New function of a well-known promoter: enhancer activity of minimal CMV promoter enables efficient dual-cassette transgene expression

Abstract

Background:

Co-expression of multiple genes in single vectors has achieved varying degrees of success by employing two promoters and/or with application of viral 2A-peptide or internal ribosome entry-site (IRES). However, promoter interference, potential functional-interruption of expressed-proteins by 2A-generated residual peptides or weaker translation of IRES-mediated downstream gene curtail their utilization. Thus, there is the need of single vectors that robustly express multiple proteins for enhanced gene therapy applications.

Methods:

We engineered lentiviral-vectors for dual-cassette expression of green fluorescent protein (GFP) and mCherry in uni- or bidirectional architectures using short-version (Es) of elongation factor 1α (EF) promoter and simian virus 40 promoter (Sv). The regulatory function of a core fragment (cC) from human cytomegalovirus (CMV) promoter was investigated with cell-lineage specificity in NIH3T3 (fibroblast) and hematopoietic cell lines U937 (monocyte/macrophage), LCL (lymphoid), DAMI (megakaryocyte), and MEL (erythroid).

Results:

The cC element in reverse-orientation not only boosted upstream Es promoter to levels comparable to full-length EF in DAMI, U937 and 3T3 cells, but also blocked the suppression of downstream Sv promoter by Es in U937 and 3T3 cells with further improved Sv activity in DAMI cells. Such lineage-restricted upregulation is likely attributed by two protein-binding domains of cC and diverse expression of related factors in different cell types for enhancer and terminator activities, but not spacing function.

Conclusion:

Such newly developed dual-cassette vector could be advantageous particularly in hematopoietic cell-mediated gene/cancer therapy by allowing for independent and robust co-expression of therapeutic gene(s) and/or with a selectable gene or imaging marker in the same cells.

1. INTRODUCTION

Vector-mediated delivery and expression of exogenous genes in biological systems have revolutionized bioscience research and its applications. Appropriate vector design is pivotal to the development of successful viral and non-viral expression systems in vitro and in vivo^1,2. Increasingly complex research and applications require robust and/or divergent expression of two or more transgenes for various purposes, such as co-expressing multiple chimeric antigen receptors (CAR) and/or cytokines into the same cells for superior cancer therapy, biological sample tracking/imaging, introducing multimeric proteins for industrial production of recombinant proteins. Unlike the challenges faced by co-transfecting a mixture of several vectors, bicistronic vectors, each expressing a single transcriptional unit (TU) from a promoter, utilize either internal ribosome entry site (IRES)³ or viral 2A or 2A-like peptide elements⁴ to generate multiple gene products from one vector. Such vectors can ensure that all transgenes are regulated/expressed together at the same time and location. However, these systems also have limitations including the number of transgenes, weaker translation of IRES-mediated downstream gene and potential functional impairment of expressed proteins by 2A-generated residual peptides^5,6.

Alternatively, independent expression of multiple genes in two (or more) TUs from different promoters in the same vector would provide the advantage of being robustly regulated in a spatio-temporal manner. For instance, such a system can potentially improve the safety and efficacy of hematopoietic stem cell (HSC)-mediated gene therapy when utilizing dual-cassette lentiviral vectors (LV), in which one cassette expresses a selectable marker or a reporter from a ubiquitous promoter and the other cassette expresses a therapeutic transgene restricted to targeted blood cell types from a lineage-specific promoter^7,8. It could also increase the number of CARs and cytokines co-expressed in the same cells when in combination with the usage of IRES or 2A element. However, transcriptional (or promoter) interference between adjacent TUs may occur and often results in reduced expression or even complete suppression of genes in one or both genes ⁹. Several factors may contribute to the interference, including the comparative strengths and relative positioning of the promoters, susceptibility of promoter(s) to epigenetic modification and the roles of other cis-acting regulatory elements^10,11, although further elucidation remains an active area in vector design research. In addition, other regulatory elements such as insulators¹², terminator or pause sequences¹³, polyadenylation (poly(A)) sequences^14–16 and DNA spacer elements¹⁷ also play roles in transgene expression. While poly(A) sequences are essential in vector designs to ensure transcription termination and enhance nuclear transport, mRNA stability, and translation of each TU, the inclusion of other elements is based on their utmost necessity to meet vector and gene expression needs. Comprehensive considerations of vector size capacity, expression configuration desired, and promoter-pair compatibility are crucial in dual-cassette expression vector design. Efforts to optimize such vector designs remain wanted to achieve effective, independently controlled expression of multiple genes for bioscience research and applications.

Promoters for human cytomegalovirus major immediate-early protein (CMV), the simian virus 40 immediate-early protein (SV40) and human elongation factor 1α (EF1α) and its short version (Es) have been widely used for transgene expressions. The advantages of their comparatively smaller size have made them more applicable¹⁸ in dual-cassette vectors and size-restricted adeno-associated viral vectors (AAV)¹⁹, although they are not the strongest of constitutive promoters and viral promoters are prone to methylation. Conversely, by combining desirable enhancer and promoter elements coordinatively, hybrid/synthetic promoters, even though inevitably larger, can achieve more controlled expressions that are either robust and sustainable avoiding silencing, or tissue restricted or responsive to external stimuli ^12,18. Moreover, reducing the size of such chimeric enhancer/promoters without sacrificing appropriate features has become highly desirable for the expression of multicistronic transgenes in vector design.

The full-length CMV promoter is the most commonly used and versatile regulatory element with a TATA-box-containing promoter region and an enhancer region from −550 bp to −39 bp relative to the transcription start site (tss) (GenBank Accession No. X03922.1)²⁰. Multiple fragments with different sizes from the enhancer region have been utilized to boost both ubiquitous and cell-specific promoters, including CBh and CAG ^21,22 as well as cardiomyocyte-, alveolar epithelia- or neuron-specific promoters ^23,24. Moreover, a 120-bp derivate of the core CMV promoter-extended region (cC element), containing nucleotides of −50bp to +70bp, has been employed as a minimal promoter for inducible transgene expression in mammalian cells in vitro ^25,26. However, no other functions have been explored for this small-size cC element.

