Engineered poly(A)-surrogates for translational regulation and therapeutic biocomputation in mammalian cells

Abstract

Here, we present a gene regulation strategy enabling programmable control over eukaryotic translational initiation. By excising the natural poly-adenylation (poly-A) signal of target genes and replacing it with a synthetic control region harboring RNA-binding protein (RBP)-specific aptamers, cap-dependent translation is rendered exclusively dependent on synthetic translation initiation factors (STIFs) containing different RBPs engineered to conditionally associate with different eIF4F-binding proteins (eIFBPs). This modular design framework facilitates the engineering of various gene switches and intracellular sensors responding to many user-defined trigger signals of interest, demonstrating tightly controlled, rapid and reversible regulation of transgene expression in mammalian cells as well as compatibility with various clinically applicable delivery routes of in vivo gene therapy. Therapeutic efficacy was demonstrated in two animal models. To exemplify disease treatments that require on-demand drug secretion, we show that a custom-designed gene switch triggered by the FDA-approved drug grazoprevir can effectively control insulin expression and restore glucose homeostasis in diabetic mice. For diseases that require instantaneous sense-and-response treatment programs, we create highly specific sensors for various subcellularly (mis)localized protein markers (such as cancer-related fusion proteins) and show that translation-based protein sensors can be used either alone or in combination with other cell-state classification strategies to create therapeutic biocomputers driving self-sufficient elimination of tumor cells in mice. This design strategy demonstrates unprecedented flexibility for translational regulation and could form the basis for a novel class of programmable gene therapies in vivo.

Fig. 1. Translational regulation mediated by engineered poly(A)-surrogates.

a Working scheme of STIF-dependent gene expression control. Translation of reporter mRNA with genetically modified 3′-UTR containing an RBP-specific aptamer region and an shRNA- or HHR-mediated cleavage site for preprogrammed poly(A) removal depends on the presence of STIFs consisting of different RBPs fused to different eIFBPs. Because STIFs are designed to mimic the role of natural PABP in simultaneously binding target mRNA and one member of the eIF4F complex (i.e., eIF4G, eIF4E, eIF4A, eIF4B, etc.) to form a “closed-loop” mRNA configuration and activate translation, poly(A) removal is essential to achieve effective and autonomous (trans)gene control and to avoid putative crosstalk with endogenous PABP-driven processes. b STIF-mediated translation of SEAP mRNA containing shRNA-216-cleavable poly(A). HEK-293 cells were co-transfected with a SEAP expression vector containing 8 tandem MCP-specific MS2-box repeats (P_hCMV-SEAP-(MS2)₈-BS_shRNA-216-pA; pSL1331), an shRNA-216 expression vector (P_hU6-shRNA-216; pSL4, 100 ng) and expression vectors for different STIF variants (PABP-MCP, pSL1315; eIF4G-MCP, pSL154; MCP-eIF4E, pSL1316; MCP-NSP3, pSL95; MCP-VPg, pLYL47) or an MCP-Coh2 protein incapable of initiating translation (pSL674, negative control). SEAP levels in culture supernatants were quantified at 48 h post transfection. Data are means ± SD of n = 4 independent experiments. c Correlation between aptamer region size and STIF-mediated SEAP expression. HEK-293 cells were transfected with SEAP expression vectors containing different tandem repeats of the MCP-specific MS2-box aptamer ((MS2-box)₈, pSL515; (MS2-box)₁₂, pSL1284; (MS2-box)₁₆, pSL516; (MS2-box)₂₄, pSL468) and constitutive expression vectors for either MCP-NSP3 (ON; pSL95) or an MCP-Coh2 protein incapable of binding eIF4F (OFF; pSL674). SEAP expression in culture supernatants was scored at 48 h post transfection. Data are presented as means ± SD; n = 4 individual experiments. d STIF-mediated translation of SEAP mRNA containing HHR-cleavable poly(A). HEK-293 cells were co-transfected with a SEAP expression vector containing 24 tandem MCP-specific MS2-box repeats (P_hCMV-SEAP-(MS2 box)₂₄-HHR-pA; pSL468) and expression vectors for different STIF variants (PABP-MCP, pSL1315; eIF4G-MCP, pSL154; MCP-eIF4E, pSL1316; MCP-NSP3, pSL95; MCP-VPg, pLYL47) or an MCP-Coh2 protein incapable of initiating translation (pSL674, negative control). SEAP levels in culture supernatants were quantified at 48 h post transfection. Data are shown as means ± SD of n = 4 independent experiments. e Translational regulation by MCP-based STIF and other state-of-the-art regulation strategies. For STIF-mediated translation (red), HEK-293 cells were co-transfected with a SEAP expression vector containing 24 tandem MCP-specific MS2-box repeats (reporter: P_hCMV-SEAP-(MS2-box)₂₄-HHR-pA, pSL468) and constitutive expression vectors for MCP-Coh2 (OFF; pSL674) or MCP-NSP3 (ON; pSL95). For translational regulation by ligand-inhibited ribozyme activity (orange; inspired by ref. ), HEK-293 cells were co-transfected with a SEAP expression vector containing bacteriophage λ N-Peptide (λN)-repressible HHR placed upstream of poly(A) (reporter: P_SV40-SEAP-HHR-pA, pMX116) and constitutive expression vectors for λN-mCherry (ON; pSA776) or mCherry (OFF; pQX183). For translational regulation by modulation of 3′-UTR stability (gray; inspired by ref. ), HEK-293 cells were transfected with expression vectors for SEAP mRNA containing no poly(A) (OFF; pMX116, SEAP-HHR-pA) or the lncRNA MALAT1 in the 3′-UTR (ON; pLZ323, SEAP-MALAT1-HHR-pA). SEAP levels in culture supernatants were scored at 48 h post transfection. Data are shown as means ± SD; n = 4 independent experiments. f STIFs activate target gene translation by associating with the endogenous eIF4F complex. HEK-293 cells were transfected with a SEAP expression vector containing 24 tandem MCP-specific MS2-box repeats (pSL468) and expression vectors for either 3×FLAG-tagged MCP (pSL1084) or MCP-NSP3 (pSL1083) before one lysate fraction was immunoprecipitated at 48 h post transfection. Target proteins in lysate fractions before (input) and after immunoprecipitation (Flag-IP) were detected with anti-FLAG, anti-eIF4G and anti-eIF4E antibodies. SEAP levels in culture supernatants were measured at 48 h post transfection (Supplementary information, Fig. S3e). Bars represent means ± SD, and filled circles show individual results; numbers above bars are fold changes (b–e).

