(Received for publication, December 24, 1996, and in revised form, March 25, 1997)
From the Department of Genetics, Stanford University School of Medicine, Stanford, California 94305-5120
The human ARD-1 (activator of RNA
decay) cDNA sequence can rescue mutations in the Escherichia
coli rne gene, which specifies the essential endoribonuclease
RNase E, resulting in RNase E-like cleavages in vivo in
rne-defective bacteria and in vitro in extracts isolated from these cells (Wang, M., and Cohen, S. N. (1994)
Proc. Natl. Acad. Sci. U. S. A. 91, 10591-10595). Recent
studies indicate that the 13.3-kDa protein encoded by ARD-1
cDNA is almost identical to the carboxyl-terminal end of the bovine
protein NIPP-1, a nuclear inhibitor of protein phosphatase 1; separate
transcripts formed by alternative splicing are proposed to encode the
discrete ARD-1 and combined ARD-1/NIPP-1 products (Van Eynde, A., Wera,
S., Beullens, M., Torrekens, S., Van Leuven, F., Stalmans, W., and
Bollens, M. (1995) J. Biol. Chem. 270, 28068-28074).
Here we show that affinity column-purified protein encoded by human
ARD-1 cDNA in E. coli is a site-specific
Mg2+-dependent endoribonuclease that binds
in vitro to RNase E substrates, cleaves RNA at the same
sites as RNase E, and, like RNase E, generates 5 phosphate termini at
sites of cleavage. Our results indicate that the ARD-1 peptide can
function as a ribonucleolytic analog of E. coli RNase E as
well as a domain of the protein phosphatase inhibitor, NIPP-1.
RNA degradation in eukaryotes, as in bacteria, appears to be a regulated process that involves multiple enzymatic steps (for reviews, see Refs. 3-5). Although a variety of ribonucleases have been identified in eukaryotes, with few exceptions little or no information is available about the specific biological role of these enzymes or the genes that encode them (for reviews, see Refs. 6-8). This is especially true of nucleases that degrade the mRNA of mammalian cells.
The Escherichia coli rne gene encodes an essential endoribonuclease, RNase E, which has a key role in the decay and processing of 9 S ribosomal RNA, RNA I, an antisense regulator of replication of ColE1-type plasmids, and a variety of messenger RNAs (for reviews, see Refs. 9-11). A human cDNA sequence designated as ARD-1 (activator of RNA decay) was identified recently by its ability to reverse the pleiotropic effects of temperature-sensitive (ts) and deletion mutations in rne (1). Expression of ARD-1 cDNA in E. coli results in RNase E-like cleavages in vivo in rnets cells shifted to a nonpermissive temperature, and extracts isolated from cells mutated in rne but expressing ARD-1 cDNA cleave 9 S RNA in vitro at locations attacked by RNase E (1). Collectively, these findings suggested that ARD-1 may encode an endoribonuclease activity similar to that of RNase E. Moreover, the 13.3-kDa ARD-1 protein sequence deduced from analysis of the cDNA has multiple structural features in common with the 118-kDa Rne protein, including similarity to domains of eukaryotic proteins implicated in endocytosis, macromolecular transport, and RNA binding (12-14).
Recently, a peptide sequence almost identical to the protein encoded by
human ARD-1 cDNA was found at the carboxyl terminus of
bovine NIPP-1 (38.5 kDa), a nuclear inhibitor of protein phosphatase 1 (2).1 Comparison of the cloned cDNA
sequences of ARD-1 and NIPP-1 showed that the
5-untranslated region of ARD-1 contains a
470-bp2 segment that is not present in
NIPP-1 cDNA and that ARD-1 cDNA lacks a
220 bp segment of the coding region of NIPP-1 (2); these
observations have led to the proposal by Van Eynde et al. (2) that ARD-1 and NIPP-1 mRNAs are generated
by alternative splicing.
Using ARD-1 protein overexpressed in E. coli and highly
purified by affinity-column chromatography, we show here that ARD-1 is
a single-strand-specific Mg2+-dependent
endoribonuclease that binds to RNase E substrates, cleaves short
oligoribonucleotides and complex substrates in A+U-rich regions at the
sites cut by RNase E, and generates 5-phosphate termini. Thus, the
amino acid residues comprising the ARD-1 sequence appear to function
both as a domain of the protein phosphatase inhibitor NIPP-1 and a
human analog of E. coli ribonuclease E.
