| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Department of Microbiology and Immunology and Sylvester Comprehensive Cancer Center, University of Miami School of Medicine, Miami, Florida, USA
1Correspondence: Room 511 Papanicolaou Building, 1550 NW 10th Ave., M710, University of Miami School of Medicine, Miami, FL 33136, USA. E-mail: gbarber{at}med.miami.edu
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONEUKARYOTIC DRBPS: THE CYTOPLASMNUCLEAR EUKARYOTIC DRBPSEDITING dsRNAA FAMILY OF dsRNA...VIRAL DRBPSINTERACTION AMONG DRBP FAMILY...ALTERNATIVE dsRNA BINDING MOTIFSCONCLUSIONSREFERENCES |
|---|
Key Words: mRNA • transcription • RNA editing • RNAi
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONEUKARYOTIC DRBPS: THE CYTOPLASMNUCLEAR EUKARYOTIC DRBPSEDITING dsRNAA FAMILY OF dsRNA...VIRAL DRBPSINTERACTION AMONG DRBP FAMILY...ALTERNATIVE dsRNA BINDING MOTIFSCONCLUSIONSREFERENCES |
|---|
65–68 amino acids (1)
. Eukaryotic DRBPs can contain up to five DRBDs, while other DRBPs, most notably those encoded by viruses, usually contain only one (2)
. The growing number of DRBPs indicates that dsRNA/protein interactions likely provide a varied and vital means of controlling gene expression in the cell. Indeed, genetic experiments with mice indicate that homozygous or even heterozygous disruption of the DRBPs such as ADAR1 (adenosine deaminase acting on RNA) and SPNR (spermatid perinuclear RNA binding protein) result in embryonic lethality (3
, 4)
. Whereas this emphasizes the critically important roles performed by many of the DRBPs, specific cellular dsRNA targets remain to be identified. It is plausible that many if not the majority of cellular RNA species comprise some form of dsRNA structure such as internal loops or bulges at one or more stages of their existence and, as a consequence, almost certainly come into contact with DRBPs. Some of the DRBPs such as DICER, the NFARs (nuclear factors associated with dsRNA), ADARs, and TRBP (TAR RNA binding protein) reside in the nucleus and probably function in RNA interference (RNAi), mRNA elongation, editing, stability, splicing, and/or export. In contrast, cytoplasmic members of the DRBPs (including PKR and PACT) have been reported to function in the regulation of translation, dsRNA signaling events, and host defense (5
, 6)
. Some of these DRBPs have been identified using similar screening procedures. For example, SPNR, PRBP, and TENR were all isolated through their ability to interact with the 3'-untranslated region (UTR) of murine protamine 1 gene (prm-1; refs 7
8
9
). However, it remains unclear whether these different DRBPs perform similar functions in the cell.
The DRBDs do not recognize specific nucleotide sequences and interact primarily with A-form double helix RNAs, which differs from the typical dsDNA B-form helix in that the minor groove is shallow and broad and the major groove is narrow and deep (1
, 2
, 10)
. This conformation allows DRBPs to bind nonspecifically to dsRNAs and ssRNAs with extensive secondary structures and ignore ssRNA, dsDNA, or ssDNA species. Indeed, the lack of nucleotide binding recognition suggests that target specificity may generally be governed through interactions with other proteins since many DRBPs will strongly bind to any dsRNA structure nonspecifically in vitro (2)
. Biochemical evidence indicates that 11–16 bp of dsRNA can span a single DRBD (11)
. Three-dimensional models of DRBDs, including those of Staufen, PKR, and RNase III, show that the motif folds in a compact 
ßßß structure, in which the -helical surface of the DRBD packs through a conserved hydrophobic core against an antiparallel ß-sheet (12)
. Nonsequence specific contacts involving 2' OH and phosphate groups likely mediate interaction between the DRBD and dsRNA. Mutation of the most conserved residues within the consensus DRBD decreases or abolishes dsRNA binding (13
, 14)
. It is not clear why some DRBDs, such as Staufen, have more than one DRBD; but some motifs have been reported to bind more efficiently to dsRNA than others, and it is postulated that low-affinity DRBDs may help in stabilizing DRBP/dsRNA conformations (15)
.
A well-characterized source of dsRNA occurs after virus infection (16)
. These structures arise directly from viral genomes comprising dsRNA, through ssRNA virus genome replication or as by-products of complementary mRNAs encoded on opposite strands of DNA viruses that anneal to form dsRNA structures (17
, 18)
. SsRNAs with extensive secondary structures such as hairpin loop formations are also effective sources of dsRNA (19)
. It is not surprising that dsRNA is a potent activator of the host’s response to virus infection. For example, dsRNA can induce the NF-
B pathways and activate interferon (IFN) through toll receptor 3 and has an inhibitory effect on protein synthesis largely through its activation of PKR and the 2–5A synthetase family (20
21
22
23)
. DsRNA is known to be a potent inducer of apoptosis, an effect exacerbated by pretreating cells with IFN. Because of these many potent effects, dsRNA has been evaluated as a potential antitumor therapy as well as an antiviral agent with contrasting success (24
25
26
27
28
29
30)
. It is now clear that dsRNA can trigger multiple cellular signaling pathways in the cell. Thus, DRBDs are important recognition motifs that facilitate interaction with RNA to assist in the regulation of gene expression. Here we present a review to help clarify the classification of known DRBPs, discuss uncharacterized DRBPs, and provide information on their diverse function (Table 1
).
|
|
EUKARYOTIC DRBPS: THE CYTOPLASM |
|---|
|
TOPABSTRACTINTRODUCTIONEUKARYOTIC DRBPS: THE CYTOPLASMNUCLEAR EUKARYOTIC DRBPSEDITING dsRNAA FAMILY OF dsRNA...VIRAL DRBPSINTERACTION AMONG DRBP FAMILY...ALTERNATIVE dsRNA BINDING MOTIFSCONCLUSIONSREFERENCES |
|---|
32
33)
. Purification studies using synthetic dsRNA and antibody affinity columns led to the isolation of a cellular kinase from IFN-treated cell extracts that could be activated after the addition of dsRNA. The kinase, now referred to as PKR, eventually cloned by screening expression libraries, was found to be IFN inducible and could potently inhibit protein synthesis (34
, 35)
.