In this study, we investigated if the small-size 120bp cC element, in cooperation with the short but weak Es (212bp; GenBank Accession No. J04617.1, nucleotides 397 to 608) and SV40 promoter (Sv) (269bp; GenBank Accession No. MT086573.1, nucleotides 4403 to 4671), would modulate transcription interference between two TUs in dual-cassette LVs and simultaneously provide relatively small cumulative size of regulatory elements which are often associated with optimal titers or required for vectors with size restriction (such as AAV). Expression levels of green fluorescent protein (GFP) and a variant of red fluorescent protein (mCherry) driven from two independent promoters were evaluated in cells from different hematopoietic lineages, including cell lines of macrophage/monocytic (U937), megakaryocytic (DAMI), lymphoid (LCL) and erythroid lineages (MEL). Distinguished from its promoter activity, a novel function of cC element was documented that, when positioned in reverse orientation proximal to the downstream TU, cC could not only block promoter interference but also increase both upstream and downstream transgene expression with lineage-restriction. The modes of actions were also studied by site-directed mutagenesis and RT-qPCR, identifying two new protein-binding domains that may provide transcription factors (TFs)-associated enhancer function as well as protein-binding terminator function, but not spacer function. Our findings establish a new promoter-pairs, small in size but efficient in strength, for dual-cassette expression especially in hematopoietic cell types that could contribute significantly to the field of hematopoietic cell-mediated gene- or cancer-therapy.

2. MATERIAL AND METHODS

2.1. Primers and oligonucleotides

Primers and oligonucleotide sequences were designed in-house with SnapGene Viewer software (GSL Biotech, Chicago, IL) and synthesized by Integrated DNA Technologies Inc (Coralville, IA).

2.2. Construction of plasmids and vectors

We received a self-inactivating lentiviral vector expressing GFP from human phosphoglycerate kinase promoter (PGK) (pCCLhPGKGFP), a kind gift from Dr. Luigi Naldini (San Raffaele University, Milan, ITALY), which served as the backbone of all vectors. To avoid excessive manipulation of the parental LV (transfer), assembly of individual components of the expression cassettes was mostly done in the pGEM^® T-Easy (T-Easy) cloning vector system (Promega, Madison, WI) prior to cloning into the LV vector. LV vectors were, however, on some occasions directly manipulated to add or remove components.

For dual-cassette expression, a bovine growth hormone (bGH) poly(A) signal sequence (nucleotides 791..1015; GenBank Accession No. JQ624676.1) was used and termed as p(A)2, while the SV40 poly(A) sequence was on the 3’LTR of LV backbone and termed as p(A)1. The EF-GFP cassette was generated when a 1181bp fragment comprising the EF promoter (nucleotides 380 to 1560; GenBank Accession No. J04617) and enhanced GFP (nucleotides 6895 to 7611; GenBank Accession No. MN517551.1) was PCR-amplified from an in-house TW-EFGFP vector previously constructed ²⁷. The mCherry (Ch) coding sequence (CDS) (nucleotides 904 to 1611; GenBank Accession No. LC311025) was obtained from Dr. Roger Y. Tsien (University of California San Diego, CA, USA). For the construction of spacer-containing vectors (LVEsG-Sp*SvCh), a 119bp fragment of the ampicillin-resistant CDS (1337 to 2197 in the backbone of T-Easy vector) was excised at the ScaI-XmnI sites and inserted as single (Sp1) and tandem repeats (Sp2) into EcoRV-site LVEsG-SvCh vector.

Detailed description for the construction of various plasmids (10) and LV-based vectors (20) used in this work is shown in Supplementary Table 1 and Supplemental Material and Methods. Restriction enzymes used were obtained from New England Biologics Inc. (Ipswich, MA), together with In-fusion HD Cloning system (Takara Bio USA Inc., Mountain View, CA, USA) in some cases.

2.3. Lentiviral vector production

Infectious viral particles of each transfer vector were produced in human embryonic kidney (HEK) 293T cells (ATCC, Manassas, VA) using the third generation (four-plasmid) split packaging system by the calcium phosphate precipitation method as previously described²⁸. In brief, this involved transient transfection of 293T cells with a transfer vector plasmid together with three helper plasmids – pMDL.g/pRRE, pMD2.VSVG and pIL.VV01(Rev) – a kind gift from the L. Naldini lab. A mixture of the four plasmids in appropriate ratios in cold calcium chloride solution was made. HEPES-buffered saline was slowly added to the DNA mixture and mixed thoroughly to form a co-precipitate of DNA and calcium phosphate. The precipitate was slowly added to freshly seeded packaging (HEK293T) cells in tissue culture plates or flasks under constant swirling to ensure even distribution of the precipitate mixture. 12–18h after incubating (at 5% CO₂, 37C) the cells-precipitate mixture the transfection medium was replaced with Gibco^™ Dulbeco’s Modified Eagle’s Medium (DMEM) supplemented with 5% fetal bovine serum (FBS). Virus supernatant was harvested 24 hours after medium change. It was passed through 0.45μm pore-size PES filter membranes (GE Healthcare, IL) and stored in aliquots in −80C until needed.

2.4. Cell culture

All cell culture media were obtained from Thermo Fisher Scientific (Waltham, MA). HEK 293T, and murine erythroleukemia (MEL) and NIH-3T3 (3T3) cell lines were cultured in DMEM supplemented with 10% fetal bovine serum (FBS) (HyClone), 1% L-glutathione and 1% penicillin-streptomycin (Invitrogen). DAMI (Human megakaryocytic) cell lines were maintained in Iscove’s Modified Dulbeco’s Medium containing 10% FBS, and 1% penicillin-streptomycin while U937 (human monoblastic leukemia) and LCL (human lymphoblastoid) cell lines were cultured in RPMI 1640 medium containing 10% and 20% FBS, respectively 1% L-glutathione and 1% penicillin-streptomycin. Cells were sub-cultured every 3–4 days depending on confluency.

