Experimental validation of machine-learning based spectral-spatial power evolution shaping using Raman amplifiers.

We experimentally validate a real-time machine learning framework, capable of controlling the pump power values of Raman amplifiers to shape the signal power evolution in two-dimensions (2D): frequency and fiber distance. In our setup, power values of four first-order counter-propagating pumps are optimized to achieve the desired 2D power profile. The pump power optimization framework includes a convolutional neural network (CNN) followed by differential evolution (DE) technique, applied online to the amplifier setup to automatically achieve the target 2D power profiles. The results on achievable 2D profiles show that the framework is able to guarantee very low maximum absolute error (MAE) (<0.5 dB) between the obtained and the target 2D profiles. Moreover, the framework is tested in a multi-objective design scenario where the goal is to achieve the 2D profiles with flat gain levels at the end of the span, jointly with minimum spectral excursion over the entire fiber length. In this case, the experimental results assert that for 2D profiles with the target flat gain levels, the DE obtains less than 1 dB maximum gain deviation, when the setup is not physically limited in the pump power values. The simulation results also prove that with enough pump power available, better gain deviation (less than 0.6 dB) for higher target gain levels is achievable.

Assessing the predictive capabilities of design heuristics and coarse-grain simulation toward understanding and optimizing site-specific covalent immobilization of ß-lactamase.

For industrial applications, covalent immobilization of enzymes provides minimum leakage, recoverability, reusability, and high stability. Yet, the suitability of a given site on the enzyme for immobilization remains a trial-and-error procedure. Here, we investigate the reliability of design heuristics and a coarse-grain molecular simulation in predicting the optimum sites for covalent immobilization of TEM-1 ß-lactamase. We utilized Escherichia coli-lysate-based cell-free protein synthesis (CFPS) to produce variants containing a site-specific incorporated unnatural amino acid with a unique moiety to facilitate site directed covalent immobilization. To constrain the number of potential immobilization sites, we investigated the predictive capability of several design heuristics. The suitability of immobilization sites was determined by analyzing expression yields, specific activity, immobilization efficiency, and stability of variants. These experimental findings are compared with coarse-grain simulation of TEM-1 domain stability and thermal stability and analyzed for a priori predictive capabilities. This work demonstrates that the design heuristics successfully identify a subset of locations for experimental validation. Specifically, the nucleotide following amber stop codon and domain stability correlate well with the expression yield and specific activity of the variants, respectively. Our approach highlights the advantages of combining coarse-grain simulation and high-throughput experimentation using CFPS to identify optimal enzyme immobilization sites.

Assessing site-specific PEGylation of TEM-1 ß-lactamase with cell-free protein synthesis and coarse-grained simulation.

PEGylation is a broadly used strategy to enhance the pharmacokinetic properties of therapeutic proteins. It is well established that the location and extent of PEGylation have a significant impact on protein properties. However, conventional PEGylation techniques have limited control over PEGylation sites. Emerging site-specific PEGylation technology provides control of PEG placement by conjugating PEG polymers via click chemistry reaction to genetically encoded non-canonical amino acids. Unfortunately, a method to rapidly determine the optimal PEGylation location has yet to be established. Here we seek to address this challenge. In this work, coarse-grained molecular dynamic simulations are paired with high-throughput experimental screening utilizing cell-free protein synthesis to investigate the effect of site-specific PEGylation on the two-state folder protein TEM-1 ß-lactamase. Specifically, the conjugation efficiency, thermal stability, and enzymatic activity are studied for the enzyme PEGylated at several different locations. The results of this analysis confirm that the physical properties of the PEGylated protein vary considerably with PEGylation site and that traditional design recommendations are insufficient to predict favorable PEGylation sites. In this study, the best predictor of the most favorable conjugation site is coarse-grained simulation. Thus, we propose a dual combinatorial screening approach in which coarse-grained molecular simulation informs site selection for high-throughput experimental verification.

Towards detection of SARS-CoV-2 RNA in human saliva: A paper-based cell-free toehold switch biosensor with a visual bioluminescent output.

