Overexpression, Purification, Characterization, and Crystallization of the BTB/POZ Domain from the PLZF Oncoprotein*

The BTB/POZ domain defines a conserved region of about 120 residues and has been found in over 40 proteins to date. It is located predominantly at the N terminus of Zn-finger DNA-binding proteins, where it may function as a repression domain, and less frequently in actin-binding and poxvirus-encoded proteins, where it may function as a protein-protein interaction interface. A prototypic human BTB/POZ protein, PLZF (promyelocytic leukemiazinc finger) is fused to RARα (retinoic acid receptor α) in a subset of acute promyelocytic leukemias (APLs), where it acts as a potent oncogene. The exact role of the BTB/POZ domain in protein-protein interactions and/or transcriptional regulation is unknown. We have overexpressed, purified, characterized, and crystallized the BTB/POZ domain from PLZF (PLZF-BTB/POZ). Gel filtration, dynamic light scattering, and equilibrium sedimentation experiments show that PLZF-BTB/POZ forms a homodimer with aK d below 200 nm. Differential scanning calorimetry and equilibrium denaturation experiments are consistent with the PLZF-BTB/POZ dimer undergoing a two-state unfolding transition with a T m of 70.4 °C, and a ΔG of 12.8 ± 0.4 kcal/mol. Circular dichroism shows that the PLZF-BTB/POZ dimer has significant secondary structure including about 45% helix and 20% β-sheet. We have prepared crystals of the PLZF-BTB/POZ that are suitable for a high resolution structure determination using x-ray crystallography. The crystals form in the space group I222 or I212121 with a = 38.8, b = 77.7, and c = 85.3 Å and contain 1 protein subunit per asymmetric unit with approximately 40% solvent. Our data support the hypothesis that the BTB/POZ domain mediates a functionally relevant dimerization function in vivo. The crystal structure of the PLZF-BTB/POZ domain will provide a paradigm for understanding the structural basis underlying BTB/POZ domain function.

The PLZF BTB/POZ domain, named for its presence in the Drosophila proteins Broad Complex, tramtrack, and bric a brac (BTB) (1) and its homology with several poxvirus proteins and zinc finger proteins (POZ) (2), is an evolutionarily conserved motif of about 120 residues ( Fig. 1) found in an increasing number of proteins having a variety of functions (1)(2)(3)(4). Proteins containing a BTB/POZ domain have been identified in poxvirus, Caenorhabditis elegans, Drosophila, and human and are generally found at the N terminus of either actin-binding or, more commonly, nuclear DNA-binding proteins (4). Proteins containing a BTB/POZ domain have been associated with diverse functions including nucleosome/chromosome disruption, pattern formation, metamorphosis, oogenesis, and eye and limb development (5)(6)(7)(8)(9)(10)(11).
Biological relevance for the function of BTB/POZ domains has come from the study of the PLZF (for promyelocytic leukemia zinc finger) oncoprotein, associated with acute promyelocytic leukemia (APL). 1 APL results from the malignant proliferation of cells blocked at the stage of promyelocytic differentiation and accounts for about 10% of all myeloid leukemias (12)(13)(14). APL is characterized by a non-random reciprocal chromosomal translocation, t (15;17), observed in greater than 95% of patients (15,16), resulting in the fusion of the retinoic acid receptor ␣ (RAR␣) gene with a gene called PML (for promyelocytic leukemia). In a small subset of APL cases, a second translocation, t (11;17), has been identified that fuses the RAR␣ gene to a previously uncharacterized gene, PLZF, resulting in the expression of the chimeric PLZF-RAR␣ fusion protein harboring the BTB/POZ domain from the PLZF protein (17). This chimeric protein appears to function as a dominant negative transcription factor (18). Remarkably, APL patients expressing the PLZF-RAR␣ fusion protein are not responsive to treatment with all-trans-retinoic acid, and thus have a poor prognosis for survival (19,20).
