- Email: [email protected]
Ancient repeated DNA elements and the regulation of the human frataxin promoter
Genomics 85 (2005) 221 – 230 www.elsevier.com/locate/ygeno
Ancient repeated DNA elements and the regulation of the human $ frataxin promoter Eriko Greene, Ali Entezam, Daman Kumari, Karen Usdin* Section on Genomic Structure and Function, Laboratory of Molecular and Cellular Biology, National Institute of Diabetes, Digestive and Kidney Diseases, National Institutes of Health, Building 8, Room 202, 8 Center Drive MSC 0830, Bethesda, MD 20892-0830, USA Received 30 June 2004; accepted 23 October 2004
Abstract Friedreich ataxia results from frataxin insufficiency caused by repeat expansion in intron 1 of the frataxin gene. Since the coding sequence is unchanged, the potential exists to ameliorate symptoms by increasing frataxin promoter activity. We therefore defined the minimal frataxin promoter in humans. Despite the fact that frataxin is an essential gene, its promoter is not well conserved in mammals, in part because it has been the frequent target of retroelement insertions. Most of the activity of the human frataxin promoter can be attributed to these retroelements, illustrating how these elements, considered parasitic by some, have been co-opted to drive critical genes. Individuals with the milder French Acadian form and those with the classic form of the disease have no biologically relevant sequence differences in the promoter or 3VUTR, suggesting that some other region of the gene, perhaps the repeat itself, is responsible for the difference in disease severity. Published by Elsevier Inc. Keywords: Frataxin promoter; Alu; MIR; L2 elements
Friedreich ataxia (FRDA), a disease found in populations of European, North African, Middle Eastern, and Indian origin, is the most common recessively inherited ataxia [22]. It is a relentlessly progressive gait and limb ataxia, which is accompanied by dysarthria, loss of tendon reflexes, and skeletal abnormalities. Onset is often early with patients on average losing the ability to walk 15 years after the first symptoms appear. Diabetes is a frequent complication that is thought to result at least in part from free radical-induced apoptosis of pancreatic h cells [27]. Early mortality is not uncommon, most frequently due to hypertrophic cardiomyopathy [6,34]. FRDA results from loss-of-function mutations in the frataxin gene on chromosome 9p13 [5]. This gene encodes an essential protein, frataxin, which is thought to play a role
$
Sequence data from this article have been deposited with the GenBank Data Library under Accession Nos. AY599001–AY599014. * Corresponding author. Fax: +1 301 402 0053. E-mail address: [email protected] (K. Usdin). 0888-7543/$ - see front matter. Published by Elsevier Inc. doi:10.1016/j.ygeno.2004.10.013
in mitochondrial iron metabolism and the biogenesis of iron–sulfur clusters [22]. More than 98% of FRDA patients are homozygous for an expansion of a GAAd TTC repeat tract in the first intron of this gene. In general there is a correlation between the number of repeats in the smaller allele and the disease severity and an inverse correlation with age of onset [22]. Expanded repeats are associated with a deficit in fulllength frataxin mRNA. This deficit has been suggested to result from a problem with transcription elongation [4,9,10,28], although an effect on transcription initiation may also be possible [32]. One recent approach to ameliorating FRDA symptoms has been to try and increase frataxin expression [8,31,36]. However, almost nothing is known about the factors normally involved in the regulation of this gene. In an effort to understand better factors important for frataxin expression we have analyzed the region 5Vof the frataxin open reading frame in a number of different primates and rodents and identified the regions of the human sequence that are important for optimal promoter activity. Our data provide a molecular explanation for the
222
E. Greene et al. / Genomics 85 (2005) 221–230
effects of some drugs already shown to increase frataxin expression and also suggest other ways to increase frataxin gene expression. To understand better the molecular basis of the milder FRDA symptoms seen in individuals of French Acadian descent [2,3,14,20,26] we have also compared the regulatory regions of these individuals with those suffering from the classic form of the disease.
Results Defining the minimal human frataxin promoter To understand better the regulation of the frataxin gene, five constructs, FRDA4356, FRDA3433, FRDA2541, FRDA1862, and FRDA1255, containing 4356, 3433, 2541, 1862, and 1255 bp, respectively, upstream of the human frataxin open reading frame were generated by cloning PCRamplified fragments into pGL-Basic. The ability of these constructs to drive luciferase activity was measured after transfection into C2C12 mouse skeletal muscle myoblasts or myotubes. These cells were chosen since frataxin expression is relatively high in skeletal muscle, the pattern of frataxin expression is similar in mouse and human [12], and transgenic mice containing YACs and BACs carrying the human frataxin locus show levels of expression similar to those of endogenous murine frataxin, particularly in skeletal muscle [25,30]. Promoter activity was measured in cells grown in either 10% fetal calf serum, conditions in which the cells remain undifferentiated, or 2% serum, conditions that promote skeletal myotube formation. All five constructs resulted in an ~20-fold increase in luciferase activity above pGL-Basic in 10% serum (Fig. 1). The same result was seen when cells were grown in 2% serum. An increase in frataxin protein has been reported when mouse embryonic carcinoma cells P19 are induced to differentiate into cardiomyocytes [29]. It is unclear at this time whether this effect results from differentiation-related changes at the transcriptional or posttranscriptional level. In any event it does not seem to be a gene-specific phenomenon since a similar increase is seen in a number of other mitochondrial proteins and has been attributed to an increase in the number of mitochondria [29].
whose activity predated the divergence of rodents and primates (Fig. 2). The second region is located immediately upstream of the translation start site. A recent survey of 46 different human promoters showed that ~94% of them had significant sequence similarity with their rodent counterparts [38]. This puts the frataxin promoter into that small category of promoters in which significant amounts of the promoter do not seem to be evolving under strong selective pressure. However, while these promoters lack strict sequence similarity in their 5V ends, both the rodent and the human promoters contain related SINE elements, Alu in the case of human and B1 in the case of mouse (data not shown). Both of these elements originate from the same region of the 7SL RNA, a component of the signal recognition particle that is involved in translation of secreted proteins in eukaryotes [37]. Thus these elements may share regulatory elements originally derived from this RNA. The B1 element in rodents is embedded in a small piece of an old rodent-specific L1 retrotransposable element, Lx (data not shown). While very little human–rodent homology is seen upstream of the start of transcription, the Catarrhine primates Homo sapiens (humans) and Pan troglodytes (chimpanzee) and the Platyrrhine primate Ateles geoffroyi (spider monkey) share significant sequence similarity (Fig. 3). Both the human and the chimpanzee promoters contain Alu retroelements belonging to the AluJb and AluY subfamilies. The activity of AluJb predates the divergence of the Catarrhine lineage from that of the Platyrrhine primates, while AluY is believed to have been active in the Catarrhine lineage after this time. The AluY element is missing in the spider monkey, consistent with fact that the AluY family arose in the Catarrhine lineage. The primate promoters also contain a member of another very ancient repeated DNA family, known as MIR (mammalian-wide interspersed repeats), which was active in the common ancestor of all modern mammals. Despite the early origin of the MIR family, no such element is found in either the rat or the mouse frataxin promoters.
Sequence conservation in the minimal frataxin promoter Since the presence of sequences farther 5V of base 1255 has little effect on promoter activity over that provided by the FRDA1255 construct, we focused our subsequent analysis on the region of the frataxin 5V end contained within this construct. Sequence similarities between humans and rodents are limited to two small regions of patchy homology downstream of the transcription start site (TSS). The first region overlaps a region with sequence similarity to the ancient L2 family of retrotransposable elements,
Fig. 1. Effects of sequences at the 5Vend of the human frataxin gene on the activity of a reporter gene in mouse myoblasts. Promoter constructs were transfected into C2C12 cells as described under Materials and methods. The luciferase activity was measured, normalized to the activity of a cotransfected plasmid expressing the Renilla luciferase, and then expressed relative to the activity of the empty pGL3-Basic vector.
E. Greene et al. / Genomics 85 (2005) 221–230
The TSS in the human gene is located 220 bases upstream of the frataxin open reading frame [5]. No TATA-like sequence is found in its vicinity. However, the TSS coincides with the A in the sequence 5V-GCAGTCT-3V (Fig. 3). This sequence has a single base deviation from the mammalian initiator (Inr) sequence consensus 5V-YYA + 1NWYY-3V. In addition to the Inr-like sequence, the human frataxin 5V end contains a downstream promoter element (DPE)-like element with the sequence 5V-GGTTG-3V. The midpoint of this sequence is situated 26 bases downstream of the reported TSS. This places the DPE-like element in the appropriate place for it to contribute together with the Inrlike element to the core promoter activity. The human DPE is not conserved in rat and mouse, although the corresponding regions of the rodent sequences still conform to the mammalian DPE consensus. The rodent TSSs have not been identified, but both mouse and rat have a sequence with one mismatch from the Inr consensus within 30 bases of the DPE. Deletion of the repeats from the human promoter has a significant effect on gene expression Deletion of the 5V end of the AluJb element in FRDA936 reduced the promoter activity by more than 20% (Fig. 4). A larger deletion that also removed part of the 3V end of this element and all of the AluY element led to a slight increase in luciferase levels, suggesting that either the AluY does not contribute significantly to the promoter activity or it might in fact have a small negative effect on transcription. Successive deletions from the 5V end that remove various amounts of the MIR element (constructs FRDA542, FRDA442, and FRDA342) resulted in progressive declines in luciferase activity relative to the intact minimal promoter in FRDA1255 and the construct lacking Alu elements (FRDA570). The MIR
223
element thus contributes significantly to the activity of the frataxin promoter. MIR elements were active prior to the mammalian radiation and, while not present in the murine frataxin gene, are plentiful in the murine genome. It may thus be that some of the regulatory factors important for the MIR-derived contribution to the human frataxin promoter are still present in mouse cells. Further deletion of bases from the 5V end to generate FRDA221 had very little effect on luciferase activity. This deletion removes most of the unique sequence upstream of the TSS. It also removes the TSS and eliminates any resemblance of that region to an Inr element. In addition, a construct in which bases 282–189 were deleted from the F1255 construct did not significantly affect promoter activity. This region includes both the Inr-like element and the DPE-like element. Furthermore, a 100-bp region that included both sites was not sufficient to drive transcription in our assay. This suggests perhaps that these elements do not contribute to the activity of the core promoter. Deletion of an additional 100 bp from the 5Vend of FRDA221 to generate FRDA121 removes the conserved L2-like fragment. This deletion drastically reduces luciferase activity, suggesting that one or more elements important for the positive regulation of frataxin production are located in the deleted region. Further deletions of the 5V end to generate FRDA64 lead to an increase in luciferase activity. This suggests that a repressor of transcription or translation is located between 121 and 64. In addition since this construct has significant luciferase activity itself, the region between 64 and the start of translation contains sequences important for the positive regulation of frataxin production. These results taken together demonstrate that the sequences downstream of the TSS, which contain the only regions conserved between rodents and humans, account for ~60% of the luciferase activity in this assay.
