ABSTRACT
Nearly 1 000 000 copies of Alu interspersed elements comprise
~
5% of human DNA. Alu elements cause gene disruptions by a process known as
retrotransposition, in which dimeric Alu RNA is a presumed intermediate.
Dimeric Alu transcripts are labile, giving rise to stable left monomeric scAlu
RNAs whose levels are tightly regulated. Induction of Alu RNA by viral
infection or cell stress leads to a dramatic increase in dimeric Alu
transcripts, while scAlu RNA increases modestly. Each monomer of the dimeric
Alu element shares sequence homology with the 7SL RNA component of the signal
recognition particle (SRP). The SRP protein known as SRP9/14 is also found in a
discrete complex with scAlu RNA, although whether dimeric Alu RNA is associated
with SRP9/14 had been unknown. Here we show that antiserum to human SRP9
immunoprecipitates both scAlu RNA and dimeric Alu RNAs and that these RNPs
accumulate after adenovirus infection, while levels of SRP9, SRP14, SRP54 and
7SL SRP RNA are unaffected. Dimeric Alu RNAs are also associated with the La
protein, indicating that these are indeed nascent RNA polymerase III
transcripts. This report documents that induced Alu transcripts are assembled
into SRP9/14- containing RNPs
in vivo
while SRP levels are unchanged. Implications for Alu RNA metabolism and
evolution are discussed.
Alu sequences comprise the most abundant class of short interspersed elements in
primate DNA (
1
-
3
).
De novo
insertion of Alu elements causes human gene disruptions and is believed to
occur via Alu RNA that is synthesized by RNA polymerase (pol) III (
4
; reviewed in
3
). Alu repeats are ~285 nt long, composed of two non-identical monomers that are connected by a 20 nt spacer and followed
by an A-rich or poly(A) tract. Primary Alu transcripts are derived from multiple
loci and vary in sequence beyond their A-rich tracts (
5
-
7
). This variability is due to the fact that Alu elements must rely on fortuitous
downstream transcription terminators for nascent RNA 3'-end formation. As a result, Alu primary transcripts are of
heterogeneous length, ranging from 300 to 450 nt (
6
-
11
). Some of these full-length transcripts are shortened to a set of ~120 nt RNAs, representing Alu left monomer transcripts that
accumulate as stable small cytoplasmic (sc)Alu RNAs of unknown function, while
the rest appear to be degraded with rapid kinetics (
7
-
9
,
12
). Although by reducing the amount of dimeric Alu RNA available for
retroposition, production of scAlu RNA represents one way to decrease Alu
transposition, preferential stabilization and cytoplasmic compartmentation
suggest an independent function for scAlu RNA.
Alu repeats are ancestrally related to the 7SL RNA component of the signal
recognition particle (SRP), a ribonucleoprotein (RNP) that recognizes signal
sequences on nascent polypeptides that are destined for secretion (
13
,
14
). Each of three SRP activities reside in a distinct domain of the particle (
15
). The first ~100 nt and last ~50 nt of 7SL RNA share nearly 90% homology with each monomer of the
Alu sequence. Two SRP polypeptides, SRP9 and SRP14, form a stable heterodimer
known as SRP9/14 that associates with the Alu-homologous region of 7SL RNA to form the translation arrest domain of SRP,
while SRP19, SRP54, SRP68 and SRP72 associate with the internal ~150 nt of 7SL RNA, referred to as the S domain, which shares no homology
with Alu (
13
,
14
,
16
-
18
).
Several lines of evidence suggest that interaction between the human SRP9/14
protein and Alu RNA influences Alu transcript metabolism and
retrotransposition. SRP9/14 accumulates to levels 10- to 20-fold higher than other SRP subunits, including 7SL RNA, specifically
in primate cells (
19
,
20
). Deregulation of SRP9/14 occurred during the evolutionary period that
encompassed a dramatic change in the rate of Alu retrotransposition in primates and was associated with a substantial structural expansion of the C-terminus of SRP9/14 (
19
). A transgene-mediated increase in the level of human SRP14 in cells is associated with
a corresponding increase in the level of scAlu RNA and this appears to occur at
the expense of full-length Alu transcripts (
7
,
21
). This observation, in conjunction with the relative lability of full-length Alu RNA, suggests that although SRP9/14 may stabilize scAlu RNA
(and/or facilitate its production), this protein may not associate efficiently
with full-length Alu RNAs
in vivo
(
6
,
12
,
20
). Although full-length Alu transcripts were found to sediment in sucrose gradients with a
mobility consistent with association with protein, no protein component was
identified (
6
). Therefore, in order to better understand the metabolism of Alu RNA and the
potential for an SRP-related function, it is important to determine the RNP nature of cellular
Alu RNAs. Yet, whether SRP9/14 associates with full-length Alu RNA
in vivo
had remained unknown (
20
).
