In silico analyses
It was documented that the C-termini of the bHLH-PAS transcription factors are responsible for the specific modulation of these proteins’ action [12]. Specific chain flexibility, predicted for most of the bHLH-PAS C-termini [15], may be a useful protein feature. To determine to what extent the D. melanogaster GCEC structure is disordered, we performed in silico analysis. We used different predictors of protein disorder: PONDR-VLXT [23], PONDR-VLS2 [24], DISOPRED2 [25], FoldIndex [26], IUPred [22] and GeneSilico MetaDisorder [27] to get the full spectrum of possible results. Since all results were comparable, we decided to only show two representative results, and in addition the result of GeneSilico MetaDisorder as a meta-server combining 13 existing methods of prediction (Fig. 1B). The GCEC seems to be mostly disordered along the entire length of the sequence. Short fragments with a tendency to order occur mainly in the area near 30 aa, between 150 and 200 aa, and near 260 aa (predicted with a high probability on the PONDR-VLXT server, Fig. 1B, top panel), and could participate in the protein-protein interactions (PPIs) or act as the molecular recognition elements (MoREs, indicated in red color).
The amino acid composition is one of the factors determining the final conformation adopted by the protein in solution [46–48]. While globular proteins are characterized by a high content of hydrophobic residues and a high hydrophobicity, intrinsically disordered proteins (IDPs) or intrinsically disordered regions (IDRs) are characterized by a high content of charged residues, causing a high net charge. The Uversky diagram [46–48] plots the mean net charge versus the mean hydrophobicity and distinguishes IDPs from ordered proteins (Fig. 1C). Both parameters determined for the GCEC (average hydrophobicity 0.4082 and average charge 0.0769) fit to the values typical for IDPs, which indicates that the GCEC sequence may present the characteristics of IDPs (Fig. 1C).
As the presented results of the in silico analyzes suggested the disordered nature of GCEC, we decided to perform structural characterization of the purified protein in vitro.
GCEC expression and purification
To perform the GCEC analysis in vitro, we developed and optimized an expression and purification protocol. We tested many vectors, introducing additional tags, which usually improve protein stability and solubility (like TrxA, MBP, SUMO and others), and different bacterial strains. Unfortunately, under all the tested conditions, we were not able to obtain GCEC in a soluble form (data not shown). This may be explained by the toxicity of this protein for bacteria, or by its disordered structure, which results in the formation of inclusion bodies. Consequently, we decided to develop a GCEC purification procedure under denaturing conditions. We focused on the pET-M11 vector, introducing a short polyhistidine tag (6xHis). After protein denaturation with GdmCl, the GCEC was refolded by dilution. The subsequent purification process was simplified, since the inclusion bodies contained mainly recombinant GCEC, with only a small amount of impurities [49]. We used Ni2+-NTA resin for the next step of purification. It enabled the refolded GCEC volume to be reduced to 3 ml, a volume equal to the volume of elution, what simultaneously it concentrated the protein. As the final step of purification, we used SEC (Fig. S2A). To verify whether the obtained GCEC sample had the correct molecular mass (MM), we performed ESI mass spectrometry measurements. Two MM values were obtained: 36 003 Da, which is compliant with the MM of the construct calculated based on the aa sequence using the ProtParam tool, and also 36 020 Da, which is oversized by 16 Da (in relation to the calculated one). Finally, we performed protein sequencing, which confirmed the GCEC identity and revealed the oxidation of two M residues: M731 (M71 in GCEC) and M909 (M249 in GCEC). The modified form of the GCEC accounted for about 16% of the preparation and appeared with every purification (data not shown).
Purified GCEC appeared as a single band on the 12% SDS-PAGE gel (Fig. S2B). Its electrophoretic mobility was decreased and corresponded to the 42 kDa protein instead of the expected 36 kDa. Such behavior is often observed for IDPs [46, 50]. Their unique amino acid composition has an impact on SDS binding, which results in an unusual mobility in the SDS-PAGE experiments [46, 50]. Existence of purified GCEC in the native, active form is ensured by the ability of GCEC to interact with FTZ-F1 (see below).
