Transcriptional Regulation of Porcine Endogenous Retroviruses Released from Porcine and Infected Human Cells by Heterotrimeric Protein Complex NF-Y and Impact of Immunosuppressive Drugs

Abstract

Recent studies revealed a significant promoter activity of porcine endogenous retrovirus (PERV) long terminal repeats (LTRs) in different human and mammalian cell lines, which is mediated by a 39-bp repeat located in the U3 region in different numbers, representing an enhancer (G. Scheef, N. Fischer, U. Krach, and R. R. Tönjes, J. Virol. 75:6933-6940, 2001). A statistical transcription factor analysis revealed putative binding sites for the CCAAT-binding transcription factor NF-Y inside the 39-bp repeat. Specific binding of NF-Y to the repeat sequence was demonstrated by electrophoretic mobility shift assays and supershift assays with specific antibodies directed against the three subunits of NF-Y. To identify further transcription-regulating elements, genetically modified LTRs lacking the repeat box, U3, R, or U5 were investigated. The results indicated a strong inhibitory element in the R region, as the deletion of R caused a significantly increased promoter activity. Since PERV might play a potential role in the application of xenogeneic cell therapy and xenotransplantation techniques, we have investigated whether immunosuppressive drugs that are routinely used in transplantation medicine have an impact on the promoter activity. Neither cyclosporine nor prednisolone had any influence on the promoter strength of the PERV LTRs. By performing a real-time PCR we were able to compare the proviral loads of porcine and infected human cells as well as the amount of released virions, which revealed a direct link between LTR activity and the number of released retroviruses.

Xenogeneic cell therapies and xenotransplantation, i.e., the therapeutic use of living cells, tissues, and organs from animals, show some promise to alleviate the limited supply of allografts in the treatment of human disorders. Pigs are the donor species of choice (15), especially since strategies to overcome rejection have been developed (17, 47, 48, 53). Recently, pigs in which the alpha-1,3-galactosyltransferase locus has been genetically knocked out were generated (10, 24).

The existence of porcine endogenous retroviruses (PERV), which are germ line transmitted (43), and of porcine DNA viruses that can persist without symptoms in their natural host (e.g., herpesviruses) (13) strengthened objections to the clinical use of pig xenografts due to the possible generation of xenozoonosis (15, 54, 55).

PERV display approximately 50 proviral integration sites in the pig genome (1, 43). Virions observed in cell lines are morphologically related to C-type viruses (2). Genetically, three classes of PERV (classes A, B, and C), which differ in their env genes, are known (25). Recent reports demonstrated that PERV which are released from different pig cell lines are able to infect human cells in vitro (32, 43, 58, 59). In a retrospective study, no cross-species transmission of PERV in 160 patients treated with pig tissue was observed (41). On the other hand, in an NOD/SCID mouse model, the diabetic and immunodeficient animals showed infection with and expression of PERV in different tissues after xenotransplantation of porcine islet cells, suggesting that PERV are xenoreactive in vivo (57). Even though a subspecies of miniature swine that failed to produce PERV was recently identified (38), the emergence of hitherto-undescribed PERV genomes (29, 36, 42) shows how difficult a comprehensive risk assessment will be.

As methods for the direct quantification of PERV are poorly developed (44), we have established a real-time PCR to determine the PERV proviral load in either porcine or infected human cells as well as a real-time one-step reverse transcription-PCR (RT-PCR) to quantify the virion release in cell culture supernatants. Recently, we demonstrated that PERV long terminal repeats (LTRs) are characterized by significant promoter activities which are linked to the existence of a repeat box in U3 representing an enhancer (49). To prove the binding of transcription factors to the 39-bp repeats, we performed electrophoretic mobility shift assays (EMSA) and supershift assays. By generation of modified LTRs lacking the repeat box, U3, R, or U5, we investigated whether further elements exist which regulate the transcriptional machinery. In regard to the significance of PERV for xenotransplantation, we investigated whether immunosuppressive drugs such as cyclosporine (CysA) or prednisolone (Pred) have any impact on the promoter activity.

MATERIALS AND METHODS

Quantification of proviral load by real-time PCR.

The genomic DNAs of the porcine cell line PK15, the human cell line 293 productively infected with PERV derived from PK15 cells (293/PERV-PK) (43), and 293 cells that were productively infected with the molecular clone 293-PERV-B(33)/ATG [293/PERV-B(33)] (9) (accession number AJ133816) were used to quantify the proviral load of PERV. After initial infection, both cell lines were passaged according to standard protocols and as described previously (9). Genomic DNA of untreated 293 cells was used as a negative control. After cultivation in appropriate cell culture medium, the cells were washed three times with phosphate-buffered saline (PBS) and harvested by scraping in PBS. The preparation of genomic DNA was subsequently performed by using the QIAamp Blood Kit (Qiagen, Hilden, Germany). The isolated DNA was finally eluted in 100 μl of water (Aqua dest) and stored at 4°C for testing.

Absolute quantification of the proviral load was performed by using an external homologous standard. Therefore, a PERV 812-bp pro-pol fragment was isolated from genomic DNA of cell line 293/PERV-B(33) by PCR with primers PK1 and PK6 (9). The PCR product was subcloned into pGEM-TEasy (Promega, Mannheim, Germany) and subsequently digested with NotI (New England Biolabs, Frankfurt, Germany). After gel electrophoresis, the pro-pol-fragment was isolated (JETsorb; Genomed, Bad Oeynhausen, Germany) and quantified by measurement of optical density at 260 nm. Finally, the copy number was calculated, and a standard curve was generated by using serial 10-fold dilutions in Aqua dest.

