The absence of a human-specific ribosomal DNA transcription factor leads to nucleolar dominance in mouse greater than human hybrid cells

Sterility and gene expression in hybrid males of Xenopus laevis and X. muelleri

BACKGROUND

Reproductive isolation is a defining characteristic of populations that represent unique biological species, yet we know very little about the gene expression basis for reproductive isolation. The advent of powerful molecular biology tools provides the ability to identify genes involved in reproductive isolation and focuses attention on the molecular mechanisms that separate biological species. Herein we quantify the sterility pattern of hybrid males in African Clawed Frogs (Xenopus) and apply microarray analysis of the expression pattern found in testes to identify genes that are misexpressed in hybrid males relative to their two parental species (Xenopus laevis and X. muelleri).

METHODOLOGY/PRINCIPAL FINDINGS

Phenotypic characteristics of spermatogenesis in sterile male hybrids (X. laevis x X. muelleri) were examined using a novel sperm assay that allowed quantification of live, dead, and undifferentiated sperm cells, the number of motile vs. immotile sperm, and sperm morphology. Hybrids exhibited a dramatically lower abundance of mature sperm relative to the parental species. Hybrid spermatozoa were larger in size and accompanied by numerous undifferentiated sperm cells. Microarray analysis of gene expression in testes was combined with a correction for sequence divergence derived from genomic hybridizations to identify candidate genes involved in the sterility phenotype. Analysis of the transcriptome revealed a striking asymmetric pattern of misexpression. There were only about 140 genes misexpressed in hybrids compared to X. laevis but nearly 4,000 genes misexpressed in hybrids compared to X. muelleri.

CONCLUSIONS/SIGNIFICANCE

Our results provide an important correlation between phenotypic characteristics of sperm and gene expression in sterile hybrid males. The broad pattern of gene misexpression suggests intriguing mechanisms creating the dominance pattern of the X. laevis genome in hybrids. These findings significantly contribute to growing evidence for allelic dominance in hybrids and have implications for the mechanism of species differentiation at the transcriptome level.

Ribosomal RNA gene silencing in interpopulation hybrids of Tigriopus californicus: nucleolar dominance in the absence of intergenic spacer subrepeats

A common feature of interspecific animal and plant hybrids is the uniparental silencing of ribosomal RNA gene transcription, or nucleolar dominance. A leading explanation for the genetic basis of nucleolar dominance in animal hybrids is the enhancer-imbalance model. The model proposes that limiting transcription factors are titrated by a greater number of enhancer-bearing subrepeat elements in the intergenic spacer (IGS) of the dominant cluster of genes. The importance of subrepeats for nucleolar dominance has repeatedly been supported in competition assays between Xenopus laevis and X. borealis minigene constructs injected into oocytes. However, a more general test of the importance of IGS subrepeats for nuclear dominance in vivo has not been conducted. In this report, rRNA gene expression was examined in interpopulation hybrids of the marine copepod Tigriopus californicus. This species offers a rare opportunity to test the role of IGS subrepeats in nucleolar dominance because the internal subrepeat structure, found in the IGS of virtually all animal and plant species, is absent in T. californicus. Our results clearly establish that nucleolar dominance occurs in F1 and F2 interpopulation hybrids of this species. In the F2 generation, nucleolar dominance appears to break down in some hybrids in a fashion that is inconsistent with a transcription factor titration model. These results are significant because they indicate that nucleolar dominance can be established and maintained without enhancer-bearing repeat elements in the IGS. This challenges the generality of the enhancer-imbalance model for nucleolar dominance and suggests that dominance of rRNA transcription in animals may be determined by epigenetic factors as has been established in plants.

Cooperative binding of the Xenopus RNA polymerase I transcription factor xUBF to repetitive ribosomal gene enhancers

