Implantable synthetic cytokine converter cells with AND-gate logic treat experimental psoriasis

See allHide authors and affiliations

Science Translational Medicine  16 Dec 2015:
Vol. 7, Issue 318, pp. 318ra201
DOI: 10.1126/scitranslmed.aac4964

A psoriatic switch

Taking pills may go the way of the horse and buggy, the rotary phone, and the Walkman, at least if synthetic biology has anything to say about it. Schukur et al. designed a circuit that would automatically sense the presence of two disease-causing molecules, called cytokines, in the body and respond by triggering the production of two other cytokines that would treat the disease. This circuit was genetically engineered in a mammalian cell; in turn, the cell was implanted in mice with psoriasis—an inflammatory skin condition that has no cure. When levels of the proinflammatory cytokines TNF and IL22 peaked in the body, the synthetic circuit kicked into gear, converting these cytokine signals into an anti-inflammatory cellular output, consisting of IL4 and IL10, which then attenuated disease. The “cytokine converter” cells not only prevented psoriasis flare-ups, as they’re called, but also treated acute (established) psoriasis, returning skin to normal in mice. In demonstrating that the converter cells were responsive to blood from psoriasis patients, the authors suggest that synthetic biology may be ready to autonomously flip therapeutic switches in people and later take on other diseases with defined disease indicators.


Psoriasis is a chronic inflammatory skin disease characterized by a relapsing-remitting disease course and correlated with increased expression of proinflammatory cytokines, such as tumor necrosis factor (TNF) and interleukin 22 (IL22). Psoriasis is hard to treat because of the unpredictable and asymptomatic flare-up, which limits handling of skin lesions to symptomatic treatment. Synthetic biology–based gene circuits are uniquely suited for the treatment of diseases with complex dynamics, such as psoriasis, because they can autonomously couple the detection of disease biomarkers with the production of therapeutic proteins. We designed a mammalian cell synthetic cytokine converter that quantifies psoriasis-associated TNF and IL22 levels using serially linked receptor-based synthetic signaling cascades, processes the levels of these proinflammatory cytokines with AND-gate logic, and triggers the corresponding expression of therapeutic levels of the anti-inflammatory/psoriatic cytokines IL4 and IL10, which have been shown to be immunomodulatory in patients. Implants of microencapsulated cytokine converter transgenic designer cells were insensitive to simulated bacterial and viral infections as well as psoriatic-unrelated inflammation. The designer cells specifically prevented the onset of psoriatic flares, stopped acute psoriasis, improved psoriatic skin lesions and restored normal skin-tissue morphology in mice. The antipsoriatic designer cells were equally responsive to blood samples from psoriasis patients, suggesting that the synthetic cytokine converter captures the clinically relevant cytokine range. Implanted designer cells that dynamically interface with the patient’s metabolism by detecting specific disease metabolites or biomarkers, processing their blood levels with synthetic circuits in real time, and coordinating immediate production and systemic delivery of protein therapeutics may advance personalized gene- and cell-based therapies.


Psoriasis is a common chronic relapsing-remitting inflammatory skin disease characterized by itchy red scaly skin lesions (1) and is associated with an increased risk of immune-mediated diseases, such as Crohn’s disease and ulcerative colitis (2, 3), as well as certain cancers (liver and pancreatic) (4), metabolic disorders (obesity and diabetes) (5, 6), and cardiovascular diseases (7). The causes of psoriasis remain largely elusive. However, psoriasis is generally considered a genetic disease (8) that is triggered and influenced by environmental factors, infections, medications, and lifestyle (9, 10).

Psoriasis results from an inflammatory cascade in the dermis involving erroneous crosstalk between keratinocytes and tissue-resident dendritic cells, which recruit immune cells from the adaptive [T helper cells (TH1 and TH17)] and innate (neutrophils, macrophages, and dendritic cells) systems to the epidermis (11), where they secrete proinflammatory cytokines, such as tumor necrosis factor (TNF) and interleukin 22 (IL22) (11, 12). IL22 synergizes with other proinflammatory cytokines (13), particularly TNF (14), to drive psoriasis-promoting activities, such as the proliferation of keratinocytes (14, 15). Additionally, TNF amplifies the biological effects of IL22 by increasing the expression of the IL22 receptor and promotes the differentiation of TH17 and TH22 cells (12), which are predominant actors in inflamed psoriatic skin (16). Because blood TNF and IL22 levels have been consistently found to be up-regulated in patients with active psoriasis (17, 18), the combination of these cytokines may serve as a specific set of biomarkers for this disease (14).

Currently, no cure is available for psoriasis. Although many strategies can help to control the symptoms, psoriasis remains challenging to treat because of its chronic, recurrent nature (19). Therapeutic antibodies targeting proinflammatory cytokines (infliximab, adalimumab, and ustekinumab) and small-molecule drugs targeting lymphocytes (methotrexate and cyclosporine) are effective but have been associated with recurrent infections and immunogenicity (20, 21). Phase 2 clinical trials using immunomodulatory cytokines IL4 (22) and/or IL10 (23) showed rapid improvements in psoriasis patients at well-tolerated doses. However, the short half-lives of these cytokines in the bloodstream require almost continuous, daily administration (IL4, 0.2 to 0.5 μg/kg per day; IL10, 8 μg/kg per day), which represents a major setback for patient compliance and treatment economics (24, 25).

To make immunomodulatory cytokine therapy possible, we designed and engineered human cells that sequentially detected elevated TNF and IL22 levels from a psoriatic flare and, in response, produced therapeutic doses of IL4 and IL10. We implanted these designer cells in mice and demonstrated that the antipsoriatic cytokine converter network could improve skin lesions and restore dermal tissue morphology. Although we apply synthetic circuits to psoriasis, such cell-based implants engineered with sensor-effector gene circuits could be applied to many chronic diseases with known markers: The cells can constantly monitor circulating disease-associated biomarkers; process their differential levels with increasingly complex Boolean logic; and coordinate in situ production, dosing, and delivery of protein therapeutics. These synthetic circuits, which program designer cells to process complex metabolic information, open the door to autonomously prevent, attenuate, or reset acute or chronic medical conditions without constant injections of drugs or cumbersome dosing schedules, and thus provide a new opportunity for personalized medicine.


