(Received for publication, March 10, 1997, and in revised form, May 19, 1997)
From the Department of Biochemistry, University of Alberta, Edmonton, Alberta T6G 2H7, Canada
We have identified two distinct cDNAs
encoding the -subunit of pig heart succinyl-CoA synthetase. The
derived amino acid sequence of one of these, PH
57, is highly similar
to the
-subunit of the rat liver precursor enzyme. The second
cDNA, PH
108, was identical throughout its sequence with PH
57
except for a stretch of 108 nucleotides which replaced a 57 nucleotide
sequence in PH
57. Coexpression of either
-subunit cDNA with a
common pig heart
-subunit cDNA produced isozymes with
GTP-specific enzyme activity. The enzyme produced by the combination of
PH
57 and the
-subunit cDNA resembled the "native" enzyme
purified from pig heart tissue. In contrast, the expressed enzyme from
the combination with PH
108 was clearly distinguishable from the
native enzyme by, for example, hydroxyapatite chromatography. Moreover,
it was now apparent that this isoform had been observed in previous
preparations of the native enzyme, but always in very low amounts and,
thus, disregarded. We have shown further that the two mRNA
transcripts arise from a single gene and are generated by mutually
exclusive splicing. The production of the PH
108 message involves the
use of a non-canonical splice site pair, AT-AA. Finally, we provide evidence for tissue specific regulation in the splicing of the PH
108
message.
Succinyl-CoA synthetase (SCS)1 catalyzes the substrate-level phosphorylation step of the citric acid cycle according to the following reaction (Reaction 1).
![]() |
![]() |
Apart from the primary role of succinyl-CoA within the citric acid
cycle, high levels of this intermediate must be maintained for
continued utilization of ketone bodies (5). In mammals, ketone bodies
such as acetoacetate and -hydroxybutyrate are produced by the liver,
and tissues such as heart muscle routinely derive much of their
metabolic energy from oxidation of these compounds (6). Although within
the citric acid cycle SCS catalyzes the conversion of succinyl-CoA to
succinate, ketone body activation requires the reverse, the conversion
of succinate to succinyl-CoA. Ottaway and co-workers (5, 6) have
suggested that such opposing functional demands could best be
accommodated by distinct forms of the enzyme. It is worth noting that
succinyl-CoA also serves as an anabolic precursor in the synthesis of
porphyrins and hemes (7).
Historically, there is evidence that diverse isoforms of the mammalian enzyme exist. Baccanari and Cha (8) showed that the GTP-specific enzyme from pig heart tissue separated into several species differing in charge. These multiple forms were found to be interconvertible indicating a common protein source (9). In a different study, two GTP-specific forms of pig heart SCS were separated upon purification using hydroxyapatite chromatography (10). A similar observation has been reported of the enzyme isolated from pigeon pectoral muscle (11). Two distinct GTP-specific forms of succinyl-CoA synthetase were detected in mouse liver, one of which was induced as a result of increased porphyrin synthesis (12). More recently, Weitzman et al. (13) demonstrated the existence of isoforms differing in their nucleotide specificity in mammalian tissues. Separable GTP- and ATP-specific forms were present in ratios that varied depending on the tissue source. Marked increases in the GTP-specific form in brain tissue occur following treatment of rats with streptozotocin, a drug used to induce diabetes (14). Diabetes invariably leads to elevated levels of ketone bodies. These and other observations have led to the suggestion that a GTP-specific form is involved exclusively in ketone body metabolism (14, 15).
The focus of the studies described herein was to gain a better
understanding of one source of the heterogeneity present with the
pig heart enzymes. Here we report on the identification and characterization of two isoforms of the -subunit of the pig heart enzyme. The two forms are generated by alternative splicing of a single
transcript. Furthermore, we provide evidence for tissue-specific regulation of the splicing event.
The following E. coli strains were used: JM109 (16) for construction and propagation of M13 and plasmid derivatives, BL21(DE3) (17) for expression studies, and Y1089 (18) and LE392 (19) for screening of cDNA and genomic libraries, respectively.
