ADGB (Q8N7X0) - do the calpain and globin domains retain their functional residues?

Generated by adgb_sites.py. Reproduce with:

uv run python adgb_sites.py --self-test              # guards, both directions
uv run python adgb_sites.py --json results.json \
                            --results-md RESULTS.md  # this file

Every number below is interpolated from the same run that printed the tables,
so this file cannot drift from the measurement. A hand-edit is reverted by the
next run.

Question

GOA gives ADGB four InterPro IEA rows, two from each domain signature:

signature	GO term
IPR001300 Peptidase_C2_calpain_cat	GO:0004198 calcium-dependent cysteine-type endopeptidase activity
IPR001300 Peptidase_C2_calpain_cat	GO:0006508 proteolysis
IPR012292 Globin/Proto	GO:0019825 oxygen binding
IPR012292 Globin/Proto	GO:0020037 heme binding

A domain's name is not an activity. Does ADGB keep the residues each activity needs?

Subject: Q8N7X0 ADGB_HUMAN / Androglobin, 1667 aa, reviewed=True.
UniProt annotates the calpain catalytic domain at 70-411 and the globin domain in two parts, (763, 890) and (936, 968) (circular permutation).

Q1. Calpain catalytic triad: NOT retained

Each of the 6 reviewed reference calpains carries three UniProt
Active site features. ADGB carries 0.

Aligning each reference's catalytic domain to ADGB 70-411 under 4 substitution/gap models (BLOSUM62, BLOSUM45, BLOSUM80, PAM250) and
reading the ADGB residue in each reference-active-site column:

triad role	aligned columns	ADGB positions hit	columns where ADGB has the catalytic residue
nucleophile Cys	15	[132, 146]	11/15, always at ['C132']
general base His	22	[267, 320, 321, 326, 332, 334, 352]	1/22 ['H267']
Asn	22	[319, 343, 344, 346, 353, 357]	0/22 (none)

Read per-residue rather than per-triad (a single model's triad count is an
alignment artefact, not a measurement):

The nucleophile column converges: whenever a reference's catalytic Cys
aligns to a Cys in ADGB it is always the same residue, C132,
in 11 of 15 columns. Which references reach it varies by model; the
ADGB position does not.
The general-base His column does not converge: 1 of 22 columns,
scattered over ADGB positions [267, 320, 321, 326, 332, 334, 352]. The lone hit
(H267) appears under one model only and at a position no
other model selects - the signature of alignment noise, not conservation.
The Asn column never yields an Asn: 0 of 22.

A papain-fold cysteine protease cannot catalyse without the general-base
histidine that deprotonates the nucleophile. A retained Cys with no His and no
Asn is a fold, not an active site.

Independent corroboration: MEROPS classifies ADGB as a non-peptidase homologue

This does not depend on the alignment at all.

accession	gene	MEROPS	catalytic type
Q8N7X0	ADGB	`C02.972`	Non-peptidase homologue
P07384	CAPN1	`C02.001`	Cysteine
P17655	CAPN2	`C02.002`	Cysteine
P20807	CAPN3	`C02.004`	Cysteine
Q9Y6W3	CAPN7	`C02.029`	Cysteine
Q9HC96	CAPN10	`C02.018`	Cysteine
O75808	CAPN15	`C02.031`	Cysteine

All 6 reference calpains: Cysteine. ADGB: Non-peptidase homologue, and ADGB is the holotype of that entry.
The identifier block is consistent: reference suffixes [1, 2, 4, 18, 29, 31] against ADGB's 972.

UniProt says the same thing in its CAUTION: "The calpain domain lacks the
conserved active site residues. Probably catalytically inactive as a
calcium-dependent cysteine-type endopeptidase."

Verdict: GO:0004198 is not supported. The measurement refines UniProt's
wording - it is not that all the active-site residues are gone; the nucleophile
survives and the general base and Asn do not.

Caveat that must be stated

Q1 CAVEAT: ADGB carries ZERO UniProt 'Active site' features, so the lands_on_subject_annotated_site condition is VACUOUS in Q1 and its all-'no' column is not a measurement. That absence is itself the signal: UniProt annotates the Calpain catalytic domain on ADGB but propagates no catalytic residues into it, whereas all 6 reference calpains carry 3 each. The condition IS load-bearing in Q2, where ADGB has 2 annotated heme-iron Binding sites and it separates the on-register Mb/Ngb/Cygb alignments from the off-register HBB one.

