A dominant negative mutation in two proteins created by ectopic ...

Somatic Cell and Molecular Genetics, Vol, 16, No. 2, 1990,pp. 151-162

A Dominant Negative Mutation in Two Proteins Created by Ectopic Expression of an AU-Rich 3' Untranslated Region Bruce A. Bunnell, 1 tlelen Fillmore, 2'3 Paula Gregory, 2 and Vincent J. Kidd t'2 Departments of IMierobiology and 2CelIBiology and Anatomy, University of Alabama at Birmingham, Birmingham, Alabama 35294 Received 29 September 1989--Final 29 November t989

Abstract-- We have found that the ectopic expression of a specific gene's 3' untranslated region

leads to the dose dependent loss, relative to gene copy number, of that specific m R N A and protein, as well as an associated protein, in a eukaryotic cell line. The loss of these proteins from the eukaryotic cell line also results in specific phenotypic changes in these cells,, suggesting that we have created a dominant negative mutant. This gene's 3' untranslated region is known to contain numerous AU sequences, reminiscent of other eukaryotic genes whose expression may be regulated by these sequences. The apparent control of gene expression by a truncated 3' untranslated region sequence provides.further evidence supporting the regulatory function of these regions. The resulting decrease in steady-state m R N A levels by the overexpression of a portion o f that gene's 3' untranslated region further suggests the possible existence of a factor(s) that may bind to this region, and thus regulate gene function via its mRNA. The use of a gene's 3' untranslated sequence to create a specific dominant negative mutation may also be applicable to other eukaryotic genes whose expression is controlled by similar regulatorT sequences.

INTRODUCTION We have been studying the possible cooperative interactions between two distinct peptides that appear to be a part of /31-4 galactosyltransferase (GalTase) enzyme activity (Adams et al., submitted; 1). To this end, we have identified a novel cell division control (cdc) related protein kinase, which we call galactosyltransferase-associated protein (GTA), that specifically phosphorylates GalTase in vitro (Adams et al., submitted). One of the striking features of this GTA cDNA isolate is a very long 3' untranslated

region that contains a number of AU-rich elements (AREs). Portions of this 3' untrans-. lated region also have been found to be highly conserved, 98% similarity in DNA sequence among humans, mice, and rats (Kidd et al., in preparation). These elements resemble the AREs found in the 3' untranslated regions of a number of protooncogenes, lymphokines, and transcriptional activating proteins (2, 3). The AREs have been found to regulate the stability of these mRNAs in eukaryotic cells, possibly through the action of additional cellular factors (4-6). These factors may include the poly(A) binding protein complex

3Currentaddress:Departmentof Anatomy,Universityof TennesseeMedicalSchool,Memphis,Tennessee. 151 0740-7750/90/0300-0151 $06.00/0©1990PlenumPublishingCorporation

152

that is found in eukaryotic cells (7). With c-fos, this rapid destabilization of the m R N A has been suggested to involve the rapid removal of the m R N A poly(A) tract, possibly directed by the AREs found in the 3' untranslated region of this gene (8). This type of gene control appears to augment other levels of eukaryotic gene regulation, i.e., transcriptional and translational, and allow the rapid regulation of potentially deleterious cellular gene products and their proteins (4). Thus, the AREs appear to be critical elements involved in eukaryotic gene regulation. In the course of our investigations of GTA function, we have found that the inclusion of its 3' untranslated region as part of a stably integrated gene construct leads to the dose-dependent loss, relative to gene copy number, of endogenous cellular GTA m R N A and protein, as well as GalTase enzyme activity. Conversely, the removal of a major portion of this 3' untranslated region results in a substantial increase in the level of this same enzyme activity (Bunnell et al., in preparation). In this report we demonstrate that elevated expression of a truncated GTA cDNA construct containing the ARE regions, as well as a substantial portion of upstream 3' untranslated sequence, is responsible for this change in enzyme activity. We demonstrate that as the steady-state level of the GTA 3' untranslated region R N A increases, endogenous steady-state GTA m R N A and protein are substantially depleted. In addition, the level of endogenous cellular GalTase protein, as well as enzyme activity, is also depleted in these cells. Alterations in endogenous GTA and GalTase also leads to pronounced phenotypic changes in the cell cycle of these cells. Thus, the ectopic expression of a portion of a gene's 3' untranslated region appears to be capable of regulating the expression of that gene, most likely by creating a dominant mutation in that gene within eukaryotic cells. The ability to potentially affect a particular genes activity by expressing a portion of its coding and/or noncoding sequence has many

Bunnell et al.

