Functional analysis of the sea urchin-derived arylsulfatase (Ars)-element in mammalian cells

Authors

  • Satoshi Watanabe,

    Corresponding author
    1. Aminal Genome Research Unit, Division of Animal Science, National Institute of Agrobiological Sciences, Ikenodai 2, Tsukuba, Ibaraki, 305-8602, Japan
      *Correspondence: E-mail: kettle@affrc.go.jp
    Search for more papers by this author
  • Sachiko Watanabe,

    1. Aminal Genome Research Unit, Division of Animal Science, National Institute of Agrobiological Sciences, Ikenodai 2, Tsukuba, Ibaraki, 305-8602, Japan
    Search for more papers by this author
  • Naoaki Sakamoto,

    1. Department of Mathematical and Life Sciences, Graduate School of Science, Hiroshima University, Kagamiyama, Higashi-Hiroshima 739-8526, Japan
    Search for more papers by this author
  • Masahiro Sato,

    1. The Institute of Medical Sciences, Tokai University, Bohseidai, Isehara, Kanagawa, 259-1193, Japan
    Search for more papers by this author
      Frontier Science Research Center, Kagoshima University, Korimoto 1-21-40, Kagoshima 890–0065, Japan.
  • Koji Akasaka

    1. Department of Mathematical and Life Sciences, Graduate School of Science, Hiroshima University, Kagamiyama, Higashi-Hiroshima 739-8526, Japan
    Search for more papers by this author
      bMisaki Marine Biological Station, Graduate School of Science, University of Tokyo, 1024, Koajiro, Misaki, Miura, Kanagawa 238–0225, Japan.

  • Communicated by: Shinichi Aizawa

  • a

    Present addresses: Frontier Science Research Center, Kagoshima University, Korimoto 1-21-40, Kagoshima 890–0065, Japan. bMisaki Marine Biological Station, Graduate School of Science, University of Tokyo, 1024, Koajiro, Misaki, Miura, Kanagawa 238–0225, Japan.

*Correspondence: E-mail: kettle@affrc.go.jp

Abstract

An insulator is a DNA sequence that has both enhancer-blocking activity, through its ability to modify the influence of neighboring cis-acting elements, and a barrier function that protects a transgene from being silenced by surrounding chromatin. Previously, we isolated and characterized a 582-bp-long element from the sea urchin arylsulfatase gene (Ars). This Ars-element was effective in sea urchin and Drosophila embryos and in plant cells. To investigate Ars-element activity in mammalian cells, we placed the element between the cytomegalovirus enhancer and a luciferase (luc) expression cassette. In contrast to controls lacking the Ars-element, NIH3T3 and 293T cells transfected with the element-containing construct displayed reduced luciferase activities. The Ars-element therefore acts as an enhancer-blocking element in mammalian cells. We assessed the barrier activity of the Ars-element using vectors in which a luc expression cassette was placed between two elements. Transfection experiments demonstrated that luc activity in these vectors was approximately ten-fold higher than in vectors lacking elements. Luc activities were well maintained even after 12 weeks in culture. Our observations demonstrate that the Ars-element has also a barrier activity. These results indicated that the Ars-element act as an insulator in mammalian cells.

Introduction

Enhancers and silencers are cis-acting elements, located either upstream or downstream of a gene, that can affect promoter activity in an orientation-independent manner. In eukaryotes, such elements can be effective even when they are located several hundreds of kilobases away from a target gene. Furthermore, the action of these cis-acting elements is not specific to the gene with which they are associated but can affect other, unrelated genes (Yaniv 1984). The apparent function of these cis-acting elements is to provide transcriptional diversity within the complex gene regulation system of eukaryotes.

In differentiated cells, the majority of genes are inactivated. Only those genes that play important roles in maintaining essential cellular functions are actively transcribed. Gene inactivation can be achieved through various mechanisms: DNA methylation; deacetylation of histones; and chromatin condensation. In inactivated chromatin, DNA is highly methylated (Cedar 1988) and the histones are deacetylated (Davie 1996). If a foreign gene is inserted into an inactive chromosomal region, its pattern of gene expression is affected by the chromosomal environment (Levis et al. 1985). This phenomenon is called the “position effect” (PE). In both cultured cells and transgenic animals, gene silencing or unexpected expression of the inserted genes is often seen and is thought to be a consequence of the PE.

The effect of cis-acting elements such as enhancers and silencers can be limited by a chromosomal boundary termed an “insulator.” Insulators can abolish the PE and thereby maintain transcriptional activity of a foreign gene (Chung et al. 1993; Gdula et al. 1996). Insulators are identified as DNA fragments that exhibit enhancer-blocking activity, a property that may be mediated by binding of the transcriptional regulator (West et al. 2002).

Analyses of a wide range of eukaryotes, such as yeast, Drosophila, sea urchins, Xenopus, chicks, mice and humans, have led to the identification of a continually increasing number of insulators reviewed by West & Fraser (2005). Insulator-binding proteins can recruit histone acetyltransferase (HAT), an enzyme that acetylates the histones of chromatin adjacent to the insulators (West et al. 2004). These acetylated histones prevent chromatin condensation, one of the principal characteristics of gene silencing. Attempts have been made to identify consensus sequences on insulators to which insulator-binding proteins may bind, but to date these have been unsuccessful. There are a number of probable reasons for this lack of success:

  • 1sequence similarity among insulators is very low;
  • 2many insulator-binding proteins have been identified;
  • 3some insulators are species-specific, while others are not.

At present, the most productive means for exploring the biological function of an insulator is by transfection of insulator-containing constructs into cultured cells.