Hydrodynamic analysis of GCEC
One of the easiest ways to identify IDPs is the determination of a protein’s hydrodynamic properties, since IDPs present a significantly overestimated hydrodynamic radius in comparison to globular proteins of the same MM [51]. During analytical size exclusion chromatography (SEC), GCEC was eluted as a single peak with an elution volume corresponding to a RS of 44.7 ± 0.3 Å (Fig. 2A, Tab. 1). The value was approximately 70% higher than the Rs calculated with the assumption of GCEC globular conformation (26.5 Å, Tab. 1), and was GCEC concentration independent (data not shown). Therefore, the experimentally determined volume (374.1 Å3) of the GCEC was much higher than the theoretical volume (77.9 Å3), and the experimentally determined density (0.10 kDa/ Å-3) was much lower than the theoretical density (0.46 kDa/Å-3) (Tab. 1). This experiment indicated that GCEC has a significantly elongated, poorly packed conformation. However, it was not possible to clearly state if GCEC exists in a monomeric form in solution. The overstated RS value may also be a consequence of protein oligomerization
Tab. 1. Characterization of GCEC by SEC
MM [kDa]
|
Rs [Å]
|
VS·103 [Å3]
|
p·10-3 [kDa/Å-3]
|
theora
|
exp
|
theorb
|
expc
|
theorb
|
expc
|
36.0
|
26.5
|
44.7
|
77.9
|
374.1
|
0.46
|
0.10
|
aCalculated from the equation: log(RS)= (0.085±0.031)+ (0.395±0.016)log(MM) [33].
bCalculated using the theoretical RS.
cCalculated using the experimental RS.
|
To definitively determine if GCEC can form oligomers, we performed analytical ultracentrifugation (AUC) experiments. We analyzed GCEC samples in three concentrations: 0.07, 0.18 and 0.33 mg/ml. The use of relatively low concentration ranges resulted from the data recording at 230 nm. It was determined by the low absorbance coefficient of the GCEC, in which the aa sequence is characterized by the low content of aromatic aa residues, and in particular no W residues (Abs2800.1%= 0.255 ml/(mg∙cm) calculated on the ProtParam server). The very high (above 1.0 AU) absorbance at 230 nm for the samples in higher concentrations would result in huge data errors.
The determined root-mean-square deviation (rmsd = 0.015, Tab. 2) values were relatively high, which could be the result of the presence of DTT in the buffer. The addition of DTT, which is highly unstable and in reduced form absorbs near 210 nm, may lead to the strong background during analysis in the absorption detection system [52]. GCEC was observed as a single signal at the 2S value (Fig. 2B, Tab. 2). Importantly, no signal at high S-values, characteristic for oligomers and aggregates, was detected (Fig. 2B). The values of the sedimentation coefficient (S20,w) were GCEC concentration independent (Tab. 2). The experimentally determined RS was approximately 45 Å (Tab. 2) and was consistent with the SEC result (44.7 ± 3.0 Å). Because of the relatively high rmsd, we decided to perform an additional experiment exploiting the Rayleigh interference detection system. This detection system significantly improves the results of the measurement of samples containing highly absorbing components, such as ATP/GTP and oxidized DTT [52]. We measured the GCEC at two concentrations: 0.31 and 0.82 mg/ml. The main signal corresponded equally to the result obtained using the absorption detection system (S=2S, Tab. 3). Importantly, the rmsd value significantly decreased to a value of 0.006 (Tab. 3), which confirmed a very good fit of results. Again, no signal at high S-values was observed (not shown).
The frictional ratio f/f0 represents the degree of deviation of the molecule from a minimum possible value of 1.0 for a hard, incompressible sphere [53]. Therefore, it allows for protein shape characterization [54]. For globular proteins, f/f0 is typically 1.05–1.30 [55]. For IDPs, the f/f0 ratio is much higher (1.75-3.0) and increases significantly with the MM [54]. The f/f0 calculated for GCEC using AUC data was over 2 (Tab. 2). This indicates a highly asymmetric and elongated shape, assigning GCEC to coil-like IDPs [54]. The experimentally determined MM is equal to the theoretical molecular weight calculated on the ProtParam server (36.7 kDa vs 36 003.0 Da, Tab. 2). To conclude, GCEC is a monomeric protein with a highly elongated shape and a high degree of asymmetry.
