Biochimie (1991) 73,729-737 © Soci6t6 fran¢aise de biochimie et biologie mol6culaire / Elsevier, Paris

729

Turnover rate of yeast PGK mRNA can be changed by specific alterations in its trailer structure P V r e k e n , R v a n d e r V e e n * * , V C H F de R e g t , A L de M a a t , RJ P l a n t a , H A R a u 6 * Biochemisch Laboratorium, Vrije Universiteit, tie Boelelaan 1083, 1081 HV Amsterdam, The Netherlands

(Received 7 November 1990; accepted 12 March 1991)

Summary m The effect of insertions in the 5'- and 3'-untranslated regions (UTR) of the Saccharomyces cerevisiae mRNA encoding phosphoglycerate kinase (PGK) on the stability of the transcript ill vivo was determined. None of the structural alterations in the 5'UTR affected mRNA turnover significantly, despite the strong negative effect on translational efficiency of some of these alterations previously observed [ 16]. We conclude that the structure of the 5'-UTR is not important for the relatively high affinity of PGK mRNA in yeast cells. Moreover, translation cannot be a major factor in determining the rate of turnover of this mRNA. Insertion of either a polyG or polyU, but not a polyA or polyC, tract into the 3'-UTR of PGK mRNA increased its half-life by a factor of about two. Introduction of a hairpin structure containing 18 G.C base pairs had only a slight stabilizing effect. We argue that the stabilization by the structural changes in the 3'-UTR is due to altered folding of the mutant mRNA which retards a rate-limiting endonucleolytic cleavage step in the normal turnover pathway of PGK mRNA. The stabilizing effect of local structural alterations in the 3'-UTR opens the possibility for further increasing the product yield of a (heterologous) gene cloned in yeast cells. yeast / Saccharomyces cerevisiae / mRNA stability / translation

Introduction The expression level of most genes is determined primarily by the abundance of the corresponding mRNA in the cell. This abundance, in turn, is a resultant of the rate at which the mRNA is produced (transcription, processing and nucleocytoplasmic transport) and the rate at which it is removed - or at least inactivated by nucleolytic degradation. However, in comparison to the enormous amount of effort spent on analysis of the regulatory mechanisms controlling transcription, mRNA degradation has drawn but scant attention. Nevertheless, the number of genes for which the rate of degradation of their transcripts has been found to be an important regulatory factor is steadily increasing [1-6] and, as a consequence, the importance of mRNA turnover in cot, troll cff the expression is now becoming generally recognized. *Correspondence and reprints **Present address: Center for Phytotechnol9gy RUL/TNO, Wassenaarseweg 64, 2333 AL Leiden, The Netherlands Abbreviations: IR, inverted repeat; mRNA, messenger RNA; PGK, phosphoglycerate kinase; REP sequences, repetitive extragenic palindromic sequences; UTR, untranslated region

It is obvious that differences in decay rates must ultimately be due, either directly or indirectly, to determinants within the mRNA structure itself. Analysis of the role of general features such as translational efficiency, the size of the poly(A) tail and the overall length of the mRNA indeed showed that changing any of these features, in particular mRNAs, did affect stability (see [3] for a recent review). This led to the suggestion that variations in these non-specific properties might underly the variation in decay rates observed in the population of mRNAs in a normal eukaryotic cell. However, more recent investigations have failed to reveal any clear-cut correlation between any of these properties and the half-life of individual members of a set mRNAs in either Dictyostelium discoideum [7] or Saccharomyces cerevisiae [8]. On the other hand, several specific structural determinants have now been identified that affect the metabolic stability of the m R N A molecule of which they are part. Among the first of these determinants discovered were the AUrich sequences present in the trailer region of several eukaryotic lymphokine and proto-oncogene mRNAs [9]. These sequences were found to act as autonomous destabilizing elements, possibly because they constitute recognition sites for a factor that facilitates (endo)-

P Vreken et al

730

nucleolytic attack or even for a specific endonuclease [10]. Further examples are the destabilizing stem-loop st~ctures in the trailer region of the transferrin receptor mRNA involved in iron-dependent regulation of its turnover [ l 1] and the stem-loop structure at the 3'end of mammalian histone mRNAs [12] as well as a region centered on the stop codon of yeast histone mRNA [131. The latter two features are responsible for cell cycle-dependent changes in the turnover rate of these messengers. Most of the research on structural elements affecting eukaryotic mRNA stability so far has been carried out on mammalian mRNAs. Recently, however, we and other groups have started to exploit the genetic and molecular biological advantages offered by yeast to address this question. Since the half-lives of yeast

