The Pre-mRNA Splicing Machinery of Trypanosomes: Complex or Simplified?

Abstract

Trypanosomatids are early-diverged, protistan parasites of which Trypanosoma brucei, Trypanosoma cruzi, and several species of Leishmania cause severe, often lethal diseases in humans. To better combat these parasites, their molecular biology has been a research focus for more than 3 decades, and the discovery of spliced leader (SL) trans splicing in T. brucei established a key difference between parasites and hosts. In SL trans splicing, the capped 5′-terminal region of the small nuclear SL RNA is fused onto the 5′ end of each mRNA. This process, in conjunction with polyadenylation, generates individual mRNAs from polycistronic precursors and creates functional mRNA by providing the cap structure. The reaction is a two-step transesterification process analogous to intron removal by cis splicing which, in trypanosomatids, is confined to very few pre-mRNAs. Both types of pre-mRNA splicing are carried out by the spliceosome, consisting of five U-rich small nuclear RNAs (U snRNAs) and, in humans, up to ∼170 different proteins. While trypanosomatids possess a full set of spliceosomal U snRNAs, only a few splicing factors were identified by standard genome annotation because trypanosomatid amino acid sequences are among the most divergent in the eukaryotic kingdom. This review focuses on recent progress made in the characterization of the splicing factor repertoire in T. brucei, achieved by tandem affinity purification of splicing complexes, by systematic analysis of proteins containing RNA recognition motifs, and by mining the genome database. In addition, recent findings about functional differences between trypanosome and human pre-mRNA splicing factors are discussed.

Trypanosomatids are protistan parasites infecting hosts as diverse as mammals, insects, and plants. In humans, vector-borne Trypanosoma brucei, Trypanosoma cruzi, and Leishmania spp. cause lethal diseases, and the strong impact of these parasites on global health has spurred investigations of the molecular processes in these organisms from early on. One of the first key discoveries in regard to gene expression was spliced leader (SL) trans splicing, which was eventually found to be an essential maturation step for all nuclear pre-mRNA in trypanosomatids.

The initial discoveries of SL trans splicing were made in T. brucei, and until now, this organism has remained the preferred trypanosomatid organism for spliceosomal studies. T. brucei is an extracellular parasite which evades the mammalian immune system by antigenic variation of its variant surface glycoprotein (VSG) coat. VSG expression has therefore been a research focus, and it was on VSG mRNAs that the 5′-terminal region was first discovered to contain a leader sequence which was not encoded in the VSG gene (10, 87). Further analysis showed that the 39-nucleotide (nt)-long leader was derived from the 5′ terminus of a separate, small nuclear RNA, which has been termed SL RNA or miniexon-derived RNA (12, 34, 56). The discovery of a Y structure intermediate which corresponds to the cis splicing intron-exon-lariat structure (Fig. 1) strongly indicated that the SL transfer functions analogously to intron removal, entailing the same two transesterification reactions (58, 79). This notion was confirmed by the demonstration that the destruction of spliceosomal uridine-rich small nuclear RNAs (U snRNAs) blocked SL transfer (84).

Fig. 1.

Schematic of the mammalian cis splicing and the trypanosome SL trans splicing reactions. Upstream exon and spliced leader are drawn as gray rectangles, and downstream exon and trypanosome gene are drawn as black rectangles. 5′ and 3′ splice sites (SSs) are represented by small open boxes, branch points (BPs) by closed circles, polypyrimidine tracts by small striped boxes, and the cap 4 structure of the spliced leader as an oval. Conserved sequences are provided below the drawing with invariant residues underlined. While in mammalian systems, 5′SSs, BPs, and 3′SSs exhibit partly conserved sequences (R, purine; Y, pyrimidine; N, any base), there is no obvious sequence conservation at trypanosome BPs (43) and 3′SSs, although it was shown for the latter that an AC dinucleotide (*) preceding the AG residues drastically reduces splicing efficiency unless a compensatory AG dinucleotide is present within the 5′ untranslated region (76). It appears that the importance of the polypyrimidine tract becomes more important when consensus sequences are lacking. Yeast has highly conserved splice site and BP sequences, and some yeast introns function without a polypyrimidine tract (not shown). The partly conserved sequences in mammals require a small polypyrimidine tract in the range of 10 to 12 residues (Y_10-12), whereas in trypanosomes, the polypyrimidine tract is large (Y_∼20), is an essential sequence determinant for efficient splicing, and typically starts just downstream of the BP (43, 76). After the first transesterification reaction, cis splicing results in a lariat intron structure, whereas a Y structure intermediate is formed in the SL trans splicing process. After the second transesterification, these intronic structures are debranched (not shown) and rapidly degraded.

SL trans splicing is not restricted to VSG mRNA but is an essential maturation step for all trypanosomatid mRNAs. In trypanosomatid genomes, coding genes are tandemly arranged in large polygenic clusters which are transcribed in a polycistronic fashion (7). Trans splicing and polyadenylation lead to precursor cleavages up- and downstream of a coding region, respectively, and therefore, are mechanistically required to process individual mRNAs from polycistronic pre-mRNA. Moreover, the SL carries a 7-methylguanosine (m⁷G) cap and the first four nucleotides of its sequence are methylated; some of these methylations are unique to trypanosomes, and the unusual 5′-terminal structure has been termed cap 4 (4). Since cap 4 is transferred onto mRNA 5′ ends as part of the SL, trans splicing represents a posttranscriptional mode of mRNA capping and, therefore, is essential in the formation of functional mRNA.

Since all trypanosomatid mRNAs are trans spliced and trypanosomatid genes typically do not harbor introns, it was long thought that these organisms use RNA splicing exclusively for SL transfer, and accordingly, trypanosome-specific deviations of splicing factors were hypothesized to be trans splicing specific. It therefore came as a surprise when the T. brucei PAP gene (TriTryp database [TriTrypDB] accession no. Tb927.3.3160), encoding poly(A) polymerase, was shown to harbor a single intron that was removed by conventional cis splicing (48). The search for further introns revealed only one more gene in T. brucei (Tb927.8.1510), encoding a putative RNA helicase (7), whose pre-mRNA was shown to be cis spliced (30). Interestingly, a recent characterization of the T. brucei transcriptome by high-throughput RNA sequencing strongly indicates that there are no other introns disrupting protein-coding genes (75).

