Brain-specific homeobox Bsx specifies identity of pineal gland between serially homologous photoreceptive organs in zebrafish

We isolated EGFP-positive retinal rod and pineal photoreceptor cells by using fluorescence-activated cell sorting from the Tg(rho:egfp) and Tg(exorh:egfp) transgenic zebrafish lines, respectively (Supplementary Fig. 1a–i). Comparison of their gene expression profiles by ordered differential display method^29,30 revealed two pineal photoreceptor-specific genes encoding transcription factors, brain-specific homeobox (bsx) and insulinoma-associated 1a (insm1a) (Supplementary Fig. 1j, k).

Whole-mount in situ hybridization analysis of zebrafish larvae demonstrated that bsx is expressed in several brain regions including the pineal gland and the hypothalamus (Supplementary Fig. 2), an expression pattern similar to that of mouse Bsx, which is localized to the pineal gland, telencephalic septum, mammillary bodies and arcuate nucleus in the hypothalamus³¹. The zebrafish pineal gland is composed of two major types of neurons: photoreceptor cells and projection neurons, the latter innervating other areas of the brain²⁵. At larval stages, exorh-positive photoreceptor cells occupy the dorsomedial portion of the pineal gland (Fig. 1a, f), while the projection neurons expressing pax6²⁵ locate laterally in the gland (Fig. 1b, g). bsx transctipts were detected in both medial and lateral regions of the pineal gland (Fig. 1c, h), indicating that bsx is expressed in both photoreceptor cells and projection neurons. Expression of bsx was also detected in the parapineal organ, which is located unilaterally to the left-side of the pineal gland (Fig. 1c, h, arrowhead). On the other hand, insm1a transcripts were detected widely in the developing central nervous system (Supplementary Fig. 2), and therefore we focused on bsx functions.

The role of bsx in pineal development was assessed by knocking down bsx function with morpholino antisense oligo (MO) (Fig. 2). Injection of bsx MO induced no gross morphological abnormality in the body (Fig. 2c) but caused significant reduction in size of the larval pineal gland at 3.5 dpf (Fig. 2d, e). The size reduction of the pineal body did not recover at later developmental stages (9.5 dpf; Fig. 2d, e), indicating that the bsx morphant exhibited a defect in pineal organogenesis rather than a transient delay of the pineal development. In addition, knockdown of bsx often led to emergence of pigmented cells within the pineal gland (Supplementary Fig. 3) and caused invasion of dermal melanophores onto its dorsal surface (Supplementary Fig. 3e, f), which normally forms a translucent pineal window^32,33. These observations indicate a developmental role of bsx facilitating efficient light transmission to the pineal photoreceptor cells.

To investigate the role of Bsx in pineal-specific differentiation of photoreceptor cells, Bsx was depleted in Tg(exorh:egfp) or Tg(rho:egfp) transgenic line. Knockdown of Bsx reduced EGFP fluorescence signals in the pineal gland of Tg(exorh:egfp) larvae (Fig. 2f, g). The fluorescence intensity decreased to about 10% of the control at 3.5 dpf and slightly recovered at later stages (to ~25% and ~27% at 6.5 and 9.5 dpf, respectively) (Fig. 2g). Furthermore, Bsx depletion induced ectopic EGFP expression in the pineal gland of Tg(rho:egfp) larvae (Fig. 2h, i), in which EGFP fluorescence is normally detected only in retinal rod photoreceptor cells²⁸. The ectopic EGFP expression was detected in all the larvae examined (n = 21) at 3.5 and 6.5 dpf but markedly decreased at 9.5 dpf (Fig. 2h, i). In Tg(exorh:egfp);Tg(rho:ntr-mCherry) double transgenic larvae (Fig. 2j), Bsx depletion similarly induced ectopic expression of the rho:ntr-mCherry transgene in a subset of pineal neurons, which were mutually exclusive from those retaining the exorh:egfp transgene expression (Fig. 2k). Consistently, whole-mount in situ hybridization of bsx morphants revealed severe reduction of endogenous exorh transcripts (Fig. 3a, f) and contrasting emergence of rho transcripts in the pineal gland (Fig. 3b, g). These effects of Bsx depletion were also confirmed in a bsx knockout mutant (bsx^m1376 line³⁴; Supplementary Figs. 4 and 5) and by using another MO (bsx e1i1 MO, Supplementary Fig. 6). These results raise the possibility that Bsx plays a commanding role in determination and/or differentiation of the pineal photoreceptor neurons by specifying them differently from the retinal counterparts.

