Demonstrating proof of concept
1. Our challenge
Protecting healthy tissue from off-tumor toxicity is a major challenge facing all strategies for cancer immuno-gene therapy. Implementing logic AND gates to strictly confine the expression of therapeutic genes or their effects to selected target cells is an intriguing concept in this direction. In a 2017 Cell paper (1), L. Nissim and colleagues in the Timothy Lu team at the MIT presented a revolutionary AND gate device comprising a two-module RNA-based circuit. Using this circuit they reported the expression of a multi-component immunostimulatory cassette only in cancer cells in which two distinct promoters are active and demonstrated a potent in-vivo antitumor response elicited by the targeted expression of this cassette.
In an interview with MIT News (October 19, 2017), Martin Fussenegger from ETH Zurich, Switzerland, a world leading scientist in the field of synthetic biology, who was not involved in this research, says: “This advance will open up a new front against cancer. First author Lior Nissim, who pioneered the very first genetic circuit targeting tumor cells, has now teamed up with Timothy Lu to design RNA-based immunomodulatory gene circuits that take cancer immunotherapy to a new level. The design of this highly complex tumor-killing gene circuit was made possible by meticulous optimization and integration of several components that target and program tumor cells to become a specific prey for the immune system - this is very smart technology.” For a broader context and significance of this new technology see also a recent review by Xie and Fussenegger on synthetic gene circuits (2).
In the same report in MIT News, Lu, who led this research, clarifies: “There has been a lot of clinical data recently suggesting that if you can stimulate the immune system in the right way you can get it to recognize cancer. Our belief is that there is a need to develop much more specific, targeted immunotherapies that work locally at the tumor site, rather than trying to treat the entire body systemically.”
Fig. 1 depicts the basic design of the two modules comprising the ‘Nissim’s’ system. The expression of the gene of interest (GOI), whether an experimental fluorescent protein such as mKate2 or a master transcription factor (TF) that controls the expression of multiple therapeutic genes, is governed by the coordinated activity of two separate modules, each driven by one of two synthetic promoters, P1 and P2. Module 1 comprises an mRNA transcript driven by promoter P1, harboring two exons which encode the intact protein product of the GOI. Upon splicing, a sequence in the intron separating these exons functions as a microRNA (miRNA, here miR1) which targets the 3’ end of module 1 transcripts in-trans for degradation, thus preventing translation. The transcript of module 2 is synthesized from promoter P2 and contains multiple miR1 reverse complementary binding sites, which function as ‘sponge’ sequences that relieve the autoinhibition of module 1.
By carefully selecting the two promoters, P1 and P2, the authors then employed this system to confine the expression of a four-component immunomodulatory cassette to cancer cells as the only cells in the body in which both promoters P1 and P2 are fully active.
Fig. 1 (copied from Fig. 2 in (1)). The Basic RNA-Only Single-Output AND Gate Design. (A) Module 1 of the AND gate is designed as an autoinhibitory loop such that it represses its own output. Module 2 is designed to relieve the autoinhibition of module 1. Module 1 and module 2 are regulated by cancer-specific promoters P1 and P2, respectively. The output from module 1 is expressed at a high level only when both P1 and P2 are active, which enhances the tumor specificity of the circuit. (B) Only state [1,1] is shown. In this state, the P2 promoter expresses multiple copies of a sponge RNA (light green) which binds miR1 and prevents its binding to its target sequence at the 3’ end of module 1 (dark green). This enables sequestration of miR1 away from inhibiting the mKate2 transcript expressed by P1, thus allowing for mKate2 expression.
Critically, safety of this system relies on complete destruction of an intact mRNA encoding a potentially lethal product by a mechanism operating in-trans (that is, miR1), which may fail to fully accomplish this task. For example, in the particular immunotherapeutic application explored by Nissim et al., expression of the TF and, consequently, the entire immunostimulatory cassette in ‘prohibited’ non-cancer cells, would result in their direct killing by cytotoxic T lymphocytes.