Upstream binding factor (UBF) is a DNA-binding transcription factor implicated in ribosomal gene promoter and enhancer function in vertebrates. UBF is unusual in that it has multiple DNA-binding domains with homology to high-mobility-group (HMG) nonhistone chromosomal proteins 1 and 2. However, a recognizable DNA consensus sequence for UBF binding is lacking. In this study, we have used gel retardation and DNase I footprinting to examine Xenopus UBF (xUBF) binding to Xenopus laevis ribosomal gene enhancers. We show that UBF has a minimum requirement for about 60 bp of DNA, the size of the short enhancer variant in X. laevis. Stronger UBF binding occurs on the longer enhancer variant (81 bp) and on multiple enhancers linked head to tail. In vivo, Xenopus ribosomal gene enhancers exist in blocks of 10 alternating 60- and 81-bp repeats within the intergenic spacer. In vitro, UBF binds cooperatively to probes with 10 enhancers, with five intermediate complexes observed in titration experiments. This suggests that, on average, one UBF dimer binds every two enhancers. A single UBF dimer can produce a DNase I footprint ranging in size from approximately 30 to about 115 bp on enhancer probes of different lengths. This observation is consistent with the hypothesis that multiple DNA-binding domains or subdomains within UBF bind independently, forming more-stable interactions on longer probes.

Cooperative binding of the Xenopus RNA polymerase I transcription factor xUBF to repetitive ribosomal gene enhancers

Upstream binding factor (UBF) is a DNA-binding transcription factor implicated in ribosomal gene promoter and enhancer function in vertebrates. UBF is unusual in that it has multiple DNA-binding domains with homology to high-mobility-group (HMG) nonhistone chromosomal proteins 1 and 2. However, a recognizable DNA consensus sequence for UBF binding is lacking. In this study, we have used gel retardation and DNase I footprinting to examine Xenopus UBF (xUBF) binding to Xenopus laevis ribosomal gene enhancers. We show that UBF has a minimum requirement for about 60 bp of DNA, the size of the short enhancer variant in X. laevis. Stronger UBF binding occurs on the longer enhancer variant (81 bp) and on multiple enhancers linked head to tail. In vivo, Xenopus ribosomal gene enhancers exist in blocks of 10 alternating 60- and 81-bp repeats within the intergenic spacer. In vitro, UBF binds cooperatively to probes with 10 enhancers, with five intermediate complexes observed in titration experiments. This suggests that, on average, one UBF dimer binds every two enhancers. A single UBF dimer can produce a DNase I footprint ranging in size from approximately 30 to about 115 bp on enhancer probes of different lengths. This observation is consistent with the hypothesis that multiple DNA-binding domains or subdomains within UBF bind independently, forming more-stable interactions on longer probes.

Involvement of DNA topoisomerase I in transcription of human ribosomal RNA genes

Treatment of HeLa cells with a DNA topoisomerase I-specific inhibitor, camptothecin, results in rapid cessation of the synthesis of the 45S rRNA precursor. The inhibition of rRNA synthesis is reversible following drug removal and correlates with the presence of camptothecin-trapped topoisomerase I-DNA abortive complexes, which can be detected as topoisomerase I-linked DNA breaks upon lysis with sodium dodecyl sulfate. These breaks were found to be concentrated within the transcribed region of human rRNA genes. No such sites can be detected in the inactive human rRNA genes in mouse-human hybrid cells, suggesting a preferential association of topoisomerase I with actively transcribed genes. The distribution of RNA polymerase molecules along the transcription unit of human rRNA genes in camptothecin-treated HeLa cells, as assayed by nuclear run-on transcription, shows a graded decrease of the RNA polymerase density toward the 3' end of the transcription unit; the density is minimally affected near the 5' start of the transcription unit. These results suggest that DNA topoisomerase I is normally involved in the elongation step of transcription, especially when the transcripts are long, and that camptothecin interferes with this role.

Human ribosomal DNA fragments amplified in hamster cells are transcribed only by RNA polymerase II and are not silver stained

Cloned human rRNA gene fragments that included the promoter region were introduced into Chinese hamster dihydrofolate reductase-deficient (dhfr-) cells by cotransformation with a dhfr minigene and amplified by selection for methotrexate resistance. The human ribosomal DNA was transcribed by RNA polymerase II, not RNA polymerase I or III. The metaphase chromosome regions containing the transcriptionally active human ribosomal DNA failed to show silver staining.

Human ribosomal DNA fragments amplified in hamster cells are transcribed only by RNA polymerase II and are not silver stained

Cloned human rRNA gene fragments that included the promoter region were introduced into Chinese hamster dihydrofolate reductase-deficient (dhfr-) cells by cotransformation with a dhfr minigene and amplified by selection for methotrexate resistance. The human ribosomal DNA was transcribed by RNA polymerase II, not RNA polymerase I or III. The metaphase chromosome regions containing the transcriptionally active human ribosomal DNA failed to show silver staining.