Fig. 2. Tailoring stimulus-responsive gene expression using a multipartite STIF design framework.

a Translational regulation by protein–protein interaction (PPI)-mediated STIF reconstitution. Bipartite STIF systems with NSP3 and RBP domains split into two independent proteins allow flexible incorporation of different PPI systems to regulate conditional STIF assembly and translation of RBP-specific mRNA. b Compatibility of STIF-mediated translational regulation with split-complementation approaches. Left, HEK-293 cells were co-transfected with a SEAP expression vector containing 24 tandem MCP-specific aptamer repeats (pSL468) and different combinations of constitutive expression vectors for MCP-NSP3 (pSL674) and (DocS)₃-NSP3 (pSL86). Right, HEK-293 cells were co-transfected with a SEAP expression vector containing bacteriophage λ N-Peptide (λN)-repressible HHR placed upstream of poly(A) (pMX116) and constitutive expression vectors for λN-Coh2 (pSL334) and DocS-mCherry (pSL1099). For (–) conditions, pcDNA3.1(+) was transfected instead of expression vectors. SEAP levels in culture supernatants were scored at 48 h post transfection. Data are shown as means ± SD; n = 4 independent experiments. c Regulation through trigger-inducible STIF reconstitution. For danoprevir-inducible SEAP translation, HEK-293 cells were co-transfected with pSL468 and constitutive expression vectors for MCP-NS3a (pSL497) and DNCR-NSP3 (pLZ72). For abscisic acid-inducible SEAP translation, HEK-293 cells were co-transfected with pSL468 and constitutive expression vectors for ABI-MCP (pPW22) and (PYL)₃-NSP3 (pPW4). For gibberellic acid-inducible SEAP translation, HEK-293 cells were co-transfected with pSL468 and constitutive expression vectors for MCP-GID (pPW23) and GAI-NSP3 (pPW2). For grazoprevir-inducible SEAP translation, HEK-293 cells were co-transfected with pSL468 and constitutive expression vectors for MCP-NS3a (pSL497) and GNCR-NSP3 (pYF3). For rapamycin-inducible SEAP translation, HEK-293 cells were co-transfected with a SEAP expression vector containing poly(A)-surrogates with 16 tandem MCP-specific MS2-box repeats (pSL516) and constitutive expression vectors for MCP-FRB (pSL1097) and FKBP-NSP3 (pSL1098). For blue light-inducible SEAP translation, HEK-293 cells were co-transfected with pSL516 and constitutive expression vectors for MCP-CIB1 (pSL1096) and Cry2-NSP3 (pSL71). For red light-inducible SEAP translation, HEK-293 cells were co-transfected with pSL516 and constitutive expression vectors for MCP-(Aff6_V18FΔN)₄ (pSL917) and DrBPhP-NSP3 (pSL901). SEAP levels in culture supernatants were scored at 48 h after addition of the corresponding inducers (danoprevir, 0.5 µM; abscisic acid, 100 µM; gibberellic acid, 100 µM; grazoprevir, 0.5 µM; rapalog, 0.1 µM) or at 48 h after exposure to blue light (450 nm; ON, 30 s at 5 mW/cm²; OFF, 30 s) or at 24 h after exposure to red light (660 nm, constantly 1 W/m²). Data are shown as means ± SD; n = 4 independent experiments. d Genetically encoded signaling-specific sensors employing phosphorylation-dependent STIF reconstitution. HEK-293 cells were co-transfected with a constitutive FLuc expression vector (pYW99), an expression vector for NanoLuc-mRNA containing MCP-specific poly(A)-surrogates (P_hCMV-NanoLuc-P2A-mCherry-(MS2-box)₂₄-HHR-pA, pSL683) and different combinations of constitutive expression vectors for MCP-(pE59)₂ (pSL637) and (ERK2)₂-NSP3 (pSL189) before cultivation in cell culture medium containing 2% fetal bovine serum (FBS) (v/v). Luciferase levels in culture supernatants were quantified at 48 h after the addition of 0 or 100 ng/mL recombinant human EGF. For (–) conditions, pcDNA3.1(+) was transfected instead of expression vectors. Data are presented as means ± SD of relative luciferase activity (NanoLuc/FLuc); n = 3 individual experiments. e Genetically encoded protein sensors employing protein association-dependent STIF reconstitution. HEK-293 cells were co-transfected with pSL468 and constitutive expression vectors for MCP-(Coh2)₃ (pSL1080), (Coh2)₃-NSP3 (pSL243) and (DocS)₃ (different amounts of pSL244). SEAP levels in culture supernatants were scored at 48 h post transfection. Data are shown as means ± SD of n = 4 independent experiments. Bars represent means ± SD, and filled circles show individual results; numbers above bars are fold changes (b–e). ns, not significant; **P < 0.01.