SL trans splicing is a more widespread phenomenon in eukaryotes, and after its initial discovery in trypanosomes, it was found to occur in a variety of organisms, including euglenids (80), nematodes (35), trematodes (70), and even lower chordates, such as the sea squirt (86). However, there is no indication that this particular mode of trans splicing occurs in the hosts of trypanosomatid parasites, e.g., vertebrates or arthropods (19), and therefore, it can be regarded as a parasite-specific process. This specificity and the ubiquitous requirement of SL trans splicing for mRNA maturation have made this process an attractive research focus. The long-term aims have been to find out how the trypanosome splicing machinery differs from its human counterpart, to identify factors or factor domains which are specifically required for the trans splicing process, and to analyze whether these features can be inactivated in a parasite-specific manner. The challenge of this research is that the splicing machinery, termed the spliceosome, is a huge, dynamic complex composed of structural RNAs and proteins that is difficult to characterize.

The spliceosome consists of the U1, U2, U4, U5, and U6 snRNAs and, in the human system, up to 170 spliceosome-associated protein factors (91). Trypanosomatids possess all five spliceosomal U snRNAs, which are typically somewhat smaller and deviate in several aspects from their human counterparts. In contrast to the well-characterized human system, until recently, our knowledge of spliceosomal protein factors in trypanosomatids was very limited. A main reason for this lack of knowledge comes from the fact that amino acid sequences of trypanosomatid proteins have diverged dramatically from their human and yeast orthologs, and thus, only a few splicing factors were identified by standard annotation of the completed L. major, Trypanosoma brucei, and Trypanosoma cruzi (TriTryp) genomes (29). In recent years, however, major progress has been made in the identification of spliceosomal proteins and the characterization of U small nuclear ribonucleoprotein particles (U snRNPs) in trypanosomes. Three factors have contributed to this success: first, the unrestricted access to the sequenced and annotated TriTryp genome databases (7) at GeneDB (http://www.genedb.org/) and, recently, also at TriTrypDB (http://tritrypdb.org/) (2); second, the systematic analysis of RNA binding proteins harboring an RNA recognition motif (RRM) (16); and third, tandem affinity purification (TAP) of splicing complexes combined with mass spectrometric identification of copurified proteins (46, 63). A current list of identified splicing factors is presented in Table 1.

Table 1.

Spliceosomal proteins of Trypanosoma brucei^g

Annotationa	Accession no.b	Mr (103)	TAPc	Description	Reference(s) or source	E valued
Sm/LSm proteins
SmB	Tb927.2.4540	12.3	1, 2, 3, 4		65
SmD1	Tb927.7.3120	11.7	1, 2		65
SmD2	Tb927.2.5850	12.5	1, 2		65
SmD3	Tb927.4.890	12.4	1, 2, 3, 4		65
SmE	Tb927.6.2700	9.6	1, 2		65
SmF	Tb09.211.1695	8.4	1, 2		65
SmG	Tb11.01.5915	8.9	1.2		65
SSm2-1/Sm15K	Tb927.6.4340	12.8	1, 2		83, 92
SSm2-2/Sm16.5K	Tb927.10.4950	14.7	1, 2		83, 92
SSm4	Tb927.7.6380	23.2	1, 2		83
LSm2	Tb927.8.5180	13.2	1, 2		46, 82
LSm3	Tb927.7.7380	10.1			41
LSm4	Tb11.01.5535	14.2	1, 2		41
LSm5	Not assignede	12.0			82
LSm6	Tb09.160.2150	9.1			41
LSm7	Tb927.5.4030	10.2	1, 2		41
LSm8	Tb927.3.1780	14.0	1, 2		41
SMN/Gemin2 and associated proteins
SMN	Tb11.01.6640	17.0	1, 2, 3, 4		63
Gemin2	Tb927.10.5640	55.4	1, 2, 3, 4		63
Coatomer α	Tb927.4.450	132.0	3, 4		47
Coatomer β	Tb927.1.2570	110.0	3, 4		47
Coatomer β′	Tb927.2.6050	93.9	3, 4		47
Coatomer γ	Tb11.01.3740	97.5	3		47
Coatomer δ	Tb927.8.5250	57.3	3, 4		47
Coatomer ε	Tb11.01.6530	34.8	3, 4		47
Coatomer ζ	Tb927.10.4270	20.5	3		47
U1 proteins
U1-70K	Tb927.8.4830	31.7	1, 2		66
U1A	Tb927.10.8280/8300	18.0	1		46
U1-24K	Tb927.3.1090	24.2	1, 2		66
U1C	Tb927.10.2120	21.7	1, 2		66
U2 proteins
U2A′ (U2-40K)	Tb927.10.2120	36.5	1, 2		15
U2B″	Tb927.3.3480	13.6	1, 2		69
SF3a60	Tb927.6.3160	61.5			TriTrypDB	1e−13
SF3b(SAP)155	Tb11.01.3690	122.0			3, 50
SF3b(SAP)145	Tb927.6.2000	52.5			50
SF3b(SAP)130	Tb927.7.6980	195.0	1		50
SF3b(SAP)49	Tb927.3.5280	29.8			50
SF3b(SAP)14b	Tb927.10.7390	12.7			50	6e−12
SF3b(SAP)10	Tb09.211.2205	10.4		SF3b10 domain	Pfam DB; 50	4e−7
SF3b14 (p14)	Tb927.10.7470	13.3			3, 50
U4 proteins
PRP3	Tb09.160.2900	63.2	1, 2	PRP3 domain	Pfam DB	2e−42
PRP4	Tb927.10.960	65.5	1, 2		46
Snu13f	Tb09.160.3670	13.6				5e−34
U5 proteins
PRP8	Tb09.211.2420	277.0	1, 2, 4		44
U5-200Kf	Tb927.5.2290	249.3	1			≈0
U5-102K	Tb11.01.7330	111.0	1			1e−5
U5-116K	Tb11.01.7080	105.5	1, 2			8e−100
U5-40K	Tb11.01.2940	35.0	1, 2		46
U5-15K	Tb927.8.2560	17.7	1			4e−26
U5-Cwc21	Tb09.160.2110	16.2	1, 2		46
PRP19 complex
PRP19	Tb927.2.5240	54.3	1, 3, 4			3e−42
CDC5f	Tb927.5.2060	80.1				1e−30
CRN/SYF3	Tb927.10.9660	87.7	1			7e−41
SYF1	Tb927.5.1340	92.2	1			7e−23
ISY1	Tb927.8.1930	31.7	1	ISY1 domain	Pfam DB	6e−7
KIAA1604/Cwc22	Tb11.01.2520	66.82	2			2e−49
Unannotated proteins that copurified in spliceosomal complexesh
Conserved hypothetical	Tb927.8.6280	27.1	1, 2	Putative cyclophilin	63
Conserved hypothetical	Tb927.8.2090	21.6	2	Putative cyclophilin	63
Conserved hypothetical	Tb927.10.11950	22.4	2	Putative Cwc15	63
Conserved hypothetical	Tb927.8.4790	26.0	1	Novel
Conserved hypothetical	Tb927.2.3400	42.0	1	Novel
Conserved hypothetical	Tb927.7.1890	31.0	1	Novel
Conserved hypothetical	Tb927.5.2910	20.0	1	Novel
Conserved hypothetical	Tb11.02.0465	12.1	1	Novel
Annotated proteins without known splicing function that copurified in spliceosomal complexes
eEF-1α	Tb927.10.2100	49.1	2
HSP70	Tb11.01.3110	75.4	1, 2, 3
Importin α	Tb927.6.2640	58.0	2
La protein	Tb927.10.2370	37.7	1, 2, 3
NORF1	Tb927.5.2140	93.3	1
PABP1	Tb09.211.2150	62.1	1, 2
TRYP1	Tb09.160.4250/80	22.4	1
Miscellaneous splicing factors
U2AF35	Tb927.10.3200	29.1			89
U2AF65	Tb927.10.3500	96.6			90
SF1	Tb927.10.9400	31.6			90
PRP17	Tb927.3.1930	52.8	2			4e−59
PRP31	Tb927.10.10700	39.7			40
PRP43	Tb927.5.1150	82.9			40
PTB1	Tb09.211.0560	37.0			78
PTB2	Tb11.01.5690	54.7			78
TSR1	Tb927.8.900	37.5		SR-like protein	28
RRM1	Tb927.2.4710	50.0		SR-like protein	51
SR protein	Tb09.160.5020	17.6			TriTrypDB	3e−06