2.5. Virus transduction

Virus transduction was done as per an in-house protocol. Briefly, adherent cells were seeded (2 × 10⁵/well) in 24-well tissue culture plates (Corning Inc. NY) containing cell culture medium and 8ng/mL polybrene (Hexadimethrine bromide, Sigma, MO). Cells were transduced at different multiplexity of infection (MOI) so to generate transduced population with <25% transduction efficiencies. After 12–18hrs exposure to the virus (in 5% CO2; 37C incubator) the (transduction) vector-containing medium was carefully replaced with fresh culture medium and maintained for 14 days. Cells were then digested with Gibco^™ 0.05% trypsin-EDTA and reconstituted in medium for flow cytometry. Suspension cells were seeded (5 ×10⁴/well) in 48-well culture plates (Corning Inc. NY) in transduction medium with various MOI. They were then incubated for 12–18hrs after which they were centrifuged at 1200rpm for 1hr at room temperature. Cell pellets were reconstituted in fresh medium and maintained in culture for 14 days after which aliquots were assayed for expression determination by flow cytometry.

2.6. Flow cytometry

Transgene expression levels were quantified using BD Canto I (Becton-Dickinson, CA) and data was processed and analyzed using FACS Diva (Becton-Dickson, CA).

2.7. Informatics and transcription factors site-specific mutagenesis

Potential TF binding domains on the cC element were searched for using two online prediction software for mammalian transcription factors – Tfsite scan (http://www.ifti.org/cgi-bin/ifti/Tfsitescan.pl) and Transfac (Alibaba 2.1) (http://gene-regulation.com/pub/programs/alibaba2/). We evaluated the predictions of the two software tools and selected two unique domains both predicted by each software. To verify the function of the selected TF binding domains, we generated two mutation primer pairs each designed to harbor mutation in one of the binding domains. PCR amplification of the cC fragment with the two primer pairs produced two mutated cC fragments listed in Supplementary Table 2. Each mutated fragment was first cloned into a T-Easy vector and their sequences verified. Subsequently, each fragment was inserted into an EcoRV-linearized and dephosphorylated LVEsG-SvCh vector via EcoRV-PmeI excision from T-Easy vector. The resulting fragments were sequence-verified for their required orientation.

2.8. RT-qPCR analysis

Total RNA was isolated from cell lines in culture using the RNeasy Kit (Qiagen). RT-qPCR was carried out as described previously²⁹. After quantification with NanoDrop 1000 spectrophotometer (Thermo Scientific), 1μg of total RNA was reverse transcribed using oligo(dT)_12~18 as primer using a SuperScript™ II Reverse Transcriptase kit (Invitrogen) in a 20 μL reaction mixture. The resulting cDNA was appropriately diluted as template for quantitative PCR using iTaq Universal SYBR Green Supermix (Bio-Rad) and primers (see Supplementary Table 3). Expression of genes of interest were acquired and analyzed using ABI 7900 (Applied Biosystems) based on the protocol provided by the manufacturer. The PCR amplification condition included 2 min at 50°C, 10 min at 95°C, followed by 40 cycles of 15 s at 95°C and 1 min at 60°C. Glyceraldehyde 3 phosphate dehydrogenase (GAPDH) was applied as internal controls for mRNA quantification. Relative expression of mRNAs was determined by the 2^−ΔΔCT method³⁰.

2.9. Statistical analysis

Data were acquired as duplicates or triplicates in at least two repeated experiments and are presented here as mean ± SD (or SEM where expressly stated). Comparison between groups were done using One-way ANOVA with Tukey’s comparison test using GraphPad Prism version 7.0 for Windows (GraphPad Software, CA). P<0.05 was considered statistically significant.

3. RESULTS

3.1. Relative positioning and strength of promoters affect expression interference in dual-cassette vector

To assess the suitability of Es, cC and Sv as a promoter for transgene expression, we tested GFP and mCherry levels from a combination of promoters in cells stably transduced with various LV vectors (Figure 1A). The functionality of each promoter was validated using single-cassette control vectors in non-hematopoietic murine fibroblast NIH3T3, and hematopoietic human monocytic U937, lymphoblastic LCL, megakaryocytic DAMI and murine erythroid MEL cells by FACS analysis (Figures 1B). Without interference from adjacent TU, the mean fluorescence intensity (MFI) of GFP from Es promoter was high in both LCL and U937 cells (up to 23-fold of untransduced cells), followed by MEL (15-fold), DAMI (11-fold) and lowest in 3T3 (~7-fold) (Figure 1C). Conversely, the promoter activity of cC element was relatively higher than Es in 3T3 (~12-fold) but lower in all hematopoietic cell lines (<9-fold) (Figure 1C). When pairing Es and cC in-cis (Es+cC) in a dual-cassette LV vector, we observed a significant increase in Es-derived GFP expression in 3T3, U937 and LCL cell lines but no change in DAMI and MEL cells (Figure 1B and 1C). On the other hand, downstream mCherry expression was completely shut down in all the hematopoietic cells while significantly reduced in 3T3 (by 77% of control), suggesting a significant suppression of cC promoter activity by upstream Es-derived TU. Interestingly, when combined in divergent bidirectional LV (cCrev+Es), GFP expression was rather reduced substantially in all the hematopoietic cell lines (down to 16% of Es only) except in U937 where Es promoter activity was similarly higher than control (Es only) as that in unidirectional setting (Es+cC). On the other hand, mCherry expression from cC was significantly enhanced in all the hematopoietic cell lines (by 2.2 to 3.4-fold) compared to cC only (Figure 1C). These results show strength- and position-dependent interference (suppression) between Es- and cC-driven TUs and indicate a position-dependent enhancer activity of cC element to “dismal” Es promoter, most notably in 3T3 and U937.