Fig. 3. Grazoprevir-inducible gene switch for translational regulation in mammalian cells and gene therapy applications.

a Regulation of therapeutic transgene expression using the FDA-approved drug grazoprevir. Delivery of the genetic components for GNCR:NS3a-dependent STIF assembly and STIF-specific target mRNA in vivo is compatible with various administration routes of non-integrative gene therapy (e.g., DNA-encoded vectors or formulated mRNA drugs), allowing oral uptake of grazoprevir to trigger in situ production of various therapeutic proteins of interest (e.g., insulin). b Dose-dependent grazoprevir-inducible SEAP expression by plasmid DNA-based delivery. HEK-293 cells were co-transfected with plasmids encoding (GNCR)₃-NSP3 (pSL1032), MCP-(NS3a)₃ (pSL1042) and SEAP mRNA containing MCP-specific poly(A)-surrogate (pSL468). SEAP levels in culture supernatants were scored at 24 h after addition of different concentrations of grazoprevir. Data are shown as means ± SD of n = 4 independent experiments. c Kinetics of grazoprevir-inducible SEAP expression. For grazoprevir-inducible translation, HEK-293 cells were co-transfected with plasmids encoding MCP-(NS3a)₃ (pSL503) and (GNCR)₃-NSP3 (pLZ74) and pSL468. For grazoprevir-inducible transcription, HEK-293 cells were co-transfected with a TetR-specific SEAP expression vector (tetO₇-P_hCMVmin-SEAP-(C/D-box)₂₄-(BS_shRNA-216)₂-pA, pLZ79) and constitutive expression vectors for TetR-NS3a (pLZ88) and (GNCR)₃-VP64 (pLZ85). 24 h after transfection, 0 μM or 0.5 μM grazoprevir was added to culture supernatants and SEAP levels were monitored for another 24 h. Data show means ± SD of fold changes calculated by dividing SEAP levels of grazoprevir-treated samples by those of non-grazoprevir-treated samples (n = 3 individual experiments). d Grazoprevir-inducible SEAP translation by mRNA delivery. HEK-293 cells were (co-)transfected with in vitro-transcribed mRNA encoding for MCP-(NS3a)₃ (from pSL1085), (GNCR)₃-NSP3 (from pYW361) and SEAP-(MS2-box)₂₄ (from pSL468). SEAP levels in culture supernatants were quantified at 48 h after addition of grazoprevir. Data are shown as means ± SD; n = 4 independent experiments. e Reversibility of grazoprevir-inducible protein secretion in mammalian cells. HEK-293_LSCCS1 (stably expressing MCP-(NS3a)₃, (GNCR)₃-NSP3 and NanoLuc-mRNA containing 24 tandem MS2-box repeats) were cultivated for 7 days while grazoprevir levels in the culture medium were successively switched between 0 nM and 500 nM. NanoLuc levels were measured every 12 h. The cell density was readjusted to 1 × 10⁵ cells/mL every 2–3 days. f, g Therapeutic efficacy of oral grazoprevir-inducible insulin expression. Plasmids encoding MCP-(NS3a)₃, (GNCR)₃-NSP3 and insulin-mRNA containing MCP-specific poly(A)-surrogate (pSL548/pLZ74/pSL685) were hydrodynamically injected into the tail vein of T1D mice. At 6 h post injection, mice were fed with the first of 3 daily administrations of 3 mg/kg grazoprevir. Fasting glycemia and blood modified insulin (mINS) levels of mice were measured at 20 h after the first grazoprevir administration (f). Intraperitoneal glucose tolerance tests (GTTs) were performed at 24 h after the first grazoprevir administration (4 h after quantification of blood insulin shown in f) (g). Data are presented as means ± SEM; n = 6 mice per group. h Long-term control of grazoprevir-inducible SEAP production in mice. C57BL/6 mice received intravenous injection of 7 × 10¹¹ AAV2/8 particles for constitutive expression of (GNCR)₃-NSP3, MCP-(NS3a)₃ and SEAP-mRNA containing MCP-specific poly(A)-surrogate. SEAP production in the bloodstream was monitored over 10 weeks. At 24 h before each measurement, mice were fed with the first of 3 daily administrations of 3 mg/kg grazoprevir. Data are presented as means ± SEM; n = 4 mice per group. Bars represent means ± SD or SEM, and filled circles show individual results; numbers above bars are fold changes (b–d, f, h). *P < 0.05; **P < 0.01; ***P < 0.001.