EcoRI

A

Kpnl

mRNAs vary from about l to over 100 min they constitute an excellent system for identifying cis-acting stability determinants and analyzing their precise function in mRNA decay. The first indications for the existence of destabilizing elements in yeast mRNAs came f,'om the analysis of chimeras encompassing (part or) a 'stable' m R N A (eg those encoding pyruvate kinase, phosphoglycerate kinase or actin) fused to part of an 'unstable' one (eg URA3, M a t a l , STE3). In many cases such chimeric mRNAs were unstable, a property shown to be conferred by the presence of the fragment of the m R N A having the shorter half-life, and not due to the loss of stabilizing sequences from the one having the longer half-life [8, 14, 15]. Only in the case of the Matttl mRNA has the destabilizing element been more pr,-cisely delineated. In this case a

Clal

Sspl

EcoRV l-lindlll BamHl

ATG I

-45 1 5' mRNA

F.coRI IL."/ //IdF.l.ul6

_ l

1251 ,~letionCbI-SspiI~ragment(76 b p ) / ' e ~

3' mRNA [ns~ttkmClal-ecoRiadapter

Eco RI !T 3

Eco RI

[7"] 1251

[7"] j T

1329/1342 3' mRNA

Sspi

T 7 I

1251

1400

1

I405/1418 3' mRNA TYEpIL~-pXconstructs

In~,ion ofhomopolymerabltoEoaRIsite

C

oligo

sequence

pG pC pA pT ClaI-EcoRI adapter

AATTTGGGGGGGGGGGGGGGGGG AATTCCCCCCCCCCCCCCCCCCA AATTCAAAAAAAAAAAAAAAAAA AATTTTTTTTTTTTTTTTTTTTG CGAAGGAATTCCTT

I

(5~3')

Fig 1. Construction of mutant PGK genes carrying various insertions in the 5'- or 3'-UTR. A. Restriction map of the PGK gene present in the plasmid YEpR5 [20, 211. The hatched bar represents the coding region of the PGK gene. Transcription initiation and termination sites are indicated. B. Enlargement of the 5'- and T-untranslated regions prior to insertion of the homopolymer oligonucleotides. The arrows indicate the points of insertion. C. Oligonucleotides used for the insertion.

Turnover of yeast PGK mRNA 42 nucleotide long segment of the coding region encompassing codons 6 8 - 8 0 appears to contain a determinant that is essential, though m a y b e not sufficient, for rapid m R N A turnover [ 15]. W e have taken a different approach towards unraveling the relationship between m R N A structure and stability in yeast by introducing predetermined structural changes into the untranslated leader and trailer regions o f the 'stable' phosphoglycerate kinase (PGK) m R N A , that in our hands has a half-life o f about 34 min. As previously shown, none of the changes made in the trailer region (3'-UTR) had any detectable effect on the efficiency with which the P G K m R N A was able to function as a template for protein synthesis, whereas several o f the alterations in the leader (5'U T R ) severely affected translational efficiency [16]. In this paper we report an analysis o f the decay rates o f the various mutant P G K m R N A s showing that alterations in the leader structure do not significantly influence m R N A stability. In contrast, three of the mutant P G K m R N A s carrying altered 3'-UTR sequences show a clearly detectable increase in stability up to about two-fold.

731

Restriction enzymes were obtained from either Bethesda Research Laboratories (Gaithersburg, MD USA) or New England Biolabs (Beverly, MA USA) and used in accordance with the suppliers' instructions. Klenow fragment of DNA polymerase I, S 1 nuclease and T4 ligase were from Bethesda Research Laboratories. Radiolabeled nucleotides were purchased from Amersham. Synthetic deoxyoligonucleotides were prepared in our laboratory using an Applied Bio~ystems 381A DNA synthesizer.

Genes can'ying an altered 3"-UTR The KpnI-BamHI fragment of plasmid YEpR5 [20], which contains the 3'-terminal portion of the PGK gene (fig IA) was cloned hlto the polylinker of plasmid pGEM4 (Promega) from which the EcoRl site had previously been removed by cutting with EcoRI, filling in the ends by treatment with Klenow polymerase and religation. The HindlII-XhoI fragment from plasmid pLGSD5 carrying the yeast URA3 gene [22] was ligated into the EcoRV site of the resulting plasmid (fig I A) to give plasmid pGEM-PGK. This plasmid was then cleaved at the unique ClaI site present in the 3'-UTR of the PGK insert and religated in the presence of a CiaI-EcoRI adaptor (fig IC) to create a unique EcoRl site in the3'-UTR. This site was subsequently used to insert the same homopo~yme~ oligonucleotides used to alter the 5'-UTR. Insertion of these oligonucleotides regenerated only one EcoRl site either upstream or downstream depending upon their orientation (fig IC). Positive transformants were selected by Southern analysis of HindllllEcoRll Kpnl digests using oligonucleotides specific for each of the four possible insertion mutants. Mutant plasmids were designated pGEM-PGKpX were X indicates the nucleotide present in the coding strand of the homopolymer insertion. In order to reconstruct complete PGK genes carrying the homopolymer insertions in the 3'-UTR, the Kpnl-BamHl fragment from each of the four pGEM-PGKpX plasmids was used to replace the KpnI-BamHI fragment of YEpR5 giving rise to plasmids YEpR5-pA, -pU, -pC and -pG. "t'E~R5-pGpC was obtained in the same way by consecutive insertion of two copies of the double-stranded 18-bp long G.C oligonucleotide in opposite orientations into the EcoRI site of pGEM-PGK. Finally, YEpR5-L containing only the Clal-EcoR1 linker in the 3'-UTR was obtained by the same procedure. The construction of plasmids YEpR5-A76 and YEpR5A76L, in which the 76-bp long Clal-SspI fragment from the 3'-UTR has been removed or replaced by an EcoRl linker (figs IA and IB) has been described elsewhere, as has concentration of its YEpR5-A76-pX derivatives carrying 18-bp long homopolymer sequences inserted into this EcoRl site [161. All plasmids contain both the original TRPI and the newly introduced URA3 marker gene. The latter is required for their selection in Saccharomyces cerevisiae strain Y260 used for determination of the half-lives of the various mutant PGK mRNAs.

