<b>Transgene‐induced gene silencing in plants</b>

Abstract

The recent development of gene transfer methods for almost all eukaryotes has revealed that transgenes can undergo silencing after integration in the genome. Host genes can also be silenced as a consequence of the presence of a homologous transgene, thus limiting the potential application of genetic transformation. Despite this limitation, transgene-induced gene silencing events were considered originally as anecdotal phenomena. However, as more and more similarities were found between transgene-induced gene silencing and natural epigenetic phenomena, considerable interest has been devoted to this subject (for recent reviews see Depicker & Van Montagu 1997;Stam et al. 1997b ). Epigenetics is commonly defined as ‘the study of mitotically and/or meiotically heritable changes in the function of a gene that cannot be explained by changes in its DNA sequence’ ( Russo et al. 1996 ). For a long time, DNA was considered as the only target for epigenetic modifications. Epigenetic changes corresponding to changes in chromatin structure and affecting transcription have been reported in almost all eukaryotes: yeast, fungi, Drosophila, plants and mammals ( Dorer 1997;Foss & Selker 1991;Rossignol & Faugeron 1994;Ye 1996). However, recent studies have suggested that, besides DNA, other molecules can be modified in a manner that resembles epigenetic DNA changes. First, it was shown that proteins can be converted into molecules of aberrant conformation called prions in yeast and mammals ( Lacroute 1971;Prusiner 1982). More recently, the involvement of RNA was hypothesized to explain post-transcriptional silencing in plants, fungi and nematodes ( Cogoni et al. 1996 ;Fire et al. 1998 ;Napoli et al. 1990 ). This review will focus on transgene-induced silencing phenomena in plants. The number of copies of a transgene that integrate into the genome of a transformed plant and the position of the integration site cannot be predicted, regardless of whether Agrobacterium-mediated transfer or direct gene transfer are used. Therefore, one or multiple intact or rearranged copies can integrate at one or multiple unlinked loci. Cis-inactivation will be used to refer to silencing events that affect the expression of transgenes integrated at a single locus, irrespective of the copy number, while trans-inactivation will be used to define the silencing effect of one locus on another. As defined by Stam et al. (1997b) , transcriptional gene silencing will be referred to as TGS, and post-transcriptional gene silencing as PTGS. Transgenes can undergo TGS in cis when one or multiple copies integrate at a locus located in or next to silent hypermethylated genomic sequences. This case closely resembles position effect variegation (PEV) in Drosophila which occurs when a euchromatic gene is moved next to heterochromatin by chromosomal rearrangement. Heterochromatin can then spread into the gene and affect its expression in a stochastic, cell-autonomous and clonal manner, thus leading to variegation (for review see Karpen 1994). Methylation in plants may spread like heterochromatin in Drosophila from the adjacent sequences into the transgene, thus leading to silencing ( Pröls & Meyer 1992). Transgenes can also undergo TGS in cis when multiple copies become methylated although they integrate at a hypomethylated locus. This case resembles local heterochromatin formation and silencing in Drosophila induced by the extension of transgene repeats (for review see Dorer 1997). Such repeat-induced gene silencing in plants, defined as RIGS by Assaad et al. (1993) , correlates with increased methylation and increased resistance to both DNase I and microccocal nuclease in transgenic Arabidopsis plants, indicating that RIGS correlates with changes in chromatin configuration ( Ye & Signer 1996). Occasionally, transgenes inserted as single copies at a hypomethylated locus can undergo TGS ( Meyer & Heidmann 1994;Meyer et al. 1992 ). Silencing was observed when a transgene derived from a monocotyledonous plant was introduced into a dicotyledonous plant, whereas it was not observed with the corresponding dicot gene ( Elomaa et al. 1995 ), suggesting that a strong discrepancy between the DNA composition of the transgene and that of the surrounding genomic sequences can be recognized by the cellular machinery, leading to the specific methylation and silencing of foreign sequences. In this case also, TGS correlated with increased methylation and increased resistance to both DNase I and microccocal nuclease ( van Blokland et al. 1997 ), thus indicating that hypermethylation and chromatin condensation are general characteristics associated with transcriptional silencing. TGS can result from the uni-directional effect of one transgene on another. An active copy of a transgene can become silent