Multiple physical forms of excised group II intron RNAs in wheat mitochondria
http://www.100md.com
《中华首席医学网》
ABSTRACT
Plant mitochondrial group II introns do not all possess hallmark ribozymic features such as the bulged adenosine involved
in lariat formation. To gain insight into their splicing pathways, we have examined the physical form of excised introns in
germinating wheat embryos. Using RT–PCR and cRT–PCR, we observed conventional lariats consistent with a two-step
transesterification pathway for introns such as nad2 intron 4, but this was not the case for the cox2 intron or nad1 intron
2. For cox2, we detected full-length linear introns, which possess non-encoded 3'terminaladenosines, as well as heterogeneous
circular introns, which lack 3' nucleotide stretches. These observations are consistent with hydrolytic splicing followed by
polyadenylation as well as an in vivo circularization pathway, respectively. The presence of both linear and circular species
in vivo is supported by RNase H analysis. Furthermore, the nad1 intron 2, which lacks a bulged nucleotide at the branchpoint
position, comprised a mixed population of precisely full-length molecules and circular ones which also include a short,
discrete block of non-encoded nucleotides. The presence of these various linear and circular forms of excised intron
molecules in plant mitochondria points to multiple novel group II splicing mechanisms in vivo.
INTRODUCTION
Almost all plant mitochondrial introns belong to the highly structured ribozymic group II class and the majority are
located within NADH dehydrogenase (nad) genes encoding complex I components of the respiratory chain. To date, a total of 25
have been identified among various flowering plants, and that includes six of the trans-splicing type [reviewed in (1,2)].
Group II introns are also found in organellar genes of other eukaryotes (excluding animals), as well as in a few bacterial
genes [reviewed in (1,3,4)]. In some cases, these introns have autocatalytic activity in vitro, even though intron-encoded
RNA maturases and/or host-encoded protein machinery is required for splicing in vivo. Only a few examples of self-splicing
introns have been reported for plant organelles, namely within algal chloroplast genes (5,6), but notably none in plant
mitochondria.
Conventional group II introns are excised through two transesterification steps, with the 2'-OH of an unpaired adenosine
7–8 nt from the 3' end of the intron acting as the first attacking nucleophile. This is the same biochemical mechanism as
for nuclear spliceosomal introns and such similarities suggest that they share a common evolutionary origin [cf. Ref. (7)].
For both classes, the intron is liberated as a lariat, although group II ones have a much shorter tail and are highly
structured with six distinctive domains (dI-dVI). From detailed biochemical and mutational analysis (primarily of fungal
mitochondrial and bacterial ribozymic introns), key interactions required for proper folding and autocatalytic activity in
vitro have been elucidated, and they include dV as part of the catalytic core and dVI which provides the first attacking
nucleophile [reviewed in (3,8)]. In particular, branching requires an adenosine in a bulged context within the dVI helix and
at a fixed distance from the 3' splice site, namely 7 nt for group IIA introns and 8 nt for group IIB (3,8–10). Mutations
within the dV/dVI region, particularly at the branchpoint, can shift the first splicing step from transesterification to
hydrolysis, and in some cases to incorrect selection of the 3' splice site (11,12).
In contrast to most other group II introns examined to date, those in plant mitochondria do not always conform to the
conventional group II secondary structural model [reviewed in (1)]. The core helices can have weak base-pairing and certain
key features sometimes appear missing. This raises questions about the mechanistic and evolutionary pathways taken by such ‘
degenerate’ introns. Interestingly, a plant chloroplast intron, trnV, which lacks a dVI bulged adenosine, is excised as a
linear molecule which is generated by a hydrolytic pathway of splicing (13). The importance of the dVI structure for
efficient splicing in plant mitochondria is also supported by mutational analysis of the wheat cox2 intron, in that
disruption of base pairing resulted in loss of splicing as monitored by in organello splicing assays (14). In the present
study, we have examined the excised physical forms of the wheat mitochondrial introns nad2 intron 4, nad1 intron 2 and cox2
intron 1, designated as nad2i1282, nad1i477 and cox2i373, respectively in the nomenclature of Dombrovska and Qiu [cf. Ref.
(15)]. We provide evidence for multiple splicing pathways in wheat mitochondria.
MATERIALS AND METHODS
RT–PCR and cRT–PCR analysis of intron RNAs
Mitochondrial RNA was isolated from 24 h germinating wheat embryos (Triticum aestivum cv. Frederick) as previously
described (16). To examine the sequences of lariat (or circularized) excised intron molecules, a synthetic oligomer mapping
close to the 5' end of the intron was used as primer in cDNA synthesis with Superscript II reverse transcriptase
(Invitrogen), which is able to cross 2'–5' bonds (17). Wheat mitochondrial RNA (5 μg) was denatured at 70°C for 10 min and
cDNA was synthesized using 10 U/μl reverse transcriptase, 0.2 vol 5x First Strand buffer (Invitrogen), 10 mM DTT and 0.5 mM
dNTPs incubated at 45°C for 2 h. Subsequently, this primer in combination with one mapping to the 3' end of the same intron
was used in PCR amplification with Taq DNA polymerase (Invitrogen). In parallel, cRT–PCR experiments (to circularize any
linear intron molecules which contain a 5' monophosphate) were performed (18) with wheat mitochondrial RNA (5 μg) being
incubated with 0.6 U/μl RNA ligase (New England Biolabs), 0.5 U/μl RNasin (Promega), 50 μg/ml BSA (Promega) and 0.1 vol
10x RNA ligase buffer (New England Biolabs) overnight at 14°C, and then recovered by ethanol precipitation prior to PCR
amplification. The resulting cRT–PCR products will also include any endogenously circularized molecules already present in
the RNA population. Prior treatment of RNA by heating at 70°C for 10 min or boiling for 5 min did not affect the nature or
yield of products. In some experiments, Superscript III reverse transcriptase (Invitrogen) was used, although it appeared
less efficient in traversing 2'–5' bonds. In the RT–PCR experiments, no product was seen in the absence of RT or when DNA
was used as template with the cRT–PCR primer pairs (because of their inverted orientation). The sole exception was an
abundant low molecular weight product for nad1 intron 2, which was determined to be co-linear with DNA by sequence analysis.
RT–PCR or cRT–PCR products were subsequently gel-purified using UltraClean 15 (MoBio Laboratories Inc.) and either
sequenced directly using Sequenase version 2.0 (US Biochemicals) and electrophoresed on urea-polyacrylamide sequencing gels
or after being cloned into pGemT-Easy (Promega) plasmid vectors (19). Automated sequencing was performed by the Ottawa Health
Research Institute DNA sequencing facility (OHRI) and the McGill University/Genome Québec Centre.
RNase H analysis of excised introns
Wheat mitochondrial RNA (5 μg) was incubated with intron-specific oligomers at 37°C for 30 min prior to treatment with
0.03 U/μl RNase H (Invitrogen), 1 U/μl RNasin (Promega), 0.27 μg/μl BSA (Promega) and 10 mM DTT at 37°C for 30 min (20).
RNase H-treated samples were phenol-extracted and ethanol-precipitated prior to northern blot analysis. Hybridizations were
performed with 32P-end-labelled (T4 polynucleotide kinase) oligomer probes designed based on wheat mitochondrial sequences.
The synthetic oligonucleotides used for RT–PCR, cRT–PCR, RNase H and as hybridization probes are shown in Supplementary
data (online).