^aAnnotation is according to the human system. Conserved hypotheticals are proteins which are conserved among trypanosomatids but dissimilar to proteins of other eukaryotes.

^bAccession numbers are from TriTrypDB (http://www.tritrypdb.org/).

^cProteins were cotandem affinity purified with SmD1 (1), SmB (2), SMN (3), or Gemin2 (4).

^dProtein sequences of identifications without experimental support were compared to the human genome and their E values determined by NCBI BLAST.

^eThe gene of LSm5 has not yet been recognized as a protein-coding gene.

^fThese genes were annotated in this study.

^gProteins shown in boldface are trypanosome specific. The accession numbers of eight putative spliceosomal DExD/H-box helicases are Tb927.6.4600, Tb927.6.4600, Tb927.10.5280, Tb927.10.7280, Tb927.10.9130, Tb11.02.3460, Tb927.7.7300, and Tb11.02.1930. More than 20 candidate putative spliceosomal peptidyl-prolyl cis/trans isomerases are not shown.

^hE values lower than 1e⁻⁰⁵ were considered not significant.

In an excellent previous review on trypanosomatid RNA splicing, Liang et al. described the discoveries and functional characterizations of the trypanosome spliceosomal U snRNAs and early characterizations of the corresponding snRNPs (39). This review omits a general discussion of the U snRNAs and instead focuses on proteins involved in splicing.

U snRNPS AND THE SPLICEOSOME IN HIGHER EUKARYOTES

Our mechanistic insight into RNA splicing and our biochemical and structural knowledge of snRNPs, splicing factors, and the spliceosome stem predominantly from work in the human and yeast systems. Unfortunately, there is no uniform nomenclature for the protein factors in the two systems, and typically, there are two distinct names for orthologous factors (listed in reference 31). In this review, the default is the denotation from the human system.

The main building blocks of the spliceosome are the U snRNPs, whose biogenesis requires several distinct assembly steps. First, all spliceosomal U snRNAs, except U6, are exported to the cytoplasm where they bind a set of seven common proteins, known as the Sm proteins B, D1, D2, D3, E, F, and G. These proteins form a heteromeric ring around a conserved Sm binding site that resides in a single-stranded region in the 3′-terminal domain of the U snRNA. This RNA-protein interaction is typically very stable, and thus, the U snRNA/Sm complex is referred to as the core snRNP. Core snRNP assembly takes place in the cytoplasm and is linked to U snRNA cap hypermethylation, which in turn codetermines the reimport of the core snRNP into the nucleus. The U6 snRNA does not have a cytoplasmic phase and, in the nucleus, binds a different complex of seven Sm-like (LSm) proteins termed LSm2 to -8 (LSm2-8). Back in the nucleus, the core snRNPs bind various snRNP-specific proteins, and overall, there are approximately 45 different proteins in the human system that interact directly with the spliceosomal U snRNAs (91). As follows, most of the spliceosome-associated proteins are considered to be non-snRNP proteins.

The RNA sequence determinants for the splicing reaction comprise the 5′ splice site (5′SS) and the 3′SS and a branch point (BP) sequence upstream of the 3′SS. In addition, a polypyrimidine tract is typically present between BP and the 3′SS (Fig. 1). Importantly, the spliceosome is assembled step-by-step on the pre-mRNA, and before and during splicing, it undergoes highly dynamic changes in which both the RNA and protein composition are altered (reference 91 and references therein). In brief, the U1 snRNP first recognizes the 5′SS, the protein factor SF1 the branch point, and the heterodimeric U2 auxiliary factor (U2AF) both the branch point and 3′SS. Subsequently, the U2 snRNP is recruited and the U2 snRNA forms base pairs with the BP sequence, displacing SF1, a process which is mediated by the U2-associated protein complexes SF3a and SF3b. At this stage, the factor assembly is called the prespliceosome or complex A. Subsequently, the U4/U6 snRNP and the U5 snRNP enter the spliceosome in the form of the U4/U6.U5 tri-snRNP, which results in the precatalytic complex B. Although in this complex, all snRNPs are on board, the spliceosome undergoes major rearrangements for activation (complex B′), including the discard of U1 and U4 snRNPs. After the first transesterification, the spliceosome is transformed into complex C, and following the second splicing step, it is disassembled. The snRNPs are then recycled for new rounds of splicing. The different spliceosomal complexes have been purified and biochemically characterized in the yeast and human systems. Besides the above-described snRNP changes, these complexes are associated with distinct sets of proteins (Fig. 2) (reviewed in references 31 and 91).