Figure 1.

Promoter activities in dual-cassette expression are dependent on their relative strength and position. A. Schematic diagrams of SIN lentiviral vectors. Single-reporter LVs expressed either EGFP (GFP) from human EF-1α core promoter (Es only) or mCherry from core-CMV element (cC only). Dual-reporter LVs expressed both GFP and mCherry from Es and cC promoters. P(A)1 and p(A)2 are polyadenylation signal sequences from SV40 virus and bovine growth hormone, respectively. B. Representative flow cytometry dot plots showing dual-expression in various cells. C. Changes of median fluorescence intensity (MFI) of GFP and mCherry in stably transduced cells (<25% transduction frequency as determined by detectable reporter%), showing expression levels (transduced/untransduced cells) of Es with cC promoters in bi-directional and uni-directional designs. Red dotted lines, no expression observed. Bars represent mean ± SD, with n=4–6 wells derived from 2–3 independent transduction experiments. *p< 0.05, **p< 0.01, ***p< 0.001 by one-way ANOVA with Tukey’s comparison test; ns, not significant.

To explore whether the suppressing or enhancing effects between Es and cC is unique, we replaced the cC promoter in the unidirectional construct with Sv promoter (also small-sized, 269bp) (Figures 2A). In its single-cassette vector, Sv promoter exhibited higher activities than both Es and cC in all (but MEL) cell lines tested (Figures 2B and 2C). In the Es+Sv construct, mCherry expression was significantly decreased in all the cell lines, suggesting again that the suppression of downstream TU by upstream Es was position-related (Figures 2B and 2C). However, unlike in the Es+cC construct, GFP expression from Es promoter was either impaired by downstream Sv promoter in U937 (by 22%), DAMI (by 28%) and MEL (by 27%), or no change in 3T3 and LCL. The data demonstrated that the suppressive effect between promoters in dual-cassette vectors may be a factor of their comparative strengths and relative position, while the enhancement for upstream Es promoter activity may be restricted to the cC element and sensitive to cell types.

Figure 2.

The cCrev element enhances both upstream and downstream promoter activities with cell type-dependence. A. Schematic diagrams of lentiviral vectors. Sv, SV40 promoter; cCrev, core-CMV element in reverse orientation. B. Representative single-channel dot plots of GFP and mCherry expression in cells with different blood lineages by flow cytometry analysis. C. Changes of median fluorescence intensity (MFI) of GFP and mCherry expression in stably transduced cells (<25% transduction frequency). Red dotted lines, no expression observed. Bars represent mean ± SD, with n=6–12 wells derived from 2–4 independent transduction experiments. *p<0.05, **p<0.01, ***p<0.001 by one-way ANOVA with Tukey’s comparison test; ns, not significant.

3.2. cC element enhances expressions of both markers in Es+Sv dual-cassette vector

We then hypothesized that cC may function as an enhancer element to boost expression from distal Es promoter and to reduce the reciprocal suppressive effects of both promoter in the Es+Sv vector. To test this hypothesis, we positioned cC element between the two TUs and in reverse orientation (cCrev) so to avoid any uncertainties with promoter competition between cC and Sv which may disrupt mCherry expression or cause an additive strength of a cC-Sv promoter which may suppress upstream Es (Figure 2A). The resulting Es+cCrevSv vector had a relatively short promoter/enhancer size (cumulative of 601bp) (Table 1). With this vector, GFP expression was not only higher than those with Es+Sv in U937 (2.2-fold), DAMI (2.2), 3T3 (1.6) and MEL cells (1.3) but also further increased from those with Es only in U937 and DAMI cells, although no significant change was observed in LCL cells (Figure 2B and 2C, Table 1). Importantly, we also observed significant elevation in downstream mCherry expression across all cell lines, with the highest increase detected in U937 cells (2.3-fold of Es+Sv) followed by DAMI (2.2), 3T3 (1.8), LCL (1.7) and MEL cells (1.6). Such levels were either even higher than baseline SV40 promoter activity (Sv only) as in DAMI cells (1.4-fold), or comparable to baseline levels (without suppression) in all other cells except for LCL where mCherry expression was still significantly lower than baseline (Figure 2C). Considering relatively minor boosting effects of cC element on both GFP and mCherry expression in LCL, we chose not to include LCL for further mechanistic evaluation. The results showed that cC element could function as an enhancer in the Es/Sv dual-cassette construct to improve expression in most blood cell types tested.

Table 1:

Dual-cassette transgene expression from unidirectional vectors for promoter strength in stably transduced cells

Vector	Cumulative size of promoters/enhancers	Average GFP MFI a (fold ± SD) b				Average mCherry MFI a (fold ± SD) b
		3T3	U937	DAMI	MEL	3T3	U937	DAMI	MEL
Es+Sv	481 bp	6.8 ±0.8	17.2 ±2.2	8.0 ±1.0	11.2 ±1.4	23.0 ±3.2	16.2 ±0.5	9.8 ±1.9	5.0 ±0.5
Es+cCrevSv	601 bp	11.1 ±4.0	38.4 ±2.2	17.5 ±0.8	14.3 ±3.9	42.2 ±10.8	37.1 ±0.3	21.6 ±1.4	7.9 ±1.9
EF+Sv	1420 bp	16.8 ± 0.3	45.3 ±3.5	17.3 ±2.1	15.3 ±2.7	15.1 ±4.1	13.5 ±0.7	9.2 ±1.8	2.2 ±0.9

^aMFI, mean fluorescent intensity.

^bFold changes were derived from MFI of transductants (<25% transduction efficiency) and un-transduced cells within the same stably transduced populations (>14 days after transduction).

