Fate of the junction phosphate in alternating forward and reverse self-splicing reactions of group II intron RNA

J. Mol. Biol. (1991) 222, 145-154

Fate of the Junction Phosphate in Alternating Forward and Reverse Self-splicing Reactions of Group II Intron RNA Manfred W. Miiller, Paul Stocker, Martin Hetzer and Rudolf J. Schweyen Institut

fiir

Mikrobiologie und Genetik, Vienna, Austria

(Received 20 May

Universitiit

1991; accepted 22 July

Wien

1991)

The RNA-catalysed self-splicing reaction of group II intron RNA is assumed to proceed by two consecutive transesterification steps, accompanied by lariat formation. This is effectively analogous to the small nuclear ribonucleoprotein (snRNP)-mediated nuclear premRNA splicing process. Upon excision from pre-RNA, a group II lariat intervening sequence (IVS) has the capacity to re-integrate into its cognate exons, reconstituting the original pre-RNA. The process of reverse self-splicing is presumed to be a true reversion of both transesterification steps used in forward splicing. To investigate the fate of the esterified phosphate groups in splicing we assayed various exon substrates (5’E-*p3’E) containing a unique 32P-labelled phosphodiester at the ligation junction. In combined studies of alternating reverse and forward splicing we have demonstrated that the labelled phosphorus atom is displaced in conjunction with the 3’ exon from the ligation junction to the 3’ splice site and vice versa. Neither the nature of the 3’ exon sequence nor its sequence composition acts as a prominent determinant for both substrate specificity and site-specific transesterification reactions catalysed by bI1 IVS. A cytosine ribonucleotide (pCp; pCOH) or even deoxyoligonucleotides could function as an efficient substitute for the authentic 3’ exon in reverse and in forward splicing. Furthermore, the 3’ exon can be minimized to a single monophosphate group. Upon incubation of 3’ phosphorylated 5’ exon substrate (5’E-*p) with lariat IVS the 3’4erminal phosphate group is transferred in reverse and forward splicing like an authentic 3’ exon, but with lower efficiency. In the absence of 3’ exon nucleotides, it appears that substrate specificity is provided predominantly by the base-pairing interactions of the intronic exon binding site (EBS) sequences with the intron binding site (IBS) sequences in the 5’ exon. These studies substantiate the predicted transesterification pathway in forward and reverse splicing and extend the catalytic repertoire of group II IVS in that they can act as a potential and sequence-specific transferase in vitro. Keywords:

Group

II intron

RNA;

RNA-catalysis; transferase

1. Introduction

0022-283e/91/220145-10

$03.00/0

transesterification:

self-splicing reactions of group II IVS is in contrast initiated by a nucleophilic attack of an introninternal 2’OH group of an adenosine residue close to the IVS%Yexon boundary. A consequence of this is accumulation of branched circular RNAs, so-called “lariat IVS”, during the reaction (Sharp, 1987; for a review, see Jacquier, 1999). Forward self-splicing of group II, as well as nuclear pre-mRNA splicing, is assumed to proceed by a consecutive two-step phosphoester transfer (transesterification) reaction (Fig. 1). First, the phosphodiester at the 5’ splice site is converted into a 2’-5’ bond at the branch site (lariat formation), and second, the phosphodiester at the 3’ splice site is transferred to the 3’OH of the 5’ exon in the exon ligation step (Van der Veen et al., 1986).

or intervening sequence Self-splicing introns, (IVSt) RNAs, are classified in group I and group II on the basis of their distinct secondary structure and the way self-splicing proceeds in vitro. As exemplified by the rRNA intron of Tetrahymena, selfsplicing of group I intron RNA is initiated by an external guanosine co-factor. The result is a liberation of the linear IVS RNA (for a review, see Cech, 1990). Similar to the spliceosome-mediated nuclear pre-mRNA splicing mechanism, RNA-catalysed

t Abbreviations used: IVS, intervening sequence; intron binding site; EBS, exon binding site.

yeast mitochondria;

IBS, 145

0 1991 Academic Press Limited

146

M. u’. Miiller

It has been demonstrated that upon excision from pre-RNA, group II lariat IVS bI1 has the intrinsic capacity to re-integrate into its cognate ligated exons by reverse self-splicing (Augustin et al., 1990; Mijrl & Schmelzer, 1990a) As to the fact that in forward splicing the same number of phosphodiester bonds are made as are broken. and therefore the energy of the existing phosphoester bonds is conserved. we have predicted that reverse selfsplicing could proceed by the true reversion of bot’h transesterification steps used in forward splicing (Augustin et al., 1990). In the first step, the 3’OH of the lariat IVS was expected to exert a nucleophilic attack at the 3’-5’ phosphodiester junction of the ligated exons. Here, transesterification would reconstitute the kinetic intermediate in splicing, the lariat IVS-3’ exon and the 5’ exon. Reversal of branch formation would be catalysed in the second reaction step. The 3’OH of the 5’ exon attacks the 2’-5’ phosphodiester at’ the branch site of the lariat IVS-3’ exon, reconstituting the pre-mRNA (5’ETVS-3’E) by transesterification (Fig. 1). Both forward and reverse splicing reactions are highly site-specific and show no requirement for the intervention of an external nucleotide cofactor or high energy source. Thus, it is evident that only the intrinsic capacity of the particular folded group IT TVS structure lowers the activation energy for both site-specific transesterification reactions, presumablv equally used in forward and in the reverse sphcing process. The reactivity towards a special phosphodiester can be explained in part by the existence of introninternal binding site(s) which specifically interacts with sequences at both splice sites. Determination of the 5’ splice-site involves base-pairing interactions of intronic sequences (EBS) with -5’ exon sequences (IBS) next to the splice site (Jacquier & Michel, 1987). Furthermore, this EBS-IBS interaction enables group II IVS to select exogenous substrate RNAs reactions in trccns (Miiller et al., 1988; Jarrell et al., 1988: Augustin et nl.. 1990; Mijrl & Schmelzer, 1990a). For specification of the 3’ splice site, in contrast, no base-pairing interaction of similar functional importance has been reported so far. Only one potential long-range base-pairing interaction (v-v’), involving a highly conserved 3’-terminal uridine residue of IVS and an adenosinr residue located between the structural domains I1 and TIT exists in all members of group TT introns (Michel et aZ., 1989). Formation of the y-y’ interaction has been considered to be rate-limiting for the exon ligation step (,Jacquier & Michel, 1990). This has been verified experimentally for the 1~11 related self-splicing intron aI5c. Studies in vitro revealed that truncation of the authentic 3’ exon sequences of bI1 group IT IVS pre-RNA to a minimum of two nucleotides has no severe effect on the efficiency of the exon ligation step in cia (Schmelzer & Miiller, 1987). We show here that group IT TVS RNA bI1 catalyses full cycles of forward and reverse splicing reactions with various substitutes of the authentic

