Amplification of the Transketolase Gene in Desensitization-resistant Mutant Y1 Mouse Adrenocortical Tumor Cells

As shown previously, mutants of the Y1 mouse adrenocortical tumor cell line that resist agonist-induced desensitization of adenylyl cyclase have elevated levels of a 68-kDa protein (designated p68), suggesting a possible relationship between p68 and the regulation of adenylyl cyclase activity. In the present study, cDNA cloning and sequencing were used to identify p68 as mouse transketolase. Cells overexpressing p68 exhibited a 17.4-fold increase in transketolase enzymatic activity relative to parental Y1 cells and a 28-fold amplification of the transketolase gene as determined by Southern blot hybridization analysis. Using fluorescent in situ hybridization analysis, the transketolase gene was mapped to mouse chromosome 16B1 and to human chromosome 3p21.2. Transketolase gene amplification was associated with telomeric fusion of the chromosome 16 pair together with the appearance of multiple copies of the transketolase gene throughout a different chromosome. The relationship between overexpression of transketolase and desensitization resistance was evaluated in somatic cell hybrids formed between a desensitization-resistant adrenal cell line and a desensitization-sensitive rat glial cell line. In these hybrids, transketolase overexpression behaved dominantly, whereas desensitization resistance behaved recessively. These results dissociate the desensitization resistance phenotype from overexpression of transketolase and suggest that desensitization resistance may have resulted from disruption of an essential regulatory gene in conjunction with the amplification event.

As shown previously, mutants of the Y1 mouse adrenocortical tumor cell line that resist agonist-induced desensitization of adenylyl cyclase have elevated levels of a 68-kDa protein (designated p68), suggesting a possible relationship between p68 and the regulation of adenylyl cyclase activity. In the present study, cDNA cloning and sequencing were used to identify p68 as mouse transketolase. Cells overexpressing p68 exhibited a 17.4-fold increase in transketolase enzymatic activity relative to parental Y1 cells and a 28-fold amplification of the transketolase gene as determined by Southern blot hybridization analysis. Using fluorescent in situ hybridization analysis, the transketolase gene was mapped to mouse chromosome 16B1 and to human chromosome 3p21.2. Transketolase gene amplification was associated with telomeric fusion of the chromosome 16 pair together with the appearance of multiple copies of the transketolase gene throughout a different chromosome. The relationship between overexpression of transketolase and desensitization resistance was evaluated in somatic cell hybrids formed between a desensitization-resistant adrenal cell line and a desensitizationsensitive rat glial cell line. In these hybrids, transketolase overexpression behaved dominantly, whereas desensitization resistance behaved recessively. These results dissociate the desensitization resistance phenotype from overexpression of transketolase and suggest that desensitization resistance may have resulted from disruption of an essential regulatory gene in conjunction with the amplification event.
In a variety of cell types, the chronic stimulation of adenylyl cyclase by hormones and neurotransmitters often desensitizes the enzyme, rendering it refractory to further stimulation. In our laboratory, this phenomenon has been investigated extensively using Y1 mouse adrenocortical tumor cells and in a family of desensitization-resistant (DR) 1 Y1 mutants (1)(2)(3)(4)(5). We have shown that the DR mutation in Y1 cells not only affects desensitization from the endogenous ACTH receptor but also affects desensitization from wild-type mouse ␤ 2 -adrenergic and human dopamine D-1 receptors when genes encoding these receptors are transfected into the mutant cell line (2)(3)(4). Using ligand binding analyses, we demonstrated that the DR muta-tion did not affect receptor internalization, a late step in the desensitization pathway, but prevented receptor uncoupling from its guanyl nucleotide-binding regulatory protein (2,3,5). On the basis of these findings, we have suggested that the DR mutation does not reside within the ACTH receptor; rather, it affects an early component of the desensitization pathway that is shared among different receptor signaling systems.