We also evaluated potential orientation specificity of cC element (forward vs. reverse) and the efficiency of dual-cassette expression from Es+cCrevSv as compared to those from EF+Sv (Figure 3A, Table 1). When positioned at the same location in the forward orientation (Es+cCSv), the cC element introduced either no enhancing activity (as in 3T3 and MEL cells) or even repressive effects (as in U937 and DAMI cells) to Es-derived GFP expression, indicating that the enhancer function of cC to dismal Es promoter was dependent on its orientation and cell types (Figure 3B). The downstream mCherry expression was similarly increased by cC element, regardless of orientation, to the levels of baseline Sv only in MEL and 3T3 cells but not DAMI cells, implicating orientation-independent rescue of promoter suppression. Surprisingly, in U937 cells, mCherry expression was further reduced by >80% in cells transduced with vector Es+cCSv compared to those with Es+Sv, suggesting that cC element would play a role as a repressing element in monocytic lineage when arranged in forward orientation. The full-length human EF1α promoter, containing the Es (~0.2 kb) with an intron (A) enhancer (~1 kb), has been widely used as a strong/sustained cellular promoter in both preclinical and clinical applications^31,32. We found that cCrev-enhanced GFP expression from weak Es promoter were comparable to those from full-length EF promoter for dual-cassette expression in all hematopoietic cell lines tested but not in the non-hematopoietic 3T3 cells (Figure 3B). Importantly vector Es+cCrevSv has much smaller cumulative size of promoter/enhancer elements than EF+Sv but exhibited similar (for upstream TU) or better (for downstream TU) promoter strength in dual-cassette expression (see Table 1). Thus, we concluded that cCrev element confers synergistic benefits in driving efficient dual-cassette expression from Es and Sv promoters in monocytic (U937), erythroid (MEL) and megakaryocytic (DAMI) cell lineages.

Figure 3.

Enhancer effects of cC on both upstream and downstream promoters are sensitive to orientation and cell types. A. Schematic diagrams of dual-expression LVs. EF, full-length human EF-1α promoter containing its intron A enhancer. B. MFI changes for GFP and mCherry in tested cell lines (<25% transduction frequency). Bars represent mean ± SD, with n=6–12 wells derived from 2–4 independent transduction experiments. * p<0.05, ** p< 0.01, *** p< 0.001 by one-way ANOVA with Tukey’s comparison test; ns, not significant.

3.3. The expression benefits of cC element are derived from its enhancer function rather than spacing effect

It has reported that increasing the space between tandem expression cassettes improved dual expression of DsRed and GFP markers from CMV and PGK promoters respectively in hair follicle-derived stem cells ¹⁷. We assessed whether spacer phenomenon would also contribute to the observed expression enhancement of cC by replacing it with a non-regulatory fragment obtained from the coding sequence of ampicillin resistance gene at the same (Sp1) or double (Sp2) sizes (Figure 4A). No improvement effects were detected from any of the spacer fragments on either GFP or mCherry expression in 3T3, U937, and MEL cell lines as compared to Es+Sv (Figure 4B). In DAMI, however, we observed significant increase of upstream GFP expression to the levels comparable to Es only but less than Es+cCrevSv. We concluded that enhancer function (protein binding), but not the spacing effect, may play a major role in cC-mediated enhancement for both upstream and downstream transgene expression.

Figure 4.

Evaluation of potential spacer effects of cC element for dual-cassette expression. A. Schematic diagrams of LVs expressing both GFP from Es promoter and mCherry from SV40 with cC (cCrevSv), a spacer fragment (Sp*Sv) or SV40 promoter alone. Sp* represents either a 119bp fragment from coding sequence of ampicillin resistance gene in pGEM T-Easy vector (Sp1) or tandem repeats of Sp1 (238 bp, Sp2). B. Change in median fluorescence intensity (MFI) of GFP and mCherry from dual promoters in stably transduced cells (<25% transduction frequency). Bars represent mean ± SD, n=6–12 wells derived from 2–4 independent transduction experiments. * p<0.05, ** p<0.01, *** p<0.001 by one-way ANOVA with Tukey’s comparison test; ns, not significant.

3.4. Lineage-restrict enhancing activities of cC are correlated with transcription factor binding and relative expression of SRF and RARα/NR1D1

Transcriptional enhancers, also known as responsive elements, boost promoter activity by recruiting their binding factors to the transcription site so to collaborate with the promoter core initiation complex³³. We then determined which transcription activator(s) may be involved in the lineage-restricted enhancer activities of cC element for both upstream Es and downstream Sv promoters. Utilizing two TF prediction algorithms, TfsiteScan and Transfac (Alibaba 2.1), we identified two presumed binding sites (S1 and S2) for non-universal TFs. Serum response factor (SRF) is assumed to bind at S1 (−30 to −21), and the retinoic acid receptor alpha (RAR-α) bind at the S2 (+33 to +42) with nuclear receptor subfamily 1 group D member 1 (NR1D1) (also known as Rev-ErbA) bind at the negative strand of the same S2 site (Figure 5A, Supplementary Table 2). To dissect potential significance of interaction between response domains within cC and these specific TFs, two dual-cassette LVs were constructed with cC element mutated at one of TF-binding sites by site-directed mutagenesis (Figure 5B, Supplementary Table 2). We evaluated expressions from these constructs together with Es+cCrevSv and Es+Sv in various stably transduced cells. Markedly, both mutated variants resulted in complete loss of the enhancing activities by cCrev element (as shown in Es+cCrevSv) for both upstream GFP and downstream mCherry in U937 cells, indicating the importance of these TF binding domains for promoter enhancement observed in monocytic cells (Figure 5C). Conversely, enhanced expressions for GFP or mCherry in murine 3T3 and MEL cells (all were moderate except for mCherry in 3T3) were not significantly affected by mutations at either of the binding domains, suggesting insignificance of the tested binding sites on its enhancer activities in fibroblastic and erythroid lineages. Interestingly in DAMI cells, the boosting ability of cCrev on upstream GFP expression was partially abolished by mutation at S1 (i.e., m1) and fully diminished by mutation at S2 (m2). For mCherry MFI, both cC variants presented significant reduction from those of wild-type cC, which were still higher than mCherry expression in Es+Sv, implicating cumulative functions of S1 and S2 for enhancer activity on Sv promoter. These results suggested that both predicted TF binding domains were essential in the enhancer functions of cC element for both upstream Es and downstream Sv promoters in human monocytic (U937) and megakaryocytic (DAMI) lineages.

