More Like this

1. Absolutely robust controllers for chemical reaction networks

   https://doi.org/10.1098/rsif.2020.0031  

   Kim, Jinsu; Enciso, German (May 2020, Journal of The Royal Society Interface)

   null (Ed.)

   In this work, we design a type of controller that consists of adding a specific set of reactions to an existing mass-action chemical reaction network in order to control a target species. This set of reactions is effective for both deterministic and stochastic networks, in the latter case controlling the mean as well as the variance of the target species. We employ a type of network property called absolute concentration robustness (ACR). We provide applications to the control of a multisite phosphorylation model as well as a receptor–ligand signalling system. For this framework, we use the so-called deficiency zero theorem from chemical reaction network theory as well as multiscaling model reduction methods. We show that the target species has approximately Poisson distribution with the desired mean. We further show that ACR controllers can bring robust perfect adaptation to a target species and are complementary to a recently introduced antithetic feedback controller used for stochastic chemical reactions. 

   more » « less
2. Robust and tunable signal processing in mammalian cells via engineered covalent modification cycles

   https://doi.org/10.1038/s41467-022-29338-w  

   Jones, Ross D.; Qian, Yili; Ilia, Katherine; Wang, Benjamin; Laub, Michael T.; Del Vecchio, Domitilla; Weiss, Ron (March 2022, Nature Communications)

   Abstract Engineered signaling networks can impart cells with new functionalities useful for directing differentiation and actuating cellular therapies. For such applications, the engineered networks must be tunable, precisely regulate target gene expression, and be robust to perturbations within the complex context of mammalian cells. Here, we use bacterial two-component signaling proteins to develop synthetic phosphoregulation devices that exhibit these properties in mammalian cells. First, we engineer a synthetic covalent modification cycle based on kinase and phosphatase proteins derived from the bifunctional histidine kinase EnvZ, enabling analog tuning of gene expression via its response regulator OmpR. By regulating phosphatase expression with endogenous miRNAs, we demonstrate cell-type specific signaling responses and a new strategy for accurate cell type classification. Finally, we implement a tunable negative feedback controller via a small molecule-stabilized phosphatase, reducing output expression variance and mitigating the context-dependent effects of off-target regulation and resource competition. Our work lays the foundation for establishing tunable, precise, and robust control over cell behavior with synthetic signaling networks. 

   more » « less
3. Siroheme synthase orients substrates for dehydrogenase and chelatase activities in a common active site

   https://doi.org/10.1038/s41467-020-14722-1  

   Pennington, Joseph M.; Kemp, Michael; McGarry, Lauren; Chen, Yu; Stroupe, M. Elizabeth (February 2020, Nature Communications)

   Abstract Siroheme is the central cofactor in a conserved class of sulfite and nitrite reductases that catalyze the six-electron reduction of sulfite to sulfide and nitrite to ammonia. InSalmonella entericaserovar Typhimurium, siroheme is produced by a trifunctional enzyme, siroheme synthase (CysG). A bifunctional active site that is distinct from its methyltransferase activity catalyzes the final two steps, NAD⁺-dependent dehydrogenation and iron chelation. How this active site performs such different chemistries is unknown. Here, we report the structures of CysG bound to precorrin-2, the initial substrate; sirohydrochlorin, the dehydrogenation product/chelation substrate; and a cobalt-sirohydrochlorin product. We identified binding poses for all three tetrapyrroles and tested the roles of specific amino acids in both activities to give insights into how a bifunctional active site catalyzes two different chemistries and acts as an iron-specific chelatase in the final step of siroheme synthesis. 

   more » « less
4. Catalase-peroxidase (KatG): a potential frontier in tuberculosis drug development

   https://doi.org/10.1080/10409238.2025.2470630  

   Liu, Aimin (February 2025, Critical Reviews in Biochemistry and Molecular Biology)

   Cox, Michael M (Ed.)

   Mycobacterium tuberculosis (Mtb) depends on the bifunctional enzyme catalase-peroxidase (KatG) for survival within the host. KatG exhibits both catalase and peroxidase activities, serving distinct yet critical roles. While its peroxidase activity is essential for activating the frontline tuberculosis drug isoniazid, its catalase activity protects Mtb from oxidative stress. This bifunctional enzyme is equipped with a unique, protein-derived cofactor, methionine-tyrosine-tryptophan (MYW), which enables catalase activity to efficiently disproportionate hydrogen peroxide in phagocytes. Recent studies reveal that the MYW cofactor naturally exists in a hydroperoxylated form (MYW-OOH) when cell cultures are grown under ambient conditions. New findings highlight a dynamic regulation of KatG activity, wherein the modification of the protein cofactor is removable-from MYW-OOH to MYW-at body temperature or in the presence of micromolar concentrations of hydrogen peroxide. This reversible modification modulates KatG's dual activities: MYW-OOH inhibits catalase activity while enhancing peroxidase activity, demonstrating the chemical accessibility of the cofactor. Such duality positions KatG as a unique target for tuberculosis drug development. Therapeutic strategies that exploit cofactor modification could hold promise, particularly against drug-resistant strains with impaired peroxidase activity. By selectively inhibiting catalase activity, these approaches would render Mtb more vulnerable to oxidative stress while enhancing isoniazid activation-a double-edged strategy for combating tuberculosis. 

   more » « less
5. Second-Order Provable Defenses against Adversarial Attacks

   Singla, Sahil; Feizi, Soheil (January 2020, International Conference on Machine Learning (ICML))

   null (Ed.)

   A robustness certificate is the minimum distance of a given input to the decision boundary of the classifier (or its lower bound). For {\it any} input perturbations with a magnitude smaller than the certificate value, the classification output will provably remain unchanged. Exactly computing the robustness certificates for neural networks is difficult since it requires solving a non-convex optimization. In this paper, we provide computationally-efficient robustness certificates for neural networks with differentiable activation functions in two steps. First, we show that if the eigenvalues of the Hessian of the network are bounded, we can compute a robustness certificate in the l2 norm efficiently using convex optimization. Second, we derive a computationally-efficient differentiable upper bound on the curvature of a deep network. We also use the curvature bound as a regularization term during the training of the network to boost its certified robustness. Putting these results together leads to our proposed {\bf C}urvature-based {\bf R}obustness {\bf C}ertificate (CRC) and {\bf C}urvature-based {\bf R}obust {\bf T}raining (CRT). Our numerical results show that CRT leads to significantly higher certified robust accuracy compared to interval-bound propagation (IBP) based training. We achieve certified robust accuracy 69.79\%, 57.78\% and 53.19\% while IBP-based methods achieve 44.96\%, 44.74\% and 44.66\% on 2,3 and 4 layer networks respectively on the MNIST-dataset. 

   more » « less