Real-time PCR was carried out with the LightCycler-FastStart DNA Master SYBR Green I kit (Roche Diagnostics, Mannheim, Germany) in a 20-μl reaction volume and was monitored after each elongation step by SYBR Green I dye binding to amplified products with the LightCycler instrument (Roche Diagnostics). To obtain specific and efficient amplification, 3 mM MgCl₂ and a 0.3 μM concentration of primers PK1 and PK6 were used. The PCR protocol consisted of an initial denaturation step at 95°C for 10 min and 45 cycles of denaturation (95°C for 10 s), annealing (58°C for 5 s), and extension (72°C for 40 s). For each step, the temperature transition rate was 20°C per s. Melting curve analyses were performed to confirm the PCR product identity and to differentiate specific amplification from nonspecific products by denaturation (95°C for 10 s), annealing (65°C for 10 s), and a slow heating to 95°C (temperature transition rate, 0.1°C/s) combined with a continuous fluorescence measurement at 0.2°C increments. After completion of PCR, the copy number of the target molecules was calculated by plotting fluorescence versus cycle number. The log-linear portion of the amplification was detected by the second-derivate maximum method. As a standard curve we used the linear regression line based on the data of standard crossing points (threshold cycle) versus the logarithm of standard sample concentrations. The mean square error and the slope of the regression line were calculated by using the LightCycler software, version 3.5 (Roche Diagnostics). Finally, the concentrations of PCR products in the samples were extrapolated from the standard curve. The slope of the standard curve was converted to amplification efficiency (E) by using the algorithm E = 10^−1/slope. As an initial retroviral infection blocks further entry of particles targeting the same receptor (19), we exclude a superinfection. Therefore, the number of proviral copies is identical to the number of stably infected cells.

To set the number of detected proviruses in relation to the cell number, we determined the amount of the single-copy housekeeping gene porphobilinogen deaminase (PBGD) (65) (accession number M95623) by performing an absolute quantification with an external homologous standard. To generate a 245-bp-long PBGD gene fragment from genomic DNA of the cell line 293 by PCR, the primers 5′-PBGD-F1 (5′-TGAAGGCTGGCTGCTCATAC-3′) and 3′-PBGD-R (5′-GTTTTCTGCCACCAGTCAACA-3′) were used. The generation of the standard curve and the real-time PCR with the primers 5′-PBGD-F1 and 3′-PBGD-R were carried out as described above with an annealing temperature of 64°C and an elongation time of 16 s. The number of detected PBGD molecules was calculated as described above. The cell amount was finally calculated by division of the PBGD gene copy number by the factor two.

Quantification of virion release by real time one-step RT-PCR.

Cell culture supernatants of the cell lines PK15, 293/PERV-PK, and 293/PERV-B(33) were used to quantify the number of released PERV virions, while supernatant of 293 cells was used as a negative control. Viral RNA was isolated by using the High Pure viral RNA kit (Roche Diagnostics) from 1 ml of cell-free membrane filtered (0.45-μm-pore size; Sartorius, Göttingen, Germany) supernatant of a total volume of 15 ml. RNA was stored at −80°C for testing.

Absolute quantification of the virion number was carried out by using an external homologous standard. Therefore, pGEM-TEasy harboring the PERV 812-bp pro-pol fragment as mentioned above was digested with restriction enzyme NdeI (New England Biolabs). After in vitro transcription by using the MaxiScript kit (AMS Biotechnology, Wiesbaden, Germany), the RNA was quantified by measurement of optical density at 260 nm. After calculation of the copy number, a standard curve was generated from serial 10-fold dilutions in Aqua dest.

Real-time one-step RT-PCR was performed with the LightCycler-RNA Master SYBR Green I kit (Roche Diagnostics) in a 20-μl reaction volume and was monitored after each elongation step by SYBR Green I dye binding to amplified products with the LightCycler instrument (Roche Diagnostics). A 3 mM concentration of Mn(OAc)₂ and a 0.3 μM concentration of primers 5′-PERV-F (5′-TCTCCCCAAGTAAAGCCTGAT-3′) and 3′-PERV-R (5′-ACTAGGATGCCCTGTTGGATTA-3′) were used, generating a 205-bp PCR product. The PCR protocol consisted of an RT step at 61°C for 20 min followed by an initial denaturation step at 95°C for 2 min. Subsequently, 50 cycles of denaturation (95°C for 5 s), annealing (64°C for 5 s), and extension (72°C for 16 s) were performed. For each step, the temperature transition rate was 20°C/s. The melting curve analysis and calculation of virion RNA copy numbers were carried out as described above. The number of virions was finally calculated by division of the virion RNA copy number by the factor two.

Cloning of LTRs with deletions and construction of expression vectors.

Based on the LTR of the molecular clone 293-PERV-B(33) (9, 49), which harbors four copies of a 39-bp repeat, each consisting of an 18-bp subrepeat and a 21-bp subrepeat, plus a solitary 18-bp repeat in U3 (Fig. 1), we generated a variety of LTRs with deletions by PCR (Fig. 1) with the primers indicated. The forward primers bear a KpnI site, while the reverse primers have a BglII site. To generate the LTR 293-PERV-B(33)-ΔF, which lacks the complete U3 region, primers 5′-PERV-ΔA and 3′-PERV-ΔFus I were used, as well as oligonucleotides 5′-PERV-ΔFus I and 3′-PERV-LTR/PBS II (49). Both amplification products were fused by PCR with primers 5′-PERV-ΔA and 3′-PERV-LTR/PBS II. Analogously, the LTR 293-PERV-B(33)-ΔG was generated by using the primers 5′-PERV-LTR II (49) and 3′-PERV-ΔFus II as well as the primers 5′-PERV-ΔFus II and 3′-PERV-LTR/PBS II, with a subsequent fusion of the respective amplification products by using primers 5′-PERV-LTR II and 3′-PERV-LTR/PBS II. The deletion LTR 293-PERV-B(43)-Zero was constructed as described previously (49). A mouse mammary tumor virus (MMTV) LTR was derived from pMAMneo-Blue (Clontech, Heidelberg, Germany) by using primers 5′-MMTV-LTR and 3′-MMTV-LTR, bearing KpnI and BglII sites, respectively. The sequences of the oligonucleotides used for PCR are available upon request.

FIG. 1.

