Studies of several group I ribozymes, but especially the intron from the large rRNA gene of Tetrahymena thermophila Tt.LSU, indicate that some domains are modular, and that the catalytic
Trang 1ribozyme from Chlamydomonas reinhardtii 23S rRNA
Tai-Chih Kuo1, Obed W Odom2and David L Herrin2
1 Department of Biochemistry, Tapei Medical University, Taiwan
2 Section of Molecular Cell and Developmental Biology and Institute for Cellular and Molecular Biology, University of Texas, Austin, TX, USA
Group I introns are cis-acting ribozymes whose
sub-strates (5¢ and 3¢ splice sites) are attached
intramolecu-larly These introns have conserved uridine and
guanosine nucleotides at the ends of the 5¢ exon and
intron segments, respectively Although sequence
con-servation of group I introns is poor, their folded forms
share a common core structure composed of two
stacked-helix domains (P5–P4–P6 and P7–P3–P8) [1,2]
Group I introns can be differentiated into five major
subgroups (IA, IB, IC, ID, and IE) with further
subdi-visions that depend on the presence of peripheral
domains that stabilize the core [3,4] Studies of several
group I ribozymes, but especially the intron from the large rRNA gene of Tetrahymena thermophila (Tt.LSU), indicate that some domains are modular, and that the catalytic site is buried inside the folded ribozyme [5–7] The tertiary structure is stabilized by domain–domain interactions, such as hydrogen bond-ing of loop–receptor pairs, base triples, and pseudo-knots [1,2]
The group I self-splicing pathway consists of two consecutive transesterification reactions with the acti-vated phosphodiesters at the splice sites First, the 3¢-OH of an exogenous guanosine nucleotide (GTP)
Keywords
Fe 2+ –EDTA; group I intron; Mn 2+ ; RNA
structure; RNA–metal interactions
Correspondence
D L Herrin, Section of Molecular Cell and
Developmental Biology, 1 University Station
A6700, University of Texas at Austin,
Austin, TX 78712, USA
Fax: +1 512 4713843
Tel: +1 512 4713843
E-mail: herrin@mail.utexas.edu
Website: http://www.biosci.utexas.edu/
MCDB/
(Received 9 February 2006, revised 3 April
2006, accepted 12 April 2006)
doi:10.1111/j.1742-4658.2006.05280.x
Group I intron ribozymes require cations for folding and catalysis, and the current literature indicates that a number of cations can promote folding, but only Mg2+and Mn2+support both processes However, some group I introns are active only with Mg2+, e.g three of the five group I introns in Chlamydomonas reinhardtii We have investigated one of these ribozymes,
an intron from the 23S LSU rRNA gene of Chlamydomonas reinhardtii (Cr.LSU), by determining if the inhibition by Mn2+ involves catalysis, folding, or both Kinetic analysis of guanosine-dependent cleavage by a Cr.LSU ribozyme, 23S.5DGb, that lacks the 3¢ exon and intron-terminal G shows that Mn2+ does not affect guanosine binding or catalysis, but instead promotes misfolding of the ribozyme Surprisingly, ribozyme mis-folding induced by Mn2+ is highly cooperative, with a Hill coefficient larger than that of native folding induced by Mg2+ At lower Mn2+ concentrations, metal inhibition is largely alleviated by the guanosine cosubstrate (GMP) The concentration dependence of guanosine cosub-strate-induced folding suggests that it functions by interacting with the G binding site, perhaps by displacing an inhibitory Mn2+ Because of these and other properties of Cr.LSU, the tertiary structure of the intron from 23S.5DGb was examined using Fe2+-EDTA cleavage The ground-state structure shows evidence of an unusually open ribozyme core: the catalytic P3–P7 domain and the nucleotides that connect it to the P4–P5–P6 domain are exposed to solvent The implications of this structure for the in vitro and in vivo properties of this intron ribozyme are discussed
Abbreviations
Cr.LSU, intron from the 23S LSU rRNA gene of Chlamydomonas reinhardtii; oligo, oligodeoxynucleotide; Tt.LSU, intron from the large rRNA gene of Tetrahymena thermophila.
