ABSTRACT
TIA-1 and TIAR are RNA binding proteins of the RNA recognition motif (RRM)/ribonucleoprotein (RNP) family that have been implicated as effectors of apoptotic cell death. We report
the structures of murine TIA-1 and TIAR (mTIA-1 and mTIAR) deduced from cDNA cloning, the mRNA and protein tissue
distribution of mTIA-1 and mTIAR, and the exon-intron structures of the mTIA-1 and mTIAR genes. Both mTIA-1 and mTIAR are comprised of three
~
100 amino acid N-terminal RRM domains and a
~
90 amino acid C-terminal auxiliary domain. This subfamily of RRM proteins is
evolutionarily well conserved; mTIA-1 and mTIAR are 80% similar to each other, and 96 and 99% similar to hTIA-1 and hTIAR, respectively. The overall exon-intron structures of the mTIA-1 and mTIAR genes are also similar to each other, as
well as to the human TIA-1 gene structure. While Northern blot analysis reveals that mTIA-1 and mTIAR mRNAs have a broad tissue distribution, mTIA-1 and mTIAR proteins are predominantly expressed in brain, testis and spleen. At
least two isoforms of both mTIA-1 and mTIAR are generated by alternative splicing. Murine TIA-1 isoforms including or lacking the exon 5 encoded sequences are expressed at a ratio of
~
1:1, whereas mTIAR isoforms including or lacking the 5
'
-end of exon 3 sequences are expressed in a
~
1:6 ratio. Molecular characterization of murine TIA-1 and TIAR RNA binding proteins provides the basis for a genetic analysis of the functional
roles of these proteins during mammalian development.
RNA binding proteins of the RNA recognition motif (RRM)/ ribonucleoprotein (RNP)
family function in diverse aspects of RNA metabolism including the biogenesis
and translation of mRNA (
1
-
4
). RRM proteins contain one to four ~100 amino acid RRM domains that have conserved hexamer and octamer peptide
sequence motifs, referred to as the RNP 2 and RNP 1 motifs, respectively, as
well as various auxiliary domains (
1
,
2
). Several RRM proteins are known to regulate development. For instance, the
Drosophila
RRM proteins Sxl and tra2 control sex determination by sex-specific alternative splicing (
5
), the
Drosophila
squid
gene product hrp40 is necessary for dorsoventral axis formation during
oogenesis (
6
,
7
), the
Drosophila
orb
gene product is required for the formation of the egg chamber and
anteroposterior as well as dorsoventral patterning during oogenesis (
8
,
9
), the
Drosophila elav
gene product is implicated in the development of the embryonic nervous system (
10
), and a family of RRM proteins is likely to regulate human spermatogenesis (
11
,
12
). Previously we identified two highly related human RRM proteins, TIA-1 and TIAR, that contain three N-terminal RRM domains and a C-terminal auxiliary domain (
13
,
14
). TIA-1 and TIAR share 80% overall identity, and both proteins bind mRNA (
15
). The second RRM domains of TIA-1 and TIAR bind RNAs containing short stretches of uracil (
15
).
Drosophila
and
C.elegans
both contain a TIA-1/TIAR-like protein that is ~47% identical to either human TIA-1 or TIAR, suggesting that these proteins have been
evolutionarily conserved (
16
,
17
). Although the function of TIA-1 and TIAR is unknown, several findings implicate these RNA binding
proteins as effectors of apoptosis: purified TIA-1 and TIAR induce apoptosis in appropriate permeabilized target cells (
13
,
14
), TIAR is translocated from the nucleus to the cytoplasm early during Fas-mediated apoptosis (
18
), and TIA-1 is a specific substrate for the Fas-activated protein serine/threonine (FAST) kinase (
19
). To develop a system to study RRM-type RNA binding proteins during mammalian development and apoptosis, we
have characterized the cDNA and gene structures of the murine homologs of the
human RNA binding proteins TIA-1 and TIAR, as well as determined their tissue distribution.