The sea urchin Ars gene is strongly transcribed from the blastula to late gastrula stages, and its expression is confined to the ectodermal cell lineage (Kiyama et al. 1998). The Ars gene contains an enhancer element, termed otx (Morokuma et al. 1997). We recently found that a 582-bp fragment, located upstream of the Ars gene, functions as an insulator in both sea urchin and Drosophila embryos and in plants (Takada et al. 2000; Nagaya et al. 2001). We therefore termed this fragment “Ars-insulator” (Akasaka et al. 1999). Although Ars-insulator-containing transgenes are actively transcribed in transgenic mice, the exact function of the insulator in mammalian cells remains unclear (Takada et al. 2000). In this study, we have extensively assessed the function of the Ars-insulator in mammalian cells, using transfection of Ars-insulator-containing constructs and found that the insulator has both enhancer-blocking and PE protection activities.

Results

Enhancer-blocking activity of the Ars-insulator

We constructed a luc expression cassette (pL) containing cytomegalovirus (CMV) minimal promoter (mp), firefly luc cDNA and SV40 poly(A) signals as the basic vector for the enhancer-blocking assay. CMV-E, or Ars-insulator, or both were ligated upstream of pL. In total, 14 different expression vectors, including control vectors such as PGK-LUC and AL, were constructed as shown in the left panel of Fig. 1(A). Vectors (EHpL+/–) carrying the HS4 insulator derived from chicken β-globin locus were also used as controls. When the vectors listed in the left panel of Fig. 1A were introduced into NIH3T3, luc activities in the cells transfected with EApL+/– were approximately three-fold lower than those in cells transfected with EpL (control vector) (right panel in Fig. 1A). A similar reduction in luc activity was seen in cells transfected with the HS4 insulator-containing constructs [EHpL+/–] (right panel in Fig. 1A). Our observations show that the Ars-insulator has enhancer-blocking activity. The orientation of the Ars-insulator did not appear to be important for its enhancer-blocking element; EApL+ and EApL– produced comparable reductions in luc activity (right panel in Fig. 1A). No difference in luc activity was seen between cells transfected with EApL+, AEApL+ or RevAEApL+ (right panel in Fig. 1A), indicating that the bi-directional activation of CMV mp by CMV-E on circular plasmids could be eliminated. Similarly, no difference in luc activity was observed between cells transfected with constructs containing HS4, EHpL+, HEHpL+ or RevHEHpL+ (right panel in Fig. 1A). Furthermore, the Ars-insulator did not appear to have either enhancer or promoter activities; the levels of luc expression were comparable in cells transfected with either ApL+/– or AL and similar to those of cells carrying pL (right panel in Fig. 1A). Moreover, the Ars-insulator did not have a silencer activity; the luc activities in cells transfected with AEpL and RevAEpL were statistcally similar to those of EpL (right panel in Fig. 1A). The luc activities were comparable in cells transfected with either HEpL or RevHEpL. Similar results were obtained when we carried out parallel experiments using 293T cells (Fig. 1B).

Figure 1.

(A) Luc activity in NIH3T3 cells transfected transiently with Ars-insulator-containing constructs. The reporter gene constructs used are shown in the left panel. Arrows represent the orientation of each insulator. CMV-E, cytomegalovirus early enhancer; HS4, HS4 insulator derived from chicken β-globin gene; Ars, Ars-insulator; LUC, firefly luciferase cDNA; mp, CMV minimal promoter; p(A), SV40 poly (A) signal sequence; PGK, mouse phosphoglycerate kinase-1 promoter. All the constructs used for transfection were in the circular form. Firefly luc activity was normalized against Renilla luc activity in order to compensate for differences in transfection efficiency and shown as RLU/mg of protein. The normalized activities were analyzed statistically by a two-tailed Student's t-test. The mean values were presented with their standard errors for each transfection group. Negative control, cells without addition of plasmid DNA. (B) Luc activity in 293T cells transfected transiently with Ars-insulator-containing constructs.

Introduction of vectors over-expressing CTCF or anti-sense CTCF

We examined the question of whether a reduction in the amount of endogenous CTCF caused a correlated decrease in the enhancer-blocking activity of the Ars-insulator. Two down-regulation vectors, anti-CTCF and siCTCF, were constructed in order to suppress production of endogenous CTCF by anti-sense inhibition. They were transfected into 293T cells with the EApL– vector (Fig. 2A). OvCTCF, a CTCF-over-expressing vector, was used as a control with EApL–. Transfection with OvCTCF +EApL– resulted in enhancement of the enhancer-blocking activity of the Ars-insulator (OvCTCF+EApL–vs EApL– in the right panel of Fig. 2A). Transfection with siCTCF +EApL–, however, caused a decrease in the enhancer-blocking activity of the Ars-insulator (siCTCF+ EApL–vs EApL– in the right panel of Fig. 2A). Transfection with anti-CTCF+EApL– failed to reduce the enhancer-blocking activity, although it did appear to cause a decrease in the endogenous level of CTCF protein (lane 3 in (a) of Fig. 2B). These data suggest a correlation of the amount of endogenous CTCF with the ability of the Ars-insulator to exhibit its enhancer-blocking activity. To confirm that expression of endogenous CTCF protein is actually suppressed by anti-sense RNA expression, Western blotting analyses were performed (Fig. 2B). Transfection with either the anti-CTCF or siCTCF vector led to a reduction in cellular levels of CTCF (lanes 4 and 5 vs lane 5 in Fig. 2Ba). The inhibitory effect of siCTCF appeared to be larger than that of anti-CTCF (lane 3 vs lane 4 in Fig. 2Ba). The level of CTCF appeared to increase when OvCTCF was introduced (lane 1 vs lane 2 in Fig. 2Ba). These observations strongly suggest that the enhancer-blocking effect of the Ars-insulator is mediated in mammalian cells through binding to cellular CTCF.