Considerable evidence confirmed that PKR plays an important role in host defense against virus infection, as described in more detail below. Computer analysis and direct experimental studies in the early 1990s indicated that PKR contained two DRBDs in its amino terminus that were also found in proteins such as Drosophila Staufen and Xenopus rbpa, prototypes of the DRBP family (1
, 36)
. In addition to the two DRBDs, PKR, which exists in humans as a 551 amino acid protein encoded from a single gene located on chromosome 2p21, contained a serine/threonine kinase domain located in the carboxyl terminus (35)
. Substantial evidence now indicates that interaction with dsRNA causes PKR to form homodimers and to autophosphorylate, in trans, on multiple serine/threonine residues (37
, 38)
. After autophosphorylation, PKR is able to catalyze the phosphorylation of target substrates, the most well characterized being the eIF2 subunit on Ser 51 (39)
. Phosphorylated eIF2 sequesters eIF2B, a rate-limiting component of translation, leading to a dramatic inhibition of protein synthesis in the cell (11
, 37
, 40
41
42)
.
Similar to other DRBPs, PKR’s interaction with activator RNA is independent of any specific RNA nucleotide motif or sequences (43)
. However, in vitro, the concentration of activator does play a role in the autophosphorylation of PKR, with low levels of dsRNA potently activating the kinase and higher concentrations of dsRNA being inhibitory (44)
. Possibly high concentrations of dsRNA prevent PKR activity by impeding intermolecular interactions and trans-autophosphorylation (11)
. Although as little as one helical turn of dsRNA (about <11 bp) has been reported to interact with a single DRBD of PKR, optimal activation requires that both DRBDs of PKR cooperate to form a single binding site that extends to interact with 30–80 bp of dsRNA (11
, 45)
. The DRBDs of PKR may negatively regulate PKR, since it appears that in the absence of dsRNA, DRBD2 of PKR interacts with the catalytic domain at the carboxyl terminus (46
47
48
49)
. Upon binding to dsRNA, a conformational change is induced that releases the interaction of DRBD2 with the carboxyl-terminal catalytic domain, allowing PKR to autophosphorylate and subsequently catalyze the phosphorylation of eIF2 (50
, 51)
.
Extensive data show that PKR functions in host defense against virus infection, and many virus-specific RNAs are probably capable of activating PKR, which would inhibit translation via eIF2 phosphorylation. To counteract the consequences of triggering PKR action, numerous viruses have devised mechanisms to inhibit the kinase, thus preventing the inhibition of protein synthesis that would be detrimental to their replication (52)
. Vaccinia virus, for example, has been reported to encode two products that impede PKR activity. One protein, referred to as E3L, itself contains a DRBD and is primarily detected in the nucleus where it competes for dsRNA activators and may even bind to and inhibit the kinase through dsRNA bridging (53
, 54)
. A second vaccinia protein, referred to as K3L, shares homology to the known PKR substrate eIF2 and is thought to function by competitively sequestering the kinase (55
56
57)
. Studies using PKR null mice or MEFs lacking PKR activity have shown that this kinase is a key component of the host early defense system that acts early in innate immunity before activation of the IFN system and the acquired immune response (58)
. PKR null mice are susceptible to normally nonlethal doses of vesicular stomatitis virus and show increased sensitivity to influenza infection (58
, 59)
.
Aside from their susceptibility to certain viruses, PKR knockout mice are developmentally normal. There is no impairment of type I IFN gene induction by dsRNA or virus in many mouse organs, suggesting that PKR does not play a major role in dsRNA/virus-mediated induction of IFN-ß in vivo. Although the induction of type I IFN has been reported to be impaired in PKR -/- MEFs in response to dsRNA, this effect can be corrected if the cells are primed with IFN- (60)
. PKR has also been implicated in affecting the induction of NF-B, a key regulator of gene expression, after exposure to viral infection and dsRNA (60
61
62)
. This may occur by PKR influencing the activity of the IB kinase (IKK) complex leading to the phosphorylation of IKKß and degradation of IB (63
, 64)
. However, PKR’s exact role in the regulation of NF-B remains to be clarified since recent studies using PKR null MEFs concluded that PKR was not solely required for dsRNA-induced activation of NF-B in response to dsRNA nor of IFN-ß, interleukin 6 (IL-6), or TNF- mRNA (65)
. PKR has been suggested to play a role in the dsRNA-signaling cascade that leads to the activation of c-Jun NH2-terminal kinase and may play a role in the antiviral activity mediated by type II IFN (IFN-
; ref 66
).
Initial experiments to study the function of PKR included attempting to overexpress the kinase in mammalian cells. These studies revealed that PKR exerted a strong growth-suppressive and toxic effect on the host cell almost certainly through translation suppression (67)
. In contrast, expression of a catalytically inactive PKR variant led to malignant transformation of NIH 3T3 cells and tumorigenicity in nude mice, suggesting that PKR may function as a suppressor of cell proliferation and tumorigenesis (68
, 69)
. Nevertheless, PKR-deficient animals do not appear to be susceptible to increased tumor development (59
, 60)
. Evidence indicates that the growth suppressive properties of PKR may involve the induction of apoptosis (54
, 70
71
72)
. Mouse 3T3 L1 fibroblasts that inducibly overexpress human PKR are sensitive to dsRNA-mediated and viral-induced cell death, whereas expression of a dominant-negative PKR variant blocked this effect (73)
. PKR-dependent apoptosis is perhaps mediated through the death adaptor protein FADD, although the exact mechanism remains to be clarified (73)
. Thus, PKR appears to be sensor for dsRNA in the cell and can regulate translation and/or induction of apoptosis largely through interacting with eIF2, although other substrates of PKR almost certainly exist and await further characterization.