Mouse TIA-1 and TIAR cDNAs were isolated from a [lambda] YES cDNA library derived from mRNA isolated from activated T-cells (
20
) by cross-hybridization with human TIA-1 (
13
) and TIAR (
14
) cDNA probes using standard techniques (
21
). DNA manipulations and DNA sequence determination using the chain termination
method were done according to standard procedures. The complete cDNA sequences
of mTIA-1 and mTIAR appear in the EMBL/GenBank/DDBJ nucleotide sequence data bases
(accession numbers U55862 and U55861, respectively).
A mouse 129 SVJ genomic DNA library constructed in the [lambda] FIX phage vector (Clontech) was screened with mTIA-1 and mTIAR cDNA probes, and positive phage clones were
characterized by restriction mapping and Southern hybridization of restriction
fragments with mTIA-1 and mTIAR cDNA probes essentially as previously described (
22
,
23
). The exact location of the exon-intron junctions was established by DNA sequencing of appropriate regions of
the murine TIA-1 and TIAR genes.
A tissue Northern blot containing about 2 [mu]g poly(A)
+
RNA per lane (Clontech) was probed with
32
P-labeled mTIA-1 and mTIAR cDNA probes using the hybridization conditions
recommended by the supplier. Murine TIA-1 specific mRNAs were detected using a mTIA-1 cDNA probe encoding amino acids 62-357 [including the exon 5 encoded TIA-1a peptide sequence (aa 93-103)] and 66 bp of 3' non-translated region, and mTIAR
specific mRNAs were detected using a cDNA probe encoding amino acids 112-357 [lacking the exon 3 encoded TIARa peptide sequence (aa 44-60)] and 37 bp of 3' non-translated sequences. The absence of cross-hybridization between mTIA-1 and mTIAR under the conditions used for
Northern blot hybridizations was confirmed by dot blot analysis (data not
shown).
The mTIA-1 expression plasmid pSR[alpha].TIA-1a (containing the alternatively spliced exon 5 sequences)
was generated by inserting TIA-1 cDNA encoding the TIA-1a isoform into the
Eco
RI site of a modified version of the eukaryotic expression vector pcDL-SR[alpha]296 (
24
), termed pSP65-SR[alpha]-PJ.Hygro, that contains the hygromycin-[beta]-phosphotransferase gene. The plasmid pMT.2.TIARb (lacking the alternatively
spliced 5' region of exon 3) was constructed by inserting the TIAR cDNA encoding
the TIARb isoform into a modified version of the eukaryotic expression vector
pMT.2 (
25
). cDNAs encoding the TIA-1b (lacking the alternatively spliced exon 5 sequence) and TIARa
(containing the alternatively spliced 5' region of exon 3) isoforms were generated by PCR using appropriate
primers (mTIA-1 sense CGGGATCCATGGAGGACGAGATGCC, mTIA-1 anti-sense TACTGGCCAATCTGTTGTGC; mTIAR sense CGGGATCCATGGAAGACGACGGACAGC, mTIAR anti-sense GCAGGTGGTTTACGTGTGG) and the T cell cDNA library as template. DNA restriction fragments derived from the PCR-generated DNA were then used to replace the corresponding
fragments in the pSR[alpha].mTIA-1a and pMT.2.mTIARb plasmids to generate
pSR[alpha].mTIA-1b and pMT.2.mTIARa which encode the mTIA-1b and mTIARa isoforms, respectively.
