DNA-mediated Gene Transfer of a Mutant Regulatory Subunit of CAMP-dependent Protein Kinase*

We have used DNA-mediated gene transfer of ge- nomic DNA to introduce into wild-type Chinese hamster ovary (CHO) cells a mutant gene that confers resistance to the growth inhibitory effect of CAMP. This dominant mutation in CHO cell line 10248 is responsible for an alteration in the RI subunit (RI*) of the type I CAMP-dependent protein kinase (Singh, T. J., Hochman, J., Verna, R., Chapman, M., Abraham, I., Pastan, I. H., and Gottesman, M. M. (1985) J. Biol. Chem. 260, 13927-13933). The transformant 11564 which was studied in detail, has the same characteristics as the original mutant 10248 including continued growth in medium containing Ei-Br-cAMP, an increase in the K,, for CAMP activation of the kinase, a greatly reduced amount of type I1 protein kinase activity, an altered incorporation of the photoaffinity label 8- N3[32P]~AMP into the RI* subunit of PKI, and an absence of CAMP-dependent phosphorylation of a M, = 52,000 protein in intact cells. In addition, analysis of the DNA of the transformant indicates the presence of an increased amount of DNA for the RI gene. These results are consistent with the transfer of a mutant gene for the RI* subunit of the CAMP-dependent protein kinase and its phenotypic expression in the trans- formant and also support the hypothesis that the mutation responsible for the defect in cell line 10248 is due to an alteration in the gene for RI.

subunits appear to be identical in both type I and type I1 kinases (9). One mutant studied in detail, 10248, has an RI with reduced affinity for CAMP (RI*) so that RI is not easily dissociated from C by cAMP (7). In this mutant, type I kinase is present but very little or no type I1 kinase is found. The CAMP-resistance phenotype is expressed even in the presence of a wild-type RI subunit (4) and it is hypothesized that under equilibrium conditions in the cell the mutant RI* subunit, which cannot be dissociated from the C subunit, excludes both wi~d-type RI and RII from binding to C (7).
We are ultimately interested in cloning the normal and mutant genes for CAMP-dependent protein kinase from CHO cells in expression vectors. The dominant mutations such as found in mutant 10248 are of special interest since they might be able to confer their phenotype on a variety of cells. As an initial step toward this goal, we have transferred the RI* gene from mutant 10248 via DNA-mediated gene transfer. The successfd gene transfer reported here formally proves the genetic nature of the mutation, confirms its p h e n o~i c expression in the presence of a wild-type RI gene, and through use of a cDNA probe for the bovine testes RI (22) which detects transfer of the CHO RI gene, confirms that the mutation is linked to the RI gene.
In order to perform the DNA-transfer experiments we have used the Capo4 technique developed by Graham and van der Eb (10) and Wigler et uZ. (11). We have previously shown that it is feasible to transform CHO cells with both purified plasmid DNA and genomic DNA at relatively high frequencies (12). In this work, we have further refined this procedure to eliminate the background of spontaneous CAMP-resistant mutants by using a co-transformation protocol with a twostep selection suggested by Kavathas and Herzenberg (13).