Indeed, as shown by Nissim and colleagues ((1) and see Fig. 2), in spite of laborious optimization attempts, at least 15% of the product of module 1 (driven by promoter P1) that is synthesized in the presence of the sponge (module 2, driven by promoter P2) can still be detected in the absolute absence of the sponge (‘Cont.’ which stands for ‘control’). The default state of module 1 is in fact not ‘0’, as desired, but rather an incompletely attenuated ‘1’. A severe safety concern therefore emerges. Lu’s statement in the same MIT News report, maintaining that “only when two of these cancer promoters are activated, does the circuit itself switch on” therefore reflects an obvious prerequisite for therapeutic application of this system and an expectation of its inventors, rather than its actual performance.
Fig. 2. The risk posed by the sponge strategy (Fig. 3 in (1)). The strategy relies on the destruction of an intact mRNA encoding a potentially lethal product by a mechanism operating in-trans (miRNA). As shown in the blue frame, following optimization, level of expression at State [1,0] (‘Cont.’) is still significant and amounts, in this case, to ~15% than that at State [1,1] (‘Sponge’).
To summarize, a revolutionary technology with numerous therapeutic applications in cancer and beyond exhibits an unacceptable and unavoidable background, which is inherent to its design. Can safety of this strategy be optimized so that the default state of module 1 and, in fact, the whole circuit, is indeed zero?
This is exactly the challenge we undertook to tackle in the iGEM project, and the achievement we are presenting here: Replacing the leaky ‘autoinhibition-sponge’ regulatory apparatus with a new strategy designed to lock the basal output of the entire circuit at ABSOLUTE ZERO, paving the way for its safe implementation in the clinical setting.
2. Our solution
We have chosen to address the safety challenge by exploiting the phenomenon of mRNA trans-splicing (TS), namely, the bimolecular joining of exons from different mRNA species, which is well documented in numerous organisms and is currently explored for diverse gene therapy applications (3). TS can be targeted to pre-selected transcripts (referred to as ‘targeted TS’) by promoting base pairing between corresponding introns flanking the exons to be trans-spliced (see, for example, (4–6) and our own previous work (7)).
Similarly to Nissim et al. we present a two-module system (Fig. 3): Module 1, driven by a synthetic promoter P1, encodes the 5’ end of the GOI (including the translation initiation codon) that is referred to as exon 1, followed by the TS ‘target’ intron. Critically, NO mature, potentially hazardous protein can be translated from this module, ensuring the desired ‘0’ state. Module 2, governed by promoter P2, encompasses the TS ‘guide’ intron, followed by exon 2, which encodes the rest of the protein. Similarly, module 2 cannot produce any functional protein and, standing alone, is permanently in state ‘0’ too. Using this design, the system becomes fully dependent on the synchronized function of both promoters P1 and P2: the risk of translation of the intact TF or the polypeptide product of any GOI in cells in which promoter P1 only or promoter P2 only is active is utterly obviated.
Fig. 3. Scheme of the proposed trans splicing-governed circuit, based on (1). GOI, gene of interest; Ex. Exon.
A major technological challenge facing the ‘conventional’ use of TS, when the target sequence in the target cell is the transcript of an endogenous gene, is how to successfully compete with cis-splicing (8). Applying TS as proposed here, when both transcripts are encoded by exogenous genes, allows maximal freedom in choosing any TS guide and target sequences and the complete avoidance of potential competing splicing events. The latter can be achieved by depriving the respective target genes of the intronic elements that are mandatory for completing an efficient splicing process: the acceptor splice site, the poly-pyrimidine tract and the branch point for module 1 and a donor splice site for module 2 (see Fig. 4).
Fig. 4. Schematic presentation of an intron (taken from https://biologydictionary.net/intron/).
3. Obtaining a proof of concept
3.1. General considerations
Our main objective in this project was to demonstrate a proof of concept of our new TS-based circuit, namely, showing inside living cells that:
The mature protein product of the two-module circuit is only synthesized when both module 1 (driven by promoter P1) and module 2 (driven by promoter P2) are co-expressed, but is absolutely absent when each of the two modules is expressed alone.