Transcriptional network for the pineal development was studied by examining genetic interactions between Bsx and other transcription factors. We found that pineal expression of otx5 and pax6b (Fig. 3d, e) was downregulated by Bsx depletion (Fig. 3i, j), suggesting that Bsx functions genetically upstream of these factors. In contrast, Bsx knockdown had no marked effect on its own transcript level in the pineal gland (Fig. 3c, h), while the bsx-positive area was reduced, consistently with the reduction of the pineal size (Fig. 2d, e). MO-based knockdown of Otx5 severely reduced exorh expression (Fig. 3k) with a marginal effect on bsx expression in the pineal gland (Fig. 3m), further supporting the idea that Bsx functions upstream of Otx5. Knockdown of Flh resulted in almost complete elimination of exorh and bsx signals at the presumed position of the pineal gland (Fig. 3n, p), in line with the pivotal role of Flh for pineal organogenesis²⁵, whereas Bsx depletion had no significant effect (by bsx knockout) or only a marginal effect (by MO-based knockdown) on flh expression at 48 hpf (Supplementary Fig. 5c, f). The results indicate that Bsx acts as a mediator downstream of Flh signaling in the developing pineal gland. On the other hand, substantial bsx expression remained in the parapineal organ of flh morphant (Fig. 3p, arrowhead). This observation is consistent with a previous report that the parapineal organ develops normally in flh mutant³⁵ and suggests that bsx expression in the parapineal organ is regulated in a manner independent of Flh activity. Unlike bsx morphant, neither otx5 nor flh morphant induced rho expression in the pineal gland (Fig. 3l, o), emphasizing the unique role of Bsx in establishing the pineal specificity.

To examine whether Bsx directly controls pineal-specific gene expression, we conducted a luciferase assay with a reporter construct harboring a 147-bp fragment of exorh promoter, which can direct pineal-specific gene expression in vivo²⁸. Although Bsx alone had little effect on the promoter activity (Fig. 4a), co-expression of Bsx with Otx5, a known regulator for pineal gene expression¹⁸, activated the transcription by 50-fold (Fig. 4a). In contrast, no synergistic action was observed when Otx5 was replaced by Crx (Fig. 4a), which mainly participates in regulation of retinal gene expression^15,18. These observations suggest that the combinatorial action of Bsx and Otx5 serves as a definitive cue for the pineal photoreceptor-specific gene expression. This idea was further examined by ectopic expression of bsx in retinal photoreceptor cells, which intrinsically express otx5¹⁸ but not bsx (Supplementary Fig. 2). Injection of rho:bsx construct into Tg(exorh:egfp) embryos induced EGFP expression in the retinal cells (Fig. 4b, c). These results demonstrate that Bsx transactivates pineal photoreceptor-specific genes in combination with Otx5 existing in both pineal and retinal photoreceptor cells.

To delineate the mechanism of Bsx action, we determined DNA-binding specificity of Bsx by a selection and amplification binding (SAAB) assay using a glutathione S-transferase (GST)-Bsx fusion protein. Repetitive rounds of selection by GST pull down and PCR amplification revealed an octanucleotide motif, TAATCGGT, as a consensus sequence for Bsx binding (Fig. 5a and Supplementary Fig. 7). This sequence harbors a pineal regulatory element PIRE (TAATC/T), which has been characterized as a putative binding site for Crx/Otx²⁴. To confirm the binding of Bsx to the sequence determined by SAAB, we performed an electrophoretic mobility shift assay (EMSA) with BSXRE probe, a synthetic oligonucleotide probe (36-mer) containing the highest matching sequence (Fig. 5b; matching score 9.54, see the legend for its definition). GST-Bsx protein showed a binding activity to BSXRE probe (Fig. 5b), whereas no protein-DNA complex was detected with a mutated probe (Fig. 5b, BSXRE-mut probe). These results indicate that Bsx binds to the TAATCGGT motif.

Based on the matching scores with the SAAB matrix (Fig. 5a), we found six potential sites for Bsx binding (P1–P6, matching scores 5.43–6.35) in the 147-bp exorh promoter region (Fig. 5c, d). By performing a competitive EMSA using BSXRE probe, we found that P3 sequence remarkably competed with BSXRE for Bsx binding, while P1 and P4 had weaker but evident competitive activities (Fig. 5d). It should be noted that these three sites harbor TAAT core motif and the others do not (Fig. 5d), indicating that the TAAT core is essential for Bsx binding. The very low, if any, competitive activity of P2 (Fig. 5d) implies that PIPE (TGACCCCAATCT), a cis-acting element participating in pineal-specific gene expression²⁸, is less likely to serve as a Bsx-binding site. We further examined four TAAT-containing sequences that locate more distally in exorh upstream region (D1–D4 in Fig. 5c). D2, D3, and D4 had substantial Bsx-binding activities, whereas D1 showed a much lower activity despite harboring the TAAT core sequence (Fig. 5d).

