control proteins a very difficult one. Fortunately, this situation has changed substantially since the advent of in vitro transcri tion systems using bacteriophage RNA polymerases (3.4? With pure RNA in abundance, progress in understanding the control of maternal mRNA expre5sion has greatly accelerated and we shall review some salient features of this topic.
Summary Early development in many animals is programmed by maternally inherited messenger RNAs. Many of these mRNAs are translationally dormant in immature oocytes, but are recruited onto polysomes during meiotic maturation, fertilization, or early embryogenesis. In contrast, other mRNAs that are translated in oocytes are released from polysomes during these later stages of development. Recent studies have begun to define the cis and trans elements that regulate both translational repression and translational induction of maternal mKNA. The inhibition of translation of some mRNAs during early development is controlled by discrete sequences residing in the 3‘ and 5’ untranslated regions, respectively. The translation of other RNAs is due to polyadenylation which, at least in oocytes of the frog Xenopus laevis, is regulated by a U-rich cytoplasmic polyadenylation element (CPE). Although similar, the CPE sequences of various mRNAs are sufficiently different to be bound by different proteins. Two of these proteins and their interactions are described here. Introduction Gene expression in most cells is regulated predominantly at the level of transcription. At some early developmental stages in animal embryos, however, therc is no nuclear gene transcription and the transfer of information from gene to protein is controlled at the level of translation. This is particularly evident during oocytc maturation and fertilization in several marine invertebrates and in the frog Xenoptis luevz~.In these cases, maternal mRNAs are recruited onto polysomes from nontranslating pools and set several developmental programs including the early cell cycle divisons and germ layer formation (cf. a thorough review of the functions of maternal mRNA in ref. 1). Maternal mRNA would therefore seem ideal for the study of translational control and, indeed, since its discovery in 1963(’’, has been the subject of much investigation. However, the identification of czs acting sequences in message that regulate translation has been hampered by one’s inability to obtain pure specific mRNA in sufficient quantity to perform standard mutagenesis type experiments. Moreover, this lack of sequence information has made the hunt for specific translational
Translational Repression Translational repression may be thought of in two ways: the maintenance of mRNA in a translationally dormant state and the removal of mRNA from polysornes. Both of these mechanisms are found in early development. Oocytes? for example, contain an amount of poly(A) RNA that far exceeds the cell’s immediate protein synthesis requirements; indeed, in Xenopus, as much as 90 % of the total poly(A) RNA is nonpolysomal. While it is clear that no single mechanism is responsible for inhibiting the translation of all maternal mRNA (reviewed in refs. 5,6) i t has been demonstrated that repressor (masking proteins do bind RNA and inhibit their translation(7p9)1. Moreover, some of the repressor proteins bind specific messages(”): although the precise nucleotides they bind still are not known. Recent work in Spisulu, however, has begun to narrow down the sequences important for translational repression. In addition, the second mechanism is also observed: several translating RNAs are released from polysomes at specific times of development, particularly during oocyte maturation and early embryogenesis(’ ). In some ways, translational control during oocyte maturation and fertilization of the surf clam Spzsufu solidissirnu typifies the regulated expression of maternal mRNA in general. The now classical experiments of Rosenthal et a/.(”) demonstrated that fertilization in this species activated the synthesis of some proteins while deactivating the synthesis of others. That these changes were due to translational control was clearly shown by the priming of a rabbit reticulocyte in vitro translation system with RNA isolated from oocytes and fertilized eggs; the exact same spectrum of proteins was synthesized irrespective of the source of the mRNA. Thus, these observations indicate that masking proteins inhibit specific mRNAs from entering polysomes in oocytes. To examine the sequenccs that are necessar for this masking to occur, Standardt and colleagues(’3 devised an in vitro system in which specific oocyte mRNAs could be unmasked. They first combined a postmitochondrial supernatant from oocytes and eggs with a rabbit reticulocyte lysate and observed that the translational repression of specific mRNAs from oocytes was maintained in this heterologous system. They then demostrated that general ‘unmasking’ of oocyte mRNAs could be accomplished in vitro if the oocyte extract was subjected to gel filtration under high salt conditions prior to its addition to the reticulocyte lysate. They next reasoned that they might be able to
Y