MATERIALS AND METHODS
Cell Lines-CHO cell line loo01 was derived from a CHO Pro-5 strain (14). Cells were grown in monolayer culture as previously described (4). The protein kinase characteristics of the wild-type strain and CAMP-resistant mutant selection and characterization of the dominant CAMP-dependent protein kinase mutant 10248 have been described (4,7). 10248 was selected with 1 pg/ml cholera toxin and 1 mM theophylline, but is also resistant to 8-Br-CAMP (41, which was exclusively used as the selective agent in this study. DNA Isolation and Cell Transformution-High molecular weight DNA was isolated from monolayer cultures as described (12). pSV2-ne0 plasmid (15) was kindly supplied by B. Howard (National Institutes of Health) and plasmid DNA was isolated by standard procedures (16,17). CHO cells were transformed by the DNA-Cap04 precipitate technique (10, 11) but the precipitate was allowed to remain on the cells from 16 to 20 h (12). We used a co-transformation strategy (13) to select the CAMP-resistant cells. Cells were transformed with 1.0 pg of pSV2-ne0 and 2 0~~8 of mutant (10248) or wildtype DNA/2.5 X 105 cells/lO-cm tissue culture plate. A total of 5 X IO6 cells were transformed with wild-type DNA, and 5 X lo6 cells were transformed with mutant DNA. The DNA-CaF'04 precipitate was applied for 16 h and then replaced with normal medium 24 h later. Cells were selected in 800 pg/ml of the neomycin analog G418, (Gibco, and a gift from Schering Corp.) for 10 days. G418-resistant colonies arose at a frequency of 7 X lo-* after transformation with pSV2-ne0 with either mutant or wild-type genomic DNA. At this stage, the G 4 1 8 -r e~s~t colonies were pooled and selected for resistance to 1 mM 8-Br-CAMP in agar for 15 days by plating 10' cells/lOcm plate containing 1 mM 8-Br-CAMP. Resistant colonies were subcloned and analyzed further. Growth Analysis-Subcloned CAMP-resistant clones were tested for growth by plating lo4 cells in 2-cmZ wells with and without 1 mM 8-Br-CAMP and counting different wells on 4 subsequent days as previously described (4). Doubling times were calculated from the average slopes of the growth curves during the first 72 h of growth. Protein Kinase Assays-Protein kinase activity was determined by measuring the transfer of =P from [yS2P]ATP to histone at varying concentrations of CAMP (5,18). Cell extracts were prepared in 10 mM Tris (pH 7 4 0 . 8 mM EDTA, 1 mM dithioth~itol. Homogenates were centrifuged for 20 min at 15,000 X g. The reaction mixture contained 25 mM MES (pH 7.0), 0.5 EGTA, 2.5 mM NaF, 5 mM magnesium acetate, 0.1 mM [y31P]ATP, 0.60 mg of histone and cell extract (60 pg of protein) in a total volume of 0.1 ml. After a 3-min preincubation at 30 "C, the reaction was initiated by the addition of cell extract. After 10 m i n , 30-0 aliquots were removed and spotted on 2-cm squares of Whatman No. 3MM filter paper, then dropped in cold 10% trichloroacetic acid. A no enzyme blank was subtracted from the total incorporation. One unit of protein kinase activity is defined as the amount of the enzyme required to catalyze the incorporation of 1 pmol of =P into histone/min at 30 "C. D~A~-c e~~s e C~~o~t o~h y -P r e p~a t i o n of cell homogenates and DEAE-ce~ulose c~o m a t o~a p h y was ~r f o~e d as previously described (5). Confluent cells (2 X l@) grown on 30 100-mm tissue culture dishes were hafvested, washed with pho~hate-huffered saline, and disrupted in 10 mi of Buf'fer A (10 mM sodium phosphate + 1 mM EDTA (pH 7.6), 1 mM dithiothreitol, and 1% aprotinin (Sigma)) (4). The 30,000 X g supernatant was then applied to a DEAE-cellulose column (0.7 X 10.5 cm) previously equilibrated with Buffer A. The column was then washed with 20 ml of Buffer A before being eluted with a linear gradient of 35 ml of Buffer A and 35 ml of Buffer A containing 0.4 M NaC1. Fractions of 1 ml were collected.
Phsphoyhtion-Intact cells were grown with or without 1 mM 8-Br-CAMP and labeled with 32P as previously described (19). Cells in monolayers were grown in serum-and phosphate-~ontaining medium and were exposed to 1 mM 8-Br-cAW at 37 "C for 3 h. After washing three times with a phosphate-and serum-free medium, cells were incubated at 37 "C with %Pi (100-200 pCi/ml) for 20 min in phosphate-free medium. Incubation was stopped by aspirating the medim and rapidly washing the cells with an excess of cold phosphatebuffered saline solution. Cells were then lysed in a small volume (100-200 pl) of boiling SDS solution (3% (w/v) SDS, 10% (v/v) glycerol, 5% fl-mercaptoethanol, 0.01% bromphenol blue, and 0.05 M Tris, pH 6.8). The samples were transferred to plastic Eppendorf tubes and boiled for 3-4 min prior to analysis by electrophoresis on 10% SDS-acrylamide gels as previously described (19,20).