The wild-type function of PLZF is not known; however, its homology with the Drosophila gap gene Kruppel (21,22) and the observation that it functions as a repressor of transcription (23,24) suggests that it plays a role in developmental programs of gene expression. Recent studies suggest that PLZF may play a specific role in hematopoiesis, perhaps in the maintenance of the phenotype of uncommitted hematopoietic progenitors and/or controlling the commitment of these cells to differentiation (25)(26)(27). Moreover, Dong and co-workers have recently shown that the BTB/POZ region of the PLZF-RAR␣ protein display a dominant-negative effect against retinoic acid receptor function (24). These observations, together with the fact that the BTB/POZ domain is present in all APL patients so far identified that express PLZF-RAR␣ fusion proteins (18), strongly suggests that the BTB/POZ domain of the PLZF protein plays an important role in the pathogenesis of APL.
We have employed gel filtration, dynamic light scattering, equilibrium analytical ultracentrifugation, equilibrium denaturation, and differential scanning calorimetry measurements to show that the isolated BTB/POZ domain from the human PLZF protein forms a very tight and highly stable homodimeric species. Analysis with circular dichroism also suggests that the * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
ʈ To whom correspondence should be addressed: BTB/POZ dimer has a high degree of secondary structure including a significant helical content. To facilitate a three-dimensional structure determination of the PLZF BTB/POZ domain, we have prepared crystals of the PLZF BTB/POZ domain that are suitable for x-ray crystallographic analysis. The crystals form in the space group I222 or I2 1 2 1 2 1 with a ϭ 38.8, b ϭ 77.7, and c ϭ 85.3 Å and contain 1 protein subunit per asymmetric unit with approximately 40% solvent. We have collected a complete 2.25 Å data set from a crystal that has been frozen at Ϫ170°C.

EXPERIMENTAL PROCEDURES
Overexpression-The DNA segment encoding residues 1-118 of PLZF comprising the BTB/POZ region (PLZF-BTB/POZ) was amplified from a pSG5 expression vector harboring the human PLZF cDNA (gift from A. Zelent, Institute of Cancer Research) using polymerase chain reaction with pairs of oligonucleotide primers in which PLZF specific primers were linked to BamHI restriction sites. This amplified fragment was subcloned into the BamHI site of the PQE30 T5-polymerasebased expression vector and transformed into Escherichia coli S9 cells (28). Plasmid from single clones were checked for insertion and correct directionality of the PLZF-BTB/POZ insert by restriction digest. The sequence of clones that contained the PLZF-BTB/POZ insert in the correct orientation were confirmed by dideoxynucleotide sequencing (Wistar DNA Core Facility). The expressed protein contained a 12residue N-terminal His 6 tag sequence MRGSHHHHHHGS, followed by residues 1-118 of the PLZF protein (PLZF-BTB/POZ).
Recombinant PLZF-BTB/POZ protein was produced by growing PLZF-BTB/POZ-transformed S9 cells to an A 595 of 0.5 and inducing with 0.5 mM isopropyl-1-thio-␤-D-galactopyranoside for 3 h. Small scale inductions in Luria broth followed by cell disruption by sonication and centrifugation to separate the soluble and insoluble protein fraction showed that the protein was partitioned about 50/50 between the soluble and insoluble fractions when cells were induced at 37°C. PLZF-BTB/POZ-transformed S9 cells that were induced at 28°C resulted in the partitioning of the majority of the PLZF-BTB/POZ protein in the soluble protein fraction.
