Creation of a synthesis-friendly inflammation-inducible promoter suitable for cell therapy

Abstract

The development of ‘smart’ cell-based therapeutics requires cells that first recognize conditions consistent with disease (e.g. inflammation) and then subsequently release therapeutic proteins, thereby reducing potential toxicity from otherwise continuous expression. Promoters containing NF-κB response elements are often used as reporters of inflammation; however, endogenous promoters have crosstalk with other pathways, and current synthetic promoters have many exact sequence repeats of NF-κB response elements which make them both difficult to synthesize and inherently genetically unstable. Herein, a synthesis-friendly inflammation-inducible promoter (named SFNp) was created by the packing of 14 NF-κB response elements, which have no repeats >9 bp, followed by a minimal cytomegalovirus promoter. In stably expressing human embryonic kidney 293 cells, we assessed the ability of SFNp to inducibly transcribe genes for reporting expression, changing cell morphology, and performing cell fusion. These experiments represent simple milestones for potentially using SFNp in the development of cell-based therapeutics. As strongly repeated DNA can compromise the long-term stability of genetic circuits, new designs used in ‘smart’ cell therapy will become more reliant on synthesis-friendly components like SFNp.

Insight Box

Using cells as therapeutics requires the construction of genetic promoters to induce gene expression in different physiological environments. A primary concern for many current inducible promoters is their inclusion of sequence repeats, which can increase the genomic instability of the promoter and compromise gene expression. Herein, we describe a novel inducible promoter that includes no exact sequence repeats and can mediate gene expression to program cellular behavior. We were able to program reporter gene expression, induce morphological changes, and mediate viral gene expression in cells containing our promoter. These programmable behaviors can have utilities in cell-based therapeutics and make our novel promoter a valuable addition to the toolbox of genetic devices, which may be used in therapeutic development.

INTRODUCTION

Advances in cellular and genetic engineering have enabled the development of ‘smart’ cell therapies for disease treatment [1]. A critical requirement of these ‘smart’ cell therapies is the conditional expression of therapeutic transgenes that have been integrated into the host cell genome and only expressed when activated by signaling cues from the microenvironment (e.g. inflammatory cytokines at the site of injury) [1]. Inducible promoters can allow this conditional expression of transgenes by the binding of signal-specific transcription factors to the promoter. The advantage of these inducible promoters is that they allow for greater regulation on therapeutic transgene expression to conditions where they are needed, thereby minimizing the side effects caused by constitutive expression [2]. In particular, the NF-κB family of transcription factors (e.g. p50, p52, p65, RelB, and c-Rel transcription factors) are responsible for the induction of endogenous inflammatory genes [3, 4], and promoters that have NF-κB response elements are commonly used for studying inflammation-inducible expression [5–7]. Upon dimerization, the NF-κB transcription factor binds to NF-κB response elements on promoters and recruits transcriptional machinery to initiate gene transcription [4]. These promoters with NF-κB response elements are frequently used as reporters of inflammation through their detection of inflammatory cytokines such as tumor necrosis factor alpha (TNF-α) [8].

To design a functional promoter, specific promoter elements are needed for the recruitment and binding of transcriptional machinery [9]. Within the sequence of a functional promoter is a core promoter which contains specific sequence motifs and the transcription start site, where the pre-initiation complex begins mRNA synthesis. For example, the minimal cytomegalovirus (CMV) promoter is the most commonly used core promoter and includes the sequence motifs for the TATA box (5′ TATAWAWR 3′) and the Initiator element (5′ YYANWYY 3′), which are bound by the TATA-binding protein and transcription factor II D, respectively [9–11]. Since the core promoter alone has negligible expression, it requires upstream enhancer sequences that bind transcription factors to recruit transcriptional machinery for more significant expression. If the binding of the transcription factor on the enhancer regions depends on specific environmental cues, then there will be an inducible increase in expression levels compared to the core promoter alone. For inflammation-induced expression in mammalian cells, this process is mediated by the NF-κB transcription factors, which dimerize and subsequently bind to NF-κB response elements on endogenous promoters.