Tissues were isolated from Balb/c mice, washed with phosphate buffered saline, minced and homogenized in a dounce homogenizer in buffer A (50 mM Tris-HCl pH 8.0, 1 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, 10 [mu]g/ml aprotinin and 10 [mu]g/ml leupeptin). The homogenates were then adjusted to 150 mM
NaCl, 1% NP-40 and incubated on ice for 20 min. Following incubation, the lysates were
subjected to sonication, and then insoluble material was removed by
centrifugation (20 000
g
). For immunoprecipitations, cell lysates containing 2-4 mg protein (
26
) were precleared with Sepharose 4B beads, and then incubated with 5 [mu]g mAb (the anti-TIA-1/TIAR mAbs ML29 or 6E3, or control isotype-matched mAb) and 25 [mu]l protein G-Sepharose slurry (Pharmacia Biotech Inc.) for 3-5 h. The anti-TIA-1 and TIAR mAbs, 2G9 (
13
), 6E3 (
18
), 3E6 (
18
) and ML29 (
18
) were all originally obtained against human proteins. Immunoprecipitates were
washed with buffer B (150 mM NaCl, 50 mM Tris-HCl pH 8.0 and 1% NP-40), and washed immunoprecipitated proteins were resolved by SDS-PAGE (10% gels) using reducing conditions. Proteins were
then transferred to Immobilon P membrane (Millipore) and probed with anti-TIA-1 2G9 mAb (5 [mu]g/ml) or anti-TIAR 3E6 mAb (1 [mu]g/ml). Immunoblots were developed with protein A/G-horseradish peroxidase (Pierce) and the
chemiluminescence reagent, luminol, essentially as described by the supplier
(DuPont/NEN). To control for the specificity of the mAbs and to generate
standards for the TIA-1 and TIAR isoforms, we produced recombinant mTIA-1a, mTIA-1b, mTIARa and mTIARb protein using the COS-7 transient transfection system. To this end, COS-7 (ATCC CRL 1651) cells were transfected with the
expression vectors pSR[alpha].mTIA-1a, pSR[alpha].mTIA-1b, pMT.2.mTIARa and pMT.2.mTIARb using the DEAE-dextran method as described (
21
). Following transfection, cells were cultured for 48-56 h, washed with phosphate-buffered saline, and then cells were lysed in lysis buffer (1% NP-40, 150 mM NaCl, 50 mM Tris-HCl pH 7.5 containing 1 mM phenylmethylsulfonyl fluoride, 10 [mu]g/ml aprotinin, 10 [mu]g/ml leupeptin). Cell lysates from pSR[alpha].mTIA-1a and pSR[alpha].mTIA-1b transfected cells were
pooled, as were cell lysates from pMT.2.mTIARa and pMT.2.mTIARb transfected
cells, and then used as controls for immunoprecipitation and immunoblotting
experiments. The relative amounts of mTIA-1 and mTIAR isoforms were determined by densitometric scanning of
autoradiographs using a Ultrascan XL laser densitometer (Pharmacia).
To determine whether mice contain TIA-1 and TIAR RNA binding proteins, we screened a murine T-cell cDNA library using low stringency hybridization conditions with
cDNA probes that contain the entire human TIA-1 and TIAR coding regions. Sequence analysis of the isolated cDNAs
revealed the existence of murine homologs of human TIA-1 and TIAR. The deduced primary structure of the 386 amino acid mouse TIA-1 (mTIA-1) is 96% identical to the 386 amino acid human TIA-1 (hTIA-1) isoform, and the 392 amino acid mouse TIAR
(mTIAR) sequence is 99% identical to the 392 amino acid human TIAR (hTIAR)
isoform (Fig.
1
A). Like their human counterparts, mTIA-1 and mTIAR consist of three RRM domains (RRM domain 1 to RRM domain 3)
and a 90 or 88 amino acid C-terminal auxiliary domain, respectively (Fig.
1
A and C). Furthermore, the mTIA-1 and mTIAR sequences share 80% overall identity, with the highest degree
of similarity residing in the RRM domain 3 (91% identity) and the lowest degree
of similarity (50%) in the auxiliary domain. mTIA-1 and mTIAR are equally similar to the
Drosophila TIA
-1-like protein rox8 (47% identity;
16
) and a
C.elegans
TIA-like protein CM13B6 (43% identity;
17
). The auxiliary domains of mTIA-1 and mTIAR do not have a significant degree of similarity with other
known proteins.
