In Vivo Analysis of Retroviral Enhancer Mutations in Hematopoietic Cells: SP1/EGR1 and ETS/GATA Motifs Contribute to Long Terminal Repeat Specificity

Abstract

The objective of this work was to identify, in the context of chromosomally integrated DNA, the contribution of defined transcription factor binding motifs to the function of a complex retrovirus enhancer in hematopoietic cells in vivo. Repopulating murine hematopoietic cells were transduced with equal gene dosages of replication-incompetent retrovirus vectors encoding enhanced green fluorescent protein. Enhancer sequences were derived from mouse spleen focus-forming virus. Destruction of GC-rich sites representing overlapping targets for SP1 or EGR1 uniformly attenuated gene expression (∼25 to 70% of wild-type levels) in all hematopoietic lineages, as shown by multicolor flow cytometry of peripheral blood and bone marrow cells at various time points posttransplantation. In contrast, a point mutation within a dual ETS/GATA motif that abolished transactivation by ETS factors but not by GATA-1 slightly increased activity in erythroid cells and significantly attenuated enhancer function in T lymphocytes. This study shows that controlled gene transfer in transplantable hematopoietic cells allows a functional analysis of distinct cis elements within a complex retrovirus enhancer, as required for the characterization and engineering of various cellular and viral regulatory sequences in basic research and gene therapy.

Understanding the role of defined transcription factor binding motifs within complex enhancers is a common goal in research dedicated to developmental biology, cellular differentiation, oncology, and virology. Identifying motifs that support overall enhancer activity and differentiation-dependent regulation in vivo is also of interest for the development of improved gene vectors for experimental or medical purposes (2, 6, 20, 41). Most of our insights into enhancer functions originate from experiments using transient or stable transfection of reporter gene constructs or using transgenic animals. Transfection assays cannot be applied to organs with high cellular turnover and multistep differentiation events, as the efficiency of stable genomic integration is low, expression levels are highly integration site dependent, and copy numbers in single cells can hardly be controlled (37). Functional heterogeneity due to unpredictable copy numbers and stochastic influences from integration sites is also a major limitation of transgenic animals (12, 14). Although this approach can still be used for functional enhancer analysis in vivo (26), it is technically cumbersome and requires the establishment and maintenance of new animal strains.

Simple retroviruses such as murine leukemia viruses (MLV) and derived gene vectors efficiently integrate at random positions of chromosomes (for a recent review see reference 17). Possibly due to some preference of the integration machinery for open chromatin, position effects on retrovirus integration sites are less variable than those observed with stably transfected transgenes (24, 30, 33). Defined mutations introduced into enhancers of replication-competent MLV alter the kinetics as well as the differentiation dependence of cellular transformation events induced by retrovirus insertional mutagenesis in hematopoietic cells in vivo, thereby indirectly reflecting a role of the potential cellular transactivators involved (8, 28, 29, 46, 47). However, a role for the transactivators of interest in cells not being transformed by the replication-competent retrovirus is easily overlooked (5).

Replication-incompetent retrovirus gene vectors have been developed by replacing the retrovirus coding sequences with transgenes of interest. As with the induction of diseases by replication-competent MLV, the enhancer plays an important role in the persistence and differentiation-dependent regulation of the vector’s transcriptional potency (2, 20, 41). Reporter genes such as that encoding enhanced green fluorescent protein (EGFP) have facilitated the monitoring of transgene expression in differentiating hematopoietic cells by multicolor flow cytometry (36). EGFP can thus be used to reveal differences between retrovirus vectors with various cis regulatory elements. However, in vivo studies published to date compared vectors that were distinguished by a number of alterations in different cis elements, thus providing only limited mechanistic insights (18, 20, 50).

Here, we demonstrate that retrovirus vectors allow a functional analysis of defined, individual enhancer mutations in hematopoietic cells in vivo. This is shown with the complex and highly condensed enhancer/promoter sequences from the U3 region of the retrovirus long terminal repeat (LTR) of mouse spleen focus-forming virus (SFFVp). These sequences are of considerable interest because they not only contribute to the disease kinetics of the erythroleukemia induced by the replication-competent Friend virus complex in mice (4) but also represent one of the most potent control regions for transgene expression in primitive hematopoietic cells and their progeny (2, 5, 21). According to electrophoretic mobility shift assays (EMSA) and transient transfection assays with transformed cells, the concert of SP1, ETS, and AML transcription factors explains about 50% of the activity of the SFFVp enhancer in various hematopoietic cells (3). Besides, there are E boxes for the binding of basic helix-loop-helix proteins (34), GATA recognition elements (1), and sites for yet-unknown DNA-binding activities FVa, FVb2, and FVc (32, 45, 46). An overview of the elements found in SFFVp is shown in Fig. 1. In addition, the upstream conserved region (UCR) of the retrovirus enhancer (located between +36 and +96; not shown in Fig. 1) has an important modulating function, possibly through binding of, among others, NFAT and YY-1 (15, 42; A. Wahlers, P. F. Zipfel, K. Itoh, B. Fehse, Z. Li, B. Schieldmeier, C. Skerka, W. Ostertag, and C. Baum, unpublished data). Although it is likely that a concert of proteins targeting these motifs explains the strong transcriptional activity of MLV enhancers in a number of hematopoietic cell types, it is unclear which sites are involved in constitutive or lineage-dependent regulation.