et al.

3’ exon like (1) monoribonucleotides and (2) deoxyoligonucleotides. irrespective of t)heir sequence. and even (3) a 3’-terminal monophosphate group. Using a 32P-labelled phosphorus atom that’ forms t’he 5’E-*p3’E junction we investigated the fate of t,he esterified phosphate group during forward and reverse splicing. These studies substantiate the predicted transesterification pathway and. furthermore. extend catalytic repertoire of group IT TVS to act’ as a potential and sequence-specific transferase in vitro.

2. Materials and Methods Full-length pre-RNAs were synthesized ia vitro with the use of phage T3 polymerasr in thr presence of [c(-35S]CTP from &oRI-digested plasmids BS/bI I and BS/bII.SG-19 (Augustin et al., 1996) as described b> Miiller et al. (1988). Pre-RNAs were gel-purified and incubated in 46 mix-Tris 80,. 2 mw 60 miv-M&I,. spermidine. 500 mM-(NH,),SG, at pH 7.5 and 45°C’ for forward self-splicing and at 26°C for reverse splicing assays, respectively. (b)

I’repamtion

of’ sa6stmtc~s

Synthetic 5’ exon RNAs (5’ (:G(:AA(IAAAUC:I:~.3.1-1’GCUGUGIlI’CACGGACA(:A: 34-mer) were svnt,hesixed on an Applied Biosystems DNA/RNA Synthesizer (39 I ), SE-*pCp was prepared by 3’-end-labelling of 5’ axon RNA with phagr T4 RNA ligase and cytidine 3’.5’-[f,‘32Pjbisphosphat’e (32pCp: Amersham) as described (Miiiler rt al.. 1988). The 5’E-*PC,,, was prepared from gel-purifird 5’E-*pCp by the 3’ phosphatase act,ivit,v of T4 polynucleotide kmase at pH 5 (Cameron 8r Ilhlenbeck, 1977). The 5’S*p was generated from 5’-*pCon by /I-rlimination according to Tanner Cy:Cech (1987). (:himeric RNA DNA substrates (5’E,,,-*Ii-3’E(,,,, wtlrp produced by T4 RNA ligase catalysed ligation of a 6’-terminal 32P labelled DNA-oligonucleotide (5’ AATTCACAATTG( ‘TCAGAACTGTAC; 25mer) to the 3’OH of the 5’ exon RNA in the presence of 26 mM-MnCl,. followed by purification on denaturing IO??, (w/v) polyacrylamide gel electrophoresis.

3. Results (a) Mononmcleotides substitutes

for

p(‘p and pC’,, as fin&ma1 the authentic d’ exon squmcrs

For studies on the reverse splicing capacity of group TI intron bT1 we replaced the authentic 3’ exon sequences of the ligated exon substrate RNA with a cytosine mononucleotide. For this purpose synthetic 5’ exon sequences (5’Eo,), containing both intron binding sites (IBSl and TBSB), were 3’.endlabelled by (32P)*pCp. The resulting ligation product (5’E-*pep) was gel-purified and used as substrate RNA in reverse splicing experiments wit)h lariat IVS bI1. The advantage of this substrata RNA is obvious: the phosphorus atom at the ligation junct’ion carries a unique 32P label. According to the proposed reverse splicing mechanism (Fig. 1) this phosphate group could be transferred in conjunction with the Cp residue as a 3’ exon substi-

Fate of the Junction Phosphate in Group II Splicing 3’E

IVS

5’E

-44

2’OH u

147 5% - 3%

Larlat - IVS

3’OH

+

m

4 ---------,

(al

Lb)

Figure 1. Forward and reverse self-splicing of group II intron RNAs. Hatched and filled boxes represent 5’ and 3’ rxons, respectively; the IVS is shown as a thin line; open circles and asterisks mark phosphodiester bonds conserved in reaction products generated in the forward and in the reverse splicing pathway. Base-pairing of exonic IRS sequences with intronic EBSsequences is indicated. (a) Forward self-splicing of group II IVS proceeds nia a consecutive 2-step transesterification pathway (Van der Veen et al., 1986). In the first step, the 2’OH group of the branch adenosine residue attacks the 3’.-.5’ phosphodiester bond at the 5’ splice site (0) leading t’o 2’-5’ branch formation oia transest,erification; the resulting kinetic intermediate in splicing consists of 2 RNAs, the lariat IVS3’ exon and the 5’ exon. In the second reaction step, the 3’OH of the 5’ exon attacks the 3’-5’ phosphodiester bond at the 3’ splice site (*) resulting in exon ligation and in the release of the lariat-IVS via transesterification. (b) Reverse self-splicing. Reintegration of the lariatTVS into t#he 3’-5’ phosphodiester junction of the ligated exons (*) p roceeds via the true reversal of both consecutive transesterification steps used in forward splicing. In the first step, the 3’OH of the lariat-IVS performs a nucleophilic attack at the 3-5’ phosphodiester bond between the ligated exons; transesterification leads to reconstitution of the kinetic int,ermediate in splicing, the lariat IVS-3’E and the 5’ exon. Reversal of branch formation is achieved in the second reaction step; the 3’OH of the 5’ exon attacks the 2’-5’ phosphodiester bond of the lariat IVS3’E (0). resulting in reconstitution of pre-RNA (5’E-TVS-3’E) by transesterification.