The synthetic cytokine converter triggers output expression with AND-gate logic

The antipsoriatic cytokine converter was designed as shown in Fig. 1A. We rewired TNF-triggered TNF receptor (TNFR) signaling through nuclear factor κB (NFκB) to a synthetic NFκB-responsive promoter that controlled the expression of human IL22 receptor α (hIL22RA), PNFκB-hIL22RA-pA (pLS25). IL22RA heterodimerizes with endogenous human IL10 receptor β (hIL10RB) to form an IL22-triggered receptor complex, which enables IL22-mediated activation of the Janus kinase (JAK)/signal transducer and activator of transcription (STAT) signaling cascade; we rewired this signaling pathway by connecting ectopically expressed human STAT3 (pLS15) to synthetic STAT3-responsive promoters driving expression of the cytokines IL4 (PSTAT3-mIL4-pA, pLS51) and IL10 (PSTAT3-mIL10-pA, pLS28) (Fig. 1A). Daisy chaining of the TNF- and IL22-responsive signaling cascades provides AND-gate logic integration of proinflammatory cytokine levels, which ensures that IL4 and IL10 are exclusively produced and secreted at the onset of a psoriatic flare. Therefore, the closed-loop designer circuit with AND-gate logic signal processing coordinates the level of the psoriasis-specific cytokines TNF and IL22 to express the anti-inflammatory cytokines IL4 and IL10 (Fig. 1B).

Fig. 1. Design and therapeutic intervention of the antipsoriasis cytokine converter.

(A) Design and assembly of the cytokine converter components. A synthetic gene network consisting of two sequentially interconnected signaling cascades constantly quantifies the inflammatory cytokines TNF and IL22, processes their relative presence with AND-gate logic, and programs the adjusted production of the anti-inflammatory cytokines IL4 and IL10 by human HEK-293T cells. In particular, (i) TNF activates endogenous or ectopically expressed human TNF receptor 1A (hTNFR1A; PhCMV-hTNFR1A-pA, pLS5), which leads to NFκB-triggered expression of the PNFκB-driven hIL22RA (PNFκB-hIL22RA-pA, pLS25). (ii) In the presence of IL22, hIL22RA heterodimerizes with endogenous hIL10RB, which triggers the corresponding JAK/STAT signaling cascade through ectopically expressed human STAT3 (PhCMV-hSTAT3-pA, pLS15). STAT3 translocates to the nucleus and activates the PSTAT3 promoters driving the exclusive expression of murine IL4 (PSTAT3-mIL4-pA, pLS51) and IL10 (PSTAT3-mIL10-pA, pLS28). The cytokine converter’s serial interconnection of the signaling cascades provides AND-gate expression logic. (B) Schematic of the AND-gate-specific cytokine converter–based psoriasis treatment. Psoriasis-associated skin inflammation is based on an erroneous crosstalk in the dermis between keratinocytes and tissue-resident dendritic cells, which hyperstimulate immune cells and result in excessive keratinocyte proliferation and the production and release of the inflammatory cytokines TNF and IL22 in the circulation, which is sensed by the cytokine converter. This process coordinates the expression and release of therapeutic levels of the anti-inflammatory cytokines IL4 and IL10 by the designer cells, which diffuse into the bloodstream and reach the affected skin areas, where they attenuate the psoriasis-associated inflammation. (C and D) Schematic representation and in vitro validation of functional hTNF and hIL22 receptors. (C) HEK-293T cells were transfected with pKR32 (PNFκB-SEAP-pA) to produce SEAP in response to TNF. SEAP expression kinetics over 96 hours using different amounts of mouse TNF (i) or SEAP expression at 48 hours with varying amounts of pKR32 and mTNF (ii). (D) (i) HEK-293T cells were cotransfected with fixed amounts of pLS15 (PhCMV-hSTAT3-pA; 100 ng) and pLS13 (PSTAT3-SEAP-pA; 300 ng) but different ratios of pLS17 (PhCMV-hIL22RA-pA) and pLS18 (PhCMV-hIL10RB-pA) and then cultivated for 48 hours in the presence of different concentrations of mouse IL22. (D) (ii) HEK-293T cells were cotransfected with fixed amounts of pLS17 (20 ng) and pLS13 (300 ng) and with different amounts of pLS15 and cultivated for 48 hours in the presence of different concentrations of mIL22. Data are means ± SD of triplicate experiments (n = 6).

To confirm that the circuit responded in a dose-dependent manner to TNF binding, we cotransfected human embryonic kidney (HEK) 293T cells with different amounts of the constitutive TNFR1A expression vectors PhCMV-hTNFR1A-pA (pLS5) and PNFΚB-SEAP-pA (pKR32). Whereas ectopic expression of TNFR1A (transfection of 5 to 450 ng of PhCMV-hTNFR1A-pA) resulted in constitutive high-level SEAP (secreted embryonic alkaline phosphatase) expression that was insensitive to TNF inputs (fig. S1A), endogenous TNFR1A levels (no transfection of PhCMV-hTNFR1A-pA) enabled TNF dose–responsive SEAP expression [Fig. 1C(i)]. When titrating the amount of transfected PNFΚB-SEAP-pA (pKR32), TNF responsiveness could be fine-tuned to the relevant sensitivity range [Fig. 1C(ii)]. The performance of the synthetic TNF-sensor cascade was also analyzed in human cervical cancer (HeLa), human fibrosarcoma (HT-1080), and Chinese hamster ovary (CHO)–K1 cells, but the HEK-293T cells showed more pronounced gene expression than these cell lines (fig. S1B).