Southern Analysis-DNA was isolated and digested by restriction enzymes according to the manufacturers' inst~ctions and as previously described (12). Restriction fragments were p r~i p i~~ and run on agarose gels and blotted to nitrocellulose (21) or Genescreen (New England Nuclear). Filters were probed with a 770-base pair PstI fragment of the pRI clone, a cDNA clone homologous to mRNA for the RI gene of CAMP-dependent protein kinase from bovine testes (22). The fragment was isolated from low-melting agarose gels (SeaPlaque, FMC Corp.) and purified through an Elutip (Schleicher and Schuell) column. The same filter was washed free of probe and rehybridized with an actin-hybridizing fragment of a cDNA chickenactin cDNA clone (23). The chicken fragment was a generous gift from G. Merlin0 and was isolated as a HindlII fragment of 550 base pairs. The probes were labeled with 32p by nick translation (17).
~n t i t a t i v e analysis of the hybridization to the Rf sequences was performed by scanning the x-ray f h in a Joyce Loebl microdensitometer and measuring the peak area. The areas of the peaks in the Southern blot probed with the RI sequences were compared with those of the same filter probed with the actin sequences.
Photoaffinity Labeling of R Subunits of Protein Kinase-The incorporation of 8-N3[32P]cAMP (ICN) into RI and RII was performed as previously described (6). 50 liters of fractions isolated on DEAEcellulose chromatography, as described above, were incubated with the 8-N3[32P]cAMP label. Incubations were carried out for 30 min at 30 "C followed by 10 min at 4 "C. Tubes were then irradiated for 10 min with a Mineralight ultraviolet lamp. Each sample was then mixed with 20 liters of SDS dissociation solution (50 mM sodium phosphate (pH 7), containiig 15% SDS, 50% glycerol, 5% 2-me~aptoethanol, and 0.05% brompheno~ blue). All tubes were heated at ,1WC for 3 min and the samples were electrophoresed on SDS-acrylamide gels.

DNA-mediated Transfer of the CAMP-resistant
Phenotype-Wild-type CHO cells can be transformed to cAMP resistance with DNA from cells carrying the RI mutation (10248) ( Table I). CHO cells were co-transformed with the plasmid pSV2-ne0 and high molecular weight DNA from either the wild-type 10001 cells or the mutant 10248. The DNA was left on for 16 h and then replaced with normal medium. 24 h later, the normal medium was 'replaced with medium containing 800 pg of the neomycin analog G418. After 10 days the G418-resistant colonies were pooled and plated at lo5 cells/lO-ern plate in 1 mM 8-Br-CAMP in agar (8). 12 days later, 24 colonies were picked from the plates of cells that had received wild-type or mutant DNA. The colonies on the plates of cells that had received the mutant DNA had many large colonies which were not seen on plates of the control cells that had received the wild-type DNA. Wild-type CHO cells will form small colonies in agar in the presence of CAMP. Mutanty CAMP-resistant cells, on the other handy will form much bigger colonies. We would therefore expect a true transformant to form a large colony, since 10248 carries a dominant mutation (4). Since the G418-resistant colonies were trypsinized and pooled before selection in 8-Br-CAMP agar, CAMP-resistant colonies do not necessarily represent unique events. That is, several CAMP-resistant colonies might be derived from the same G418-resistant colony. For this reason, we cannot determine the true frequency of CAMPresistant transformants among the G418-resistant colonies,  (10001). The cells were selected for G418 resistance'for 10 days, pooled, and selected for resistance to 1 mM 8-Br-CAMP in agar (4,8). Both CAMP-resistant and non-CAMP-resistant cells w i l l survive on 1 mM 8-Br-cAMP, but only resistant cells will form large colonies on agar. The largest colonies from each experiment were selected after 14 days and some were retested for growth and morphology on microwell plates in the presence of 1 mM 8-Br-CAMP. Of the largest colonies selected from each experiment, those from the transformation with the m u~t 10248 DNA were much larger than those from the transfo~ation with the wild-type lo001 DNA. pSV2-neo psv2-neo but there must have been at least one c A~-r e s i s t a n t colony/ 3500 G418-resistant colonies, for an overall frequency of 2 2 X 10-7/original transformed cell.