Recombinant DNA technology

SI mapping the 3"-ends of the various mutant PGK transcripts

All DNA manipulations were carried out essentially as described in Maniatis et al [17]. Escherichia coil strain DHI (F", fecAl, endAl, GyrA96, thi, hsdRl 7, supE44, relAl, ~,-; [181) was used for the propagation of all plasmids. Transformation was carried out using the MOPS-RbCi method [ 19].

The EcoRI-HindlIl fragment isolated from YEpR5-L (fig I A), labeled with [0t-32p]ATP at the EcoRI site with the aid of Klenow polymerase and hybridized to 15-25 I.tg of total RNA isolated from cells transformed with a mutated PGK gene.The hybrids were treated with 75 units of S 1 nuclease for 2 h at ! 6°C. The products were analyzed on 6% denaturing polyacrylamide gels. Maxam and Gilbert sequencing reaction mixtures of the EcoRI-Hindlll fragment were run alongside to determine the length of the protected region. The 3'-ends of wild-type and YEpR5-A76L PGK RNA were determined in the same way using the EcoRI-Hindlll fragment from YEpRS-L or YEpRSA76L as probe.

Materials and methods Materials

Construction of mutant PGK genes Genes carrying an altered 5"-UTR The construction of mutant PGK genes carrying homopolymer sequences in the 5'-UTR has been described in detail elsewhere [ 16]. Briefly, synthetic duplex deoxyoligonucleotides consisting of their 18 G.C of 18 AoT base pairs (fig IC) were inserted in either orientation into the unique EcoRI site present in the 5'UTR of the PGK gene of plasmid dELul6 (fig IB). This plasmid was derived from YEpR5 [201 by deletion of 29 of the original 45 bp from the 5'-UTR of the (slightly modified) PGK gene [21] present in the plasmid. The resulting plasmids were designated dlNpA, dlNpU, dlNpG and dlNpC to indicate the homopolymer sequence present in their PGK transcript.

Determination of mRNA half-life

YEpR5 derivatives carrying the various mutant PGK genes were transformed to Saccharomyces cerevisiae strain Y260 (Mata, ura3-52, rpbl-l), which carries a temperature-sensitive mutation in the largest subunit of RNA polymerase It [23], using the freeze-thaw method [24]. Transformants were grown

732

P Vreken et al

B

A

•

"i 1

10

"

I

"

30

D

30

30

45

45

60

60

aid

1

,ID

75

75

~

2

~~

90

90

" ....

105

105

,,.

"~

I

o

,,.

0 i 15

.... .

min

15

PG.~

.

i°'~

0

2 3

100-k

C

min

.

'~..

.

I.tg R N A

~

4 ID

6

.... I

tv2 = 34 + 2 m m I

I

"

45

I

"

60

I

~

I

5

~

I

o

minutes at 36°C

"

I

a

o

I

I

,i

,

8

I~g R N A

Fig 2, Determination of the half-life of wild-type PGK mRNA. Total RNA from YEpR5-transformed Y260 cells was isolated at different times after arresting transcription by an increase in temperature from 25 to 37°C, and analyzed by hybridization to a PGK-specific probe. A. Northern hybridization. Lane 1:0 rain, lane 2 : 2 0 rain, lane 3 : 4 0 min, lane 4 : 6 0 min after transcriptional arrest. In each lane 2.5 I~g of RNA was applied. B and C. Slot-blot hybridization. Duplicate filters were hybridized with excess probe specific for PGK mRNA (B) or 17S rRNA (C) respectively, Time after transcriptional arrest is indicated. In each ~lot 2.5 I~g of RNA was applied. The hybridization signals were quantified either densitometrically using an LKB laser densitometer (Northern blot) or by liquid scintillation counting (slot-blot). The data are shown in E: A, PGK mRNA, slot-blot analysis; n, PGK mRNA, Northern analysis, *, 17S rRNA. The lines represent a best-fit analysis for each of the various sets of data. D. Increasing amounts of total RNA isolated from YEpR5-transformed Y260 cells prior to transcriptional arrest (time 0), were slot-blotted and hybridized with the PGK-specific probe under the same conditions as used in B. The quantitative data are shown in F.

