Efficient in vivo knock-down of estrogen receptor alpha: application of recombinant adenovirus vectors for delivery of short hairpin RNA.

BACKGROUND
Adenovirus (Ad) mediated gene transfer is a well-established tool to transiently express constructs in livers of mice in vivo. In the present study, we determined the specificity and efficiency of Ad vectors expressing short hairpin (sh) RNA constructs to knock-down the estrogen receptor alpha (ERalpha).


RESULTS
Two different shRNA constructs derived from the murine ERalpha coding sequence were designed (shERalpha). In vitro, transfection of three mouse cell lines with pSUPER-shERalpha constructs resulted in up to 80% reduction of endogenous ERalpha activity. A single mismatch in the target sequence eliminated the reduction of ERalpha activity, demonstrating the specificity of shERalpha. The subsequently generated Ad.shERalpha vectors were equally effective in vitro. In vivo, intravenous administration of Ad.shERalpha resulted in 70% reduced hepatic mouse ERalpha mRNA levels. Co-injection of Ad.shERalpha with an Ad vector containing a luciferase (luc) gene driven by an estrogen responsive element (ERE) containing promoter resulted in a significant (90% on day five) down-regulation of hepatic luciferase activity, as determined by non-invasive optical imaging. Down-regulation was sustained up to day seven post-injection.


CONCLUSION
Ad mediated transfer of shERalpha expression constructs results in efficient and specific knockdown of endogenous ERalpha transcription both in vitro and in vivo.


Background
Estrogen exerts various biological effects in numerous organs throughout the body and has been implicated in the pathophysiology of a number of diseases including breast cancer, osteoporosis and cardiovascular disorders. Most of the estrogenic effects are mediated via the two known estrogen receptors, ERα and ERβ. These estrogen receptors are ligand-dependent transcription factors that can modulate gene transcription directly but also indirectly. Thus far, there is a relative paucity in the description of the role of estrogen and estrogen receptors in specific organs. Most studies have been performed using Affectivity and specificity of pSUPER mediated expression of shERα in mouse cell lines Figure 1 Affectivity and specificity of pSUPER mediated expression of shERα in mouse cell lines. (A+C) The indicated mouse cell lines were co-transfected with, ERE-Luc, CMV-LacZ, and pSUPER-empty, pSUPER-shERα_1395, pSUPER-shERα_1103, or pSUPER-shERα_tandem. Subsequently, the cells were treated 24 hours with 10 -9 M 17-β-estradiol. Luciferase activity was measured 48 hours after transfection and after correction for LacZ expression, represented as the mean (n = 3) ± SD relative to the transfection with pSUPER-empty. (A) Endogenous mouse ERα mediated transcription in EOMA, H5V and MXT cells after introducing pSUPER +/-shERα. (B) The 19-nt target-recognition sequence of ERα_1395 contains one mismatch with human ERα and five mismatches with the mouse ERβ sequence. (C) ERα mediated transcription in EOMA cells after over expression of either mouse ERα-or human ERα-expression vectors in presence of pSUPER empty or pSUPER shERα_1395 (D) Western blot analysis of H5V cells co-transfected with pCMV-mERα and pSUPER-empty or pSUPER-shERα_1395. The lysates were analysed by immunoblotting (insert-photo) with anti-mouse ERα and anti-p38. The intensity of the bands was quantified and normalized to cells transfected with pSUPER-empty. The relative ERα protein levels are presented (bar-diagram) as mean (n = 3) +/-SD.