tute to the 3’OH group of the lariat TVS by transesterification. Consequently, products generated by partial and by complete reverse self-splicing (lariat TVS-*pCp and 5’E-IVS-*pCp, respectively; Fig. 1) should be 3’.end-labelled via transesterification. Lastly, by virtue of the 3’-terminal *pCp label these products should be suitable for direct RNA sequencing by enzymatic methods. As shown in Figure 2, incubation of gel-purified lariat IVS with 5’E-*pCp substrate generated 32P-labelled products which migrated in polyacrylamide gels similar to 5’E-IVS-*pCp RNA and the lariat IVR-*pCp intermediate (or its y-shaped linearized form; IVS-*pCp). The reconstituted 5’I!-IVS-*pCp pre-RNA was functionally identical to the authentic bT1 pre-RNA. After incubation of the 5’E-IVS-*pCp RNA under forward splicing conditions the original input RNAs (lariat IVS and

the 5’E-*pCp) were reconstituted (see below). The reaction kinetics were essentially not affected by the presence of the substrate’s 3’-terminal phosphate group. This is shown in the results of experiments performed with 5’E-*pCou RNAs (Fig. 2). Identical results were obtained with mutant S6-19 lariat IVS, after substitution of the 3’-terminal TVS sequence (Au) by UCU (Augustin et al., 1990). Due to the sequence substitutions of the mutant IVS, however, the completely reverse spliced product (5’E-IVS-*pCp) was obtained in significantly higher concentrations (Miiller et al., unpublished results). The basic conclusion of these experiments is that the cytosine mononucleotide (*pCp or *PC,,) is a functional substitute for the authentic 3’ exon sequences in forward and in reverse self-splicing. Interestingly, the natural 3’ exon sequence of bI1

M. W. Miiller

148

I

5&p

et al.

I

5’E-+pCgH

S/E-;

‘#@is

sm”

-

Lariat

IVS-3’E*

5’E-IVS-3’E*

-

5 43

ii

4

i 4

43

2

I

S’E-3’E*

54321 4

Figure 2. <:omplete reconstitution of 3’ and 5’ splice junctions by reverse self-splicing with 5’E -*J)($. 5’li:~-*~)(‘,, OI .i’&*p substrates. Gel-purified lariat. TVS (5 nM), generated in splicing assays by bT1 wild-type prc-RSA. were incubated under reverse splicing conditions for 90 min with increasing concentrations (lane I to 5) of exogenous ligated suhstrak RXAs (5’EX’E*). comprising a unique 32P-labelled phosphorus atom at the ligation junction (5’E -3’E*: !i’E-*p(‘~). 5’EP*pCon or 5’E-*p, respectively). Lariat IVS-3X*: intermediate product in reverse splicing. generated upon sitrphosphorus atom. IVS-3X*: y-shaped intermediate: 5’lGTVS--3’E*. specific transesterification at the 321’-labelled constituted pre-RPu’A upon debranching of the lariat IVW’E* by the 5’EoH. Reaction products w-ere elect)rophoresed on denaturating 5% polyacrylamide gels. Lane 1 to 5: 2.5 nM (l), 50 nM (2). l@O nM (3). 15.0 nM (4) and 2@0 nM (5).

(14-mer) starts with an uridine residue. Thus. it) seems likely that neither the sequence composition and size of the 3’ exon, nor its first’ nucleotide are prominent determinants for specificity in forward and in reverse splicing reactions catalysed by bT1 group II IVS RNA in vitro. (b) Fate

of the phosphorus atom ligation junction

at thr 5’1’M’ti:

Group II lariat IVS-mediated reverse self-splicing has been assumed to be initiated by nucleophilic attack of the lariat IVS 3-oxygen atom on the phosphodiester bond at the 5’E-3’E junction. This is accompanied by the displacement of the 3’ exon in conjunction with its 5’-terminal phosphate group to the 3’-oxvgen atom of the lariat TVS by transesterifica;on (Fig. 4). We have substantiated this prediction by a sequence analysis of products generated in reverse splicing experiments with lariat IVS of mutant S6-19 and 5’E-*pCp substrate RNA (Fig. 2). As noted above, a transesterification reaction at the 32-labelled phosphorus atom of the substrate RPc’A renders the product of partial (lariat IVS-*pCp) and of complete reverse splicing (5’E-IVS-*pCp) amenable to direct RNA sequencing by enzymatic methods.

According to the scheme shown in Figure 3(a), digestion with RKase rZ is expected to generate 3’-labelled pentanucleotides from both analysed RNAs (5’E:-IVS-*pCp and lariat TVS-*pCp). In contrast. digestion with R?u’ase T, should result in the appearance of a labelled 1‘I-mrr or 1%mer. depending on whether the linear 5’E-TVS-*pCp RSA (17-mer) or the branched lariat TVS-*p(‘p intermediate (I 8-mer) is assayed The results. as presented in Figure 3(a), art’ consistent wit’h t)hese predictions. Thcb clearly majority of products generated by the sequencespecific RNase digestions had the expected sizes. These data confirm the covalent ligation of t~hr junction phosphate group from the substrate RXA (5’E-*p3’E) to the 3’OH of the lariat TVS bl transesterification. Some minor reaction products were also obtained that were exactly one nucleotide longer than predicted (6-mer upon U, 19-mer upon T, digestion~~~~~v1R;nerT;;L~~ products are consist’ent with an alternatibe version of a transesterification reaction catalysed by bI1 lariat IVS. In this case, a dinucleotide (the *p(:p pseudo 3’ exon plus the 3’-terminal nucleotide of t’he 5’ exon) is transferred to the 3’OH of the lariat IVS. The simultaneous occurrence of accurat*ely and inaccurately reverse-spliced products indicates that t)hcb