Viral infection as well as heat shock and other forms of cell stress stimulate
Alu RNA expression (
10
,
22
,
23
). Under these conditions, as well as in other cases where Alu RNA is induced,
full-length Alu transcripts increase dramatically while scAlu RNAs increase
less than 5-fold (
6
,
11
,
12
,
24
). Some of the multiple adenoviral proteins required for Alu induction were
previously known to act as RNA processing and transport factors (
22
; reviewed in
25
). Thus, although Alu RNA induction was demonstrated at the transcriptional
level, RNA accumulation may also involve alterations in Alu RNA-associated proteins, however, this aspect of Alu RNA induction had not
been investigated (
11
,
22
-
24
).
Certain fundamental issues regarding an SRP9/14-like protein in human cells had also remained unresolved. Biochemical
assays revealed an scAlu RNA binding activity that co-purified with two polypeptides of ~18 and ~10 kDa from HeLa cells (
21
). The ~18 kDa polypeptide was identified as human-specific SRP14 polypeptide, which contains an extended C-terminus accounting for its increased size relative to rodent
SRP14, while evidence that the ~10 kDa polypeptide is SRP9 was indirect (
19
,
21
,
26
). Thus, in light of the variability of SRP9/14 in primates, its regulation
independent of other SRP subunits and propensity to undergo degradation to
smaller proteins, it is important to identify the ~10 kDa protein that co-purified with human SRP14 and scAlu RNA binding protein (
9
,
19
-
21
).
We also wanted to examine whether full-length Alu RNAs are associated with this Alu RNA binding protein
in vivo
. Further, because mechanisms that induce Alu RNA expression could conceivably
lead to changes in the association of 7SL RNA and SRP9/14, we also examined
whether adenoviral infection leads to an altered form of SRP. Our results
demonstrate that both full-length Alu and scAlu RNAs are assembled into SRP9/14-containing RNPs in uninfected cells and that these are substantially
increased after infection with adenovirus. The amount of 7SL RNA that remains
associated with SRP9/14 as well as the amount of total SRP9/14 appears to be
unchanged by infection.
Human SRP9 cDNA was overexpressed as a glutathione S-transferase (GST)-SRP9 fusion protein from a pGEX-4T-2 plasmid (Pharmacia, Piscataway) designated pGST-hSRP9 (
26
). After purification by glutathione affinity chromatography, SRP9 was released
from immobilized GST fusion protein by cleavage with thrombin (Pharmacia) and
was used to immunize rabbits. Anti- SRP serum was provided by F. Miller (Federal Drug Administration,
Bethesda, MD); this and other anti-SRP autoimmune sera recognize SRP54 specifically and do not recognize
SRP9/14 (
27
,
28
; data not shown). Anti-La and anti-Sm autoimmune sera were provided as standards from the Centers for
Disease Control (CDC, Atlanta, GA). Affinity-purified antibodies raised against an N-terminal peptide of human SRP14 and their use in chemiluminescent
Western blotting were as described previously (
19
).
RNA electrophoretic mobility shift assays (EMSA) were performed as described (
9
). Briefly, scAlu and scB1
d40
[
32
P]RNAs were synthesized from T7 promoter-containing scAlu and scB1
d40
DNA templates in the presence of [[alpha]-
32
P]GTP and gel purified prior to use (
29
). A mixture of scAlu and scB1
d40
[
32
P]RNAs and protein were incubated in 15 [mu]l reactions containing 10 mM Tris-HCl, pH 7.5, 80 mM KCl, 5 mM MgCl
2
, 0.1% Triton X-100, 1 mM DTT, 1 mM EDTA, 4 U RNasin, 5% glycerol (EMSA buffer) and 100 ng
poly(rG). After a 40 min incubation at room temperature, samples were analyzed
on non-denaturing 6% polyacrylamide gels as previously described for EMSA (
9
) or diluted to 300 [mu]l with NET-2 (150 mM NaCl, 50 mM Tris-HCl, 0.05% Nonidet-P40) for immunoprecipitation (
30
). The source of SRP9/14 used for
in vitro
RNP reconstitutions was the heparin-agarose fraction purified from HeLa cells (
21
).
