Do natural antisense transcripts make sense in eukaryotes?

Gene 211 (1998) 1–9

Review

Do natural antisense transcripts make sense in eukaryotes? Christine Vanhe´e-Brossollet, Catherine Vaquero * CNRS UPR0415, ICGM, 22 rue Me´chain, 75014 Paris, France Received 15 October 1997; received in revised form 15 January 1998; accepted 16 January 1998

Abstract The existence of naturally occurring antisense RNAs has been illustrated, in eukaryotes, by an increasing number of reports. The following review presents the major findings in this field, with a special focus on the regulation of gene expression exerted by endogenous complementary transcripts. A large variety of eukaryotic organisms, contains antisense transcripts. Moreover, the great diversity of genetic loci encoding overlapping sense and antisense RNAs suggests that such transcripts may be involved in numerous biological functions, such as control of development, adaptative response, viral infection. The regulation of gene expression by endogenous antisense RNAs seems of general importance in eukaryotes as already established in prokaryotes: it is likely to be involved in the control of various biological functions and to play a role in the development of pathological situations. Several experimental evidences for coupled, balanced or unbalanced expression of sense and antisense RNAs suggest that antisense transcripts may govern the expression of their sense counterparts. Furthermore, documented examples indicate that this control may be exerted at many levels of gene expression (transcription, maturation, transport, stability and translation). This review also addresses the underlying molecular mechanisms of antisense regulation and presents the current mechanistic hypotheses. © 1998 Elsevier Science B.V. All rights reserved. Keywords: Endogenous; RNA; Regulation; Expression

1. Introduction Natural antisense RNAs are endogenous transcripts that exhibit complementary sequences to transcripts of an already known function, named sense transcripts. Most antisense transcripts, so-called cis-encoded, are issued from the same locus as sense transcripts. Transcribed from opposite strands of DNA, sense and cis-encoded antisense transcripts overlap each other at * Corresponding author. Present address: INSERM U313, La Pitie´Salpeˆtrie`re, 91 Bd de l’Hoˆpital, 75013 Paris, France. Tel: +33 1 40 77 97 36; Fax: +33 1 45 83 88 58; e-mail [email protected] Abbreviations: 3∞UTR, 3∞ untranslated region; ASP, antisense protein; bFGF, basic fibroblast growth factor; BrdU, 5-bromo-2∞-deoxyuridine; dsRNA, double-stranded RNA; Env, envelope; FIV, feline immunodeficiency virus; HIV, human immunodeficiency virus; HSV, human simplex virus; IBP, Inr-associated binding protein; LAT, latency associated transcript; LCE, lin-4-complementary element; MBP, myelin basic protein; mld, myelin-deficient; ORF, open reading frame; PKR, ds activated, eIF-2 a- specific protein kinase; RRE, Revresponsive element; RT-PCR, reverse transcription-polymerase chain reaction; T3, triiodothyronine; TS, thymidylate synthetase; u/m, unwindase/modificase. 0378-1119/98/$19.00 © 1998 Elsevier Science B.V. All rights reserved. PII: S0 3 7 8 -1 1 1 9 ( 9 8 ) 0 0 09 3 - 6

least partially and display perfect complementarity. In contrast, trans-encoded antisense RNA originate from a different locus and may display only partial complementarity with the sense transcript. Natural antisense RNAs have first been described in prokaryotes, where they were found to be part of a general mechanism of control of gene expression. They are widely distributed and are involved in the control of biological functions as diverse as transposition, plasmid replication, incompatibility and conjugation, bacteriophage temporal control of development and bacterial gene expression [see review by Wagner and Simons (1994)]. In all the prokaryotic examples studied so far, antisense transcripts were found to downregulate the expression of sense transcripts. Although there is no experimental evidence so far, mechanisms for positive regulation are, none the less, quite plausible. Moreover, the introduction of artificial complementary oligonucleotides and expression of transduced antisense RNA or ribozymes have been extensively used to inhibit gene expression. Although variable degrees of success were achieved, these techniques have, indeed,

C. Vanhe´e-Brossollet, C. Vaquero / Gene 211 (1998) 1–9

2

Table 1 Representative natural antisense transcripts thought to regulate sense gene expression Organism

Sense transcript

Antisense

Antisense ORF

Sense ORF function

Level of antisense regulation

References

EBV HTLV-1 HSV-1

ebna

Viral transactivation Viral transactivation Viral replication

Post-transcriptional

lat

+ + +

HIV-1

bzlf1 tax 1 ICP0 et ICP c34.5 Env

AS01

+

FIV

Env

C. elegans

lin 14

−

Heterochronicity control

Dictyostelium

eb4-spv or PSV-A HcB.12 dopa decarboxylase micropia bFGF or FGF-2 a1(I) collagen c-erbA a

−

Structure

Post-transcriptional: translation? Cytoplasmic stability

Prang et al. (1995) Larocca et al. (1989) Stevens et al. (1987); Croen et al. (1987) Michael et al. (1994); Vanhe´e-Brossollet et al. (1995) Vanhe´e-Brossollet and Vaquero (unpublished observations) Lee et al. (1993); Wightman et al. (1993) Hildebrandt and Nellen (1992)

− −

Choriogenesis

Nuclear?

Skeiky and Iatrou (1990) Spencer et al. (1986)

− +

Retrotransposition Development and repair

Cytoplasmic Cytoplasmic stability

Lankenau et al. (1994) Kimelman and Kirschner (1989)

−

Structure

Transcriptional

Farrell and Lukens (1995)

Rev-erb

+

Hormonal response

Splicing

gfg

+

Development and repair

Cytoplasmic stability?

−

Myelination

−

Transcription control

Nuclear processing and/or transport Post-transcriptional nuclear event: splicing?

