Purification and Biological Characterization of an Adenovirus Type 2 E1A Protein Expressed in Escherichia coli*

The adenovirus 2 E1A gene encodes a multifunc- tional protein of 289 amino acids that can immortalize primary rodent cells and transcriptionally activate a number of viral and cellular genes. To facilitate an understanding of the molecular basis for the various actions of ElA, we have redesigned our bacterial expression vector (KO, J.-L., and Harter, M. L. (1984) Mol. Cell. Biol. 4, 1427-1439) containing the cloned E1A gene such that a soluble authentic E1A protein now constitutes approximately 1.5% of the total Escherichia coli cellular protein. Further, we have devel- oped a simple rapid purification scheme without the use of detergents or denaturants and show a purity of >98% with a yield of approximately 53%. The E1A so purified is biologically active, stimulating cellular DNA synthesis following microinjection into quiescent NIH 3T3 and REF52 cells. In another report (Span- gler, R., Bruner, M., Dalie, B., and Harter, M. L. (1987) Science 237, 1044-1046) we have also shown that our purified ElA protein activates transcription from appropriate promoters in an in vitro system. The E1A protein encoded by the Ad2l ElA gene is polypeptide of amino acids that been mutational analysis to be composed of at least three functional domains (for review, see Refs. 1 and 2). Each of these domains

The E1A protein encoded by the Ad2l ElA gene is a polypeptide of 289 amino acids that has been shown by mutational analysis to be composed of at least three functional domains (for review, see Refs. 1 and 2). Each of these domains appears to be capable of functioning independently, and each has been shown to be associated with a biochemical activity that is sufficient to either regulate RNA transcription (3), induce cellular DNA synthesis (4, 5), or immortalize primary rat cells in tissue culture (6, 7).
The mechanisms by which the E1A protein operates to fulfill its biological functions are still poorly understood, particularly at the molecular level. Therefore, the availability of this protein in a highly purified form would be quite useful to study its molecular details both in vivo and in vitro. A purified Ad2-Ad5 hybrid E1A-like protein expressed in Esch-* This work was supported in part by National Institutes of Health Grant CA28414, Grant MV-334 from the American Cancer Society, and by Grant 84-567-CCR from the New Jersey State Cancer Commission. 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. erichiu coli (8) has in fact shown some biological activity in vivo. However, this protein was found to be produced in an insoluble form and required the use of strong denaturants for its solubilization. Under these circumstances, the physical condition of the hybrid E1A-like protein might prove to be unacceptable for analysis of its protein structure and function.
We have previously cloned an authentic Ad2 E1A gene in a bacterial plasmid (pKHAO) and have demonstrated its expression in E. coli (9). Moreover, the E1A protein produced in this system was found to be of the exact length and correct amino acid sequence as that synthesized in an Ad2-infected cell. Here we describe the use of a new plasmid expression vector (pKHAO-T) that allows for the efficient expression and purification of our E1A protein in bacterial cells. The 289-amino acid E1A protein is soluble upon cell lysis (constituting approximately 1.0-1.5% of the total E. coli protein) and most importantly, it is purified without the use of detergents or chaotropic agents. We also present evidence that our purified E1A protein microinjected into quiescent NIH 3T3 or REF52 cells is capable of inducing cellular DNA synthesis and is, therefore, biologically active. Furthermore, we have demonstrated that this E1A protein is also functionally active in an in vitro transcriptional system as described elsewhere (10).

