The β-globin Replicator greatly enhances the potential of S/MAR based episomal vectors for gene transfer into human haematopoietic progenitor cells

Specific human chromosomal elements enhance the performance of episomal gene-transfer vectors. S/MAR-based episomal vector pEPI-eGFP transfects CD34+ haematopoietic cells, but only transiently. To address this issue we reinforced (1) transgene transcription by replacing the CMV promoter driving eGFP with the EF1/HTLV or SFFV promoters to produce vectors pEPI-EF1/HTLV and pEPI-SFFV, respectively; and (2) plasmid replication by inserting the replication-Initiation Region (IR) from the β-globin locus into vector pEPI-SFFV to produce vector pEP-IR. All vectors supported stable transfections in K562 cells. Transfections of CD34+ cells from peripheral blood of healthy donors reached 30% efficiency. Upon evaluation of CD34+/eGFP+ cells in colony-forming cell (CFC) assays, vector pEP-IR showed superior performance after 14 days, by fluorescent microscopy: 100% eGFP+-colonies against 0% for pEPI-eGFP, 56.9% for pEPI-SFFV and 49.8% for pEPI-EF1/HTLV; 50% more plasmid copies per cell and 3-fold eGFP expression compared to the latter two constructs, by quantitative (q)PCR and RT-qPCR, respectively. Importantly, the establishment rate in CFC assays was 15% for pEP-IR against 5.5% for pEPI-SFFV and 5% for pEPI-EF1/HTLV. Vector pEP-IR shows extremely low delivery rate but supports eGFP expression in thalassaemic mouse haematopoietic progenitor cells. The IR is a novel human control element for improved episomal gene transfer into progenitor cells.


Results
Our first objective was to ensure transcription through eGFP-S/MAR in CD34 + cells. In vector pEPI-eGFP, transcription through eGFP-S/MAR is driven by promoter pCMV, a strong promoter in a variety of cells, but with variable activity in haematopoietic progenitor cells. In our experience, pCMV-driven eGFP expression within the pEPI-eGFP episomal vector is very low in the progeny of cord blood CD34 + cells and detectable only by RT-PCR 20 . As transcription of the pCMV-eGFP-S/MAR cassette is a prerequisite for S/MAR function, we hypothesized that vector performance could be improved by the replacement of pCMV by promoters that are active in human haematopoietic progenitor cells, such as the hybrid promoter Elongation Factor 1α /human T-cell leukemia virus (EF1α /HTLV) 31 (Genbank accession number HG530137.1) and the U3 region of the long terminal repeat (LTR) from spleen focus -forming virus (SFFV) 32 (Genbank accession number AJ224005.1).
Our second objective was to reinforce the DNA replication capacity of these plasmids, which may contribute to higher rate of episome establishment. As mentioned previously, in our experience, the inclusion of the β -globin Replicator-(see Suppl Fig. S0) can rescue episome's replication and can result in higher episome retention 27 . We therefore used the β -globin Replicator in a plasmid with altered promoter, in order to investigate its performance as part of S/MAR-pEPI-eGFP based vectors.

Constructs and Stress-Induced Duplex Destabilization (SIDD) analysis. Constructs were made
by introducing modifications in the prototype vector pEPI-eGFP, which served as control in all experiments (Fig. 1a). We used promoter EF1/HTLV which has been shown to support homogeneous, high expression of the reporter gene eGFP 31 , and promoter SFFV which has been shown to enhance the long-term expression of a transgene in primary human haematopoietic cells in vivo 32 . Specifically, we replaced the pCMV in the transcription cassette CMV-eGFP-S/MAR of plasmid pEPI-eGFP, with the EF1/HTLV promoter to construct plasmid pEPI-EF1/HTLV, and with the SFFV promoter to construct plasmid pEPI-SFFV. Subsequently, the β -globin Replicator was added to the pEPI-SFFV plasmid (Fig. 1a), to construct plasmid pEP-IR. As shown below, the performance of vectors pEPI-SFFV and pEPI-EF1/HTLV was found to be comparable and therefore, only one plasmid, the pEPI-SFFV, was used for the insertion and analysis of the role of the IR element.