Analyzing the dependence of the RS on the relative MM, globular proteins can be divided into four states: native proteins (N), molten globules (MGs), pre-molten globules (PMGs) and 6-M GdmCl-unfolded proteins (coil). Two additional IDP states are known: coil-like IDPs) and pre-molten globules-like IDPs (PMG-like) [56, 57]. RS determined for the GCEC with the SEC and AUC experiments place GCEC on the plot relating RS and MM in the area occupied by coil-like IDPs (Fig. 2D). Such a result is consistent with previous in silico and SEC analysis. We performed additional calculations based on equations derived by Tcherkasskaya et al. [55], correlating the MM and the RS for different conformational states of the protein. For GCEC (MM of 36.0 kDa), RS calculated with the assumption of the PMG-like conformation was 36.4 ± 0.4 Å, and of the coil-like it was 51.6 ± 0.7 Å. The experimentally determined RS (44.7 Å SEC and 45 Å AUC) indicates that GCEC conformation corresponds to coil-like IDPs.
Tab. 2. Characterization of GCEC by sedimentation velocity AUC using an absorption detection system
Concentration [mg/ml]
|
rmsd
|
f/f0
|
S20,w (S)
|
S (S)
|
Rs [Å]
|
App MM [kDa]
|
0.07
|
0.01435
|
2.09
|
2.118
|
2.043
|
46.1
|
38.3
|
0.18
|
0.01517
|
2.09
|
2.063
|
1.990
|
44.8
|
36.2
|
0.33
|
0.01606
|
2.03
|
2.069
|
1.995
|
43.8
|
35.5
|
Tab. 3. Characterization of GCEC by sedimentation velocity AUC using a Rayleigh interference detection system
Concentration [mg/ml]
|
rmsd
|
f/f0
|
S20,w (S)
|
S (S)
|
Rs [Å]
|
App MM [kDa]
|
0.31
|
0.00566
|
2.01
|
2.041
|
1.966
|
42.6
|
34.1
|
0.82
|
0.00635
|
2.06
|
2.009
|
1.935
|
44.2
|
34.8
|
Based on hydrodynamic analyzes, we conclude that GCEC exhibits IDPs properties. It has a highly elongated shape, does not oligomerize in solution, and can be assigned to coil-like IDPs.
Far-UV CD analysis
CD spectroscopy is commonly used for the determination of the secondary structure content and folding properties of proteins [58]. The shape of the curve makes it easy to distinguish between α-helical structures (negative peaks at 222 nm and 206 nm)[59], β-strands (negative peaks at 218 nm)[60] and non-regular secondary structures (negative peak at near 200 nm)[61]. The CD spectrum of GCEC (Fig. 3A, Tab. 4) shows a clear minimum near 200 nm (‑7.2×10−3 deg·cm2·dmol-1) and a small negative signal near 222 nm (-1.9×10−3 deg·cm2·dmol‑1). Such a result indicates the disordered character of GCEC and highlights the presence of a residual ordered structure. Deconvolution of the CD spectrum performed with CDPro software (CONTIN/LL algorithm, SPD48 base) confirmed that GCEC is mainly disordered (49.0 ± 5.5 %). It also revealed the existence of some ordered structures, mainly β-strands (31.9 ± 6.0%), partially distorted (9.2 ± 2.5%) (Tab. 4). Moreover, small amounts of totally distorted (5.4 ± 3.7%) α‑helixes are estimated (Tab. 4).
The changes in the CD spectrum observed in the presence of denaturing agents can provide important information regarding protein structure and the degree of protein compaction [62, 63]. To determine the impact of denaturing agent on the GCEC’s secondary structure, we recorded spectra in the presence of 1 M, 2 M and 4 M GdmCl (Fig. 3A). All data, due to the strong absorbance of GdmCl in high concentrations, were collected in a narrow wavelength interval. For this reason quantitative data deconvolution was not performed. The presence of GdmCl resulted in signal blanking at 222 nm (to ‑0.6×10‑3deg·cm2·dmol-1 in the presence of 4M GdmCl, Fig. 3A). Such an observation clearly confirmed the presence of the residual ordered secondary structure in GCEC in the absence of GdmCl. After incubation with the denaturing agent, the GCEC conformation becomes much more disordered, indicating the loss of the residual ordered secondary structure.