Enzymes and ReagentsThe polynucleotide kinase, ligase, Klenow DNA polymerase, and restriction enzymes were obtained from Life Technologies (as Gibco BRL products) and New England Biolabs Inc. The Taq DNA polymerase was purchased from Cetus Corp. Reagents used for sequencing nucleic acids were ordered from U. S. Biochemical Corp. The dNTPs were purchased as 100 mM solutions from Pharmacia Biotech Inc. Unless otherwise specified, all other chemicals were supplied by Sigma and BDH Chemicals Ltd.
cDNA and Genomic LibrariesThe generation of the pig
heart cDNA library used in this study has been reported elsewhere
(20). The unamplified library was screened with a biotin-labeled probe
prepared from a cDNA encoding the precursor -subunit of rat
liver SCS (21). Potential clones were identified with
streptavidin-conjugated alkaline phosphatase (22, 23). The cDNA
fragments were subcloned into M13 vectors (16), and their sequences
were determined by the dideoxynucleotide chain termination method
(24).
High molecular mass DNA was prepared from the liver of a mature pig
(25), and used to construct a genomic library in GEM11 following
standard procedures (26). The
GEM11 arms and packaging extracts were
bought from Promega Corp. A second genomic library, constructed in
EMBL-3, was purchased from CLONTECH
Laboratories. Both genomic libraries (1 × 106 plaques
each) were screened under conditions of high stringency (26) with the
cDNA clone, PH
57. This probe was radioactively labeled with
[
35S]dATP (Amersham Canada Ltd.) by random priming
(27). Southern blot analyses were carried out using standard methods
with GeneScreen membranes (DuPont NEN) and oligonucleotide probes
end-labeled with 32P.
Total RNA was
purified from fresh tissues of newborn piglets (generously provided by
Dr. G. Foxcroft, Faculty of Agriculture, Forestry, and Home Economics).
Trizol reagent (Life Technologies) was used to purify the RNA. An
aliquot of RNA (1 µg) was reverse transcribed with SuperScript II
(Life Technologies) and oligo(dT). Reverse transcribed PCR for each
isoform was carried out pairing oligonucleotide 6, below, with either
oligonucleotide 3 (specific for PH57) or oligonucleotide 5 (specific
for PH
108) as the primers. The cycling conditions used were as
follows: 20 s at 94 °C, 30 s at 57 °C, and 60 s at
68 °C (40 cycles for PH
57, 45 cycles for PH
108, and 35 cycles
for
-actin); followed by one cycle of 10 min at 72 °C. The PCR
was performed as described by Saiki et al. (28).
The procedures of Frohman et al. (29) were followed for the
5-RACE analyses using the appropriate reagents purchased from Life
Technologies. RNA (1 µg) was reverse transcribed with SuperScript II
and oligonucleotide 2, below, as the primer. This primer was used for
first strand synthesis of specifically
-subunit cDNAs carrying
either the 57 or the 108 nucleotide sequence. These cDNAs were
purified through a GlassMax spin cartridge (Life Technologies) and were
tailed with dCTP and TdT. Anchored PCR was carried out with a 5
-RACE
anchor primer and oligonucleotides specific for PH
57 (
oligonucleotide 3 below) and PH
108 (oligonucleotide 5 below). The
cycling conditions were as follows: 42 cycles of 20 s at 94 °C,
30 s at 57 °C, and 60 s at 68 °C; followed by one cycle
of 10 min at 72 °C. The amplified products were purified and cloned
directly into TA vectors (Invitrogen Corporation).
The sequences of the oligonucleotides used in all these studies are as follows.