Trang 2attacks the 5¢ splice site (G-dependent cleavage),
gener-ating 5¢ exon and intron-3¢ exon intermediates Then,
the 3¢-OH of the 5¢ exon attacks the 3¢ splice site,
forming ligated exons and a free intron [9] The
liber-ated intron can react with itself, forming a circular
RNA [9], or with another RNA [10], via the
3¢-ter-minal G (XG) of the intron; it can attack a
phosphodi-ester that becomes properly positioned in the catalytic
center [9,10]
Owing to the complexity of the ribozyme reactions
and the polyanionic nature of RNA, the catalytic
chem-istry and folding of group I ribozymes (and other large
ribozymes) require divalent metals [11,12] Whereas
only Mg2+and Mn2+are able to support the chemistry
of the Tetrahymena ribozyme [13], several divalent
cati-ons (e.g Mg2+, Mn2+, Ca2+, Sr2+, and Ba2+) [14],
and even monovalent cations [15], are able to promote
the formation of a native, or native-like, structure With
divalent and monovalent salts as the only aids to RNA
folding, however, the formation of alternative,
nonpro-ductive base pairs can trap a fraction of a large
ribo-zyme in inactive conformations [16–18] The conversion
of these forms into an active ribozyme is sometimes
hampered, ironically, by native domain–domain
interac-tions or high Mg2+ concentrations [19] Thus, in vivo,
the folding of most large ribozymes is probably assisted
by proteins In a few cases, it has been shown that in
the presence of the proper protein, group I ribozymes
that would otherwise be inactive, or become active only
at high temperatures and high Mg2+ concentrations,
perform catalysis in vitro under mild conditions [20,21]
We have been studying the group I ribozyme,
Cr.LSU, from the chloroplast 23S (LSU) rRNA gene
of the green alga Chlamydomonas reinhardtii [22–24]
Splicing of Cr.LSU in vivo is required for ribosome
formation, and could be a limiting step in ribosome
biogenesis, as it is one of the slowest steps in rRNA
maturation [24] This subgroup IA3 intron self-splices
efficiently in vitro, but requires higher Mg2+
concen-trations than the model intron, Tt.LSU, and it is more
sensitive to nucleotide substitutions in the core [23,25]
Kinetic analysis indicated that a Cr.LSU pre-RNA
containing the full-length intron and relatively long
exon sequences tends to misfold in vitro, although the
active fraction self-spliced rapidly [18] Self-splicing of
Cr.LSU occurs only with Mg2+, and is inhibited by
equivalent concentrations of Ca2+ or Mn2+ [26] The
Mn2+ inhibition was unexpected, because Mn2+ is
similar to Mg2+ [27,28], and has been shown to
sup-port the activity of group I and other large and small
ribozymes [27,29,30] Thus, this group I intron exhibits
several properties that distinguish it from the more
well-studied Tt.LSU and phage introns
We wished to know whether Mn2+ inhibits the for-mation of active Cr.LSU or whether it interferes with catalysis, and have addressed this question using ribo-zyme kinetics We also wanted to probe the folding and tertiary structure of the ribozyme using Fe2+ -EDTA, which promotes cleavage of the sugar–phos-phate backbone, and can determine, for example, if the active site of Cr.LSU is internalized like those of other group I ribozymes [7,31] The wild-type Cr.LSU intron was unsuitable for this, because of its large size (888 nucleotides) Moreover, high concentrations of
NH4+ and Mg2+ were required for efficient self-spli-cing of the large 23S.1 precursor [23] Hence, a smaller RNA, 23S.5 (448 nucleotides), in which the intron was shortened by replacing the long P6 extension (which encodes the I-CreI endonuclease [32]) with a short stem–loop, and the exons were reduced, was generated This pre-RNA self-splices efficiently without monova-lent salt, and approximately 85% of the RNA is of the same kinetic competence (kcat¼ 1 min)1 and K1/2G¼
26 lm) [18] For this study, we have taken the 23S.5 pre-RNA and generated a ribozyme, 23S.5DGb, that performs GMP cleavage at the 5¢ splice site, but not exon ligation or intron circularization, as it lacks the 3¢ exon and the XG This RNA was used to assay ribozyme activity, and to generate end-labeled RNA for structural probing
Results
Inhibition of self-splicing and G-dependent cleavage by Mn2+
In splicing reactions with 23S.5 pre-RNA, substituting part (> 1⁄ 3) of the Mg2+ with Mn2+ reduced the amount of products, which were undetectable when