Organizations and structures of the 293-PERV-B(33) LTR and the variants with artificial deletions used for dual luciferase reporter assays. The 293-PERV-B(33) LTR U3 region harbors four copies of a 39-bp repeat, consisting of an 18-bp subrepeat and a 21-bp subrepeat, plus an additional solitary 18-bp subrepeat located downstream of the repeat box. U3 (open box), R (black box), U5 (light gray box), the 39-bp repeats (gray box [18-bp subrepeat] and dark gray box [21-bp subrepeat]), the cap site (black triangle), and the TATA box (open triangle) are indicated. The results of a statistical transcription factor analysis are indicated by black rectangles, which represent the putative transcription factor binding site in U3. All transcription factors shown have a core similarity of 100%. 1, LPOLYA (nt 46 to 53); 2, TH1E47 (nt 71 to 86); 3, FKH1 (nt 92 to 105); 4, SOX5 (nt 120 to 129); 5, GATA1 (nt 151 to 164, 180 to 192, and 249 to 261); 6, ETS2 (nt 159 to 172); 7, AP1FJ (nt 265 to 275); 8, NF-Y (nt 322 to 332, 361 to 371, 400 to 410, 439 to 449, and 478 to 488); 9, CAAT (nt 479 to 490). Numbers indicate the locations of elements in the LTRs. Primers that were used to generate LTRs with artificial deletions are indicated by arrows, showing position and orientation.

All PCR products were subcloned into pGEM-TEasy. KpnI-BglII inserts were further cloned into the luciferase reporter vector pGL3 Basic (Promega). The generation of pGL3 Basic harboring the LTRs of the molecular clones 293-PERV-B(33) and 293-PERV-A(42) (accession number AJ133817) was described previously (49). A pGL3/human interleukin-2 (IL-2) promoter construct (kindly provided by A. Avots, Würzburg, Germany) was used as a control. DNA sequences of both strands of constructs were confirmed by primer walking by the double-stranded dideoxy-chain termination method as described previously (49).

Cell lines and transient transfection.

The cell lines 293, PK15, and 293/PERV-PK were kindly provided by R. Weiss (London, Great Britain). The cell line A3.01 was purchased from the National Institutes of Health AIDS Research and Reference Reagent Program. HeLa cells (ECACC 93021013), Bai/NJ cells (ECACC 9168), and MRC-5 cells (ECACC 84101801) were purchased from the European Collection of Cell Cultures. All cell lines tested negative for mycoplasmas. Plasmids used for transfection were prepared by using the EndoFree system (Qiagen). Transfection of the suspension cell lines A3.01 and Bai/NJ was carried out with DMRIE-C (Life Technologies, Karlsruhe, Germany), while Lipofectamine (Life Technologies) was used for the adherent cells. Transfection was performed in six-well plates by using 1 μg of plasmid DNA cotransfected with 333 pg of the internal standard pRL-CMV (Promega) or 1 μg of pRL-Null (Promega). At 5 h posttransfection, the transfection medium was replaced with appropriate culture medium or medium containing immunosuppressive drugs. CysA (Sigma, Deisenhofen, Germany) and Pred (Sigma) were used at a concentration of 300 ng/ml (10-mg/ml [CysA] and 20-mg/ml [Pred] stock solutions in dimethyl sulfoxide). Stimulation with 10 ng of tetradecanoyl phorbol acetate (TPA) (Sigma) per ml and 0.5 μM ionomycin (Iono) (Sigma), each at 2 μl per well, was carried out at 29 h posttransfection. Altogether, the cells were grown for 43 h after the first replacement of medium.

Dual luciferase reporter assay system.

To investigate the LTR activity, the Dual-Luciferase Reporter Assay System (Promega) was used as described previously (49). After being washed with PBS, the transfected cells were lysed with passive lysis buffer (Promega) and supernatants of the cell lysate were stored at −20°C for testing. The luciferase assay was performed by using 20 μl of the cell lysate incubated with LAR II (Promega) and a subsequent addition of Stop & Glo (Promega) for light-counting times of 10 s each.

Oligonucleotide probes.

Putative transcription factor binding sites of the 293-PERV-B(33) LTR were analyzed by using MatInspector software (Genomatix, Munich, Germany). Since putative binding sites for nuclear factor Y (NF-Y), also designated CCAAT-binding factor (CBF) (28), were found inside of the 39-bp repeat, we used the oligonucleotide NF-Y+ (5′-AAATGATTGGTCCAC-3′ (nucleotides [nt] 319 to 334; nucleotide positions refer to accession number AJ133816, matrix binding sites are underlined, and the core binding site is in italics), bearing the binding site. Complementary single strands were annealed by mixing 200 μg of each single-stranded oligonucleotide in 200 μl of annealing buffer (0.5 M NaHCO₃, 1.65 M NaCl, 1 mM EDTA), heating to 100°C for 2 min, and incubating at 4°C over night. The double-stranded oligonucleotides were purified by using ELUTIP D columns (Schleicher & Schuell, Dassel, Germany) according to the instructions of the manufacturer. Unspecific competition was performed by using an AP1 consensus oligonucleotide (Promega). As a control, the CBF consensus oligonucleotide (Santa Cruz, Heidelberg, Germany) was used for supershift experiments. Double-stranded oligonucleotides (3.5 pmol) were labeled with 1 μl of [γ-³²P]ATP (3,000 Ci/mmol; Amersham Pharmacia Biotech, Freiburg, Germany), using T4 polynucleotide kinase (Promega), at 37°C for 10 min. Labeled oligonucleotides were purified by using Nick G50 columns (Amersham Pharmacia Biotech).

EMSA.

Crude nuclear protein extracts were prepared as described previously (51), using a modified protocol. For EMSA, 5 μg of nuclear proteins was preincubated on ice with 2 μg of poly(dI-dC) (Roche Diagnostics) as an unspecific competitor and 1 μg of bovine serum albumin in band shift buffer (50 mM Tris, 150 mM KCl, 5 mM EDTA, 2.5 mM dithiothreitol, 20% Ficoll) for 40 min. ³²P-labeled oligonucleotides (50,000 cpm) were added in a total volume of 20 μl, incubated on ice for 20 min, and loaded onto 5% native polyacrylamide gels in 0.5× Tris-borate-EDTA buffer. Upon fractionation, gels were dried and exposed for autoradiography.

