|Home | About | Journals | Submit | Contact Us | Français|
The nonpathogenic human adeno-associated virus type 2 (AAV-2) has adopted a unique mechanism to site-specifically integrate its genome into the human MBS85 gene, which is embedded in AAVS1 on chromosome 19. The fact that AAV has evolved to integrate into this ubiquitously transcribed region and that the chromosomal motifs required for integration are located a few nucleotides upstream of the translation initiation start codon of MBS85 suggests that the transcriptional activity of MBS85 might influence site-specific integration and thus might be involved in the evolution of this mechanism. In order to begin addressing this question, we initiated the characterization of the human MBS85 promoter region and compared its transcriptional activity to that of the AAV-2 p5 promoter. Our results clearly indicate that AAVS1 is defined by a complex transcriptional environment and that the MBS85 promoter shares key regulatory elements with the viral p5 promoter. Furthermore, we provide evidence for bidirectional MBS85 promoter activity and demonstrate that the minimal motifs required for AAV site-specific integration are present in the 5′ untranslated region of the gene and play a posttranscriptional role in the regulation of MBS85 expression. These findings should provide a framework to further elucidate the complex interactions between the virus and its cellular host in this unique pathway to latency.
Human adeno-associated virus type 2 (AAV-2) is a nonpathogenic parvovirus that requires helper functions provided by viruses such as adenovirus or herpes simplex virus for efficient replication of its genome (2, 60). In the absence of such helper functions, AAV-2 establishes latency by site-specifically integrating its genome into a region of human chromosome 19 (19q13.42) termed AAVS1 (27-30, 51).
The AAV-2 genome is a single-stranded DNA molecule of 4.7 kb and contains two open reading frames (ORFs), rep and cap, flanked by two inverted terminal repeats (ITRs) (56). The rep and cap ORFs encode four overlapping regulatory proteins (Rep78, Rep68, Rep52, and Rep40) and three structural capsid proteins (VP1, VP2, and VP3), respectively (6). The two large Rep proteins, Rep78 and Rep68, which are expressed from the p5 promoter, are involved in every step of the viral life cycle, i.e., replication, site-specific integration, rescue, splicing, and regulation of viral-gene expression (6, 47, 48).
The biochemical properties of Rep78 and Rep68, such as DNA binding (11, 12, 22, 40) and ATPase (63), helicase (20, 23-25, 63), and endonuclease activities (23), are essential for AAV DNA replication. It was further demonstrated that Rep78 and Rep68 can specifically bind to the Rep binding site (RBS) (11, 12, 41, 50) and introduce a nick in a site- and strand-specific manner at the terminal resolution site (TRS) (7, 23, 55, 67). The TRS-RBS motifs, present in the ITR, serve as the minimal origin for Rep-mediated AAV DNA replication (54, 61). The function, of the TRS-RBS, however, is not limited to AAV DNA replication; it is also one of the major components in Rep-mediated site-specific integration. Indeed, similar TRS-RBS motifs are also present within AAVS1 (62), where they represent the minimal requirements for targeted integration (39). Biochemical studies have demonstrated that Rep68 is able to simultaneously bind to both viral and cellular RBS motifs (62) and introduces a nick at the AAVS1 TRS site (34, 59). Subsequently, AAV DNA integration is speculated to occur through limited viral and cellular DNA synthesis following template strand switches, resulting in partial duplication of MBS85 sequences (19, 38, 39).
Several lines of evidence suggest that AAVS1 contains a transcriptionally active region. Kotin et al. originally reported the presence of several putative transcription factor binding sites upstream of the RBS and a CpG island, which is often the hallmark of a TATA-less promoter (28). Further studies by Lamartina et al. identified a DNase I-hypersensitive site within the same AAVS1 region, which displays transcriptional activities in an orientation-independent manner (35). It has therefore been suggested that an enhancer is present upstream of the RBS (35, 36). Studies by Tan et al. have shown that the minimal motifs necessary for AAV site-specific integration, the TRS and RBS, are located only a few nucleotides upstream of the translation initiation site of the myosin binding subunit 85 gene (MBS85), also termed PPP1R12C (for protein phosphatase 1 regulatory protein) (57). To date, there are only a few reports describing either MBS85 regulation or the function of the resulting protein. MBS85 is ubiquitously expressed in human and mouse tissues and appears to be highly expressed in the heart (16, 57). The protein is thought to be a component of the regulatory subunit of the myosin light chain phosphatase, which is involved in myosin phosphorylation, indicating that MBS85 might play a role in the regulation of assembly and disassembly of the actin cytoskeleton (57).