The COVID-19 pandemic has illustrated the global demand for rapid, low-cost, widely distributable and point-of-care nucleic acid diagnostic technologies. Such technologies could help disrupt transmission, sustain economies and preserve health and lives during widespread infection. In contrast, conventional nucleic acid diagnostic procedures require trained personnel, complex laboratories, expensive equipment, and protracted processing times. In this work, lyophilized cell-free protein synthesis (CFPS) and toehold switch riboregulators are employed to develop a promising paper-based nucleic acid diagnostic platform activated simply by the addition of saliva. First, to facilitate distribution and deployment, an economical paper support matrix is identified and a mass-producible test cassette designed with integral saliva sample receptacles. Next, CFPS is optimized in the presence of saliva using murine RNase inhibitor. Finally, original toehold switch riboregulators are engineered to express the bioluminescent reporter NanoLuc in response to SARS-CoV-2 RNA sequences present in saliva samples. The biosensor generates a visible signal in as few as seven minutes following administration of 15 µL saliva enriched with high concentrations of SARS-CoV-2 RNA sequences. The estimated cost of this test is less than 0.50 USD, which could make this platform readily accessible to both the developed and developing world. While additional research is needed to decrease the limit of detection, this work represents important progress toward developing a diagnostic technology that is rapid, low-cost, distributable and deployable at the point-of-care by a layperson.

Inverse design of a Raman amplifier in frequency and distance domains using convolutional neural networks.

We present a convolutional neural network architecture for inverse Raman amplifier design. This model aims at finding the pump powers and wavelengths required for a target signal power evolution in both distance along the fiber and in frequency. Using the proposed framework, the prediction of the pump configuration required to achieve a target power profile is demonstrated numerically with high accuracy in C-band considering both counter-propagating and bidirectional pumping schemes. For a distributed Raman amplifier based on a 100 km single-mode fiber, a low mean set (0.51, 0.54, and 0.64 dB) and standard deviation set (0.62, 0.43, and 0.38 dB) of the maximum test error are obtained numerically employing two and three counter-, and four bidirectional propagating pumps, respectively.

Rapid RNase inhibitor production to enable low-cost, on-demand cell-free protein synthesis biosensor use in human body fluids.

Human body fluids contain biomarkers which are used extensively for prognostication, diagnosis, monitoring, and evaluation of different treatments for a variety of diseases and disorders. The application of biosensors based on cell-free protein synthesis (CFPS) offers numerous advantages including on-demand and at-home use for fast, accurate detection of a variety of biomarkers in human fluids at an affordable price. However, current CFPS-based biosensors use commercial RNase inhibitors to inhibit different RNases present in human fluids and this reagent is approximately 90% of the expense of these biosensors. Here the flexible nature of Escherichia coli-lysate-based CFPS was used for the first time to produce murine RNase Inhibitor (m-RI) and to optimize its soluble and active production by tuning reaction temperature, reaction time, reduced potential, and addition of GroEL/ES folding chaperons. Furthermore, RNase inhibition activity of m-RI with the highest activity and stability was determined against increasing amounts of three human fluids of serum, saliva, and urine (0%-100% v/v) in lyophilized CFPS reactions. To further demonstrate the utility of the CFPS-produced m-RI, a lyophilized saliva-based glutamine biosensor was demonstrated to effectively work with saliva samples. Overall, the use of CFPS-produced m-RI reduces the total reagent costs of CFPS-based biosensors used in human body fluids approximately 90%.

Coarse-grained simulation of PEGylated and tethered protein devices at all experimentally accessible surface residues on ß-lactamase for stability analysis and comparison.

PEGylated and surface-tethered proteins are used in a variety of biotechnological applications, but traditional methods offer little control over the placement of the functionalization sites on the protein. Fortunately, recent experimental methods functionalize the protein at any location on the amino acid sequence, so the question becomes one of selecting the site that will result in the best protein function. This work shows how molecular simulation can be used to screen potential attachment sites for surface tethering or PEGylation. Previous simulation work has shown promise in this regard for a model protein, but these studies are limited to screening only a few of the surface-accessible sites or only considered surface tethering or PEGylation separately rather than their combined effects. This work is done to overcome these limitations by screening all surface-accessible functionalization sites on a protein of industrial and therapeutic importance (TEM-1) and to evaluate the effects of tethering and PEGylation simultaneously in an effort to create a more accurate screen. The results show that functionalization site effectiveness appears to be a function of super-secondary and tertiary structures rather than the primary structure, as is often currently assumed. Moreover, sites in the middle of secondary structure elements, and not only those in loops regions, are shown to be good options for functionalization-a fact not appreciated in current practice. Taken as a whole, the results show how rigorous molecular simulation can be done to identify candidate amino acids for functionalization on a protein to facilitate the rational design of protein devices.