and methylated when brought by crossing into the presence of a silenced homologous copy, and can acquire the capacity to inactivate another copies in subsequent crosses ( Meyer et al. 1993 ). This phenomenon resembles ‘paramutation’, a natural epigenetic phenomenon affecting host genes in the absence of transgenes. Paramutator (inactive) alleles inhibit the expression of paramutable (active) alleles which themselves become paramutators ( Brink 1956). Paramutation indicates that homologous chromosomes can exchange information in somatic cells. The mechanism invoked for paramutation involves DNA–DNA pairing and transmission of the chromatin structure from the silent copy to the active copy, as shown in Drosophila where PEV can be transmitted in a dominant manner from the rearranged silent chromosome to the wild-type active chromosome (for review see Karpen 1994). Active transgenes can also become silent and methylated when brought into the presence of an unlinked silenced homologous transgene ( Matzke et al. 1989 ;Vaucheret 1993). This phenomenon can be defined as ‘ectopic trans-inactivation’. It differs from paramutation because the target-silenced transgenes do not acquire the capacity to inactivate in trans other unlinked transgenes ( Park et al. 1996 ). It can affect any transgene that is expressed under the control of the same promoter, irrespective of the coding sequence being expressed ( Matzke et al. 1994 ;Vaucheret 1993). This specificity indicates that the promoter of the transgene is the target for this form of transcriptional silencing ( Thierry & Vaucheret 1996). Deletion analysis indicated that 90 bp of homology between a silencing locus and the promoter of a target transgene is sufficient for directing silencing and methylation ( Vaucheret 1993). Two silencing loci showing transcriptional silencing of unlinked promoter–homologous transgenes irrespective of their positions within the genome have been identified ( Matzke et al. 1994 ;Vaucheret 1993). These two loci consist of multiple and rearranged copies of the transgene and are each located on or near a telomere ( Park et al. 1996 ). These data suggest that the ability to trigger ectopic trans-inactivation may depend on the ability of the silencing loci to interact with any other position of the genome by direct DNA–DNA pairing. Alternatively, it could involve the production of diffusible RNA by the silencing locus that leads to heritable silencing and methylation of homologous target loci via an RNA–DNA interaction ( Park et al. 1996 ;Wassenegger & Pélissier 1998). However, since one of these two silencing loci was shown to be unable to silence the expression of extra-chromosomal copies of a target transgene ( Vaucheret 1994), the transmission of silencing is more likely to occur through DNA–DNA pairing between stably integrated homologous copies. When the silenced target transgenes are separated from a silencing locus by segregation, they reactivate more or less quickly, i.e. during a period that ranges between immediately after segregation and two or three generations following the segregation. The fact that target transgenes which remain silent in the absence of the silencing locus are still methylated ( Park et al. 1996 ;Vaucheret 1994) indicates that methylation is probably involved in the maintenance of the silent state. Target transgenes driven by a 35S promoter devoid of CG and CNG methylation acceptor sites were as susceptible to silencing in trans as those driven by a wild-type promoter, indicating that methylation at CG and CNG sites is not a prerequisite for silencing. However, using this promoter, silencing was immediately relieved in the absence of the trans-silencer, thus further supporting the hypothesis that CG/CNG methylation is essential for the maintenance of silencing ( Diéguez et al. 1998 ). Transgene silencing is defined as occurring at the post-transcriptional level when RNA does not accumulate even though transcription occurs. As opposed to TGS, which is meiotically heritable ( Assaad et al. 1993 ;Matzke et al. 1989 ;Mittelsten Scheid et al. 1998 ;Park et al. 1996 ;Vaucheret 1993), PTGS is reset (i.e. affected genes are reactivated) after meiosis ( Balandin & Castresana 1997;Dehio & Schell 1994;Dorlhac de Borne et al. 1994 ;Hart et al. 1992 ;Vaucheret et al. 1995 ). However, in affected lines, PTGS recurs every generation at some time during plant development. Up until now, post-transcriptional cis-inactivation has been observed when foreign (bacterial) transgenes (uidA, npt, rolB) were introduced under the control of the strong viral 35S promoter ( Dehio & Schell 1994;Elmayan & Vaucheret 1996;English et al. 1996 ;Ingelbrecht et al. 1994 ). In all cases, PTGS occurred more efficiently (or exclusively) in haploids and homozygous plants as compared with hemizygous plants, suggesting a transgene