The fact that AAV has evolved to integrate site specifically into a ubiquitously transcribed region raises several questions with regard to the nonpathogenic character of AAV. How can the virus integrate, and thus disrupt a transcriptional unit, in the absence of any apparent deleterious effects on the cell? Using a mouse model for site-specific integration, Henckaerts et al. have recently demonstrated that integration into one allele of diploid embryonic stem (ES) cells does not interfere with either in vitro or in vivo differentiation of these cells, indicating a complex mechanism by which a functional copy of the MBS85 transcription unit is maintained (19). Through the analyses of multiple integrants, these studies also provided indirect evidence that the p5 and MBS85 promoters might interact to form an initial integration complex, introducing the possibility that shared promoter elements might be directly involved in the integration mechanism (19). A further question that arises is based on the challenge that in order to maintain latency AAV needs to put in place mechanisms that secure a level of regulation of the transcriptional activity surrounding the integrated viral genome.
In order to provide a framework for addressing these questions and to gain a better understanding of the mechanisms underlying AAV site-specific integration, we initiated the characterization of the MBS85 promoter and compared its transcriptional activities to those of the AAV p5 promoter. Our results clearly indicate that AAVS1 is defined by a complex transcriptional environment and that the MBS85 promoter shares key regulatory elements with the viral p5 promoter. Furthermore, we provide evidence for bidirectional MBS85 promoter activity and demonstrate that the minimal motifs required for AAV site-specific integration (TRS-RBS) are present in the 5′ untranslated region (UTR) of the gene and play a posttranscriptional role in the regulation of MBS85 expression.
293T (HEK_293T), RD, A673, and MCF7 cells were grown in Dulbecco's modified Eagle's medium (Mediatech Inc., Manassas, VA), 10% fetal bovine serum (Gemini Bio-Products, West Sacramento, CA). HeLa cells were grown in minimal essential medium (Mediatech Inc.), 10% fetal bovine serum, 0.1 mM nonessential amino acids, and 1 mM sodium pyruvate (Invitrogen, Carlsbad, CA). The human ES cell line H1 was grown on DR4 mouse embryonic feeder cells in Dulbecco's modified Eagle's medium-F12 supplemented with 20% (vol/vol) Knockout Serum Replacement (both from Invitrogen), basic fibroblast growth factor (20 ng/ml; R&D Systems), 50 U/ml penicillin, 50 μg/ml streptomycin, 2 mM l-glutamine, 0.1 mM nonessential amino acids (all from Invitrogen), and 0.1 mM β-mercaptoethanol (Sigma-Aldrich, St. Louis, MO). Prior to RNA isolation, H1 cells were grown under feeder-free conditions in matrigel-coated plates (Biocoat; Becton Dickinson, San Jose, CA).
Total RNA was extracted using the RNeasy kit (Qiagen, Valencia, CA). Ten micrograms of RNA was separated on a 1% formaldehyde-agarose gel and transferred onto a Magna nylon membrane (Osmonics, Minnetonka, MN). All membranes were hybridized with [α-32P]dCTP-labeled probes (Prime-It RmT Random Primer Labeling Kit; Stratagene, Cedar Creek, TX). Northern blots were first hybridized to red fluorescent protein (RFP) or MBS85 (exons 18 to 22) probes; stripped by being boiled for 15 min in 0.05× SSC (1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate), 10 mM EDTA (pH 8), 0.1% sodium dodecyl sulfate (SDS); and rehybridized to a β-actin cDNA probe. The RFP, MBS85, and β-actin probes were generated by PCR performed on plasmids pRFP, pND83, and pβ-actin, respectively (see Tables S1 and S2 at http://www.kcl.ac.uk/linden).
Transcription start sites (TSS) were identified using the 5′ rapid amplification of cDNA ends (5′ RACE) GeneRacer Kit (Invitrogen) according to the manufacturer's instructions. Total RNA was extracted from 293T cells (MBS85 TSS) and from 293T cells transfected with construct pND29 (antisense TSS) (see Table S1 at http://www.kcl.ac.uk/linden). Briefly, DNase I-treated total RNA (5 μg) was dephosphorylated with calf intestinal phosphatase, and the cap structure was removed with tobacco acid pyrophosphatase. The decapped RNA was then ligated to the GeneRacer RNA primer. RNA was reverse transcribed for 1 h at 50°C using an oligo(dT) primer and SuperScriptIII reverse transcriptase (Invitrogen). The resulting cDNAs were amplified by PCR with the GC-rich PCR kit (Roche Applied Science, Indianapolis, IN) and GeneRacer 5′ and ND189 primers (MBS85 TSS) or GeneRacer 5′ and ND192 primers (antisense TSS) (see Table S2 at http://www.kcl.ac.uk/linden). The PCR conditions were as follows: 95°C for 3 min; 10 cycles at 95°C for 30 s, 58°C for 30 s, and 72°C for 1 min 20 s (or 1 min for the antisense TSS); 25 cycles at 95°C for 30 s, 58°C for 30 s, and 72°C for 1 min 20 s with an elongation time of 5 s per cycle; and a final extension at 72°C for 7 min. The PCR products were cloned into the PCR2.1 Topo vector (Invitrogen), and the resulting clones were sequenced using a 3730xl DNA Analyzer (Applied Biosystems, Foster City, CA).