Fig. 4. Genetically encoded sensors for intracellular target proteins.

a Engineering of transcription- and translation-based sensors for detection of differentially localized intracellular proteins. The synthetic target protein EGFP-NS3a(H1) was targeted to different intracellular compartments through fusion with different localization signals (1, NLS (nuclear localization signal); 2, NES (nuclear export signal); 3, CAAX (prenylation motif); 4, transmembrane localization signal; 5, secretory signal peptide) to allow detection with co-expressed genetic sensors engineered on the basis of LaG16 (an EGFP nanobody) and ANR (a peptide motif binding NS3a(H1)). For transcription-based sensing, LaG16 was fused to TetR and ANR was fused to VP64 to allow EGFP-NS3a(H1)-dependent activation of TetR-specific promoters. For translation-based sensing, LaG16 was fused to MCP and ANR was fused to NSP3 to allow EGFP-NS3a(H1)-dependent STIF reconstitution and translation of MCP-specific mRNA. b Dose-dependent detection of differentially translocated EGFP-NS3a(H1). HEK-293 cells were co-transfected with plasmids encoding the translation-based EGFP-NS3a(H1) sensor (pSL776/pSL582/pSL468: for expression of MCP-LaG16, (ANR)₈-NSP3 and SEAP-mRNA with MCP-specific poly(A)-surrogate) or the transcription-based EGFP-NS3a(H1) sensor (pSL834/pSL836/pMF111: for expression of TetR-LaG16, (ANR)₈-VP64 and a TetR-specific promoter controlling SEAP transcription) and different amounts of constitutive expression vectors for different target proteins (0, native EGFP-NS3a(H1), pSL775; 1, nuclear NLS-EGFP-NS3a(H1), pSL797; 2, cytosolic NES-EGFP-NS3a(H1), pSL824; 3, prenylated EGFP-NS3a(H1)-CAAX, pSL799; 4, membrane-localized TM-EGFP-NS3a(H1), pSL798; 5, secretory SP-EGFP-NS3a(H1), pSL796). At 48 h after transfection, fluorescence images of cellular EGFP signals were acquired (scale bars, 10 μm) and SEAP levels in the culture supernatant were profiled. Data are shown as means ± SD fold change of SEAP activity relative to basal SEAP levels detected with no EGFP-NS3a(H1) expression (0 ng columns). Filled circles represent individual results (n = 4 independent experiments).