FIG. 1.

Alignment of regulatory elements present in the central enhancer region of the LTRs of SFFVp, MoMLV, and MPSV. Potential sites within the UCR (+36 to +97) are not shown. Dots, homology; dashes, deletions. The direct repeat (DR) of MoMLV and MPSV is shown as a unique sequence. 3′ FR, 3′ flanking region of the DR. For abbreviations of binding sites, see the introduction. Thick boxes, enhancer elements addressed in the present study. Core sequences are shaded.

In the present study, we addressed the role of GC-rich SP1 motifs and of a dual ETS/GATA motif (LVb; Fig. 1) within this enhancer assembly in hematopoietic cells in vivo. SP1 motifs have been chosen because they were implicated as an important regulator of retrovirus enhancer activity in studies with embryonic stem (ES) cells (19, 38) and because they may be involved in chromatin reorganization, thus mediating the general accessibility and long-term expression of the enhancer in vivo (7, 11, 31, 44). The ETS/GATA motif was of interest because, although it represents the only conserved ETS motif in the SFFVp enhancer, a point mutation destroying the ETS consensus can be found in the Axelrad strain of SFFVp (see alignment in reference 35), thus raising questions as to its contribution to overall enhancer activity.

(This study contains parts of the doctoral thesis of A. Wahlers.)

MATERIALS AND METHODS

Construction of retrovirus vectors.

Retrovirus vectors expressing EGFP were derived from pSF11EGFPrev (51). Enhancer mutations were inserted by site-directed mutagenesis or PCR-based methods, with subcloned LTR sequences as templates and proofreading Pfu polymerase (QuikChange site-directed mutagenesis kit; Stratagene, Heidelberg, Germany). Enhancer mutants were exchanged for the corresponding sequences in pSF11EGFPrev by using unique restriction sites NheI and XhoI. The resulting plasmids were sequenced using the M13rev primer, confirming the identities of the enhancer mutants contained in the 3′ LTR of the plasmidal provirus. Oligonucleotides used for creation of enhancer mutations were dSpDF and dSpDR for dSpD, dSpUF and dSpUF for dSpU, dEtsF and dEtsR for dETS, dFVaF and dFVaR for dFVA, and finally dFVb2F and dFVb2R for dFVb2. Sequences of oligonucleotides were as follows: M13rev, 5′-TGACCGGCAGCAAAATG-3′; SpDF, 5′-GTCACCAGATATGGCCCAACCCTCAGCAGTTTC-3′; SpDR, 5′-TTAAGAAACTGCTGAGGGTTGGGCCATATCTG-3′; dSpDF, 5′-GTCACCAGATATGGCCCAAAACTCAGCAGTTTC-3′; dSpDR, 5′-TTAAGAAACTGCTGAGTTTTGGGCCATATCTG-3′; SpUF, 5′-AGAAGTTCAGATCAAGGGCGGGTACATGAAAATAGCTAACG-3′; SpUR,5′-CGTTAGGCTATTTTCATGTACCCGCCCTTGATCTGAACTTCT-3′; dSpUF, 5′-AGAAGTTCAGATCAAGTTTTGGTACATGAAAATAGCTAACG-3′; dSpUR, 5′-CGTTAGGCTATTTTCATGTACCAAAACTTGATCTGAACTTCT-3′; dEtsF, 5′-TGGGCCAAACAAGATATCTGCGG-3′; dEtsR, 5′-CCGCAGATATCTTGTTTGGCCCA-3′; dFVAF, 5′-GAAAATAGCTAACGTTTGTCCAAACAGGATATCTGC-3′; dFVAR, 5′-GCAGATATCCTGTTTGGACAAACGTTAGCTATTTTC-3′; dFVb2F, 5′-TTCGGCCCCGGCCCGTGTCCAAGAACAGATGGTCAC-3; dFVb2R, 5′-GTGACCATCTGTTCTTGGACACGGGCCGGGGCCGAA-3′.

Retrovirus vector production target cell transduction.