Plasmids pDsRed2.1, pDsRed2-N1, and pIRES2-EGFP were purchased from Clontech (Mountain View, CA). EST R35625 (catalog number 363135) was purchased from the ATCC (Manassas, VA). Plasmid pDsRed2.1 was used to clone the MBS85 and p5 promoter regions. The p5 promoter used in this study is the region described by Chang et al. (reference 10 and Table S2 at http://www.kcl.ac.uk/linden). All constructs have been sequenced.
Transfection experiments were performed in 60-mm plates. At 50% confluence, 293T cells were transfected with 6 μg of reporter construct using Lipofectamine Plus reagent (Invitrogen). HeLa cells were transfected at 80% confluence with 6 μg of reporter construct using Fugene 6 reagent (Roche Applied Science). 293T and HeLa cells were harvested 48 h and 72 h posttransfection, respectively, and assayed for plasmid DNA uptake, along with Northern blot, Western blot, and fluorescence-activated cell sorter (FACS) analyses. Transfection efficiencies were normalized by plasmid DNA uptake, as previously described (31). All transfection experiments were repeated four times using plasmids that were independently prepared at least twice.
Transfected cells were lysed in 0.2 M NaOH, 10 mM EDTA. Samples were boiled for 15 min at 90°C and loaded onto a Hybond XL nylon membrane (Amersham Biosciences, Piscataway, NJ) using a slot blot manifold (Bio-Rad). The membranes were hybridized to an RFP or green fluorescent protein (GFP) probe, generated by PCR, to determine the amount of reporter plasmid taken up by the cells.
Cells were solubilized in RIPA buffer (50 mM Tris-HCl [pH 8], 150 mM NaCl, 0.1% SDS, 1% Nonidet P-40, 0.5% sodium deoxycholate, 1× Complete protease inhibitor cocktail) (Roche Applied Science). Samples (10 μg) were loaded on a 15% SDS-polyacrylamide gel and transferred onto a Hybond C extra nitrocellulose membrane (Amersham Biosciences, Piscataway, NJ). The membranes were blocked in 5% fat-free milk and incubated with anti-DsRed polyclonal antibody at a dilution of 1:16,000 (catalog number 632397; Clontech) or anti-actin monoclonal antibody at a dilution of 1:10,000 (catalog number 612656; Becton Dickinson Biosciences). After being washed in Tris-buffered saline-Tween 20 buffer, the blots were incubated with horseradish peroxidase-conjugated secondary antibody at a dilution of 1:10,000 (Jackson ImmunoResearch Laboratories, West Grove, PA). RFPs were visualized by the enhanced-chemiluminescence method using Pico and Femto detection kits (Pierce, Rockford, IL) for 293T and HeLa cell extracts, respectively. Actin proteins were visualized with the Pico detection kit. Membranes were first incubated with anti-DsRed antibody, stripped with the Restore Western Buffer Stripping Buffer (Pierce), and blotted with anti-actin antibody for normalization.
Protein quantification was performed using the Li-Cor Odyssey infrared imaging system (Li-Cor Biosciences UK Ltd., Cambridge, United Kingdom). Anti-mouse IRDye 680 (catalog number 926-32220; Li-Cor) or anti rabbit IRDye 800 (catalog number 926-32211; Li-Cor) fluorescent secondary antibody was used at a dilution of 1:5,000 in 1% milk, 0.5% Tween 20.
Cell suspensions from transfected 293T and HeLa cells were prepared and analyzed for RFP expression. The data were acquired using a FACScalibur (Becton Dickinson) and analyzed by Flowjo software (Tree Star, Inc., Ashland, OR). FACS data are presented as dot plots with linear axes for forward/side scatter and logarithmic axes for FL1 (empty; x axis) and FL2 (RFP; y axis). Gates were set to exclude dead cells based on forward/side scatter. This gated population was analyzed for RFP expression. The gate set to determine the percentage of RFP-expressing cells for each sample was obtained by the analysis of a negative population (cells transfected with a promoterless vector, pDsRed2.1) and was verified by a positive population (cells transfected with a cytomegalovirus [CMV]-controlled RFP vector, pDsRed2-N1). The geometric mean fluorescence intensity (Geo MFI) was calculated for this gated population of RFP-positive cells. FACS data collected from four experiments were used to generate graphs (see Fig. Fig.3,3, ,4,4, and and66).
Sorting experiments were performed on a Moflo cell sorter (Cytomation, Ft. Collins, CO).
DNase I-treated RNA was subjected to reverse transcription-PCR (RT-PCR) using SuperScriptIII reverse transcriptase (Invitrogen) and the KCL1 primer. The resulting cDNAs were amplified by PCR with GoTaq polymerase (Promega, Madison, WI) and the KCL2 and KCL4 primers. The PCR conditions were 94°C for 2 min; 35 cycles at 94°C for 30 s, 62°C for 30 s, and 72°C for 40 s; and a final extension at 72°C for 10 min. The PCR products were cloned into the PCR2.1 Topo vector (Invitrogen), and the resulting clones were sequenced.
