Assembly of the ‘ligand trap’ (LT) vector

Construction of the ligand trap (LT) vector (Figure 1a) involved an extensive series of sub-cloning events and configuration comparisons, with the chosen vector described below. The 3X FLAG [23], Strep II [24] and 6X His tags were combined into a single triple-tag cassette, collectively referred to as the ‘FSH’-tag, and inserted in front of a minimal GAL4 DNA-binding domain (DBD: residues 1–132). Following the Gal4 DBD sequence, a series of restriction sites, was used for the in-frame addition of NR cDNAs (hinge + LBD). To reduce potential lethality that might be caused by constitutive expression of the GAL4-LBD fusion proteins, a zebrafish heat shock 70 gene promoter [25] was used for inducible expression in any tissue and at any developmental stage. Unique restriction sites flanking the hsp70 promoter can be used to swap in tissue-specific promoters, if so desired.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1. The Ligand Trap (LT) system.

(a) Schematic diagram of the multi-component ligand trap (LT) construct. Upon heat pulse, the zebrafish hsp70 promoter directs ubiquitous expression of the GAL4 DNA-binding domain (DBD) fused in-frame to a human nuclear receptor ligand-binding domain (LBD) and an affinity tag cassette (FSH-tag). Upon binding of this fusion protein to the GAL4 UAS (upstream activating sequence) response elements, in the presence of active hormone and cofactors, expression of the reporter gene (nuclear enhanced Green Fluorescent Protein (eGFP)) occurs. Expression of nuclear GFP is used to monitor receptor ligand sensor activity in a cell- and tissue autonomous manner in live zebrafish. The second component makes use of the tags to co-purify bound hormones or cofactors. I  =  insulator elements; pA  =  polyadenylation signal; NLS  =  nuclear localization signal. (b) Western blots of GAL4-NR fusion proteins. Embryos (F2; 72 hpf) were heat pulsed for 30 min at 37°C and recovered for 1 h at room temperature. 10 embryos were pooled and lysed in 50 µl of FSH buffer (see material and methods) followed by adding SDS buffer and boiling. Proteins were detected using the FLAG-M2 antibody. (c) Time course of fusion protein expression. TRβ embryos (F2; 72 hpf) were heat induced as in 1b) and recovered for the times indicated. Each sample contained 10 embryos.

https://doi.org/10.1371/journal.pone.0009797.g001

Also within the vector is an enhanced GFP (eGFP; includes a nuclear localization signal) reporter gene that is expressed under the control of a 14x UAS_GAL4-containing promoter. Expression of this reporter requires binding and activation of the Gal4-NR-LBD fusion protein. Bracketing the GAL-LBD fusion gene and the eGFP reporter are insulator elements that were introduced to discourage the influence of neighboring gene regulatory elements. Finally, I-SceI meganuclease recognition sites were added at the ends of the vector to facilitate integration into the genomes of transgene and meganuclease co-injected one-cell stage embryos [26].

Induction of LT fusion protein expression

To test the heat inducible expression and activity of LT fusion proteins, and to determine the best parameters for activity screening, we generated stable transgenic lines for each human nuclear receptor by co-injecting the LT-vector together with SceI meganuclease into one-cell stage zebrafish embryos. Injected F0 fish were raised to adulthood and crossed with wild type fish to identify germline transformed animals. Positive progeny (F1) were identified either by target PCR or GFP screening (see Material and Methods). At least two independent lines for each nuclear receptor were obtained, unless otherwise noted. Transgenic LT embryos (F2) were heat pulsed at different developmental stages, temperatures and durations. Western blot detection using a M2-Flag antibody that recognizes the FSH-tag of each LT fusion protein revealed uniquely sized proteins with expected molecular weights (Figure 1b). Figure 1c shows the robust expression observed in homozygous transgenic (F2) Thyroid Receptor-β (TRβ; NR1A2) embryos at successive time points following a temperature shift to 37°C. As with most LT fusion proteins, peak expression was observed ∼1 hr following the heat pulse (Figure 1c). A thirty minute heat pulse at 37°C also proved adequate for the production of robust GFP fluorescence in the presence of endogenous or exogenously added ligands (see below).