Figure 2.

(A) Luc activity in 293T cells transfected with vectors for CTCF over-expression and down-regulation. The vectors used here are shown in the left panel. OvCTCF is a CTCF-over-expressing vector in which CTCF cDNA is placed in a sense orientation. AntiCTCF is an anti-sense CTCF-expressing vector in which CTCF cDNA is placed in a reverse orientation. siCTCF is a vector that expresses anti-sense CTCF siRNA. The sequences below siCTCF indicate oligonucleotides used for expression of siRNA. Arrows indicate direction of the target sequence for CTCF. CAG, CMV early enhancer and chicken β-actin promoter; mCTCF, mouse CTCF cDNA; p(A), SV40 poly(A) signals; U6, mouse RNA polymerase III (PolIII) promoter. Negative control, cells without addition of plasmid DNA. Luc activity was measured at 48 h after transfection and expressed as RLU/mg of protein. (B) (a) Western blotting analysis of CTCF. Nuclear extracts prepared from the transfected cells were subjected to Western blotting analysis using anti-CTCF antibody. A band specific for CTCF is indicated by an arrow. siEGFP is a vector that expresses anti-sense EGFP siRNA and was used as negative control. pRL-TK is a control plasmid used for dual assay of luc activity. Signal intensity of each band is shown below the blot. Lane 1, cells transfected with EApL– + pRL-TK; lane 2, cells transfected with EApL– + OvCTCF + RL-TK; lane 3, cells transfected with EApL– + anti-CTCF + pRL-TK; lane 4, cells transfected with EApL– + siCTCF + pRL-TK; lane 5, cells transfected with EApL– + siEGFP + pRL-TK. (b) The CBB stained pattern of a gel prior to Western blotting shown as a loading control.

PE protection activity of the Ars-insulator

To test whether the Ars-insulator possesses a PE protection activity, we first constructed a firefly luc expression basic cassette, PGK-LUC (left panel in Figs. 1A and 3), in which the mouse phosphoglycerate kinase (PGK-1) promoter controls luc expression. Eight vectors (FF, FR, RF, RR, IL, OL, SL2 and SL3) were then constructed by adding the Ars-insulator either to one end or to both ends of the PGK-LUC vector (left panel in Fig. 3). Prior to transfection of the vectors into NIH3T3 cells, the plasmid backbones of the constructs were removed by restriction enzyme digestion to avoid any possible effects by contaminating plasmid components on gene expression of Ars-insulator-containing expression units. After transfection of each of the Ars-insulator-containing plasmids with the PGK-puro-p(A) cassette, the cells were selected in the presence of puromycin. The resulting recombinant cells were used to assay luc activity without being subjected to cloning (right panel in Fig. 3). Cells transfected with both the PGK-LUC vector and the Ars-insulator(s) exhibited higher luc activities than those transfected with PGK-LUC alone. Furthermore, the orientation of each of the two Ars-insulators appeared to play an important role in the effective expression of luc from a PGK-LUC cassette. For example, OL exhibited the strongest luc activity of the constructs tested (right panel in Fig. 3). OL has two insulators with opposite orientations. SL2 and SL3, which also have two Ars-insulators but with different orientations to those of OL, have much lower luc activities (right panel in Fig. 3).

Figure 3.

Reporter gene constructs used for analysis for position-effect protection and luc activity in stable transfectants obtained after gene transfer. Several Ars-insulator-containing vectors, shown in the left panel, were constructed by adding an Ars-insulator to one or both ends of the PGK-LUC cassette. Arrows represent the Ars insulator and its direction (5′ to 3′ orientation of the sequence). These vectors were removed from plasmid backbone and then introduced into cells with PGK-puro-p(A). NIH3T3 cells transfected with each construct were selected using puromycin, and stable transfectants obtained. At least eight clones for each construct were obtained. Luc activity is expressed as RLU/mg of protein. Note that cells transfected with OL exhibit remarkably high degree of luc activity. LUC, firefly luciferase cDNA; p(A), SV40 poly(A) signal sequence; PGK, mouse phosphoglycerate kinase-1 promoter.

Why did cells transfected with OL exhibit a higher degree of luc activity than those transfected with similar constructs?

To investigate this question, we examined the possible correlation between luc activity and copy number of transgenes integrated into host chromosomes. We prepared several stable transfectants after transfection with OL + GK-puro-p(A) or PGK-LUC + GK-puro-p(A). Transfectants carrying OL were termed “OL,” and those carrying PGK-LUC were termed “L.” The lysates prepared from these clones were used for measurement of luc activity. The amount of protein was determined at the same time. Genomic DNA was isolated from each clone and subjected to real-time PCR to determine copy numbers of the introduced transgenes. Luc activity was expressed as RLU/mg of protein/copy number. The average luc activity/copy number from OL transfectants (OL1 to OL9) was approximately 42-fold higher than that from L transfectants (L1 to L9) (Fig. 4). Notably, only three of the OL transfectants exhibited luc activity less than 10 RLU/mg of protein/copy number, while seven of nine L transfectants had low activities. The average copy number of OL transfectants was approximately four-fold higher than that of L transfectants (Fig. 4).

Figure 4.

Luciferase activities in each recombinant clone for OL (upper panel) and PGK-LUC (designated as “L”) (lower panel). Nine clones for each construct were obtained and propagated. Luc activity was measured for each clone. Copy number of the transgenes in each clone was concomitantly determined using real-time PCR. Luc activity is plotted as RLU/mg of protein/one copy of the transgene (right panel). NC indicates negative control (cells without addition of plasmid DNA).