For expression of the
histidine-tagged ARD-1 protein in E. coli, two
oligonucleotides, 5ACGTGACTCATATGGTGCAAACTGCAGTGGTC-3
(primer 1) and
5
-TCACGTCTGGATCCTCAAATCAGCAAGGAAGGTG-3
(primer 2),
corresponding to the 5
-end and 3
-end of ARD-1
cDNA sequence, respectively, were synthesized for PCR amplification
of the ARD-1 cDNA sequence. An NdeI site and
a BamHI site were added to the 5
ends of primers 1 and 2, respectively. The PCR product, after digestion with NdeI and
BamHI, was ligated to the NdeI- and
BamHI-cleaved pET16 expression vector (Novagen, WI), which
contains a polyhistidine region and Factor Xa protease-cleavable
region, and introduced into the protease-deficient
DE3 lysogenic
strain BL21, which expresses T7 RNA polymerase under control of an IPTG
(isopropyl-1-thio-
-D-galactopyranoside)-inducible lacUV5 promoter (15, 16).
E. coli cells containing the expression plasmid or its derivatives were grown at 35 °C in LB medium containing 200 µg/ml ampicillin to A600 = 0.6-0.8, except where indicated. Expression of the ARD-1 protein was induced by addition of IPTG to a final concentration of 1 mM.
Cell LysateExtracts of cells allowed to grow for 2 h at 30 °C following induction of ARD-1 expression by IPTG were analyzed by lysing the cells directly in the loading buffer for sodium dodecyl sulfate-polyacrylamide gel electrophoresis, as described by Sambrook et al. (17). For purification of ARD-1 by affinity column chromatography, cells were harvested and sonicated in 20 mM Tris-HCl, pH 7.9, containing 0.5 M NaCl and 5 mM imidazole. Sonications were repeated for 20-s intervals, with 30-s rest periods at 0 °C until the solution was no longer viscous. For some preparations, 6 M guanidine HCl was added. Samples were then centrifuged at 10,000 × g for 10 min to remove debris. The lysate was filtered through a 0.45-µm filter before applying to a column.
Purification of the ARD-1 ProteinThe histidine-tagged fusion protein was purified by binding to divalent cations (Ni2+) immobilized on His-Bind (Novagen) metal chelation resin at 0 °C. The column was washed at 0 °C in 0.5 M NaCl, 20 mM Tris-HCl, pH 7.9, containing 60 mM, and then 100 mM, imidazole. The protein was eluted in buffer containing 400 mM imidazole at a flow rate of 0.4 ml/min. Buffers used for purification under protein-denaturing conditions contained 6 M guanidine HCl and 20 mM imidazole in the washing buffer. The eluted fraction was concentrated by centrifugation using a Centricon-10 (Amicon) concentrator and dialyzed for 24 h at 4 °C against three changes of storage buffer (18) at 4 °C. Further protein purification on a denaturing preparative gel and electro-elution was as described previously (19). Alternatively, ARD-1 protein was also purified under nondenaturing conditions using a imidazole gradient to elute the protein from the affinity column.
Factor Xa Cleavage of the Fusion ProteinProteolytic cleavage of the ARD-1 fusion proteins was carried out using a ratio of Factor Xa (New England Biolabs) to fusion protein of 1 to 100 (w/w) in a storage buffer (50 mM Tris-HCl, 20% glycerol, 1 mM EDTA, 0.5 M NaCl, and 0.5% Triton X-100). Incubations were carried out for 2-4 h at room temperature or overnight at 16 °C.
In Vitro Transcription and RNA Purification32P-Labeled 9 S RNA and GGG·RNA I were
synthesized in vitro from plasmids pTH90 (20) and pM21 (21),
respectively, using the T7 MEGAshortscriptTM in vitro
transcription kit (Ambion, Inc.) in the presence of
[-32P]GTP (DuPont NEN) as recommended by the vendor.
The oligoribonucleotides BR10 and BR13, containing sequences present at
the 5
end of RNA I and GGG·RNA I, respectively (18), were labeled by
treatment with T4 polynucleotide kinase (Life Technologies Inc.) (17) in the presence of [
-32P]ATP. All radioactively
labeled RNA substrates were purified on urea-polyacrylamide gels (6%
for 9 S RNA and GGG·RNA I, and 20% for oligonucleotides) and
extracted with phenol-chloroform prior to use.