As demonstrated, the temperature and selected chemical reagents (i.e. osmolytes, binding partners, crowding agents, counter ions) can affect the structure of some IDPs [63]. Usually, a more ordered structure can be observed. To determine GCEC conformation changes under certain conditions, the corresponding CD spectra were collected after incubation with TFE or in the function of temperature increase (Fig. 3A and 3B). First, we studied the influence of 15% and 30% TFE, which is known as ordered secondary structure stabilizer [64]. The presence of TFE significantly affects the shape of the GCEC CD spectrum: the signal around 200 nm decreases, while negative signals around 222 nm and 206 nm, characteristic for ordered secondary structures, appear (Fig. 3A). Data deconvolution revealed a significant increase in the content of α-helical structures (from 5.4 ± 3.7% to 17.6 ± 4.7% and 27.2 ± 8.5% for 15% and 30% TFE respectively, Tab. 4). Simultaneously, a decrease in the content of β-type structures was observed (Tab. 4). We suppose that some of the β-structures can be transformed into α-type structures, which is often observed for TFE [65, 66]. However, the decrease in the quantity of β-structures and the increase in the quantity of α-helixes are not proportional and some of the α-helices can be formed from the disordered GCEC fragments. Finally, in the presence of 30% TFE, a significant part of the GCEC (34.3 ± 1.1%) still exhibits a disordered character (Tab. 4).
Tab. 4. Characterization of GCEC by CD
Factor
|
α-helises (%)
|
β-strands (%)
|
Turns (%)
|
Ua (%)
|
Ra
|
Db
|
Σ
|
Ra
|
Db
|
Σ
|
-
|
0.5 ± 1.6
|
4.9 ± 2.1
|
5.4 ± 3.7
|
22.7 ± 3.5
|
9.2 ± 2.5
|
31.9 ± 6.0
|
13.8 ± 1.5
|
49.0 ± 5.5
|
15% TFE
|
8.7 ± 3.2
|
8.9 ± 1.5
|
17.6 ± 4.7
|
16.9 ± 1.6
|
9.7 ± 2.1
|
26.2 ± 3.7
|
19.4 ± 2.0
|
36.4 ± 2.0
|
30% TFE
|
14.9 ± 6.2
|
12.8 ± 2.3
|
27.2 ± 8.5
|
10.7 ± 3.0
|
7.7 ± 2.8
|
18.4 ± 5.8
|
19.1 ± 3.4
|
34.3 ± 1.1
|
1M
GdmCl
|
0.2 ± 2.5
|
1.8 ± 3.7
|
2.0 ± 6.2
|
14.1 ± 3.4
|
8.7 ± 3.1
|
22.8 ± 6.5
|
12.8 ± 5.4
|
62.3 ± 4.8
|
2M GdmCl
|
0.0 ± 4.0
|
2.0 ± 5.2
|
2.0 ± 9.2
|
11.5 ± 5.1
|
6.4 ± 5.7
|
17.9 ± 10.8
|
10.0 ± 6.0
|
68.1 ± 5.2
|
4M GdmCl
|
-
|
-
|
-
|
-
|
-
|
-
|
-
|
79.8 ± 7.8
|
a Regular structure
b Distorted structure
c Unstructured
|
Some of coil-like and PMG-like IDPs present a unique temperature response. In contrast to globular proteins, which denature in higher temperatures, such IDPs in the same conditions can adopt a more ordered conformation [63]. This can be explained by the increase of the strength of hydrophobic interactions promoting protein folding [63]. Such behavior can be observed for GCEC. In spectra recorded for GCEC as a function of temperature increase, the signal around 200 nm was gradually reduced and shifted toward higher wavelengths (Fig. 3B). In addition, a characteristic negative maximum around 222 nm appeared, indicating an increase in the content of ordered secondary structures (Fig. 3B). Importantly, these induced structural changes were completely reversible. After cooling the sample to 20°C, the GCEC spectrum returned to its original shape (Fig. 3B). Since the signal changes observed at 222 nm are linear, there is no cooperative transition between extreme conformational states (Fig. 3B, inset). We analyzed the obtained data with CDPro software, however the observed changes were not big enough to get quantitative deconvolution, indicating an increase in secondary structures.
SAXS analysis
SAXS is commonly used for the characterization of the low-resolution structure of macromolecules in solution [67, 68]. Importantly, SAXS is especially useful for the analysis of the IDPs with elongated and flexible chains, where other methods fail [69]. Therefore, we performed SAXS studies to get additional information regarding the structure and conformational dynamics of GCEC in solution.
