- Email: [email protected]
SH2 domains: elements that control protein interactions during signal transduction
TIBS 1 6 - DECEMBER 1991
JOURNALCLUB Growth factors such as platelet-derived growth factor (PDGF), epidermal growth factor (EGF), fibroblast growth factor (FGF) and colony stimulating factor-I (CSF-I) stimulate cell proliferation by binding to receptors with intrinsic protein tyrosine kinase activities. Ligand binding activates the receptor kinases by a process which often involves receptor dimerization (for review see Ref. I). Kinase activation leads to autophosphorylation of the receptors and to the phosphorylation of certain cytoplasmic substrates, some of which are likely to be involved in the transduction of the mitogenic signal from the cell membrane to the nucleus. The identification of substrates for tyrosine kinase receptors has been greatly aided by the observation that many substrates associate with activated receptors. Recent findings have demonstrated that this association occurs between specific autophosphorylation sites in the tyrosine kinase receptors and specific conserved domains in the substrates. These domains are known as SH2 or Src homology region 2 since they were first identified in non-receptor tyrosine kinases of the v-Src family, and in v-Fps and v-Abl (Ref. 2; for review see Ref. 3). Recent data discussed here support the notion that SH2 domains function as recognition motifs that link components of the mitogenic pathway to activated growth factor receptors in a process which is regulated by phosphorylation/dephosphorylation of specific tyrosine residues in the receptors. SH2 domains consist of about 100 amino acids with about five regions that are highly conserved between proteins containing this domain. One or more SH2 domains have been found in several proteins (for examples see Fig. 1). Many of the proteins that contain SH2 domains also tontain another conserved motif (SH3 domain) that has been implicated in interactions with the cytoskeleton 3. Some of the proteins that contain SH2 domains can be linked to early effects of growth factors in responder cells. Phospholipase C-7 (PLC-7) and the regulatory subunit (p85) of phosphatidylinositol 3' kinase (PI-3'-K) have
45O
SH2 domains: elements that control protein interactions
Ras molecules, thus providing a link to signalling pathways controlled by GTPbinding proteins TM. Src, and other members of the Src family, are themselves tyrosine kinases and may be part of the cascade of phosphorylation events which characterizes growthfactor-stimulated cell division TM. However, none of these substrates have yet been shown to be essential for growth stimulation, and their exact roles in the mitogenic pathway remain to be determined. Each activated growth factor receptor complex contains several autophosphorylation sites which may interact with SH2-containing proteins. been found to associate with growth Moreover, phosphorylated tyrosine factor receptors including the activated residues within the SH2 proteins may PDGF receptor and to become phospho- provide additional interaction sites for rylated on tyrosine residues (see Refs proteins with SH2-domains. Thus, it is 1,3 and 4 and references therein). PLC-7 possible that large complexes conand PI-3'-Kare involved in the turnover taining various enzymatic activities of phospholipids, including the gener- (denoted signal transduction particles') ation of second messengers of potential assemble around activated growth importance in mitogenic stimulation. factor receptors. GTPase activating protein (GAP) interIt is possible that the SH2 domains of acts with tyrosine kinases as well as the cytoplasmic protein tyrosiae kinases,
during signal transduction
Ik A
A~I
n
m
Src
"1
PLC-Y
p85
~
I"
~
.
....
GAP
"
i
• ,.A...,, i
-
. . . . . .
v-crk
=
Vav
-.~
_e =
.
J
~
, ,
i
=!
"1
"
PTPIC
_
=--,,-=~-~
•
--
Fps
Abl
.
I
//
I
~• " " ' = " I
I
Rgure 1 Schematic illustration of the proteins containing SH2 domains that are discussed in the text. The black boxes with an imprinted 'P' symbolize SH2 domains, and the other black boxes SH3 domains. Different types of effector domains are symbolized by open boxes.