RESULTS
Nature of dVI branchpoint region of plant mitochondrial group II introns
Most plant mitochondrial introns fall in the group IIA category, in which the bulged adenosine of dVI is located 7 nt
from the 3'end of the intron (Figure 1A, boxed) and the two unpaired terminal nucleotides (5'-AY-3') are involved in long-
range interactions including –' with the dII/III linker [reviewed in (3,8,10,21)]. Figure 1B shows such an example in nad2
intron 4, which conforms to this conventional structure and its sequence is highly conserved among plants (Figure 1B, inset
box). In contrast, a subset of plant mitochondrial introns, of both cis- and trans-splicing types, deviate from the classical
group II structure to varying extents (Figure 1C–H). The dVI helices of some have a lower predicted thermodynamic stability
than for classical group II introns, and this can be seen by A:G mispairs in the cox2 intron (Figure 1C), adjacent A:A and
G:G mispairs in nad5 intron 2 (Figure 1D), A:A mispairs in nad4 intron 2 (Figure 1F) and to a much more pronounced extent in
nad1 intron 1 [Figure 1H, Ref. (22)]. It is notable that plant mitochondrial intron folding can be improved by RNA editing,
which converts A–C mis-pairs to A–U pairs, although not all such mispairs in dV/dVI regions are seen to undergo editing in
the introns of mitochondria (23) or chloroplasts (13), and very few of the dVI mispairs shown in Figure 1 fall in that
category (Figure 1D and H, but cf. underlined lower case U in dVI of nad5 intron 2). Even those introns possessing strongly
paired dVI helices, in several cases deviate from the conventional group II format of having a purine stretch opposite a
string of downstream pyrimidines (Figure 1D–E, G–H). The cox2 intron contains an A:G mispair immediately distal to the dVI
branchpoint, as do a number of barley chloroplast introns, yet all of those were found to be excised as conventional lariats
(13).
With respect to the lariat-generating branchpoint, the majority of plant mitochondrial introns possess a bulged adenosine
at the expected position, however there are several interesting exceptions (Figure 1E–H). For example, a bulged nucleotide
is completely absent from the dVI helix of nad1 intron 2 and its short tight structure is conserved among all flowering
plants examined to date. In contrast, in the very weakly structured dVI of nad1 intron 1, the first candidate adenosine is
located at least 11 nt from the 3' end (Figure 1H) and considerable variation in sequence is seen among plants. In the case
of nad2 intron 1 and nad4 intron 2 (Figure 1E–F), although a bulged nucleotide is present at the expected location (albeit
flanked by pairs of varying strength among plants) it is not adenosine (cf. nad4 intron 2) or adenosine in only some plants
(cf. nad2 intron 1).
Indeed the extent of sequence variation among orthologous introns from various plants in the vicinity of the branchpoint
and its impact on potential base pairing (Figure 1E, F and H, shaded residues and insets) is surprising given the crucial
role of dVI in lariat formation, as well as the low rate of nucleotide substitution in plant mitochondrial DNA (24). This is
not the behaviour expected of sequences under strong functional constraint. In this regard, we had previously noted that dV
helices also tend to show weaker pairing than classical group II introns (23). In some cases, dV structure is improved by
editing, but in other instances the mispairs either remain unedited or they are not editing candidates (cf. Figure 1D and F,
boxed Cs and cox2 intron U:U). Incidentally, nad4 intron 2 and nad1 intron 1 also have aberrantly long dV loops, and this is
correlated with weak dVI helices and high sequence variation among plants (Figure1F and H). Based on such examination of dVI
features, we selected three cis-splicing introns from wheat for this study, namely, nad2 intron 4 which has a conventional
dVI (Figure 1B), nad1 intron 2, which has a short tight dVI helix (Figure 1G), and the cox2 intron, which has an A:G mispair
adjacent to the branchpoint and mispairing in dV (Figure 1C).
Atypical circularized and linear introns detected by RT–PCR and cRT–PCR
To assess the presence of lariat (and/or circularized) excised introns in wheat embryo RNA populations in vivo, we used
an RT–PCR strategy to amplify and then subsequently sequence the junction regions of nad2 intron 4, nad1 intron 2 and the
cox2 intron. These experiments were accompanied in parallel with RNA ligase-treated samples (cRT–PCR) to determine whether
linear intron molecules might also be present in the population (Figure 2). We included rps7 mRNA as a control, as it is
known to be a 5' processed transcript (25) and thus can be circularized with RNA ligase. As expected, the rps7 mRNA template
resulted in an RT–PCR product of 450 bp only when pre-treated with RNA ligase (Figure 2A, lane 1 versus lane 2, arrowhead).
For nad2 intron 4, nad1 intron 2 and the cox2 intron, we observed RT–PCR products for both the RNA ligase-treated and
untreated templates, and their respective sizes were as expected for excised introns (Figure 2A, lanes 4–9, arrowheads).
Several larger products were also observed (Figure 2A, lanes 4, 6, 7); however only nad2 intron 4 (Figure 2A, lane 4)
exhibited discrete hybridization signals with intron-specific probes in Southern blot analysis (Figure 2B, lane 4). The
latter may reflect amplification products which include neighbouring nad2 sequences, whereas those generated with the cox2
intron primers appear to be due to mis-priming from an unrelated genomic region. Similarly, higher molecular weight products
were reported in the analysis of barley chloroplast introns (13). The apparent shift from discrete to heterogeneous high
molecular weight products when RNA ligase-treated template was used with nad2 intron 4 primers (Figure 2A, lane 4 versus lane
5) may relate to contribution from circularized nad2 trans-splicing precursor molecules. In the case of nad1 intron 2, the
abundant low molecular weight species (Figure 2A lanes 8, 9) were confirmed by sequence analysis to be mis-primed products
containing an internal section of nad1 intron 2 that is co-linear with mitochondrial DNA (Figure 2B, lanes 8, 9), and this is
consistent with the absence of Southern hybridization signal because of the intron oligomer probe maps outside that region.
Comparable results with those shown in Figure 2 were reproducibly obtained with templates from different mitochondrial RNA
preparations, different cDNA preparations and when alternative primers were used for cDNA synthesis.
To determine the precise nature of the excised intron junction, these RT–PCR products were cloned and sequenced. Such data
are shown in Table 1. As expected, the nad2 intron 4 junction sequence was consistent with the excised intron being a lariat
(with a 6 nt tail) based on six out of six RT–PCR clones and four out of four cRT–PCR clones. At the branchpoint position,
we observed mis-incorporation of adenosine instead of thymidine, as has been previously reported for templates with 2'–5'
bonds (13,17,22).
In contrast, for the cox2 intron only one out of five RT–PCR clones was consistent with a lariat model of splicing
(Table 1). This was unexpected as this intron contains a bulged A in dVI (Figure 1B). The other four clones displayed
heterogeneous circles each missing different lengths (from 10–29 nts) from the 3' terminal region of the intron. It should
be noted that 50% of the clones obtained from multiple experiments had the correct insert size but sequencing revealed
heterogeneous stretches of upstream and/or downstream exon sequences in addition to all or part of the intron (data not
shown). This was not seen in our examination of any of the other introns. Such species might be derived from degraded RNA
molecules, yet it is curious that they appear to be endogenously circularized. In this regard, evidence has been presented
for a potential cryptic 5' splice site within cox2 exon 1 of petunia (26), but that appeared to be discrete rather than
heterogeneous in nature.
When the cox2 intron cRT–PCR (RNA ligase-treated) products were cloned and sequenced to assess the presence of any
linear intron RNA species in the population, we found that six out of eight clones had short non-encoded polyA tracts (1–5
nt in length) and four of these were full-length introns whereas the other two lacked 3' nucleotide stretches. In addition,
one clone represented a full-length circularized intron species (end-to-end) and one simply lacked 3'terminal nucleotides.
The latter presumably reflects contribution from the pre-existing non-RNA-ligase treated population of endogenously
circularized forms. Notably in all cases, the precise 5' end of the cox2 intron was observed. The presence of a mixed
population of excised cox2 intron species was corroborated by direct sequencing of the cRT–PCR products (Figure 2D). It can
be seen that the sequence is homogeneous until the 5' end of the intron, but immediately past the junction there is a mixture
of A-rich sequences (Figure 2D, asterisk).