in minimal medium (2% (w/v) glucose, 0.7% (w/v) yeast nitrogen base without amino acids (DIFCO)) in a starter culture for 48 h at 24°C. A flask containing 200 ml of minimal medium

prewarmed to 24°C was inoculated with 2.5 ml of the starter culture and grown for approximately 16 h until the culture had reached an optical density of 0.6 at 600 nm (1 cm path length).

Turnover of yeast PGK mRNA The culture was then shifted to 37°C to block further mRNA transcription. Aliquots (20 ml) were taken at 15-min intervals and immediately frozen in liquid nitrogen. RNA was isolated from the samples as described by Zitomer et al [251. Samples (2.5 l.tg) were denatured in a solution containing 6.15M formaldehyde and 10 X standard saline citrate during 15 min at 65°C [26]. Samples were then placed on ice and spotted onto nitrocellulose (Schleicher and Schuell) in triplicate, using a Schleicher and Schuell spot-blot apparatus. The filters were bak~,d for 2 h at 80°C in a vacuum oven and hybridized with a probe spanning the 5'-region of the PGK gene from EcoRI to the KpnI site (fig I A)~ The probe was labeled by the random primer method as described previously [21]. Precautions were taken to carry out the il:~bridization under conditions of excess probe (fig 2, panels D and F). In all cases a duplicate filter was analyzed using a probe specific for 17S rRNA as internal control. Furthermore, in each experiment we included a determination of the half-life of wild-type PGK RNA in YEpRS-transformed Y260 cells, as a check on the experimental conditions. After exposure on an X-ray film the spots were cut out and counted in a liquid scintillation counter. The half-lifes of the transcr;pts were calculated from the slope of the semi-log curves obtained by best-fit analysis. In some cases the RNA samples were analyzed by Northern blotting after gel electrophoretic separation on !.5% agarose gels [211. At least three independent transformants were analyzed for each of the mutant PGK genes.

Results and discussion

Determination of mRNA half-lives In order to study the effect of specific changes in the structure of a ye,,~:~ mRNA on its half-life in vivo, we inserted various t~omopolymer seqaences into either the 5'- or 3'-UTR of the gene encodin~ phosphoglycerate kinase (PGK). This gene was chosen for two reasons. Firstly, it has unique transcription initiation [21] and termination ([27]; however, see next section) sites thus producing a transcript with well-defined leader and trailer regions. Secondly, reported half-lives of PGK mRNA indicate it to be a moderately stable transcript [8, 15] thus allowing the detection of stabilizing as well as destabilizing effects of the insertions. The half-lives of the various structurally altered PGK mRNAs were determined by introducing the mutated genes on a multi-copy plasmid into Saccharomyces cerevisiae strain Y260, which carries a temperature-sensitive mutation in the gene encoding the largest subunit of RNA polymerase II [23]. Increasing the temperature of the culture from 24 to 37°C blocks m R N A transcription and thus allows one to assay the rate of degradation of a specific mRNA by hybridization with a radiolabeled probe. Despite the necessity of a temperature shift, this method has been shown to give results that generally are in good agreement with the data obtained by other techniques such as 'approach to steady state labeling' or the use of transcription inhibitors [8].

733

Figure 2 compares stability assays of wild-type PGK mRNA in YEpR5-transformed Y260 cells by either Northern (panel A) or slot-blot hybridization (panel B) using a probe complementary to the 5'terminal one-third of the PGK coding region (up to the KpnI site; fig 1A). Panel C shows hybridization of a duplicate slot-blot filter with excess probe specific for 17S rRNA. This type of internal control was used in all experiments reported in this paper. Furthermore, the data shown in panel D and F demonstrate that under the conditions used for hybridization (2.5 l,tg of RNA per slot) the PGK-specific probe was in excess. As can be seen in panel E, the two types of stability of PGK mRNA are in good agreement with each other, as well as data in the literature [15], both giving a value of 34 + 2 min for the half-life. Consequently, for reasons of convenience we used slot-blot hybridization in all further experiments. Although the probe used also detects the transcript derived from the chromosomal PGK gene, this will not significantly affect half-life values obtained because this transcript represents less than 10% of the total amount of PGK mRNA present in the transformants [16,21].

