In the nucleus and cytoplasm of chicken erythroleukemic cells, prosomes containing the p23K subunit are found in centers of globin (pre-)mRNA processing and accumulation

Primary transcripts: From the discovery of RNA processing to current concepts of gene expression - Review

The main purpose of this review is to recall for investigators - and in particular students -, some of the early data and concepts in molecular genetics and biology that are rarely cited in the current literature and are thus invariably overlooked. There is a growing tendency among editors and reviewers to consider that only data produced in the last 10-20 years or so are pertinent. However this is not the case. In exact science, sound data and lucid interpretation never become obsolete, and even if forgotten, will resurface sooner or later. In the field of gene expression, covered in the present review, recent post-genomic data have indeed confirmed many of the earlier results and concepts developed in the mid-seventies, well before the start of the recombinant DNA revolution. Human brains and even the most powerful computers, have difficulty in handling and making sense of the overwhelming flow of data generated by recent high-throughput technologies. This was easier when low throughput, more integrative methods based on biochemistry and microscopy dominated biological research. Nowadays, the need for organising concepts is ever more important, otherwise the mass of available data can generate only "building ruins" - the bricks without an architect. Concepts such as pervasive transcription of genomes, large genomic domains, full domain transcripts (FDTs) up to 100 kb long, the prevalence of post-transcriptional events in regulating eukaryotic gene expression, and the 3D-genome architecture, were all developed and discussed before 1990, and are only now coming back into vogue. Thus, to review the impact of earlier concepts on later developments in the field, I will confront former and current data and ideas, including a discussion of old and new methods. Whenever useful, I shall first briefly report post-genomic developments before addressing former results and interpretations. Equally important, some of the terms often used sloppily in scientific discussions will be clearly defined. As a basis for the ensuing discussion, some of the issues and facts related to eukaryotic gene expression will first be introduced. In chapter 2 the evolution in perception of biology over the last 60 years and the impact of the recombinant DNA revolution will be considered. Then, in chapter 3 data and theory concerning the genome, gene expression and genetics will be reviewed. The experimental and theoretical definition of the gene will be discussed before considering the 3 different types of genetic information - the "Triad" - and the importance of post-transcriptional regulation of gene expression in the light of the recent finding that 90% of genomic DNA seems to be transcribed. Some previous attempts to provide a conceptual framework for these observations will be recalled, in particular the "Cascade Regulation Hypothesis" (CRH) developed in 1967-85, and the "Gene and Genon" concept proposed in 2007. A knowledge of the size of primary transcripts is of prime importance, both for experimental and theoretical reasons, since these molecules represent the primary units of the "RNA genome" on which most of the post-transcriptional regulation of gene expression occurs. In chapter 4, I will first discuss some current post-genomic topics before summarising the discovery of the high Mr-RNA transcripts, and the investigation of their processing spanning the last 50 years. Since even today, a consensus concerning the real form of primary transcripts in eukaryotic cells has not yet been reached, I will refer to the viral and specialized cellular models which helped early on to understand the mechanisms of RNA processing and differential splicing which operate in cells and tissues. As a well-studied example of expression and regulation of a specific cellular gene in relation to differentiation and pathology, I will discuss the early and recent work on expression of the globin genes in nucleated avian erythroblasts. An important concept is that the primary transcript not only embodies protein-coding information and regulation of its expression, but also the 3D-structure of the genomic DNA from which it was derived. The wealth of recent post-genomic data published in this field emphasises the importance of a fundamental principle of genome organisation and expression that has been overlooked for years even though it was already discussed in the 1970-80ties. These issues are addressed in chapter 5 which focuses on the involvement of the nuclear matrix and nuclear architecture in DNA and RNA biology. This section will make reference to the Unified Matrix Hypothesis (UMH), which was the first molecular model of the 3D organisation of DNA and RNA. The chapter on the "RNA-genome and peripheral memories" discusses experimental data on the ribonucleoprotein complexes containing pre-mRNA (pre-mRNPs) and mRNA (mRNPs) which are organised in nuclear and cytoplasmic spaces respectively. Finally, "Outlook " will enumerate currently unresolved questions in the field, and will propose some ideas that may encourage further investigation, and comprehension of available experimental data still in need of interpretation. In chapter 8, some propositions and paradigms basic to the authors own analysis are discussed. "In conclusion" the raison d'être of this review is recalled and positioned within the overall framework of scientific endeavour.