© 1991,ElsevierSciencePublishers, (UK) 0376-5067/91/$02.00
TIBS 1 6 - D E C E M B E R 1 9 9 1
such as Src, Fps and Abl, are also involved in control of intrinsic kinases activities. For example, Src contains a carboxy-terminal tyrosine residue (Tyr527), which can be phosphorylated by exogenous kinase(s) leading to a suppression of the kinase activity. The amino-terminal SH2 domain of Src might interact with the phosphorylated Tyr527, thereby blocking the active site of the kinase (for review see Ref. 4). Although other mechanisms have not been excluded, there are several observations that support this possibility; mutations in the SH2 domain of Src can enhance its kinase activity5,e, the SH2 domains of repressed and derepressed Src differ in protease susceptibility, and the amino-terminal part of Src is required for the phosphorylated Tyr527 to exert its repressive effect7. Derepression of the kinase activity of Src could thus occur by dephosphopjlation ol Tyr527 and/or interaction of the SH2 domain of Src with other molecules. Interestingly, there is one example of an oncoprotein (v-Crk) that contains an SH2 domain but no catalytic domain (Fig. I) which forms stable complexes with several tyrosine-phosphorylated proteins s. v-Crk may bind to phosphorylated sites involved in negative regulation of tyrosine kinases, thereby derepressing their activities. However, this requires further investigation. A recent report extends the notion that interactions between SH2 domains and phosphorylated sites can result in the deregulation of growth control pathways; the SH2 domain of Abl was shown to interact with phosphorylated sites in c-Bcr as well as in the Bcr-Abl fusion protein that is implicated in the pathogenesis of human leukemia9. It is thus possible that the Bcr sequences in the fusion protein interfere with the negative regulation of Abi. Notably, the interaction occurred with phosphorylated serine and threonine residues, suggesting that the interaction of the
SH2 domains is not restricted to phosphorylated tyrosine residues. Protein phosphorylation is balanced by dephosphorylation. Recent data indicate the presence of a large family of phosphotyrosine protein phosphatases. Interestingly, one member in this growing family contains an SH2 domain (PTPIC; Ref. I0). Since the specificity of the phosphatase activity of this enzyme is unknown, the functional implications of this finding can not yet be evaluated. However, the
enzyme might dephosphorylate sites involved in negative regulation of kinases (e.g. Tyr527 of Src) thus promoting tyrosine phosphorylation; alternatively, the phosphatase could serve a feedback function and balance the phosphorylation, e.g. by the activated growth factor receptor in the signal transduction particle. Another interesting category of factors with SH2 domains in exemplified by the oncoprotein Vav, which contains sequence motifs commonly found in transcription factors, like helix-loophelix, leucine zipper and zinc fingers (Ref. II; J. Schlessinger, pers. commun.). Vav interacts with and is phosphorylated on tyrosine residues by the EGF receptor (£ Schlessinger, pers. commun.). It is possible that phosphorylated Vav can be released from the receptor complex, and exert direct transcriptional control. The key to a deeper understanding o,~ the ro|e of SH2 domains in signal transduction lies undoubtly in the elucidation of the specificities in the interactions between various SH2 domains and individual phosphorylation sites. There is evidence accumulating for some specificity; growth factor receptors generally associate with some but not all of the proteins with SH2 domains that are listed in Fig. I. The PDGF receptor binds several of the substrates; an important question is thus whether the substrates bind to the same or different regions of the receptor. Analysis of bacterially expressed SH2 domains revealed some competition between PLC-y and GAP for binding to the receptor ~2. In another study, analysis of the abilities of various peptides to interfere with the binding of substrates to the PDGF ~-receptor demonstrated that PI-3'-K interacts with two regions in the sequence that divides the kinase domain into two parts (the kinase insert) (Ref. 13; L.T. Williams, pers. commun.). Only tyrosine-phosphorylated peptides showed activity, which is consistent with the idea that PI-3'-K interacts with autophosphorylated sites in the receptor. One of the two implicated tyrosine residues was previously identified as an autophosphorylation site in the PDGF ~receptor ~4and it seems likely that the other site can also be autophosphorylated. A necessary, but not suffid.ent, prerequisite for PI-3'-Kbinding seems to be a methionine residue in the third position downstream of a phosphorylated tyrosine residue. By the same strategy, it was concluded that GAP
interacts with another phosphorylated t~osine residue, located further downstream in the kinase insert (L. T. Williams, pers. commun.). Autophosphorylated tyrosine residues in the carboxy-terminal tails of the EGF receptor ~5 and the FGF receptor z6 have been found to be important for the interaction with PLC-¥. It is likely that many more proteins containing SH2-domains exist. The autophosphorylated region of the EGF receptor shows great promise as a probe to identify cDNAs coding for proteins with SH2 domains in ~,gtl I expression libraries ~7. The use of autophosphorylated regions of other receptors as probes in similar approaches could help to identify components of their respective signalling pathways. In conclusion, SH2 domains appear to play a fascinating role in interactions between activated growth factor receptors and cytoplasmic components in the mitogenic pathway. Data have also been presented that suggest a role for SH2 domains in the control of the activities of cytoplasmic tyrosine kinases. Although other molecular mechanisms for the interaction between, and control of, components in the signalling pathway may exist, the studies on SH2 domains have provided insight both into a possible molecular mechanisn-, for the interactions between components involved, and also a method for the identification of additional components in the mitogenic pathway. Methods have also established which will make it possib;e to eetermine the structural requirements around phosphorylated residues for binding of different SH2 domains. The stage is set for rapid progress in this research field; it is anticipated that the results will significantly extend our understanding of the signal transduction via tyrosine kinase receptors.