The nad1 intron 2, which contains a short tight dVI helix, showed yet a different set of features in its population of
excised intron molecules and none were lariats. Rather 5/9 clones showed a discrete 5'-ACAAAAC-3' insert at the junction of
full-length introns, and its relative abundance is evident in the directly sequenced population of RT–PCR products (Figure
2E, boxed nts). Of the remaining four clones, one represented a full-length circularized intron, while two clones also
contained 5'-ACTGT-3' which matches the first 5 nt of nad1 exon 2, and one clone lacked the 3' terminal 32 nt of the intron
(Table 1). Similar results were observed for the cRT–PCR analysis, in that 3/10 clones showed precise end-to-end
circularized species, and the others also contained either the non-encoded ACAAAAC block (4/10 clones) or ACTGT (3/10
clones). When the cRT–PCR products were examined directly, the sequence after the junction (Figure 2F) appears to be
somewhat more heterogeneous than what was seen for the in vivo population (Figure 2E).
Linear and circular excised introns in vivo supported by RNase H analysis
Although the above strategy provides information about the physical form of excised introns, it does not directly address
their relative abundance, especially since the yields of RT–PCR products may be influenced by reverse transcriptase being
less efficient at extending over the 2'–5' bond of a lariat species compared with the 3'–5' bond generated by ligation of
the ends of a previously linear molecule (cf. Figure 2C). To assess the relative levels of spliced lariat (and/or
circularized) versus linear physical forms of introns in vivo, wheat mitochondrial RNA was annealed with intron-specific
oligomers prior to incubation with RNase H and subsequent northern blot hybridization. In our analysis, we included nad7
intron 4 which we had previously shown to be predominantly lariat, but which also exhibited a minor population of fully
circularized (end-to-end) intron molecules (16,27), and as expected, the vast majority appear to be lariat rather than linear
(Figure 3A, black vs. open arrowhead). In the case of nad2 intron 4, when we used two annealing oligomers prior to RNase H
treatment, again the shift in size was consistent with a lariat (or circularized) excised intron (Figure 3B, black arrowhead
denoting 900 nt product detected by probe), when compared with the native 1.4 kb intron (Figure 3B, black dot). A minor
signal was detected at the position corresponding to a linear form (400 nt, Figure 3B, open arrowhead). Also, major products
of 4.8 kb derived from the linear precursor of 6 kb can be seen (Figure 3B) in addition to the precursor and intron species
which were not completely digested by RNase H.
RNase H analysis of the cox2 intron using two intron-specific primers revealed conversion of the 1.1 kb intron (Figure
3C, black dot) to products consistent with circular (or lariat) molecules of 600 nt (Figure 3C, black arrowhead) and linear
ones of 300 nt (Figure 3C, open arrowhead), as well as a shift in the 2.6 kb precursor to a prominent species of 800 nt. The
relative strengths of the hybridization signals for the linear and circular forms (Figure 3C, open versus black arrowheads)
suggest that both species are rather equally represented in the population, and this corroborates the results obtained with
RT–PCR and cRT–PCR (Figure 2 and Table 1).
In the case of wheat nad1 intron 2, more complex northern profiles are observed because there are multiple precursors due
to its co-transcription with upstream atp6 and rps13 genes, and this results in RNA species of 6, 4.6 and 3.6 kb (Figure 3D).
After RNase H treatment, when only one intron-specific primer was used (oligomer g), the signal for the excised intron of 1.4
kb shifts to a 1.2 kb product consistent with the presence of a linear species (Figure 3D, black dot versus open arrowhead).
This outcome was also seen when oligomer h was used, the shift being to 960 nt (Figure 3D, open arrowhead). In both cases,
the linear species appear relatively abundant. Lariat (or circular) species when converted to Y-shaped structures by RNase H,
would be expected to co-migrate with the native form under these electrophoretic conditions, but there is very little signal
at that position (Figure 3D, black arrowhead and black dot). The high molecular weight precursors shift to the expected sizes
of 2.6 and 2.3 kb. It is worth noting that the RNA ligase (cRT–PCR) experiments (cf. Figure 2A, lane 9) might be influenced
by protective structures at the intron termini, such as the 3' terminal hairpin of nad1 intron 2.
DISCUSSION
Our analysis of the cox2 and nad1 excised introns in RNA isolated from germinating wheat embryos has uncovered novel
physical forms which deviate from the classical lariat structure of group II introns. In contrast to introns such as nad2
intron 4 (this work), nad7 intron 3 (21,27) and nad4 intron 3 (N. Niknejad and L. Bonen unpublished data) which possess a
bulged adenosine branchpoint in dVI and are excised as lariats, our observations in the case of nad1 intron 2 and the cox2
intron support the presence of both circularized and linear molecules in vivo (Figure 4). Because the cox2 intron has an
apparently conventional branchpoint adenosine, it was unexpected that the lariat form was not predominant, but instead we
observed heterogeneous circular excised introns which lack various lengths at the 3'terminus. Moreover, the population
included full-length linear intron molecules with short, non-encoded polyA stretches.
Taken together, our data for the cox2 intron support a hydrolytic pathway of splicing generating linear molecules which
undergo either (i) polyadenylation at the 3' terminus [perhaps mediated by the bacterial-type tagging of transcripts for
degradation observed in plant organelles, Ref. (28)] or (ii) endogenous circularization after 3' exonucleolytic attack
(Figure 4B). Linear molecules have also previously been reported for the Streptococcus GBS intron, which possesses a
conventional branchpoint adenosine (29), and for the barley chloroplast trnV intron, which in addition is polyadenylated
(13); however, the wheat cox2 intron population includes endogenously circularized molecules as well. This might occur
through nucleophilic attack by 2'-OH groups at other positions within dV or dVI to generate aberrant lariats with longer
tails of varying lengths (Figure 4C). However, the cloned sequences correspond identically to the genomic sequence at such
putative novel branchpoints, whereas typically an adenosine branchpoint is mis-read as T by reverse transcriptase (13,17,22).
Alternatively, circularization might occur via an endogenous RNA ligase. It is also not excluded that the linear molecules
which we observed are derived from the debranching of lariats with subsequent protection from exonuclease attack.
The nad1 intron 2 appears to be present as linear and circular molecules in vivo, but they have quite different physical
features compared with the cox2 intron (Figure 4B and C). Notably, the excised introns fall into three categories: (i)
precisely full-length molecules, (ii) those which in addition have a discrete non-encoded ACAAAAC block and (iii) ones which
have the full-length intron plus ACTGT (which corresponds to the first 5 nt of the downstream exon). The latter could be
derived from incorrect 3' splice site selection [cf. Bacillus cereus, Ref. (30) and Bacillus anthracis, Ref. (31)] and such
alternative splicing would lead to defective mRNA molecules. Because exactly the same non-encoded nucleotide blocks were
consistently seen (and with or without RNA ligase treatment), these observations differ from those for the cox2 intron and do
not easily conform to a model of being generated by a polyA ragged tailing mechanism (Figure 4B). The endogenous linear
molecules, which comprise the majority based on RNase H analysis, appear resistant to 3' exonuclease attack (unlike the cox2
intron) and this may be related to the short tight dVI helix [reminiscent of the 3' termini of certain plant mitochondrial
mRNAs, Ref. (32)]. In all cases, the correct 5' splice site was seen both for nad1 intron 2 as well as the heterogeneous cox2
intron molecules. The apparent absence of exonuclease attack at the 5' end of putative linear molecules might relate to
protective RNA structures or close association with proteins.
Our nad1 intron 2 data suggest a combination of splicing mechanisms which exploit various internal or external
nucleophiles to explain the discrete non-encoded block (Figure 4B and C, hatched box). The latter would be analogous to group
I intron splicing where an external guanosine acts as the attacking nucleophile [reviewed in (33)]. The origin of the non-
encoded ACAAAAC block is unknown. If it is added to the 3' end of the intron, it seems unlikely to be mediated by tRNA
nucleotidyl transferase or an RNA-directed RNA polymerase, as was suggested for tobacco chloroplast ndhD mRNA processing
(34), since there is no obvious template. Other models consistent with a precisely full-length circularized intron (which we
observed at low levels for nad1 intron 2, nad7 intron 4 and the cox2 intron) include reverse splicing or a ‘spliced exon
reopening’ pathway [cf. yeast mitochondrial aI2 intron circles in vivo, Ref. (35)] whereby the 3'-OH of the liberated 5'-
exon acts as first attacking nucleophile in trans and then the hydroxyl group of the intron terminal pyrimidine attacks the
5' splice site (Figure 4C). Interestingly, in plant mitochondria, uncoupled splicing intermediates have been observed to be
relatively abundant in some cases (16) and thus might contribute to such a pathway.