Fig. 2. Alignment of the regions of similarity in the human, mouse, and rat frataxin 5Vends identified by HUMOR (human–mouse–rat), a special variant of the Multiz multiple sequence alignment program. The region shown begins 25 bases downstream of the reported transcription start site in humans. Gaps in the alignment are shown as dashes. Those bases conserved in rodents but that differ from humans are shown in the lowercase gray letters. Those bases differing in all three species and those found in humans and either rat or mouse but not both are shown in the gray capital letters. The numbering refers to the number of bases 5Vof the translation start site in the human sequence. The (N) in the mouse sequence denotes a run of an undetermined number of unknown bases in the sequence database.
224
E. Greene et al. / Genomics 85 (2005) 221–230
Fig. 3. Comparison of the 5Vends of the frataxin gene of H. sapiens, P. troglodytes, and A. geoffroyi. The last base shown in each instance corresponds to the base immediately preceding the beginning of the frataxin open reading frame. The black bars above the 3Vend of the sequence mark the regions conserved between primates and rodents. The gray bars below the sequence mark the locations of the two protein-binding sites identified in this study (see Fig. 5). The underlined sequences indicate the Inr-like sequence in which the TSS is embedded. The sequences shown in bold indicate the DPE-like sequences downstream of the Inr-like elements.
E. Greene et al. / Genomics 85 (2005) 221–230
225
Fig. 4. Analysis of the 5Vend of the frataxin gene. (A) Diagrammatic representation of the 1255 bases 5Vof the beginning of the human frataxin open reading frame. Note that the HindIII site shown in parentheses is derived from the reverse primer originally used to clone the frataxin 5Vend. The arrows at the bottom indicate the positions of the 5Vend of the primers used to generate some of the constructs used in the experiments shown in (B). The numbers refer to the position of the 5Vend of the primer relative to the beginning of the open reading frame. The remaining constructs were made using the restriction sites indicated on the diagram. The two primate-specific elements as well as the L2 element fragment are oriented with the direction of transcription being the same as that of the frataxin transcript, while the MIR element is transcribed in the opposite direction. (B) Luciferase activities of the indicated promoter deletion constructs. Equal amounts of the indicated plasmids were cotransfected with the pRL-null plasmid into mouse C2C12 cells as described under Materials and methods. Extracts from these cells were assayed using the Dual Luciferase assay system. The amount of luciferase detected was expressed relative to pGL-Basic after normalization to the cotransfected pRL-null plasmid. The X at the 5Vend of construct FRDA221 indicates that the TSS has been deleted in this construct.
Novel transcription protein factors may contribute to FRDA promoter activity To identify proteins that bind to the region of the promoter conserved in rodents and primates, we used the 221-bp PstI– HindIII fragment from FRDA1255 as a probe in electrophoretic mobility shift assays (EMSA). At 308C, only one band was observed in a variety of different extracts tested, including those from C2C12 cells and mouse liver and brains. At 48C, small amounts of two shifted products were seen with NIH 3T6 cells (Fig. 5). A similar pattern was seen with extracts from mouse brain and liver and C2C12 cells. The relatively small amounts of these shifted products may reflect suboptimal reaction conditions or preferential loss of these transcription factors during making of the nuclear extracts. Several possible transcription factor binding sites were identified by MatInspector, including AP4, SP1, OCT-1, E box, and SREBP. Oligonucleotides corresponding to the consensus binding sites for these transcription factors as well as those for AP1, CREBP, and TFIID were unable to compete for protein binding (data not shown), suggesting that these factors do not account for the observed shifted fragments. To identify the specific regions responsible for protein binding, smaller restriction fragments of the FRDA221 construct were used as competitors. The DraIII–XmnI
fragment shown in Fig. 5A was able to compete out the faster migrating DNA–protein complex, while the XmnI– HindIII fragment successfully competed out the slower migrating complex (Fig. 5B). Double-stranded oligonucleotides corresponding to the regions labeled 1 to 7 in Fig. 5A were also tested for their ability to compete. Two 30-bp regions were identified that were able to compete with the FRDA221 fragment for protein binding. The faster migrating shifted product was competed by an oligonucleotide homologous to the region 168–139 bp upstream of the coding sequence (oligonucleotide 3). This region is entirely contained within the small region that shares sequence similarity with L2. An oligonucleotide homologous to the region 85–56 bp from the translation start (oligonucleotide 6) competed with the slower band for binding. Neither of these regions contains transcription factor binding sites that are identified by MatInspector using the Transfac 5.0 database. The promoter and 3V UTR in individuals with the French Acadian vs. the classic form of FRDA A number of variant forms of FRDA have been described, including one seen in descendants of 17th century French settlers in Canada [3]. The French Acadian form of FRDA shows a later age of onset and generally
226
E. Greene et al. / Genomics 85 (2005) 221–230
Fig. 5. Electrophoretic gel mobility shift analysis of the region immediately 5Vof the start of the frataxin open reading frame. (A) Sequence of the PstI–HindIII fragment analyzed in (B) illustrating the relevant restriction sites and locations of the competitor oligonucleotides used. Note that the HindIII site shown in parentheses is derived from the reverse primer used to clone the frataxin 5Vend. (B) The PstI–HindIII fragment from the 5Vend of the frataxin gene was used as a probe. With the exception of the lane marked b Q, which contains the probe alone, all lanes contain the probe mixed with nuclear extract. The lane marked b+Q contains the probe, NIH 3T6 nuclear extract, and nonspecific competitor DNA as described under Materials and methods. The remaining lanes contain, in addition to nonspecific competitor DNA, the indicated specific competitors. The lanes labeled DraIII–XmnI and XmnI–HindIII contain the DraIII–XmnI and XmnI–HindIII fragments indicated in (A) as specific competitors. The lanes labeled 1–7 are the result of competition using double-stranded oligonucleotides whose locations correspond to the regions marked 1–7 in (A).
slower disease progression than in individuals with the classic form of the disease [2]. The genetic basis of the milder symptoms is not known, but its segregation is consistent with it being tightly linked to the frataxin locus [14,26] and does not seem to be related to repeat length [20]. A number of different mutations might be responsible for the milder symptoms. For example promoter mutations may increase transcription initiation, while mutations in the repeat tract or 3VUTR may affect transcription elongation or mRNA stability, respectively. Since cloning of long repeat tracts in bacteria and yeast frequently results in deletions, and amplification and sequencing of these tracts can also be very difficult, we decided to first examine the promoter and 3V UTR of
individuals with the French Acadian form of the disease and those with the classic form of FRDA. We thus cloned the promoter region from unaffected individuals, three individuals with the classic form of FRDA, and three affected individuals of Acadian descent. Sequence comparison showed a small number of positions in which these regions differed in the three groups but none that was specific to a particular group (data not shown). Moreover, none of these changes fell into those areas we have shown to be important for promoter activity. Comparison of the promoter activity using luciferase assays also revealed no significant differences between either of the patient groups and unaffected individuals (data not shown).
E. Greene et al. / Genomics 85 (2005) 221–230
The 3VUTR region was also cloned, this time downstream of the luciferase ORF in the pGL-Control vector. As was seen for the promoter region, there were several positions in the sequence at which individuals with the French Acadian form of FRDA differed from those with the classic form of the disease and from unaffected individuals. However, no mutations characteristic of any patient group were seen. In addition none of these changes affected mRNA stability as assessed by their effect on luciferase activity (data not shown).