Whereas zero background is inherent to our design, maximizing its actual yield, that is, level of the mature protein product synthesized when both modules are fully active, would require careful optimization steps (see, for example, the original Nissim paper (1) and, especially the supplementary material). While these have naturally been beyond the scope of our current iGEM endeavor, they are certainly in the focus of our future plans. One should remember that this system is UNIVERSAL. When the optimal composition of the TS guiding elements in both modules is determined, it could serve in all possible circuits, regardless of the actual therapeutic GOI, identity of promoters or target cells chosen for a given clinical application.
3.2. Experimental design
To achieve this goal we have assembled an experimental in-cell system comprising the following components:
Cells: We have chosen human HEK293T cells, which are easy to propagate and transfect.
Gene delivery: In our preparatory experiments we have found that transient expression of plasmid DNA in HEK293T cells using improved lipid reagents, such as FuGENE HD transfection reagent (Promega), can approach 100% efficacy with practically no cell death (Fig. 5).
Fig. 5. Determining efficiency of transient transfection of HEK293T cells with plasmid DNA. HEK293T cells were transfected with 1 μg of the pEGFP-N3 plasmid (Clontech) or irrelevant DNA (Mock) using FuGENE HD (Promega) according to the manufacturer’s instructions. 48 hours post-transfection cells were analyzed for EGFP fluorescence by flow cytometry using FACSCalibur (BD Biosciences).
Gene of interest: For demonstrating a proof of concept we have used the same mKate2 reporter gene used by Nissim et al. (see (1) and Fig. 1), which is easy to detect by flow cytometry. We have preserved the original division of the gene to two exons as in (1).
Circuits: In the current study we have assembled and analyzed two different circuits, each comprising two modules (see below for details). While sharing elements such as promoters, poly A sites, etc. the intronic sequences selected to induce TS in each of the two circuits completely differ. In the two modules of the first circuit, designated TS1 and TS2, these intronic sequences are based on the original intron separating the two mKate2 exons used in (1) and represent a ‘basic’ TS design. Unlike, the intronic sequences in the two modules comprising the second circuit, TS11 and TS12, are based on a thorough analysis performed by the group of Volker Patzel from the National University of Singapore, a world leader in mRNA TS, in a study which aimed at maximizing the efficacy of targeted TS for cancer therapy (8).
Vector: For expressing Module 1 in each circuit we have used the expression vector pLN193 and for Module 2 we have used pLN75. Description of the different components incorporated into these vectors, including their full DNA sequence, is available in the suppl. material of (1).
Gene composition of the two circuits:
Module TS1: Expression of module TS1 is driven by the SSX1p promoter, which was used by Nissim et al. and is active in HEK293 cells (1). As argued above, our circuits are based on TS events designed to take place between the transcripts of two exogenously introduced modules, thus allowing full degree of freedom in selecting the optimal guiding sequences. Here, downstream to exon 1 of mKate2 we have introduced the first 198 bp of the miR1 intron employed by Nissim et al. (1), including the 5’ basal stem to serve as the TS guiding sequence to module TS2. This sequence is followed by the 34 bp linker of the first module in (1) and the HSV1pA poly A site.
Module TS2: This module is designed to facilitate optimal targeted TS with the transcript of module TS1 to generate the intact mKate2 coding sequence. Its expression is driven by the H2A1p promoter (1), followed by a 40 bp linker (positions 421-460 in the kanamycin resistance gene of thermophilic bacillus from plasmid pTB913 (GenBank accession K02551.1)) and the remaining 219 bp of the miR1 intron, including the 3’ basal stem as the TS1 TS binding domain (BD) and the full acceptor splice site, followed by mKate2 exon 2, 40 bp linker (positions 361-400 in the kanamycin resistance gene), BamHI and HindIII restriction sites and the HSV1pA poly A site.