We performed a Stress Induced Duplex Destabilization (SIDD) profile analysis of plasmids' DNA ( Fig. 1b), in order to determine the distribution of the strand separation potential along the various sites in the DNA of each plasmid constructed 33 . Strand separation is an important contributor to the two major DNA functions, namely transcription and replication. Any change along the plasmid sequence may result in an energetic transition, leading to redistribution of the strand separation potential among the different parts of the circular plasmid DNA 33 , which, in turn, may affect the efficiency of transcription and/or replication of the plasmid. SIDD analysis here served the purpose of investigating the possibility that replacement of pCMV by other promoters, or the insertion of the β -globin Replicator, may introduce drastic changes in the distribution of destabilization potential along the DNA of the respective plasmid, as compared to the control plasmid. Our analysis shows that no changes in the SIDD profiles of backbone and promoter sequences are brought about by the promoter replacements (Fig. 1b). Additionally, the β -globin Replicator site presents a new position of high destabilization potential in the DNA of plasmid pEP-IR and, according to our previous data 27 , this may indicate higher capacity for the vector's retention in the cell nucleus. Column (a) shows control plasmid pEPI-eGFP, constructs pEPI-EF1/HTLV and pEPI-SFFV containing promoter EF1/HTLV and SFFV respectively, in place of the CMV promoter driving eGFP expression and construct pEP-IR with the insertion of the IR 5′ to the SFFV promoter. Column (b) shows the corresponding SIDD profiles of the plasmids used. G(x): destabilization energy (for explanation see Methods). All SIDD profiles show the highly destabilized area of S/MAR; the lightly destabilized site at 4.7 kb falls within the origin of replication of the pUC plasmid backbone. The SIDD profile for construct pEP-IR also shows the strongly destabilized areas of the IR element. Transfections in K562 cells. We performed gene transfer experiments into cells of the haemopoietic cell line K562, in order to investigate the ability of the novel vectors to establish stably transfected cell lines, as episomes. The percentage of fluorescent cells containing one of the plasmids pEPI-eGFP, pEPI-EF1/HTLV, pEPI-SFFV and pEP-IR was determined 48 hours after transfection, by flow cytometry, in, at least, three transfections and was found to be on average, 26.7 ± 7.3%, 27.3 ± 3.1%, 27.8 ± 8% and 44.5 ± 8.5%, respectively (see Suppl Fig. S1). G418 was applied 48 hours post-transfection, and stably transfected cell lines were obtained 7 to 10 days later. Cells in stably transfected cultures were eGFP + for 5 months of continuous culture, with and without selection. Cells growing without selection were tested for resistance to G418 at 3 and 5 months of culture and were found to be resistant to the antibiotic, while untransfected K562 cells died 4-5 days after the addition of G418 (see Suppl Fig. S2). Results show that all three novel vectors are functional in establishing stable transfection in K562 cells, as does the control. Establishment in K562 cells of vectors pEPI-eGFP, pEPI-EF1/HTLV and pEPI-SFFV is in the area of 4% to 5% (data not shown) and that of vector pEP-IR was found (see Methods) to be around 8 ± 2.3%.
The possibility of integration of plasmid pEP-IR into the endogenous genetic material of the K562 cells was investigated by Fluorescent In Situ Hybridization (FISH) (Fig. 2a). No integration events were detected in about one hundred metaphase cells examined. It is deduced that expression of eGFP in this study derives solely from episomal vector pEPI-IR.
Expression of eGFP in transfected K562 cells was estimated at the mRNA level by RT-qPCR (Fig. 2b). Expression of eGFP for each vector was found to be steadily maintained throughout the period of 5 months of culture, and Fig. 2b depicts data at the end of 5 months. Similar results were obtained at the protein level by the use of flow cytometry to estimate the mean fluorescence intensity 3 and 5 months post-transfection (see Suppl Fig. S3). Analysis of transgene expression in K562 cells, transfected with the novel vectors, verified consistent functionality for all vectors.
Transfections in CD34 + cells. Transfections in CD34 + haematopoietic progenitor cells were carried out to study the various aspects of the vectors' performance in these cells.
Transfected CD34 + cells were sorted by fluorescence-activated cell sorting (FACS) and eGFP + cells were placed in colony-forming cell culture (CFC) in order to investigate a) whether the plasmids are retained and to what extend and b) whether they maintain expression during erythroid differentiation of HSCs.