![]() |
The plasmid, pT7-6 (30) was used to
construct vectors capable of co-producing both subunits of SCS. The
configuration of the recombinant vector, pT7-6
Ec/Ec
, carrying the genes for the E. coli enzyme is illustrated in Fig. 3A. Both cDNAs
of the pig heart
-subunits were modified to include a
ClaI restriction site at the equivalent position relative to
their bacterial counterpart. These modified sequences were cloned into
pT7-6 Ec
/Ec
in place of the E. coli gene. The resulting plasmids, pT7-6 Ec
/PH
57
and Ec
/PH
108, were used as negative controls; although
the expression of the individual E. coli
-subunit and the
pig heart
-subunit can occur, the subsequent formation of SCS
hybrids does not. A cDNA sequence encoding the pig heart
-subunit (31) was altered to include NdeI and
EspI restriction sites at the appropriate locations. The
E. coli gene coding for the
-subunit in pT7-6 Ec
/PH
57 and Ec
/PH
108 was replaced
with the modified pig heart
-subunit sequence creating the
expression plasmids, pT7-6 PH
/PH
57 and pT7-6 PH
/PH
108. The
above mentioned sequence alterations were achieved using standard
mutagenesis procedures (32) on uracil-enriched templates (33). The
mutagenic oligonucleotides were as follows.
![]() |
Protein Purification and Related Techniques
The frozen
bacterial cultures were thawed and resuspended in 100 ml of
"sonication" buffer (50 mM potassium phosphate, pH 7.4, 1 mM EDTA, 0.1 mM PMSF). These suspensions were
sonicated over crushed ice and centrifuged to pellet the cell debris.
Each supernatant was subjected to a two-step ammonium sulfate
precipitation. The protein fraction precipitating between 15% and 40%
(w/v) ammonium sulfate was collected by centrifugation and redissolved
in a minimal volume of 10 mM potassium phosphate, 1 mM EDTA, 0.1 mM PMSF at pH 7.4. These solutions
were desalted by gel filtration through Sephadex G25 equilibrated and
eluted with the same dissolving buffer. Eluted fractions that contained
high enzyme activity were pooled and loaded directly onto a 50-ml
hydroxyapatite column equilibrated with 10 mM potassium
phosphate, pH 7.4, 1 mM EDTA, 0.1 mM PMSF. The
column was washed with another 200 ml of the buffer. Two columns were
set up and run simultaneously in this manner, one for expressed
PH/PH
57 and the other for expressed PH
/PH
108. Elution (from
a common 500-ml gradient reservoir) was carried out with a salt
gradient of 10 mM to 400 mM potassium phosphate, pH 7.4. All of the above purification procedures were conducted at 4 °C. Protein extracts made from minced pig heart tissue were fractionated in a similar manner. Succinyl-CoA synthetase activity was measured by following the change of absorbance at 235 nm
using the method of Cha (34).
Approximately 105 plaques from an unamplified pig heart cDNA library were screened as described under "Materials and Methods." A total of 25 clones were recovered from the library. Of these, 12 were found to be full-length and were selected for further analysis. It was observed that for all these clones two fragments were produced upon EcoRI digestion, indicative of a single, internal EcoRI site. The clones were classified into two groups on the basis of the size of the larger EcoRI fragment (Fig. 1A). Class I accounted for nine of these clones and contained a 1.15-kb EcoRI fragment. The remaining Class II clones contained a slightly larger 1.2-kb fragment. Representatives from each group were selected for sequencing. The two classes of cDNAs were identical throughout their sequence, with the exception of one short region midway through their open reading frames (Fig. 1A). A stretch of 108 nucleotides in the Class II clones replaced a 57-nucleotide sequence in the Class I clones.
The nucleotide and deduced amino acid sequence of a Class I clone,
PH57, is shown in Fig. 1B. The cDNA PH
57 encodes a
333-amino acid protein with a high degree of similarity to the
precursor
-subunit of the rat liver enzyme (21): 305 of the 333 amino acids are identical in the two subunits. Both mammalian proteins carry a putative signal peptide of 27 amino acids, which would be
removed upon entry into the mitochondria (35). Part of the sequence
derived from a Class II clone, PH
108, is shown in Fig. 1C. As stated previously, PH
108 was identical to PH
57
except for the replacement of a short stretch of sequence within its open reading frame. Substitution of this sequence preserves the reading
frame and results in a novel 36 amino acid stretch within the
-subunit. The identities of the remaining clones were designated using the presence of a unique NdeI site at the 3
end of
the PH
108-specific sequence (Fig. 1A) as a discriminating
characteristic of Class II clones.