TAR RNA binding protein (TRBP)
Attempts to identify cellular proteins that could interact with HIV-1 trans-activating region (TAR) RNA led to the isolation of a 313 amino acid product referred to as the TAR RNA binding protein (TRBP; ref 74
). Subsequently, TRBP was found in a screen designed to isolate proteins that could interact with the Rev-responsive element RNA of HIV-1 (75)
. Analysis indicated that TRBP contained three typical DRBDs, with no other identifiable domains being apparent (Fig. 1
). In HeLa cells, TRBP was shown to be present in the nucleus, as well as the cytoplasm, where it specifically associated with ribosomes and the endoplasmic reticulum (76)
. Although studies indicated that TRBP may play a role in the activation of HIV-1 in part by assisting HIV-encoded Tat to bind to TAR RNA sequences in the long terminal repeat (LTR), several studies indicated that TRBP may play a role in the regulation of translation through modulating PKR (74)
. TRBP can inhibit the autophosphorylation of PKR in vitro, possibly by competing for dsRNA activator, thereby preventing PKR autophosphorylation (75)
. It has been reported that TRBP can bind PKR in an RNA binding independent manner to inhibit the growth regulatory properties of the kinase (75
, 77
, 78)
. Overexpression of TRBP in NIH 3T3 cells leads to a transformed phenotype arguably due to counteracting PKR’s regulation of cell growth (75
, 78)
. Three TRBP species of 50, 43, and 40 kDa have actually been reported in mammalian cells, two of which appear to be generated from a single gene on human chromosome 12 through alternative splicing events that vary in the 5' nucleotide sequence (79
, 80)
. The third TRBP species appears to be the product of a processed pseudogene and maps to chromosome 8 (80)
.
|
Further clues as to the role of TRBP came from the use of mouse genetic techniques. The mouse homologue of TRBP, referred to as Prm-1 RNA binding protein (PRBP), was isolated through its interaction with the 3'-UTR of the mouse protamine gene (Prm-1) and is 93% identical at the amino acid level to human TRBP (8
, 81)
. PRBP has been shown to be a highly expressed cytoplasmic protein involved in the general repression of translation in testis, which is a requirement for functional spermiogenesis (8)
. Human TRBP has been reported in the cytoplasm of elongating spermatids at steps 3–4 of spermiogenesis (82)
. Mice lacking PRBP have been generated, though usually die at the time of weaning for unknown reasons, clearly implying a role for PRBP in functions other than those involved in spermatogenesis (83)
. However, PRBP -/- animals can be maintained with a dietary supplement, revealing that surviving males are severely oligospermic and sterile. Alternate studies using a Prm1 3'-UTR transgenic mouse model have complemented mouse knockout studies by demonstrating that PRBP is required for the translational activation of protamine mRNAs. Thus, TRBP/PRBP may be required for maintenance and assembly of specific, translationally regulated ribonucleoprotein particles in spermatogenesis (83)
. The severe phenotype of PRBP null mice indicates that this protein is almost certainly involved in processes other than spermatogenesis and may be usurped to play a role in regulating HIV expression. Protein partners other than PKR remain unknown. It is not clear whether PKR, reportedly regulated by TRBP/PRBP, may play a role in spermatid development. However, PKR-deficient mice are viable and apparently have normal spermiogenesis.
Protein activator of PKR (PACT)
The use of yeast two-hybrid screens using PKR as bait led to the isolation of a protein referred to as PACT. Human PACT was found to be a 35 kDa protein that contains three DRBDs similar to TRBP (Fig. 1)
. Studies indicated that PACT could induce the autophosphorylation of PKR in a dsRNA-independent manner (6)
. Transient overexpression of PACT in yeast and mammalian cells, in conjunction with cellular stress, was found to enhance the level of PKR autophosphorylation and to induce the phosphorylation of eIF2 (6)
. In response to diverse stress signals, such as serum starvation and arsenite treatment, PACT becomes phosphorylated and increases its affinity with PKR, leading to enhanced phosphorylation of eIF2 and apoptosis (84)
. The first and second DRBDs of PACT have been shown to bind dsRNA and to mediate the interaction with PKR. However, the third DRBD was reported unable to bind strongly with dsRNA but was found to be required for triggering PKR activation (85)
. The role of PACT in dsRNA signaling has been addressed, and although overexpression of PACT alone has not been found to effect IFN-ß expression, PACT could augment IFN-ß gene activation induced by Newcastle disease virus (NDV). This event coincided with an increase in activation of NF-B, IRF-3, and IRF-7 and was specific to NDV, since gene induction by dsRNA, EMCV, or Sindbis did not appear to be affected (86)
. PACT localizes in a diffuse cytoplasmic pattern, but after NDV infection becomes associated with a cytoplasmic structure thought be part of a viral replication complex (86)
.
The murine homologue of PACT, referred to as PKR activator X (RAX), was isolated in a yeast two-hybrid screen using PKR as bait. Similar to PACT, cell stress signals were found to induce rapid phosphorylation of RAX and to mediate association with and activation of PKR (87)
. RAX is phosphorylated by an unknown stress-activated protein kinase (SAPK). Ceramide was suggested to perhaps play a role in inhibiting protein synthesis by activating a SAPK that phosphorylates RAX, leading to the activation of PKR and regulation of translation through phosphorylation of eIF2 (88)
.
Xlrbpa is a ubiquitously expressed 33 kDa protein originally characterized as the Xenopus homologue of TRBP (76)
. However, although Xlrbpa is 48% identical at the amino acid level to human and mouse TRBP/PRBP, it shares 60% amino acid identity to PACT and RAX. Xlrbpa has been reported to be highly expressed in oocytes, where it localizes mainly in the cytoplasm, though has also been found in small nuclear domains. Xlrbpa has been found to associate with ribosomes in the cytoplasm, localize with rRNAs in nucleoli, and immunoprecipitate with heterogeneous nuclear (hn)RNP particles (76)
. Drosophila also have an apparent homologue, CG6866, that shares 33% amino acid identity with PACT, RAX, and Xlrbpa and 29% identity with TRBP and PRBP (Table 1)
. Data would indicate that PACT/RAX may play a role in responding to stress signals such as virus infection and may modulate host defense in vertebrates through regulating the activity of IFN and PKR.
Staufen
The DRBP Staufen was originally isolated in 1991 from Drosophila and is a 1026 amino acid protein important for oocyte development (89)
. Staufen functions in mRNA localization and translational regulation of select mRNAs in Drosophila. Staufen was reported to anchor bicoid mRNA at the anterior of the Drosophila oocyte and oskar mRNA to the posterior of the oocyte, but at two distinct stages of development (89
90
91)
. Staufen is conserved from Caenorhabditis elegans to humans and has four or five DRBDs, depending on the species. For Drosophila Staufen, only DRBDs 1, 3, and 4 appear to bind dsRNA strongly in vitro (1
, 92)
. In the case of DRBD2, the inability to interact with dsRNA is due to the presence of a conserved proline-rich insertion that splits the domain in half. Removal of this insertion, however, allows DRBD2 to again bind dsRNA (92)
. DRBD5 does not strongly interact with RNA, due to being less conserved, and is thought to function via protein–protein interactions, although the exact nature of these associations remains to be determined. DRBD2 has been reported to be required for microtubule-dependent localization of oskar mRNA, whereas DRBD5 is required for the activation of oskar mRNA translation after localization. DRBDs 2 and 5 are necessary for the anchoring of bicoid mRNA whereas DRBD5 also directs the actin-dependent localization of prospero mRNA (92)
.
