code RESULTS.md).

Gates

Reproduction of the sibling ADPRH panel (origin/paint/ADPRH:genes/human/ADPRH/ADPRH-bioinformatics/results.json; P54922, Q6AZR2, Q8BGK2, Q8NDY3, Q9NX46 at positions [54, 55, 56, 302, 305]):
residue and mapped position: identical on all 25 calls -- the two independent alignments agree
class label: 7 differences, all of metric, none of data.
paint/ADPRH scores substitutions with hand-defined conservative groups; this
script uses BLOSUM62 > 0. Both are defensible and they are not the same
measurement. Every difference is a BLOSUM62-positive substitution that
nonetheless deletes a metal-coordinating oxygen, which is why this script
reports a third, mechanism-anchored column (donor_group) and rests its
conclusion on that rather than on either generic scheme.
- Q8NDY3 P54922:56D->N: paint/ADPRH (hand-defined groups) calls it disruptive, BLOSUM62>0 calls it conservative; donor-group test says lost
- Q8NDY3 P54922:305S->A: paint/ADPRH (hand-defined groups) calls it disruptive, BLOSUM62>0 calls it conservative; donor-group test says lost
- Q8BGK2 P54922:56D->N: paint/ADPRH (hand-defined groups) calls it disruptive, BLOSUM62>0 calls it conservative; donor-group test says lost
- Q8BGK2 P54922:305S->A: paint/ADPRH (hand-defined groups) calls it disruptive, BLOSUM62>0 calls it conservative; donor-group test says lost
- Q6AZR2 P54922:55D->N: paint/ADPRH (hand-defined groups) calls it disruptive, BLOSUM62>0 calls it conservative; donor-group test says lost
- Q6AZR2 P54922:56D->N: paint/ADPRH (hand-defined groups) calls it disruptive, BLOSUM62>0 calls it conservative; donor-group test says lost
- Q6AZR2 P54922:305S->A: paint/ADPRH (hand-defined groups) calls it disruptive, BLOSUM62>0 calls it conservative; donor-group test says lost
Self-test: passed

External checks on the alignment method

comparison	published	measured (aligned-column identity)	source
ADPRH_vs_ADPRHL1_human	46%	46.6%	PMID:32726316
ADPRHL1_human_vs_Xenopus	75%	74.6%	PMID:32726316

Identity here is identities / aligned columns (gaps excluded), which runs a little
higher than an alignment-length denominator. It is applied identically to every row.

Part 1 - against ADPRH / ARH1 (P54922), 20 UniProt-annotated ligand sites

Reference sites (all from UniProt's own feature table, nothing hand-assigned):

position	residue	ligand
54	S	Mg(2+)
55	D	Mg(2+)
56	D	Mg(2+)
85	K	substrate
101	G	substrate
102	A	substrate
103	S	substrate
124	S	substrate
130	G	substrate
163	H	substrate
164	H	substrate
165	H	substrate
263	Y	substrate
264	S	substrate
265	G	substrate
269	S	substrate
270	S	substrate
302	D	Mg(2+)
304	D	Mg(2+)
305	S	Mg(2+)

clade	n	% id to ADPRH	identical of 20	disruptive+gap of 20	Mg(2+) donor kept of 6
ARH1 / ADPRH (active, arginine; POSITIVE CONTROL)	5	48.4-100.0	13-20	0-3	6
ARH2 / ADPRHL1 (SUBJECT clade)	7	42.6-47.7	6-7	7-8	2-3
ARH3 / ADPRS (active, serine+PAR+OAADPr; DISCRIMINATING CONTROL)	4	26.6-28.1	7-8	7-9	6
bacterial DraG (active, arginine; LOW-IDENTITY POSITIVE CONTROL)	1	27.5	9	8	6

The identity-matched control is what makes this an argument rather than an
observation. Dictyostelium ADPRH is a genuine ARH1 at 48.4% identity -- the
same distance from human ADPRH as ADPRHL1's 42.6-47.7% -- and it retains 13 of the
20 sites with 3 disruptive. Every ADPRHL1 orthologue, at that same distance, retains
6-7 with 7-8 disruptive. Retention therefore is not tracking sequence distance here.
Two low-identity active enzymes point the same way: DraG at 27.5% keeps 9, and the
ARH3 orthologues at ~28% keep 7-8 -- i.e. ADPRHL1, at nearly twice the identity,
retains no more of ADPRH's site set than proteins half as similar do.

Per-protein detail:

accession	entry	organism	SwissProt	% id	identical	conservative	disruptive	gap	own annotated sites
P54922	ADPRH_HUMAN	Homo sapiens	yes	100.0	20	0	0	0	20
P54923	ADPRH_MOUSE	Mus musculus	yes	82.4	19	0	1	0	9
Q02589	ADPRH_RAT	Rattus norvegicus	yes	82.9	19	0	1	0	9
Q32KR8	ADPRH_BOVIN	Bos taurus	yes	88.5	20	0	0	0	9
Q54H71	ADPRH_DICDI	Dictyostelium discoideum	yes	48.4	13	4	3	0	8
Q8NDY3	ARHL1_HUMAN	Homo sapiens	yes	46.6	7	6	7	0	0
Q8BGK2	ARHL1_MOUSE	Mus musculus	yes	46.1	7	6	7	0	0
Q5XIB3	ARHL1_RAT	Rattus norvegicus	yes	45.6	7	6	7	0	0
Q3ZBM1	ARHL1_BOVIN	Bos taurus	yes	42.6	7	6	7	0	0
Q5RCJ0	ARHL1_PONAB	Pongo abelii	yes	46.3	7	6	7	0	0
Q6AZR2	ARHL1_XENLA	Xenopus laevis	yes	47.7	7	6	7	0	0
Q5XJB9	ARHL1_DANRE	Danio rerio	yes	44.4	6	6	8	0	0
Q9NX46	ADPRS_HUMAN	Homo sapiens	yes	27.8	7	6	7	0	17
Q8CG72	ADPRS_MOUSE	Mus musculus	yes	27.9	7	6	7	0	17
Q28FQ6	ADPRS_XENTR	Xenopus tropicalis	yes	28.1	8	3	9	0	17
Q66HT8	ADPRS_DANRE	Danio rerio	yes	26.6	8	5	7	0	16
P14300	DRAG_RHORU	Rhodospirillum rubrum	yes	27.5	9	3	4	4	9