The apparent loss of classical group II features of mitochondrial introns in the flowering plant lineage over
evolutionary time is supported by comparisons with homologues from early-diverging lineages of land plants. For example, nad1
intron 1 in Isoetes lacustris (22,36) and the cox2 intron in Acorus calamus (37) have conventional dVI structures. Moreover,
the dynamic nature of plant mitochondrial genomes [reviewed in (2)] can have impact on intron sequences, such as the genomic
rearrangements which have led to trans-splicing. It also appears that gene conversion (or swapping) events can occur between
group II core structures. For example in Figure 1, it can be seen that dV and part of dVI of nad1 intron 2 and nad2 intron 1
are identical. This also underscores the plasticity and tolerance that the splicing machinery must have in light of such
deviation from conventional structure.
The mixture of physical forms of excised introns observed in this study suggests that multiple biochemical mechanisms of
splicing occur in plant mitochondria, such as a hydrolytic pathway as well as unusual circularization events. Their relative
contributions appear to vary among introns and it will be interesting to learn if there are developmental effects, perhaps
due to differences in the availability of nuclear-encoded machinery especially early in germination when seeds leave dormancy
(16). In this regard, it is interesting that the steady state levels of excised introns show marked differences during
germination, typically being present at higher levels in embryos than in seedlings. The apparently ‘degenerate’ nature of
the folding of these group II introns raises the possibility that additional proteins and/or small RNAs have been recruited
for the splicing process in plant mitochondria.
SUPPLEMENTARY DATA
ACKNOWLEDGEMENTS
The authors thank C. Carrillo, N. Niknejad, S. Subramanian and J. Crosthwait for their earlier work on wheat and rice
mitochondrial introns in our laboratory. The authors also thank Dr R. Pandeya (Agriculture and Agri-Food Canada, Ottawa) for
kindly providing the wheat seeds used in this study. Financial support in the form of a Discovery Grant from the Natural
Sciences and Engineering Research Council of Canada (L.B.) and an Ontario Graduate Scholarship (J.L.P.T.) is gratefully
acknowledged. Funding to pay the Open Access publication charges for this article was provided by the Natural Sciences and
Engineering Research Council of Canada.
Conflict of interest statement. None declared.
REFERENCES
Bonen, L. and Vogel, J. (2001) The ins and outs of group II introns Trends Genet, . 17, 322–331
Knoop, V. (2004) The mitochondrial DNA of land plants: peculiarities in phylogenetic perspective Curr. Genet, . 46, 123–
139
Michel, F. and Ferat, J.L. (1995) Structure and activities of group II intron Annu. Rev. Biochem, . 64, 435–461
Lambowitz, A.M. and Zimmerly, S. (2004) Mobile group II introns Annu. Rev. Genet, . 38, 1–35 .
Sheveleva, E.V. and Hallick, R.B. (2004) Recent horizontal intron transfer to a chloroplast genome Nucleic Acids Res, .
32, 803–810
Odom, O.W., Shenkenberg, D.L., Garcia, J.A., Herrin, D.L. (2004) A horizontally acquired group II intron in the
chloroplast psbA gene of a psychrophilic Chlamydomonas: in vitro self-splicing and genetic evidence for maturase activity
RNA, 10, 1097–1107
Seetharaman, M., Eldho, N.V., Padgett, R.A., Dayie, K.T. (2006) Structure of a self-splicing group II intron catalytic
effector domain 5: parallels with spliceosomal U6 RNA RNA, 12, 235–247
Lehmann, K. and Schmidt, U. (2003) Group II introns: structure and catalytic versatility of large natural ribozymes Crit.
Rev. Biochem. Mol. Biol, . 38, 249–303.
Chu, V.T., Adamidi, C., Liu, Q., Perlman, P.S., Pyle, A.M. (2001) Control of branch-site choice by a group II intron EMBO
J, . 20, 6866–6876.
Toor, N., Hausner, G., Zimmerly, S. (2001) Coevolution of group II intron RNA structures with their intron-encoded
reverse transcriptases RNA, 7, 1142–1152
Podar, M., Chu, V.T., Pyle, A.M., Perlman, P.S. (1998) Group II intron splicing in vivo by first-step hydrolysis Nature,
391, 915–918
Boulanger, S.C., Faix, P.H., Yang, H., Zhuo, J., Franzen, J.S., Peebles, C.L., Perlman, P.S. (1996) Length changes in the
joining segment between domains 5 and 6 of a group II intron inhibit self-splicing and alter 3' splice site selection Mol.
Cell. Biol, . 16, 5896–5904
Vogel, J. and Borner, T. (2002) Lariat formation and a hydrolytic pathway in plant chloroplast group II intron splicing
EMBO J, . 21, 3794–3803
Farre, J.C. and Araya, A. (2002) RNA splicing in higher plant mitochondria: determination of functional elements in group
II intron from chimeric cox II gene in electroporated wheat Plant J, . 29, 203–213
Dombrovska, O. and Qiu, Y.L. (2004) Distribution of introns in the mitochondrial gene nad1 in land plants: phylogenetic
and molecular evolutionary implications Mol. Phylogenet. Evol, . 32, 246–263
Li-Pook-Than, J., Carrillo, C., Bonen, L. (2004) Variation in mitochondrial transcript profiles of protein-coding genes
during early germination and seedling development in wheat Curr. Genet, . 46, 374–380
Vogel, J., Hess, W.R., Borner, T. (1997) Precise branch point mapping and quantification of splicing intermediates
Nucleic Acids Res, . 25, 2030–2031
Vogel, J. and Hess, W.R. (2001) Complete 5' and 3' end maturation of group II intron-containing tRNA precursors RNA, 7,
285–292
Fedor, M.J. and Williamson, J.R. (2005) The catalytic diversity of RNAs Nature Rev. Mol. Cell Biol, . 6, 399–412
Zandueta-Criado, A. and Bock, R. (2004) Surprising features of plastid ndhD transcripts: addition of non-encoded
nucleotides and polysome association of mRNAs with an unedited start codon Nucleic Acids Res, . 32, 542–550
Murray, H.L., Mikheeva, S., Coljee, V.W., Turczyk, B.M., Donahue, W.F., Bar-Shalom, A., Jarrell, K.A. (2001) Excision of
group II introns as circles Mol. Cell, 8, 210–211 .
Malek, O. and Knoop, V. (1998) trans-splicing group II introns in plant mitochondria: the complete set of cis-arranged
homologs in fersn, fern allies, and a hornwort RNA, 4, 1599–1609
Albertazzi, F.J., Kudla, J., Bock, R. (1998) The cox2 locus of the primitive angiosperm plant Acorus calamus: molecular
structure, transcript processing and RNA editing Mol. Gen. Genet, . 259, 591–600
Farré, J.C. and Araya, A. (1999) The mat-r open reading frame is transcribed from a non-canonical promoter and contains
an internal promoter to co-transcribe exons nad1e and nad5III in wheat mitochondria Plant Mol. Biol, . 40, 959–967
Wissinger, B., Schuster, W., Brennicke, A. (1991) Trans splicing in Oenothera mitochondria: nad1 mNAs are edited in exon
and trans-splicing group II intron sequences Cell, 65, 473–482
Ogihara, Y., Yamazaki, Y., Murai, K., Kanno, A., Terachi, T., Shiina, T., Miyashita, N., Nasuda, S., Nakamura, C., Mori,
N., Takumi, S., Murata, M., Futo, S., Tsunewaki, K. (2005) Structural dynamics of cereal mitochondrial genomes as revealed by
complete nucleotide sequencing of the wheat mitochondrial genome Nucleic Acids Res, . 33, 6235–6250(Jennifer Li-Pook-Than and)
Plant mitochondrial group II introns do not all possess hallmark ribozymic features such as the bulged adenosine involved
in lariat formation. To gain insight into their splicing pathways, we have examined the physical form of excised introns in
germinating wheat embryos. Using RT–PCR and cRT–PCR, we observed conventional lariats consistent with a two-step
transesterification pathway for introns such as nad2 intron 4, but this was not the case for the cox2 intron or nad1 intron
2. For cox2, we detected full-length linear introns, which possess non-encoded 3'terminaladenosines, as well as heterogeneous
circular introns, which lack 3' nucleotide stretches. These observations are consistent with hydrolytic splicing followed by
polyadenylation as well as an in vivo circularization pathway, respectively. The presence of both linear and circular species
in vivo is supported by RNase H analysis. Furthermore, the nad1 intron 2, which lacks a bulged nucleotide at the branchpoint
position, comprised a mixed population of precisely full-length molecules and circular ones which also include a short,
discrete block of non-encoded nucleotides. The presence of these various linear and circular forms of excised intron
molecules in plant mitochondria points to multiple novel group II splicing mechanisms in vivo.