Effect of Ars-insulator on transgene silencing

To evaluate the anti-silencing activity of the Ars-insulator, we introduced OL, IL and PGK-LUC vectors with PGK-puro-p(A) into NIH3T3 cells. The transfected cells were selected with puromycin for 1 week and, finally, four clones for each construct were isolated. These clones were then cultured in the absence of puromycin and periodically tested for luc expression. Luc expression activity from PGK-LUC started to fall soon after removal of puromycin, and the rate declined to approximately 40% of puromycin-selected counterparts (Fig. 5A). In contrast, luc expression from OL and IL was maintained throughout the assay period (Fig. 5A). A similar result was obtained in three independent experiments. Interestingly, luc expression from the Ars-insulator containing vector continued to increase 1 week after puromycin removal, but then fell and stabilized. Notably, OL had a stronger effect on the enhancement of luc expression than IL (Fig. 5B). These findings indicate that the Ars-insulator is effective for stabilizing gene expression from the integrated transgenes as well as for blocking the PE.

Figure 5.

Luciferase expression in NIH3T3 cells transfected with PGK-LUC, OL, and IL together with PGK-puro-p(A). After transfection, these cells were selected by puromycin for 1 week, and then cultured in the absence of puromycin. Sequential changes of luc activity after removal of puromycin are shown by the percentage of luc activity. (A) RLU of an original clone/RLU of a clone cultured for up to 12 weeks in the absence of puromycin. (B) Mean luc activity.

Discussion

Insulators have been isolated from invertebrates such as Drosophila (Geyer & Corces 1992; Kellum & Schedl 1992; Hagstrom et al. 1996; Ohtsuki & Levine 1998; Vazquez & Schedl 2000) and the sea urchin (Palla et al. 1997; Akasaka et al. 1999), and from vertebrates such as the chick (Chung et al. 1993; Robinett et al. 1997; Saitoh et al. 2000), the mouse (Holmgren et al. 2001; Farrell et al. 2002; Gombert et al. 2003) and humans (Li & Stamatoyannopoulos 1994; Namciu et al. 1998; Antes et al. 2001; Filippova et al. 2001). Su(Hw) of Drosophila is the most extensively analyzed of the insulators, and there are reports that it functions in mammalian cells (Yao et al. 2003). The HS4, which is one of the insulators well proven to be functional in mammalian species, exhibits both enhancer-blocking and PE protection activities (Bell & Felsenfeld 1999; Inoue et al. 1999). We demonstrate that Ars-element can act as an insulator in mammalian cells similar to the HS4 insulator. Ars-insulator has the enhancer-blocking activity when it was positioned only between the CMV enhancer and the promoter. When located “outside,” the Ars-insulator did not affect luc expression, indicating that the Ars-insulator may act as an insulator and not as a silencer. These results could be confirmed also by introducing a linearized plasmid in order to avoid the possibility of bi-directional activation of the enhancer in circular plasmids.

The binding of CTCF to the HS4 insulator is required for manifesting the enhancer-blocking activity of the HS4 insulator itself (Bell et al. 1999). CTCF is a cis-acting transcription inhibitor that is thought to play important roles in “globin gene switching” during late developmental stages (Bell et al. 1999). The over-expression experiment of CTCF increased the enhancer-blocking effect of the Ars-insulator, while forced expression by anti-sense CTCF RNA or CTCF siRNA weakened its effect (see Fig. 2A). Our data suggested a possibility that mammalian CTCF can bind to Ars-insulator through which it exhibits the enhancer-blocking. However, unfortunately, we failed to demonstrate CTCF binding to the Ars-insulator by using ChIP assay and electrophoresis mobility shift assay (EMSA) (data not shown). Hino et al. (2006) also reported that in stable transformants carrying the Ars-insulator-containing vector, CTCF binding to the Ars-insulator could not be detected by a ChIP assay using an anti-CTCF antibody. It has been reported that CTCF can bind to the relatively GC rich sequences of the HS4 insulator, but the exact sequence to which CTCF binds remains to be determined (Ohlsson et al. 2001). Since the Ars-insulator contains GC-rich sequences (Akasaka et al. 1999), the HS4 and Ars-insulators appear to share similar properties. Further analysis is required for possible interaction between CTCF and Ars-insulator. To date, two sea urchin-derived insulators, histone H2A sns (Palla et al. 1997) and Ars, have been reported. Di Simone et al. (2001) first reported that the sns insulator is functional in mammalian cells and showed the possibility that CTCF binds to the sns insulator when the enhancer-blocking event occurs. We anticipate that the enhancer-blocking activity of an insulator in the sea urchin cells may be promoted through interaction with insulator-binding factors such as CTCF, although sea urchin CTCF has not yet been identified.