Mn2+ was the only divalent cation (not shown [26]) Varying the Mn2+ concentration (0.1–50 mm), pH (5.5–7.5), monovalent salt, temperature (37 or 47C), and reaction time (0.25–60 min) also did not yield any splicing products (data not shown) Mn2+inhibits self-splicing of the 23S.3 and 23S.4 pre-RNAs, which have different lengths of 5¢ exon [23], and it inhibits a trans-reaction [10] that involves the free intron reacting with 5.8S rRNA (not shown) Together, these data sugges-ted that inhibition by Mn2+ probably involved the core ribozyme, and not the intron open reading frame (ORF) or exon sequences
23S.5DGb pre-RNA is a truncated version of 23S.5 that terminates 3 nucleotides before the end of the intron (Fig 1A; see Fig 5D for the intron sequence and structure) Core catalytic activity is preserved, however, in the form of G-dependent cleavage at the
Trang 35¢ splice site Incubation of 23S.5DGb with saturating
Mg2+ (25 mm) and GMP (150 lm) produces the InDGband 5¢ exon (not shown) molecules as expected (Fig 1B) In the presence of 10 mm Mn2+, however, less of the pre-RNA reacts (compare Fig 1B,C) Thus, the inhibition of 23S.5 self-splicing by Mn2+is recapit-ulated by the G-dependent cleavage of 23S.5DGb pre-RNA Mixed-metal titrations over a range of total metal concentrations showed that the ratio of the two metals is somewhat more important than the absolute concentrations; a ratio of Mn2+⁄ Mg2+ of approxi-mately 1 : 2 or higher inhibited G-dependent cleavage
of 23S.5DGb RNA (and self-splicing of 23S.5, not shown)
Quantification of time-course reactions similar to those in Fig 1B,C, except at two different GMP con-centrations (Fig 1D), show that approximately 85%
of the 23S.5DGbRNA is kinetically homogeneous and highly active (kobs approximately 0.9 min)1 at 150 lm GMP) The remaining fraction (approximately 15%) is relatively inactive, reacting 20–30 times more slowly (kobs¼ 0.032 min)1 at 150 lm GMP, and 0.017 min)1
at 20 lm GMP) It can also be inferred from Fig 1D that the inactive RNA fraction increases substantially when 10 mm Mn2+ is added, from 15% to 32% at
150 lm GMP, or 50% at 20 lm GMP The inverse is true for the active fraction, which decreased from 85%
to 68% and 50%, respectively The observed rate of G-dependent cleavage by the active fraction is not sub-stantially affected by Mn2+: the kobs at 150 lm GMP
is 0.96 min)1 with Mn2+ and 0.87 min)1 without it, and at 20 lm GMP, the kobsis 0.49 min)1 with Mn2+ and 0.55 min)1 without it We conclude that Mn2+ increases the proportion of 23S.5DGbpre-RNA that is inactive, whereas GMP increases the proportion that is active
An extensive kinetic analysis was performed at 5–300 lm GMP and 10 or 15 mm Mn2+ in the pres-ence of 25 mm Mg2+ Figure 2A shows that Mn2+ has little or no effect on the KG1/2or kcatfor the active fraction of the ribozyme However, as Fig 2B shows quite dramatically, the metal decreases the size of this fraction It should be noted that the proportion of act-ive ribozyme without Mn2+ is approximately 88% at all GMP concentrations tested In the presence of
10 mm Mn2+, the maximum size of this fraction is 66% (> 100 lm GMP), and it decreases dramatically
at GMP concentrations < 100 lm (Fig 2B) At 15 mm
Mn2+, the percentage of active ribozyme is even lower (approximately 30% at > 25 lm GMP) and it decrea-ses further at GMP < 20 lm These data extend the above result, and support the conclusion that Mn2+ affects mainly the correct folding of the ribozyme The
Ct 0 3 7 1 1.5 3 5 10 25 60
0 3 6 1 1.5 3 5 10 20 40 60
Fig 1 Mn 2+ inhibits the activity of the 23S.5DGb ribozyme.
(A) Schematic diagram of 23S.5DGbpre-RNA and the reaction being
assayed 23S.5DG b pre-RNA contains a partial 5¢ exon (rectangle)
and a shortened Cr.LSU intron (line), which lacks the large P6
extension and the last three nucleotides (AU XG) of the intron The
sizes of the intron (InDG b ) and 5¢ exon (5E) are indicated The arrow
indicates cleavage of the pre-RNA at the 5¢ splice site by GMPG*.
(B, C) G-dependent cleavage reactions with 0 m M (B) or 10 m M (C)
MnCl 2 The reactions with 32 P-labeled 23S.5DG b pre-RNA (Pre)
included 25 m M MgCl 2 and 150 l M GMP The large product, InDG b ,
was separated on a denaturing gel and phosphorimaged (D)
Quan-tification of 23S.5DG b pre-RNA decay in the presence (10 m M ) or
absence (0 m M ) of Mn2+, and either 20 or 150 l M GMP The GMP
cleavage reactions were performed as in (B) and (C), except at two
different concentrations of GMP The %23S.5DGbpre-RNA
remain-ing was measured, and the data fitted to an equation for
two-phase, exponential decay kinetics (see Experimental procedures).