Retrovirus vectors were produced from transiently transfected Phoenix-ampho packaging cells cotransfected with an expression vector for the glycoprotein of vesicular stomatitis virus as described previously (51). Thawed aliquots of cryopreserved supernatants were diluted in tissue culture medium to obtain equivalent multiplicities of infection (MOIs). Replication-defective amphotropic supernatants were harvested from polyclonal cultures of GP&envAM12 packaging cells, as described previously (2). Target cells were suspended at a density of 10⁵/ml in media containing retrovirus supernatants and protamine sulfate (4 μg/ml). Then, cells were plated in single wells of six-well tissue culture plates and centrifuged for 1 h at 800 × g. Vector-containing media were exchanged with fresh media the following day. Cells were grown for another 48 h prior to determining the frequency of EGFP⁺ cells by flow cytometry. Cells (and media used) were as follows: FDCP-mixA4 (Iscove’s modified Dulbecco medium [IMDM], 20% HS, murine interleukin-3 [mIL-3] conditioned medium), WEHI-3B (IMDM, 10% fetal calf serum [FCS]), F4-N (minimal essential medium [MEM], 10% FCS), EL-4 (IMDM, 10% FCS), K562 (RPMI 1640, 10% FCS), SC1 (MEM, 10% FCS). All media were supplemented with 2 mM glutamine, 100 U of penicillin/ml, and 100 μg of streptomycin/ml (Sigma, Deisenhofen, Germany).

Mouse experiments.

Mouse bone marrow (BM) cells were harvested from long hind limb bones of C57BL/6 mice, aged 3 to 4 months, 4 days after intraperitoneal injection of 5-fluorouracil (Sigma) at 150 mg/kg of body weight. Mononuclear cells were cultivated in IMDM supplemented with 10% FCS and recombinant growth factors mIL-3, human IL-6 (hIL-6), and murine stem cell factor (mSCF), as described previously (51). After 2 days, cells were transduced twice with retrovirus supernatants (MOI ≈ 2) and reinjected into irradiated (950 cGy) syngeneic recipients, each receiving 2 × 10⁶ mononuclear cells. Peripheral blood (PB) was collected from the tail vein and processed for fluorescence-activated cell sorter analysis. Experiments were performed in a pathogen-free animal facility of the Heinrich-Pette-Institute according to animal experimental ethics guidelines.

Flow cytometry.

Detection of cellular EGFP expression was accomplished on a FACScalibur cytometer (Becton Dickinson, Heidelberg, Germany) equipped with a 488-nm argon laser and 530/30- (FL1 and EGFP) and 575/38-nm (phycoerythrin [PE]) bandpass filters. Transformed cells cultured in vitro and leukocytes of the peripheral blood were gated according to scatter criteria. Primary murine mononuclear cells were labeled with PE-conjugated lineage-specific monoclonal antibodies (PharMingen, Hamburg, Germany) CD3, CD4, CD8, CD11b, CD34, TER119, and B220. Gate criteria were adjusted according to isotype controls. Mean fluorescence of EGFP expression was determined using CellQuest software (Becton Dickinson).

Statistical analysis.

EGFP fluorescence levels are expressed as mean values ± standard deviations, calculated with Microsoft Excel software.

EMSA.

One to 5 pM double-stranded oligonucleotides were end labeled with [γ-³³P]ATP (specific activity, 3,000 Ci/mM) by a kinase reaction. Double-stranded oligonucleotides were obtained by annealing equimolar amounts of the forward and reverse oligonucleotides above. For example, the SpU double-stranded oligonucleotide resulted from annealing SpUF and SpUR. Labeled oligonucleotides (10 to 30 μM) were incubated at 4°C with recombinant EGR1 or SP1 proteins. Recombinant EGR proteins were expressed in the baculovirus system as described previously (9); recombinant SP1 was obtained from Promega. Binding reaction mixtures were incubated in 20 μl of buffer containing 20 mM HEPES, pH 7.9, 50 mM KCl, and 0.5 μg of poly(dI-dC) (Amersham Pharmacia Biotech) for 20 min. Antiserum directed against EGR1 or SP1 (Santa Cruz Biotechnology) was added to the incubation mixture. The resulting DNA-protein complexes were separated in a 5% nondenaturing gel at 4°C in 0.25× Tris-borate-EDTA (TBE) at 150 V and 20 mA. Gels were dried and exposed to X-ray films.

RESULTS

Genetic and functional characterization of DNA-binding motifs.