Transfected 293T cells were visualized for RFP and GFP expression 48 h posttransfection. Epifluoresecnce microscopy was carried out by using an inverted epifluorescence microscope (Leica; DM IRB) and an ×20 magnification lens. Images were acquired with a Mintron digital camera. Exposure times and camera gain values were kept strictly identical in all pictures. All images were assembled and processed identically (adjustment level correction and contrast enhancement were identical for each image file) with Adobe Photoshop 7.0 (Adobe Systems Inc., Mountain View, CA).
Nucleotide sequence accession numbers were as follows: Homo sapiens MBS85 expressed sequence tag (EST), R35625; IMAGE clone identifier, 38310; H. sapiens MBS85 mRNA, AF312028; H. sapiens β-actin, NM_001101; AAV2 complete genome, AF043303; and upstream sequences from the MBS85 ATG start codon for H. sapiens (human) chromosome 19, NT_011109.15 from nucleotide (nt) 27897099 to nt 27898224, for Pan troglodytes (chimpanzee) chromosome 19, NW_001228247.1 from nt 1307249 to nt 1307515 and from nt 1305832 to nt 1306173 (there is a gap between the two contigs), for Bos taurus (cattle) chromosome 18, NW_001493632.2 from nt 1526744 to nt 1527802, for Equus caballus (horse) chromosome 10, NW_001867363.1 from nt 24348902 to nt 24349949, for Canis lupus familiaris (dog) chromosome 1, NW_876270.1 from nt 32959175 to nt 32960143, for Mus musculus (mouse) chromosome 7, NW_001030825.1 from nt 1246105 to nt 1247455, and for Rattus norvegicus (rat) chromosome 1, NW_047555.2 from nt 13365737 to nt 13367100.
In order to map the promoter region of the human MBS85 gene and identify potential regulatory elements, the monkey and mouse MBS85 genes (3, 16) were used to retrieve additional MBS85 orthologous genes from the NCBI database. The availability of genomic sequences from an increasing number of species indicated that MBS85 is evolutionarily conserved among higher eukaryotes. Analysis of the highly conserved regions from the human, monkey, cattle, horse, dog, mouse, and rat sequences upstream of the translation initiation site demonstrated that the MBS85 gene lacks the typical TATA and CAAT boxes, while several putative DNA motifs for transcription factors are conserved, i.e., Sp1, Staf, Brn, Egr-1, ATF-CREB, E4F, YY1, and zinc finger 5 (ZF5) (Fig. (Fig.1).1). Importantly, not only the sequence of the RBS is conserved among all these species, but also its position relative to the ATG codon (located 17 to 37 nt upstream from the ATG site), suggesting a possible role in MBS85 gene regulation.
In order to characterize the MBS85 TSS, we first investigated the expression of MBS85 by Northern blotting in six human cell lines. Total RNA was extracted from H1 (ES), 293T (embryonic kidney), HeLa (cervix epitheloid carcinoma), MCF7 (breast adenocarcinoma), and RD and A673 (rhabdomyosarcoma) cell lines. Northern blots hybridized to a probe containing MBS85 exons 18 to 22 revealed a unique transcript of 3 kb, indicating that MBS85 is expressed in all six cell lines (Fig. (Fig.2A).2A). This result is consistent with the previously described findings of Tan et al., in which a unique transcript of 3 kb was detected in various human tissues. However, the possibility that additional transcripts (i.e., alternative splicing) expressed at low levels might be generated from the MBS85 promoter cannot be excluded. The relative expression level of MBS85 transcripts compared to that of the β-actin gene suggests that MBS85 is expressed at higher levels in 293T than in HeLa cells. These two cell lines were chosen for further experiments based on their endogenous MBS85 expression (Fig. (Fig.2A)2A) and their relatively high transfection efficiencies.
We subsequently determined the MBS85 mRNA TSS by 5′ RACE on total RNA extracted from 293T cells. A single RT-PCR product was detected (Fig. (Fig.2B).2B). Sequence analysis of several clones assigned the MBS85 TSS to an adenine located 94 bp upstream from the translation initiation site (Fig. (Fig.2B2B and Fig. S1 and S2 at http://www.kcl.ac.uk/linden). This result also confirmed that the AAVS1 TRS-RBS motifs are located in the 5′ UTR of the MBS85 gene (Fig. (Fig.11 and Fig. S2 at the URL given above).
In order to identify regulatory regions involved in MBS85 promoter activity, a series of progressive 5′-truncated sequences were cloned in the sense orientation into the pDsRed2.1 promoterless RFP vector. These constructs contained MBS85 sequences all the way down to the ATG translation initiation site, including the TSS and 5′ UTR (Fig. (Fig.3A).3A). To compare MBS85 and p5 promoter activities, the p5 promoter was also cloned into the pDsRed2.1 vector (Fig. (Fig.3A3A).