LT fusion proteins interact with fish ligands and cofactors

Different NR-derived lines should exhibit unique GFP responses during development in response to their respective ligand and cofactor distributions. To verify this, we induced expression of our Peroxisome Proliferator-Activated Receptor-gamma (PPARγ; NR1C3), Retinoic acid-related Orphan Receptor-gamma (RORγ: NR1F3), Neuron-derived Orphan Receptor-1 (NOR1: NR4A3) and Rev-erb alpha (Rev-erbα: NR1D1) transgenic lines and documented their patterns of GFP expression. The PPARγ embryos show strong GFP expression in the tail bud epidermis, as well as the brain, posterior spinal cord and heart (Figure 2; left panel). RORγ embryos show ubiquitous GFP expression over the entire embryonic epidermis during the first three days of development, along with brain and retina later on (Figure 2 middle panel). GFP expression in NOR1 embryos occurs in the epidermis and CNS (Figure 2; upper right picture). In contrast, Rev-erbα LT embryos show no GFP expression (Figure 2; lower right), consistent with its role as a transcriptional repressor [27].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 2. Human NRs interact with fish ligands and cofactors.

Transgenic NR zebrafish show unique GFP patterns at different stages of development. Activity patterns of the PPARγ (first panel), RORγ (middle panel), NOR1 (upper right picture) and Rev-erbα (lower right picture) LT constructs. Embryos were heat pulsed at 37°C for 30 min and images taken 24 h later. PPARγ (48 hpf, F2) embryos show GFP expression in cells of the epidermis and heart, as well as in the posterior spinal cord. Strong GFP expression occurs in the epidermis and retina of RORγ embryos (72 hpf, F2). NOR1 embryos (F3) express GFP in the posterior spinal cord, hatching gland, epidermis and yolk syncictial layer (24 hpf, F2). Rev-erbα embryos show no GFP expression. Overlay pictures of bright field and GFP (75% transparent) are shown. Views are lateral with anterior to the left.

https://doi.org/10.1371/journal.pone.0009797.g002

To verify that GFP expression effectively reflects endogenous signaling activity, we injected a morpholino (MO) oligo-nucleotide, complementary to the translational start site of the transgene, into one-cell stage PPARγ, RORγ and TRβ embryos (Figure S1). GFP expression was lost or dramatically reduced in these MO injected embryos, even in the presence of control agonist.

Further documentation of these sites of hormone activity, and those of the other LT lines, should provide many new insights into novel NR functions and relationships during development. They will also serve as useful tools to genetically and chemically probe corresponding cellular and developmental processes, and to mark cells and tissues for co-expression and lineage analyses.

Purification of an in vivo TRβ protein complex from early stage embryos

While a large number of nuclear receptor cofactors have been identified, their means of identification have generally been limited to immunoprecipitation from cultured cell extracts or yeast two-hybrid screens. The ability to identify cofactors throughout development or in different tissues where NRs are responding to unique cofactors, and/or ligands, would lead to the identification of many new cofactors, cofactor complexes and associated functions. To validate the usefulness of our tags, we induced expression of the TRβ fusion protein in 5–7 hpf embryos (Blastula/Gastrula stage) and purified the receptor protein complex from a whole embryo extract (see Methods for details).

Figure 3a shows the silver-stained SDS PAGE gel of the final elution of the TRβ purification. Several proteins (bands 1–5) in addition to the bait protein (band 6; Gal-hTRβ) can readily be seen. These and others were identified by MALDI-ToF and Orbitrap mass spectrometry (Dataset S1). Identified proteins included several known nuclear receptor interacting proteins, such as specific heat shock proteins and lung resistance protein [28], [29]. Nuclear proteins were in less abundance, due most likely to the use of whole cell extracts for this purification.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 3. Co-purification of TRβ cofactors.

