ABSTRACT
[alpha]
-Amanitin is a well-known specific inhibitor of RNA polymerase II (RNAPII)
in vitro
and
in vivo
. It is a cyclic octapeptide which binds with high affinity to the largest
subunit of RNAPII, RPB1. We have found that in murine fibroblasts exposure to
[alpha]
-amanitin triggered degradation of the RPB1 subunit, while other RNAPII subunits, RPB5 and RPB8, remained almost unaffected. Transcriptional inhibition in
[alpha]
-amanitin-treated cells was slow and closely followed the disappearance of
RPB1. The degradation rate of RPB1 was
[alpha]
-amanitin dose dependent and was not a consequence of transcriptional arrest.
[alpha]
-Amanitin-promoted degradation of RPB1 was prevented in cells exposed to
actinomycin D, another transcriptional inhibitor. Epitope-tagged recombinant human RPB1 subunits were expressed in mouse
fibroblasts. In cells exposed to
[alpha]
-amanitin the wild-type recombinant subunit was degraded like the endogenous protein, but a mutated
[alpha]
-amanitin-resistant subunit remained unaffected. Hence,
[alpha]
-amanitin did not activate a proteolytic system, but instead its binding to
mRPB1 likely represented a signal for degradation. Thus, in contrast to other
inhibitors, such as actinomycin D or 5,6-dichloro-1-
[beta]
-D-ribofuranosylbenzimidazole, which reversibly act on transcription,
inhibition by
[alpha]-amanitin cannot be but an irreversible process because of the destruction
of RNAPII.
Transcription by RNA polymerase II (RNAPII) can be inhibited by exposure of
cells to various drugs which act at different steps. The most widely used,
actinomycin D, intercalates into double-stranded DNA and blocks transcription elongation by all three polymerases
(I, II and III) indiscriminately (
1
,
2
). The nucleoside analog 5,6-dichloro-1-[beta]-D-ribofuranosylbenzimidazole (DRB) inhibits
elongation of transcription by RNAPII (
3
-
6
). DRB and the isoquinoline sulfonamides H-7 and H-8 inhibit protein kinases which phosphorylate the C-terminal domain (CTD) of the RNAPII largest subunit (RPB1)
in vitro
(
7
-
11
) and
in vivo
(
12
). However, other kinases are also inhibited by the same compounds (
13
,
14
).
In contrast to the above-mentionned inhibitors, which display pleiotropic effects, [alpha]-amanitin selectively and specifically inhibits transcription by RNAPII (
15
,
16
) and RNAPIII (
17
), RNAPII being the most sensitive. [alpha]-Amanitin is a cyclic peptide which binds to the RPB1 subunit with
high affinity (
K
d
~10
-9
M) (
18
,
19
). Several mutations confering resistance to [alpha]-amanitin have been isolated (
20
-
22
). They all map in the RPB1 subunit coding sequence and the mutant RNA
polymerases show a decreased affinity for the toxin.
In this paper, we show that in mouse fibroblasts exposed to [alpha]-amanitin the endogenous RNAPII largest subunit (henceforth
designated mRPB1) is degraded. Experiments involving epitope-tagged recombinant human RNAPII RPB1 subunits (hRPB1) expressed in
fibroblasts establish that degradation results from the interaction between
RNAPII and [alpha]-amanitin, rather than from activation of a specific proteolytic
system.
NIH 3T3 cells were cultured in Dulbecco's modified Eagle's medium supplemented
with 10% fetal calf serum (Gibco, Grand Island, NY). Cells were seeded in
tissue culture tubes or Petri dishes 24 h before experiments. [alpha]-Amanitin (stock solution at 1 mg/ml; kindly provided by Prof.
H.Faulstich), actinomycin D (stock solution at 10 mg/ml; Sigma) or DRB (stock
solution at 200 mM in dimethylsulfoxide; Sigma) were added to the medium for
the times and concentrations indicated.
The plasmid pAT7 was derived from pEVRF1 (
23
) by insertion into the
Sma
I site of an oligonucleotide encoding the B10 epitope of the human estrogen
receptor (
24
) followed by a unique
Nhe
I restriction site, in-frame with six histidines and a stop codon. The HeLa cell cDNA (5910 bp)
encoding the largest subunit of human RNAPII (
25
), hRPB1, was modified by site-directed mutagenesis so as to flank it with
Nhe
I restriction sites (one just in front of the ATG and the other in place of the
stop codon of the hRPB1 open reading frame). A point mutation was then
introduced into the coding sequence (AAC -> GAC), resulting in the replacement of Asn792 by aspartate. A similar
mutation has previously been described in a mouse cell line to give rise to an [alpha]-amanitin-resistant RNAPII (
21
). The
Nhe
I restriction fragment corresponding to the coding sequences of either the wild-type or the [alpha]-amanitin-resistant hRPB1 subunit was then inserted into the
Nhe
I site of pAT7 to yield pAT7h1 and pAT7h1[alpha]Am
r
.