Lazar et al. (1989, 1990); Munroe and Lazar (1991) Murphy and Knee (1994); Li et al. (1996) Tosic et al. (1990); Okano et al. (1991) Khochbin and Lawrence (1989); Khochbin et al. (1992) Bedford et al. (1995) Nepveu and Marcu (1986); Kindy et al. (1987); Spicer and Sonenshein (1992); Celano et al. (1992) Miyajima et al. (1989) Knee et al. (1994); Murphy and Knee (1994) Silverman et al. (1992); Noguchi et al. (1994) Krystal et al. (1990); Armstrong and Krystal (1992) Lerner et al. (1993) Laabi et al. (1994) Eccles et al. (1994); Campbell et al. (1994); Malik et al. (1995) Hervieu and Nahon (1995) Chang et al. (1991); Celano et al. (1992) Celano et al. (1992)

Silkmoth Drosophila Drosophila Xenopus Chicken Rat Rat

+ lin 4

gfg

Mouse

bFGF or FGF-2 mbp

Mouse

p53

Mouse Mouse

Hoxd-3 c-myc

Dxoh-3

+ −

Development Proliferation

Human Human

ear-1 gfg

+ +

Hormonal response Development and repair

Splicing? Cytoplasmic stability?

Human

ear-7 bFGF or FGF-2 eIF2 a

−

Translation initiation

Transcription

Human

N-myc

N-cym

+

Transcription control

Transcription? Splicing?

Human Human Human

CD3 e/g/h bcma WT1

Oct1

+ + −

Immune response Immune response Proliferation control

Splicing?

Human Human

MCH c-myc

− −

Neurotransmission Proliferation

Human

c-myc, N-myc, p53, TK TS SC35 GnRH bcl-2

ASM-1

+

Growth control

c-myb SH bcl-2/IgH

+ + + −

Proliferation Splicing Hormonal response Control of apoptosis

Human Human Human Human

WIT1

proved that antisense nucleic acids were able to modulate gene expression in eukaryotes as well as in prokaryotes. Over the last 10 years, some endogenous antisense RNAs have been reported in eukaryotes, thus raising

Nuclear Nuclear event: transcription?

Stability?

Dolnick (1993) Fu and Maniatis (1992) Adelman et al. (1987) Capaccioli et al. (1996)

the following questions. Are natural antisense transcripts widely distributed in eukaryotes? Do they play a role in gene expression as is the case in prokaryotes? If so, what are the precise mechanisms involved?

C. Vanhe´e-Brossollet, C. Vaquero / Gene 211 (1998) 1–9

2. Are natural antisense transcripts widely distributed among eukaryotes? The transcription of both strands of human and mouse mitochondrial DNA was first reported in 1981 (Anderson et al., 1981; Bibb et al., 1981). Since then and since the description of overlapping sense and antisense transcripts in Drosophila (Spencer et al., 1986), an increasing number of endogenous antisense RNA has been reported ( Table 1). Examples of antisense transcripts have been described in numerous organisms: viruses, slime molds, insects, amphibians and birds as well as mammals (rats, mice, cows and humans). These antisense RNAs are complementary to sense transcripts encoding proteins involved in extremely diverse biological functions: hormonal response, control of proliferation, development, structure, viral replication, etc. ( Table 1). These characteristics suggest that antisense transcripts are found throughout the eukaryotic world and might play a role in general antisense-mediated gene regulation as is the case in prokaryotes. In addition, some antisense RNA are conserved between species as shown by the following examples. Antisense transcription over the c-myc locus was detected in rodent ( Kindy et al., 1987), bovine (Nepveu and Marcu, 1986) and human (Bentley and Groudine, 1986) species. In lentiviruses, an antisense open reading frame (ORF ) is transcribed antisense to the Rev-responsive element (RRE) for both human (HIV ) and feline (FIV ) immunodeficiency viruses. However, the RRE region is located either within ( HIV ) or at the 3∞ terminus of (FIV ) the envelope (env) sequence (Michael et al., 1994; Vanhe´e-Brossollet et al., 1995; CV-B and CV, unpublished data). A genomic arrangement for the c-erbAa locus that yields two alternatively spliced sense mRNA and an antisense transcript overlapping the last exon of only one of the sense messenger is conserved in rats (Lazar et al., 1989) and in humans (Miyajima et al., 1989). An antisense transcript to the basic fibroblast growth factor (bFGF ) mRNA encodes a protein highly conserved from frog to man ( Kimelman and Kirschner, 1989; Murphy and Knee, 1994). All these data showing wide representation and conservation of endogenous antisense transcripts strongly suggest that these antisense RNA are not fortuitous and may play a general role in gene expression.

3. What are the potential roles of antisense transcripts? Even though most reported endogenous antisense RNAs have unknown functions or significance to date, pertinent proposals and data are accumulating, strongly implying that natural antisense transcripts may, a priori, fulfil two non-exclusive major functions: template for translation and regulation of sense gene expression.

