Page 239 - IJB-10-2

International Journal of Bioprinting G40T60@WNT5A promotes osteoblast differentiation

transcription factors were predicted through the TTRUST 2.8. PPI analysis
website; binding sites for transcription factors and target Protein–protein interaction analysis was performed using
genes through the JASPAR website; genes involved in E3 sequences of differentially expressed genes available from
ubiquitin ligase activity through the browser database; the STRING database(https://cn.string-db.org/cgi/),
and potential ubiquitination sites for genes through with the species set to human. The analyzed results were
the BioGRID database. STRING database was used for imported into Cytoscape 3.6.0 software for visualization,
analyzing protein–protein interaction (PPI) among genes, and the top 30 nodes in the PPI network were ranked
while Cytoscape 3.5.1 software was used to visualize the based on their degree values. 50,51
PPI network. Kyoto Encyclopedia of Genes and Genomes
(KEGG) pathway enrichment analysis on differentially 2.9. Preparation of brackets
expressed genes was performed using the “ClusterProfiler” A 12% w/v gelatin solution was prepared by adding gelatin
powder to distilled water, followed by homogenization at
package in R software. A significance level of P <0.05 was 65°C for 15 min to make an aqueous gelatin solution. The
considered statistically significant.
beta-tricalcium phosphate (β-TCP) powder and gelatin
2.5. Screening of differential gene expression solution were mixed in different ratios to prepare scaffolds
Differential gene expression analysis of mRNA was (G80T20, gelatin:β-TCP, 80:20; G60T40, gelatin:β-TCP,
performed using the “edgeR” package in R language, 60:40; G40T60, gelatin:β-TCP, 40:60), followed by stirring
based on read count numbers derived from mRNA. The at 60°C for 2 h. Brackets in various sizes were printed using
criteria for selecting differentially expressed genes were a 3D bioprinting system, with a speed of 300 mm/min and
set as |log2FC| >0.5 and P <0.05. The “ggplot2” package a syringe size of 21 G. Each of the brackets had a thickness
in R was used to create volcano plots and boxplots of of 500 µm and a spacing of 700 µm. After bioprinting, the
differentially expressed genes. Additionally, the “heatmap” samples were cross-linked in 25% glutaraldehyde vapor at
package in R was used to generate heat maps of differential room temperature for 24 h. Unreacted aldehyde groups
gene expression. 42,43 were capped with a 0.1 M glycine solution. 42
2.6. GO and KEGG functional enrichment analyses 2.10. Rheological testing
Functional enrichment analysis on differentially expressed The rheology of composite hydrogel scaffolds comprising
genes was performed. The “ClusterProfiler” R package gelatin and β-TCP was tested using a rheometer (MARS
was used for Gene Ontology (GO) enrichment analysis, 40; Haake, Germany) with parallel plates (diameter: 60
which performs GO functional enrichment analysis at mm). The viscosity and storage modulus (G’) at 25°C were
three levels—biological process (BP), cellular component measured, and the gel point at a heating rate of 1°C/min
52
(CC), and molecular function (MF). KEGG pathway from 25 to 45°C was determined.
analysis was performed using the PANTHER database 2.11. Printability assessment
(http://pantherdb.org/), with a significance threshold of The printability of hydrogels prepared was evaluated with
P <0.05, and the results were then visualized using the the aid of optical imaging. The bioprinting conditions for
ImageGP website. 44 composite hydrogels, despite the different mixing ratios,
remained the same, while the optimal pressure for each
2.7. Machine learning algorithms for screening ratio was determined to ensure uniform extrusion. 52
CTO&BD-related genes
Two machine learning algorithms were applied to screen 2.12. Fourier-transform infrared spectroscopy, X-ray
the candidate genes for CTO&BD. Least absolute shrinkage diffraction, and scanning electron microscopy
and selection operator (LASSO) is a regression method To analyze the physical characteristics of the G40T60
used for variable selection to improve prediction accuracy bracket, attenuated total reflection infrared spectroscopy
and a regression technique used for variable selection and (ATR-IR; ALPHA, Bruker, Belgium) was used to analyze the
regularization to enhance statistical models’ prediction spectral range of β-TCP in the wavelength region of 400–
accuracy and interpretability. Random forest (RF), 4000 cm with a resolution of 2 cm . The sample’s crystal
-1
-1
which has no restrictions on variable conditions, has the and amorphous phase regions were verified using an X-ray
advantages of better accuracy, sensitivity, and specificity. It diffractometer (SmartLab; Rigaku Corporation, Japan)
could be used to predict continuous variables and provide with an X-ray diffraction (XRD) working power of 9 kW
predictions with no apparent changes. The R packages and using a field emission scanning electron microscope
“glmnet” and “randomForest” were used for performing (FE-SEM; Nova Nano SEM 200; FEI Corporation, USA).
LASSO regression and RF analysis. 45-49 Before the scanning electron microscopy (SEM) analysis,

Volume 10 Issue 2 (2024) 231 doi: 10.36922/ijb.1461