32P-Labeled 9 S RNA, GGG·RNA I, and oligoribonucleotides BR10 and BR13 were synthesized as described above. Endoribonuclease activity was assayed in 20-µl reaction mixtures containing 20 mM Tris-HCl, pH 7.6, 1 mM dithiothreitol, 10 mM MgCl2, 3 units/ml RNasin ribonuclease inhibitor, and 10,000-30,000 cpm of the 32P-labeled RNA substrate. The mixtures were incubated at 30 °C for 1 h, and then extracted with phenol/chloroform. After adding 20 µl of formamide sequencing loading buffer and heating for 3 min at 85 °C, samples were loaded in 6% or 20% sequencing gels. Gels were dried and then examined by autoradiography. The purified NH2-terminal half of RNase E, which contains the catalytic domain (22), was provided by K. J. McDowall (Stanford University). Placental RNase A, mung bean nuclease, and placental RNase inhibitor (RNAsinTM) were purchased from Ambion, Sigma, and Promega, respectively.
Northwestern AssaysProteins from SDS-polyacrylamide gels were blotted onto nitrocellulose filter membranes as described by Dunn (23). Filters were then prehybridized in buffer containing yeast total tRNA or E. coli total RNA at 250 µg/ml before being probed with radioactively labeled 9 S RNA or GGG·RNA I as described by Cormack et al. (19).
Gel Mobility Shift AssaysPurified ARD-1 protein
(concentrations indicated in figure legends) was incubated in 20-µl
reactions with ~30,000 cpm of 32P-radiolabeled 9 S RNA in
the presence of 20 mM Tris-HCl (pH 7.6), 200 mM
NaCl, 1 mM dithiothreitol, 10 mM
MgCl2, 240 µg/ml tRNA, 3 units/ml placental ribonuclease
inhibitor, and 240 µg/ml bovine serum albumin for 15 min at room
temperature. Reaction mixtures for RNase E binding studies were carried
out under the same conditions except that the buffer contained 40 mM EDTA. The mixtures were then adjusted to 3% Ficoll
(400) and loaded on 6% nondenaturing polyacrylamide gels with a
cross-linking ratio of 37.5:1 in 100 mM Tris-HCl (pH 8.0),
50 mM glycine buffer. Electrophoresis was carried out at
4 °C at 240 V for 3 h. The gels were dried and then exposed to
x-ray film (DuPont) with an intensifier screen at 70 °C.
For RNA competition, radiolabeled RNA was first mixed with varying amounts of unlabeled RNA competitor and then subjected to the gel mobility shift assay as described above. Total RNA and mRNA from B cells were purified using the RNA STAT-60TM reagent (Tel-Test "B", Inc., Friendswood, TX), and the Poly(A)Tract mRNA isolation system (Promega), respectively. Synthesis of unlabeled 9 S RNA was carried out as described for the 32P-labeled 9 S RNA, but in the absence of radionucleotide. Poly(A) oligoribonucleotides and yeast tRNA were purchased from Pharmacia Biotech (Uppsala, Sweden) and Boehringer Mannheim, respectively.
Analysis of 5Unlabeled GGG·RNA I was synthesized and purified as
described for 32P-labeled GGG·RNA I but in the absence of
[-32P]GTP. GGG·RNA I (1 µg) was incubated at
30 °C with ARD-1 (2.5 µg), full-length RNase E (0.5 µg), or
RNase A (2 pg) in 50-µl reactions for 3 h, 10 min, and 15 min,
respectively. The ARD-1 and RNase A reactions contained 20 mM Tris-HCl, pH 7.6, 1 mM dithiothreitol, and
10 mM MgCl2. The RNase E reaction contained 20 mM Tris-HCl, pH 8.0, 5 mM MgCl2,
100 mM NaCl, 5% glycerol, 0.1% Triton X-100, and 0.1 mM dithiothreitol. The ARD-1 and RNase E reactions
contained 100 units of ribonuclease inhibitor. After incubation,
samples were extracted with phenol/chloroform, precipitated with
ethanol, and split into two aliquots. One aliquot from each sample was treated with calf intestine alkaline phosphatase (Boehringer Mannheim) to dephosphorylate the 5
termini, extracted with phenol/chloroform, and precipitated with ethanol. Then all samples were incubated with T4
polynucleotide kinase in the presence of [
-32P]ATP
(6000 Ci/mmol). After phenol/chloroform extraction and ethanol precipitation, the RNA was resuspended in 50 µl of diethyl
pyrocarbonate-treated H2O and passed through a MicroSpin
S-200 HR column (Pharmacia) to remove unincorporated radionucleotides.