Since colony size is a relatively subjective measure and influenced by various factors, including the microenvironment in the agar, we tested these colonies further for growth as monolayers in 1 mM 8-Br-CAMP. The results are also shown in Table I. Six of the colonies derived from the transformation with the mutant DNA were resistant to the growth inhibition of 8-Br-CAMP in monolayer culture. We have chosen one of these transformants, 11564, for further analysis. A growth curve, comparing growth of this t r~s f o r m a n t with one of the control transformants, 11572 (G418-resistant clone transformed with 10001 DNA and pSVz-neo), is shown in Fig. 1. The parental 10001 strain has a growth response to cAMP very similar to the control 11572 (data not shown). It is clear that the transformant is very resistant to inhibition of growth by 8-Br-CAMP. Two clones derived from this transformation and two derived from the transformation with wild-type DNA showed slight resistant to 8-Br-CAMP. These variants might be the result of spontaneous mutations causing a very marginal CAMP resistance. Alternatively, the transfer of wildtype DNA sequences themselves may in some way alter the CAMP resistance phenotype possibly by a gene dosage effect.
Protein Kinase Activity in the Transformants-Transfonnants selected on G418 and then 1 mM 8-Br-CAMP were tested for the level of CAMP-dependent protein kinase activity in extracts of whole cells (Fig. 2).
The transformant 11564, transformed with the mutant DNA, has a cAMP activation curve more like that of the mutant cells than that of wildtype cells. Both the mutant and transformant show a decreased sensitivity to CAMP which is shown by the rightward shift of their CAMP activation curves. This increased & for CAMP is ch~acteristic of mutant 10248 (7).

DEAE Analysis of Type I and Type If Protein Kinases-
The wild-type CHO cell has two types of CAMP-dependent protein kinase, type I and type 11, that are present in many  (2, 23). The mutant 10248, however, has mainly type I (7) (Fig. 3). The protein kinase profile of the transformant 11564 looks very similar to the mutant, having also very little type I1 kinase.

Analysis of CAMP-dependent Phosphorylations in Intact
Celk-Previous studies have demonstrated that the mutant cell 10248 does not phosphorylate a 52,000-dalton protein in a CAMP-dependent manner (19) although the wild-type does. To test for this defect in the transformant, we labeled intact cells with C A M P . Homogenates of the mutant, wild-type, and transformant 11564 were electrophoresed on o n e -~e n s i o n a l SDS-acrylamide gels. The results (Fig. 4) show the presence of a phosphorylated band at 52,000 daltons in the wild-type. This band is not phosphorylated in mutant 10248 and is also not phosphorylated in the transformant. With respect to phosphorylation, the transformant is behaving like the mutant, as would be expected if we have transferred the gene responsible for the behavior of the mutant.

The Regulatory Subunit of the T~n s f o r~~n t
Is like That of the Mutant-The amount of RI present in each of the fractions eluted from the D E m column can be measured by the incorporation by the RI subunit of the p h o t o~l n i~ label, 8-N3[3zP]cAMP. When the RI l i e d to the photoaffinity label from representative fractions is electrophoresed on SDSacrylamide gels, a characteristic profile is seen. This is shown in. Fig. 5. The largest peak of label represents free RI (6). The small peak that is evident in the wild-type, at fractions 16-20, represents the RI present in the holoenzyme PKI. Note that the mutant 10248 and the transformant 11564 do not have this peak of RI in these fractions. This absence of normally binding RI in the holoenzyme PKI is characteristic of the mutant 10248 (7) and has not been seen for any of our other protein kinase mutants. Under these labeling conditions, RII does not bind well to the p h o t o~~~t y label and the upper bands, representing RII, can only be seen faintly starting at fraction 28. from every other fraction were assayed for protein kinase activity in the absence and presence of 10 p~ CAMP. The CAMP-dependent protein kinase activity, derived by subtracting the CAMP-independent activity from the total kinase activity, is plotted. A, 10001, wild-type; B, 10248, mutant; C, 11564, transformant. One unit is defined as the amount of enzyme required to catalyze the incorporation of a pmol of 32P into histone/min at 30 "C. mutant tubulin gene, we had observed an amplification of the transferred gene. To test for extra copies of the RI gene in the transformant, we isolated DNA