The molecular masses of the mature proteins (without signal sequence)
encoded by PH57 and PH
108, were calculated to be 32.11 kDa and
34.45 kDa, respectively. The value predicted for the PH
57-encoded protein is consistent with that estimated for the
-subunit of "native" SCS purified from pig heart tissue (36). Expression of
PH
57 in bacteria produced a protein that co-migrated with the
-subunit of the native enzyme on SDS-polyacrylamide gel
electrophoresis, whereas PH
108 produced a significantly larger
protein. Amino acid analyses of peptide fragments generated by cyanogen
bromide treatment of the native enzyme identified the
-subunit
encoded by PH
57 as that corresponding to the subunit present in
native enzyme. Moreover, the amino acid sequence derived from the 57 nucleotides present in the PH
57 cDNA is conserved in known forms of SCS (Fig. 2A). The unique
sequence from PH
108 was not found in a search of the various data
banks.
The crystallographic model of the -subunit of the E. coli
enzyme (illustrated in Fig. 2B) was examined focusing on the
corresponding location of the unique segments encoded by PH
57 and
PH
108. The amino acid sequence replaced in the two pig heart
isoforms corresponds to a polypeptide stretch that interconnects the
two domains of the
-subunit: the CoA-binding domain and the
phosphohistidine domain (Fig. 2B). Note: as can be seen in
Fig. 2A, the amino acid residues are highly conserved on
either side of both interconnecting sequences and are thought to play a
role in catalysis. The sequence SRSGTLTYE to the right forms an
-helix that stabilizes the phosphohistidine loop, whereas the
residues to the left come in close contact with the reactive thiol
group of CoA (4). This comparison to the model of the E. coli
-subunit serves to demonstrate that the longer amino acid
sequence encoded by PH
108 can be accommodated within the structure
of the protein.
Prompted
by the earlier reports of heterogeneity in the enzyme (10-13), we
considered both pig heart -subunits as authentic proteins
contributing to enzyme activity. To better understand the functional
significance of the novel PH
108-encoded protein, a bicistronic
expression system in bacteria was developed for the production of the
pig heart enzymes. The recombinant vector, pT7-6
Ec
/Ec
(Fig.
3A), had been used previously
for the production of the E. coli enzyme (41). In this
vector, the two genes for the E. coli enzyme overlap in such
a manner that translation of the
-subunit is coupled to that of the
-subunit, ensuring that equal levels of both are produced (37). The
nucleotide sequence responsible for this coupling is referred to here
as the translational coupling region (TCR) and occurs within the
overlapping ends of the two genes as illustrated in Fig. 3A. The
-subunit of SCS from either E. coli or pig heart was
produced in combination with each
-subunit isoform (see "Materials
and Methods"). The TCR was retained in all of these cDNA
constructs (Fig. 3B). Although high levels of the subunits
were produced by expression, greater than 90% of this protein was
insoluble. Despite this, we could clearly measure elevated levels of a
GTP-specific activity in the bacterial lysates following the expression
of both PH
/PH
57 and PH
/PH
108. The enzyme of the host
E. coli cells uses ATP or GTP as the nucleotide substrate
(42), contributing a basal SCS activity in these experiments. As shown
in Fig. 3C, when measured from the time of induction to
5 h, there was little increase above basal levels of the ATP- or
GTP-specific activities by the controls (Ec
/PH
57 and
Ec
/PH
108). In contrast, the GTP-specific activity associated with the expression of the two plasmids PH
/PH
57 and PH
/PH
108 increased by a factor of two over the course of the experiment, while the ATP-specific activity remained low and relatively constant as seen in the controls. We conclude that two separate isoenzymes with GTP-specific activities are produced by the combination of each of the two
-subunit isoforms with the pig heart
-subunit.