Figure 5.

Enhancer function of cC depends on its binding sites for transcription factors. A. Diagram of predicted binding sites (S1 and S2) and their mutated sites (m1 and m2) for transcription factors in cC element. The location of cC (−50 to +70) are indicated with reference to tss of CMV IE gene. B. Schematic diagrams of LVs expressing GFP from Es promoter and mCherry from either SV40 (Sv), hybrid cC-SV40 (cCrev-Sv) or its mutated variants. C. Changes in mean fluorescence intensity (MFI) for GFP and mCherry expression from dual promoters in cells stably transduced with different LVs (<25% transduction frequency). Bars represent mean ± SD, n=6–12 wells derived from 2–4 independent transduction experiments. *p< 0.05, **p<0.01, ***p< 0.001 by one-way ANOVA with Tukey’s comparison test; ns, not significant.

To further assess the significance of S1 and S2 on the differential enhancer effects of cC element among the various cell lineages, we quantified the relative abundance of their associated TFs, i.e., SRF, RARα and NR1D1 by RT-qPCR analysis (Figure 6). The specificities of designed primer sets were confirmed by the dissociation curves of reactions which showed highly repeatable, single peak among different samples and cells (Figure 6A). In 3T3 fibroblastic cells, relative expression of RARα and NR1D1 was the highest; however, their binding at S2 site seemed inconsequential as shown in mutagenesis study, suggesting other factors may contribute to the observed improvement of mCherry with Es+cCrevSV in fibroblasts (Figure 6B and 6C). Among blood cell lines tested, all three TFs showed the highest expression in monocytic U937 cells (Figure 6B and 6C). These were consistent with the observation in the mutagenesis study that S1 and S2 were essential domains for robust enhancer effects of cC element in association with both distal (Es) and proximal (Sv) promoters. Additionally, megakaryocytic DAMI cells exhibited much higher expression of RARα and NR1D1 than MEL (4- and 1.5-fold) or LCL (3- and 5-fold) but similarly low levels of S1-binding SRF among them, which coincides with the more important role of S2 domain for cC enhancer activities found by mutagenesis evaluation in DAMI cells. Unsurprisingly, the lowest expression of all three TFs was observed in LCL cells where enhancer effect of cC was minimal. Taken together the data demonstrated that the differential enhancer effects of cC element among various blood cell types were closely associated with the cooperation of two specific domains within the cC element and the availability of their correspondent binding factors.

Figure 6.

Relative expression of cC element-related transcription factors in various hematopoietic cell lines. A. Representative dissociation curves of RT-qPCR for transcription factors SRF, RARα and NR1D1. GAPDH was used as reference gene for relative comparison. B-C. Relative mRNA levels of SRF (in B, for S1 site) as well as RARα and NR1D1 (in C, for S2 site) in different cell lines. Data were derived from 3 independent reverse transcriptions, each with qPCR in triplicate, and shown as mean±SD. * p < 0.05, ** p < 0.01, *** p < 0.001 by one-way ANOVA with Tukey’s comparison test; ns, not significant.

4. Discussion

In this study, we identified a cell-type restricted, promoter-enhancing function of a 120bp element (i.e., cC) from human CMV MIE promoter (−70…+50 bp) that has not been reported previously. The fragment is effective in boosting transcription from both downstream proximal promoter (Sv) and upstream distal promoter (Es) in cells stably transduced with a dual-cassette lentiviral vector (Table 1). Specifically, the cC element can increase Es promoter activity in all hematopoietic cells tested and reach levels as high as full-length EF promoter in monocytic U937 and megakaryocytic DAMI with orientation sensitivity. The element ascertained that downstream mCherry expression was sustained at baseline expression (without repression from upstream TU) in some cell types while exceeding baseline levels in others (DAMI). Moreover, site-specific mutagenesis studies demonstrated that responsive domains at two sites (S1 and S2) of the cC fragment play key roles in its enhancer/terminator activities detected. This observation was further supported by relative expression levels of correspondent transcription factors among different types of cells as determined by RT-qPCR. Notably, the small size of the cC element (120bp) is a desirable feature in vector design as reducing the cumulative size of regulatory elements is often associated with optimal titers or required for vectors with size restriction (i.e., AAV). The cC element represents a potent and cell type-restricted enhancer fragment that boosts the expression and reduces promoter interference in dual-cassette vectors.