3

3.1. Template for translation Antisense transcripts, like sense mRNAs, may encode proteins, if they contain an ORF and if once polyadenylated, they are transported into the cytoplasm. Some do meet these requirements such as the mRNA antisense to thymidylate synthetase (TS ) found in the polyadenylated cytoplasmic fraction of RNA (Dolnick, 1993). Some antisense mRNA are also translatable in vitro such as rev-ErbAa (Lazar et al., 1989; Miyajima et al., 1989), gfg ( Kimelman and Kirschner, 1989) or n-cym transcripts (Armstrong and Krystal, 1992). Others are actually translated in vivo such as two ORF encoded by different regions of the latency associated transcript (LAT ) of the human simplex virus (HSV-1): the Latency-Associated Antigen of type 1 (Doerig et al., 1991) and the P-ORF (Lagunoff and Roizman, 1994) as well as the antisense protein ASP in HIV ( Vanhe´eBrossollet et al., 1995) and Gfg in both Xenopus oocytes and rat tissues (Li et al., 1996). 3.2. Regulation of sense expression Apart from their capability of encoding proteins, antisense transcripts may play a role in the regulation of expression of their sense counterparts. The finding that antisense-mediated gene regulation occurs in prokaryotes and even in archaebacteria (Stolt and Zillig, 1993) strongly suggests that such regulation is also functional in eukaryotes. Indeed, because of their complementarity, antisense transcripts may hybridize to sense transcripts and thus modify sense expression. We will now review experimental evidence accounting for antisense-mediated gene regulation. Some publications simply suggest antisense-mediated gene regulation based upon spatial and/or temporal distributions of sense and antisense transcripts. For instance, tissue distribution studies showed that high levels of sense and antisense dopa decarboxylase transcripts never occur together in Drosophila (Spencer et al., 1986). Along the same line, the only viral transcripts expressed in HSV-1 latently infected cells are antisense RNAs that are absent from productively infected cells, thus displaying a temporal pattern opposite to that of sense transcripts (Croen et al., 1987; Stevens et al., 1987). Stronger experimental data are given by reports showing that changes in sense gene expression are correlated with the presence of antisense RNA. Indeed, an inverse relationship between levels of accumulation of sense and antisense messengers has been documented in the case of the eb4-psv gene during development of Dictyostelium ( Hildebrandt and Nellen, 1992). A similar inversion in the ratio of sense to antisense a1(I) collagen transcripts was also described in chondrocytes upon pharmacological treatment (Farrell and Lukens, 1995), as discussed in Section 3.2.1.1.

4

C. Vanhe´e-Brossollet, C. Vaquero / Gene 211 (1998) 1–9

Other compelling evidences of antisense-mediated gene regulation concern the development of some pathologies. For example, nervous disorders observed in myelin-deficient (mld ) mice results from a tandem duplication of the gene encoding the myelin basic protein (MBP) with the upstream gene containing an inversion from exons 3 to 7, thus yielding an antisense transcript leading to a marked down-regulation of the myelin messenger and protein ( Tosic et al., 1990). Endogenous antisense transcripts able to downregulate Wt1 levels may also be involved in dysregulation of the normal expression of the tumour suppressor gene wt1 observed in some Wilm’s tumours ( Eccles et al., 1994; Malik et al., 1995). In some lymphomas containing the t(14;18) (q32;q31) translocation resulting in bcl-2/IgH loci fusion, an antisense transcript to the bcl-2 messenger seems to be responsible for an increase in the cellular level of the antiapoptotic protein Bcl-2, thus leading to upregulated cellular proliferation (Capaccioli et al., 1996). This mediated upregulation of the sense transcripts will be discussed in Section 4.2. Apart from the above data only correlating sense and antisense transcript expressions, an in-depth analysis of antisense-mediated gene regulation was provided by more detailed experiments examining where and how such a possible regulation could take place. Indeed, due to the location of the antisense transcripts in the cells, either in the nucleus or the cytoplasm, the antisensemediated regulation might actually occur at any step of gene expression, from transcription to translation. 3.2.1. Nucleus Evidence for nuclear regulation usually comes from antisense transcripts that are non-polyadenylated and consequently not transported to the cytoplasm. Antisense RNA may regulate sense expression in the nucleus either at the level of transcription, processing or nucleocytoplasmic transport. 3.2.1.1. Transcriptional regulation. A case of transcriptional regulation has been described for a1(I ) collagen, an important structural component of skin, bones, ligaments and tendons (Farrell and Lukens, 1995). In chicken chondrocytes, the steady state of a1(I) collagen mRNA increases upon 5-bromo-2∞-deoxyuridine (BrdU ) treatment although both mRNA half-life and overall transcription rate, using double-stranded probes for run-on assays, remain constant. However, using singlestranded probes, antisense transcription was detected across a major portion of a1(I) collagen messenger. This natural antisense transcription predominantly occurs in uninduced cells while the sense transcription is low. Upon BrdU induction, antisense transcription decreases while sense transcription rises, the level of sense transcription being exactly correlated with overall a1(I) collagen mRNA accumulation. These results sug-

gest that either the activity of sense and antisense promoters is differentially regulated by cellular conditions and/or that antisense transcription impedes sense transcription. This interference would involve the collision of two transcription complexes, resulting in premature termination or in reduced elongation of transcription, the transcripts with the highest rate of transcription being predominant. The other example of transcriptional regulation is provided by the translation initiation factor eIF-2a whose level dramatically rises upon entry into the cell cycle. Antisense RNA to eIF-2a mRNA was found by RT-PCR in human G T cells when sense transcripts 0 were barely detectable. However, in G T lymphocytes, 1 sense mRNA is induced, and antisense RNA disappear (Noguchi et al., 1994). In addition, an antisense InR promoter was described in the first intron of eIF-2a gene (Silverman et al., 1992). Conditions, such as mutation of InR, that result in a decreased antisense transcription actually increase sense transcription, thus showing that, in this instance, sense and antisense transcription are coupled. In addition, this antisense transcription is under the control of a positive regulatory element that binds a protein (IBP) probably controlling the access of RNA polymerase II to the antisense promoter. It may thus be inferred that sense transcription induced by entry in the cell cycle impedes IBP binding, subsequently reducing antisense transcription. 3.2.1.2. Post-transcriptional nuclear regulation. As we have seen, post-transcriptional nuclear antisense regulation has been suggested for mld mice ( Tosic et al., 1990) in which hypomyelination of the central nervous system results from a tandem duplication of mbp with the upstream gene containing a partial inversion. MBP is a functionally important constituent of myelin, the specialized membrane that surrounds and insulates axons. In brains of mld mice, a marked decrease in both MBP and cytoplasmic mbp messenger levels (the level of mbp mRNA in the nucleus being similar to that of control cells) is observed with the concomitant presence of abundant nuclear antisense transcripts. The drastic difference between cytoplasmic sense messengers in mld mice compared to wild-type animals cannot be accounted for either by nuclear degradation of mRNA or by transcriptional interference, even though sense transcription is slightly reduced. Thus, antisense regulation seems to operate at a post-transcriptional level probably by impairing either maturation and/or transport of the sense transcript. An example of an in-depth analysis of post-transcriptional nuclear antisense regulation is provided by the study of the locus c-erbAa encoding three structurally related proteins, belonging to the thyroid/steroid hormone receptor family, on both DNA strands in rats (Lazar et al., 1989) and in humans (Miyajima et al.,