Acknowledgements I thank L. Claesson-Welsh, K. Miyazono, T. Pawson, J. Schlessinger, B. Westermark and L. T. Williams for valuable comments.
~eferences
1 UIIrich,A. and Schlessinger,J. (1990) Cell 61, 203-212 2 Sedowski,I., Stone,J. C. and Pawson,T. (1986) Mol. Cell. Biol. 6, 4396-4408 3 Koch,C. A. et al. (1991)Science 252, 668-674 4 Cantley,L. C. et al. (1991)Cell 64, 281-302 5 Hirai, H. andVarmus,H. E. (1990) Mol. Cell. Biol. 10, 1307-1318 60'Brian, M. C., Fukui,H. andHanafusa,H. (1990) MoL Cell. Biol. 10, 2855-2862
451
TIBS 1 6 - D E C E M B E R 1 9 9 1 7 MacAuley,A. and Cooper,J. A, (1989) MoL Cell. BioL 9, 2648-2656 8 Matsuda,M., Mayer,B. J., Fukui,Y. and Hanafusa,H. (1990) Science 248,1537-1539 9 Pendei~ast,A. M. et al. (1991)Cell 66,161-171 10 Shen,S. H., Basti~=n,L., Posner,P. I. and Chr6tien,P. (1991) Nature 352, 736-739 11 Coppola,J. et aL (199i) Cell Growth Differ. 2, 95-105
17 Skolnik,E. Y. et al. (1991) Cell 65, 83-90
12 Andersson,D. et al. (1990) Science 250,
979-982 13 Escebedo,J. A. et al. (1991) MoL Cell. Biol. 11, 1125--1132 14 Kazalauskas,A. and Cooper,J. A. (1989) Cell 58,1121-1133 15 Margolis,B. et al. (1990) EMBOJ. 9, 4375-.4390 16 Mehammadi,M. et al. (1991) Mol. Cell. Biol. 11, 5068-5078
CARL-HENRIK HELDIN Ludwig Institute for Cancer Research, Box 595, Biomedical Center, S-751 24 Uppsala, Sweden.
ASF/SF2 Most mRNA precursors (pre-mRNAs) in higher eukaryotes contain introns which must be precisely removed to generate functional mRNA products. Many laboratories have studied the mechanism of pre-RNA splicing in vitro using nuclear extracts prepared from human HeLa cells. From these studies it has emerged that a ribonucleoprotein site signals. Two recent papers reportcomplex, a spliceosome, is assembled ing the cloning of a gene encoding a on pre-mRNA in an apparently ordered mammalian factor that can influence fashion. Five of the abundant nuclear U the choice between competing 5' splice snRNPs (u,'idine-rich, small nuclear sites represent an important advance ribonucleoprotein particles) are the towards this goal4,5. major subunits of spliceosomes I which A well-documented example of alternaalso contain non-snRNP protein factors. tive splice site choice occurs in SV40Spliceosomes containing equivalent infected cells. In this case, either of two snRNPs are assembled in a similar path- alternative 5' splice sites can be joined way in yeast splicing extracts 2, under- to a common 3' splice site (Fig. I). Use lining the highly conserved nature of of the first 5' splice site produces mRNA the splicing mechanism and the funda- encoding large T antigen, while use of mental role that snRNPs play in it. the second produces mRNA encoding While the snRNPs represent basic com- small t antigen. The amino-terminal ponents of the splicing machinery, how- sequences of these two proteins are ever, there is now also considerable identical but the remainder of their interest in characterizing additional sequences are different due to a shift in trans-acting factors which might coop- the reading frame caused by the alternaerate with snRNPs and possibly regu- tive splicing events. This illustrates the late splice site choice. flexibility in coding potential that alternaMany pre-mRNAs contain multiple tive splicing can generate.