(a) Affinity purification of TRβ from zebrafish embryos. Blastula-/Gastrula-stage embryos were heat induced for 20 min and recovered for 2 h at room temperature. After purification, the eluate was run on an 8% SDS PAGE and then silver stained. Protein bands (indicated as 1–6) were cut and in gel digested followed by mass spectrometry. Identified protein IDs are shown. M =  protein marker. (b) Immunodetection of HIPK3. Western blotting using a HIPK3 antibody (lower panel) shows that a subfraction of the protein co-purifies with the bait protein. Presence of the TRβ protein in the same fractions, detected using Flag-M2 antibody, is shown above. (c) HEK293 cells were transfected with FLAG-HIPK3 and one of three different Gal expression constructs containing the FSH-tag: Gal, Gal-TRβ (aa 189–461) and TRβ_ΔAF2 (aa 189–451). After 48 h of transfection, the cells were harvested in IP buffer and Gal and Gal fusion proteins immunoprecipitated using Streptactin Sepharose. Western blotting using a FLAG M2 antibody shows HIPK3 in the Gal-TRβ and Gal-TRβ_ΔAF2 pull downs. (d) Transcriptional activity of TRβ proteins. 293 cells were transfected with a 2xUAS-Luciferase reporter construct (0.2 µg/well) and Gal or Gal-TRβ constructs (0.05 µg/well): Gal-TRβ (aa 189–461) and TRβ_ΔAF2 (aa 189–451) in the presence of 100 nM TRIAC. Fold activation was measured in the presence (0.05 or 0.1 µg/well) or absence of HIPK3. Luciferase values were normalized against β-Gal. (e) Effects of HIPK3 on Gal-TRβ induced repression. 293 cells were transfected with a 2xUAS-Luciferase reporter construct and either Gal, Gal-TRβ or GAL-TRβ_ΔAF2. Fold repression was measured in the presence or absence of HIPK3. Luciferase values were normalized against β-Gal. Transfections as under 3d. (f) The effect of HIPK3 was investigated in the absence or presence of the co-repressor N-CoR. Transfections were performed as in 3d using the Gal-TRβ_ΔAF2 (aa 189–451) construct. Fold repression of the TRβ mutant in the presence of HIPK3 and/or N-CoR (0.05 µg/well) are shown.

https://doi.org/10.1371/journal.pone.0009797.g003

To test the validity and relevance of one of the novel TRβ cofactors identified, we focused on the protein Homeodomain-interacting Protein Kinase 3 (HIPK3). HIPK3 is a Ser/Thr kinase that affects transcriptional regulation, cell differentiation, growth and apoptosis [30], [31], [32], [33]. Notably, it has already been shown to bind and modulate two other NRs; Steroidogenic Factor-1 (SF1) and Androgen Receptor (AR) [30], [31].

Western blot analysis of the zebrafish extract and purified fractions (Figure 3b) confirms the presence of HIPK3 in the elution. Clearly though, the majority of HIPK3 present in the whole animal extract is not bound to the TRβ bait protein, consistent with its enzymatic participation in numerous other protein complexes.

The specificity and nature of this interaction was also tested using transiently expressed GAL, GAL-TRβ or GAL-TRβ_ΔAF2 fusion proteins. The latter protein lacks the C-terminal AF2 helix, which is required for co-activator binding. The fusion proteins were expressed together with HIPK3 in human HEK 293 cells. Both full-length and AF2-deleted GAL-TRβ fusion proteins were able to pull down the HIPK3 protein (Figure 3c). Although the intensity of the HIPK3 band co-purified by the TRβ_ΔAF2 deleted protein appears less intense, this is largely due to the presence of additional co-migrating bands in the full-length TRβ pull down, which may be due to subsequent modifications of the non-deleted LBD protein. We conclude that the AF-2 helix is not required for HIPK3 complex formation.

HipK3 negatively modulates TRβ transcriptional activity

Ligand activation of TR is associated with the displacement of co-repressors and recruitment of co-activators. To investigate whether HIPK3 affects the transcriptional activity of TRβ, a cell culture co-transfection assay was employed using Gal4-TRβ and Gal-TRβ_ΔAF-2 fusion protein. Deletion of the AF2 domain of TRβ results in continued co-repressor binding even in the presence of hormone, a phenomenon referred to as ‘Resistance to Thyroid Hormone’ (RTH) [34]. This type of deletion represents the most severe genetic form of RTH.

Addition of the thyroid hormone analog, TRIAC (3,5,3′-triiodothyroacetic acid) strongly stimulates the activity of the full LBD fusion protein, but has no effect on the non-activating AF2-deleted form (Figure 3d). Addition of HIPK3 led to a modest (∼1.3X) increase in the transcriptional activation activity of the WT fusion protein (Figure 3d). No effect was seen upon addition of HIPK3 on the activation activity of the AF2 deleted protein, consistent with the need for the AF2 motif for agonist-based transcriptional activation.

As TRβ represses the expression of target genes in the absence of hormone, we also looked for effects of HIPK3 on Gal4-LBD fusion protein activity using a reporter plasmid with significant basal transcription activity, making it suitable for the observation of repression (Figure 3e and Figure S2a). As with the transcriptional activation assay with ligand present, HIPK3 had only a modest affect on GAL4-TRβ activity (∼1.3X lower). However, with the AF2 removed, HIPK3 increased transcriptional repression by an impressive 5–6 fold. This suggests that HIPK3 acts primarily as an enhancer of TRβ- mediated transcriptional repression.