It was first reported by Cha et al. (10) that conventional
preparations of SCS from pig heart tissue were resolved into two
separate peaks of GTP-specific activity by hydroxyapatite column
chromatography. We have consistently made the same observation during
routine purification of the native pig heart enzyme. A profile
monitoring the GTP-specific activity of one such purification is
presented in Fig. 4C. The
first peak of enzymatic activity to emerge from the column is never
more than 10% of the total, but is consistently present and invariably
discarded. The second peak to emerge often accounts for greater that
90% of the total enzyme; these fractions are pooled for further
purification. In the present studies, bacterial cell extracts from
cultures with the two expression combinations PH/PH
57 and
PH
/PH
108 were prepared for hydroxyapatite column chromatography
as described under "Materials and Methods." Their simultaneous
fractionation on identical columns is illustrated in Fig. 4
(A and B, respectively). The first peak of
enzymatic activity that emerged from both columns with 65 mM phosphate was specific for both ATP and GTP, a
characteristic of the host bacterial SCS. The second peak of activity
detected, albeit at different points of elution in both fractionations, showed an absolute specificity for GTP, that expected of the pig heart
enzymes. As can be seen by comparing Fig. 4A with Fig.
4C, the GTP-specific activity produced by the PH
/PH
57
combination eluted from the column in the range of 170 mM
phosphate, coincident with the main peak of a native pig heart SCS
preparation. It is not surprising, then, that the native enzyme
purified from pig heart tissue had been found to comprise the PH
57
protein. It was, however, a surprise to find that the GTP-specific
activity produced by the PH
/PH
108 combination emerged earlier
with 110 mM phosphate, colinear with the minor peak of a
native preparation (compare Fig. 4B with Fig.
4C). Unfortunately, attempts to purify this minor isozyme
from the others in pig heart tissue have met with limited success.
Studies of the
Further experiments were
carried out with the aim of delineating the molecular mechanisms
involved in generating these two nearly identical cDNAs. We
screened two pig genomic DNA libraries (see "Materials and
Methods"). Two clones, designated E4 and
E6, were recovered
from a
EMBL-3 library. Fragments of
E6 hybridized to a probe
specific for the 57-nucleotide sequence (oligonucleotide 3 under
"Materials and Methods") but not to a probe specific for the
108-nucleotide sequence (oligonucleotide 4). In contrast, fragments of
E4 hybridized only to the probe specific for the 108-nucleotide
sequence. The positive hybridization results are presented in Fig.
5A. Furthermore, it appeared
that these clones represented two parts of the
-subunit gene since
E6 also contained the remaining upstream sequence while
E4
contained the remaining downstream sequence of the
-subunit. The two
clones failed to hybridize to each other, and thus did not overlap. A
third clone,
3551, was recovered from a
GEM11 library and proved
to overlap both
E6 and
E4. As shown by Southern blot analyses, an
11-kb fragment of
3551 hybridized to both probes (Fig.
5A). This clearly established that the two isoform sequences
originate from within a common gene.
A map generated from extensive trials of restriction digestions and hybridization analyses of these clones has been compiled in Fig. 5B. The sizes of the relevant introns were estimated using methods involving PCR (see Fig. 5B). The intron upstream of the 57-nucleotide sequence was 200 bp, and the intron downstream of the 108-nucleotide sequence was 1.3 kb. PCR carried out directly on genomic DNA with the same primer pairs yielded similarly sized fragments. The size of the intervening region between the 57- and 108-nucleotide exons was found to be 7.5 kb by restriction digest analyses.
The nucleotide sequences of the exon-intron boundaries were established
by sequence comparison of the cDNA and genomic clones. The
sequences unique to PH57 and PH
108 are represented as exons within the gene (the boxed sequences in Fig.
6A). With one exception, the
5
donor and 3
acceptor splice sites of the introns conform to the
"GT-AG" rule (43). The deviation from this rule involves the 5
donor and 3
acceptor splice sites of the intron that is immediately
downstream of the 108-nucleotide exon, and pertains to the splicing of
PH
108. The dinucleotide AT rather than GT is used at the 5
donor
site; to maintain reading frame, the 108-nucleotide exon must use the
dinucleotide AA at the 3
acceptor site (Fig. 6A). Note: the
dinucleotide AG at this acceptor site provides the perfect junction for
splicing to the 57-nucleotide exon. Thus, the 3
acceptor site is being
used differently by the two exons (Fig. 6B).
