Cosmetics-triggered percutaneous remote control of transgene expression in mice

Synthetic biology has significantly advanced the rational design of trigger-inducible gene switches that program cellular behavior in a reliable and predictable manner. Capitalizing on genetic componentry, including the repressor PmeR and its cognate operator OPmeR, that has evolved in Pseudomonas syringae pathovar tomato DC3000 to sense and resist plant-defence metabolites of the paraben class, we have designed a set of inducible and repressible mammalian transcription-control devices that could dose-dependently fine-tune transgene expression in mammalian cells and mice in response to paraben derivatives. With an over 60-years track record as licensed preservatives in the cosmetics industry, paraben derivatives have become a commonplace ingredient of most skin-care products including shower gels, cleansing toners and hand creams. As parabens can rapidly reach the bloodstream of mice following topical application, we used this feature to percutaneously program transgene expression of subcutaneous designer cell implants using off-the-shelf commercial paraben-containing skin-care cosmetics. The combination of non-invasive, transdermal and orthogonal trigger-inducible remote control of transgene expression may provide novel opportunities for dynamic interventions in future gene and cell-based therapies.


INTRODUCTION
Synthetic trigger-controlled gene switches that enable spatio-temporal fine-tuning of transgene expression have been instrumental for functional genomic research (1), drug discovery (2) and the manufacturing of difficult-toproduce drug targets (3) and protein therapeutics (4). Dur-ing the past decade synthetic biology, the science of reassembling cataloged and standardized biological items in a systematic, rational and predictable manner to create, engineer and program functional biological designer devices, systems and organisms with novel and useful functions (5-10) has significantly advanced the design of gene switches. They evolved from simple control devices providing trigger-inducible transgene expression (11)(12)(13)(14)(15) to complex transcription/translation networks enabling oscillating expression dynamics (16), intercellular communication (17) and fundamental arithmetic operations (18,19). Today, gene switches form the basis for the design of therapeutic gene networks that have been successfully validated in cell-based therapies using animal models of prominent human disorders (2,4,(20)(21)(22)(23)(24)(25)(26)(27)(28)(29).
We have engineered paraben-repressible and -inducible transgene expression systems based on the genetic componentry of the Gram-negative bacterium Pseudomonas syringae pathovar tomato DC3000, a plant pathogen that causes bacterial specks of tomato (38). Expression of P. syringae's major multidrug efflux pump MexAB-OprM is regulated by PmeR (Pseudomonas multidrug efflux regulator), a TetR-type transcriptional repressor that binds to an inverted repeat (O PmeR ) overlapping with the promoters driving mexAB-oprM and pmeR (39,40). Parabens

Construction and characterization of stable cell lines
The HEK-PAR OFF cell line, transgenic for parabenrepressible secreted alkaline phosphatase (SEAP) expression, was constructed by co-transfection of HEK-293 cells with a 10:5:1 (w/w/w) mixture of pWH8 (P SV40 -PMA-pA), pWH10 (P PMA -SEAP-pA) and pZeoSV2(+) (P SV40zeo-pA), followed by selection in culture medium containing 1 mg/ml zeocin (Invitrogen, cat. no. R250-05) and FACS-mediated single-cell cloning. Six out of 30 cell clones were randomly picked and the best-in-class HEK-PAR OFF1 was used for all follow-up studies. Likewise, the HEK-PAR ON cell line, transgenic for paraben-inducible SEAP expression, was constructed by co-transfection of HEK-293 cells with an 8:8:1 (w/w/w) mixture of pWH9 (P SV40 -PMS-pA), pWH5 (P PMS -SEAP-pA) and pZeoSV2(+) (P SV40 -zeo-pA) followed by selection in culture medium containing 1 mg/ml zeocin and FACS-mediated single-cell cloning. Six out of 30 cell clones were randomly chosen and the bestin-class HEK-PAR ON6 cell line was used for all follow-up studies.

Quantification of reporter protein production
Production of human placental SEAP was quantified in the culture supernatant by measuring the colorimetric absorbance time course of the SEAP-mediated p-nitrophenylphosphate to p-nitrophenolate conversion, as described previously (44). In brief, 120 l of buffered substrate solution (100 l of 2x SEAP assay buffer [20 mM homoarginine, 1 mM MgCl 2 , 21% diethanolamine, pH9.8] and 20 l substrate solution [120 mM p-nitrophenylphosphate]) was added to 80 l heatinactivated (65 • C, 30 min) cell culture supernatant and the light absorbance time course was measured at 405 nm (37 • C). The SEAP levels in the bloodstream were profiled using a chemiluminescence-based assay (Roche Diagnostics GmbH, Mannheim, Germany; cat. no. 11 779 842 001).