Discussion The 1255-bp region extending 5Vfrom the frataxin open reading frame seems to contain most, if not all, of the components necessary for the activity of the human promoter in transient transfection assays in mouse skeletal muscle cells. Deletion studies indicate that this promoter, which lacks a TATA sequence, is not controlled by the Inr/ DPE-like elements found in the vicinity of the TSS. The frataxin promoter is thus unusual among mammalian promoters in not using some of the most common core promoter regulatory elements. It is also unusual in that the region upstream of the TSS shows no sequence conservation between primates and rodents. The frataxin promoter in both primates and rodents is rich in repetitive elements of different types and ages. Some, like MIR found in the human promoter, are very ancient retroposons that have not been active for many millions of years. Their insertion into the frataxin locus thus probably predates the mammalian radiation. Others, like Alu, are primate-specific while the related B1 elements are rodentspecific. The AluJb elements are relatively old Alu elements, having being active in retroposition 40–80 million years ago [13]. The AluJb element was thus presumably inserted into the frataxin locus after the primate ancestor diverged from the other mammals. AluY elements are younger members of the Alu family and are thought to have arisen about 20 million years ago in the common ancestor of Catarrhine primates (Old World monkeys, apes, and humans) [13]. The variety of different retroelement families and the different times at which they were inserted into the promoter suggest that remodeling of the frataxin promoter has occurred repeatedly during the evolution of mammals and fairly recently in the primate lineage. However, insertion of lineage-specific retroelements alone does not explain the lack of sequence similarity between rodents and humans in the region upstream of the transcription start site. For example, since the MIR element in the human promoter was probably inserted into this locus before the mammalian radiation but is missing from rodent promoters, other genome rearrangements and/or deletions also presumably helped shape this region. Deletion of the repetitive DNA elements, in particular the MIR and L2-like elements, had a significant effect on the
227
human frataxin promoter activity. This illustrates how repetitive DNA elements, considered by many to be bjunk DNAQ or bDNA parasites,Q have been adapted to play positive roles in the regulation of an essential gene. Alu, B1, and MIR elements are transcribed from internal Pol III promoters but RNA polymerase III promoter elements are known to enhance transcription of RNA polymerase II genes [21,35], which may account for some of the observed effect on frataxin promoter activity. Moreover, Alu and related elements like B1 are also thought to affect gene expression via an effect on nucleosome positioning [7]. Similarities in the factors that regulate the different repeats or a general effect of the repeats on chromatin organization may help explain the similarity in the regulation of the rodent and human promoters in the absence of strict sequence conservation. Repetitive elements tend to be heavily methylated [1,11,19,33] and are associated with histones carrying posttranslational modifications typical of heterochromatin [16]. In the case of Alu elements, this association with silent chromatin has been shown to be reversible by treatment with 5-aza-2V-deoxycytidine [16]. DNA-demethylating agents thus have the potential to increase the levels of frataxin mRNA. In this regard the recent demonstration that butyric acid, an inhibitor of histone deacetylases, also increases frataxin expression [31] is particularly interesting. Reversal of the heterochromatinization of the repeats may contribute to this effect. Other factors that increase the transcriptional activity of Alu elements such as heat or oxidative stress [15,17,18], or viral infection [23,24], may also lead to an increase in frataxin expression. We have demonstrated the presence of two proteinbinding sites in the region downstream of the TSS. Both binding sites fall into the only two regions of the frataxin 5V end that are conserved between rodents and primates. One binding site falls within the L2-like region and one is located downstream of the first binding site, 56–85 bp upstream of the frataxin ORF. These sites do not correspond to any known transcription factor-binding site. Since the L2-like region has a positive effect on gene expression, the as yet unidentified factor that binds this element is a positive regulator of frataxin gene expression. The factor that binds the downstream site, on the other hand, probably acts negatively since deletion of part of this region causes an increase in promoter activity. We have thus described two major regions of the frataxin 5V end, one upstream and one downstream of the TSS, that contribute to frataxin gene regulation in mouse skeletal cells. It is possible that other regions of the frataxin 5Vend will prove to have a role in other cell types or under other growth conditions. With respect to the molecular basis of the milder symptoms in individuals with the Acadian form of FRDA, we conclude that neither changes in the 5V end that lead to increased promoter initiation nor changes in the 3VUTR that increase mRNA stability are responsible. These observations lend support to the idea that differences in the GAA d TTC
228
E. Greene et al. / Genomics 85 (2005) 221–230
repeat tract in these individuals may ameliorate the frataxin deficit.
Materials and methods Genomic DNA Lymphoblasts from an individual without FRDA (GM09145) were obtained from the Coriell Cell Repository (Camden, NJ, USA). DNA was isolated from these cells according to standard procedures. DNA isolated from lymphocytes of FRDA patients of French Acadian descent (FA32, FA64, FA288) and those of non-Acadian descent (FA471, FA504, and FA510) were a gift from Dr. Ed Grabczyk (Louisiana State University, New Orleans, LA, USA). Genomic DNA from P. troglodytes and A. geoffroyi was obtained from the Coriell Cell Repository. FRDA promoter and 3VUTR constructs To minimize the possibility of PCR-induced sequence changes, Expand High Fidelity enzyme mix (Roche, Basel, Switzerland) and a maximum of 20 cycles of PCR were used for amplification. The region upstream of the human frataxin translation start site was amplified by PCR from the genomic DNA of an unaffected male. Five different upstream primers, F1255, F1862, F2541, F3433, and F4356, whose 5V ends are 1255, 1862, 2541, 3433, and 4356 bp upstream of the translation start site were used in conjunction with the same downstream primer (FRDA-R1, 5V-TTTAAGCTTCTGCTCCGGGTCTGCCGCCC-3V). Each of the upstream primers contained an XhoI site at their 5V end. The 5V end of the frataxin sequence in the FRDA-R1 primer corresponds to the region 2 bp upstream of the initiator methionine. This primer contained a HindIII site at its 5V end. MasterAmp PCR Buffer E (Epicentre, Madison, WI, USA) was used with the F1255-FRDA-R1 primer pair and buffer F (Epicentre) was used for the remaining primer combinations. PCR conditions were 958C for 5 min, followed by 20 cycles of denaturation at 958C for 30 s, annealing at 658C for 1 min, and extension at 728C for 5 min. This was followed by incubation at 728C for 10 min. The amplified products were then digested with XhoI and HindIII and cloned into pGL3-Basic (Promega, Madison, WI, USA). The clones were designated FRDAX, where X is the number of frataxin-derived bases in the clone as well as the distance of the 5Vend of the frataxin sequence from the start of translation. The clones were verified by sequencing. Deletions were made in the FRDA1255 construct by subcloning individual restriction or PCR fragments into pGL3-Basic. The primers FRDA1255 and FRDA-R1 were also used to amplify the 5V UTR from three individuals of French Acadian descent with FRDA and three individuals with FRDA who were not of French Acadian ancestry. The resultant PCR fragments were digested with XhoI and
HindIII and cloned into pGL3-Basic as described above. These clones were sequenced and the sequences of both affected and unaffected individuals were deposited with GenBank (GenBank Accession Nos. AY599001, AY 599002, AY599003, AY 599004, AY599005, AY599006, AY599007). A similar approach was used to clone the 5Vend of the frataxin gene from A. geoffroyi (spider monkey)and P. troglodytes (chimpanzee). The GenBank accession numbers for these sequences are AY599013 and AY599014, respectively. The frataxin 3V UTR was amplified using MasterAmp PCR premix Buffer C (Epicentre) and the primers Frataxin UTR-F (5V-TTTTCTAGATGCCCAGCCCCGTTTT-3V) and Frataxin UTR-R (5V-TTTTCTAGAAGATAGAACAGTGAGC-3V). The PCR conditions were 20 cycles of 958C for 30 s, 558C for 1 min, and 728C for 1 min. The PCR products were purified, digested with XbaI, and cloned into pGL3-Control downstream of the luciferase gene. The clones were verified by sequencing (GenBank Accession Nos. AY599008, AY599009, AY599010, AY599011, AY599012). These clones were then transfected into C2C12 cells and the luciferase activity measured as described above. In silico promoter analysis The sequence of the rat frataxin 5Vend was identified in the working draft of the rat (Rattus norvegicus, June 2003 assembly) at the UCSC Genome Bioinformatics server (http://genome.ucsc.edu) using the Blat alignment tool [39] and exon 1 of the mouse frataxin gene as the query sequence. The region upstream of the start of translation of the rat frataxin gene was used to search NCBI build 30 (UCSC version mm3) and build 32 (UCSC version mm4) of the mouse sequence database. In build 32 homology was found only to a region in an unordered contig from chromosome 19. However, in build 30, in which a different algorithm was used to assemble the contigs, a region of homology was identified immediately upstream of the mouse frataxin open reading frame on chromosome 19. Delineation and identification of the repeated sequences were carried out using RepeatMasker on the UCSC Genome Bioinformatics server. Comparisons of the rodent and primate promoters were carried out using Blast 2 (http://www.ncbi.nlm.nih.gov/blast/ bl2seq/bl2.html) and PipMaker (http://bio.cse.psu.edu/cgibin/pipmaker?basic [40]). HUMOR, a special variant of the Multiz multiple sequence alignment program [41], was used to identify regions of weak homology between rodents and humans. The resultant sequences were then aligned with ClustalW and the alignments optimized manually to minimize arbitrary gaps. Cell culture, transient transfections, and promoter assays Mouse C2C12 myoblasts were grown in Dulbecco’s modified Eagle’s medium (Invitrogen Life Technologies,
E. Greene et al. / Genomics 85 (2005) 221–230
Inc., Carlsbad, CA, USA) supplemented with 10 or 2% fetal bovine serum. Cotransfections using pRL-null and various test constructs were carried out using Fugene 6 (Roche) according to the supplier’s instructions. Luciferase activity was measured 30 h posttransfection using the Dual Luciferase Reporter assay system (Promega) and a Microlumat LB 96 P luminometer (Berthold Systems, Inc., Aliquippa, PA, USA).
[10]
[11]
[12]
EMSAs [13]
The 221-bp PstI–HindIII fragment from FRDA1255 was gel purified and radiolabeled using [a-32P]dCTP and Klenow (-exo) DNA polymerase (Fermentas, Vilnius, Lithuania) according to standard procedures. Nuclear extracts were prepared from different cell lines, including C2C12 and NIH 3T6, using NE-PER nuclear extraction reagents (Pierce, Rockford, IL, USA) according to the supplier’s instructions. Binding was carried out at 48C in 30 Al of reaction buffer containing 25 mM Hepes (pH 7.5), 40 mM NaCl, 1 mM EDTA, and 4 mM DTT with 0.25 ng/Al labeled probe in the presence of 60- to 1000-fold molar excess of various unlabeled competitors.
[14]
[15]
[16] [17]
[18]
[19]
Acknowledgments Thanks to Donna Karolchik and Terry Furey (University of California, Santa Cruz, CA, USA) with their help in navigating the different builds of the UCSC mouse genome database. Thanks also to Anthony Furano (NIDDK, NIH) for his careful reading of the manuscript.