Purification-For protein purification, 300 ml of saturated PLZF-BTB/POZ-transformed S9 culture was used to inoculate a total of 6 liters of Luria broth (50 ml of saturated culture/1 liter of Luria broth) that had been pre-warmed to 28°C. Cells were grown to an A 595 of 0.5 and induced with 0.5 mM isopropyl-1-thio-␤-D-galactopyranoside, and growth was continued until the A 595 reading plateaued. This typically took about 8 h. In our early purification attempts, we had noted that, once partially purified, the PLZF-BTB/POZ protein became insoluble when the pH was below 8.5. We subsequently used a purification strategy in which the protein was isolated in a buffer solution containing 40 mM Tris, pH 8.5. Following cell disruption by sonication in 100 ml of a buffer containing 40 mM Tris, pH 8.5, 100 mM NaCl, 1 mM ␤mercaptoethanol, and 100 g/ml phenylmethylsulfonyl fluoride (LS buffer), the cell extract was centrifuged, and the supernatant was applied directly to a 10-ml Q-Sepharose column that had been preequilibrated with LS buffer. The column was washed with 10 column volumes of LS buffer containing 0.25 M NaCl, and protein was eluted with a linear gradient of salt (0.25-0.4 M NaCl) in the same buffer. The fractions containing PLZF-BTB/POZ were pooled and adjusted to 0.75 M (NH 4 ) 2 SO 4 prior to application to a 10-ml phenyl-Sepharose column, which had been pre-equilibrated with LS buffer containing 0.75 M (NH 4 ) 2 SO 4 . After washing the column with 10 column volumes of LS buffer containing 0.75 M (NH 4 ) 2 SO 4 , protein was eluted with a linear gradient of salt (0.75-0 M (NH 4 ) 2 SO 4 ) in LS buffer. Peak fractions containing PLZF-BTB/POZ protein were pooled and concentrated to about 2 ml by ultracentrifugation using a Centriprep-10 microconcentrator (Amicon Inc.). The concentrated protein was purified further using size exclusion chromatography on a Superdex-200 FPLC column (Pharmacia Biotech Inc., 16 ϫ 600 mm). Peak fractions containing PLZF-BTB/POZ protein were concentrated to about 50 mg/ml by centrifugation using a Centricon-10 microconcentrator (Amicon Inc.) in LS buffer. 50-l protein aliquots were finally frozen at Ϫ70°C and thawed for use as needed. We typically obtained about 60 mg of purified protein from one 6-liter preparation of bacterial culture.
Protein concentration was determined spectroscopically by absorbance at 280 nm using a molar extinction coefficient of 7450 M Ϫ1 cm Ϫ1 predicted from the amino acid sequence (29). The identity of the purified PLZF-BTB/POZ protein was confirmed by MALDI mass spectroscopy FIG. 1. Sequence alignment of the BTB/POZ domains. Among the 40ϩ BTB/POZ domains known to date, a panel of 19 representative sequences that are most homologous with PLZF are aligned (CLUSTAL program) and displayed (BOXSHADE program). Black and gray backgrounds are used to indicate identical and/or conserved residues found in at least 50% of the proteins at a given position, respectively. and N-terminal sequencing (Wistar Institute Protein Microchemistry Facility).
Analytical Ultracentrifugation-The PLZF-BTB/POZ protein was dialyzed against a buffer containing 40 mM boric acid, pH 8.5, and 100 mM NaCl prior to analysis. Sedimentation equilibrium experiments were carried out at 4°C in a Beckman XL-A analytical ultracentrifuge with two-or six-sector cells using three rotor speeds (20,000, 30,000, and 42,000 rpm) and at three different protein concentrations (0.2, 0.1, and 0.05 mg/ml). After samples had reached equilibrium (typically after 15 h), they were scanned at 280 and 230 nm. Data were analyzed with a nonlinear least squares fitting program based on IGRO-PRO (provided by P. Hensley, Smith Kline Beacham Pharmaceuticals) using a partial specific volume of 0.7347 ml/g calculated from the protein amino acid composition (30).
Differential Scanning Calorimetry-DSC measurements were performed using a MicroCal MCS Calorimeter. Protein solutions were dialyzed against a buffer containing 40 mM boric acid, pH 8.5, and 100 mM NaCl prior to loading the sample cell. An aliquot of the dialysate was used in the reference cell. The protein concentration was 1.0 mg/ml (70 M). All scans were from 10 to 90°C at a scan rate of 90 degrees/h. ORIGIN software (MicroCal, Inc.) was used for data analysis, which involved subtracting a buffer base line from the raw data and then defining a base line using a progress curve fitted to the end of the transition. The constructed base-line was then subtracted, and the data were curve-fitted using standard models to determine T m , calorimetric heat change, van't Hoff heat change, and the heat capacity associated with the thermal unfolding.