broad applications. Eukaryotic genes are now being isolated at a rapid rate, but their biological function is not always obvious. This is particularly true for cellular oncogene homologs, where their function can only be inferred from the predicted protein sequence and/or its homology with another gene product. In addition, genes corresponding to well-characterized proteins have been shown to have functions and properties not associated with the previously isolated protein (9). One way to address the problem of biological function for an isolated gene product is to inactivate the protein or corresponding gene locus in a cell and examine the change in cellular phenotype. Elegant methodologies have been devised to accomplish this task, but these techniques are not applicable to all systems (10-13). Herskowitz (14) has suggested an additional approach that involves the utilization of truncated or highly mutated cDNAs that are expressed in a cell, inactivating the normal endogenous gene product and thereby creating a dominant negative mutation. In this study we demonstrate that ectopic expression of an ARE containing 3' untranslated region leads to the loss of m R N A and protein expression for a specific gene as well as the concomitant loss of an associated protein and its enzyme activity. In addition, the possible implications of this type of dominant negative mutation may have importance with regard to one potential mechanism of ARE gene regulation. MATERIALS AND METHODS

DNA Constructs, Transfections, and Cell Lines. The human GTA cDNA construct has been completely sequenced and characterized (Adams et al., submitted). The hGTA2 cDNA fragment was cloned into the EcoRI site of p91023 (B) (15). Proper orientation was determined by restriction endonuclease analysis of miniprep DNA samples. The human chromosomal GTA gene isolate, cos hGTA 17-2, was isolated with

Dominant Mutation in GTA and GT

purified X hGTA 5 cDNA insert and has been characterized by restriction mapping and partial DNA sequence analysis. Its complete structure will be reported elsewhere. The cosmid vector containing this chromosomal fragment has been described by Lau and Kan (16). Both p91023 (B) and pCV107 contain a dihydrofolate reductase (dhfr) minigene to allow selection and amplification in the CHO (dhfr-) cell line originally described by Urlaub and Chasin (17). DNA constructs were transfected into cells by conventional Ca-PO 4 DNA precipitation (18). Cotransfection of p9t023 (B)-hGTA 2 and p91023 (B) was done using a 10:1 ratio of the respective DNAs. Transfected cells were selected and amplified with methotrexate (MTX) as described previously (16, t 9). DNA and RNA Isolations and Blot Analysis. All genomic DNA samples were isolated as previously described (1). Cellular RNA was isolated by treating cells with 5 mt of 50 mM NaOAc, pH 5.5, 1% SDS, and immediate extraction by the hot-phenol technique (20). Equivalent amounts of total R N A in each lane were verified by ethidium bromide staining and rehybridization with a tubulin cDNA (21). Southern and Northern blots were probed with either the X hGTA 2 cDNA insert (containing only the GTA 3' untranslated region), or a 1.5-kb EcoRI fragment containing the entire human GTA coding region and missing this 3' untranslated region in its entirety (Adams et al., submitted), after random-primer labeling with [32P]dNTPs (22). A murine ill-4 galactosyltransferase cDA clone was isolated from a murine B cell cDNA library using a 27 bp oligonucleotide probe, 5' CTTTATCTCTGGGTTCTTCTTTGCCAA 3', based on the nucleotide sequence of murine /31-4 galactosyltransferase, nucleotides 464-490 reported by Shaper et al. (23). The identity of the cDNA isolate was confirmed by DNA sequence analysis, indicating we had 100% identity to a region of/31-4 gatactosyltransferase extending from nucleotides 241 to 1884

t53

of the reported sequence (23). The blots were normally exposed for 1-3 h to XAR-5 film, because of the high copy-number of the DNA sequences. Cross-hybridization experiments between the human GTA 1.5-kb cDNA region probe and the CHO cellular RNAs were performed under reduced-stringency at 37°C for 18 h, washed in 2×SSC, 0.5% SDS two times at 22°C followed by 2xSSC 0.5% SDS at 55°C two times for 30 min. Filters were then exposed to XAR-5 film for 3-5 days. Gatactosyltransferase Enzyme Assays. All enzyme assays were done using 10 ~g of total protein cell extract in a 50-~1 reaction containing 100 #M UDP-galactose, 1 uCi UDP-[3H]galactose (Amersham), 30 mM GlcNAc, and 10 mM MnCI 2. Samples were incubated for the indicated times at 37°C and the enzymatic hydrolysis of UDP-Gal into galactose- 1-phosphate, free galactose, and the addition of galactose to the exogenous GlcNAc substrate determined by high-voltage borate paper electrophoresis (24). Sample values were corrected by subtraction of 0 time point values. Immunofluorescence and Western Blot Analysis. For the indirect immunofluorescence studies, 450 ~g/ml of anti-GalTase IgG (25), kindly provided by Dr. B. Shur, in PBS was added to monolayer cell cultures for I h at 37°C. Control cultures were treated similarly with normal rabbit IgG. The cells were washed three times in PBS/BSA after each incubation and fixed in 10% formalin in P B S Fluorescein isothiocyanate (FITC) -conjugated goat anti-rabbit IgG (Boehringer Mannheim) was added for 1 h at 23°C. The ceils were then washed three times in PBS/BSA, covered with glycerol and Hoechst stain, mounted, and examined. Western blots were performed using 50 ug of total cellular protein as previously described with either the rabbit anti-GlaTase IgG provided by Dr. B. Shur (25) or the rabbit anti-GTA peptide IgG (Adams et al., submitted) as probe and [125I]goat anti-rabbit IgG as