a splice site "selector
introns, allowing several different mRNAs to be produced by alternative splicing of a single primary transcript. Alternative splicing has been shown to occur for both cellular and viral RNAs and can play an important regulatory role in controlling gene expression. For example, recent studies in D r o s o p h i l a have shown that sex determination is principally regulated through modulation of splice site selection, governed by products of the sex lethal, transformer and transformer-2 genes 3.
Progress in understanding splice site utilization in mammalian cells clearly requires the identification and characterization of bans-acting factors that determine the use of alternative splice
452
Interestingly, the ratio of small t to large T mRNA produced from the common pre-mRNA transcript was observed to be greatly increased in an adenovirus-transformed human embryonic kidney cell line (293 cell@. This suggested that these cells either contained a factor that stimulated use of the proximal, small t 5' splice site or lacked a repressor which normally reduced selection of the small t splice site. The increased level of small t splicing was also seen in vitro, using nuclear extracts of 293 cells. This in vitro system provided Ge and Manley with a convenient assay for purification of a factor (Alternative Splicing Factor or ASF) responsible for the increase in small t mRNA~. Highly purified ASF could also stimulate small t splicing when added to HeLa splicing extracts. These results showed that the increase in small t splicing in 293 cells was indeed due to an activator, rather than the absence of a repressor. Ge and Manley also showed that when small t antigen splicing was activated in vitro by ASF there was a concomitant decrease in production of large T mRNA. This indicated that ASF was
T
m
~
m
~
Figure1 Cartoon (not to scale) showing the altemative 5' splice sites in SV40 pre-mRNA which are used, in conjunction with a common 3' splice site, to generate distinct mRNAs encoding large T and small t antigens. High concentrations of the splicing factor ASF-1/SF2 promotes selection of the small t 5' splice site over the large T one.
© 1991,ElsevierSciencePublishers,(UK) 0376--5067/91/$02.00
JOURNALCLUB Growth factors such as platelet-derived growth factor (PDGF), epidermal growth factor (EGF), fibroblast growth factor (FGF) and colony stimulating factor-I (CSF-I) stimulate cell proliferation by binding to receptors with intrinsic protein tyrosine kinase activities. Ligand binding activates the receptor kinases by a process which often involves receptor dimerization (for review see Ref. I). Kinase activation leads to autophosphorylation of the receptors and to the phosphorylation of certain cytoplasmic substrates, some of which are likely to be involved in the transduction of the mitogenic signal from the cell membrane to the nucleus. The identification of substrates for tyrosine kinase receptors has been greatly aided by the observation that many substrates associate with activated receptors. Recent findings have demonstrated that this association occurs between specific autophosphorylation sites in the tyrosine kinase receptors and specific conserved domains in the substrates. These domains are known as SH2 or Src homology region 2 since they were first identified in non-receptor tyrosine kinases of the v-Src family, and in v-Fps and v-Abl (Ref. 2; for review see Ref. 3). Recent data discussed here support the notion that SH2 domains function as recognition motifs that link components of the mitogenic pathway to activated growth factor receptors in a process which is regulated by phosphorylation/dephosphorylation of specific tyrosine residues in the receptors. SH2 domains consist of about 100 amino acids with about five regions that are highly conserved between proteins containing this domain. One or more SH2 domains have been found in several proteins (for examples see Fig. 1). Many of the proteins that contain SH2 domains also tontain another conserved motif (SH3 domain) that has been implicated in interactions with the cytoskeleton 3. Some of the proteins that contain SH2 domains can be linked to early effects of growth factors in responder cells. Phospholipase C-7 (PLC-7) and the regulatory subunit (p85) of phosphatidylinositol 3' kinase (PI-3'-K) have
45O
SH2 domains: elements that control protein interactions
Ras molecules, thus providing a link to signalling pathways controlled by GTPbinding proteins TM. Src, and other members of the Src family, are themselves tyrosine kinases and may be part of the cascade of phosphorylation events which characterizes growthfactor-stimulated cell division TM. However, none of these substrates have yet been shown to be essential for growth stimulation, and their exact roles in the mitogenic pathway remain to be determined. Each activated growth factor receptor complex contains several autophosphorylation sites which may interact with SH2-containing proteins. been found to associate with growth Moreover, phosphorylated tyrosine factor receptors including the activated residues within the SH2 proteins may PDGF receptor and to become phospho- provide additional interaction sites for rylated on tyrosine residues (see Refs proteins with SH2-domains. Thus, it is 1,3 and 4 and references therein). PLC-7 possible that large complexes conand PI-3'-Kare involved in the turnover taining various enzymatic activities of phospholipids, including the gener- (denoted signal transduction particles') ation of second messengers of potential assemble around activated growth importance in mitogenic stimulation. factor receptors. GTPase activating protein (GAP) interIt is possible that the SH2 domains of acts with tyrosine kinases as well as the cytoplasmic protein tyrosiae kinases,
during signal transduction
Ik A
A~I
n
m
Src
"1
PLC-Y
p85
~
I"
~
.