A B C D
non-tissue specific manipulation of ER signaling such as complete knockouts either via deletion of the estrogen receptor or via deletion of estrogen production by ovariectomy. The availability of tools to specifically address the role of ER signaling in individual tissues would thus fill a void.
Short synthetic duplexes of 21 nucleotides long RNA molecules can specifically inhibit gene expression in mammalian cells [1]. Because of their efficacy and specificity, siRNA molecules provide a powerful tool to dissect gene function. To expand the applicability of the siRNA approach, Brummelkamp and co-workers [2] have introduced vector-based siRNA expression systems. By directing the synthesis of shRNA via the polymerase-III H1 RNA gene promoter, effective siRNA molecules are formed intracellular after transfection of shRNA expression constructs. To further expand the applicability of the siRNA approach, recombinant retro-and adenoviral based vectors have been designed [3,4]. Of these, adenoviral vectors offer the advantage of highly efficient infection of a broad range of cells, independent of active cell division. Moreover, high titers can be obtained and intravenous injection results in efficient transduction of the liver.
The present study was designed to generate tools to address the role of ERα in a tissue-and time-specific manner. To this end, we have developed recombinant Ad vectors encoding shRNA's directed against mouse ERα (Ad.shERα). Introduction of shERα, either by transfection or by Ad mediated gene transfer into different murine cell lines, led to efficient sequence specific repression of ER mediated transcription. Furthermore, intravenously administration of Ad.shERα resulted in efficient reduction of hepatic ERα mRNA levels (P < 0.005) and ERα functionality.

Efficient and specific knock-down of endogenous mERα in vitro: Transfection with pSUPER-shERα constructs
Three pSUPER-derived vectors [2] designed to drive expression shERα sequences were constructed. Two vectors contained sh sequences derived from the boundary of the DNA binding domain and the hinge region (shERα_1103), or from the ligand binding domain (shERα_1395) of mERα, respectively. A third expression vector contained both the shERα_1103 and shERα_1395 expression cassettes in series (shERα_tandem).
The efficiency of the shERα constructs for reducing endogenous ERα activity in vitro was determined using a luciferase reporter assay. For this purpose, the pSUPER-shERα_1395, pSUPER-shERα_1103, or pSUPER-shERα_tandem were transfected together with a reporter plasmid carrying a trimer of ERE plus TATA box upstream of luciferase (pERE-Luc) into endothelial cell lines (EOMA and H5V) and in mouse breast cancer cells (MXT). As shown in Figure 1A, upon transfection with shERα_1395, relative luciferase activity in lysates of all three cell lines was reduced by 70-80%. A similar result was obtained with shERα_1103 in EOMA's. In addition, the shERα_tandem expression construct proved to be more efficient than either of single shERα contructs alone in the EOMA cells, adding some 15% to the 70% reduction observed with shERα_1395 (Fig. 1A).
To evaluate the specificity of the shERα construct, shERα_1395 was introduced into EOMA cells overexpressing either mouse ERα or human ERα. The ERα_1395 target sequence contains only a single mismatch with the human ERα (Fig. 1B). Significant suppression of ERα mediated transcription was only observed in lysates of cells that were transfected with mouse ERα but not with human ERα (Fig. 1C). Thus, the observed effects of shERα_1395 are specific for mouse ERα. Moreover, changing a single nucleotide in shERα_1395 completely abolished the silencing effect (data not shown). By western blotting, the effect of shERα on ERα protein expression was studied (Fig. 1D). In the presence of shERα_1395, ERα protein levels were reduced to 33% as compared to control transfected cells. This reduction correlated well with our findings in the luciferase reporter assay. Thus, the observed inhibition of luciferase activity upon treatment with shERα_1395 or shERα_1103 is caused by reduced accumulation of mERα protein. All together, the shERα_1395 and shERα_1103 expression vectors are effective and specific in repression of murine ERα expression.

