BIOLOGICAL OVERVIEW

Nucleolar protein at 60B [Nop60B, alternatively termed Minifly (mfl)] plays a central role in ribosomal RNA processing and pseudouridylation. Eukaryotic homologs include yeast Cfb5p, rat NAP57 and human dyskerin, which encodes a gene responsible for the X-linked dyskeratosis congenita disease. Nop60B/mfl genetic analysis (Phillips, 1998; Giordano, 1999; Kauffman, 2003) represents the first in vivo functional characterization of a member of this highly conserved gene family from higher eukaryotes. In addition, Nop60B hosts an intron encoded box H/ACA snoRNA gene, the first member of this class of snoRNAs identified so far from Drosophila (Giordano, 1999). Genetic, molecular, and functional characterization of Nop60B/mfl shows that the gene is essential for Drosophila viability and fertility. While P-element induced total loss-of-function mutations cause lethality, mfl partial loss-of-function mutations cause pleiotropic defects, such as extreme reduction of body size, developmental delay, hatched abdominal cuticle, and reduced female fertility. Morphological abnormalities characteristic of apoptosis are found in the ovaries, and a proportion of eggs laid by mfl mutant females degenerates during embryogenesis. mfl has an intriguing molecular organization, hosting within its intronic sequence an intron-encoded box H/ACA snoRNA that represents the first member of this class thus far described in Drosophila. This gene has been named RNA snoH1; it may be functionally equivalent to the human U70 snoRNA (Giordano, 1999). Although null mutations in Nop60B are lethal, P element-induced alleles generate viable, but sterile flies, exhibiting severe testicular atrophy. Testicular atrophy is due to severe loss of germ cells, including stem cells, but much milder effects on the somatic cells, which are themselves maintained by a stem cell lineage. Nop60B activity is required intrinsically for the maintenance of germ-line stem cells. These phenotypes can be related to the human syndrome Dyskeratosis congenita, caused by mutations in a Nop60B homolog (Kauffman, 2003).

In eukaryotic cells, synthesis, maturation and modification of rRNA take place in the nucleolus; ribonuclearprotein particles, consisting of a variety of nucleolar proteins and small nucleolar RNAs (snoRNAs), are responsible for these essential cellular processes. Almost 100 different snoRNAs species have been identified so far in yeast and mammalian cells. Most of these snoRNAs can be classified into two major distinct families, each defined by common associated proteins and by the presence of conserved sequences, designated as either C/D or H/ACA boxes. The C and D box-containing snoRNAs display extensive sequence complementarity to conserved rRNA regions and are associated with a conserved nucleolar protein, fibrillarin or, in yeast, with the fibrillarin homolog Nop1p. Some fibrillarin-associated snoRNAs are required for rRNA processing, but most of them function as a guide in site-specific ribose methylation of rRNA (Giordano, 1999 and references therein).

Members of the other large class of snoRNAs share H and ACA elements and have only short rRNA complementary motifs, brought together by a conserved stem-loop secondary structure. This structure, composed of two hairpins connected and followed by short single-stranded regions containing the H and ACA elements, directs the site-specific pseudouridylation event with the short (5-9 nucleotide [nt]) regions of snoRNA; rRNA complementarity flanks both sides of the target site (Giordano, 1999 and references therein).

In yeast, members of the box H/ACA class of snoRNAs are specifically associated with two essential nucleolar proteins, Gar1p and Cbf5p. Gar1p, a glycine-arginine-rich protein required for accumulation of mature 18S rRNA and for rRNA pseudouridylation, is thought to play a crucial role in structuring box H/ACA sno-RNPs and favoring association of H/ACA snoRNAs to the pre-rRNA. In a two hybrid yeast assay, Gar1p interacts with Cbf5p which, in turn, coprecipitates with box H/ACA snoRNAs and is required for their stability (Lafontaine, 1998). Interestingly, Cbf5p is the yeast member of a highly conserved protein family that includes homologs from at least 18 organisms. Among eukaryotes, genetic analysis has so far been restricted to two members of this family: the yeast Cbf5 (Jiang, 1993) and the DKC1 human gene, whose mutations cause the X-linked dyskeratosis congenita disease (Heiss, 1998). Whereas little information is available on human dyskerin, Cbf5p and its rat homolog, NAP57, are known to be proteins with prevalent nucleolar localization. However, whereas it has been proposed that NAP57 may be involved in nucleo-cytoplasmatic shuttling (Meier, 1994), the yeast protein has been shown to be required for transcription, processing and efficient rRNA pseudouridylation (Cadwell, 1997; Lafontaine, 1998). This last finding raises the possibility that Cbf5p might act as eukaryotic rRNA pseudouridine synthase, a role originally suggested (Cadwell, 1997) by its homology with E. coli TruB/P35 synthase. Considering the multiple, essential functions played by Cbf5p in yeast cells, the definition of the roles played by members of this family in multicellular organisms appears to be a relevant issue that deserves extensive investigation (Giordano, 1999 and references therein).