Potential binding sites for SP1 and ETS/GATA motifs in the SFFVp enhancer were identified by alignment with related MLV enhancer sequences and by sequence analysis in a transcription factor database (http://genomatix.gsf.de/cgi-matinspector_prof/mat_fam.pl). The binding sites for SP1 in the upstream region of the enhancer and for ETS beside the AML/CBF-binding enhancer core motif are known from our earlier work (3). We now checked whether the CCAACCC motif in the SFFVp sequences located 17 nucleotides 3′ of the direct repeat (FVc motif) allowed the binding of SP1. The corresponding sequences in myeloproliferative sarcoma virus (MPSV) (CCCGCCC) represent a high-affinity SP1 site, and those found in Moloney murine leukemia virus (MoMLV) (CCAGCCC) represent a weaker one (19, 38). EMSA with nuclear extracts from murine erythroleukemia F4-N cells revealed binding of SP1 at FVc (Fig. 2A). However, EMSA performed with recombinant SP1 revealed that this downstream SP1 site (referred to as SpD) had reduced activity compared with the corresponding sequence of MPSV or the upstream SP1 site (SpU) of SFFVp (Fig. 2 and data not shown). Interestingly, both SP1 sites of SFFVp also bound recombinant EGR1 but not EGR4 (data not shown), indicating that SP1 is not the only cellular protein potentially targeting these motifs. The affinity of EGR1 was also higher at SpU than at SpD. Point mutations introduced into SpD and SpU motifs strongly reduced binding of both SP1 and EGR1 (Fig. 2B). Residual binding activities of elevated motility in EMSA noted with these probes (dSpU and dSpD) were unspecific. Therefore, we addressed SpU and SpD as dual SP1/EGR1 sites, implying that only one factor, acting as a monomer, binds at a given site and time point.

FIG. 2.

SP1 and EGR1 bind to the FVc motif (SpD region) and to the GC box in the SpU region. (A) SpD contains an SP1 site. ∗∗, supershift of SP1 complexed with the antibody. (B) SpD and SpU bind both SP1 and EGR1. ∗∗∗, supershift of EGR1 complexed with the antibody. Oligonucleotides with destroyed SP1 sites (dSpD and dSpU) bind SP1 and EGR1 with at least 10-fold-reduced affinity. EMSAs were performed using recombinant protein SP1 or EGR1 and radiolabeled oligonucleotides SpD and SpU. A 50-fold excess of cold SpD oligonucleotides or the SV oligonucleotide containing multiple SP1 motifs derived from the promoter of simian virus 40 (3) or antibodies against the hemagglutinin tag present in recombinant SP1 and EGR1 were added as indicated.

EMSA with the LVb motif indicated potential binding of GATA-1 in murine erythroleukemia (MEL) cell extracts, confirming recently published data (results not shown) (1). The binding of ETS factors at the LVb motif has previously been shown by others (49, 53).

Generation of retrovirus vectors containing defined enhancer alterations.

To address the functional significance of the sites of interest, retrovirus vectors were constructed on the basis of a recently published version completely lacking retrovirus coding sequences (22, 51). All vectors used in this report were identical with the exception of the defined enhancer alterations in the LTR. The different LTRs (Fig. 3) thus contained the wild-type SFFVp enhancer/promoter (SF) or variants with a destroyed upstream SP1/EGR1 site (dSpU), a destroyed downstream SP1/EGR1 motif (dSpD), a combination of dSpU and dSpD (dSpUD), or the dETS motif at LVb, with the SP1/EGR1 sites left intact. A control derived from the dSpUD enhancer contained additional mutations in FVa and FVb2 sites. As reporter cassettes, we used either EGFP or a bicistronic cassette (DIN) linking truncated nerve growth factor receptor (ΔLNGFR) to the neomycin resistance gene (neoR) via an internal ribosomal entry site of poliovirus.

FIG. 3.

Scheme representing retrovirus vectors used in this study. Specific alterations introduced into the regulatory boxes are in boldface. All enhancer/promoter sequences were based on a variant of SFFVp in which duplicated sequences of the direct repeat (DR; Fig. 1) were deleted. IRES, internal ribosomal entry site.

Replication-defective particles of EGFP vectors pseudotyped with the glycoprotein of vesicular stomatitis virus G were generated by transfection in Phoenix packaging cells and stored in aliquots at −70°C (51). Each batch was titrated by limiting dilution on SC1 fibroblasts; flow cytometry was used to quantify the numbers of cells expressing the vector-encoded phenotype. All subsequent transductions with EGFP vectors were performed at comparable MOIs. Amphotropic DIN vectors were produced from clones of transduced and G418-selected retrovirus GP&envAM12 packaging cells.

Impact of retrovirus enhancer alterations in transformed cells.

The differentiation dependence of the enhancer motifs was first analyzed with a panel of cell lines representing different developmental stages (2, 3, 51): multipotent mouse FDCP-mixA4 hematopoietic progenitor cells, F4-N mouse erythroleukemia cells, WEHI-3B mouse monocytic cells, K562 human CML-derived cells, EL4 mouse T lymphoblasts, and SC1 mouse fibroblasts. Cells were transduced with the EGFP vectors containing the enhancer variants indicated above at comparable MOIs (Fig. 3). EGFP expression was monitored in unselected polyclonal cultures 3 to 5 days posttransduction, as described previously (51), yielding highly reproducible results.