Regulation of gene expression and the transcription factor cycle hypothesis

Post-genomic data show unexpected extent of the transcribed genome and the size of individual primary transcripts. Hence, most cis-regulatory modules (CRMs) binding transcription factors (TFs) at promotor, enhancer and other sites are actually transcribed within full domain transcripts (FDTs). The ensemble of these CRMs placed way upstream of exon clusters, downstream and in intronic or intergenic positions represent a program of gene expression which has been formally analysed within the Gene and Genon concept [1,2]. This concept has emphasised the necessity to separate product information from regulative information to allow information-theoretic analysis of gene expression. Classically, TFs have been assumed to act at DNA level exclusively but evidence has accumulated indicating eventual post-transcriptional functions. The transcription factor cycle (TFC) hypothesis suggests the transfer of DNA-bound factors to nascent RNA. Exerting downstream functions in RNA processing and transport, these factors would be liberated by RNA processing and cycle back to the DNA maintaining active transcription. Sequestered on RNA in absence of processing they would constitute a negative feedback loop. The TFC concept may explain epigenetic regulation in mitosis and meiosis. In mitosis control factors may survive as single proteins but also attached to FDTs as organised complexes. This process might perpetuate in cell division conditioning of chromatin for transcription. As observed on lampbrush chromosomes formed in avian and amphibian oogenesis, in meiosis the genome is fully transcribed and oocytes conserve high Mr RNA of high sequence complexity. When new interphase chromosomes form in daughter cells and early embryogenesis, TFs and other factors attached to RNA might be reinserted onto the DNA.

Proteins tightly bound to DNA: new data and old problems

Proteins tightly bound to DNA (TBP) comprise a group of proteins that remain bound to DNA after usual deproteinization procedures such as salting out and treatment with phenol or chloroform. TBP bind to DNA by covalent phosphotriester and noncovalent ionic and hydrogen bonds. Some TBP are conservative, and they are usually covalently bound to DNA. However, the TBP composition is very diverse and significantly different in different tissues and in different organisms. TBP include transcription factors, enzymes of the ubiquitin-proteasome system, phosphatases, protein kinases, serpins, and proteins of retrotransposons. Their distribution within the genome is nonrandom. However, the DNA primary structure or DNA curvatures do not define the affinity of TBP to DNA. But there are repetitive DNA sequences with which TBP interact more often. The TBP distribution within genes and chromosomes depends on a cell's physiological state, differentiation type, and stage of organism development. TBP do not interact with DNA in the sites of its association with nuclear matrix and most likely they are not components of the latter.

In embryonic chicken erythrocytes actively transcribed alpha globin genes are not associated with the nuclear matrix

The spatial organization of a 250 Kb region of chicken chromosome 14, which includes the alpha globin gene cluster, was studied using in situ hybridization of a corresponding BAC probe with nuclear halos. It was found that in non-erythroid cells (DT40) and cultured erythroid cells of definite lineage (HD3) the genomic region under study was partially (DT40 cells) or fully (HD3 cells) associated with the nuclear matrix. In contrast, in embryonic red blood cells (10-day RBC) the same area was located in the crown of DNA loops surrounding the nuclear matrix, although both globin genes and surrounding house-keeping genes were actively transcribed in these cells. This spatial organization was associated with the virtual absence of RNA polymerase II in nuclear matrices prepared from 10-day RBC. In contrast, in HD3 cells a significant portion of RNA polymerase II was present in nuclear matrices. Taken together, these observations suggest that in embryonic erythroid cells transcription does not occur in association with the nuclear matrix.

Gene and genon concept: coding versus regulation. A conceptual and information-theoretic analysis of genetic storage and expression in the light of modern molecular biology

We analyse here the definition of the gene in order to distinguish, on the basis of modern insight in molecular biology, what the gene is coding for, namely a specific polypeptide, and how its expression is realized and controlled. Before the coding role of the DNA was discovered, a gene was identified with a specific phenotypic trait, from Mendel through Morgan up to Benzer. Subsequently, however, molecular biologists ventured to define a gene at the level of the DNA sequence in terms of coding. As is becoming ever more evident, the relations between information stored at DNA level and functional products are very intricate, and the regulatory aspects are as important and essential as the information coding for products. This approach led, thus, to a conceptual hybrid that confused coding, regulation and functional aspects. In this essay, we develop a definition of the gene that once again starts from the functional aspect. A cellular function can be represented by a polypeptide or an RNA. In the case of the polypeptide, its biochemical identity is determined by the mRNA prior to translation, and that is where we locate the gene. The steps from specific, but possibly separated sequence fragments at DNA level to that final mRNA then can be analysed in terms of regulation. For that purpose, we coin the new term "genon". In that manner, we can clearly separate product and regulative information while keeping the fundamental relation between coding and function without the need to introduce a conceptual hybrid. In mRNA, the program regulating the expression of a gene is superimposed onto and added to the coding sequence in cis - we call it the genon. The complementary external control of a given mRNA by trans-acting factors is incorporated in its transgenon. A consequence of this definition is that, in eukaryotes, the gene is, in most cases, not yet present at DNA level. Rather, it is assembled by RNA processing, including differential splicing, from various pieces, as steered by the genon. It emerges finally as an uninterrupted nucleic acid sequence at mRNA level just prior to translation, in faithful correspondence with the amino acid sequence to be produced as a polypeptide. After translation, the genon has fulfilled its role and expires. The distinction between the protein coding information as materialised in the final polypeptide and the processing information represented by the genon allows us to set up a new information theoretic scheme. The standard sequence information determined by the genetic code expresses the relation between coding sequence and product. Backward analysis asks from which coding region in the DNA a given polypeptide originates. The (more interesting) forward analysis asks in how many polypeptides of how many different types a given DNA segment is expressed. This concerns the control of the expression process for which we have introduced the genon concept. Thus, the information theoretic analysis can capture the complementary aspects of coding and regulation, of gene and genon.