EMSA using a probe of the 147-bp exorh promoter region (Fig. 5e) showed that Bsx directly interacts with the pineal-specific promoter (lanes 1 and 2). Bsx binding was remarkably reduced when P3 site was mutated, whereas mutations in P1 and P4 had only marginal effects on the formation of Bsx-bound complexes (Fig. 5e, lanes 3–5). Introduction of double mutations into P3 and P4 sites almost completely abrogated the Bsx-bound complex formation (Fig. 5e, lane 8). These results demonstrated that Bsx binds to the exorh promoter primarily through P3 and P4, both of which harbor the known cis-element PIRE. PIRE has long been considered as a binding site for Crx/Otx-family proteins, but the present data indicate that certain types of these pineal-specific cis-elements serve as binding sites for Bsx.

RIKEN WT, EkkWill, and ABTL were used as wild-type strains. In the present study, Ex(-1055) transgenic line (ja1Tg strain), in which EGFP was driven by a 1055-bp fragment of exorh promoter²⁸, was renamed Tg(exorh:egfp) according to the zebrafish nomenclature guideline. Rh(-1084) transgenic line (ja2Tg strain), in which EGFP was driven by a 1084-bp fragment of rho promoter²⁸, was also renamed Tg(rho:egfp). The Tg(rho:NTR-mCherry) line (ja83Tg strain) was generated by introducing a transgene having the 1084-bp rho promoter²⁸ and the nitroreductase (NTR)-mCherry fusion gene⁴². The bsx^m1376 mutant line³⁴ was a kind gift from Dr. Wolfgang Driever. All research described here adhered to local guidelines of the University of Tokyo, and all appropriate ethical approval and licenses were obtained from Institutional Animal Care and Use Committees of The University of Tokyo.

To minimize potential cross reactivities to homologous genes, digoxigenin (DIG)-labeled RNA probes (216–306 bases in length) were designed for 3′ UTRs of the zebrafish mRNAs for exorh (NM_131212.2↗, nucleotide 1112 – 1371), rho (AF109368↗, 1207–1422), insm1a (NM_205644.1↗, 1293–1588), otx5 (NM_181331↗, 1300–1601) and bsx (GRCz11, chromosome 10, nucleotide 29,890,307–29,890,615). Because zebrafish pax6b mRNA has a short 3′ UTR, the pax6b probe was designed for the region covering 168 bases of the ORF and the following 134 bases of the 3′ UTR (NM_131641↗, 1407–1708). Each DNA fragment was cloned into the EcoRV site of a modified plasmid vector, termed pGEM15H, in which the multiple cloning sites of pGEM 3Zf(+) was substituted by the EcoRI/HindIII fragment excised from pBluescript II SK(+). The RNA probes were synthesized by using DIG RNA Labeling Kit (Roche Diagnostics).

Whole-mount in situ hybridization was performed as described previously⁴³. Briefly, embryos at 48 or 78 hpf were fixed in 4% PFA in phosphate-buffered saline overnight at 4 °C and dehydrated in methanol. The samples were rehydrated and pre-treated with proteinase K and fixed again in 4% PFA in phosphate-buffered saline for 20 min at room temperature. The pre-treated larvae were hybridized with the DIG-labeled RNA probes, and the hybridization signals were visualized by staining with BM purple (Roche) or NBT/BCIP mixture. After staining, the specimens were immersed in 80% glycerol and examined with an upright microscope (Axioplan2, Zeiss).