For supershift EMSA, a set of antibodies directed against the three subunits of NF-Y (CBF-A, CBF-B, and CBF-C; Santa Cruz) were used together with a control antibody (NF-κB p65; Santa Cruz). Five micrograms of nuclear proteins was preincubated on ice with 2 μg of poly(dI-dC) and 1 μg of bovine serum albumin in band shift buffer together with 6 μg of the respective antibodies on ice for 40 min. ³²P-labeled oligonucleotide (50,000 cpm) was added in a total volume of 20 μl and incubated on ice for 20 min. Subsequently, the DNA-protein complexes were separated on 5% native polyacrylamide gels as described above.

Statistical analysis.

Luciferase assays were carried out in at least three separate triplicate experiments for each cell line and vector. The luciferase activity of the LTRs was normalized with regard to the internal control (pRL vectors). Those values were set as 1 and by calculation of the ratio RLU_LTR/RLU_pRL-vector, where RLU is relative light units. For each triplicate experiment, the mean value and standard deviation were calculated.

Since real-time PCR was performed in single measurements with respective external homologous standards, no statistical analysis was performed. Nevertheless, the results were confirmed by at least two further independent experiments.

RESULTS

Proviral loads of porcine and infected human cells.

The numbers of proviral copies in the genomic DNAs of porcine and infected human cells were quantified by real-time PCR amplification of a PERV pro-pol fragment (Table 1). For genomic DNAs of cell lines PK15, 293/PERV-PK, and 293/PERV-B(33), PERV proviruses were detected in different copy numbers ranging from 6.95 × 10⁵ copies/μl [293/PERV-B(33)] (9) to 3.50 × 10⁶ copies/μl (PK15) (Fig. 2A; Table 1). As melting curve analyses revealed only a single peak for all products, the amplification of misprimed products and primer dimers could be excluded (data not shown). In contrast, the increase of fluorescence for DNA of cell line 293 after cycle 32 is linked to the generation of primer dimers due to the lack of PERV sequences, which can be clearly distinguished from the proper products by their significantly lower melting temperature (T_m) (data not shown).

TABLE 1.

Results of real-time PCR experiments^a

Cell line	PERV proviral DNAb	Cell amtc	PERV DNA copiesd	PERV virion RNAe	PERV virions released per infected cellf
Human 293	NPg	1.55 × 107	0.00	NP	0.00
293/PERV-PK	1.16 × 106	5.76 × 107	2.01 × 104	2.46 × 104	1.59 or 3.18i
293/PERV-B(33)	6.95 × 105	7.23 × 107	9.61 × 103	4.18 × 104	4.51
Porcine PK15	3.50 × 106	∼1.34 × 105h	∼26 × 106h	1.30 × 104	∼7.28

^aQuantification of cell numbers by amplifying the single-copy housekeeping gene PBGD and determination of proviral load by amplifying an 812-bp pro-pol fragment of PERV, both from genomic DNA. The amount of virions was detected by real-time one-step RT-PCR from cell culture supernatants by amplifying a 205-bp PERV pro-pol fragment.

^bNumber of PERV copies per microliter of genomic DNA.

^cNumber of PBGD copies/per microliter of genomic DNA divided by the factor 2.

^dPer 10⁶ cells.

^eNumber of PERV RNA copies per microliter of cell culture supernatant.

^fNumber of virions in cell culture supernatant; total volume divided by total number of infected cells (superinfection is excluded).

^gNP, no specific PCR product due to melting curve analysis.

^hCalculation based on the copy number of PERV containing pol in the porcine genome (36) (see Discussion).

ⁱSingle infection by a PERV-A or PERV-B clone (1.59) versus simultaneous PERV-A and PERV-B infection (3.18).

FIG. 2.

Real-time PCR. Quantification of proviral PERV DNA (A) and the human single-copy housekeeping gene PBGD (B) in genomic DNA by real-time PCR and quantification of PERV virion RNA (C) in cell culture supernatant by real-time one-step RT-PCR are shown. The standard curves generated with 10-fold dilutions of the respective PCR products are indicated. The fluorescence is plotted versus cycle number. The genomic DNA and cell culture supernatants were harvested from porcine cell line PK15; human kidney cell line 293; cell line 293/PERV-PK (43), representing 293 cells productively infected with PERV derived from PK15 cells; and cell line 293/PERV-B(33), representing 293 cells productively infected with the molecular clone 293-PERV-B(33)/ATG (9).

PCR amplification of the human single-copy housekeeping gene PBGD was used to quantify the number of cells from which the genomic DNA was isolated. Infected 293 cells, 293/PERV-PK cells, and 293/PERV-B(33) cells harbor copy numbers ranging from 1.15 × 10⁸ copies/μl (293/PERV-PK) to 1.45 × 10⁸ copies/μl [293/PERV-B(33)] (Fig. 2B; Table 1). Based on melting curve analyses, the existence of primer dimers and misprimed products could be excluded. In contrast, for the genomic DNA of the porcine cell line PK15, no specific PBGD product was amplified. The increase of fluorescence after cycle 38 rather is linked to the generation of primer dimers due to the lack of PBGD sequences, as the clearly reduced T_m of the product indicated (data not shown). In regard to the respective PCR amplification efficiencies, the results can be related to each other, revealing 2.01 × 10⁴ PERV DNA copies per 10⁶ cells in the cell line 293/PERV-PK and 9.61 × 10³ PERV DNA copies per 10⁶ cells in the cell line 293/PERV-B(33) (Table 1).

Absolute quantifications of the proviral load and of the PBGD copy number were performed by using serial dilutions of the PERV pro-pol fragment and of the PBGD PCR product as external homologous standards. After real-time PCR was completed, the logarithmic values of fluorescence for each sample and dilution were plotted against cycle numbers (Fig. 2A and B). The direct relationship between the cycle number and the logarithmic concentration of DNA molecules of the serial dilution was demonstrated. The amplification of the PERV standard showed a conserved linear relationship, with the serial dilutions ranging from 10³ to 10⁷ copies/μl resulting in an error of the regression line of 0.056 (data not shown). The efficiency of PCR amplification was controlled by the slope of the standard curve and was calculated to be 1.80, representing a high efficiency rate (data not shown). For the PBGD standard, a conserved linearity for the serial dilutions from 10⁵ to 10⁹ copies/μl was found, resulting in an error revealed by the regression line of 0.052 and a high PCR efficiency rate of 1.94 (data not shown).

