SURVEY AND SUMMARY: The pre-ribosomal network

Abstract

Recent achievements in yeast functional proteomics have significantly advanced our knowledge about ribosome biogenesis. Here, we present a program developed to integrate data from various proteome analyses with cell biological data on components present in the ribosome producing factories. This program allows users to attribute factors to certain complexes and to specific steps of ribosome biogenesis. Thus, it helps to gain novel insights into the complex network involved in maturation of ribosomal subunits. The database can be accessed at the URL http://www.pre-ribosome.de.

INTRODUCTION

The machineries that trigger assembly, maturation and nucleo-cytoplasmic transport of ribosomes are thought to be various ribonucleoprotein complexes that form distinct ribosomal precursor particles containing rRNAs and ribosomal proteins. Apparently, these large structures assemble and mature in a temporally and spatially regulated manner: they originate in the nucleolus and reach the nuclear pore through which they are then finally exported to the cytoplasm, resulting in the mature 40S and 60S ribosomal subunits (1). Although several individual factors have already been attributed to ribosome biogenesis, a complete picture of the whole pre-ribosomal network is only slowly emerging. A major obstacle to the identification of factors present in pre-ribosomes was that the methods were not sensitive enough, even though pre-ribosomal particles or their substructures could be identified (2). Pre-ribosomes were characterised by their sedimentation behaviour in sucrose gradients using either isolated HeLa nucleoli or yeast whole cell extracts containing radiolabelled rRNA. However, resolution into different particles and identification of the protein constituents was not possible. This hindrance can now be overcome by applying affinity purification methods which allow the isolation of specific protein complexes under physiological conditions using epitope-tagged proteins (3–10). Components of such purified complexes can be identified by mass spectrometry. Now, a new level of complexity arises with the analysis of multifactor machineries, such as the pre-ribosomal network, which change their composition in a highly dynamic fashion. If single affinity purified complexes of epitope-tagged pre-ribosomal factors are compared, the heterogeneity in their subunit composition can be attributed to methodical uncertainties or that single components can associate with different complexes in a more or less transient fashion. To distinguish between these possibilities, it is necessary to perform a more thorough biochemical analysis of the single affinity purified complexes or to apply a statistical evaluation to as many isolated complexes as possible.

In general, to compare different affinity purifications of pre-ribosomal complexes conducted under physiological conditions the following problems have to be taken into account. (i) Affinity purification using a tagged protein may result in the isolation of a mixture of complexes that each contain the tagged protein. If a tagged protein is present in several complexes, not all complexes may be isolated with the same efficiency. For example, it is conceivable that the tag is not equally accessible in different complexes under physiological conditions. (ii) Some proteins are not detected using the SDS–PAGE/mass spectrometry approaches applied. (iii) The variability in the protein composition of various affinity purifications using the same epitope-tagged protein as bait has to be considered. Gavin et al. asserted that only 70% of the proteins found in two independent experiments were identical (3). An even higher variability is likely if different tags and/or protocols are applied. (iv) In the past, pre-ribosomes were defined by their sedimentation coefficient and content of rRNA species that are characteristic for a certain stage of ribosome maturation. Since the genome-wide analyses of protein complexes containing ribosome biogenesis factors do not include (quantitative) data about co-purifying rRNA fragments, it is difficult to attribute the complexes to thepreviously characterised pre-ribosomes. (v) Ribosomal proteins, which are commonly used as markers for (pre)ribosomal complexes, appear to contaminate all sorts of affinity purified complexes, including those not involved in ribosome biogenesis (3). Therefore, in the genome-wide screens, they were pragmatically excluded from further analysis (3,4). Consequently, it is impossible to attribute ribosomal proteins to specific pre-ribosomal complexes although this would facilitate determining the order of assembly of the ribosomes.

