ABSTRACT
Jurkat cells, a human T lymphocyte line that can be induced to synthesize and
secrete interleukin 2, contain a factor that binds interleukin 2 mRNA. Binding
can be demonstrated by formation of a complex detectable by gel
electrophoresis. The binding is sequence specific and occurs in the 3
'
-non-coding region, within 160 nt of the end of the coding region, at or
near a site on the mRNA that is rich in A and U residues. However, it appears not to be due to known AU binding factors. The
factor is protease sensitive and binds non-covalently to interleukin 2 mRNA. It behaves like a protein of molecular weight 50 000-60 000 after UV-induced cross-linking to the mRNA. Preparations of the binding factor
also protect interleukin 2 mRNA against degradation by a recently described
RNasin-resistant endoribonuclease activity in Jurkat cells. Protection occurs
under the same conditions required to generate the gel-retarded complex.
Rates of transcription and degradation control the steady-state levels of cellular mRNAs. The lifetimes of mRNAs in mammalian cells
vary from minutes to days, but those encoding proteins that change
concentration rapidly are often short lived, allowing for pre-translational control (
1
,
2
). The synthesis of the lymphokine interleukin 2 (IL2) and its mRNA is rapidly
induced by antigenic or mitogenic stimulators of T helper lymphocytes and upon
their removal the mRNA dies away equally rapidly. This is consistent with a
relatively short half-life of the mRNA under these conditions, of between 40 and 80 min (
3
). Other cytokines having short lived mRNAs include GM-CSF (
4
) and interferon [beta] (
5
). The stabilities of a number of short lived cytokine mRNAs are regulated. For
example, the mRNAs for IL2, IFN[gamma], GM-CSF and TNF[alpha] (but not those for c-Myc and c-Fos) are stabilized by signaling of T helper
lymphocytes through the CD28 cell surface structure (
6
). The same signaling pathway can also stabilize the [alpha] and [beta] transcripts of the IL2 receptor (
7
). Induction of GM-CSF synthesis in some cell lines is caused by stabilization of its mRNA,
rather than an increase in its transcription rate (
8
). Similarly, stabilization of constitutively transcribed mRNA by calcium influx
induces IL3 production in a mast cell line (
9
).
Only limited information is available about the ribonucleases responsible for
the degradation of mRNA
in vivo
. All cells contain high levels of RNase A-type RNases or neutral, non-specific RNases, but under reductive intracellular conditions (
10
,
11
) these are efficiently inhibited by a tightly bound ubiquitous protein
inhibitor (
12
,
13
). This inhibitor is commercially available as placental RNasin. The
endoribonuclease RNase E, the product of the
ams
/
rne
/
hmp1
gene, is at least partly responsible for the degradation of mRNA in
Escherichia coli
(
14
,
15
). The nature of the ribonucleases responsible for mRNA degradation in
eukaryotic cells is not known. Given the broad distribution and effectiveness
of RNasin, RNases sensitive to it seem unlikely to be directly involved and
there is therefore a particular interest in RNasin-resistant RNases as candidate mRNases. The Vhs protein of herpes simplex
virus-1 is a known RNasin-insensitive mRNase, which is responsible for the degradation of mRNA
upon viral infection (
16
). Candidate mRNases include an RNasin-resistant exonuclease that preferentially degrades polysome-associated histone mRNA
in vitro
(
17
) and the RNase in a related system that preferentially attacks polysome-associated c-Myc mRNA (
18
). We recently described an RNasin-resistant RNase, present in a T lymphocyte cell line, that preferentially
attacks IL2 mRNA, relative to [beta]-globin mRNA (
19
). A
Xenopus
RNase has been described that cleaves the
HoxB7
gene product endonucleolytically (
20
). A segment of the 3'-NC part of the target mRNA directs the RNase sensitivity in that
system.