To implement the second part of the synthetic cytokine-sensor cascade, we engineered HEK-293T cells to respond to IL22, which included expression of the cognate IL22 receptor complex, which is composed of IL22RA and IL10RB subunits and STAT3, for downstream signaling (Fig. 1A). Therefore, we cotransfected HEK-293T cells with different combinations and relative amounts of plasmids encoding IL22RA (PhCMV-hIL22RA-pA, pLS17), IL10RB (PhCMV-hIL10RB-pA, pSL18), and STAT3 (PhCMV-hSTAT3-pA, pLS15), along with the SEAP expression vector driven by a synthetic STAT3 promoter, PSTAT3-SEAP-pA (pLS13). These cells were then exposed to various IL22 concentrations [Fig. 1D(i) and (ii)]. The synthetic signaling cascade could be fine-tuned for optimal sensitivity by adjusting the ratio of IL22RA and IL10RB responsiveness, which revealed that ectopic IL10RB expression was expendable [Fig. 1D(i)]. Target gene expression could also be fine-tuned and substantially boosted by ectopic expression of STAT3 in the relevant IL22 concentration range without compromising the device’s tightness [Fig. 1D(ii)]. Although other cell lines were also tested, HEK-293T remained the best-in-class cell type, with optimal response dynamics (fig. S1C).

After validation and optimization of the individual TNF- and IL22-sensor components, we daisy-chained both signaling cascades by placing IL22RA expression under the control of NFκB (PNFκB-hIL22RA-pA, pLS25), thereby providing AND-gate type–exclusive target gene expression in the presence of TNF and IL22 (Fig. 2A). Cotransfection of the HEK-293T cells with PNFκB-hIL22RA-pA (pLS25), PhCMV-hSTAT3-pA (pLS15), and PSTAT3-SEAP-pA (pSL13) confirmed that heterologous IL22RA and STAT3 cooperate with endogenous TNFR1A and IL10RB to drive target gene expression only in the presence of both TNF and IL22 (Fig. 2A). The AND-gate expression logic was also fully reversible and could be repeatedly switched on and off by the addition and withdrawal of the TNF/IL22 cytokine inducer set (Fig. 2B), which is a premise to react to the changing levels of these pathological cytokines during the relapsing-remitting disease course of psoriasis.

Fig. 2. Validation of the cytokine converter’s AND-gate logic in vitro.

(A and B) HEK-293T cells were cotransfected with AND-gate components pLS25 (PNFκB-SEAP-pA), pLS15 (PhCMV-hSTAT3-pA), and pLS13 (PSTAT3-SEAP-pA) to produce SEAP in response to TNF and IL22. (A) HEK-293T cells were cotransfected with pLS25, pLS15, and pLS13 and cultivated for 48 hours in the presence (1) or absence (0) of mouse TNF (0.5 ng/ml) and mouse IL22 (1 ng/ml) according to the truth table. (B) Reversibility of the TNF/IL22-responsive SEAP expression. HEK-293T cells were cotransfected with pLS25, pLS15, and pLS13, and the SEAP expression kinetics were profiled for 96 hours while alternating the presence (+) or absence (−) of TNF (0.5 ng/ml) and IL22 (1 ng/ml) every 24 hours. (C and D) HEK-293T cells were cotransfected with pLS25, pLS15, pLS28 (PSTAT3-mIL10-pA), and pLS51 (PSTAT3-mIL4-pA) and exposed to mouse TNF and/or IL22 for 48 hours before the IL4 (C) and IL10 (D) levels were quantified in the culture supernatant. HEK-293T cells transfected with constitutive mIL4 (PhCMV-mIL4-pA; pLS1) and mIL10 (PhCMV-mIL10-pA; pLS2) expression vectors served as positive controls. Data are means ± SD of triplicate experiments (n = 6).

To link the psoriasis-associated inflammatory input to a therapeutic cytokine output, we designed a synthetic cytokine converter that coordinates the presence of the proinflammatory cytokines TNF and IL22 to the corresponding expression of the anti-inflammatory cytokines IL4 and IL10 using AND-gate logic. Indeed, HEK-293T cells engineered for PNFκB-hIL22RA-pA (pLS25) and PhCMV-hSTAT3-pA (pLS15) as well as for PSTAT3-mIL4-pA (pLS51) and PSTAT3-mIL10-pA (pLS28) exclusively produced anti-inflammatory cytokines when exposed to the proinflammatory cytokines TNF and IL22 in vitro (Fig. 2, C and D). The functionality of the anti-inflammatory cytokines IL4 and IL10 produced by the cytokine converter were profiled using specific reporter cell lines containing IL4 and IL10 receptors linked to SEAP expression (fig. S2, A and B).

Validation of the cytokine converter in mouse models of inflammation

After validation of the cytokine-response dynamics in cultured human cells, we tested the performance of the synthetic cytokine converter to sense TNF and IL22 under inflammatory conditions in vivo in a mouse model of imiquimod-induced psoriasis-like lesions (26). Engineered HEK-293T cytokine converter cells were intraperitoneally implanted into mice in coherent alginate-(poly-l-lysine)-alginate capsules, a biomaterial that is known for its optimal pore size tenability (27), lack of immunogenicity (27), and ability to be vascularized (28), and which has been successfully used in humans (28). To assess whether the implantation itself may elicit an inflammatory response, we treated wild-type mice receiving either no implant or intraperitoneal HEK-293T–containing implants for five consecutive days with topical administration of either imiquimod or control (Vaseline). No significant differences in TNF (Fig. 3A) or IL22 (Fig. 3B) blood levels were observed between the nonimplanted/implanted animals from both treatment groups after 3 and 5 days. Additionally, neither macro- nor microscopic analysis of the skin surface and tissue morphology (Fig. 3C) revealed any significant differences among the treatment groups, confirming that neither the injection-based implantation procedure nor the implant material triggered an inflammatory response, confirming that the synthetic cells were insulated from the mouse immune system inside the alginate capsules.

Fig. 3. Validation controls of cytokine converter components in psoriatic mice.