154

Bunnell et al.

detector (26). Blots were exposed to XAR-5 film for 3-4 days at - 80°C.

Region of GTA Results in Specific Loss of GTA and GalTase Protein. We have previ-

FH]Thymidine Incorporation and Cell Cycle Analysis. A total of 106 cells were

ously reported the complete cloning and characterization of a novel cell division control (cdc) -related gene that we call galactosyltransferase-associated protein (GTA) (1; Adams et al., submitted). While these studies were underway, we discovered that the expression of a specific portion of the GTA cDNA (or chromosomal gene) resulted in the dose-dependent loss of GTA and GalTase proteins. Initially, we were examining the expression of a human GTA chromosomal gene isolate, cos-hGTA 17-2, in CHO (dhfr-) cells (Eipers and Bunnell, unpublished observations). This GTA chromosomal gene fragment was found to be missing a portion of its 5' coding region, but contained the entire 3' untranslated region (Eipers and Bunnell, unpublished observations). Expression of this chromosomal gene fragment in CHO (dhfr-) cells led to the dose-dependent loss, relative to cos-hGTA17-2 gene copy number, of GalTase enzyme activity (Fig. 1). However, expression of the amplified vector sequences alone appeared to have no effect on GalTase enzyme activity (Fig. 1). At the same time, expression studies with the human GTA eDNA lacking a major portion of its 3' untranslated region resulted in a dramatic increase in GalTase enzyme activity (Bunnell

labeled with 5 # Ci of [3H]thymidine/ml cell medium for 6 h. The cell medium was then removed and the cells were rinsed three times with PBS. The cell monolayers were then removed by scraping, the cells pelleted, and the DNA extracted. The DNA from equivalent numbers of cells was then precipitated with cold 10% TCA and the incorporated [3H]Td quantitated and normalized to total cellular protein. For cell cycle analysis by the fluorescenceactivated cell sorter, cells were grown in DMEM containing 10% FCS. When cells reached approximately 50% confluency the medium was removed and replaced with DMEM containing 0.5% FCS. After 48 h of growth in this medium, cells were either analyzed or switched back to their normal growth media and grown for an additional 12 h before analysis. Cells were removed by trypsinization and stained with propidium iodide. Cell cycle analysis was carried out on a Becton Dickensen FACStar. RESULTS

Expression of Amplified Truncated Genes or cDNAs Containing 3' Untranslated • CHOW.T, ~09~

x

• GTA cos 17-2,10 pM MTX • GTA oOS 17-2, 20 /~M MTX

/ /

54-

A

10

20

40 60 Reaction Time (rain)

80

Fig. ]. Galactosyltransferase assays on wild type, C H O ( d h f r - ) cells containing the pCV107 vector only, and cos h G T A

17-2 containing CHO (dhfr-) cells, The amount of labeled galactose added to excess exogenous GIcNAc acceptor by the endogenous GalTase contained in the cellular extracts is shown for specified reaction times. The level of gene amplification is indicated by the concentration of methotrexate (MTX).

t 55


et al., in preparation). These results suggested that a specific regulatory function might be associated with the G T A 3' untranslated sequence. Complete D N A sequence analysis of this region has shown that the last 400 bp of the G T A c D N A contains a preponderance of A and T nucleotides similar to those found in genes known to be regulated by similar A R E s (Adams et at., submitted; 2, 3). Additionally, hybridization studies have shown that the G T A 3' untranslated region is extremely well conserved in mammalian species and has been used to isolate the rat and murine G T A c D N A s (Kidd et al., in preparation). We, therefore, decided to examine the potential role of this 3' untranslated region in the loss of GalTase activity using the selectable C H O (dhfr-) cell line as a model (17). A G T A 3' untranslated region construct was made using the eukaryotic expression vector p91023(B) (Fig. 2) (15). This construct contains 1.6 kb of the 3.5-kb human G T A cDNA, including the AREs. This sequence is under the control of a strong adenovirus major late promoter in this vector and can be amplified in the C H O (dhfr-) cells by selection and addition of methotrexate to the culture media (15, 19). We cotransfected the p91023-hGTA 2 construct and p91023 (B) into the C H O (dhfr-)