The luciferase reporter plasmid pCMVL was constructed and kindly provided by Dr Uzan (Institut Pasteur): the
Hin
dIII-
Bam
HI fragment from plasmid pRSVL (
26
), coding for firefly luciferase, was placed under the control of the
cytomegalovirus early promoter from plasmid pCMV-CAT (
27
).
Transfections were performed following the standard calcium phosphate
precipitation procedure (
28
). One microgram of luciferase reporter (pCMVL) and 10 [mu]g of either a recombinant hRPB1 expression vector (pAT7h1 or pAT7h1[alpha]Am
r
) or the pSP64 plasmid (Promega), as a carrier, were used for each 20 cm
2
dish. [alpha]-Amanitin was added to the medium 48 h after transfection and the
cells were lysed 24 h later, in a buffer containing 50 mM Tris-phosphate, pH 7.8, 5 mM MgCl
2
, 15% glycerol, 1% Triton X-100, 1 mM 2-mercaptoethanol. Luciferase activity in the lysates was determined with a Berthold LUMAT luminometer (
29
).
The RPB1 subunit was detected by the POL3/3 monoclonal antibody kindly provided
by Prof. E.Bautz (
30
). Identical results (data not shown) were obtained using the monoclonal
antibody 8WG16 (
31
), which is directed against an epitope of the C-terminal domain. Histidine-tagged hRPB5 was overproduced in
Escherichia coli
using a PET3a-derived vector. The soluble protein was purified by metal chelate affinity
chromatography (Pharmacia) to near homogeneity. The purified fraction was injected into mice as described previously (
32
). Hybridoma culture supernatants were screened by Western blot analysis, using either partially
purified RNAPII (
33
) or recombinant hRPB5 protein. The selected hybridoma (4H7) was subcloned twice
on soft agar. Ascites fluid was produced by classical methods (
34
) and precipitated with 50% ammonium sulfate. In a Western blot analysis of a whole
HeLa cell lysate, the 4H7 monoclonal antibody recognized a single band co-migrating with the bacterially produced recombinant hRPB5.
Two peptides derived from the sequence of the hRPB8 subunit (NH
2
-DIFDVKDIDPEGKKFC-COOH and NH
2
-DGTLDDGEY- NPTDDH
2
RPSRADQC-COOH) were synthesized and coupled to ovalbumin through their extra C-terminal cysteine residue. Polyclonal antibodies were raised by serial subcutaneous injection into rabbits. The sera
were tested on Western blots of either partially purified RNAPII (
33
) or recombinant protein. One serum (853b) was capable of detecting the
endogenous hRPB8 subunit as a major band in a Western blot assay using a HeLa
whole cell lysate.
After drug treatment, the cells were rapidly washed with ice-cold phosphate-buffered saline and lysed in Laemmli buffer (60 mM Tris-HCl, pH 6.8, 10% glycerol, 2% SDS, 1% 2-mercaptoethanol and 0.002% bromophenol blue). The lysates were heated
for 8 min at 90oC and loaded on 5 or 15% polyacrylamide-SDS gels. After electrophoresis, proteins were electrotransfered onto nitrocellulose membranes (0.45 [mu]m; Schleicher & Schuell). The membranes were blocked in Tris-buffered saline (20 mM Tris-HCl, pH 7.6, 140 mM NaCl, 0.2% Tween 20)
containing 5% non-fat dry milk before incubation with the appropriate antibody. The
endogenous RNAPII mRPB1, mRPB5 and mRPB8 subunits were detected with the
monoclonal POL 3/3, the monoclonal 4H7 and the polyclonal 853b antibodies respectively, while the tagged recombinant subunit (hRPB1) was visualized with the monoclonal antibody B10 directed against the B region of the human estradiol receptor (
24
).
Nuclei were isolated from NIH 3T3 cells treated or not with 5 [mu]g/ml [alpha]-amanitin for 8 or 12 h and run-on assays were performed following established procedures
(
35
) with previously described adaptations (
36
). In brief, 2 * 10
7
nuclei were allowed to transcribe for 20 min at 30oC in the presence of [[alpha]-
32
P]UTP and the resulting RNAs were isolated and hybridized to nitrocellulose
strips on which 10 [mu]g of the indicated linearized and denatured plasmids had been transfered
after electrophoresis in 1% agarose gel. Final washes were at 65oC in 2* SSC (0.3 M NaCl/citrate buffer). Transcription of the various genes
was probed with plasmids pAL41 (
37
) for the cytoplasmic actin gene, pGPDH (
38
) for the glyceraldehyde phosphate dehydrogenase gene and pMSE2 (
39
) (kindly provided by Dr J.P.Bachellerie) for the 18S ribosomal RNA. Probing
with pBR322 was used as a negative control.
