Highly efficient site-specific transgenesis in cancer cell lines

We modified our basic docking site vector (DockZ) was described by Monetti et al.[39]. This new vector designated as DZL contains a fusion gene comprising puromycin N-acetyl-transferase (PAC) and luciferase linked via the T2A peptide [40], i.e. PAC-T2A-luciferase (Figure 1A). The T2A junction along with part of the C-terminus of PAC and N-terminus of luciferase was synthesized and cloned into pBluescript between Bst API and Cla I sites. This fragment was then subcloned into DockZ using Bst API and Cla I. The C-terminus of luciferase was subcloned from a pBluescript (pBS) luciferase-containing vector after digestion with Bsr GI and Eco NI.

The basic incoming vector, Inc-basic, as well as Inc-CAG-MCS-pA and Inc-CAG-EGFP-pA were described by Monetti et al.[39] and shown in Figure 1B. After digestion of Inc-CAG-MCS-pA with Hpa I, the Gateway RfA cassette was inserted to generate Inc-CAG-Gateway-pA. Inc-TRE-EGFP-pA was constructed by excision of the pCAGGs promoter and insertion of second-generation tetracycline-regulated promoters, TRE [41] by Bam HI and Hind III digestion. Subsequent excision of EGFP by Bsu 36I and Nhe I and cloning of MCS gave rise to Inc-TRE-MCS-pA. The Gateway RfA cassette was inserted after Hpa I digestion in order to generate Inc-TRE-Gateway-pA. The IRES-EGFP digested with Eco RI (polished) and Cla I was inserted into Inc-TRE-MCS-pA after Bsu 36I (polished) and Cla I digestion generating Inc-TRE-MCS-IRES-EGFP-pA. The Gateway RfA cassette was inserted after Hpa I digestion, constructing Inc-TRE-Gateway-IRES-EGFP-pA.

The reverse-transactivator of tetracycline [rtTA; Tet-ON Advance [42] was subcloned from the pTet-On Advanced vector (Clontech) into pBluescript using Eco RI and Bam HI sites. Subsequently, it was subcloned into pcDNA6 digested with Hind III and Sac II deriving pcDNA6-rtTA.

The cancer cell lines Pc-3 (prostate cancer), Du145 (prostate cancer) and SKOV-3 (ovarian cancer) were purchased from ATCC and maintained in RPMI-1640 media containing 10% fetal bovine serum (Invitrogen). Cells were transfected with plasmids using ExGen500 (Fermentas) according to the manufacturer's protocol. Stable clones were selected using 1.0 μg/ml puromycin (Sigma), 7.0 μg/mL blasticidin (Sigma) and 750 μg/mL G418 (Invitrogen). 2.0 μg/mL of doxycycline was used to induce gene expression under the control of the TRE promoter. Cell lines were incubated at 37°C with 5% CO₂.

Stables lines for DockZ and DZL were derived by transfection of 10⁶ cells (Pc-3 for DockZ; DU145 and SKOV-3 for DZL) with 10 μg DNA linearized with Eam 1105I. Forty-eight hours after transfection, cells were trypsinized and replated at limited dilutions. Resistant colonies were selected with puromycin and picked using cloning cylinders after two weeks of selection.

Single-copy stable lines were screened by southern blot. Ten micrograms of genomic DNA was digested with Eco RI overnight, resolved by gel electrophoresis and transferred to Hybond N+ (GE Healthcare). Single copy integrants were detected using a neomycin phosphotransferase (neo^R) probe.

Recombination was performed by cotransfection of 10⁶ cells with 7.5 μg of incoming vector and 2.5 μg of pCAGGS-PhiC31 using ExGen 500 (Fermentas). Forty-eight hours after transfection, resistant colonies were selected with G418. For 96-well format PhiC31-mediated integration, 0.66 μg of incoming vector and 0.33 μg of pCAGGS-PhiC31 were used. Correct integrants were verified with standard PCR using primers recognizing the attL site (CCAGGGCGTGCCCTTGAGTTCTCTC) and neo^R gene (CGATGAATCCAGAAAAGCGGCCATTTTTC) and/or southern using the thymidine kinase probe (Figure 2A).

The Luciferase assay (Promega) was performed according to the manufacturer’s protocol. In brief, cells grown on 24-well plates were washed twice with phosphate-buffered saline (PBS) and incubated in 200 μl lysis buffer for 30 min on ice. Twenty microliters of cell lysate was mixed with 25 μl of luciferase reagent and luciferase activity was read on the luminometer. Luciferase activity was normalized to total protein measured using the BCA kit (Pierce).

The FACSAria™ cell sorter (BD biosciences) was used for single cell analysis of fluorescent expression. Cells grown on 6-well dishes were trypsinized and suspended in PBS. 7-AAD or propidium iodide, diluted 1:100 in PBS, was used for detection of apoptotic cells in the case of GFP and dsRed stably transfected cells, respectively. Ten thousands cells were analyzed per sample.

Soluble Flt1-Fc and free human VEGF were measured using commercially available sandwich ELISAs (R&D systems, catalog # MVR100 and DVE00, respectively) following the manufacturer’s protocol.