dose effect. PTGS was observed in a larger proportion of transformants using a 35S promoter with a double enhancer as compared to the classical 35S promoter ( Elmayan & Vaucheret 1996;English et al. 1996 ). These results initially suggested that PTGS was due to the over-production of transgene RNA above a putative threshold level that triggers the irreversible degradation of RNA ( Dehio & Schell 1994). However, the level of transgene transcription was not always found to be significantly higher in silenced lines as compared to non-silenced lines ( English et al. 1996 ), thus suggesting that some other parameters may also play a role in the triggering of PTGS. The presence of repeats at the transgene locus of the silenced lines was proposed to play such a role ( English et al. 1996 ). Multiple models of PTGS have been proposed, considering mainly the roles of RNA thresholds and DNA repeats (reviewed by Baulcombe 1996). These models may not be exclusive if we consider that only a particular subpopulation of transgene RNA is important for the triggering of PTGS. Transgene RNA could be specifically degraded if tagged by specific molecules. To account for sequence-specific RNA degradation, these tag molecules are hypothesized to be small complementary RNA (cRNA). They could be synthesized by a plant RNA dependent RNA polymerase (RdRp) using transgene RNA as template ( Dougherty & Parks 1995). Alternatively, they could be internal fragments of transgene RNA produced by pairing-cleavage cycles between aberrant poly (A)– RNA and normal transgene RNA as a result of internal sequence complementarity ( Metzlaff et al. 1997 ). These cRNA could interact with mRNA, thus forming duplexes that behave as targets for cellular enzymes like double-strand RNA-specific RNase. Not all transgenic lines undergo PTGS. Therefore, the transgene loci that trigger PTGS may have some specific characteristics. First, the presence of repeats could allow DNA–DNA interactions and subsequent changes in methylation and/or chromatin structure that impedes correct transcription and leads to the production of malformed (aberrant) RNA that are a better template for RdRp than mRNA. Second, the utilisation of a very strong promoter to drive the transgene may contribute to elevate the amount of malformed (aberrant) RNA produced spontaneously by RNA polymerase errors up to a threshold that triggers PTGS. These aberrant RNA and/or the high amount of mRNA accumulated in the cytoplasm due to the use of a strong promoter may also trigger methylation of the coding sequence of the corresponding transgene by a feedback mechanism ( Wassenegger et al. 1994 ). Feedback methylated transgenes may therefore produce aberrant RNA as might do methylated transgene repeats. Therefore, either the use of a strong promoter or the presence of transgene repeats may lead to post-transcriptional silencing as a consequence of the production of aberrant RNA and subsequently of cRNA. PTGS was originally discovered as the reciprocal and co-ordinated silencing of transgenes and homologous host genes, and is usually referred to as ‘co-suppression’ ( Napoli et al. 1990 ). Since that time, a number of transgenes encoding part or the entire transcribed sequence of a host gene have been shown to trigger co-suppression of homologous host genes (for recent reviews see Depicker & Van Montagu 1997;Stam et al. 1997b ). Co-suppression occurs more efficiently (or exclusively) in haploids and homozygous plants as compared with hemizygous plants, suggesting a transgene dose effect ( de Carvalho et al. 1992 ;Dorlhac de Borne et al. 1994 ;Hart et al. 1992 ). As shown by a large scale analysis of petunia plants transformed with a chalcone synthase transgene, the efficiency of co-suppression correlates with the strength of the promoter driving the transgene, suggesting a transgene product dose effect rather than a transgene dose effect ( Que et al. 1997 ). Co-suppression of nitrate reductase is inhibited when the transgenes homologous to the host genes are themselves silenced at the transcriptional level, thus indicating that transgene transcription is required ( Vaucheret et al. 1997 ). In addition, the efficiency of co-suppression is reduced or delayed when host genes are not expressed ( Dorlhac de Borne et al. 1994 ;Smith et al. 1990 ), or when transgenes are introduced into mutants lacking a functional host gene ( Vaucheret et al. 1997 ). These results suggest that co-suppression cannot be considered as the unidirectional silencing effect of transgenes onto host genes, but rather as a reciprocal and synergistic phenomenon where host genes and transgenes can co-operate to produce aberrant RNA and/or cRNA above the threshold level that activates the RNA degradation of the co-suppression events result from the of expressed and data PTGS of host genes by transcribed or transgenes