Samples (5 µl) were mixed with formamide sequencing loading buffer (5 µl), heated for 3 min at 85 °C, and loaded in 6% sequencing
gels.
A PCR-amplified cDNA
segment containing the full-length ARD-1 translational open
reading frame was introduced into plasmid pET16b (Fig.
1) and expressed from the bacteriophage T7 promoter in
E. coli as a His/ARD-1 fusion protein. Total protein
extracted from cells containing the fusion construct contained a 19-kDa band that increased approximately 20-fold in relative concentration 2 h after IPTG induction of T7 RNA polymerase. This protein
species, which migrated more slowly than was predicted for a protein
formed by joining of the 13.3-kDa ARD-1 peptide to the polyhistidine tag and Factor Xa-cleavable region, was purified by
Ni2+-column chromatograply as described under
"Experimental Procedures."
The purified His/ARD-1 fusion protein migrated in SDS-polyacrylamide
gels stained by silver nitrate as a single 19-kDa band (Fig.
2); treatment of this protein by Factor Xa generated a
product migrating at 17 kDa, which was slower than expected from the
calculated molecular mass of the 13.3-kDa protein encoded by the ARD-1
open reading frame. The observed retarded migration in gels of ARD-1, which has a pI of 10.43 and contains highly charged regions at both the
NH2-terminal and COOH-terminal ends (pI = 10.43) (1), parallels the anomalous migration properties of Rne, which is a highly
charged 118-kDa protein that migrates in SDS-polyacrylamide gels at the
180-kDa position (12, 24).
Binding of ARD-1 to RNase E Substrates
Both Northwestern and
gel mobility shift assays showed that ARD-1 is an RNA-binding protein.
In initial Northwestern blots using crude extracts from E. coli overexpressing the His/ARD-1 fusion (data not shown), a
19-kDa protein band was observed to bind to the RNase E substrates 9 S
RNA, GGG·RNA I, and the BR13 oligonucleotide. Fig. 3
(lanes 1 and 4) shows binding of the
affinity-purified His/ARD-1 fusion protein to 9 S RNA and RNA I. E. coli extracts of cells expressing RNase E but lacking
ARD-1 cDNA contained bands that in Northwestern blots
were observed to bind to these same RNAs (Fig. 3, lanes 3 and 6), as has been reported previously (19, 25, 22, 26);
however, these extracts did not show a 19-kDa RNA-binding protein
species. Gel mobility shift assays showed that ARD-1 fusion protein
purified under nondenaturing conditions and incubated with
radioactively labeled 9 S ribosomal RNA retarded migration of the
target in native polyacrylamide gels (Fig. 4,
top); unlabeled 9 S RNA (lanes 3 and
4), total B cell RNA (lanes 5 and 6),
which consists largely of ribosomal RNA, and poly(A) mRNA isolated
from B cells (lanes 10 and 11) efficiently
competed with the 32P-labeled 9 S RNA probe for binding to
ARD-1. While the presence of unlabeled polyadenylic acid (poly(A)) in
10-fold excess (lanes 7 and 8) or tRNA in 50-fold
excess (lane 9) did not prevent binding of ARD-1 to 9 S RNA,
both of these RNAs reduced 9 S-specific binding of ARD-1 when added in
still greater excess. As shown in Fig. 4A
(bottom), the RNA-binding specificity of ARD-1 was identical to that seen for RNase E.
ARD-1 Cleaves Oligoribonucleotides and Complex RNAs at the Same Sites as RNase E
ARD-1 protein purified under nondenaturing conditions by affinity chromatography and shown to migrate as a single band in silver nitrate-stained gels (see above) was assayed for endoribonuclease activity on 9 S RNA, GGG·RNA I, and the BR13 and BR10 oligoribonucleotides, all of which are cleaved by purified RNase E at highly specific sites (18, 19, 22, 27-29). Control reactions that show the cleavages produced by RNase A, a common eukaryotic single-strand-specific endoribonuclease, were included for comparison; these controls demonstrated that the cleavage patterns of ARD-1 and RNase A are different, and also indicated the absence of RNase A activity in our purified ARD-1 preparations.