Unfortunately, during irradiation, GCEC exhibited radiation damage, which resulted in protein aggregation. Since the SAXS scattering signal is a function of molecular weight, this technique is sensitive to the presence of even a very small fraction of aggregates (higher oligomers or larger impurities). These phenomena significantly affect the measurement results and make them not interpretable [70]. However, the radiation effect was significantly reduced when we performed the SAXS experiment immediately after the protein purification without any further concentration. To get an insight into the structure of GCEC, we collected three 10 min scans and combined them for further analyzes. Although the GCEC was measured in relatively low concentrations, the collected data were of a good enough quality for low resolution modelling (EOM) and to get an insight into the protein’s structural properties.
The Kratky plot presents the scattering intensity (I(s)) multiplied by the square of scattering vectors s (s2) as a function of the scattering vector s, and was used for SAXS data qualitative analysis [71]. The shape of the Kratky plot is sensitive to protein conformation and is used for the assessment of the protein’s flexibility and degree of its unfolding. The SAXS profile obtained for GCEC (Fig. 4A) does not present the maximum characteristic for globular proteins and reaches a plateau at higher values of the scattering vector. Such a shape is characteristic for IDPs [72]. The Gunier function designated for the GCEC shows a linear character, which is a good indicator of the GCEC’s monodispersity in solution (Fig. 4B). The radius of gyration Rg calculated for the GCEC from the Gunier plot (function) was 52.2 Å.
The pair distance distribution function p(r), representing the distribution of all interatomic distances within the molecule [71], was calculated for GCEC using GNOM [43, 73] and experimental SAXS data in the s-range from 0.0104 to 0.1522 Å-1 (Fig. 4C). The Rg calculated independently from the Gunier function was 54.1 Å. The maximal intramolecular distance (Dmax) was 247 Å. All the determined parameters indicate the highly asymmetric and expanded GCEC conformation [74].
Finally, we performed EOM analysis [44] in order to define the most representative conformations adopted by GCEC in the solution. The EOM algorithm was used to generate a pool of 10 000 random conformers of random coil conformation. Then, a sub-ensemble that fits best to the experimental scattering profile was selected. The Rg determined for the final conformational sub-ensemble is slightly moved towards higher Rg values (56.0 Å, Fig. 5A) in comparison to the random pool. Moreover, the shape of the histogram is asymmetric and irregular. It means that the sub-ensemble conformations differ in the degree of compaction. An additional peak near 65 Å corresponds to more extended conformations (Fig. 5A). The fit between the experimental data and the back-calculated EOM sub-ensemble is good (χ2 of 0.759) (Fig. 5B). The obtained representative models present two types of conformations: strongly bent in the middle of the length (Fig. 6A), corresponding to the main peak in Fig. 5A, and a longer, highly tangled conformation at both termini (Fig. 6B), corresponding to the additional peak near 65 Å.
NMR analysis
The interactions comprising full length FTZ-F1 and GCE were studied previously [18]. In our study we focused on the binding capacity between selected protein domains: LBD FTZ-F1, comprising the AF2 motif, and the GCEC region comprising the novel NR-box. Additionally, we tested the interactions between the LBD FTZ-F1 and short GCEPEP peptide (LRLIQNLQK), representing the novel NR-box (LIXXL). We aimed to determine in vitro if the LIXXL motive alone is able to create an interaction surface with the AF2 motif presented in the FTZ-F1.
The NMR spectrum of the GCEC was recorded in order to directly confirm the intrinsically disordered character of this protein. The obtained GCEC spectrum is typical for IDPs (Fig. 7; blue). Most proton signals are observed in a narrow frequency range (8-8.5 ppm) and strongly overlap (Fig. 7; blue) with little dispersion. Such a result is caused by the narrow diversity range of the chemical environments experienced by the observed nuclei [75, 76]. The signals of around 6.7 and 7.5 ppm correspond to the side chains of Q and N (Fig. 7; blue). The similar size of all the observed signals confirms the lack of stable, ordered longer fragments in the GCEC structure. The single and more dispersed signals may correspond to the amino acid residues involved in the formation of the short, local and transient motifs of the secondary structure (Fig. 7). The result of NMR analysis is compatible with the above presented CD denaturation data and SAXS experiments. The GCEC spectrum was used as a reference for the chemical shift perturbation experiment, aimed at verifying the GCEC and LBD FTZ-F1 interactions. The adequate spectrum was recorded after incubation with an equimolar amount of unlabeled LBD FTZ-F1. The observed signals show significant changes in intensity (Fig. 7; red). Importantly, multiple signals are shifted, clearly indicating the interaction between proteins (Fig. 7; compare blue and red spectra).