Fig. 2.

Comparison of known spliceosomal factors of humans and trypanosomes. Schematic drawing of spliceosomal complexes during a splicing reaction as described in the mammalian and yeast systems. For each complex, proteins are listed that enter the spliceosome at the outlined stage (slightly modified human protein repertoire is according to reference 91). Please note that only incoming proteins are listed and proteins leaving the spliceosome in the transitions are not recognized. Bold blue lettering indicates proteins for which orthologs have been found in trypanosomes, whereas red lettering specifies trypanosome-specific factors. ¹, Highly divergent, putative cyclophilin orthologs have been copurified with SmD1 and SmB1 (Table 1); ², U5-100K is a DExD/H-box helicase, and it is unclear whether one of the putative trypanosome DExD/H-box helicases (Table 1) represents a U5-100K ortholog; ³, U5-Cwc21 is possibly the ortholog of human SRM300 but seems to have a trypanosome-specific function (see text); ⁴, the trypanosome exon junction complex has recently been characterized (6), but its specific function in RNA splicing or metabolism remains unclear.

Is the spliceosome different for SL trans splicing? Interestingly, as first shown for the nematode Caenorhabditis elegans, the SL RNA splicing substrate itself is assembled into a core snRNP binding the Sm proteins (11). This finding led to the hypothesis that the SL RNP activates its own splice site and that trans splicing does not require the U1 snRNP (81, 88). Indeed, in vitro studies of the parasitic nematode Ascaris lumbricoides showed that the destruction of U1 snRNA affected only cis and not trans splicing (27). Moreover, in the same system, two specific SL RNP proteins were identified and termed SL175 and SL30 according to their molecular masses (17). The results of protein-protein interaction experiments suggested that these proteins bridge the SL RNP and, thus, the 5′SS of the SL RNA, via SF1 to the BP, a function which in cis splicing is mediated by the U1-specific FBP11/Prp40p (human/yeast nomenclature) subunit (17). SL175 and SL30 are indispensable for SL trans splicing, but they have no function in cis splicing or in an SL-independent mode of trans splicing (17) which has also been described in the human system (reviewed in reference 22). While these factors and their interactions are potential antiparasitic targets, the amino acid sequences of these proteins are not conserved, and putative orthologs have not been identified outside nematodes.

TRYPANOSOME Sm AND LSm PROTEINS AND Sm CORE VARIATION IN U2 AND U4 snRNPs

Sm and LSm proteins are small proteins with a molecular mass typically of ∼10 to 20 kDa that share a highly conserved bipartite Sm motif. The corresponding Sm fold characteristically consists of an N-terminal helix and a strongly bent, antiparallel beta-sheet of five strands. While antibodies directed against the Sm domain of human proteins recognize Sm proteins in a wide range of organisms, they did not cross-react with trypanosome proteins (55, 61, 62). Hence, it required affinity purification of U snRNPs and protein analysis to show that trypanosome U snRNAs and the SL RNA bind a set of common proteins (62). The identity of five of these proteins was revealed in the classic way: U snRNPs were affinity purified, amino acid sequence information from common proteins was obtained by protein microsequencing, and the respective genes were cloned with the help of degenerate primers and PCR (65). In the same study, the missing SmB and SmD3 orthologs, however, could already be identified by mining the growing T. brucei genome database (65). Later, this was the exclusive route to identify the orthologs of LSm2-8 (41). However, while the Sm motifs were readily identifiable in all these proteins, the remaining amino acid sequences exhibited limited similarity to their putative orthologs in other eukaryotes and, therefore, needed functional verification. In the case of the Sm proteins, SmG was shown to complement an SmG-deficient yeast strain (65), whereas the others exhibited protein-protein interactions which were consistent with the known arrangement in the ring structure (63, 65). For the LSm proteins, only LSm8 and LSm3 were functionally analyzed at first, and all others were identified by sequence similarity alone (41). This approach backfired because LSm2 and LSm5 turned out to be very interesting Sm proteins (see below) but not LSm proteins. Eventually, a second study clarified the trypanosome LSm repertoire, identified new LSm2 and LSm5 orthologs, and provided strong evidence through functional studies that the correct set of LSm proteins was identified (82). The formation of core snRNPs stabilizes the U snRNAs, and expression silencing of a single Sm or LSm protein leads to a loss of the cognate U snRNA. Accordingly, conditional RNA interference (RNAi) experiments targeting each of the seven LSm proteins resulted in a specific loss of U6 snRNA, confirming the new identifications (41, 46, 82).

In initial studies of trypanosome snRNPs, it was found that the trypanosome U2 core snRNP, in contrast to its human counterpart and other trypanosome U snRNPs, disassembled completely in a cesium chloride density gradient (14) and in high-salt buffers (25). While this instability with exposure to salt was originally attributed to a deviating U2 Sm binding site, which in T. brucei contains an unusual central guanosine residue, it was found only recently that the different core includes the Sm complex as well. U2 snRNP purification revealed two U2-specific proteins with apparent sizes of 15 and 16.5 kDa that contained the bipartite Sm motif (92). This was odd because the whole Sm repertoire had already been characterized and, moreover, Sm15K had at that time been identified as LSm5. However, a comprehensive tagging and coimmunoprecipitation analysis clarified the issue and showed that Sm15K and Sm16.5K are paralogs of SmB and SmD3, respectively; they replace these proteins specifically in the U2 Sm core and do not bind other U snRNAs. Furthermore, snRNP reconstitution assays with recombinant Sm proteins and synthetic RNAs demonstrated that the guanosine residue of the U2 Sm binding site is the recognition determinant of the U2-specific Sm core complex (92). In an independent study, the identification of the two U2-specific Sm paralogs was confirmed and the previously misannotated LSm2 was shown to be a second SmD3 paralog that replaces SmD3 in the U4 snRNP (83). Importantly, the study by Tkacz et al. (83) provided an in vivo analysis demonstrating that RNAi-mediated expression silencing of the specific Sm paralogs reduced the abundance of only the corresponding U snRNA. The U4-specific Sm core was subsequently characterized at the biochemical level, verifying the U4 association of the SmD3 paralog (30). Unfortunately, the studies on Sm core variation established different nomenclatures, and Sm15K is also referred to as specific spliceosomal Sm2-1 protein (SSm2-1), Sm16.5K as SSm2-2, and the U4-specific protein as SSm4.