unmask specific mRNAs in vitro with antisense RNAs or antisense oligodeoxynucleotides plus RNase H. That is, specific antisense RNAs would form double stranded molecules with their complements and prevent masking proteins from binding. Simjlarly. the oligodeoxynucleotides would form double stranded molecules, but in this case the RNase H would cleave the RNA in the duplexed region. The remainder of the mRNA would then be liberated from a sequence that would be the binding site for a repressor protein and be free to enter polysomes. The mRNAs chosen for these experiments were cyclin A and ribonucleotide reductase, both of which are known to be translated specifically after fertilization(’“”’). The data from an extensive series of experiments have shown that sequences within the 3’ untranslated region (UTR) of each RNA were required for masking. Moreover, some regions of homology between the two 3’UTRs suggest the existence of a common ‘unmasking’ box (i.e., prevention of a protein from binding this sequence unmasks the mRNA). The candidate sequences for the unmasking box include the pentanucleotide UUUUA, which is present in multiple copies in both 3’UTRs as well as the nonanucleotide GUGCAAUAA. At the present time, however, neither of these sequence elements has been shown experimentally to function as a binding site for a repressor protein. Deadenylation and Message Release from Polysomes Tt has been known for a number of years that mRNAs undergo polyadenylation and deadenylation during oocytc maturation and fertilization (reviewed in refs. 5,6). Furthermore, there was a strong correlation between polyadenylation and message recruitment and deadenylation and message release from polysomes. One message that is deadenylated and released from polyomes during Xenopus oocyte maturation is ribosomal protein L l mRNA. To examine the sequences that might be important for deadenylation and possib~6:ranslationaI inhibition, Hyman and Wormington synthesized L1 mRNA in vitro and injected it into oocytes that were subsequently induced to mature with progesterone. Like its endogenous counterpart, the injected RNA was translated in control (immature) oocytes but was deadenylated and released from polysomes following maturation. Those investigators then fused the 3’UTR of L1 mRNA to the S’UTR and coding sequence of Xenopus B-globin. which was considered a ‘nonspecific’ RNA devoid of any deadenylation signal. Upon injection into oocytes. this chimeric mRNA was deadenylated and releaced from polysomes following maturation. They concluded from this work that discrete sequences within the L1 3’UTR direct deadenylation and translational repression. However, more recent work suggests that this conclusion might be erroneous because other ‘nonspecific’ injected RNAs are also deadenylated during matu-
ration(17).Based on these results, deadenylation is now viewed as a ‘default’ mechanism. That is, RNAs are either programmed for adenylation by defined cis elements (see below) or, if they contain no such element, are automatically deadenylated(18).However, it does appear that the extent of deadenylation is influenced by the sequence of the message(16,18).Tn either case, it is still unclear whether deadenylation actually induces translational repression or is merely coincident with it. In contrast to the apparent lack of a cis elcment govcrning L1 mRNA translation. other ribosomal protein mRNAs (rpmRNAs) do contain such sequences. Consider, for examplc. Xenopus S19 mRNA, which is maternally inherited, stored during embryogenesis, and translated at the tailbud stage (2011 postfertilization). The cis elements regulating translation of this message were investigated by Mariottini and Amaldi(”)), who injected chimeric plasmid DNAs containing the S’UTR of S19 mRNA fused to chloramphenicol acetyltransferasc (CAT) coding sequences into fertilized eggs. Following embryogenesis up to the tailbud stage, they assessed CAT mRNA polysome profiles and determined CAT activity. The data showed that the S’UTR of S19 mRNA rcpressed CAT mRNA expresqion until the tailbud stage. Although it is not known where the putative ‘repressor’ sequence resides within the 256 base element used for the experiments described above, other rpmRNAs that display a translational profile similar to that of S19 mRNA have some common features in their ~’UTRS(~’), which might indicate binding sites for repressor proteins. These includc a run of 8-12 pyrimidines, the sequence GAGAAG, and a G + C region. The relative importance of these separate regions to translational control. however, has yet to be described. Induction of Translation by Polyadenylation Although some maternal mRNAs can be translated only after the removal of dominant repressor proteins. others require primary structural processing before they can enter polysomes. This was suggested initially by the experiments of Rosenthal ef al. (21) using Spisula material, and to a large extent is exemplified by the Xenopus maternal mRNA G10. Coincident with its translation during oocyte maturation, the poly(A) tail of this message is elongated from 90 residues to about 150 residues. To determine what sequences were required for polyadenylation and whether this modification was actually required for translation, McGrew et ~ l . ( ~synthesized *) radiolabeled G10 RNA in vitro and injected it into Xmopus oocytes. Following progesterone-induced maturation, the polyadenylation and translational status of the RNA was examined. Like the endogenous C 10 RNA, the injected transcript was polyadenylated and recruited for translation. Deletion mutations in the injected RNA were then made to
Table 1. Seqrrerrce of cytoplasmic poljadenvlation elenienis
RNA
CPE
Reference
GI0 €34 c-mos HPRT
UUUUUUAUAAAG UUUUUA AU UUUUAU UUUUAAAU UUUUAU
22,25 27
D7
27 27 26
With the exception of HPRT mRNA, all of the sequences shown above are froin Xenopus and were analyzed in injected Xenopiis oocytes following maturation. HPRT mRNA ia from mouse. but was also analyzed in injected Xenopxr oocytes following maturation; HPRT mRNA is known to be polyadenylatcd during mouse oocyte rnaturati~d~‘).