C. Vanhe´e-Brossollet, C. Vaquero / Gene 211 (1998) 1–9

1989). The sense primary c-erbAa transcript may yield two different mRNA upon alternative splicing and two proteins, R-erbAa1 and R-erbAa2 with different properties. Whereas R-erbAa1 binds to the thyroid hormone triiodothyronine ( T3), R-erbAa2 is a non-T3 binding inhibitor of T3 action. The antisense rev-erbAa transcript is complementary to the last exon of r-erbAa2 mRNA but not complementary to the r-erbAa1 messenger. The antisense protein, Rev-erbAa that shows a low but appreciable binding to T3 is able to modulate gene transcription as for any member of this erbA super family of regulatory proteins and is thought to be involved in adipocyte differentiation. Tissue expression and pharmacological induction of rev-erbAa mRNA are always associated with lower levels of r-erbAa2 and higher levels of r-erbAa1 messengers, without any change in either transcription or stability of c-erbAa pre-mRNA, suggesting that antisense expression may alter primary transcript processing (Lazar et al., 1989, 1990; Chawla and Lazar, 1993). In-vitro reconstitution experiments indeed showed that rev-erbAa mRNA can inhibit rerbAa2 transcript splicing (Munroe and Lazar, 1991). These experimental data suggest that in-vivo expression of antisense rev-erbAa messenger prevents sense c-erbAa primary transcript splicing into r-erbAa2 mRNA, thus tilting the balance towards R-erbAa1 synthesis and ultimately modulating cellular response to hormones. 3.2.2. Cytoplasm Although the former examples have shown that antisense regulation may occur in the nucleus, antisense regulation is generally described as a cytoplasmic event operating mostly at the messenger stability level. The best examples of such regulation are the eb4-psv mRNA in Dictyostelium and the bFGF transcript in Xenopus. 3.2.2.1. Messenger stability. In spite of constitutive transcription during growth and development, eb4-psv transcripts that encode a protein associated with the membrane of prespore vesicles transiently accumulate only when cells aggregate to establish the presporeprestalk pattern (slug). If this structure is mechanically disaggregated, the messenger disappears rapidly without any change in transcription rate, thus suggesting regulation at the stability level. During early phases of aggregation and upon mechanical disruption, an antisense RNA is transcribed, and its subsequent mainly cytoplasmic accumulation correlates with the disappearance of sense transcripts. This antisense RNA, devoid of any open reading frame, initiates within exon 3 of the sense gene and ends in the promoter region of eb4-psv gene. It seems to be transcriptionally regulated and rather unstable. Indeed, inhibition of cellular transcription prevents both antisense transcript accumulation and sense messenger destabilization. From these observations, a model of regulation was proposed. Due to the intrinsic instabil-

5

ity of antisense RNA, when sense and antisense transcriptions are equivalent, as in slugs, antisense RNA accumulation is too small to interfere with sense expression. However, in vegetative cells and in disaggregated slugs, antisense RNA transcription increases and becomes sufficient to destabilize and downregulate sense expression (Hildebrandt and Nellen, 1992). The growth factor bFGF is a highly conserved broad spectrum mitogen for cells of mesodermal origin that is involved in a range of developmental and tissue repair processes (embryogenesis, neurite outgrowth, differentiation, angiogenesis, wound healing). Its locus also encodes an antisense transcript, gfg, overlapping the entire exon 3 of bFGF gene and coding for a protein thought to be involved in removal of damaged nucleotides from the intracellular pool (Li et al., 1996). In Xenopus oocytes, both sense and antisense transcripts are constitutively expressed until maturation. Upon fertilization, bFGF transcripts are abruptly deadenylated and degraded, whereas the level of gfg transcripts only slightly decreases. Isolation of overlapping regions between sense and antisense transcripts shows an extensive modification of adenosine residues into inosine residues. This modification is probably achieved by an unwindase/modificase (u/m) specific for double-stranded RNA ( Wagner et al., 1989). This enzymatic activity is present in oocyte nuclei and would be released in the cytoplasm upon germinal vesicle breakdown. From these observations, it was hypothesized that most sense mRNA would be annealed to antisense transcripts present in excess in oocytes. Upon fertilization, the RNA duplexes would be modified and thus become susceptible to nuclease attack ( Kimelman and Kirschner, 1989). 3.2.2.2. Translation. Cytoplasmic antisense regulation may also occur at translation level as shown by the example of lin-4/lin-14 transcripts in C. elegans. The heterochronic lin-14 gene encodes a protein (Lin-14) involved in the control of early development stages of the nematode. During development, the level of its mRNA remains constant even though the level of Lin-14 dramatically decreases after L2 stage, suggesting posttranscriptional regulation at the translation level. This downregulation is controlled by the 3∞UTR of the lin-14 mRNA and by another heterochronic gene, lin-4 ( Wightman et al., 1993). The lin-4 locus yields two small transcripts of 22 and 61 nt, without ORF, that are complementary to seven well-conserved lin-4-complementary-elements (LCE ) of lin-14 messenger 3∞ untranslated region (3∞UTR) (Lee et al., 1993). The model inferred from the experimental data suggests that the lin-4 transcripts may anneal in the cytoplasm to the lin-14 mRNA 3∞UTR and therefore efficiently downregulate the synthesis of Lin-14. Inhibition of translation may, for instance, result from structural features of the lin-4/lin-14 duplex interfering with a critical interaction

6

C. Vanhe´e-Brossollet, C. Vaquero / Gene 211 (1998) 1–9

for translation between the 5∞ and 3∞ ends of the lin-14 mRNA. Indeed, the 3∞UTR region and the polyA tail are believed to be able to modulate translation efficiency probably via direct or indirect interaction between 3∞-proximal elements and far upstream sequences or structures [see review by Jackson and Standart (1990)]. Recently, lin-4 transcripts were shown to control another heterochronic gene lin-28 via one LCE in its 3∞UTR, suggesting that regulation of gene expression by small antisense transcripts may be a general mechanism in nematode development (Moss et al., 1997).