To further probe the mechanistic nature of this HIPK3 affect, we examined its ability to augment repression mediated by the known TRβ co-repressor, N-CoR [35], [36], [37]. Co-expression of HIPK3 or N-CoR with the WT TRβ fusion protein increased repression of the UAS-containing reporter approximately 4- and 3- fold respectively (Figure 3f and Figure S2b). When added at the same time, a further ∼2-fold increase in repression mediated by the TRβ_ΔAF2 mutant was observed. These interactions and effects were not observed with a kinase-defective HIPK3 protein (Figure 3f). These results are consistent with the activities of other HIPK family members on target gene expression and co-repressor function [38], [39].

Zebrafish ligand trap activities are hormone-inducible

To test whether our LT lines are responsive to externally provided hormones or drugs, we treated the PPARγ line with readily available agonists and antagonists (Figure 4a–b). PPARγ fish treated with the receptor specific agonists Rosiglitazone (RGZ), Pioglitazone (PGZ) or Troglitazone (TGZ)) increased the GFP reporter response from the relatively restricted signal in tail epidermis and posterior spine to include strong expression in the CNS, heart, blood, renal tube and eye (Figure 4a). Higher magnification of the embryo head shows specific and detailed activation in the presence of drug in the brain, eye and heart (Figure 4b panels 2 and 4). Older embryos (5–6 dpf) also show strong GFP expression in other tissues including heart, blood and intestine (Figure 4b, panels 5–8). Treatment of PPARγ fish with PPARα- or PPARδ-specific agonists showed no increase in reporter activation when used at their receptor-specific EC₅₀ concentrations (Figure S3a). However, as seen in Gal4-PPAR cell based assays [40], higher concentrations of the beta or alpha agonist do result in partial activation of the PPARγ LT embryos (Figure S3a). Treatment with the selective PPARγ antagonist GW9662 either decreases or completely blocks the GFP responses to endogenous ligand(s) (Figure 4b-13 and Figure S3b). When added together with Rosiglitazone, in a classical agonist/antagonist competition experiment, GW9662 was clearly able to displace the potent PPARγ agonist (Figure S3b).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 4. Ligand trap activities are drug responsive and reveal SNRM activities.

(a) LT-PPARγ embryos show strong reporter activation in the presence of receptor specific agonist (uses as indicated). After heat pulse followed by overnight incubation with solvent or drug GFP expression in cells of the epidermis, CNS, eye, blood and heart (48 hpf; lateral view) is induced. Yellow arrow indicates GFP expression in heart. (b) Upper panel shows magnifications of embryo heads in ventral (b-1-2) and dorsal (b-3-4) orientation of 2 dpf fish treated with solvent or 50 nM Rosiglitazone (RGZ). B-2 shows reporter expression in eye (e), forebrain (fb), heart (h) and b-4 indicates GFP expression in hindbrain (hb), tectum (t) and anterior spinal cord (s). Middle panel shows drug responses in tissues of older embryos: Expression in heart (b-5; 5 dpf ventral view), in blood (b-6; 8 dpf), in the spinal cord (s) and endothelium layer ((el), b-7; 4 dpf, lateral view) and in the intestine ((I, b-8; 6 dpf). Embryos were subjected to a 30 minute heat pulse at 37°C, followed by incubation with 50 nM Rosiglitazone for 24 h. The lower panels show responses to: endogenous hormone in the tail epidermis of 3 dpf embryos treated with solvent (b-9); 500 nM Pioglitazone (b-11), 500 nM RGZ (b-12) and treatment with the antagonist GW9662 (500 nM). (c) TRβ LT embryos show tissue-specific responses to hormones and drugs. After 40 min heat induction, embryos (24 hpf) were incubated with compounds for 28 h in a 28°C incubator in the dark. Solvent (Ethanol) treated embryos show no GFP expression. T₄ (2.5 µM) treated embryos show strong reporter activity in muscle and to a lesser amount in epidermis, brain and eye retina. T₃ (2.5 µM) treatments result in similar responses but GFP induction in epidermis, brain and eye is stronger, and signal is also seen in blood cells. TRIAC (100 nM) treatments also induce GFP responses in brain, heart, eye, epidermis and muscle, and in addition, anterior spinal cord. Overlay pictures of bright field and GFP (75% transparent) are shown. Arrows indicate tissue-specific GFP responses: grey  =  muscle and red  =  brain. (d) TRβ responses in transgenic fish are dependent on a functional LBD. TRβ_wt, TRβ_E457A and TRβ_dAF2 embryos (24 hfp, F1) were heat pulsed and soaked in TRIAC (100 nM) as described in Figure 4c. The upper row shows strong (wt), weak (E457A) and no (dAF2) epidermal GFP expression in the tail. Western blot detected proteins show similar levels of TRβ transgene expression (right side).