INTRODUCTION
Almost all plant mitochondrial introns belong to the highly structured ribozymic group II class and the majority are
located within NADH dehydrogenase (nad) genes encoding complex I components of the respiratory chain. To date, a total of 25
have been identified among various flowering plants, and that includes six of the trans-splicing type [reviewed in (1,2)].
Group II introns are also found in organellar genes of other eukaryotes (excluding animals), as well as in a few bacterial
genes [reviewed in (1,3,4)]. In some cases, these introns have autocatalytic activity in vitro, even though intron-encoded
RNA maturases and/or host-encoded protein machinery is required for splicing in vivo. Only a few examples of self-splicing
introns have been reported for plant organelles, namely within algal chloroplast genes (5,6), but notably none in plant
mitochondria.
Conventional group II introns are excised through two transesterification steps, with the 2'-OH of an unpaired adenosine
7–8 nt from the 3' end of the intron acting as the first attacking nucleophile. This is the same biochemical mechanism as
for nuclear spliceosomal introns and such similarities suggest that they share a common evolutionary origin [cf. Ref. (7)].
For both classes, the intron is liberated as a lariat, although group II ones have a much shorter tail and are highly
structured with six distinctive domains (dI-dVI). From detailed biochemical and mutational analysis (primarily of fungal
mitochondrial and bacterial ribozymic introns), key interactions required for proper folding and autocatalytic activity in
vitro have been elucidated, and they include dV as part of the catalytic core and dVI which provides the first attacking
nucleophile [reviewed in (3,8)]. In particular, branching requires an adenosine in a bulged context within the dVI helix and
at a fixed distance from the 3' splice site, namely 7 nt for group IIA introns and 8 nt for group IIB (3,8–10). Mutations
within the dV/dVI region, particularly at the branchpoint, can shift the first splicing step from transesterification to
hydrolysis, and in some cases to incorrect selection of the 3' splice site (11,12).
In contrast to most other group II introns examined to date, those in plant mitochondria do not always conform to the
conventional group II secondary structural model [reviewed in (1)]. The core helices can have weak base-pairing and certain
key features sometimes appear missing. This raises questions about the mechanistic and evolutionary pathways taken by such ‘
degenerate’ introns. Interestingly, a plant chloroplast intron, trnV, which lacks a dVI bulged adenosine, is excised as a
linear molecule which is generated by a hydrolytic pathway of splicing (13). The importance of the dVI structure for
efficient splicing in plant mitochondria is also supported by mutational analysis of the wheat cox2 intron, in that
disruption of base pairing resulted in loss of splicing as monitored by in organello splicing assays (14). In the present
study, we have examined the excised physical forms of the wheat mitochondrial introns nad2 intron 4, nad1 intron 2 and cox2
intron 1, designated as nad2i1282, nad1i477 and cox2i373, respectively in the nomenclature of Dombrovska and Qiu [cf. Ref.
(15)]. We provide evidence for multiple splicing pathways in wheat mitochondria.
MATERIALS AND METHODS
RT–PCR and cRT–PCR analysis of intron RNAs
Mitochondrial RNA was isolated from 24 h germinating wheat embryos (Triticum aestivum cv. Frederick) as previously
described (16). To examine the sequences of lariat (or circularized) excised intron molecules, a synthetic oligomer mapping
close to the 5' end of the intron was used as primer in cDNA synthesis with Superscript II reverse transcriptase
(Invitrogen), which is able to cross 2'–5' bonds (17). Wheat mitochondrial RNA (5 μg) was denatured at 70°C for 10 min and
cDNA was synthesized using 10 U/μl reverse transcriptase, 0.2 vol 5x First Strand buffer (Invitrogen), 10 mM DTT and 0.5 mM
dNTPs incubated at 45°C for 2 h. Subsequently, this primer in combination with one mapping to the 3' end of the same intron
was used in PCR amplification with Taq DNA polymerase (Invitrogen). In parallel, cRT–PCR experiments (to circularize any
linear intron molecules which contain a 5' monophosphate) were performed (18) with wheat mitochondrial RNA (5 μg) being
incubated with 0.6 U/μl RNA ligase (New England Biolabs), 0.5 U/μl RNasin (Promega), 50 μg/ml BSA (Promega) and 0.1 vol
10x RNA ligase buffer (New England Biolabs) overnight at 14°C, and then recovered by ethanol precipitation prior to PCR
amplification. The resulting cRT–PCR products will also include any endogenously circularized molecules already present in
the RNA population. Prior treatment of RNA by heating at 70°C for 10 min or boiling for 5 min did not affect the nature or
yield of products. In some experiments, Superscript III reverse transcriptase (Invitrogen) was used, although it appeared
less efficient in traversing 2'–5' bonds. In the RT–PCR experiments, no product was seen in the absence of RT or when DNA
was used as template with the cRT–PCR primer pairs (because of their inverted orientation). The sole exception was an
abundant low molecular weight product for nad1 intron 2, which was determined to be co-linear with DNA by sequence analysis.
RT–PCR or cRT–PCR products were subsequently gel-purified using UltraClean 15 (MoBio Laboratories Inc.) and either
sequenced directly using Sequenase version 2.0 (US Biochemicals) and electrophoresed on urea-polyacrylamide sequencing gels
or after being cloned into pGemT-Easy (Promega) plasmid vectors (19). Automated sequencing was performed by the Ottawa Health
Research Institute DNA sequencing facility (OHRI) and the McGill University/Genome Québec Centre.
RNase H analysis of excised introns
Wheat mitochondrial RNA (5 μg) was incubated with intron-specific oligomers at 37°C for 30 min prior to treatment with
0.03 U/μl RNase H (Invitrogen), 1 U/μl RNasin (Promega), 0.27 μg/μl BSA (Promega) and 10 mM DTT at 37°C for 30 min (20).
RNase H-treated samples were phenol-extracted and ethanol-precipitated prior to northern blot analysis. Hybridizations were
performed with 32P-end-labelled (T4 polynucleotide kinase) oligomer probes designed based on wheat mitochondrial sequences.
The synthetic oligonucleotides used for RT–PCR, cRT–PCR, RNase H and as hybridization probes are shown in Supplementary
data (online).