For immunoprecipitations, antibodies were first adsorbed onto protein A-Sepharose beads, washed with NET-2 and then incubated with cell-derived extracts or RNP reconstitution reactions for 90 min
at 4oC, washed four times with NET-2 and RNA purified by phenol/chloroform extraction and ethanol
precipitation (
30
). Carrier tRNA (1 [mu]g) was included just prior to ethanol precipitation of RNP reconstitutions.
The precipitated RNA was analyzed directly by polyacrylamide gel electrophoresis and autoradiography or by Northern blot analyses after hybridization to
oligo [
32
P]DNA probes complementary to Alu and 7SL RNAs as described previously (
21
).
Adenovirus type 2 was derived from high titer stocks provided by B. Howard's
laboratory (
11
). Cytoplasmic extracts were produced from adenovirus-infected and control HeLa cells by a standard hypotonic lysis procedure
followed by removal of nuclei by low speed centrifugation (
30
,
31
). Extracts from infected and uninfected cells were quantitated for protein
content by a BioRad colorimetric assay and visually compared by SDS-PAGE with Coomassie blue staining.
Previous attempts to identify the ~10 kDa polypeptide that co-purified with scAlu RNA binding activity by direct amino acid
sequencing were unsuccessful (
21
). We raised antisera against purified recombinant ~10 kDa protein that was expressed in bacteria from human SRP9 cDNA. This
antiserum was used to probe samples representative of various stages of
purification of Alu RNA binding activity by Western blotting (
21
,
26
). Both anti-SRP9 and a previously characterized anti-SRP14 serum (
19
) recognized polypeptides of the appropriate size (Fig.
1
A). By comparing the relative amounts of SRP9 and SRP14 in the ammonium sulfate
(AS) lane with the heparin-agarose (Hp) lane it appears that the amount of SRP9 antigen increased
relative to the amount of SRP14 during the multiple chromatographic steps
employed. This suggests that SRP14 is in excess over SRP9 in the crude extract
represented by the AS fraction. In any case, since 48 [mu]g protein in the AS fraction contained less SRP9 (Fig.
1
A, lower panel, lane AS) than 10 ng protein in the final Hp fraction (lane Hp),
the increase in antigenic specific activity observed here is somewhat higher
than the increase in RNA binding activity that was observed previously,
although the significance of this is unknown (
21
).
Preliminary experiments revealed that detection of Alu RNA from anti-SRP9 precipitates of uninduced HeLa cell cytoplasm would require
substantial amounts of extract as well as antiserum. This was expected since
Alu and scAlu RNAs accumulate only to 100-1000 copies/cell (
6
,
9
) and theoretically should occupy no more than 0.01% of the total cellular
SRP9/14 antigen. After immunoprecipitation, RNA was purified and analyzed by
Northern blot using an oligonucleotide probe specific for Alu RNA (Fig.
2
). This reproducibly revealed that both the 300-450 nt Alu transcripts and scAlu RNA were immunoprecipitated by anti-SRP9 (lane 2) but not by preimmune serum (lane 3). As alluded to in
the Introduction, full-length Alu primary (1o) transcripts appear diffuse because of their expected size
heterogeneity (
7
). The broad band of Alu RNA precipitated by anti-SRP9 suggested that nascent unprocessed Alu transcripts are associated
with SRP9/14. To confirm this we used antiserum to the human La protein, a pol
III transcription termination factor that transiently associates with all pol
III nascent transcripts by binding to their common oligo(U) 3'-termini (
33
-
35
). Alu RNAs precipitated by anti-La also exhibited a broad size distribution (lane 5), but markedly less
scAlu species as compared with anti-SRP9. We also noted an Alu-homologous RNA slightly larger than scAlu that is barely detectable
in HeLa input RNA (lane 1) which was reproducibly enriched by anti-La serum (Fig.
2
, lane 5, scAlu*) (see Discussion). Another control serum, anti-Sm, precipitated little if any Alu-homologous RNA (lane 4). These experiments showed for the first time
that uninduced HeLa cells accumulate full-length Alu transcripts in the form of SRP9/14-containing RNPs as well as scAlu RNPs.
Cytoplasmic extract isolated from adenovirus-infected HeLa cells was also subjected to immunoprecipitation. It was
previously shown that 7SL RNA levels do not change after infection with
adenovirus, although other subunits of SRP have not been examined (
11
,
22
). Extracts from infected and uninfected cells that were equalized for their
protein and 7SL RNA content (not shown) were immunoprecipitated separately with
anti-SRP54 and anti-SRP9 and subjected to Northern blot analysis (Fig.
3
). The results indicate for the first time that the 300-450 nt Alu transcripts and the scAlu RNAs that are induced by adenovirus
infection were associated with SRP9/14 (Fig.