have been reported ( van Blokland et al. 1994 ). These data can be with the aberrant hypothesis above if we consider that these transgene loci always consist of transgene repeats ( Stam et al. ). Therefore, as proposed DNA–DNA pairing could play a role in the of production of aberrant between transgene repeats or between the transgene repeats and the homologous host genes could therefore trigger changes in methylation or chromatin leading to the production of aberrant RNA either by a transcribed transgene or by the host genes only when the transgene does not a When the transgene which PTGS part of the genome of a plant RNA silenced transgenic plants become (reviewed by Baulcombe In some cases, resistance can be after a of a phenomenon called ( et al. 1993 ). In this the silent is not in the transgenic and only the by the triggers transgene silencing and It is that transgene transcription and both contribute to the level of of RNA that triggers PTGS. a transgene inserted within their genome to transgenic plants that PTGS of this transgene ( English et al. 1996 ). These results are with a in which viral RNA and transgene RNA the same sequence are tagged by the same cRNA and are subsequently degraded by the RNA degradation a transgene inserted within their genome in transgenic plants that do not PTGS of this However, the transgene may become silenced after a phenomenon called gene ( et al. 1998 ). PTGS as a ( et al. 1994 ;Hart et al. 1992 et al. 1996 ). and in the of the silencing during plant development suggest the of a silencing through the Co-suppression of nitrate reductase and host genes and transgenes in lead to particular or with the of the corresponding that can be the Co-suppression as or on one and then to the with silencing efficiency ( et al. 1994 et al. 1996 ). Since these of are found in all transgenic lines silenced for a this that a sequence-specific involved in the control of PTGS through the plant in a specific the transmission of a PTGS was in Silencing was transmitted with efficiency from silenced to target the corresponding transgene ( et al. 1997 ), indicating the of a The transmission of co-suppression also occurred when silenced and non-silenced target were separated by up to of of a wild-type These results therefore the hypothesis that a the of de post-transcriptional silencing long within the plant, a phenomenon called silencing or ( et al. 1997 ). were when PTGS was after of one of a transgenic plant a non-silenced transgene by an the same transgene ( & Baulcombe 1997). using nitrate reductase silenced and a of transgenic and revealed for the of RNA degradation, and RNA degradation occurred in both transgenic mRNA due to the presence of a transgene and host mRNA due to However, RNA degradation not occur in wild-type thus indicating that mRNA above the level of wild-type plants rather than the presence of a transgene in the is required for the RNA degradation of co-suppression ( & Vaucheret 1998). When silenced were from the silenced and onto wild-type plants, silencing was not in plants and in transgenic lines that are not to trigger co-suppression silencing was in transgenic lines that are to trigger co-suppression thus indicating that only the transgene loci that are to co-suppression can also a silent ( silencing silencing. transgene loci are to produce a that sequence-specific RNA degradation in the This to through and at long through the then triggering degradation of homologous this silencing is it is likely that it from the it is a direct product of the transgene (for an aberrant or a product (for a and whether such RNA through the plant or with a host to be The absence of maintenance of silencing in plants that undergo RNA degradation that the silencing is not of RNA degradation It also indicates that the transgene locus as a to co-suppression in each that the In the absence of a transgene locus, a production of the silencing by the silenced is required to RNA degradation in the This phenomenon could therefore be as silencing because it is not when the silenced are in a transgene locus, the could an epigenetic of the transgene locus that the of the production of both the RNA degradation and the silencing thus leading to silencing in the a silencing is involved in all PTGS events reported to be In were produced by transgenes that control As proposed by co-suppression may of plant to of This hypothesis is in with the fact that wild-type onto transgenic silenced do not undergo whereas which accumulate the host mRNA above the level of the wild-type do undergo silencing ( & Vaucheret 1998). Therefore, both a sequence-specific silencing and a high amount of target mRNA (or a high of to be required to a silent state. Dougherty et al. not the transmission of resistance in plants a viral thus suggesting that and/or is not a general However, this result be in