4. What are the mechanisms of antisense-mediated regulation? 4.1. Two possible mechanisms Two mechanisms of antisense-mediated gene regulation may be envisioned. First, antisense transcripts displaying very similar structural features to sense transcripts may bind proteins actually interacting with their sense counterparts, thus depriving sense messengers from proteins necessary for their function. Such an explanation was proposed when an artificial transcript complementary to the RRE region was shown to downregulate HIV replication via its interaction with the Rev protein ( Kim et al., 1996). However, sense RRE was as active as antisense RRE in regulating HIV replication presumably because both share a very similar three-dimensional structure, and hence interact in the same way with Rev. This regulation was also often reported for gene silencing approaches using oligonucleotides (Sharma and Narayanan, 1995) since trapping of proteins is thought to be a general mechanism of synthetic oligonucleotide action. The other mechanism of antisense-mediated regulation is thought to operate via duplex formation between complementary sense and antisense transcripts. Indeed, in-vivo duplexes have been detected by RNAse protection assays in two examples: between antisense n-cym RNA and a subset of sense n-myc transcripts retaining intron I ( Krystal et al., 1990) and between sense and antisense mbp transcripts (Okano et al., 1991). The conversion of adenosine residues into inosine residues detected in overlapping regions of sense and antisense bFGF transcripts interpreted as providing indirect evidence for the actual existence of RNA duplexes ( Kimelman and Kirschner, 1989) remains controversial (Saccomanno and Bass, 1994). Another piece of indirect evidence is given by the existence of a lin-4 mutant nematode unable to downregulate Lin-14 cellular levels. A single nucleotide change in the region of lin-4 transcript expected to be base-paired to lin-14 mRNA 3∞UTR would result in a decreased stability of the sense/antisense duplex, again strongly suggesting that

lin-4 transcripts are probably duplexed to regions of lin-14 3∞UTR (Lee et al., 1993). In spite of these few experimental examples, sense/antisense RNA duplexes are usually not detected in vivo, and thus are believed to be rapidly degraded. What would be the consequences of such duplex formation? Two hypotheses have been formulated so far. In the first one, by simple steric hindrance, RNA duplexes would prevent sense RNA from interacting with diverse cellular components required for normal sense expression, thus impairing maturation, nucleocytoplasmic transport, transcript stability (either nuclear or cytoplasmic), or translation depending on the cellular components involved. Alternatively, duplexes may represent substrates for double-stranded RNA (dsRNA) specific enzymes [for a more complete review, see Nellen and Lichtenstein (1993)] such as the u/m enzyme thought to modify bFGF duplexes in Xenopus oocytes. Other potentially involved dsRNA-dependent enzymes are the protein kinase (PKR) and the 2-5A synthetase, that would play a role in the decrease of translation and mRNA stability, respectively. The activated PKR is responsible for the phosphorylation of the initiation factor eIF-2a that leads to inhibition of initiation of protein synthesis (Proud, 1995). The activated 2-5A synthetase leads to 2-5A production and subsequent RNA degradation via activation of a latent nuclease ( Kerr, 1987). Therefore, it is commonly believed that most duplexes will become targeted for degradation by RNAses either specific for dsRNA or specific for singlestranded RNA but activated by dsRNA. That would explain why RNA duplexes are so difficult to demonstrate. From this hypothesis, it could be inferred that only the most abundant transcripts, either sense or antisense, will persist in the cells. 4.2. Lasting coexpression of sense and antisense transcripts How, then, can concomitant expression of sense and antisense transcripts, often seen in vivo, be explained? Although a number of cases describe a strict inverse relationship between sense and antisense accumulation, there are examples of lasting coexpression of both transcripts: bcma sense and antisense transcripts in human B cells (Laabi et al., 1994), sense gnrh and antisense sh RNA expression in rat hypothalamus (Adelman et al., 1987) and in Dictyostelium for the eb4-psv complementary RNAs (Sadiq et al., 1994). In addition to persistent expression, sense and antisense transcripts may even be coordinately regulated in some instances as described for hcb12 sense and antisense RNA in silkmoth (Skeiky and Iatrou, 1990), n-myc and n-cym transcripts in tumour cell lines (Armstrong and Krystal, 1992), and sense and antisense micropia RNA during spermatogenesis in Drosophila (Lankenau et al.,