[20]
[21]
References [22] [1] P. Arnaud, C. Goubely, T. Pelissier, J.M. Deragon, SINE retroposons can be used in vivo as nucleation centers for de novo methylation, Mol. Cell. Biol. 20 (2000) 3434 – 3441. [2] A. Barbeau, M. Roy, M. Sadibelouiz, M.A. Wilensky, Recessive ataxia in Acadians and bCajunsQ, Can. J. Neurol. Sci. 11 (1984) 526 – 533. [3] A. Barbeau, M. Sadibelouiz, M. Roy, B. Lemieux, J.P. Bouchard, G. Geoffroy, Origin of Friedreich’s disease in Quebec, Can. J. Neurol. Sci. 11 (1984) 506 – 509. [4] S.I. Bidichandani, T. Ashizawa, P.I. Patel, The GAA triplet-repeat expansion in Friedreich ataxia interferes with transcription and may be associated with an unusual DNA structure, Am. J. Hum. Genet. 62 (1998) 111 – 121. [5] V. Campuzano, L. Montermini, M.D. Molto, L. Pianese, M. Cossee, F. Cavalcanti, E. Monros, F. Rodius, F. Duclos, A. Monticelli, et al., Friedreich’s ataxia: autosomal recessive disease caused by an intronic GAA triplet repeat expansion, Science 271 (1996) 1423 – 1427. [6] J. Chaco, Friedreich’s ataxia with myocardial involvement and myopathy, Isr. J. Med. Sci. 2 (1966) 52 – 54. [7] E.W. Englander, B.H. Howard, Nucleosome positioning by human Alu elements in chromatin, J. Biol. Chem. 270 (1995) 10091 – 100916. [8] M. Ghazizadeh, Cisplatin may induce frataxin expression, J. Nippon Med. Sch. 70 (2003) 367 – 371. [9] E. Grabczyk, K. Usdin, Alleviating transcript insufficiency caused by
[23]
[24]
[25]
[26]
[27]
[28]
229
Friedreich’s ataxia triplet repeats, Nucleic Acids Res. 28 (2000) 4930 – 4937. E. Grabczyk, K. Usdin, The GAA*TTC triplet repeat expanded in Friedreich’s ataxia impedes transcription elongation by T7 RNA polymerase in a length and supercoil dependent manner, Nucleic Acids Res. 28 (2000) 2815 – 2822. U. Hellmann-Blumberg, M.F. Hintz, J.M. Gatewood, C.W. Schmid, Developmental differences in methylation of human Alu repeats, Mol. Cell. Biol. 13 (1993) 4523 – 4530. S. Jiralerspong, Y. Liu, L. Montermini, S. Stifani, M. Pandolfo, Frataxin shows developmentally regulated tissue-specific expression in the mouse embryo, Neurobiol. Dis. 4 (1997) 103 – 113. V. Kapitonov, J. Jurka, The age of Alu subfamilies, J. Mol. Evol. 42 (1996) 59 – 65. B.J. Keats, L.J. Ward, J. Shaw, A. Wickremasinghe, S. Chamberlain, bAcadianQ and bclassicalQ forms of Friedreich ataxia are most probably caused by mutations at the same locus, Am. J. Med. Genet. 33 (1989) 266 – 268. C. Kim, C.M. Rubin, C.W. Schmid, Genome-wide chromatin remodeling modulates the Alu heat shock response, Gene 276 (2001) 127 – 133. Y. Kondo, J.P. Issa, Enrichment for histone H3 lysine 9 methylation at Alu repeats in human cells, J. Biol. Chem. 278 (2003) 27658 – 27662. T.H. Li, C.W. Schmid, Differential stress induction of individual Alu loci: implications for transcription and retrotransposition, Gene 276 (2001) 135 – 141. W.M. Liu, W.M. Chu, P.V. Choudary, C.W. Schmid, Cell stress and translational inhibitors transiently increase the abundance of mammalian SINE transcripts, Nucleic Acids Res. 23 (1995) 1758 – 1765. W.M. Liu, C.W. Schmid, Proposed roles for DNA methylation in Alu transcriptional repression and mutational inactivation, Nucleic Acids Res. 21 (1993) 1351 – 1359. L. Montermini, A. Richter, K. Morgan, C.M. Justice, D. Julien, B. Castellotti, J. Mercier, J. Poirier, F. Capozzoli, J.P. Bouchard, B. Lemieux, J. Mathieu, M. Vanasse, M.H. Seni, G. Graham, F. Andermann, E. Andermann, S.B. Melancon, B.J. Keats, S. Di Donato, M. Pandolfo, Phenotypic variability in Friedreich ataxia: role of the associated GAA triplet repeat expansion, Ann. Neurol. 41 (1997) 675 – 682. S. Oliviero, P. Monaci, RNA polymerase III promoter elements enhance transcription of RNA polymerase II genes, Nucleic Acids Res. 16 (1988) 1285 – 1293. M. Pandolfo, The molecular basis of Friedreich ataxia, Adv. Exp. Med. Biol. 516 (2002) 99 – 118. B. Panning, J.R. Smiley, Activation of RNA polymerase III transcription of human Alu elements by herpes simplex virus, Virology 202 (1994) 408 – 417. B. Panning, J.R. Smiley, Activation of expression of multiple subfamilies of human Alu elements by adenovirus type 5 and herpes simplex virus type 1, J. Mol. Biol. 248 (1995) 513 – 524. M.A. Pook, S. Al-Mahdawi, C.J. Carroll, M. Cossee, H. Puccio, L. Lawrence, P. Clark, M.B. Lowrie, J.L. Bradley, J.M. Cooper, M. Koenig, S. Chamberlain, Rescue of the Friedreich’s ataxia knockout mouse by human YAC transgenesis, Neurogenetics 3 (2001) 185 – 193. A. Richter, J. Poirier, J. Mercier, D. Julien, K. Morgan, M. Roy, F. Gosselin, J.P. Bouchard, S.B. Melancon, Friedreich ataxia in Acadian families from eastern Canada: clinical diversity with conserved haplotypes, Am. J. Med. Genet. 64 (1996) 594 – 601. M. Ristow, H. Mulder, D. Pomplun, T.J. Schulz, K. MullerSchmehl, A. Krause, M. Fex, H. Puccio, J. Muller, F. Isken, J. Spranger, D. Muller-Wieland, M.A. Magnuson, M. Mohlig, M. Koenig, A.F. Pfeiffer, Frataxin deficiency in pancreatic islets causes diabetes due to loss of beta cell mass, J. Clin. Invest. 112 (2003) 527 – 534. N. Sakamoto, P.D. Chastain, P. Parniewski, K. Ohshima, M. Pandolfo, J.D. Griffith, R.D. Wells, Sticky DNA: self-association properties of
230
[29]
[30]
[31]
[32]
[33] [34]
E. Greene et al. / Genomics 85 (2005) 221–230 long GAAd TTC repeats in Rd Rd Y triplex structures from Friedreich’s ataxia, Mol. Cell 3 (1999) 465 – 475. M.M. Santos, K. Ohshima, M. Pandolfo, Frataxin deficiency enhances apoptosis in cells differentiating into neuroectoderm, Hum. Mol. Genet. 10 (2001) 1935 – 1944. J.P. Sarsero, L. Li, T.P. Holloway, L. Voullaire, S. Gazeas, K.J. Fowler, D.M. Kirby, D.R. Thorburn, A. Galle, S. Cheema, M. Koenig, R. Williamson, P.A. Ioannou, Human BAC-mediated rescue of the Friedreich ataxia knockout mutation in transgenic mice, Mamm. Genome 15 (2004) 370 – 382. J.P. Sarsero, L. Li, H. Wardan, K. Sitte, R. Williamson, P.A. Ioannou, Upregulation of expression from the FRDA genomic locus for the therapy of Friedreich ataxia, J. Gene Med. 5 (2003) 72 – 81. A. Saveliev, C. Everett, T. Sharpe, Z. Webster, R. Festenstein, DNA triplet repeats mediate heterochromatin-protein-1-sensitive variegated gene silencing, Nature 422 (2003) 909 – 913. C.W. Schmid, Human Alu subfamilies and their methylation revealed by blot hybridization, Nucleic Acids Res. 19 (1991) 5613 – 5617. C. Thor’En, Cardiomyopathy in Friedreich’s ataxia: with studies of cardiovascular and respiratory function, Acta Paediatr. 53 (Suppl. 153) (1964) 1.
[35] N.V. Tomilin, Control of genes by mammalian retroposons, Int. Rev. Cytol. 186 (1999) 1 – 48. [36] M. Turano, A. Tammaro, I. De Biase, M.S. Lo Casale, G. Ruggiero, A. Monticelli, S. Cocozza, L. Pianese, 3-Nitropropionic acid increases frataxin expression in human lymphoblasts and in transgenic rat PC12 cells, Neurosci. Lett. 350 (2003) 184 – 186. [37] E. Ullu, C. Tschudi, Alu sequences are processed 7SL RNA genes, Nature 312 (1984) 171 – 172. [38] E.T. Dermitzakis, A.G. Clark, Evolution of transcription factor binding sites in Mammalian gene regulatory regions: conservation and turnover, Mol. Biol. Evol. 19 (2002) 1114 – 1121. [39] W.J. Kent, C.W. Sugnet, T.S. Furey, K.M. Roskin, T.H. Pringle, A.M. Zahler, D. Haussler, The human genome browser at UCSC, Genome Res. 12 (2002) 996 – 1006. [40] S. Schwartz, Z. Zhang, K.A. Frazer, A. Smit, C. Riemer, J. Bouck, R. Gibbs, R. Hardison, W. Miller, PipMaker–a web server for aligning two genomic DNA sequences, Genome Res. 10 (2000) 577 – 586. [41] S. Schwartz, W.J. Kent, A. Smit, Z. Zhang, R. Baertsch, R.C. Hardison, D. Haussler, W. Miller, Human-mouse alignments with BLASTZ, Genome Res. 13 (2003) 103 – 107.