Knock-down of hepatic ERα expression in vivo: using Ad.shERα vectors
To repress ERα activity in vivo, Ad vectors expressing either shERα_1395 (Ad.shERα_1395), shERα_1103 (Ad.shERα_1103) or both (Ad.shERα_tandem) were generated ( Fig. 2A). The H1 RNA promoter plus shERα expression cassettes were sub-cloned from the corresponding pSUPER into pAdTrack [5], which is engineered to co-express GFP enabling the tracking of infected cells. In addition, we constructed a control AdTrack plasmid, carrying only the H1 RNA promoter, which allowed for the generation of Ad.Empty. Prior to the evaluation of recombinant Ad vectors in vivo, we tested the functionality of the vectors in vitro. EOMA and MXT cells were transfected with pERE-luc, and subsequently infected with Ad.Empty or the Ad.shERα vectors. Fluorescence analysis indicated a near 100% infection percentage. The luciferase experiments (Fig. 2B) were comparable to those obtained with transfection of the pSUPER constructs ( Fig. 1A): both Ad.shERα vectors repressed luciferase reporter activity up ER-mediated luciferase activity after Ad-mediated transfer of shERα in vitro Figure 2 ER-mediated luciferase activity after Ad-mediated transfer of shERα in vitro. (A) Schematic representation of the recombinant Ad vectors, carrying GFP and shERα expression cassettes that were used in this study. (B) EOMA and MXT cells were co-transfected with pERE-Luc and pCMV-LacZ and than infected either with Ad.Empty, Ad.shERα_1395, or Ad.shERα_1103. 10 -9 M Estrogen was administrated for 24 hours. Luciferase activity was measured 48 hours after infection. Data represented as mean ± SD relative to infection with Ad.Empty.

A B
to 90%. Thus, Ad.shERα vectors were found to be fully functional with respect to repression of mERα activity.
We then proceeded with the application of our vectors in vivo. The Ad vectors (Ad.Empty, Ad.shERα_1395, or Ad.shERα_tandem) were injected in the tail vain of C57Bl/6 mice. This allowed examination of inhibition by shERα of endogenous hepatic mERα. Four days postinjection, animals were sacrificed, and the livers were studied for GFP expression and ERα mRNA level. Similar GFP expression patterns were observed in all groups, indicating equally efficient transduction (data not shown). ERα mRNA levels were studied by real time PCR analysis (Fig. 3). Administration of Ad.shERα_1395 reduced ERα mRNA levels 70%, whereas hepatic expression of shERα_tandem resulted in an 85% reduction.
Subsequently, we sought to examine the extent of shERαmediated repression of hepatic mERα transcription activity. For this purpose, we constructed an Ad vector carrying the estrogen responsive luciferase reporter gene (Ad.ERE-Luc). First the estrogen-responsiveness of this vector was determined in vivo (Fig. 4A). Five days post-injection of 8 × 10 8 pfu Ad.ERE-Luc, the mice were injected s.c with increasing concentrations of estrogen, ranging from 0 to 50 µg/kg. As shown in Fig 6, 6 hours post-injection, estrogen induced hepatic luciferase activity in a dose-dependent manner. Maximal stimulation was reached after applying 25 µg/kg estrogen. Then, we determined to what extend Ad.shERα down-regulates the transcriptional activity of hepatic ERα. Ad.shERα together with Ad.ERE-Luc reporter vector was administrated intravenously to C57BI/ 6 mice. Luciferase expression was detected by a CCD camera in living mice. Without estrogen treatment, all mice exhibited the same basal expression of the reporter construct (data not shown). Administration of 5 µg/kg estrogen, three and seven days after transduction with Ad.shERα_1103, resulted in a significant repression of hepatic ERα-mediated luciferase activity (Fig. 4B). These data were confirmed by measuring luciferase activity in liver extracts of mice that received estrogen (5 µg/kg, sc) five days post-injection with Ad.ERE-Luc plus Ad.Empty or Ad.shERα_1103 (Fig. 4C).
We conclude that Ad-mediated introduction of shERα in vivo results in an almost complete repression of hepatic mERα mRNA levels, as well as mERα-mediated transcription activity.