Some Staufen homologues have now been isolated, including those from human (hStau), mouse (mStau), and rat (rStau). Using human ESTs with sequence similarity to dStau as probes to screen cDNA libraries, hStau was isolated and shown to lack the first DRBD, present in dStau, and to contain a putative microtubule binding domain that binds tubulin (93)
. Staufen has been shown to bind dsRNA through its two full-length DRBDs (2+3) since DRBDs 4 and 5 are not strongly conserved (93)
. The human gene, which is 38% identical at the amino acid level to Drosophila Staufen, was localized to chromosome 20q13.1 and gives rise to four potential spliced variants, which encode two proteins of 60 and 65 kDa that differ in their amino terminus (93
94
95)
. hStau is ubiquitously expressed and localizes to the rough endoplasmic reticulum, microtubules, and polysomes, implicating hStau in transporting mRNAs to sites of translation (93
, 95)
. hStau has been reported to become incorporated into retroviruses, including HIV-1, and has even been suggested to play a role in genomic RNA encapsidation (96)
.
mStau is 91% identical to the human protein and has been shown to bind dsRNA and tubulin (93)
. A spliced variant has been cloned, referred to as mStaui, that contains an 18 bp insertion within the third DRBD (97)
. mStaui is ubiquitously expressed at slightly lower levels than mStau and is impaired in its dsRNA binding activity. mStau is a component of RNA–protein complexes, which consist of Staufen isoforms associated with RNAs. Association with mStaui has been reported to reduce the RNA content of these complexes (97)
.
rStau was detected in hippocampal neurons associated with large RNP complexes, which suggested a potential role for this protein in neuronal mRNA transport (98
, 99)
. mStau is enriched in the vicinity of the smooth endoplasmic reticulum and microtubules near synaptic contacts and is associated with RNPs in distal dendrites known to contain mRNA, ribosomes, and translation factors (99)
. It has been proposed that in developing neurons, rStau might be responsible for the formation of large RNPs that transport specific mRNAs along microtubules to the elaborating dendrite; in differentiated neurons rStau might transport specific mRNAs to the synapse to produce synaptic microenvironments through local protein synthesis (100)
.
Thus, Staufen plays an important role in the localization of maternal determinants such as bicoid RNA in Drosophila oogenesis and likely plays a role in the transportation and localization of mRNAs in mammalian cells. The importance of Staufen in mammalian development and its cellular functions awaits further genetic experiments, including elimination of the Staufen gene in mice.
|
NUCLEAR EUKARYOTIC DRBPS |
|---|
|
TOPABSTRACTINTRODUCTIONEUKARYOTIC DRBPS: THE CYTOPLASMNUCLEAR EUKARYOTIC DRBPSEDITING dsRNAA FAMILY OF dsRNA...VIRAL DRBPSINTERACTION AMONG DRBP FAMILY...ALTERNATIVE dsRNA BINDING MOTIFSCONCLUSIONSREFERENCES |
|---|
, 102)
. Database analysis and experimentation indicated that NF90 appeared to be a 90 kDa product that contained two DRBDs that could indeed interact with dsRNA (103)
. Independent studies using the yeast two-hybrid system and PKR as bait isolated a protein referred to as DRBP76/NFAR, which was noted to share amino acid similarity with the amino terminus of NF90 (104
105
106)
. However, subsequent studies of NFAR clones isolated from two-hybrid screens using PKR as bait indicated that NFAR actually existed as two related species, NFAR-1 and 2, which were identical at the amino terminus of the open reading frame but had unique carboxyl termini (Fig. 1
; ref 107
). Upon comparison of cDNA clones and through sequencing the entire NFAR genomic region, it was confirmed that a single NFAR gene exists on chromosome 19 that generates two alternatively spliced variants (107)
. NFAR-1 is a 90 kDa protein that is 99% identical to DRBP76; NFAR-2 exists as a 110 kDa protein and is related to the NF-90 clone, the latter product probably being artificial due to the introduction of two nucleotides (nt 1798 and 1799) in the NF-90 coding sequence that shifts the reading frame and results in premature termination of translation (106
107
108)
. NFAR-1 and NFAR-2 share 98% homology, though NFAR-2 lacks exon 18 and encodes three novel exons 19
20
21
.
The NFAR proteins are ubiquitously expressed in the nucleus of many cell types and tissues and do not appear to be exclusive to T cells. Moreover, even though NF-90 was reported to interact specifically with NF-AT DNA, these proteins do not exhibit any typical DNA binding motifs nor have they been demonstrated to play a role in IL-2 transcription, although it was recently suggested that NF90 might play a role in the stability of IL-2 mRNA in T cells after their activation (109)
. Aside from physically associating with PKR, NFAR-1 and 2 appear to be substrates for PKR at least in vitro, indicating they may function in PKR-mediated signaling events in the cell (106)
. Phosphorylation by PKR does not require the DRBDs of the NFAR proteins (107)
. Whereas adenovirus VA RNAI has been shown to bind to and inhibit PKR, another adenovirus structural RNA, VA RNAII, was identified as binding to NF90, perhaps to prevent its function (104)
. In transfected cells, both NFARs were found to regulate the transcription of cotransfected reporter genes, probably at the level of mRNA elongation and/or splicing (106)
. It is not clear why two NFAR proteins exist, but both typically appear present in relatively stoichiometric amounts in the nucleus of cells. Transfection studies indicate that NFAR-2 rather than NFAR-1 appeared to have a more potent ability to regulate gene transcription. The carboxyl-terminal region unique to NFAR-2 has been shown to interact with proteins such as p68, SMN, and FUS, which are implicated in mRNA splicing. In support of this, NFAR-2 has recently been identified as a component the spliceosome. Collectively, these data would indicate that the NFARs may be involved in mRNA processing events in the cell (106
, 110
111
112)
.