A potential insight into regulation of the desensitization pathway came from our observations that the DR phenotype is associated with the overexpression of a 68-kDa protein designated p68 (1, 6 -8). Among 18 independent subclones of the Y1 adrenal cell line, the level of p68 correlated with the level of ACTH-responsive adenylyl cyclase activity and those with high levels of p68 desensitized more slowly and recovered from the desensitized state more quickly than clones with low levels of p68 (1,6). Inasmuch as p68 has not been identified, we have undertaken the cloning and sequencing of the cDNA encoding this protein. We report that p68 is the mouse transketolase (EC 2.2.1.1; TKT). We show that TKT activity in the DR mutant is 20-fold higher than in parental Y1 cells and that the overexpression of TKT results from amplification of a chromosome segment derived from mouse chromosome 16. Using somatic cell hybridization analyses, we are able to dissociate TKT overexpression from the DR phenotype, suggesting that the DR phenotype likely resulted from a reciprocal gene deletion that accompanied amplification of the TKT gene.

MATERIALS AND METHODS
cDNA Library Screening-Custom cDNA libraries in the bacteriophages gt11 and gt10, respectively (Clontech Laboratories, Inc., Palo Alto, CA) were prepared from poly(A) ϩ RNA (9) isolated from Y1 mouse adrenocortical tumor cells (10) and from the Y1 derivative, Kin-8 (11). The gt11 library was probed for expression of p68 using a rabbit polyclonal p68 antiserum (8) and 125 I-labeled protein A as described (12). Subsequent screenings of the gt11 and gt10 libraries were performed by DNA hybridization as described by Maniatis et al. (13) using the 600-bp EcoRI fragment from gt11 clone 16 (Fig. 1). The probe was labeled by nick translation in the presence of [␣-32 P]dCTP using a kit from Life Technologies, Inc. (Canadian Life Technologies, Inc., Burlington, Ontario, Canada).
Rapid Amplification of 5Ј cDNA ends (5Ј-RACE)-The 5Ј-RACE procedure (14) was used to clone the 5Ј end of the p68 transcript. Total RNA was prepared from DR cells overexpressing p68 (1) using guanidine thiocyanate for extraction, followed by centrifugation through CsCl (15). cDNA was synthesized with SuperScript™ reverse transcriptase and C-tailed with terminal deoxynucleotidyltransferase and dCTP using a 5Ј-RACE System from Life Technologies, Inc. The oligodeoxynucleotide primer used for first strand cDNA synthesis (5Ј-GGTAT-GGAAAAACAGGACAGCCAT-3Ј) was complementary to the mouse p68 (TKT) sequence from positions 233-256 (Fig. 3). The cDNA was then amplified by polymerase chain reaction (PCR) using a kit containing AmpliTaq® DNA polymerase (Perkin Elmer (Canada) Ltd., Rexdale, Ontario, Canada) together with the Anchor Primer provided with the 5Ј-RACE System and an internal primer (5Ј-CATGATCTCG-GCAGCGCTGCAGCATGATGT-3Ј; Kronem Systems Inc., Missisauga, Ontario, Canada) complementary to the p68 sequence at positions 206 -235. PCR was carried out for 36 cycles with a hot start (16); the timing for each cycle consisted of 1-min incubations at 94, 54, and 72°C. At the end of the reaction, samples were incubated at 72°C for 10 min to ensure that cDNA synthesis went to completion. An aliquot of the reaction was amplified for an additional 35 cycles under similar conditions, except that the reaction contained the universal amplification primer from the 5Ј-RACE system instead of the Anchor Primer and the annealing temperature was raised to 65°C. The PCR product was end-filled with Klenow DNA polymerase (Pharmacia, Baie d'Urfe, Quebec, Canada) and cloned into the SmaI sites of pBluescript SK ϩ and KS ϩ (Stratagene, La Jolla, CA).
DNA Sequence Analysis-Sequences from p68 cDNA downstream from the internal EcoRI site (Fig. 1) were determined using fragments generated following digestion with EcoRI and BamHI; sequences 5Ј of the internal EcoRI site were determined using nested deletion fragments generated by digestion with exonuclease III and mung bean nuclease (Stratagene). All DNA fragments were subcloned into the phagemids pBluescript SK ϩ and KS ϩ , and single-stranded templates were prepared following propagation in Escherichia coli JM101 cells with VCR-M13 helper phage (Stratagene). DNA was sequenced by the dideoxynucleotide chain termination method (17) using Sequenase 2.0 (U. S. Biochemical Corp.) with T7 primers (U. S. Biochemical) or T3 primers (Pharmacia). Reactions carried out in the presence of dGTP or c 7 dITP were compared to resolve sequence ambiguities. Data base searches were carried out using the FASTA algorithm (18), and alignments were performed using Geneworks® version 2.21 (IntelliGenetics, Inc., Mountain View, CA).
