INTRODUCTION
Since the discovery of the nucleolus, many works have focud on understanding the function of this dynamic nuclear structure. Ribosome biogenesis is the major function known to be associated with the nucleolus (Hadjiolov, 1985), although recent work suggest that this nuclear compartment might be involved in other cellular process (Pederson, 1998).
The nucleolus is the site of ribosomal RNA (rRNA)transcription, rRNA modification, maturation and asmbly with ribosomal proteins to form the pre-ribosomal particles that are then exported to the cytoplasm to form the mature ribosomes. Ribosome biogenesis is probably one of the most complex pathways of ribonucleoparticle synthesis, involving a complex rRNA maturation and an ordered asmbly of about 80 ribosomal proteins in eukaryotes. In addition to the component found in the mature ribosomes, the nucleolus contains many other RNAs and proteins, most of which transiently associate with pre-ribosomal particles and appear to be involved in various aspects of ribosome biogenesis (transcription, maturation, modification and ribosome asmbly). In this review we focus on one of the major and most studied nucleolar proteins, nucleolin, who multiple functions in ribosome biogenesis are beginning to be characterized.
Properties of the different domains of nucleolin and ‘nucleolin-like proteins’
Nucleolin was first described by Orrick et al. (1973) and was initially called C23 becau of its mobility on a two-dimensional gel (Prestayko et al., 1974). The same protein was then described and purified from Chine Hamster Ovary cells (Bugler et al., 1982) and veral other eukaryotic cells. The name nucleolin is now widely ud for this nucleolar protein,which can reprent as much as 10% of total nucleolar protein in CHO (Bugler et al., 1982). This protein is also often described as a 100-110 kDa protein; however, cloning of the cDNA of hamster nucleolin revealed that it contained 713amino acids, giving ri to a predicted molecular mass of 77kDa (Lapeyre et al., 1987, 1985). This discrepancy was latter attributed to the amino acid composition of the N-terminal domain of nucleolin. Homologous proteins were then identified in man (Srivastava et al., 1989), rat (Bourbon et al., 1988),mou (Bourbon et al., 1988), chicken (Maridor and Nigg,1990) and Xenopus laevis (Caizergues-Ferrer et al., 1989;Rankin et al., 1993). Nucleolin is highly phosphorylated (Olson et al., 1975; Rao et al., 1982) and methylated (Lischwe et al., 1982), and could also be ADP-ribosylated (Leitinger and Wesierska-Gadek, 1993).
The human nucleolin gene is prent at one copy per haploid genome. The human gene consists of 14 exons and 13 introns on chromosome 2q12-qter (Srivastava et al., 1990). The four acidic stretches of the N-terminal domain lie within exons 2 to 4 and the nuclear localization signal (NLS) is
encoded in exon 5. Interestingly, each of the four RNA-binding domains is encoded by two independent and concutive exons. For the four RNA-binding domains, the intron position is similar and interrupts the conrved RNP1 quence of each RNA-binding domain, suggesting that the RNA-binding domains are duplications of the same ancestor RBD domain. This genomic
761
Journal of Cell Science 112, 761-772 (1999)
Printed in Great Britain ©The Company of Biologists Limited 1999JCS5011
Nucleolin is an abundant protein of the nucleolus.Nucleolar proteins structurally related to nucleolin are found in organisms ranging from yeast to plants and mammals. The association of veral structural domains in nucleolin allows the interaction of nucleolin with different proteins and RNA quences. Nucleolin has been implicated in chromatin structure, rDNA transcription,rRNA maturation, ribosome asmbly and nucleo-cytoplasmic transport. Studies of nucleolin over the last 25years have revealed a fascinating role for nucleolin in ribosome biogenesis. The involvement of nucleolin at multiple steps of this biosynthetic pathway suggests that it could play a key role in this highly integrated process.