determine what sequences werc necessary for polyadenylation and translation. Two sequences in the G10 3’UTR were found to be necessary for these events: the nuclear pre-mRNA cleava e and polyadenylation hexanucleotide AAUAAA(’33”) and the sequence UUUUUUAU. Subsequent nucleotide substitution experiments have more completely defined this latter element as UUUUUUAUAAAG, which is now referred to as the cytoplasmic polyadenylation element (CPE)(25).Variants of this CPE are present in other RNAs that are polyadenylated during Xerzopus oocyte maturation are listed in Table 1. In the RNAs studied to date, the CPE and polyadenylation hexanucleotide are usually in close proximity with the CPE being the most 5’. However, polyadenylation occurs even when the hexanucleotide is upstream of the CPE(26)and when the two elements are immediately adjacent(27) or at a distance(25). Does polyadenylation actually stimulate G10 RNA translation? One might suspect that it does when it results in a tail of the ‘mature’ length of 150A residues. If this tail is sufficient, then an injected RNA already containing the mature size tail should be translated even in control (immature) oocytes. However, such an RNA is not translated in control oocytes, but is translated in mature oocytes when it is further adenylated with another 100 residues(22).This. then. could indicate that the process of polyadenylation and not poly(A) tail length per se determines translatability. To test this, McGrew et a1.(22)synthesized a modified G10 RNA that was 3’ terminally blocked by cordycepin (3’-dATP). When this RNA was injected, it was not translated in either control or mature oocytes. This result strongly suggests that rnaturation-specific translation of G10 RNA requires the active process of poly(A) elongation. One might ask whether the phenomenon of translational regulation by poly(A) elongation occurs in other systems as well as for other mRNAs in Xenopus. In one definitive study, Vassalli, Strickland and colleagues(28)have shown that the translation of tissue plasrninogen activator (tPA) mRNA during mouse oocyte maturation is accompanied by an elongation of its poly(A) tail. In this case, however, it is clear that
poly(A) tail length, and not the process of polyadenylation, regulates translation(2y~.This is similar to another RNA in Xeizopus, B4 RNA. whose translation during oocyte maturation is also regulated by poly(A) tail length(27). Thus. cytoplasmic polyadenylation appears to make some mRNAs competent for translation in two different ways: by the process of adenylate polymer forination and by the absolute number of adenylate residues. Regulation of Polyadenylation: CPE Binding Proteins Xenopus egg extracts that were developed initially for the study of cell cycle control(30)have proven to be remarkably efficient in adenylating exogenous RNA(z5.27). As a consequence, they could be used €or the identification of proteins that interact with the CPE. Using 32P-UTP labeled G10 RNA, McGrew and Richter@’) showed that an 82 kD protein UV photocrosslinked to the CPE in egg, but not oocyte, extracts. This suggested that the G10 CPE binding protein either had to be synthesized de novo during maturation, or was stored in oocytes but in an inactive (i.e.. unbound) form. An additional result that bears on this is that the activation of G10 RNA polyadenylation in vivo requires the synthesis of a protein within one hour following exposure of oocytes to progesterone. Thus, one could hypotheske that the mRNA encoding the 82 kD CPE binding protein enters polysomes soon after progesterone exposure and that the newly synthesized protein binds the G10 CPE and effects polyadenylation. Put in other terms. polyadenylation, which controls translation. would itself be under translational control. Similar crosslinking experiments with B4 RNA failed to identify an 82 kD CPE binding protein. In fact, initial experiments could identify the B4 CPE binding protein only by a gel mobility shift assay(27).Fortunately, recent improvements in the crosslinking procedure have shown that a 58 kD protein binds the B4 CPE (J. Paris, K. Swcnson, H. Piwnica-Worms, and J. D. Richter, in preparation). Unlike the 82 kD G10 CPE binding protein, this protein crosslinks in both oocyte and egg extracts, but has an apparent molecular size about 2 kD greater in the egg extract. This suggests posttranslational modification of the protein that could be important for its function. In fact, recent results have demonstrated this modification to be phosphorylation and indicated that it could be crucial for polyadenylation (see below). Regulation of Polyadenylation: Lessons from
Cell Cycle Control Oocyte maturation consists of an ordered cascade of molecular events that culniinates in germinal vesicle breakdown. One of these events is the activation of maturation promoting factor (MPF), a heterodimer of cyclin and ~ 3 4 “ ~ ‘the * , latter of which is a serine/threo-