Many of the uncertainties arising from individual co-purification experiments can be overcome using internal statistical evaluations of as many independent experiments as possible. The prediction that certain proteins belong to one complex is likely to be true if they are often found to co-purify regardless of which component was used as bait. Moreover, the use of statistical ‘marker’ proteins helps to discriminate between distinct groups of complexes.

Examination of two recently published yeast proteome analyses (3,4), as well as of five reports on the composition of different pre-60S and pre-40S ribosomes (5–9), reveals that the picture of nascent ribosomes is strikingly complex. According to these analyses, overall 54 multiprotein complexes containing 388 different proteins are presumably involved in ribosome biogenesis. Eighty-eight of these proteins (∼20%) have already been described to act at the level of pre-ribosomes, 41 are involved in 60S and 38 in 40S biogenesis and nine factors were suggested to participate in both pathways. We developed a program that allows users to study the composition of the protein networks involved in maturation of ribosomal subunits independently of possible methodical failures. With the help of the published data about factors required for ribosome biogenesis, putative novel ribosomal factors can be predicted and yet uncharacterised proteins can be attributed to specific steps in ribosome biogenesis. For instance, if a protein of unknown function is frequently found in complexes containing similar factors that are all involved in early steps of biogenesis, it probably plays a role during the onset of maturation. On the other hand, factors can be associated with several pre-ribosomal complexes of diverse composition containing either ‘early’ or ‘late’ pre-ribosomal proteins. They can be considered as core factors that accompany pre-ribosomes through several stages of maturation. In this way, a crude reconstruction of the dynamics of (dis)assembly of pre-ribosomes is possible. Accordingly, the proteins involved in ribosome biogenesis can be arranged into separate units, which might represent functional entities in ribosomal maturation.

MATERIALS AND METHODS

For an initial arrangement of protein complexes putatively involved in ribosome biogenesis we used the data obtained from two systematic analyses of protein complexes in yeast (3,4). In each analysis more than 10% of randomly selected yeast proteins were successfully used as bait to purify protein complexes. Altogether, we chose 49 of these protein complexes that all contained factors known to be involved in ribosome biogenesis. To this random selection of protein complexes we added five individually isolated multiprotein complexes, which have recently been described as involved in ribosome biogenesis, to obtain the complete data set (5–9). For the classification into different groups we used the following criteria: (i) known function of the individual constituent; (ii) similarities in complex composition; (iii) most importantly, distribution according to marker proteins. Altogether, 19 proteins were found in more than 20% of the initially selected group of 49 complexes. They were designated as marker proteins and can be explained as the major components of pre-ribosomal intermediates that are well detected using the experimental strategy of protein affinity purification with subsequent mass spectrometry analysis. Consequently, if complexes are compared through the appearance of marker proteins rather than through their entire subunit composition it facilitates their classification irrespective of the exact reproducibility of the single experiment. The distribution of the marker proteins within the complexes yielded two groups (group 2 and group 5) in which either exclusively one or another set of marker proteins was detected. Accordingly, the corresponding marker proteins were designated as class A and class B markers. Two groups of complexes contain complexes with both marker A and marker B proteins (group 1 and group 4 complexes): group 1 contains more of the class A markers while group 4 includes more of the class B markers. The remaining two groups contain complexes in which only few marker proteins were identified (group 3 and group 6 complexes) and which were sorted according the presence of either 60S or 40S subunit factors.

A Pascal program was established that reads these data in the form of several external data files: composition and classification of the selected complexes and the functional features of their components. The program calculates the similarities of complexes, the proportion of ribosome biogenesis factors in a complex and the probability that two proteins are found in the same complex. Furthermore, the program exports all the data into a format that is compatible with standard database software. Filemaker Pro 5.5 was used to make the database available for the Internet. The source code of the Pascal software and further information are available upon request.