RESULTS
Expression of the E l A Protein in Bacteria-We have previously reported (9) the construction of a vector (pKHAO) that produces in E. coli a corresponding adenovirus E1A protein of exact length (289R) and putative correct amino acid sequence as that synthesized in an adenovirus-infected cell. In that plasmid, transcription, of the E1A protein was directed by the lacUV5 promoter, and the amount of E1A ultimately produced in conventional strains of E. coli constituted approximately 0.1-0.3% of the total cellular protein (9).
In order to increase the production of E1A in bacteria, we converted the lacUV5 promoter in the pKHAO plasmid to a promoter (tuc) consisting of the tryptophan promoter and the lac operator and Shine-Dalgarno sequence. This hybrid promoter has already been shown to be significantly more efficient than the lac promoter (15), and like its predecessor, it can be induced to full strength by the lac inducer, isopropyl-@-D-thiogalactopyranoside (details of construction of the new ElA expression vector, designated pKHAO-T, are given under "Experimental Procedures").
Portions of this paper (including "Experimental Procedures," Figs. 1-4, and Tables I and 11) are presented in miniprint at the end of this paper. Miniprint is easily read with the aid of a standard magnifying glass. Full size photocopies are included in the microfilm edition of the Journal that is available from Waverly Press.
The production of the E1A protein in bacterial cells bearing the pKHAO-T plasmid minus the lucIq gene was initially demonstrated in strains of E. coli carrying the lacIq allele on the chromosome. Although the overall production of E1A in these bacterial cells appeared to increase, the bacterial response in allowing this protein to remain stable was far less than we had anticipated (data not shown). We decided then to transfer the pKHAO-T plasmid carrying the hcIq gene to a number of lon-strains (protease-deficient mutants) in the hope of finding a bacterial environment that would confer a lower rate of E1A degradation. To determine whether E1A could accumulate to high levels in any one of those mutants, cultures of cells either with or without pKHAO-T were pulselabeled with [35S]methionine after isopropyl-8-D-thiogalactopyranoside induction and then analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis. One mutant in particular, namely MC102, produced a 45-kDa protein (Fig.  2, lune I ) of considerable amount which was noticeably absent in control cells lacking the pKHAO-T plasmid (Fig. 2, lune  4 ) . Moreover, this protein was of the same size as that previously shown to be synthesized by the original expression vector (pKHAO) when in maxicells (9)  labeled products when immunoprecipitated with the same antibody ( Fig. 2, lune 5 ) , and the 45-kDa protein was not precipitated from extracts of cells either with or without the pKHAO-T plasmid when the control antibody was used (Fig.  2, lunes 3 and 6, respectively). We also examined the effects of other E1A-specific monoclonal antibodies, namely M-1, M-3, and "73 (ll), and all of these behaved similar to M-2 in that each one of them identified a 45-kDa protein only in those cells that carried that pKHAO-T plasmid (data not shown). We can safely conclude then that the modified vector, pKHAO-T, like its predecessor, encodes an authentic E1A protein, and when synthesized in a lon-system the protein appears to remain relatively unaffected by proteolysis.
Purification of the ElA Protein-The purification procedures that we found to be most useful for isolating a biologically active E1A protein are summarized in Table I. This relatively simple and rapid method for purifying E1A was developed with the help of a dot-immunobinding assay (see "Experimental Procedures") that included the use of the three E1A-specific monoclonal antibodies described above. The assay was routinely performed in triplicate after each step of the purification, and this enabled us then to monitor the protein at all times.
The most important aspect of the purification is that the E1A protein is found almost exclusively in the soluble portion of the cell extract and, therefore, did not require at any time the use of strong denaturants for its recovery. The details of the purification of E1A are given under "Experimental Procedures,'' but briefly, proteins in the extracts of MC102 cells were first fractionated by precipitation with ammonium sulfate. After assaying with E1A-specific antibody, we found that E1A precipitated in the 15-40% ammonium sulfate fraction with a yield of 98% (Table I). The partially purified protein was then dialyzed extensively to lower the pH to 5.9 and afterwards applied to a CM-Sepharose CL-GB column. Fig. 3 shows the pattern of the distributed E. coli proteins as a function of absorbance after stepwise elution with KC1 in buffer and also shows the locations of the E1A protein (fractions under the arrow) as determined by dot-immunobinding analysis. Almost all of the contaminating protein (96%) and at least 39% of the total amount of E1A protein passed through the column without binding (Table I; Fig. 3, fractions 1-20). The remainder of the E1A protein was eluted at salt concentrations of 0.15 M (Fig. 3, fractions 71-78) and 0.3 M KC1 (Fig. 3, fractions 90-97) with calculated yields of 6.2 and 52.5%, respectively (Table I). A sodium dodecyl sulfate-polyacrylamide gel electrophoresis analysis of the E1A-containing fractions resulting from each step of the purification is shown in Fig. 44. Unquestionably, the most purified form of the E1A protein can be found in the 0.3 M eluate fractions derived from the CM-Sepharose column (Fig. 4A, lune 5), and judging from the gel scan (data not shown), the purity was more than 98%, even when 5 pg of purified E1A was loaded onto the gel. As expected, the M-2 monoclonal antibody readily identified by immunoblot analysis the purified form of the E1A protein without the appearance of proteolytic products (Fig. 4B, lune  1 ). The CM-Sepharose column then clearly enriches for the E1A protein and more importantly does so under elution conditions which favor high yield both in sample mass (approximately 1 mg/liter of bacterial culture) and biological activity.
Microinjection of Purified ElA into Quiescent Cells-It has been previously shown that a microinjected plasmid encoding the E1A protein can stimulate cellular DNA synthesis in quiescent cells (5). We, therefore, decided to test by microinjection whether our purified E1A could also induce cellular DNA synthesis in quiescent cells. A fundamental way to assay for DNA synthesis is to label growth-arrested cells with [3H] thymidine after microinjection and afterwards screen for the incorporation of radioactivity into cell nuclei by autoradiography. For our study we examined two different established cell lines. As shown in Table 11, the purified E1A protein is able to stimulate cellular DNA synthesis when microinjected into quiescent mouse NIH 3T3 or rat REF52 cells. Compared to the injection of bovine serum albumin which had no effect on DNA synthesis, purified E1A protein at its highest concentration (0.8 mg/ml) triggered DNA synthesis in about 36% of the REF52 cells and 18% of the NIH 3T3 cells. Addition of pure platelet-derived growth factor or 10% fetal calf serum to both of the arrested cell lines induced DNA synthesis in 90% of the cells (data not shown). Collectively, these results show that our purified E1A protein is biologically active, and its efficiency in stimulating cellular DNA synthesis appears to depend on the cell type examined.

