Marilen Benner

GESTATIONAL ANTIBIOTICS AFFECT IMMUNITY 199 7 Placental and intestinal mRNA-expression analysis Total RNA was isolated from maternal intestinal tissues and placenta using the RNeasy mini kit (Qiagen, Germantown, USA) and cDNA was prepared using the iScript cDNA synthesis kit (Bio Rad, Veenendaal, the Netherlands), according to the manufacturer’s instructions. For quantitative real-time PCR, the reaction mixture was prepared by adding specific forward and reverse primers and iQSYBR Green Supermix (Bio-Rad Laboratories, Hercules, CA, USA) to the cDNA samples, and amplifications were performed according to the manufacturer’s instructions using the CFX96 Touch™ Real-Time PCR Detection System (Bio-Rad Laboratories, Hercules, CA, USA). Validated qPCR primers for FOXP3, T-bet, GATA3, ROR- γ T, β -Actin and IL-10 were obtained from SABiosciences (Qiagen, Germantown, USA). mRNA expression levels were calculated relative to the expression of β -actin reference gene with CFX Manager software (version 1.6). Determination of cytokine profiles (in amniotic fluid, and after ex vivo stimulation of splenocytes) Splenocytes collected from pregnant mice were cultured at a concentration of 4.10 6 cells/ ml RPMI 1640 culture medium in 96-well U-bottom culture plates at 37°C in a humidified environment containing 5% CO 2 , in the presence or absence of 10 µg/ml lipopolysaccharide (LPS) (Sigma). Cell culture supernatants were collected after 24 hours and stored at −20°C until further analysis. A ProcartaPlex multiplex protein assay kit (Invitrogen, Thermo Fisher Scientific, Waltham, MA, USA) was used to assess the concentrations of interleukin (IL)-1 β , IL-2, IL-4, IL- 6, IL-10, IL-22, tumor necrosis factor (TNF)- α , and interferon (IFN)- γ in amniotic fluid and cell culture supernatants, according to manufacturer’s instructions. To calculate expression levels of cytokines by maternal splenocytes, cytokine concentrations in supernatants of LPS-stimulated cells were corrected for those measured in supernatants of unstimulated cells. Recursive automatic ensemble feature selection To discover the selection of immunological parameters that allow classification as control or antibiotics-treated cohort, a previously established ensemble feature selection was used (33, 34). This strategy allows for a more general selection of stratifying features than a single classifier, overcoming the bias of each individual algorithm. In brief, 8 classifiers (Bagging, Gradient Boosting, Logistic regression, Passive-Aggressive regression, Random Forest, Ridge Regression, SGD (Stochastic Gradient Descent on linear models), and SVC (Support Vector Machines Classifier with a linear kernel) classifier) generated a list of relative feature importance that is scored for a combined summary of top most relevant features. To ensure generality of the results, each classifier was run 10 times together with a 10-fold cross validation. This was repeated in a stepwise reduction of the 129 initial features by 20%, while determining the accuracy for each classifier.