Key words: Nucleolin, Ribosome biogenesis, Nucleolus, Ribosomal RNA, RNA-binding protein
SUMMARY
COMMENT ARY
Structure and functions of nucleolin
Hervé Ginisty*, Hélène Sicard*, Benoit Roger and Philippe Bouvet ‡
Laboratoire de Biologie Moléculaire Eucaryote, Institut de Biologie Cellulaire et de Génétique du CNRS, UPR 9006, 118 route de Narbonne, 31062 Toulou Cedex, France
*The authors contributed equally to this work
‡Author for correspondence at prent address: CNSR-IPBS, UPR 90062, 205 route de Narbonne, 31077 T oulou, France (E-mail: Bouvet@ipbs.fr).
Published on WWW 25 February 1999
762
organization is extremely well conrved in hamster, mou and rat (Bourbon and Amalric, 1990; Bourbon et al., 1988). Another intriguing feature of nucleolin gene organization (in tetrapod and vertebrates) is that two snoARNs (U20 and U23 snoRNA) are encoded in introns 11 and 12, respectively (Nicoloso et al., 1994; L. H. Qu and J. P. Bachellerie, personal communication). U20 and U23 belong to the two large families of snoRNAs, the box C/D antin snoRNAs and the H/ACA snoRNAs, respectively, which guide the two major types of rRNA nucleotide modifications, namely ribo methylation and pudouridine formation (Bachellerie and Cavaillé, 1998; Smith and Steitz, 1997). In vertebrates, most modification-guiding snoRNAs are intron-encoded and procesd from pre-mRNA introns, and their host genes are usually directly involved in ribosome biogenesis (Bachellerie and Cavaillé, 1998; Bachellerie et al., 1995).
Analysis of the amino acid quence of nucleolin reveals the prence of three different structural domains (Fig. 1). The N-terminal domain is made up of highly acidic regions intersperd with basic quences and contains multiple phosphorylation sites. The central domain contains four RNA-binding domains called RBD or RRM. The C-terminal domain called GAR or RGG domain is rich in glycine, arginine and phenylalanine residues. This domain contains high levels of N G,N G-dimethylarginines (Lischwe et al., 1982).
Several nucleolar proteins of different eukaryotic species exhibit a similar tripartite structural organization (Table 1 and e legend of Fig. 1 for details). In this review, we will call the proteins ‘nucleolin-like proteins’. This does not imply that they are all orthologs of nucleolin in the different species; however, bad on what is known of nucleolin’s properties, we can propo that all the proteins are multifunctional nucleolar proteins involved in different aspects of ribosome biogenesis.
N-terminal domain
The length of this domain is very variable among the different nucleolin-like proteins (Fig. 1). The prence of highly acidic regions parated from each other by basic quences is one feature of this domain. Yeast gar2 and Nsr1p N-terminal domains have long stretches of rine residues in the acidic regions. Acidic domains have been propod to bind histone H1 (Erard et al., 1988), and could be responsible for a displacement of H1 from its interaction with linker DNA. This interaction with histone H1 would induce chromatin decondensation (Erard et al., 1988). The acidic stretches also determine the Ag-NOR stainability of nucleolin (Rousl et al., 1992). The prence of the basic and repeated octapeptide motifs (XTPXKKXX, X being a non-polar residue) bears strong similarity to an analogous quence of histone H1 (Erard et al., 1990). The motifs could be responsible for the capacity of nucleolin to modulate DNA condensation in chromatin (Erard et al., 1990; Olson and Tho
mpson, 1983). In addition to its interaction with chromatin and histone H1, the N-terminal domain of nucleolin has been involved in many protein-protein interaction, such as with the U3 snoRNP (Ginisty et al., 1998) and some ribosomal proteins (Bouvet et al., 1998; Sicard et al., 1998).