nine kinase (cyclin and ~ 3 4 " ~ "have ' been reviewed recently in refs. 31-33). In addition to stimulating Xenopcis oocyte maturation, activated MPF and cyclin mRNA both stimulate polyad~nylation(~~). Because these agents act mainly (only?) through ~ 3 4 " ~ "kinase ' activity, one would suspect that activation of polyadenylation proceeds through serine/threonine phosphorylation. Indeed. because the B4 CPE binding protein is phosphorylated during maturation, it is tempting to think that it is this event that is key to polyadenylation. Supporting evidence comes from experiments in which baculovirus-expressed cyclin and ~ 3 4 " ~ ' 'each stimulated polyadenylation when added to oocyte extracts (J. Paris, K. Swenson, H. Piwnica-Worms, and J. D. Richter). At the present time, however, it is unclear whether ~ 3 4 ~ ~is' ' the physiological kinase that phosphorylates the B4 CPE binding protein during oocyte maturation. Moreover, the functional signifi-
5'
,
-
AAUAAA
A -50
3'
Oocyfe maturation
cance of B4 CPE binding protein phosphorylation remains to be determined. One model for the control of B4 RNA polyadenylation is shown in Fig. 1. In oocytes. the CPE and polyadenylation hexanucleotide are each bound by a protein. Because in v i m RNA competition studiedz7) suggest that these proteins exist in a complex, a bridge protein between the two is also represented. The CPE binding protein, when unphosphorylated, is inactive and there is no polyadenylation. The phosphorylation of this protein during oocyte maturation induces a conformational changc such that it enters into a second complex that includes a cytoplasmic poly(A) polymerase (PAP). Formation of this complex, in turn, initiates adenylate polymerization. It is important to note that although the figure shows a physical interaction of the RNA binding proteins following 58 kD protein phosphorylation, this is merely conjecture. In addition. the existence of a polyadenylation hexanucleotide binding protein is only surmised and the poly(A) polymeraye, while surely present. has not been identified in any manner other than by its activity. Finally, it should be noted that the poly(A) polymerase might have 'reading' as well as polymerization activity. This was suggested by experiments of McGrew and Richter(25) who showed that an RNA containing a poly(A) tail terminating in guanylate residues would not serve as a substrate for polyadenylation. Clearly, the isolation of all these factors is required before a clearer picture of maturation-specific polyadenylation can be drawn. Acknowledgements I thank Tim Hunt and Mike Wormington for communication of unpublished material and Barbara StebbinsBoaz and Jeannie Paris for comments on the manuscript. References
5'
CPE
Fig. 1. A proposal for phosphorylation-induccd polyadenylation during oocyte maturation. RNAs with a cytoplasmic polyadenylation element (CPE) identical to B4 RNA (see Table 1) would be bound by a 58kD protein as identified by Paris et ul. (in preparation) in Xenopus oocytes. The polyadenylation hexanucleotide would also be bound by a distinct protein and thesc two would interact via a hrid e protein, whose presence is suggested by Paris and Richter'"'. Following the induction of maturation, ~34'~'' kinase or another kinase with a similar substrate specificity, would phosphorylate the 58 kD CPE binding protein. This would induce a conformational change such that the 58 kD protein would enter into a complex that would include a cytoplasmic poly(A) polymerase (PAP). This latter protein is then activated and begins scanning in the 3' direction before it begins adenylate polymerization. The polyadenylation hexanucleotide binding protein is idcntified with a question mark since its existence is by inference only. It is also noted that B4 KNA contains a 50 residue poly(A) tail in oocytes.
1 DWORKIN, M. R . AND DWORKIN-R,\STL. E. (1990). Functions ol maternal
mRNA in early development. Mo1t.c. Reprod. Develop. 26. 261-297. 2 BRACHFT, J., FrcQ. A. AND T'EYCFR, R. (1963). Amino acid incorporation into protein&of nucleate and anucleolate fragment3 of sea urchin eggs: Effect of parthenogenetic activation. Exp. C ' d l Res. 32. 168-170. 3 KRIEO.P. A. AND MELTON,D. A . (1983). Fcinctional ineslenger RNAs are produced hy SP6 in virro transcription of cloned DNAs. Nfrcl. Acids Res. 12. 7057-7070, 4 MELTOY.11. A , . KKKEG, P. A , , KEBAGLCATI, M. R . . MA.ui.tns, T.. ZINX.K . . 4 i m GREON. M. R . (1924). Efficient in vilro synrheais of hiologically active R N R and RNA hyhridization probcs from plasmids contidining a bacteriophage SP6 promoter. Nucl. Acid3 Rw. 12, 7035-7056. 5 RICHTEX. J. D. (1987). Molccular mechanisms of translational control during the early development of Xcnopus laevis. I n Translational Regulurion of Gene Expresxion. (ed. J. Ilan), pp 111-139. Plenum Press, Kew York. 6 ROSENTFU., E. A N D WILT, F. H. (3987). Srlectivc messenger RNA translation in marine invertebrate oocytes, eggs. and zygotes. In Trnndarional Regulation o,f Gene Expwsion. (ed. .I. Ilan), pp 87-110. Plenum Presa, New York. 7 RIrHTEX. J. D. AND SMITH, L. D. (1984). The rcversihle inhibition of translation by Xenopus oocyte-specific proteins. Narurp 309, 378-3130. 8 KICK, D.. BARRETT.P.. CLlMMtNOS: A. AND SOMMERVIL~. J . (1987). Phosphorylation u l a 60 kDw polypeptide from Xenopus oocyter hlocks messenger RNA tramlation. A'ncl. Acids Re.