This analysis indicated that the mutation at LVb (dETS) had a differentiation-dependent effect (Table 1). While expression in F4-N cells was not affected, a slight reduction (around 10%) occurred in FDCP-mix cells, EL4 lymphoblasts, and premonocytic WEHI cells. EGFP expression was more strongly reduced in K562 cells (by 30%) and SC1 fibroblasts (by 25%). The differentiation-dependent impact of dETS was confirmed using three independently established mass cultures of cells transduced with the DIN vectors and selected for transduction with G418 (Fig. 4A and B). Southern blot analysis of individual clones derived from these mass cultures revealed single-copy integrations in the majority of clones, thus excluding an effect of different gene dosages (Fig. 4C).

TABLE 1.

Evaluation of enhancer mutants in transformed cells

Cell line	Mean % expression ± SD (no. of expts) for mutant:
	dETS	dSpUD	dSpUDFVab2
SC1 (mouse fibroblast)	75.1 ± 10.5 (27)	55.4 ± 14.3 (14)	52.5 ± 7.9 (7)
FDCP-mixA4 (mouse pluripotent hemato-poietic progenitor)	91 ± 0.6 (6)	65.9 ± 4.6 (10)	48.6 ± 9.8 (6)
WEHI-3B (mouse mye-loid leukemia)	85.4 ± 1 (4)	50.3 ± 12.2 (4)	29.4 ± 5.9 (4)
F4-N (MEL)	101 ± 1 (4)	68.2 ± 3.1 (6)	39.9 ± 14.1 (6)
K562 (human myeloid leukemia)	69.5 ± 7.7 (4)	57.5 ± 0.1 (4)	38.3 ± 1.3 (4)
EL-4 (mouse T-cell lymphoma)	88.5 ± 8.7 (6)	77.2 ± 6.7 (8)	47.6 ± 19.5 (6)

FIG. 4.

The dETS mutation attenuates retrovirus expression in myeloid, but not in erythroid, cells. (A) Representative histograms of flow cytometry results obtained with mass cultures expressing vector SF11DIN (boldface line) or dETS11DIN (dotted line). (B) Summary of results achieved with DIN vectors in F4-N (mouse erythroid), WEHI-3B (mouse myeloid), and K562 (human CML) cells. Shown are means and standard deviations of three independently established mass cultures per cell line. (C) Southern blot revealing that the majority of individual clones derived from mass cultures (A) had unique integration sites of the vector; the radioactive probe was derived from neoR. ∗, clones with two integrations or biclonal cultures; dash, empty lanes; +, positive control, derived from plasmid DNA of pSF11DIN; M, marker for molecular weight.

Vectors with mutations in SP1/EGR1 sites showed a more uniform reduction of activity in all cell types investigated. The results shown in Table 1 indicate that the SP1/EGR1 motifs are required for full activity of the SFFVp enhancer in all cell types investigated, as mutations of both sites decreased enhancer function by 25 to 50%. Vectors with individual mutations of SpU or SpD were analyzed in SC-1 fibroblasts. The presence of the SpD motif was less important, as the mutation reduced enhancer function by roughly 10% ± 8% (n = 8) whereas destruction of the SpU motif reduced enhancer activity by 46.5% ± 8% (n = 8), revealing a highly significant difference of activity (P < 0.01). This difference was in line with the stronger affinity observed by EMSA in vitro (Fig. 2 and data not shown). Combined destruction of both SP1 sites (dSpUD) had effects similar to those of the destruction of the SpU motif alone. These results suggested a strong contribution of the SpU motif, likely binding SP1, EGR1, or related proteins, to retrovirus enhancer activity, almost irrespective of the cell type under investigation.

To verify that the enhancer motifs addressed so far were not the only activating sites of the SFFVp enhancer, we used the dSpUD variant to introduce further mutations into the FVa and FVb2 motifs. These motifs have been shown to attract cellular DNA-binding factors in the context of the related enhancer of Friend MLV (32). However, the corresponding cellular activities remain to be identified. The resulting mutant (dSpUDdFVab2; Fig. 3) was further attenuated and had a more differentiation-dependent profile, with the strongest loss of expression observed in WEHI-3B cells and almost no additional effect seen in fibroblasts (Table 1). Taking into consideration the fact that further potent cellular transactivators can still bind to other motifs such as the core-element, the E boxes, and the NFAT site in the UCR of the SFFVp enhancer (3; Wahlers et al., unpublished), these results clearly demonstrate that the mutations of the SP1/EGR1 sites or the ETS/GATA motif were analyzed in the context of an otherwise fully functional retrovirus enhancer orchestra.

In vivo fate map of enhancer alterations: differential impact of mutations in SP1/EGR1 and ETS/GATA motifs.

Finally, we analyzed in vivo, in primary hematopoietic cells, the differential impact of the mutations in the SP1/EGR1 sites and the ETS/GATA motif. In contrast to the analysis in cell lines, this experimental setting is not biased by potential changes in transcription factor activities resulting from cellular transformation and involves numerous differentiation events from the level of a repopulating, multipotent progenitor to a variety of mature cell types found in the PB. BM cells of C57BL/6 mice were transduced with cell-free supernatants of EGFP vectors SF11E, dSpUD11E, and dETS11E, taking care to use comparable MOIs. Three experiments were performed, two addressing the role of the SP1/EGR1 motifs by comparing SF11E with dSpUD11E, and one addressing the function of the mutation in LVb (SF11E versus dETS11E).