Animal experiments
Designer cell implants were produced by microencapsulating pWH9/pWH5-transgenic HEK-293 into coherent alginate-poly-(L-lysine)-alginate beads (400 m diameter; 200 cells/capsule) using an Inotech Encapsulator Research Unit IE 50R (Buechi Labortechnik AG, Flawil, Switzerland) set to the following parameters: 0.2 mm nozzle with a vibration frequency of 1025 Hz, 25-ml syringe operated at a flow rate of 410 units and 1.12 kV for bead dispersion (26). One hour after intraperitoneal (IP) or SC (lower dorsum) implantation of 1×10 4 microcapsules into eight-week-old female OF1 mice (oncins France souche 1; Charles River Laboratories, France), the animals were treated with PP injections (0-10 mg/kg in 50 l DMSO, once every 24 h) or topical application of commercial hand cream (Kamill R , 600 mg, 3x every 8 h) and cleanser (Lancaster R , 600 l, 3x every 8 h) or solutions containing three different concentrations of PP (50 l, 3x every 8 h for a total of 0, 10, 100 mg/kg day −1 ). Blood samples were collected 48 h after implantation and serum was isolated using microtainer serum separating tubes (SST) tubes according to the manufacturer's instructions (centrifugation for 5 min at 10 000xg; Becton Dickinson, Plymouth, UK; cat. no. 365967). Semi-quantitative analysis of blood-paraben levels was performed by injecting mice once with 300 mg/kg of PP, collecting blood samples after 24 h and adding 10 l of serum to 2.5x 10 4 pWH9/pWH5-transfected HEK-293 cells, before SEAP expression was profiled and compared to a standard curve after 48 h.
All experiments involving animals were performed according to the directive of the European Community Council (2010/63/EU), approved by the French Republic (no.

Design, construction and validation of paraben-repressible and -inducible mammalian transgene expression switches
In Pseudomonas syringae pv. tomato DC3000, the repressor PmeR (Pseudomonas multidrug efflux regulator) is released from promoters containing specific O PmeR operator sites upon interaction with plant defense metabolites of the paraben class, to induce the paraben-eliminating multidrug efflux pump MexAB-OprM and establish resistance to plant-derived antimicrobial compounds (39,40).
Capitalizing on the paraben-sensitive PmeR-O pmeR interaction, we have designed two isogenic synthetic mammalian gene switches that either repress (PAR OFF ) or induce (PAR ON ) transgene expression in response to FDAapproved parabens. PAR OFF consists of the synthetic mammalian transcription factor PMA (paraben-mediated transactivator; pWH8, P SV40 -PMA-pA; PMA, PmeR-VP16), engineered by fusing PmeR's C-terminus to the Herpes simplex virus-derived transactivation domain (VP16), that modulates the activity of synthetic P PMA promoters (pWH10, P PMA -SEAP-pA; P PMA , O PmeR2 -P hCMVmin ), containing the PMA-specific tandem operator module O PmeR2 5' of a minimal version of the human cytomegalovirus immediate-early promoter (P hCMVmin ), in a paraben-responsive manner: In the absence of parabens, PMA binds and activates P PMA -driven transgene expression while paraben derivatives prevent the PMA-P PMA interaction and repress target-gene expression ( Figure 1A).
PAR ON consists of a synthetic mammalian transcription silencer PMS (paraben-mediated transsilencer; pWH9, P SV40 -PMS-pA; PMS, PmeR-KRAB), engineered by fusing PmeR's C-terminus to the Krueppel-associated box (KRAB) domain of the human kox-1 gene, that modulates the activity of synthetic P PMS promoters (pWH5, P PMS -SEAP-pA; P PMS , P SV40 -O PmeR2 ), containing the PMSspecific tandem operator module O PmeR2 3' of the constitutive simian virus 40 promoter (P SV40 ), in a parabenresponsive manner. In the absence of parabens PMS binds and represses P PMS while paraben derivatives prevent the PMS-P PMS interaction and induce target-gene expression ( Figure 1B).
To assess the impact of PMA and PMS expression on the metabolic integrity and the viability of mammalian cells, we co-transfected HEK-293 with pSEAP2-Control and increasing concentrations of either the PMA-encoding vector pWH8 (P SV40 -PMA-pA) or the PMS-encoding vector pWH9 (P SV40 -PMA-pA) and profiled SEAP production ( Figure 1C) as well as the viable cell number of the cell population for up to 60 h ( Figure 1D). Likewise, to assess the impact of parabens on the viability and the metabolic integrity of mammalian cells, we cultivated HEK-293 cells in medium containing different concentrations (0-200 M) of MP, EP PP, BP and iBP, which are most commonly used as cosmetics additives, and profiled the percent viable cell number ( Figure 1E) and the overall SEAP production capacity of treated cell lines ( Figure 1F). Collectively, these results show that none of the parabens impaired cell viability and that only parabens with longer alkyl chains such as BP and iBP decreased constitutive SEAP expression, suggesting that they have a negative impact on the cell physiology within the tested concentration range (0-200 M). However, neither MP, EP nor PP did impair the SEAP production capacity of HEK-293 ( Figure 1F). When using increasing concentrations (0-200 M) of MP, EP and PP to control SEAP expression in PAR OFF -(pWH8/pWH10) or PAR ON -(pWH9/pWH5)-engineered HEK-293 cells product gene expression was repressed ( Figure 1G) or induced ( Figure 1H) in a dose-dependent manner, respectively. Since PP showed the tightest repression and highest induction for PAR OFF as well as PAR ON , this paraben derivative was used in all follow-up experiments.

