Purification and Characterization of a Functionally Homogeneous 60-kDa Species of the Retinoblastoma Gene Product*

The retinoblastoma susceptibility gene (RB) encodes a 928-amino acid protein (pRB) that is hypothesized to function in a pathway that restricts cell proliferation. The immortalizing proteins from three distinct DNA tumor viruses (SV40 large T antigen, adenovirus Ela, and human papilloma virus Type 16 E7) have been shown to interact with RB protein through two noncontiguous regions comprised of amino acids 393-572 (domain A) and 646-772 (domain B). We constructed a truncated form of RB (RB p60) that retains these two domains but eliminates the N-terminal386 amino acids of RB. RB p60 was expressed in Escherichia coli in inclusion bodies. After solubilization, it was refolded in the presence of magnesium chloride, and the active protein was isolated with an E7 peptide affinity col- umn. The protein that elutes from this column is functionally homogenous in its ability to bind immobilized E7 protein. Thermal denaturation studies provide additional evidence for the conformational homogeneity of the isolated protein. This purification scheme allows the isolation of significant amounts of RB p60 protein that is suitable for structural and functional studies.

The retinoblastoma susceptibility gene (RB) encodes a 928-amino acid protein (pRB) that is hypothesized to function in a pathway that restricts cell proliferation. The immortalizing proteins from three distinct DNA tumor viruses (SV40 large T antigen, adenovirus Ela, and human papilloma virus Type 16 E7) have been shown to interact with RB protein through two noncontiguous regions comprised of amino acids 393-572 (domain A) and 646-772 (domain B). We constructed a truncated form of RB (RB p60) that retains these two domains but eliminates the N-terminal386 amino acids of RB. RB p60 was expressed in Escherichia coli in inclusion bodies. After solubilization, it was refolded in the presence of magnesium chloride, and the active protein was isolated with an E7 peptide affinity column. The protein that elutes from this column is functionally homogenous in its ability to bind immobilized E7 protein. Thermal denaturation studies provide additional evidence for the conformational homogeneity of the isolated protein. This purification scheme allows the isolation of significant amounts of RB p60 protein that is suitable for structural and functional studies.
The retinoblastoma susceptibility gene is a member of a growing collection of genes that encode proteins that perform a growth suppressor function in the cell (for a recent review, see Ref. 1). Mutational inactivation of the retinoblastoma susceptibility gene has been linked to the development of pediatric retinoblastomas and has more recently been associated with other types of cancers, including small cell lung carcinoma and breast cancer (2, 3). In normal cells, pRB undergoes changes in its phosphorylation state that correlate to specific stages in the cell cycle (4-6). It has been hypothesized that the retinoblastoma gene product regulates growth of normal cells, since its inactivation by genetic changes leads to uncontrolled cellular proliferation (2, 3). An alternative * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
7 To whom correspondence should be addressed Dept. of Cancer Research, Bldg. WP16-101, Merck Sharp and Dohme Research Laboratories, West Point, PA 19486. Tel.: 215-661-3074;Fax: 215-661-7320. mechanism for RB' "inactivation" has been proposed to explain the immortalizing effects of proteins from three different viruses, SV40 large T antigen, adenovirus ElA, and HPV-16 E7. Immunoprecipitation studies demonstrate that these viral proteins form complexes with pRB under physiological conditions (7-10). This interaction may be crucial to the immortalizing function of these proteins, since by binding pRB, the viral proteins may prevent pRB from interacting with normal cellular partners, thus effectively inactivating it. Several candidate cellular pRB-binding proteins have recently been identified, including the cellular transcription factor E2F (11-16).
Antagonists of the E7/pRB interaction might be useful for the treatment of cervical cancer (17). An understanding of the molecular basis of this interaction would facilitate the design of these antagonists. Sequence homologies between E7, ElA, and SV40 large T antigen suggest a relatively small contact site on these proteins (9, 18). Peptides based on this sequence homology proved effective at blocking binding of E7 to pRB (19,20). Mapping of the contact domain on pRB has proved more difficult. Deletion analysis indicates the existence of two noncontiguous domains within pRB that are necessary for viral protein binding: amino acids 393-572 (domain A) and 646-772 (domain B) (21-23). The individual amino acids that make actual contact with the viral proteins during complex formation are currently unknown.
In order to isolate sufficient quantities of RB protein for structural and functional studies, we constructed a truncated form of RB protein that deletes the N-terminal 386 amino acids but retains domains A and B (RB p60). The protein was expressed in Escherichia coli and isolated from inclusion bodies. A novel refolding and affinity purification scheme afforded functionally homogeneous RB p60.