All sense constructs were transiently transfected into 293T and HeLa cells. The expression of the RFP reporter gene was monitored by measuring the amounts of RFP mRNA and protein by Northern blot and Western blot analyses, respectively. Since a direct correlation between the relative mRNA and protein levels could be observed, the reporter protein activity directly reflects activity at the transcriptional level. As shown in Fig. Fig.3B,3B, a reporter construct containing sequences from nt −3302 to +94 exhibited moderate RFP expression in 293T and HeLa cells. Deletions of sequences spanning from −3.3 kb to −137 bp resulted in only minor changes in MBS85 promoter activities in the two cell lines. However, a further deletion from nt −137 to −99 almost completely abrogated MBS85 promoter activity (Fig. (Fig.3B).3B). Together, these data indicate that the proximal region spanning from −137 bp to +94 bp is sufficient for constitutive MBS85 transcription in both 293T and HeLa cells. Sequence analysis revealed that several putative response elements for transcription factors are located within this region, e.g., Sp1, ATF-CREB, E4F, Staf, YY1, and ZF5 (see Fig. S2 at http://www.kcl.ac.uk/linden). Of these potential cis-acting elements, the YY1 site at position −110 is of particular interest, since deletion of nt −137 to −99 almost completely abolished MBS85 promoter activity.
Since flow cytometry provides direct readout at the single-cell level, it was used to determine the percentage of RFP-expressing cells and to quantify the promoter activity in the RFP-expressing cell population. These data indicated that 20% to 30% of 293T cells and 10% of HeLa cells expressed the reporter gene (Fig. (Fig.3B).3B). Comparison of the Geo MFI of RFP indicated that the MBS85 promoter activity was significantly stronger in 293T than in HeLa cells (Geo MFI, 50 in 293T cells compared to 20 in HeLa cells) (Fig. (Fig.3B).3B). These data were correlated with the higher endogenous MBS85 mRNA levels detected in 293T cells in comparison to HeLa cells, as observed by Northern blotting (Fig. (Fig.2A2A).
We next compared the p5 and MBS85 promoter activities. As shown in Fig. Fig.3B,3B, the p5 promoter led to a significant increase in RFP expression compared to the MBS85 promoter in both 293T and HeLa cells. This increase in RFP expression is due not only to a higher number of expressing cells, but also to stronger promoter activity, as determined by FACS analysis (p5 Geo MFI, 150 and 40 in 293T and HeLa cells, respectively).
It has previously been reported that a reporter construct containing MBS85 sequences from nt −332 to +20 exhibited transcriptional activities in an orientation-independent manner, suggesting the presence of an enhancer element within this region (35, 36). However, our data clearly indicate that this putative enhancer-like sequence lies within the MBS85 minimal promoter region. It has recently become evident that bidirectional promoters might not be as uncommon as previously thought (1, 43), and a large number of transcripts are indeed arising from the opposite direction of protein-coding genes (13, 18, 26, 46, 52, 66). Therefore, we hypothesized that the MBS85 promoter might display bidirectional promoter activities rather than enhancer activities. In order to address this hypothesis, 5′-truncated sequences were cloned in the antisense orientation into the pDsRed2.1 promoterless vector, and the resulting antisense constructs were transiently transfected into 293T and HeLa cells (Fig. (Fig.4A).4A). As shown in Fig. Fig.4B,4B, reporter constructs containing sequences from nt +94 to −332 exhibited transcriptional activities in both 293T and HeLa cells, whereas no promoter activity could be detected by additional sequences from nt −332 to −3302. Flow cytometry analyses further indicated weaker promoter activity for constructs +94/−137 and +94/−332 compared to constructs +94/−99 and +94/−236, suggesting possible negative regulatory elements located within the regions −332 to −236 and −137 to −99. Sequence analysis of these two regions revealed a putative YY1 binding site within each region (see Fig. S2 at http://www.kcl.ac.uk/linden). Fluctuations of the percentage of cells expressing the RFP reporter construct could be observed between constructs, as was previously reported with different cellular promoters (17). Importantly, constructs exhibiting a lower number of RFP-positive cells also concomitantly resulted in a lower MFI. Even though the significance of these differences is not well understood, it may be that cells expressing the lowest RFP level cannot be detected under the conditions used in the assay.
As observed for the MBS85 sense promoter activity, the antisense promoter activity was also stronger in 293T cells than in HeLa cells (Geo MFI, 40 and 20 in 293T and HeLa cells, respectively). Interestingly, weak p5 antisense promoter activity could be detected in both 293T cells and HeLa cells.
These results demonstrate that the MBS85 region spanning from nt +94 to −332 is able to drive the expression of a reporter gene from the antisense orientation. Our data further indicate that the MBS85 promoter might have bidirectional transcriptional activities, as detected from reporter constructs in which the same fragments were individually cloned in opposite directions.