Mapping of transcription initiation and termination points We have previously shown that the structural alterations present in the 5'-UTR of the PGK genes used in this study did not affect the position of the transcription initiation site and, thus, lead to the expected change in mRNA structure [16]. In order to determine whether the structural changes introduced into the 3'-UTR left the 3'-end of the PGK mRNA unaffected, we carried out S 1 nuclease mapping experiments. Figure 3 shows the results obtained for wild-type PGK mRNA (lane 3) and the YEpR5-pG and -pU mutants (lanes 1 and 2). In all three cases two major 3'-ends were found at positions corresponding to the U~405and AI4t8 residue located 157 and 170 nt downstream of the stop codon in the wild-type sequence respectively. The same result was obtained for all other mutant PGK transcripts used in this study (data not shown). The 3'-end points determined by us differ from the one previously reported by Hitzeman et al [27] whid~ was located at a position between 86 and 93 nt downstream of the stop codon (positions 1337-1344). The latter site is detected in our experiments as well (fig 3) but represents less than 10% of the total number of PGK transcripts in each case. Since the probe used by Hitzeman et al [27] would have allowed detection of the end points at positions 1405 and 1418, we have no explanation for the discrepancy. However, our results clearly show that all mutant PGK mRNAs have the trailer structure expected on the basis of the mutation introduced into the gene.

734

P Vreken et al

Effect o f structural alterations in the 5"- and 3"-UTR on PGK RNA stabili~

A

"x.

Effect of homopolymer insertions in the 5"-UTR As shown previously, structural alterations in the 5'UTR can significantly affect the translational efficiency of PGK mRNA in yeast cells [16, 21]. As far 10~ ,

GC GAT

. ,,.., . . . . . . . .

-,

0 15 30 45 60 75 90 105 mlnutu at 36°C

12

3

x

1418

" - ' U 1405

to

G' is' i0' ~S'/0' ~s' 60' lhs'

tOO.

1oo

-"A

,01

¢

~' is' ~0' ~s'/0' ~s' ~0' t~s' migratesat 36+C

~

mlnut~ at 36°C

|

.

I0.

6' i5" ~o' is'/~' i5" io' l~' mlnutemat 36"13

Fig 4. Effect of homopolymer insertions in the 5'-UTR on PGK mRNA half-life. A. dlNpA; B. dlNpC; C. dlNpU; D. dlNpG. Each panel shows a representative experiment for the type of transformant in question. The numerical value shown in each panel is the mean of multiple assays on at least three independently obtained transformants.

--- U 1341

Fig 3. Mapping of the 3'-end of PGK mRNA. The EcoRlHindIII fragment of YEpR5-L, labeled at the EcoRl site, was hybridized to 15-25 ~tg total RNA isolated from YEpR5-L (lane 3), YEpR5-pG (lane 1) or YEpR5-pU (lane 2) and treated with 75 units of S I nuclease. The products were analyzed on a 6% denaturing polyacrylamide gel alongside a Maxant and Gilbert sequence ladder of the same fragment. End points ere indicated. The right-hand part of the autoradiogram was deliberately overexpressed to visualize the signal around U1341, which was previously identified as the 3'-end of PGK mRNA by Hitzeman et al [271.

as insertion of homopolymer stretches is concerned, polyG (18 nt) reduced translation to a virtually undetectable level, while a similarly long polyU stretch lowered translation by a factor of about two [ 16]. The data presented in figure 4 demonstrate that despite their reduced rate of translation the stability of the mutant mRNAs carrying the altered leaders is hardly, if at all, affected. The PGK transcripts encoded by dlNpU and dlNpG (panels C and D, respectively) do not appear to differ significantly from either wild-type transcript (fig 2) or those derived from dlNpA and dINpC (fig 4A, B), which display a normal translational efficiency [16]. Insertion of a polyG stretch may even have a slight stabilizing effect since the dlNpG transcript shows a half live of 41 versus 34 rain for the wild-type mRNA (fig 4D). Similarly, other changes in the nucleotide composition of the PGK 5'-UTR that reduced translational efficiency by a factor of three to four failed to have any detectable effect on the steady state level, and thus probably the rate of degradation, of the mutant transcripts [ 16]. These results are at variance with an earlier report by Hoekema et al [20] who found that blocking translation of PGK mRNA by introduction of a stop codon near the 5'-end of the coding region dramatically reduced its steady state level in yeast cells, presumably because of an increased turnover of the untranslated transcripts. Similar results have been reported for other yeast mRNAs such as those derived