Western, Northern, and Southern Blot Analyses-For Western blot analysis, purified p68 and adrenal cell extracts were separated by electrophoresis on polyacrylamide gels, blotted onto nitrocellulose using a Bio-Rad Transblot apparatus and probed with an affinity-purified p68 antiserum and 125 I-protein A essentially as described (8,19). The affinity-purified p68 antiserum was obtained by passing a polyclonal p68 antibody (8) over nitrocellulose filters containing the epitope expressed in purified bacteriophage gt11 clone 16 ( Fig. 1) as detailed previously (12). For Northern blot hybridization analysis, total RNA from adrenal cell cultures was prepared as described above, fractionated on a 1.2% agarose, 2.2 M formaldehyde gel, transferred to a Nytran Plus nylon filter (Xymotech Biosystems, Toronto, Ontario) and probed with a 32 Plabeled cDNA probe prepared by nick translation of the 600-bp EcoRI fragment from the bacteriophage gt11 clone 16 using a kit from Canadian Life Technologies Inc. For Southern blot hybridization analysis, high molecular weight DNA was prepared as described by Wigler et al. (20) and collected by spooling onto glass rods. DNA samples digested with restriction endonucleases were fractionated by electrophoresis on 0.7% agarose gels, blotted onto nylon membranes, and hybridized with a 32 P-labeled mouse TKT cDNA from nucleotide 153 to nucleotide 2062 ( Fig. 2) or with genomic probes for the mouse immunoglobulin light chain gene and the mouse immunoglobulin 5 gene.
Transketolase Activity-Transketolase activity was assayed in a coupled enzymatic assay as described previously (21). Briefly, cell monolayers were rinsed with phosphate-buffered saline, scraped in buffer containing 50 mM potassium phosphate, pH 7.8, 1 mM EDTA, 1 mM 2-mercaptoethanol, and 0.1% (w/v) Triton X-100. Cells were homogenized using a Dounce homogenizer with a tight-fitting pestle, and homogenates were centrifuged 100,000 ϫ g for 90 min at 4°C to obtain cell supernatant fractions. Assays were carried out in quartz cuvettes in a 1-ml reaction volume containing 1 mM MgCl 2 , 0.1 mM NADH (Boehringer Mannheim Canada, Laval, Quebec), 2.5 mM D-xylulose-5-phosphate (Sigma), glycerol-phosphate dehydrogenase (1 unit) and triosephosphate isomerase (11 units) (Sigma), 100 mM Tris-HCl, pH 7.6, and 5-30 g of cell supernatant. The assay mixture was equilibrated at room temperature for 5 min, and the reaction was initiated by adding ribose 5-phosphate (Sigma) to a final concentration of 2.5 mM to the sample cuvette. Reactions were monitored by measuring the change in absorbance at 340 nm over a period of 20 min; results were converted to units of activity, where 1 unit of activity represents 1 mol of NADH oxidized ϫ min Ϫ1 ϫ mg protein Ϫ1 . Activity was linear with respect to both enzyme protein and incubation times.
Gene Mapping-Gene mapping was performed by SeeDNA Biotech Inc. (North York, Ontario) as described previously (22,23). Normal mouse chromosomes were prepared from synchronized cultures of splenic lymphocytes (24). Human chromosomes were prepared from syn- chronized cultures of lymphocytes isolated from cord blood (22). Chromosomes from parent and mutant mouse adrenocortical tumor cell lines were prepared from cells arrested in metaphase with 2 g/ml vinblastin for 18 h (25).
Mouse chromosomes were probed with a partially characterized 20-kb TKT genomic clone isolated from an EMBL3 mouse genomic library. Human chromosomes were probed with mouse TKT cDNA from nucleotide 153 to nucleotide 2062. Probes were labeled with biotinylated dATP, hybridized to the chromosome spreads, and detected with FITC-avidin. Signals were amplified by incubation with biotinylated goat anti-avidin followed by a second round of incubation with FITCavidin. Chromosome banding patterns were obtained with the chromatin-binding fluorescent dye 4Ј-6-diamidino-2-phenylindole (DAPI). Chromosomal localization of TKT was made by superimposing photographs of the hybridization signals with photographs of the DAPI banding patterns.