The regulation of mRNA turnover, particularly of short lived species, may be
mediated by factors that recognize sites on the mRNAs. A number of binding
factors for short lived mRNAs have been described and some correlations have
been made between stability and the presence or absence of factors binding such
sites. One of the most thoroughly studied
cis
-acting mRNA motifs is the `AU-rich element' (ARE), which is based on the pentanucleotide AUUUA (
21
,
22
). Several different AU binding factors (AUBF), present under different
conditions and differing in their fine specificity of binding, have been
identified (
22
-
25
). The ARE of GM-CSF mRNA destabilizes it in intact cells (
4
). A coding region site on c-Myc mRNA is responsible for destabilizing it
in vivo
(
26
,
27
) and
in vitro
(
28
) and a specific binding factor for this site has recently been purified and
characterized (
28
).
The RNasin-resistant RNase that we previously described (
19
) attacks IL2 mRNA at a small number of sites in an
in vitro
assay. Reasoning that this activity might at least reflect the natural
sensitivity of IL2 mRNA to degradation, we examined crude extracts of T cells
for any factors that could interact with IL2 mRNA or alter its sensitivity to
RNase. We have discovered a factor that binds to a segment of the 3'-NC region of IL2 mRNA, but it differs from known AU binding factors
that also recognize sites in this region. Preparations of the IL2 mRNA binding
factor block its degradation by the RNasin-resistant RNase, while not significantly affecting the stability of [beta]-globin mRNA. This report describes the binding and protective
properties of the factor(s) specific for IL2 mRNA.
Cell lines were cultured in RHFM (RPMI 1640 supplemented with 5 * 10
-5
M 2-mercaptoethanol, 20 mM sodium bicarbonate, 0.34 mM pyruvate, 0.02 M HEPES
buffer, 9% fetal bovine serum). Jurkat is a human T leukemia cell line that can
be induced to synthesize IL2 (
29
). Cytosolic extracts of cells (S130) were prepared as described previously (
19
), aliquoted and stored at -70oC. Briefly, cells were broken in hypotonic medium using a Teflon in
glass homogenizer and the S130 fraction was obtained by centrifugation of the
cell lysate in a TLA 100.3 rotor (Beckman TL100 Ultracentrifuge) at 53 000
r.p.m. for 2.5 h.
Full-length IL2 mRNA was generated from the plasmid pHIL2.CA2, a modified
version of plasmid pHIL2.GA described earlier (
19
). This plasmid (see Fig.
1
) comprises full-length human IL2 cDNA with an upstream T7 promoter and a 62 nt downstream
poly(A) region, within vector pGEM-1. A primer containing a
Pvu
II site, a T7 minimal promoter and 18 nt of the human IL2 5'-NC region was used for the 5'-end of a PCR reaction. The 3' primer contained an
Eco
RI site, followed by a complement to the 3'-end of IL2 mRNA. The PCR product generated with these primers from
pHIL2.GA treated with
Pvu
II and
Eco
RI was cloned into pHIL2.GA from which the segment between the
Pvu
II and
Eco
RI sites had been removed, thereby inserting the complete IL2 sequence between a
T7 promoter and a 62 nt poly(A) segment, as shown in Figure
1
. Plasmid pHIL2.CA2 was digested with
Bam
HI before transcription to generate IL2 mRNA carrying a poly(A) tail or by
Eco
RI,
Sty
I or
Stu
I to generate shorter RNAs lacking parts of the 3'-NC region. RNA containing only the 3'-NC region was synthesized from a cloned PCR product
generated with a 5' primer comprising a
Pvu
II site, the T7 promoter sequence and 18 nt of IL2 sequence beginning at
position 507 (the start of the 3'-NC sequence). The 3' primer was that used for producing pHIL2.CA2 and the
product was also cloned into
Pvu
II/
Eco
RI cut pHIL2.GA to generate pHIL2.3N1. This plasmid is identical to pHIL2.CA2
except that it lacks nt 1-506 of the IL2 sequence (see Fig.