https://doi.org/10.1371/journal.pone.0009797.g004

As further validation of drug responsiveness, we also treated TRβ fish with known agonists (Figure 4c). Vehicle treated TRβ embryos show the same response seen in untreated embryos. TRβ fish treated with either Thyroxine (T₄) or Triiodothyronine (T₃) induced reporter activation in the eye, heart, epidermis, blood, muscle and brain (Figure 4c). As expected, T₃ activates the reporter more strongly than T₄. Treatments with TRIAC, also yielded strong responses at dosages similar to those used in mammalian tissues. The eye activity of TRβ ligand trap fish correlates well with previously reported expression of endogenous TRβ in retina at 48 hpf [19]. However, our finding of receptor activity in other tissues such as the brain and anterior spinal cord are novel. Interestingly, in a recent profiling of zebrafish NR expression patterns, Bertrand and colleagues found that nuclear receptors expressed in the retina, except of TRβ, are also expressed in the brain and/or anterior spinal cord. We expect that many of the novel activity patterns observed in our LT lines will represent novel roles in early development, opening the door for new avenues of study.

LT drug treatments reveal tissue selective nuclear receptor activities

Recent drug discovery programs have been heavily focused on the identification of compounds with activities in a subset NR-regulated tissues. Such Selective NR Modulators (SNRMs) would yield fewer side effects and have additional uses. Interestingly, each of the TR agonists tested above exhibited significant tissue-selective activities, as indicated by the colored arrows in Figure 4c. For example, T₃ and T₄ elicited strong GFP responses in dorsal muscle while TRIAC did not. Conversely, TRIAC elicited responses in the spinal cord while T₃ and T₄ did not. Differential responses were also seen in the eye, heart, epidermis brain and blood. Many of these sites of action have not been previously documented.

To further verify that GFP expression in the TRβ lines are ligand- and cofactor-dependent, we made two additional transgenic lines carrying mutations in the TRβ LBD (Figure 4d). These include the previously described LT-TRβ_ΔAF2 and a construct carrying a single AF2 point mutation, LT-TRβ_E457A. Both mutations result in impaired co-activator recruitment upon ligand binding [34]. As already shown (Figure 4c), TRβ WT embryos respond strongly to the addition of TRIAC. However, TRIAC-treated TRβ_E457A fish show dramatically reduced GFP expression, with signal restricted mainly to epidermal cells of the tail bud, and TRβ_ΔAF2 embryos show no detectable GFP whatsoever (Figure 4d). Western blot analysis using anti-FLAG antibody shows that both fusion proteins are well expressed. We conclude that the signals observed with these lines are ligand and co-activator dependent.

GFP signals are readily quantified and suitable for drug screening

To test whether drug effects can be quantified with our ligand trap fish, we subjected TRβ LT embryos to increasing concentrations of the agonists T₃, TRIAC or KB2115 [41], and the antagonist Amiodarone. Figure 5a shows a series of tail regions of the treated TRβ embryos. Note that the number and intensity of GFP responding nuclei increase in proportion to hormone concentration but decrease again as levels become teratogenic.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 5. LT drug responses are quantifiable.

(a) Dose dependent GFP expression Shown are the dose responses for T₃, KB2115 and TRIAC in TRβ tail epidermis (48 hpf) treated embryos. Treatments were done from 0.1–500 nM. GFP images show anterior-posterior tail view (imaged at 150× magnification) (b) The graph indicates average intensity of GFP signals for each treatment in percent. GFP expression was quantified using nuclei count software from MetaX files. Values for drug treated fish are shown in yellow (TRIAC), in blue (KB2115) and in red (T₃).

https://doi.org/10.1371/journal.pone.0009797.g005

Figure 5b shows averaged GFP response curves for increasing concentrations of all three ligands. TRIAC proved to be the most effective, then KB2115, and lastly T₃. Amiodarone treatment yielded no detectable GFP response as expected. Importantly our dose-response curves for the various TRβ agonists tested correlate well with previously determined animal and human data [42], [43].