....
GAP
"
i
• ,.A...,, i
-
. . . . . .
v-crk
=
Vav
-.~
_e =
.
J
~
, ,
i
=!
"1
"
PTPIC
_
=--,,-=~-~
•
--
Fps
Abl
.
I
//
I
~• " " ' = " I
I
Rgure 1 Schematic illustration of the proteins containing SH2 domains that are discussed in the text. The black boxes with an imprinted 'P' symbolize SH2 domains, and the other black boxes SH3 domains. Different types of effector domains are symbolized by open boxes.
© 1991,ElsevierSciencePublishers, (UK) 0376-5067/91/$02.00
TIBS 1 6 - D E C E M B E R 1 9 9 1
such as Src, Fps and Abl, are also involved in control of intrinsic kinases activities. For example, Src contains a carboxy-terminal tyrosine residue (Tyr527), which can be phosphorylated by exogenous kinase(s) leading to a suppression of the kinase activity. The amino-terminal SH2 domain of Src might interact with the phosphorylated Tyr527, thereby blocking the active site of the kinase (for review see Ref. 4). Although other mechanisms have not been excluded, there are several observations that support this possibility; mutations in the SH2 domain of Src can enhance its kinase activity5,e, the SH2 domains of repressed and derepressed Src differ in protease susceptibility, and the amino-terminal part of Src is required for the phosphorylated Tyr527 to exert its repressive effect7. Derepression of the kinase activity of Src could thus occur by dephosphopjlation ol Tyr527 and/or interaction of the SH2 domain of Src with other molecules. Interestingly, there is one example of an oncoprotein (v-Crk) that contains an SH2 domain but no catalytic domain (Fig. I) which forms stable complexes with several tyrosine-phosphorylated proteins s. v-Crk may bind to phosphorylated sites involved in negative regulation of tyrosine kinases, thereby derepressing their activities. However, this requires further investigation. A recent report extends the notion that interactions between SH2 domains and phosphorylated sites can result in the deregulation of growth control pathways; the SH2 domain of Abl was shown to interact with phosphorylated sites in c-Bcr as well as in the Bcr-Abl fusion protein that is implicated in the pathogenesis of human leukemia9. It is thus possible that the Bcr sequences in the fusion protein interfere with the negative regulation of Abi. Notably, the interaction occurred with phosphorylated serine and threonine residues, suggesting that the interaction of the
SH2 domains is not restricted to phosphorylated tyrosine residues. Protein phosphorylation is balanced by dephosphorylation. Recent data indicate the presence of a large family of phosphotyrosine protein phosphatases. Interestingly, one member in this growing family contains an SH2 domain (PTPIC; Ref. I0). Since the specificity of the phosphatase activity of this enzyme is unknown, the functional implications of this finding can not yet be evaluated. However, the
enzyme might dephosphorylate sites involved in negative regulation of kinases (e.g. Tyr527 of Src) thus promoting tyrosine phosphorylation; alternatively, the phosphatase could serve a feedback function and balance the phosphorylation, e.g. by the activated growth factor receptor in the signal transduction particle. Another interesting category of factors with SH2 domains in exemplified by the oncoprotein Vav, which contains sequence motifs commonly found in transcription factors, like helix-loophelix, leucine zipper and zinc fingers (Ref. II; J. Schlessinger, pers. commun.). Vav interacts with and is phosphorylated on tyrosine residues by the EGF receptor (£ Schlessinger, pers. commun.). It is possible that phosphorylated Vav can be released from the receptor complex, and exert direct transcriptional control. The key to a deeper understanding o,~ the ro|e of SH2 domains in signal transduction lies undoubtly in the elucidation of the specificities in the interactions between various SH2 domains and individual phosphorylation sites. There is evidence accumulating for some specificity; growth factor receptors generally associate with some but not all of the proteins with SH2 domains that are listed in Fig. I. The PDGF receptor binds several of the substrates; an important question is thus