RESULTS AND DISCUSSION

Marker proteins specific for distinct stages versus core proteins

We sorted the pre-ribosomal protein network using the following criteria (Table 1; see also Materials and Methods): components which appeared in the two yeast proteome analyses (3,4) in more than 20% of the various complexes containing ribosome biogenesis factors were designated as ‘markers’. In both genome-wide approaches, proteins had been excluded that are generally considered as contaminants, such as heat shock proteins and ribosomal proteins (3). Thus, proteins here designated as ‘marker proteins’ are most likely not abundant contaminants; they are well detectable in many distinct complexes containing ribosome biogenesis factors that can be efficiently purified. Accordingly, they probably represent stably associated factors with a possible role in ribosome biogenesis. ‘Marker’ proteins co-appear in complexes of two different categories: complexes that contain predominantly class A markers and those consisting of more class B markers. All characterised class A markers have been described to be involved in 60S biogenesis. In contrast, class B markers are proteins that are more generally involved in early steps of ribosome biogenesis or in modification of rRNAs, like Cbf5 and Nop1, or components required for 40S biogenesis. With the exception of two markers, all class A and B markers are essential for growth, which underlines their fundamental role in ribosome biogenesis.

Table 1. Criteria for sorting pre-ribosomal factors.

If the relative abundance of class A and class B markers in the various protein complexes is evaluated and the ascertained functions of associated factors from the literature are taken into account, the complexes can be roughly sorted into six different groups (Table 2). In most cases, the ratio of class A to class B markers clearly defines the affiliation of a complex with a certain group. If a complex contains more of class A than of class B markers it is sorted into group 1. For instance, a complex purified via epitope-tagged Nop4 contains more (86%) of the class A markers than of the class B markers (57%). Furthermore, 62% of the components have been described to be involved in 60S biogenesis whereas only 4% are considered to play a role in the 40S pathway. Accordingly, the Nop4 complex is a representative member of group 1. Only one complex of group 1 (Ela1) contains more of the class B than class A markers. Even so, this complex was attributed to group 1, since it exclusively contains factors described to be required for 60S biogenesis.

Table 2. Pre-ribosomal complexes can be sorted into six groups.

The complexes were named according to their tagged component and sorted according to the appearance of marker proteins in the complexes. Yellow: class A markers (60S factors). Blue: class B markers (early factors, 40S factors) (see also text). Note that three complexes of group 6 do not contain marker proteins. These complexes were assigned to the 40S pathway according to the literature. Proteins such as Erb1, Nop7, Ssf1, Cbf5, Nop2 and Tsr1 served as baits in either two or three independent purification procedures using different epitope tags (3,4,13). The complexes were sorted according to the appearance of markers and listed in the corresponding groups as individual complexes.

Members of complexes assigned to group 2 contain only class A markers and almost exclusively factors with a described role in 60S biogenesis. Group 3 complexes also contain only class A markers; however, their compositions vary significantly from those of groups 1 and 2. Some of their components, like Nmd3, are considered to represent nuclear export factors for the 60S subunit (11,12). Consequently, group 3 complexes could be involved in late export steps.

Complexes with more class B than A markers are sorted into groups 4–6. Group 4 represents complexes that also contain class A markers, whereas groups 5 and 6 completely lack class A markers. Complexes listed in group 6 differ significantly in their composition from those of groups 4 and 5 and contain only few class B markers. Some of them, for example epitope-tagged Nob1 and Rio2, lack any markers. However, since these complexes contain factors that have been attributed to 40S biogenesis, they were charted in group 6.

Complexes sorted into groups 1 and 4, 2 and 5, and 3 and 6 can be interpreted to be involved in ribosome biogenesis of early, intermediate and late stages, respectively. Alternatively, they can be considered as mixed (groups 1 and 4) or pure (groups 2 and 5) pre-40S and pre-60S assemblies, respectively, or distinct pre-40S and pre-60S assemblies (groups 3 and 6) which vary significantly from the majority of pre-40S and pre-60S complexes (Table 2). Recently, the components of a large pre-40S ribonucleoprotein complex which was designated as the small subunit processome (SSU processome) were identified (9). Many constituents of the SSU processome are present in complexes of group 5. Accordingly, group 5 could also be considered as a mixture of pre-40S and SSU processome-containing complexes.