Dynamic Light Scattering-All measurements were carried out on a DynaPro-801 molecular sizing instrument at 20°C. Samples were typically run at a concentration of 4 mg/ml (280 M) and pre-filtered through a 0.02-m membrane filter (Whatman, Anodisc 13) prior to analysis. Sample buffer typically contained 50 mM Tris-HCl, pH 8.5, 50 mM NaCl, 1 mM ␤-mercaptoethanol.
Circular Dichroism Spectroscopy-CD spectra were carried out on a Jasco J-720 spectropolarimeter at 25°C. The far-UV CD spectra were recorded using a 100-l cell containing a 0.2-mm path length. The sample was typically at a concentration of 1 mg/ml (67 M) in a buffer containing 20 mM boric acid, pH 8.5, 50 mM NaCl. Spectra were analyzed using the SOFTSPEC software supplied by the manufacturer. [ 222 ] f and [ 222 ] u were obtained at the base lines of the transition curves, at which [ 222 ] obs became relatively invariant at changing guanidine hydrochloride concentrations. The Gibbs free energy for each of the partially unfolded states was calculated assuming a two-state dimer denaturation as supported by the DSC results described above. K u for unfolding was calculated using the equation were P t is the total protein concentration and f u is the fraction of unfolded protein at a particular guanidine hydrochloride concentration. The Gibbs free energy for unfolding at a particular guanidine hydrochloride concentration was then calculated with the equation ⌬G ϭ ϪRT ln(K u ), and the Gibbs free energy for protein unfolding in water (⌬G U (H 2 O)) was calculated by extrapolating to a value of ⌬G at zero guanidine hydrochloride (31).
Crystallization-Crystallization screens employed the vapor diffusion crystallization technique using 24-well culture plates (Linbro) (32). Typically, a 2-l solution containing protein, salts, buffer, and precipitating agent were equilibrated against a 1-ml reservoir containing salts, buffer, and precipitating agent at twice the concentration contained in the protein drop. We employed three broad factorial crystallization screens (33)(34)(35). Each of the factorial screens were conducted at two temperatures, 20 and 4°C.
Crystallographic Data Collection and Processing-Diffraction data were collected on an R-AXIS II image plate with an MSC/YALE focusing mirror system using CuK ␣ radiation from a Rigaku RU-200 x-ray generator operating at 100 mA and 50 kV. Data were processed with the programs DENZO and SCALEPACK (36).

Purification of the BTB/POZ Domain from the PLZF Protein-
We have successfully overexpressed and purified to homogeneity the BTB/POZ region from the PLZF protein (residues 1-118) in bacteria using a PQE30 expression plasmid in S9 cells (Fig. 2A). The protein preparation yields about 10 mg of purified protein per liter of bacterial cell culture. The availability of large amounts of purified recombinant protein has allowed us to study the biochemical and biophysical properties of this highly conserved BTB/POZ region from the PLZF protein.
Oligomerization Properties of the BTB/POZ Domain from the PLZF Protein-Several biochemical/biophysical techniques, including gel filtration chromatography, dynamic light scattering, and equilibrium analytical ultracentrifugation, show that the recombinant PLZF-BTB/POZ protein exists as a stable dimeric species under near physiological conditions. Size exclusion chromatography of the PLZF-BTB/POZ protein at a concentration above 2 mg/ml (140 M) shows that it elutes between the horse myoglobin (17.0 kDa) and the chicken ovalbumin (44.0 kDa) protein standards (Fig. 2B). A plot of the logarithm of the molecular mass of the protein standards against the elution volume predicts that the molecular mass of the PLZF-BTB/POZ species is 30 kDa. This predicted molecular mass is most consistent with a dimeric state for the PLZF-BTB/ POZ protein since the molecular mass for a PLZF-BTB/POZ dimer calculated from the protein sequence is 29.97 kDa.
Analysis of the PLZF-BTB/POZ protein by dynamic light scattering, using a DynaPro-801 molecular sizing instrument, shows that the protein has a Stokes radius of 2.7 nm, corresponding to a globular molecular mass of 32 kDa. As with the gel filtration analysis, this experiment suggests that the PLZF-BTB/POZ protein exists as a dimer.