Adenylyl Cyclase Activity-Adenylyl cyclase activity was measured in cell homogenates by measuring the conversion of [2, H]ATP to cAMP in a 5-min reaction at 37°C as described previously (26). The reaction mixture contained 1 mM disodium ATP (approximately 1.6 ϫ 10 6 cpm; DuPont Canada), 2 mM MgCl 2 , 6 mM theophyllin, 50 g of albumin, 12.5 mM Tris-HCl, pH 7.7, and approximately 140 g of enzyme protein in a final volume of 85 l.

RESULTS
Isolation of p68 cDNA-A gt11 cDNA library prepared from Y1 cells was screened for p68 cDNA by expression using a polyclonal p68 antiserum (8). An immunoreactive isolate, gt11-16, was plaque-purified, and found to contain a 900-bp insert with an internal EcoRI restriction site that encoded an open reading frame (Fig. 1). The identity of the insert as a p68 cDNA clone was confirmed using epitope selection and Northern blot hybridization techniques (Fig. 2). As shown in Fig. 2, p68 antiserum affinity-purified by passage over the epitope expressed in gt11-16 reacted with purified p68, and specifically recognized single bands of 68 kDa in Y1 cell extracts on Western blots. The signals obtained when cell extracts were probed with the affinity-purified antiserum correlated with the differential levels of expression of p68 expected in parental Y1 cells and mutant DR cells overexpressing p68. Similarly, Northern blots probed with a cDNA fragment from the gt11-16 isolate gave a signal at 2.2 kb that was more intense in mutant DR cells overexpressing p68 than in the parental Y1 cell line (Fig. 2). The cDNA from the gt11-16 isolate was then used as a probe to clone three additional p68 cDNA fragments from mouse adrenal cell libraries prepared in gt11 and gt10 as described under "Materials and Methods." These cDNA fragments spanned 1910 bp of p68 cDNA plus the 3Ј poly(A) tail. Isolation of an additional 152 bp of cDNA including the initia-tor ATG from the 5Ј end of the p68 transcript was achieved using the 5Ј-RACE procedure (Fig. 1).
Identification of p68 as Mouse TKT-The p68 cDNA fragments were sequenced completely from both DNA strands and shown to encode a large open reading frame corresponding to 623 amino acids with a calculated molecular mass of 67,553 Da (Fig. 1). A search of the GenBank™ (27) data bank showed that the p68 sequence was 83% identical with human TKT at the DNA level, 94% identical at the protein level and included conserved amino acids implicated in binding thiamine pyrophosphate (28). The sequencing data thus indicate that p68 is the mouse TKT. Of the 38 amino acids that differed between the mouse and human TKT, the substitutions S30T, S31T, V46E, and A426P in the mouse protein also are seen as polymorphisms in the human gene (28).
The identity of p68 as TKT was further confirmed by demonstrating that extracts from the DR mutant exhibited a 17.4fold higher level of TKT activity (0.40 Ϯ 0.04 units) compared to extracts from parental Y1 cells (0.023 Ϯ 0.002 units), consistent with the observed amplification of p68 in DR clones.
Chromosomal Localization of TKT-Strong TKT signals were detected on more than 80% of mitotic figures examined and were localized to mouse chromosome 16 and human chromosome 3; background signals were minimal, and positive signals were not detectable on any other chromosomes. A representative example for mouse TKT is shown in Fig. 3. Detailed characterization of 10 mitotic figures further localized the TKT gene to the B1 region of mouse chromosome 16 and to the p21.2 region of human chromosome 3 (Fig. 4).

Basis for Overexpression of TKT in DR Mutant
Clones-In order to gain further insight into the relationship between the DR phenotype and overexpression of TKT, we examined TKT gene copy number by Southern blot hybridization using mouse TKT cDNA as a probe. As shown in Fig. 5, Southern blotting genomic DNA from DS and DR cells using TKT cDNA gave similar bands of hybridization that ranged in size from 1 to 12 kb, depending on the restriction endonuclease used for digestion. In each case, hybridization signals for the TKT gene were, on average, 28-fold more intense in the DR mutant than in parental Y1 cells, suggesting that overexpression of TKT resulted from gene amplification. In control experiments, hybridization of enzyme-digested genomic DNA from DS and DR cells with cDNA probes for the ACTH receptor (26) or the regulatory subunit of the type 1 cAMP-dependent protein kinase (29) gave signals of approximately equal intensity (data not shown).