3
A, lane 1). Because Alu RNA is easily detectable in infected cells, we chose to
conserve infected extract by using less than required to detect Alu RNA from
immunoprecipitates of uninfected extract. Thus, although this amount was
clearly sufficient for detection of the Alu RNA that was immunoprecipitated by
anti-SRP9 from infected cells, Alu RNA was barely detectable in the uninfected
cells in this experiment, as expected (lane 2). The high degree of Alu RNA
induction may be seen by comparing lanes 1 and 2 of Figure
3
A and B, which represent different probings of the same blot. The association of
full-length Alu RNA and scAlu RNA with SRP9/14 was specific, since antisera
directed to the SRP54 subunit did not precipitate these RNAs (lanes 3 and 4).
The full-length Alu transcripts that accumulated in infected cells were also
efficiently precipitated by anti-La (not shown).
It was previously shown that SRP can be assembled
in vitro
in the absence of SRP9/14 and that these RNPs recognize signal sequences and
promote nascent polypeptide translocation but exhibit no translation arrest
activity (
18
). This raises the possibility that SRP9/14-deficient SRPs might exist in cells under certain conditions even though
the level of 7SL RNA may appear unchanged (
11
,
22
). Therefore, we wanted to examine whether Alu RNA induction might lead to a
significant amount of SRPs that lack SRP9/14. As one approach to evaluate this,
we reprobed the blot shown in Figure
3
for 7SL RNA (Fig.
3
). In addition to providing a control for Figure
3
, these results revealed that anti-SRP9 and anti-SRP54 each immunoprecipitated similar amounts of 7SL RNA from
uninfected (Fig.
3
, lanes 1 and 3) and infected (lanes 2 and 4) cell extracts, although anti-SRP54 immunoprecipitated 7SL RNA more efficiently than did anti-SRP9 from both extracts. Since anti-SRP9 immunoprecipitated comparable amounts of 7SL RNA from
infected and uninfected cell extracts, these results suggested that infection
did not lead to a significant amount of SRP9/14-deficient SRPs.
As another approach to examine whether viral infection leads to SRP9/14-deficient SRP complexes, we examined SRP isolated from infected and
uninfected cell extracts for their SRP9 and SRP14 content. For this we
immunoprecipitated SRPs from the two extracts with antiserum specific for SRP54
and compared their SRP9 and SRP14 content by Western blot analysis. As shown in
Figure
4
, both SRP9 and SRP14 polypeptides were co-immunoprecipitated in comparable amounts from infected and uninfected cell
extracts by anti-SRP54 (lanes 3 and 4) but not by non-immune serum (lanes 5 and 6). Since the anti-SRP54 serum does not recognize either SRP9 or SRP14 directly (
28
; unpublished observation), these data indicate that SRP9/14 was co-immunoprecipitated with SRP54 by virtue of its association with 7SL RNA as
a subunit of SRP. These results, together with those shown in Figure
3
provide the first evidence to indicate that SRP9/14 remains associated with 7SL
RNA in extracts from adenovirus-infected cells.
As alluded to above, SRP protein levels had not previously been examined in
adenovirus-infected cells. The levels of total SRP9, SRP14 and La antigen were
determined by Western blot analyses of equal amounts of protein isolated from
infected and uninfected cell extracts. The levels of these three Alu RNA
binding proteins remained virtually unchanged after infection (Fig.
5
A), as did SRP54 (not shown).
The major conclusion from this work is that the Alu primary transcripts as well
as the scAlu RNAs that are induced by adenovirus are assembled into SRP9/14-containing RNPs. In addition, we observed that HeLa cells contain a
sufficient capacity of accessible SRP9/14 to accommodate a large increase in
Alu RNA without affecting SRP integrity as it has been assayed here. Moreover,
the fact that SRP9/14 RNA binding activity (Fig.
5
B) was essentially unchanged after Alu induction supports the idea that SRP9/14
is in substantial excess over its RNA ligands, 7SL RNA and Alu RNA, even in
viral-infected HeLa cells. Thus, after induction, full-length Alu transcripts are organized into SRP9/14-specific RNPs in the context of a fixed amount of SRP.
The findings reported here provide an opportunity to consider mechanisms of Alu
RNA metabolism. First, the fact that anti-La serum immunoprecipitates a small amount of scAlu RNA but a relatively
large amount of full-length Alu transcripts as compared with anti-SRP9/14 (Fig.
