Combinatorial discovery of antibacterials via a feature-fusion based machine learning workflow.

The discovery of new antibacterials within the vast chemical space is crucial in combating drug-resistant bacteria such as methicillin-resistant Staphylococcus aureus (MRSA). However, the traditional approach of screening the entire chemical library in an ergodic manner can be laborious and time-consuming. Machine learning-assisted screening of antibacterials alleviates the exploration effort but suffers from the lack of reliable and related datasets. To address these challenges, we devised a combinatorial library comprising over 110 000 candidates based on the Ugi reaction. A focused library was subsequently generated through uniform sampling of the entire library to narrow down the preliminary screening scale. A novel feature-fusion architecture called the latent space constraint neural network was developed which incorporated both fingerprint and physicochemical molecular descriptors to predict the antibacterial properties. This integration allowed the model to leverage the complementary information provided by these descriptors and improve the accuracy of predictions. Three lead compounds that demonstrated excellent efficacy against MRSA while alleviating drug resistance were identified. This workflow highlights the integration of machine learning with the combinatorial chemical library to expedite high-quality data collection and extensive data mining for antibacterial screening.

A paradigm for high-throughput screening of cell-selective surfaces coupling orthogonal gradients and machine learning-based cell recognition.

The combinational density of immobilized functional molecules on biomaterial surfaces directs cell behaviors. However, limited by the low efficiency of traditional low-throughput experimental methods, investigation and optimization of the combinational density remain daunting challenges. Herein, we report a high-throughput screening set-up to study biomaterial surface functionalization by integrating photo-controlled thiol-ene surface chemistry and machine learning-based label-free cell identification and statistics. Through such a strategy, a specific surface combinational density of polyethylene glycol (PEG) and arginine-glutamic acid-aspartic acid-valine peptide (REDV) leads to high endothelial cell (EC) selectivity against smooth muscle cell (SMC) was identified. The composition was translated as a coating formula to modify medical nickel-titanium alloy surfaces, which was then proved to improve EC competitiveness and induce endothelialization. This work provided a high-throughput method to investigate behaviors of co-cultured cells on biomaterial surfaces modified with combinatorial functional molecules.

Identification of potent antimicrobial peptides via a machine-learning pipeline that mines the entire space of peptide sequences.

Systematically identifying functional peptides is difficult owing to the vast combinatorial space of peptide sequences. Here we report a machine-learning pipeline that mines the hundreds of billions of sequences in the entire virtual library of peptides made of 6-9 amino acids to identify potent antimicrobial peptides. The pipeline consists of trainable machine-learning modules (for performing empirical selection, classification, ranking and regression tasks) assembled sequentially following a coarse-to-fine design principle to gradually narrow down the search space. The leading three antimicrobial hexapeptides identified by the pipeline showed strong activities against a wide range of clinical isolates of multidrug-resistant pathogens. In mice with bacterial pneumonia, aerosolized formulations of the identified peptides showed therapeutic efficacy comparable to penicillin, negligible toxicity and a low propensity to induce drug resistance. The machine-learning pipeline may accelerate the discovery of new functional peptides.

A deep-learning-based workflow to deal with the defocusing problem in high-throughput experiments.

The increasing throughput of experiments in biomaterials research makes automatic techniques more and more necessary. Among all the characterization methods, microscopy makes fundamental contributions to biomaterials science where precisely focused images are the basis of related research. Although automatic focusing has been widely applied in all kinds of microscopes, defocused images can still be acquired now and then due to factors including background noises of materials and mechanical errors. Herein, we present a deep-learning-based method for the automatic sorting and reconstruction of defocused cell images. First, the defocusing problem is illustrated on a high-throughput cell microarray. Then, a comprehensive dataset of phase-contrast images captured from varied conditions containing multiple cell types, magnifications, and substrate materials is prepared to establish and test our method. We obtain high accuracy of over 0.993 on the dataset using a simple network architecture that requires less than half of the training time compared with the classical ResNetV2 architecture. Moreover, the subcellular-level reconstruction of heavily defocused cell images is achieved with another architecture. The applicability of the established workflow in practice is finally demonstrated on the high-throughput cell microarray. The intelligent workflow does not require a priori knowledge of focusing algorithms, possessing widespread application value in cell experiments concerning high-throughput or time-lapse imaging.