1
). RNA containing only the 5'-most 161 nt of the 3'-NC was transcribed from pHIL2.3N1 digested with
Sty
I.
S130 (1.0 ml at 10 mg/ml protein) was loaded onto a Sephacryl S-200 (Pharmacia) column (54 * 1.0 cm) for separation of the IL2-selective RNase from the inhibitor factor described in this
paper. This procedure was done at 4oC. The column was equilibrated with a buffer containing 100 mM NaCl and 25
mM HEPES, pH 7.2. Blue dextran and thymidine were used to determine the void
volume and the bed volume respectively and bovine serum albumin (BSA) to establish the position of the 68 000 mol. wt protein. Fractions were assayed for both RNase and the RNase
inhibitor.
Ten nanograms (5 * 10
5
c.p.m.) of
32
P-labeled mRNA were heated at 85oC for 5 min and incubated with binding factors at 30oC for 10 min in 10 [mu]l reactions containing 100 mM KAc, 0.50 mM Mg(Ac)
2
, 2 mM DTT, 10 mM Tris-HCl, pH 7.6, 5% glycerol and 1 mg/ml tRNA. RNase T1 was added to a final
concentration of 1 U/[mu]l and the sample was incubated for 10 min at 30oC. Heparin was added (final concentration 5 mg/ml) and incubation was
continued for 10 min (
28
). The reaction mixture was then loaded onto a 6% native polyacrylamide gel.
For UV cross-linking analysis the same binding reactions were performed and then the
reaction mixtures were exposed to UV light (Stratagene UV Stratalinker Model 1800, run for 5 min at 3 cm) before being loaded onto a 12% polyacrylamide-SDS gel.
The T leukemia cell line Jurkat contains an endoribonuclease that is RNasin
resistant and that degrades IL2 mRNA 5-10 times faster
in vitro
than [beta]-globin mRNA (
19
). It cleaves IL2 mRNA at a small number of sites, including two in the coding
region and one in the 3'-NC segment (Fig.
1
). This RNase was reproducibly demonstrable in extracts of Jurkat cells when
these were tested at moderate protein concentrations. However, assays at
increasing concentrations of Jurkat S130 showed an unexpected effect; namely
IL2 mRNA was selectively stabilized by high concentrations of S130. Thus at
concentrations of S130 protein of ~1 mg/ml IL2 mRNA was almost completely stable, whereas [beta]-globin mRNA was degraded at a rapid rate. In other words, the
stability of IL2 mRNA relative to [beta]-globin mRNA was inverted at high S130 concentrations compared with
the results obtained at 20- to 100-fold lower protein levels. This suggested the presence of an IL2-selective protective factor in the crude extracts (Fig.
2
).
The protection of IL2 mRNA by Jurkat S130 suggested that it contained a
component, apart from the RNasin-resistant mRNase, that interacted with IL2 mRNA. Gel shift analysis was
therefore performed in an attempt to detect a binding factor for IL2 mRNA.
Labeled IL2 mRNA was incubated with S130 at 1 mg/ml (a concentration showing
protection of IL2 mRNA) and then subjected to degradation by T1 RNase. After
competing out non-specific complexes with heparin the products were analyzed on a native
acrylamide gel. A single band was seen with S130 (Fig.
3
).
Sequence-specific competition for the binding factor, as detected by gel shift
analysis, was observed (Fig.
4
). The addition of unlabeled IL2 mRNA diminished the signal intensity of the
shifted band due to saturation of the binding factor. No competition was
observed with [beta]-globin, GAPDH or the c-Myc CRD RNAs under the same conditions. It should be noted
that the binding reactions routinely contained 1 mg/ml tRNA, compared with 1 [mu]g/ml labeled target mRNA, further supporting the notion of specificity.
The complex between IL2 mRNA and Sephacryl S-200 fraction 15 was not affected by the non-ionic detergent NP40, whether the detergent was present throughout
the binding reaction or was added after formation of the complex (Fig.