FIG. 3. Fluorescent in situ hybridization analysis of the mouse TKT gene.
Chromosome spreads were prepared from mouse splenocytes and probed with biotinylated TKT genomic DNA. Signals were amplified and detected with FITC-avidin (panel A). Panel B shows the same mitotic spread stained with DAPI. The fluorescent banding patterns obtained were used to identify the chromosome labeled with the TKT probe as chromosome 16. In chromosome spreads prepared from parental Y1 mouse adrenocortical tumor cells, TKT signals also were observed on chromosome 16, and there was no evidence of TKT gene amplification (data not shown). In the DR mutant, however, the TKT gene seemed to be amplified over a large region on a single chromosome (Fig. 6). Since the level of amplification was very high, the morphology of the affected chromosome was completely changed and identification of the affected chromosome was not possible. TKT signals also were evident on the chromosome 16 pair, which showed an abnormal telomeric fusion in the DR mutant (Fig. 6). Additional faint signals seen scattered throughout the chromosome spread are not reproducible and represent background.
As determined by Southern blot hybridization analysis (Fig.  5), other genes associated with the proximal region of mouse chromosome 16, i.e. immunoglobulin 1 and mouse immunoglobulin 5 (30), were not amplified in the DR mutant clone.
Effects on Adenylyl Cyclase Activity-The identification of p68 as TKT raises interesting questions about its contribution to the DR phenotype and the mechanisms responsible for its overexpression in DR cells (6). We considered the possibility that this enzyme, when overexpressed, protects adenylyl cyclase from agonist-induced desensitization. TKT has not been implicated in the regulation of the adenylyl cyclase system previously, and its function in this regard was not readily apparent. We were unable to modify adenylyl cyclase activity in broken DR cells by treatment with TKT antiserum or by adding back purified enzyme to homogenates of parental Y1 cells (data not shown), suggesting that increased expression of TKT per se is not responsible for the DR phenotype.
To further address the relationship between TKT overexpression and desensitization resistance, we evaluated the linkage of these two phenotypes in somatic cell hybrids formed between a DR derivative, Kin-8HGPRT Ϫ , and the rat glioma cell line, C6TK Ϫ (25). As we reported previously, Kin-8 cells, like the DR parent, resist ACTH-induced desensitization and produce elevated levels of p68 (approximately 10% of total protein; Ref. 1), whereas in C6 cells, the adenylyl cyclase system is readily desensitized upon continuous exposure to ␤-adrenergic agonists such as isoproterenol (31) and the levels of p68 are low (approximately 0.1% of total protein; Ref. 8).
As determined from Southern blot hybridization analysis using the mouse TKT probe, the TKT genes in two independently isolated Kin-8HGPRT Ϫ ϫ C6TK Ϫ hybrid clones, H7 and H8, were amplified to the same extent and gave the same restriction patterns as the Kin-8HGPRT Ϫ fusion partner (Fig.  7A) and parental DR cells (Fig. 5). Under these same conditions of hybridization stringency, the mouse TKT probe did not give a detectable signal for the TKT gene from the rat glial cell line. As determined by Northern blot hybridization (Fig. 7B), TKT transcripts were markedly abundant in the H7 and H8 hybrids, reaching levels comparable to those seen in Kin-8HGPRT Ϫ and parental DR cells; these levels of TKT transcript were much higher than those seen in the C6TK Ϫ fusion partner or in DS cells. These results indicate that the hybrid clones acquired and expressed the amplified TKT gene from the DR parent and that TKT overexpression behaves dominantly in the hybrids. In panel A, genomic DNA from DS and DR cells was digested to completion with the restriction endonucleases indicated, electrophoresed on 0.8% agarose, blotted onto a Hybond ϩ nylon membrane, and hybridized to a TKT cDNA probe. In panel B, EcoRI-digested genomic DNA from DS and DR cells was electrophoresed on a 0.4% agarose gel, blotted and probed for mouse immunoglobulin light chain ( 1) and mouse immunoglobulin 5. Fragment sizes were estimated using HindIII-digested bacteriophage and HaeIII-digested X-174 as standards.
