Dual-responsive synthetic gene circuit for dynamic biologic drug delivery via inflammatory and circadian signaling pathways

Abstract

Background: Engineered cells provide versatile tools for precise, tunable drug delivery, especially when synthetic stimulus-responsive gene circuits are incorporated. In many complex disease conditions, endogenous pathologic signals such as inflammation can vary dynamically over different time scales. For example, in autoimmune conditions such as rheumatoid arthritis or juvenile idiopathic arthritis, local (joint) and systemic inflammatory signals fluctuate daily, peaking in the early morning, but can also persist over long periods of time, triggering flare-ups that can last weeks to months. However, treatment with disease-modifying anti-rheumatic drugs is typically provided at continuous high doses, regardless of disease activity and without consideration for levels of inflammatory signals. In previous studies, we have developed cell-based drug delivery systems that can automatically address the different scales of flares using either chronogenetic circuits (i.e., clock gene-responsive elements) that can be tuned for optimal drug delivery to dampen circadian variations in inflammatory levels or inflammation-responsive circuits (i.e., NF-κB-sensitive elements) that can respond to sustained arthritis flares on demand with proportional synthesis of drug. The goal of this study was to develop a novel dual-responsive synthetic gene circuit that responds to both circadian and inflammatory inputs using OR-gate logic for both daily timed therapeutic output and enhanced therapeutic output during chronic inflammatory conditions.

Results: We developed a synthetic gene circuit driven by tandem inflammatory NF-κB and circadian E'-box response elements. When engineered into induced pluripotent stem cells that were chondrogenically differentiated, the gene circuit demonstrated basal-level circadian output with enhanced stimulus-responsive output during an inflammatory challenge shown by bioluminescence monitoring. Similarly, this system exhibited enhanced therapeutic levels of biologic drug interleukin-1 receptor antagonist (IL-1Ra) during an inflammatory challenge in differentiated cartilage pellets. This dual-responsive therapeutic gene circuit mitigated both the inflammatory response as measured by bioluminescence reporter output and tissue-level degradation during conditions mimicking an arthritic flare.

Conclusions: The dual-responsive synthetic gene circuit developed herein responds to input cues from two key homeostatic transcriptional networks, enabling dynamic and tunable output. This proof-of-concept approach has the potential to match drug delivery to disease activity for optimal outcomes that addresses the complex environment of inflammatory arthritis.

Keywords: Arthritis; Cartilage; Cell engineering; Chronogenetic; Circadian rhythms; Gene circuits; Synthetic biology.

Fig. 1

The inflammatory and circadian dual-responsive gene circuit produces a biologic using OR-gate logic. Inflammatory-driven biologic production is programmed by NF-κB response elements (REs) that promote downstream gene activation during inflammatory signaling, such as when IL-1 binds its receptor (IL-1R). This leads to IL-1Ra production that blocks inflammatory signaling, providing on-demand, feedback controlled therapeutic delivery. Chronogenetic delivery is driven by the central internal circadian feedback loop, where binding of BMAL1-CLOCK dimers in E’-box elements activates downstream gene expression, as well as the PER and CRY proteins that dimerize to block BMAL1-CLOCK activation, generating a self-regulated timing mechanism. Using E’-box elements upstream of the therapeutic gene permits times daily delivery that can coincide with daily changes in inflammation. Combing these approaches, the dual-input circuit utilizes NF-κB REs followed by E’-boxes to drive therapeutic gene expression that responds to inflammatory signaling or programmed time of day.

Fig. 2

Distinct, disease-relevant cues activate inflammatory or chronogenetic gene circuits. (A) The inflammatory-driven NF-κB circuit activates on-demand in response to IL-1β challenge, whereas the circadian E’-box circuit is autonomously regulated by clock-controlled feedback on a 24-hour on/off cycle. (B) Dynamic tracking of gene circuit response after inflammatory challenge (indicated by the red line at 24 hours) showed an increased response for NFκB-LUC, (C) quantitatively demonstrated by increased area under the curve (AUC). (D) Chronogenetic output by E’box-GFP-LUC was dampened with IL-1β challenge, demonstrated by reduced (E) AUC and (F) baseline of the sinusoidal curve fit and (G) lengthened period. Figures show mean and SEM, n = 4–5/condition. Groups not sharing the same letter are significant (p < 0.05) by one-way ANOVA with Tukey’s multiple comparisons test (C– 0 vs. 0.1: p < 0.0001, 0 vs. 1: p < 0.0001, 0.1 vs. 1: p < 0.0001; E– 0 vs. 0.1: p = 0.0001, 0 vs. 1: p < 0.0001, 0.1 vs. 1: p = 0.0049; F– 0 vs. 0.1: p = 0.0002, 0 vs. 1: p < 0.0001, 0.1 vs. 1: p = 0.0048; G– 0 vs. 0.1: p < 0.0001, 0 vs. 1: p = 0.0010, 0.1 vs. 1: p = 0.1900)

