The first version of this article inadvertently neglected to include Ting Gong as third author. This has now been corrected online. The authors regret the error.

The control of gene expression noise is important for improving drug treatment and the performance of synthetic biological systems. Previous work has tuned gene expression noise by changing the rate of transcription initiation, mRNA degradation, and mRNA translation. However, these methods are invasive: they require changes to the target genetic components. Here, we create an orthogonal system based on CRISPR-dCas9 to tune gene expression noise. Specifically, we modulate the gene expression noise of a reporter gene in Escherichia coli by incorporating CRISPR activation and repression (CRISPRar) simultaneously in a single cell. The CRISPRar uses a single dCas9 that recognizes two different single guide RNAs (sgRNA). We build a library of sgRNA variants with different expression activation and repression strengths. We find that expression noise and mean of a reporter gene can be tuned independently by CRISPRar. Our results suggest that the expression noise is tuned by the competition between two sgRNAs that modulate the binding of RNA polymerase to promoters. The CRISPRar may change how we tune expression noise at the genomic level. Our work has broad impacts on the study of gene functions, phenotypical heterogeneity, and genetic circuit control.

Protein networks can be assembled in vitro for basic biochemistry research, drug screening, and the creation of artificial cells. Two standard methodologies are used: manual pipetting and pipetting robots. Manual pipetting has limited throughput in the number of input reagents and the combination of reagents in a single sample. While pipetting robots are evident in improving pipetting efficiency and saving hands-on time, their liquid handling volume usually ranges from a few to hundreds of microliters. Microfluidic methods have been developed to minimize the reagent consumption and speed up screening but are challenging in multifactorial protein studies due to their reliance on complex structures and labeling dyes. Here, we engineered a new impact-printing-based methodology to generate printed microdroplet arrays containing water-in-oil droplets. The printed droplet volume was linearly proportional (R² = 0.9999) to the single droplet number, and each single droplet volume was around 59.2 nL (coefficient of variation = 93.8%). Our new methodology enables the study of protein networks in both membrane-unbound and -bound states, without and with anchor lipids DGS-NTA(Ni), respectively. The methodology is demonstrated using a subnetwork of mitogen-activated protein kinase (MAPK). It takes less than 10 min to prepare 100 different droplet-based reactions, using <1 μL reaction volume at each reaction site. We validate the kinase (ATPase) activity of MEK1 (R4F)* and ERK2 WT individually and together under different concentrations, without and with the selective membrane attachment. Our new methodology provides a reagent-saving, efficient, and flexible way for protein network research and related applications.

The recent development of synthetic biology has expanded the capability to design and construct protein networks outside of living cells from the bottom-up. The new capability has enabled us to assemble protein networks for the basic study of cellular pathways, expression of proteins outside cells, and building tissue materials. Furthermore, the integration of natural and synthetic protein networks has enabled new functions of synthetic or artificial cells. Here, we review the underlying technologies for assembling protein networks in liposomes, water-in-oil droplets, and biomaterials from the bottom-up. We cover the recent applications of protein networks in biological transduction pathways, energy self-supplying systems, cellular environmental sensors, and cell-free protein scaffolds. We also review new technologies for assembling protein networks, including multiprotein purification methods, high-throughput assay screen platforms, and controllable fusion of liposomes. Finally, we present existing challenges towards building protein networks that rival the complexity and dynamic response akin to natural systems. This review addresses the gap in our understanding of synthetic and natural protein networks. It presents a vision towards developing smart and resilient protein networks for various biomedical applications.