The N-terminal domain of nucleolin is highly phosphorylated (Bourbon et al., 1983; Mamrack et al., 1979; Rao et al., 1982). Nucleolin is a substrate for veral kinas,including cain kina II (CK2) (Caizergues-Ferrer et al., 1987), p34cdc2(Belenguer et al., 1989; Peter et al., 1990) and protein kina C-ζ(Zhou et al., 1997). CK2 co-purifies with nucleolin (Caizergues-Ferrer et al., 1987), and recent work indicates that α or α′subunits of CK2 interact directly with nucleolin (Li et al., 1996). CK2 phosphorylates nucleolin in vitro and in vivo at rine residues found predominantly in two highly acidic regions (Caizergues-Ferrer et al., 1987). p34cdc2 phosphorylation occurs on threonine residues within the basic TPXKK repeat (Belenguer et al., 1989; Peter et al., 1990). All potential sites for p34cdc2kina are not ud with the same efficiency in vivo as in vitro (Belenguer et al., 1989). Phosphorylation of nucleolin by CK2 and p34cdc2is highly regulated during the cell cycle (Belenguer et al., 1989; Peter et al., 1990). Extensive phosphorylation by CK2 occurs in interpha and by p34cdc2in mitosis, and this regulated phosphorylation of nucleolin probably regulates nucleolin function during the cell cycle.
In plants and yeast, phosphorylation sites in the N-terminal domain have been diverly conrved. Nucleolin-like proteins from plants (Bogre et al., 1996; de Carcer et al., 1997) and yeast (Gulli et al., 1997) exhibit connsus CK2 phosphorylation sites, or are highly phosphorylated by this kina. Plant nucleolin-like proteins are also phosphorylated by p34cdc2. Although S. pombe gar2 is most likely phosphorylated by p34cdc2in mitosis on a single rine residue (Gulli et al., 1997), a gar2 protein with this rine changed into a non-phosphorylated alanine is able to complement perfectly a gar2-
H. Ginisty and others
法兰西第二帝国
Fig. 1.Schematic reprentation of the organization of three nucleolin
and ‘nucleolin-like proteins’. RBDs are defined from the β1 and β4
strand (Kenan et al., 1991). (A) Organization of hamster nucleolin:
mou, rat chicken, human and Xenopus laevis nucleolin have the
same organization. (B) Alfalfa and pea ‘nucleolin-like protein’ have
the same organization. (C) Organization of gar2. Nsr1p posss
only 3 long acidic stretches. N-, N-terminal; C-, C-terminal.
763 Nucleolin structure and functions
null strain (Gulli et al., 1997). In S. cerevisiae, Nsr1p lacks connsus p34cdc2phosphorylation sites, so it was propod that p34cdc2phosphorylation had become uless in cells where mitosis is not accompanied by nucleolar disasmbly/ reasmbly.
The RNA-binding domains
Nucleolins from hamster, mou, rat, human, chicken and Xenopus laevis posss four RNA-binding domains called RBDs, also known as RRM (RNA recognition motif) (e Fig.
1 and Table 1). The domains, found in a large number of proteins implicated in various functions (Burd and Dreyfuss, 1994), are known to confer an RNA-binding specificity to the protein they belong to (Nagai, 1996). For example, one of the RBDs found in the U1A protein is responsible for the RNA-binding specificity of the U1A protein towards two structurally different RNA targets (Jovine et al., 1996). No structural data are yet available for nucleolin RBDs. One interesting feature of nucleolin RBDs is that they are less conrved within the same protein than between RBDs from divergent species (Table 2). For example, RBDs 1 and 4 are less than 10% identical within the same protein. According to their quence identity, nucleolins and nucleolin-like proteins can be divided in three groups. In the first group is found nucleolin from hamster, mou, rat, human, chicken, Xenopus laevis and fish. In the cond group are the yeast proteins gar
2 and Nsr1p, and in the third group the plant proteins NucMs1 and nucleolin from pea. Within each of the groups, the RBDs are very similar. Mou and human nucleolin RNA-binding specificity is beginning to be well characterized, but there is very little information on the RNA-binding properties for the other nucleolin-like proteins.
