TNFα is responsible for the canonical offspring number-size trade-off

There is a canonical life-history trade-off between quantity and quality of offspring, but molecular determinants for this are unknown. Here, we show that knockout of tumor necrosis factor (TNF-KO) in mice switched a relation between the number and size of developing embryos from expectedly negative to unexpectedly positive. Depletion of TNFα imbalanced humoral and trophic maintenance of embryo growth during gestation with respect to the litter size. The levels of embryotrophic GM-CSF cytokine and placental efficiency attained positive correlations with the number and size of embryos in TNF-KO females. Thus, TNFα oversees mother’s resource allocations to balance embryo growth with the number of offspring. Consequently, this suggests an intricate link between the number-size trade-off and immunity given a pivotal role of TNFα in immune homeostasis.


Fig. S1. Zygotic TNFα is required for blastocyst implantation.
A) TNFα functions in the development of immune system. Spleen weights were reduced in both TNF-KO males and females, while thymus weights were lower in TNF-KO males, but not in females. Comparisons were based on Student's t-test.

B)
Pre-implantation losses were estimated from the number of ovulated eggs (corpora lutea -CL), and the numbers of live (LE) and resorbed (RE) embryos as 1-(LE+RE)/CL; post-implantationas RE/(LE+RE), and totalas 1-LE/CL. Embryo losses were compared for TNF +/+ embryos with embryos homo-or heterozygous for TNF-KO obtained from the crosses of TNF -/or TNF +/+ males with TNF-KO females, or TNF -/males with TNF +/+ females. One-tailed p values were calculated with Fisher's exact test on contingency tables. Note that pre-implantation losses were also significantly lower for TNF +/+ embryos with two-tailed Fisher's exact test (p = 0.024). A-B) Correlations of embryo weights with (A) mother weight to litter size ratio and (B) the number of offspring for TNF +/+ (left panel) and TNF -/-(right panel) females. Regression lines are shown for matings of control females with TNF +/+ (purple) and TNF-KO (green) males, and of TNF-KO females with control (blue) and TNF-KO (red) males. Correlation coefficients (r) for each group are also indicated in the same colour. Correlation coefficients (r), significance (p) and the number of studied embryos (n) for TNF +/+ (left) and TNF -/-(right) females are shown in the top left corners and the corresponding regression lines are plotted in black. Note, that in (A) 5% outliers with respect to mother weight to litter size ratio were removed using minimum covariance determinant algorithm from the MASS package of R. See also Table S2 for the statistical details.   Table S3 for statistical details. Significant correlations (p < 0.05) are outlined. Negative correlations are shown in blue, positivein red.  A) Boxplots of log10-scaled concentrations of steroid hormones (progesterone, corticosterone, testosterone) in blood plasma (top panel) and amniotic fluid (bottom panel). One-way ANOVA followed by LSD revealed a significant increase in serum corticosterone levels in TNF-KO females mated with TNF-KO males. Serum testosterone levels were marginally affected (p < 0.1), while the levels of other hormones remained unchanged. Letters indicate statistically significant differences (p < 0.05) for plasma corticosterone and testosterone.

Fig. S5. Function of TNFα in male's reproduction
Serum testosterone levels, weights of androgen-dependent tissues (testis, epididymis, seminal vesicles), sperm differentiation (% spermatogonia, spermatocytes, spermatids), the number (per mg of epididymis) of spermatozoa and percentage of motile spermatozoa, and their velocities and shape (sperm head elongation, size) characteristics were assessed for the control and TNF-KO males housed with females. All parameters except for testosterone were compared with Student's t-test, * -p < 0.05, ** -p < 0.01. Given a significant departure of log10-scaled testosterone levels from the normal distribution for TNF-KO males due to a substantial increase in variations, testosterone concentrations were compared with non-parametric Mann-Whitney U-test. The number of ovulated eggs was estimated from the number of corpora lutea (CL) and one-way ANOVA showed a significant effect of TNF-KO on ovulation. Main effects and interaction of the parental genotypes (TNF +/+ or TNF -/-) on the number of implanted embryos (sum of live and resorbed embryos), live embryos, reproductive output (litter size, weight) and embryo growth were assessed by two-way ANOVA (F-statistics and p values are shown).
Reproductive output was calculated only for females carrying live embryos at day 16.5 of gestation.  Variations in embryo, placental weights (mg) and embryo to placenta weight ratios (E:P) with respect to parental genotype and linear covariates (Z) were modelled as: parental genotypes (TNF +/+ or TNF -/-), Z -mother weight to litter size ratio (Mfemale:L) or the litter size (L). ANCOVA revealed 1) significant effects of parental genotype on all response variables and 2) interaction effects of ♀ x Z on embryo weight and E:P ratio. The latter is due to a switch in correlation signs or gain in correlation between embryo weight or E:P ratio and the covariates in TNF-KO females as compared to TNF +/+ females (Figure 1, S1). Table S3. Maternal and paternal effects of TNF-KO along with covariates (litter size, embryo weight, placental weight or E:P ratio) on the log10 levels (g/l) of GM-CSF in amniotic fluid.
Parameter: Covariate (Z): Factor: Variations in amniotic log10 GM-CSF levels with respect to parental genotype and linear covariates (Z) were modelled as:    Table S5. Maternal and paternal effects of TNF-KO along with covariates on the log10 levels (g/l) of progesterone, corticosterone and testosterone in amniotic fluid.