the of recent studies showing that PTGS of et al. reported that PTGS of nitrate reductase or transgenes in of and Arabidopsis plants, suggesting that the spread of has sequence to the or may inhibit the spread of the PTGS genes are likely to be involved in both TGS and PTGS. In Drosophila, of genes PEV has the of more than loci (for review see Karpen 1994). These loci to two that PEV and that of the corresponding genes have been of the are involved in the formation of heterochromatin while genes of the transcription in which PTGS is have been in These called (for define three genetic loci ( Cogoni & 1997). in the control of TGS have been in Arabidopsis and are called mutants (for These mutants which are to reactivate a silent transgene ( Scheid et al. 1998 ). These mutants also a in the methylation of sequences of the as do methylation mutants called in DNA ( et al. 1993 ). is to The function by the gene is It is that it a because the plants a normal plants lacking were using an transgene the gene TGS was not in such plants ( Scheid et al. 1998 ). Therefore, these results that in the between TGS and affected in PTGS have also been in Arabidopsis These mutants to two The to mutants called (for enhancer of gene in which PTGS of a transgene is These are and define two genetic loci ( Dehio & Schell 1994). The to mutants called (for of gene which PTGS of a transgene as as post-transcriptional co-suppression of nitrate reductase host genes and transgenes. These are and define two genetic loci ( Elmayan et al. 1998 ), which probably to loci in and are not to reactivate a silent transgene, and do not trigger a in the methylation of sequences of the genome. Therefore, genes are to specifically control PTGS. The of and genes will the of the of TGS and PTGS. gene silencing in plants can occur at the transcriptional or post-transcriptional TGS occurs mainly when multiple repeats of a transgene are inserted in the genome of transgenic plants. It correlates with condensation of chromatin and with The transfer of methylation and silencing from one locus to another indicates that of the genome and exchange This transfer may occur through direct DNA–DNA pairing. Alternatively, it could involve the production of diffusible RNA by one locus, leading to of homologous targets via an RNA–DNA interaction ( Park et al. 1996 ;Wassenegger & Pélissier 1998). Epigenetic of sequences may be an important because it might or between transgene repeats may be for the between induced methylated and DNA molecules may not be thus This hypothesis may not only to the genome of transformants transgene but also to wild-type plants in which an in the number of can the of the genome if they can with each Such a to in other In the fungi and DNA that the are very silenced and In addition, in methylation of is by a very high of in a phenomenon called repeat-induced or ( et al. 1989 ). Such methylation and are proposed to also occur in to the homology between repeats and to efficiently ( et al. 1992 ). In plants, for such a has been observed ( Scheid et al. 1994 ), suggesting that may be sufficient to the genome transgene-induced PTGS occurs mainly when transgene RNA is produced at high under the control of the 35S promoter of the it has been shown that and initially when by from which they by of This phenomenon correlates with the absence of of and 35S RNA although their of transcription remain ( et al. 1998 et al. 1997 ). plants spontaneously from by the ( et al. 1997 ). These results suggest that plants can in a post-transcriptional Therefore, of the of PTGS of transgenes may result from the of high of RNA transcription and from the subsequent degradation of this RNA by the cellular involved in post-transcriptional However, since or transcribed transgenes can lead to post-transcriptional silencing of homologous host genes, some molecules (for the aberrant RNA and/or cRNA defined involved in the of events leading to RNA degradation might be produced in an manner, for by and subsequently methylated that diffusible can the plant to trigger silencing in the other ( et al. 1997 ), as aberrant RNA and/or cRNA as diffusible of silencing and whether they can or with to be The similarities between transgene-induced PTGS and in plants ( et al. 1997 ), and the between the spread of and the spread of PTGS ( et al. 1998 ), the hypothesis that transgene-induced PTGS from a natural post-transcriptional TGS and PTGS phenomena may natural of plant at the DNA or RNA level or like TGS may cellular DNA that into the while PTGS may cellular DNA that in the or RNA that in the large of transgene and/or or can be introduced in plants to their silencing and as plant mutants affected in the control of silencing become it is to further in the analysis of the and the natural roles of epigenetic control in plants. from the on Silencing for and and for of the