Even before the identification of RBDs in nucleolin, it was known that nucleolin interacted with nucleic acids (Herrera and Olson, 1986; Olson et al., 1983). The high concentration of nucleolin in the den fibrillar region of the nucleolus (Escande et al., 1985; Lischwe et al., 1981) suggested that nucleolin could be associated with rRNA. It was latter shown that nucleolin associated with nascent pre-ribosomal RNA (Herrera and Olson, 1986), and could even be detected on Christmas trees (Ghisolfi-Nieto et al., 1996). Nucleolin binds with high affinity and specificity with rRNA fragments from the 5′ETS (Ghisolfiet al., 1992b; Serin et al., 1996). A SELEX experiment performed with hamster nucleolin purified from CHO cells identified an 18-nt RNA connsus motif (Ghisolfi-Nieto et al., 1996). Sequence comparison between this RNA motif and quences of the rRNA 5′ETS previously identified as nucleolin binding sites, in addition to extensive mutagenesis, allowed the detailed characterization of one nucleolin binding site (Ghisolfi-Nieto et al., 1996; Serin et al., 1996). A small stem-loop structure compod of a short stem (5 ba pairs) and a 7-10 nt loop containing the motif U/G CCCGA forms a minimal RNA-binding site (NRE; Nucleolin Recognition Element). The binding affinity (K d) of nucleolin for this RNA target can vary from 5 nM for the SELEX quence to 100 nM for a natural quence found in rRNA (Ghisolfi-Nieto et al., 1996). The factors responsible for this binding affinity difference are not known.
Taken parately, none of the four individual RBDs interact significantly with this RNA target, but a peptide that contains the first two RBDs (RBD 12) is sufficient to account for nucleolin RNA-binding specificity and affinity towards this RNA target (Serin et al., 1997). A SELEX experiment
Table 1. Characteristics of nucleolin and ‘nucleolin-like proteins’ in veral organisms
Molecular
mass Number of
Organism Name(kDa)*RBDs pI References Remarks
Hamster Nucleolin77.0 (100)4 4.49Lapeyre et al. (1985)
Mou Nucleolin76.7 (105)4 4.48Bourbon et al. (1988)
Human Nucleolin76.3 (100) 4 4.39Srivastava et al. (1989)
Rat Nucleolin77.1 (110)4 4.46Bourbon and Amalric (1990)
Chicken Nucleolin75.64 4.68Maridor et al. (1990)
Xenopus laevis Nucleolin75.52 (95) 4 4.58Rankin et al. (1993);Two different forms.
70.20 (90)4 4.64Caizergues-Ferrer et al. (1989)Difference esntially in the N-terminal
domain
Alfalfa NucMs167.1 (95)2 4.62Bogre et al. (1996)Higher expression in root meristematic
cells
Pea Nucleolin64.8 (90)2 4.71Tong et al. (1997)Expression is induced by red light
Onion Nop64A (64)2ND de Carcer et al. (1997)Two differents forms. Only Nop61A is Nop61A(61)2phosphorylated. cDNA not characterized Arabidopsis thaliana FMV3bp57.92 4.43Didier and Klee (1992)Very short C-terminal domain with only
钢琴入门指法教程one RGG motif. N-terminal acidic
stretches are very short
Tetrahymena thermophila Nopp5251.7 (52)2 6.16McGrath et al. (1997)Only one RGG motif in C-ter
minal domain S. pombe gar253.02 4.82Gulli et al. (1995)gar2−strain is cold-nsitive
S. cerevisiae Nsr1p44.5 (67)2 4.65Lee et al. (1992)Overexpression of Nsr1p is lethal. Nsr1−
strain is cold-nsitive.
成人英语怎么说*Apparent molecular mass are given in parenthes.
ND, not determined.
764
performed with this recombinant RBD 12 protein lected exactly the same 18-nt motif as full-length nucleolin (G. Serin and P. Bouvet, unpublished result), demonstrating that the two domains were sufficient for this interaction. A genetic study of this RBD12-NRE interaction provided more details on this interaction (Bouvet et al., 1997). The constraints provided by the studies allowed us to propo a three-dimensional model of this interaction. In this model, both domains participate in a joint interaction with the NRE using a different surface to contact the RNA. The high-resolution determination of the structure of this RNA protein complex is now currently being investigated by NMR and X-ray crystallography. Human, mou and hamster nucleolin interact with the same bindin
g affinity and specificity with a mou 5′ETS RNA fragment that contains a NRE motif. A similar human 5′ETS RNA fragment is also efficiently recognized by mou nucleolin (Serin et al., 1996). This conrved binding specificity for RBD 12 of human, mou and hamster is in agreement with the high percentage of identity of this domain between the different species (Table 2). Interestingly, Xenopus laevis RBD 1 and 2, which are only 53% identical to the hamster protein, do not interact significantly with the NRE motif (G. Serin and P. Bouvet, unpublished data).