We carried out transfections (4 to 5) with each of the three novel vectors and the control vector in parallel, under the same conditions and with the same pool of cells. We used, on average, 1000 eGFP + /CD34 + FACS sorted cells per dish placed in CFC assay. Fourteen days later, all colonies deriving from CD34 + cells transfected with the control vector pEPI-eGFP were eGFP − in fluorescence microscopy, similarly to previous experiments with cord blood CD34 + cells 19 . In sharp contrast, intense fluorescence was observed in colonies produced from cell cultures carrying vectors pEPI-EF1/HTLV, pEPI-SFFV and pEP-IR (Fig. 4a). The colonies obtained were, on average, 115.8 for pEPI-eGFP, 100.3 for pEPI-EF1/HTLV, 98 for pEPI-SFFV and 154 for pEP-IR ( Table 1). The fluorescent colonies obtained in each case were 0%, 49.8%, 56.9% and 100% respectively. These data reveal two new findings: firstly, the replacement of CMV promoter by either EF1/HTLV or SFFV promoter has a positive effect on eGFP expression in CD34 + cells and the presence of fluorescent colonies represent cells with efficient plasmid retention, leading to establishment; secondly, the presence of IR has a catalytic effect on further increase in plasmid retention, as, virtually, every colony developing from cells transfected with vector pEP-IR is fluorescent. Importantly, the differences in colony numbers referring to vector pEP-IR versus vector pEPI-eGFP are statistically significant (p = 0.0008). This was not documented for any other pair of vectors.
These data allow the evaluation of the rates of establishment of the novel vectors in CD34 + cells, on the basis of reported, similar gene transfer in CHO cells 34 , given by the ratio of colonies appearing after 14 days of culture over transfected, seeded cells. Here, the ratio of fluorescent, stably transformed colonies over eGFP + /CD34 + seeded cells after 14 days of the CFC assay (16 days after transfection) is represented by the percentages 0% for pEPI-eGFP, 5%, for pEPI-EF1/HTLV, 5.5% for pEPI-SFFV and 15% for pEP-IR. These values show that the relative capacity for establishment is comparable between vectors pEPI-EF1/HTLV and pEPI-SFFV as well as between them and the respective values for the control plasmid pEPI-eGFP in permissive conditions, while for vector pEP-IR it is increased 3-fold, compared to that of the other two vectors.
Noticeably, the CFC data obtained here allow for the definition of the establishment rates of the respective vectors in HSCs, presenting a platform for comparison amongst them, but they do not provide a proof for long-term establishment of the vector plasmids in HSCs, for which long-term culture studies are needed.
The data reported in Table 1 also reveal an unexpected finding, namely that, in the case of CFU-GEMM, the fraction of number of (non-fluorescent) colonies over total colonies for pEPI-eGFP (39/463) appear to be significantly higher than those of vectors pEPI-SFFV (0/490) and pEP-IR (0/770) (p = 0.0077). It is possible that the vector pEPI-eGFP supports CFU-GEMM survival or, alternatively, the derived vectors have a detrimental effect on it, but no further experiments were carried out on this issue. The novel vectors support transgene expression and are maintained as free plasmids in the progeny of CD34 + cells. The presence of fluorescent cell colonies under fluorescent microscopy is a strong indication of the expression of the transgene eGFP in CD34 + cells, transfected by each of the three novel vectors. As the determination of eGFP expression is crucial in this work, we confirmed the above observation by Western blot analysis of cell lysates, prepared from fluorescent colonies only, with an anti-eGFP antibody (Fig. 4b). eGFP was clearly detectable in preparations derived from cells transfected with vectors pEPI-SFFV or pEP-IR, in two different experiments, but not in preparations from cells transfected with the control vector pEPI-eGFP.