The gene and the genon concept: a functional and information-theoretic analysis

'Gene' has become a vague and ill-defined concept. To set the stage for mathematical analysis of gene storage and expression, we return to the original concept of the gene as a function encoded in the genome, basis of genetic analysis, that is a polypeptide or other functional product. The additional information needed to express a gene is contained within each mRNA as an ensemble of signals, added to or superimposed onto the coding sequence. To designate this programme, we introduce the term 'genon'. Individual genons are contained in the pre-mRNA forming a pre-genon. A genomic domain contains a proto-genon, with the signals of transcription activation in addition to the pre-genon in the transcripts. Some contain several mRNAs and hence genons, to be singled out by RNA processing and differential splicing. The programme in the genon in cis is implemented by corresponding factors of protein or RNA nature contained in the transgenon of the cell or organism. The gene, the cis programme contained in the individual domain and transcript, and the trans programme of factors, can be analysed by information theory.

RNA-dependent nuclear matrix contains a 33 kb globin full domain transcript as well as prosomes but no 26S proteasomes

Previously, we have shown that in murine myoblasts prosomes are constituents of the nuclear matrix; a major part of the latter was found to be RNase sensitive. Here, we further define the RNA-dependent matrix in avian erythroblastosis virus (AEV) transformed erythroid cells in relation to its structure, presence of specific RNA, prosomes and/or proteasomes. These cells transcribe but do not express globin genes prior to induction. Electron micrographs show little difference in matrices treated with DNase alone or with both, DNase and RNase. In situ hybridization with alpha globin riboprobes shows that this matrix includes globin transcripts. Of particular interest is that, apparently, a nearly 35 kb long globin full domain transcript (FDT), including genes, intergenic regions and a large upstream domain is a part of the RNA-dependent nuclear matrix. The 23K-type of prosomes, previously shown to be co-localized with globin transcripts in the nuclear RNA processing centers, were found all over the nuclear matrix. Other types of prosomes show different distributions in the intact cell but similar distribution patterns on the matrix. Globin transcripts and at least 80% of prosomes disappear from matrices upon RNase treatment. Interestingly, the 19S proteasome modulator complex is insensitive to RNase treatment. Only 20S prosomes but not 26S proteasomes are thus part of the RNA-dependent nuclear matrix. We suggest that giant pre-mRNA and FDTs in processing, aligning prosomes and other RNA-binding proteins are involved in the organization of the dynamic nuclear matrix. It is proposed that the putative function of RNA within the nuclear matrix and, thus, the nuclear dynamic architecture, might explain the giant size and complex organization of primary transcripts and their introns.

The upstream area of the chicken α-globin gene domain is transcribed in both directions in the same cells

It was demonstrated previously that in erythroid chicken cells an extended upstream area of the alpha-globin gene domain is transcribed in both directions as a part of ggPRX gene and a part of a full domain transcript of the alpha-globin gene domain. Here, we show that both DNA chains of the above-mentioned region are transcribed in the same cells and that the corresponding transcripts coexist in nuclei. The data obtained suggest that cells possess a molecular mechanism which in some cases prevents the formation of dsRNA and subsequent destruction of both transcripts in spite of the presence of complementary RNA chains in the cell nucleus.

The 33 kb transcript of the chicken ?-globin gene domain is part of the nuclear matrix

Giant nuclear transcripts, and in particular the RNAs of the globin gene domains which are much larger than their canonical pre-mRNAs, have been an enigma for many years. We show here that in avian erythroblastosis virus (AEV)-transformed chicken erythroleukaemic cells, where globin gene expression is abortive, the whole domain of alpha-globin genes is transcribed for about 33 kb in the globin direction and that this RNA is part of the nuclear matrix. Northern blot hybridisation with strand-specific riboprobes, recognising genes and intergenic sequences, and RT-PCR with downstream primers, show that the continuous full domain transcript (FDT) starts in the vicinity of a putative LCR and includes all the genes as well as known regulatory sites, the replication origin, and the DNA loop anchorage region in the upstream area. Absent in chicken fibroblasts, the globin FDT overlaps the major part of the ggPRX housekeeping gene that is transcribed in the opposite direction. RT-PCR and in situ hybridisation with genic and extra-genic globin probes demonstrated that the globin FDT is a component of the nuclear matrix. We suggest that the globin FDTs keep the domain in an active state, and the globin RNAs on the processing pathway are a component of the nuclear matrix. They may take part in the dynamic nuclear architecture when productively processed, or turn over slowly when globins are not synthesised.