P

Pv

Pv

cell line and subsequently selected and amplified the cells containing these linearized plasmids. The cotransfection of the vector and v e c t o r - c D N A construct should avoid potential problems involving rearrangements a n d / or deletions of c D N A sequences encoding normal potyadenytation splice sites (19; R. Kaufman, personal communication). Appropriate control celt lines containing plasmid vector only, transfected into the same cells, also were established and amplified with methotrexate. All transfected cell lines, both vector control and h G T A 2 transfectants, were selected and amplified in the same manner by incremental increases in methotrexate concentration as described previously (16, 19). Genomic D N A and total cellular R N A were isolated from these cells during amplification and examined for integration and expression of the p 9 1 0 2 3 - h G T A 2 construct (Fig. 3). EcoRI digestion of the genomic D N A releases the 1.6-kb human G T A c D N A sequence (Fig. 3, panel A). In addition, larger

Pv

p

5' ~

3'

hGTA 2

2.,.,.°°.°o°,°r =100 bp

Fig. 2. GTA cDNA construct used in amplification experiments. The complete restriction map of the human GTA cDNA as well as the portion of the 3' untranslated region used in the expression vector construct is shown. The coding region is denoted by the thick black box and its transcriptional orientation indicated by 5' and 3' symbols. Restriction sites: P = PstI; Bg = BglII; Pv = PvuII; B = BamHI; K = KpnI; E = EcoRI.

Fig. 3. DNA and RNA analysis of transfected cells. (A) Southern blot analysis of 10 #g of total genomic DNA digested with EcoRI and probed with the X hGTA 2 insert. The degree of amplification (gM MTX) is shown above each lane. Molecular weight markers are shown on the left in kb. (B) Northern blot analysis of total RNA from the same cellsanalyzed in A. Total RNA, 10 ~g, was probed with the X hGTA 2 insert. The degree of gene amplification is shown above each lane. Estimated molecular weight sizes are shown on the left in kh.

156

EcoRI fragments were found to be partially amplified and hybridized with the probe but were not found in the control DNA. Since calcium phosphate transfection of the DNA was employed, these bands most likely represent multiple, tandem integrants that may be rearranged. We found that the 1.6-kb hGTA cDNA sequence was amplified appropriately in the chromosomal DNA and that it was expressed as an appropriate RNA transcript (Fig. 3, panels A and B). Equal loading of total RNA in each lane was verified by both ethidium bromide staining of the gel and rehybridization of the blot with a ~-tubulin cDNA probe (data not shown). The hGTA 2 sequence contains a normal polyadenylation splice site, AATAAA (Adams et al., submitted) and should generate a 1.7-kb transcript using this vector system (Fig. 2, Fig. 3, panel B) (15). tn addition, the position of the murine dhfr minigene results in a longer 3.0-kb transcript in the event that the RNA polymerase II reads through the hGTA polyadenylation site and terminates in the SV40 polyadenylation site found flanking the dhfr minigene (Fig. 3, panel B). Careful examination of the steady-state RNA levels of these two transcripts by densitometer scanning suggests that a larger 3.0-kb transcript readily accumulates during amplification, while the smaller 1.6-kb transcript does not (data not shown). This finding may be relevant to potential position effects in relation to the polyadenylation signal site, previously noted for other AREs (2). We then examined the effect of hGTA 2 overexpression on endogenous CHO GTA mRNA levels. This was possible since the amplified hGTA 2 fragment contains only a portion of the 3' untranslated region of this gene and contains none of the protein coding sequences (Fig. 2) (Adams et al., submitted). We, therefore, could examine the steady-state levels of endogenous CHO GTA mRNA transcripts in vector-transfected cells as well as hGTA 2-transfected cells (Fig. 4). We found that in the CHO cell line containing

Bunnell et al.