Release of virions from porcine and infected human cells.

The quantification of PERV virion release from porcine and infected human cells was performed by using a real-time one-step RT-PCR (Fig. 2C). For the cell culture supernatants, virion RNA copy numbers range from 1.30 × 10⁴ copies/μl (PK15) to 4.18 × 10⁴ copies/μl [293/PERV-B(33)] (Table 1). Amplification of misprimed products and primer dimers could be excluded based on melting curve analyses (data not shown). In contrast, the supernatant of noninfected 293 cells revealed no specific PCR product. The increase of fluorescence after cycle 37 can be linked to the generation of primer dimers, which can be clearly distinguished from the proper products by their significantly lower T_m (data not shown).

Absolute quantification of the virion RNA copy number was performed by using serial dilutions of a PERV pro-pol fragment as external homologous standards. After the completion of real-time PCR, the logarithmic values of fluorescence for each sample and dilution were plotted against cycle numbers (Fig. 2C). The direct relationship between the cycle number and the logarithmic concentration of DNA molecules present in serial dilutions was demonstrated. The amplification of the PERV standard showed a conserved linear relationship, with the serial dilutions ranging from 10³ to 10⁷ copies/μl resulting in an error revealed by linear regression of 0.102 and a high PCR efficiency rate of 1.71 (data not shown).

Promoter activity of modified LTRs.

To assess the roles of different elements of PERV LTRs in transcriptional regulation, we have generated a set of modified LTRs by deletion of structural motifs (Fig. 1), and their promoter activities were investigated by using the dual luciferase reporter assay system in the human cell lines 293, A3.01, and HeLa as well as in the porcine cell line PK15 (Table 2). In all cell lines tested, the deletion of the U3 region upstream of the repeat box [293-PERV-B(33)-ΔA] and the deletion of the U5 region [293-PERV-B(33)-ΔC] did not result in a significant change of activity compared to the unmodified LTR. Similarly, the deletion of the U3 region except the repeat box [293-PERV-B(33)-ΔF] in A3.01 and HeLa cells and the deletion of the R region together with the U5 region [293-PERV-B(33)-ΔD] in HeLa and PK15 cells showed the same results (Table 2). In contrast, the deletion of the repeat box [293-PERV-B(43)-Zero] resulted in a significant decrease of the promoter activity in human cells (up to 11-fold in A3.01 cells), which was potentiated by additional deletion of the U3 region located upstream of the repeat box [293-PERV-B(33)-ΔB] (up to 130-fold in 293 cells) or the deletion of the complete U3 region [293-PERV-B(33)-ΔE] (up to 138-fold in 293 cells). Vice versa, the absence of only the repeat box caused no significantly reduced activity in the porcine cell line PK15, while deletion mutants 293-PERV-B(33)-ΔB and 293-PERV-B(33)-ΔE showed a clearly reduced activity as well (Fig. 1; Table 2). Interestingly, the deletion of the R region [293-PERV-B(33)-ΔG] resulted in a strong increase of promoter activity in all cell lines (up to 4.5-fold in 293 cells), which was found for a combined deletion of the R and U5 regions [293-PERV-B(33)-ΔD] in 293 and A3.01 cells as well (Table 2).

TABLE 2.

Activities of native and modified PERV LTRs^a

Vector	Mean no. of RLUb ± SD in the following cell line:
	Human 293	Human A3.01	Human HeLa	Porcine PK15
pGL3 Basic	1.50 ± 0.13	0.55 ± 0.07	0.08 ± 0.01	0.04 ± 0.00
293-PERV-B(33)	158.53 ± 0.66	111.17 ± 14.03	10.72 ± 0.18	12.17 ± 0.71
293-PERV-B(43)-Zero	31.88 ± 2.06	9.93 ± 0.35	3.68 ± 0.49	12.52 ± 1.72
293-PERV-B(33)-ΔA	174.80 ± 30.84	145.54 ± 19.06	15.87 ± 0.13	14.46 ± 0.97
293-PERV-B(33)-ΔB	1.22 ± 0.21	3.87 ± 0.25	0.26 ± 0.04	0.09 ± 0.01
293-PERV-B(33)-ΔC	135.28 ± 3.34	130.07 ± 54.95	6.40 ± 0.57	13.39 ± 1.09
293-PERV-B(33)-ΔD	218.94 ± 1.66	261.24 ± 26.43	8.99 ± 0.95	13.25 ± 0.63
293-PERV-B(33)-ΔE	1.15 ± 0.19	2.12 ± 0.11	0.28 ± 0.01	0.24 ± 0.05
293-PERV-B(33)-ΔF	32.56 ± 0.71	130.86 ± 7.76	6.21 ± 1.04	3.98 ± 0.62
293-PERV-B(33)-ΔG	707.49 ± 23.96	206.09 ± 15.39	60.43 ± 0.12	21.80 ± 1.80

^aActivities of LTRs cloned into reporter gene luciferase vector pGL3 Basic in relation to internal control pRL-CMV in human and porcine cells.

^bThe luciferase activity of PERV LTRs (Fig. 1) is shown in relation to that of the internal standard pRL-CMV.

Impact of immunosuppressive drugs on PERV LTR activity.