Based on this classification and the abundance of single factors within the different groups, it is possible to discriminate among elements specific for a certain (maturation) stage and (core) elements that are associated with many forms of the network involved in ribosome biogenesis (or which stay bound to pre-ribosomes from early to late stages). Association of each component within the single complexes, including the corresponding reference, can easily be screened using the database (http://www.pre-ribosome.de). Moreover, it is possible to see how often components were found together in complexes and thereby to distinguish compositions of very similar complexes from those which differ significantly in their subunit composition. Of course, not all of the isolated complexes have to be part of pre-ribosomal structures. Some of them might be only indirectly related to ribosome biogenesis or play a role in a different cellular function. Similarly, proteins that are found in only few complexes do not necessarily have to be ribosome biogenesis factors. For instance, factors required for replication and cell proliferation were recently reported to co-purify with ribosome biogenesis factors, suggesting a link between these pathways (13). Therefore, such rarely appearing factors that are not directly involved in ribosome biogenesis might point towards possible overlaps between ribosome biogenesis and a second, supposedly separate, cellular pathway. On the other hand, it is likely that most of the isolated complexes are required for ribosome biogenesis since the majority of their components were described as playing a role in either small or large ribosomal subunit formation. Supposedly most of these complexes participate in subsequent steps of ribosome maturation, and it can be followed up whether components are involved in one specific and/or several subsequent steps of ribosome biogenesis and whether they act early and/or late in the pathway. According to the statistical assumption, examples for core factors involved in several steps of 60S biogenesis are Erb1, Has1, Nug1 and Noc2, which are found in many complexes of different composition in which 60S factors are predominant. In contrast, typical proteins of pure pre-60S complexes (Table 2, group 2) are, for instance, Rix1, Rlp7, Nug2 and Noc3, since the compositions of the complexes in which they are found are closely related. Examples of core factors of the 40S pathway are Kre33 and Cbf5, whereas Noc4, Nop14 and Emg1 are typical 40S proteins of closely related (pure) pre-40S/SSU processome complexes (Table 2, group 5).

In a few cases, factors which are ascribed to a specific stage of ribosome biogenesis appear also in complexes that apparently function in a completely different step. For instance, two components, Nmd3 and Tif6 (11,12,14,15), which are found in pre-60S ribosomes and are required for 60S synthesis, are also found together in a complex (DIA2) predominantly consisting of 40S factors. On the other hand, according to our database (see ‘component data file’) there is also no doubt that both factors are required for 60S biogenesis: Nmd3 and Tif6 were identified in 4 and 15 different 60S complexes, respectively, while only one of each was sorted to the 40S pathway. Further experiments have to examine whether these designated ‘60S factors’ participate in more roles of ribosome biogenesis than anticipated. To get an overview about the established role of the single factors as well as to speculate about further possible roles, it is necessary to perform a search in both data files, the ‘complex data file’ and the ‘component data file’.

The validity and potential of the database can be easily demonstrated by consideration of the subunit composition of the recently published SSU processome (9). The SSU processome was isolated using consecutive affinity purifications by two different tags on two subunits, Mpp10 and Nop5. If the database is searched for either Mpp10 or Nop5, the proteins appear in either five or four, respectively, differently composed complexes. However, if complexes which contain both components are looked for, only one other complex appears (purified by epitope-tagged PWP2). This complex consists of 54 proteins and contains more than 85% of the factors present in the published SSU processome and therefore probably represents a major part of the SSU processome. Thus, in this case, the simultaneous search for two different factors in the database mimics the biochemical isolation using tags on two different subunits of a common complex. On the other hand, the database indicates that Mpp10 and Nop5 participate in at least one common and also in several divergent steps of ribosome biogenesis, which still have to be biochemically investigated. Interestingly, most of the SSU processome components are only found in group 4 or group 5 complexes, suggesting a specific role for 40S biogenesis. Only five proteins, UTP10, Nop5, Nop1, Sik1 and Rrp5, also appear in complexes of groups other than group 4 or 5. Since implied roles in 60S biogenesis have been already described for Nop1, Sik1 and Rrp5 (16–18), it is conceivable that these five proteins, although components of the SSU processome, play a more ubiquitous role in ribosome biogenesis. These applications of the database underline that the predictions of function and affiliation of the single components are meaningful.