We performed an analogical experiment for the labeled LBD FTZ-F1 (Fig. 8, blue). We decided to analyze possible interactions of this domain with short peptide GCECPEP (LRLIQNLQK), corresponding to the LIXXL motif present in the GCEC sequence [18]. The recorded reference spectrum is similar to the spectrum presented by Daffern et al. [19], and shows a good peaks dispersion appropriate for globular protein (Fig. 8, blue). The spectrum recorded after incubation witch GCEPEP (Fig. 8, red) presents specific signal perturbations and confirms its binding to the LBD FTZ-F1 (Fig. 8; compare blue and red spectra). The specific signals, representing aa residues experiencing major changes, are marked. All the results indicated that the intrinsically disordered GCEC (or GCEPEP representing the novel GCE NR-box) is sufficient to form an interaction interface with the LBD of FTZ-F1 in vitro in the absence of JH. Significantly, these interactions can force GCEC to adopt a more fixed structure. We suggest that GCEC could be sufficient to modulate the FTZ-F1 nuclear receptor activity in the FTZ-F1 LBD dependent manner.
FLAG Pull-down and Assay fluorescence analysis
To confirm the results of the NMR studies in more natural milieu, i.e. in cells, and to determine the effect of GCEC and LBD FTZ-F1 interactions on the subcellular localization of these proteins, we performed dedicated experiments in COS-7 cells.
First, the immunoprecipitation experiment with the ANTI-FLAG M2 Affinity gel was executed. The expressed C-terminally FLAG-tagged GCEC protein or FLAG-tagged GCEC with a partner of interaction (LBD FTZ-F1) were pulled-down from the cell extracts using ANTI-FLAG M2 Affinity gel and then analyzed by Western-blot. The GCEC was additionally tagged on the N-terminus with YFP and LBD-FTZ-F1 with CFP, allowing further localization analyses. Cells transfected with the LBD FTZ-F1 were used as a negative control. The LBD FTZ-F1 with no FLAG-tag is not able to bind to the ANTI-FLAG M2 Affinity gel (Fig. 9A, no bands in the elution fraction). Cells transfected with GCEC-FLAG were used as a positive control. GCEC-FLAG was bound to the ANTI-FLAG M2 Affinity gel and then observed in the immunoblotting as a single band at the appropriate high (Fig. 9B). The observed additional bands are poorly marked and non-specific. When the COS-7 cells were co-transfected with GCEC-FLAG and LBD FTZ-F1, both proteins were observed in the fraction eluted from the ANTI-FLAG M2 Affinity gel and detected with anti-GFP antibody (Fig. 9C). Such a result clearly shows that the intrinsically disordered GCEC comprising the novel NR-box (LIQNL) is sufficient to form an interaction interface with the FTZ-F1 ligand binding domain (LBD) which forms AF2.
Simultaneously, we performed localization studies using proteins of interest N-terminally tagged with YFP (GCEC-FLAG) or CFP (LBD FTZ-F1). As demonstrated by Chalfie et al. [77], green fluorescent protein (GFP) can be used to monitor protein expression and localization in living cells. The labeling of the protein with different fluorescent tags is currently a widely used method that does not affect the localization or the function of fused proteins [78]. Twenty-four hours after transfection (with pEYFP-C1/GCEC-FLAG or pECFP-C1/LBD FTZ F1), or co-transfection (simultaneously with pEYFP-C1/GCEC-FLAG and pECFP-C1/LBD FTZ F1) of the COS-7, we analyzed the expression and subcellular localization of the fluorescently tagged proteins using fluorescent microscopy. While the CFP-LBD FTZ-F1 was distributed within the whole cell (Fig. S5A), YFP-GCEC-FLAG was observed exclusively in the nuclei (Fig. S5A). Simultaneous expression of the GCEC and LBD FTZ-F1 did not affect the GCEC nuclear localization (Fig. S5B), while the LBD FTZ-F1 was shifted to be predominantly nuclear (Fig. S5B). As a control, we transformed COS-7 cells with empty vectors (pEYFP-C1 or pECFP-C1) to express YFP or CFP. As expected, the expression of YFP or CFP resulted in the ubiquitous localization of the proteins (Fig. S5C) and did not influence the fused proteins’ localization.