What is the significance of Sm core variation? It has been speculated that it assists U2 and U4 snRNP assembly (83, 92), and indeed, this has recently been demonstrated for the U2 snRNP. Earlier it was found that stable protein binding to the 3′-terminal region of U2 snRNA, which included the Sm binding site, was dependent on residues in the 3′-terminal loop IV sequence. This suggested that the U2 core snRNP was not formed by Sm protein binding alone but required cooperative binding of Sm and loop IV-binding proteins (25). This model was recently verified by the demonstration that the U2-specific Sm15K/Sm16.5K doublet interacts with the U2 snRNP protein U2A′ which in turn interacts with the loop IV-binding protein U2B″ (69). Only this ternary complex efficiently and specifically interacts with the 3′-terminal U2 snRNA region. In the human system, U2A′ is separated from the Sm core by stem-loop III. Since this structure is completely missing in trypanosome U2 snRNA, it is likely that the Sm15K/Sm16.5K-U2A′ interaction occurs through an essential, parasite-specific protein-protein interface that compensates for the lack of this RNA structure.

Furthermore, it was suggested that Sm core variation may facilitate snRNP function in the splicing process (92). For example, human and yeast U2 snRNAs share a conserved motif which is complementary to the BP sequence. Conversely, in trypanosomes, there is no conserved BP sequence and, typically, no complementarity between BP and U2 snRNA sequences (43). Possibly the U2-specific Sm paralogs undergo specific protein-protein interactions that position the U2 snRNP at the BP in the absence of sequence complementarity. A third speculation stated that the different Sm cores may be connected to different U snRNA cap structures (83). In vertebrate and yeast systems, the U6 snRNA carries a γ-monomethyl phosphate cap, whereas U1, U2, U4, and U5 snRNAs obtain cotranscriptionally an m⁷G cap which is further methylated to a 2,7,7 trimethylguanosine (m₃G) cap after the formation of the core snRNP. It was shown that the binding of the Sm proteins to the U snRNAs is a prerequisite for the recruitment of the enzyme trimethylguanosine synthase 1 (54, 67). In trypanosomes, U1, U2, U4, and U6 snRNAs have the same caps as their yeast and human orthologs (18, 57, 64), whereas SL RNA has cap 4 (4) and U5 snRNA lacks a cap (20, 95). Theoretically, Sm core variation could facilitate the recruitment of different cap-modifying enzymes into the core RNP, but there is no correlation between the type of cap and the type of Sm core. For example, U1, U2, and U4 snRNAs bind different Sm complexes but share the same m₃G cap. Moreover, it was demonstrated experimentally for the T. brucei U2 snRNA that cap trimethylation does not depend on the presence of the Sm binding site or on formation of the core RNP, thus excluding the possibility that the U2-specific Sm core is involved in cap hypermethylation (24).

SMN-MEDIATED ASSEMBLY OF CANONICAL Sm CORES

When Wang et al. (92) reconstituted core snRNPs with recombinantly expressed Sm proteins, they detected specific binding of the canonical and U2-specific Sm cores to their cognate Sm binding sites only with short RNA fragments, whereas full-length U snRNAs did not discriminate between the two Sm complexes. This suggested that other activities in the cell confer specificity of Sm core binding. A candidate for such an activity was the SMN (survival motor neuron) complex, which in the human system was shown to act as a catalyst for core snRNP formation (reviewed in references 33 and 59). The human SMN complex consists of the SMN protein and seven additional subunits, termed Gemin2 to -8, and it binds the SmD1/SmD2-SmE/SmF/SmG and SmD3/SmB subcomplexes in an open ring formation (13). This interaction then leads to U snRNA binding, ring closure, and dissociation of the SMN complex. While standard annotation of the Tritryp genomes did not identify SMN or Gemin homologs, tandem affinity purification of the T. brucei SmB protein (see below) led to the identification of highly divergent orthologs of SMN and Gemin2 (63). No other Gemin orthologs were found, and it is possible that they do not exist in trypanosomes, because lower eukaryotes in general have a strongly reduced Gemin repertoire and, in Drosophila melanogaster, the SMN/Gemin2 complex was sufficient to mediate core RNP assembly in vitro (36). In trypanosomes, in vitro core snRNP assembly experiments functioned efficiently in the absence of SMN, but when the factor was added it exhibited a striking discriminatory role: in its presence the canonical Sm core was efficiently loaded onto its cognate U5 snRNA but not onto U4 and U2 snRNAs or a U5 snRNA with a mutated Sm site (63). In contrast, the SMN complex had no effect on the binding of the U2-specific Sm core. These findings suggested that the SMN complex specifically bound the canonical SmD3/SmB subcomplex and directly interacted with SmB, because this protein, in contrast to SmD3, is replaced in both the U2- and U4-specific Sm cores. This was indeed the case. SMN purification coisolated only the SmB and SmD3 proteins and not their paralogs, and pulldown assays with recombinant, tagged SMN proteins identified a direct interaction with SmB and the N-terminal part of SMN (63). The latter finding, again, is highly significant because it identified an important, potentially parasite-specific protein-protein interaction: human SMN utilizes an internal Tudor domain and C-terminal regions to interact with dimethylated arginines in the RG-rich C-termini of Sm proteins (73), whereas in trypanosomes, neither Tudor domain nor RG-rich domains are present in SMN and Sm proteins, respectively (63). Another striking difference from the human system was found in regard to SMN localization. In the human system, core snRNP assembly takes place in the cytoplasm and, accordingly, human SMN is primarily localized in this compartment. Conversely, trypanosome SMN was found almost exclusively in the nucleus, suggesting that U snRNP assembly in this organism is nuclear, a finding which is consistent with localizations of SL RNA and U2 snRNA by fluorescence in situ hybridization (8, 83). In summary, it appears that despite its small size, the trypanosome SMN complex is mechanistically complex, entailing both chaperone and specificity functions in core snRNP assembly.