A mouse homologue of the NFARs, termed mILF3, was cloned and mapped to chromosome 9 near D9Mit160, a region that is conserved with human chromosome 19p13.2 (107
, 113
, 114)
. mILF3 shares 91% amino acid identity with NFAR-2 (106
, 113)
. The NFARs also appear to have homologues in Xenopus, with Xenopus 4F.1 being 74% identical at the amino acid level to NFAR-1; the partial clone 4F.2 is 61% identical at the amino acid level to NFAR-2 (106
, 115)
. The Xenopus proteins were isolated in a screen for DRBPs and shown to bind dsRNA as well as RNA-DNA hybrids (115)
. The full-length Xenopus 4F.2, also referred to as CCAAT box transcription factor (CBTF), has been described as a maternal DNA binding protein that is essential for GATA-2 transcription during early development (116)
. It was shown that 4F.2/CBTF translocates from the cytoplasm to the nucleus in embryonic cells between stages 8 and 9 of development, and both the CBTF mRNA and protein are expressed in the ectoderm and mesoderm (116)
. An NFAR-1 homologue in Xenopus was found to be associated with mRNAs and to sediment with translationally quiescent mRNP complexes (117)
. The ability of the 4F.2/CBTFs to bind RNA maintains these proteins in the cytoplasm before their nuclear translocation at the midblastula transition corresponded to the degradation of maternal mRNAs (117)
. Thus, similar to the Staufen proteins, it may be that the Xenopus homologues of NFAR transport, localize, or translationally repress maternal mRNAs in the cytoplasm in the early embryo. After degradation of maternal mRNAs, the proteins may translocate to the nucleus where they activate transcription of GATA-2 (117)
. The importance of the NFARs in development is underscored by data from our own laboratory indicating that heterozygote deletion in mice results in embryonic lethality (L. Saunders and G. N. Barber, unpublished results).
Spermatid perinuclear RNA binding protein (SPNR)
Translational regulation of spermatid-specific protamine 1 (Prm-1) mRNA is critically important in male germ cell development (118)
. To identify proteins that may play a role in the translational control of Prm-1, the 3'-UTR of the Prm-1 transcript was used to isolate RNA interacting proteins, one of which was termed PRBP, the murine homologue of TRBP, as described above. Another DRBP found to interact with Prm-1 3'-UTR was a 71 kDa protein termed spermatid perinuclear RNA binding protein (SPNR). SPNR was subsequently found to contain two DRBDs toward its carboxyl terminus as well as a putative leucine zipper motif (7)
. SPNR was found to have an amino-terminal domain with high homology to the NFAR proteins and to the carboxyl terminus of a protein called ZFR (119)
. The single murine SPNR gene has been localized to chromosome 2 region C1 whereas the human homologue assigns to chromosome 9q34 (7)
. Similar to other DRBPs, SPNR binds to dsRNA structures nonspecifically in vitro. Several SPNR mRNA species are highly expressed in testis, ovary, and brain, with lower amounts detected in several other tissues. Although it was reported that SPNR protein expression is limited to testis, perhaps localizing to cytoplasmic microtubules, possibly functioning in transporting mRNAs along the manchette, SPNR homozygous null mice exhibit retarded growth and have a high rate of mortality (4
, 7
, 120)
. Defects in spermatogenesis, reduced fertility, and neurological defects are characteristic of mice lacking SPNR function, again indicating other roles for SPNR outside of sperm development.
A rat homologue of the murine SPNR protein, referred to as p74, has been isolated in two-hybrid yeast screens using PKR as bait, similar to the way PACT was discovered (121)
. Coexpression in yeast of p74 with a catalytically inactive PKR variant (K296R) led to an abnormal morphology and cell death, and it was suggested that in mammalian cells the competition between p74-PKR heterodimers and PKR-eIF2 could potentially regulate cell growth (121)
. Despite rSPNR sharing 92% identity with mSPNR, characterized as being testis specific, detailed Northern blot analysis revealed the possible existence of several widely expressed p74 transcripts, which might explain the severe phenotype in SPNR–/– mice (4
, 121)
. Thus, SPNR is almost certainly essential for normal spermatogenesis and sperm function, perhaps through regulating the translation of selected mRNAs, although it may play a role in PKR-mediated cell-growth similar to PACT and TRBP.
SPNR shares 85% sequence identity with the NFARs outside of the DRBDs in an upstream conserved region (USCR) encompassing amino acids 193–288 of NFAR and 185–280 of SPNR (Fig. 2
; refs 106
, 121
). This USCR is also found in ZFR, a protein that binds dsRNA through a C2H2 type zinc finger (ZF) motif, as discussed in detail later. Amino acids 1–332, 1–340, and 700–1014 of mSPNR, NFAR, and mZFR, respectively, share between 54 and 64% identity. NFAR and p74/SPNR have been shown to interact with PKR in a yeast two-hybrid assay, and NFAR is a substrate for the kinase. This conserved region may be involved in binding to PKR, indicating that SPNR and ZFR may also be substrates of PKR (106)
.
|
RNA helicase A (RHA)
RNA helicase A (RHA) is a 130 kDa DEAD (Asp-Glu-Ala-Asp) box protein originally isolated from calf thymus by assaying for RNA unwinding activity (122)
. The molecule responsible for this activity was subsequently cloned and exhibited RNA and DNA helicase activity, allowing it to unwind dsRNA in a 3' to 5' direction (123)
. RHA has two DRBDs in its amino terminus region and an arginine-glycine-rich domain (RGG box) at the carboxyl terminus (Fig. 1
; ref 124
). RHA localizes in a diffuse nucleoplasmic pattern and associates with pre-mRNA and mRNA, but does not appear to function as a component of the spliceosome (125)
. RHA mediates the interaction between CREB binding protein (CBP) and RNA pol II, and has been shown to link BRCA1 to the pol II holoenzyme (126
, 127)
. Besides binding dsRNA through two DRBDs, RHA interacts with ssRNA and ssDNA through a carboxyl-terminal RGG box (125
, 128)
. The localization of RHA in the nucleus, its association with RNA pol II, and its ability to interact with dsRNA, ssRNA, and potentially melt DNA-RNA hybrids suggest that RHA may induce local changes in chromatin to promote transcription of specific targets (126)
. The human RHA gene maps to the major susceptibility locus for prostate cancer on chromosome 1q25 and a pseudogene maps to chromosome 13q22 (129)
. Similar to NFAR (NF90), RHA has been shown to bind to adenovirus VA RNAII, leading to a decrease in its RNA helicase activity, and to bind HIV TAR RNA, similar to PKR and TRBP (104
, 130)
. Overexpression of RHA reportedly enhances viral mRNA synthesis and HIV virion production and can influence HIV LTR-directed reporter gene expression. This effect is dependent on RHA’s ability to bind TAR RNA through its DRBDs (130)
. RHA was shown to release unspliced genomic HIV RNA and incompletely spliced mRNA, which requires exportation to the cytoplasm for packaging and translation (131)
. Similar to NFAR, RHA has been found to associate with SMN; together, they associate with a complex containing pol II, and snRNPs (132)
. It remains to be seen whether NFAR and RHA directly interact in such mRNA processing complexes.