(A to C) Back-shaved mice were treated daily with topical administration of imiquimod (IMQ; psoriasis induction) or Vaseline (VAS; negative control) and either received no implant (−implant) or control implants (+implant) consisting of microencapsulated pcDNA3.1(+)-transfected HEK-293T cells. Resulting blood TNF (A) and IL22 (B) levels as well as surface and section morphologies of the skin were analyzed for all treatment groups (C). (D) Response of the cytokine converter to psoriasis-unrelated inflammation. Back-shaved mice were treated with imiquimod and Vaseline for 3 days; then implanted with microencapsulated HEK-293T cells cotransfected with pLS25 (PNFκB-hIL22RA-pA), pLS15 (PhCMV-hSTAT3-pA), and pLS13 (PSTAT3-SEAP-pA); and injected daily with phosphate-buffered saline (PBS), different concentrations of S. enterica–derived LPS, 100 μg of poly(I:C), or 1% (w/v) thioglycollate (Thio) before blood SEAP levels were profiled. Data are means ± SD (n = 8 mice). *P < 0.05, **P < 0.005, ***P < 0.0001, Student’s t test.

To analyze the sensitivity and specificity of the cytokine converter, we intraperitoneally implanted AND-gate circuit–modified HEK-293T cells containing the circuitry required to produce SEAP in response to both TNF and IL22 (PNFκB-hIL22RA-pA, pLS25; PhCMV-hSTAT3-pA, pLS15; and PSTAT3-SEAP-pA, pSL13) into mice with imiquimod-induced psoriasis. Control mice also received the engineered cells but were either topically administered Vaseline control or injected with different doses of a Salmonella enterica–derived lipopolysaccharide (LPS) to simulate a bacterial infection; polyinosinic-polycytidylic acid potassium salt [poly(I:C)], a synthetic double-stranded RNA analog to mimic viral infection (29); or thioglycollate, which triggers a psoriasis-unrelated sterile inflammation (30). The SEAP produced by the implanted designer cells in response to the circulating TNF and IL22 was substantially higher in the animals with psoriasis compared to all controls (Fig. 3D), indicating that the AND-gate circuit was specifically activated by psoriasis-like inflammation.

The cytokine converter prevents psoriasis-like plaque formation

To assess the cytokine converter cell’s potential to prevent the onset of a psoriatic flare-up, we intraperitoneally implanted engineered cells into mice and subsequently administered topical imiquimod. The psoriatic symptoms and disease activity of the treated mice were compared to those of control animals receiving an anti-inflammatory treatment with the synthetic cortisol derivative prednisolone (Fig. 4). Whereas the blood levels of the psoriasis-associated cytokines TNF and IL22, as well as other cytokines associated with the pathogenesis of psoriasis, such as IL17 (31), interferon-α (IFNα) (32), and C-X-C motif chemokine 9 (CXCL9) (33), decreased substantially in the cytokine converter–treated animals, in the prednisolone-treated mice, TNF, IFNα, and CXCL9 levels decreased, but the levels of IL22 and IL17 remained unchanged (Fig. 4A).

Fig. 4. Prophylactic psoriasis treatment in mice.

(A to B) Back-shaved mice were treated with topical administration of imiquimod (psoriasis induction) or Vaseline (negative control) and received either the anti-inflammatory drug prednisolone (PDS) or intraperitoneal implants of microencapsulated HEK-293T cells containing the cytokine converter (CC; pLS25, pLS15, pLS28, pLS51; see Fig. 2, C and D). The inflammatory (TNF, IL22, IL17, IFNα, and CXCL9) (A) and anti-inflammatory (IL4 and IL10) (B) blood cytokine levels of all the treatment groups (IMQ, IMQ + CC, IMQ + PDS, and VAS) were profiled on days 3 and 5. (C to F) Representative skin surface morphology, skin section [hematoxylin and eosin (H&E) staining], and cell proliferation (Ki67 staining) images of all the treatment groups were taken on days 3 and 5 (C), and epidermal cell numbers (D), thicknesses (E), and T cell (CD3+) and neutrophil (Ly-6G+) infiltrates (F) were analyzed by quantitative image analyses of eight different mice (the bars indicate the epidermal thickness). ND, not detectable; ns, not significant. Data are means ± SD (n = 12). *P < 0.05, **P < 0.005, ***P < 0.0001, ****P < 0.00001, Student’s t test.

We further profiled the blood levels of anti-inflammatory cytokines in the animals. At baseline, and in line with previous studies characterizing imiquimod-induced psoriasis-like skin inflammation in mice (26), there were higher levels of IL4 and IL10 in the imiquimod-treated mice than in the Vaseline-treated controls (Fig. 4B), which may be explained by the anti-inflammatory response mounted by the immune cells to restore tissue homeostasis (34, 35). The mice with psoriasis that had been implanted with the cytokine converter did not produce substantially higher levels of IL4 compared with baseline, but did produce more IL10 (Fig. 4B); however, prednisolone treatment also resulted in greater IL10 production compared with baseline. The relatively small increase in the IL4 and IL10 levels most likely results from the lower input levels of the proinflammatory cytokines TNF and IL22 during the onset of psoriatic flares in the prophylactic setting (Fig. 4A). Thus, as expected, the cytokine converter responds to lower proinflammatory cytokine input by producing lower output levels of the anti-inflammatory cytokines IL4 and IL10. Additionally, the ramp-up of the anti-inflammatory cytokine production by the cytokine converter is expected to dampen the inflammation and so further reduce the input levels of the proinflammatory cytokines. This feedback control in which anti-inflammatory cytokines decrease production of proinflammatory is well known (36) and has been confirmed to occur after implantation of the cytokine converter during the acute phase (Fig. 5, A and B).

Fig. 5. Cytokine converter–based treatment of acute psoriasis in mice.