Promoter interference, defined as the influence (mostly suppressive) of one transcriptional process by the transcriptional activity of a nearby promoter, is mostly undesirable in transgene expression, although it is a natural phenomenon of gene regulation in both prokaryotic and eukaryotic genomes³⁴. Several models have been proposed for this poorly understood phenomenon that may contribute to the transcription suppression observed, mainly related to the comparative strengths of the promoters and their positioning relative to each other. The “sitting duck” model refers to the dislodging of RNA Polymerase complex from a downstream weak promoter (TU) by the transcription complex of an upstream stronger promoter due to the weak promoter’s slow rate of transitioning from its transcription initiation complex. The interference in the unidirectional Es+cC construct, but not the bidirectional cCrev+Es, is consistent with this model. In single-cassette control LVs, the cC element (as a promoter) exhibited relatively ubiquitous but lower activity than upstream Es promoter in all hematopoietic cells tested, except for non-hematopoietic 3T3 cells in which cC was stronger (12-fold) than Es (7-fold) (Figure 1C). In the Es+cC vector, the stronger upstream Es-driven TU shut down the promoter activity of the weaker downstream cC in all hematopoietic cells, but not in 3T3 cells where Es reduced cC promoter activity. In the bidirectional cCrev+Es construct, mCherry expression from cC was not only detectable but even significantly higher than baseline cC promoter activity in all hematopoietic cells tested, suggesting beneficial effect of relative positioning/orientation when both TUs proceed away from each other. However, bidirectional cCrev+Es was less optimal in viral vector design as it consistently produced very low vector titer yields as compared to unidirectional Es+cC (data not shown), consistent with observations by others for bidirectional transgene expression^35,36. Surprisingly, despite the impaired activity of cC in the unidirectional LV, cC upregulated upstream Es which suggested its potential role as an enhancer element.

Another interference model is the promoter occlusion, i.e., transcription initiation in one TU is disrupted by the arrival of an elongation complex from another TU that interrupting the assembly of DNA (promoter)-binding factors³⁴. Occlusion is often the result of a lack of adequate terminating space for the upstream TU which therefore runs into the second TU. However, due to the rate of transcription, this interference can be brief and some promoter activity from the downstream unit may be reserved. Examples could be seen in the constructs of Es+Sv (Figure 2C) and EF+Sv (Figure 3B and Table 1). In both constructs the upstream promoter activity impaired downstream Sv promoter – the stronger EF promoter presenting more suppression – but without completely shutting down the high rate of transcription from the Sv promoter.

Finally, interference also occurs when promoters compete with each another for recruitment of binding proteins to responsive elements common to them^34,37,38. Depending on the abundance of a specific factor in a cell, competition may reduce availability of the factor for optimum use by two strong adjacent promoters simultaneously and therefore result in reduced expression from both TUs. The Es and Sv promoters have at least one binding site for the transcription factors Sp1 and CEBPα, while both EF and Sv have additional domains to bind AP-1, AP-2α, NF-kB, c-Jun and ETF. Thus, competition for binding proteins may contribute to the observations derived from Es/EF+Sv constructs where downstream Sv promoter activity significantly suppressed the upstream promoter in most of the hematopoietic cell lines (Figure 3). Interference between two adjacent TUs has been commonly observed and may not be adequately explained by the simplified models above⁹ as other factors beyond the models also contribute to the phenomenon.

Inefficient transcription termination also negatively affects both upstream and downstream TUs when an elongation complex runs into and disrupts the transcription complex of the downstream adjacent unit^16,39. Poly(A) sequences are essential in vector designs to facilitate transcriptional termination of each TU and enhance their nuclear transport, mRNA stability, and translation¹⁸. There are two elements important for the full functioning of poly(A) signal sequence – the hexanucleotide AAUAAA and a downstream GU or U-rich motif ¹⁵. Poly(A) sequence from SV40 contains both domains and proven to be one of the most efficient terminator ⁴⁰. The bGH poly(A) sequence has no recognizable GU or U-rich motif and is considered relatively weak even though it may have other motifs that make it “sufficient” for the termination process^15,41. Relatively inefficient termination of the GFP transcription complex by the bGH poly(A) may have contributed to the interference/reduction in expression observed for both GFP and mCherry TUs in the unidirectional constructs. It may also be a factor to the lower titer observed in the bidirectional constructs when mCherry transcription complex ran into the vector 5’ LTR interrupting transcription of full-length vector RNA genome. On the other hand, the strong SV40 poly(A) located in the 3’ LTR of all the vectors ensured efficient termination of the preceding TU and would likely prevent disruption to endogenous TUs at the site of vector integration into the host genome. The results thus highlight the importance of the choice of poly(A) in such vector designs and function¹⁵.

Other elements without consensus motifs have also been suggested to be important for the transcription termination process, such as terminator or pause sequences ⁴². They are thought to be either bound by proteins which form DNA-protein complexes or are intrinsic DNA sequences that form structural units to affect the site delineation for final disassembly of the termination complex⁴³. Some investigators have affirmed the functional importance of terminators in the efficiency of poly(A)-dependent termination with varied sequence motifs and sizes for different genes ⁴⁴. However, Orozco et al. and others have argued that apart from the poly(A) signal sequence no other regulatory element is required except for sufficient space to process the termination signal^13,14. Sufficient spacing was also found to benefit dual-cassette expression by reducing interference between the TUs as it allowed both TUs to effectively complete their transcription processes independently¹⁷. The potential spacing effect of cC element in boosting both TUs was evaluated using scramble spacer fragments and resulted in no or minimal enhancing benefits in all cell types tested (Figure 4B). The prospective terminator function of cC, where DNA-binding factors block the run-through of the upstream transcription complex, is likely to be associated with NR1D1 which has a single binding site (S2) on the positive strand of cCrev in Es+cCrevSv (i.e., negative strand of cC element). NR1D1 regulates transcription in several biological processes such as coordinating the circadian rhythm, metabolic pathways and cell differentiation in a heme-dependent manner⁴⁵ in multiple cell types. For example, it has been reported to repress expression of inflammatory cytokines in macrophages and reduce the severity of peritoneal inflammation. Importantly, this factor functions as a transcriptional repressor only when binding as a homodimer to tandem repeats of its recognition sequence within promoter regions but remains inactive as a monomer ⁴⁶. In the Es+cCrevSv construct therefore, we speculate that NR1D1 may likely function as a DNA-binding protein to facilitate efficient termination of the GFP transcription complex, block a run-through to the downstream mCherry TU and contribute to the enhanced expression observed in U937 and DAMI cells (Figure 2C). Such NR1D1-mediated termination benefits (for both up- and down-stream TUs) would be orientation-sensitive (strand-related), which is consistent with the loss of improvement on both TUs when cC was placed in the sense orientation as in Es+cCSv (Figure 3B). This is also supported by the relatively high mRNA levels of NR1D1 found in U937 and DAMI cells by RT-qPCR.