If the SMN complex only chaperones the assembly of the canonical Sm core, how, then, are the U2- and U4-specific Sm cores put together? One possibility is that they require a different, yet-to-be-determined assembly chaperone. Alternatively, the specific interactions of the U2 Sm paralogs SmK15/SmK16.5 with U2A′/U2B" may facilitate correct assembly of the U2-specific Sm complex onto the U2 Sm binding site. On the other hand, U4 core snRNP formation appears to be independent of snRNP-specific proteins, because in core RNP reconstitution assays, SSm4 alone determined efficient and specific assembly of the U4-specific Sm complex onto the U4 snRNA (30).

TANDEM AFFINITY PURIFICATION OF SPLICING COMPLEXES IN T. BRUCEI

Until recently, only two snRNP-specific proteins had been studied in trypanosomes, namely, the orthologs of human U2A′ (originally termed U2-40K) (15) and the U5-specific PRP8 (44). While the latter was identified by sequence homology, U2A′ was copurified with the U2 snRNA in high-stringency U snRNP purifications which typically left only the core structures intact (62). For a more comprehensive biochemical characterization of U snRNPs and/or of the spliceosome, it was therefore essential to purify the RNA-protein complexes under conditions of lower stringency. A method well-suited for this purpose is tandem affinity purification (TAP), which is based on expressing a known protein factor fused to a composite TAP tag. TAP comprises two consecutive high-affinity chromatography steps which are carried out under nearly physiological conditions. Since the advent of this technology (71), the TAP tag and the TAP method have been modified in various ways to accommodate different systems, extracts, or protein complexes (26). For the purification of nuclear protein complexes in trypanosomes, the PTP (protein C epitope-TEV protease cleavage site-protein A domains) modification of TAP has proven to be very useful (26, 72). One of the first applications of the PTP tag was the purification of the trypanosome U1 snRNP. A first characterization of this snRNP had revealed a protein with sequence homology to the human U1-70K protein (64). And indeed, PTP tagging and purification of T. brucei U1-70K resulted in the specific copurification of the U1 snRNA (66). The protein profile of the purification comprised the Sm proteins, the tagged protein, two additional proteins of major abundance, and several proteins of minor abundance. The proteins of the two major bands were identified by mass spectrometry and found to be annotated as “conserved hypotheticals,” meaning that they were conserved among trypanosomatids but had no obvious similarity to proteins of other eukaryotes. However, when kinetoplastid sequences of one of the new proteins were compared to those of known U1-specific proteins of model organisms, the protein was identified as the ortholog of human U1C (66). In contrast, the second protein, termed U1-24K, could not be meaningfully aligned to known U1 proteins and, therefore, probably represents a trypanosome-specific U1 snRNP subunit.

Since this initial tandem affinity purification of a snRNP was successful, more comprehensive proteomic analyses of trypanosomal splicing complexes were carried out by PTP tagging and purification of the common proteins SmD1 (46) and SmB (63). Overall, mass spectrometry identified 53 proteins in these purifications, and the majority of the proteins copurified in both studies (Table 1). Moreover, with the exception of three LSm proteins and two non-snRNP proteins, all known trypanosomal snRNP proteins were identified in these proteomic analyses, confirming the high significance of the proteomic data sets. Consequently, bioinformatic analyses of the amino acid sequences of unannotated proteins revealed several new orthologs of known splicing factors (Table 1), and for LSm2 (U6), U1A, PRP4 (U4), and U5-40K, the bioinformatic identifications were confirmed by coimmunoprecipitation experiments which showed that these proteins were bound to their predicted snRNAs (46).

While these proteomic analyses increased the number of spliceosomal protein orthologs in trypanosomes considerably and identified potentially novel splicing factors (see below), the number of proteins that copurified with SmD1 or SmB is ∼3-fold lower than the protein count in human spliceosomes. Proteomics of yeast spliceosomal complexes revealed about 90 proteins (21), which is lower than the count in the human system but still about 2-fold higher than the identified trypanosome repertoire. One possible interpretation of this finding is that the splicing machinery of trypanosomatids that diverged early is simplified. However, this is unlikely because the vast majority of newly identified proteins are snRNP proteins, and non-snRNP proteins are strongly underrepresented (Fig. 2). In fact, trypanosome orthologs have been identified for nearly all known bona fide snRNP proteins, indicating that a trypanosome spliceosome comprises additional non-snRNP proteins, possibly in numbers comparable to those in yeast and humans. Hence, the question arises of why only a few non-snRNP proteins copurified with SmB and SmD1. Both proteomics studies were carried out according to the standard PTP protocol, including the extract preparation procedure (37, 72). Since these extracts contain an estimated overall salt concentration of 250 to 300 mM, it is likely that the spliceosome did not withstand the extract preparation procedure. Accordingly, a sucrose gradient sedimentation analysis of SmD1-PTP-purified material showed that the U snRNPs did not cosediment as part of a larger complex, and complexes with Svedberg values greater than 20 were not detected (46). In contrast, spliceosomal 45S complexes were characterized by a combination of glycerol gradient sedimentation and native gel electrophoresis in extracts of lower salt concentration (40). It is therefore likely that modifying the extract preparation procedure will result in the formation of higher-order spliceosomal complexes which possibly can be isolated by tandem affinity purification and characterized by mass spectrometry in the future.

As discussed above, in both proteomic studies, the highly divergent SMN and Gemin2 orthologs copurified. To better understand the trypanosome SMN complex, Palfi et al. (63) PTP tagged and tandem affinity purified both proteins and identified copurified proteins by mass spectrometry. While no other Gemin proteins were detected, which supports the idea that an SMN/Gemin2 complex is sufficient for chaperone function, surprisingly, all coatomer subunits copurified. While the coatomer complex functions in vesicular transport between Golgi apparatus and endoplasmic reticulum (47), the significance of the coatomer-SMN/Gemin2 interaction is not understood. Possibly, the trypanosome SMN/Gemin2 complex has a cytoplasmic function independent of snRNP core assembly (63), or there is a cytoplasmic component of the core snRNP assembly process which is vesicular and has not yet been detected.