Ancient repeated DNA elements and the regulation of the human $ frataxin promoter Eriko Greene, Ali Entezam, Daman Kumari, Karen Usdin* Section on Genomic Structure and Function, Laboratory of Molecular and Cellular Biology, National Institute of Diabetes, Digestive and Kidney Diseases, National Institutes of Health, Building 8, Room 202, 8 Center Drive MSC 0830, Bethesda, MD 20892-0830, USA Received 30 June 2004; accepted 23 October 2004
Abstract Friedreich ataxia results from frataxin insufficiency caused by repeat expansion in intron 1 of the frataxin gene. Since the coding sequence is unchanged, the potential exists to ameliorate symptoms by increasing frataxin promoter activity. We therefore defined the minimal frataxin promoter in humans. Despite the fact that frataxin is an essential gene, its promoter is not well conserved in mammals, in part because it has been the frequent target of retroelement insertions. Most of the activity of the human frataxin promoter can be attributed to these retroelements, illustrating how these elements, considered parasitic by some, have been co-opted to drive critical genes. Individuals with the milder French Acadian form and those with the classic form of the disease have no biologically relevant sequence differences in the promoter or 3VUTR, suggesting that some other region of the gene, perhaps the repeat itself, is responsible for the difference in disease severity. Published by Elsevier Inc. Keywords: Frataxin promoter; Alu; MIR; L2 elements
Friedreich ataxia (FRDA), a disease found in populations of European, North African, Middle Eastern, and Indian origin, is the most common recessively inherited ataxia [22]. It is a relentlessly progressive gait and limb ataxia, which is accompanied by dysarthria, loss of tendon reflexes, and skeletal abnormalities. Onset is often early with patients on average losing the ability to walk 15 years after the first symptoms appear. Diabetes is a frequent complication that is thought to result at least in part from free radical-induced apoptosis of pancreatic h cells [27]. Early mortality is not uncommon, most frequently due to hypertrophic cardiomyopathy [6,34]. FRDA results from loss-of-function mutations in the frataxin gene on chromosome 9p13 [5]. This gene encodes an essential protein, frataxin, which is thought to play a role
$
Sequence data from this article have been deposited with the GenBank Data Library under Accession Nos. AY599001–AY599014. * Corresponding author. Fax: +1 301 402 0053. E-mail address: [email protected] (K. Usdin). 0888-7543/$ - see front matter. Published by Elsevier Inc. doi:10.1016/j.ygeno.2004.10.013
in mitochondrial iron metabolism and the biogenesis of iron–sulfur clusters [22]. More than 98% of FRDA patients are homozygous for an expansion of a GAAd TTC repeat tract in the first intron of this gene. In general there is a correlation between the number of repeats in the smaller allele and the disease severity and an inverse correlation with age of onset [22]. Expanded repeats are associated with a deficit in fulllength frataxin mRNA. This deficit has been suggested to result from a problem with transcription elongation [4,9,10,28], although an effect on transcription initiation may also be possible [32]. One recent approach to ameliorating FRDA symptoms has been to try and increase frataxin expression [8,31,36]. However, almost nothing is known about the factors normally involved in the regulation of this gene. In an effort to understand better factors important for frataxin expression we have analyzed the region 5Vof the frataxin open reading frame in a number of different primates and rodents and identified the regions of the human sequence that are important for optimal promoter activity. Our data provide a molecular explanation for the
222
E. Greene et al. / Genomics 85 (2005) 221–230
effects of some drugs already shown to increase frataxin expression and also suggest other ways to increase frataxin gene expression. To understand better the molecular basis of the milder FRDA symptoms seen in individuals of French Acadian descent [2,3,14,20,26] we have also compared the regulatory regions of these individuals with those suffering from the classic form of the disease.
Results Defining the minimal human frataxin promoter To understand better the regulation of the frataxin gene, five constructs, FRDA4356, FRDA3433, FRDA2541, FRDA1862, and FRDA1255, containing 4356, 3433, 2541, 1862, and 1255 bp, respectively, upstream of the human frataxin open reading frame were generated by cloning PCRamplified fragments into pGL-Basic. The ability of these constructs to drive luciferase activity was measured after transfection into C2C12 mouse skeletal muscle myoblasts or myotubes. These cells were chosen since frataxin expression is relatively high in skeletal muscle, the pattern of frataxin expression is similar in mouse and human [12], and transgenic mice containing YACs and BACs carrying the human frataxin locus show levels of expression similar to those of endogenous murine frataxin, particularly in skeletal muscle [25,30]. Promoter activity was measured in cells grown in either 10% fetal calf serum, conditions in which the cells remain undifferentiated, or 2% serum, conditions that promote skeletal myotube formation. All five constructs resulted in an ~20-fold increase in luciferase activity above pGL-Basic in 10% serum (Fig. 1). The same result was seen when cells were grown in 2% serum. An increase in frataxin protein has been reported when mouse embryonic carcinoma cells P19 are induced to differentiate into cardiomyocytes [29]. It is unclear at this time whether this effect results from differentiation-related changes at the transcriptional or posttranscriptional level. In any event it does not seem to be a gene-specific phenomenon since a similar increase is seen in a number of other mitochondrial proteins and has been attributed to an increase in the number of mitochondria [29].
whose activity predated the divergence of rodents and primates (Fig. 2). The second region is located immediately upstream of the translation start site. A recent survey of 46 different human promoters showed that ~94% of them had significant sequence similarity with their rodent counterparts [38]. This puts the frataxin promoter into that small category of promoters in which significant amounts of the promoter do not seem to be evolving under strong selective pressure. However, while these promoters lack strict sequence similarity in their 5V ends, both the rodent and the human promoters contain related SINE elements, Alu in the case of human and B1 in the case of mouse (data not shown). Both of these elements originate from the same region of the 7SL RNA, a component of the signal recognition particle that is involved in translation of secreted proteins in eukaryotes [37]. Thus these elements may share regulatory elements originally derived from this RNA. The B1 element in rodents is embedded in a small piece of an old rodent-specific L1 retrotransposable element, Lx (data not shown). While very little human–rodent homology is seen upstream of the start of transcription, the Catarrhine primates Homo sapiens (humans) and Pan troglodytes (chimpanzee) and the Platyrrhine primate Ateles geoffroyi (spider monkey) share significant sequence similarity (Fig. 3). Both the human and the chimpanzee promoters contain Alu retroelements belonging to the AluJb and AluY subfamilies. The activity of AluJb predates the divergence of the Catarrhine lineage from that of the Platyrrhine primates, while AluY is believed to have been active in the Catarrhine lineage after this time. The AluY element is missing in the spider monkey, consistent with fact that the AluY family arose in the Catarrhine lineage. The primate promoters also contain a member of another very ancient repeated DNA family, known as MIR (mammalian-wide interspersed repeats), which was active in the common ancestor of all modern mammals. Despite the early origin of the MIR family, no such element is found in either the rat or the mouse frataxin promoters.
Sequence conservation in the minimal frataxin promoter Since the presence of sequences farther 5V of base 1255 has little effect on promoter activity over that provided by the FRDA1255 construct, we focused our subsequent analysis on the region of the frataxin 5V end contained within this construct. Sequence similarities between humans and rodents are limited to two small regions of patchy homology downstream of the transcription start site (TSS). The first region overlaps a region with sequence similarity to the ancient L2 family of retrotransposable elements,
Fig. 1. Effects of sequences at the 5Vend of the human frataxin gene on the activity of a reporter gene in mouse myoblasts. Promoter constructs were transfected into C2C12 cells as described under Materials and methods. The luciferase activity was measured, normalized to the activity of a cotransfected plasmid expressing the Renilla luciferase, and then expressed relative to the activity of the empty pGL3-Basic vector.
E. Greene et al. / Genomics 85 (2005) 221–230
The TSS in the human gene is located 220 bases upstream of the frataxin open reading frame [5]. No TATA-like sequence is found in its vicinity. However, the TSS coincides with the A in the sequence 5V-GCAGTCT-3V (Fig. 3). This sequence has a single base deviation from the mammalian initiator (Inr) sequence consensus 5V-YYA + 1NWYY-3V. In addition to the Inr-like sequence, the human frataxin 5V end contains a downstream promoter element (DPE)-like element with the sequence 5V-GGTTG-3V. The midpoint of this sequence is situated 26 bases downstream of the reported TSS. This places the DPE-like element in the appropriate place for it to contribute together with the Inrlike element to the core promoter activity. The human DPE is not conserved in rat and mouse, although the corresponding regions of the rodent sequences still conform to the mammalian DPE consensus. The rodent TSSs have not been identified, but both mouse and rat have a sequence with one mismatch from the Inr consensus within 30 bases of the DPE. Deletion of the repeats from the human promoter has a significant effect on gene expression Deletion of the 5V end of the AluJb element in FRDA936 reduced the promoter activity by more than 20% (Fig. 4). A larger deletion that also removed part of the 3V end of this element and all of the AluY element led to a slight increase in luciferase levels, suggesting that either the AluY does not contribute significantly to the promoter activity or it might in fact have a small negative effect on transcription. Successive deletions from the 5V end that remove various amounts of the MIR element (constructs FRDA542, FRDA442, and FRDA342) resulted in progressive declines in luciferase activity relative to the intact minimal promoter in FRDA1255 and the construct lacking Alu elements (FRDA570). The MIR
223
element thus contributes significantly to the activity of the frataxin promoter. MIR elements were active prior to the mammalian radiation and, while not present in the murine frataxin gene, are plentiful in the murine genome. It may thus be that some of the regulatory factors important for the MIR-derived contribution to the human frataxin promoter are still present in mouse cells. Further deletion of bases from the 5V end to generate FRDA221 had very little effect on luciferase activity. This deletion removes most of the unique sequence upstream of the TSS. It also removes the TSS and eliminates any resemblance of that region to an Inr element. In addition, a construct in which bases 282–189 were deleted from the F1255 construct did not significantly affect promoter activity. This region includes both the Inr-like element and the DPE-like element. Furthermore, a 100-bp region that included both sites was not sufficient to drive transcription in our assay. This suggests perhaps that these elements do not contribute to the activity of the core promoter. Deletion of an additional 100 bp from the 5Vend of FRDA221 to generate FRDA121 removes the conserved L2-like fragment. This deletion drastically reduces luciferase activity, suggesting that one or more elements important for the positive regulation of frataxin production are located in the deleted region. Further deletions of the 5V end to generate FRDA64 lead to an increase in luciferase activity. This suggests that a repressor of transcription or translation is located between 121 and 64. In addition since this construct has significant luciferase activity itself, the region between 64 and the start of translation contains sequences important for the positive regulation of frataxin production. These results taken together demonstrate that the sequences downstream of the TSS, which contain the only regions conserved between rodents and humans, account for ~60% of the luciferase activity in this assay.