Turnover of yeast PGK mRNA from the URAI [28], URA3 [29] and SUC2 [30] genes. On the other hand, our data are in agreement with the observation of Bairn et al [31 ] that a structural change at the 5'-end of the coding region of yeast CYC1 mRNA which introduces the potential for secondary structure, causes a considerable drop in ribosome loading without significantly affecting the steady state level (and thus probably the half-life) of the mRNA. Similarly, Purvis et al [32] did not observe any effect on stability of mutations in yeast pyruvate kinase mRNA that severely curtailed its translation. A caveat in the interpretation of these data, however, is that structural alterations introduced into the mRNA might cause it to be degraded by a pathway different from the one entered by the wildtype transcript [33]. Nevertheless, also in studies on a set of wild-type yeast mRNAs displaying wide variations in stability, Santiago et al [34] were unable to detect an obvious correlation between the degree of ribosome loading and half-life. In fact, translation may even be required for rapid mRNA degradation in yeast, at least in some cases, as suggested by the results of studies by Parker and Jacobson [15] on stability determinants in yeast Matod mRNA. Apart from demonstrating that the rate of translation is not the dominant factor in determining the metabolic stability of PGK mRNA, our results also show that the structure of the 5'-UTR is of minor, if any, importance in this respect. Clearly, the 5'-UTR of PGK mRNA does not contain any determinants ensuring the relatively long half-life of this messenger. Effect of structural alterations in the 3'- UTR Although each of the three main segments of a eukaryotic mRNA (5'-UTR, coding region and 3'-UTR) may contain stability determinants (see [4] for a review), in a large number of cases the 3'-UTR appears to be of major importance ([3]; and references therein, 9, 11, 12, 35-41). In order to analyze the role of the 3'-UTR in PGK mRNA degradation, we first introduced a deletion of the ClaI-SspI fragment (fig 1). On the basis of the 3'-end of wild-type PGK mRNA previously reported by Hitzeman et al [27], this deletion should have removed all but about 10 nt of the 81 nt long trailer region. The S 1 mapping experiments described above, however, demonstrated the 3'-UTR of all PGK transcripts to be 154-167 instead of 81 nt in length (fig 3). Thus, the YEpRS-A76 and YEpRS-A76L PGK transcripts still contain about half of the original 3'-UTR. As shown in figure 5D and E, the half-lives of these transcripts are identical to that of wild-type PGK mRNA, demonstrating that the deleted sequences are not important in determining the metabolic stability of this messenger. In a further set of experiments we inserted homopolymer sequences into either the wild-type or the shortened 3'-UTR at the unique EcoRI site introduced by means

735

of insertion of a linker into the ClaI site (fig 1). Figure 5 (panels B, C and F) shows that insertion of an 18-nt long polyA or polyC stretch did not alter the stability of the resulting mRNA. Insertion of an equally long poiyG or polyU stretch into the intact or partially deleted 3'-UTR, however, has a clear effect (fig 6A-D). PGK transcripts carrying either of these sequences in their 3'-UTR were found to have a halflife around 60 min, a value about twice that of their wild-type counterpart and approaching that of the most stable yeast mRNA reported so far [3]. Three different models for degradation of eukaryotic mRNA have been proposed in the literature (reviewed in [3]). One of these, based upon experimental evidence from studies using a mammalian in vin'o

100-

100~..

•

"

"

a

I

. 3 6 + 6 min

10 . , ,

•

,

.

0

.

15

30

,',

,~

.

,

,

i

- ,

-

lOJ

|

.

45 60 T$ minutesat 3 6 ° C

90

105

0

-

15

|

-

30

w

-

45

w'

-

60

•

-

•

T$

-

90

w

wj

-

105

minutes at 36"°C 1 0 0

L

,,

|1 10

,

0

15

30

,

lOJ, . . . . . . . . . . . . . . . . 0 15 30 45 60 '75 90 105 minutesat ~°C

,

4.5 60 ?$ 90 lOS minutesat 36°C

!

i

F!

in1

m,n

10.

o

15

3o

45 e

minutesat ~

75 9o ~os

~)" is" ~o' ,~s' ~o' 'a' 9 o ' ~ ' minutes at 36°C

Fig 5. Effect of deletion and polyA or polyC insertion in the 3'-UTR on PGK-mRNA half-life. A. YEpRS-L; B. YEpRSpA; C. YEpRS-pC; D. YEpRS-A76; E. YEpR5-A76L; F. YEpRS-A76-pC. Each panel shows a representative experiment for the type of transformant in question. The numerical values shown in each panel are the mean of multiple assays on at least three independently obtained transformants.

736

P Vreken et al 1oo

A

f

ib. Z ee t ~ = SS _+S rain @

t~-= 58

l

lOJ 0

IS 30 45 60 75 90 105 m i n u t e s at 360C

mlnulesat 36*(:

100L

I

•

D

r

!ltm-

1

~ ]tm=SS±4mln

]o~ ............ 0

, ..~

15 30 45 60 75 90 105 ~*C

~0 ] 11/2= 44

mlnutuat

lOJ . . . , . . . . . . . . . . . . 0

.