(A to B) Back-shaved mice were treated with topical administration of imiquimod (psoriasis induction) or Vaseline (negative control) for 72 hours before microencapsulated HEK-293T cells containing the cytokine converter (CC; pLS25, pLS15, pLS28, pLS51; see Fig. 2, C and D) were implanted intraperitoneally. The inflammatory (TNF, IL22, IL17, IFNα, and CXCL9) (A) and anti-inflammatory (IL4 and IL10) (B) blood cytokine levels of all the treatment groups (IMQ, IMQ + CC, IMQ + PDS, and VAS) were profiled on days 3, 5, and 7. (C to F) Representative skin surface morphology, skin section (H&E staining), and cell proliferation (Ki67 staining) images of all the treatment groups were taken on days 5 and 7 (C), and the epidermal cell numbers (D), thicknesses (E), and T cell (CD3+) and neutrophil (Ly-6G+) infiltrates (F) were scored by quantitative image analyses of eight different mice (the bars indicate the epidermal thickness). (G) Treatment of acute psoriasis by intraperitoneal injections of recombinant IL4 and IL10. Representative skin surface morphology of back-shaved mice treated with topical administration of imiquimod or Vaseline for 72 hours before they received intraperitoneal injections of recombinant IL4 and IL10 (rIL4 and rIL10; 100 ng per mouse) every 6 hours for up to 72 hours. Data are means ± SD (n = 12). *P < 0.05, **P < 0.005, ***P < 0.0001, ****P < 0.00001, Student’s t test.

Despite the small increase of IL4 and IL10 levels produced by the cytokine converter, macroscopic analysis of the mice receiving the engineered cells showed normal skin morphology comparable to that in the healthy (Vaseline) group, and they lacked the erythema formation that is typical of animals with experimental psoriasis even when receiving prednisolone (Fig. 4C). Quantitative analyses of the skin sections revealed significant decreases in cell number (Fig. 4D) and epidermis thickness (Fig. 4E) in the animals treated with the cytokine converter as well as reduced immune cell infiltrations (T cells and neutrophils; Fig. 4F) and/or reduced hyperproliferation of resident keratinocytes, as confirmed by Ki67 staining (Fig. 4C). Overall, these findings indicate that prophylactic treatment using implanted designer cells transgenic for the synthetic cytokine converter prevented the onset of psoriatic inflammation more so than did prednisolone, even if systemic reduction of proinflammatory cytokines did not fully reach the levels of healthy animals that have never suffered from psoriasis (Vaseline treatment group) (Fig. 4A).

The cytokine converter attenuates established psoriasis-like plaques

To assess whether the antipsoriatic cytokine converter could diminish the inflammation of an established psoriasis plaque, we profiled blood cytokine concentrations and skin samples from mice implanted with the engineered cells on day 3 of imiquimod induction, when IL22 reached its peak level (Fig. 3B) and the TNF levels continued to rise (Fig. 3A). When implanted at a later, acute stage of the disease, the cytokine converter cells reduced the proinflammatory cytokines TNF and IL22 (Fig. 5A) and substantially increased the production of the anti-inflammatory cytokines IL4 and IL10 on day 5 (Fig. 5B). There was a substantial drop in blood TNF, IL22, IL17, and IFNα levels on day 7, suggesting that the cytokine converter continued to work for the relevant time span (37); conversely, prednisolone had a shorter-term effect, reducing IL17 and IFNα only up to day 5.

Only the skin of the animals implanted with designer cells containing the antipsoriatic cytokine converter showed reduced psoriasis-like symptoms, such as erythema, scaling, and thickening, compared to the control treatment group (Fig. 5C). This observation was confirmed by histochemical analysis of the corresponding skin sections, which showed reductions of more than 50% in the epidermal cell number (Fig. 5D) and thickness (Fig. 5E) with the engineered cells implanted after disease onset. Furthermore, animals implanted with the cytokine converter showed decreased immune cell infiltrations characterized by lower numbers of T cells and neutrophils in the imiquimod-treated skin sections (Fig. 5F). Prednisolone-treated animals also showed a rapid improvement in skin morphology. Unlike the cytokine converter cells, prednisolone reduced neither IL22 levels (Fig. 5A) nor keratinocyte proliferation (Fig. 5C), suggesting that prednisolone provides more of a general anti-inflammatory response than a specific antipsoriatic response (38). Furthermore, animals suffering from psoriasis that received frequent high-dose injections of recombinant IL4 and IL10 showed no improvement of skin morphology (Fig. 5G), corroborating recent human clinical trials showing that the limited half-life of these antipsoriatic cytokines obviates their success in vivo (24, 25).

The cytokine converter detects cytokines in blood samples from psoriasis patients

Having demonstrated function in vitro and in vivo in mice, we lastly tested the ability of the AND-gate circuit (Fig. 1A) to sense and respond to pathological levels of the proinflammatory cytokines TNF and IL22 in patient samples. Therefore, we established a human blood culture assay in which the microencapsulated circuit-transgenic designer cells engineered to produce SEAP in response to human TNF and IL22 [with PNFκB-hIL22RA-pA (pLS25), PhCMV-hSTAT3-pA (pLS15), and PSTAT3-SEAP-pA (pSL13)] were cultivated in medium containing blood samples from either psoriasis patients or healthy individuals (Fig. 6, B to D). The observation that SEAP levels were exclusively increased in the blood cultures of the psoriasis patients (Fig. 6D) suggests that the cytokine converter is sufficiently sensitive to detect circulating TNF and IL22 in humans.

Fig. 6. Validation of the cytokine converter in blood cultures of psoriasis patients.

(A) Performance of the cytokine converter in response to human inflammatory cytokines hTNF and hIL22. HEK-293T cells were cotransfected with pLS25 (PNFκB-hIL22RA-pA), pLS15 (PhCMV-hSTAT3-pA), and pLS13 (PSTAT3-SEAP-pA) and cultivated for 48 hours in different combinations and concentrations of hTNF and hIL22 before SEAP levels were profiled in the culture supernatant. (B and C) hTNF (B) and hIL22 (C) levels of blood samples from psoriasis patients with different psoriasis area and severity index (PASI) scores. (D) Microencapsulated HEK-293T cells cotransfected with pLS25, pLS15, and pLS13 were cocultivated with blood samples from psoriasis patients or healthy individuals for 48 hours before SEAP levels were profiled in the blood culture supernatant. Data are means ± SD of each patient sample (n = 4) measured in duplicate. Data for healthy donors are averages ± SD (n = 3 pooled donor samples). *P < 0.05, **P < 0.005, ***P < 0.0001, Student’s t test.