Trang 4results also show that most of the inhibition by 10 mm
Mn2+is reversed by saturating GMP (Fig 2B)
More-over, the fact that this GMP activation curve is similar
to the GMP cleavage plot (Fig 2A) indicates that
GMP is promoting ribozyme folding via the G binding
site in P7
The rate of G-dependent cleavage by the inactive
fraction that forms in Mg2+ increases slowly with
GMP concentration (K1/2G¼ 60 lm and kcat¼ 0.035 min)1, not shown) However, the inactive fraction
in Mn2+(10 mm in Fig 2B) reacts much more slowly (approximately 0.007 min)1) and independently of the GMP concentration (not shown) This result suggests that Mn2+ induces a distinctive slow-reacting fraction that must go through a rate-limiting conformational change before it can bind GMP and catalyze cleavage
Model for the effect of Mn2+on the 23S.5DGb ribozyme
Since Mn2+ does not substantially affect the kinetic parameters for the active ribozyme, but instead reduces the size of this fraction, the following scheme (Scheme 1) is proposed to describe the inhibition of G-dependent cleavage by the metal ion
Uþ nMg2þ+( pre-RNAnMg2þ
active
Uþ mMn2þ+( pre-RNAmMn2þ
inactive
Scheme 1
U is unfolded 23S.5DGbpre-RNA, and the binding of
a minimum of n Mg2+ ions leads to formation of the active complex, whereas binding of a minimum of m
Mn2+ ions forms the inactive complex The sizes of the active and inactive fractions are the result of com-petitive metal binding to RNA The values of n and m are estimated from Hill analysis of G-dependent clea-vage of 23S.5DGb It should be noted that Scheme 1 indicates only the initial and final states of the pre-RNA; it does not invoke or rule out any misfolded intermediates that might form
To determine n, G-dependent cleavage of 23S.5DGb was analyzed at varying MgCl2 concentrations (0–50 mm) and either 0 or 12 mm MnCl2 (plus satur-ating GMP) Figure 3A shows that in the presence
of Mn2+, a much higher concentration of Mg2+ is required to form the same amount of cleavage product (InDGb) A quantitative analysis of similar experiments (Fig 3B), but using 0–100 mm Mg2+ and several fixed
Mn2+ concentrations (0, 7, 12 and 17 mm), reveals that cleavage increases cooperatively with increasing
Mg2+, in the absence or presence of Mn2+ The mid-point of the Mg2+ titration curve in the absence of
Mn2+ is approximately 4.5 mm, and nearly full activity is reached by 10 mm Mg2+ Hill analysis of the data gives n-values of 2.6, 2.2, 2.7 and 2.7, for reactions in 0, 7, 12 and 17 mm Mn2+, respectively These results indicate that formation of the active RNA–Mg2+ complex involves the binding of at least three Mg2+ions by the ribozyme The data also show that Mg2+ can completely block the inhibition caused
by Mn2+
Fig 2 Ribozyme activity at varying GMP and fixed Mn 2+
concentra-tions The G-dependent cleavage reactions were performed at
dif-ferent concentrations of GMP (0–300 l M ) and Mn 2+ (0, 10 or
15 m M ) in the presence of 25 m M MgCl2 The reactions were
ana-lyzed as described in Experimental procedures, and the observed
rate constants (A) and percentages (B) of active ribozyme were
plotted versus GMP concentration In (A), the line was fitted using
the 0 m M Mn 2+ data In (B), the k cat values are 1.1, 1.0 and
1.0 min)1, respectively, for the reactions at 0, 10 and 15 m M Mn2+,
and the corresponding K G
1/2 values are 22, 24 and 21 l M , respect-ively.
Trang 5To determine the minimal number of Mn2+
involved in forming the inactive RNA–metal
com-plex, G-dependent cleavage reactions were performed
at varying Mn2+ and fixed Mg2+ concentrations
Figure 4A shows representative gels of reactions that
were performed at varying (0–30 mm) MnCl2
con-centrations and either 15 or 25 mm MgCl2 The
GMP-dependent cleavage decreases sharply above 5 and 7 mm MnCl2, respectively Quantitative analysis (Fig 4B) gives a Hill value (m) of 5.7 for the experi-ments with 15 and 25 mm MgCl2 Thus, formation
of the inactive ribozyme is highly cooperative The data also suggest that binding of a minimum of six
Mn2+ ions is involved in the misfolding that forms the inactive ribozyme
Structure of the Cr.LSU intron in the 23S.5DGb pre-RNA
The unusual metal specificity, as well as other atypical features of Cr.LSU (see Discussion), led us to study the global tertiary structure of the intron using
Fig 3 Mg 2+ dependence of ribozyme activity at fixed Mn 2+
con-centrations (A) G-dependent cleavage of 23S.5DGb pre-RNA at
varying Mg2+ concentration, and 0 m M (top) or 12 m M (bottom)
MnCl2; the reactions also contained 150 l M GMP, and were
incuba-ted for 40 s They were separaincuba-ted on a denaturing polyacrylamide
gel, which was phosphorimaged (B) Mg 2+ concentration curves at
fixed Mn2+ concentrations G-dependent cleavage of 23S.5DG b
was performed as in (A), except for using the indicated Mn 2+
con-centrations The cleavage product (InDGb) was quantified, and
expressed as a percentage of total RNA [Relative InDG b (%)] The
data were curve-fitted to obtain Hill coefficients as described in
Experimental procedures.