Fate

qf the

Junction

Phosphate

in Group II 8plicing

149

5’E- ;Cp S’E-D/S-&p --

Lariat IVS-;Cp

l7(T,)

I(T,)

- (6) -5 ;‘I

I

I

7

(Ue)f~

6

I

to 1

(b)

Figure 3. Sequence analysis of products generated in reverse and in forward splicing reactions. (a) RNase digestion of reverse splicing products. “P-1abelled RKAs (5’E-IVS-*pCp, lariat IVS-*pCp intermediate), generated in initial reverse splicing experiments with 5’E-*pCp substrate in the presence of mutant S&19 lariat TVS, were gel-extracted and subjected to direct RNA sequencing with RZu’aseU, and T,, respectively. The schematics represent 3’.terminal sequences of the IVS. constituted by the covalent transfer of the 32P-labelled phosphorus atom (shown by an asterisk) in conjunction with the Cp residue to the 3’OH of the IVS; arrows mark the positions of the digestion sites. Xote that two RPc’ase T, digestion reactions at the lariat IVG*pCp RKA are necessary for the release of the 3’-terminal-labelled fragment as indicated. (b) RNase T, digestion analysis of the 5%*pCp product, reconstituted in forward splicing reactions with gel-purified lariat IVS-*pCp RNA in the presence of 5’E,, (S’E-*pCp: 5’E UCGUUGUGUUUAU($GACAGA-*pCp). Enzymatic sequencing was performed as described in Materials and Methods. The reaction products were fractionated on 20% polyacrylamide/8 M-urea gels and autoradiographed. M, marker ladder.

3’017 of the lariat IVS has two potential targets for nucleophilic attack at the 5’E-*pCp substrate: (1) the phosphodiester at the ligation junction; (2) the phosphodiester one nucleotide 5’ to it. This alternative transesterification reaction has also been observed in reverse splicing assays with the authentic 5’E-3’E substrate RNA (Miiller et al., unpublished results). To test the fate of the phosphate group at the TVS-3’E ligation junction we re-incubated the lariat TVS-*pep intermediate (Fig. 2) with 5’ exon substrates under forward splicing conditions (not shown); the resulting 32P-labelled 5’E-*pCp RNA reaction product was gel-extracted and sequenced by enzymatic methods. As shown in Figure 3(b), the products of the RSase T, digestion assay were in agreement with the assumption that the *pCp mononucleotide was spliced to the 5’Eo,. These results clearly substantiate the mechanistic similarity bet’ween the second step of the forward reaction (exon ligation) and the formation of the lariat TVS-3’E intermediate in the first transesterification step of the reverse splicing reaction (Fig. 1).

(c) Reverse splicing 6 E,,,-3’E,,,

,reactions with chime+ substrate molecules

The reactions discussed above suggest that the nature of the 3’ exon is not a major determinant for group II bI1 catalysed splicing reactions in citro. Apparently, site-specific integration into a substrate molecule depends essentially on base-pairing interactions of 5’ exon sequences (TBSl and IBSB) with complementary intronic sequences (EBSl and EBSS). Thus, it seems likely that) the catalytic apparatus of the IVS can transfer any 3’ exon substitute, assuming that it is linked via a 3’45’ phosphodiester to the 5’ exon (reverse splicing) or bo the 3’ terminus of the IVS (forward splicing). We have tested this prediction by using a chimeric substrate molecule, composed of 5’ exon RNA sequences (comprising IBS2 and TBSl) covalently ligated to a 3’ exon substitute composed of 25 deoxynucleotides. The sequence of the DNA oligonucleotide was not related to t’he original 3’ exon RNA sequence. For experimental purposes, the 3’-5’ phosphodiester at the RNA-DNA ligation

M. W. Miiller

150

-II

Wild-type

et’ al.

S6-19 Lariat

Ivs- EDNA

Lariat IVS

I Lariat-IVS-c

-

S’E-IVS-;

\-

I

IVS-p‘/ 5’E-IVS-;DNA

IVS 0

I

210

I

2

Figure 4. Deoxyoligonucleotides as 3’ exon substitutes in reverse splicing. Chimaeric 5’E cfNA’- 3’E@,,, substrates (5 nM). comprising a unique 3 P-labelled phosphorus atom at the ligation junction, were assayed under reverse splicing conditions for 90 min with gel-purified lariat IVS of wild-type and mutant SC-19 (5 nM). Reaction products were electrophoresed on denaturing 5 y0 denaturing polvacrylamide gels. Lane 0, incubation of [cr-YSS]UTP-labelled lariat IVS under assay conditions in absence of substrate. Lanes 1 and 2. incubation of [cc-3SS]UTP-1abe11ed lariat IVS with 32P-1abe11ed substrate: lane 1 is identical to lane 2 ~'E(RNA,-~'~(DNA, with bhe exception that the 35S label was selectively eliminated in lane 2 by screening. The ligation junctionspecific 32P label is seen with the lariat TVS-*p DrjA int,ermediate and with the Fi’E-IVS-pDNA end product,. Additional products, so far undefined. are marked b> arrows.