RESULTS
Nature of dVI branchpoint region of plant mitochondrial group II introns
Most plant mitochondrial introns fall in the group IIA category, in which the bulged adenosine of dVI is located 7 nt
from the 3'end of the intron (Figure 1A, boxed) and the two unpaired terminal nucleotides (5'-AY-3') are involved in long-
range interactions including –' with the dII/III linker [reviewed in (3,8,10,21)]. Figure 1B shows such an example in nad2
intron 4, which conforms to this conventional structure and its sequence is highly conserved among plants (Figure 1B, inset
box). In contrast, a subset of plant mitochondrial introns, of both cis- and trans-splicing types, deviate from the classical
group II structure to varying extents (Figure 1C–H). The dVI helices of some have a lower predicted thermodynamic stability
than for classical group II introns, and this can be seen by A:G mispairs in the cox2 intron (Figure 1C), adjacent A:A and
G:G mispairs in nad5 intron 2 (Figure 1D), A:A mispairs in nad4 intron 2 (Figure 1F) and to a much more pronounced extent in
nad1 intron 1 [Figure 1H, Ref. (22)]. It is notable that plant mitochondrial intron folding can be improved by RNA editing,
which converts A–C mis-pairs to A–U pairs, although not all such mispairs in dV/dVI regions are seen to undergo editing in
the introns of mitochondria (23) or chloroplasts (13), and very few of the dVI mispairs shown in Figure 1 fall in that
category (Figure 1D and H, but cf. underlined lower case U in dVI of nad5 intron 2). Even those introns possessing strongly
paired dVI helices, in several cases deviate from the conventional group II format of having a purine stretch opposite a
string of downstream pyrimidines (Figure 1D–E, G–H). The cox2 intron contains an A:G mispair immediately distal to the dVI
branchpoint, as do a number of barley chloroplast introns, yet all of those were found to be excised as conventional lariats
(13).
With respect to the lariat-generating branchpoint, the majority of plant mitochondrial introns possess a bulged adenosine
at the expected position, however there are several interesting exceptions (Figure 1E–H). For example, a bulged nucleotide
is completely absent from the dVI helix of nad1 intron 2 and its short tight structure is conserved among all flowering
plants examined to date. In contrast, in the very weakly structured dVI of nad1 intron 1, the first candidate adenosine is
located at least 11 nt from the 3' end (Figure 1H) and considerable variation in sequence is seen among plants. In the case
of nad2 intron 1 and nad4 intron 2 (Figure 1E–F), although a bulged nucleotide is present at the expected location (albeit
flanked by pairs of varying strength among plants) it is not adenosine (cf. nad4 intron 2) or adenosine in only some plants
(cf. nad2 intron 1).
Indeed the extent of sequence variation among orthologous introns from various plants in the vicinity of the branchpoint
and its impact on potential base pairing (Figure 1E, F and H, shaded residues and insets) is surprising given the crucial
role of dVI in lariat formation, as well as the low rate of nucleotide substitution in plant mitochondrial DNA (24). This is
not the behaviour expected of sequences under strong functional constraint. In this regard, we had previously noted that dV
helices also tend to show weaker pairing than classical group II introns (23). In some cases, dV structure is improved by
editing, but in other instances the mispairs either remain unedited or they are not editing candidates (cf. Figure 1D and F,
boxed Cs and cox2 intron U:U). Incidentally, nad4 intron 2 and nad1 intron 1 also have aberrantly long dV loops, and this is
correlated with weak dVI helices and high sequence variation among plants (Figure1F and H). Based on such examination of dVI
features, we selected three cis-splicing introns from wheat for this study, namely, nad2 intron 4 which has a conventional
dVI (Figure 1B), nad1 intron 2, which has a short tight dVI helix (Figure 1G), and the cox2 intron, which has an A:G mispair
adjacent to the branchpoint and mispairing in dV (Figure 1C).
Atypical circularized and linear introns detected by RT–PCR and cRT–PCR
To assess the presence of lariat (and/or circularized) excised introns in wheat embryo RNA populations in vivo, we used
an RT–PCR strategy to amplify and then subsequently sequence the junction regions of nad2 intron 4, nad1 intron 2 and the
cox2 intron. These experiments were accompanied in parallel with RNA ligase-treated samples (cRT–PCR) to determine whether
linear intron molecules might also be present in the population (Figure 2). We included rps7 mRNA as a control, as it is
known to be a 5' processed transcript (25) and thus can be circularized with RNA ligase. As expected, the rps7 mRNA template
resulted in an RT–PCR product of 450 bp only when pre-treated with RNA ligase (Figure 2A, lane 1 versus lane 2, arrowhead).
For nad2 intron 4, nad1 intron 2 and the cox2 intron, we observed RT–PCR products for both the RNA ligase-treated and
untreated templates, and their respective sizes were as expected for excised introns (Figure 2A, lanes 4–9, arrowheads).
Several larger products were also observed (Figure 2A, lanes 4, 6, 7); however only nad2 intron 4 (Figure 2A, lane 4)
exhibited discrete hybridization signals with intron-specific probes in Southern blot analysis (Figure 2B, lane 4). The
latter may reflect amplification products which include neighbouring nad2 sequences, whereas those generated with the cox2
intron primers appear to be due to mis-priming from an unrelated genomic region. Similarly, higher molecular weight products
were reported in the analysis of barley chloroplast introns (13). The apparent shift from discrete to heterogeneous high
molecular weight products when RNA ligase-treated template was used with nad2 intron 4 primers (Figure 2A, lane 4 versus lane
5) may relate to contribution from circularized nad2 trans-splicing precursor molecules. In the case of nad1 intron 2, the
abundant low molecular weight species (Figure 2A lanes 8, 9) were confirmed by sequence analysis to be mis-primed products
containing an internal section of nad1 intron 2 that is co-linear with mitochondrial DNA (Figure 2B, lanes 8, 9), and this is
consistent with the absence of Southern hybridization signal because of the intron oligomer probe maps outside that region.
Comparable results with those shown in Figure 2 were reproducibly obtained with templates from different mitochondrial RNA
preparations, different cDNA preparations and when alternative primers were used for cDNA synthesis.
To determine the precise nature of the excised intron junction, these RT–PCR products were cloned and sequenced. Such data
are shown in Table 1. As expected, the nad2 intron 4 junction sequence was consistent with the excised intron being a lariat
(with a 6 nt tail) based on six out of six RT–PCR clones and four out of four cRT–PCR clones. At the branchpoint position,
we observed mis-incorporation of adenosine instead of thymidine, as has been previously reported for templates with 2'–5'
bonds (13,17,22).
In contrast, for the cox2 intron only one out of five RT–PCR clones was consistent with a lariat model of splicing
(Table 1). This was unexpected as this intron contains a bulged A in dVI (Figure 1B). The other four clones displayed
heterogeneous circles each missing different lengths (from 10–29 nts) from the 3' terminal region of the intron. It should
be noted that 50% of the clones obtained from multiple experiments had the correct insert size but sequencing revealed
heterogeneous stretches of upstream and/or downstream exon sequences in addition to all or part of the intron (data not
shown). This was not seen in our examination of any of the other introns. Such species might be derived from degraded RNA
molecules, yet it is curious that they appear to be endogenously circularized. In this regard, evidence has been presented
for a potential cryptic 5' splice site within cox2 exon 1 of petunia (26), but that appeared to be discrete rather than
heterogeneous in nature.
When the cox2 intron cRT–PCR (RNA ligase-treated) products were cloned and sequenced to assess the presence of any
linear intron RNA species in the population, we found that six out of eight clones had short non-encoded polyA tracts (1–5
nt in length) and four of these were full-length introns whereas the other two lacked 3' nucleotide stretches. In addition,
one clone represented a full-length circularized intron species (end-to-end) and one simply lacked 3'terminal nucleotides.
The latter presumably reflects contribution from the pre-existing non-RNA-ligase treated population of endogenously
circularized forms. Notably in all cases, the precise 5' end of the cox2 intron was observed. The presence of a mixed
population of excised cox2 intron species was corroborated by direct sequencing of the cRT–PCR products (Figure 2D). It can
be seen that the sequence is homogeneous until the 5' end of the intron, but immediately past the junction there is a mixture
of A-rich sequences (Figure 2D, asterisk).