Fig. 4. Northern blot analysis of endogenous CHO GTA mRNA. (A) Total RNA, 40 #g, isolated from CHO (dhfr-) cells containing the p91023(B) vector only [CHO ( d h f r ) ] , amplified to a final concentration of 10 ~M MTX, as well as from CHO ( d h t ¥ ) cells containing the p91023(B) vector with the hGTA 2 sequences (hGTA 2), amplified to a final concentration of 10 ~M MTX. These RNAs were then probed with the human GTA coding region. This probe has no overlapping homology with the hGTA 2 3' untranslated region used in these experiments. (B) The same Northern blot shown in A was stripped and then rehybridized with a 13-tubulin cDNA probe to demonstrate equal loading of RNA, followed by a murine GalTase cDNA probe.

only the p91023 (B) vector sequences, the normal endogenous 3.9-kb transcript, and a very small amount of a smaller 1.7-kb transcript, were observed (Fig. 4, panel A). However, in the CHO cells containing the ampli-


157

fled p91023(B)-hGTA 2 construct very little of these endogenous GTA transcripts could be detected (Fig, 4, panel A). Equal loading of RNA samples was verified by rehybridization of this same blot with a fi-tubulin cDNA probe (Fig. 4, panel B). Thus, overexpression of the hGTA 2 construct apparently was leading to the loss of the endogenous CHO GTA mRNA. We decided to examine this effect further by analyzing steady-state GTA protein levels. Cellular protein extracts were prepared from each of the cell lines for enzyme assay and Western blot analysis. Once again, a specific dose-dependent loss of GalTase enzyme activity was observed in the hGTA 2 containing cell lines, but not in the amplified vector-only cell lines (Fig. 5). The absence of any effect on GalTase expression in the amplified vector-only ceil line suggests that this loss of activity may be associated with the expression of the truncated GTA mRNA, and not the vector sequences or the amplification process. Assays of unrelated glycosyltrans-

CHO, W. T. m GTA 3' UT, 1 /IM MTX ® GTA 3' UT, t0 ~M MTX GTA 3' UT, 20 ~M MTX o Vector, 1 gM MTX

10

9

I

8

~.

7

t

oT - -

6

×

5

CM

E CL. 4 O

3 2

/

J

1 I

20

40 Reaction Time (rain)

f

__

6O

Fig. 5. Galactosyltransferase assays on p91023(B)- and pg1023(B) hGTA 2-containing cells. Assays are identical to those described in Fig. 1. The vector-only cells were amplified with 1, 10, and 20 #M methotrexate. All of the vector-containing cell lines were identical withregard to GalTase activity, therefore, only the t IzM MTX amplified cell line is shown.

ferase activities demonstrated no change during hGTA 2 or vector-only amplification (data not shown). We then decided to examine any possible physical changes in GalTase and GTA protein levels by Western blot analysis. All Western blots were performed using identical amounts of total cellular protein, as verified by Coomassie gel staining (Fig. 6, panel A). Using a previously characterized polyclonal antibody to bovine GalTase known to react with murine GalTase (25), we found that two different proteins, 76 and 40 kDa in size, were affected by ectopic hGTA 2 expression (Fig. 6, panel B). In cells containing only the amplified vector sequences, the levels of both proteins were unaffected; however the larger 76-kDa species completely disappeared, while the smaller 40-kDa species was substantially reduced at even low-copy amplification of the hGTA 2-containing construct (Fig. 6, panel B, lanes 1 and 2). The dimunition of the presumed 40-kDa GalTase protein species appeared to plateau in the high-copy hGTA 2 amplification cell lines (Fig. 6, panel B, lanes 3 and 4), while no such effect was seen in high-copy vector containing cell lines that were grown in the presence of identical amounts of methotrexate (Fig. 6, panel B, lane t). When a specific human GTA peptide antibody was used (Adams et al., submitted), which has been shown to specifically cross-react with this protein in a number of mammalian species, only a 76-kDa protein was identified (Fig. 6, panel C, lane 1). In addition, this protein was completely lost in all subsequent amplified cell lines examined (Fig. 6, panel C, lanes 2-4). These results suggest that the ectopic expression of the human GTA 3' untranslated region leads to the rapid loss of the endogenous CHO 76-kDa GTA protein and the concomitant toss of an associated 40-kDa GalTase protein. When Northern blots containing total RNA from cells with various amounts of the amplified hGTA 2 sequence were probed with a routine GalTase cDNA probe, no significant differences were

158

Bunnell et ai.