While it is possible to use endogenous promoters (e.g. the IL-6 promoter [12] or iNOS promoter [13]) that have NF-κB response elements, these promoters contain other response elements that may be inducible by other pathways. To avoid these cross-pathway effects, it may be preferable to design synthetic promoters with only NF-κB response elements. Crystal structure analyses of the DNA binding domain of NF-κB revealed contacts with major and minor grooves of the DNA helix at a 10-bp response element [14]. For example, commonly used commercial NF-κB promoters were designed to include several repeats of NF-κB response elements (5′-GGGRNWYYCC-3′) placed upstream of a minimal promoter [5–7, 15]. However, these response elements, combined with flanking subsequences, are often exact sequence repeats, making them difficult to synthesize by Gibson assembly [16]. More concerning, these exact sequence repeats make the genetic material unstable due to a greater potential for replication slippage, misalignment, and homologous recombination [17]. Thus, to allow more economical de novo gene synthesis and more stable genetic material, a synthesis-friendly alternative was designed (hereafter, SFNp) that had no repeats >9 bp or extreme GC ratios that would be difficult for Gibson assembly [16]. To demonstrate SFNp as a viable alternative, we showed that our promoter can be induced to express proteins for fluorescence imaging, pathway rewiring, and cell fusion. Our experiments demonstrate the use of SFNp as an alternative to current, commercially available NF-κB promoters.

RESULTS AND DISCUSSION

Design of a synthesis-friendly inflammation-inducible promoter

A synthesis-friendly inflammation-inducible promoter (SFNp) was created by the packing of 14 NF-κB response elements (i.e. 5′-GGGRNWYYCC-3′) that have no repeats >9 bp followed by a minimal CMV promoter (CMVp) (Fig. 1A). The NF-κB response elements provide docking sites for the NF-κB transcription factor, while the minimal promoter allows the initiation at the desired transcriptional start site. Dot plot analysis of the commonly used commercial NF-κB promoter from the plasmid pGL4.32 (Promega) shows repeated segments (i.e. diagonal lines on the dot plot) arising from the 5 NF-κB response elements and flanking sequence, which are identical to each other (Fig. 1B, right). Since the consensus sequence of the NF-κB response element has some flexibility in the sequence space, it is not necessary to have exact sequences repeated. Using this flexibility in the sequence space, our SFNp includes 14 NF-κB response elements, while having GC content in the synthesis-friendly range of 25–75% and no repeats >9 bp (Fig. 1B, left; Fig. 1C left). High GC content (e.g. 80% is considered very high) is undesired for gene synthesis because PCR itself becomes inefficient [18]. Neither our SFNp nor NF-κB promoter from the plasmid pGL4.32 has high GC content (Fig. 1C). Through this promoter design, our SFNp acquired nine more NF-κB response elements than the NF-κB promoter from the plasmid pGL4.32. Previous studies into NF-κB promoters have shown that inducibility improves with more NF-κB response elements that are more tightly packed [19].

Figure 1 .

(A) SFNp sequence. NF-κB response elements with consensus sequence 5′ GGGRNWYYCC 3′ are underlined beneath sequence. TATA Box with sequence 5′ TATATAA 3′ is underlined and labeled. Initiator sequence 5′ TCAGAT 3′ is underlined and labeled Inr. Start codon ATG is located at the end of the sequence. (B) Dot plots demonstrating sequence repeats for SFN promoter (left) and commercial promoter from the plasmid pGL4.32 (right). Window size was 10 bp. (C) GC-content plot for SFNp (left) and reference promoter pGL4.32 (right). SFNp resides within the synthesis-friendly range of 25–75%. Window size was 35 bp.

SFN can inducibly express a fluorescent protein when stimulated by tumor necrosis factor alpha