Discussion
In this paper, we demonstrate that efficient silencing of mouse ERα can be achieved in vitro as well as in vivo by use of Ad-mediated transfer of shRNA molecules that target the ERα mRNA. Two independent shERα plasmid and Ad vector expression constructs were generated and shown to be effective in repressing endogenous ERα activity up to 80% in several different cell lines and in vivo (Fig. 1A, 2B  and 3). In addition, a construct was made expressing both shERα sequences simultaneously. In vitro as well as in vivo, this construct was shown to be more effective (Fig. 1A and 3) than either of the two shERα constructs alone. Noninvasive optical imaging of living mice, allowed us to quantify shERα activity in vivo. Significant reduction of mouse ERα transcription levels were observed up to seven days post-transduction (Fig. 4B).
Thus far, bystander effects caused by shRNA constructs targeted to an unrelated gene have not been reported, and the specificity of the shERα_1395 construct was verified by the observation that human ERα, which has a single mismatch with the murine ERα target sequence, is not down-regulated (Fig. 1C). The number of mismatches Hepatic ERα mRNA levels after Ad-mediated transfer of shERα in vivo Figure 3 Hepatic ERα mRNA levels after Ad-mediated transfer of shERα in vivo. Male C57Bl/6 mice (n = 5) were injected with 4 × 10 9 pfu Ad.Empty, Ad.shERα_1395 or Ad.shERα_tandem. Livers were harvested Four days after Ad. administration and subjected to taqman analysis. The cyclophillin gene was used as internal standard. Data represented as mean ± SD.
Hepatic ERα activity Figure 4 Hepatic ERα activity. (A) Male wt/ C57Bl/6 mice, pre injected with 5 × 10 8 pfu Ad.LacZ, were injected with 8 × 10 8 pfu Ad.ERELuc. Five days later, the recipients were treated for 6 hours with increasing amounts of estrogen (0-50 µg/kg, s.c). Then, the mice were sacrificed, and the livers were processed for luciferase assays. Luciferase activity is expressed as relative luciferase units (RLU) per mg total liver protein. (B) Male C57Bl/6 mice (n = 5) were injected with Ad.ERE-Luc (5 × 10 8 pfu) plus Ad.Empty or Ad.shERα_1103 (3 × 10 9 pfu). Three or seven days post-infection, the mice were injected with 5 µg/kg estrogen. The (inset) photo shows the result of optical imaging of the bioluminescence at day three, the bar-diagram is a quantitative representation of hepatic luciferase activity at day three or day seven. (C) Male C57Bl/6 mice (n = 5) were co-injected with Ad.ERELuc (5 × 10 8 pfu) + Ad.Empty or Ad.shERα_1103 (3 × 10 9 pfu). Five days later, the mice received 0 or 5 µg/kg estrogen. After 6 hours, the animals were sacrificed, and hepatic luciferase activity was determined. Luciferase activity is expressed as relative luciferase units (RLU) per mg total liver protein. Data represented as mean ± SD. with the murine ERβ sequence totals five, making it unlikely that the shERα_1395 construct would affect expression of ERβ. Similarly, the shERα_1103 construct has three mismatches with the human ERα and nine mismatches with murine ERβ, making it unlikely that the shERα_1103 construct would interfere with either of them. A single mismatch in the shERα_1395 sequence did render the construct ineffective in down-regulating murine ERα (data not shown). Thus, the two independent shERα constructs described here are exquisitely suited to demonstrate that a specific effect is mediated by downregulation of ERα expression and not by down-regulation of a related sequence.
A key challenge in the application of an shRNA based approach is efficient delivery of the shRNA constructs to target cells in vitro and in vivo. For application of shRNA in vivo, the sh oligopair, driven by H1 RNA polymerase [2] or U6 promoter [6], can be cloned in viral vectors. Here, the Ad vector was chosen as delivery vector, because of the relative ease of generation and amplification. Moreover, the natural tropism of Ad vectors for the liver enables the rapid analysis of the hepatic knock-down phenotype.
Since Ad vectors predominantly infect the parenchymal cells [7,8], it is important to note that most abundant hepatic ERα expression was detected in parenchymal cells while ERα expression was barely detected in hepatic endothelial cells or kupffer cells (data not shown). This supported the rationale for application of shERα Ad vectors in vivo. Another interesting observation was that upon administration of 4 × 10 9 pfu Ad.shERα, an 85% reduction of ERα mRNA levels was obtained (Fig. 5), whereas co-injection of 3 × 10 9 pfu Ad.shERα_1103 with 5 × 10 8 pfu Ad.ERE-Luc resulted in an almost complete absence of luciferase activity (Fig. 4C). The ratio of Ad.ERE-Luc versus Ad.shERα_1103 (1:6) should ensure that all cells that were transduced by Ad.ERE-Luc also received Ad.shERα_1103. Thus, the remainder of ERα expression determined by real-time PCR likely reflects ERα expression in non-parenchymal and non-infected cells.
Thus far, relative few reports describe the application of Ad vectors as delivery system for RNAi in vitro [9][10][11][12]. Similarly, relative few studies on effective RNA interference in vivo using Ad mediated gene transfer have been reported [13][14][15]. One potential explanation for this relative paucity in the application of Ad mediated gene transfer for shRNA expression constructs could lie in the recent observations of Lu and Cullen [16], that VA1 non-coding RNA, expressed by wild type adenovirus is a potent inhibitor of RNA interference. However, replication-incompetent adenovirus vectors such as the vectors used in our study have been reported to express low levels of VA1 [16]. Moreover, in our hands the effect of the pSUPER shRNA construct shERα_1935 on reduction of ERα activity in vitro was not affected by super-infection with the Ad.empty vector (data not shown). Thus, the Ad vectors applied in this study seem to have no or a minor inhibitory effect on the RNAi response in vitro and in vivo. Whether this effect is also insert specific and/or depends on the particular target gene remains to be investigated.
The strongest evidence for efficient reduction of endogenous hepatic ERα RNA levels in vivo was obtained by coinjection of Ad.ERE-luc and advanced non-invasive in vivo optical imaging. Administration of Ad.ERE-luc led to readily detectable levels of luciferase activity from day 3 up to day 7 and disappeared at day 10 (data not shown).
In agreement with this, the Ad.shERa mediated knockdown effect was present at day three, five, and seven postinjection ( Fig 4B). This represents a 4 to 5-day window of expression to determine the phenotypic effects of hepatic shRNA-mediated reduction of mRNA levels.