15 30 45 60 75 90 10.5 ~*C

minutesat

+3

o 'is 3o'is m 7+ +o tin' minutes

at360C

Fig 6. Effect of polyU, polyG or G.C hairpin insertion in the 3'-UTR on PGK mRNA half-life. A. YEpR5-pU; B. YEpR5-pG; C. YEpR5-A76-pU; D. YEpRS-A76-pG; E. YEpR5-pGpC. Each panel shows a representative experiment tbr the type of transtbrmant in question. The numerical value shown in each panel is the mean of multiple assays on at least three independently obtained transformants.

degradation system [42], envisages degradation by a 3' ~ 5' exonuclease after (partial) deadenylation. The stabilizing effect of the polyG and polyU tracts in the 3'-UTR of PGK mRNA might, thus, be due to the fact that they impede the progress of this exonuclease by formation of secondary structure, an effect similar to that observed for the so-called REP or IR hairpin structures in a number of prokaryotic [43] and chloroplast [44] messengers. We have previously shown [ 16] that both a polyG and a polyU tract, when present in the 5'-UTR prevent 40S ribosomal subunits from scanning through to the translation start codon, probably because the polyG tract is able to form a self-complementary structure [45] while the polyU sequence might base-pair to a region (most likely the polyA tail) elsewhere in the mRNA.

In order to analyze the possible blocking of exonuclease by the polyG and polyU tracts further, we introduced a sequence into the 3'-UTR that should give rise to a thermodynamically highly stable hairpin structure. This hairpin, consisting of 18 consecuUve G.C pairs, was created by insertion of two copies of the double-strande• homopolymer G.C oligonucleotide in opposite orientations into the E c o R I site of YEpR5-L (fig 1). As shown in figure 6E this insertion did increase the stability of the mutant transcript, but only by about 30%, which is significantly less than the two-fold stabilization observed for the two homopolymer insertions. Moreover, REP elements, which have approximately the same thermodynamic stability as the G.C hairpin inserted into the PGK mRNA, cause a much larger (three-to ten-fold) stabilization of E coli mRNAs [41]. Consequently, we do not consider blocking of 3' ~ 5' exonuclease degradation a satisfactory explanation for the stabilizing effect of the homopolymer or hairpin insertions in the 3'-UTR of PGK mRNA. This conclusion is supported by the observation of Herrick et al [8], which has been confirmed in our laboratory (P Vreken, HA Rau6, manuscript in preparation), that deadenylation of yeast PGK mRNA does not decrease its metabolic stability and cannot be the rate-limiting step in the decay of this transcript, as required by the model described above. It should also be noted that introduction of a highly stable hairpin structure into the 3'-UTR of yeast pyruvate kinase mRNA failed to retard its decay [45]. A second model, applied in particular to 'stable' yeast mRNAs like the one encoding PGK, envisages ratelimiting decapping of the transcript followed by rapid 5' ~ 3' exonucleolytic degradation [3]. This model, however, does not seem to be tenable any more because the observation that an inverse relationship exists in yeast cells between mRNA length and stability [3], which forms one of its cornerstones, could not be confirmed when mRNA half-lives were assayed by any one of three methods different from the one originally used [8]. This leaves the third model in which endonucleolytic cleavage at a specific target site is the rate-limiting step. The insertions in the 3'-UTR that increase the half-life of PGK mRNA would then have to cause a (locally) altered folding of the transcript that impedes (or blocks) access of the endonuclease to the target site. This could either slow down the rate-limiting cleavage or target the mutant mRNA to an alternative, less efficient, degradation pathway [32]. Experiments in our laboratory (P Vreken, HA Rau6, unpublished observations) support the first of these two alternatives and indicate the primary endonucleolytic cleavage to occur closely upstream of the 3'-end of the PGK coding region. An important conclusion from the experiments described in this paper is that the metabolic stability

Turnover of yeast PGK mRNA of a yeast m R N A can be increased by local structural alterations that do not affect its translational efficiency. This could be of considerable practical interest since it would provide a further means for increasing the product yield of a (heterologous) gene in yeast cells. We have found that YEpR5-pG or -pU transformants contain significantly higher levels of PGK than cells transformed with the YEpR5-L parent plasmid although the copy numbers of all three plasmids are identical (P Vreken, V C H F de Regt, unpublished experiments). We are now analyzing the effect of these homopolymer insertions into the 3'UTR of other yeast m R N A s in order to determine whether their effect on decay rate is specific to PGK m R N A or whether these sequences can act as autonomous stabilizing elements.

Acknowledgments We thank Drs AW Hoekema and R Young for supplying us with plasmid YEpR5 and S cerevisiae strain Y260 respectively. This work was supported by the Programmacommissie Industri~le Biotechnologie (PclB) with financial aid from the Ministry of Economic Affairs.