Fig. 5. Engineering of target-specific protein sensors driving self-sufficient cancer gene therapies.

a Validation of therapeutic efficacy using xenograft mouse models. Cell lines with different molecular signatures hypothetically distinguishable by a cytoplasmic target protein were subcutaneously implanted into the lower back of mice and tumors were allowed to grow for 17 days. On day 7, gene circuits engineered for cancer-selective activation of apoptosis were injected into the tumors every 2 days. Tumor size was monitored over the entire experimental timespan. b–d EGFP-NS3a(H1)-specific activation of apoptosis in mice. Mice harboring subcutaneous B16-F10_{EGFP-NS3a(H1)}-derived tumors received local injections of pcDNA3.1(+) (negative control, n = 5 mice per group) or plasmid DNA mixture comprising pSL831 (P_hCMV-mBax-(MS2-box)₂₄-HHR-pA), pSL776 (P_hCMV-MCP-LaG16-pA) and pSL582 (P_hCMV-(ANR)₈-NSP3-pA) (treatment group, n = 5 mice per group). Daily changes of tumor size were assessed by calculating V = (length × width)²/2 (b). Tumors were harvested on the final experimental day for weight analysis (c) and measurement of Bax protein levels by western blot (d). e–g No activation of apoptosis by a P_hCMV-driven EGFP-NS3a(H1) sensor in EGFP-NS3a(H1)-deficient tissues. Mice harboring subcutaneous B16-F10-derived tumors received local injections of pcDNA3.1(+) (negative control, n = 5 mice per group) or plasmid DNA mixture comprising pSL831/pSL776/pSL582 (treatment group, n = 5 mice per group). Daily changes of tumor size were assessed by calculating V = (length × width)²/2 (e). Tumors were harvested on the final experimental day for weight analysis (f) and measurement of Bax protein levels by western blot (g).

Fig. 6. Therapeutic biocomputer for self-sufficient elimination of malignant cells in mice.

a Design principle of a gene circuit with customized detection algorithms. To precisely distinguish complex cell signatures within heterogenous tissues (e.g., tumors), STIF-based protein sensors with programmable target-specificity could be coupled with TSP driving the production of STIF-specific mRNA for suicide gene expression (e.g., Bax). Upon delivery of such gene circuits into living tissues in vivo, self-sufficient apoptosis is exclusively triggered in cells that meet the criteria of both tissue- (TSP-triggered transcription of poly(A)-deficient STIF-dependent Bax mRNA) AND target-specificity (translation of Bax mRNA only upon STIF-mediated detection of intracellular protein markers). b Tissue-specific sensing of EGFP-NS3a(H1) in vitro. Native (WT) or stable EGFP-NS3a(H1)-transgenic N2A and Hepa1-6 cells were co-transfected with P_MusAFP-driven (pSL813) or constitutive expression vectors for NanoLuc-P2A-mRNA containing an MCP-specific poly(A)-surrogate (pSL683) and constitutive expression vectors for MCP-LaG16 (pSL776), (ANR)₈-NSP3 (pSL582) and FLuc (pYW99). Luciferase levels were quantified at 48 h post transfection. Data are presented as means ± SD of relative luciferase activity (NanoLuc/FLuc); n = 3 individual experiments. c Validation of tissue-specific EGFP-NS3a(H1) sensing in mice. Plasmids encoding P_hCMV-driven (pSL468/pSL582/pSL776) and P_MusAFP-driven (pSL857/pSL582/pSL776) EGFP-NS3a(H1) sensors and EGFP-NS3a(H1) expression vectors (pSL775) were hydrodynamically injected into the tail vein of C57BL/6 mice. SEAP levels in the bloodstream were measured after 24 h. Mice receiving pcDNA3.1(+) instead of pSL775 were used as negative controls with no hepatic EGFP-NS3a(H1) expression (WT). Data are presented as means ± SEM; n = 5 mice per group. d–f P_MusAFP- AND EGFP-NS3a(H1)-specific activation of apoptosis in mice. Mice harboring subcutaneous Hepa1-6_{EGFP-NS3a(H1)}-derived tumors received local injections of pcDNA3.1(+) (negative control, n = 5 mice per group) or plasmid DNA mixture comprising pSL886 (P_MusAFP-mBax-(MS2-box)₂₄-HHR-pA), pSL776 and pSL582 (treatment group, n = 5 mice per group). Daily changes of tumor size were assessed by calculating V = (length × width)²/2 (d). Tumors were harvested and weighed on the final experimental day (e, f). Bars represent means ± SD, and filled circles show individual results (b, c, e). ns not significant; *P < 0.1; **P < 0.01; ***P < 0.001.