In regard to the potential significance of PERV for xenotransplantation, we have investigated whether the immunosuppressive drugs Pred and CysA have any impact on the activity of the 293-PERV-B(33) LTR and the 293-PERV-A(42) LTR in the cell lines A3.01 and Bai/NJ by performing dual luciferase reporter assays. In the lymphoblastoid T-cell line A3.01, the 293-PERV-B(33) LTR activity was not significantly influenced by Pred, showing a weak, 1.2-fold activation. In contrast, the activity of the MMTV LTR used as a positive control was elevated 121.3-fold by the glucocorticoid hormone (Fig. 3A). In the lymphoblastoid B-cell line, the PERV LTR activity was even decreased by Pred (0.3-fold), while the MMTV LTR activity was increased 9.3-fold (Fig. 3A). Accordingly, the promoter strength of the 293-PERV-A(42) LTR was not significantly influenced by Pred, resulting in a 0.6-fold decrease in A3.01 cells and a 1.5-fold increase in Bai/NJ cells. In contrast, in both cell lines the MMTV LTR activity was significantly elevated (40.3-fold in A3.01 cells and 5.8-fold in Bai/NJ cells) (data not shown).

FIG. 3.

Impact of the immunosuppressive drugs Pred (A) and CysA (B) on the activity of the 293-PERV-B(33) LTR compared to those of an MMTV LTR and human IL-2 promoter. All promoters were cloned into the luciferase reporter gene vector pGL3 Basic, and activities are given with respect to that of the internal control, pRL-CMV, for the lymphoblastoid T-cell line A3.01 and for the lymphoblastoid B-cell line Bai/NJ. Error bars indicate standard deviations.

In accordance with the results for Pred, we have found no significant influence of CysA on the promoter strength of the PERV LTR (Fig. 3B). In A3.01 cells stimulated by CysA, the activity of the 293-PERV-B(33) LTR was increased only 1.3-fold compared to that in unstimulated cells. The human IL-2 promoter was used as positive control. The facultative IL-2 promoter was induced by incubation with TPA and Iono via the Ca²⁺ pathway, leading to a 301.2-fold-increased activity. With additional application of CysA, this activity was significantly decreased (53.4-fold). In Bai/NJ cells, the results were comparable, with an almost unchanged activity of the PERV LTR and an increased activity of the IL-2 promoter after treatment with TPA and Iono, which could be significantly reduced by subsequent application of CysA (Fig. 3B).

Binding of transcription factors to repeat elements.

A statistical transcription factor analysis was performed with MatInspector software. The analysis revealed a number of putative binding sites for transcription factors in the PERV LTR, with a core similarity of 100% and a matrix similarity of at least 95.2% (Fig. 1). The 39-bp repeat box harbors five putative binding sites for NF-Y, binding to the sequence GGACCAATCAT (the core binding site is underlined), on the LTR reverse strand with a 95.6% similarity for the matrix sequence, encompassing the core sequence residing within the 18-bp subrepeat.

To confirm the results of the statistical transcription factor analysis, we performed EMSA by using a 15-mer specific oligonucleotide bearing the putative NF-Y binding site present in the PERV LTR. Crude nuclear extracts of porcine cell line PK15 and of human cell lines 293, HeLa, A3.01, and MRC-5 were used. Except for MRC-5 cells, the PERV LTRs revealed strong promoter activity in those cell lines (48).

Incubation of 293 nuclear proteins with ³²P-labeled oligonucleotide NF-Y+ revealed specific protein complexes (Fig. 4A, lane 2), which were not detectable in absence of the nuclear proteins (Fig. 4A, lane 1). The binding of the nuclear proteins to the oligonucleotide NF-Y+ was competed by a 100-fold molar excess of the unlabeled oligonucleotide NF-Y+ (Fig. 4A, lane 3), while surplus addition (100- and 300-fold molar excesses) of an unlabeled oligonucleotide bearing the binding sequence of the AP1 transcription factor caused a band pattern not distinguishable from that of the noncompeted reaction (Fig. 4A, lanes 4 and 5).

FIG. 4.

DNA-protein interaction. (A) EMSA representing DNA-protein interactions of the nuclear extracts of 293 cells with the putative NF-Y binding site located in the U3 39-bp repeat of the PERV LTR (lane 2). Probe NF-Y+ without nuclear extracts served as negative control (lane 1). To demonstrate binding specificity, the unlabeled probe was added in a 100-fold molar excess to the reaction mixture (lane 3), while an unspecific competitor (AP1 consensus sequence) was added in 100-and 300-fold molar excesses (lanes 4 and 5). (B) Binding of transcription factor NF-Y to the respective predicted sequence was demonstrated by supershift assays. Nuclear proteins were preincubated with specific IgGs which are directed against the C-terminal ends of the three NF-Y subunits A, B, and C or the N-terminal end of subunit C (lanes 3 to 6, respectively). Lanes 1 and 2 are as in panel A. Controls were performed with oligonucleotide NF-Y+ combined with specific CBF-A IgG without nuclear extracts (lane 7) and with a commercial NF-Y oligonucleotide (lane 8). The shifted protein complex is indicated by an arrow.

To prove the binding of transcription factor NF-Y to the respective PERV LTR sequence, we performed supershift assays by using a set of antibodies directed against the three subunits of the heterotrimeric NF-Y (28). Supershifts were generated by antibodies directed against the C-terminal ends of the NF-Y subunit A (Fig. 4B, lane 3), subunit B (Fig. 4B, lane 4), and subunit C (Fig. 4B, lane 5), as well as by an antibody directed against the N-terminal end of NF-Y subunit C (Fig. 4B, lane 6), showing a shifted protein complex. In contrast, supershifts performed with anti-NF-Y immunoglobulin Gs (IgGs) but without nuclear proteins revealed no bands and were identical to those of the negative control (Fig. 4B, lane 7), while the use of an anti-NF-κB antibody revealed the same band pattern as did the uncompeted assay (data not shown). In addition, the use of the CBF consensus oligonucleotide together with anti-NF-Y IgG revealed a shifted protein complex identical to the one described above (Fig. 4B, lane 8). Supershifted protein complexes were also detected by using nuclear extracts from the cell lines HeLa, A3.01, and PK15, while nuclear extracts from the cell line MRC-5 revealed no supershifted proteins (data not shown).

DISCUSSION

PERV virions are released in significant numbers from infected human cells.