Modern medicine consists of taking pills with small-molecule drugs or receiving injections of biologics at regular intervals—both of which have several conceptual limitations that inhibit advanced care of patients suffering from diseases with complex and relapsing dynamics as well as symptom-free flare-up and progression that require daily changing dosing regimens of biopharmaceuticals. Standard drug dosing remains rudimentary and consists of systemic administration of drugs at fixed intervals based on body weight. Furthermore, diagnosis and treatment often occur late in disease pathogenesis, in the acute phase. Thus, current treatment strategies do not address the unmet clinical need whereby metabolic disturbances would be detected at an early phase and reset by controlled interventions. With the advent of personalized medicine, the need to combine diagnostics with therapeutic interventions has been recognized by inherently linking the two (also known as “theranostics”).

Synthetic biology is one field that is capable of tethering therapy to early diagnosis, by reassembling biological parts in a systematic, rational, and predictable manner to program novel cellular behavior. Synthetic biology has recently enabled the design of complex gene circuits that process molecular input and output with Boolean logic and near-digital precision (39) and is currently moving toward medical applications (40, 41). Advances in therapeutic network design have resulted in the successful coupling of biosensor-based detection of disease-specific metabolites or biomarkers (diagnosis) to expression of therapeutic transgenes in a closed-loop manner. We therefore sought to design for the first time a circuit that can be integrated into theranostic “designer cells” and dynamically interface with host metabolism in real time (42) to treat a chronic inflammatory condition for which modern medicine is lacking: psoriasis. The hope is that such cells could be used to treat many inflammatory and immunological diseases where early detection is crucial but for which symptoms often appear long after metabolic damage has been done.

Boolean AND gates are exclusively activated when all trigger compounds are simultaneously present (39) and are therefore particularly suited to increase the specificity of the gene circuit and insulate it from unrelated inputs. In the context of psoriasis, IL22 has different functionalities that depend on the inflammatory context and the combination of cytokines present in a specific microenvironment (43). Likewise, excessive TNF levels alone are also not specific for psoriasis because they also increase during infection and psoriasis-unrelated inflammation. We therefore designed our cytokine converter based on AND-gate logic to require simultaneous input of IL22 and TNF, which made the synthetic circuit exclusively activated in the context of psoriasis, insensitive to bacterial and viral infection as well as to psoriasis-unrelated inflammation. In this way, the device also did not interfere with the immune system’s capacity to mount inflammatory responses.

Through detailed characterization in vitro, in vivo in an animal model of psoriasis, and ex vivo with human blood, we have demonstrated that the cytokine converter operates with strict AND-gate logic and is exclusively induced when TNF and IL22 reach clinically relevant levels; is sufficiently sensitive to score the levels of TNF and IL22 in the blood of psoriasis patients, which correlate with the PASI values of the patient; is insensitive to bacterial and viral infections and psoriasis-unrelated inflammation, and enables TNF and IL22 level–dependent production of therapeutic concentrations of the clinically validated antipsoriatic cytokines IL4 and IL10. Human clinical trials have recently established IL4 and IL10 as effective antipsoriatic cytokines (22, 23), but their short half-lives requiring almost continuous administration (24, 25) have stopped further industry development. Indeed, we demonstrated that frequent high-dose injections of recombinant IL4 and IL10 were unable to improve the skin morphology of mice suffering from acute psoriasis, which suggests that in situ production, dosing, and delivery of biopharmaceuticals by designer cell–based therapies such as the cytokine converter may drive the use of certain biopharmaceuticals into a viable therapy. The production of IL4 and IL10 by our cytokine converter decreased TNF and IL22 concentrations in a closed-loop fashion; reduced associated cytokines IL17, IFNα, and CXCL9 as well as corresponding immune cell infiltrations; and restored skin morphology, epidermal thickness, and keratinocyte quiescence.

Although the cytokine converter prevented the development of psoriatic lesions in mice, it apparently produced higher levels of the anti-inflammatory cytokines IL4 and IL10 when attenuating established psoriasis. Because the levels of the proinflammatory input cytokines TNF and IL22 are higher during an established psoriasis than during flare-up, the cytokine converter is expected to produce lower levels of the anti-inflammatory/psoriatic cytokines IL4 and IL10 in the preventive setting. However, early detection of a medical condition followed by rapid production, secretion, and systemic release is the hallmark process of our designer cell–based theranostic treatment strategy (44). The cytokine converter is sufficiently sensitive to detect the flare-up of psoriasis and to produce therapeutic levels to attenuate the disease in an animal model.

Although the cytokine converter exclusively contains human genetic components, is engineered into human cells, and responds to circulating cytokines as we intended, there are still translational hurdles before theranostic designer cells will be routinely used for the treatment of psoriasis in humans. To adapt our synthetic cells to the clinic, we will need to address critical design parameters, including the use of autologous cells, scaling of the system to provide therapeutic levels of the cytokines for a human (versus a mouse), and development of an implant that stores the designer cells in a single device. We envision that the final therapy will be based on patient-derived autologous cell batches that are produced, engineered with the cytokine converter, validated for optimal patient-specific response and dosing performance, and frozen for storage. The designer cells will be filled into appropriate containers and implanted into the body where they automatically connect to the bloodstream (28), monitor the psoriasis-specific cytokine levels, and coordinate the therapeutic response. The designer cell implants will be preferably placed subcutaneously because they can be removed by a minimal ambulant intervention in case of complications or replaced at regular intervals (for example, every 3 to 4 months) because of fibrosis. Freeze-thaw cycles have been established for autologous cells (45), and hydrogels such as alginate have also been validated in human clinical trials (28). However, because hydrogels have poor mechanical properties and a risk of leakage (46), semipermeable plastic containers, such as Encaptra (47), are currently in human clinical trials for cell-based therapies (NCT02239354). The results of these pioneering clinical trials, evaluating in vitro differentiated human stem cells inside an immune-protecting and retrievable encapsulation medical device for diabetes therapy, will teach us to what extent cell-based therapies translate and scale from mouse to man.

