The in vitro system used here was first employed in studies of poly(A)-dependent transcriptional pausing (53
), then it was optimized for transcription-coupled 3′-end processing (60
), and now we have shown that it robustly recapitulates all major activities of the “mRNA factory” (79
), including splicing. Although we have not explicitly characterized capping, mammalian capping has been reproduced numerous times in vitro under a wide variety of experimental conditions (12
), so we assume that it occurs normally here as well. The splicing efficiencies we report are comparable to those of recently reported transcription-processing systems optimized exclusively for splicing (15
), and the cleavage/polyadenylation activity reported here is comparable to that of transcription-processing systems optimized exclusively for cleavage and polyadenylation (1
). Thus, it appears that no compromises have been made to the individual processing activities in order to drive both of them simultaneously, along with transcription, under the in vitro conditions reported here. We believe that this is an important indication of the physiological significance of the functional interactions observed by use of this system because in vivo, of course, all of these processes share the same conditions.
As emphasized previously (60
), simply achieving high rates and efficiencies of concurrent reactions in vitro does not establish that the reactions are functionally coupled. To establish functional coupling, it is necessary to show that the individual reactions affect each other according to meaningful functional criteria. Figure summarizes the functionally coupled relationships (arrows 1 to 5) established experimentally for the transcription/processing system described here. Four of these coupling interactions (arrows 1 to 4) involve second-intron splicing. Of these, the first three involve coupling between second-intron splicing and upstream (first-intron splicing) or downstream (cleavage/polyadenylation) processing reactions and have previously been described for systems that do not include concurrent transcription (5
). However, arrow 4 in Fig. denotes a new coupling interaction in which second-intron splicing is linked directly to the integrity of the ternary transcription complex. The fifth coupling interaction in Fig. is that between the 3′-end processing apparatus and the transcription complex, which occurs here, in the presence of splicing, just as with the nonsplicing substrates described previously (60
). We now consider individually each of the coupling interactions noted in Fig. .
FIG. 6. Coupling of transcription to terminal exon definition. (A) Experimentally determined functional connections among transcription, splicing and poly(A) site cleavage in this in vitro system. (B) Cartoon of terminal exon definition in the context of the (more ...)
Arrow 1 in Fig. refers to the disabling effect on first-intron splicing caused by mutating the 5′ splice site of the second intron (Fig. ). This shows that the 5′ splice site of the second intron couples first- and second-intron splicing via exon definition (5
) in this in vitro system. Most transcription-splicing studies of which we are aware have been done under conditions of intron definition (15
). However, exon definition is the predominant mode of vertebrate splicing (5
). According to the exon definition model, 3′ splice sites do not commit to splicing until the polymerase traverses the entire downstream exon, including its flanking 5′ splice site. Thus, for example, the splice sites flanking exon 2 in the globin pre-mRNA direct formation of an exon definition complex, mediated by U2AF, SR proteins, and the U1 snRNP, before being juxtaposed for splicing with the previously transcribed exon 1 upstream (5
). Importantly, exon definition is robust in our coupled system (Fig. ) even though the first intron is short and fully capable of intron definition (29
) and is transcribed in its entirety well before the downstream 5′ splice site of exon 2 emerges from the polymerase.
Arrow 2 in Fig. refers to the impaired second-intron splicing caused by mutation of the poly(A) signal (Fig. ). This reflects participation of the poly(A) signal in second-intron splicing via its role in third-exon definition (56
). This situation is analogous to the role of the downstream 5′ splice site in second-exon definition discussed above (5
). Arrow 3 refers to the need for an intact 3′ splice site in the second intron to achieve full activity of the poly(A) signal (Fig. ). This confirms the expected coupling of second-intron splicing to 3′-end processing in this system (20
). The mutual enhancements of splicing and 3′-end processing summarized by arrows 2 and 3 reflect the joint participation of these two activities in definition of the third exon. The molecular interactions allowing this cooperation appear to be based on binding between the U2AF-U2 snRNP complex of the spliceosome and the CFIm
-CPSF complex of the cleavage/polyadenylation apparatus (Fig. ) (36
). Presumably, these four proteins constitute the core of the exon definition complex for exon 3 (Fig. , red).
Arrows 2 and 3 of Fig. reflect interactions that are established very early during transcription and before the processing events themselves occur, consistent with findings from previous work (36
). Thus, a 3′ splice site enhances poly(A) site cleavage on transcripts that have not been spliced (Fig. , lane 1), and a poly(A) signal enhances second-intron splicing on transcripts that have not been cleaved (band H4 in Fig. , lanes 8 to 10; see Fig. S5 in the supplemental material). The ability of the 3′ splice site to enhance poly(A) site cleavage in this coupled system shows that the splicing apparatus forges connections with the cleavage apparatus, not only prior to splicing but also prior to poly(A) site cleavage. Thus, in less than 5 min after transcription of the third exon, the 3′ splice site is recognized, the poly(A) signal is recognized, and the two establish functional connections with each other. Indeed, as pointed out in Results, the SV40 late poly(A) signal establishes connections with the 3′ splice site in less than 2 min.
