tusc2a / tusc2b protein, expression and synteny comparison

Script: pair_analysis.py (run from the repo root with
uv run python genes/DANRE/tusc2a/tusc2a-bioinformatics/pair_analysis.py > genes/DANRE/tusc2a/tusc2a-bioinformatics/output.txt).
Raw output: output.txt (run 2026-09-28). Gar-bridge synteny: gar_bridge.py, run as
uv run python genes/DANRE/tusc2a/tusc2a-bioinformatics/gar_bridge.py ENSLOCG00000014230 ENSDARG00000099817 ENSDARG00000025340 > genes/DANRE/tusc2a/tusc2a-bioinformatics/gar_bridge_output.txt. Zebrafish sequences come from the cached UniProt records
(tusc2a Q08BH0, 111 aa; tusc2b Q6DH03, 111 aa). Human TUSC2 (O75896) comes from UniProt; gar and
medaka proteins are the Ensembl canonical proteins of the orthologues Ensembl Compara assigns (gar
ENSLOCG00000014230, one gar gene for both copies; medaka ENSORLG00000027802 for tusc2a and
ENSORLG00000027402 for tusc2b, each one-to-one). Alignments are Biopython global, BLOSUM62, gap open
-10 / extend -0.5; identity = identical columns / alignment length.

Protein

Comparison	Identity
tusc2a vs tusc2b	80.2% (111 columns)
tusc2a / tusc2b vs human TUSC2 (O75896, 110 aa)	70.3% / 65.8%
tusc2a / tusc2b vs gar TUSC2	81.2% / 77.7%
tusc2a vs its medaka orthologue / vs the other medaka copy	77.5% / 76.6%
tusc2b vs its medaka orthologue / vs the other medaka copy	79.3% / 75.7%

The protein is short (about 110 residues), so these differences between orthologue and paralogue
distances are only a few residues and do not by themselves resolve the tree.

Functional features (from UniProt O75896 and the calcium-binding motif described in
PMID:42314984):

The N-myristoylation site (Gly2 of human TUSC2) is kept in both copies: both start MGGSGSK.
The Ensembl gar model starts with one extra residue (KMGGSGSK), so the Gly is at position 3
there, but it aligns to human Gly2.
The calcium-binding motif DEDGDLAHEFYEE is present, exactly, in both zebrafish copies (starting
at residue 55), in human TUSC2, in gar and in both medaka copies.
The phosphoserine at human Ser50 aligns to Ser in both copies and in gar.
Over the mature chain (human residues 2-110), tusc2a is 70.6% and tusc2b 66.1% identical to human.

Both copies keep the myristoylation site and the exact calcium-binding motif.

Relative rate (gar outgroup): of 111 aligned gar positions, 5 changed only in tusc2a and 9 only
in tusc2b (chi2 = 1.14, not significant).

Expression

Whole-embryo time course (E-ERAD-475, median TPM).

Stage	tusc2a	tusc2b
zygote	26	10
2-cell	13	13
blastula (128-cell to dome)	5-6	31-38
gastrula (50% to 75% epiboly)	1-2	26-38
segmentation	4-9	47-51
pharyngula	7-11	42-45
hatching to larval day 5	14-18	37-45

tusc2a is the larger maternal transcript (26 vs 10 TPM in the zygote) but falls to 1-2 TPM by
gastrulation and recovers only partly (14-18 TPM in larvae). tusc2b is the main zygotic copy,
at 26-51 TPM from blastula to larva.

Bgee calls (only "expressed" calls are returned; absence is not proven absence). tusc2a has 25
calls, tusc2b 31. All 22 RNA-Seq-supported entities of tusc2a are also called for tusc2b; RNA-Seq
calls for tusc2b only are heart, ovary and tail. Of 25 entities called for both, tusc2a scores
higher in 4 (median difference tusc2a minus tusc2b = -16.9). tusc2a's highest scores are testis
(89.7) and mature ovarian follicle (87.3); tusc2b's are muscle (91.8), somite and liver.
Gar TUSC2 has 14 calls across adult tissues, embryo and larva, highest in eye, muscle and brain, so
broad expression is the ancestral state.

ZFIN curated expression. tusc2a has no curated wild-type expression. tusc2b has one row: whole
organism by in situ hybridization from zygote to pec-fin stage (ZDB-PUB-040907-1), with no
restricted domain recorded.

Interpretation

Protein: both copies keep the two features known to matter for TUSC2 (N-myristoyl glycine and
the calcium-binding motif), and neither is evolving faster. No evidence of protein divergence.
Expression: overlapping and broad in adults, with tusc2b higher in most tissues. The difference
is in timing and level: tusc2a is mainly a maternal transcript and a lower-level adult copy with
its highest calls in gonads; tusc2b carries zygotic expression through development. This is a
quantitative difference, not a tissue partition. No gar developmental time course was available,
so whether the maternal/zygotic split is derived cannot be tested here.

Synteny

Direct comparison (pair_analysis.py, section 7 of output.txt). tusc2a is on chr6 (54.1 Mb)
and tusc2b on chr22 (10.7 Mb). Of 72 protein-coding neighbours of tusc2a (within 1.5 Mb) 30 have a
teleost-level zebrafish paralogue, and of 102 neighbours of tusc2b 16 do, but none of these
paralogue partners lies near the other copy or on the other copy's chromosome. Neighbour-to-neighbour
paralogy therefore gives no signal, probably because most duplicated neighbours were lost from one
side.

Gar bridge (gar_bridge.py; output in gar_bridge_output.txt). Gar TUSC2 is on LG5. Of 52
protein-coding gar genes within 1 Mb, 47 have zebrafish orthologues.

Near tusc2a (chr6, within 3 Mb): orthologues of 5 gar neighbours, including hyal1 and hyal2a
(54.06 and 54.09 Mb, next to tusc2a at 54.12 Mb), cacna2d2a and mst1.
Near tusc2b (chr22, within 3 Mb): orthologues of 6 gar neighbours, including bap1, rad54l2,
cyb561d2, rassf1, hyal2b and abhd14a (10.59-10.70 Mb, around tusc2b at 10.66 Mb).
One gar neighbour, hyal2b, has zebrafish orthologues next to both copies (hyal2a beside tusc2a,
hyal2b beside tusc2b).
The largest block of gar-neighbour orthologues (24) is on zebrafish chr8, away from both copies.

The gar TUSC2 neighbourhood maps onto two separate zebrafish regions, each holding one tusc2 copy
and a different subset of the ancestral neighbours (a HYAL gene next to each copy, RASSF1 and BAP1
next to tusc2b). This is double conserved synteny, the signature expected of a whole-genome
duplication, and supports a TGD origin of the pair.