We next investigated whether the MBS85 antisense promoter would function in a truly bidirectional manner. Fragment −236/+94, which exhibited sense promoter activity together with the strongest antisense promoter activity, was cloned in both orientations into a dual reporter construct carrying the RFP and enhanced GFP (EGFP) genes in opposite directions (see Table S1 at http://www.kcl.ac.uk/linden). To exclude any promiscuous transcription occurring from the bidirectional vector, a 0.3-kb fragment originating from the first intron of the human MBS85 gene was cloned in both directions into the bidirectional vector. All bidirectional constructs were transiently transfected into 293T cells, and reporter gene expression was detected by Western blot analysis. As shown in Fig. Fig.4C,4C, expression of both RFP and EGFP genes was detected from each MBS85 promoter construct, whereas no EGFP or RFP expression could be detected with all three negative controls. Two constructs driving the expression of only one reporter gene (RFP or EGFP) were also transfected to assess the specificity of each antibody. These results show that the MBS85 promoter can drive the expression of two inversely oriented reporter genes. Taken together, our data provide strong evidence that the MBS85 promoter displays bidirectional promoter activities.
We next mapped the antisense TSS by 5′ RACE in 293T cells transfected with the +20/−332 antisense construct. The lengths of the RT-PCR products were in the range from 200 bp to 400 bp, suggesting the existence of multiple antisense start sites (Fig. (Fig.4D).4D). Sequence analysis of several RT-PCR products revealed four antisense start sites spread over 70 bp (see Fig. S2 at http://www.kcl.ac.uk/linden). However, since antisense promoter activity could be detected with shorter promoter fragments (+94/−99, +94/−137, and +94/−236), it is more likely that additional start sites will be identified and it is therefore difficult to ascertain which site is the predominant antisense TSS. Furthermore, it remains unclear whether the MBS85 and antisense transcripts overlap.
In order to identify potential genes that are transcribed in the opposite direction from MBS85, we searched the NCBI human EST database for the genomic sequences that are located between the MBS85 and TNNT1 genes (15). No full-length EST was found to perfectly match the intergenic region between these two genes. However, three short sequences upstream from the MBS85 TSS are highly homologous to the 3′ end of the nuclear transport factor 2 (NUTF2) mRNA, suggesting the presence of a NUTF2 pseudogene (ψNUTF2) (Fig. (Fig.5A).5A). To first assess whether NUTF2 pseudogene-derived antisense transcripts could be generated by the MBS85 antisense promoter, we performed an RT reaction on total RNA extracted from 293T cells using a primer that could anneal only to the NUTF2 pseudogene antisense transcripts. As shown in Fig. Fig.5B,5B, a product of 650 bp was detected by RT-PCR. Nucleotide sequence analysis of this 650-bp RT-PCR product showed 100% homology with the NUTF2 pseudogene present upstream of the MBS85 gene. This result further suggests that ψNUTF2 is transcribed in the antisense orientation with respect to the NUTF2 gene. We next determined whether the ψNUTF2 antisense transcripts could hybridize to NUTF2 sense transcripts and lead to suppression of NUTF2 protein synthesis through an antisense repression mechanism. Constructs containing different lengths of the intergenic region between MBS85 and TNNT1 were transfected in 293T cells (Fig. (Fig.5C).5C). All constructs contained the antisense promoter driving ψNUTF2 antisense transcription and the RFP gene under the control of the MBS85 promoter to selectively enrich for ψNUTF2 antisense transcripts. Total protein was extracted from the RFP-positive cell population, and NUTF2 expression was assessed by Western blot analysis. As shown in Fig. Fig.5D,5D, overexpression of ψNUTF2 antisense transcripts did not significantly decrease NUTF2 protein expression compared to endogenous NUTF2 expression. These results indicate that under the conditions used in this assay, detectable ψNUTF2 antisense transcripts do not affect NUTF2 expression.
Since the RBS is located within the 5′ UTR of the MBS85 gene and is conserved among all the species analyzed, we investigated whether the TRS-RBS region is involved in the regulation of MBS85 expression. Constructs with the TRS-RBS sequence deleted were transiently transfected into 293T and HeLa cells (Fig. (Fig.6A).6A). As shown in Fig. Fig.6B,6B, promoter constructs in which the TRS-RBS sequence was deleted exhibited higher RFP expression than control promoters harboring the TRS-RBS sequence in both 293T and HeLa cells. In order to further confirm the effect of the TRS-RBS region on gene expression, we performed quantitative Western blots using a Li-Cor Odyssey system. Deletion of the TRS-RBS sequence resulted in a 4- to 5.8-fold increase in promoter activity in 293T cells (Fig. S3A and S3B at http://www.kcl.ac.uk/linden). This higher RFP expression is directly correlated with a higher number of RFP-expressing cells and stronger promoter activity, as shown by FACS analysis (Fig. (Fig.6B6B and Fig. S3C and S3D at http://www.kcl.ac.uk/linden). Despite an increase in the amount of reporter protein, Northern blot analysis showed constant mRNA levels (Fig. (Fig.6B).6B). Together, these results indicate that the TRS-RBS region present in the 5′ UTR of MBS85 contains an inhibitory element and that this inhibition likely occurs at the posttranscriptional level.