PB cells were analyzed for EGFP expression by flow cytometry at various time points posttransplantation, and different cell lineages were identified by scatter characteristics (red blood cells [RBC]) or by staining with lineage-specific B220 monoclonal antibodies (B cells) or CD3 or a simultaneous staining with CD4 and CD8 (T cells) and CD11b (myeloid cells). A representative analysis with cells harvested 13 weeks posttransplantation is shown in Fig. 5. At the end of each experiment, BM cells were also analyzed with the antibodies described above and also CD34 (mixed progenitor cells) and Ter119 (erythroid progenitor cells). Gene transfer efficiency was determined by PCR, which detected the EGFP provirus in a single CFU-GM obtained from unselected BM cells at the end of each experiment.

FIG. 5.

The dSpUD mutant shows reduced expression in different blood cell lineages. Set of representative histograms showing multilineage analysis of EGFP expression in cells derived from the PB of transplanted mice (chosen from a set of data represented in Fig. 6A). The upper two triplets show myeloid cells (stained by CD11b) and T cells (stained by CD4 or/and CD8); the lower triplet shows RBC, as defined by forward scatter (FSC).

The first in vivo analysis addressing the role of the SP1/EGR1 sites was performed up to week 13. Seven mice were transplanted per vector. Due to the rather low gene-marking efficiency in this experiment (12 to 15% of BM-derived CFU-GM were PCR positive), some mice had low rates of EGFP-expressing cells (<5%) and, therefore, had to be omitted from statistical analysis of vector expression. The data described below (Fig. 6A) show the means of expression and standard deviations for at least four individual mice per vector. The second in vivo analysis with the SP1/EGR1 mutant enhancer involved eight mice per group, all qualified for fluorescence-activated cell sorter analysis due to a higher gene-marking rate, as confirmed by PCR analysis of CFU-GM (38% PCR positive for the wild-type enhancer and 24% PCR positive for the mutant).

FIG. 6.

Lack of SP1/EGR1 sites in the retrovirus enhancer of SFFVp (dSpUD mutant) reduces expression in all lineages and stages of blood cell differentiation examined. (A) Mean fluorescence intensity (MFI) of EGFP expression obtained with wild-type enhancer SF (white bars) and the dSpUD mutant (black bars) in the first experiment, with four mice per group. The ratio of both mean values is shown below each pair of columns. The frequencies of PCR-positive CFU-GM obtained at the end of the experiment are shown for comparison on the right. CD4 and -8, T cells; B220, B cells; CD11b, myeloid cells. All were obtained from PB. BM Lin-, BM cells not stained by the lineage marker cocktail consisting of CD4, CD8, B220, CD11b, and TER119. Ter119, nucleated BM erythroid cells. Error bars indicate standard deviations. (B) Same graph as in panel A, summarizing data from the second experiment. This was performed with seven mice per group, allowing statistical analysis. ∗∗, P < 0.005; ∗, P < 0.01. CD34⁺ BM progenitor cells were included in this analysis. (C) The frequencies of EGFP-expressing cells with both constructs were comparable.

In vivo, to a greater degree than previously observed in vitro, destruction of the SP1/EGR1 motifs (dSpUD) reduced retrovirus enhancer activity in all cell types examined. Similar results were obtained in two independent sets of experiments (Fig. 6A and B). The strongest decrease (to around 30% of wild-type levels) was observed in myeloid cells and RBC and their respective precursors in the BM. T cells showed a reduction to at least 50%, whereas expression in B cells was less significantly affected (Fig. 6A and B). For the two vectors, the frequencies of EGFP⁺ cells in the various populations were equivalent (Fig. 6C), a finding compatible with the PCR analysis; this emphasizes that the differences in reporter gene expression could not be explained by differences in transgene dosage. Both enhancer variants showed constant and consistent expression levels over the time of the experimental follow-up (data not shown).