To test whether these vectors were actually maintained as free plasmids in the transfected cultures, we performed a plasmid rescue assay with DNA isolated from eGFP positive cell colonies (Fig. 4c). Results show that the restriction pattern of the DNA from rescued plasmids is the same as that of input plasmid DNA, used for the transfection of the CD34 + cells, and this demonstrates that vectors exist as free, non-integrated plasmids. Table 1, 50.2% of colonies carrying plasmid pEPI-EF1/HTLV and 43.1% of colonies carrying plasmid pEPI-SFFV were non-fluorescent. To test whether the vector is absent or present but silenced as a result of epigenetic alterations in these cells, we performed PCR for the eGFP gene in DNA from fluorescent and non-fluorescent colonies, both deriving from eGFP positive, CD34 + FACS-sorted cells, transfected with vector pEPI-SFFV (Fig. 4d). Single, non-fluorescent-colony PCR 19 applied to about 10 colonies, was negative for plasmid retention. Extended analysis was carried out with DNA from at least 90 pooled, non-fluorescent colonies in each PCR, so as to maximize chances of plasmid detection in the case retention is a rare event and PCR using equal number of fluorescent colonies were carried out, as control. The data showed that non-fluorescent colonies were negative for the presence of the transgene, while control PCR was positive for the same gene. To determine the detection threshold of this assay, PCRs using DNA from the 90 pooled, fluorescent colonies in dilutions 1/100, 1/150, 1/250, 1/300 and 1/500 were performed. Plasmid presence was documented for dilutions up to 1/250, while results for the remaining dilutions were negative (data not shown). The data document that non-fluorescent colonies do not carry plasmid DNA. Evidently, the non-fluorescent colonies of cells that are devoid of vectors derive from transfected, eGFP + cells, as a result of inefficient mitotic segregation of plasmids and plasmid loss during culture.

Only fluorescent cell colonies deriving from transfected CD34 + cells carry vectors. As shown in
Plasmid copy number per fluorescent cell in the progeny of transfected CD34 + cells. The determination of the plasmid copy number per cell provide a measure of the vector potential and is necessary for estimating the transgene expression per plasmid copy. For each novel vector in this study the plasmid copy number per cell was determined, by qPCR for Absolute Quantification (see Methods) using eGFP-specific primers and input DNA from whole fluorescent colonies. Plasmid copy number per cell was found to be 0.58 for vector pEPI-EF1/HTLV, 0.69 for vector pEPI-SFFV and 1.78 for vector pEP-IR.
The estimations for vectors pEPI-EF1/HTLV and pEPI-SFFV show that the respective fluorescent colonies include a high number of cells that do not contain vector plasmid. This is also documented by fluorescent microscopy, in which fluorescent colonies for these two vectors are mixed, containing fluorescent as well as non-fluorescent cells (Fig. 4a). Taking into account that only fluorescent cells carry plasmids, the plasmid copy number for each vector should be determined per fluorescent cell, provided that one can estimate the percentage of fluorescent cells in mixed colonies. We assumed that, for each vector (pEPI-EF1/HTLV or pEPI-SFFV) -cell culture, the probability of plasmid loss should be the same among mitoses occurring at any time in transfected CD34 + -derived/eGFP + , FACS-sorted cells. In particular, the likelihood of plasmid loss is not expected to be different between mitoses in the original seeded cells that produced fluorescent and non-fluorescent colonies on one hand and mitoses within a growing, mixed colony that produced fluorescent and non-fluorescent cells on the other hand. We can, therefore, consider that within a mixed colony, the percentage of fluorescent cells over total number of cells is equal to the percentage of mixed colonies over total number of colonies (mixed and non-fluorescent) that is given for the respective vector-cell culture. Subsequently, we carried out normalization of the plasmid copy number per cell that was obtained by absolute quantification, taking into account the respective percentage of fluorescent colonies for each vector, namely 49.8% for vector pEPI-EF1/HTLV and 56.9% for vector pEPI-SFFV (Table 1). This gave the values 1.16 plasmid copy number per fluorescent cell for vector pEPI-EF1/HTLV and 1.21 plasmid copy number per fluorescent cell for vector pEPI-SFFV (Fig. 5a).  For vector pEP-IR the plasmid copy number per cell was determined as 1.78 (Fig. 5a). This is also the plasmid copy number per fluorescent cell, as 100% of colonies are fluorescent. These data are in accordance with the data from fluorescent microscopy, in which not only all colonies but all cells within a colony, as far as it can be determined by observation, are fluorescent. Therefore, the presence of IR in the plasmid pEP-IR results in an increase in plasmid copy number per fluorescent cell by 50%.
Relative capacity of novel vectors for eGFP expression in the progeny of transfected CD34 + cells. The eGFP mRNA per plasmid copy was determined for each of the three novel vectors by RT-qPCR, taking into account the plasmid copy number per fluorescent cell (described in the previous section), derived from CD34 + cells that were transfected accordingly. The relative abundance of each mRNA was estimated considering the value of eGFP mRNA per plasmid copy for vector pEPI-EF1/HTLV as one (1) and by calculating the respective values for vectors pEPI-SFFV and pEP-IR, each as a fraction over that of vector pEPI-EF1/HTLV (Fig. 5b). Thus, it is shown that cells carrying either vector pEPI-EF1/HTLV or vector pEPI-SFFV have almost the same capacity for eGFP expression per plasmid copy. Significantly, cells carrying vector pEP-IR are capable of 3-fold higher eGFP expression per plasmid copy, than cells carrying either of the two vectors that do not carry the IR. It is possible that the presence of the IR element has an enhancing effect on nearby eGFP transcription, resulting in triplication of steady-state mRNA level of the transgene.