Fig. 6. Western blot analysis of cellular proteins fromvector and hGTA 2 mutant cells. (A) Fifty microgramsof total cellular protein was resolvedby 10%SDS-PAGEand stained with Coomassieblue to demonstrateequivalent loadingof protein. Molecular weight markers are shown on the left. (B) The same proteins shown in panel A were transferred to nitrocellulose and analyzed with a bovineGalTase antibody. The estimated sizesof the two detected protein species are shown on the left. (C). The same proteins shown in A were transferred to nitrocellulose and analyzed with a human GTA-peptide antibody. The estimated size ofthe protein is shown on the left. In all blots the degree of amplification (~M MTX) of the p91023(B)-hGTA 2 construct is indicated above the lanes. The cells containing the p91023(B) vectoronly were amplified to a final concentrationof 20 #M MTX. observed i n steady-state GalTase m R N A levels (data not shown). Therefore, the apparent loss of immunoreactive GalTase protein and enzyme activity must be due to some unknown posttranscriptional or posttranslational event. Interestingly, both the GalTase and G T A antibodies detect the 76-kDa protein species whose expression is lost in these experiments. Finally, we then decided to examine the potential affect of ectopic hGTA 2 expression on cell-surface GalTase staining. Indirect immunofluorescence analysis of amplified vector control cells and hGTA 2 RNA-containing cells indicated that the amount of cell-surface associated GalTase was dramatically reduced in the h G T A 2 cells (Fig. 7). Similarly, permeabilized hGTA 2 mutant cells stained with the human GTA peptide antibody confirmed the loss of G T A protein from the cytoplasm of these cells as no significant cell staining was observed (data not shown), Thus, the presence of cell-surface-associated GalTase

appears to be either directly or indirectly correlated with the level of G T A protein expression.

Alterations in Normal Growth Characteristics of GTA/GalTase Mutants. Finally, the mutant cell phenotype we have generated by removing endogenous G T A protein may provide further information regarding the cellular function of the GTA protein kinase. The GTA protein kinase contains a large cdc-related domain that may be relevant to its function (Adams et al., submitted). We therefore, decided to examine the changes in the normal cellular phenotype in these hGTA 2 mutant cells to ascertain possible functions of the G T A gene product. We and others have noted that specific increases in cell-surface associated GalTase enzyme activity are associated with 3-agonistinduced hypertrophy of rat parotid glands (24, 26) and in certain transformed cells (27, 28). Concomitant increases in G T A m R N A levels have also been noted in the hypertrophied rat


159

Fig. 7. Immun••u•res•enceana•ysis•fwi•d-typeandhGTA2mutantsusingtheb•vineGa•Taseantisera.Cell-surface staining with the GalTase antisera is shown in the top panels for vectoronly, indicated as CHO(dhfr-) (20 gM MTX) and p91023(B)-hGTA2 (20 vM MTX), indicatedas GT-, cells. The correspondingimage of the nuclei stained with Hoechst is showndirectlybelow. parotid gland (27). These studies have shown that specific modifiers of the GatTase enzyme can block this transition to growth dramatically (24, 26). We looked at the incorporation of [3H]thymidine (Td) into DNA as a measure of cell growth. Quantitation of [3H]Td incorporation by trichloroacetic acid (TCA) precipitation demonstrated a threefold increase in DNA synthesis in the mutant cells versus control or amplified vector treated cells (Fig. 8, panel A). This result was also confirmed by in situ analysis of [3H]Td incorporation into cellular nuclei (data not shown). In addition, this effect could not be attributed to the amplification process since cells containing only the amplified vector sequences had no cell cycle abnormalities (Fig. 8, panel A). To confirm this observation, we took asynchronous, as well as synchronized, normal and

mutant hGTA 2-containing cells and analyzed their respective cell cycles using fluorescence-activated cell sorting (FACS). As can be seen, asynchronously growing hGTA 2containing mutant cells had a much higher proportion of cells in S phase (60%) versus the control cell population (40%) (Fig. 8, panel B). When these cells were treated with low serum for 48 h, the control ceil population responded by shifting approximately 70% of the cells to G0/G ~ phase, while the hGTA 2 mutants had only 38% of its cells in G0/G I (Fig. 8, panel B). Finally, when the two ceil populations were released from this quiescent state of growth by addition of 10% serum and cultured for 12 h, the control cells demonstrated a large increase in the number of cells entering mitosis, while the hGA 2 mutant cells demonstrated a slight increase in the proportion of cells entering S phase (Fig. 8, panel B).

160

Bunnell et al.

GT-

CHO ( d h f r - ) B.