8
). The ionic detergent Sarkosyl, however, prevented formation of the complex
and/or destroyed the complex once formed. These results are consistent with the
idea that the binding factor is a protein that forms a non-covalent complex with IL2 mRNA.
The binding factor obtained by Sephacryl S-200 chromatography (fraction 15) blocked the ability of the RNasin-resistant RNase (fraction 17) to degrade IL2 mRNA under conditions
leading to formation of the complex seen in gel retardation experiments (Fig.
9
). Protection was not observed for [beta]-globin mRNA that was present in the same reactions. The protein
concentrations stabilizing IL2 mRNA were of the order of 1 mg/ml, similar to
those used in binding experiments (cf. Fig.
6
). The protective effect of the binding factor was saturable by an excess of IL2
mRNA, but not by [beta]-globin or c-Myc CRD RNA (data not shown). It was destroyed by treatment
with proteinase K (data not shown), suggesting that it contains an essential
protein component.
Strong circumstantial evidence indicates that the factor that binds IL2 mRNA in
the gel shift assay also protects it against the RNasin-resistant mRNase described in our previous study (
19
). Binding and protection are both specific for IL2 mRNA relative to the other
RNAs tested, including a sequence that binds AU binding factors (AUBF) (
22
,
31
,
32
). The concentrations of S130 or Sephacryl S-200 fraction 15 protein needed to form the binding complex and to protect
against mRNase are similar. Both have properties suggesting they are proteins;
the protective factor is sensitive to proteinase K and the binding assay is
sensitive to ionic detergents (Fig.
8
). The size of the UV cross-linked complex in the binding assay (~60 kDa; Fig.
7
) is similar to the size of the binding and protective factors on Sephacryl S-200 (near the elution position of BSA).
The binding factor recognizes a specific segment of the IL2 3'-NC segment, between positions 507 and 667, within which there are two long oligonucleotides lacking G residues. One of
these, comprising nt 543-580, contains several AU-rich elements. Under appropriate conditions this fragment binds one of the AUBF (
22
,
33
), of mol. wt ~36 000. A second AUBF, AU-B, is found in stimulated normal human T lymphocytes and exhibits a
high affinity for ARE in lymphokine mRNAs, but not for those in c-Myc mRNA (
33
). It is of lower molecular weight. A third component, AU-C, resembles AU-B (
24
), but is of mol. wt ~43 000. Two immunologically cross-reactive proteins, of mol. wt 37 000 and 40 000, are present in the
AU binding factor AUF1 (
25
). AUF1 binds the 3'-NC regions of c-Myc and GM-CSF RNAs and may be involved in their ARE-dependent degradation. The 37 000 mol. wt protein
has been cloned (
25
). These factors have varying affinities for naturally occurring AREs, but have
in common the ability to bind synthetic RNAs containing three or more repeats
of the type AUUUA, including the sequence we have used in this study, which
contains four repeats (
22
). This probe sequence did not compete for binding of our factor to IL2 mRNA
(Figs
5
and
7
) nor did it form a complex with any factor under the conditions of our
experiments (possible reasons for this are discussed below).
Our factor probably binds to one of the two long G-free segments within the region 507-667 of the IL2 RNA probe. These segments have lengths of 38 and 35
nt respectively. The UV cross-linked complex seen in Figure
7
presumably comprises a protein binding factor and one or both of these T1-resistant oligonucleotides. It is noteworthy that binding to IL2 mRNA
lacking a poly(A) tail, or indeed a segment carrying only the fragment 507-667, formed a complex with the same mobility as that seen with full-length poly(A)
+
mRNA (Fig.