To obtain an estimate of the molecular weight of the multimeric PLZF-BTB/POZ species that is independent of molecular shape and to investigate potential monomer-dimer equilibria, we performed equilibrium sedimentation experiments using analytical ultracentrifugation (Fig. 3). We performed these experiments in a buffer that was relatively transparent at 230 nm so that we could obtain information in the nanomolar concentration range of protein. Equilibrium sedimentation experiments were carried out at three rotor speeds (20,000, 30,000, and 42,000 rpm) and at three different initial protein concentrations (0.2, 0.1, and 0.05 mg/ml). The data from each run were fitted to a single molecular weight species yielding an average molecular mass of 29,956 Ϯ 593 Da over a protein concentration range of 33 M to 200 nM. A representative curve fit at a rotor speed of 30,000 rpm and at a protein concentration of 0.05 mg/ml is shown in Fig. 3. This experiment shows that the PLZF-BTB/POZ domain is dimeric above a concentration of 200 nM.
Taken together, gel filtration, dynamic light scattering, and sedimentation analytical ultracentrifugation experiments show that the predominant oligomeric state for the PLZF-BTB/ POZ protein is dimeric above a concentration of 200 nM under near physiological conditions. Stability of the BTB/POZ Domain from the PLZF Protein-We performed both DSC and guanidine hydrochlorideinduced unfolding that was monitored by CD spectroscopy to investigate the stability of the PLZF-BTB/POZ dimer.
DSC experiments showed that the PLZF-BTB/POZ fragment undergoes a single irreversible two-state transition with precipitation of the sample (Fig. 4A). The protein is surprisingly heat stable with an estimated T m of 70.4°C and an estimated enthalpy change of 48 kcal/mol (Fig. 4B). The ratio of calorimetric heat change over the van't Hoff heat change was approximately 0.47, suggesting that the protein dimer undergoes a single coupled two-state transition during unfolding.
The PLZF-BTB/POZ protein shows a strong mean residue ellipticity at 222 nm, suggesting a significant helical content. The ellipticity at 222 nm was used to evaluate unfolding of the PLZF-BTB/POZ protein in the presence of the denaturant guanidine hydrochloride. Varying concentrations of guanidine hy- ). All data were collected using a scan rate of 90°C/h. drochloride were added, and the mean residue ellipticity was monitored at 222 nm. This analysis shows that unfolding is sigmoidal and single phase (Fig. 5). Moreover, the midpoints of the transition curves increased with increasing protein concentrations, further supporting the DSC experiments and suggesting a coupled two-state unfolding transition from folded dimer to unfolded monomers. The denaturation profiles of the PLZF-BTB/POZ protein were used to calculate an average ⌬G U (H 2 O) of 12.8 Ϯ 0.4 kcal/mol. Comparable Gibbs free energies for unfolding have been observed for other proteins of about the same size that form stable homodimers, including HIV-1 protease, and Trp aporepressor (31). Taken together the DSC and CD measurements show that the PLZF-BTB/POZ domain of PLZF forms an extremely stable, potentially intertwined dimer.
Secondary Structure Content of the BTB/POZ Domain from the PLZF Protein-CD spectra of the PLZF-BTB/POZ domain shows that the protein has a high degree of secondary structure (Fig. 6). Analysis of the spectra using the SOFTSPEC software supplied by the manufacturer shows an excellent agreement with a protein containing 45% helix, 20% ␤-sheet, 11% turn, and 24% random coil. These experimental values are consistent with secondary structure predictions suggesting that about 50% of the BTB/POZ domain is helical (27).