Module TS11: Similarly to TS1, the promoter of this module is SSX1p. In module TS11, immediately 3’ to exon 1 of mKate2 we inserted the first 50 bp of the mKate2 intron (1) in order to fully preserve the strong donor splice site and to provide a linker preceding the TS guiding sequence. Here we have chosen to adapt the optimized TS elements which have recently been reported in (8). This paper describes in full detail the design of RNA structures, which improve both activity and specificity of TS-mediated targeting of the herpes simplex virus thymidine kinase coding sequence to the endogenous alpha-fetoprotein (AFP) transcript, as suicide gene therapy of cancer. In particular, we have chosen the optimized 50 bp sequence derived from intron 5 of AFP (NCBI Reference Sequence: NG_023028.1). Yet, adapting exactly the same sequence as in (8) poses a problem as the AFP gene is expressed in HEK293 cells (as shown in Fig. S1C in (8)) and the expected TS may be dampened by binding to the endogenous transcript. In order to avoid such an undesired outcome yet preserve the favorable properties of the selected stretch, we have simply chosen to invert the 50 bp sequence from intron 5. The inverted sequence also preserves an artificial 2-base mismatch, introduced by the Patzel group “to prevent formation of long nuclear double-stranded RNA, which could trigger antisense effects” (8). The final TS guiding sequence in module 1 is (mismatch is marked yellow):
5’TGGAGAGATTTGGATTTTTTTTAAAAGAAGAGATTTGGAGAAAGGATCAA 3’
This TS guiding sequence is followed by a 34 bp linker and the HSV1pA poly A site, as in (1).
Module TS12: Expression is driven by the H2A1p promoter. The promoter sequence is followed by the 40 bp linker of the kanamycin resistance gene (positions 421-460). The TS binding domain (BD) in module TS12 is the reverse complementary of the AFP-based TS guiding sequence of module 1, with the 2 nucleotide mismatch (in yellow):
5’TTGATCCTTTCTCCAAATCTCTTCTTTTAAATTAAATCCAAATCTCTCCA 3’
This TS BD is followed by the 34 bp linker which precedes the HSV1pA poly A (1) and the intron splice enhancer, branch point, polypyrimidine tract and the acceptor splice site, all taken from (8), continuing with mKate2 exon 2, 40 bp linker (positions 361-400 in the kanamycin resistance gene) and the HSV1pA poly A site.
For plasmid maps of all four new modules see Fig. 6.
First circuit:
Second circuit:
Fig. 6. Maps of the four modules as drawn with the SnapGene software.
3.3. Results
To obtain a proof of concept, we have designed a co-transfection experiment in HEK293 cells. We first confirmed that the two promoters we have selected for our system, SSX1p (modules TS1 and TS11) and H2A1p (modules TS2 and TS12), are indeed active in HEK293T cells. To this end we have transiently transfected HEK293 cells with either plasmid pLN74, which encodes the intact mKate2 protein under the SSX1p promoter or plasmid pLN75 in which ECFP is driven by the H2A1p promoter (1). Indeed, as shown in the flow cytometry analysis (Fig. 7), transfection of either plasmid, but not irrelevant DNA, resulted in intense fluorescence of the respective protein.
Fig. 7. The two selected promoters are highly active in HEK293T cells. HEK293T cells were transiently transfected with 1 μg either pLN74 (A) or pLN75 (B) encoding mKate2 or ECFP, respectively, or with irrelevant DNA (Mock), using FuGENE HD. 48 hours later cells were subjected to flow cytometry analysis for the respective fluorophore using FACSCalibur.
We then continued and assessed the function of our two new circuits. Our design guarantees no expression of the full GOI (in our case the mKate2 fluorophore) upon introduction of each of the two modules alone. The only scenario, which allows expression of the GOI, is the successful TS between the transcripts of the two modules. Accordingly, we have transiently transfected HEK293T cells with either Module 1 or Module 2 plasmids of the two circuits alone, or with a 1:1 mixture of the two, using irrelevant DNA to keep the final amount of DNA in all transfections constant. The mKate2-encoding plasmid pLN74 served for evaluation of transfection efficacy. Results of two independent experiments performed along the same scheme are shown in Fig. 8.
Indeed, as expected, absolutely no mKate2 fluorescence could be detected when cells were transfected with each module separately; yet, a clear signal was evident following co-transfection of the cells with the two modules.