Label-Free and In Situ Identification of Cells via Combinational Machine Learning Models.

Cell identification and counting in living and coculture systems are crucial in cell interaction studies, but current methods primarily rely on complicated and time-consuming staining techniques. Here, a label-free method to precisely recognize, identify, and instantly count cells in situ in coculture systems via combinational machine learning models s presented. A convolutional neural network (CNN) model is first used to generate virtual images of cell nuclei based on unlabeled phase-contrast images. Coordinates of all the cells are then returned according to the virtual nucleus images using two clustering algorithms. Finally, phase-contrast images of single cells are cropped based on the coordinates and sent into another CNN model for cell-type identification. This combinational approach is highly automatic and efficient, which requires few to no manual annotations of images in the training phase. It shows practical performance in different cell culture conditions including cell ratios, densities, and substrate materials, having great potential in real-time cell tracking and analyzing.

A facile method for high-throughput screening of drug-eluting coatings in droplet microarrays based on ultrasonic spray deposition.

Coating modification such as drug-eluting coating is one of the most important approaches for the functionalization of biomedical devices. However, the throughputs are limited in conventional coating methods and the concept of miniaturization is rarely fulfilled. A droplet microarray (DMA), as a unique high-throughput platform, can avoid cross-contamination and reduce the consumption of materials which is inherently suitable for coating research yet is difficult to apply with coating materials via traditional methods. Here, we bring up a facile method based on ultrasonic spray deposition to integrate coating materials into a DMA. Several common polymer materials were selected to fabricate a DMA, and the obtained DMA showed the ability to anchor water droplets and form specific patterns. Coating arrays with a typical sandwich structure were also prepared for the high-throughput screening of drug-eluting coatings to demonstrate the potential of the platform in coating research. This developed method is efficient and compatible and enriches the choices of materials that can be applied in DMAs.

High-throughput hyaluronic acid hydrogel arrays for cell selective adhesion screening.

As a component of extracellular matrix (ECM), hyaluronic acid (HA) has plenty of applications in the biomedical field such as tissue engineering. Due to its non-adhesive nature, HA requires further grafting of functional molecules for cell related study. RGD and YIGSR are two kinds of cell adhesion peptides. YIGSR enhances endothelial cell (EC) adhesion, which is important for endothelialization after implantation of stents to prevent in-stent restenosis. However, the effect of combined densities of these peptides for EC and smooth muscle cell (SMC) adhesion has not been explored in a quantitative and high-throughput manner. In this work, single or orthogonal gradient densities of RGD and YIGSR were grafted onto the HA hydrogel array surfaces using thiol-norbornene click chemistry. Optimized peptide combinations for EC preponderant adhesion were found in hydrogel arrays and confirmed by scaling samples.

Rapid build-up of high-throughput screening microarrays with biochemistry gradients via light-induced thiol-ene "click" chemistry.

Microarrays have become extremely powerful experimental tools for high-throughput screening of cell behaviors in multivariate microenvironments. Herein, a microarray-based high-throughput platform with biochemistry gradients was developed using poly(limonene carbonate) (PLimC) as a substrate through thiol-ene click chemistry. ATR-IR, XPS, Raman spectrum, and water contact angle results demonstrated that the sulfhydryl molecules, including PEG (polyethylene glycol) and RGD (arginine-glycine-aspartic acid) peptide, could be grafted onto PLimC substrates, while the grafting density could be well controlled by regulating the intensity of UV irradiation. Then, microarrays with a gradient of RGD grafting density were fabricated by using UV irradiation patterned by a photomask and a gradient light filter. Adhesion experiments of smooth muscle cells and 3T3-L1 mouse embryonic fibroblast cells proved that the cell behaviors were highly determined by the RGD density. This platform puts forward a facile, high-throughput method to study the effect of biochemical signal density on cell behaviors.