Conclusion
We have shown significant repression of hepatic ERα activity in mice utilizing Ad.shERα vectors. In addition, using advanced non-invasive optical imaging technology, the dynamics of the knock-down effect in vivo have been demonstrated. Thus, our data confirm that application of shRNA represents a powerful tool for targeted gene silencing. We conclude that Ad-mediated delivery of shERα constructs represents an elegant tool to gain more insight in the role of the hepatic ERα.
The H1prom plus or minus shERα were cloned from the pSUPER into the promoter less pAdTrack vector [5] by use of XbaI and XhoI restriction sites. The Ad.shERα_tandem construct was generated by ligation of H1prom-shERα_1103 between the NotI and KpnI sites of pTrack-H1prom-shERα_1395.

Luciferase reporter assays
Transient transfections were performed in triplicate in 12wells plates (1.10 5 cells per well) using Lipofectamine (Invitrogen). The effect of shERα on ERα mediated transcription regulation was determined by co-transfecting the cells with 100 ng of reporter construct (ERE) 3 TATA-LUC and 500 ng expression vector pSUPER-shERα or an empty pSUPER control vector together with 100 ng pCMV-LacZ. After 24 hours, the cells were stimulated with complete DMEM containing 10 -9 M Estrogen for an additional 24 hours. The cells were lysed with reporter lyses buffer (Promega) and after centrifugation of 2 min, supernatant was used for determining β-galactosidase normalized luciferase activity by adding 100 µl luciferyl-CoA (Promega) to 20 µl of cell extract in a monolight luminometer (BD Biosciences). β-galactosidase was measured in a 96-well microtiter plate using the β-Galactosidase Enzyme Assay System in reporter lyses buffer (Promega). Absorbance at 450 nm was determined in a microplate reader. Luciferase activities were normalized for transfection efficiency with the β-galactosidase activity and expressed as a percentage relative to expression levels induced by endogenous estrogen receptor (ER). Expression of endogenous ERα in those cells was verified by real time PCR.