from our transformant, wild-type cells, and mutant cells, digested with the restriction endonucleases HindIII, EcoRI, and BanHI, and electrophoresed the fragments on agarose. To probe the DNA fragments, we used a fragment isolated from a cDNA clone to bovine testis CAMP-dependent protein kinase type I (22). As can be seen in Fig. 6, in the transformant (lanes 3, 7, and 11) there appears to be an amplification of the one or two bands that hybridized to the cDNA clone of bovine testis RI. This amplification is not seen in either the wild-type cell, a mock transformant, or the mutant. We have verified that the amount of DNA in all of the lanes is approximately the same by removing the RI probe and hybridizing the filter with chicken actin probe (23). Since the actin gene should be present in the same amount per cell in the three cell lines, the density of the bands is proportional to total DNA in each lane. All bands hybridizing with this probe were of approximately equal intensity. This was confirmed by densitometry. Table I1 shows a quantitation by densitometer of the density of the bands on the filter after probing with actin or RI, normalized to the amount of actin DNA. The amplification  of the band or bands that hybridized with the pRI probe is between 1.4-and 1.7-fold. While the amount of amplification is small, on the order of 50-loo%, it is consistently present with all three enzyme digestions and on several blots.  We confirmed this result by a slot blot analysis of DNA from wild-type cells and the transformant, 11564, shown in Fig. 7. In this experiment we used MEP cDNA as a control (24). For each level of DNA tested, the signal in the lanes probed with the RI cDNA probe is stronger for the transformant DNA (lane d ) then for the wild-type DNA (lane c). These data were quantitated by densitometry and normalized to the small differences seen between transformant (lane b) and wild-type DNA (lane a) using the MEP cDNA probe. The signals were directly proportional to the amount of DNA applied. The relative amounts of RI DNA in the transformant compared to wild-type cells for each DNA level were: 1.9 (1 pg), 2.1 (2 pg), 2.2 (3 pg), 2.0 (4 pg), and 1.9 (5 pg), with an average of 2.0. An increase of 50-100% in the RI band could be accounted for by the transfer of one or two extra gene copies, assuming the CHO cell has two alleles of the RI gene. CAMP. In addition, we have shown the presence of DNA fragments in the transformant that hybridized at least 50% more to a RI subunit gene fragment from a cDNA clone of bovine testis RI. This last piece of data is also a confirmation that the mutation is linked to the RI subunit, as well as indicating the success of the transfer. These data do not completely rule out the unlikely possibility that the mutation affecting the RI* subunit is closely linked to the RI subunit, but not in the gene encoding this subunit. Examination of the Southern blots shown in Fig. 6 indicates that CHO DNA cut with both Hind111 and BarnHI give only one band when hybridized to the 770-base pair PstI fragment of pRI. The simplest explanation for this result is that CHO cells carry only one RI gene, although other explanations are possible. If this were true, then the two forms of RI seen in

1-
FIG. 6. Hybridization of transformant DNA to cDNA probe of the bovine testes RI subunit gene and to the &actin gene. DNA was extracted and digested with HindIII (lanes 1 4 ) , EcoRI (lanes 5 4 , or BamHI (lanes 9-12), and electrophoresed on agarose gels. After electrophoresis the DNAs were transferred to GeneScreen filters using the procedure of Southern (15). Filters were probed with a 32P nick-translated PstI fragment of the bovine testes RI cDNA clone, pRI @anel A), or with a 32P nick-translated HindIII fragment of &actin @anel B ) ("Materials and Methods"). Autoradiograms of the filters are shown. Panels A and B: lanes I ,5,9,  wild-type and mutant cells (7) could be the products of two different RI alleles. This conclusion is similar to what has been found on the basis of biochemical and genetic analysis in S49 cells (25). Alternative explanations include a single allele with alternate mRNA splicing or a single mRNA which can be translated in two different ways.
The ability to transform CHO cells with DNA from mutant cells will enable us to rescue and identify this and other mutant genes affecting cAMP metabolism by techniques already developed (11,26). Since many of the proteins involved in cAMP metabolism are present in the cell in small amounts, a method to clone these genes by means other than isolation of mRNA is very advantageous.