Adding to the complexity of the function of this gene is the observation that dyskerin, the human homolog of Drosophila Nop60B, is associated not only with H/ACA small nucleolar RNAs, but also with human telomerase RNA. Sufferers of dyskeratosis congenita (DKC) have defects in highly regenerative tissues such as skin and bone marrow, chromosome instability and a predisposition to develop certain types of malignancy. Telomerase is an enzyme that adds simple sequence repeats to chromosome ends using an internal region of its associated RNA (telomerase RNA) as a template; telomerase is required for the indefinite proliferation of primary human cells. Human telomerase RNA contains an H/ACA RNA motif. Primary fibroblasts and lymphoblasts from DKC-affected males are not detectably deficient in conventional H/ACA small nucleolar RNA accumulation or function; however, DKC cells have a lower level of telomerase RNA, produce lower levels of telomerase activity and have shorter telomeres than matched normal cells. The pathology of DKC is consistent with compromised telomerase function leading to a defect in telomere maintenance, which may limit the proliferative capacity of human somatic cells in epithelia and blood (Mitchell, 1999).

The developmental time at which lethality is achieved in mfl/Nop60B mutants correlates well with MFL level. Given the similarity existing between the mfl phenotype and that caused by mutations affecting the synthesis of ribosomal components, the role of mfl on rRNA processing was examined. Electrophoresis of larval rRNA labeled by [³H]uridine incorporation shows that pre-rRNA processing is inefficient in mfl mutants. In fact, with respect to wild-type flies, increased levels of the pre-rRNA and 28S rRNA and reduced amounts of the 18S, 28Sa, and 28Sb mature species were observed. MFL over-expression in mfl transgenic flies is sufficient to reduce rRNA precursor accumulation and to increase the level of the newly synthesized 18S and 28S species (Giordano, 1999).

Northern blot analysis with three different probes derived from the rDNA internal transcribed spacer (ITS) the abnormal rRNA processing occurring in mfl/Nop60B mutants to be defined in more detail. In Drosophila the rRNA primary transcript (pre-rRNA) undergoes two alternative types of initial cleavages. The most predominant type occurs in the external transcribed spacer, at site 1, and generates the large type 'a' molecule, from which both 18S and 28S are derived. An alternative cleavage occurs within ITS, at site 3, generating the intermediate 'd' and 'b' forms which are, respectively, 18S and 28S rRNA precursors. Hybridization to a probe derived from the ITS 5' end reveals that the accumulation of the pre-rRNA observed in mfl mutants is accompanied by a reduction of the type 'a' precursor and by an increase of the 'd' form; both effects become more evident with progression of the larval development. Thus, mfl mutations specifically affect site 1 cleavage, inhibiting the formation of type 'a' molecules and the processing of the 'd' intermediate. With the pathways blocked, pre-rRNA processing proceeds mainly through pathway ß, generating equimolar amounts of 'd' and 'b' intermediate molecules. This is confirmed by hybridization to probe II, which shows that while in wild-type animals the amount of form 'b' largely exceeds that of 'd' (as expected, since the 'b' molecule is actively produced by both alpha and ß pathways), in mfl mutants these two forms are detected in similar amounts. However, since the processing of form 'd' is inhibited, this species accumulates progressively along larval development. Conversely, hybridization to probe III indicates that mfl genetic depletion does not impair site 4 cleavage of type 'b' molecule, since the amount of form c observed in the mutants exceeds even that of the control. It is concluded that form c is generated properly, but its further processing is inhibited by mfl mutations. In mfl transgenic flies, MFL over-expression leads to a reversal of all of the effects observed, although the efficiency of pre-rRNA processing is not fully restored. In heat-shocked transformed animals, in fact, MFL expression causes a decrease of pre-rRNA accumulation and an increase in the production of the type 'a' molecule. Processing of the type 'a' precursor also occurs properly, since, these larvae show an excess in form 'b' versus form d, although the amount of the 'b' molecule does not reach that observed in wild-type animals. Finally, the amount of form c appears reduced after the heat-shock, indicating that its processing is at least partially restored (Giordano, 1999).