The experiment addressing the impact of the dETS mutation in the LVb site involved five mice per group. The gene-marking rate in CFU-GM was 43% for the wild-type enhancer and 46% for the mutant. Accordingly, the frequencies of EGFP⁺ cells were almost identical (Fig. 7C). The analysis was performed up to 9 weeks posttransplantation. In marked contrast to what was found for the SP1/EGR1-defective enhancer, the dETS mutation had a highly differentiation-dependent effect (Fig. 7A and B). While expression in myeloid and B cells was only slightly reduced, enhancer function was significantly impaired (reduction by about 50%) in T cells. The strong attenuation of expression in primary T cells, induced by a single point mutation, contrasted with unaltered expression in myeloid BM cells (CD11b⁺) and multipotent progenitors (CD34⁺). Even more strikingly, in early TER119⁺ erythroid progenitor cells, the dETS mutant showed increased activity compared to the wild-type enhancer (up to 1.6-fold in large TER119⁺ cells; Fig. 7B). For this analysis, Ter119⁺ cells were further subdivided according to side scatter characteristics (size) into earlier (Ter119⁺ large), medium (Ter119 middle), and more mature (Ter119 small) progenitors (Fig. 7D) (18, 51). Importantly, the results achieved with the dETS variant also represent a control for the studies addressing the function of the SP1/EGR1 sites (Fig. 5 and 6; Table 1), as clearly distinct patterns of enhancer activity were obtained.

FIG. 7.

The dETS mutation increases expression in erythroid progenitor cells in vivo, while strongly attenuating expression in T cells. (A) Analysis of PB cells similar to that shown in Fig. 6 but using EGFP vectors containing the dETS enhancer (black bars) beside wild-type enhancer SF (white bars), showing results obtained by week 4 (wk4) and week 9. (B) In contrast to mature PB cells, BM cells analyzed by week 9 showed no difference in expression (CD11b⁺ and CD34⁺) or even an increase in favor of the dETS enhancer in large (Ter119L), medium-sized (TER119M), and small (TER119S) erythroid progenitors. (C) The frequencies of EGFP-expressing cells were comparable in both groups. (D) Gate criteria used to identify TER119⁺ subpopulations.

DISCUSSION

To our knowledge, this is the first report demonstrating the utility of using replication-incompetent retrovirus vectors encoding marker genes to analyze in vivo, in hematopoietic and mature blood cells, the impact of defined point mutations in a complex enhancer array. Specifically, we show here a differentiation-independent, constitutive function of GC boxes representing overlapping SP1/EGR1 motifs as opposed to a more lineage-specific impact of a single point mutation in an overlapping ETS/GATA consensus. The activity of the enhancer mutants thus reflects the expression and function of the corresponding transcription factors in hematopoiesis.

SP1/EGR1.

Evidence for a crucial role of GC-rich regions in MLV enhancers was first obtained in studies addressing retrovirus enhancer activity in primitive embryonic cells, including ES cells (19, 38). It has been shown that a point mutation increasing the affinity to SP1 significantly increased the activity of the MPSV enhancer in ES cells, where many hematopoietic transcription factors are poorly expressed or absent. It has also been suggested that this high-affinity binding site for SP1 leads to an increased probability of transcription in various types of hematopoietic cells (48), although this hypothesis has never been addressed experimentally. We show here that SP1 is not the only cellular transcription factor that may bind to the retrovirus GC boxes and identify a second putative SP1/EGR1 site in the enhancer of SFFVp, located downstream of the central enhancer array, as in MPSV. This motif, formerly called FVc in the context of enhancers related to Friend MLV (32), was shown by EMSA to bind both SP1 and EGR1, although the affinity was reduced compared with that for the corresponding sequences found in MPSV or the upstream GC box of SFFVp. A similar, slightly degenerated GC box still able to bind SP1 with reduced affinity is found in MoMLV (19, 38). To address the contribution of GC boxes to the in vivo regulation of the SFFVp enhancer, we expanded our previous work performed with cell lines (3; Wahlers et al., submitted) to an analysis of mice undergoing transplantation with retrovirus vector-transduced BM cells.

Although our functional analysis of a panel of transformed cell types revealed that the upstream GC box of SFFVp (SpU) had a stronger effect than the slightly degenerated downstream motif (SpD) identified in this study, we decided to perform the in vivo analysis with a mutant defective in both sites. Thus, compensation of function by access to a similar motif was less likely. Considering that we cannot exclude the possibility that SP1/EGR1 may bind to other, yet-unknown, sites that may have escaped our database search and functional investigations, the implications of our results are twofold. First, the SP1/EGR1 motifs addressed here make an important and rather ubiquitous contribution to the strength of the SFFVp enhancer, with the strongest effect observed in myeloid and erythroid cells and the weakest observed in B cells. Second, the SP1/EGR1 sites examined here are not strictly required to maintain accessibility of the enhancer during differentiation and over time. To further elucidate the role of SP1/EGR1 motifs and of the potential cellular binding activities, it may be interesting to search for related, yet-unknown sites located closer to the promoter region. Such sites may be more crucial for the modification of chromatin and repression or derepression of promoter activity, as suggested from studies with the promoter of human immunodeficiency virus type 1 or cellular genes (7, 10, 11, 31, 44). Moreover, it will be interesting to determine to what extent SP1 and EGR1 fulfill different functions in the retrovirus enhancer/promoter, as described for cellular binding sites (23). Irrespective of the outcome of such experiments, an important and rather constitutive role of GC boxes for retrovirus enhancer activity in transplanted hematopoietic cells, including progenitor cells, is established here.