pLT plasmid construction

To generate the ligand trap vector, p-LT, we first combined the SV40 late polyadenlyation site from pCS2+ (cloned XbaI/Hind III) and the Gal4 DBD (aa 1–132; cloned into XbaI) into pBluescript (Stratagene). PmeI and NheI restriction sites were introduced in the primer as cloning sites for NR ligand binding domains. An EcoRV restriction site was designed N-terminal to the Gal4DBD for introducing the FSH-triple tag. Oligos for the triple tag were designed with EcoRI restriction sites on both ends and annealed together followed cloning into EcoRI digested pBSII. The tag was then PCR amplified with primers containing EcoRV restriction sites and a Kozak consensus sequence followed by an ATG start codon. This fragment was then introduced into the EcoRV restriction site of pBSGal1-132-pA, resulting in pBS-tripletag-Gal1-132-pA. Next, two I-SceI mega nuclease recognition sites were introduced by site directed mutagenesis flanking the fragment.

The eGFP-NLS gene was PCR amplified from the pUAS Stinger vector [52] and cloned into pBluescript II containing a SV40 polyA to generate pB-eGFP-pA. A fragment containing multi-UAS Gal DBD binding sites followed by the basal adenovirus promoter, E1b, was PCR amplified from pBUASEIB [13] and inserted into pB-eGFP-pA. The UASE1b-eGFP-pA fragment was then cut out by KpnI digestion and cloned into KpnI digested pB SceI-Tag-Gal1-132-pA. The resulting plasmid was SacII/NotI cut to insert a SacII/NotI fragment of the zebrafish Hsp70 promoter amplified from pzHSP70/4prom [25]. Finally, the gypsy insulator elements from the UAS-Stinger GFP transformation vector were PCR amplified and inserted upstream of the Hsp70 promoter into SacII restriction sites or downstream of the eGFP reporter into HindIII/ApaI.

Fish microinjections and maintenance

Zebrafish were maintained at 28.5°C on a 14/10 hour light/dark cycle and staged according to hours (h) or days (d) postfertilization [53]. To generate pLT lines, plasmids were pre-digested with I-SceI (0.8 µg DNA, 1 µl I-SceI (New England Biolabs) 2 µl 10x I-SceI buffer, 1 µl BSA) for 20 min at 37°C. Digested DNA was adjusted with 0.1% phenol red. 4.6 nl of DNA solution was injected into the blastomeres of early one-cell stage embryos [26]. Morpholino (MO) injection: 4.6 nl of MO (0.5 mM) in 0.1% phenol red was injected into the one-cell stage.

F0 fish were crossed with WT fish to identify germ-line transformed animals, as determined by fin cut PCR with GFP primers or control agonist treatments. F1 progeny showing strong and consistent GFP responses were selected for F2 homozygote production. To avoid reporter GFP silencing of stable transgenic LT lines [54], [55], F2 lines showing strong and consistent GFP responses were selected for continued propagation of the line. These lines have been maintained for several generations with no loss of responsiveness. At least two independent stable lines of each LT construct were generated, however for LT-PPARγ only one germline fish was identified.

All experiments were approved by the institutional review board at University of Toronto.

Plasmids, cell culture transfection

FSH-Gal1-132 from pLT plasmid (NotI/PmeI) was cloned in frame into pcDNA3.1 (Invitrogen) to generate pcDNA3 FSH-GalDBD. The N-CoR plasmid has been described previously [36]. pFLAG-HIPK3 (2–1191) and pFLAG-HIPK3 (K226R) were provided by J.J. Palvimo (University of Helsinki). The 14xUASE1B plasmid was a gift of Dr. Sue (Banting and Best Department of Medical Research/originally provided by Dr. R.W. Köster; Helmholtz Zentrum, Germany). The plasmids for reporter gene assays 2X-GAL4 binding-site luciferase, pSV40 β-Gal) have been described previously [37], [56]. 293 cell line transfections were done with Lipofectamin™ 2000 (Invitrogen).

Compound screening

Embryos were heat induced (28→37°C) for 30 minutes and then arrayed in 24 well plates (10 per well). The water was removed and 500 µl of embryo water/well including dissolved small molecules or solvent was added shortly after heat pulse. Embryos were incubated at 28°C for 14–20 h and then monitored using a fluorescent dissection scope (SteREO Lumar.V12 Carl Zeiss). For analyzing GFP fluorescent pattern, embryos, larvae and adult fish were anesthetized with Tricaine (Sigma, Cat.# A-5040). Embryos were mounted in 1% AgarPlaque PlusTM Agarose (# 21403A) purchased from PharminGen.