Fig 4 Mn 2+ dependence of ribozyme inhibition at fixed Mg 2+ con-centrations G-dependent cleavage of 23S.5DGbpre-RNA was per-formed with the indicated concentrations of MnCl 2 (0–30 m M ), 10
or 25 m M MgCl2, and 150 l M GMP for 40 s The reactions were analyzed as in Fig 3.
Trang 6hydroxyl radical cleavage This analysis was carried
out by incubating end-labeled, Mg2+-folded intron
(InDGb) with Fe2+-EDTA [7,8] It should be
empha-sized that the RNA structure revealed by Fe2+-EDTA
is an averaged image of the RNA molecules in
solu-tion However, since the kinetic data indicate that 85–
90% of 23S.5DGbpre-RNA is functionally similar, we
assume that the predominant signal in the protection
pattern is from active ribozyme
Figure 5A shows an Fe2+-EDTA cleavage analysis
performed at 0–25 mm Mg2+ In the absence of Mg2+
(lanes 1 and 12), cleavage occurs throughout the
mole-cule; this profile was defined as background As the
Mg2+ concentration increased, different regions of the
RNA became either more sensitive to, or more
protec-ted from, cleavage or cleavage was relatively
unchanged Figure 5A shows that the overall cleavage
profile changes gradually with Mg2+, a trend that
appears to hold for most nucleotide positions (solid
and open bars in Fig 5A, lanes 2–11) It should be
noted, however, that the degree of change at some
nucleotides, such as the protection of P3¢ (Fig 5A),
appears to be greater at the higher concentrations of
Mg2+(> 5 or 6 mm), suggesting that this region may
have a higher Mg2+ requirement for folding There is
also a general increase in the amplitudes of the
clea-vage⁄ protection peaks at the highest (25 mm) Mg2+
concentration tested A similar result is apparent in
some of the Tt.LSU L-21 ScaI protection data [33],
and may reflect a greater overall stability of the RNA
at high Mg2+
Since 23S.5DGb pre-RNA is fully active at 25 mm
Mg2+, we inferred the native structure of the ribozyme
by comparing the cleavage profiles at 0 and 25 mm
Mg2+ (Fig 5B) For those nucleotides that showed
differential cleavage, the extent of the difference was
2–4-fold at most positions; Fig 5C is the difference profile (for 0 and 25 mm Mg2+) plotted by nucleotide position It should be noted that similar results were obtained when cleavage time, or concentration of
Fe2+-EDTA, was varied over a four-fold range, or when dithiothreitol was replaced by ascorbate and hydrogen peroxide (not shown)
To better visualize the locations of exposed and pro-tected regions of the InDGb intron, data from the difference plot were converted to a color palette and mapped onto the proposed secondary structure (Fig 5D) Focusing on the critical P7–P3–P8 domain, the 5¢ strand of P7, which includes the G-binding site,
is cleaved, whereas the 3¢ strand is protected The 3¢ strand of P3 is also protected, but the 5¢ strand is neither protected nor particularly exposed The part of J8⁄ 7 proximal to P8 is protected, whereas the residues close to P7 are strongly cleaved; also, P8 itself is pro-tected, but L8 is not For the P9 subdomains, most of P9 is protected, but most of P9.1 is neutral The 5¢ strand of P9.0 is protected despite the absence of the 3¢ strand For the idiosyncratic P7.1 and P7.2 domains, the helices are weakly protected, except for the 3¢ strand of P7.2, which is strongly protected, and both loops are cleaved
For the other major stacked-helix domain, P5–P4–P6,
as well as P2, the extent of protection or exposure was mostly neutral or relatively weak compared with the P7–P3–P8 domain However, P5, P6 and minor parts of P5a and P6a are protected Also, J4⁄ 5 is weakly protec-ted, but J5⁄ 4 is weakly cleaved Of particular signifi-cance is the observed cleavage of J6⁄ 7 and the neutrality
of J3⁄ 4; these joining segments link the two major domains and form base triples with P4 and P6
The overall Fe2+-EDTA cleavage⁄ protection pat-tern for this group IA3 intron (InDGb) has many
Fig 5 Hydroxyl radical cleavage with Fe 2+ -EDTA of the intron ribozyme (A) Fe 2+ -EDTA cleavage and protection pattern of 5¢ end-labeled InDGbRNA as a function of Mg 2+ concentration The InDGbRNA, labeled at its 5¢ end through G-dependent cleavage of 23S.5DG b pre-RNA, was incubated with Mg2+(final concentration indicated above lanes 2–12) and then cleaved with Fe2+-EDTA The reactions were resolved
on gels of 5–12% polyacrylamide with ladders and other markers The locations of structural elements (P5, P6, etc.) are marked to the left
of the gel image, and RNA sizes to the right Rectangular bars indicate areas where cleavage is enhanced (filled rectangles), or reduced (open rectangles), with increasing Mg2+ Other lanes are: S, starting RNA; Mn, S RNA cleaved with Mn2+at GAAA sequences (nucleotides
56, 257 and 323); A, G, A ⁄ U, and C, enzymatic sequence ladder of S RNA; OH, partial alkaline hydrolysis of S RNA; and M, 5¢ end-labeled DNA size markers The composite figure is from 8% (upper) and 12% (lower) polyacrylamide gels (B) Phosphorimager scans of Fe 2+ -EDTA cleavage of InDG b RNA with 0 and 25 m M Mg 2+ The RNA was probed as in (A) and the samples resolved on a 12% polyacrylamide gel, which was phosphorimaged (C) Difference plot of Fe2+-EDTA cleavage at 0 and 25 m M Mg2+ Regions of cleavage are positive (> 0), and regions of protection are negative (< 0) The plot was compiled from cleavage profiles obtained on a series of 5–12% polyacrylamide gels (D) Protection and cleavage patterns mapped onto the predicted secondary structure; the data are from (C) Protected nucleotides are shades of red and orange, whereas cleaved nucleotides are blue shades; a color bar is given on the bottom right The analysis under these conditions was repeated twice with similar results The accuracy of the protection data at the nucleotide level is ± 0 for residues £ 180, ± 1 for residues 181–270, and ± 2 for residues 271–342 The cleavage ⁄ protection patterns of nucleotides 1–14 and 343–378 were insufficient relative to background, and are not indicated The red arrowheads indicate sites of cleavage by Mn2+-GAAA ribozymes [38].