ETS/GATA.

In marked contrast to what was found for the GC box-defective enhancers, the dETS mutation, which destroyed the consensus for ETS factors while potentially improving accessibility for GATA proteins, created a more lineage-specific, namely, erythrotropic, enhancer variant. The mutation tested here is present in the Axelrad strain of SFFVp and differs from that introduced by Speck and colleagues to generate a replication-competent MoMLV with increased pathogenicity in myelo-erythroid cells (46). Interestingly, the mutant tested here still has profound activity in progenitor cells, while being significantly attenuated in T cells. Our study reveals that the ETS site at LVb, known to be an intimate co-operator of CBF/AML-related transcription factors targeting the neighboring core motif (3, 49), is essential for full activation of the retrovirus enhancer in T cells but is dispensable for expression in the erythroid lineage. Improved binding of other transcription factors to neighboring enhancer sequences and/or a repressive effect of the ETS site during early erythroid differentiation may contribute to the improved performance of the dETS enhancer in erythroid cells. An alternative hypothesis is compensatory binding of GATA factors in erythroid cells (1), where expression of GATA-1 and downregulation of ETS proteins are required for differentiation (40, 54). Thus, the dETS mutation uncovers a notable plasticity in a single retrovirus enhancer motif. This plasticity appears to contribute strongly to the broad activity of murine retrovirus enhancers in hematopoietic and nonhematopoietic cell types. Interestingly, different GATA factors may bind to identical sites in different tissues (27). Therefore, it will be interesting to determine whether GATA factors can bind to this motif in T cells or other pluripotent hematopoietic cells where different members of this family are functional (13, 16, 25) and whether these proteins can interact with CBF/AML family members targeting the core motif. If so, it is likely that there is a competition with ETS factors, as described previously on the basis of results for other enhancers (40, 54), and the relative concentration, target site affinity, and interactions with co-operating transcription factors determine which factor actually binds to this motif and exerts its function.

Experimental strategy.

Although the differences evoked by single point mutations within a complex enhancer array may appear moderate, they are nevertheless highly informative, because results were obtained in the full context of a functional enhancer with chromosomally integrated transgenes. When mutations are present in replication-competent retrovirus, such alterations of enhancer activity may be augmented by several orders of magnitude, thus explaining their well-known impact on the kinetics and specificity of leukemia or lymphoma induction (8, 28, 29, 46, 47). For example, a mutation destroying the core element results in a less than 50% reduction of reporter gene activity (3), although this site plays a central role in disease induction by replication-competent retroviruses (46).

Variations in enhancer activity between different mice were likely caused by the influence of the integration site. Differences in transgene dosage were unlikely, as we took care to perform the work with replication-incompetent vectors and equivalent MOIs. The side-by-side comparison of the three in vivo experiments shown here reveals that higher marking rates translate into reduced interanimal variability, possibly because higher marking rates support polyclonal reconstitution of transgenic hematopoiesis in single mice.

It is also worth mentioning that not all results obtained in vivo were predicted by the data from transformed cell lines. The reduced impact of the SP1/EGR1 mutations in B cells and the enhancing effect of the dETS mutation on gene expression in erythroid cells became apparent only by in vivo analysis. Elucidating the impact of enhancer mutations on maintenance of expression over time and differentiation dependence certainly requires in vivo analysis, as transformed cells likely differ from their untransformed counterparts by alterations in their transcription factor setup and chromatin organization. Carefully chosen cell lines may still be valuable for screening.

Outlook.

The present study reveals new insights into the mechanisms underlying retrovirus vector-mediated gene expression in hematopoietic cells in vivo. This may have interesting consequences for the design of improved gene vectors as well as for basic research addressing molecular virology, hematopoiesis, and oncology. Specifically, the strong impact of the SP1/EGR1 sites not only may suggest a molecular mechanism underlying the role of the LTR in the transformation of primitive hematopoietic cells by MPSV (48) but also opens new perspectives for vector design. An interesting implication of our results achieved with the dETS point mutation is the identification of an erythrotropic retrovirus enhancer which still has substantial activity in progenitor cells. This may explain why this mutation was tolerated in a strain of SFFVp. In the course of the infection with the Friend virus complex, SFFVp is predominantly propagated in erythroid progenitors due to its expression of a glycoprotein that stimulates the cellular receptor for erythropoietin (reviewed in references 35 and 43). If the same mutation were introduced into a more lymphotropic MLV (e.g., MoMLV or SL3), it may significantly reduce virus lymphomagenicity. The dETS enhancer or variations thereof may be of practical interest for in vivo selection of hematopoietic cells (39, 52) in gene therapy of inborn defects affecting erythroid cells. Beyond these findings and suggestions, the experimental system described here may find more widespread application in the analysis of viral and cellular sequences controlling transcription or posttranscriptional processing of RNA.