Summary Early development in many animals is programmed by maternally inherited messenger RNAs. Many of these mRNAs are translationally dormant in immature oocytes, but are recruited onto polysomes during meiotic maturation, fertilization, or early embryogenesis. In contrast, other mRNAs that are translated in oocytes are released from polysomes during these later stages of development. Recent studies have begun to define the cis and trans elements that regulate both translational repression and translational induction of maternal mKNA. The inhibition of translation of some mRNAs during early development is controlled by discrete sequences residing in the 3‘ and 5’ untranslated regions, respectively. The translation of other RNAs is due to polyadenylation which, at least in oocytes of the frog Xenopus laevis, is regulated by a U-rich cytoplasmic polyadenylation element (CPE). Although similar, the CPE sequences of various mRNAs are sufficiently different to be bound by different proteins. Two of these proteins and their interactions are described here. Introduction Gene expression in most cells is regulated predominantly at the level of transcription. At some early developmental stages in animal embryos, however, therc is no nuclear gene transcription and the transfer of information from gene to protein is controlled at the level of translation. This is particularly evident during oocytc maturation and fertilization in several marine invertebrates and in the frog Xenoptis luevz~.In these cases, maternal mRNAs are recruited onto polysomes from nontranslating pools and set several developmental programs including the early cell cycle divisons and germ layer formation (cf. a thorough review of the functions of maternal mRNA in ref. 1). Maternal mRNA would therefore seem ideal for the study of translational control and, indeed, since its discovery in 1963(’’, has been the subject of much investigation. However, the identification of czs acting sequences in message that regulate translation has been hampered by one’s inability to obtain pure specific mRNA in sufficient quantity to perform standard mutagenesis type experiments. Moreover, this lack of sequence information has made the hunt for specific translational
Translational Repression Translational repression may be thought of in two ways: the maintenance of mRNA in a translationally dormant state and the removal of mRNA from polysornes. Both of these mechanisms are found in early development. Oocytes? for example, contain an amount of poly(A) RNA that far exceeds the cell’s immediate protein synthesis requirements; indeed, in Xenopus, as much as 90 % of the total poly(A) RNA is nonpolysomal. While it is clear that no single mechanism is responsible for inhibiting the translation of all maternal mRNA (reviewed in refs. 5,6) i t has been demonstrated that repressor (masking proteins do bind RNA and inhibit their translation(7p9)1. Moreover, some of the repressor proteins bind specific messages(”): although the precise nucleotides they bind still are not known. Recent work in Spisulu, however, has begun to narrow down the sequences important for translational repression. In addition, the second mechanism is also observed: several translating RNAs are released from polysomes at specific times of development, particularly during oocyte maturation and early embryogenesis(’ ). In some ways, translational control during oocyte maturation and fertilization of the surf clam Spzsufu solidissirnu typifies the regulated expression of maternal mRNA in general. The now classical experiments of Rosenthal et a/.(”) demonstrated that fertilization in this species activated the synthesis of some proteins while deactivating the synthesis of others. That these changes were due to translational control was clearly shown by the priming of a rabbit reticulocyte in vitro translation system with RNA isolated from oocytes and fertilized eggs; the exact same spectrum of proteins was synthesized irrespective of the source of the mRNA. Thus, these observations indicate that masking proteins inhibit specific mRNAs from entering polysomes in oocytes. To examine the sequenccs that are necessar for this masking to occur, Standardt and colleagues(’3 devised an in vitro system in which specific oocyte mRNAs could be unmasked. They first combined a postmitochondrial supernatant from oocytes and eggs with a rabbit reticulocyte lysate and observed that the translational repression of specific mRNAs from oocytes was maintained in this heterologous system. They then demostrated that general ‘unmasking’ of oocyte mRNAs could be accomplished in vitro if the oocyte extract was subjected to gel filtration under high salt conditions prior to its addition to the reticulocyte lysate. They next reasoned that they might be able to
Y