As it was not known whether sea urchin-derived insulators would exhibit PE protection activity in mammalian cells, we examined this possibility using the Ars-insulator. To this end, we constructed eight PGK-LUC-based vectors carrying one or two Ars-insulators at the end of a PGK-LUC cassette (left panel of Fig. 3). After these constructs were introduced into NIH3T3 cells with the PGK-puro-p(A) cassette, transfectants were screened by puromycin and their luc activities were determined. Regardless of the number or orientation of Ars-insulators, the presence of an Ars-insulator resulted in increased expression of the PGK-LUC cassette (see right panel of Fig. 3). Interestingly, OL, which possesses Ars-insulators with opposite orientations at each end of the PGK-LUC cassette, exhibited the strongest luc activity (see right panel of Fig. 3). For example, luc activity in OL transfectants was approximately 100-fold higher than that in L transfectants carrying the PGK-LUC construct. Similar observations have previously been obtained in the sea urchin system (Akasaka et al. 1999; Takada et al. 2000; Nagaya et al. 2001; Hino et al. 2004). These findings indicate that the Ars-insulator may be a valuable tool for stabilizing and increasing gene expression in the fields of gene therapy and transgenic animal studies. Indeed, the HS4 insulator has already been used for construction of improved viral vectors for use in gene therapy (Neff et al. 1997; Inoue et al. 1999; Emery et al. 2000) and for production of transgenic mice (Potts et al. 2000; Pantano et al. 2002; Rival-Gervier et al. 2003). However, it remains to be determined whether increased gene expression in transfectants carrying Ars-insulator-containing constructs is a result of increased efficiency of gene transfer or of increased efficiency of transgene expression itself. To assess this question, we isolated stable transfectants carrying PGK-LUC (L) or OL and determined the copy number of the integrated transgenes and also assessed the luc activity of the transfectants (Fig. 4). When copy numbers were compared between the two types of transfectants, OL transfectants had approximately four-fold more copies of the transgenes than the L transfectants (see Fig. 4). Furthermore, OL transfectants exhibited luc activity 42-fold higher than L transfectants (see Fig. 4). These findings suggest that the Ars-insulator affects both integration and expression efficiency of a transgene. The reason why transgene integration efficiency is higher when an Ars-insulator-containing construct is introduced remains unknown. Jeffreys et al. (1998) reported that GC-rich minisatellite repeats affected the efficiency of recombination in the human genome. As the Ars-insulator contains GC-rich sequences, this may indicate a possible means by which the insulator can promote efficient transgene integration.

In recent studies aimed at examining the role of insulators in protecting the PE (Emery et al. 2000; Takada et al. 2000; Villemure et al. 2001), CMV and elongation factor (EF)-1α promoters and viral LTRs have been employed for driving transgene expression. However, the effects of the insulators were not clearly shown. This is in part due to the use of such strong promoters. To avoid this problem, we used the weaker mouse PGK-1 promoter in this study.

As previously described, the HS4 insulator has been reported to be effective for protection against transgene silencing in a transgenic system (Potts et al. 2000; Rivella et al. 2000; Ciana et al. 2001; Pantano et al. 2002; Frazar et al. 2003; Guglielmi et al. 2003; Pant et al. 2003; Rival-Gervier et al. 2003). The Ars-insulator was also shown to be useful for avoiding transgene silencing in transgenics (Takada et al. 2000; Nagaya et al. 2001; Hino et al. 2004). Our present finding that the Ars-insulator can increase the efficiency of gene expression from a transgene supports the previous reports. Interestingly, there are indications that the efficiency of transgene expression is greatly affected by topological differences between insulators on the transgene construct (Yannaki et al. 2002; Hino et al. 2004). For example, Yannaki et al. (2002) pointed out that the orientation of the HS4 insulator strongly affected the efficiency of transgene expression, and suggested that the distance between the core region of HS4 and the promoter is important for this effect. In this study, we observed variable levels of luc activity in IL, OL, SL2 and SL3, although the orientation of only one insulator differed between the constructs (see right panel of Fig. 3). Since the Ars-insulator contains a specific region exhibiting 65% homology to the G-rich HS4 core region (Chung et al. 1997; Recillas-Targa et al. 2002), the distance between this specific region of the Ars-insulator and the PGK-1 promoter could have affected luc activity in the manner suggested by Yannaki et al. (2002).

We observed that almost all OL recombinants retained their original luc activities even after culture for over 3 months, whereas luc activity fell in PGK-LUC recombinants (see Fig. 5). This suggests that the Ars-insulator can continue to express its anti-silencing activity even after chromosomal integration. In addition, among OL recombinants recovered, we obtained clones that exhibited low luc activities but that possessed high copy numbers of transgenes; other clones exhibited high luc activities but possessed low copy numbers of transgenes (Fig. 4). This suggests that the Ars-insulator does not interfere with an extremely strong PE. In the assays of PE blocking using FF, FR, RF, RR, IL, OL, SL2 and SL3 constructs, the Ars element does appear to increase the average luc expression levels especially in the OL construct. However, as shown in Fig. 4, there is a wide variation in expression at different integration sites. In other words, the OL construct is highly susceptible to PE even though some integration sites can express to very high levels. This variation in expression of exogenous genes appeArs to be an important aspect of PE that has not been studied extensively. Giraldo et al. (2003) pointed out in their review that the expression level of insulator-containing transgenes was unrelated to the copy number of integrated transgenes, which is in agreement with our results.

Hino et al. (2004) reported that, in cells carrying insulator-containing transgenes, histone acetylation occurs more frequently than in cells carrying insulator-free transgenes. In yeast, histone acetylation is thought to be controlled by the balance between histone acetylases (HATs) and histone deacetylases (HDACs) (Kimura et al. 2002; Suka et al. 2002). When both enzymes are localized to the nuclear matrix of a cell (Davie 1996, 1997), histone acetylation may occur frequently. Interestingly, Yusufzai & Felsenfeld (2004) reported that as CTCF expression increases, the HS4 insulator becomes localized in the nuclear matrix fraction. In an attempt to examine the possible interaction between CTCF and histone acetylation-related enzymes (HATs and HDACs), we introduced CTCF over-expressing and CTCF down-regulating vectors (anti-CTCF and siCTCF) into OL recombinant clones. However, no change in luc activity was noted in any of the OL recombinants transfected with these vectors (data not shown). Thus, other protein(s) beside CTCF may be involved in insulator-mediated protection of the PE. Our present results appear to be consistent with the reports of Felsenfeld's group who demonstrated that PE blocking of the HS4 insulator is independent of CTCF (Recillas-Targa et al. 2002; West et al. 2004) and that USF could bind to HS4 and recruit histone modification enzymes for PE protection. Moritani et al. (2004) found that Unichrom, a sea urchin ortholog of RSF-1, could bind to a G-stretch in the Ars-insulator. Since Loyola et al. (2003) demonstrated that RSF-1 functioned as a chromatin assembly factor, RSF-1 may also play an important role in PE protection.