MATERIALS AND METHODS
Construction and Expression of RB p60-A human lung Xgtll cDNA library (Clontech, Palo Alto, CA) was screened for RB clones using end-labeled oligonucleotide-specific probes based on the published RB sequence (24). One of the clones identified contained a 1.9kilobase EcoRI insert that encompassed amino acids 300-928 of the published RB sequence. This fragment was subcloned into M13, and single-stranded DNA was prepared. Using site-specific mutagenesis, a unique SpeI site was engineered in at amino acid position 386. Cutting with SpeI makes the methionine at amino acid position 387 the initiating methionine. The SpeIIEcoRI fragment was ligated into a PUC19 vector, and then the pTAC-RB195 plasmid was generated by digesting the PUC19 plasmid with Hind111 and EcoRI and subcloning into the pTAC bacterial expression plasmid (25) (Fig. 1). The resultant clone was identified by hybridization to labeled oligonucleotide probes and by sequencing.
Fermentation-Fermentation of E. coli containing the plasmid pTAC-RB195 was performed from frozen vials of cells. The cells were grown at 37 "C in 3 X LS medium (15 g/liter yeast extract (Difco), 30 g/liter soya peptone (Sheffield), 10 g/liter sodium chloride) plus 100 mg/liter ampicillin, until they reached midlog growth (&w = 5, 6 h). Isopropylthiogalactoside was added at a final concentration of 1 mM, and 2 h later (A600 = lo), the cells were harvested by membrane filtration and subsequent centrifugation. The cell paste, about 2 kg from 200 liters, was then frozen and stored at -70 "C.
Purification and Refolding-All steps were carried out at 4 "C The The plasmid used to express RB p60 in E. coli pTAC-RB195 contains a pTAC promoter followed by an open reading frame that begins with 12 amino acids resulting from cloning procedures, followed by RB-(387-928).
unless otherwise noted. Cell paste (approximately 200 g) was resuspended in phosphate-buffered saline (PBS, 6 mM sodium phosphate, 150 mM NaCI, pH 7.2) containing 0.1 mM phenylmethylsulfonyl fluoride, 1 mM MgC12,O.Ol mg/ml DNase I, and 0.1 mg/ml lysozyme at 2.5 ml/g cell paste. The mixture was stirred for 30 min, and the cells were then lysed by sonication. An additional 0.5 volume of PBS was added, and the mixture was centrifuged for 20 min a t 13,000 X g. PBS and solid urea were added to the pellet to give a final urea concentration of 6 M and a volume of 3 ml/g cell paste. Triton X-100 was added to 1%, and the mixture was stirred for 1 h. After dilution with 1 volume of PBS, the solution was centrifuged for 20 min a t 13,000 X g. The pellet was washed with PBS and then resuspended in 50 mM Tris-CI, pH 8.0,50 mM NaCI, 1 mM EDTA. Solid guanidine hydrochloride was added to give a final concentration of 6 M guanidine hydrochloride and approximately 2 ml/g cell paste. Following a brief incubation, the solution was centrifuged for 20 min a t 13,000 X g. The supernatant was then frozen.
Subsequent purification steps were typically performed on a 10-ml aliquot of the guanidine extract (resulting from approximately 7 g of cell paste). Solid dithiothreitol was added to 0.1 M, and the mixture was stirred a t 30 "C for 30 min. Refolding was achieved by a dropwise 100-fold dilution of the extract into 1 liter of buffer containing 0.1 M Tris-C1, pH 8.0, 0.4 M MgCI,, 0.6 M guanidine hydrochloride, 0.1 mM EDTA, 0.1 mM DTT, and a protease inhibitor mixture (10 pg/ml benzamidine and 5 pg/ml leupeptin, pepstatin A, and aprotinin). The refolding mixture was stirred for at least 5 h, and then centrifuged for 60 min at 10,000 X g. The buffer was changed to 50 mM HEPES, p H 7.0, 250 mM NaCl, 0.1% Nonidet P-40, and 1 mM D T T by crossflow filtration using an Enka Lab cross-flow cartridge (Microdyne, Raleigh, NC). The dialysate was centrifuged for 1 h a t 11,000 X g and was then sequentially filtered through 0.45-and 0.2-pm filters.
The filtrate (approximately 1 liter) was sequentially passed through a 2-ml ethanolamine-capped Affi-Gel15 (Bio-Rad) guard column and a 2-ml E7-(20-29) peptide amide Affi-Gel 15 column a t a flow rate of approximately 1 ml/min. The columns were washed with 50 ml of dialysis buffer, and the guard column was then disconnected. The affinity column was eluted with unbuffered 50 mM Na2C03, 1 mM DTT. Three-ml fractions were collected in tubes containing 0.3 ml of 1 M HEPES, pH 7.0, to neutralize the elution buffer. The active RB p60 typically eluted in fractions 2-4. Protein was quantitated by the Bradford assay (26) with standardization by amino acid composition analysis.