Q2. Globin heme ligands: proximal His retained, distal His replaced by Gln

Same method against reviewed globins whose heme-iron ligands UniProt annotates.
Here ADGB does have its own annotated sites, so the
lands_on_subject_annotated_site condition is live and does the discriminating.

reference	%id	reference site	-> ADGB	residue match	on an ADGB-annotated site	role
P02144 MYG_HUMAN	21.9	H65	Q792	no	yes	O2
P02144 MYG_HUMAN	21.9	H65	Q792	no	yes	nitrite
P02144 MYG_HUMAN	21.9	H94	H824	yes	yes	proximal binding residue
P69905 HBA_HUMAN	22.7	H59	Q792	no	yes	O2
P69905 HBA_HUMAN	22.7	H88	Q821	no	no	proximal binding residue
P68871 HBB_HUMAN	20.3	H3	H763	yes	no	(2R)-2,3-bisphosphoglycerate
P68871 HBB_HUMAN	20.3	H64	H830	yes	no	distal binding residue
P68871 HBB_HUMAN	20.3	H93	Q854	no	no	proximal binding residue
P68871 HBB_HUMAN	20.3	H144	E887	no	no	(2R)-2,3-bisphosphoglycerate
Q9NPG2 NGB_HUMAN	22.7	H64	Q792	no	yes	distal binding residue; reversible
Q9NPG2 NGB_HUMAN	22.7	H96	H824	yes	yes	proximal binding residue
Q8WWM9 CYGB_HUMAN	18.0	H81	Q792	no	yes	distal binding residue
Q8WWM9 CYGB_HUMAN	18.0	H113	H824	yes	yes	proximal binding residue

Distal position (ADGB Q792). 5 reference columns land on it, from 4 distinct globins, and the reference
residue is H in every one. ADGB has Q.
Proximal position (ADGB H824). 3 reference columns land on it and 3 of 3 match: ADGB retains the proximal histidine.
5 columns are off-register (they do not land on either ADGB
annotated site). Their apparent matches are noise - the HBB alignment in
particular 'matches' a distal His at a position UniProt does not annotate while
failing the proximal one. Sequence identity does not predict register here
(CYGB at 18.0% is on-register for both sites; HBB at 20.3% is on-register for neither) - the
annotated-site condition does.

Verdict: the globin domain is intact as a heme protein, but it is a His/Gln
hexacoordinate globin, not a bis-histidyl one.

This is confirmed independently by the primary literature, in three places that
agree with the sequence and with each other:

Reeder et al. 2024 (PMID:38725499) state the ferric coordination as
His-Fe(III)-X with "X = His for Cygb and Gln for Adgb", and label the
alignment position "the distal histidine/glutamine".
Nie et al. 2025 (PMID:39719941) label Fig. 1B "the distal Gln30" and conclude
"Gln30 in the heme distal site may coordinate to the heme iron in the ferrous
state" - their construct starts at UniProt 763, so their His62/res30 are
UniProt H824/Q792 exactly (spacing 32 in both numberings).
Ren et al. 2026 (PMID:42372363) name it in the title and mutate it: "the heme
distal residue Gln12 through the generation of Q12H and Q12Y mutants".

UniProt's own FUNCTION line is internally inconsistent with its feature table:
it says ADGB is a "chimeric globin with a bis-histidyl six-coordinate heme-iron
atom" while the feature table annotates the distal ligand as Q792. Worth reporting to UniProt.

What this does and does not settle

It settles that the calpain-domain-derived catalytic MF term has no residue
support, and that MEROPS and UniProt independently agree.
It settles that heme binding is structurally well founded (the proximal ligand
is retained and is the residue UniProt annotates).
It does not test oxygen binding, which is a ligand-binding measurement, not
a sequence one. See the notes file: an oxy complex is formed and its
association rate was measured (PMID:38725499), so GO:0019825 is true as
chemistry, while the same paper's autoxidation rate argues it is not a
physiological role.
It does not test whether ADGB contributes to SEPT10 cleavage by a non-catalytic
route, which is what the cell-based data actually show.