Drugs used

Sigma-Aldrich: Amiodarone hydrochloride (A8423); T₃ (T2877); T₄ (T2376) and TRIAC (T7650); Alexis Biochemicals: Pioglitazone (ALX-270-367); Cayman Chemicals: Rosiglitazone (#71740); Troglitazone (#71750); GW 9662 (#70785); KB2115 (#10011054), GW 0742 (#10006798) and GW 590735 (#10009880).

Affinity purification of proteins

Purifications were performed at 4°C. 5-6 hpf LT-TRβ embryos (128 g) were heat induced (37°C) for 20 min and recovered for 2 h. Homogenization was carried out in 100 ml FSH buffer (100 mM Tris-Cl pH 8, 150 mM NaCl, 0.1% Triton X-100 and Roche complete Mine protease inhibitor cocktail) including 2.5 mM DTT, 10 nM Avidin and 1 mM EDTA. After 20 min incubation the lysate was centrifuged for 10 min at 9000 rpm. 15 ml of prewashed (FSH buffer) Agarose-bead slurry (ABT; 2% plain Agarose beads) was incubated with the centrifugation supernatant (220 ml) for 30 minutes. After this incubation and again centrifugation the supernatant (200 ml) was incubated with 5 ml prewashed Strep Tactin® (IBA; Cat: 2-1201-0251) slurry for 2 h. After incubation the beads were transferred to 2 gravity disposable columns (Bio-Spin columns, Cat: 732–6008) and washed 6 times with 2 ml of FSH-buffer including 1 mM EDTA. Bound proteins were eluted using volume 0.5 ml of 2.5 mM d-desthiobiotin containing FSH-buffer and 1 mM EDTA. For triple step purifications the Strep elutions were incubated for 2 hrs with FLAG M2 monoclonal antibody matrix (Sigma), and then washed 3times with FSH buffer followed by elution with FSH buffer containing 300 µg/ml 3x FLAG peptide (Sigma). Flag elutions were incubated for 1 hrs with 60 µl of Talon beads in an Eppendorf tube, and then washed two times with FSH buffer followed elution with 2xSDS buffer. For immunoprecipitations, HEK 293 cells were grown in 10 cm dishes and DNA (9 µg of FLAG-HIPK3 with either 4 µg of pcDNA3-FSH-GAL, -GAL-TRβ or -GAL-TRβ_ΔAF2/empty vector was used to adjust the DNA concentration) was transfected with Lipofectamin following the manufacturer's instructions. Cells were harvested in 1 ml IP buffer (50 mM Tris pH 8, 150 mM NaCl, 1 mM EDTA, 0,75% NP-40, 1 mM DTT and Roche complete Mine protease inhibitor cocktail). After 10 min centrifugation at 13,000 rpm, 600 µl of cleared extract was incubated with 100 µl Strep Tactin® slurry for 2 h; followed by 3 washes with 1 ml of IP buffer. Immunoprecipitated proteins were separated by SDS PAGE and detected by FLAG-M2 Western blotting.

Western blotting

Expression of ligand trap constructs was verified using a monoclonal mouse αM2 antibody (dilution 1∶10 000; Sigma). Heat pulsed and recovered transgenic embryos were homogenized in 5 µl FSH buffer/embryo and denatured with SDS PAGE loading buffer followed separation on 10% SDS-polyacrylamide gels. Detection of HIPK3 was done with αHIPK3 from Abgent (AP 7500b).

Mass spectrometry

Eluted proteins were separated by SDS PAGE, silver stained and individual bands excised, digested with Trypsin and analyzed by MALD-TOF or ESI-MS tandem mass spectrometry. All samples for ESI-MS were analyzed by a 2-hour LC gradient using a split-free nano-LC (EasyLC, Proxeon, Odense, Denmark) coupled to a LTQ-Orbitrap XL, as recently described [57]. Raw data was converted to m/z XML using ReAdW and searched on a Swquest SorcererTM, using both Sequest and X!Tandem, against a zebrafish IPI protein sequence database (version 3.38) also containing the protein sequences of known contaminants (i.e. human keratins) and trypsin and the bait protein (human thyroid receptor). Searches were performed with a fragment ion mass tolerance of 0.8 Da and a parent ion tolerance of 50 ppm. Complete tryptic digestion was assumed. The iodoacetamide derivative of cysteine was specified in Sequest and X!Tandem as a fixed modification. The oxidation of methionine was specified as a variable modification. Scaffold (version Scaffold 2.1.1, Proteome Software Inc., Portland, OR) was used to validate MS/MS based peptide and protein identifications. Peptide identifications were accepted if they could be established at greater than 95.0% probability as specified by the Peptide Prophet algorithm and contained at least 1 identified peptide. Proteins that contained similar peptides and could not be differentiated based on MS/MS analysis alone were grouped to satisfy the principles of parsimony.