Hierarchical Capillary Coating to Biofunctionlize Drug-Eluting Stent for Improving Endothelium Regeneration.

The drug-eluting stent (DES) has become one of the most successful and important medical devices for coronary heart disease, but yet suffers from insufficient endothelial cell (EC) growth and intima repair, eventually leading to treatment failure. Although biomacromolecules such as vascular endothelial growth factor (VEGF) would be promising to promote the intima regeneration, combining hydrophilic and vulnerable biomacromolecules with hydrophobic drugs as well as preserving the bioactivity after harsh treatments pose a huge challenge. Here, we report on a design of hierarchical capillary coating, which composes a base solid region and a top microporous region for incorporating rapamycin and VEGF, respectively. The top spongy region can guarantee the efficient, safe, and controllable loading of VEGF up to 1 µg/cm2 in 1 minute, providing a distinctive real-time loading capacity for saving the bioactivity. Based on this, we demonstrate that our rapamycin-VEGF hierarchical coating impressively promoted the competitive growth of endothelial cells over smooth muscle cells (ratio of EC/SMC~25) while relieving the adverse impact of rapamycin to ECs. We further conducted the real-time loading of VEGF on stents and demonstrate that the hierarchical combination of rapamycin and VEGF showed remarkable endothelium regeneration while maintaining a very low level of in-stent restenosis. This work paves an avenue for the combination of both hydrophobic and hydrophilic functional molecules, which should benefit the next generation of DES and may extend applications to diversified combination medical devices.

Humidity-Triggered Relaxation of Polyelectrolyte Complexes as a Robust Approach to Generate Extracellular Matrix Biomimetic Films.

Generating a biofunctional film that can mimic the extracellular matrix (ECM) in an efficient and robust technique that may have great potential for medical devices, tissue engineering, and regenerative medicines. Herein, a facile approach to generate ECM biomimetic films based on the humidity-triggered relaxation of polyelectrolyte complex (PEC) nanoparticles is reported. The poly(l-lysine) and hyaluronan are precomplexed and sprayed onto a substrate, which, via a trigger of vaporous water, can be transformed into an even and stable film. The spontaneous polymer chain interfusion (diffusion coefficient ≈1.01 × 10-9  cm2  s-1 ) under saturated humidity, allowing for the rapid reorganization (within 30 min) of film morphology and structure is demonstrated. A controllable and scalable way for the loading of diversified bioactive agents, as well as on-demand modulation of stiffness is further presented. Moreover, the high-throughput arrays and programmed patterns can be easily completed, suggesting huge potentials that surpass those of state-of-the-art methods. Combined with high efficiency and flexible functionalization, it is believed that this approach should be beneficial for extending the practical applications of PEC films, such as medical implants, chip detectors, and so on.

Biodegradable phosphorylcholine copolymer for cardiovascular stent coating.

Phosphorylcholine (PC) based polymer coatings with excellent biocompatibility have shown successful commercialization in drug-eluting stents. However, poor degradability represents a challenge in the application of biodegradable stents. Herein, a biodegradable phosphorylcholine copolymer is developed based on one-step radical ring-opening polymerization (RROP). This copolymer was synthesized by copolymerization of a PC unit, degradable ester (2-methylene-1,3-dioxepane, MDO) unit and non-degradable butyl methacrylate (BMA) unit, which showed ratio controllability by changing the monomer ratio during polymerization. We demonstrated that the copolymer with the ratio of 34% MDO, 19% MPC and 47% BMA could form a stable coating by ultrasonic spray, and showed good blood compatibility, anti-adhesion properties, biodegradability, and rapamycin eluting capacity. In vivo study revealed its promising application as a biodegradable stent coating. This work provides a facile path to add biodegradability into PC based polymers for further bio-applications.