junction (5’ .GA-*p-AA .3’) was selectively labelled by 32P. Subjection of the chimaeric substrate to reverse splicing conditions revealed that the 5’EcRNA)-*p 3’E(,,,, molecule can be efficiently processed as substrates entirely composed of ribonucleotides (Fig. 4). Both, wild-type and mutant S6-19 lariat, IVS catalysed the covalent transfer of the DNA sequence to the 3’OH of the lariat IVS to almost equal efficiency. After nucleophilic attack of the bond at the ligation IVSO” on the phosphodiester junction, the DNA component of the substrate was covalently ligated to the IVS by transesterification. This leads to the formation of the kinetic intermediate in splicing, the lariat IVS-3’E,,,,, and the 5’ exon RNA. Reversal of branch formation by the constituted a chimeric 5’E-IVS-3’E(,,,, 5’Eo, molecule. The accuracy of the complete TVS integration into the 5’E (RNA)-*p3’EcnNA) substrate was

(0) Figure 5. Characterization of products generated by lariat IVS-mediated phosphotransferase reactlion. (a) Reverse splicing experiments with 5%*p substrate RKA. Gel-purified lariat IVS of wild-type and mutant, SC-19 were incubated with an excess of 5X--*p substrate under reverse self-splicing conditions for 90 min. The reaction products were electrophoresed on denaturating 5O,, polyacrylamide/8 M-urea gels and autoradiographed. Lariat IVS-*p: 3’ phosphorylated IV8 int.ermediatr RNA: IVS-*p: y-shaped intermediate (broken lariat TV%*p): Fi’E~IVS--*p: end product. constitut,ed by cxomplete reverse splicing. Preparation of 5’E--*p subst,ratr RKAs. having minimized the 3’ axon Imrtion to a single [32P]monophosphate group. is described in Materials and Methods. (b) Sequencing of the linear 5%lVS~*p. Reconstituted S’E-IV%*p pre-RNA, generated in r’evrrsc splicing experiments with Se-19 lariat TVS. was gelextracted and subjected to RKase T, and RKase r2 digestion, respectively. For comparison. the 5’WT\‘S*pCp RNA. constituted in assays wit,h wild-typr lariat IVS and 5’E-*pCp substrate (Fig. 2) was analysed in parallel. Reaction products were fractionated on ZOq,, polyacrylamide/8 M-urea gels and autoradiographed. M. marker ladder.

confirmed by sequence analysis (not shown). Re-incubation of the chimeric 5’E--TVS-3’EtoNa, under forward splicing conditions regenerated thtb lariat TVS,, and the 5’E(,,,)-*l)-3’E~oNA) input molecules (not shown). Chimaeric RNA-DNA substrates, having substituted the 3’-terminal ribonucleotide of the 5’ exon RNA by a deoxyadenosine residue, were found to be non-reactive for the site-specific t,ransest,erilication reaction at the natural splice junction (not’ shown). Thus, it appears that the 2’OH group adjacent to the attacked phosphodiester bond strongly affects

Fate of the Junction Phosphate in Group II Splicing

3

4

5

6

7

8

9

IO

PH

Figure 6. The pH dependence of the diester transfer reaction (m) as compared to the monophospho transfer reaction (A), Reverse splicing reactions of wild-type lariat IVS RNA was done as described for Fig. 2, with the exception that lariat IVS was kept saturated with substrate at, all the pH values tested. Following gel electrophoresis. the relative activity of reaction products was determined by an automated gel scanner (LKB, I%roScan XL) and plotted against the pH.

the activity reaction.

for the site-specific

(d) Reverse splicing with 5’E-*p

transesterification

substrate RLVA

In the reverse splicing reactions discussed above, group 11 IVS acted on the specific 3’-5’ phosphodiester that forms the junction of the ligated exon. Tn continuation of our studies we have assayed a single phosphate monoester with regard to its function as a 3’ exon substitute. For this purpose 3’-phosphorylated 5’ exon RNAs (5’E-*p), after reduction of the 3’ exon portion to a single 32P-labelled monophosphate. were examined under reverse splicing conditions in the presence of bI1 lariat TVS. As seen in Figures 2 and 5(a), [ 32P]phosphorvlated TVS int,ermediate RNA (lariat TVS*p; linearized y-shaped IVS-*p) as well as the end-product, of t,he complete reverse splicing reaction (5’E-TVS-*p) were generated. Treatment of the phosphorylated lariat’ TVS-*p RNA with T4 polynucleotide kmase, which exhibits a specific 3’ phosphatase activity at pH 54 to 69 (Cameron & Chlenbeck, 1977), lead to quantitative dephosphorylation of the lariat IVS (not, shown). This finding (combined with data presented below) is consistent with the notion that the transferred phosphate group is esterified through the 3’OH of the 3’-terminal uridine residue (U765) of the lariat TVS.

151

The putative 3’-terminal phosphorylated RNA (5’E-IVS-*p), generated by reverse splicing experiments with lariat IVS from mutant S&19, was gelextracted and subjected to direct RNA sequencing by enzymatic methods. For comparison, the 5’EIVS-*pCp wild-type pre-RNA, constituted in reverse splicing assays with lariat IVS and 5’E-*pCp substrate, was analysed in parallel (Fig. 5(b)). The majority of the RNase digestion products at position 4 (RNase U,) and at position 16 (RNase T,) represent 3’-terminal IVS sequences of mutant S&19 expected to be constituted by accurate transesterification at the 32P-labelled 3’ monophosphate group of the 5’E-*p substrate. Minor reaction products at position 5 (RNase U,) and at position 17 (RNase T,) are consistent with the alternative version of the IVS transesterification reaction mentioned above. The monophospho transfer reaction catalysed by group II IVS is specific for the 3’-terminal phosphate group: 5’-phosphorylated 5’ exon RNAs (*~-5’~,,) are non-reactive substrates (not shown). The pH dependence of the 3’-specific phosphotransferase reaction was examined under reverse splicing conditions at 26°C. The 5’E-*p substrate was incubated with lariat IVS at a pH range from 3.5 to 94. The conversion rate of the input) 5’E-*p to 3’-phosphorylated IVS RNA (lariat TVS-*p. 5’E-IVS-*p) and 5’Eou was found to be pHdependent with a maximum around pH 6.0, displayed in Figure 6. In contrast, the transfer of *pCp from 5’E-*pCp to the lariat’ TVS (diester transfer reaction) has a maximum at pH Ml. Based on comparison at pH 7.5 (Fig. 2), we evaluated that the reaction rate is at least one order of magnitude higher than that of the monophospho transfer reaction (Fig. 2). Phosphorylation of the IVS RNA was found to be reversible: upon incubation of the lariat TVS-*p intermediate with 5’ exon RNA (LYE,,) the original input RNAs (5’E-*p substrate and the lariat IVSoH) were reconstituted (not shown). Thus, group TT lariat TVS has the potential to act as a true catalyst by transferring a monophospho group from a 5’E-*p donor to a 5’E,u acceptor RNA. We suggest that the monophospho transfer reaction: (1) uses a path identical to that of the RNA catalysed diester reaction, (2) takes place at the same active site and apparently, (3) relies on the IBS-EBS interaction.