Validation of PAR OFF -and PAR ON -controlled transgene expression in different mammalian cell lines
To assess the potential of the paraben-responsive transgene expression devices for a wide range of applications, we tested the performance of the PAR OFF and PAR ON systems in different cell types. Therefore, the PAR OFF system (pWH8/pWH10) was transfected into four human  Figure 2B).
Collectively, both ON-type and OFF-type PP-controlled transgene expression systems are functional in different mammalian cell types from various species, suggesting that these control devices will have broad utility for a wide range of applications.

Construction, selection and characterization of stably transgenic paraben-regulated mammalian cell lines
We have generated six clonal double-transgenic cell lines (HEK-PAR OFF1-6 ), in which SEAP is controlled by the PAR OFF system, by stably co-transfecting pWH8 and pWH10 into HEK-293. All of these transgenic HEK-PAR OFF cell lines showed PP-repressible regulation pro- files, but differed in their overall SEAP expression performance characterized by specific maximum and basal transgene expression signatures ( Figure 3A). Due to the combination of the highest induction ratio and the lowest IC 50concentration of PP ( Figure 3B), HEK-PAR OFF1 emerged as the best-in-class transgenic cell line that was chosen for all follow-up studies. Also, HEK-PAR OFF1 showed dosedependent SEAP repression ( Figure 3B), dose-dependent SEAP induction kinetics ( Figure 3C) and completely reversible SEAP expression profiles when alternating the presence and absence of PP in the culture medium ( Figure 3D). Likewise, we have generated six clonal double-transgenic cell lines, in which SEAP is controlled by the PAR ON system, by stably co-transfecting pWH9 and pWH5 into HEK-293. All of these transgenic HEK-PAR ON cell lines showed PP-inducible regulation profiles, but differed in their overall SEAP induction ratio ( Figure 3E). The cell clone HEK-PAR ON6 showed (i) a near perfect induction ratio characterized by almost undetectable basal expression in the absence of PP and high maximum expression levels in the presence of PP ( Figure 3F), (ii) robust PP dose-dependent SEAP production kinetics ( Figure 3G) and completely reversible SEAP expression kinetics when alternating presence and absence of the trigger compound in the culture medium ( Figure 3H).
Because of the integration of the transgene expression units into random chromosomal loci by illegitimate recombination, gene switch performance is dependent on the chromosomal context and can therefore dramatically vary among different stable cell clones ( Figure 3A and E) (45,46).