C. Vanhe´e-Brossollet, C. Vaquero / Gene 211 (1998) 1–9

1994). Even more intriguing, examples have even been described where sense and antisense transcripts are differentially expressed in some tissues but permanently co-expressed in others. For example, the HIV env sense and antisense transcripts accumulate in parallel in chronically infected T lymphocytic cell lines, but inversely in a chronically infected promonocytic cell line upon mitogenic activation ( Vanhe´e-Brossollet et al., 1995, and CV-B and CV unpublished observations). In most cells and tissues, low expression of bFGF mRNA is usually associated with high levels of antisense gfg messenger and vice versa ( Kimelman and Kirschner, 1989; Murphy and Knee, 1994) with the exception of kidney and colon where both transcripts are equally expressed ( Knee et al., 1994). Several explanations may account for the apparent permanent expression of both sense and antisense transcripts in a given cellular population. First, sense and antisense transcripts may not be expressed in exactly the same cell. Second, the intracellular environment and organization may hamper duplex formation, such as physicochemical conditions (ionic strength, temperature, pH ), local concentrations of complementary RNA as well as accessibility of transcripts for efficient annealing. Inaccessibility of transcripts to duplex formation may be due to either RNA packaging or RNA secondary structure. For instance, only one class of n-myc precursor RNA is involved in duplexes with antisense n-cym transcripts whereas another class of n-myc precursor differing only by 17 additionnal bp at the 5∞ end is unable to do so, suggesting that the 17 bp alter the secondary structure, thereby, impairing stable duplex formation ( Krystal et al., 1990). This assumption is strengthened by the fact that RNA stem-loop structures have been defined as essential for endogenous antisense transcript action in prokaryotes. Indeed, loops were described as essential for the initiation of the pairing reaction, whereas the stems confer metabolic stability [see review by Wagner and Simons (1994)]. A lack of duplex formation may also result from different compartmentalization of sense and antisense transcripts (association of one of the transcripts with polysomes, for example) and/or compartmentalization of enzymes required for duplex formation or modification, as in the case of u/m localized in oocyte nuclei and released in the cytoplasm only upon fertilization ( Kimelman and Kirschner, 1989). Differential expression of such enzymes would also explain why sense and antisense transcripts could coexist in some tissues but not in others. Third, in contrast to the general belief that duplexed sense and antisense transcripts are rapidly degraded, stabilization of transcripts via duplex generation might be quite possible. For instance, an impaired binding of destabilizing proteins to their target sequence, would

7

allow stable expression of both sense and antisense transcripts. No experimental evidence for such a phenomenon exists so far. However, the proposal for positive regulation of gene expression by an antisense transcript was put forward for the bcl-2/IgH fusion locus (Capaccioli et al., 1996). Oligonucleotides directed against the putative antisense transcript arising from the translocation event led to decreased bcl-2 mRNA and protein levels, thus suggesting that the natural antisense transcript actually upregulates sense gene expression. The hypothesis proposed by the authors was that the antisense transcript impaired normal degradation of the sense transcript by masking at least part of the bcl-2 3∞UTR that contains AU-rich destabilizing elements (Chen and Shyu, 1995).

5. Conclusion Numerous examples have clearly shown that endogenous antisense RNAs are not restricted to the prokaryotic world but can also be found in all eukaryotes. They may well be the emerging tip of an iceberg of as-yet unexplored, but real, general antisense-mediated gene regulation. Apart from being translated, these natural antisense transcripts may regulate the expression of their complementary sense transcripts at any step from transcription to translation, probably via either depletion of regulatory RNA-binding proteins or RNA duplex formation. They are generally believed to downregulate sense gene expression. Indeed, antisense-mediated gene regulation is a way of lowering the abundance of stable transcripts more rapidly than the cessation of transcription. However, upregulation has been proposed and may also be another general mechanism of gene regulation in eukaryotes. In addition to making sense in normal regulation of gene expression, endogenous antisense RNA may also participate in the alteration of gene regulation leading to different pathologies. Thus, understanding how natural antisense transcripts regulate gene expression is of great interest not only for fundamental knowledge of gene expression but also to comprehend the development of pathological situations.

Acknowledgement We thank Christian Doerig, Anne-Lise Haenni, Andreas Tsapis and Jennifer Richardson for critical reading of the manuscript and helpful discussions. This work has been supported by a fellowship from the French Agence Nationale de Recherche contre le SIDA (ANRS) to CV-B and by grants from ANRS to CV.

8

C. Vanhe´e-Brossollet, C. Vaquero / Gene 211 (1998) 1–9

References Adelman, J.P., Bond, C.T., Douglass, J., Herbert, E., 1987. Two mammalian genes transcribed from opposite strands of the same DNA locus. Science 235, 1514–1517. Anderson, S., Bankier, A.T., Barrell, B.G., Bruijn, M.H.L., Coulson, A.R., Drouin, J., Eperon, I.C., Nierlich, D.P., Roe, B.A., Sanger, F., Schreier, P.H., Smith, A.J., Staden, R., Young, I.G., 1981. Sequence and organization of the human mitochondrial genome. Nature 290, 457–465. Armstrong, B.C., Krystal, G.W., 1992. Isolation and characterization of complementary DNA for N-cym, a gene encoded by the DNA strand opposite to N-myc. Cell Growth Differ. 3, 385–390. Bedford, M., Arman, E., Orr, U.A., Lonai, P., 1995. Analysis of the Hoxd-3 gene: structure and localization of its sense and natural antisense transcripts. DNA Cell. Biol. 14, 295–304. Bentley, D.L., Groudine, M., 1986. A block to elongation is largely responsible for decreased transcription of c-myc in differentiated HL60 cells. Nature 321, 702–706. Bibb, M.J., Van Etten, R.A., Wright, C.T., Walberg, M.W., Clayton, D.A., 1981. Sequence and organization of the mouse mitochondrial DNA. Cell 26, 167–180. Campbell, C.E., Huang, A., Gurney, A.L., Kessler, P.M., Hewitt, J.A., Williams, B.R., 1994. Antisense transcripts and protein binding motifs within the Wilms tumour ( WT1) locus. Oncogene 9, 583–595. Capaccioli, S., Quattrone, A., Schiavone, N., Calastretti, A., Copreni, E., Bevilacqua, A., Canti, G., Gong, L., Morelli, S., Nicolin, A., 1996. A bcl-2/IgH antisense transcript deregulates bcl-2 gene expression in human follicular lymphoma t(14; 18) cell lines. Oncogene 13, 105–115. Celano, P., Berchtold, C.M., Kizer, D.L., Weeraratna, A., Nelkin, B.D., Baylin, S.B., Casero, R.J., 1992. Characterization of an endogenous RNA transcript with homology to the antisense strand of the human c-myc gene. J. Biol. Chem. 267, 15092–15096. Chang, Y., Spicer, D.B., Sonenshein, G.E., 1991. Effects of IL-3 on promoter usage, attenuation and antisense transcription of the c-myc oncogene in the IL-3-dependent Ba/F3 early pre-B cell line. Oncogene 6, 1979–1982. Chawla, A., Lazar, M.A., 1993. Induction of Rev-ErbA alpha, an orphan receptor encoded on the opposite strand of the alpha-thyroid hormone receptor gene, during adipocyte differentiation. J. Biol. Chem. 268, 16265–16269. Chen, C.Y., Shyu, A.B., 1995. AU-rich elements: characterization and importance in mRNA degradation. Trends Biochem. Sci. 20, 465–470. Croen, K.D., Ostrove, J.M., Dragovic, L.J., Smialek, J.E., Strauss, S.E., 1987. Latent herpes simplex virus in human trigeminal ganglia. N. Engl. J. Med. 317, 1427–1432. Doerig, C., Pizer, L.I., Wilcox, C.L., 1991. An antigen encoded by the Latency-Associated transcript in neuronal cell cultures latently infected with herpes simplex virus type 1. J. Virol. 65, 2724–2727. Dolnick, B.J., 1993. Cloning and characterization of a naturally occurring antisense RNA to human thymidylate synthase mRNA. Nucleic Acids Res. 21, 1747–1752. Eccles, M.R., Grubb, G., Ogawa, O., Szeto, J., Reeve, A.E., 1994. Cloning of novel Wilms tumor gene ( WT1) cDNAs; evidence for antisense transcription of WT1. Oncogene 9, 2059–2063. Farrell, C.M., Lukens, L.N., 1995. Naturally occurring antisense transcripts are present in chick embryo chondrocytes simultaneously with the down-regulation of the alpha 1 (I ) collagen gene. J. Biol. Chem. 270, 3400–3408. Fu, X.D., Maniatis, T., 1992. Isolation of a complementary DNA that encodes the mammalian splicing factor SC35. Science 256, 535–538. Hervieu, G., Nahon, J.L., 1995. Pro-melanin concentrating hormone messenger ribonucleic acid and peptides expression in peripheral tissues of the rat. Neuroendocrinology 61, 348–364.