Fig. 8. Confirming that the new circuit is functional. HEK293T cells (105) were transfected with either Module 1 or Module 2 plasmid alone, each at 1μg or both plasmids together using the FuGENE HD reagent. Irrelevant DNA was used to complement the total amount of DNA to 2 μg in each transfection. Irrelevant DNA (Mock) was used as a negative control and the mKate2-coding plasmid pLN74 was used to assess transfection yield. The results of two independent experiments are shown. Transfected cells were prepared and analyzed by FACS (Attune NxT (Thermo Fisher Scientific)) 48 hours post-transfection.
3.4. Conclusions
These experiments clearly demonstrate a proof of concept of our approach:
- In both circuits, transfection of each module alone gave rise to zero fluorescence (as expected, since it simply CANNOT result in a closed circuit under any such circumstances), so that our TS-based concept provides a solution to the safety challenge.
- Co-transfection of the two modules closed the circuit, yielding clear mKate2 fluorescence.
- The second circuit, which included several components that had been designed to improve TS (based on (8)), indeed exhibited greater efficacy than the first circuit which only directed ‘basic’ TS.
Of note, this is a preliminary demonstration of the expected mode of action of our new TS-based circuit, which was already achieved in our very first attempt to test the system in its entirety, using two different circuits.
We have no doubt that several rounds of optimization steps could substantially improve the circuit efficacy (while not impairing safety, whatsoever). These steps would assess the effect of size and composition of the reverse complementary BDs, the number of copies of the BDs in each module, length and sequence of the linkers and the incorporation of additional elements such as optimally suited TS-ribozymes (8, 9).
Reference
1. Nissim, L., M.-R. Wu, E. Pery, A. Binder-Nissim, H. I. Suzuki, D. Stupp, C. Wehrspaun, Y. Tabach, P. A. Sharp, and T. K. Lu. 2017. Synthetic RNA-Based Immunomodulatory Gene Circuits for Cancer Immunotherapy. Cell 171: 1138-1150.
2. Xie, M., and M. Fussenegger. 2018. Designing cell function: assembly of synthetic gene circuits for cell biology applications. Nat. Rev. Mol. Cell Biol. 19: 507–525.
3. Yang, Y., and C. E. Walsh. 2005. Spliceosome-Mediated RNA Trans-splicing. Mol. Ther. 12: 1006–1012.
4. Puttaraju, M., S. F. Jamison, S. G. Mansfield, M. A. Garcia-Blanco, and L. G. Mitchell. 1999. Spliceosome-mediated RNA trans-splicing as a tool for gene therapy. Nat. Biotechnol. 17: 246–252.
5. Puttaraju, M., J. DiPasquale, C. C. Baker, L. G. Mitchell, and M. A. Garcia-Blanco. 2001. Messenger RNA repair and restoration of protein function by spliceosome-mediated RNA trans-splicing. Mol. Ther. 4: 105–14.
6. Suñé-Pou, M., S. Prieto-Sánchez, S. Boyero-Corral, C. Moreno-Castro, Y. El Yousfi, J. Suñé-Negre, C. Hernández-Munain, and C. Suñé. 2017. Targeting Splicing in the Treatment of Human Disease. Genes (Basel). 8: 87.
7. Schlesinger, J., D. Arama, H. Noy, M. Dagash, P. Belinky, and G. Gross. 2003. In-cell generation of antibody single-chain Fv transcripts by targeted RNA trans-splicing. J Immunol Methods 282: 175–186.
8. Poddar, S., P. S. Loh, Z. H. Ooi, F. Osman, J. Eul, and V. Patzel. 2018. RNA Structure Design Improves Activity and Specificity of trans -Splicing-Triggered Cell Death in a Suicide Gene Therapy Approach. Mol. Ther. - Nucleic Acids 11: 41–56.
9. Sullenger, B. A., and T. R. Cech. 1994. Ribozyme-mediated repair of defective mRNA by targeted trans-splicing. Nature 371: 619–622.