With the increasing refinement of metabolite sensors (48), the integration and processing capacities of designer cells reaching the complexity of digital electronics (39), and the clinically validated protein therapeutics available from the biologics era, the time has now come to program therapeutic gene networks for various diseases. Given the modular configuration of these therapeutic networks, they may be readily tailored within months for a particular disease phenotype and thus provide new opportunities in future gene- and cell-based therapies.


Study design

The objective of the study was to capitalize on synthetic biology design principles to engineer human cells with a closed-loop therapeutic cytokine converter that constantly monitors the levels of the proinflammatory cytokines TNF and IL22 in circulation (increased concomitantly in psoriasis) and uses AND-gate logic to program dose-dependent expression of the clinically relevant, therapeutic anti-inflammatory cytokines IL4 and IL10. The cytokine converter was assembled from human genetic components and engineered into human cells. After full characterization of the individual components and the tuning of the specificity, sensitivity, and control dynamics in cell culture, transgenic human cells containing the cytokine converter were microencapsulated in autovascularizing, immunoprotective, and safe alginate beads and implanted intraperitoneally into mice topically treated with imiquimod to induce human-like psoriasis (26). The designer cells were either implanted at the start of imiquimod treatment, to evaluate disease prevention, or implanted into mice that had already been treated with imiquimod for 3 days and had developed psoriatic lesions characterized by disease-specific morphological skin alterations and a systemic increase of proinflammatory cytokines. The cytokine converter was also tested in blood samples from psoriasis patients.

Throughout the study, animals were randomly allocated to the individual treatment groups, and the experimenter was blinded to the analysis of all samples. Neither animals nor samples were excluded from the study. In vitro experiments were done in triplicate each containing six samples, and animal studies included 8 to 12 mice per treatment group as specified in the figure legends. Blood samples of healthy donors (n = 3) and psoriasis patients (n = 4) were measured in duplicate. At least 4 weeks before PASI values were determined and blood samples were taken, the psoriasis patients did not receive any systemic treatment.

Components of the antipsoriatic cytokine converter

Comprehensive design and construction details for all the expression vectors are provided in table S1.

Cell culture and transfection

HEK-293T [American Type Culture Collection (ATCC): CRL-11268], HeLa (ATCC: CCL-2), and HT-1080 (ATCC: CCL-121) were cultivated in Dulbecco’s modified Eagle’s medium (DMEM; Invitrogen) supplemented with 10% fetal bovine serum (FBS; cat. no. F7524, lot no. 022M3395, Sigma-Aldrich) and 1% (v/v) penicillin/streptomycin solution (Sigma-Aldrich). Wild-type CHO-K1 (ATCC: CCL-61) were cultured in ChoMaster HTS (Cell Culture Technologies) supplemented with 5% FBS and 1% penicillin/streptomycin solution. All cell types were cultivated at 37°C in a humidified atmosphere containing 5% CO2. Cell concentration and viability were profiled using a CASY Cell Counter and Analyzer System Model TT (Roche Diagnostics). For the cotransfections (39), 0.5 × 105 cells were diluted in 0.4 ml of culture medium and seeded in a 24-well plate 24 hours before cotransfection. The cells were then incubated for 6 hours with 200 μl of FBS-free DMEM containing 3 μg of polyethyleneimine (molecular weight, 40,000; Polysciences Inc.) and 750 ng of total DNA (for the cotransfections, an equal amount of plasmid DNA was used unless otherwise indicated). After cotransfection, the culture medium was replaced, and the engineered cells were used for a dedicated experiment.

Cytokine profiling

The following recombinant cytokines were purchased from PeproTech Inc.: hIL22 (cat. no. 200-22, lot. no. 0102246), mIL22 (cat. no. 210-22, lot. no. 0602257), hTNF (cat. no. 300-01A, lot. no. 0906CY25), and mTNF (cat. no. 315-01A, lot. no. 121054). Recombinant mIL4 (cat. no. 404-ML-010) and mIL10 (cat. no. 417-ML-005) were purchased from R&D Systems. Cytokine concentrations were quantified in both the cell culture supernatants and the bloodstream of the treated animals using the following enzyme-linked immunosorbent assays (ELISAs) according to the manufacturer’s instructions: mTNF (cat. no. 900-M53), mIL22 (cat. no. 900-M246), mIL4 (cat. no. 900-M14), mIL10 (cat. no. 900-M21), and mIL17 (cat. no. 900-K392), purchased from PeproTech Inc.; mouse IFNα (cat. no. MBS2506010), supplied by MyBioSource; and mouse CXCL9 (cat. no. MCX900), provided by R&D Systems. Plasma cytokine levels of the psoriasis patients and the healthy individuals were profiled using the following ELISAs: hTNF (Chongqing Biospes Co., cat. no. BEK1212) and hIL22 (Mabtech AB Biotechnologie, cat. no. 3475-1H-6).

Implant production

Cell implants were produced by microencapsulating antipsoriatic (pLS15/pLS25/pLS28/pLS51) and control [pcDNA3.1(+) or pLS15/pLS25/pLS13] circuit-transgenic HEK-293T cells into coherent alginate-(poly-l-lysine)-alginate capsules (400 μm, 200 cells per capsule) using an Inotech Encapsulator Research IE-50R (Büchi Labortechnik AG) set to the following parameters: 0.2-mm single nozzle, stirrer speed control at 5 U, 20-ml syringe with a flow rate of 410 U, nozzle vibration frequency of 1024 Hz, and 900 V for capsule dispersion (49). The integrity and quality of the capsules were confirmed by microscopy (Leica DMIL LED, 5× objective; Leica Microsystems AG).