As shown in Table I, the hybrid clones responded to ACTH, isoproterenol, and NaF with increases in adenylyl cyclase activity. The response to ACTH reflected the contribution of Kin-8HGPRT Ϫ cells, whereas the response to isoproterenol reflected the contribution of C6TK Ϫ (Table I). Despite the presence of the amplified TKT gene and overexpression of TKT transcripts, adenylyl cyclase in the hybrid clones was rapidly desensitized upon exposure to ACTH (Fig. 8). Within 1 h of exposure to ACTH, the hybrid clones lost 85% of their hormoneresponsive adenylyl cyclase activity. In contrast, the Kin-8HG-PRT Ϫ parent resisted ACTH-induced desensitization and retained 70% of its ACTH-responsive activity after 6 h of continuous exposure to the hormone (Fig. 8). The desensitization induced by ACTH in the hybrid clones was homologous, since the hybrids retained 85-90% of their isoproterenol-stimulated adenylyl cyclase activity after treatment with ACTH (not shown).

DISCUSSION
In order to further understand the biochemical and molecular causes of the DR phenotype in mutant Y1 adrenocortical tumor cells, we sought to identify p68 and determine the basis for its overexpression in DR cells. Based on cDNA sequencing results and direct assays of enzymatic activity, we have estab-lished that p68 is the mouse TKT. TKT is a thiamine-requiring enzyme that is part of the pentose phosphate metabolic pathway responsible for the synthesis of pentoses and for the generation of NADPH (32). Defects in TKT have been described in a population of alcoholic patients and may contribute to the neuropathological disturbances associated with Wernicke-Korsakoff syndrome (33,34).
Chromosomal mapping experiments localized human TKT to chromosome 3p21.2 (Fig. 4) and mouse TKT to the B1 region of chromosome 16 (Figs. 3 and 4), a region that appears to be poorly defined and not yet established as syntenic with human chromosome 3p21 (35). Previous gene mapping studies also had localized the human TKT to chromosome 3p (34,36); however, the earlier results had placed the TKT gene at 3p14 (36) rather than in the adjacent 3p21.2 region as reported here.
As evidenced from Southern blot and fluorescent in situ hybridization analyses (Figs. 5 and 6), the overexpression of TKT in DR mutant clones resulted from an approximate 28-fold amplification of the TKT gene. In most examples of gene amplification, the regions involved (referred to as amplicons) are very large and can involve as much as 10,000 kb of DNA (37). Other markers of chromosome 16 proximal to the centromere, such as the immunoglobulin genes, are not amplified in the DR mutant (Fig. 5) and thus must be too far away from the TKT gene to have been included in the amplicon. Although the basis for amplification of the TKT gene is unknown, it is interesting that TKT shares structural and functional homology with the RecP protein of Streptococcus pneumoniae, a protein required for genetic transformation that functions to promote insertion-duplication mutations in the prokaryotic chromosome (38,39). As reviewed elsewhere (40,41), gene amplification may occur through a number of different mechanisms, may involve recombination events (including gene insertions, deletions, and inversions), and is sometimes associated with telomeric fusions (e.g. Fig. 6). Amplified genes can exist as selfreplicating minute chromosomes or as arrays of amplified segments on one or more chromosomes (40), as seen in the case of TKT amplification in the DR mutants (Fig. 6).
To further explore the relationship of TKT gene amplification to desensitization resistance, we examined the linkage of these two phenotypes in somatic cell hybrids between a DR isolate and desensitization-sensitive C6 glioma cells. The hybrids acquired the TKT amplicon and overexpressed TKT (Fig.  7) but failed to resist ACTH-induced desensitization of adenylyl cyclase (Fig. 8). These results clearly dissociate TKT gene amplification from the DR phenotype and indicate that desensitization resistance behaves recessively in the hybrid. On the basis of these results, we suggest that desensitization resistance may have resulted from a recombination event that disrupted or mutated a gene required for the desensitization proc-ess rather than from amplification of TKT itself or from coamplification of a closely linked gene.