unmask specific mRNAs in vitro with antisense RNAs or antisense oligodeoxynucleotides plus RNase H. That is, specific antisense RNAs would form double stranded molecules with their complements and prevent masking proteins from binding. Simjlarly. the oligodeoxynucleotides would form double stranded molecules, but in this case the RNase H would cleave the RNA in the duplexed region. The remainder of the mRNA would then be liberated from a sequence that would be the binding site for a repressor protein and be free to enter polysomes. The mRNAs chosen for these experiments were cyclin A and ribonucleotide reductase, both of which are known to be translated specifically after fertilization(’“”’). The data from an extensive series of experiments have shown that sequences within the 3’ untranslated region (UTR) of each RNA were required for masking. Moreover, some regions of homology between the two 3’UTRs suggest the existence of a common ‘unmasking’ box (i.e., prevention of a protein from binding this sequence unmasks the mRNA). The candidate sequences for the unmasking box include the pentanucleotide UUUUA, which is present in multiple copies in both 3’UTRs as well as the nonanucleotide GUGCAAUAA. At the present time, however, neither of these sequence elements has been shown experimentally to function as a binding site for a repressor protein. Deadenylation and Message Release from Polysomes Tt has been known for a number of years that mRNAs undergo polyadenylation and deadenylation during oocytc maturation and fertilization (reviewed in refs. 5,6). Furthermore, there was a strong correlation between polyadenylation and message recruitment and deadenylation and message release from polysomes. One message that is deadenylated and released from polyomes during Xenopus oocyte maturation is ribosomal protein L l mRNA. To examine the sequences that might be important for deadenylation and possib~6:ranslationaI inhibition, Hyman and Wormington synthesized L1 mRNA in vitro and injected it into oocytes that were subsequently induced to mature with progesterone. Like its endogenous counterpart, the injected RNA was translated in control (immature) oocytes but was deadenylated and released from polysomes following maturation. Those investigators then fused the 3’UTR of L1 mRNA to the S’UTR and coding sequence of Xenopus B-globin. which was considered a ‘nonspecific’ RNA devoid of any deadenylation signal. Upon injection into oocytes. this chimeric mRNA was deadenylated and releaced from polysomes following maturation. They concluded from this work that discrete sequences within the L1 3’UTR direct deadenylation and translational repression. However, more recent work suggests that this conclusion might be erroneous because other ‘nonspecific’ injected RNAs are also deadenylated during matu-
ration(17).Based on these results, deadenylation is now viewed as a ‘default’ mechanism. That is, RNAs are either programmed for adenylation by defined cis elements (see below) or, if they contain no such element, are automatically deadenylated(18).However, it does appear that the extent of deadenylation is influenced by the sequence of the message(16,18).Tn either case, it is still unclear whether deadenylation actually induces translational repression or is merely coincident with it. In contrast to the apparent lack of a cis elcment govcrning L1 mRNA translation. other ribosomal protein mRNAs (rpmRNAs) do contain such sequences. Consider, for examplc. Xenopus S19 mRNA, which is maternally inherited, stored during embryogenesis, and translated at the tailbud stage (2011 postfertilization). The cis elements regulating translation of this message were investigated by Mariottini and Amaldi(”)), who injected chimeric plasmid DNAs containing the S’UTR of S19 mRNA fused to chloramphenicol acetyltransferasc (CAT) coding sequences into fertilized eggs. Following embryogenesis up to the tailbud stage, they assessed CAT mRNA polysome profiles and determined CAT activity. The data showed that the S’UTR of S19 mRNA rcpressed CAT mRNA expresqion until the tailbud stage. Although it is not known where the putative ‘repressor’ sequence resides within the 256 base element used for the experiments described above, other rpmRNAs that display a translational profile similar to that of S19 mRNA have some common features in their ~’UTRS(~’), which might indicate binding sites for repressor proteins. These includc a run of 8-12 pyrimidines, the sequence GAGAAG, and a G + C region. The relative importance of these separate regions to translational control. however, has yet to be described. Induction of Translation by Polyadenylation Although some maternal mRNAs can be translated only after the removal of dominant repressor proteins. others require primary structural processing before they can enter polysomes. This was suggested initially by the experiments of Rosenthal ef al. (21) using Spisula material, and to a large extent is exemplified by the Xenopus maternal mRNA G10. Coincident with its translation during oocyte maturation, the poly(A) tail of this message is elongated from 90 residues to about 150 residues. To determine what sequences were required for polyadenylation and whether this modification was actually required for translation, McGrew et ~ l . ( ~synthesized *) radiolabeled G10 RNA in vitro and injected it into Xmopus oocytes. Following progesterone-induced maturation, the polyadenylation and translational status of the RNA was examined. Like the endogenous C 10 RNA, the injected transcript was polyadenylated and recruited for translation. Deletion mutations in the injected RNA were then made to
Table 1. Seqrrerrce of cytoplasmic poljadenvlation elenienis
RNA
CPE
Reference
GI0 €34 c-mos HPRT
UUUUUUAUAAAG UUUUUA AU UUUUAU UUUUAAAU UUUUAU
22,25 27
D7
27 27 26
With the exception of HPRT mRNA, all of the sequences shown above are froin Xenopus and were analyzed in injected Xenopiis oocytes following maturation. HPRT mRNA ia from mouse. but was also analyzed in injected Xenopxr oocytes following maturation; HPRT mRNA is known to be polyadenylatcd during mouse oocyte rnaturati~d~‘).