Noticeably, fluorescence in the cell-colonies carrying vector pEP-IR is brighter and more homogeneous, compared to the colonies derived from cells transfected with either of the other two experimental vectors (Fig. 4a).
Transfer of the vector pEP-IR in thalassaemic mouse, haematopoietic, progenitor cells. We transferred vector pEP-IR into haematopoietic progenitor cells of the Hbb th-3 mouse model which resembles human thalassaemia intermedia 35 for the sole purpose of investigating whether or not it can be delivered and it is capable of supporting trangene expression in these cells.
Whole bone marrow cells from the thalasseamic Hbb th-3 mouse as well as from wild type C57Bl6 mouse were transfected with pEP-IR vector and 24 hours later cells were analyzed by flow cytometry and transfection efficiency was comparable, from 7% to 9%. Cells were stained as described in Methods and results show that viability of the transfected, thalassaemic cells was 18%, while that of the control, untransfected, thalassaemic cells was 68%. eGFP + cells were found only in the transfected, thalassaemic cell cultures and were 2.5% of the live cells. Amongst eGFP + cells, 0.4% were Lin − /Sca + cells (characterizing haematopoietic progenitor cells) (see Suppl Fig. S4). It is shown that vector pEP-IR can achieve some uptake in haematopoietic progenitor cells, derived from a thalassaemic mouse model, but at extremely low rate. Importantly, it can support transgene expression in these cells.

Discussion
In this work, we used episomal vectors pEPI-EF1/HTLV, pEPI-SFFV and pEP-IR and control vector pEPI-eGFP to investigate vector establishment, transgene expression and other aspects of episomal gene transfer in K562 and haematopoietic progenitor cells: 1. Transfections of K562 cells determined the functionality of vectors pEPI-EF1/HTLV, pEPI-SFFV and pEP-IR, as they produced stably transfected cell lines, just like the ones obtained with the control plasmid pEPI-eGFP. This was also expected by the vectors' SIDD profiles, which maintain the same site of strong destabilization potential corresponding to the S/MAR element, despite the promoter exchange and the addition of the IR. The presence of IR element in vector pEP-IR enhances vector's establishment and eGFP expression in K562 cells, but to a lesser degree than in the progeny of CD34 + cells (Figs 2b and 5b). The haematopoietic K562 cell line is a somewhat abnormal cellular system, in that it presents karyotype inconsistency in long-term culture, but nevertheless, a tendency for higher eGFP expression in the presence of IR in K562 is clear. 2. Vector pEP-IR is delivered at extremely low efficiency, but supports transgene expression in mouse haematopoietic progenitor cells. 3. Finally, transfections of CD34 + human haematopoietic progenitor cells revealed that the β -globin Replicator greatly enhances all main properties of this S/MAR-based episomal vector. The specific features of vector pEP-IR with regard to transfections into CD34 + haematopoietic progenitor cells are presented below.

The novel vectors are capable of mediating stable transfection of CD34 + cells. Although all vec-
tors transfect CD34 + cells with satisfactory rates, only the three novel vectors pEPI-EF1/HTLV, pEPI-SFFV and pEP-IR, but not the control vector pEPI-eGFP, give fluorescent colonies in the progeny of CD34 + cells, over the two-week time period of the CFC assay (Fig. 3, Table 1) and actually, the number of fluorescent colonies concerning the vector pEP-IR is statistically significantly higher versus the control vector pEPI-eGFP. The data with control vector pEPI-eGFP are similar to our previous ones, in that the vector is perfectly functional in a number of established cell lines but not in CD34 + cells. It seems that pCMV, driving eGFP transcription through the S/ MAR, is silenced to non-functionality in transfected CD34 + cells, whether these derive from cord blood, as in our previous study, or from adult bone marrow, as in this study, most probably by epigenetic mechanisms. To the contrary, promoters EF1/HTLV and SFFV are competent in supporting sustained transcription through eGFP-S/ MAR in CD34 + cells, as well as in the differentiated cells produced in CFC assay, thus providing the necessary condition for the S/MAR to facilitate nuclear retention of episome(s) in these cells. Evidently, conclusions on pCMV function in K562 cells, as in this setting, cannot be extrapolated to primary cell cultures. Furthermore, the CFC analyses for total CFU-GEMM numbers raised the possibility that replacement of the pEPI-eGFP promoter may interfere with survival of more rudimentary HSCs. While this does not affect in any way the essence of this work, nevertheless, possible confirmation of this finding may require specific interventions e.g. in vector structure or culture conditions, for future application in HSCs.