4. Discussion We have analysed the fate of the phosphorus originating from the junction of an exogenous 5’E-*p3’E substrate during alternating processes of reverse and forward splicing a a group II IVS RNA. Our results substantiate the previously postulated splicing pathway and provide the first experimental evidence for the mechanistic similarity between the init’ial transesterification step of the reverse reaction and the exon ligation step in forward splicing.

152 (a) Fate of the junction phosphate and its reversion

Ill. IV. Miiller in exon ligation

With an exogenous 5’E-*pCp substrate, where *pCp substitutes for the 3’ exon sequence. we demonstrated that this [32P]phosphorus atom is transferred in conjunction with the Cp residue as 3’ exon substitute to the attacking nucleophile (Figs 2 and 3). In detail, the junction phosphate is esterified through t’he 3’ oxygen of the lariat TVS in reverse splicing, or through the 3’ oxygen group of the 5’ axon in forward splicing. These data confirm our previous prediction that the phosphodiester that forms the IVS-3’E and .5’E-3’E junction are constituted by the same phosphate group (Augustin rt al., 1990). It is derived from the 5’ position of t,he first nucleotide of the 3’ rxon and. furthermore. it is conserved in reaction products generated in multiple cycles of forward and reverse splicing reactions. Thus, from the mechanistic as well as from t,he chemical point of view. formation of the lariat IVRXE) in t)he initial t’ransesterification step of reverse splicing is t)he true reversion of the exon ligation reaction used in forward splicing and Gee cersa (Fig. 1). The reversible reaction process can be shown schematically: .?‘E-p3’E + lariat IVSOH ti lariat IVS-p3’E

+ 5’EoH.

Although not rigorously proven experimentally. we assume that the phosphate group that’ forms the d’-5’ linkage at the branch site is converted into the 3’4 phosphodiester at the authentic 5 splice site (reverse splicing) and f:ice versa (forward splicing). Furthermore, our data clearly show a mechanistic similarity between the RNA-catalysed splicing reartions of group I and group TI intron RN& and the mechanism as pre-mRNA splicing nuclear previously suggested (Sharp, 1987; Price, 1987). In all cases, splicing proceeds by a consecutive twostep transesteritication pathway where the phosphate moieties derived from both t,he 5’ and 3’ splice sites are conserved in the reaction products. Self-splicing of group I intron RNAs is dependent on an external guanosine residue that triggers splicing by transesterification upon nucleophilit attack of its 3’-hydroxyl on the 5 splice site (Cech. 1987, 1990). In contrast, the RNA-catalysed selfsplicing of group TX IVR. as well as the nuclear premR?r’A splicing is initiated by nucleophilic attack of an intron-internal 2’-hydroxyl group of the branch adenosine residue, located near the IVS-3’ exon boundary. As a consequence, t’he intermediate product in splicing (lariat’ IV&3’ exon and Fi’E,,) is generated after conversion of the 3’4 phosphodiester at the 5’ splice site to the 2’--5’ phosphodiester at the branch adenosine residue (Jacquier, 1990). On the basis of these similarities a close relationship between self-splicing of group II intron RNAs and that of nuclear pre-mRNA splicing can be drawn. This relation would reflect a common evolutionary origin, suggesting that the splicing process of nowadays nuclear pre-mRNA introns

et al.

could be descended from a filndamentally R&A-\based splicing mechanism (Sharp. 1987); probably directly descended from self-splicing group I I TVS RNAs. ln contrast to the intramolecular self-splicing reactions catalysed by group IT and group I IVS RSA, nuclear pre-mRNA splicing is mrdiat’ed by a group of exogenous snRNAs (small nuclear RNAs) complexed wit,h proteins (snRNPs) (Maniat’is & Reed, 198i; (ireen, 1986). It) has been suggest,ed that the assembly of the splicing caomplex from snRNPa primarily assists to bring the precursor into the proper geometry thr t,he two-step t,ransesterification process. This then (aan be viewed as being basically RNA-catalysed (C’rcah.1986).