As shown in Fig. 5A, the affinity-purified
His/ARD-1 fusion protein produced the same cleavages in 9 S RNA as
RNase E, although cleavage at the "a" site (30, 31) was relatively
more prominent for ARD-1; an identical cleavage pattern was observed
for the discrete ARD-1 peptide generated by Factor Xa treatment of the fusion protein (results not shown). GGG·RNA I was cleaved by ARD-1 at
a site located 8 nt from the 5 end, leaving a product of 103 nt (Fig.
5B), as has been observed for RNase E (18, 26, 32). While
the major product resulting from RNase A digestion of GGG·RNA I was
also a 103-nt species, RNase A additionally cleaved this substrate at
other sites (Fig. 5B).
Analysis of ARD-1 and RNase E cleavages of two synthetic
oligoribonucleotides, BR10 and BR13, containing the same sequence as
the 5 single-strand region of RNA I and GGG·RNA I, respectively (18)
provided further evidence of the identical cleavage specificity of the
two enzymes. McDowall et al. (18) have shown that BR10 and
BR13 are each cleaved by RNase E at the intranucleotide bonds cleaved
in RNA I; as seen in Fig. 5 (C and D, lane
6 of both), purified ARD-1 cleaved these oligoribonucleotides at
the same sites as E. coli RNase E. However, differences in
relative amounts of the products generated by the two enzymes, as
indicated by the intensity of gel bands corresponding to individual RNA
species, were observed. Cleavage of BR10 by ARD-1 occurred
approximately equally at phosphodiester bonds located 5 and 6 nt from
the 5
end, whereas this oligonucleotide was cleaved by RNase E
predominantly at a site 5 nt from the 5
end (Fig. 5C; see
also Ref. 18).
RNase A cleaved oligonucleotides BR10 and BR13 at different locations
than either ARD-1 or RNase E (Fig. 5, C and D).
Additionally, RNase A generated 5-OH termini that could be
phosphorylated without further treatment using radioactively labeled
ATP (Fig. 6, lane 7), as has been shown
previously (33, 49); in contrast, both RNase E (cf. Refs. 29
and 34) and ARD-1 generate 5
-phosphate termini that cannot be
phosphorylated in vitro unless treated with alkaline
phosphatase (Fig. 6, lanes 3 and 5). Cleavages by ARD-1 and RNase E did not occur in reaction mixtures that contain 0.01 M EDTA and lack Mg2+, whereas cleavages by
RNase A do not require divalent cations (35).
The human gene ARD-1 was identified by the ability of ARD-1 cDNA to rescue temperature-sensitive and deletion mutants of the E. coli rne gene (1), which is essential for bacterial viability as well as for RNA decay and processing (36, 37). rne-defective mutants expressing ARD-1 can carry out RNase E-like cleavages in vivo, and extracts of cells that express ARD-1 but are deleted for the rne gene segment encoding the catalytic domain of RNase E (22), were observed to cleave 9 S RNA in vitro (1). Like Rne, ARD-1 was inferred by DNA sequence analysis to be a highly proline-rich peptide that has similarity to segments of the small ribonucleoprotein RNA-binding proteins (38) and to dynamin, an eukaryotic protein implicated in endocytosis, membrane trafficking, and microtubule-based organelle transport (39, 40). Analogous features of the 13.3-kDa ARD-1 and 118-kDa Rne proteins are further evident from our demonstration that ARD-1 has the same RNA binding properties as RNase E. However, antibodies raised against RNase E did not cross-react with ARD-1 (data not shown), consistent with the observation that these two proteins do not show regions of amino acid sequence homology (1, 12, 24).
The studies reported here show that highly purified ARD-1 protein is an endoribonuclease that, like RNase E (19, 28, 41), cleaves single-strand RNA segments in A+U-rich regions. ARD-1 cleavage of native substrates and synthetic oligonucleotides containing RNase E cleavage sites occurs at the same phosphodiester bonds cleaved by E. coli RNase E. However, the relative rate of ARD-1 cleavage and RNase E cleavage differed at different sites within some substrates. These differences provide additional evidence that the activity we assayed for affinity-purified ARD-1 preparations, which had been synthesized in E. coli and were >99% pure by silver stain analysis (Fig. 2), does not result from contamination by RNase E.