Rapid Buildup Arrays with Orthogonal Biochemistry Gradients via Light-Induced Thiol-Ene "Click" Chemistry for High-Throughput Screening of Peptide Combinations.

The concept of high-throughput screening sheds new light on fabrication and analysis of materials. Herein, a combinatorial surface-modified platform with biochemical gradients was developed through thiol-ene "click" chemistry by adjusting the intensity of ultraviolet (UV) irradiation. Contact angle, X-ray photoelectron spectroscopy, and ellipsometry measurement results demonstrated that the sulfhydryl molecules including polyethylene glycol and RGD (arginine-glycine-aspartic acid) and REDV (arginine-glutamic acid-aspartic acid-valine) peptides can be directly attached onto alkene-modified substrates, in which the graft density can be well controlled by the intensity of UV irradiation. The multistep attachment of different molecules onto substrates is archived via the multistep UV-initiated thiol-ene "click" reaction. The high-throughput arrays with the gradient density of single ligand and the orthogonal gradient density of two ligands were rapidly fabricated via the one-step UV gradient irradiation and the two-step orthogonal UV gradient-initiated thiol-ene "click" reaction. Endothelial cells (ECs) and smooth muscle cells (SMCs) were cocultured on the array with the orthogonal gradient density of RGD and REDV to screen the peptide combination with high EC selectivity, which is essential for in situ endothelialization during stent implant. From 64, 8 × 8, combinations investigated, a special combinatorial surface representing the really high competitiveness of ECs over SMCs was screened. This platform puts forward a facile, high-throughput method to study the combinatorial variation of biochemical signals to cell behavior.

A photodynamic antibacterial spray-coating based on the host-guest immobilization of the photosensitizer methylene blue.

In the "post-antibiotic era", healthcare-associated infection has become a global problem that threatens public health and causes huge economic losses. The development of antibacterial coatings based on non-antibiotic strategies is particularly important as drug-resistant bacteria continue to evolve. Photodynamic coatings are a high potential method to treat bacteria, however, the aggregation of photosensitizers on the coating affects the photodynamic capacity seriously. Herein, a photodynamic coating is developed based on the host-guest interaction between ß-cyclodextrin and the photosensitizer methylene blue (MB). The host-guest interaction avoids aggregation of MB and results in a high singlet oxygen quantum yield. Consequently, efficient photoantibacterial activity towards methicillin-resistant Staphylococcus aureus is achieved by the photodynamic coating with very low MB density (0.53 ± 0.06 µg cm-2).

Thermo-triggered ultrafast self-healing of microporous coating for on-demand encapsulation of biomacromolecules.

Medical coatings cooperated with biomacromolecules can regulate biological events and tissue responses, thus increasing medical implant longevity and providing improved and/or new therapeutic functions. In particular, medical coatings, which can load the correct species and doses of biomacromolecules according to individual diagnoses, will significantly optimize treatment effects and satisfy the rising clinical need of "precision medicine". Herein, we report on a dynamic microporous coating with an ultrafast self-healing property to fulfill the "load-and-play" concept for "precision medicine". A structure-switchable coating based on poly(ε-caprolactone)-poly(ethylene glycol)-poly(ε-caprolactone) triblock copolymer network is constructed. The coating can be switched to microporous morphology via a water swelling and freeze-drying process. Then, through a mild thermo-trigger as low as 40 °C, this spongy coating can undergo self-healing to switch back to a pore-free structure within minutes to even 5 s. Based on this dynamic coating, we suggest a simple and versatile method to encapsulate biomacromolecules for surface-mediated delivery. The ultrafast self-healing of the microporous structure enables uniform incorporation of biomacromolecules with an easily achieved high loading of albumin of 16.3 µg/cm2 within 1 min. More importantly, controllable encapsulation can be realized by simple control of the concentration of the loading solution. We further demonstrate that the encapsulated biomacromolecules retained their bioactivity. This work may benefit clinicians with flexibility to provide personalized medical coatings for individual patients during treatment.