Adenoviral vectors
Recombinant adenoviral plasmids were generated by homologous recombination of pAdtrack or pShuttle vectors with pAdEasy1 in BJ5183 cells as described previously [5]. Correct clones were propagated in DH5α cells (Life Technologies). For the generation of the Ad.shERα vectors, Ad.Empty and Ad.ERE-Luc, PERC6 cells were transfected with 4 µg Pac-I-linearized adenoviral construct using LipofectAMINE PLUS (Life Technologies). After 16 hours transfection medium was replaced by growth medium. Transfected cells were harvested at day seven post-transfection and after three freeze-thaw cycles the lysate was used for large-scale production of Ad vectors in PERC6 cells. Virus was purified by double CsCl centrifugation and subsequently dialysed as described previously [21]. Infection cells 24 hours before transfection, 1.10 5 cells per well were seeded into12 wells-plate. Cells were transiently transfected by use of lipofectamine with a total of 450 ng of DNA per well (150 ng of reporter plasmid (ERE) 3 TATA-LUC and 300 ng pCMV-LacZ). After 4 hours cells were infected with either Ad.shERα or control Ad.Empty (MOI 5.000). Additionally, they received 10 -9 M estrogen for 24 hours. Cells were lysed in 300 µl reporter lyses buffer. βgalactosidase and luciferase activity was determined as described above.

Animals and Ad Injection
The Ethics Committee for Animal Experiments of the Leiden University approved all animal work and the experimental protocols complied with the national guidelines for use of experimental animals. Male C57Bl/6JIco (Charles river, The Netherlands) were given a standard m diet Chow (Hope Farms, Woerden, NL) and housed under standard conditions in conventional cages with free access to water and food.
Recombinant Ad, with a maximum of 4 × 10 9 pfu in 200 µl of PBS, were administered by injection into the tail vein of mice at the age of 14 weeks. Within five days post-infusion, mice were sacrificed; liver pieces were removed and immediately deep-frozen in liquid nitrogen and stored at -80°C.

Pharmacological treatment
The experiment was carried out in 12-wks old C57BL/6 male mice. To prevent sequestration of low doses of Ad.ERE-Luc by liver Kupffer cells, mice were pre-injected with Ad.LacZ (5 × 10 8 pfu) 4 hours before administration of 8 × 10 8 pfu Ad.ERE-Luc. 17β-estradiol (Sigma, E8875) was dissolved in sesame oil (Sigma). In the dose-response experiment, five days post-injection of Ad.ERE-Luc, 0, 5, 25 and 50 µg/kg 17β-estradiol was injected for 6 hours. Then liver pieces were rapidly dissected and immediately deep-frozen in liquid nitrogen and stored at -80°C for further analysis.

Bioluminescent reporter imaging
The experiment was carried out in 12 wks old C57BL/6 male mice co-injected with Ad.ERELuc (5 × 10 8 pfu) plus either Ad.Empty or Ad.shERα (3 × 10 9 pfu). Bioluminescent signals (BLS) were performed at time 0 and at several days after 6 and 24 hours s.c injections of 5 µg/kg 17 β-estradiol with the Xenogen IVIS imaging system (IVIS 100). The living mice were intraperitoneal (ip) injected with the luciferase substrate, luciferin, at a dose of 150 mg/kg body weight approximately 5 minutes before imaging. The mice were anaesthetized with isoflurane/oxygen and placed on the imaging stage. Total photon emission of each animal was acquired for 1 minute. Captured images were then quantified by using the Living Image software (Xenogen Corp, Almeda, CA) and the IGOR software (WaveMetrics Corp, Lake Oswego, OR). BLS from the region of interest (ROI) was expressed using the pseudo colour scale (Red most intense and Blue least intense luminescence) and the data were presented as the cumulative photon counts collected within each ROI. Because layers of tissue may limit photon emission from inner organs, the experiment was repeated. Of these mice the livers were rapidly dissected at day 5, 6 hours after 17β-estradiol administration, verifying the results from the bioluminescent reporter imaging by determining the luciferase activity in liver lysates

Luciferase enzymatic assay
The liver extracts were prepared by homogenisation with the minibead beater in reporter lyses buffer (Promega), two cycles of freeze-thawing and 2 min. of centrifugation at maximum speed. Supernatants were used for determining protein-normalized luciferase activity by adding 100 µl luciferyl-CoA (Promega) to 20 µl of liver extract in a monolight luminometer (BD Biosciences). Protein content was measured in a 96-well microtiter plate using the BCA protein assay kit (Pierce). Absorbance at 562 nm was determined in a microplate reader.