Data were analyzed using the General Linear Models (GLM) least squares analysis of variance procedures of Statistical Analysis Systems (SAS). All parameters were averaged within gilt and analyzed in a statistical model that included treatment group, day of gestation, and their interaction. Data for fetal weight, fetal liver weight, placental weight, fetal total RBC count, fetal EPO levels, BE, OE, and all 2D-PAGE data were log-transformed, and data from fetal hemoglobin levels were square root-transformed before analysis to alleviate heterogeneity of variance.
When significant effects of day or treatment were detected, data were further analyzed by orthogonal or nonorthogonal contrasts where appropriate. For effects of day, contrasts were 1) Day 24 vs. Day 30, 2) Day 30 vs. Day 40, and 3) Day 24 vs. Day 40. For effects of treatment, contrasts were 1) INT vs. UHO, 2) WC (INT and UHO) vs. MS (if contrast 1 was not significant), and 3) INT vs. MS (if contrast 1 was significant). When an interaction between day and treatment was detected, the treatment contrasts as above were conducted within each day of gestation. For fetal survival percentage, the day-by-treatment interaction was partitioned into four one-degree-of-freedom contrasts using orthogonal comparisons. Analysis of covariance was conducted on fetal, fetal liver, and placental weight data, with fetal weight and placental weight used separately as co-variates for the other two parameters and together for fetal liver weight. Regression analysis was conducted on data from Day 40 fetuses to determine relationships between fetal weight and either total RBC counts, hematocrit, hemoglobin, or EPO.
FIG. 1. Representative fluorograph from 2D-PAGE analysis of cultured Day 30 fetal swine liver. Spots punched for scintillation analysis are indicated by labels placed directly above the spot punched from each gel. Spots labeled 4 and 5 are proteins 4 and 5, respectively; B, background spot (no Coomassie staining).