Comparative performance analysis of the PAR ON and PEACE ON systems
ON-type gene control systems that induce target-gene expression in response to a transient molecular cue is the preferred gene switch design, because the trigger compound only needs to be administered upon induction. In contrast, OFF-type switches require continuous presence of control compounds for repression and active removal for induction, which limits applications of this control topology in vivo. However, ON-type switches are more challenging to design, as they have to be extremely tight so that the target protein does not accumulate to significant levels even in the absence of the trigger compound. We have therefore redesigned the phloretin-adjustable control element (PEACE), the pioneering OFF-type design that enabled transdermal control by the apple metabolite phloretin (12), into an isogenic ON-type design (pMX101, P SV40 -TtgR-KRAB-pA; pWH19, P SV40 -O TtgR2 -SEAP-pA) for comparative performance analysis with the PAR ON system (pWH9/pWH5) ( Figure 4). To assess the impact of phloretin on the viability and the metabolic integrity of mammalian cells, we performed the same reporter protein-based assay as for parabens ( Figure 1E and F) and found that although phloretin did not impair the viable fraction of treated cells within the standard PEACE-inducing concentration range (0-50 M, (12)) ( Figure 4A), higher concentrations decreased constitutive SEAP production capacity of mammalian cells ( Figure 4B). Therefore, the phloretin-inducible transgene expression switch ( Figure 4C) may not reach optimal peak expression levels in various human cell lines. Also, the induction factor reached by the PEACE ON system was lower in all tested mammalian cell lines ( Figure 4D) when compared to the PAR ON system ( Figure 2B). Moreover, the PP-controlled transgene expression system delivers precisely adjustable induction kinetics ( Figure 4E) and shows improved dose-dependent transgene-induction characteristics compared to its phloretin-regulated counterpart ( Figure 4F). Collectively, these data suggest that the PAR ON gene expression system will be the preferred control design for percutaneous control of transgene expression.

Cosmetics-controlled transgene expression
In order to evaluate whether the paraben levels in cosmetics approved by the Personal Care Products Council (<0.8%, (32)) matches the sensitivity range of the PAR ON system, we exposed pWH9/pWH5-transgenic HeLa cultures to different amounts of commercial skin-care products including toner solutions (cleanser (Lancaster R ), skin toner (Lancôme R )), emulsion creams (hand cream (Kamill R ) and shower gels (Dove R , Cien R )) ( Figure 5). Although all of the skin-care products had to be diluted to reduce the cytotox-icity of the soap components, all paraben-containing products were able to dose-dependently induce PAR ON -driven SEAP expression. Dove R shower gel, which was explicitly declared as paraben-free, was indeed not inducing the PAR ON device and served as a negative control ( Figure 5). The results indicated that the PAR ON system could be regulated by paraben-containing skin-care and hygiene products in vitro with Lancaster R cleanser and Cien R shower gel showing the best dose-dependent induction performance among all tested toner solutions and emulsions, respectively ( Figure 5). However, since the induction performance   of the cosmetics could be confounded by the cytotoxicity of its soap components for cells grown in culture the true paraben-based control capacity of the cosmetics can only be assessed by percutaneous control of SC PAR ON -transgenic designer cell implants in an animal model.

Percutaneous control of SC implants by topical administration of cosmetic skin-care products
To assess the performance of the PAR ON system in vivo, we microencapsulated pWH9/pWH5-transgenic HEK-293 cells into coherent, semi-permeable (allowing free diffusion of nutrients, waste metabolites and SEAP) and immunoprotective (pore-size tuned to prevent transfer of immunoglobulins) beads made of alginate-poly-(L-lysine)alginate, a clinically licensed material that was shown to enable vascularization and connection of entrapped designer cells to the bloodstream (47,48) and has been successfully tested in human-clinical trials (49). Paraben-inducible SEAP expression of microencapsulated PAR ON -transgenic designer cells was validated in cell culture ( Figure 6A) before the same batch was either intraperitoneally ( Figure 6B) or subcutaneously implanted into mice ( Figure 6C). Animals treated with IP implants received one-dose-per-day of three different concentrations of PP (0-10 mg/kg). Analysis of blood-paraben levels confirmed regulation-effective paraben concentrations in circulation for up to 24 h (60.6 ± 9.1 M), corroborating established paraben pharmacokinetics (50,51). Forty-eight hours after paraben administration, SEAP expression was profiled in the bloodstream of treated mice ( Figure 6B). This data set confirmed dosedependent high-level performance of the preferred PAR ON gene switch in vivo yet did not reveal whether the device was sufficiently sensitive to accept percutaneous control input or whether parabens contained in commercial skin-care products would cross the skin to program transgene expression in SC designer cell implants. Therefore, we treated mice with SC PAR ON -transgenic designer cell implants with three topical applications per day of undiluted Lancaster R cleanser or Kamill R hand cream or solutions containing three different concentrations of PP (0-100 mg/kg) as controls (Figure 6C). Collectively, this data show that paraben (i) is able to cross the skin, (ii) can remote control cellular behavior inside the body in a non-invasive manner simply by topical application of paraben and that (iii) commercial skin-care and hygiene products contain sufficient paraben to program designer cell implants using a typical three-times-per-day application frequency.