Trang 7similarities to the introns from other subgroups [7,8]
(see also below) However, it is atypical because of the
relative lack of protection of the catalytic domain and
the junction nucleotides that link it to the other major domain We attempted to probe the InDGb RNA with
Fe2+-EDTA in the presence of Mn2+, but cleavage was
Trang 8too poor to obtain a clear result This is most likely due
to the similar affinities of EDTA for Mn2+and Fe2+,
and the fact that Mn2+ is in large excess in these
reactions [34]
Discussion
Kinetic analysis of Mn2+inhibition
The group I ribozyme literature indicates that divalent
metals are required for tertiary folding and catalysis,
and that Mg2+ or Mn2+ can satisfy these functions
However, the activity of some group I introns,
inclu-ding three of the five introns in Chlamydomonas
rein-hardtii (Cr.psbA2, Cr.psbA3, and Cr.LSU) is inhibited
by Mn2+ [26] We wanted to determine for the
Cr.LSUintron if this metal specificity is due to a more
stringent requirement for catalysis, or folding, or both
The data show that Mn2+ does not have a significant
effect on the kinetic parameters (K1/2Gand kcat) of the
23S.5DGb ribozyme Thus, Mn2+ does not inhibit
binding of the guanosine nucleotide, and nor does it
suppress cleavage chemistry, suggesting that, as in the
case of Tt.LSU and some other group I ribozymes,
Mn2+ can support catalysis by Cr.LSU The data do
indicate, however, that Mn2+inhibits formation of the
active ribozyme, presumably by causing misfolding
This result was unexpected, because of the extensive
work with the Tt.LSU ribozyme indicating that the
metal requirement for ribozyme folding is less
restrict-ive than that for catalysis [14,15]
It is also surprising that Mn2+-induced misfolding of
this ribozyme is highly cooperative, with a Hill
coeffi-cient of approximately 6 In fact, it is more cooperative
than Mg2+ induction of active ribozyme (Hill
coeffi-cient of approximately 3) High cooperativity is
consis-tent with the great stability of the RNA formed in
Mn2+, which converts very slowly to a form capable of
binding GMP and reacting The higher Hill coefficient
could also indicate that Mn2+ binds to additional
(three) sites on the ribozyme that do not bind Mg2+
However, the fact that Mg2+ can completely block
Mn2+ inhibition would suggest that they bind to the
same sites Hence, it may be that they bind to the same
basic locations, but that Mn2+ binds with slightly
dif-ferent configurations at some key sites, resulting in
inhibition In this respect, it should be noted that,
based on mixed-metal titrations of 23S.5 self-splicing,
Mn2+can functionally replace Mg2+at some sites [26]
Mn2+ inhibition of 23S.5DGb is partially alleviated
by GMP, especially at lower metal concentrations
The GMP concentration dependence of this effect
indicates that the nucleotide is acting via the G
binding site in P7 An obvious explanation for this result is that binding of GMP prevents an inhibi-tory Mn2+ from binding to this site Interestingly, 2¢-dGTP inhibits Pb2+ cleavage of the T4-td intron
at the bulge nucleotide in P7 [30], which is very close
to the bound XG in recent crystal structures [35–37] With Cr.LSU, we did not see a similar specific clea-vage with Pb2+ or other metals [38], so the same experiment could not be performed It should be noted, however, that metal-dependent cleavage reveals only a small fraction of metal-binding sites in RNA [29] An alternative explanation for GMP-induced folding of the ribozyme in Mn2+ is that binding of the cosubstrate to its site induces a conformational change in the RNA that inhibits Mn2+ binding at another inhibitory site(s) It may be relevant that
in vitro-evolved Tt.LSU ribozymes capable of using