ANALYSIS OF PROTEINS CARRYING AN RRM

Besides by tandem affinity purification, trypanosome splicing factors have been identified through a focus on RRM-containing proteins. Since trypanosomatid protein coding genes are typically arranged in long tandem gene arrays which are transcribed polycistronically, differential gene expression is typically regulated posttranscriptionally, for example, at the level of RNA stability. Many proteins which affect RNA stability bind to mRNAs directly by virtue of an RRM motif and thus, RRM-containing proteins have become a research focus in gene expression studies of both T. brucei and Trypanosoma cruzi (16). Since the spliceosome comprises several RRM proteins, their identification came as a benefit from the attempt to determine the role of RRMs in the regulation of gene expression.

One RRM protein that was identified as a splicing factor was a subunit of the U2-associated SF3b complex. SF3a and SF3b are two essential multisubunit splicing factors that interact with the U2 snRNP after its recognition of the BP. The trypanosome protein was identified as the ortholog of human SF3b49, and accordingly, expression silencing of the corresponding gene was lethal and affected RNA splicing in T. brucei (50). Moreover, TAP tagging and purification of the protein, using the original TAP method, led to the complete characterization of the trypanosome SF3b complex. Orthologs of all seven human SF3b subunits were identified, including the RRM protein SF3b14, often referred to as p14 (50). The SF3a complex appears to also be present in trypanosomatids, because a putative homolog of the SF3a60 subunit was annotated in the genome database (Table 1).

Other RRM proteins that were found to be splicing factors comprise the snRNP protein U1A (46) and the U2 auxiliary factor components U2AF65 and U2AF35 (89, 90). In addition, RRM-containing serine-arginine-rich (SR) proteins have been identified. SR proteins comprise a phylogenetically conserved protein family and, as has been shown in other systems, play significant roles in constitutive and alternative splicing of pre-mRNA (reviewed in reference 42). SR proteins contain one or two N-terminal RRMs and a C-terminal RS domain, rich in arginine-serine dipeptides. The first such protein discovered in trypanosomes was termed RRM1 (51). While RRM1 was shown to be encoded by an essential gene and located in the nucleus, its specific function has not yet been determined (52). A second SR protein, termed trypanosomal SR-rich protein 1 (TSR1), was localized to the nucleus and shown to bind to the heterologous human U2AF complex, and in a yeast three-hybrid system, it appeared to interact with the SL RNA (28). While these findings led to the speculation that TSR1 may facilitate recognition of the SL RNA by the trans spliceosome (28), a functional characterization of TSR1 strongly indicated that the factor has an essential role in cis splicing but not in SL trans splicing (Christian Tschudi, Yale University, personal communication). This result is in accordance with the results of a previous study of the T. cruzi ortholog TcSR which showed that TcSR was functional in cis splicing in a heterologous system (68). Finally, RRM protein analysis in trypanosomes revealed two homologs (PTB1 and PTB2) of the mammalian polypyrimidine tract binding protein. While mammalian PTB did not copurify with spliceosomal complexes and has several nonsplicing functions, it negatively affects the splicing process, presumably by binding to the polypyrimidine tract near the 3′SS, thereby interfering with U2AF65 function (77). Functional characterization of trypanosome PTB1 and PTB2 did not reveal a repressor function of these proteins in splicing. In contrast, a detailed study provided very strong evidence that both proteins are essential for trans splicing of pre-mRNAs that contain C-rich polypyrimidine tracts (78). In addition, expression silencing of PTB1 but not of PTB2 affected cis splicing, indicating that both proteins have distinct activating functions in trypanosome RNA splicing (78).

BIOINFORMATIC IDENTIFICATION OF TRYPANOSOME SPLICING FACTORS

A third route to identify RNA splicing factors has been data mining. Some of the splicing factors in trypanosomes are conserved enough to be identified by in silico analysis alone. For example, it was straightforward to identify the missing orthologs of the human snRNP proteins Snu13 and U5-200K for this study (Table 1). Similarly, the important CDC5 subunit of the PRP19 complex, which is an essential component of the active spliceosome, was readily identifiable in the trypanosome genome database (Table 1). Two splicing factors which had previously been identified bioinformatically are PRP43 and PRP31 (40). PRP43 is a conserved spliceosomal helicase with essential functions in intron lariat release from the spliceosome (53) and in spliceosome disassembly (1), whereas PRP31 is a factor of the U4/U6.U5 tri-snRNP that is important for tri-snRNP formation and assembly into the spliceosome (49). The trypanosome PRP31 and PRP43 appear to be functionally equivalent, because both proteins were shown to be essential for both cis and SL trans splicing, and PRP31 was specifically associated with the trypanosome U4/U6.U5 tri-snRNP (40). Nevertheless, the bioinformatics route of identifying trypanosome splicing proteins has not been exploited extensively and it is very likely that a systematic approach will reveal additional (putative) orthologs of non-snRNP proteins.

Overall, our knowledge of the spliceosomal protein repertoire of trypanosomes has greatly increased in the past years. While the set of snRNP proteins appears to be nearly complete, most of the non-snRNP factors have probably not been identified yet. However, the identification of individual components of spliceosomal protein complexes, such as PRP19 and SF3a (Fig. 2), indicate that these complexes are present and that they can be further analyzed. For example, in yeast, more than 20 splicing proteins were identified by tandem affinity purification of the CDC5 ortholog Cef1p (reviewed in reference 31). The identification of CDC5 in this study will enable a comparable analysis in trypanosomes.

Another important aspect of the newly identified trypanosome splicing factors is that they strongly indicate that trypanosomes form a spliceosome that possesses the same basic components and undergoes the same dynamic rearrangements as its human and yeast counterparts. It should be kept in mind that, with the exception of a 45S spliceosome detection by native gel electrophoresis (40), there is so far no biochemical evidence that trypanosomes do form complexes that correspond to the well-characterized spliceosome E, A, B, or C complexes in the yeast and human systems. On the other hand, a comparison of human proteins that enter the spliceosome at these defined stages and of the known trypanosome repertoire shows that for each spliceosome transition, characteristic trypanosome orthologs have been identified (Fig. 2).