The mouse homologue of RHA is 87% identical to the human protein, ubiquitously expressed, and possesses both RNA and DNA helicase activities (133)
. Unlike human cells where RHA is nuclear and excluded from the nucleoli, mouse RHA was reported to localize specifically to nucleoli and colocalize with RNA pol I, suggesting an involvement in rRNA biosynthesis (134)
. Homozygous disruption of RHA in mice led to early embryonic lethality, with significant apoptotic cell death in embryonic ectodermal cells during gastrulation (135)
.
The Drosophila maleless (MLE) gene shares 49% amino acid identity with RHA and similarly contains two copies of the DRBD (136
, 137)
. MLE localizes to the X chromosome, is required for X chromosome dosage compensation, and is essential for male viability (138
139
140)
. Association of MLE with the X chromosome is RNA dependent, and it is thought that nascent transcripts or a hypothetical RNA component of chromatin plays a critical role in the biochemical mechanism of dosage compensation (141)
.
Two rice (O. sativa) cDNAs that share 60% amino acid identity are 25% identical to RHA (Table 2
). They contain a single DRBD as well as a DEAD box helicase domain, although their function in plants has not been examined. Although the precise role of human RHA and its homologues in other organisms remains to be fully clarified, the data imply an important role for RHA in regulation of gene expression in the nucleus.
|
Negative regulatory element binding protein (NREBP/SON/DBP-5)
Database analysis performed in our laboratory has revealed that a 2386 amino acid human protein referred to as negative regulatory element binding protein (NREBP) contains a single DRBD (Fig. 1
; refs 142
, 143
). Several homologues that appear to be partial clones or spliced variants of NREBP, called SON and DBP-5, were found to contain a single DRBD (144
, 145)
. The NREBP gene has been mapped to human chromosome 21q22.1 and contains 13 exons and 12 introns (142
, 146)
. NREBP contains at least six areas of tandem repeats and has been reported to be a potential transcriptional repressor that binds to the negative regulatory element in human hepatitis B virus (HBV; refs 142
, 147
). InterPro search results indicate that in addition to a single DRBD at the extreme carboxyl terminus, NREBP contains a D111/G patch domain (IPR000467) found in putative ssRNA binding proteins (143)
. Several groups have reported that NREBP is localized to the nucleus (142
, 145)
. A single 8.3 kb mRNA is ubiquitously expressed in all human tissues and the apparent full-length cDNA gives rise to a 257 kDa protein (142)
. The only function assigned to NREBP to date involves the repression of HBV core promoter activity. For example, overexpression of NREBP repressed transcription of HBV genes and the production of HBV virions (142)
. Since NREBP may be a nuclear protein that can bind dsRNA as well as ssRNA, it will be interesting to examine its interaction with nucleic acids and nucleic acid binding proteins and evaluate its potential role in influencing virus replication.
Kidney anion exchanger adaptor protein (kanadaptin)
Another protein that contains a single DRBD whose function has not yet been studied is kanadaptin. This protein was identified in a yeast two-hybrid screen through an interaction with the kidney Na+-independent Cl-/HCO3- anion exchanger 1 (kAE1) and was suggested to play a role in targeting kAE1-containing vesicles to the basolateral plasma membrane in the rabbit kidney (148)
. However, subsequent analysis indicated that the mouse kanadaptin protein was a widely expressed nuclear protein (149)
. The partial sequence of mouse kanadaptin appears to be missing the amino terminus, which is conserved in the apparent homologues in human, Drosophila, C. elegans, A. thaliana, and O. sativa (see Table 1
for accession numbers). The nuclear function of this DRBP and its potential interaction with other DRBPs remain to be studied.
Hyponastic leaves (HYL1)
Database searches indicate that plants encode numerous DRBPs, most of which remain uncharacterized. One Arabidopsis gene, designated hyponastic leaves (HYL1), encodes a 419 amino acid protein that contains two DRBDs able to bind to dsRNA (150)
. HYL1 has a bipartite nuclear localization signal, was shown to localize to the nucleus, and is expressed in most or all tissues. HYL1 expression is hormonally regulated, and a transposon insertion mutation in HYL1 leads to altered responses to several hormones. These mutant plants show a short stature, delayed flowering, leaf hyponasty, reduced fertility, decreased root growth rate, and altered root gravitropic responses. This phenotype suggests that HYL1 plays a role throughout the Arabidposis life cycle and indicates that dsRNA may be involved in hormone signaling that controls growth and development (150)
. B. oleracae and O. sativa appear to encode homologues that share 75% and 31% amino acid identity with HYL1 and contain two DRBDs (Table 1)
.
|
EDITING dsRNA |
|---|
|
TOPABSTRACTINTRODUCTIONEUKARYOTIC DRBPS: THE CYTOPLASMNUCLEAR EUKARYOTIC DRBPSEDITING dsRNAA FAMILY OF dsRNA...VIRAL DRBPSINTERACTION AMONG DRBP FAMILY...ALTERNATIVE dsRNA BINDING MOTIFSCONCLUSIONSREFERENCES |
|---|
152
153)
. An A-to-I editing enzyme was discovered in Xenopus through its ability to unwind long dsRNAs by deaminating multiple A’s to I’s, which results in unstable I:U base pairs (154)
. Although several well characterized RNA targets are edited in their coding region, it was recently found that A-to-I editing also occurs in 3'-UTR, intron and noncoding RNA sequences in human and C. elegans RNA substrates (155)
. Possibly, hyperediting of viral RNAs by ADARs may provide a mechanism to remove dsRNA from cells, perhaps in conjunction with cytoplasmic endonucleases now known to specifically cleave hyperedited dsRNA species (156)
.