whether the substrates bind to the same or different regions of the receptor. Analysis of bacterially expressed SH2 domains revealed some competition between PLC-y and GAP for binding to the receptor ~2. In another study, analysis of the abilities of various peptides to interfere with the binding of substrates to the PDGF ~-receptor demonstrated that PI-3'-K interacts with two regions in the sequence that divides the kinase domain into two parts (the kinase insert) (Ref. 13; L.T. Williams, pers. commun.). Only tyrosine-phosphorylated peptides showed activity, which is consistent with the idea that PI-3'-K interacts with autophosphorylated sites in the receptor. One of the two implicated tyrosine residues was previously identified as an autophosphorylation site in the PDGF ~receptor ~4and it seems likely that the other site can also be autophosphorylated. A necessary, but not suffid.ent, prerequisite for PI-3'-Kbinding seems to be a methionine residue in the third position downstream of a phosphorylated tyrosine residue. By the same strategy, it was concluded that GAP
interacts with another phosphorylated t~osine residue, located further downstream in the kinase insert (L. T. Williams, pers. commun.). Autophosphorylated tyrosine residues in the carboxy-terminal tails of the EGF receptor ~5 and the FGF receptor z6 have been found to be important for the interaction with PLC-¥. It is likely that many more proteins containing SH2-domains exist. The autophosphorylated region of the EGF receptor shows great promise as a probe to identify cDNAs coding for proteins with SH2 domains in ~,gtl I expression libraries ~7. The use of autophosphorylated regions of other receptors as probes in similar approaches could help to identify components of their respective signalling pathways. In conclusion, SH2 domains appear to play a fascinating role in interactions between activated growth factor receptors and cytoplasmic components in the mitogenic pathway. Data have also been presented that suggest a role for SH2 domains in the control of the activities of cytoplasmic tyrosine kinases. Although other molecular mechanisms for the interaction between, and control of, components in the signalling pathway may exist, the studies on SH2 domains have provided insight both into a possible molecular mechanisn-, for the interactions between components involved, and also a method for the identification of additional components in the mitogenic pathway. Methods have also established which will make it possib;e to eetermine the structural requirements around phosphorylated residues for binding of different SH2 domains. The stage is set for rapid progress in this research field; it is anticipated that the results will significantly extend our understanding of the signal transduction via tyrosine kinase receptors.
Acknowledgements I thank L. Claesson-Welsh, K. Miyazono, T. Pawson, J. Schlessinger, B. Westermark and L. T. Williams for valuable comments.
~eferences
1 UIIrich,A. and Schlessinger,J. (1990) Cell 61, 203-212 2 Sedowski,I., Stone,J. C. and Pawson,T. (1986) Mol. Cell. Biol. 6, 4396-4408 3 Koch,C. A. et al. (1991)Science 252, 668-674 4 Cantley,L. C. et al. (1991)Cell 64, 281-302 5 Hirai, H. andVarmus,H. E. (1990) Mol. Cell. Biol. 10, 1307-1318 60'Brian, M. C., Fukui,H. andHanafusa,H. (1990) MoL Cell. Biol. 10, 2855-2862
451
TIBS 1 6 - D E C E M B E R 1 9 9 1 7 MacAuley,A. and Cooper,J. A, (1989) MoL Cell. BioL 9, 2648-2656 8 Matsuda,M., Mayer,B. J., Fukui,Y. and Hanafusa,H. (1990) Science 248,1537-1539 9 Pendei~ast,A. M. et al. (1991)Cell 66,161-171 10 Shen,S. H., Basti~=n,L., Posner,P. I. and Chr6tien,P. (1991) Nature 352, 736-739 11 Coppola,J. et aL (199i) Cell Growth Differ. 2, 95-105
17 Skolnik,E. Y. et al. (1991) Cell 65, 83-90
12 Andersson,D. et al. (1990) Science 250,
979-982 13 Escebedo,J. A. et al. (1991) MoL Cell. Biol. 11, 1125--1132 14 Kazalauskas,A. and Cooper,J. A. (1989) Cell 58,1121-1133 15 Margolis,B. et al. (1990) EMBOJ. 9, 4375-.4390 16 Mehammadi,M. et al. (1991) Mol. Cell. Biol. 11, 5068-5078
CARL-HENRIK HELDIN Ludwig Institute for Cancer Research, Box 595, Biomedical Center, S-751 24 Uppsala, Sweden.