Site-by-site, human ADPRHL1 vs human ADPRH

ADPRH pos	ADPRH aa	ligand	ADPRHL1 pos	ADPRHL1 aa	BLOSUM62 call	donor group
54	S	Mg(2+)	56	S	identical	retained
55	D	Mg(2+)	57	D	identical	retained
56	D	Mg(2+)	58	N	conservative	lost
85	K	substrate	85	R	conservative	n/a
101	G	substrate	101	P	disruptive	n/a
102	A	substrate	102	A	identical	n/a
103	S	substrate	103	T	conservative	retained
124	S	substrate	125	E	disruptive	lost
130	G	substrate	131	G	identical	n/a
163	H	substrate	164	H	identical	n/a
164	H	substrate	165	N	conservative	n/a
165	H	substrate	166	H	identical	n/a
263	Y	substrate	265	S	disruptive	n/a
264	S	substrate	266	E	disruptive	lost
265	G	substrate	267	G	identical	n/a
269	S	substrate	271	R	disruptive	lost
270	S	substrate	272	R	disruptive	lost
302	D	Mg(2+)	304	E	conservative	retained
304	D	Mg(2+)	306	A	disruptive	lost
305	S	Mg(2+)	307	A	conservative	lost

Three things to read off this table.

The Mg(2+) site is dismantled. Of ADPRH's six Mg(2+) ligands, ADPRHL1 keeps the
oxygen donor at only three (S54->S56, D55->D57, D302->E304) and loses it at three
(D56->N58, D304->A306, S305->A307). The lost one that matters most is D56->N58:
aspartate-to-asparagine at a vicinal catalytic aspartate is the exact substitution
that abolishes activity in the active paralogue (PMID:17075046 mutates ARH3 D77/D78
to N and the reaction stops). BLOSUM62 calls D->N conservative; the mechanism does not.
The adenosine-ribose subsite is not merely lost, it is replaced by the opposite
chemistry. ADPRH's two adjacent substrate-binding serines S269/S270 align to
ADPRHL1 R271/R272 -- small hydroxyls replaced by two long cationic side chains.
That di-arginine is the experimentally validated functional element of ADPRHL1:
PMID:32726316 finds that CRISPR deletions of 1-4 residues from the Arg271-Arg272
loop abolish ventricular myofibril assembly. So this alignment independently
reproduces that paper's structural claim -- 'the critical Adprhl1 deletion covers
the exact structural position where in the active enzyme Adprh, two adjacent serines
that support adenosine-ribose substrate binding are located' -- from UniProt features
and a pairwise alignment alone, with no reference to the paper's own model.

Part 2 - against ADPRS / ARH3 (Q9NX46), 17 UniProt-annotated ligand sites

Asked separately because ARH1 and ARH3 have different specificities: an
annotation that is right for one is wrong for the other. If ADPRHL1 were a
serine/PAR/OAADPr hydrolase rather than an arginine one, it would score here.

clade	n	% id to ARH3	identical of 17	disruptive+gap of 17	Mg(2+) donor kept of 7
ARH1 / ADPRH (active, arginine; POSITIVE CONTROL)	5	20.4-27.8	7-9	5-8	6-7
ARH2 / ADPRHL1 (SUBJECT clade)	7	20.7-26.9	6-7	7-9	1-3
ARH3 / ADPRS (active, serine+PAR+OAADPr; DISCRIMINATING CONTROL)	4	54.8-100.0	14-17	0	7
bacterial DraG (active, arginine; LOW-IDENTITY POSITIVE CONTROL)	1	28.5	11	5	6

The ARH3 clade recovers 14-17 of its own 17 sites, so the reference set is not
intrinsically hard to hit. ADPRHL1 scores 6-7 -- it fails against both active
references. There is therefore no reading of these data on which ADPRHL1 is a
mis-assigned ARH3-type (serine / PAR / O-acetyl-ADP-ribose) hydrolase rather than a
mis-assigned ARH1-type (arginine) one; the correct conclusion is neither.

What this does and does not establish

It measures residue retention at positions UniProt annotates as ligand sites on
the active paralogues. It is not an activity assay; the activity evidence is the
in-vitro negative in PMID:17075046 and PMID:36497109.
ADPRHL1 has no annotated binding or active sites of its own, so the stronger
'lands on the target's own annotated site' test used elsewhere in this campaign
cannot be applied. That is reported rather than silently dropped: the residue-identity
call is the weaker of the two available tests.
A retained residue is not evidence of activity, and a lost one is not proof of
its absence in vivo -- see PMID:36497109's own caveat that the negative results were
obtained with model substrates.