Interestingly, the addition of forward-orientated cC element in Es+cCSv further reduced mCherry expression as compared to Es+Sv in U937 cells (Figure 3B). This observation could be related to promoter suppression reported in double-promoter monocistronic expression vector⁴⁷. Colella et al. evaluated promoter strength in various cells or organs by combining two promoters in a tandem arrangement of liver-, and muscle- and neuron-specific promoters in AAV vectors⁴⁷. The combination of liver- (hAAT) and neuron-specific (hSYN) promoters resulted in significant reduction of transgene expression as compared to single-promoter alone in hepatic and neuronal cell lines. However, protein expression was synergistically increased in hepatic or myoblast cells when liver- and muscle-specific (spC5.12) promoters were placed side-by-side. Thus, the outcome of double-promoter, either enhancement or suppression, is likely dependent on characteristics of the promoter pairs and particular cell types, although the specific mechanisms remain undefined.

The full-length CMV promoter has a known enhancer region between −550bp and −39bp relative to the transcription start site +1 ²⁰ from which different fragments (distal and proximal) have been used in the construction of hybrid promoters. For example, a 288bp CMV enhancer fragment (−517bp to −230bp) has been combined with different variants of chicken β-actin promoter in CAG or CBh ^21,22,48, and the fusion of a 380bp CMV enhancer fragment (−541bp ~ −160bp) with PDGFb promoter formed the neuron-specific CPDGFb²³. Whereas the 120bp cC element (−50bp ~ +70bp) studied here has only been used as a minimal promoter with very low baseline expression in combination with responsive elements to drive inducible transgene expression ^25,26. For instance, the cC element functioned as the minimal promoter in combination with a tetracycline-response element for expression of transgenes in CHO cells²⁵. Recently, synthetic promoters containing the cC fragment and various regulatory elements responsive to different activated TFs were generated for ligand-induced reporter expression, resulting in high signal-to-noise ratios in response to their specific ligands in human cell lines ²⁶. In our studies, however, we found a novel role of the 120bp cC fragment as an enhancer element that improves lineage-restricted dual-cassette reporter expression in monocytic (U937) and megakaryocytic (DAMI) cells.

In this report, we identified two key domains responsible for the cell-type restricted dual-enhancer function of cC fragment as demonstrated by the site-specific mutagenesis (Figure 5) and supported by the relative expression among cells of related transcription factors, SRF and RARα (Figure 6). The predicted SRF is considered an essential transcription factor for mature myeloid cell functions⁴⁹ and megakaryopoiesis⁵⁰, which was consistent with the RT-qPCR data showing (relatively) highest expression of SRF in U937 cells. Mutation to the SRF binding domain in the cC element (Es+cCrevm1Sv) completely abolished enhancer function of cC for both TUs in human monocytic U937 and partially reduced in megakaryocytic DAMI cells (Figure 5C), further supporting functional significance of SRF-binding site in the observed lineage-restricted enhancer activity. The RARα protein has been reported to play important roles in myeloid differentiation⁵¹. Its expression pattern in the hematopoietic cells tested, where it was most abundant in U937 and DAMI but not in MEL and LCL (Figure 6C), was also correlated with the observation that cC element exhibited stronger enhancer effects in U937 and DAMI. Moreover, the fact that the binding sites for RARα and NR1D1 (both showed high expression in U937 and DAMI cells) were overlapped and on the opposite strands of responsive domain (S2) would likely provide a possibility of synergistic outcome from both RARa binding for enhancer activity and NR1D1 binding for terminator effect. This, in turns, may compensate for their lower expression levels than SRF in U937 cells. The mutation to the S2 site in cC (in Es+cCrevm2Sv) abrogated all enhancer benefits for both TUs in U937 and DAMI except for downstream mCherry expression in DAMI cells (Figure 5C), demonstrating the importance of this domain on the lineage-restrictive enhancer function observed. Additionally, LCL cells presented the lowest expression levels of all three factors among hematopoietic cells tested, which was consistent with the minimal effects of cC found in LCL cells. Our results do not exclude the possibility that other co-factors may cooperate with SRF, RAR and NR1D1 at these sites, or the involvement of other untested responsive domains (especially for 3T3 cells). However, the data highlight the importance of two domains within cC elements and their correspondent factors on improving promoter activities of dual-cassette expression and substantiate the preference of such enhancer function in monocytic (U937) and megakaryocytic (DAMI) lineages.

In summary, we have documented a small fragment within the major immediate early promoter region of human CMV which functions as a terminator sequence and an enhancer element in a dual-cassette lentiviral vector. This 120bp fragment not only prevented promoter reciprocal suppression but also further increased transgene expression particularly in monocytic (U937) and megakaryocytic (DAMI) cells from two short promoters – Es and SV40 promoters. Recently, the Es promoter has shown in vivo efficacy and safety in a gene therapy clinical trial⁵², although it is much less potent than EF promoter. Markedly the enhancer effect of cC on the Es promoter resulted in gene expression levels comparable to those from the full-length EF1α promoter. Importantly, the cC element significantly reduces the cumulative size of the regulatory elements in the dual-cassette vector without compromising functionality and would therefore be beneficial for vectors with limited size- or packaging-capacity. The specific regions of the Es and Sv promoters that interact with the cC fragment remain to be elucidated, which may shred lights on identifying other promoters (especially tissue-specific types) with similar motifs to be paired with the cC fragment for advances potency. Overall, the novel design of a potent and lineage-restricted dual-cassette vector developed herein would be constructive to the growing repertoire of robust tools for multi-gene expression in hematopoietic cell-mediated gene and/or cancer therapy when co-expression of multiple CARs is needed.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author on reasonable request.