In yeast, lack of Cbf5 gene activity affects not only rRNA processing, but also rRNA pseudouridylation. Thus, the level of modification was examined in wild-type and mfl/Nop60B mutants at several 28S and 18S pseudouridylate- (psi-) specific sites. With this aim, oligonucleotide primers complementarity was used to selected 28S or 18S regions to perform primer extension analyses on CMC-treated Drosophila rRNA, in order to localize psi residues. CMC blocks reverse transcription, resulting in a gel band terminating in one residue 3' of the Psi site. In planning these experiments, advantage was taken of the known location of Drosophila 28S rRNA pseudouridines. However, none of the 18S Psi sites examined in these experiments was previously known. In spite of the persistence of maternal rRNA, pseudouridylation appears reduced in mfl⁰⁵ larvae at several 28S sites, such as the U2442, U2444, and U2499 residues. Similar reduction was observed at various 18S rRNA sites, such as U830/U831, U840, U841, and U885, indicating that, as Cbf5, mfl is required for efficient rRNA pseudouridylation (Giordano, 1999).

An unexpected feature of the mfl/Nop60B gene structure was revealed by the finding that a small RNA species, ~0.1 kb in length, hybridized specifically with the genomic sequences of the fourth mfl intron, while it was not detected by any cDNA probe. This small RNA was detected in total RNA preparations from all developmental stages and was specifically enriched in the poly(A) minus RNA fraction. The length of the small RNA species was accurately determined on denaturing 6% polyacrylamide gels and its 5' end precisely mapped by primer extension analysis of total larval RNA using two different oligonucleotides. These experiments pointed out that this transcript was ~140 nt long and derived from position +37 to about +176 of the 235-nt-long fourth mfl intron. Since a large number of small nucleolar RNAs are intron encoded, the presence of conserved snoRNA elements within the 0.14-kb RNA sequence was examined. Two H boxes (consensus ANANNA) and a 3' terminal ACA element were found; in addition, the predicted secondary structure of the mfl intron-encoded RNA conforms well to the hairpin-hinge-hairpin-tail architecture common to most yeast and vertebrate box H/ACA snoRNAs. Two short regions of complementarity between the mfl intron encoded RNA and Drosophila 18S rRNA were also found. Short regions of pairing with rRNA are known to flank the site of pseudouridylation, allowing the positioning of the residue to be isomerized at the base of the stem, at the first unpaired position before the 3' snoRNA helical segment. The pseudouridine selected is found to be separated from the H or ACA box by 14 or, in a few cases, by 15 nucleotides. On the basis of these observations, the rRNA pairing properties of the mfl intron-encoded RNA predicted it may direct pseudouridylation of Drosophila 18S rRNA at position U1820. Primer extension analysis on CMC-treated Drosophila rRNA shows that the potentially selected residue is actually pseudouridylated. The selected U1820 residue is equivalent to U1698 of human 18S rRNA, whose pseudouridylation has been related to the U70 snoRNA. As for U1698 in human rRNA, the Drosophila U1820 residue is the first of three consecutive uridines, all of which are pseudouridylated (Giordano, 1999).