ASF/SF2 Most mRNA precursors (pre-mRNAs) in higher eukaryotes contain introns which must be precisely removed to generate functional mRNA products. Many laboratories have studied the mechanism of pre-RNA splicing in vitro using nuclear extracts prepared from human HeLa cells. From these studies it has emerged that a ribonucleoprotein site signals. Two recent papers reportcomplex, a spliceosome, is assembled ing the cloning of a gene encoding a on pre-mRNA in an apparently ordered mammalian factor that can influence fashion. Five of the abundant nuclear U the choice between competing 5' splice snRNPs (u,'idine-rich, small nuclear sites represent an important advance ribonucleoprotein particles) are the towards this goal4,5. major subunits of spliceosomes I which A well-documented example of alternaalso contain non-snRNP protein factors. tive splice site choice occurs in SV40Spliceosomes containing equivalent infected cells. In this case, either of two snRNPs are assembled in a similar path- alternative 5' splice sites can be joined way in yeast splicing extracts 2, under- to a common 3' splice site (Fig. I). Use lining the highly conserved nature of of the first 5' splice site produces mRNA the splicing mechanism and the funda- encoding large T antigen, while use of mental role that snRNPs play in it. the second produces mRNA encoding While the snRNPs represent basic com- small t antigen. The amino-terminal ponents of the splicing machinery, how- sequences of these two proteins are ever, there is now also considerable identical but the remainder of their interest in characterizing additional sequences are different due to a shift in trans-acting factors which might coop- the reading frame caused by the alternaerate with snRNPs and possibly regu- tive splicing events. This illustrates the late splice site choice. flexibility in coding potential that alternaMany pre-mRNAs contain multiple tive splicing can generate.
a splice site "selector
introns, allowing several different mRNAs to be produced by alternative splicing of a single primary transcript. Alternative splicing has been shown to occur for both cellular and viral RNAs and can play an important regulatory role in controlling gene expression. For example, recent studies in D r o s o p h i l a have shown that sex determination is principally regulated through modulation of splice site selection, governed by products of the sex lethal, transformer and transformer-2 genes 3.
Progress in understanding splice site utilization in mammalian cells clearly requires the identification and characterization of bans-acting factors that determine the use of alternative splice
452
Interestingly, the ratio of small t to large T mRNA produced from the common pre-mRNA transcript was observed to be greatly increased in an adenovirus-transformed human embryonic kidney cell line (293 cell@. This suggested that these cells either contained a factor that stimulated use of the proximal, small t 5' splice site or lacked a repressor which normally reduced selection of the small t splice site. The increased level of small t splicing was also seen in vitro, using nuclear extracts of 293 cells. This in vitro system provided Ge and Manley with a convenient assay for purification of a factor (Alternative Splicing Factor or ASF) responsible for the increase in small t mRNA~. Highly purified ASF could also stimulate small t splicing when added to HeLa splicing extracts. These results showed that the increase in small t splicing in 293 cells was indeed due to an activator, rather than the absence of a repressor. Ge and Manley also showed that when small t antigen splicing was activated in vitro by ASF there was a concomitant decrease in production of large T mRNA. This indicated that ASF was
T
m
~
m
~
Figure1 Cartoon (not to scale) showing the altemative 5' splice sites in SV40 pre-mRNA which are used, in conjunction with a common 3' splice site, to generate distinct mRNAs encoding large T and small t antigens. High concentrations of the splicing factor ASF-1/SF2 promotes selection of the small t 5' splice site over the large T one.
© 1991,ElsevierSciencePublishers,(UK) 0376--5067/91/$02.00