Ca2+ as sole divalent cation had several nucleotide substitutions clustered about the G binding site and the triple base pairs at the P4–P6 junction [39] We tried unsuccessfully to select variants of Cr.LSU act-ive with Mn2+ from pools of mutants generated by error-prone PCR (T.-C Kuo & D L Herrin, unpub-lished results) Based on the high Hill coefficient reported here, however, the failure of that experiment may be attributed to the difficulty in overcoming the relatively high number of inhibitory Mn2+-binding sites in Cr.LSU
We previously identified six Mn2+-binding sites in Cr.LSU based on site-specific cleavages at pH > 7 [38] These sites all contain the sequence GAAA, and cleavage occurs between G and A The cleavage effi-ciency, however, varied between sites, and correlated with the predicted secondary structure Also, the addi-tion of sufficient Mg2+ to induce self-splicing did not affect the Mn2+cleavage rates at the various sites, sug-gesting that most of the intron’s secondary structure forms correctly in Mn2+ The experiments herein were performed at pH 6 (self-splicing of Cr.LSU is efficient
at pH 6–9 [23]), and for shorter times to limit the
Mn2+-induced cleavages Thus, these data would sug-gest that Mn2+ is probably inhibiting tertiary folding
of Cr.LSU In this respect, we note the published evi-dence [40] that correct tertiary folding of the Azoarcus intron is specific for Mg2+ However, it is possible that subtle but important changes in secondary structure could also be involved For example, the crystal struc-tures of yeast tRNAPhe in Mg2+, or in a mixture of
Mg2+ and Mn2+, are quite similar but not identical; residue D16 (D loop) forms a base pair with U59 (TWC loop) only in the latter condition [41]
It is unlikely that the three remaining GAAA-Mn2+ cleavage sites in the 23S.5DGb ribozyme (three are
Trang 9deleted) are principal targets of Mn2+inhibition, since
Mn2+ cleavage is not cooperative [38]; however, one
of them could be a site of Mn2+ inhibition The best
candidate would probably be J4⁄ 5, which is strongly
cleaved with Mn2+ [38], and is involved in substrate
recognition during splicing [35,36] Changing the
GAAA in J4⁄ 5 to GACA blocked Mn2+ cleavage and
doubled the Mg2+ requirement for Cr.LSU
self-spli-cing (T.-C Kuo, S P Holloway & D L Herrin,
unpublished results), suggesting that it might be an
important metal-binding site Evidence for a functional
metal interaction with the J4⁄ 5-GAAA region of an
Anabaenaintron was reported recently [42]
Structural and functional idiosyncrasies of the 23S.5DGbintron ribozyme
To help us understand the noncanonical properties of Cr.LSU, the intron’s (InDGb) structure was analyzed using Fe2+-EDTA, which has been used extensively to view tertiary-folded group I ribozymes The chelated iron generates hydroxyl radicals that cleave riboses, unless they are protected by RNA–RNA interactions [7] Figure 6 compares the Cr.LSU protection pattern, which is the first for a subgroup IA3 intron, with five ribozymes from other subgroups: L-21 ScaI of Tt.LSU [7], T4.nrdD [8], T4.td [8], Sc.bI5 [43], and Azoarcus
Fig 6 Comparison of the Fe 2+ -EDTA protection patterns of group I ribozymes Residues protected from Fe 2+ -EDTA cleavage are indicated
by filled squares with white letters The data for Cr.LSU (A) are from Fig 5; for Tt.LSU (B) from [7]; for T 4 sunY (C) and T 4 td (D) from [8]; for bI5 (E) from [43]; and for Azoarcus pre-tRNAIle(F) from [44] Domain–domain interactions in introns B–F are indicated by dashed lines In (A), the lightly shaded nucleotides in the L9.1 and P7.1 loops may form a novel base-pairing interaction; the gray dashed line between L2 and P8 also indicates a possible interaction Solid arrows indicate sites of Mn 2+ cleavage in Cr.LSU and T4-td; open arrows are sites of Pb 2+ cleavage in Tt.LSU, T4.td, and T4.nrdD; and arrowheads in Azoarcus indicate the 5¢ and 3¢ splice sites.