The prence of veral RBDs in nucleolin suggests that nucleolin is potentially able to interact with multiple RNA targets. Indeed, studies on nucleolin function in pre-rRNA processing (Ginisty et al., 1998) identified an RNA target within the mou 5′ETS, that was distinct from the NRE motif. The RNA quences and nucleolin RBDs involved in this interaction remain to be determined. RNA quences containing the telomeric repeat UUAG (Ishikawa et al., 1993), quences in the 3′UTR of the amyloid protein precursor (Zaidi and Malter, 1995) and in the 3′noncoding region of poliovirus (Waggoner and Sarnow, 1998) have also been described as human nucleolin binding sites. However, little information on binding affinity, specificity or mechanism of nucleolin interaction with the RNA motifs, or the biological relevance of the interactions, is available.The C-terminal GAR/RGG domain
This domain is defined as spaced Arg-Gly-Gly (RGG) repeats intersperd with amino acids, which are often aromatics. It ems that the motif RGGF is particularly frequent in nucleolar proteins. The length of the GAR/RGG domain is variable among nucleolins, with its quence and arrangement of the repeats not well conrved. For example, plant nucleolin-like proteins have a longer GAR domain than mammalian nucleolins (Bogre et al., 1996). The prence of this domain in a protein is associated with the prence of an RNA-binding domain (RBD or others) (Burd and Dreyfuss, 1994). Structural studies of this domain indicate that it can adopt repeated β-turns (Ghisolfiet al., 1992a). The GAR domain of hamster nucleolin interacts nonspecifically with RNA, leading to unstacking and unfolding of this RNA (Ghisolfiet al., 1992a). The prence of this GAR domain does not influence the binding affinity and specificity for the NRE quence (Ghisolfi-Nieto et al., 1996; Serin et al., 1997), but one function of this domain could be to facilitate the interaction of nucleolin RBD domains with targets located within large and complex RNA, such as rRNA (Ghisolfiet al., 1992b; Heine et al., 1993). Recently it was shown that the GAR domain is also a protein-protein interaction domain. The hnRNP A1 GAR domain interacts in vitro with itlf and with other hnRNP proteins (Cartegni et al., 1996). The GAR domain of mou nucleolin, and probably also of the yeast gar2 protein, interacts with veral ribosomal proteins (Bouvet et al., 1998; Sicard et al., 1998). It is not known how the GAR domain mediates protein-protein interaction, but since the GAR domain of nucleolin interacts with only a subt of ribosomal proteins, the interactions are likely to be specific.
Soon after its discovery, it was shown that nucleolin contained high levels of N G,N G-dimethylarginines (Lischwe et al., 1982). This post-translational modification is found on arginine located in the GAR domain (Lapeyre et al., 1986; Lischwe et al., 1985). This modification is not absolutely required for the non-specific interaction of the GAR domain with RNA (Serin et al., 1997), or for the interaction with other proteins (Bouvet et al., 1998). However, it could be an important signal for the regulation of the interactions, the stability of the protein, or its localization. It is not known if this post-translational modification is reversible. Localization of nucleolin
The intracellular localization of nucleolin has been extensively studied by electron microscopy analysis and/or immunofluorescence ultrastructural localization in vertebrates (Escande et al., 1985; Lischwe et al., 1981; Spector et al., 1984), plants (Martin et al., 1992; Minguez and Moreno Diaz de la Espina, 1996; Tong et al., 1997) or even S. pombe yeast cells (Leger-Silvestre et al., 1997). There is a general agreement that nucleolin is mainly found in the fibrillar component around the fibrillar centres. However, a small portion has also been detected in the granular component in higher eukaryotes (Biggiogera et al., 1990; Escande et al., 1985), but is rarely en in the fibrillar centres (Escande et al., 1985; Martin et al., 1992).