Factors and nucleotide sequences that direct ribosomal DNA transcription and their relationship to the stable transcription complex

We have studied the protein components and nucleic acid sequences involved in stably activating the ribosomal DNA (rDNA) template and in directing accurate transcription of mammalian rRNA genes. Two protein components are necessary to catalyze rDNA transcription, and these have been extensively purified. The first, factor D, can stably associate by itself with the rDNA promoter region and is responsible for template commitment. The second component, factor C, which appears to be an activated subset of polymerase I, can stably bind to the factor D-rDNA complex but not to the rDNA in the absence of factor D. A third component which had been previously identified as a rDNA transcription factor is shown to be a RNase inhibitor. Extending our earlier observation that the approximately 150-base-pair mouse rDNA promoter consists of a minimal essential region (residues approximately -35 to approximately +9) and additional upstream stimulatory domains, we now report that each of these promoter domains acts to augment the binding of the polymerase I transcription factors. A minimum core region (residues approximately -35 to approximately -15) is capable of stable complex formation and of binding transcription factor D. Factor C can also bind to this D-core region complex.

Isolation, fractionation and reconstitution of a nuclear extract capable of transcribing ribosomal DNA

A procedure for preparing a nuclear extract that efficiently transcribes rat rDNA in vitro has been developed. This procedure, which is based on the protocol described by Dignam et al. (Nucl Acids Res 11:1475, 1983), allows the preparation of extract from large or small amounts of material and requires neither ultracentrifugation nor column chromatography. These extracts were found to be more efficient than other transcription systems. Extract prepared as described routinely synthesize 1-2 transcripts per linear template, and could synthesize upto 6 transcripts per linear template at an elongation rate of 2.1 nucleotides per second. 0.3 M NaCl extracts of nuclei contained RNA polymerase I, but did not transcribe rat rDNA in vitro, whereas extract prepared with 0.42 M NaCl did. The 0.42 M NaCl extract of nuclei was fractionated by chromatography on DEAE-Sephadex and heparin-Sepharose. Two activities were identified that were required for accurate in vitro transcription by endogenous RNA polymerase I. One of these activities was required for accurate initiation, and the second inhibited non-specific transcription. The fraction required for accurate initiation by the endogenous RNA polymerase I is that factor which directs species specific transcription, as it also directed the transcription of rat rDNA by nuclear extracts of HeLa cells. Combining that same chromatographic fraction of the 0.42 M NaCl extract with the 0.3 M NaCl extract resulted in specific transcription. These results suggest that a fraction of the RNA polymerase I molecules may exist in a complex with some, or all, of the factors required for transcription.

Factors and nucleotide sequences that direct ribosomal DNA transcription and their relationship to the stable transcription complex

We have studied the protein components and nucleic acid sequences involved in stably activating the ribosomal DNA (rDNA) template and in directing accurate transcription of mammalian rRNA genes. Two protein components are necessary to catalyze rDNA transcription, and these have been extensively purified. The first, factor D, can stably associate by itself with the rDNA promoter region and is responsible for template commitment. The second component, factor C, which appears to be an activated subset of polymerase I, can stably bind to the factor D-rDNA complex but not to the rDNA in the absence of factor D. A third component which had been previously identified as a rDNA transcription factor is shown to be a RNase inhibitor. Extending our earlier observation that the approximately 150-base-pair mouse rDNA promoter consists of a minimal essential region (residues approximately -35 to approximately +9) and additional upstream stimulatory domains, we now report that each of these promoter domains acts to augment the binding of the polymerase I transcription factors. A minimum core region (residues approximately -35 to approximately -15) is capable of stable complex formation and of binding transcription factor D. Factor C can also bind to this D-core region complex.