determine what sequences werc necessary for polyadenylation and translation. Two sequences in the G10 3’UTR were found to be necessary for these events: the nuclear pre-mRNA cleava e and polyadenylation hexanucleotide AAUAAA(’33”) and the sequence UUUUUUAU. Subsequent nucleotide substitution experiments have more completely defined this latter element as UUUUUUAUAAAG, which is now referred to as the cytoplasmic polyadenylation element (CPE)(25).Variants of this CPE are present in other RNAs that are polyadenylated during Xerzopus oocyte maturation are listed in Table 1. In the RNAs studied to date, the CPE and polyadenylation hexanucleotide are usually in close proximity with the CPE being the most 5’. However, polyadenylation occurs even when the hexanucleotide is upstream of the CPE(26)and when the two elements are immediately adjacent(27) or at a distance(25). Does polyadenylation actually stimulate G10 RNA translation? One might suspect that it does when it results in a tail of the ‘mature’ length of 150A residues. If this tail is sufficient, then an injected RNA already containing the mature size tail should be translated even in control (immature) oocytes. However, such an RNA is not translated in control oocytes, but is translated in mature oocytes when it is further adenylated with another 100 residues(22).This. then. could indicate that the process of polyadenylation and not poly(A) tail length per se determines translatability. To test this, McGrew et a1.(22)synthesized a modified G10 RNA that was 3’ terminally blocked by cordycepin (3’-dATP). When this RNA was injected, it was not translated in either control or mature oocytes. This result strongly suggests that rnaturation-specific translation of G10 RNA requires the active process of poly(A) elongation. One might ask whether the phenomenon of translational regulation by poly(A) elongation occurs in other systems as well as for other mRNAs in Xenopus. In one definitive study, Vassalli, Strickland and colleagues(28)have shown that the translation of tissue plasrninogen activator (tPA) mRNA during mouse oocyte maturation is accompanied by an elongation of its poly(A) tail. In this case, however, it is clear that
poly(A) tail length, and not the process of polyadenylation, regulates translation(2y~.This is similar to another RNA in Xeizopus, B4 RNA. whose translation during oocyte maturation is also regulated by poly(A) tail length(27). Thus. cytoplasmic polyadenylation appears to make some mRNAs competent for translation in two different ways: by the process of adenylate polymer forination and by the absolute number of adenylate residues. Regulation of Polyadenylation: CPE Binding Proteins Xenopus egg extracts that were developed initially for the study of cell cycle control(30)have proven to be remarkably efficient in adenylating exogenous RNA(z5.27). As a consequence, they could be used €or the identification of proteins that interact with the CPE. Using 32P-UTP labeled G10 RNA, McGrew and Richter@’) showed that an 82 kD protein UV photocrosslinked to the CPE in egg, but not oocyte, extracts. This suggested that the G10 CPE binding protein either had to be synthesized de novo during maturation, or was stored in oocytes but in an inactive (i.e.. unbound) form. An additional result that bears on this is that the activation of G10 RNA polyadenylation in vivo requires the synthesis of a protein within one hour following exposure of oocytes to progesterone. Thus, one could hypotheske that the mRNA encoding the 82 kD CPE binding protein enters polysomes soon after progesterone exposure and that the newly synthesized protein binds the G10 CPE and effects polyadenylation. Put in other terms. polyadenylation, which controls translation. would itself be under translational control. Similar crosslinking experiments with B4 RNA failed to identify an 82 kD CPE binding protein. In fact, initial experiments could identify the B4 CPE binding protein only by a gel mobility shift assay(27).Fortunately, recent improvements in the crosslinking procedure have shown that a 58 kD protein binds the B4 CPE (J. Paris, K. Swcnson, H. Piwnica-Worms, and J. D. Richter, in preparation). Unlike the 82 kD G10 CPE binding protein, this protein crosslinks in both oocyte and egg extracts, but has an apparent molecular size about 2 kD greater in the egg extract. This suggests posttranslational modification of the protein that could be important for its function. In fact, recent results have demonstrated this modification to be phosphorylation and indicated that it could be crucial for polyadenylation (see below). Regulation of Polyadenylation: Lessons from
Cell Cycle Control Oocyte maturation consists of an ordered cascade of molecular events that culniinates in germinal vesicle breakdown. One of these events is the activation of maturation promoting factor (MPF), a heterodimer of cyclin and ~ 3 4 “ ~ ‘the * , latter of which is a serine/threo-