Since the production of the PH108 message involved the use of a
non-consensus splice site pair (AT-AA), we sought direct evidence for
the presence of the PH
108 message in tissues. Total RNA prepared
from four tissues of a newborn piglet was reverse transcribed with
oligo(dT) and subjected to PCR specific for each isoform according to
the scheme outlined in Fig. 7. The
PH
57 message was readily detected in all four tissues, liver, brain, heart, and skeletal muscle, whereas, the PH
108 message was only detected in heart and skeletal muscle (Fig. 7). We carried out 45 cycles of PH
108-specific PCR (compared with 40 cycles for PH
57)
to be certain about the absence of this message in liver and brain
tissue. Reverse transcribed-PCR was also carried out with each tissue
type to monitor the levels of the
-actin message as an internal
control. These observations led us to believe that production of the
PH
108 message is regulated in a tissue-specific manner. Furthermore,
the presence of the two messages in heart and skeletal muscle was
consistent with earlier reports of heterogeneity of GTP-specific SCS in
these two tissues (10, 11).
The full-length clones recovered in the library screen had identical 3
UTRs followed by short poly(A) tails (Fig. 1). The longest 5
UTR from
each class of clones was similar in size and identical in sequence. The
5
UTR was unexpectedly short, containing just 36 nucleotides (Fig.
1B). In contrast, for example, the 5
UTR of the corresponding rat
liver cDNA is over 500 bp long (21). To rule out the possibility
that these clones had truncated 5
UTRs due to anomalies in the
creation of the libraries, we carried out 5
-RACE reactions specific
for each isoform (for details see "Materials and Methods"). The
cDNA used in these reactions was synthesized from heart RNA, since
both messages were produced in this tissue. The longest 5
-RACE
products derived from the two messages were no longer than 36 nucleotides, the length already present in the cDNA clones. Thus,
we concluded that the only difference between the two messages was the
57- and 108-nucleotide sequences.
The derived amino acid sequence of the 57-nucleotide exon is
conserved in known forms of SCS (see Fig. 2A). Furthermore,
this is consistent with the finding that the PH57 protein is the
predominant form of the
-subunit in preparations of the enzyme from
pig heart tissue. The identification of the equivalent segment as an
external loop in the structure of the E. coli enzyme (Fig.
2B) suggests that changes in amino acid sequence and or
length (as found in PH
108) can be readily accommodated. Although
part of the active site region in the
-subunit, this sequence can
not dictate the specificity of the substrate nucleotide since both the
PH
57/PH
and PH
108/PH
constructs expressed GTP-specific SCS
activities (Fig. 3). Luo and Nishimura (44) created a mutant of
E. coli SCS in which one histidine residue (in
bold, Fig. 2A), corresponding to His-168 in the
PH
57-subunit sequence, was replaced by Asn. This mutant was devoid
of full catalytic activity but was able to catalyze some of the partial
reactions at significant rates. In particular, phosphorylation with ATP
occurred at a rate approximately 10-fold greater than that in the
wild-type enzyme, while the formation of succinyl-CoA proceeded at
one-tenth that of the wild-type rate (44). Thus, this loop is
positioned in the
-subunit to have possible influence on the
catalytic properties of SCS. And in turn, its substitution would have
possible influence on the metabolism of that cell.
Any standard biochemistry textbook will point out that ketone bodies
are routinely being used for fuel in cardiac and skeletal muscle of
higher animals (for example, see Ref. 45). The brain only uses ketone
bodies when glucose supplies are very low, such as during starvation;
liver tissue is unable to breakdown ketone bodies since it lacks the
central enzyme, CoA transferase. The pattern of PH108 levels in
various tissues matches that expected of an enzyme involved in ketone
body metabolism. On this basis we hypothesize that succinyl-CoA
synthetase is produced, in part, as two separate GTP-specific isoforms
in pig heart tissue. The major form contains the
-subunit encoded by
the PH
57 message and the minor form by PH
108. The major isoform
of the enzyme plays the role of substrate-level phosphorylation within
the citric acid cycle, while the minor form plays a specialized role
providing succinyl-CoA in ketone body activation. This would be
consistent with the proposal that two separate enzymes could best
accommodate the opposing functional demands of the citric acid cycle
and ketotic pathways (5, 6).