In yeast, genetic depletion of most of the box H/ACA snoRNAs has been reported to inhibit pseudouridylation of the specifically selected sites. When modification was checked of the U1820 residue in rRNA preparations obtained from mfl⁰⁵ first instar larvae, pseudouridylation was found to be reduced not only at U1820, but also at U1821 and U1822 residues. This result may be explained by the widespread inhibition of rRNA pseudouridylation observed in mfl/Nop60B mutants. Further experiments are thus required to define the specific functional role, if any, played by the mfl intron-encoded RNA (Giordano, 1999).

Finally, the localization of the mfl/Nop60B intron encoded RNA was checked by in situ hybridization experiments to whole mount ovary preparations. This analysis showed that a 0.14-kb RNA-specific antisense probe exclusively labeled the nucleoli; the signal occurs in each tested embryonic or larval tissue. Specific nucleolar localization may also be observed for MFL, whose ubiquitous expression was established using both immunolocalization and histochemical staining of lacZ activity in mfl¹ flies. In ovarian tissue preparations it was noticed that the protein occasionally diffuses into the cytoplasm in several patches of follicle cells. As judged by the presence of well defined, round-shaped nuclei having morphologically well distinguishable nucleoli, these cells should not be in or around mitosis. Moreover, cytoplasmic diffusion can also be observed after stage 10b of oogenesis, when follicular cells endocycles are reported to be terminated. It is thus plausible that occasional MFL cytoplasmic localization may be related to ability to carry out nucleolus-cytoplasmic shuttling, as proposed for NAP57 in rat cells (Giordano, 1999 and references therein).

It is concluded that mfl/Nop60B encodes an ubiquitous nucleolar protein essential for Drosophila viability and female fertility. The data also show that mfl is closely related to the other members of the Cbf5 family so far characterized from higher eukaryotes, the rat Nap57 and the human gene responsible for the X-linked dyskeratosis congenita disease. As predicted (Luzzatto, 1998), flies carrying mutations in the Drosophila DKC1 ortholog show a pleiotropic phenotype very similar to that caused by mutations that affect the synthesis of ribosomal RNA. In fact, mfl loss-of-function mutations impair rRNA processing and lead to accumulation of rRNA precursors. Although these effects are very similar to those caused by Cbf5 genetic depletion, yeast mutations preferentially affect the production of mature 18S rRNA (Lafontaine, 1998), while mfl mutations cause similar reduction of 18S and 28S rRNA species. It would be of interest to know whether this is due to a distinctive feature of Drosophila rRNA processing pathways, or whether it reflects a general property of rRNA processing in higher eukaryotes (Giordano, 1999).

In addition to affecting rRNA maturation, mfl/Nop60B loss-of-function causes reduced levels of pseudouridylation at several 28S and 18S Psi sites, suggesting that gene activity might be required for fully efficient rRNA pseudouridylation. Again, these results are reminiscent of those obtained in yeast (Lafontaine, 1998), and outline the existence of a link between rRNA processing and rRNA pseudouridylation in eukaryotes. By mapping the protein domains conserved among members of the Cbf5p family and investigating the definition of their functional roles, significant information should be generated about the functional role played by rRNA pseudouridylation, which still remains elusive. Although pseudouridylation of eukaryotic rRNAs occurs predominantly on the primary rRNA transcripts before nucleolytic processing, this type of modification is not required for efficient processing of 25S yeast rRNA. It has been suggested that pseudouridylation can contribute to rRNA folding, rRNPs assembly, and ribosomal subunit assembly. Other hypotheses, such as subtle enhancing of ribosomal functions or influencing fidelity of codon recognition, have also been proposed (Giordano, 1999 and references therein).

An additional role that could be suggested for MFL/Nop60B is based on the observation that it can occasionally diffuse within the cytoplasm. As previously suggested for NAP57 in rat cells, it is tempting to speculate that this may possibly reflect the ability of MFL to structure and export pre-ribosomal RNP particles into the cytoplasm. If confirmed, this would strongly support the view that members of this family are multifunctional proteins involved in different aspects of ribosome biogenesis. It is possible that these proteins may constitute essential components of a single multifunctional complex or, alternatively, they represent common components of structurally and functionally different RNP particles. The definition of the functional interactions required to carry out such a variety of functions will help to clarify this point (Giordano, 1999).