The first human family member isolated, ADAR1, localizes to chromosome 1q21 and was found to encode a protein with three DRBDs as well as a nuclear localization signal (157
, 158)
. Two major species of ADAR1 have been reported: an IFN-inducible 150 kDa protein present in the cytoplasm and nucleus, and a second 110 kDa protein that is constitutively expressed and predominantly nuclear (159)
. The two species arise through the use of different promoters that initiate transcription at alternative sites in the first exon, so that the IFN-inducible form comprises a novel 300 amino acids in the amino terminus (160)
. ADAR1 has two Z-DNA (negatively supercoiled dsDNA) binding domains, the first of which is absent in the constitutively expressed spliced variant. This domain perhaps allows ADAR1 to form a specific structure that can interact with Z-DNA species (161
, 162)
. ADAR1 is widely expressed in many tissues and has been shown to be involved in the editing of specific glutamate-gated ion channels (152
, 163
, 164)
. There are three spliced isoforms of ADAR1 that show tissue-specific expression and different affinities for dsRNA that affect the efficiency of editing, but not substrate site selection (165
166
167)
. ADAR1 was also shown to be able to edit hepatitis delta virus (HDV) in vitro, an event reportedly required for the virus’ life cycle (168)
. HDV encodes two forms of a single protein—HDAg, with the larger molecule being produced by RNA editing, which converts a stop codon to extend the open reading frame. HDV genotype III has alterations in conserved base pairs of the unbranched rod structure required for editing of genotype I and must rearrange its genome to form the branched double hairpin required for editing (169)
. Editing of HDV is required for virion formation but not for replication; since HDB genotype III is associated with the most severe disease, RNA editing may be related to pathogenic differences (169)
.
The rat homologue of ADAR1, termed DSRAD is 79% identical to its human counterpart. DSRAD is ubiquitously expressed, including in the brain, and like hADAR1 is reported to be involved in editing the serotonin-2C receptor pre-mRNA (151
, 167
, 170)
. Attempts to generate an ADAR1 knockout mouse revealed a heterozygous embryonic-lethal phenotype (3)
. Indeed, most of the heterozygous embryos died before embryonic day (E) 14 and revealed defects in erythropoiesis in the liver. Expression levels of ADAR1/DSRAD were reported to increase at E13 to E14 in the liver; plausibly, RNA target(s) at this point in development may be underedited by a lack of the ADAR1 protein (3)
.
A Xenopus ADAR gene has also been cloned that exhibits 41% amino acid identity with hADAR1. The Xenopus homologue was shown to have adenosine deaminase activity that could edit specific sites in the mammalian glutamate receptor mRNA (171
172
173)
. Like its mammalian homologue, Xenopus ADAR1 contains three DRBDs (174)
. Analysis of Xenopus lampbrush chromosomes indicated that ADAR1 localizes to the nascent ribonucleoprotein matrix and is enriched in a transcriptionally silent RNA-containing loop (175)
. Although there is a putative nuclear localization signal near the amino terminus of Xenopus ADAR1, a basic region upstream of the DRBDs has been shown to be important for the nuclear localization of ADAR1 (176)
. C. elegans has also been found to encode an adenosine deaminase gene with a single DRBD at the extreme amino terminus and a deaminase domain at the extreme carboxyl terminus (177)
. This 495 amino acid protein has 25% amino acid identity with hADAR1 (177)
.
ADAR2 and its rat homologue RED1, which share 94% amino acid identity, are 23% identical to ADAR1 and contain a 54 amino acid amino-terminal extension that includes an arginine-rich motif (151)
. Human ADAR2 maps to chromosome 21q22.3 and contains two DRBDs as well as the characteristic deaminase domain (178)
. ADAR2 is expressed in many tissues throughout development, with the highest levels being in the thalamus (179
, 180)
. Several mRNA species give rise to four major protein isoforms that have been reported to differ in their RNA editing capabilities in vitro (181)
. A Drosophila homologue of ADAR2 has been cloned that specifically edits the Q/R site in the pre-mRNA encoding the glutamate receptor subunit GluR-B (182)
. Drosophila with ADAR2 deletions were found to lack editing in three ion channel transcripts (referred to as para, cac, and Dros Glu Cl-) and exhibited extreme behavioral deficits (defects in motor control, mating, and flight), suggesting that A-to-I editing of pre-mRNAs in Drosophila acts predominantly through nervous system targets to affect adult nervous system function, integrity, and behavior (183)
. Enzymatically active ADAR1 and ADAR2 have recently been shown to be associated with large nuclear ribonucleoprotein (lnRNP) particles (184)
. Formation of dsRNA editing targets requires base pairing of exon and intron sequences, so editing is expected to occur before or at the same time as splicing (184
185
186)
.
The Q/R site of the glutamate receptor (GluR) is edited nearly 100%, and mice heterozygous for an uneditable GluR transcript have early-onset epilepsy and premature death (187
, 188)
. ADAR2 is the only adenosine deaminase that edits the Q/R site in GluR in vitro (181
, 189
, 190)
. Although ADAR2 knockout mice were born at expected frequencies, they become seizure prone and die by 20 days after birth (191)
. Q/R editing of GluR was solely responsible for death, as the mice were rescued by introducing two edited copies of GluR, thus identifying the GluR transcript as the most critical substrate of ADAR2 (191)
.
ADAR3 was cloned and found to contain an arginine- and lysine-rich domain that interacts with ssRNA in addition to its two DRBDs (153)
. Human ADAR3 is an 81 kDa protein encoded on chromosome 10p15 whose expression is limited to the brain, specifically to the amygdala and thalamus (153
, 192)
. Although ADAR3 was not found to catalyze the deamination of any known editing targets, it was suggested that the dsRNA and ssRNA binding domains of ADAR3 allow it to interact in a unique manner with its unidentified RNA substrates (153)
. The rat homologue RED2 shares 60% amino acid identity with hADAR3; it is expressed solely in the brain and has not yet been shown to edit any known targets (193)
. Although hADAR1 and 2 have been shown to be functional adenosine deaminases that are ubiquitously expressed, hADAR3 is expressed only in the brain and has not yet been shown to be an active adenosine deaminase. Collectively, the ADAR family of DRBPs plays crucial roles in the cell by editing RNA, an event critical for development and, perhaps, similar to other DRBP members, host defense.