The indirect proof and quantification of virions in cell culture supernatants by use of RT assays is well established (9, 43, 58), while a direct quantification by real-time PCR has been scarcely described (44). Therefore, we developed real-time PCR experiments to quantify the PERV proviral DNAs in infected human and porcine cells, as well as the total cell number by amplifying the single-copy housekeeping gene PBGD, both from genomic DNA (Table 1). The virion RNA copy number in cell culture supernatants was quantified by performing a one-step RT-PCR (Table 1). All PCRs based on binding of the dye SYBR Green I to the double-stranded amplicon (60) were performed by using a LightCycler instrument (Roche Diagnostics) (61).

As we used primers specific for the human gene PBGD, PCR amplifications revealed no specific product for the porcine cell line PK15. Nevertheless, cell numbers were calculated, since the screening of a BAC library from the porcine genome (36) revealed 13 complete or partially truncated PERV proviruses harboring the complete pol gene per haploid chromosomal set. Consequently, the number of cells was calculated by dividing the number of PERV copies by the factor 26. As the efficiency of PCR amplification is almost identical for both targets, PERV and PBGD (data not shown), the results can be related to each other, revealing 2.01 × 10⁴ PERV DNA copies per 10⁶ 293/PERV-PK cells and 9.61 × 10³ PERV DNA copies per 10⁶ 293/PERV-B(33) cells. By excluding a superinfection and assuming that 293/PERV-B(33) cells harbor only one copy of an integrated PERV-B provirus, this result corresponds to 0.96% infected cells. For 293/PERV-PK cells, the simultaneous infection with PERV-A and PERV-B is conceivable, resulting in 1.05 or 2.01% infected cells. As immunostaining using nucleocapsid antibodies (36) revealed numerous PERV-positive cells (data not shown), the low proviral load might be due to a weak integration efficiency of PERV. In fact, nonintegrated viral DNA appears to be obligatory to the life cycle of retroviruses (35). Therefore, the proof of positive cells by immunostaining might be linked to an episomal expression of PERV proteins, as described for human immunodeficiency virus (66). Results of serial passaging experiments (49) show that viruses produced by 293 cells are continuously infectious for this cell line, resulting in spreading of PERV in susceptible human cell cultures. Thus, the low proviral load cannot be linked to a lack of infectivity.

PERV RNA copy numbers detected in supernatants of infected human cells range from 2.46 × 10⁴ copies/μl (293/PERV-PK) to 4.18 × 10⁴ copies/μl [293/PERV-B(33)]. The amount of virions released per infected cell was calculated with regard to the diploid retroviral RNA genome (6). Infection with molecular clone 293-PERV-B(33) revealed an elevated virus release of up to 2.8-fold compared to cells which were infected with a mixture of different viruses released from porcine PK15 cells (Table 1). The LTR of molecular clone 293-PERV-B(33), due to its large number of 39-bp repeats, was identified by an increased activity relative to other molecular clones isolated from the cell line PK15 (49). Consistently, we found a direct correlation between promoter strength and virion release as the cell line 293/PERV-B(33) produced a higher number of particles. During the serial passaging of the molecular clone 293/PERV-B(33), LTRs with a reduced number of the 39-bp repeat were generated (49). As promoter activity is decreased for LTRs with fewer repeats (49), it remains to be shown if the amount of virions released from 293 cells in the course of serial passaging is decreased.

Promoter activity of PERV LTRs can be enhanced by deletion of the R region.

On the basis of the different activities of PERV LTRs with artificial deletions (Fig. 1) compared to native LTRs in human cells (293, A3.01, and HeLa) and porcine cells (PK15) (Table 2), we believe that regulatory elements are not located exclusively in the U3 region but also in the R region.

As the majority of the cis-acting elements are located in the U3 region (6, 46), the deletion of that region [293-PERV-B(33)-ΔE] subsequently caused a dramatic decrease of promoter activity as demonstrated in all tested cell lines. As shown previously (49), the repeat box of the 293-PERV-B(33) LTR located in the U3 region acts as a retroviral enhancer. Therefore, the LTR 293-PERV-(43)-Zero, lacking the repeat box, showed a clearly decreased activity in human cell lines. Since the decrease was potentiated by an additional deletion of the U3 elements located upstream of the repeat box [293-PERV-B(33)-ΔB] and since the deletion of all U3 elements except the repeat box [293-PERV-B(33)-ΔF] caused a reduced activity in 293 and PK15 cells, we believe that further cis-acting elements mediating the transcriptional activity, in addition to the already-described enhancer, are located in the U3 region upstream of the repeat box. This hypothesis corresponds to the statistical transcription factor analysis (Fig. 1) that identified putative other binding sites in that region.

A strong increase of the promoter activity was shown after deletion of the R region [293-PERV-B(33)-ΔG] in all cell lines and after the deletion of the R and U5 regions [293-PERV-B(33)-ΔD] in 293 and A3.01 cells (Fig. 1; Table 2). In the proviral context, the R region is defined by the transcriptional start site in the 5′ LTR (46). Deletion or mutation of the R region of complex retroviruses which encode transactivators, such as human immunodeficiency virus type 1 (18), human T-cell leukemia virus type 1 (20), and bovine leukemia virus (12), causes a decreased reporter gene expression (11, 37) or transcriptional activity (18). Similar to the case for PERV, murine leukemia virus (MuLV) does not encode transactivators (8, 9). Nevertheless, deletion of the MuLV R region caused a decrease of the respective LTR activity (8) due to the loss of the R region stem-loop (RSL), which is located at the 5′ end of the R region in many mammalian type C retroviruses (7, 56) and mediates LTR activity by posttranscriptional processing of the transcripts by cytoplasmatic accumulation of unspliced RNA (56). Nucleotides predicted to be involved in base pairing for RSL stem generation are usually conserved among MuLV, feline leukemia virus, and type C primate retroviruses in general (8). For PERV, the predicted RSL stem-loop differs in shape and size (data not shown). Therefore, it might be not as essential for the LTR activity as the RSL, causing no decrease by deletion of the R region. Otherwise, as strong inhibitory elements which might suppress the enhancing function of the stem-loop structure in native LTRs obviously exist, those deletions enclosing the complete R region [293-PERV-B(33)-ΔG] caused increased LTR activities. Such inhibitory elements have been reported for the human T-cell leukemia virus type 1 LTR, where an unconventional interaction of a CREB factor with the R region causes a suppressed activity (63).