The splicing mechanism used to generate PH108 would certainly be
novel, since the intron immediately downstream of the 108-nucleotide exon carries the splice site dinucleotides, AT-AA. This would seem to
contravene the invariant GT-AG rule followed in pre-mRNA splicing
(43). Nevertheless, as reported by Carothers et al. (46),
the dinucleotide pair, AT-AA, has been demonstrated
experimentally to function within the context of a major
class intron. The fourth intron of the dihydrofolate reductase gene
normally contains the dinucleotide pair GT-AG commonly associated with
splicing. However a change in the last nucleotide of the intron from G
to A crippled the splicing (46). A suppressor mutation due to a
single-base substitution at the second site was identified and
isolated. The revertant was found to have an A in place of G in the
first position of the intron (46). Thus a major class GT-AG intron was
capable of splicing properly with donor and acceptor dinucleotides of AT and AA, respectively. The splicing of the intron immediately downstream of the 108-specific exon now represents an example of a
"natural" AT-AA intron.
The use of the dinucleotide AT as a 5 donor in splicing pre-mRNA
is not new. A minor class of introns containing the dinucleotide pair
AT-AC has recently been described (47, 48). Five such AT-AC introns are
known. All share conserved sequences at their 5
donor and branch sites
that are distinct from those found in major class introns.
Consequently, it was discovered that these introns use the
correspondingly rare small nuclear RNAs, U11 and U12, in place of the
major species, U1 and U2 (49, 50). We believe that the major small
nuclear RNAs, U1 and U2, are used in splicing the 57- and 108-exons to
the common downstream exon. Five of the six nucleotides defining their
5
donor sites, GTGAGT and ATAAGT, respectively, are identical to the
consensus binding sequence for U1, GTAAGT (51). Although we have not
yet identified the branch site, U2 is likely to be used in branch
formation. Splicing of the 57-exon involves a typical GT-AG
dinucleotide pair and would therefore be expected to use the major
splicing components, U1 and U2. Since AT-AA dinucleotide pairs have
previously been found to operate within the context of a major class
intron (46), it is conceivable that splicing of the 108-exon also uses the major splicing machinery.
The differential use of the common downstream exon deserves special
mention. It is believed that the formation of a non-Watson-Crick base
pair interaction between the first and last nucleotides of an intron is
necessary for splicing (52). Such interactions are only possible
between G-G, A-C, and A-A base pairs given the geometrical disposition
of the RNA strand within the spliceosome (53). Splicing in the majority
of cases, including the 57-exon, involves the formation of a G-G base
pair. Splicing of the minor class of AT-AC introns are possible due to
their ability to form A-C base pairs (53). If the splicing of the
108-exon were to use the same 3 acceptor dinucleotide as the 57-exon,
it would be forced to form an A-G base pair, an interaction that is
unfavorable given the constraints of geometry. This is consistent with
the observations that intron splicing is crippled by G to A changes in
the first nucleotide (46, 52). Thus, the splicing of the 108-exon
selects AA over AG at the 3
acceptor site to form an allowed A-A
non-Watson-Crick base pair. We speculate that the differential use of
the acceptor site can be accommodated by the "normal" splicing
machinery, and that the selection of the 108-exon is governed by
tissue-specific trans acting factors.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBankTM/EMBL Data Bank with accession number(s) AF008430-AF008433, AF008588, and AF008589.
We thank Warren Henning for able technical assistance and Roger Bradley for help with the photography. We are indebted to Marie Fraser, Michael Joyce, and Steven Rice for their thoughtful suggestions on the preparation of this manuscript.