Rapid sensing of clinically relevant glutamine concentrations in human serum with metabolically engineered E. coli-based cell-free protein synthesis.

Bioavailable glutamine (Gln) is critical for metabolism, intestinal health, immune function, and cell signaling. Routine measurement of serum Gln concentrations could facilitate improved diagnosis and treatment of severe infections, anorexia nervosa, chronic kidney disease, diabetes, and cancer. Current methods for quantifying tissue Gln concentrations rely mainly on HPLC, which requires extensive sample preparation and expensive equipment. Consequently, patient Gln levels may be clinically underutilized. Cell-free protein synthesis (CFPS) is an emerging sensing platform with promising clinical applications, including detection of hormones, amino acids, nucleic acids, and other biomarkers. In this work, in vitro E. coli amino acid metabolism is engineered with methionine sulfoximine to inhibit glutamine synthetase and create a CFPS Gln sensor. The sensor features a strong signal-to-noise ratio and a detection range ideally suited to physiological Gln concentrations. Furthermore, it quantifies Gln concentration in the presence of human serum. This work demonstrates that CFPS reactions which harness the metabolic power of E. coli lysate may be engineered to detect clinically relevant analytes in human samples. This approach could lead to transformative point-of-care diagnostics and improved treatment regimens for a variety of diseases including cancer, diabetes, anorexia nervosa, chronic kidney disease, and severe infections.

Engineering Cell-Free Protein Synthesis for High-Yield Production and Human Serum Activity Assessment of Asparaginase: Toward On-Demand Treatment of Acute Lymphoblastic Leukemia.

Acute lymphocytic leukemia (ALL) is a common childhood cancer in the United States, with over 6000 new cases diagnosed each year. Administration of bacterial asparaginase (ASNase) has improved survival rates to nearly 80%, however these therapeutics have high incidence of immunological neutralization and serum activity must be monitored for most effective treatment regimens. Here, a 72% improvement in cell-free protein synthesis (CFPS) of FDA approved l-asparaginase (crisantaspase) is demonstrated by employing an aspartate-fed-batch reactor format. A CFPS-based ASNase activity assay as a tool for therapeutic regimentation and production quality control is also presented. This work suggests that shelf-stable and low-cost Escherichia coli-based CFPS reactions may be employed on-demand to 1) synthesize biologics on-site for patient administration, 2) verify biologic activity for dosage calculations, and 3) monitor therapeutic activity in human serum during the treatment regimen. The combination of both therapeutic production and activity assessment introduces a concept of synergistic utility for bacterial cell lysates in modern medical treatment. Indeed, recent work with CFPS biosensors supports a not-too-distant future when shelf-stable E. coli CFPS systems are used to diagnose, treat, and monitor treatment of diseases in the clinical setting.

Streamlining the preparation of "endotoxin-free" ClearColi cell extract with autoinduction media for cell-free protein synthesis of the therapeutic protein crisantaspase.

An "endotoxin-free" E. coli-based cell-free protein synthesis system has been reported to produce therapeutic proteins rapidly and on-demand. However, preparation of the most complex CFPS reagent - the cell extract - remains time-consuming and labor-intensive because of the relatively slow growth kinetics of the endotoxin-free ClearColiTMBL21(DE3) strain. Here we report a streamlined procedure for preparing E. coli cell extract from ClearColi™ using auto-induction media. In this work, the term auto-induction describes cell culture media which eliminates the need for manual induction of protein expression. Culturing Clearcoli™ cells in autoinduction media significantly reduces the hands-on time required during extract preparation, and the resulting "endotoxin-free" cell extract maintained the same cell-free protein synthesis capability as extract produced with traditional induction as demonstrated by the high-yield expression of crisantaspase, an FDA approved leukemia therapeutic. It is anticipated that this work will lower the barrier for researchers to enter the field and use this technology as the method to produce endotoxin-free E. coli-based extract for CFPS.