RB p60/E7 Binding Assays-RB p60 activity was determined by assessing binding to HPV16-E7 in two different ELISA assays. The assays have been described in detail elsewhere (19,20). Briefly, in the standard ELISA, binding of RB p60 to immobilized E7 protein was quantitated using an anti-RB p60 monoclonal antibody (E. Harlow, Massachusetts General Hospital, Boston). Evidence for a single class of E7-binding sites on RB p60 was obtained in the "reverse" ELISA, in which binding of E7 to RB p60 immobilized via an anti-RB p60 monoclonal antibody was quantitated using an anti-E7 polyclonal antibody.
Circular Dichroism Spectroscopy-Circular dichroism spectra on purified RB p60 were obtained on an Aviv 62-DS thermostatted instrument. The protein was dialyzed into either 6 mM sodium phosphate, 0.1 mM DTT (Fig. 4), or into PBS containing 0.1 mM D T T (Fig. 5). Isothermal spectra were obtained a t 4 "C. Estimates of protein secondary structure were obtained by fitting of the experimental spectrum into a combination of basis set protein spectra using the variable selection method (27). The best fit was selected that met the following criteria: ( a ) the root mean square errors between the experimental and calculated values were less than or equal to 0.2 and ( b ) the sum total of fractional contributions of different secondary structures was between 0.9 and 1.10. Thermal denaturation in the presence or absence of peptides was obtained by measuring the ellipticity a t 222 nm as the temperature was increased from 6-90 "C in 2-degree intervals.
Peptide Chemistry-Peptides were synthesized on an Applied Biosystems 430A peptide synthesizer using a t-butoxycarbonyl protection scheme and standard solid-phase peptide synthesis methodology. as previously described (19,20,28). After purification, peptides were >95% pure, as determined by high performance liquid chromatography. The identity of the peptides and peptide content were determined by amino acid composition and fast atom bombardment mass spectroscopy. The sequence of the E7-(20-29) peptide amide is TDLY-CYEQLN-amide. The sequence of the scrambled peptide is YNELCQYDL-amide.

RESULTS
Upon induction of the pTAC-195 E. coli expression system with isopropylthiogalactoside, RB p60 was produced a t approximately 2% of the total cell protein ( Fig. 2 -4 ) . After lysing the cells by sonication, virtually all the induced protein was recovered in the centrifugation pellet. The RB p60 in the lysis pellet remained insoluble when resuspended in 6 M urea, 1% Triton-X100, whereas other contaminating materials dissolved and were removed by centrifugation. Solubilization of RB p60 protein was achieved in 6 M guanidine hydrochloride, 0.1 M DTT.
Many attempts were made to refold RB p60 by dilution or dialysis from the guanidine/DTT solution into a wide variety of buffers, but this generally resulted in precipitation of the protein, with no soluble RB p60 capable of binding E7. Under certain conditions, the presence of 0.4 M MgC12 in the diluting buffer appeared to facilitate proper folding of RB p60, as determined by the E7 binding assay (Table I). The role of the MgC12 in promoting refolding of the protein is unclear, since bound Mg" does not appear essential to RB p60's E7-binding activity (data not shown).
Although the majority of the RB p60 was soluble after the refolding step, it was assumed to be conformationally heter-  As determined by the Bradford assay (26). As determined in the ELISA assay using immobilized E7 protein. Expressed in units X IOd6; 1 unit is defined as the amount of pRB giving the same absorbance in the ELISA assay as 50 pl of T24 cell lysate at 4 X 1 0 ' cells/ml. Expressed in units X 10-6/mg of protein.
ND, not determined. e Activity recovered upon dilution of guanidine extract into assay buffer.
." ogeneous. An affinity purification step was devised that selects for the specific biochemical interaction of most interest, namely binding to the HPV-16 E7. Because this interaction requires the participation of two sequentially discontinuous domains on pRB, we hypothesized that it would provide a rigorous conformationally sensitive purification step.
An E7-(20-29) peptide amide Affi-Gel 15 column was pre- pared in order to affinity purify the active RB p60. Magnesium chloride in the RB p60 refolding buffer was first removed by dialysis against a HEPES/NaCl buffer, since magnesium chloride was inhibitory to the RB p60-E7 interaction in the pRB/E7 binding assay (data not shown). When the refolded mixture was passed over the E7-(20-29) peptide amide Affi-Gel 15 column, the contaminating proteins, as well as most of the RB p60, did not bind to the column. The column flowthrough contained virtually no E7-binding activity, however. Upon elution of the column with 50 mM Na2C03, a small amount of the original RB p60 was recovered. Analysis of the column eluate in the binding assay indicated that 70% of the activity in the column load was recovered. A single dominant band is present in the SDS-PAGE profile (Fig. 2B) with a number of lower molecular weight protein bands evident as well. Western blot analysis of the eluate verifies that most of these lower molecular weight bands represent RB p60 degradation products (data not shown). This purification scheme yields approximately 0.5 mg of active RB p60/g of cell paste. The E7-binding activity of the affinity column eluate was verified by rebinding the eluted RB p60 to the affinity resin in a batchwise manner. After binding at 4 "C for 20 min, the protein was again eluted with Na2C03. Virtually all the RB p60 protein was recovered upon elution (Fig. 2C). These results confirm that the affinity column elution protocol causes no significant irreversible loss of E7-binding activity.