We next tested whether the TRS-RBS region could also regulate the expression of a reporter gene construct under the control of a heterologous promoter. A construct containing the TRS-RBS sequence cloned between the CMV promoter and the RFP reporter gene was transiently transfected into 293T and HeLa cells (Fig. (Fig.6A).6A). Insertion of the TRS-RBS sequence decreased the expression of the RFP reporter gene compared to the CMV promoter control in both 293T and HeLa cells (Fig. (Fig.6C).6C). These results were further confirmed by quantitative Western blot and FACS analyses (see Fig. S4 at http://www.kcl.ac.uk/linden). The TRS-RBS sequence significantly reduced CMV promoter activity by 3.7-fold and 16-fold in 293T and HeLa cells, respectively. However, the 16-fold decrease in promoter activity might be overestimated, since in HeLa cells, the signal is near the limit of detection. When placed between the promoter and the reporter gene, the TRS-RBS sequence led to a significant decrease in protein levels whereas the mRNA levels were unchanged or even slightly increased (Fig. (Fig.6C).6C). These data strongly suggest that the TRS-RBS region present in the 5′ UTR of MBS85 is able to regulate the expression of a reporter gene at a posttranscriptional level, regardless of the promoter used.
We further tested whether the TRS-RBS region can interfere with the expression of a reporter gene in the context of the bidirectional promoter vector. The TRS-RBS sequence was therefore removed from the bidirectional promoter constructs used previously. The bidirectional promoter constructs were transiently transfected in 293T cells, and reporter gene expression was visualized by epifluorescence microscopy. As shown in Fig. Fig.7A,7A, higher gene expression was observed in 293T cells with a reporter construct with the TRS-RBS region deleted than with the control construct containing the TRS-RBS region (compare subpanels b and c [RFP expression] and subpanels i and j [GFP expression]). However, the TRS-RBS sequence decreased the expression level of only the sense reporter gene (compare subpanels b and c to i and j) without affecting the expression level of the antisense reporter gene (compare subpanels g and h [GFP expression] to d and e [RFP expression]). To confirm the effect of the TRS-RBS sequence on gene expression, quantitative Western blot analyses were performed using the Li-Cor Odyssey system (Fig. 7B and C). Deletion of the TRS-RBS sequence resulted in 5.4-fold (RFP expression) and 3-fold (GFP expression) increases in the promoter activity of only the sense reporter gene. The images generated by epifluorescence microscopy, furthermore, confirm the bidirectional promoter activity of the MBS85 promoter, since simultaneous expression of both RFP and GFP genes can be detected with all constructs. Together, these results indicate that the TRS-RBS region can inhibit MBS85 expression in an orientation-dependent manner, consistent with a possible role in posttranscriptional regulation. It remains to be determined whether a secondary structure or cis-acting elements are involved in the TRS-RBS-mediated inhibition of MBS85 expression.
As AAV has evolved to integrate site-specifically into a ubiquitously transcribed region (57), the question arises whether integration and the maintenance of latency are associated with MBS85 gene expression, and ultimately whether the p5 promoter of AAV has coevolved with the MBS85 regulatory elements. In order to gain a better understanding of this potential relationship, we initiated a characterization of the transcriptional activities of MBS85.
Interestingly, the RBS sequence and its position relative to the MBS85 translation initiation site are conserved among the different species, suggesting that this motif, which is essential for the viral life cycle, might also play a critical role in MBS85 regulation. It is further noteworthy that the consensus binding site for the ubiquitously expressed ZF5 transcription factor (44) overlaps with the RBS (8, 9, 12) and is one of the most frequently occurring motifs within core promoter regions (4). However, the function and biological significance of ZF5 sites have not yet been elucidated. The identification of RBS motifs within 5′ UTRs of a number of cellular genes strengthened the hypothesis of a functional role for RBS-like sequences (5, 14, 64, 65). Lackner and Muzyczka have provided evidence that the p5 RBS acts as a repressor in the presence of the adenovirus and Rep proteins (33). However, currently there is no insight into the potential role of RBS sequences in cellular transcription.
In contrast, in conjunction with the RBS, the TRS motif shows little sequence homology among the different species. However, even though the primate and mouse MBS85 RBS-TRS sequences are somewhat different, we have previously shown by in vitro endonuclease assays that Rep68 can introduce a specific nick at the mouse TRS (16). More importantly, we have further demonstrated that Rep can mediate site-specific integration into the Mbs85 locus in mouse ES cells (19). In-depth analyses using ES cell differentiation assays, as well as the generation of chimeric mice with an Mbs85-targeted transgene, indicated no detectable effects on transcription or, in fact, the ES cell potential in any of the assays employed.
Characterization of the MBS85 regulatory region in this report revealed a single TSS located 94 nt upstream from the ATG translation initiation site, indicating that the minimal signals required for AAV site-specific integration, the TRS-RBS motifs, are located within the 5′ UTR of the gene. We further demonstrated that the MBS85 promoter has bidirectional activity and provided evidence that the TRS-RBS sequence contains an inhibitory signal that affects gene expression at a posttranscriptional level. Using bidirectional promoter constructs, we also showed that the TRS-RBS region inhibits gene expression of only the sense reporter gene without affecting the expression level of the antisense gene. Although the mechanism of inhibition remains unclear, it is possible that the TRS-RBS region might form a stable RNA secondary structure around the translation initiation site, thus potentially affecting ribosome accessibility. An alternative scenario is that trans-acting RBS binding factors might be involved in the inhibition of MBS85 expression. We are currently investigating the mechanisms underlying this regulation.