Crystallization and Diffraction Properties of PLZF-BTB/ POZ Crystals-The crystallization trials for the PLZF-BTB/ POZ protein produced several different crystal forms at 20°C; however, none of these crystals diffracted x-rays to beyond 10 Å resolution. We obtained two crystal forms at 4°C that were small but nicely shaped. We optimized crystal growth conditions for one of these forms and were able to reproducibly prepare crystals of typical size 0.2 ϫ 0.2 ϫ 0.5 mm in size. The crystals were prepared using 2-l hanging drops containing 10 mg/ml PLZF-BTB/POZ protein, 8% isopropyl alcohol, 600 mM MgCl 2 , 50 mM Tris, pH 8.5, 50 mM HEPES, pH 7.5, equilibrated over a reservoir containing 2 times the concentration of salts, buffer, and precipitating agent. Crystals were transiently transferred (for about 5 min) to a harvest solution composed of salts, buffer, and precipitating agent at the same concentrations as the reservoir solution with the addition of 25% glycerol to facilitate x-ray data collection at cryogenic temperature (Ϫ170°C). The diffraction quality of the crystals show that they are well ordered, showing strong diffraction to beyond 2.3 Å resolution, indicating that they would be suitable for structure determination using x-ray crystallographic techniques.
A native diffraction data set (2.25 Å, R sym ϭ 6.8%) was collected from a crystal at Ϫ170°C (Table I). Analysis and reduction of the x-ray data using the DENZO and SCALE-PACK programs (36) shows that the crystal forms in the space group I222 or I2 1 2 1 2 1 with a ϭ 38.8, b ϭ 77.7, and c ϭ 85.3 Å and contain 1 protein subunit per asymmetric unit, with approximately 40% solvent content in the crystals.

DISCUSSION
Oligomerization Properties of Proteins Containing a BTB/ POZ Domain-The biochemical/biophysical properties of the BTB/POZ region from the PLZF protein presented in this study show that it forms a highly stable homodimer. We have established an upper limit of 200 nM as a dissociation constant for the dimer, strongly suggesting that the PLZF protein exists as a dimer in vivo. Consistent with this is the apparent instability of the BTB/POZ monomer as determined by DSC and equilibrium denaturation experiments. The extensive homology within the BTB/POZ region of homologous proteins (Fig. 1) suggest that they also form dimers in vivo. This is supported by several in vivo and in vitro studies showing that the BTB/POZ domain from a variety of proteins mediate homodimerization and in some cases heterodimerization.
Specifically, Laski and co-workers (14) have employed GST fusion proteins and cross-linking experiments to show that the BTB/POZ region of the Drosophila bric a brac protein mediates dimerization in vitro. These investigators also localize the dimerization surface to the N-terminal 51 residues of the BTB/ POZ domain and show that mutation of several highly conserved residues in this region disrupt the proteins oligomerization properties. Bardwell    proteins can form homodimers in vitro. Moreover, they demonstrate that the BTB/POZ domain of the Drosophila Ttk protein form homodimers as well as heterodimers with the BTB/POZ region from the Drosophila GAGA protein. Taken together, a key property of the BTB/POZ domain appears to be to direct formation of particular homo and heterodimeric protein complexes.
Involvement of the BTB/POZ Domain in PLZF-mediated APL-The PLZF-RAR␣ fusion receptor generated as a result of a t(11;17) chromosomal translocation that occurs in a small subset of APL patients has been shown to display a dominantnegative effect against retinoic acid receptor function (18). Recently, Dong et al. (24) have shown that the BTB/POZ region of the PLZF protein mediates this activity. These authors also show that the PLZF-RAR␣ fusion protein could heterodimerize in vitro with wild-type PLZF protein, suggesting that the BTB/ POZ domain may play a significant role in leukemogenesis by antagonizing not only the retinoid receptors but also PLZF and possible other BTB/POZ-domain containing regulators. Taken together, the dimerization properties of the PLZF BTB/POZ domain appears to be strongly linked to the molecular pathogenesis of APL.
The studies presented here have contributed substantially to our understanding of the dimerization properties of the PLZF protein. The structure of the PLZF BTB/POZ region that will be afforded by the crystals obtained in this study will provide the high resolution chemical details underlying dimer stability and specificity. This information will provide a framework for understanding the structural basis underlying PLZF-RAR␣-mediated APL and for understanding the function and dimerization properties of other BTB/POZ-containing proteins.