For xenograft assays, 5×10⁶ cells were suspended into 150 μl RPMI-1640 media containing 33% of matrigel and injected subcutaneously in both dorsal flanks of SCID mice. Doxycycline was administered using doxycycline-containing pellets (0.625g/kg, Harlan Laboratories). Tumour size was monitored using calipers and the volume was calculated using the formula V= (LxWxH)π/6.

In this study, we modified our recently developed PhiC31 integrase-mediated transgene insertion docking site, DockZ [39] to derive DZL (Figure 1A). Both docking sites allow for introduction of a transgene in their predetermined genomic integration of the desired cell line. Cell with docking site insertion can be identified with puromycin selection. The docking site also carries an inverted promoter-less neomycin phosphotransferase (neo^R) gene at the 5’ end of the attP site. The transgene is inserted with the incoming vector through PhiC31-mediated integration. The incoming vector contains an attB site (Figure 1B) that is subjected to recombination along with the attP site of the docking site. Selection for the correct integrants is achieved through activation of the promoter-less neo^R gene in the docking site. The incoming vector contains a promoter that becomes positioned upstream of the neo^R gene upon correct integration and starts its expression (Figure 1B). In addition, the original system allows for removal of all selection markers through Flp recombinase-mediated deletion [39] and negative selection for removal of the thymidine kinase gene by means of selection for FIAU resistance (Figure 1B). In our cancer cell lines, however, we did not perform this resolution step since our aim was to take advantage of a built-in positive and negative selectable marker system for future drug selection for or against the cancer cells. In the DZL docking site we added a luciferase transgene linked to the puromycin N-acetyl-transferase, in a bicistronic manner through a T2A peptide [40]. Thus the luciferase allows in vivo imaging of tumours. Both proteins were expected to remain functional despite the addition of the T2A peptide resulted 10 amino acids at the carboxy terminus of PAC and a proline at the amino terminus of luciferase genes. To test the docking site and establish the properties of this system, we used a series of incoming vectors (Figure 1C) and three human cancer cell lines (Pc-3, DU145 and SKOV-3).

After establishing stable lines for the docking site transgene, single transgene insertion colonies were identified by Southern blotting using a neo^R probe on genomic DNA cut with the diagnostic Eco R1 restriction enzyme (Figure 1A). For the Pc-3 line, a total of 36 clones were screened, three of which contained a single copy of DockZ (data not shown). For the SKOV-3 and DU145 lines, 8 out of 48 and 12 out of 48 clones, respectively, contained a single copy of DZL (data not shown), showing that the PAC was active. All single-copy containing clones showed the same proliferation rate as the original lines, indicating that the insertion did not have any detrimental effects (data not shown). Furthermore, we performed the luciferase assay, demonstrating that the luciferase gene was also active in those clones (Figure 2A and 2B).

Single-copy docking site clones were screened for PhiC31-mediated integration of an incoming vector encoding for EGFP. Integration was successful for one out of three clones for the Pc-3 line (i.e. Pc-3-A7) and two out of eight for the SKOV-3 line (i.e. SKOV-3-13 and SKOV-3-33). We were not able to achieve successful integration in any of the DU145 DZL lines. Subsequently, these clones were subjected to PhiC31-mediated integration of various incoming vectors and the fidelity of the integration was characterized. A total of 66 colonies (60 for Pc-3-A7 and 6 for SKOV-3-13) were analyzed for correct integration using pairs of PCR primers specific to the attL site and the neo^R gene (Figure 3A). The expected 700 bp fragment was successfully amplified in all screened colonies (Figure 3B), proving that all integration events into the docking sites occurred in a correct manner. Furthermore, 6 Pc-3-A7 and 5 SKOV-3-13 colonies for IncCAG and IncTRE, respectively, were also analyzed by Southern blot using a tk probe in order to demonstrate single integration of the incoming vector (Figure 3A). A single band corresponding to the expected size (9.2Kb and 7.5Kb for Inc-CAG and Inc-TRE, respectively) was detected; indicating that after PhiC31-mediated integration, a single recombination event between the attP site of the docking site and the attB site of the incoming vector occurred (Figure 3C).

We used two different reporter assays in order to examine the expression levels between clones derived from the same parental line. The incoming vector, expressing luciferase under the control of pCAGGs, was introduced into the DockZ docking site containing Pc-3-A7 line and nine clones were isolated. Luciferase activity was measured and normalized to total protein (Figure 4A). One-way ANOVA analysis indicated that the expression levels were similar between isogenic clones (P=0.509) in contrast to the remarkable expression differences observed between SKOV-3 and DU145 lines (Figure 2A and 2B) with different genomic integration for the luciferase transgene.