Sequences of morpholinos (MO) are as follows: bsx MO, 5′-ATTTA ACGCA ATTAC CTGTG TTTCC-3′; bsx e1i1 MO, 5′-TATAG GCTGC ACTTA CCAGA GGTGA-3′; otx5 MO, 5′-CATGA CTAAA CTCTC TCTCT CTCTC-3′; flh MO, 5′-AATCT GCATG GCGTC TGTTT AGTCC-3′. The bsx MO was synthesized by GeneTools (Philomath, USA). The otx5 and flh MOs were obtained from Open Biosystems (Huntsville, USA). These MOs synthesized were dissolved in distilled water containing 0.05% phenol red at a concentration of 8 mg/ml (bsx MO), 4 mg/ml (bsx e1i1 MO) or 0.5 mg/ml (otx5 and flh MOs). One-cell-stage embryos of wild-type or transgenic fish were injected with 2 nl of the MO solutions into the yolk immediately beneath the cell body. For control experiments, we used the standard control morpholino (GeneTools) in the same condition. The effects of the MO injection on bsx transcripts were evaluated at 48 hpf by RT-PCR with a pair of BSX3F (5′-GACGC ACGGA TTGTT CGC-3′) and BSX3R (5′-GTTGA TTTAG TGTAA TAATA GC-3′) primers. ZBAF1 and ZBAR2 primers⁴⁴ were used for RT-PCR of β-actin 2 (actb2).

For experiments using bsx^m1376 mutants, embryos at 48 or 78 hpf were collected from crossing pairs of the heterozygotes, and dissected into anterior and posterior segments on ice; the posterior segments were used for genotyping of bsx^m1376 allele³⁴, while the anterior segments were soaked in RNAlater and stored at 4 °C. After genotyping, four or six of the anterior segments for each genotype were pooled as a biological replicate. For MO-injected embryos, four or six of the anterior segments were pooled as a biological replicate as well. Total RNA was extracted and purified with RNeasy Mini Kit (Qiagen). An equal amount of RNA from each sample was reverse-transcribed into cDNA with a mixture of the oligo (dT)₁₅ primer and the random primers with GoScript^TM Reverse Transcriptase (Promega). The reverse-transcribed cDNA was then subjected to quantitative PCR using GoTaq qPCR Master Mix (Promega) and the StepOnePlus^TM Real-time PCR system (Applied Biosystems) according to the manufacturers’ protocols. Relative expression levels were calculated with a relative standard curve created with serially diluted cDNA samples of the 48-hpf anterior segments. The expression levels were normalized to β-actin 2 (actb2) expression levels in all the figures. The primers used for quantitative PCR are 5′-TGTG TTCAT GACG GTTCC TG-3′ and 5′-ACACC GTTGA CAACA TGCAG-3′ for exorh, 5′-ACAGG CTATC CCGTG TTCTG-3′ and 5′-GCCAG TACTG CATCT TGTGC-3′ for flh, 5′-TCCGA GACCA CACAG CG-3′ and 5′-CTGCT TGTTC ATGCA GATG-3′ for rho, and 5′-GGCAA TGAGA GGTTC AGGTG-3′ and 5′-GTGGT ACCAC CAGAC AATAC-3′ for actb2.

A full-length coding sequence of bsx was obtained by RT-PCR from adult zebrafish pineal gland cDNA with a pair of primers, BSX3F and BSX3R, and cloned into pCR2.1-TOPO vector (Invitrogen). To generate an expression construct rho:bsx, the coding sequence of bsx was PCR-amplified from the cDNA plasmid with primers, 5′-CAAGC TTCAT GAATC TGAAC TACAC G-3′ and 5′-CTAGA GCGGC CGCTA CTAGA GTAAA TGTTC-3′. Then the PCR product was digested with BspHI and NotI, and inserted into the NcoI-NotI site of the Rh(-1084) plasmid²⁸. The rho:bsx plasmid was linearized by SacI digestion, and dissolved at a concentration of 25 ng/µl in distilled water containing 0.05% phenol red. The linearized DNA solution was microinjected into the cell body of one-cell stage embryos.

Full-length coding sequences of the zebrafish bsx, otx5, and crx were inserted into pcDNA3.1/V5-His-TOPO expression vector (Invitrogen). To generate an exorh promoter-reporter construct, the 147-bp fragment upstream from the exorh translation initiation site was amplified by PCR with expro147-Fw (5′-TCTCT GGATT ATCCC CCTGT C-3′) and expro147-Rv (5′-GATGG AGAAG TGGAC GATCG-3′) primers from the template plasmid, and inserted into the SmaI site of pGL3-Basic vector (Promega). Human embryonic kidney 293 T cells were plated at 4 × 10⁵ cells per well in 12-well plates, and transfected with 250 ng of the reporter construct, 250 ng of each expression construct and 0.5 ng of pRL-CMV internal control vector (Promega) using Lipofectamine 2000 reagent (Invitrogen). The total amount of transfected DNA was adjusted to 750 ng/well by mixing with pcDNA3.1/V5-His empty vector. Luciferase reporter activity was measured 24 h after the transfection by using Dual-Luciferase Reporter Assay System (Promega), although the firefly luciferase activity was not normalized by Renilla luciferase activity.