DISCUSSION
We have maximized the expression of an authentic Ad2 E1A protein in E. coli by converting the lacUV5 promoter on our former expression vector pKHAO (9) to a promoter (tac) of greater strength (15). The decision to use the tuc promoter instead of the XPL or trp promoter for controlling the expression of E1A was to prevent overproduction of the protein to an insoluble form. Indeed, proteins produced in an insoluble form always require the use of strong denaturants for their solubilization, and treatment of this type usually proves unacceptable for detailed analysis of a protein in an in vitro environment. As demonstrated under "Results," intracellular aggregation of the E1A protein does not appear to be the case, and when the tuc promoter on the new plasmid vector (pKHAO-T) functions in a lon-strain, E1A is synthesized in an amount that approximates 1.2-1.5% of the total cellular protein. The E1A protein also appears to be quite stable in the lon-mutant: proteolytic products do not appear after

P u r i f~~t i o n of the Adenovirus EIA Protein Made in E. coli
precipitating E1A with a number of E1A-specific monoclonal antibodies (11). A protein that is soluble in bacteria can usually be recovered by extractions with neutral aqueous buffer. Such was the case for our E. coli-produced E1A protein, and without the assistance of harsh eluants the protein was purified to approximately 98% homogeneity. The purification procedure described in this paper for E1A is relatively rapid and simple. The first step of the procedure involves an ammonium sulfate fractionation which removes over 62% of the endogenous protein with full recovery of the E1A protein in the 1540% cut. In the next and final step, E1A is enriched by passage of the 15-40% fraction over a CM-Sepharose column. The flowthrough and the 0.15 M eluate fractions from the column remove virtually all of the protein contaminants, and the remaining 0.3 M fractions contain almost exclusively E1A with an overall yield of 53%. It is interesting that a portion of the E1A (39%) passes through the column without initially binding to the CM-Sepharose matrix. This finding is reminiscent of our earlier work (20) involving the study of an unpurified preparation of our E1A on a variety of DNA-cellulose columns. In these experiments, E1A would invariably exhibit at least two types of salt-sensitive activities upon association with DNA-cellulose; however, a certain percentage of the protein would also wash through the column at a low ionic strength. In both of these situations E1A may be interacting with other proteins that are themselves not bound to the columns, or alternatively ElA could be interacting with itself, possibly in the form of a dimer. In fact, we have preliminary evidence3 which suggests that E1A can form such a complex under certain experimental conditions.
Our purified E. coli-produced ElA protein exhibits both in vitro and in vivo biological activities that are characteristic of the Ad2 E1A gene product. In experiments described elsewhere (10) we have demonstrated that our purified plasmidencoded E1A protein has the ability to up-regulate RNA transcription from promoters of Ad2 origin in an in vitro transcription^ system. These results were gratifying since it has been shown in uiuo that the Ad-2 E1A protein can in fact transcriptionally activate (transactivate) other adenovirus genes (21, 22) and certain cellular genes as well (23-25). The structural part of the E1A protein that appears to be responsible for RNA activation has been recently located between amino acid positions 140 and 188 (26, 27). Interestingly, this functional domain contains a potential metal binding sequence of the form Cys-X2-Cys-X13-Cys-X2-Cys, where X may be any amino acid (28). Whether this sequence forms metalbinding units for interacting with DNA or whether the sequence provides a structural site for linking E1A to another type of protein is an important question, particularly since E1A appears to interact with a cellular factor@) during transcriptional activation, both in uiuo and in vitro (10,29). This observation not withstanding, in recent experiments which will be reported elsewhere4 we found that our purified E1A protein binds directly to double-stranded DNA with a fairly high affinity but in a nonspecific fashion.
In this report, we have also demonstrated the capability of our purified E1A to induce cellular DNA synthesis when microinjected into quiescent NIH 3T3 and REF52 cells. These findings support those of a recent study which showed the stimulation of cellular DNA synthesis in growth-arrested cells after microinjection of a plasmid encoding the Ad2 E1A protein (5). Of course, the capability of E1A in triggering the transition from Goto S-phase in the cell cycle of a rodent cell (5) ties in well with its ability to also induce proliferation of quiescent cells (4). The domain that allows E1A to induce cellular DNA synthesis has also been recently identified, and its boundaries begin and end at amino acid positions 40 and 80 in the N-terminal half of the protein (4). Finally, it is worth mentioning that another domain, which stretches from amino acid positions 121 to 140, functionally enables E1A to rescue rat cells from senescence (27, 29), induce mitosis (4), and to repress enhancer-stimulated transcription (27). The enhancers that have been shown to be negatively regulated by E1A are usually viral in origin (30, 31), but cell-specific enhancers have also been shown to be actively repressed by E1A as well (32, 33). In fact, we have found quite recently5 that addition of our purified E1A to an in uitro transcriptional system is sufficient to repress transcription from a promoter with an upstream viral enhancer. Currently, we are investigating the nature of this repression by E1A and its role in possibly displacing a cellular factor that may be binding to the enhancer region for RNA stimulation.
The mechanism by which the Ad2 E1A protein operates is still poorly understood. Our ability to efficiently express and purify an authentic E1A protein that is biologically active provides an opportunity for us to study in detail the physical and biochemical parameters associated with this protein.