Hildebrandt, M., Nellen, W., 1992. Differential antisense transcription from the dictyostelium EB4 gene locus: implications on antisensemediated regulation of mRNA stability. Cell 69, 197–204. Jackson, R.J., Standart, N., 1990. Do the poly(A) tail and 3∞ untranslated region control mRNA translation? Cell 62, 15–24. Kerr, I.M., 1987. The 2-5A system: a personnal view. J. Interferon Res. 7, 505–510. Khochbin, S., Lawrence, J., 1989. An antisense RNA involved in p53 mRNA maturation in murine erythroleukemia cells induced to differentiate. EMBO J. 8, 4107–4114. Khochbin, S., Brocard, M.P., Grunwald, D., Lawrence, J.J., 1992. Antisense RNA and p53 regulation in induced murine cell differentiation. Ann. NY Acad. Sci. 660, 77–87. Kim, J.H., McLinden, R.J., Mosca, J.D., Vahey, M.T., Greene, W.C., Redfield, R.R., 1996. Inhibition of HIV replication by sense and antisense rev response elements in HIV-based retroviral vectors. J. Acquir. Defic. Syndr. Hum. Retrovirol. 12, 343–351. Kimelman, D., Kirschner, M.W., 1989. An antisense mRNA directs the covalent modification of the transcript encoding fibroblast growth factor in Xenopus oocytes. Cell 59, 687–696. Kindy, M.S., McCormack, J.E., Buckler, A.J., Levine, R.A., Sonensheim, G.E., 1987. Independent regulation of transcription of the two strands of the c-myc gene. Mol. Cell. Biol. 7, 2857–2862. Knee, R.S., Pitcher, S.E., Murphy, P.R., 1994. Basic fibroblast growth factor sense (bFGF ) and antisense (gfg) RNA transcripts are expressed in unfertilized human oocytes and in differentiated adult tissues. Biochem. Biophys. Res. Commun. 205, 577–583. Krystal, G.W., Armstrong, B.C., Battey, J.F., 1990. N-myc mRNA forms an RNA–RNA duplex with endogenous antisense transcripts. Mol. Cell. Biol. 10, 4180–4191. Laabi, Y., Gras, M.P., Brouet, J.C., Berger, R., Larsen, C.J., Tsapis, A., 1994. The BCMA gene, preferentially expressed during B lymphoid maturation, is bidirectionally transcribed. Nucleic Acids Res. 22, 1147–1154. Lagunoff, M., Roizman, B., 1994. Expression of a herpes simplex virus 1 open reading frame antisense to the gamma(1)34.5 gene and transcribed by an RNA 3∞ coterminal with the unspliced latencyassociated transcript. J. Virol. 68, 6021–6028. Lankenau, S., Corces, V.G., Lankenau, D.H., 1994. The Drosophila micropia retrotransposon encodes a testis-specific antisense RNA complementary to reverse transcriptase. Mol. Cell. Biol. 14, 1764–1775. Larocca, D., Chao, L.A., Seto, M.H., Brunck, T.K., 1989. Human T-cell leukemia virus minus strand transcription in infected T- cells. Biochem. Biophys. Res. Commun. 163, 1006–1013. Lazar, M.A., Hodin, R.A., Darling, D.S., Chin, W.W., 1989. A novel member of the thyroid/steroid hormone receptor family is encoded by the opposite strand of the rat c-erbAa transcriptional unit. Mol. Cell. Biol. 9, 1128–1136. Lazar, M.A., Hodin, R.A., Cardona, G., Chin, W.W., 1990. Gene expression from the c-erbAa/Rev-ErbAa genomic locus. J. Biol. Chem. 265, 12859–12863. Lee, R.C., Feinbaum, R.L., Ambros, V., 1993. The C. elegans heterochronic gene lin-4 encodes small RNAs with antisense complementarity to lin-14. Cell 75, 843–854. Lerner, A., D’Adamio, L., Diener, A.C., Clayton, L.K., Reinherz, E.L., 1993. CD3e/g/h locus is colinear with and transcribed antisense to the gene encoding the transcription factor Oct-1. J. Immunol. 151, 3152–3162. Li, A.W., Seyoum, G., Shiu, R.P., Murphy, P.R., 1996. Expression of the rat bFGF antisense RNA transcript is tissue-specific and developmentally regulated. Mol. Cell Endocrinol. 118, 113–123. Malik, K.T., Wallace, J.I., Ivins, S.M., Brown, K.W., 1995. Identification of an antisense WT1 promoter in intron 1: implications for WT1 gene regulation. Oncogene 11, 1589–1595. Michael, N.L., Vahey, M.T., d’Arcy, L., Ehrenberg, P.K., Mosca, J.D., Rappaport, J., Redfield, R.R., 1994. Negative-strand RNA tran-