In summary, in this study we have demonstrated that the Ars-insulator has both enhancer-blocking and PE protection activities in mammalian cells. The former activity is independent of orientation of the insulator, but the latter activity depends on the topology of the insulator. The Ars-insulator increases both the efficiencies of gene expression from the integrated transgenes and of transgene integration, although it is still unclear why the Ars-insulator can affect the efficiency of transgene integration. The Ars-insulator does not have either enhancer or promoter activity. Since the length of the Ars-insulator is very short (582 bp), just half of that of the HS4 insulator (1201 bp), it is convenient for use by researchers constructing insulator-containing transgenes. Although further research is still required for exploring other Ars-insulator-binding proteins to elucidate the PE protection activity of the Ars-insulator, we believe that the Ars-insulator will be a powerful tool for increasing the efficiency of gene expression in transgenic systems and for improving expression vectors in the field of gene therapy.

Experimental procedures

Vectors for analysis of enhancer-blocking activity

The CMV mp of the pTRE2 plasmid was PCR-amplified using the primers 5′-ccgctcgagacgcgaattccggtaggcgtg-3′ and 5′-cccaagcttctatggaggtcaaaacag-3′ (Clonetech Laboratory, Palo Alto, CA, USA). The PCR product, digested with XhoI and HindIII, was subcloned between the XhoI and HindIII sites of a pica gene basic vector 2 (Dainippon Ink Co., Tokyo, Japan) carrying a luc expression cassette. The resulting vector was sequenced for confirmation of fidelity of construction and named pL (left panel in Fig. 1A). Ars-insulator, excised from pBSK-SmBM+ (Akasaka et al. 1999), was placed at the 5′ end of the pL cassette to obtain ApL+/– (left panel in Fig. 1A). The symbol +/– in ApL+/– denotes the orientation (forward or reverse) of the Ars-insulator. The CMV early enhancer (E), isolated by digestion with SalI and NcoI from the pCAGGS expression vector (Niwa et al. 1991), was subcloned into the EcoRV site of pBluescript II SK(+) (Stratagene, La Jolla, CA, USA) by blunt-end ligation. The orientation of the enhancer in the resulting clone was confirmed and the plasmid named pCMV-E(+/–). The symbol (+/–) in pCMV-E(+/–) denotes the orientation (forward or reverse) of the CMV E enhancer. The pL cassette was separated from its backbone vector by digestion with XhoI and SalI, and was subcloned into the XhoI site of the pCMV-E(+) so as to be located downstream of CMV E. The resulting plasmid was named EpL (left panel in Fig. 1A). Ars-insulator, excised from pBSK-SmBM+ by digestion with BamHI and PstI, was inserted into the XhoI site between the enhancer and promoter of EpL. The resulting clones were named EApL+/– (left panel in Fig. 1A). The presence of the insulator was confirmed and its orientation determined. The symbols +/– in EApL+/– denote the orientation (forward or reverse) of the Ars-insulator for the luc gene. HS4 insulator, isolated from EcoRI-digested pCRBlunt-HS4 (Hino et al. 2004), was subcloned into the XhoI site of EpL to obtain EHpL+/– (left panel in Fig. 1A). A control vector, AL (left panel in Fig. 1A), was also constructed by placing the Ars-insulator at the 5′ end of a promoterless luc expression cassette LUC-p(A). AEApL+ and RevAEApL+ were constructed by blunt-end ligation of BamHI and PstI digested Ars-insulator into the SalI site of EApL+. Similarly, HEHpL and RevHEHpL were constructed by blunt-end ligation of EcoRI-digested HS4 into the SalI site of EHpL+. AEpL and RevAEpL were constructed by blunt-end ligation of BamHI and PstI-digested Ars-insulator into the SalI site of the 5′ end of EpL. Similarly, HEpL and RevHEpL were constructed by blunt-end ligation of EcoRI-digested HS4 into the SalI site of the 5′ end of EpL.

Vectors for analysis of PE protection activity

PGK-1 promoter, isolated from SacI and HindIII-digested pPGK-puro (Watanabe et al. 1995), was inserted between the SacI and HindIII sites of a pica gene basic vector 2 to obtain PGK-LUC (left panel in Fig. 1A). The PGK-LUC cassette was isolated from SalI-digested PGK-LUC and then inserted into the SalI site of pBSK-SmBM+ or pBSK-SmBM– plasmids. The resulting plasmids were named FF, FR, RF and RR (left panel in Fig. 3). FF and FR have an Ars-insulator in either sense or reverse orientation upstream of the PGK-LUC cassette, respectively. RF and RR have an Ars-insulator in either sense or reverse orientation downstream of the PGK-LUC cassette, respectively. An additional Ars-insulator, isolated from pBSK-SmBM+ digested with BamH1and PstI, was inserted into these four vectors to obtain vectors IL, OL, SL2 and SL3 (left panel in Fig. 3) containing two copies of the Ars-insulator, one at each end of the PGK-LUC cassette. DNA sequencing was performed on all constructs to confirm that the insulator had been correctly inserted into each vector.

