sh3glb2a / sh3glb2b protein, expression and synteny comparison

Script: pair_analysis.py (run from the repo root with
uv run python genes/DANRE/sh3glb2b/sh3glb2b-bioinformatics/pair_analysis.py > genes/DANRE/sh3glb2b/sh3glb2b-bioinformatics/output.txt).
Raw output: output.txt (run 2026-09-28). Zebrafish sequences come from the cached UniProt
records used for the reviews (sh3glb2a A0A8M1N8K5, 381 aa, RefSeq NP_001035087.2; sh3glb2b
A0A8M9Q303, 421 aa, RefSeq XP_021331719.1 "isoform X1") plus the sh3glb2b RefSeq NP isoform
(Q802U5, NP_957413.1, 373 aa). Human SH3GLB2 (Q9NR46) and SH3GLB1 (Q9Y371) come from UniProt;
gar and medaka proteins are the Ensembl canonical proteins of the orthologues that Ensembl
Compara assigns. Alignments: Biopython global, BLOSUM62, gap -10/-0.5; identity = identical
columns / alignment length. The script is the same as the mapre3a/mapre3b one with a
different CONFIG.

Orthology (Ensembl Compara)

sh3glb2a (ENSDARG00000008983, chr8) and sh3glb2b (ENSDARG00000035470, chr5) are
within-species paralogues with the duplication node at Osteoglossocephalai.
Both share one spotted gar orthologue (ENSLOCG00000006193) and the human orthologue SH3GLB2.
Each has its own one-to-one medaka orthologue (ENSORLG00000016839 for sh3glb2a,
ENSORLG00000014325 for sh3glb2b), so medaka kept both copies.

Protein

Comparison	Identity
sh3glb2a vs sh3glb2b (reviewed accessions)	73.6% (421 columns)
sh3glb2a vs sh3glb2b NP isoform (Q802U5)	70.4% (402 columns)
sh3glb2a / sh3glb2b vs human SH3GLB2	72.9% / 69.2%
sh3glb2a / sh3glb2b vs human SH3GLB1	58.2% / 55.3%
sh3glb2a / sh3glb2b vs gar SH3GLB2 orthologue	74.9% / 73.7%

Both copies are SH3GLB2 (endophilin-B2) orthologues, closer to SH3GLB2 than to SH3GLB1.
The reviewed sh3glb2b entry (isoform X1) is longer than sh3glb2a because it includes two
alternatively spliced insertions: about 15 residues inside the BAR domain after the
"SASA" motif and a proline/serine-rich stretch in the linker before the SH3 domain. These
insertions lower the whole-length identity; the domains themselves are about equally
conserved.

Per region of human SH3GLB2 (identical residues / region length):

Region (UniProt features of Q9NR46)	sh3glb2a	sh3glb2b	gar
Membrane-binding amphipathic helix (1-27)	85.2%	85.2%	33.3%
BAR domain (24-287)	76.9%	78.8%	80.3%
Coiled coil (206-240)	85.7%	91.4%	91.4%
SH3 domain (335-395)	88.5%	86.9%	86.9%

The low gar value for the N-terminal helix reflects a different N-terminus in the gar gene
model (only 21 of 27 positions align), not necessarily a real difference. Both zebrafish copies
end in the same C-terminal SH3 sequence as human SH3GLB2 (...GKVPVTYLELLS).
Both copies keep the N-terminal amphipathic helix, the full BAR domain and the SH3 domain of endophilin-B2.

Relative rate (gar outgroup). Of 374 gar positions aligned in both copies, 36 changed only
in sh3glb2a and 23 only in sh3glb2b (chi2 = 2.86, not significant at 0.05).

Expression

Whole-embryo time course (E-ERAD-475, median TPM). sh3glb2b is strongly maternal (71 TPM
in zygote, 69 at 2-cell) and stays at 10-15 TPM from segmentation through day 5. sh3glb2a is
low in zygote (2 TPM), rises to 27-28 TPM at blastula stages (so it is transcribed early), falls
to 2-3 TPM from 75% epiboly onward and stays at 2-4 TPM through day 5.
sh3glb2b is the predominant copy in the zygote and from segmentation onward; sh3glb2a is comparable to sh3glb2b only at blastula stages.

Bgee calls (only "expressed" calls are returned; absence is not proven absence).
sh3glb2a has 20 calls, sh3glb2b 27; every sh3glb2a entity also has a sh3glb2b call. The seven
entities called only for sh3glb2b (somite, presomitic and tail-bud paraxial mesoderm, cleaving
embryo, caudal fin, integument, mesonephros) come mostly from Affymetrix microarray data, and
sh3glb2a has no Affymetrix calls at all.
Adult tissues are shared: both copies are called in brain, retina, heart, muscle, gill, intestine, liver, spleen, testis and ovarian follicle.

Gar (pre-duplication state). Gar SH3GLB2 has 14 calls, highest in eye (94.6) and brain
(91.9), then larva, skin, embryo, liver, mesonephros, testis, ovary, bone, heart, gill, muscle
and intestine: broad, with neural maxima, like both zebrafish copies.

ZFIN. High-throughput in situ records (ZDB-PUB-040907-1): sh3glb2b in central nervous
system and eye/lens from 14-19 somites to long-pec; sh3glb2a annotated only as "whole
organism" (1-cell to pec-fin). No published expression study of either copy.

Synteny

sh3glb2a is on chr8 (2.76 Mb) and sh3glb2b on chr5 (31.64 Mb). Method as for mapre3a/mapre3b:
neighbours within 1.5 Mb, their teleost-level (Teleostei, Osteoglossocephalai or Clupeocephala)
zebrafish paralogues, and where the partners lie.

Conserved neighbours: three teleost-level paralogue pairs flank both copies, found in both
directions: zdhhc12a/zdhhc12b, slc25a25a/slc25a25b and eeig1a/eeig1b, all within about 180 kb of
sh3glb2a on chr8 and within about 200 kb of sh3glb2b on chr5.
Chromosome level: 5 teleost-level pairs from the sh3glb2a neighbourhood have their partner on
chr5, and 7 from the sh3glb2b neighbourhood have their partner on chr8.
Three neighbouring gene pairs (zdhhc12, slc25a25, eeig1) are duplicated alongside sh3glb2, so the two copies sit in paralogous chromosome segments.

Interpretation

Protein: both copies keep the endophilin-B architecture and evolve at statistically similar
rates. No evidence of protein-level divergence.
Expression: largely overlapping and broad in adults, like gar SH3GLB2; the main difference is
quantitative (sh3glb2b higher in the zygote and from segmentation onward). This is
compatible with dosage retention, redundancy, or quantitative subfunctionalization; no
functional data exist for either copy.