Fig. 2. Alignment of the regions of similarity in the human, mouse, and rat frataxin 5Vends identified by HUMOR (human–mouse–rat), a special variant of the Multiz multiple sequence alignment program. The region shown begins 25 bases downstream of the reported transcription start site in humans. Gaps in the alignment are shown as dashes. Those bases conserved in rodents but that differ from humans are shown in the lowercase gray letters. Those bases differing in all three species and those found in humans and either rat or mouse but not both are shown in the gray capital letters. The numbering refers to the number of bases 5Vof the translation start site in the human sequence. The (N) in the mouse sequence denotes a run of an undetermined number of unknown bases in the sequence database.
224
E. Greene et al. / Genomics 85 (2005) 221–230
Fig. 3. Comparison of the 5Vends of the frataxin gene of H. sapiens, P. troglodytes, and A. geoffroyi. The last base shown in each instance corresponds to the base immediately preceding the beginning of the frataxin open reading frame. The black bars above the 3Vend of the sequence mark the regions conserved between primates and rodents. The gray bars below the sequence mark the locations of the two protein-binding sites identified in this study (see Fig. 5). The underlined sequences indicate the Inr-like sequence in which the TSS is embedded. The sequences shown in bold indicate the DPE-like sequences downstream of the Inr-like elements.
E. Greene et al. / Genomics 85 (2005) 221–230
225
Fig. 4. Analysis of the 5Vend of the frataxin gene. (A) Diagrammatic representation of the 1255 bases 5Vof the beginning of the human frataxin open reading frame. Note that the HindIII site shown in parentheses is derived from the reverse primer originally used to clone the frataxin 5Vend. The arrows at the bottom indicate the positions of the 5Vend of the primers used to generate some of the constructs used in the experiments shown in (B). The numbers refer to the position of the 5Vend of the primer relative to the beginning of the open reading frame. The remaining constructs were made using the restriction sites indicated on the diagram. The two primate-specific elements as well as the L2 element fragment are oriented with the direction of transcription being the same as that of the frataxin transcript, while the MIR element is transcribed in the opposite direction. (B) Luciferase activities of the indicated promoter deletion constructs. Equal amounts of the indicated plasmids were cotransfected with the pRL-null plasmid into mouse C2C12 cells as described under Materials and methods. Extracts from these cells were assayed using the Dual Luciferase assay system. The amount of luciferase detected was expressed relative to pGL-Basic after normalization to the cotransfected pRL-null plasmid. The X at the 5Vend of construct FRDA221 indicates that the TSS has been deleted in this construct.
Novel transcription protein factors may contribute to FRDA promoter activity To identify proteins that bind to the region of the promoter conserved in rodents and primates, we used the 221-bp PstI– HindIII fragment from FRDA1255 as a probe in electrophoretic mobility shift assays (EMSA). At 308C, only one band was observed in a variety of different extracts tested, including those from C2C12 cells and mouse liver and brains. At 48C, small amounts of two shifted products were seen with NIH 3T6 cells (Fig. 5). A similar pattern was seen with extracts from mouse brain and liver and C2C12 cells. The relatively small amounts of these shifted products may reflect suboptimal reaction conditions or preferential loss of these transcription factors during making of the nuclear extracts. Several possible transcription factor binding sites were identified by MatInspector, including AP4, SP1, OCT-1, E box, and SREBP. Oligonucleotides corresponding to the consensus binding sites for these transcription factors as well as those for AP1, CREBP, and TFIID were unable to compete for protein binding (data not shown), suggesting that these factors do not account for the observed shifted fragments. To identify the specific regions responsible for protein binding, smaller restriction fragments of the FRDA221 construct were used as competitors. The DraIII–XmnI
fragment shown in Fig. 5A was able to compete out the faster migrating DNA–protein complex, while the XmnI– HindIII fragment successfully competed out the slower migrating complex (Fig. 5B). Double-stranded oligonucleotides corresponding to the regions labeled 1 to 7 in Fig. 5A were also tested for their ability to compete. Two 30-bp regions were identified that were able to compete with the FRDA221 fragment for protein binding. The faster migrating shifted product was competed by an oligonucleotide homologous to the region 168–139 bp upstream of the coding sequence (oligonucleotide 3). This region is entirely contained within the small region that shares sequence similarity with L2. An oligonucleotide homologous to the region 85–56 bp from the translation start (oligonucleotide 6) competed with the slower band for binding. Neither of these regions contains transcription factor binding sites that are identified by MatInspector using the Transfac 5.0 database. The promoter and 3V UTR in individuals with the French Acadian vs. the classic form of FRDA A number of variant forms of FRDA have been described, including one seen in descendants of 17th century French settlers in Canada [3]. The French Acadian form of FRDA shows a later age of onset and generally
226
E. Greene et al. / Genomics 85 (2005) 221–230
Fig. 5. Electrophoretic gel mobility shift analysis of the region immediately 5Vof the start of the frataxin open reading frame. (A) Sequence of the PstI–HindIII fragment analyzed in (B) illustrating the relevant restriction sites and locations of the competitor oligonucleotides used. Note that the HindIII site shown in parentheses is derived from the reverse primer used to clone the frataxin 5Vend. (B) The PstI–HindIII fragment from the 5Vend of the frataxin gene was used as a probe. With the exception of the lane marked b Q, which contains the probe alone, all lanes contain the probe mixed with nuclear extract. The lane marked b+Q contains the probe, NIH 3T6 nuclear extract, and nonspecific competitor DNA as described under Materials and methods. The remaining lanes contain, in addition to nonspecific competitor DNA, the indicated specific competitors. The lanes labeled DraIII–XmnI and XmnI–HindIII contain the DraIII–XmnI and XmnI–HindIII fragments indicated in (A) as specific competitors. The lanes labeled 1–7 are the result of competition using double-stranded oligonucleotides whose locations correspond to the regions marked 1–7 in (A).
slower disease progression than in individuals with the classic form of the disease [2]. The genetic basis of the milder symptoms is not known, but its segregation is consistent with it being tightly linked to the frataxin locus [14,26] and does not seem to be related to repeat length [20]. A number of different mutations might be responsible for the milder symptoms. For example promoter mutations may increase transcription initiation, while mutations in the repeat tract or 3VUTR may affect transcription elongation or mRNA stability, respectively. Since cloning of long repeat tracts in bacteria and yeast frequently results in deletions, and amplification and sequencing of these tracts can also be very difficult, we decided to first examine the promoter and 3V UTR of
individuals with the French Acadian form of the disease and those with the classic form of FRDA. We thus cloned the promoter region from unaffected individuals, three individuals with the classic form of FRDA, and three affected individuals of Acadian descent. Sequence comparison showed a small number of positions in which these regions differed in the three groups but none that was specific to a particular group (data not shown). Moreover, none of these changes fell into those areas we have shown to be important for promoter activity. Comparison of the promoter activity using luciferase assays also revealed no significant differences between either of the patient groups and unaffected individuals (data not shown).
E. Greene et al. / Genomics 85 (2005) 221–230
The 3VUTR region was also cloned, this time downstream of the luciferase ORF in the pGL-Control vector. As was seen for the promoter region, there were several positions in the sequence at which individuals with the French Acadian form of FRDA differed from those with the classic form of the disease and from unaffected individuals. However, no mutations characteristic of any patient group were seen. In addition none of these changes affected mRNA stability as assessed by their effect on luciferase activity (data not shown).