The expression of mTIA-1 and mTIAR mRNA was examined by Northern blot analysis using poly(A)
+
RNA isolated from various mouse tissues and cDNA probes specific for either
mTIA-1 or mTIAR. The Northern blot was sequentially probed for mTIA-1 expression and then for mTIAR expression. Multiple mTIA-1 mRNAs ranging in length from ~3.5 to 9.5 kb are detected in all of the tissue mRNAs
examined, with the exception of liver mRNA which has significantly lower levels
of mTIA-1 mRNA (Fig.
2
A). The predominant mTIA-1 mRNAs are ~4, 4.4, 7 and 9.5 kb in length in most of the tissues analyzed, but
the relative ratios of the various mTIA-1 mRNAs varies depending on the tissue mRNA analyzed. For instance, brain
has comparable levels of the 4.0 and 9.5 kb mTIA-1 mRNA species, whereas testis has significantly more of the 4.0 kb mRNA
than the 9.5 kb mRNA. The molecular basis for the various mTIA-1 mRNA forms is not known, but may include accumulation of mRNAs with
retained introns and/or alternative length 3' non-translated regions. Northern blot analysis of mTIA-1 mRNA isolated from cell lines that express mTIA-1 reveals only the 3.5 and 4.0 kb mRNA species,
suggesting that these shorter mTIA-1 mRNAs are the mature isoforms encoding mTIA-1 (data not shown). mTIAR mRNA, like mTIA-1 mRNA, has a broad tissue distribution with multiple length
mRNAs (Fig.
2
B). In all of the tissue mRNAs examined (including liver), there are 1.6 and 4.5
kb mTIAR mRNA species, and in several tissues (i.e. spleen, lung and testis)
there is an additional 2.3 kb form. In brain, the predominant mRNA isoform is
4.5 kb, whereas in testis it is 1.6 kb. Because brain and testis express mTIAR
protein (see below) it is likely that both the 4.4 and 1.6 kb mTIAR mRNAs
encode mTIAR. Control hybridization of the Northern blot with a [beta]-actin cDNA probe indicates that approximately equal amounts of mRNA
are present from each tissue (Fig.
2
C). Thus, both mTIA-1 and mTIAR mRNAs are expressed in diverse tissues.
To establish the exon-intron organization of the mTIA-1 gene, a [lambda]gt11 murine genomic DNA library was screened using mTIA-1 cDNA probes containing the entire coding regions.
Positive clones were characterized by restriction mapping, Southern blotting and DNA sequencing
to establish the exon-intron junction of individual exons. Characterization of mTIA-1 genomic clone [lambda]mTIA-1-5.2 and the overlapping clones [lambda]mTIA-1-f and [lambda]mTIA-1-o
revealed that the mTIA-1 gene contains 13 exons that span >20 kb (Fig.
4
A). The exon-intron organization of the mTIA-1 gene, as well as the exon-intron junction sequences, are shown in Figures
4
A and
5
A, respectively. The mTIA-1 RRM domains 1, 2 and 3 are encoded by exons 1-4, 5-8 and 9-11, respectively; and the C-terminal auxiliary domain is encoded by exons
12 and 13. TIA-1 exon 5 encodes the alternatively used 11 amino acid TIA-1a peptide that distinguishes the mTIA-1a isoform from the mTIA-1b isoform. This exon-intron organization is conserved between the murine and human TIA-1 genes (
23
). The 11 amino acid TIA-1a peptide is directly N-terminal to the RNP2 motif consensus sequence of RNP domain 1. This
location is similar to an alternatively used 19 amino acid peptide of hnRNP D0
(
28
). hnRNP D0 contains two RRM domains, and the alternatively spliced 19 amino acid sequence affects RNA sequence binding specificity, suggesting that the alternative usage of the TIA-1a peptide may also modify RNA binding specificity (
28
).
The exon-intron organization of the mTIAR gene was established in a similar manner as the mTIA-1 gene. Characterization of the overlapping mTIAR genomic clones [lambda]mTIAR-32, [lambda]mTIAR-5, [lambda]mTIAR-17 and [lambda]mTIAR-13
demonstrated that the mTIAR gene is comprised of 12 exons that span ~20 kb (Fig.