The nad1 intron 2, which contains a short tight dVI helix, showed yet a different set of features in its population of
excised intron molecules and none were lariats. Rather 5/9 clones showed a discrete 5'-ACAAAAC-3' insert at the junction of
full-length introns, and its relative abundance is evident in the directly sequenced population of RT–PCR products (Figure
2E, boxed nts). Of the remaining four clones, one represented a full-length circularized intron, while two clones also
contained 5'-ACTGT-3' which matches the first 5 nt of nad1 exon 2, and one clone lacked the 3' terminal 32 nt of the intron
(Table 1). Similar results were observed for the cRT–PCR analysis, in that 3/10 clones showed precise end-to-end
circularized species, and the others also contained either the non-encoded ACAAAAC block (4/10 clones) or ACTGT (3/10
clones). When the cRT–PCR products were examined directly, the sequence after the junction (Figure 2F) appears to be
somewhat more heterogeneous than what was seen for the in vivo population (Figure 2E).
Linear and circular excised introns in vivo supported by RNase H analysis
Although the above strategy provides information about the physical form of excised introns, it does not directly address
their relative abundance, especially since the yields of RT–PCR products may be influenced by reverse transcriptase being
less efficient at extending over the 2'–5' bond of a lariat species compared with the 3'–5' bond generated by ligation of
the ends of a previously linear molecule (cf. Figure 2C). To assess the relative levels of spliced lariat (and/or
circularized) versus linear physical forms of introns in vivo, wheat mitochondrial RNA was annealed with intron-specific
oligomers prior to incubation with RNase H and subsequent northern blot hybridization. In our analysis, we included nad7
intron 4 which we had previously shown to be predominantly lariat, but which also exhibited a minor population of fully
circularized (end-to-end) intron molecules (16,27), and as expected, the vast majority appear to be lariat rather than linear
(Figure 3A, black vs. open arrowhead). In the case of nad2 intron 4, when we used two annealing oligomers prior to RNase H
treatment, again the shift in size was consistent with a lariat (or circularized) excised intron (Figure 3B, black arrowhead
denoting 900 nt product detected by probe), when compared with the native 1.4 kb intron (Figure 3B, black dot). A minor
signal was detected at the position corresponding to a linear form (400 nt, Figure 3B, open arrowhead). Also, major products
of 4.8 kb derived from the linear precursor of 6 kb can be seen (Figure 3B) in addition to the precursor and intron species
which were not completely digested by RNase H.
RNase H analysis of the cox2 intron using two intron-specific primers revealed conversion of the 1.1 kb intron (Figure
3C, black dot) to products consistent with circular (or lariat) molecules of 600 nt (Figure 3C, black arrowhead) and linear
ones of 300 nt (Figure 3C, open arrowhead), as well as a shift in the 2.6 kb precursor to a prominent species of 800 nt. The
relative strengths of the hybridization signals for the linear and circular forms (Figure 3C, open versus black arrowheads)
suggest that both species are rather equally represented in the population, and this corroborates the results obtained with
RT–PCR and cRT–PCR (Figure 2 and Table 1).
In the case of wheat nad1 intron 2, more complex northern profiles are observed because there are multiple precursors due
to its co-transcription with upstream atp6 and rps13 genes, and this results in RNA species of 6, 4.6 and 3.6 kb (Figure 3D).
After RNase H treatment, when only one intron-specific primer was used (oligomer g), the signal for the excised intron of 1.4
kb shifts to a 1.2 kb product consistent with the presence of a linear species (Figure 3D, black dot versus open arrowhead).
This outcome was also seen when oligomer h was used, the shift being to 960 nt (Figure 3D, open arrowhead). In both cases,
the linear species appear relatively abundant. Lariat (or circular) species when converted to Y-shaped structures by RNase H,
would be expected to co-migrate with the native form under these electrophoretic conditions, but there is very little signal
at that position (Figure 3D, black arrowhead and black dot). The high molecular weight precursors shift to the expected sizes
of 2.6 and 2.3 kb. It is worth noting that the RNA ligase (cRT–PCR) experiments (cf. Figure 2A, lane 9) might be influenced
by protective structures at the intron termini, such as the 3' terminal hairpin of nad1 intron 2.
DISCUSSION
Our analysis of the cox2 and nad1 excised introns in RNA isolated from germinating wheat embryos has uncovered novel
physical forms which deviate from the classical lariat structure of group II introns. In contrast to introns such as nad2
intron 4 (this work), nad7 intron 3 (21,27) and nad4 intron 3 (N. Niknejad and L. Bonen unpublished data) which possess a
bulged adenosine branchpoint in dVI and are excised as lariats, our observations in the case of nad1 intron 2 and the cox2
intron support the presence of both circularized and linear molecules in vivo (Figure 4). Because the cox2 intron has an
apparently conventional branchpoint adenosine, it was unexpected that the lariat form was not predominant, but instead we
observed heterogeneous circular excised introns which lack various lengths at the 3'terminus. Moreover, the population
included full-length linear intron molecules with short, non-encoded polyA stretches.
Taken together, our data for the cox2 intron support a hydrolytic pathway of splicing generating linear molecules which
undergo either (i) polyadenylation at the 3' terminus [perhaps mediated by the bacterial-type tagging of transcripts for
degradation observed in plant organelles, Ref. (28)] or (ii) endogenous circularization after 3' exonucleolytic attack
(Figure 4B). Linear molecules have also previously been reported for the Streptococcus GBS intron, which possesses a
conventional branchpoint adenosine (29), and for the barley chloroplast trnV intron, which in addition is polyadenylated
(13); however, the wheat cox2 intron population includes endogenously circularized molecules as well. This might occur
through nucleophilic attack by 2'-OH groups at other positions within dV or dVI to generate aberrant lariats with longer
tails of varying lengths (Figure 4C). However, the cloned sequences correspond identically to the genomic sequence at such
putative novel branchpoints, whereas typically an adenosine branchpoint is mis-read as T by reverse transcriptase (13,17,22).
Alternatively, circularization might occur via an endogenous RNA ligase. It is also not excluded that the linear molecules
which we observed are derived from the debranching of lariats with subsequent protection from exonuclease attack.
The nad1 intron 2 appears to be present as linear and circular molecules in vivo, but they have quite different physical
features compared with the cox2 intron (Figure 4B and C). Notably, the excised introns fall into three categories: (i)
precisely full-length molecules, (ii) those which in addition have a discrete non-encoded ACAAAAC block and (iii) ones which
have the full-length intron plus ACTGT (which corresponds to the first 5 nt of the downstream exon). The latter could be
derived from incorrect 3' splice site selection [cf. Bacillus cereus, Ref. (30) and Bacillus anthracis, Ref. (31)] and such
alternative splicing would lead to defective mRNA molecules. Because exactly the same non-encoded nucleotide blocks were
consistently seen (and with or without RNA ligase treatment), these observations differ from those for the cox2 intron and do
not easily conform to a model of being generated by a polyA ragged tailing mechanism (Figure 4B). The endogenous linear
molecules, which comprise the majority based on RNase H analysis, appear resistant to 3' exonuclease attack (unlike the cox2
intron) and this may be related to the short tight dVI helix [reminiscent of the 3' termini of certain plant mitochondrial
mRNAs, Ref. (32)]. In all cases, the correct 5' splice site was seen both for nad1 intron 2 as well as the heterogeneous cox2
intron molecules. The apparent absence of exonuclease attack at the 5' end of putative linear molecules might relate to
protective RNA structures or close association with proteins.
Our nad1 intron 2 data suggest a combination of splicing mechanisms which exploit various internal or external
nucleophiles to explain the discrete non-encoded block (Figure 4B and C, hatched box). The latter would be analogous to group
I intron splicing where an external guanosine acts as the attacking nucleophile [reviewed in (33)]. The origin of the non-
encoded ACAAAAC block is unknown. If it is added to the 3' end of the intron, it seems unlikely to be mediated by tRNA
nucleotidyl transferase or an RNA-directed RNA polymerase, as was suggested for tobacco chloroplast ndhD mRNA processing
(34), since there is no obvious template. Other models consistent with a precisely full-length circularized intron (which we
observed at low levels for nad1 intron 2, nad7 intron 4 and the cox2 intron) include reverse splicing or a ‘spliced exon
reopening’ pathway [cf. yeast mitochondrial aI2 intron circles in vivo, Ref. (35)] whereby the 3'-OH of the liberated 5'-
exon acts as first attacking nucleophile in trans and then the hydroxyl group of the intron terminal pyrimidine attacks the
5' splice site (Figure 4C). Interestingly, in plant mitochondria, uncoupled splicing intermediates have been observed to be
relatively abundant in some cases (16) and thus might contribute to such a pathway.