Two forms of the largest subunit of mammalian RNAPII coexist in similar
proportions in fibroblasts (
12
,
40
). Both forms are readily distinguished by Western blot: the IIo form, which is
multiphosphorylated on its C-terminal domain (CTD), migrates as a 240 kDa protein, whereas the
unphosphorylated IIa form migrates as a 214 kDa protein (
41
). Using the POL3/3 monoclonal antibody directed against the core of the largest subunit (
30
), we noticed that the signal corresponding to both forms of mRPB1 in mouse NIH
3T3 cells disappeared gradually when [alpha]-amanitin was added to the culture medium (Fig.
1
A). The disappearance rate was dependent upon the concentration of the toxin:
the largest subunit was no longer detectable after 24 h in the presence of 5 [mu]g/ml [alpha]-amanitin (~8 h half-life), after 8 h in the presence of 20 [mu]g/ml h [alpha]-amanitin (~4 h half-life) and after 4 h in
the presence of 100 [mu]g/ml [alpha]-amanitin (<2 h half-life). The quantity of a protein present in a cell is a
steady-state resulting from synthesis and degradation. [alpha]-Amanitin also triggered the disappearance of the mRPB1
subunit in the presence of cycloheximide (10 [mu]g/ml), to inhibit protein synthesis (Fig.
1
B). Hence, the disappearance of the mRPB1 subunit is due to a toxin-dependent degradation process.
RNAPII is composed of several subunits (reviewed in
42
-
45
). Therefore, we next questioned whether other RNAPII subunits would be degraded
at the same rate as mRPB1 upon [alpha]-amanitin treatment (Fig.
1
A). In lysates from [alpha]-amanitin-treated cells, the levels of mRPB5 and mRPB8 detected by
Western blot decayed much slower than subunit mRPB1; the corresponding signals
remained close to control after 24 h treatment with 20 [mu]g/ml amanitin.
Thus, [alpha]-amanitin specifically triggers degradation of the mRPB1 subunit. The mRPB5 and the mRPB8 subunits are present in all three
classes of RNA polymerases (
46
). It is possible, therefore, that degradation of these common subunits is
somehow prevented as long as RNAPI and RNAPIII are preserved.
Degradation of the mRPB1 subunit might have been a consequence of transcriptional arrest. Therefore, the fate of mRPB1 was investigated in
cells exposed to two transcriptional inhibitors acting through different mechanisms. Actinomycin D is a DNA intercalator which inhibits transcription (
1
,
2
). DRB is an inhibitor of the TFIIH-associated CTD kinase (
47
), which inhibits elongation of transcription (
3
-
6
).
In the presence of actinomycin D, the mRPB1 signals decayed, but much more
slowly than in the presence of [alpha]-amanitin (Fig.
2
). Indeed, mRPB1 was still visible in cells treated for 24 h with actinomycin D,
while it had disappeared in cells treated with [alpha]-amanitin, even at 5 [mu]g/ml (Fig.
1
A). Similarly, a slow decay in the mRPB5 and mRPB8 signals was also observed
after 24 h in the presence of actinomycin D (Fig.
2
A). When both [alpha]-amanitin and actinomycin D were added simultaneously, the decay
rate was the same as with actinomycin D alone (Fig.
2
A); it was slower than in the presence [alpha]-amanitin alone (Fig.
1
A). Whatever the mechanism involved, these results clearly indicate that [alpha]-amanitin-promoted degradation of the mRPB1 subunit was not a
consequence of inhibition of transcription. On the contrary, it was prevented
in the presence of actinomycin D.
It had been reported that
in vivo
transcriptional inhibition by [alpha]-amanitin is a slow process (
48
,
49
). Several hours exposure are required and the higher the [alpha]-amanitin concentration, the faster inhibition occurs. On the basis
of the values obtained from
in vitro
transcription assays (RNAPII activity being 100% inhibited in the presence of
0.1 [mu]g/ml [alpha]-amanitin;
15
), these observations have been interpreted as a consequence of the slow entry
of [alpha]-amanitin into the cells: higher [alpha]-amanitin concentrations would increase the drug
diffusion rate through the cell membrane and expedite transcriptional
inhibition.
To examine whether transcriptional arrest is nevertheless related to the
disappearance of subunit mRPB1, nuclei were isolated from [alpha]-amanitin-treated cells and allowed to transcribe
in vitro
(run-on assays). Transcription of two class II genes (cytoplasmic actin and glyceraldehyde phosphate dehydrogenase) was followed and compared with that of a class I gene (18S ribosomal RNA)
taken as a control (Fig.