In chicken leukemia cells globin genes are fully transcribed but their rnas are retained in the perinucleolar area

Using hybridization in situ with a ribo-probe recognizing transcripts of the chicken alpha A globin gene, we show here that in proliferating AEV-transformed erythroblasts this gene is strongly transcribed, but the corresponding transcripts are retained in the nuclei. Most surprisingly, this globin RNA accumulates in the perinucleolar areas in a pattern never observed before. Upon induction of cells to differentiate, leading to productive expression of the hemoglobins, the transcripts of the alpha A globin gene were found for the most part in the cytoplasm. In the nuclei of differentiated cells, the globin RNA is concentrated in one or two specific spots, which are likely to represent the "processing centers" (PCs) of the globin RNA. The results presented indicate that posttranscriptional steps of regulation involving in particular the perinuclear areas are of major importance for erythroid differentiation.

Prosomes form sarcomere-like banding patterns in skeletal, cardiac, and smooth muscle cells

Prosomes (20S proteasomes) constitute the catalytic core of the 26S proteasomes, but were first observed as factors associated with unstranslated mRNA. Recently, their RNase activity was discovered together with the fact that their proteolytic function is dispensable in adapted human cells. By indirect immunofluorescence using monoclonal antibodies, we demonstrate as a general phenomenon, regular intercalation of specific types of prosomes into the sarcomeric structure of all types of striated muscle. Surprisingly, in cultured smooth muscle cells without sarcomeric organization, some prosomes also form regular striations in extended projections of cytoplasmic regions. The significance of their sarcomeric distribution is not understood as yet, but the pattern we observe is very similar to that shown by others for muscle-specific mRNAs, identified by in situ hybridization, and that of the cognate proteins. A role of prosomes in the cotranslational assembly of the myofibrillar proteins is suggested, since prosomes organize into pseudo-sarcomeric patterns prior to formation de novo of the actin-myosin arrangement.

In mouse myoblasts nuclear prosomes are associated with the nuclear matrix and accumulate preferentially in the perinucleolar areas

Prosomes are the core of 26S proteasomes, although they were originally observed as 20S particles associated with cytoplasmic mRNPs. Here we show for the first time that prosomes are also genuine constituents of the nuclear matrix, chromatin and the nuclear RNP networks. Using mouse myoblasts we tested three monoclonal antibodies recognising the prosomal subunits p23K, p27K and p30K, and found that the corresponding prosome subclasses are characterised by a variable distribution pattern within the nuclei. Their presence on the nuclear matrix, and most abundantly in the perinucleolar area, is of particular importance. When myoblasts fuse into myotubes, the distribution pattern of certain types of prosomes on the nuclear matrix changes drastically. Surprisingly, DNA strongly interferes with the detection of prosomal antigens by immunofluorescence methods, whereas RNA, histones and other proteins soluble in 2 M NaCl have no such effect. This ‘masking’ of prosomes can be completely overcome by extensive or even mild digestion with DNase I or restriction enzymes. Many nuclear prosomes can be solubilized by combined treatment with 0.5% Triton X-100 and 2 M NaCl, and others can be released by digestion of DNA and/or RNA, and about 10–20% of nuclear prosomes remain tightly bound to the protein-based nuclear matrix.

Specific types of prosomes distribute differentially between intermediate and actin filaments in epithelial, fibroblastic and muscle cells

First observed as components of non-translated mRNP complexes, prosomes harbour RNase and several proteinase activities; they are also the central constituent of the "Multicatalytic Proteinase (MCP) complexes" or "26S-proteasomes". In two recent publications (Arcangeletti et al., 1997b; De Conto et al., 1997) we have shown, by applying a new fixation technique, that these particles distribute differentially between the cytoskeletal networks of intermediate filament (IF) and actin types; previously they had been observed exclusively on the intermediate filaments. Here we further investigate the distribution of prosomes of several types, distinct by their subunit composition, between the IF of vimentin type and the actin network, as well as in the 3D space of the cell. It is shown that subtypes of prosomes occupy specific networks of the cytoskeleton, and that this pattern is specific for a given cell type. Confocal microscopy shows that prosome cytodistribution is not homogeneous in the 3D space: in the perinuclear area they colocalize most strongly with the IF, and more peripherally with the microfilament/stress fiber system; connections may exist between the two networks. Furthermore, new data indicate that the prosome-actin interaction may participate in the molecular structure of the stress fibers.