Transcription of mouse rDNA and associated formation of the nucleolus organizer region after gene transfer and amplification in Chinese hamster cells

Mouse rDNA can initiate transcription by using only Chinese hamster cell components, and this is associated with nucleolus organizer activity. To demonstrate this, we transferred a 3.2-kilobase segment of mouse rDNA containing the promoter, the transcription initiation site, and part of the external transcribed spacer to dihydrofolate reductase-deficient Chinese hamster cells by cotransformation with an abbreviated mouse dhfr gene. Stepwise selection for methotrexate resistance produced sublines in which the mouse rDNA was usually coamplified with the donor dhfr DNA and occupied the same site or sites in the hamster genome, as shown by in situ hybridization. Transcription from mouse rDNA was demonstrated in two such lines, and S1 protection mapping indicated faithful initiation of the transcript. In some cells from both lines, the chromosome segments containing amplified mouse rDNA showed multiple silver-staining regions (i.e., active nucleolus organizers). Although the transferred mouse rDNA was able to use the rDNA transcriptional machinery of the Chinese hamster, the level of transcription was much lower than expected from the rDNA copy number, and a large fraction of each amplified region showed no silver staining. Since the absence of silver staining is generally correlated with the absence of transcription, many copies of the amplified mouse rDNA may have been in a chromatin conformation in which they could not be transcribed. This was not associated with the extensive methylation seen in other amplified, inactive rDNA sequences.

Transcription of mouse rDNA and associated formation of the nucleolus organizer region after gene transfer and amplification in Chinese hamster cells

Mouse rDNA can initiate transcription by using only Chinese hamster cell components, and this is associated with nucleolus organizer activity. To demonstrate this, we transferred a 3.2-kilobase segment of mouse rDNA containing the promoter, the transcription initiation site, and part of the external transcribed spacer to dihydrofolate reductase-deficient Chinese hamster cells by cotransformation with an abbreviated mouse dhfr gene. Stepwise selection for methotrexate resistance produced sublines in which the mouse rDNA was usually coamplified with the donor dhfr DNA and occupied the same site or sites in the hamster genome, as shown by in situ hybridization. Transcription from mouse rDNA was demonstrated in two such lines, and S1 protection mapping indicated faithful initiation of the transcript. In some cells from both lines, the chromosome segments containing amplified mouse rDNA showed multiple silver-staining regions (i.e., active nucleolus organizers). Although the transferred mouse rDNA was able to use the rDNA transcriptional machinery of the Chinese hamster, the level of transcription was much lower than expected from the rDNA copy number, and a large fraction of each amplified region showed no silver staining. Since the absence of silver staining is generally correlated with the absence of transcription, many copies of the amplified mouse rDNA may have been in a chromatin conformation in which they could not be transcribed. This was not associated with the extensive methylation seen in other amplified, inactive rDNA sequences.

Presence of a limited number of essential nucleotides in the promoter region of mouse ribosomal RNA gene

Point mutations are introduced into a mouse rDNA fragment containing the promoter region by a sodium bisulfite method and the mutants are tested for the ability of accurate transcription initiation in vitro. The results indicate that the change, G to A, at -7 completely eliminates the promoter activity, and those at -16 and at -25 decrease it to about 10% and 50%, respectively. On the other hand, the substitutions at +9, +4, -2, -9 and -39 do not alter the template activity significantly. It is concluded that there are limited but distinct nucleotides that are essential for the transcription initiation of this gene. This sort of absolute requirement for single specific bases is not reported in protein coding genes transcribed by RNA polymerase II. We propose that these rigid recognition signals which we have found are the molecular basis for the strong species-dependency of the transcription machinery of RNA polymerase I system. A model is presented in which a transcription factor interacts with the rDNA promoter from one side of the DNA double-helix with essential contacts at these bases.

Characterization of a factor that can prevent random transcription of cloned rDNA and its probable relationship to poly(ADP-ribose) polymerase

A factor which eliminated nonspecific transcription of cloned rat rDNA was extensively purified from rat mammary adenocarcinoma ascites cells by successive fractionations on DEAE-Sephadex and heparin-Sepharose columns. The fractions containing RNA polymerase I (HS-B) and fractions eluting thereafter (HS-C) from the heparin-Sepharose column were pooled separately. Addition of HS-C to HS-B prevented random transcription of rDNA and yielded an accurate rDNA transcript with negligible non-specific transcription. The factor was essentially homogeneous and corresponded to Poly(ADP-ribose) polymerase with respect to molecular weight, dependence on DNA for its activity and its ability to undergo auto ADP-ribosylation. The total amount of protein in the transcription assay was approximately 2 micrograms, which indicates a high degree of purity of all the factors required for specific transcription of rDNA.