The β-globin Replicator has a positive effect on vector rate of establishment in CD34 + cells. The estimated ratios of fluorescent, stably transfected colonies to transfected, seeded cells are 5% for pEPI-EF1/HTLV, 5.5% for pEPI-SFFV and 15% for pEP-IR, documenting that the relative capacity of vectors pEPI-EF1/HTLV and pEPI-SFFV for establishment is virtually the same, while for vector pEP-IR it is 3-fold higher.
These estimates can be of clinical importance, particularly in the case of vector pEP-IR, which, in each of a series of seven transfections of CD34 + cells routinely produced 100% fluorescent cells. Practically, vector pEP-IR is expected to persist in all cells that give rise to the differentiated cell-progeny of CD34 + cells, that is in 10% to 20% of the total sorted CD34 + /eGFP + cells.
Furthermore, we have shown that vector pEP-IR is capable of ensuring 50% increase in copy number per fluorescent cell, compared to the parent vector pEPI-SFFV (Fig. 5a).
These data establish that the β -globin Replicator exerts a definite positive effect on vector establishment rate and the number of vector copies in the nucleus of the CD34 + haematopoietic progenitor cells and their differentiated progeny.
The β-globin Replicator has a strong positive effect on eGFP expression in the progeny of CD34 + cells. Vector pEP-IR ensures three times the eGFP mRNA per plasmid copy compared to vectors pEPI-EF1/HTLV and pEPI-SFFV (Fig. 5b). This is a novel information concerning IR function within an episomal vector. It is not obvious how enhancing plasmid DNA replication per se can increase transcription of eGFP, but, considering that the IR is positioned immediately upstream the proximal SFFV promoter, a possible replicative activity at the IR site may bring about a higher degree of DNA destabilization and of open chromatin state at this site, which, in turn, may allow a higher degree of transcriptional activity at the SFFV promoter. So far, the IR has only been used within S/MAR-based vectors. Our data from previous work 27 have shown that the existence of this particular combination of high destabilization potential sites, S/MAR and IR, in an episome derived from plasmid pCEP4, is associated with high efficiency of nuclear retention of the plasmid in Jurcat cells. It is possible that the IR operates in synergy with the S/MAR element, as it has been shown that the coexistence of a MAR element and the β -globin Rep-P replicator, i.e., the AT-rich region, the MAR-like element, and the AG-rich region, is required for extrachromosomal amplification 36 .
In vector pEP-IR the IR is located next to the 5′ side of the SFFV promoter, a position relatively far from the S/MAR, while from our previous work 27 we know that the IR can operate when placed 3′ downstream the poly-A signal of the transgene transcription cassette, a position close to the S/MAR.
On the whole, it is documented that the IR can contribute to plasmid persistence (1) from two different plasmid backbones, pCEP4 and pEPI-1, (2) in established cell lines as well as in cultures of primary cells, and (3) from two different positions within the vector DNA sequence in relation to proximity to S/MAR element.
The IR may exert these effects as a bona fide mammalian replication element, but alternative or complementary mechanisms may contribute to its function. The IR element includes DNA sequences of the β -globin promoter, with binding sites for a number of ubiquitous and tissue-specific transcription factors. Additionally, the IR bears a predicted, consensus binding site for the DNA-binding protein CTCF (CCCTC-binding factor), a well-known, ubiquitously expressed nuclear protein, highly conserved from Drosophila to mouse and man. CTCF is prominently involved in the three-dimensional organization of chromatin and a key player in transcriptional regulation in vertebrates 37 , while it appears to possess strong insulating activity in vitro 38 . Importantly, it has been found to confer episomal stability to herpesvirus genomes in latently infected human T cells 38 . Transfection of murine HSCs. Vector pEP-IR is capable of achieving delivery in murine HSCs, but to an extremely low level. On the other hand, vector pEP-IR is capable of supporting transgene expression in these cells, which may prove to be important in pre-clinical studies.