2
) supports the proposal that most scAlu RNA is not the nascent product of pol
III, but is derived by RNA processing from Alu primary transcripts (
7
,
8
). Since SRP9/14 is primarily cytoplasmic, it was previously unclear if it would
be able to associate with nascent Alu transcripts (
20
). The existence of nascent Alu-SRP9/14 RNPs suggests that scAlu-SRP9/14 RNPs are derived from these. After adenovirus infection
Alu sequences become derepressed to yield an ~50-fold increase in full-length Alu transcripts, while scAlu RNA levels increase <5-fold (
11
,
22
). Increases in Alu expression induced by transfection, heat shock and protein
synthesis inhibitors also lead to preferential accumulation of full-length Alu transcripts, indicating that scAlu RNA levels are more tightly
regulated than nascent Alu transcripts (
6
,
10
,
12
). An increase in the level of SRP9/14 in cells is associated with an ~5-fold increase in scAlu RNA levels, suggesting that SRP9/14 may be
able to exert a modest regulatory influence over scAlu RNA (
21
). These cumulative results are consistent with a pathway for scAlu RNA
expression that becomes overwhelmed by the large amount of nascent Alu
transcripts that are induced during viral infection and cell stress and that
SRP9/14 levels alone do not determine Alu RNA levels. Presumably, limiting
amount of an as yet unidentified factor governs the levels to which scAlu RNA
can accumulate. Identification of this putative factor(s) and its relationship
to SRP9/14 and the Alu RNA expression pathway may shed light on Alu
retrotransposition as well as provide clues to the function of Alu RNPs.
An unexpected finding was that anti-La serum immunoprecipitated an Alu left monomer transcript that is
slightly larger than scAlu (Fig.
2
, lane 5, scAlu*). A similar left monomer transcript was previously detected as
a prominent RNA, for which the coding DNA was localized to human chromosome 15
(see
7
, fig. 3A, lane 15, and
21
, fig. 7B). This La-associated small Alu RNA most likely represents a nascent pol III
synthesized transcript, whose biogenesis may be due to termination in the Alu
intermonomeric A+T-rich spacer. Nucleotide substitutions in this linker region of Alu
elements may create a (dT)
4
pol III termination signal (
36
). Alternatively, scAlu* RNA may represent a unique locus that harbors an Alu
free-left monomer (
37
,
38
). Since La is found associated with nuclear precursors of small cytoplasmic
RNAs, it may be of interest to determine whether La-associated scAlu* RNA is efficiently compartmentalized to the cytoplasm,
as is genuine scAlu RNA, or if it is primarily nuclear (
7
,
9
,
39
).
Finally, we would like to understand why changes in the structure and
deregulation of the SRP9/14 protein occurred during primate evolution.
Specifically, whether selection for these traits was related to Alu
retrotransposon activity (
19
). Amplification of the majority of Alus in the human genome ceased ~30 000 000 years ago and a substantial proportion of these have since
accumulated mutations, including in their transcriptional control elements.
Therefore, ancestral primates may have been able to produce higher levels of
Alu RNA as compared with human cells. In addition, non-mutated Alu RNA sequences might have been higher affinity ligands for
SRP9/14 as compared with the Alu sequences induced in human cells. In the
ancestral primate that presumably existed prior to the deregulation of SRP9/14,
a massive induction of Alu RNA might indeed have interfered with SRP function.
In this scenario, the genetic deregulation that accompanied the 10- to 20-fold increase in SRP9/14 could plausibly have been an adaptive
response that allowed induction of Alu RNAs while protecting the integrity of
the SRP. This reasoning implies that the ability to induce Alu RNA was
beneficial to the species. We wish to emphasize that although Alu RNA induction
is not accompanied by disruption of SRP in human cells, this does not exclude
the possibility that Alu-SRP9/14 RNPs may play a role in translation (
40
). It has been reported that cellular mRNAs are blocked at translational
elongation after infection by adenovirus (
41
,
42
). Thus, it is tempting to speculate that Alu-SRP9/14 RNPs may be involved in the modulation of translational
elongation that occurs upon infection of HeLa cells. The results reported here
suggest that it is reasonable to examine this possibility in the future.
We are grateful to G. Humphrey and members of the Howard laboratory for generously providing adenovirus and extracts, to F. Miller for anti-SRP sera, to the CDC for standard sera, to E. Englander for critical reading and to A. Sakulich, LMGR members and
anonymous reviewers for discussion and suggestions. K.H. was supported by the NIH Research Scholars Program of the Howard Hughes Medical Institute. D.Y.C. was supported by an Interpersonnel Act between NICHHD and the Department of Biochemistry,
University of Maryland Medical School.
REFERENCES
Return