Testis nuclear RNA binding protein (TENR)
Testis nuclear RNA binding protein (TENR) was isolated by its ability to bind to the 3'-UTR of the mouse Prm-1 sperm specific mRNA, similar to SPNR and PRBP (9)
. TENR encodes a 72 kDa protein and has one DRBD that mediates interaction with dsRNA. TENR is reportedly expressed specifically in testis, where its mRNA is restricted to cells in the mid-pachytene to -round spermatid stage (9)
. TENR mRNA may itself be under translational control since the protein is not detected until the haploid stage of spermatid development. TENR is a predominantly nuclear protein whose expression is localized to round and early elongating spermatid cells in a lattice-like nuclear distribution. The localization and dsRNA binding characteristics of TENR led to the hypothesis that TENR is involved in a testis-specific nuclear posttranscriptional process such as hnRNA packing, alternative splicing, or mRNA transport (9)
. Southern blot results suggest that TENR is conserved in the human and rat genomes; recently, the human homologue, which shares 88% amino acid identity, has been identified in GenBank (9)
. TENR has an adenosine deaminase domain from amino acids 291 to 617, as shown in Fig. 1
. However, it has been reported that critical amino acids required for deaminase activity are changed in this domain and TENR may therefore not have deaminase activity (194
, 195)
. Thus, TENR may bind to substrates but not have the ability to edit them, and may even act as an inhibitor of mRNA editing.
|
A FAMILY OF dsRNA SPECIFIC ENDORNASES THAT ARE CONSERVED FROM E. COLI TO HUMANS |
|---|
|
TOPABSTRACTINTRODUCTIONEUKARYOTIC DRBPS: THE CYTOPLASMNUCLEAR EUKARYOTIC DRBPSEDITING dsRNAA FAMILY OF dsRNA...VIRAL DRBPSINTERACTION AMONG DRBP FAMILY...ALTERNATIVE dsRNA BINDING MOTIFSCONCLUSIONSREFERENCES |
|---|
, 197)
. RNase III has an amino-terminal endonuclease domain (9 amino acid RNase III signature motif) in addition to a single carboxyl-terminal DRBD. It has recently been shown that the DRBD is dispensable for catalytic activity, but assists in cleavage specificity (198
199
200)
. E. coli RNase III is involved in processing pre-rRNA, tRNA, and phage polycistronic mRNA and has been suggested to play a global role in gene regulation since depletion of RNase III alters the expression level of 10% of bacterial proteins (201
, 202)
. RNase III homologues have also been identified in Saccharomyces cerevisiae (Rnt1) and Schizosaccharomyces pombe (Pac1) and have been reported to be involved in processing pre-rRNA, small nuclear RNAs (snRNAs), and small nucleolar RNAs (snoRNAs; refs 203
204
205
).
Several RNase III family members have likewise been identified in C. elegans and Drosophila. The carboxyl terminus of Drosophila Drosha and its C. elegans homologue contains a single RNase III motif and a DRBD similar to E. coli RNase III (206)
. Drosha contains a second RNase III signature sequence, which suggest that these RNase III proteins in Drosophila and C. elegans may be able to form active catalytic centers as monomers. This is in contrast to E. coli RNase III, which forms homodimers and requires a divalent metal ion (Mg2+) to hydrolyze phosphodiesters (198
, 206
, 207)
. The Drosophila and C. elegans RNase III proteins also have an extended amino terminus of 800 amino acids, with Drosophila RNase III having a proline-rich region at the amino terminus and a serine-rich stretch at the carboxyl terminus not found in the C. elegans homologue (206)
. Significantly, Drosha is likely an essential gene in Drosophila development since microdeletions in this region are lethal (206)
. A human homologue of RNase III has been cloned that has an amino-terminal region that is proline-rich, followed by a serine/arginine-rich domain, and two RNase III signature motifs as well as a DRBD at the carboxyl terminus (208)
. This 160 kDa nuclear protein is ubiquitously expressed and is involved in the processing of pre-rRNA. Depletion of RNase III through antisense oligos results in accumulation of pre-rRNAs and induction of cell death, suggesting an essential role for this RNase III protein in the cell (208)
.
An additional group of proteins containing Rnase III motifs and a DRBD, referred to as Dicer, are found in S. pombe, A. thaliana (CAF), C. elegans (DCR-1), Drosophila (DCR-1 and DCR-2), mouse, and humans (209
210
211
212)
. Dicer contains an amino-terminal helicase domain, a PAZ (Pinwheel, Argonaut, Zwille) domain, two RNase III signature motifs, a carboxyl-terminal DRBD, and plays a role in RNAi, as discussed below (210
211
212
213)
.
RNA interference: dsRNA, siRNA, and miRNA
It is conceivable that some cellular responses to dsRNA are mediated by defense systems designed to recognize virus infection and the presence of transposons or other similar threats to the host (214)
. Short dsRNA structures are now known to induce potent homology-dependent gene silencing in a process known as RNAi. This event occurs through dsRNA becoming processed into 22-nucleotide sequences, referred to as short interfering RNAs (siRNAs). Should these sequences correspond to a complementary mRNA, the latter is targeted for degradation (215
216
217
218)
. RNAi and the related phenomenon of quelling and post-transcriptional gene silencing (PTGS) have been shown to exist in fungus (Neurospora), plants (Arabidopsis), invertebrates (Drosophila and C. elegans), and vertebrates (zebrafish, mouse, and human cells; refs 219
220
221
222
223
224
). However, RNAi is difficult to detect in mammalian cells since dsRNA structures greater than 30 bp may stimulate the IFN system, including activating PKR, which nonspecifically inhibits mRNA translation. Nevertheless, certain exceptions apparently exist and long dsRNAs (500 nt) have been shown to elicit sequence specific silencing in mouse oocytes and early embryos as well as in mouse embryonic stem cells and embryonal teratocarcinoma cell lines (223
, 225
226
227)
. It is not clear whether PKR is absent in these cells, but it has recently been shown that synthetic 21 nt siRNAs transfected into mammalian cells can effectively bypass activation of PKR and 2',5'-oligoadenylate synthetase and cause sequence specific RNAi (224)
.
The finding that many viruses encode inhibitors of RNAi supports evidence that RNAi may indeed represent a viral defense system. For example, the tombusvirus p19 protein suppresses PTGS in