Vectors over-expressing CTCF or anti-sense CTCF cDNA

Vectors over-expressing CTCF or anti-sense CTCF cDNA were constructed by inserting mouse transcription factor CTCF cDNA (Filippova et al. 1996) into the expression vector pCAGGS. Briefly, CTCF cDNA was amplified by RT-PCR, with the primers 5′-ggaattccggagaggaagggggagtgg-3′ and 5′-ggaattccaaggccccagcatcaccgg-3′, using mouse liver mRNA and KOD polymerase (TOYOBO Co., Tokyo, Japan). The 2.2 kb amplified fragment was subcloned into the BamHI site of pBluescript II SK(+), and fidelity of the insert was confirmed by sequencing. The CTCF cDNA was next subcloned into the EcoRI site of pCAGGS via blunt-end ligation. The resulting plasmids, termed OvCTCF and anti-CTCF (left panel in Fig. 2A), have CTCF cDNA in sense and reverse orientations, respectively.

Anti-CTCF siRNA expression vector

The synthetic oligonucleotides 5′-aaaagggccc TCAACAGCAGT-GTACAGATTTCAAGAGAATCTGTACACTGCTGTTGATT T TTgaattccg-3′ and 5′-cggaattcAAAAATCAACAGC AGTGTA-CAATTTCTTGAAATCTGTACACTGCTGTTGAgggccctttt-3′, which correspond to a 5′ portion of mouse CTCF cDNA (Filippova et al. 1996), were annealed together and then digested with ApaI and EcoRI. The enzyme-digested fragments were inserted into the ApaI and EcoRI sites of the pSilencer1.0 vector (Ambion, Austin, TX, USA) to obtain siCTCF (left panel in Fig. 2A). The synthetic oligonucleotides 5′-aaaagggcccCCACTACCTGAGCACCCAGTTCAAGAGACT-GGGTGCTCAGGTAGTGGTTTTTgaattccg-3′ and 5′-cggaattcCGGGATCCCTAGTTTCCAAAAAAACCACTACCTGAGCACCCAGTCTCTTGAACTGGGTGCTCAGGTAGTGGTTGGGTGCAAAGCTTGGGgggccctttt-3′, which correspond to a 5′ portion of enhanced green fluorescent protein (EGFP) cDNA (Hasuwa et al. 2002), were annealed and then also digested with ApaI and EcoRI. The fragments were inserted into the pSilencer1.0 to obtain siEGFP (left panel in Fig. 2A).

Cell culture and transfections

NIH3T3 cells (a gift from Dr Tadashi Furusawa of the National Institute of Agrobiological Sciences, Japan) and 293T cells (a gift from Dr Shoji Tajima of Osaka University, Japan) were cultured in Dulbecco's modified Eagle's medium (DMEM; Sigma Co. Ltd, St. Louis, MO, USA) supplemented with 10% fetal bovine serum (FBS), 50 units of penicillin and 50 µg of streptomycin per ml at 37 °C under an atmosphere of 5% CO2.

For the evaluation of enhancer-blocking activity, cells were seeded in 24-well plates (Nunc Co., Roskilde, Denmark) at a density of 2.5 × 104 cells per well. The next day, 400 ng of each of circular plasmid construct and 40 ng of pRL-TK, a Renilla luc expression vector (Promega Co., Madison, WI, USA), were mixed with Lipofect Amine Plus reagent (Invitrogen Co., Carlsbad, CA, USA) and added to each well, according to the method described by the manufacturer. Twelve wells were used for transfection of each construct. Wells that did not receive constructs were considered “negative controls.” Culture media were removed 48 h after transfection. The cells were lyzed in situ by adding 100 µL of lysis buffer (Promega Co.) at room temperature. They were then kept at v80 °C until the luc assay was performed.

For the evaluation of the position-effect protection activity, 100-mm dishes (Becton Dickinson, Franklin Lakes, NJ, USA) were each seeded with 1 × 106 cells. DNA, completely lacking vector backbone, was isolated by restriction enzyme digestion and purification in agarose gels (Nacalai Tesque Co., Kyoto, Japan) prior to use for transfection. On the day following cell seeding, 8 µg of the purified DNA fragments and 0.8 µg of SalI-linearized PGK-puro-p(A) cassette were mixed with Lipofect Amine Plus reagent and the mixture was added to each dish. The cells were trypsinized and seeded on to 5 × 6-well plates (Nunc Co.) 24 h after cultivation. One microgram/milliliter of puromycin (Sigma Co. Ltd) was added to the medium for selection. After seven days, surviving cells on some plates were trypsinized, passaged into 100-mm dishes, cultured for 3-4 days until confluency, and then subjected to the luc assay. On other plates, single colonies surviving after selection with puromycin for 7 days were picked up using yellow tips and transferred to Nunc 24-well plates and cultured for 3–4 days until confluency. Finally, the cells were transferred into a 60-mm dish (Becton Dickinson) and subcultured until required for the luc assay of recombinant clones.

The CTCF over-expression vector (OvCTCF), the anti-sense RNA expression vector (anti-CTCF) and the siRNA expression vector (siCTCF) were introduced into NIH3T3 or 293T cells along with EApL– to examine the possible involvement of CTCF in the enhancer-blocking event. Two hundred nanograms of each vector, 200 ng of EApL– and 40 ng of pRL-TK per well were transfected using the Lipofect Amine Plus reagent into cells in Nunc 24-well plates. Two hundred nanograms of EpL and 40 ng of pRL-TK were also transfected as a positive control. All plasmids used were in a circular form. After 48 h incubation, these cells were subjected to lysis, and luc activity and CTCF expression were investigated.