References 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

RaghowR (1987) TIBS 12, 358-360 MarzluffWF, Pandey NB (1988) TIBS 13, 49-51 Brown AJP (1989) Yeast 5, 239-257 Ross J (1989) SciAm 260, 28-35 Yen TJ, Gay DA, Pachter JS, Cleveland DW (1988) Mol CellBiol 8, 1224-1235 Yen TJ, Machlin PS, Cleveland DW (1988) Nature (Lond) 334, 580-585 ShapiroRA, Herrick D, Manrow RE, Blinder D, Jacobson A (1988) Moi Cell Biol 8, 1957-1967 Herrick D, Parker R, Jacobson A (1990) Mol Cell Biol 10, 2269-2284 Shaw G, Kamen R (1986) Cei146, 659-667 Beutler B, Thompson P, Keyes J, Hagerty K, Crawford D (1988) Biochem Biophys Res Commun 152, 973-980 Casey JL, Koeller DM, Ramin VC, Klausner RD, Harford JB (1989) EMBO J 8, 3693-3699 Pandey NB, Marzluff WF (1987) Moi Cell Biol 7, 4557-4559 Xu HX, Johnson L, Grustein M (1990) Mol Cell Biol 10, 2687-2694 Brown AJP, Purvis IJ, Santiago TC, Bettany AJE, Loughlin L, Moore J (1988) Gene 72, 151-160 Parker R, Jacobson A (1990) Proc Natl Acad Sci USA 87, 2780-2784

737

16 Van den Heuvel JJ, Planta RJ, Rau6 HA (1990) Yeast 6, 473-482 17 Maniatis T, Fritsch EF, Sambrook J (1982) Molecular Cloning: A Laboratory Manual. Cold Spring Harbor Lab, Cold Spring Harbor, NY 18 Hanahan D (1983)JMolBiol 166, 557-580 19 Lopes TS, Kiootwijk J, Veenstra AE, Van der Aar PC, Van Heefikhuizen H, Rau6 HA, Planta RJ (1989) Gene 79, 199-206 20 Hoekema A, Kastelein RA, Vasser M, DeBoer H (1987) Mol Cell Biol 7, 2914--2924 21 Van den Heuvel JJ, Bergkamp RJM, Planta RJ, Rau6 HA (1989) Gene 79, 83-95 22 Guarente L, Yocum RR, Gifford P (1982) Proc Nati Acad Sci USA 79, 7410-7414 23 Nonet M, Scafe C, Sexton J, Young R (1987) 11401Cell Biol 7, 1602-1611 24 Klebe RJ, Harriss JV, Sharp ZD. Douglas MG (1983) Gene 25. 333-341 25 Zitomer RS, Montgomery DL, Nichols DL, Hall BD (1979) Proc Natl Acad Sci USA 76, 3627-3631 26 White BA, Bancroft FC (1982) J Biol Chem 257, 8569-8572 27 HitzernanRA, Hagie FE, HayflickJS, Chen CY, Seeburg PH, Derynck R (1982) Nucleic Acids Res 10, 7791-77808 28 Pelsy F, Lacroute F (1984) Curr Genet 8, 277-282 29 Losson R, Lacroute F (1979) Proc Natl Acad Sci USA 76, 5134-5137 30 Golzalbo D, Hohmann S (1990) Curr Genet 17, 77-79 31 Baim SB, Pietras DF, Eustice DC, ShermanF (1985) Mol Cell Biol 5, 1839 32 Purvis IJ, Bettany AJE, Loughlin L, Brown AJP (1987) Nucleic Acids Res 15, 7951-7962 33 Shyu AB, Greenberg ME, Belasco JG (1989) Genes Develop 3, 60-72 34 Santiago TC, Bettany AJE, Pur,~ ~ IJ, Brown AJP (1987) Nucleic Acids Res 6, 2417-2429 35 Jones TR, Cole MD (1987) Moi Ii Biol 7, 4513-4521 36 Petersen DD, Koch SR, Granne~~ DK (1989) Proc Natl Acad Sci USA 86, 7800-7804 37 Binder R, Hwang SPL, Ratnasabapathy R, Williams DL (1989) J Biol Chem 264, 16910--i6918 38 Tonouchi N, Miwa K, Karasuyama H, Matsui M (1989) Biochem Biophys Res Commun 163, 1056-1062 39 Kabnick KS, Housman DE (1988) Mol Cell Biol 8, 3244-3250 40 Wilson T, Treisman R (1988) Nature (Lond) 336, 396-399 41 Petersen RB, Lindquist S (1989) Cell Regulation 1, 135-149 42 Bemstein P, Peltz SW, Ross J (1989) Mol Cell Biol 9, 659-670 43 Higgins CF, McLaren RS, Newbury SF (1988) Gene 72, 3-14 44 Stem DB, Jones H, Gruissen W (1989) J Biol Chem 264, 18742-18750 45 Zimmermann SB, Cohen GH, Davies DR (1975)J Mol Bio192, ! 81-192