Remarkably, the identification and the characterization of mutations disrupting mfl gene expression has led to establishing the first animal model system for the study of the X-linked dyskeratosis congenita human disease. Some of these may immediately provide useful information for the comprehension of the molecular basis of the DKC disease. A first relevant point concerns the observation that none of the mfl mutations so far isolated disrupts the gene coding region. Thus, each Drosophila mutant line has certainly quantitative and not qualitative alterations of the gene product which causes the pleiotropic abnormalities observed. The level of MFL protein was found to be critical, and a simple dose-effect rule may be derived: when the protein level is below a crucial threshold, mortality ensues. Instead, while the protein level is lowered but still stands above a critical threshold, the viable, hypomorphic mfl¹ phenotype is reached. By analogy, it can be suggested that in humans the level of dyskerin activity may be one of the critical parameters able to trigger the DKC disease. The finding that DKC mutations mapped so far all affect the dyskerin coding region (Heiss, 1998) is in only apparent contrast with that found in Drosophila. In fact, it is reasonable to suppose that, as observed in Drosophila, total or severe loss-of-function mutations should not be compatible with life. Mutations recovered in patients might be those causing partial loss-of-function, so that the level of dyskerin activity is still compatible with survival. Accordingly, DKC patients might carry hypomorphic mutations, the human counterparts of the viable mfl¹ phenotype. Whether these hypomorphic phenotypes are simply a consequence of the inadequate mature rRNA level or are, at least partially, caused by abnormal accumulation of intermediate rRNA species is an important point which deserves further investigation. A further issue concerns the observation that, although MFL and dyskerin are ubiquitous proteins, phenotypic abnormalities are, in Drosophila as in humans, restricted to only certain tissues. Since the gene product is presumed to be critically important for protein synthesis in every cell of the body, the finding that abnormalities are developed only by selected cell types is quite surprising. However, if it is accepted that the level of protein activity may be a critical parameter, then it is reasonable to suppose that the amount of properly processed rRNA may be sufficient in cells having a slow growth rate, while in highly proliferating tissues or in cells sustaining a high rate of protein synthesis this would not the case, and degenerative cell defects could progressively be accumulated. Interestingly, inhibition of protein synthesis is known to be one of the stimuli capable of inducing apoptotic cell death, probably by decreasing the levels of essential proteins or by inhibiting the synthesis of proteins that normally suppress the spontaneous activation of apoptosis. In mfl¹ ovary, one of the Drosophila tissues where morphological abnormalities can be observed, it was found that degeneration is specifically accompanied by apoptotic cell death. This observation might also suggest a role for apoptosis in the progressive clinical manifestations of the DKC disease (Giordano, 1999).

Finally, it has been shown that mfl gene organization is intriguing; this led to the identification of the first member of the box H/ACA class of snoRNAs described so far in Drosophila. As in the case of the snoH1 gene described here, most of the snoRNAs are intron encoded, and snoRNA host genes often encode proteins involved in translation or ribosome biogenesis (reviewed by Smith, 1997). These intron-encoded snoRNAs are cotranscribed with their host pre-mRNA and their accumulation is splicing-dependent, since they are released from the excised intron by exonucleolytic processing. The observation that snoH1 RNA and mfl mRNAs levels are reduced in parallel in each mfl mutant line strongly suggests that snoH1 RNA processing is linked to the splicing of the mfl primary transcript. This feature, which allows coordinated regulation of the host protein and the intron encoded snoRNA, may hinder a precise definition of the specific functional role played by each product. With regard to snoH1, it cannot be excluded, in principle, that it may be required for Drosophila viability and that its depletion might contribute to the generation of mfl phenotype. However, snoH1 has little, if any, effect on mfl phenotypic rescue when over-expressed in mfl transgenic flies, either in the presence or in the absence of MFL overexpression. This is not surprising, given that each box H/ACA snoRNAs found in yeast, with the exception of snR30, is dispensable for viability. It will now be interesting to determine whether this type of gene organization is restricted to mfl or is shared by other members of this conserved gene family (Giordano, 1999).