A.

iI il "~° !',\,. . . . . . Il r~s

unsynchronized o lo Q.

o X E Q.

s

~L2 0.5% s e r u m , 4 8 hours

-o 0 Z 0

(D

rL-,

r e l e a s e d , 12 h o u r s

Fig. 8. Alterations in DNA synthesis and cell cycle regulation in hGTA 2 mutant cells. (A) The amount of [3H] Td incorporated into cellular DNA (normalized to total cellular protein) was assayed in wild-type (open bar), vector-transfected (stipled bar), and p91023(B)-hGTA2 transfected (black bar) cells. The vector-only and hGTA 2 mutant cells were both amplified by methotrexate to a final concentration of 20 ~M. (B) Flow cytometric analysis of wild-type (CHO (dhfr-) and mutant (GT-) cells. The relative distribution of cells within the cell cycle is shown for (1) unsynchronized cell populations, (2) cells arrested in G 0 / G 1by serum depletion, and (3) cells restimulated for 12 h after 48 h of serum depletion. DISCUSSION In this report we demonstrate that aberrant expression of a gene's 3' untranslated region leads to the quantitative loss of not only t h e specific m R N A and protein of a gene whose untranslated region is misexpressed, but also of an associated protein as well. This loss of protein and enzyme activity occurs in a dose-dependent manner, apparently related to the level of ectopic R N A expression. These results are independent of the vector, or the amplification process, since cells containing only the amplified vector sequences were unaffected with regard to G T A or GalTase expression. The 3' untranslated region of the G T A region used in these experiments con-

tains an A U - r i c h motif, as well as substantial upstream sequence, that is reminiscent of other genes whose expression is regulated by their 3' untranslated regions (2, 3, 8). However, because of the size of the c D N A fragment utilized in these experiments, we cannot definitively state that these A R E s regulate this gene or that additional sequences within the misexpressed h G T A 2 R N A are not required for these effects. W e are currently generating deletion mutants of the 1.6-kb h G T A 2 sequence for further, more definitive analysis. Our data suggest that we m a y have, in effect, created a dominant negative m u t a n t by flooding the cell with a truncated, but homologous, untranslated region. One might conjecture that, normally,


this region of the endogenous CHO GTA mRNA may interact with a rate-limiting cellular factor that precludes its rapid turnover. The vast excess of truncated 3' untranslated sequences could circumvent this protective system by competing for the available cellular protective factor(s), thereby displacing the endogenous mRNA and accelerating its removal from the cell, The result would be identical to one type of dominant negative mutation described by Herskowitz (14). This type of model might also explain why genes containing AREs are not regulated similarly within a certain cell (5). However, more rigorous examination of these control region sequences and the proteins that may bind to these regions is necessary to prove these assumptions. The loss of endogenous cellular GalTase activity and protein in response to the loss of GTA protein is also intriguing. In plants, expression of antisense RNA sequences specific for the small subunit of ribulose biphosphate carboxylase results in a reduction of not only small subunit mRNA and protein, but also of the heterologous large subunit protein (29). These effects appear to be mediated by translational and posttranslational factors in this system (29). Our observations with GTA and GalTase may involve a similar mechanism, although that has not been rigorously proven. Since GTA has been found to specifically phosphorylate GalTase in vitro, we might suggest that aberration of this posttranslation modification event might lead to the observed effects. In view of the high degree of homology between a large domain of the GTA protein kinase and the cdc protein kinases cdc 28 and cdc 2Hs (Adams et al., submitted), the apparent effects in cell cycle associated with the loss of endogenous GTA protein kinase are quite interesting. These results suggest that the normal CHO cell cycle was affected in the hGTA 2 mutants, presumably by either a block between the S phase and the G2/M phase of these cells or by the accelerated

161

passage of these cells through some phase(s) of the cell cycle. Changes in cell cycle regulation of the hGTA 2 mutants in some ways resemble the conditional dominant negative mutants of the yeast cdc protein kinase genes (30). Our studies have shown that the GTA protein kinase phosphorylates a number of substrates in vitro, including GalTase and histone H1, and that it binds to calmodulinSepharose (Adams et al., submitted). Calmodulin has also been shown to be importan~t for normal celt cycle progression (31). The apparent alteration in normal progression from S phase to M phase in these cells suggests that the GTA protein kinase might be an important part of this process. Further analysis of these mutant cells and the role of the GTA protein in cellular processes will, it is hoped, provide an answer. ACKNOWLEDGMENTS This research was supported by a grant from the American Cancer Society (CD-389) and by a Basil O'Connor Scholar Award (5-577) from the March of Dimes/Birth Defects Foundation to V.J.K.B.A.B. is a student of the Cell and Molecular Biology Graduate Training Program (UAB). The authors would also like to thank Dr. B. Shur for providing the GalTase antisera and Dr. R. Kaufman for providing the p91023 (B) expression vector. The authors would like to thank Ms. Cynthia Webster for preparing this manuscript. UTERATURE CITED 1. Humphreys-Beher,M.G.,Bunnelt,B., Van Tuinen, P., Ledbetter, D., and Kidd, V.J. (1986). Proc. Natl. Acad. SeL U.S.A. 83:8918-8922. 2. Shaw,G., and Kamen,R. (1986). Cell 46:659-667. 3. Caput, D., Beutter, B., Hartog, K., Thayer, R., Brown-Shimer, S., and Cerami, A. (1986). Proc. Natl. Acad. Sci. U.S.A. 83:1670-I674. 4. Rahmsdorf,HA., Schontal, A., Angel, P., Litfin, M., Ruther, U., and Herlich, P. (1987). Nucleic Acids Res. 15:1643-1660.