C. Vanhe´e-Brossollet, C. Vaquero / Gene 211 (1998) 1–9 scripts are produced in human immunodeficiency virus type 1-infected cells and patients by a novel promoter downregulated by Tat. J. Virol. 68, 979–987. Miyajima, N., Horiuchi, R., Shibuya, Y., Fukushige, S., Matsubara, K., Toyoshima, K., Yamamoto, T., 1989. Two erbA homologs encoding proteins with different T3 binding capacities are transcribed from opposite DNA strands of the same genetic locus. Cell 57, 31–39. Moss, E.G., Lee, R.C., Ambros, V., 1997. The cold shock domain protein LIN-28 controls developmental timing in C. elegans and is regulated by the lin-4 RNA. Cell 88, 637–646. Munroe, S.H., Lazar, M.A., 1991. Inhibition of c-erbA mRNA splicing by a naturally occurring antisense RNA. J. Biol. Chem. 266, 22083–22086. Murphy, P.R., Knee, R.S., 1994. Identification and characterization of an antisense RNA transcript (gfg) from the human basic fibroblast growth factor gene. Mol. Endocrinol. 8, 852–859. Nellen, W., Lichtenstein, C., 1993. What makes an mRNA antisense-itive? Trends Biochem. Sci. 18, 419–423. Nepveu, A., Marcu, K.B., 1986. Intragenic pausing and anti-sense transcription within the murine c-myc locus. EMBO J. 5, 2859–2865. Noguchi, M., Miyamoto, S., Silverman, T.A., Safer, B., 1994. Characterization of an antisense Inr element in the eIF-2a gene. J. Biol. Chem. 269, 29161–29167. Okano, H., Aruga, J., Nakagawa, T., Shiota, C., Mikoshiba, K., 1991. Myelin basic protein gene and the function of antisense RNA in its repression in myelin-deficient mutant mouse. J. Neurochem. 56, 560–567. Prang, N., Wolf, H., Schwarzmann, F., 1995. Epstein–Barr virus lytic replication is controlled by posttranscriptional negative regulation of BZLF1. J. Virol. 69, 2644–2648. Proud, C.G., 1995. PKR/ a new name and new roles. Trends Biochem. Sci. 20, 241–246. Saccomanno, L., Bass, B.L., 1994. The cytoplasm of Xenopus oocytes contains a factor that protects double-stranded RNA from adenosine-to-inosine modification. Mol. Cell. Biol. 14, 5425–5432. Sadiq, M., Hildebrandt, M., Maniak, M., Nellen, W., 1994. Develop-

9

mental regulation of antisense-mediated gene silencing in Dictyostelium. Antisense Res. Dev. 4, 263–267. Sharma, H.W., Narayanan, R., 1995. The therapeutic potential of antisense oligonucleotides. BioEssays 17, 1055–1063. Silverman, T.A., Noguchi, M., Safer, B., 1992. Role of sequences within the first intron in the regulation of expression of eukaryotic initiation factor 2a. J. Biol. Chem. 267, 9738–9742. Skeiky, Y.A., Iatrou, K., 1990. Silkmoth chorion antisense RNA. J. Mol. Biol. 213, 53–66. Spencer, C.A., Gietz, R.D., Hodgetts, R.B., 1986. Overlapping transcription units in the dopa decarboxylase region of Drosophila. Nature 322, 279–281. Spicer, D.B., Sonenshein, G.E., 1992. An antisense promoter of the murine c-myc gene is localized within intron 2. Mol. Cell Biol. 12, 1324–1329. Stevens, J.G., Wagner, E.K., Devi-Rao, G.B., Cook, M.L., Feldman, L.T., 1987. RNA complementary to a herpesvirus a gene mRNA is prominent in latently infected neurons. Science 235, 1056–1059. Stolt, P., Zillig, W., 1993. Antisense RNA mediates transcriptional processing in an archaebacterium, indicating a novel kind of RNase activity. Mol. Microbiol. 7, 875–882. Tosic, M., Roach, A., de Rivaz, J.C., Dolivo, M., Matthieu, J.M., 1990. Post-transcriptional events are responsible for low expression of myelin basic protein in myelin deficient mice: role of natural antisense RNA. EMBO J. 9, 401–406. Vanhe´e-Brossollet, C., Thoreau, H., Serpente, N., d’Auriol, L., Levy, J.P., Vaquero, C., 1995. A natural antisense RNA derived from the HIV-1 env gene encodes a protein which is recognized by circulating antibodies of HIV+ individuals. Virology 206, 196–202. Wagner, R.W., Smith, J.E., Cooperman, B.S., Nishikura, K., 1989. A double-stranded RNA unwinding activity introduces structural alterations by mean of adenosine to inosine conversions in mammalian cells and Xenopus eggs. Proc. Natl. Acad. Sci. USA 86, 2647–2651. Wagner, E.G., Simons, R.W., 1994. Antisense RNA control in bacteria, phages, and plasmids. Annu. Rev. Microbiol. 48, 713–742. Wightman, B., Ha, I., Ruvkun, G., 1993. Posttranscriptional regulation of the heterochronic gene lin-14 by lin-4 mediates temporal pattern formation in C. elegans. Cell 75, 855–862.