To assess the ability of SFNp to inducibly express a reporter gene, we created human embryonic kidney (HEK)293 cells expressing the cyan fluorescent protein cerulean (hereafter, Ceru) under the control of SFNp, which was stably integrated via lentiviral infection (Fig. 2A). Since previous studies used HEK293 cells in the characterization of NF-κB promoter activity [5–7, 15], this cell line was chosen for our study as well. Furthermore, the HEK293 is a standard cell line for gene expression studies due to their transfection efficiency, fast growth rate, and efficient protein production [20]. To test that SFNp behaved similarly to the commercial NF-κB promoter in pGL4.32, transient transfection experiments were performed in HEK293 cells upon induction with active tumor necrosis factor alpha (TNF-α) (Supplementary Fig. S1). Our stably expressing HEK293 cells were serially diluted to select a colony grown from a single cell, thereby ensuring genotypic and phenotypic uniformity and a consistent behavior in response to stimulation. To induce expression of Ceru, these cells were stimulated by active TNF-α, which then interacts with the monomeric TNF receptors (TNFRs) [21] to induce TNFR trimerization followed by NF-κB-mediated transcription (Fig. 2B). Prior to TNF-α stimulation, there was no visible fluorescence from the cells (Fig. 2C). However, 24 h after stimulation with TNF-α, all stable HEK293 cells within the field of view displayed cyan fluorescence (Fig. 2D). As expected, we observed that the cyan fluorescence was distributed throughout the cell and not localized to any specific cellular compartment (Fig. 2E). This cell-wide distribution is due to the lack of localization sequences and the relatively small size of Ceru (26.8 kDa). Lastly, the expression was reversible by removal of the stimulus and subsequent re-culturing of cells. Using this stable cell line, a dose response to TNF-α was performed, showing responsiveness in physiological relevant concentrations [22] (Supplementary Fig. S2). The ability of SFNp to inducibly express Ceru upon TNF-α stimulation indicates that SFNp may be able to express other proteins of interest as well.

Figure 2.

(A) DNA construct for Ceru expression under the control of SFNp for HEK293 lentiviral infection. (B) Cartoon of expression and subcellular localization of Ceru regulated by SFNp in response to TNF-α stimulation. (C) Brightfield (left) and fluorescence image (right) of stable HEK293 cell lines without TNF-α stimulation (10× magnification). Fluorescence image of stable HEK293 cell lines 24 h after TNF-α stimulation at 10× magnification (D) and 40× magnification (E). Scale bar represents 100 μm (C, D) and 10 μm (E). Ceru fluorescence is pseudocolored cyan.

SFN can inducibly express engineered Ca²⁺-sensitive RhoA (CaRQ) for pathway rewiring

To assess the ability of SFNp to rewire cell morphology changes, we created stable HEK293 cells expressing our previously engineered Ca²⁺-activated RhoA (CaRQ) under the control of SFNp (Fig. 3A). Our group engineered CaRQ as a Ca²⁺-activated RhoA that causes cellular blebbing in HEK293 cells in response to a cytosolic Ca²⁺ increase [23]. When CaRQ is paired with receptors that generate a Ca²⁺ increase after receptor-ligand binding, directed cell migration can be rewired to that ligand [23]. For fluorescence imaging, CaRQ has a C-terminal fusion with the yellow fluorescent protein Venus; for increased Ca²⁺ sensitivity, CaRQ has an N-terminal fusion with the membrane localization tag from Lyn kinase (i.e. ¹MGCIKSKGKDSA¹²). As stated previously, all our stably expressing HEK293 cells were serially diluted after lentiviral infection to select a colony grown from a single cell, thereby ensuring consistent behavior in response to stimulation. After 24 h of TNF-α stimulation, these stable HEK293 cells expressed CaRQ as observed by plasma membrane localization of Venus fluorescence in every cell (Fig. 3C and D). Since the Lyn membrane tag requires the first methionine to be cleaved and lipid-modified to properly tether on the plasma membrane, the observed membrane localization of CaRQ means transcription occurred at SFNp rather than by read-through expression. As expected, shortly after stimulation with ATP that caused a cytosolic Ca²⁺ increase, CaRQ is activated as observed by dynamic blebbing at the plasma membrane (Fig. 3E and Supplementary Video S1). As expression under SFNp retained the activity of CaRQ, a ‘smart’ cell-based therapy could theoretically be developed with SFNp that recognizes an inflammatory microenvironment and then expresses a multi-component gene circuit to rewire cell migration for seeking local pathogenic signals (e.g. gm-csf [24] or IL6 [25]). To demonstrate this cell migration possibility for gm-csf, we tested whether induced cells transfected with GMRchi [24] could pass through a gradient of gm-csf in a Boyden-chamber transwell migration assay (Supplementary Fig. S3). Only cells induced by TNF-α and transfected with GMRchi for rewiring gm-csf to Ca²⁺ signaling could direct migration across the transwell.