nine kinase (cyclin and ~ 3 4 " ~ "have ' been reviewed recently in refs. 31-33). In addition to stimulating Xenopcis oocyte maturation, activated MPF and cyclin mRNA both stimulate polyad~nylation(~~). Because these agents act mainly (only?) through ~ 3 4 " ~ "kinase ' activity, one would suspect that activation of polyadenylation proceeds through serine/threonine phosphorylation. Indeed. because the B4 CPE binding protein is phosphorylated during maturation, it is tempting to think that it is this event that is key to polyadenylation. Supporting evidence comes from experiments in which baculovirus-expressed cyclin and ~ 3 4 " ~ ' 'each stimulated polyadenylation when added to oocyte extracts (J. Paris, K. Swenson, H. Piwnica-Worms, and J. D. Richter). At the present time, however, it is unclear whether ~ 3 4 ~ ~is' ' the physiological kinase that phosphorylates the B4 CPE binding protein during oocyte maturation. Moreover, the functional signifi-
5'
,
-
AAUAAA
A -50
3'
Oocyfe maturation
cance of B4 CPE binding protein phosphorylation remains to be determined. One model for the control of B4 RNA polyadenylation is shown in Fig. 1. In oocytes. the CPE and polyadenylation hexanucleotide are each bound by a protein. Because in v i m RNA competition studiedz7) suggest that these proteins exist in a complex, a bridge protein between the two is also represented. The CPE binding protein, when unphosphorylated, is inactive and there is no polyadenylation. The phosphorylation of this protein during oocyte maturation induces a conformational changc such that it enters into a second complex that includes a cytoplasmic poly(A) polymerase (PAP). Formation of this complex, in turn, initiates adenylate polymerization. It is important to note that although the figure shows a physical interaction of the RNA binding proteins following 58 kD protein phosphorylation, this is merely conjecture. In addition. the existence of a polyadenylation hexanucleotide binding protein is only surmised and the poly(A) polymeraye, while surely present. has not been identified in any manner other than by its activity. Finally, it should be noted that the poly(A) polymerase might have 'reading' as well as polymerization activity. This was suggested by experiments of McGrew and Richter(25) who showed that an RNA containing a poly(A) tail terminating in guanylate residues would not serve as a substrate for polyadenylation. Clearly, the isolation of all these factors is required before a clearer picture of maturation-specific polyadenylation can be drawn. Acknowledgements I thank Tim Hunt and Mike Wormington for communication of unpublished material and Barbara StebbinsBoaz and Jeannie Paris for comments on the manuscript. References
5'
CPE
Fig. 1. A proposal for phosphorylation-induccd polyadenylation during oocyte maturation. RNAs with a cytoplasmic polyadenylation element (CPE) identical to B4 RNA (see Table 1) would be bound by a 58kD protein as identified by Paris et ul. (in preparation) in Xenopus oocytes. The polyadenylation hexanucleotide would also be bound by a distinct protein and thesc two would interact via a hrid e protein, whose presence is suggested by Paris and Richter'"'. Following the induction of maturation, ~34'~'' kinase or another kinase with a similar substrate specificity, would phosphorylate the 58 kD CPE binding protein. This would induce a conformational change such that the 58 kD protein would enter into a complex that would include a cytoplasmic poly(A) polymerase (PAP). This latter protein is then activated and begins scanning in the 3' direction before it begins adenylate polymerization. The polyadenylation hexanucleotide binding protein is idcntified with a question mark since its existence is by inference only. It is also noted that B4 KNA contains a 50 residue poly(A) tail in oocytes.
1 DWORKIN, M. R . AND DWORKIN-R,\STL. E. (1990). Functions ol maternal
mRNA in early development. Mo1t.c. Reprod. Develop. 26. 261-297. 2 BRACHFT, J., FrcQ. A. AND T'EYCFR, R. (1963). Amino acid incorporation into protein&of nucleate and anucleolate fragment3 of sea urchin eggs: Effect of parthenogenetic activation. Exp. C ' d l Res. 32. 168-170. 3 KRIEO.P. A. AND MELTON,D. A . (1983). Fcinctional ineslenger RNAs are produced hy SP6 in virro transcription of cloned DNAs. Nfrcl. Acids Res. 12. 7057-7070, 4 MELTOY.11. A , . KKKEG, P. A , , KEBAGLCATI, M. R . . MA.ui.tns, T.. ZINX.K . . 4 i m GREON. M. R . (1924). Efficient in vilro synrheais of hiologically active R N R and RNA hyhridization probcs from plasmids contidining a bacteriophage SP6 promoter. Nucl. Acid3 Rw. 12, 7035-7056. 5 RICHTEX. J. D. (1987). Molccular mechanisms of translational control during the early development of Xcnopus laevis. I n Translational Regulurion of Gene Expresxion. (ed. J. Ilan), pp 111-139. Plenum Press, Kew York. 6 ROSENTFU., E. A N D WILT, F. H. (3987). Srlectivc messenger RNA translation in marine invertebrate oocytes, eggs. and zygotes. In Trnndarional Regulation o,f Gene Expwsion. (ed. .I. Ilan), pp 87-110. Plenum Presa, New York. 7 RIrHTEX. J. D. AND SMITH, L. D. (1984). The rcversihle inhibition of translation by Xenopus oocyte-specific proteins. Narurp 309, 378-3130. 8 KICK, D.. BARRETT.P.. CLlMMtNOS: A. AND SOMMERVIL~. J . (1987). Phosphorylation u l a 60 kDw polypeptide from Xenopus oocyter hlocks messenger RNA tramlation. A'ncl. Acids Re.