In order to further examine the expression between isogenic clones, we used an incoming vector encoding EGFP or dsRed under the control of pCAGGs either as a single or bicistronic gene, coupled with an IRES to a truncated form of CD4 (tCD4), missing the intracellular and extracellular domains responsible for interaction with other proteins (I. P. Michael and A. Nagy, unpublished data) (Figure 4B). Using flow cytometry for single cell analysis, we assessed the GFP expression after a period of 2-month continuous culture of isogenic clones. We showed uniform GFP expression in each clone (Figure 4Bi and 4Bii, Additional file 1: Figure S1) as well as a similar mean fluorescent value between isogenic clones derived from Pc-3-A7 and SKOV-3-13 (Table 1 and data not shown). Furthermore, essentially all cells (99%) were positive for GFP (Table 1).

Mean GFP expression was one order of magnitude lower when EGFP was expressed as a bicistronic gene using IRES (Table 1). After establishing xenografts, we were able to show that the difference in expression level between pCAGGs EGFP and pCAGGs tCD4-IRES-EGFP was maintained at the tumor site as well (Figures 4Ci and 4Cii). Similar results were obtained when dsRed was used instead of EGFP (data not shown).

To obtain temporal control on transgene expression, we coupled the IMSI strategy with the doxycycline-inducible system. We created a range of incoming vectors containing second-generation tetracycline responsive elements coupled with the minimal CMV promoter (TRE [41]) (Figure 1C). The second-generation reverse tetracycline transactivator was randomly integrated in lines Pc-3-A7 and SKOV-3-13 using linearized pcDNA6-rtTA-Adv (Figure 5A) and clones resistant to blasticidin were selected. An incoming vector for EGFP under the control of TRE was introduced through PhiC31-IMSI. After addition of doxycycline, strong GFP expression was observed for both Pc-3-A7 and SKOV-3-13 rtTA lines, while no GFP was expressed in the absence of doxycycline (Figure 5B and Additional file 1: Figure S2). Withdrawal of doxycycline resulted in the disappearance of GFP expression, indicating that this system is tightly regulated (Additional file 1: Figure S3).

Then an incoming vector, encoding a secreted protein able to trap Vascular Endothelial Growth Factor (VEGF trap), i.e. Flt1-Fc [43, 44], under the control of the TRE promoter and coupled to EGFP through IRES was stably introduced into line Pc-3-A7-23. Using flow cytometry, we showed that after 72 h of doxycycline administration, uniform expression of GFP was observed in three different sister clones (Figure 5C). Furthermore, using a sensitive ELISA for Flt1-Fc, we demonstrated that its expression was tightly regulated; no protein was detected in the absence of doxycycline while high levels were measured after 72h of dox administration (Figure 5Di). Flt1-Fc can also trap VEGF secreted by Pc-3 cells. By using an ELISA that detects only free VEGF, we were able to show that after 72h of doxycycline treatment and secretion of Flt1-Fc, VEGF was ablated from the supernatant, while no reduction of its levels was observed in the absence of doxycycline (Figure 5Dii). Inducible expression of EGFP alone did not affect the levels of VEGF in the media (Figure 5Dii).

In order to examine if inducibility is maintained in vivo, we established xenografts of Pc-3-A7-23 expressing Flt1-Fc or EGFP under the control of TRE. Since previous studies indicate that Flt1-Fc can slow tumour growth through inhibition of angiogenesis [43], we used this as a functional assay to test the stability and inducibility of our system in vivo. After establishing xenografts for Pc3-A7-23 transfected with Inc-TRE-EGFP (control) and Pc-3-A7-23 transfected with Inc-TRE-Flt1Fc-IRES-EGFP, no difference in tumour growth rate was observed during the first five weeks, indicating that tight regulation is maintained (Figure 5E). After inducing the expression of the transgene by feeding the xenograft bearing animals with doxycycline containing pellets, a rapid reduction in tumour size was observed for Pc-3-A7-23 stably transfected with Flt1-Fc, indicating that inducibility is maintained in the xenograft as well (Figure 5E).

The generation of a series of incoming vectors for derivation of isogenic clones would require both the cloning of GOIs in the desired vector as well as the establishment of stable cell lines. To facilitate and accelerate this procedure, we coupled our system with the Gateway cloning system [45]. We inserted reading frame A (RfA) of the Gateway system into three different incoming vectors, generating Inc-pCAGGs-RfA-pA (Inc-CAP), Inc-TRE-RfA-pA (Inc-TAP) and Inc-TRE-RfA-IRES-EGFP-pA (Inc-TAG) (Figure 1C). This allows for high-throughput insertion of cDNA libraries into the desired incoming vector.

We then examined if it was feasible to achieve PhiC31-mediated integration in a 96-well format. Using the Pc-3-A7 clone, we were able to establish stable clones using three different incoming vectors expressing EGFP, dsRed and luciferase. On average, four G418 resistant colonies were observed per well (Table 2 and Additional file 1: Figure S3). All of the colonies transfected with EGFP or dsRed were positive, while the luciferase colonies were not characterized (data not shown). One-way ANOVA analysis indicated that the number of resistant colonies was independent from the transgene that was introduced (P=0.298). Out of 48 wells, 47 had at least one resistant colony (98%, Table 2). Similar results were observed when an incoming vector for Flt-1Fc-IRES-EGFP under the control of TRE was introduced in line Pc-3-A7-23, with all colonies expressing EGFP upon addition of doxycycline (data not shown).