Discussion The 1255-bp region extending 5Vfrom the frataxin open reading frame seems to contain most, if not all, of the components necessary for the activity of the human promoter in transient transfection assays in mouse skeletal muscle cells. Deletion studies indicate that this promoter, which lacks a TATA sequence, is not controlled by the Inr/ DPE-like elements found in the vicinity of the TSS. The frataxin promoter is thus unusual among mammalian promoters in not using some of the most common core promoter regulatory elements. It is also unusual in that the region upstream of the TSS shows no sequence conservation between primates and rodents. The frataxin promoter in both primates and rodents is rich in repetitive elements of different types and ages. Some, like MIR found in the human promoter, are very ancient retroposons that have not been active for many millions of years. Their insertion into the frataxin locus thus probably predates the mammalian radiation. Others, like Alu, are primate-specific while the related B1 elements are rodentspecific. The AluJb elements are relatively old Alu elements, having being active in retroposition 40–80 million years ago [13]. The AluJb element was thus presumably inserted into the frataxin locus after the primate ancestor diverged from the other mammals. AluY elements are younger members of the Alu family and are thought to have arisen about 20 million years ago in the common ancestor of Catarrhine primates (Old World monkeys, apes, and humans) [13]. The variety of different retroelement families and the different times at which they were inserted into the promoter suggest that remodeling of the frataxin promoter has occurred repeatedly during the evolution of mammals and fairly recently in the primate lineage. However, insertion of lineage-specific retroelements alone does not explain the lack of sequence similarity between rodents and humans in the region upstream of the transcription start site. For example, since the MIR element in the human promoter was probably inserted into this locus before the mammalian radiation but is missing from rodent promoters, other genome rearrangements and/or deletions also presumably helped shape this region. Deletion of the repetitive DNA elements, in particular the MIR and L2-like elements, had a significant effect on the
227
human frataxin promoter activity. This illustrates how repetitive DNA elements, considered by many to be bjunk DNAQ or bDNA parasites,Q have been adapted to play positive roles in the regulation of an essential gene. Alu, B1, and MIR elements are transcribed from internal Pol III promoters but RNA polymerase III promoter elements are known to enhance transcription of RNA polymerase II genes [21,35], which may account for some of the observed effect on frataxin promoter activity. Moreover, Alu and related elements like B1 are also thought to affect gene expression via an effect on nucleosome positioning [7]. Similarities in the factors that regulate the different repeats or a general effect of the repeats on chromatin organization may help explain the similarity in the regulation of the rodent and human promoters in the absence of strict sequence conservation. Repetitive elements tend to be heavily methylated [1,11,19,33] and are associated with histones carrying posttranslational modifications typical of heterochromatin [16]. In the case of Alu elements, this association with silent chromatin has been shown to be reversible by treatment with 5-aza-2V-deoxycytidine [16]. DNA-demethylating agents thus have the potential to increase the levels of frataxin mRNA. In this regard the recent demonstration that butyric acid, an inhibitor of histone deacetylases, also increases frataxin expression [31] is particularly interesting. Reversal of the heterochromatinization of the repeats may contribute to this effect. Other factors that increase the transcriptional activity of Alu elements such as heat or oxidative stress [15,17,18], or viral infection [23,24], may also lead to an increase in frataxin expression. We have demonstrated the presence of two proteinbinding sites in the region downstream of the TSS. Both binding sites fall into the only two regions of the frataxin 5V end that are conserved between rodents and primates. One binding site falls within the L2-like region and one is located downstream of the first binding site, 56–85 bp upstream of the frataxin ORF. These sites do not correspond to any known transcription factor-binding site. Since the L2-like region has a positive effect on gene expression, the as yet unidentified factor that binds this element is a positive regulator of frataxin gene expression. The factor that binds the downstream site, on the other hand, probably acts negatively since deletion of part of this region causes an increase in promoter activity. We have thus described two major regions of the frataxin 5V end, one upstream and one downstream of the TSS, that contribute to frataxin gene regulation in mouse skeletal cells. It is possible that other regions of the frataxin 5Vend will prove to have a role in other cell types or under other growth conditions. With respect to the molecular basis of the milder symptoms in individuals with the Acadian form of FRDA, we conclude that neither changes in the 5V end that lead to increased promoter initiation nor changes in the 3VUTR that increase mRNA stability are responsible. These observations lend support to the idea that differences in the GAA d TTC
228
E. Greene et al. / Genomics 85 (2005) 221–230
repeat tract in these individuals may ameliorate the frataxin deficit.
Materials and methods Genomic DNA Lymphoblasts from an individual without FRDA (GM09145) were obtained from the Coriell Cell Repository (Camden, NJ, USA). DNA was isolated from these cells according to standard procedures. DNA isolated from lymphocytes of FRDA patients of French Acadian descent (FA32, FA64, FA288) and those of non-Acadian descent (FA471, FA504, and FA510) were a gift from Dr. Ed Grabczyk (Louisiana State University, New Orleans, LA, USA). Genomic DNA from P. troglodytes and A. geoffroyi was obtained from the Coriell Cell Repository. FRDA promoter and 3VUTR constructs To minimize the possibility of PCR-induced sequence changes, Expand High Fidelity enzyme mix (Roche, Basel, Switzerland) and a maximum of 20 cycles of PCR were used for amplification. The region upstream of the human frataxin translation start site was amplified by PCR from the genomic DNA of an unaffected male. Five different upstream primers, F1255, F1862, F2541, F3433, and F4356, whose 5V ends are 1255, 1862, 2541, 3433, and 4356 bp upstream of the translation start site were used in conjunction with the same downstream primer (FRDA-R1, 5V-TTTAAGCTTCTGCTCCGGGTCTGCCGCCC-3V). Each of the upstream primers contained an XhoI site at their 5V end. The 5V end of the frataxin sequence in the FRDA-R1 primer corresponds to the region 2 bp upstream of the initiator methionine. This primer contained a HindIII site at its 5V end. MasterAmp PCR Buffer E (Epicentre, Madison, WI, USA) was used with the F1255-FRDA-R1 primer pair and buffer F (Epicentre) was used for the remaining primer combinations. PCR conditions were 958C for 5 min, followed by 20 cycles of denaturation at 958C for 30 s, annealing at 658C for 1 min, and extension at 728C for 5 min. This was followed by incubation at 728C for 10 min. The amplified products were then digested with XhoI and HindIII and cloned into pGL3-Basic (Promega, Madison, WI, USA). The clones were designated FRDAX, where X is the number of frataxin-derived bases in the clone as well as the distance of the 5Vend of the frataxin sequence from the start of translation. The clones were verified by sequencing. Deletions were made in the FRDA1255 construct by subcloning individual restriction or PCR fragments into pGL3-Basic. The primers FRDA1255 and FRDA-R1 were also used to amplify the 5V UTR from three individuals of French Acadian descent with FRDA and three individuals with FRDA who were not of French Acadian ancestry. The resultant PCR fragments were digested with XhoI and
HindIII and cloned into pGL3-Basic as described above. These clones were sequenced and the sequences of both affected and unaffected individuals were deposited with GenBank (GenBank Accession Nos. AY599001, AY 599002, AY599003, AY 599004, AY599005, AY599006, AY599007). A similar approach was used to clone the 5Vend of the frataxin gene from A. geoffroyi (spider monkey)and P. troglodytes (chimpanzee). The GenBank accession numbers for these sequences are AY599013 and AY599014, respectively. The frataxin 3V UTR was amplified using MasterAmp PCR premix Buffer C (Epicentre) and the primers Frataxin UTR-F (5V-TTTTCTAGATGCCCAGCCCCGTTTT-3V) and Frataxin UTR-R (5V-TTTTCTAGAAGATAGAACAGTGAGC-3V). The PCR conditions were 20 cycles of 958C for 30 s, 558C for 1 min, and 728C for 1 min. The PCR products were purified, digested with XbaI, and cloned into pGL3-Control downstream of the luciferase gene. The clones were verified by sequencing (GenBank Accession Nos. AY599008, AY599009, AY599010, AY599011, AY599012). These clones were then transfected into C2C12 cells and the luciferase activity measured as described above. In silico promoter analysis The sequence of the rat frataxin 5Vend was identified in the working draft of the rat (Rattus norvegicus, June 2003 assembly) at the UCSC Genome Bioinformatics server (http://genome.ucsc.edu) using the Blat alignment tool [39] and exon 1 of the mouse frataxin gene as the query sequence. The region upstream of the start of translation of the rat frataxin gene was used to search NCBI build 30 (UCSC version mm3) and build 32 (UCSC version mm4) of the mouse sequence database. In build 32 homology was found only to a region in an unordered contig from chromosome 19. However, in build 30, in which a different algorithm was used to assemble the contigs, a region of homology was identified immediately upstream of the mouse frataxin open reading frame on chromosome 19. Delineation and identification of the repeated sequences were carried out using RepeatMasker on the UCSC Genome Bioinformatics server. Comparisons of the rodent and primate promoters were carried out using Blast 2 (http://www.ncbi.nlm.nih.gov/blast/ bl2seq/bl2.html) and PipMaker (http://bio.cse.psu.edu/cgibin/pipmaker?basic [40]). HUMOR, a special variant of the Multiz multiple sequence alignment program [41], was used to identify regions of weak homology between rodents and humans. The resultant sequences were then aligned with ClustalW and the alignments optimized manually to minimize arbitrary gaps. Cell culture, transient transfections, and promoter assays Mouse C2C12 myoblasts were grown in Dulbecco’s modified Eagle’s medium (Invitrogen Life Technologies,
E. Greene et al. / Genomics 85 (2005) 221–230
Inc., Carlsbad, CA, USA) supplemented with 10 or 2% fetal bovine serum. Cotransfections using pRL-null and various test constructs were carried out using Fugene 6 (Roche) according to the supplier’s instructions. Luciferase activity was measured 30 h posttransfection using the Dual Luciferase Reporter assay system (Promega) and a Microlumat LB 96 P luminometer (Berthold Systems, Inc., Aliquippa, PA, USA).
[10]
[11]
[12]
EMSAs [13]
The 221-bp PstI–HindIII fragment from FRDA1255 was gel purified and radiolabeled using [a-32P]dCTP and Klenow (-exo) DNA polymerase (Fermentas, Vilnius, Lithuania) according to standard procedures. Nuclear extracts were prepared from different cell lines, including C2C12 and NIH 3T6, using NE-PER nuclear extraction reagents (Pierce, Rockford, IL, USA) according to the supplier’s instructions. Binding was carried out at 48C in 30 Al of reaction buffer containing 25 mM Hepes (pH 7.5), 40 mM NaCl, 1 mM EDTA, and 4 mM DTT with 0.25 ng/Al labeled probe in the presence of 60- to 1000-fold molar excess of various unlabeled competitors.
[14]
[15]
[16] [17]
[18]
[19]
Acknowledgments Thanks to Donna Karolchik and Terry Furey (University of California, Santa Cruz, CA, USA) with their help in navigating the different builds of the UCSC mouse genome database. Thanks also to Anthony Furano (NIDDK, NIH) for his careful reading of the manuscript.