The pineal gland and the retina were dissected from adult fish of the Tg(exorh:egfp) and Tg(rho:egfp) transgenic lines, respectively. These tissues were digested in 0.25% trypsin/1 mM EDTA in phosphate-buffered saline (0.8% NaCl, 0.02% KCl, 20 mM phosphate, pH 7.3) at 37 °C for 30 min, and then the reaction was stopped by addition of an equal volume of phosphate-buffered saline containing 2% (w/v) soybean trypsin inhibitor and 20% (v/v) fetal bovine serum. After filtration through a 35-µm nylon mesh, the dissociated cells were sorted by using EPICS ELITE ESP cytometer equipped with Expo32 software (Beckman Coulter). Forward scatter, side scatter and 525-nm green fluorescence intensity were monitored for gating. Sorted cells were directly collected into 1.5-ml microtubes containing TRIzol reagent (Invitrogen) for subsequent extraction of total RNA.

Ordered differential display was carried out as described³⁰ with several modifications. For each sample, total RNA (~45 ng) was used for the synthesis of double-strand cDNA by using Superscript Plasmid System (Invitrogen) with 0.2 µM of T-primer (5′-CGCAG TCGAC CGTTT TTTTT TTTTT-3′). The cDNA samples were digested by RsaI for 2 h at 37 °C and then ligated with adapters (a mixture of two oligonucleotides, 5′-TGTAG CGTGA AGACG ACAGA AAGGG CGTGG TGCGG AGGGC GGT-3′ and 5′-ACCGC CCTCC G-3′) by using DNA Ligation Kit Ver.2 (TaKaRa Shuzo, Kyoto, Japan). Using the cDNA sample as a template, first PCR amplification was performed in PC2-buffer [50 mM Tricine-KOH, 16 mM (NH₄)₂SO₄, 3 mM MgCl₂, 150 µg/ml bovine serum albumin (BSA), pH 8.7 at 20 °C] with 250 µM dNTPs, 0.1 µM of each DAd-primer (5′-TGTAG CGTGA AGACG ACAGA A-3′) and T-primer, and 6.25 units KlenTaq1 polymerase (Ab peptides, St. Louis, MO) in a total volume of 25 µl. The amplification reaction was composed of 18 cycles of 94 °C for 40 s, 65 °C for 30 s, and 72 °C for 90 s. The PCR products were then diluted 25-fold in QT buffer (10 mM Tris-HCl, pH 8.0), and the aliquot (1 µl) was used as a template for the second PCR amplification in combination with a pair of two-base-anchored primers, 5′-GCGTG GTGCG GAGGG CGGT(G/T) CNN-3′ and 5′-CGCAG TCGAC CGTTT TTTTT TTTTT (A/C/G)N-3′, in which N represents any base. The combination of the former (16 variants) and the latter (12 variants) primers yields 192 subsets of the amplified products. Amplification reaction was performed in PC2-ODD buffer [20 mM Tricine-KOH, 16 mM (NH₄)₂SO₄, 2.5 mM MgCl₂, 150 µg/ml BSA, pH 8.7 at 20 °C] with 250 µM dNTPs, 0.2 µM each of the primers and 2.5 units KlenTaq1 polymerase in a volume of 10 µl for 25 cycles. Then the entire reaction mixture was subjected to 7.5% native PAGE, and the PCR product bands stained with SYBR Green I (Molecular Probes) were quantified by FLA-2000 Bioimage analyzer (Fuji Film, Tokyo, Japan). The selected bands were cut out from the gel and the DNA was eluted from the gel pieces by incubation in 50 µl of QT buffer for 2 h at 55 °C. The eluted DNA was re-amplified by PCR with a pair of Ad2-primer (5′-GCGTG GTGCG GAGGG CGGT-3′) and T-primer, and then the products were separated by agarose gel electrophoresis and purified by phenol extraction and ethanol precipitation. Nucleotide sequences of the purified DNA fragments were determined by direct sequencing with Ad2- and T-primer.