Although virtually all the protein eluted from the E7 peptide affinity column displays E7-binding activity, this is not conclusive evidence of functional homogeneity, since the eluted RB may display a range of affinities for the immobilized E7 peptide. To investigate this possibility, RB p60 was immobilized by adsorption to a monoclonal anti-pRB antibody in a 96-well plate. The E7 titration curve of this immobilized RB p60 protein is consistent with a single class of E7-binding sites (Fig. 3) (29). These results suggest that the RB p60 eluted from the E7 affinity column is functionally homogeneous.
As a first step toward the elucidation of the structure of RB p60, we performed a series of circular dichroism studies on the affinity-purified protein. The ultraviolet circular dichroism spectrum of affinity-purified RB p60 is shown in Fig.  4. Analysis of this spectrum by the variable selection method (see " Materials and Methods") suggests that this protein contains approximately 39% a-helix, 16% j3-sheet, 22% turn, and 30% unordered conformations. The addition of E7-(20-29) peptide amide to RB p60 at a 10-fold molar excess (17 PM) had no effect on the spectrum (data not shown).
The thermal stability of the secondary structure of RB p60 was examined by monitoring the ellipticity at 222 nm as a function of temperature. This wavelength was selected because of the relatively large percentage of helical structure in RB p60. A sharp melting transition was observed, suggestive of a cooperative thermal denaturation process (Fig. 5A). The midpoint of this transition (T,) was estimated to be at 46 "C from the peak of the first derivative plot of the denaturation data (Fig. 5B). The addition of a 20-fold molar excess of E7-(20-29) peptide amide (30 p~) caused a 12-degree shift in the T,, to 58 "C. At the same concentration, a scrambled E7 peptide (for sequence, see "Materials and Methods") that does not bind RB p60 had no effect on the transition temperature. The same results were obtained whether the experiment was performed in PBS or 6 mM sodium phosphate, pH 7.4. These results demonstrate that E7-(20-29) peptide amide specifically binds to RB p60 and significantly stabilizes it to thermal denaturation. A single melting transition is observed in the presence or absence of E7 peptide, consistent with the hypothesis that the affinity-purified RB p60 contains a single conformational population.

DISCUSSION
As a member of the growing family of growth suppressor genes that have been linked to cancer, there is great interest in understanding the mechanism of action of the retinoblastoma gene product. Mutagenesis studies of RB have provided a great deal of structural information, including the elucidation of two discontinuous domains that appear critical for the high affinity binding of pRB to several virally encoded proteins (21-23). Alternative methods of probing these interactions can be utilized if sufficient quantities of functionally active protein are available. While pRB and fragments of pRB have been previously expressed in baculovirus (30), and as a fusion protein in E. coli ( l l ) , none of these expression/ purification schemes were capable of providing sufficient quantities of chemically and functionally homogenous protein for structural and biophysical characterization.
In the current study, we describe a system for the expression and purification of a functionally homogenous, truncated form of pRB from E. coli. Binding studies with recombinant HPV E7 protein are consistent with a single, high affinity binding site, and thermal denaturation studies exhibit a single thermal transition. In addition, the affinity-purified RB p60 exhibits DNA-binding properties that are virtually identical with those of rabbit reticulocyte transcribed/translated RB p60 (31). Taken together, these studies provide evidence for the functional and conformational integrity of our affinity-purified protein.
Circular dichroism spectroscopy on purified RB p60 provides the first estimate of the secondary structure of func-tional RB protein. In solution, RB p60 appears to contain a relatively high percentage of a-helix, with smaller amounts of @-sheet and turns (27). It is interesting to note that binding of E7-(20-29) peptide amide to RB p60 causes no significant change in the circular dichroism spectrum of the protein, which implies that the overall conformation of RB p60 is unaffected by the binding of this E7 fragment. Nevertheless, the interaction of RB p60 with the peptide affords significant stabilization of the protein to thermal denaturation, with a shift in the a-helix melting temperature of approximately 12 "C.
In conclusion, we have demonstrated a method for the isolation of active, functionally homogeneous RB p60 in sufficient quantities for structural studies. The availability of this protein should allow the initiation of a number of studies that will help clarify, at the molecular level, the mechanism by which the retinoblastoma gene product contributes to the regulation of cell growth.