Luciferase assay

To test enhancer-blocking activity, luc activity was measured using the Dual-Luciferase Assay System (Promega Co.) and a luminometer (CT-9000D; Dia-Iatron, Tokyo, Japan), following the protocols described by the manufacturers. First, the range of linearities for measuring both firefly and Renilla luc activities were determined, and then the appropriate amount of sample was used for measurement of firefly and Renilla luc. For each construct, cells from 12 independent wells were separately subjected to measurement of firefly and Renilla luc activities. In order to compensate for the various degrees of gene transfer efficiency, firefly luc activity was normalized against the Renilla luc activity in each well. Luc activity was finally expressed as RLUs per mg of protein. The resulting RLUs for 12 independent cells were compared statistically by a two-tailed Student's t-test.

To test position-effect protection activity, luc activity was measured using the Luciferase Assay System (Promega Co.) and Dia-Iatron luminometer, following the protocols described by the manufacturers. Firefly luc activity was normalized against protein concentration in each pool, and luc activity was expressed as RLUs per mg of protein. The resulting RLUs for four independent transfectants of each vector were compared statistically by a two-tailed Student's t-test. For individual clones, the luc activity was normalized against copy number (evaluated by real-time PCR as described below) in addition to protein concentration and finally expressed as RLU/mg of protein/copy number. Protein concentrations were determined by the Bedford method (Bio-Rad, Hercules, CA, USA). The resulting RLUs for nine independent clones of each vector were compared statistically by a two-tailed Student's t-test.

Isolation of nuclear proteins and immunoblot analysis

Nuclear proteins were extracted from NIH3T3 and 293T cells by Dignam's method (Dignam et al. 1983). Briefly, cellular pellets (1 × 106 cells) were resuspended and agitated in 0.4 mL of cold lysis buffer [10 mm HEPES-KOH (pH 7.8), 10 mm KCl, 0.1 mm EDTA and 0.01% (v/v) NP-40 containing 1 × protease inhibitor cocktails (Roche Diagnostics, Mannheim, Germany)]. Nuclei were collected by centrifugation (5000 r.p.m. at 4 °C for 5 min) and re-suspended in 100 µL of extraction buffer [50 mm HEPES-KOH (pH 7.8), 420 mm KCl, 0.1 mm EDTA, 5 mm MgCl2 and 1 × protease inhibitor cocktails]. The proteins were extracted by mild shaking at 4 °C for 30 min. After centrifugation (4 °C, 15 000 r.p.m. for 15 min), supernatants were used for measurement of protein concentration prior to gel electrophoresis. Nuclear proteins (30 µg) were treated with β-mercaptoethanol, separated by electrophoresis under reducing conditions on a 10% polyacrylamide-SDS-gel and finally transferred to Immobilon-P nylon membranes (Millipore, Bedford, MA, USA). Blots were blocked with 5% non-fat dry milk in Tris-buffered saline [TBS; 50 mm Tris-HCl (pH 7.4) and 150 mm NaCl], and then incubated at room temperature for 1 h with anti-CTCF antibody (Upstate Co., Lake Placid, NY, USA) diluted 10 000-fold in 5% non-fat dry milk/TBS. After washing with TBS, the blots were incubated at room temperature for 1 h with horseradish peroxidase (HRP)-conjugated anti-rabbit IgG (Amersham Pharmacia Biotech., Little Chalfont, UK) diluted 5000-fold in TBS containing 5% non-fat dry milk. Blots were then re-washed with TBS. The CTCF protein was detected by treatment of the blots with ECL Plus detection kit (Amersham Pharmacia Biotech.) and exposure to an X-ray film for several minutes at room temperature. The signal intensities were measured with the densitometer function of a ChemiDoc XRS image analyzer (Bio-Rad Co., Hercules, CA, USA).

Genomic DNA analysis

Determination of the number of transgene copies integrated into the chromosomes of recombinant cells was performed using the FastStart DNA Master Hybridization Probe kit (Roche Diagnostics) and LightCycler (Roche Diagnostics). The following oligonucleotides were used: 3′-fluorescein isothiocyanate (FITC)-labeled probe (termed Luciferase Flu) (5′-TCTCTAAGGAAGTCGGGGAAGCGGT-3′); 5′- LightCycler-Red-640-N-hydroxy-succinimide ester (LC Red 640)-labeled probe (termed Luciferase LC) (5′-GCCAAGAGGTTCCATCTGCCAGGTATCA-3′); and primer Luciferase R (5′-CTCGGGTGTAATCAGAATAGC-3′). All these oligonucleotides correspond to portions of the luc gene. Genomic DNA (30 ng), isolated from recombinant cells, was PCR-amplified with 60 cycles of denaturation at 95 °C for 10 s, annealing at 60 °C for 10 s and extension at 72 °C for 5 s. The amount of resulting PCR product was monitored using LC Run (Roche Diagnostics), which can store real-time data during the reaction. Calculations of the transgene copy numbers in each sample were made according to the methods reported by Ballester et al. (2004) and Tesson et al. (2002).

Acknowledgements

We would like to thank Drs Shoji Tajima, Tadashi Furusawa and Masao Matsuoka for gift of 293T, NIH3T3 cells and chicken HS4 insulator fragment, respectively. We also thank Dr Daisuke Honma for technical advice and Dr Takashi Nagai for his critical reading of the manuscript. This research was supported by a grant from the Bio-oriented Technology Research Advancement Institution.

Ancillary