DISCUSSION
With the ambition to use orthogonal gene switches for the control of therapeutic transgene expression dosing in future gene-and cell-based therapies, the quest for the ideal inducer compounds has just started. While clinically licensed drugs such as antibiotics (13,(52)(53), hormones (54) and antidiabetics (55) have secondary therapeutic effects and collateral side effects, amino acids (56,57), vitamins (44,58) and metabolites (59) are non-orthogonal and require control concentrations that permanently exceed physiologic levels, food components and food additives such as phloretin (12), preservatives (60) and flavors (60,61) that limit the choice of diet and traceless inducers such as temperature (62,63), light (4) and radiowaves (64) are ubiquitous environmental cues that are impossible to avoid and use for exclusive therapeutic control.
Besides the type of the trigger compound, the administration route will be of prime importance for dosing and The same batch of microencapsulated pWH9/pWH5-transgenic HEK-293 cells (2×10 6 cells, 10 000 capsules, 200 cells/capsule) was subcutaneously implanted into mice, which received thrice daily topical applications of paraben-containing cosmetics (three daily administrations: 3×600 g Kamill R hand cream; 3×600 l Lancaster R cleanser). Solutions containing different concentrations of propylparaben (three daily administrations: 3×50 l for a total of 0, 10, 100 mg/kg day −1 ) were used as controls. The SEAP levels in the bloodstream of treated animals were profiled after 48 h. The data are shown as the mean ± SEM, statistics by two-tailed t test, n = 8 mice. **P < 0.01, ***P < 0.001 versus control. patient compliance. In contrast to conventional compound injections, which require medical care, and oral administration, which is limited by the hepatic first-pass effect, transdermal delivery of trigger compounds would improve patient compliance, enable local administration and eliminate the hepatic first-pass effect as well as the need for assistance by trained medical personnel. Pioneering efforts to establish transdermal gene expression control culminated in the design of the phloretin-repressible control element (PEACE), whose expression could be fine-tuned in SC implants by topical application of the penetration enhancer phloretin (12). However, since the PEACE system shows lower trigger sensitivity compared to the paraben control switch in vivo, phloretin needs to be administered at much higher concentrations (1680 mg/kg, (12); compared to 100 mg/kg for PP), which are not present at control-effective levels in off-theshelf cosmetics.
Collectively, an ideal control compound for therapeutic transgene control should be (i) physiologically inert to prevent any metabolic crosstalk, (ii) clinically licensed to guarantee safety, (iii) rapidly cleared from peripheral circulation to support reversibility of transgene expression and (iv) enable percutaneous control following topical administration.
With their validated generic design principle, their highperformance ON-/OFF-type switch dynamics, combining adjustability and reversibly with low basal as well as high maximum expression profiles, and their responsiveness to the physiologically inert cosmetics preservative PP, the PAR OFF and PAR ON devices meet with all criteria of an ideal gene switch for therapeutic transgene control at a high standard (65). In direct comparison with the ON-type PEACE system providing phloretin-modulated expression control, the isogenic PAR ON gene switch shows higher induction factors and faster induction kinetics in the tested mammalian cell line. Despite the long history of commonplace application as cosmetics preservative, the finding of increased paraben levels in breast cancer tissue has triggered discussions about the role of parabens in the development of breast cancer (66)(67)(68). Although the causal connection has not been scientifically proven and authorities have left the NOAEL (1000 mg/kg day −1 ) unchanged, an increasing number of consumer product companies feel the consumer pressure and produce paraben-free skin-care products. The PAR ON system is sufficiently sensitive to reliably test the presence of parabens in cosmetics. For example, using the PAR ON system we could confirm the presence of parabens in Kamill R hand cream, Lancaster R cleanser and Lancôme Paris R toner as well as the absence of regulation-effective paraben levels in the Dove R shower gel. The parabeninducible gene switch may therefore also be used to augment animal models for toxicology studies. Furthermore, using paraben-spiked paraben-free commercial skin-care products would allow orthogonal percutaneous remote control of transgenes for therapeutic purposes without interfering with the patients' lifestyle or hygiene habits. Because of the duty of declaration for parabens in consumer care products and cosmetics, the risk of accidental exposure of individuals to parabens is considered negligible.
Collectively, due to the combination of high skin permeation capacity, low toxicity and a regulation-effective concentration range that fully matches FDA-approved doses (<0.1%), parabens particularly qualify for remote control of therapeutic transgene expression in SC designer cell implants for safe transdermal therapeutic applications in the future.