Figure 3.

(A) DNA construct for expression of plasma membrane-tagged CaRQ-Venus fusion protein under the control of SFNp for HEK293 lentiviral infection. pm is the membrane localization tag from Lyn kinase. (B) Cartoon of the expression, subcellular localization, and activity of CaRQ mediated by SFNp in response to TNF-α-induced expression and ATP-induced Ca2+ release followed by CaRQ-mediated cellular blebbing. Brightfield (top) and fluorescence image (bottom) of stable HEK293 cell lines without TNF-α stimulation (C) and after 24 h of TNF-α stimulation (D) at 10× magnification. Scale bar represents 100 μm. (E) Fluorescence images of TNF-α stimulated HEK293 cells taken from time-lapse at 40× magnification following bolus addition of 10 μM ATP at 4 min and 30 s. Individual blebs are denoted by white arrows. Scale bar represents 10 μm. Venus fluorescence is pseudocolored green.

SFNp can inducibly express vesicular stomatitis virus G protein for cell fusion

To further extend the functionality of SFNp into potentially therapeutic behavior, we created stable HEK293 cell lines expressing vesicular stomatitis virus G (VSVG) protein under the control of SFNp and nuclear-exported Ceru under the control of the CMVp (Fig. 4A). VSVG is the envelope protein that facilitates the entry of the vesicular stomatitis virus into the cell by binding to a putative target receptor (i.e. low-density lipoprotein receptor) and subsequently mediates low-pH-dependent membrane fusion during endocytosis [26, 27]. For visualization of syncytia (i.e. multinucleated cells resulting from cell fusion), Ceru was C-terminally fused with a nuclear export signal (NES) from the HIV-1 Rev protein (i.e. ¹LQLPPLERLTLD¹²). When stimulated with TNF-α for 24 h, the stable HEK293 cells expressed VSVG and its cell fusion function was detected in a low-pH environment (i.e. pH: 6). Upon association with an opposing cell membrane at a moderately acidic pH of 6, VSVG mediates membrane fusion, creating a multinucleated cell (Fig. 4B). As expected, syncytia were observed by the clustering of the nuclei and a uniform cytoplasmic Ceru fluorescence distribution after low pH stimulation, but not before (Fig. 4C). In the absence of TNF-α pre-stimulation, the stable HEK293 cells will not undergo any cell fusion. Since NF-κB is often upregulated and activated in the tumor microenvironment due to surrounding tissue damage [4, 21], we could potentially engineer a therapeutic cell that detects local inflammation using SFNp to express VSVG for cell fusion and subsequent death [28] with tumor cells in the microenvironment which typically has a low pH due to hypoxia [29, 30].

Figure 4.

(A) DNA construct for inducible VSVG expression under the control of SFNp and for constitutive Ceru expression under the control of CMVp for HEK293 lentiviral infection. (B) Cartoon of the expression and subcellular localization of VSVG and Ceru-NES mediated by SFNp and CMVp, respectively. VSVG expression occurs in response to TNF-α stimulation. Cell fusion to form a multinucleated cell occurs in response to addition of a +6 pH buffer. (C) Fluorescence microscopy image of cells before low pH stimulation. (D) Fluorescence microscopy image of cells after low pH stimulation. Scale bar represents 10 μm. Ceru fluorescence is pseudocolored cyan.

CONCLUSION

Herein, we have designed and characterized a synthesis-friendly synthetic promoter (SFNp), which is inducible through the NF-κB signaling pathway. SFNp encodes 14 NF-κB response elements and no sequence repeats >9 bp while keeping the GC content within the synthesis-friendly range of 2%–75%. We have also demonstrated the utility of SFNp to inducibly drive reporter expression with Ceru, rewire cell morphology changes with CaRQ, and perform cell fusion with VSVG. These applications demonstrate the potential for SFNp to inducibly alter cell behaviors, such as migration and therapeutic protein expression, both of which may have clinical application in the development of cell-based therapeutics. For example, an engineered cell could detect inflammation in the tumor microenvironment [4, 21] and express a CaRQ-based gene circuit to rewire cell migration for seeking local pathogenic signals or express VSVG for cell fusion and subsequent suicide with tumor cells. Future studies should focus on demonstration of SFNp inducibility in cellular therapeutics in vivo. Since strongly repeated DNA is inherently more unstable and therefore more corruptible over time, genetic circuits used in ‘smart’ cell therapy will become more reliant on synthesis-friendly components like SFNp.