In order to study the molecular mechanisms responsible for the transcriptional regulation of MBS85, we cloned the promoter region of the human MBS85 gene. We demonstrated that the region from nt −137 to +94 relative to the TSS is sufficient for basal MBS85 expression. Comparison of proximal promoter regions from MBS85 orthologs indicated that several transcription factor binding sites, such as Sp1, CRE-ATF, E4F, Staf, YY1, and ZF5, are conserved among the different species, suggesting a possible role in the regulation of MBS85 gene expression.
In order to determine whether common regulatory mechanisms might be involved in viral and target gene regulation, we compared the transcriptional activities of the human MBS85 promoter to those of AAV p5. Interestingly, of the three cis-acting regulatory elements that are involved in the regulation of p5 promoter activity, two (the RBS and YY1 sites) are also present within the minimal MBS85 promoter. In the absence of helper functions, the YY1 and Rep78/68 proteins repress the p5 promoter activity by direct binding to the recognition site (21, 31, 32, 45, 53). It has further been demonstrated that ZF5 can repress p5 promoter activity, as well. However, the mechanism for this repression does not require the p5 RBS, but instead, the RBS motif present within the viral ITR (9).
Our study further indicates that both MBS85 and p5 promoter activities are stronger in 293T cells than in HeLa cells. It can be speculated that the increased gene expression of MBS85 is attributable to the presence of the adenovirus E1A protein in 293T cells, although we have not yet found evidence in support of this hypothesis.
Previously, an enhancer-like activity was reported to be present in the region upstream from the minimal signals required for AAV site-specific integration (35). Our data show that this element is located within the minimal MBS85 promoter region. Emerging evidence indicates that as many as 20% of all human promoters have bidirectional activities (1, 43); it is therefore possible that the activities reported here can be attributed to promoter rather than to enhancer functions. Many bidirectional promoters contain shared transcription factor binding sites, and many genes regulated by bidirectional promoters appear to be coexpressed (58). It has therefore been suggested that a bidirectional arrangement provides a unique mechanism to regulate the expression of two divergently transcribed genes (58). Interestingly, YY1 binding sites are significantly overrepresented, and CpG islands often encompass the TSS within bidirectional promoters (1, 37), as we also observed for the MBS85 promoter.
To date, there is no evidence for any biological function of the antisense transcripts in vivo. We found that a possible transcript from this promoter could be the NUTF2 pseudogene, which is located between the TNNT1 and MBS85 genes. The NUTF2 protein facilitates the import of proteins into the nucleus through the nuclear pore complex and, in particular, mediates the nuclear import of RanGDP (49). However, overexpression of antisense ψNUTF2 transcripts did not significantly decrease NUTF2 protein synthesis in our hands, suggesting that this particular NUTF2 pseudogene has no biological relevance for NUTF2 expression. Several reports are converging toward the idea that short, unstable noncoding transcripts are frequently initiated in the opposite direction from protein-encoding genes, in close proximity to the TSS. It still needs to be determined whether these short transcripts, which are widespread throughout the human and yeast genomes, have any biological functions (13, 18, 26, 46, 52, 66) and whether the activity observed with the MBS85 promoter could fall into this category.
The ability of AAV to site-specifically integrate its genome into the MBS85 gene ultimately raises the question of the uniqueness of this locus or, alternatively, the contribution of the locus to integration and/or the maintenance of viral latency. Despite the fact that Rep78/68 can interact with a large number of RBSs scattered throughout the human genome (64, 68), MBS85 has so far remained the only locus satisfying the requirements for AAV integration (42, 62). One component, which is also shared by other integrating viruses, is the open chromatin structure (35), which is consistent with the ubiquitous expression of MBS85. A further intriguing aspect is that AAV has evolved to share what we have demonstrated to be a regulatory element, the RBS, with its target locus, highlighting the possibility of coregulation of the viral and cellular promoters. This particular aspect raises the possibility that through this and possibly more shared regulatory elements, the viral Rep protein might be able to influence the transcriptional activity of the target locus, which in turn could aid in the maintenance of viral latency in the absence of helper virus infection and rescue under permissive conditions. In addition, our finding that AAV integrates in an oriented manner, with the p5 promoter consistently forming the initial junction (19), invites the hypothesis that shared promoter binding factors (such as Rep and YY1) could be involved in the assembly of the integration complex, thus directly linking transcription, Rep-mediated replication, and subsequent integration into one mechanistic framework.
This work was supported by U.S. National Institutes of Health grants GM071023 and GM073901 to R.M.L. E.H. was supported by a Senior Fellowship in Biomedical Science from the Charles H. Revson Foundation.
We are grateful to all members of the laboratory for helpful comments and discussions. We thank Chad Swanson for discussions and critical reading of the manuscript.
Published ahead of print on 16 September 2009.