GENE STRUCTURE

Genomic probes have identified on Northern blots two main transcripts of 1.8 and 2.0 kb in length. While the 1.8-kb species is constitutively expressed throughout the life cycle, the 2.0-kb RNA was specifically found in adult female and embryonic RNA preparations, in which a further transcript of ~4.0 kb is also occasionally detected. However, no cDNA representative of this mRNA subform was isolated after extensive screening of an adult female cDNA library, so that it remains unclear whether it actually derives from the mfl gene. In contrast, several cDNAs representative of the 1.8 and 2.0 kb were isolated from adult female and larval libraries. The longest cDNAs of each class, respectively, of 1,833 and 2,034 bp, including the poly(A) tail, represent almost full-length transcripts and allow the mfl gene structure to be defined by Southern blot hybridization and alignment with nucleotide sequence of the genomic region. In each mfl mutant line, a copy of P was inserted at the 5' common end of 1.8- and 2.0-kb transcription units: in mfl⁰⁶, the insertion site was mapped 18 nt upstream from the 5' end of the longest cDNAs obtained, in mfl¹ 18 nt downstream, within the 5' leader sequence, while in mfl⁰⁵ the insertion occurred within the first intron of the gene. The 1.8- and 2.0-kb mfl mRNA subforms share a common coding region and differ from each other only at their alternatively spliced 3' untranslated region, where two additional exons (7 and 8) are specifically included in the 2.0-kb mRNA (Mitchell, 1999).

PROTEIN STRUCTURE

Amino Acids - 508

Structural Domains

The Cbf5 protein of Saccharomyces cerevisiae was originally identified as a low-affinity centromeric DNA-binding protein, and chf5 mutants have a defect in rRNA synthesis. A closely related protein from mammals, NAP57, is a nucleolar protein that coimmunoprecipitates with the nucleolar phosphoprotein Nopp140. To study the function of this protein family in a higher eukaryote that is amenable to genetic approaches, the gene encoding a Drosophila melanogaster homolog, Nop60B, was identified. The predicted Drosophila protein shares a high degree of sequence identity over a 380-residue region with both the mammalian and yeast proteins, and shares several conserved motifs with the prokaryotic tRNA pseudouridine 55 synthases (Phillips, 1998).

The mfl open reading frame (ORF), identically present in both mRNA subforms, encodes a predicted protein of 508 amino acids with a calculated molecular mass of 56 kD. Database searches reveal that this protein belongs to the Cbf5p/NAP57/dyskerin family. The MFL polypeptide shows a significant degree of conservation to other members of the family, particularly with the two very similar rat and human proteins (66% identity, 79% similarity to human dyskerin). The conservation increases remarkably within several specific domains, strongly underlining that their function has been preserved during evolution. Total identity exists among Drosophila and human proteins within the two TruB motifs which have homology with bacterial and yeast tRNA pseudouridine synthases (Heiss, 1998). A repeated hydrophobic domain, possibly involved in the nucleo-cytoplasmatic shuttling postulated for the rat protein is also highly conserved. This domain is immediately followed by a block of >20 amino acids having a central tyr motif that is identical in Drosophila, rat, and human proteins. Although no function has been suggested so far for this domain, its conservation suggests that it might play a relevant role in protein activity. Within the tyr domain, there is a RX-x(2,3)-DE-x(2,3)-Y central core motif highly conserved among uracil-DNA glycosylases from different organisms as part of the rigid uracil-binding pocket present in these repair enzymes. Within the pocket, the tyrosine residue has been shown to be directly involved in uracil recognition (Kavli, 1996; Slupphaug, 1996). By analogy, it is reasonable to suggest that the highly conserved tyr motif might play a similar role in uracil recognition. A highly charged lysine-rich COOH-terminal region containing a nuclear localization signal is found in MFL, as in NAP57 and dyskerin, and the NH₂-terminal nuclear localization signal observed in rat and human proteins is also preserved. Finally, it is interesting to note that all five missense mutations thus far identified in DKC patients fall into regions that are conserved between the human and the Drosophila gene (Giordano, 1999).

date revised: 30 March 2003

Home page: The Interactive Fly © 1995, 1996 Thomas B. Brody, Ph.D.

The Interactive Fly resides on the
Society for Developmental Biology's Web server.