Towards pEP-IR in preclinical studies. Successful transfections in murine HSCs are of critical importance for pre-clinical, in vivo studies of episomal gene transfer, and non-delivery is a major problem, particularly for plasmids of long DNA sequence, like pEP-IR. Thus, our system is restricted by what is known as a major experimental shortcoming for episomal gene transfer, namely, the lack of sufficient plasmid delivery into cells, in this case murine progenitor cells. A somewhat similar situation is also encountered with transfections into the human CD34 + cells, as the respective transfection efficiency is adequate for current experimentation, but most probably will not be sufficient for in vivo studies involving the transfer of transfected CD34 + cells into appropriate mouse disease model. To address this issue, the development of a non-integrating lenti-viral vector 5 engineered for the delivery of episomes, could offer a valued solution. Additionally, work on the β -globin Replicator and, in general, on the pEP-IR DNA topology, may lead to a more confined sequence for this element's function, and to shorter plasmid DNA, which, on the whole, may be more suitable for gene therapy applications.

Conclusions.
It is shown that promoters SFFV and EF/HTLV are efficient in driving transgene and S/MAR transcription in episomes derived from pEPI-eGFP and that the respective vectors can be maintained in CD34 + cells and their progeny in CFC assays. Additionally, it is documented, for the first time, that the presence of the IR in S/MAR-based vectors for gene transfer in CD34 + human, haematopoietic progenitor cells, has a definitive, enhancing effect on episomal vector performance, namely the rate of establishment, vector copy number per cell Scientific RepoRts | 7:40673 | DOI: 10.1038/srep40673 and transgene expression per plasmid copy. The present study also points to the need of addressing the delivery issue for episomes.
The IR element is thus emerging as one more chromosomal element along with other such elements 39,40 , like the extensively studied S/MAR, the ubiquitous chromatin opening elements (UCOEs) and the cHS4 insulator, which contribute to the formation of autonomously replicating genetic elements. Such autonomous replicons are considered to be suitable for therapeutic and biotechnological applications as well as for the study of the compartmentalization of the nucleus.

Methods
Human and animal cells. Informed consent was obtained from all CD34 + cell donors. CD34 + cells from adults were isolated from leukapheresis products, collected from G-CSF-mobilized healthy donors in accordance with the relevant guidelines of the Bone Marrow Transplantation Unit, and all experimental protocols were approved by the Ethics Committee of the Medical School University of Patras (Nr. 525/07.09.06).
All animal experiments were approved by the Animal Care and Use Committee of the Veterinary Directorate, Prefecture of Thessaloniki, Region of Central Macedonia, Greece and performed in accordance with relevant guidelines and regulation.
Plasmid construction. Plasmids pEPI-EF1/HTLV (6.6 kb) and pEPI-SFFV (6.5 kb) (Fig. 1a) were constructed by deleting the CMV promoter (0.7 kb) from plasmid pEPI-eGFP (6.7 kb) after DNA digestion with the restriction enzymes NheI and PciI in the first case and AgeI and PciI in the second case, resulting in linear vectors with sticky ends. Promoters EF1/HTLV and SFFV were PCR amplified from plasmids pBC-EF1/HTLV and SFFV-pRRL, respectively. The PCR products EF1/HTLV (0.6 kb) and SFFV (0.5 kb) were also restricted with the corresponding enzymes above and were inserted into the respective linearized plasmid vector with the use of DNA Ligation Kit (Takara) according to the manufacturer's instructions. Plasmid pEP-IR (8.2 kb) (Fig. 1a) was constructed by digestion of the pEPI-SFFV (6.5 kb) with the NheI and PciI restriction enzymes, and the linear vector was ligated to the IR (1.6 kb) PCR fragment digested also with NheI and PciI.
Cloning PCR. PCR reactions were performed using 0. 25 u Taq polymerase (Invitrgen), 0.5 pmol dNTPs (Fermentas) and 0.25 pmol of each specific primer. All primers contained the restriction site required for subcloning, underlined in the primer sequence below. Sequence fidelity of clones was confirmed by DNA sequencing (outsourced).