Nucleolin relocalization during mitosis is a specific feature of higher eukaryotes. A general agreemen
t is that nucleolin is prent at the periphery of metaphasic chromosomes (Medina et al., 1995), in the nucleolar remnant (NR) of CHO cells
当心机械伤人
H. Ginisty and others
Table 2. Percentage of quence identity of the RDB domain of nucleolin from different species
RDB 1RBD 2RDB 3RBD 4 Hamster100100100100 Mou969993100 Human85839699 Chicken62647089 Xenopus laevis53536785 Carp*ND516485 Nsr1p100100
Gar25861
NucMs1100100
Nucleolin of pea8686
FMV3bp (Arabidopsis)2222
According to their quence identity, nucleolin and nucleolin-like proteins have been divided into three groups. All RBDs are compared to the corresponding domain of one member of each group (Hamster in group 1, Nsr1p in group 2 and NucMs1 in group 3).
The percentages have been calculated for each domain between the predicted β1 and β4 strand, excluding the linker between each domain. RBD domains of FMV3bp prent low quence identity with each of the other three groups.
*P. Ganot and J. P. Bachellerie, personal communication.
ND, quence not determined.
765 Nucleolin structure and functions
(Azum-Gelade et al., 1994) and in cytoplasmic nucleolus-
derived foci (NDFs) (Dundr et al., 1997). The NDFs
disappear at telopha and are replaced by nucleolin-
containing pre-nucleolar bodies (PNBs) (Gas et al., 1985;
Medina et al., 1995; Ochs et al., 1983). The NR, NDFs and
PNBs also contain other factors necessary for pre-rRNA
processing, whereas the Nucleolar Organir Regions (NORs)
only gather factors responsible for rDNA transcription
(Hernandez-Verdun and Gautier, 1994). In spite of some
controversial obrvations (Gas et al., 1985; Lischwe et al.,
1981; Ochs et al., 1983; Spector et al., 1984), the mitotic NORs
are almost certainly devoid of nucleolin (Azum-Gelade et al.,
1994; Scheer and Weinberger, 1994; Weinberger and
Scheer, 1995). At the end of mitosis, during nuclear membrane
formation, the NR, NDFs and PNBs gather around the NORs
to form new nucleoli so that rRNA transcription and then
processing of the newly transcribed pre-rRNA can start again
李威威
very rapidly.
Structural domains necessary for the proper nuclear
昙鸾targeting and nucleolar localization of nucleolin in
mammalian cells as well as Nsr1p in S. cerevisiae have also
been extensively studied (Creancier et al., 1993; Meβmer and
Dreyer, 1993; Schmidt-Zachmann and Nigg, 1993; Yan and
Mele, 1993), but the results are somewhat puzzling. First,
they clearly demonstrate the prence of a conrved
functional bipartite NLS, which is necessary and sufficient for
nuclear targeting. Nucleolin, Nsr1p and gar2 from S. pombe
are also able to bind to NLS peptides in vitro, thanks to their
amino-terminal domain (Lee et al., 1991; Xue et al., 1993).
This has led to the proposal that the proteins may be
implicated in nucleo-cytoplasmic transport (Xue et al., 1993).
However, the nucleolar accumulation of nucleolin and Nsr1p
ems to be determined by a complex interplay between
veral domains and is therefore not well understood.
Basically, the RNA-binding domains of nucleolin or
Nsr1p/gar2 are crucial for their nucleolar localization
(Creancier et al., 1993; Meβmer and Dreyer, 1993; Schmidt-
Zachmann and Nigg, 1993; Yan and Mele, 1993), but are
unable by themlves to target hybrid proteins to the
nucleolus. Moreover, the recent finding that RNa treatment
induces the relea of nucleolin from nucleoli in A6 cells
(Schwab and Dreyer, 1997) confirms that nucleolar
accumulation of nucleolin requires the prence of RNA. The
N-terminal domain of Nsr1p, fud to its NLS, is able to direct
核心价值观的作用β-galactosida to the nucleolus, but progressive deletions result in nucleoplasmic localization (Yan and Mele, 1993).