Immunosuppressive drugs have no impact on PERV LTR activity.

As PERV might play a potential role for the application of xenotransplantation techniques, we have investigated whether immunosuppressive drugs such as Pred and CysA, both routinely used in transplantation medicine (3), have any impact on the promoter activity.

The treatment with the glucocorticoid hormone Pred had no significant influence on the activities of the 293-PERV-B(33) LTR and the 293-PERV-A(42) LTR in either the lymphoblastoid T-cell line A3.01 or the lymphoblastoid B-cell line Bai/NJ (Fig. 3A and data not shown). As the activity of an MMTV LTR used as positive control was elevated 121.3-fold in A3.01 cells and 9.3-fold in Bai/NJ cells, both cell lines were obviously susceptible to Pred. The induction of MMTV transcription by glucocorticoid hormones, which is mediated by interaction of the hormone receptor with DNA regulatory sequences (glucocorticoid regulatory elements [GRE]), is well known (4). The analysis of GRE in both vertebrate genes and retroviral promoters revealed a conserved GRE sequence, 5′-GGTACANNNTGTTCT-3′ (4), encompassing the hexanucleotide core motif 5′-TGTTCT-3′ that is located in the U3 region of the MMTV LTR (34, 50). For both PERV LTRs, which differ only in the number of the 39-bp repeat located in U3 (49) and a single nucleotide in U5, only the core motif located in the U5 region [nt 687 to 692, where nucleotide positions refer to 293-PERV-B(33)] was detected. In silico analysis revealed a putative hormone-responsive region (HRR) bearing an estrogen-responsive element (ERE) in U5 of the PERV LTR (45). The HRR is located upstream of the core motif 5′-TGTTCT-3′ and corresponds to nucleotide positions 632 to 646 and thereby the single-nucleotide polymorphism of the 293-PERV-A(42) LTR (nt 639). Due to the single-nucleotide polymorphism, the 293-PERV-A(42) LTR harbors two EREs, in contrast to other PERV LTRs which have only one ERE. The 293-PERV-A(42) LTR binds the glucocorticoid receptor (GR), the estrogen receptor, and the progesterone receptor (45). As neither the promoter strength of the 293-PERV-A(42) LTR nor that of the 293-PERV-B(33) LTR was influenced by the glucocorticoid Pred, we conclude that even binding of the GR dimer to the core motif 5′-TGTTCT-3′ or the HRR with respect to their localization downstream of the transcriptional start site has no impact on the regulation of the promoter activity. Regulation occurs either by the classical pathway mediated by GR binding to regulatory elements or by a transcriptional cross talk, resulting in the GR interaction with transcription factors (16). Due to the localization of the HRR in U5, we conclude that the PERV gene expression cannot be influenced by other steroid hormones such as estrogen or progesterone. For human endogenous retrovirus type K, a second GRE core motif, 5′-TGTTAT-3′, located in the U3 region has been described (39, 40). However, no analogue was detected in PERV LTRs.

Similar to the case for Pred, incubation of A3.01 and Bai/NJ cells with CysA had no significant influence on the promoter strength of the PERV LTR (Fig. 3B). In contrast, the activity of the TPA- and Iono-stimulated human IL-2 promoter used as a positive control was decreased by additional application of CysA (53.4-fold in A3.01 cells and 4.5-fold in Bai/NJ cells). The induction of the facultative IL-2 promoter in T cells by TPA and Iono via the Ca²⁺ signaling pathway and its suppression by CysA is well known (21, 52).

As the promoter strength of the 293-PERV-B(33) LTR, which shows strong activity in human cells per se (49), is not further increased by Pred and CysA, we believe that the inevitable treatment of xenotransplant recipients with immunosuppressive drugs might not increase the possible LTR-mediated risks, such as activation of proto-oncogenes (14) or recombination with human endogenous retroviral sequences that might generate pathogenic variants (27, 64).

Transcription of PERV is correlated with binding of the transcription factor NF-Y to repeat elements in U3.

PERV LTRs harboring a repeat box are characterized by an extraordinary promoter strength in mammalian and human cells which is mediated exclusively by the repetitive sequences representing an enhancer (49). As enhancer elements typically bear binding sites for transcriptionally regulatory proteins (5), we performed a statistical transcription factor analysis which revealed putative binding sites for the CCAAT-binding transcription factor NF-Y, also designated CBF (28), inside each 39-bp repeat (Fig. 1). The specific binding of nuclear proteins to the respective LTR sequences was demonstrated by performing EMSA (Fig. 4A). Supershift assays using antibodies directed against the three NF-Y subunits revealed that the banding pattern demonstrated in EMSA was generated by the specific interaction of DNA with NF-Y (Fig. 4B and data not shown) in all cells tested except the human lung cell line MRC-5.

NF-Y is a ubiquitous and highly conserved heterotrimeric transcription factor (28, 30, 31), although differential expression of the NF-Y subunits and cell-specific alterations of the synthesis of subunit A may result in variable NF-Y activity in individual cell lines (33). As the histone-like subunits NF-Y-B and NF-Y-C form a heterodimer, DNA binding is possible exclusively for the heterotrimer, which is generated by an interaction of NF-Y-A with the NF-Y-B-NF-Y-C heterodimer (26, 28, 30). In accordance with the lack of NF-Y-interaction with the PERV LTR in MRC-5 cells, we found only a basal LTR activity in that cell line and an inability of the 39-bp repeat box to activate a heterologous simian virus 40 promoter (49). Since all three NF-Y subunits are required for DNA binding, the lack of subunit A would effectively block DNA interactions. As functional NF-Y appears to interact with other factors, thereby mediating and regulating transcriptional processes (10, 22, 62), those results might be explained by the lack of such interacting factors in MRC-5 cells. Indeed, cell or tissue specificity is a standard property of enhancer sequences due to the existence of particular sets of transcription factors (23).