No ARD-1 activity was detected in reaction mixtures that lack
Mg2+ and contain 0.01 M EDTA, indicating that
ARD-1 activity, like RNase E activity, requires divalent cations.
Neither ARD-1 nor RNase E was inhibited by placental RNase inhibitor,
which inhibits RNase A and its analogs (35). Both RNase E (18, 31, 32, 41) and ARD-1 (this paper) cleave single-strand segments of RNA 3 to
both purines and pyrimidines in A+U-rich regions, whereas RNase A cuts
single-stranded RNAs predominantly 3
to pyrimidine residues (C or U
nucleotides) (33).
Cleavage of RNA I by ARD-1 generates 5-phosphate termini on the
product 3
to the cleavage site (Fig. 6), as does cleavage of RNA I and
9 S RNA by RNase E (Refs. 29 and 34; also Fig. 6). Other endonucleases
generating 5
-phosphate ends include the bacterial enzymes RNase III
and RNase P (42, 43). In contrast, the mammalian ribonuclease, RNase A,
and analogous endoribonucleases of E. coli that degrade
substrates by cleavage at a large number of sites (i.e.
RNase M, I, I*, and R) generate products that contain 5
-OH ends (33,
44).
During the course of these experiments, we observed that a small amount of ARD-1 was produced in the absence of IPTG treatment; presumably, this was due to incomplete repression of the T7 RNA polymerase gene that controls production of ARD-1 mRNA from the T7 promoter on the expression construct. Populations of cells containing the uninduced ARD-1 gene expressed from a high copy number plasmid lose the plasmid or show DNA rearrangements that inactivate ARD-1 production (1), suggesting that excess ARD-1 protein is detrimental to E. coli cells. Similar findings have been reported for E. coli expressing RNase E from a high copy number plasmid (13). By growing cells containing the ARD-1 expression construct at 30 °C, we minimized these effects, and upon addition of IPTG, routinely achieved a 20-30-fold induction of ARD-1 expression.
Recent studies have shown that three regions of homology exist between
the cDNA sequences of ARD-1 and NIPP-1, a nuclear inhibitor of
protein phosphatase 1 (2). Two of these regions are located in a
segment that corresponds to the 5-untranslated region of ARD-1
mRNA, and are separated by a 470-bp fragment that is not present in
mRNA encoding NIPP-1 (Fig. 7). On the other hand,
NIPP-1 cDNA contains a specific 220-bp segment not found
in ARD-1 cDNA. The remaining sequences of both cDNAs
suggest that the ARD-1 polypeptide is identical to the COOH terminus of
NIPP-1 (2). Chromosome mapping studies (45) have shown that the common
3
-untranslated regions of the gene(s) encoding ARD-1 and NIPP-1 are
located on human chromosome 1. The sequences of the ARD-1 (1), NIPP-1 (2), and RNase E (13, 12) proteins all encode regions that resemble the
highly conserved 70-kDa RNA binding component of the U1 small
ribonucleoprotein complex, which is involved in mRNA splicing in
eukaryotes. Moreover, PP-1, which interacts with NIPP-1 and is the
target of NIPP-1 inhibition, has been shown to have a role in both in
the regulation of spliceosome assembly and in the splicing process
itself (46). Thus, it is tempting to speculate that ARD-1, which we
have now shown to be a site-specific single-strand endonuclease having
the cleavage specificity of RNase E, may be implicated in mRNA
splicing.
Recently, a ~65-kDa enzyme having RNase E-like cleavage specificity in vitro has been identified and partially purified from human cells (47). This enzyme and its E. coli counterpart were both shown to cleave in vitro the pentanucleotide motif (i.e. AUUUA), which has been proposed as a determinant of c-myc mRNA stability in mammalian cells. The relationship of this enzyme to ARD-1 is unknown.
We thank P. Cohen for communicating unpublished data showing the ARD-1/NIPP-1 relationship, K. J. McDowall for providing the purified NH2-terminal half of RNase E, H. Huang for providing purified full-length RNase E for some experiments, S. Lin-Chao for a gift of the BR10 oligoribonucleotide, A. C. Y. Chang for providing mRNA from B cells, and G. Aversa for providing B lymphocytes transformed by Epstein-Barr virus.