splice-sitr

rfmqnatzon

by group I I I 1’S

We have addressed, in detail. the question of the functional importancae of the 3’ exon for sit,e-specific transesterification reactions. (Qatalysed by group II IVS RNA. Published sequence comparisons of 70 organellar group 11 introns and their 3’ exons ha,ve revealed a putative long-range base-pair interaction of the tirst nuoleot,ide of t,he 3’ exon with the nucteotide immediately upst’ream from EHSl: t,his criterion. however. does not fit t)o group 11 IVS hll (Michel rt al., 1989). We tint1 that a cytosine mor~oribonu~~leotitle (p( ‘1) and p(‘OH) (aan substitute t,hr a,uthent,ic* bT1 3’ t’son sequences. originally starting with an uridine residue. 11~csycles of alternating rwwse and forward splicing reactions we have documented t’hat this cytosine rriono-nucleotitie is eficiently accepted as il 3’ axon subst,itute. Essentially identical result)s werf‘ obtained with A. G or I’ ribonuc+ot~idrs serving as 3’ exon substit,utes. however. with signiticant differences in the rate of lariat IVS-3’N formation (reverse splicing) or IULP f’er.sa in the rate of tason ligation (Miiller rt al.. unpublished result,s). Worth noting, that, also in RNA caatalysis t’xernplified k)y the Tetrahyrrwrra group 1 1VS RNA. a mononucleotide (aan substit’ute thr t,htx natural 3’ exon sequences. A shortened form of this group 1 intron RNA (I,- 19) catalyses nnc~leotiti~lt~ral~st’erasr reactions with oligo(cytid,vlic acid) substrates with multiple t urnowr (Zauy C! (‘~1~. 1986). Furthermore, mini-rxon ligation reactions between (Iplr and Gplv (were N is (:, A. I’ or (‘) was reported for thcl I,-21 Sac1 IVS R’NA by Ka.y & lnoue (19X7). Tn t,hin case. (“~1~ acts as 5’ exon and the phosphodiest,er of (:pN kcts as the 3’ splice sit,r. Even deosynucleotides can suhstit,utt, tor t hr authentic 3’ exon. Again. neither t,he sequence c-onposition of the 3’ rxon (DKA). nor its first) nucleot,ide was related to t’he downst,ream exon of bI1. WP have shown that the DIVA portion originating f’rom a chimarric RN&DNA substrate (5’En,,,-3’l$,,,,)

Fate of the Junction Phosphate in Group II Splicing can be transferred in reverse and forward splicing reactions as efficiently as a 3’ exon entirely composed of ribonucleotides. Interestingly, chimaeric RNA-DNA substrates, where the 3’-terminal ribonucleotide of the 5’ exon RNA was replaced by a deoxyadenosine residue, are almost unreactive in the initial step of the reverse splicing reaction. This suggests that the 2’OH of the terminal ribose of the 5’ exon strongly affects the reactivity of the site-specific transesterification at the adjacent phosphorus atom. (iii) Monophospho

tmnsferase

reaction

In combined studies of reverse and forward splicing reactions we have further shown that group II IVS catalyses the transfer of a 3’ exon portion minimized to a single monophosphate group. According to the RNA-catalysed diester reaction, the phosphate monoester derived from the 3’ position of an exogenous 5’ exon substrate RNA (5’IC*p), is esterified through t,he 3’OH group of the lariat’ IVS. generating the “charged” lariat IVS-*p intermediate. Trans.phosphorylation of the JVS RNA was found to be reversed by the 5’ exon terminating with a 3’-hydroxyl group, leading to reconstitution of the original input) RNAs (5’-*p and lariat JVS). The reactivity towards monophosphate groups, substituting for the 3’ exon, shows a pH maximum around 6.0; the overall reaction is at least one order of magnitude lower than that of the RNA-catalysed diester reaction. which exhibits a pH maximum around 8.0. As noted previously (Cech. 1987), the acid pH optimum probably reflects the demand for protonation of a phosphate monoester, rendering it’ to react like a diester. The sequence specificity of the reaction relies entirely on t)hr TBS-ERS interaction; therefore, we predict that the monoester transfer reaction takes place in the same active site and presumably uses the same path as the RNA-catalysed diester reaction. Besides the phosphotransferase, group 11 lariat, TVS also exhibits a sequence-specific 3’ phosphatase activitl (not shown). Prolonged incubation of 3’-terminal 32P-labelled 5’ exon RXA (5’E-*p) leads to conversion of the input radioactivity into inorganic phosphate in the presence of lariat IVS RNA only. In contrast. neither 5’ phosphorylated 5’ exon sub&rates (*I)-5’E). nor the 3’ phosphorylated lariat I\‘%*p intermediate in absence of the 5’ exon are reactive. These findings suggest that by virtue of t’he EBS-TBS interaction the phosphomonoester 3’ adjacent t,o 5’ axon is subjected t’o site-specific hydrolysis by H,O or OH- of the solvent. In summary. it turns out that the RNA catalysis of group IT intron R,NA is fundamentally based on the reactivity towards phosphodiesters or even phosphomonoesters designed to be special for a transesterification reaction as well as for site-specific hydrolysis. Croup I1 TVS bll has then a catalytic potential comparable to the Tetrahymena group 1 intron RNA. Shortened versions of this group I IVS (L-l 9) facilitate cleavageligation reactions on exogenous substrate RNAs with mult,iple turnover:

153

the group I ribozyme can act as a nucleotidyl transferase and phosphotransferase, both by the sitespecific transesterification mechanism (for a review, see Cech, 1990).

