Neural Network Analysis Identifies Scaffold Properties Necessary for In Vitro Chondrogenesis in Elastin-like Polypeptide Biopolymer Scaffolds

Abstract

The successful design of biomaterial scaffolds for articular cartilage tissue engineering requires an understanding of the impact of combinations of material formulation parameters on diverse and competing functional outcomes of biomaterial performance. This study sought to explore the use of a type of unsupervised artificial network, a self-organizing map, to identify relationships between scaffold formulation parameters (crosslink density, molecular weight, and concentration) and 11 such outcomes (including mechanical properties, matrix accumulation, metabolite usage and production, and histological appearance) for scaffolds formed from crosslinked elastin-like polypeptide (ELP) hydrogels. The artificial neural network recognized patterns in functional outcomes and provided a set of relationships between ELP formulation parameters and measured outcomes. Mapping resulted in the best mean separation amongst neurons for mechanical properties and pointed to crosslink density as the strongest predictor of most outcomes, followed by ELP concentration. The map also grouped formulations together that simultaneously resulted in the highest values for matrix production, greatest changes in metabolite consumption or production, and highest histological scores, indicating that the network was able to recognize patterns amongst diverse measurement outcomes. These results demonstrated the utility of artificial neural network tools for recognizing relationships in systems with competing parameters, toward the goal of optimizing and accelerating the design of biomaterial scaffolds for articular cartilage tissue engineering.

Screening of hyaluronic acid–poly(ethylene glycol) composite hydrogels to support intervertebral disc cell biosynthesis using artificial neural network analysis

“…The sulfated glycosaminoglycan (sGAG) and DNA contents for papain digests of cell-hydrogel samples cultured for 7 days (n=5–6 per hydrogel formulation/cell type) were determined [41, 66]. In brief, papain digests of each cell seeded HA-PEG hydrogel (10 units/mg papain: Sigma Aldrich #P4762, St. Louis, MO; 1X PBS, 5mM cysteine HCL, 5mM EDTA, pH=6.0; mixed for 2h at 37°C then filtered; digested for less than 24 hours at 60°C) were analyzed by a dimethylmethylene blue (DMMB) assay for sGAG content using commercially available chondroitin-4-sulfate as a standard (Sigma).…”

“…Statistical approaches such as principal components analysis or cluster analyses have been used to group biomaterial features based on multiple “outcomes” for material selection, particularly when large datasets are generated. Artificial neural network (ANN) analysis is another, probability-based approach to identify relationships between material formulations parameters and biological outcomes [41–43]. ANN analyses to generate a “self-organizing map” will identify connectivity amongst formulations based on similarities in a diverse array of measured outcomes [44–46].…”

“…This iterative process is repeated for all measured arrays (samples) until a trained network of connected neurons is generated, leading to a “map” showing similarities across sets of measured variables that has parallels to outcomes from cluster analyses. In our prior work that used ANN to identify a suitable elastin-like polypeptide scaffolds for cartilage tissue engineering [41], polypeptide scaffolds formed from moderately crosslinked polymers and low protein density were mapped together based on the observation that they provide for the greatest matrix production and cell viability; nevertheless, the physical properties in these formulations were considered “weaker” than those of the native cartilage tissue but factored less into the material selection than the weighting assigned by ANN to the biological outcomes. A specific advantage to using this ANN in material optimization for biomedical applications is that ANN analysis does not require prior input parameters or defined assumptions of relationships among data components [45–50].…”

Nano and Micro-Structures of Elastin-Like Polypeptide-Based Materials and Their Applications: Recent Developments

“…[65][66][67][68][69][70] In a study by Nettles et al, 16 lysine-containing ELP formulations were evaluated for their mechanical properties and their ability to promote cartilage matrix synthesis by encapsulated primary chondrocytes. 69 These ELP formulations di®ered in the frequencies of lysine along the ELP backbone (termed the K-period), their solution concentrations and molecular weights. An arti¯cial neural network analysis was used to evaluate relationships between the properties of ELP-associated biological and mechanical outcomes.…”

Biologically derived scaffolds

“…Elastin tubes fi lled with agarose gel containing basic fi broblast growth factor for sustained release of the growth factor showed signifi cantly more cell infi ltration at 28 days compared to those without growth factor (Kurane et al, 2007). Elastin scaffolds formed from cross-linked elastin-like polypeptide hydrogels were investigated to identify relationships between scaffold formulation parameters (crosslink density, molecular weight, and concentration) and properties including mechanical, matrix accumulation, metabolite use and production, and histological appearance (Nettles et al, 2010). Crosslink density was the strongest predictor of most outcomes related to neuron functions, followed by elastin-like polymer concentration (Nettles et al, 2010).…”