To facilitate an initial search through the statistically estimated pre-ribosomal network, we chose a graphical presentation in which the components of the various complexes identified by the two genome-wide approaches (3,4) are sorted according to their abundance and appearance together with other factors involved in ribosome biogenesis (depicted in Fig. 1) (http://www.pre-ribosome.de). Though this illustration provides an overview of the variety, complexity and dynamics of the pre-ribosomal network, the detailed interactions of the participating components should be viewed in the database itself.

Figure 1.

Complex pathways of ribosome biogenesis according to the two proteome approaches (3,4). The components of groups 1–6 are shown. Each component is depicted as an oval. The size of the oval corresponds to the abundance with which the factor appears in complexes of that group. Components which are present in one, two, three or four groups are marked yellow, green, light blue or blue, respectively. For instance, large blue ovals designate proteins that are frequently found within the same group but which are also present in three other groups. They can be considered as core subunits, which accompany pre-ribosomal structures through several stages of maturation. In contrast, large yellow or green ovals stand for components that are often found in complexes belonging to only one or two groups. They are considered to be specific for a certain stage of ribosomal maturation. Components in small yellow ovals stand for either transiently associated factors or possible contaminants.

Divergent tracks toward mature 40S and 60S ribosomal subunits

Our statistical examination of the various protein complexes containing ribosomal factors supports previous findings that pre-ribosomal ribonucleoprotein complexes change their composition in a temporal manner (19) and that the pathways leading to mature 40S and 60S subunits can proceed independently from each other once they have separated (10,20,21). After this branching, factors known to be involved in 60S biogenesis were not found in complexes together with 40S biogenesis factors and vice versa, which points towards different requirements for the completion, quality control and transport of the two ribosomal subunits. Surprisingly, factors implicated in very early steps of rRNA processing, such as RNA pseudouridylation and/or methylation, were often found in complexes together with 40S biogenesis factors but not with 60S factors. A simple explanation is that the pre-40S factors assemble together with early pre-rRNA processing factors and that assembly of pre-60S proteins onto the 27S rRNA occurs after the two pathways have separated (10). Since the end products of the rRNA cleavage at site A₂, 20S rRNA and 27S rRNA are the characteristic intermediates for pre-40S and pre-60S ribosomes, respectively, cleavage at site A₂ could be the signal for association of 60S factors onto 27S rRNA, which then initiates formation of a pre-60S ribosomal particle. Congruently, transport and biogenesis of 60S subunits can proceed independently of 40S biogenesis unless a defect in the 40S pathway inhibits the cleavage required for 27S rRNA formation.

The pre-ribosomal network as deduced from the proteome analysis can only produce a rough frame of the ribosome network. However, it cannot substitute for a thorough biochemical/cell biological investigation. A first insight into the action of some participating factors recently became possible through the combination of cell biological techniques with biochemical assays that monitor subsequent steps in ribosome maturation (5,6,10–12,15,22,23). But the real work starts now: more methods have to be developed that allow one to analyse individual maturation steps and most of the different players have to be examined regarding their specific role in ribosome biogenesis. Understanding the molecular and cell biological details of their function will be necessary to explain how the complex network of ribosome biogenesis is coordinated to achieve the high efficiency of ribosome synthesis required for the maintenance of cellular functions.