162

5. Schuler, G.D., and Cole, M.D. (1988). Cell 55:1115-1122. 6. Shyu, A.-B., Greenberg, M.E., and Belasco, J.G. (1989). Genes Dev. 3:60-72. 7. Bernstein, P., Peltz, S.W., and Ross, J. (1989). Mol. Cell. Biol. 9:659-670. 8. Wilson, T., and Treisman, R. (1988). Nature 336:396-399. 9. Morgan, D.O., Edman, J.C,, Standring, D.N., Fried, V.A., Smith, M.C., Roth, R.A., and Rutter, W.J. (1987). Nature 329:301-306. 10. Blose, S.H., Meltzer, D.I., and Feramisco, J.R. (1984). J. Cell Biol. 98:847-858. 11. Izant, J.G., and Weintraub, H. (1984). Cell 36:1007-1015. 12. Meeks-Wagner, D., and Hartwell, L.H. (1986). Cell 44:43-52. 13. Smithies, O., Gregg, R.G., Boggs, S.S., Korakwski, M.A., and Kucherlapati, R.S. (1985). Nature 317:230-234. 14. Herskowitz, I. (1987). Nature 329:219-222. 15. Wong, G.G., Witek, J.-A.S., Temple, P.A., Wilkens, K.M., Leary, A.C., Luxenberg, D.P., Jones, S.S., Brown, E.L., Ray, R.M., Orr, E.C., Shoemaker, C., Golde, D.W., Kaufman, R.J., Hewick, R.M., Ward, E.A., and Clark, S.C. (1985). Science 228:810-815. 16. Lau, C.-F., and Kan, Y.W. (1983). Proc. Natl. Acad. Sci. U.S.A. 80:5225-5229. 17. Urlaub, G., and Chasin, L.A. (1980). Proc. Natl. Acad. Sci. U.S.A. 77:4216-4220. 18. Wigler, M., Silverstein, S., Lee, L.-S., Pellicer, A., Cheng, Y.-C. and Axel, R. (1977). Cell 11:223232. 19. Kaufman, R.J., Wasley, L.C., Furie, B.C., Furie,

Bunnell et al.

20. 21. 22.

23. 24. 25. 26. 27. 28. 29. 30. 31.

B., and Shoemaker, C.B. (1986). J. Biol. Chem. 261:9622-9628. Scherrer,K. (1969). In Fundamental Techniques in Virology, (eds.) Habl, E.K., and Salzman, N.P. (Academic Press, New York), pp. 317-330. Cowan, N.J., Wilde, D.C., Chow, L.T., and Welfald, R.W. (1981). Proc. Natl. Acad. Sci. U.S.A. 78:4877-4881. Ausubel, F.M., Brent, R., Kingston, R.E., Moore, D.D., Smith, J.A., Seidman, J.G., and Struhl, K. (1987). In Current Protocols in Molecular Biology, (John Wiley & Sons, New York). Shaper, N.L., Hollis, G.F., Douglas, J.G., Kirsch, LR, and Shaper, J.H. (1988). J. Biol. Chem. 263:10420-10428. Marchase, R.B., Kidd, V.J., Rivera, A., and Humphreys-Beher, M.G. (1988). J. Cell. Biochem. 36:453-465. Bayna, E.M., Shaper, J.H., and Shur, B.D. (1988). Cell 53:145-I 57. Humphreys-Beher, M.G., Schneyer, C.A., Kidd, V.J., and Marchase, R.B. (1987). 3". Biol. Chem. 262:11706-11713. Podolsky,D.K., and Weiser, M.M. (1979). or. BioL Chem. 2544:3983-3990. Klohs, W.D., Wilson, J.R., Weiser, M.M., Frankfurt, O., and Bernacki, R.J. (1984). J. Cell. Physiol. 119:23-28. Rodermel, S.R., Abbott, M.S., and Bogorad, L. (1988). Cell 55:673~581. Mendenhall, M.D., Richardson, H.E., and Reed, S.I. (1988). Proc. Natl. Acad. Sci. U.S.A. 85:44264430. Rasmussen, C,D., and Means, A.R. (1989). EMBO J. 8:73-82.