Oligos used to clone pLT:

pB Gal1-132-pA:

5’ATTCATCTAGAGATATCAAGCTACTGTCTTCTATCGAACAAGC 3’ATTATCTAGAGTTTAAACAGCTAGCTGATGATGTCGCACTTATTCTATGC

pB II triple tag:

sense –AATTCGACTACAAAGACCATGACGGTGATTATAAAGATCATGACATCG -

ACTACAAGGATGACGATGACAAGGAGAACCTGTACTTCCAGTCCAACTGGAGCC -

ACCCGCAGTTCGAAAAGCATCACCATCACCATCACG

-antisense -ATGATCTTTATAATCACCGTCATGGTCTTTGTAGTCG

-AGTTGGACTGGAAGTACAGGTTCTCCTTGTCATCGTCATCCTTGTAGTCGATGTC

-AATTCGTGATGGTGATGGTGATGCTTTTCGAACTGCGGGTGGCTCC

pB triple tag-Gal1-132-pA

5’ ATTATGATATCgccaccatgGACTACAAAGACCATGACGG

3’ ATTATGATATCGTGATGGTGATGGTGATGC

SceI pB triple tag-Gal1-132-pA

T7 5’ GACTCACTATAGGGCTAGGGATAACAGGGTAATGAATTGGGTACCGGG-

T7 3’ CCCGGTACCCAATTCATTACCCTGTTATCCCTAGCCCTATAGTGAGTC-

T3 5’ CGGTGGAGCTCCAGTAGGGATAACAGGGTAATCTTTTGTTCCCTTTAGTG-

T3 3’ CACTAAAGGGAACAAAAGATTACCCTGTTATCCCTACTGGAGCTCCACCG-

pB eGFP-pA

5' ATTATCTAGAACCATGGTGAGCAAGGGC

3' ATTATCTAGATTACTTGTACAAGTAGCG

pB UASE1b-eGFP-pA

5' ATTATCCGCGGGGTACCCTCCAAGGCGGAGTACTGTCC

3’ ATAATCGGCCGGTGTGGAGGAGCTCAAAGTGAGGC

pLT ; SceI pB zHsp70 triple tag Gal-1-132-pA-UASE1b-eGFP-pA)

5’ATTATCCGCGGTCAGGGGTGTCGCTTGG 3’ATTATGCGGCCGCGATATCGAATTCCTGCAGG

pLT gypsy; SceI gypsy pB zHsp70 triple tag Gal-1-132-pA-UASE1b-eGFP-pA)

5’ Hsp ATAACCGCGGTCACGTAATAAGTGTGCG

3’ Hsp ATAACCGCGGAGATCTATACTAGAATTGATCGGC

5’ Gfp ATAAAAGCTT TCACGTAATAAGTGTGCG

3’ Gfp ATAAGGGCCCATACTAGAATTGATCGGC

Morpholino oligo sequence:

5' to 3' and complementary to the translational start of the FSH Gal-NR:

ACCGTCATGGTCTTTGTAGTCCATG

Primes used to clone NR cDNAs:

NR1A2_wt/ _E457A/ _dAF2

5’ ATTAGCTAGCATGACTCCCAACAGTATGAC

NR1A2wt

3’ ATTAGTTTAAACCTAATCCTCGAACACTTCC

NR1A2_E457A

3’ ATTAGTTTAAACCTAATCCTCGAACACTGCCAAGAACAAAGG

NR1A2_dAF2

3’ ATTAGTTTAAACCTACTATTCTGTGGGGCATTCCACC

NR1C3

5’ ATTAGCTAGCGAGAAGGAGAAGCTGTTGG

3’ ATTAGTTTAAACCTAGTACAAGTCCTTGTAGATCTCC

NR1F3

5’ ATTAGCTAGCATGTCCAAGAAGCAGAGGG

3’ ATTAGTTTAAACTCACTTGGACAGCCCCACAGG

NR1D1

5’ ATTAGCTAGCATGCTTGCTGAGATGCAGAGTGCC

3’ ATTAGTTTAAACTCACTGGGCGTCCACCCGGAAGG

NR4A3

5’ GATCGCTAGCCCATTACAACAGGAACCTTCTCAG

3’ CACCGTTTAAACTTAGAAAGGTAGGGTGTCCAGG