The apparent loss of classical group II features of mitochondrial introns in the flowering plant lineage over
evolutionary time is supported by comparisons with homologues from early-diverging lineages of land plants. For example, nad1
intron 1 in Isoetes lacustris (22,36) and the cox2 intron in Acorus calamus (37) have conventional dVI structures. Moreover,
the dynamic nature of plant mitochondrial genomes [reviewed in (2)] can have impact on intron sequences, such as the genomic
rearrangements which have led to trans-splicing. It also appears that gene conversion (or swapping) events can occur between
group II core structures. For example in Figure 1, it can be seen that dV and part of dVI of nad1 intron 2 and nad2 intron 1
are identical. This also underscores the plasticity and tolerance that the splicing machinery must have in light of such
deviation from conventional structure.
The mixture of physical forms of excised introns observed in this study suggests that multiple biochemical mechanisms of
splicing occur in plant mitochondria, such as a hydrolytic pathway as well as unusual circularization events. Their relative
contributions appear to vary among introns and it will be interesting to learn if there are developmental effects, perhaps
due to differences in the availability of nuclear-encoded machinery especially early in germination when seeds leave dormancy
(16). In this regard, it is interesting that the steady state levels of excised introns show marked differences during
germination, typically being present at higher levels in embryos than in seedlings. The apparently ‘degenerate’ nature of
the folding of these group II introns raises the possibility that additional proteins and/or small RNAs have been recruited
for the splicing process in plant mitochondria.
SUPPLEMENTARY DATA
ACKNOWLEDGEMENTS
The authors thank C. Carrillo, N. Niknejad, S. Subramanian and J. Crosthwait for their earlier work on wheat and rice
mitochondrial introns in our laboratory. The authors also thank Dr R. Pandeya (Agriculture and Agri-Food Canada, Ottawa) for
kindly providing the wheat seeds used in this study. Financial support in the form of a Discovery Grant from the Natural
Sciences and Engineering Research Council of Canada (L.B.) and an Ontario Graduate Scholarship (J.L.P.T.) is gratefully
acknowledged. Funding to pay the Open Access publication charges for this article was provided by the Natural Sciences and
Engineering Research Council of Canada.
Conflict of interest statement. None declared.
REFERENCES
Bonen, L. and Vogel, J. (2001) The ins and outs of group II introns Trends Genet, . 17, 322–331
Knoop, V. (2004) The mitochondrial DNA of land plants: peculiarities in phylogenetic perspective Curr. Genet, . 46, 123–
139
Michel, F. and Ferat, J.L. (1995) Structure and activities of group II intron Annu. Rev. Biochem, . 64, 435–461
Lambowitz, A.M. and Zimmerly, S. (2004) Mobile group II introns Annu. Rev. Genet, . 38, 1–35 .
Sheveleva, E.V. and Hallick, R.B. (2004) Recent horizontal intron transfer to a chloroplast genome Nucleic Acids Res, .
32, 803–810
Odom, O.W., Shenkenberg, D.L., Garcia, J.A., Herrin, D.L. (2004) A horizontally acquired group II intron in the
chloroplast psbA gene of a psychrophilic Chlamydomonas: in vitro self-splicing and genetic evidence for maturase activity
RNA, 10, 1097–1107
Seetharaman, M., Eldho, N.V., Padgett, R.A., Dayie, K.T. (2006) Structure of a self-splicing group II intron catalytic
effector domain 5: parallels with spliceosomal U6 RNA RNA, 12, 235–247
Lehmann, K. and Schmidt, U. (2003) Group II introns: structure and catalytic versatility of large natural ribozymes Crit.
Rev. Biochem. Mol. Biol, . 38, 249–303.
Chu, V.T., Adamidi, C., Liu, Q., Perlman, P.S., Pyle, A.M. (2001) Control of branch-site choice by a group II intron EMBO
J, . 20, 6866–6876.
Toor, N., Hausner, G., Zimmerly, S. (2001) Coevolution of group II intron RNA structures with their intron-encoded
reverse transcriptases RNA, 7, 1142–1152
Podar, M., Chu, V.T., Pyle, A.M., Perlman, P.S. (1998) Group II intron splicing in vivo by first-step hydrolysis Nature,
391, 915–918
Boulanger, S.C., Faix, P.H., Yang, H., Zhuo, J., Franzen, J.S., Peebles, C.L., Perlman, P.S. (1996) Length changes in the
joining segment between domains 5 and 6 of a group II intron inhibit self-splicing and alter 3' splice site selection Mol.
Cell. Biol, . 16, 5896–5904
Vogel, J. and Borner, T. (2002) Lariat formation and a hydrolytic pathway in plant chloroplast group II intron splicing
EMBO J, . 21, 3794–3803
Farre, J.C. and Araya, A. (2002) RNA splicing in higher plant mitochondria: determination of functional elements in group
II intron from chimeric cox II gene in electroporated wheat Plant J, . 29, 203–213
Dombrovska, O. and Qiu, Y.L. (2004) Distribution of introns in the mitochondrial gene nad1 in land plants: phylogenetic
and molecular evolutionary implications Mol. Phylogenet. Evol, . 32, 246–263
Li-Pook-Than, J., Carrillo, C., Bonen, L. (2004) Variation in mitochondrial transcript profiles of protein-coding genes
during early germination and seedling development in wheat Curr. Genet, . 46, 374–380
Vogel, J., Hess, W.R., Borner, T. (1997) Precise branch point mapping and quantification of splicing intermediates
Nucleic Acids Res, . 25, 2030–2031
Vogel, J. and Hess, W.R. (2001) Complete 5' and 3' end maturation of group II intron-containing tRNA precursors RNA, 7,
285–292
Fedor, M.J. and Williamson, J.R. (2005) The catalytic diversity of RNAs Nature Rev. Mol. Cell Biol, . 6, 399–412
Zandueta-Criado, A. and Bock, R. (2004) Surprising features of plastid ndhD transcripts: addition of non-encoded
nucleotides and polysome association of mRNAs with an unedited start codon Nucleic Acids Res, . 32, 542–550
Murray, H.L., Mikheeva, S., Coljee, V.W., Turczyk, B.M., Donahue, W.F., Bar-Shalom, A., Jarrell, K.A. (2001) Excision of
group II introns as circles Mol. Cell, 8, 210–211 .
Malek, O. and Knoop, V. (1998) trans-splicing group II introns in plant mitochondria: the complete set of cis-arranged
homologs in fersn, fern allies, and a hornwort RNA, 4, 1599–1609
Albertazzi, F.J., Kudla, J., Bock, R. (1998) The cox2 locus of the primitive angiosperm plant Acorus calamus: molecular
structure, transcript processing and RNA editing Mol. Gen. Genet, . 259, 591–600
Farré, J.C. and Araya, A. (1999) The mat-r open reading frame is transcribed from a non-canonical promoter and contains
an internal promoter to co-transcribe exons nad1e and nad5III in wheat mitochondria Plant Mol. Biol, . 40, 959–967
Wissinger, B., Schuster, W., Brennicke, A. (1991) Trans splicing in Oenothera mitochondria: nad1 mNAs are edited in exon
and trans-splicing group II intron sequences Cell, 65, 473–482
Ogihara, Y., Yamazaki, Y., Murai, K., Kanno, A., Terachi, T., Shiina, T., Miyashita, N., Nasuda, S., Nakamura, C., Mori,
N., Takumi, S., Murata, M., Futo, S., Tsunewaki, K. (2005) Structural dynamics of cereal mitochondrial genomes as revealed by
complete nucleotide sequencing of the wheat mitochondrial genome Nucleic Acids Res, . 33, 6235–6250(Jennifer Li-Pook-Than and)