Psoriatic mouse model

To induce psoriasis-like skin inflammation, 62.5 mg of the clinically licensed Aldara cream (5%; 3M Pharmaceuticals) containing 3.125 mg of the active compound imiquimod was topically administered daily onto the skin of back-shaved 8-week-old female BALB/c mice (Charles River Laboratories). The control mice were treated with Vaseline (Unilever).

In vivo treatment of psoriasis

All the experiments involving animals were performed according to the directives of the European Community Council (2010/63/EU) and were approved by the French Republic (no. 69266309). On day 1 (prophylactic psoriasis treatment) or 72 hours (acute psoriasis therapy) after the administration of imiquimod or Vaseline, 800 μl of serum-free DMEM containing 2 × 106 microencapsulated (200 cells per capsule) antipsoriatic circuit-transgenic HEK-293T cells were intraperitoneally implanted. The levels of cytokines (mTNF, mIL22, mIL17, CXCL9, mIL4, and mIL10) in blood were profiled on days 3, 5 and 7, unless indicated otherwise.

To confirm that the antipsoriatic designer network was insensitive to acute inflammatory responses, such as those triggered by bacterial and viral infections or psoriasis-unrelated inflammation, 8-week-old female BALB/c mice were intraperitoneally implanted with 800 μl of serum-free DMEM containing 2 × 106 microencapsulated (200 cells per capsule) antipsoriatic circuit-transgenic HEK-293T cells and were treated with daily 200-μl injections of S. enterica–derived LPS (cat. no. L7770) (0, 0.01, 0.1, or 1.0 μg/ml in PBS), poly(I:C) (cat. no. P9582-5MG, Sigma-Aldrich) (0.5 μg/ml in PBS), or thioglycollate [1% (w/v) thioglycollate in distilled H2O; cat. no. 70157, Sigma-Aldrich]. Blood SEAP levels of the animals from all the treatment groups were quantified 24 and 48 hours after administration of LPS, poly(I:C), and thioglycollate. The imiquimod-treated mice served as positive controls, and the Vaseline-treated animals served as negative controls. Prednisolone (Streuli Pharma AG), used as a treatment control, was reconstituted in 200 μl of PBS and administered at a dose of 0.5 mg/kg per mouse.


To score the cell numbers and visualize the tissue morphology, 3-μm paraffin-embedded mouse skin sections were stained with hematoxylin (J. T. Baker) and eosin (Thermo Scientific). Dividing cells were stained for Ki67 (cat. no. RM-9106-S1, Thermo Scientific), T cells for CD3 (cat. no. ab5690, Abcam), and neutrophils for Ly-6G (cat. no. bs-2576R, Bioss Antibodies) using a Ventana DiscoveryUltra immunohistochemistry device (Roche Diagnostics) according to the manufacturer’s instructions and were visualized by light microscopy (Leica DMIL LED, 20× objective; Leica Microsystems AG). The light micrographs of eight randomly selected skin sections from each treatment group were analyzed using ImageJ (National Institutes of Health; to determine the thickness of the epidermis and the cell numbers in the dermal and epidermal layers.

Human blood culture assay

All the experiments involving human blood samples were approved by the Ethics Committee of the ETH Zurich (EK 2012-N-42). The enrolled patients showed moderate to severe plaque-type psoriasis with a PASI value of ≥10, and none of the patients had received any systemic treatment for a minimum of 4 weeks before the blood analysis. Whole blood from healthy individuals (n = 3) and psoriasis patients (n = 4, with different PASI values, analyzed individually) was collected using standard venous blood collection tubes (BD Vacutainer, cat. no. 366480; Becton Dickinson AG), and 0.5 ml of each blood sample was immediately added to 0.5 ml of cell culture medium [RPMI 1640, PAA Laboratories; supplemented with 1% (v/v) penicillin/streptomycin] containing 2 × 106 microencapsulated pLS15/pLS25/pLS13-transgenic HEK-293T cells (same cell batch as those used as control mouse implants). Blood samples were then incubated in a 24-well plate for 48 hours at 37°C in a humidified atmosphere containing 5% CO2 before scoring the SEAP levels in the blood culture supernatants, according to the protocol for cell culture supernatants (50).

Statistical analyses

In vitro data were obtained from triplicate experiments (n = 2 samples per experiment). In vivo treatment groups included at least eight animals. After assessing normality assumptions and group SDs, a t test was considered more appropriate than nonparametric group comparison statistics such as the Wilcoxon rank sum test. Therefore, all group comparisons were analyzed by Student’s t test (cutoff of P < 0.05) using GraphPad Prism 6 software (GraphPad Software Inc.).



Fig. S1. Validation of the cytokine converter components.

Fig. S2. Validation of the production of functional anti-inflammatory cytokines IL4 and IL10 using an autocrine mammalian signal transduction assay.

Table S1. Plasmids and oligonucleotides used and designed in this study.


  1. Acknowledgments: We thank N. Yawalkar for obtaining human blood samples and corresponding patient PASI values; P. Saxena, M. Mueller, and S. Auslaender for generous advice; K. Roessger for providing the pKR32 plasmid; M. Daoud-El Baba and M. Gilet for support with the animal study; S. Bichet, A. Bogucki, S. Kondo, and J. Scholl for processing immunohistochemistry samples; B. M. Lang for assistance with the statistical analysis. M.F. thanks the “Région Alsace” and the “Communauté Urbaine de Strasbourg” for the award of a Gutenberg Excellence Chair. Funding: This work was supported by a European Research Council (ERC) advanced grant (no. 321381) and in part by the National Centre of Competence in Research (NCCR) Molecular Systems Engineering. Author contributions: L.S., B.G., and M.F. planned the study, designed the circuit, analyzed results, and wrote the manuscript. L.S. assembled and tested the circuit and performed the in vitro work. G.C.-E.H. performed the in vivo studies. Competing interests: The authors declare that they have no competing interests. Data and materials availability: All data pertaining to this study are in the paper. All genetic components are available with a material transfer agreement (contact M.F.).
View Abstract

Navigate This Article