(b) Catalytic potential of group II intron We report on new catalytic activities of group II bI1 lariat IVS RNA in vitro. On the basis of the principle of the second transesterification reaction used in forward splicing (exon ligation) and its reversibility by the 3’OH of the lariat IVS, group II TVS acts as a potential nucleotidyl transferase. RXA-RNA and RNA-DNA recombinase and phosphotransferase. We have shown that monoribonucleotides. polyribonucleotides and deoxyoligonucleotides, irrespective of their sequence, and even a monophosphate group can be transferred from one substrate to another. As no high energy cofactor is consumed during the RNA-catalysed transfer reautions, group II lariat IVS RN,4 can be envisaged as a potential isomerase, entirely composed of specititally arranged polyribonucleotides. A key feature in all these transfer reactions is the alternating charging and discharging reaction of the lariat, IVS by transesterification. Similar to reactions catalysed by protein enzymes, initially a noncovalent enzyme substrate complex of the lariat IVS is formed with its exogenous substrate RNA win ICBS-IBS interaction. Secondly, a covalent intermediate is generated in the charging reaction upon nucleophilic attack of the lariat 3’-oxygen atom at t,he specified phosphorus atom (reversal of exon ligation). Lastly, the lariat IVS is regenerated into its original form by discharging the lariat intermediate by transesterification upon nucleophilic attack of an exogenous 5’EoH at the 3’ splice sit’e (exon ligation). Substrate specificity in the charging as well as in the discharging reaction is provided 1)~ interactions between the intronic exon binding sites (EBS) and the complementary sequences (TBS) of the donor and acceptor RNAs. Authentic 5’ exon sequences contain t,wo intron binding sites (IBSl and IBS2) (Jacquier bz Michel, 1987). Several in vitro data, however, have revealed, that only one of them, the EBSl-IBSI base-pairing interaction is sufficient to promote reactions in. cix (forward splicing) and in tram (reverse splicing) (Jacquier & Michel, 1987; Miiller et al., 1988:,Miirl & Schmelzer, 1990a,b). This suggests that’ speciticity of the RNA-catalysed transfer reactions is provided by the complementarity of six base-pairs only (EKSIIBS1 interaction). Therefore, one might speculate that’ the intronic EBSl motif can be adapted to an: RNA sequences of a substrate molecule of choice. substit,utinp as pseudo-IBS sequence. The phoxphorus atom 3’ adjacent to t’he pseudo-IRS1 site would be particularly prone to a transesterification reaction initiated by the 3’OH of the lariat TVS, leading to the transfer of any molecule 3’ adjacent to the IRS1 sequence from an exogenous substrate t’o another with multiple turnover.

154

M. W. Miiller

We thank Professor Dr John Dittami for critical reading of the manuscript; Dr Susanne Augustin for many helpful discussions, comments and criticisms. Special thanks are due to Sylvia Kaltenbrunner for her expert technical assistance. This work was supported by the Austrian Ministry of Science and Research.

References Augustin, S., Miiller, M. W. & Schweyen. J. R. (1990). Reverse self-splicing of group II intron RNAs l;n vitro.

Xature

(London),

343. 383-386.

Cameron. V. & Uhlenbeck. 0. C. (1977). 3’-Phosphatase activity in T4 Polynucleotide Kinase. Biochemistry, 16. 5120-5126. Cech, T. R. (1986). The generality of self-splicing RNA: Relationship to nuclear mRNA splicing. (‘ell, 44, 207-210. Tech, T. R. (1987). The chemistry of self-splicing RKA and RNA enzymes. Science, 236. 1532-1539. Tech, T. R. (1990). Self-splicing of group I introns. dnnu. Rec. Biochem.

59. 543-568.

Green, M. (1986). Pre-mRNA splicing. Annu. Rev. Uenet. 20. 671-680. ,Jacquier. A. (1990). Self-splicing group IT and nuclear pre-mRNA introns: How similar are they! Trends Hiochem. Ski. 15, 351-354. .Jacyuier. A. & Michel, F. (1987). Multiple rxon-binding sites in class II self-splicing introns. (‘ell. 50. 17-21). .Jacyuier. A. & Michel. F. (1990). Base-pairing inttiactions involving the .5’ and 3’-terminal nucleotides of group II self-splicing introns. ./. Mol. Biol. 213, 437447. Jarrell. A. K.. Peebles, c’. L.. Dietrich, R. (:.. Romiti. S. L. & Pearlman. P. S. (1988). Group II intron selfsplicing: Alternative reaction conditions yield novel products. J. Biol. Chem. 263, 3432-3439. Kay. P. S. & Inoue, T. (1987). Catalysis of splicing-related

et al. reactions Nature

between

(London).

dinucleotides

by

a ribozyme.

327, 343-346.

Maniatis, T. & Reed. It. (1987). The role of small nuclenl ribonucleoprotein particles in pre-RNA splicing. Nature

(London),

325, 673-678.

Michel, F.. I’mesono, K. &, Ozeki. H. (1989). (‘omparativr and functional anat,omy of group II catalytic int,rons---a review. Gene. 82. 5530. ,Miirl, MM.& Schmelzer. (‘. (1990a). Integration of’ group 11 intron bll into a foreign RNA by reversal of the selfsplicing reaction in vitro. Cell. 60, 629--636. MGrl, M. 8r Schmelzer. (‘. (199Ob). Group IT intron KXA catalyzed recombination of RXA in vitro. .v/tc/. .-lcids IZrs. 18, 6545-6551. Miller, 11. W.. Schweven, R. ,J. $ S(ahmrlzer. (‘. (19%). Selection of cryptic 5’ splice sites by group r1 intron RN& itt &To. Sucl. Acids Ke,s. 16. 7383- 7395. Price, ,I IT. (I 987). 5’ Axon requirement, for self-splicing of the TPtrrchymena th,ermophiZa pre-ribosomal RNA and identitication of a crypt,ic 5’ splice-&e in the 3 PROXY. .I. Mol. Hid. 196. 217-221. Schmelzer. (‘. B Mtiller. 11. WT. (1987). Self-splic?ng of group IT introns in &ro: Lariat form&on and 3’ splicae site selection in mutant R,NAs. /‘e/l. 51. 753-762. Sharp, 1’. A. (1987). Splicing of messenger RSA prvvursors. Science, 235. 76&771. Tanner. N. K. & (“ech. T. R. (1987). (:uanosinr binding required for cyclisation of t#he self-splicaing int,rrvrning sequence ribonucleic acid from ‘frtrahqlmurr~~ fhermophila.

Biwhemistry,

26. 3330-3340.

Van der Vern. R.. Xrnbrrg. .A. (‘.. Van der Horst. (I.. Ronrn. I,.. Tabak. H. F. & (:rivrll. 1,. ;\. (I!)%). Excised group II int)rons in Beast mitorhondria arc> lariats and can be formed by self-splicing in Gfro. (‘PI1 44 22-234. . <

Zaug. .L\. ,I: & (‘ech. ‘I‘. R. (1986). ‘I’hr illtrrvrnlny sequence KNA of Tetrr@nma is an enzyme. Scit,nc~, 231. l-70- 475.