For CD34 + Cell isolation. CD34 + cells were isolated from low-density mononuclear cells (MNC) derived from peripheral blood of initiated healthy donors, using gradient isolation (1.077 g/ml Ficoll-Paque, Biochrom) and immunomagnetic selection, using a combination of the Miltenyi CD34 MultiSort kit (Miltenyi Biotec) and the LS Columns (Miltenyi Biotec) in accordance with the manufacturer's instructions. In all preparations, CD34 + cell purity exceeded 98%, as estimated by flow cytometry (data not shown).
Cell culture and transfection. K562 cells were cultured in DMEM (Gibco) supplemented with 10% FCS, 2 mM L-glutamine, 50 u/ml penicillin and 50 mg/ml streptomycin. All reagents were purchased from Invitrogen.
CD34 + cells were cultured at a concentration of 2 to 5 × 10 5 cells/ml in IMDM with 10% FCS supplemented with the recombinant human (rh) cytokines: 20 ng/ml rh stem cell factor (SCF), 50 ng/ml rh thrombopoietin (TPO) and 50 ng/ml rh Flt-3 ligand (FL) (complete, serum-free medium). Cell culture reagents were purchased from Invitrogen and recombinant cytokines were purchased from BioSource. K562 cells were transfected by electroporation in a GenePulserII electroporator (BioRad). Cell suspension 2 × 10 6 up to a final volume of 500 μ l was mixed with 15 μ g of plasmid DNA, was placed in a 0.4 cm electroporation cuvette (BioRad,) and electroporated at 220 V, 975 mF (t: 23.2-25.1). After electroporation, cells were immediately transferred to 10 ml preheated, complete medium and incubated at 37 °C in 5% CO 2 . Transfected cells were selected 48 hours after transfection by the addition of Geneticin (G418) (Invitrogen) in the medium at an initial concentration of 800 mg/ml. 14-19 days later, stably transfected, selected cells were split in two and cultured further, either with G418 (at a concentration of 400 mg/ml) or without G418.
CD34 + cells were transfected with nucleofection using the Amaxa NucleofectorII (programme U008) with Human CD34 + Cells Nucleofector solution following the company's instructions. Approximately 5 × 10 5 cells were washed and incubated with an anti-CD34-phycoerythrin (PE)-conjugated monoclonal antibody (Pharmingen) for 30 min at 4 °C. Dead cells were excluded by gating based on propidium iodide (PI) (Pharmingen). Transfected cells were sorted with a FACS Vantage (BD) as previously described 19 .
Estimation of vector pEP-IR % establishment in K562 cells. Two parallel transfections were performed and GFP + cells were FACS sorted after 24 h. Cultures of known number of these GFP + cells were set up and flow cytometry was applied for the recording of percentage of surviving eGFP + cells in consecutive days, taking into account the cell death for each day. This percentage was falling steadily until day 7 or 8, when it started increasing again, giving rise to a full, stably transfected culture one week later. The percentage of GFP + cells on days 7 or 8 provides the rate of establishment in K562 cells.
Calculation of stress-induced destabilization profiles. The analysis of the SIDD profile was performed as described previously 27 , using the WebSIDD internet application 41 . By this method two parameters are evaluated: the equilibrium probability (p) of denaturation at single base pair (x) resolution and the energy G(x), also called destabilization energy, which is the incremental free energy required for the denaturation of the base pair at position x. The calculation of G(x) makes possible the detection of fractionally destabilized sites -locations that have a high probability of denaturation-and duplex opening can be driven by relatively small incremental energy. The data set produced is an approximate statistical mechanical equilibrium distribution of a population of identical molecules among its accessible states, using structural and energy parameters that have been verified experimentally.
DNA fluorescence in situ hybridisation (FISH). K562 cells stably transfected with the pEP-IR, were hybridized with pEPI-SFFV (6.5 kb) construct plasmid used as a hybridization probe that detects episomal and/or integrated transgenes, labelled with Fluorescein-12-2′ -deoxy-uridine-5′ -triphosphate (Roche). Plasmid hβ -S/MAR (13.3 kb) plasmid was used as a hybridisation probe labelled with Tetramethyl-Rhodamine-5-dUTP (Roche) that detects the endogenous HBB locus. The procedure is reported in detail previously 30 . Probe labelling was performed by Nick Translation Mix (Roche) following the manufacturer's instructions. A total of 100 metaphase and interphase spreads were analysed.