Fig. 3

The dual-responsive inflammatory and chronogenetic design activates in response to either cue. (A) Continuous bioluminescence monitoring of NFκB.E’box-GFP-LUC response reveals basal-level circadian output prior to challenge and in the absence of stimulation, shown by the signal’s (B) 24-hour period. (C) When challenged, circuit activation increased, demonstrated by AUC. In alignment with bioluminescence, (D) Il1rn expression and (F) IL-1Ra accumulation was increased with challenge in comparison to non-challenged controls, resulting a significant difference between groups by 24 hours post-challenge and (E, G) increased AUC. Figures show mean and SEM (A-D, F) or standard error (E, G), n = 3–7/condition. Different letters indicate significant differences (p < 0.05) by one-way ANOVA with Tukey’s multiple comparisons test (B, C) or two-way ANOVA with Sidak’s multiple comparisons test (D, F) (C– 0 vs. 0.1: p < 0.0001, 0 vs. 1: p < 0.0001, 0.1 vs. 1: p < 0.0001; D– 0 vs. 1 for 0 h: p > 0.9999, 4 h: p = 0.6287, 8 h: p = 0.7287, 12 h: p = 0.9887, 24 h: p = 0.0025; F– 0 vs. 1 for 0 h: p > 0.9999, 4 h: p > 0.9999, 8 h: p > 0.9999, 12 h: p = 0.9979, 24 h: p = 0.0183)

Fig. 4

Therapeutic production from NFκB.E’box-IL1Ra mitigated inflammatory activation response. Continuous bioluminescence monitoring of NFκB.E’box-GFP-LUC response (A) without or (C) with the inclusion of NFκB.E’box-IL1Ra demonstrated reduction of the circuit activation following challenge quantified by (B, D, E) AUC and (F) relative peak activation. Figures show mean and SEM, n = 3–6/condition. Groups not sharing the same letter are significant (p < 0.05) by one-way ANOVA with Tukey’s multiple comparisons test (B, D, F) or t-test (E). (B– pre-IL-1β 0 vs. pre-IL-1β 1: p = 0.9419, pre-IL-1β 0 vs. post-IL-1β 0: p > 0.9999, pre-IL-1β 0 vs. post-IL-1β 1: p < 0.0001, pre-IL-1β 1 vs. post-IL-1β 0: p = 0.9419, pre-IL-1β 1 vs. post-IL-1β 1: p < 0.0001, post-IL-1β 0 vs. post-IL-1β 1: p < 0.0001; D– pre-IL-1β 0 vs. pre-IL-1β 1: p = 0.9897, pre-IL-1β 0 vs. post-IL-1β 0: p > 0.9999, pre-IL-1β 0 vs. post-IL-1β 1: p = 0.0080, pre-IL-1β 1 vs. post-IL-1β 0: p = 0.9897, pre-IL-1β 1 vs. post-IL-1β 1: p = 0.0081, post-IL-1β 0 vs. post-IL-1β 1: p = 0.0080; E– p < 0.0001; F– pre-IL-1β 0 vs. pre-IL-1β 1: p < 0.0001, pre-IL-1β 0 vs. post-IL-1β 0: p > 0.9999, pre-IL-1β 0 vs. post-IL-1β 1: p = 0.9985, pre-IL-1β 1 vs. post-IL-1β 0: p < 0.0001, pre-IL-1β 1 vs. post-IL-1β 1: p < 0.0001, post-IL-1β 0 vs. post-IL-1β 1: p = 0.9990)

Fig. 5

The dual-responsive gene circuit protected a model of arthritis from tissue-level damage. IL-1Ra production was significantly increased in comparison to non-transduced (NT) controls and with challenge at both (A) 24 hours and (B) 72 hours post-challenge. This contributed to protected cartilage matrix, shown by (C) quantified sulfonated GAG/DNA and (D) representative images of histological sections stained with Safranin-O, Fast-Green, and hematoxylin (n = 2/condition; scale: 200 μm). (E) At a gene expression level, loss of cartilage associated genes aggrecan (Acan) and type-II collagen (Col2a1) and increase in inflammatory associated genes C-C Motif Chemokine Ligand 2 (Ccl2) and interleukin-6 (Il6) were improved by the therapeutic circuit. Figures show mean and SEM, n = 3–8/condition (A-C, E). Groups not sharing the same letter are significant by one-way ANOVA with Tukey’s multiple comparison test (detailed statistical p-values can be found in Table S1-S4)