TRYPANOSOME-SPECIFIC ASPECTS OF THE SPLICEOSOME

Despite the rapidly increasing number of identified trypanosomal RNA splicing factors, only a few functional protein characterizations have been published thus far. Nevertheless, several trypanosome-specific characteristics of the splicing machinery have already been identified. As discussed above, Sm core variation, including the trypanosome-specific interaction between SmB and SMN, as well as the particular architecture of the U2 RNP core, involving potentially unique interactions between Sm15K/Sm16.5K and U2A′, are trypanosome-specific U snRNP features. Other notable differences, shown in the T. cruzi system, include the demonstration that the U2AF subunits U2AF35 and U2AF65 exhibit weak or no interaction (90), that instead, U2AF65 forms a stable complex with the BP binding protein SF1 (90), and that, within the U2-related SF3b complex, the protein interface between the SF3b155 and SF3b14 subunits appears to be larger and more complex than in the human system (3).

Another interesting trypanosome splicing factor is U5-Cwc21. This protein shares a highly conserved N terminus with the human SRm300/SRRM2 protein and yeast Cwc21p (complexed with Cef1p protein 21). Coimmunoprecipitation analysis showed that the trypanosome protein is predominantly associated with U5 snRNA, and expression silencing of U5-Cwc21 was lethal and affected both cis and trans splicing (46). In contrast, yeast Cwc21p and human SRm300 have redundant, nonessential roles in RNA splicing, because CWC21 is a nonessential gene and SRm300 can be immunodepleted from extract without affecting splicing efficiency in vitro (9). Moreover, while yeast Cwc21p does interact with the U5-protein PRP8 (23), it is predominantly associated with U2 snRNA and not with U5 snRNA (32). These findings therefore strongly indicate that trypanosome U5-Cwc21 has an essential function in RNA splicing that is unique to trypanosomes.

A further peculiarity of trypanosome splicing factors is the expression level of U1 snRNP components. In the nematode system, the U1 snRNP functions exclusively in cis splicing and not in SL trans splicing (27). If the trypanosome U1 snRNP functions analogously, it would be required only for the removal of a single intron from two different pre-mRNAs. However, the U1-specific proteins U1-70K, U1-24K, and U1C are among the most abundant proteins that copurified with SmD1 (46). This discrepancy between intron number and U1 snRNP expression level suggests that the trypanosome U1 snRNP has functions beyond intron removal. There is evidence that the trypanosome 45S spliceosome contains both SL and U1 snRNA, and it was suggested that there may only be one kind of spliceosome for both cis and trans splicing (40). If this is true, the U1 snRNP may be essential for spliceosome integrity or it may have a yet-undetected, trypanosome-specific function in trans splicing. Alternatively, the trypanosome U1 snRNP, like its human counterpart, may function beyond intron removal in transcription initiation and/or elongation (reviewed in reference 5).

Finally, there seems to be a difference in SL RNP recruitment to the BP in nematode and trypanosome systems. While the nematode SL RNP apparently docks on SF1 via a protein bridge (17), immunoprecipitation of trypanosome SF1 under low-stringency conditions did not coprecipitate SL RNA (D. L. Ambrósio and A. Günzl, unpublished results). It is therefore likely that other proteins and protein-protein interactions than in the nematode system mediate the recruitment of the trypanosome SL RNP. Possibly, U5 and U6 snRNPs play a role in this process, because U5 and U6 snRNAs were convincingly shown to interact with the 5′SS of the SL RNA (93, 96). If SL RNP recruitment to pre-mRNA requires trans splicing-specific factors, as in the nematode system, potential candidates for such proteins are listed in Table 1; there are currently eight proteins which copurified with trypanosomal splicing complexes but could not be annotated convincingly. Three of these proteins may be the orthologs of cyclophylins and of the PRP19-related factor Cwc15, although the sequence similarities are very weak. The remaining five proteins are novel in sequence because they do not exhibit any sequence similarity to nontrypanosomatid proteins.

PERSPECTIVES

The spliceosome is one of the most complex molecular machineries in the cell, and it is a great challenge to functionally characterize this dynamic RNP-protein machinery. In recent years, major progress has been made in the biochemical and structural analysis of the human spliceosome (45, 91). If corresponding studies can be carried out in trypanosomes, it will be possible to determine in detail essential differences between trypanosome and human spliceosomes. While such differences may be the consequence of evolutionary divergence or may represent SL trans splicing-specific requirements, they are potential antiparasitic drug targets. This notion is not remote since the spliceosome has been validated as a drug target, for example, for anticancer treatment (reviewed in reference 85). Although the undertaking of comprehensively analyzing the trypanosome spliceosome appears overwhelming, the prospects are nevertheless good because all necessary tools are in place. As shown in Fig. 2, there are now several new spliceosomal proteins which can serve as bait in tandem affinity purification to broadly characterize the trypanosome splicing factor repertoire. The conditional RNAi-based expression-silencing system in T. brucei (94), in combination with established reverse transcription-PCR and primer extension assays for the analysis of trans and cis splicing defects, provides an in vivo platform for determining the splicing functions of individual proteins. Moreover, a homologous in vitro trans splicing system was recently established in T. brucei which will allow the functional dissection of important splicing factors (74). Finally, the recent demonstration that the tandem affinity-purified trypanosome transcription factor complex TFIIH was sufficiently intact and pure to determine its molecular structure by macromolecular electron microscopy (38) strongly indicates that similar structures can be obtained from tandem affinity-purified splicing complexes.

And there is another potentially exciting perspective. While introns and alternative splicing greatly enhance the protein repertoire in higher eukaryotes (recently reviewed in reference 60), the functional role of introns in lower eukaryotes is not well understood. Trypanosomes appear to have reduced their intron repertoire to only two (75). Why did they not eliminate these two introns as well? The fact that the insertion site of the PAP intron is conserved in trypanosomatids argues that the intron has a specific and essential function which was retained throughout trypanosomatid evolution. Since it should be straightforward to test the outcome of [conditionally] deleting these intron sequences in the trypanosome genome, it may be possible to determine the specific function of these introns and understand the functional significance of cis splicing in these early-diverged organisms.