3
, upper part, and quantification in Table
1
). As expected, the rate of ribosomal gene transcription was unaffected by drug
treatment. In contrast, run-on transcription of the selected class II genes was slowly abolished with
increased time of treatment, reflecting the decreased number of active
polymerase molecules engaged in transcription of these genes. In parallel, the
amount of mRPB1 subunit present in the same nuclei was followed by Western blot
analysis (Fig.
3
, lower part, and quantification in Table
1
). Strikingly, this reduction in class II gene transcription roughly followed
the drop in both forms of the mRPB1 protein detected in nuclei by Western blot.
In the presence of [alpha]-amanitin, mRPB1 decayed with a concentration-dependent rate: the mRPB1 subunit became undetectable within
a few hours when concentrations up to 100 [mu]g/ml were used, whereas at least two other subunits, mRPB5 and mRPB8, remained unaffected. Decay of the mRPB1 polypeptides paralleled inhibition of nuclear RNAPII transcription activity, suggesting that
binding of [alpha]-amanitin by mRPB1 results in very rapid degradation of this
subunit. Inhibition of transcription by [alpha]-amanitin is a slow process in cultured cells (
49
). Slow and drug concentration-dependent [alpha]-amanitin-promoted RPB1 degradation was observed in all cultured cell systems that we
tested (mouse embryos and murine, rat and human cells; data not shown). The
presence of the RPB1 subunit might be used as a simple test for the efficiency
of [alpha]-amanitin inhibition.
[alpha]-Amanitin-promoted RPB1 degradation is consistent with previous
pioneering observations. Indeed, treatment of HeLa cells with [alpha]-amanitin was reported to result in loss of RPB1 immunohistostaining
(
52
). Lysates from [alpha]-amanitin-treated CHO cells showed decreasing RNAPII enzymatic activity
and this decrease was accompanied by a loss of [alpha]-amanitin binding capacity of cell extracts (
53
). Immunoprecipitation indicated that the mRPB1 subunit disappeared from the
lysates at higher rates than other subunits (
54
). However, both latter studies used low salt cell lysates in which <10% of the mRPB1 subunit molecules are extractible (
55
) and many artefacts may affect immunohistochemistry. Hence, these pioneering
observations did not indicate degradation. For that reason, [alpha]-amanitin has been extensively used in nuclear ultrastructural
studies with little consideration of the possible disappearance of RNA
polymerase II.
The degradation process is specific and directly related to interaction of the
largest subunit with [alpha]-amanitin, since a recombinant subunit carrying an [alpha]-amanitin resistance mutation is not degraded. [alpha]-Amanitin binds to and rapidly inhibits
RNAPII in
in vitro
assays. However, we could not observe any disappearance of the mRPB1 subunit after incubating nuclei prepared for run-on assays for up to 16 h with [alpha]-amanitin (data not shown). This observation demonstrates that
[alpha]-amanitin has no proteolytic action
per se
; it binds and inhibits RNAPII first, then targets the mRPB1 subunit for
degradation. Since its affinity for RNAPII is very high (
K
d
~10
-9
M), entry of [alpha]-amanitin into the cells is likely to be the limiting step. Once the
mRPB1 subunit is bound to [alpha]-amanitin, it is likely to be degraded rapidly. Genetic studies
showed that mutations of the yeast endoprotease KEX2 located in the Golgi
apparatus suppresses numerous rpb-1 mutations in yeast (
56
); unfortunately [alpha]-amanitin does not enter yeast cells (A.Sentenac, personal
communication) and the mammalian KEX2 homolog is presently uncharacterized.
Further studies are required to establish the RPB1 degradation mechanism.
In cultured cells, transcriptional inhibition by actinomycin D (
57
,
58
), DRB (
3
-
5
,
59
-
62
) or isoquinolinesulfonamides (H-8 and H-7) (
12
) is fast and reversible. In contrast, transcriptional inhibition by [alpha]-amanitin is slow and irreversible: once RPB1 has been degraded,
cell death should follow, since the largest RNAPII subunit is very likely
required for transcription of its own gene.
We are much indebted to Prof. H.Faulstich for his generous gift of [alpha]-amanitin and to Drs J.Acker and Y.Lutz for the production of
antibodies. Many thanks to Dr O.Jeanjean for stimulating discussions. This work
has been supported by grants from the Association pour la Recherche sur le
Cancer (nos 6250 and 6353), the Human Frontier Science Program (no. RG-496/93), the Ligue Nationale contre le Cancer and the Korean Science and Engineering Foundation (KOSEF).
REFERENCES
Return