4
B). The exon-intron organization of the mTIAR gene, as well as the exon-intron junction sequences, are shown in Figures
4
B and 5B, respectively. The mTIAR RRM domains 1, 2 and 3 are encoded by exons 1-4, 5-7 and 8-10, respectively; and the C-terminal auxiliary domain is encoded by exons 11 and
12. The 5'-end of TIAR exon 3 encodes the alternatively used 17 amino acid
TIARa peptide sequence that distinguishes the mTIARa and mTIARb isoforms from each other. The 3'-end of exon 3 encodes amino acids 61-93, which are common to both mTIAR isoforms. Thus, the mTIARa
and mTIARb isoforms are generated by alternative 3' splice acceptor site selection of intron 2. The TIARa peptide is located
in a loop region in RNP domain 1 and may be important for RNA binding
specificity (
4
,
29
).
The overall exon-intron organization of the mTIA-1 and mTIAR genes, as well as the hTIA-1 gene (
23
), are well conserved suggesting that the TIA-1 and TIAR genes were created by a gene duplication event that occurred
prior to the divergence of mice and humans. However, the alternative usage of
the TIA-1a peptide sequence (encoded by mTIA-1 exon 5) is specific for the TIA-1 gene, and the alternate usage of the TIARa peptide sequence
(encoded by the 5'-end of mTIAR exon 3) is specific for the TIAR gene. It is unlikely
that the mTIAR gene has a mTIA-1 exon-5-like exon, as the 133 bp mTIAR intron located between mTIAR
exons 4 and 5 does not contain sequences that could encode an ~11 amino acid peptide similar to the TIA-1a peptide. Sequences 5' of mTIA-1 exon 3 also cannot be analogous to the TIARa peptide
as alternative 3' splice selection more 5' of mTIA-1 exon 3 would result in the inclusion of a termination
codon just 5' of mTIA-1 exon 3. Thus, it is likely that alternative usage of the TIA-1a and/or TIARa peptide is a feature that was acquired (or
lost) following the gene duplication event. The functional significance of
different TIA-1 and TIAR isoforms generated by alternative splicing remains to be
demonstrated, but is likely to regulate the RNA binding specificities of these proteins.
The role of mTIA-1 and mTIAR in RNA metabolism remains to be established. In most tissues
the expression of mTIA-1 and mTIAR is coincident (i.e. both proteins are expressed in brain,
spleen and testis, and are absent in heart, skeletal muscle and kidney). This
coincident expression pattern of mTIA-1 and mTIAR may suggest that these highly related proteins serve interdependent functions. However, in liver only mTIAR expression is observed, indicating that mTIAR can function independently of mTIA-1. There appears to be tissue-specific translational and/or transcriptional control of mTIA-1 and mTIAR expression. In skeletal muscle and kidney
both mTIA-1 mRNA and mTIAR mRNA are expressed, but neither protein is detected. Furthermore, the absence of mTIA-1 mRNA in liver suggests that tissue-specific expression of these proteins can also be transcriptionally
regulated. Given the role of RRM RNA binding proteins in
Drosophila
development (
5
-
10
), ongoing experiments with targeted disruptions of mTIA-1 and mTIAR should provide further insights into the functions of these
mammalian RRM proteins.
We thank Drs Nancy Kedersha, Haruo Saito and Walter Blattler for critical review
of the manuscript, Drs Stuart F. Schlossman and Kasper H. Winterhalter for
encouragement and support, Dr S. Elledge for providing the murine cDNA library,
and Dr Mark Boothby for help with isolating mTIA-1 cDNA. This work was supported by a Dana-Farber Cancer Institute/Apoptosis Technology Inc. drug discovery grant, National Institutes of Health Grants
AI33600 and CA67929, an ETH training fellowship to A.R.P.B, a Medical Research
Council of Canada Fellowship to Q.G.M, and a Pew Scholar in the Biomedical
Sciences Award to M.S.
REFERENCES
Return