MATERIALS AND METHODS

Promoter design

NF-κB response elements with the consensus sequence 5′ GGGRNWYYCC 3′ were used to design the enhancer region of SFNp. Sequence repeats were identified using the EMBOSS dottup online tool with a window size of 10 bp, plotting each sequence against itself. GC-content plots were generated using the EMBOSS Explorer isochore online tool, with a window size of 35 bp and a shift increment of 0.

Plasmid construction

All synthesis and subcloning of plasmids was done by Genscript. The SFNp sequence is: GGATCCACGGGATACCCCAGGGGCTCTCCAGGGAATCTCCGGGGATACTCCAGGGGGTTTCCGGGGAATCCCCCGGGAGTTTCCTGGGAATTTCCCGGGATTTCCCCGGGGCATCCCGGGGACTCTCCTGGGATTTTCCAGGGACATTCCTGGGACTTTCCTGCGCGGTAGGCGTGTACGGTGGGAGGTCTATATAAGCAGAGCTCGTTTAGTGAACCGTCAGATCGCCTGGAGACGCCATCCACGCTGTTTTGACCTCCATAGAAGACACCGGGACCGATCCAGCCTCTCGACATTCGTGCCACCATG. Similarly, the Ceru, CaRQ, Venus, and VSVG sequences were human codon-optimized, inserted into the SFNp plasmids by Genscript.

Stable cell culture creation

The above designed plasmids were used to create stable HEK293 cells by lentiviral infection as previously described [31]. Cells were selected using 1 μg/ml of puromycin resistance, and colonies were derived from single cells after serial dilution in 96-well plates, to ensure genotypic consistency. Individual colonies were screened for inducibility after a 24-h stimulation with 10 ng/ml of active TNF-α (Cedarlane).

Imaging

Prior to imaging, cells were plated in 96-well glass-bottom plates (Mattek). Images were taken with the Olympus IX81 microscope by using a Lambda DG4 xenon lamp for the light source and a QuantEM 512SC CCD camera with a 10× objective or 40× objective (Olympus). Excitation (EX) and emission (EM) filter bandpass specifications were as follows: for fluorescent proteins: Ceru (EX: 438/24, EM: 482/32) and Venus (EX: 500/24, EM: 524/27) (Semrock). Images were analyzed via ImageJ and μManager software. Time-lapse images were taken 10 s apart. Imaging was conducted with cells maintained in PBS. Blebbing was induced in HEK293 cells expressing CaRQ with stimulation of 10 μM ATP. Cell fusion was induced in HEK293 cells expressing VSVG by replacing growth media (DMEM +10% FBS) with a pH 6 PBS buffer for 1 min and then returning back to the growth media.

Boyden-chamber assay

We used the Boyden-Chamber assay to test cell migration across a porous membrane. For all experiments, 24-well format transwells (8-μm pore size; Greiner) were used. Migrating cells were quantified visually by fluorescence microscopy. The migration index was calculated by dividing the amount of cell migration in the sample with 100 ng/ml of gm-csf by the amount of migration in a well with no gm-csf (basal level of cell migration).

Statistical analysis

All data with normal distribution and similar variance were analyzed using a one-factor ANOVA with Tukey–Kramer post-hoc test. The α was set at 0.05, and P-values <0.05 were considered significant. Data were expressed as mean ± s.d unless otherwise stated.

Authors’ contributions

The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript.

Funding

This work was supported by the Canadian Institutes of Health Research (#PJT-156317) and the Natural Science and Engineering Research Council NSERC (#RGPIN-2019-04183).

List of abbreviations

NF-κB, SFNp, CMV, HEK293, TNFα, IL6, TNFR, CaRQ, GMCSF, VSVG, NES, ATP, DMEM, PBS, FBS

Competing interests

The authors declare no financial or competing interests.

Supporting information

The following files are available free of charge: Supporting Document: Supplemental Figures 1–3.

Conflict of interest statement

The authors declared no conflicts of interests.