[20]
[21]
References [22] [1] P. Arnaud, C. Goubely, T. Pelissier, J.M. Deragon, SINE retroposons can be used in vivo as nucleation centers for de novo methylation, Mol. Cell. Biol. 20 (2000) 3434 – 3441. [2] A. Barbeau, M. Roy, M. Sadibelouiz, M.A. Wilensky, Recessive ataxia in Acadians and bCajunsQ, Can. J. Neurol. Sci. 11 (1984) 526 – 533. [3] A. Barbeau, M. Sadibelouiz, M. Roy, B. Lemieux, J.P. Bouchard, G. Geoffroy, Origin of Friedreich’s disease in Quebec, Can. J. Neurol. Sci. 11 (1984) 506 – 509. [4] S.I. Bidichandani, T. Ashizawa, P.I. Patel, The GAA triplet-repeat expansion in Friedreich ataxia interferes with transcription and may be associated with an unusual DNA structure, Am. J. Hum. Genet. 62 (1998) 111 – 121. [5] V. Campuzano, L. Montermini, M.D. Molto, L. Pianese, M. Cossee, F. Cavalcanti, E. Monros, F. Rodius, F. Duclos, A. Monticelli, et al., Friedreich’s ataxia: autosomal recessive disease caused by an intronic GAA triplet repeat expansion, Science 271 (1996) 1423 – 1427. [6] J. Chaco, Friedreich’s ataxia with myocardial involvement and myopathy, Isr. J. Med. Sci. 2 (1966) 52 – 54. [7] E.W. Englander, B.H. Howard, Nucleosome positioning by human Alu elements in chromatin, J. Biol. Chem. 270 (1995) 10091 – 100916. [8] M. Ghazizadeh, Cisplatin may induce frataxin expression, J. Nippon Med. Sch. 70 (2003) 367 – 371. [9] E. Grabczyk, K. Usdin, Alleviating transcript insufficiency caused by
[23]
[24]
[25]
[26]
[27]
[28]
229
Friedreich’s ataxia triplet repeats, Nucleic Acids Res. 28 (2000) 4930 – 4937. E. Grabczyk, K. Usdin, The GAA*TTC triplet repeat expanded in Friedreich’s ataxia impedes transcription elongation by T7 RNA polymerase in a length and supercoil dependent manner, Nucleic Acids Res. 28 (2000) 2815 – 2822. U. Hellmann-Blumberg, M.F. Hintz, J.M. Gatewood, C.W. Schmid, Developmental differences in methylation of human Alu repeats, Mol. Cell. Biol. 13 (1993) 4523 – 4530. S. Jiralerspong, Y. Liu, L. Montermini, S. Stifani, M. Pandolfo, Frataxin shows developmentally regulated tissue-specific expression in the mouse embryo, Neurobiol. Dis. 4 (1997) 103 – 113. V. Kapitonov, J. Jurka, The age of Alu subfamilies, J. Mol. Evol. 42 (1996) 59 – 65. B.J. Keats, L.J. Ward, J. Shaw, A. Wickremasinghe, S. Chamberlain, bAcadianQ and bclassicalQ forms of Friedreich ataxia are most probably caused by mutations at the same locus, Am. J. Med. Genet. 33 (1989) 266 – 268. C. Kim, C.M. Rubin, C.W. Schmid, Genome-wide chromatin remodeling modulates the Alu heat shock response, Gene 276 (2001) 127 – 133. Y. Kondo, J.P. Issa, Enrichment for histone H3 lysine 9 methylation at Alu repeats in human cells, J. Biol. Chem. 278 (2003) 27658 – 27662. T.H. Li, C.W. Schmid, Differential stress induction of individual Alu loci: implications for transcription and retrotransposition, Gene 276 (2001) 135 – 141. W.M. Liu, W.M. Chu, P.V. Choudary, C.W. Schmid, Cell stress and translational inhibitors transiently increase the abundance of mammalian SINE transcripts, Nucleic Acids Res. 23 (1995) 1758 – 1765. W.M. Liu, C.W. Schmid, Proposed roles for DNA methylation in Alu transcriptional repression and mutational inactivation, Nucleic Acids Res. 21 (1993) 1351 – 1359. L. Montermini, A. Richter, K. Morgan, C.M. Justice, D. Julien, B. Castellotti, J. Mercier, J. Poirier, F. Capozzoli, J.P. Bouchard, B. Lemieux, J. Mathieu, M. Vanasse, M.H. Seni, G. Graham, F. Andermann, E. Andermann, S.B. Melancon, B.J. Keats, S. Di Donato, M. Pandolfo, Phenotypic variability in Friedreich ataxia: role of the associated GAA triplet repeat expansion, Ann. Neurol. 41 (1997) 675 – 682. S. Oliviero, P. Monaci, RNA polymerase III promoter elements enhance transcription of RNA polymerase II genes, Nucleic Acids Res. 16 (1988) 1285 – 1293. M. Pandolfo, The molecular basis of Friedreich ataxia, Adv. Exp. Med. Biol. 516 (2002) 99 – 118. B. Panning, J.R. Smiley, Activation of RNA polymerase III transcription of human Alu elements by herpes simplex virus, Virology 202 (1994) 408 – 417. B. Panning, J.R. Smiley, Activation of expression of multiple subfamilies of human Alu elements by adenovirus type 5 and herpes simplex virus type 1, J. Mol. Biol. 248 (1995) 513 – 524. M.A. Pook, S. Al-Mahdawi, C.J. Carroll, M. Cossee, H. Puccio, L. Lawrence, P. Clark, M.B. Lowrie, J.L. Bradley, J.M. Cooper, M. Koenig, S. Chamberlain, Rescue of the Friedreich’s ataxia knockout mouse by human YAC transgenesis, Neurogenetics 3 (2001) 185 – 193. A. Richter, J. Poirier, J. Mercier, D. Julien, K. Morgan, M. Roy, F. Gosselin, J.P. Bouchard, S.B. Melancon, Friedreich ataxia in Acadian families from eastern Canada: clinical diversity with conserved haplotypes, Am. J. Med. Genet. 64 (1996) 594 – 601. M. Ristow, H. Mulder, D. Pomplun, T.J. Schulz, K. MullerSchmehl, A. Krause, M. Fex, H. Puccio, J. Muller, F. Isken, J. Spranger, D. Muller-Wieland, M.A. Magnuson, M. Mohlig, M. Koenig, A.F. Pfeiffer, Frataxin deficiency in pancreatic islets causes diabetes due to loss of beta cell mass, J. Clin. Invest. 112 (2003) 527 – 534. N. Sakamoto, P.D. Chastain, P. Parniewski, K. Ohshima, M. Pandolfo, J.D. Griffith, R.D. Wells, Sticky DNA: self-association properties of
230
[29]
[30]
[31]
[32]
[33] [34]
E. Greene et al. / Genomics 85 (2005) 221–230 long GAAd TTC repeats in Rd Rd Y triplex structures from Friedreich’s ataxia, Mol. Cell 3 (1999) 465 – 475. M.M. Santos, K. Ohshima, M. Pandolfo, Frataxin deficiency enhances apoptosis in cells differentiating into neuroectoderm, Hum. Mol. Genet. 10 (2001) 1935 – 1944. J.P. Sarsero, L. Li, T.P. Holloway, L. Voullaire, S. Gazeas, K.J. Fowler, D.M. Kirby, D.R. Thorburn, A. Galle, S. Cheema, M. Koenig, R. Williamson, P.A. Ioannou, Human BAC-mediated rescue of the Friedreich ataxia knockout mutation in transgenic mice, Mamm. Genome 15 (2004) 370 – 382. J.P. Sarsero, L. Li, H. Wardan, K. Sitte, R. Williamson, P.A. Ioannou, Upregulation of expression from the FRDA genomic locus for the therapy of Friedreich ataxia, J. Gene Med. 5 (2003) 72 – 81. A. Saveliev, C. Everett, T. Sharpe, Z. Webster, R. Festenstein, DNA triplet repeats mediate heterochromatin-protein-1-sensitive variegated gene silencing, Nature 422 (2003) 909 – 913. C.W. Schmid, Human Alu subfamilies and their methylation revealed by blot hybridization, Nucleic Acids Res. 19 (1991) 5613 – 5617. C. Thor’En, Cardiomyopathy in Friedreich’s ataxia: with studies of cardiovascular and respiratory function, Acta Paediatr. 53 (Suppl. 153) (1964) 1.
[35] N.V. Tomilin, Control of genes by mammalian retroposons, Int. Rev. Cytol. 186 (1999) 1 – 48. [36] M. Turano, A. Tammaro, I. De Biase, M.S. Lo Casale, G. Ruggiero, A. Monticelli, S. Cocozza, L. Pianese, 3-Nitropropionic acid increases frataxin expression in human lymphoblasts and in transgenic rat PC12 cells, Neurosci. Lett. 350 (2003) 184 – 186. [37] E. Ullu, C. Tschudi, Alu sequences are processed 7SL RNA genes, Nature 312 (1984) 171 – 172. [38] E.T. Dermitzakis, A.G. Clark, Evolution of transcription factor binding sites in Mammalian gene regulatory regions: conservation and turnover, Mol. Biol. Evol. 19 (2002) 1114 – 1121. [39] W.J. Kent, C.W. Sugnet, T.S. Furey, K.M. Roskin, T.H. Pringle, A.M. Zahler, D. Haussler, The human genome browser at UCSC, Genome Res. 12 (2002) 996 – 1006. [40] S. Schwartz, Z. Zhang, K.A. Frazer, A. Smit, C. Riemer, J. Bouck, R. Gibbs, R. Hardison, W. Miller, PipMaker–a web server for aligning two genomic DNA sequences, Genome Res. 10 (2000) 577 – 586. [41] S. Schwartz, W.J. Kent, A. Smit, Z. Zhang, R. Baertsch, R.C. Hardison, D. Haussler, W. Miller, Human-mouse alignments with BLASTZ, Genome Res. 13 (2003) 103 – 107.