Trang 10pre-tRNAIle [44] All of these introns can self-splice,
although the Mg2+requirement for bI5 is higher than
that for the others Focusing on the catalytic domain
(P8–P3–P7), the P3 3¢-strand is the only element that is
uniformly protected in all ribozymes, although they all
have parts of P7, J8⁄ 7 and P9.0 protected The
Cr.LSU pattern is distinct in that the first three
nucle-otides of P7, and the 3¢-half of J8 ⁄ 7, are not protected
The first three nucleotides of P7 are part of the G
binding site, as are the terminal nucleotides of J6⁄ 7
and J8⁄ 7 [35–37]; the latter nucleotides are also not
protected in the InDGb RNA It should be noted that
the three terminal nucleotides of Cr.LSU, including
the XG, are not present in the InDGb RNA, which
presumably also lacks P9.0 (Fig 6A) It is possible this
has an effect on the protection pattern However, the
kinetic parameters (kcat and K1/2G) for G-dependent
cleavage by 23S.5DGb are very similar to those
obtained for the first step of self-splicing by 23S.5 [18],
indicating that P9.0 is not important for core ribozyme
activity (it is probably important for the second step
of splicing [45]) It is also noteworthy that the
3¢-ter-minal nucleotides were also absent from the Tt.LSU
ribozyme mapped with Fe2+-EDTA [7] To conclude,
the G binding site and flanking nucleotides of Cr.LSU
are more accessible to solvent (i.e less internalized)
than are those of other group I ribozymes studied to
date It is also intriguing that the kinetic data implicate
the G binding site as playing an important role in
Mn2+inhibition
Why is the active site less internalized in the
InDGb RNA? The lack of protection of J6⁄ 7 and
J3⁄ 4, which are involved in triple base pairs with P4
and P6 [3,35–37], and the relatively weak overall
protection of the P4–P5–P6 domain (Fig 5C),
indi-cate that this domain is not tightly packed against
the catalytic domain Analysis of the predicted
sec-ondary structure suggests that Cr.LSU may be
some-what deficient in interactions between these domains
For example, it seems to lack the L9· P5
interac-tion found in the other ribozymes (Fig 6) This
interaction is primarily between the L9 tetraloop and
the second and third nucleotides of P5 [3]; however,
in Cr.LSU, P5 is only two base pairs Although
deletion of the P4–P5–P6 domain from the T4.td
intron suggests that it is not essential [46], disruption
of P6 in Cr.LSU obliterated self-splicing, and point
mutations in P4 strongly decreased splicing in vitro
and in vivo, indicating that the P5–P4–P6 domain is
important for Cr.LSU [25] The Li et al study [25]
also indicates that Cr.LSU is more sensitive to single
nucleotide substitutions in the core than Tt.LSU
or T4.td, which is consistent with fewer tertiary
interactions Li et al [25] also isolated nuclear gene suppressors of the P4 mutations; thus, based on these data, it is reasonable to speculate that one or more
of these suppressors promote interaction between the two major domains
There are other functional differences between Cr.LSU and the Tt.LSU and phage T4 introns besides
Mn2+ inhibition that may reflect the distinctive struc-tures It was shown that Pb2+promotes specific cleav-ages in the P7 and J8⁄ 7 regions of Tt.LSU and the phage introns (Fig 6B–D) in the presence of Mg2+ [30] However, we did not observe similar specific cleavages in Cr.LSU with Pb2+(or other cations) [38] Cr.LSU is also more resistant to inhibition by polycat-ionic aminoglycoside antibiotics, such as neomycin B [47], requiring 25–50-fold higher concentrations to inhibit self-splicing in vitro by 50% (T.-C Kuo,
Y Bao & D L Herrin, unpublished results) It is noteworthy that neomycin inhibits the T4.td intron by binding at the G binding site and displacing one or two critical Mg2+ ions [48] We propose that the lack
of Pb2+cleavage, and the apparent absence of a high-affinity, inhibitory site for neomycin, could be conse-quences of the more open structure of Cr.LSU, which should present a less electronegative environment at the active site It would be interesting to know if other group I introns that are inactive with Mn2+[24] have properties similar to those of Cr.LSU, including the more stringent metal requirement for ribozyme forma-tion
The tertiary-folded structure of the InDGb RNA in
Mn2+ would probably have been instructive, but unfortunately, the Fe2+-EDTA cleavage pattern was strongly inhibited by Mn2+ under these conditions, presumably due to the similar affinities of EDTA for
Mn2+ and Fe2+ (Kd approximately 10)14.1) [34] It may be possible to accomplish this with synchrotron X-ray [49] or peroxynitrous [50] cleavage, although the former requires special equipment, and the latter rea-gent is not as easy to use as Fe2+-EDTA
An intriguing, though speculative, implication of the relatively open ground-state structure of InDGbis that the ribozyme might undergo a transient internalization
of P7, J8⁄ 7 and J6 ⁄ 7 after binding GMP to start the reaction Binding of the guanosine nucleotide by the Tt.LSU ribozyme is very slow, and is believed to induce a local rearrangement of the G binding site [51]; these authors also argued that an incompletely preformed G binding site could promote specificity The posited conformational rearrangement of Cr.LSU would seem to be more extensive, but if it does happen and is necessary, then that dynamic change could be a key step that is inhibited by Mn2+