A full-length coding sequence of the zebrafish bsx was obtained by PCR amplification with a pair of primers (5′-CCCGG AATTC ATGAA TCTGA ACTAC ACG-3′ and 5′-CTGCA GGTCG ACTAG AGTAA ATGTT CAGG-3′) and the bsx cDNA plasmid as a template. The amplified fragment was digested with EcoRI and SalI, and inserted into EcoRI-SalI site of pGEX-5X-1 vector (GE Healthcare). Recombinant GST-Bsx protein was expressed in Escherichia coli BL21 strain and purified by affinity chromatography using glutathione-Sepharose 4B resin (GE Healthcare). GST-Bsx protein bound to the resin was washed three times with a Mg²⁺- and ATP-containing buffer (50 mM Tris-HCl, 10 mM ATP, 10 mM MgSO₄, pH 7.5) at room temperature to eliminate contaminating proteins such as bacterial chaperon proteins. After overnight elution with 50 mM Tris-HCl (pH 8.0) containing 5 mM glutathione (reduced form) at 4 °C, excess glutathione was removed by gel filtration with a PD-10 column (GE Healthcare). GST protein for control experiments was similarly prepared.

SAAB experiments based on the GST pull-down method were performed as described previously⁴⁵ with modifications. For the first round of binding sequence selection, we used double-stranded random oligonucleotides (5′-GAGTC CAGCG AATTC TGTCG-N₂₀-GAGTC CTCGA GAGTG TCAAC-3′) having a pair of 20-bp invariant flanking regions⁴⁶, which enable PCR amplifications under more stringent conditions and hence prevent concatemerization of the PCR products. The random oligonucleotides (500 ng) were incubated with GST-Bsx (500 ng) or GST (263 ng) protein pre-bound to glutathione-Sepharose 4B beads (5 µl bed volume) in 500 µl of SAAB binding buffer (20 mM Tris-HCl, 50 mM KCl, 0.5 mM EDTA, 1 mM DTT, 50% glycerol, pH 8.0) containing 20 µg/ml bovine serum albumin and 2 µg/ml poly(dI-dC) at 4 °C for 1 h. The pelleted beads were collected by centrifugation at 500 × g for 1 min, washed three times with 1 ml of the SAAB binding buffer, and then resuspended in 30 µl of distilled water. A 10 µl aliquot of the suspension was used as a template for PCR amplification by AmpliTaq Gold DNA polymerase (Applied Biosystems) with a pair of primers, 5′-GTTGA CACTC TCGAG GACTC-3′ and 5′-GAGTC CAGCG AATTC TGTCG-3′, in a 25 µl total reaction volume. The cycling protocol of the PCR reaction was 30 cycles of 94 °C for 30 s and 68 °C for 30 s. A 10-µl aliquot of the PCR product was used for the subsequent round of selection. After five repetitive rounds, the amplified products were cloned into the pCR2.1-TOPO vector and sequenced. Among 89 sequences obtained, eight sequences showing far lower similarities to the others were eliminated from the determination of the consensus binding sequence. Two sequences having multiple potential binding sites were also eliminated.

EMSA was performed by using DIG Gel Shift Kit (Roche Diagnostics) according to the manufacturer’s instructions. Short double-stranded oligonucleotides for BSXRE probe, BSXRE-mut probe or the competitors were prepared by annealing a pair of synthetic oligonucleotides. Native or mutated DNA probe corresponding to the 147-bp exorh promoter was obtained by PCR amplification using a template of the plasmid containing the promoter sequence with or without introduction of mutations. Mutations of P3 and P4 sites (TAATCTG to TACGATG) were introduced into the plasmid having a 1.1-kbp sequence of the exorh promoter by using QuikChange Site-Directed Mutagenesis Kit (Stratagene). Probes without P1 mutation were amplified with expro147-Fw (5′-TCTCT GGATT ATCCC CCTGT C-3′) and expro147-Rv (5′-GATGG AGAAG TGGAC GATCG-3′) primers, while those carrying the P1 mutation were obtained by using expro147-mut-Fw (5′-TCTCT ATCGT ATCCC CCTGT CTG-3′) instead of expro147-Fw. Then the PCR products were purified with QIAquick PCR Purification Kit (QIAGEN) and subjected to digoxigenin-labeling procedures. For binding reactions, 2.0 pmol of GST-Bsx (107 ng) or GST (56 ng) was incubated with 32 fmol of DIG-labeled oligonucleotides or 8 fmol of DIG-labeled DNA probes.

The statistical data reported include results from at least three biological replicates, whose numbers are stated in each figure. Statistical analyses (Welch’s two-sided t-test, Games-Howell two-sided test, and Fisher’s two-sided exact test) were performed in R (Figs. 2, 4, Supplementary Fig. 3) or Microsoft Excel (Supplementary Figs. 5, 6).

Further information on research design is available in thelinked to this article. Nature Research Reporting Summary

All the data are available on request from the authors.