Arrow 4 in Fig. refers to the impairment of second-intron splicing that results from cutting the tether of nascent RNA between the poly(A) signal and the polymerase during transcription (Fig. and ). This tether is a unique feature of the coupled state in which it physically links processing and transcription. The role of the tether has been discussed in detail previously in the context of 3′-end processing (60
), for which it was shown to facilitate assembly of the nascent 3′-end processing complex by keeping the poly(A) signal close to the polymerase. If the tether is cut prior to stable attachment of the cleavage apparatus to the polymerase, then the attachment does not occur and cleavage does not take place (60
). Similarly, in the present study, the tether apparently facilitates assembly during transcription of the exon definition complex across the third exon and, therefore, allows for efficient second-intron splicing (Fig. ).
Assembly of the exon definition complex is presumably initiated by factors associated with the 3′ splice site, such as U2AF and the U2 snRNP. U2AF65 is known to be associated with the polymerase (61
), and the two probably interact directly (69
). However, this binding appears to be weakened upon the interaction of U2AF65 with RNA (42
). In this context, the tether may serve to promote the 3′ splice site/U2AF/polymerase association while also retaining the poly(A) signal in the same vicinity (Fig. ). This would facilitate recruitment of CFIm
and CPSF to the nascent exon definition complex assembling on the polymerase (Fig. ). This scenario is consistent with various suggestions that CFIm
and CPSF are among the first of the cleavage and polyadenylation factors to be recruited to the poly(A) signal (54
). If the tether is cut before the necessary interactions have been established, then the polymerase cannot participate in the assembly of the exon definition complex, and splicing of the last intron is impaired.
This model predicts that second-intron splicing will be rescued from the adverse effects of tether cutting if a stronger poly(A) signal that can recruit the exon definition factors faster, before the tether is cut, is used. This is the result that was obtained in the experiment illustrated in Fig. (and Fig. S4B in the supplemental material) when the strong SV40 late poly(A) signal was used in place of the β-globin poly(A) signal from the experiment illustrated in Fig. . This is reminiscent of in vivo data showing that strong poly(A) signals are faster than weak poly(A) signals to establish resistance to inactivation by antisense elements in the RNA (11
). Hence, splicing gives the appearance of being mostly tether independent for strong poly(A) signals like the SV40 late signal (Fig. ) but largely tether dependent for weaker ones like the β-globin signal (Fig. ). Apparently, only about half the normal number of the slower β-globin poly(A) signals are able to contribute to the formation of an exon definition complex before the process is interrupted by the tether being cut with RNase H (Fig. ).
Arrow 5 in Fig. refers to the tether-cutting experiments illustrated in Fig. (and Fig. S3 in the supplemental material) which confirm, as described previously (60
), that the tether is required for efficient poly(A) site cleavage. More importantly, the tether-cutting experiments illustrated in Fig. indicate that the two tether-dependent steps indicated by arrows 4 and 5 in Fig. can be experimentally distinguished. This follows from the observation that although cutting the tether blocks poly(A) site cleavage (Fig. and , arrow 5) (60
), a significant proportion of the RNase H-cut RNAs continue to exhibit poly(A) signal-mediated enhancement of second-intron splicing (Fig. ; see Fig. S5 in the supplemental material), a step which was shown in separate experiments also to be tether dependent (Fig. , line 2, and 6B, arrow 4). Thus, the tether-dependent step that defines the exon and enhances splicing (Fig. , arrow 4) occurs at a distinctly earlier time than the tether-dependent step required for commitment to cleave at the poly(A) site (arrow 5). These results agree well with in vivo results showing that cleavage/polyadenylation complex assembly goes through at least two stages (11
) and that cleavage/polyadenylation is more vulnerable than splicing to cutting of the tether (6
The second of the two steps described above, commitment to poly(A) site cleavage, coincides with attachment of the cleavage/polyadenylation apparatus to the polymerase (60
). This suggests that the nascent apparatus commits to cleavage (after having previously committed to definition of the terminal exon) as a composite exon definition/cleavage/polyadenylation apparatus attached to the polymerase CTD (Fig. ). This idea is consistent with the recent demonstration that Pcf11 (a component of CFIIm
) (Fig. ) is not required for enhancement of splicing but is required for cleavage at the poly(A) site (in a transcriptionally uncoupled system) (36
). As illustrated by the RNA molecules of band H4 in Fig. , the events that enhance splicing and that do not require Pcf11 (36
) have already occurred in the immature cleavage/polyadenylation complexes whose cleavage is abrogated by cutting the tether. The CTD is an essential participant in poly(A) site cleavage (30
), and Pcf11 is a polymerase CTD-binding protein (46
). Perhaps Pcf11 is recruited late to the maturing cleavage/polyadenylation apparatus (75
) to provide the final contacts that secure this apparatus to the polymerase (60
) as part of the larger exon definition/cleavage/polyadenylation complex (Fig. ). If so, it is likely that cutting the tether impairs poly(A) site cleavage, because cutting occurs before Pcf11 has a chance to consolidate the attachment of the final mature apparatus to the CTD.