The same domains (N-terminal + NLS) of nucleolin are able
to target chloramphenicol acetyl transfera (CAT) only to the
nucleus (Creancier et al., 1993), although Schmidt-Zachmann史提芬
and Nigg (1993) find it in the nucleolus. The C-terminal GAR
domain allows total nucleolar accumulation when it is
associated with at least one RBD (Creancier et al., 1993). The
nucleolar localization of nucleolin lacking only its RGG
repeats is also controversial (Creancier et al., 1993; Schmidt-
Zachmann and Nigg, 1993). In conclusion, although nuclear
targeting of nucleolin-like proteins depends on a classical
bipartite NLS, we assume that nucleolar accumulation is a
conquence of the affinity of the various domains for
nucleolar factors, possibly rRNA (Olson and Thompson,
1983; Schwab and Dreyer, 1997) or other nucleolar
components with which nucleolin interacts (Li et al., 1996).
A relationship between the function of nucleolin NLS and the phosphorylation of this protein has also
been pointed out in Xenopus egg extract (Schwab and Dreyer, 1997). The prence of p34cdc2phosphorylation sites improves the efficiency of nuclear translocation when they are dephosphorylated and enhances cytoplasmic localization when phosphorylated. The results therefore indicate that the changes in the phosphorylation state of nucleolin during the cell cycle might be a possible regulatory element of nucleolin localization.
In addition to the nucleolar localization of nucleolin, veral reports indicate the prence of nucleolin at the cell surface (e Table 3). However, further experiments are needed to demonstrate unambiguously that the protein detected in the cell membrane is nucleolin.
Functions of nucleolin in ribosome biogenesis
The almost exclusive localization of nucleolin within the nucleolus and its transient interaction with rRNA and pre-ribosomes strongly point to a role in ribosome biogenesis. Indeed, experimental data suggest that nucleolin is involved in multiple steps of this process. However, most of the data are bad on in vitro studies and only correlate the prence or abnce of nucleolin with a function. The expression of nucleolin (and nucleolin-like proteins) and rRNA em well coordinated (Bogre et al., 1996; Meyuhas et al., 1990; Tong et al., 1997), and the accumulation of nucleolin is correlated to
the proliferative activity of the cell (Derenzini et al., 1995; Mehes and Pajor, 1995; Sirri et al., 1995, 1997; Yokoyama et al., 1998). Moreover, during Xenopus laevis early development, the appearance of nucleolin precedes the transcription of rDNA and synthesis of ribosomal proteins, suggesting that nucleolin is involved in the early steps of ribosomes biogenesis (Caizergues-Ferrer et al., 1989). Nucleolin and rRNA transcription
A small fraction of nucleolin tightly binds to chromatin (Olson et al., 1975; Olson and Thompson, 1983). Moreover, in vitro, nucleolin interacts with single-stranded DNA with higher affinity than with double-stranded DNA, and binds preferentially to a DNA fragment containing the region of the non-transcribed spacer of rDNA located upstream of the site of transcription initiation (Olson et al., 1983). Taken together with the ability of nucleolin to interact with histone H1 and to modulate chromatin structure (Erard et al., 1998, 1990), this suggested that nucleolin could be involved in the regulation of rRNA transcription. Indeed, following inhibition of rRNA transcription, a rapid relea of nucleolin from the den fibrillar zone of the nucleolus is obrved (Escande-Geraud et al., 1985) demonstrating that its prence in this subnucleolar compartment was dependant upon rRNA transcription. In an in vitro transcription system, full-length nucleolin inhibits rRNA transcription from a plasmid template (Bouche et al., 1984). Furthermore, injection of anti-nucleolin antibodies in Chiro
nomus tentans salivary gland cells stimulate 2- to 3.5-fold the synthesis of pre-rRNA (Egyhazi et al., 1988). Altogether, the data indicate that nucleolin could be involved in a control of rRNA transcription, probably without its having a direct role in transcription initiation since it is not found in the polymera I initiation complex. The relevance of the interaction of nucleolin with rDNA quences upstream of the initiation site