6
). The binding factor thus does not resemble poly(A) or poly(U) binding factors
(
34
). However, we have noted an effect of poly(A) on binding of our factor that
cannot yet be explained, namely the presence of poly(A), either as part of the
IL2 mRNA or unattached, reduces affinity of the factor for the binding site in
the 3'-NC region of the IL2 sequence. The saturation level of binding is
the same as in the absence of poly(A). One possibility is that the factor is
associated with a poly(A) binding factor, which leads to competition between the IL2 3'-NC region site and poly(A). Another possibility is that poly(A)
interacts with the factor binding site in the 3'-NC region and thereby blocks factor binding (the binding site is
high in A+U). An explanation of the poly(A) effect is currently being sought.
It might seem surprising that we saw no evidence for the ARE binding factors
that are known to exist in Jurkat cell extracts and that have an affinity for
the 3'-NC region of IL2 mRNA. The reason probably lies in the conditions
used. Compared with studies on AUBF mentioned above (
24
,
33
) we used a higher temperature (30oC rather than room temperature), a higher salt concentration (100 rather
than 40 mM K
+
), a lower Mg
2+
(0.5 rather than 3 mM) and a high concentration of tRNA (1 mg/ml compared with
none in the earlier work) to compete for non-specific binding. In comparing our conditions with those of Malter, in
which an AU binding protein was first described (
22
), our binding experiments are higher in salt, lower in Mg
2+
, 5-fold higher in tRNA and include 5 mg/ml heparin to reduce low affinity complexes. Under our conditions,
which we designed to mimic closely those of the functional assay (i.e.
inhibition of the RNasin-resistant mRNase), we would have reduced lower affinity binding. This
probably accounts for the lack of AUBF activity bound to either the IL2 3'-NC probes or to the AU-rich probe.
There are several other examples of stabilization of mRNAs by binding factors.
Stabilization of an mRNA in an
in vitro
polysome-based assay has been shown for c-Myc, where a factor binding to a coding region determinant
stabilized the mRNA against degradation (
27
,
28
). Using the same system it was found that GM-CSF mRNA is stabilized by an AUBF, presumably interacting with AU-rich 3'-NC sites (
31
). An endoribonuclease has been purified from
Xenopus
oocytes which cleaves Xlhbox2B mRNA in the 3'-NC region. This attack is blocked by a factor that appears to bind
at or near the same sequence recognized by the nuclease, a site that does not
contain the classical ARE motif (
20
). An extract that selectively degrades TGF[beta]-1 mRNA has been described and also the action of a protective factor
that is induced by stimulation of the cells by PMA (
35
,
36
). The ability of the factor to stabilize
in vitro
parallels
in vivo
stabilization. Sequence-specific binding factors that may influence mRNA stability include a 65
kDa protein that binds at two sites in the 3'-NC region of [beta]-IFN mRNA (
5
), an AU-rich region that is involved in destabilizing the mRNA. Another binding
factor for a 3'-NC region is found in the tyrosine hydroxylase system (
37
), where a 75 kDa protein can be cross-linked. Vitellogenin mRNA is stabilized by estrogen, which induces a
protein with binding sites in the 3'-NC region of this mRNA (
38
). A 75 kDa binding factor recognizes a site in the 3'-NC region of the R2 subunit of ribonucleotide reductase. This
factor is induced by TGF[beta] and is not cross-reactive with other short lived mRNAs or poly(A) (
39
). A 57 kDa PMA-inhibited binding factor apparently destabilizes the ribonucleotide
reductase R1 subunit mRNA (
40
).
There is evidence that for some short lived mRNAs instability sequences
identified by functional studies in intact cells are targets for the binding of
factors
in vitro
. This is so for c-Fos (
41
) and c-Myc (
28
,
42
). In the study of IL2 mRNA stabilization the linkage between binding and
protection
in vitro
and the mechanism of mRNA turnover
in vivo
is not yet established. This work provides a focus for examining such a
linkage.
This work was supported by a grant from the National Cancer Institute of Canada.
We are grateful to Drs Jeffrey Ross and James Malter of the University of
Wisconsin and Dr Kathryn Calame of Columbia University for providing plasmids.
REFERENCES
Return