Reproduction has immediate effects on female mortality, but no discernible lasting physiological impacts: A test of the disposable soma theory

Significance

The dominant evolutionary theory about why animals age and die is the disposable soma theory (DST). This posits that during reproduction, individuals are forced to trade-off allocation of resources between the reproductive event and somatic maintenance. This leads to the prediction that investments in reproduction should lead to greater somatic damage and shortened lifespan. We tested this idea using mice and showed that while reproduction did increase the immediate risks of mortality, there were no long-lasting impacts on survival once breeding ended. These data question the foundation of the DST.

Abstract

The disposable soma theory (DST) posits that organisms age and die because of a direct trade-off in resource allocation between reproduction and somatic maintenance. DST predicts that investments in reproduction accentuate somatic damage which increase senescence and shortens lifespan. Here, we directly tested DST predictions in breeding and nonbreeding female C57BL/6J mice. We measured reproductive outputs, body composition, daily energy expenditure, and oxidative stress at peak lactation and over lifetime. We found that reproduction had an immediate and negative effect on survival due to problems encountered during parturition for some females. However, there was no statistically significant residual effect on survival once breeding had ceased, indicating no trade-off with somatic maintenance. Instead, higher mortality appeared to be a direct consequence of reproduction without long-term physiological consequences. Reproduction did not elevate oxidative stress. Our findings do not provide support for the predictions of the DST.

The disposable soma theory (DST) suggests that aging, senescence, and ultimately death are a consequence of strategic allocation of resources between reproduction and somatic maintenance (1–3). The DST predicts that when animals allocate energy toward reproduction, they devote less to protect their somatic tissue. Hence, they sustain damage that accumulates in relation to reproductive output (4). This leads to a predicted relationship between early life investment and later life mortality. The idea potentially explains why species of animals that reproduce more slowly live longer (5), and why under calorie restriction animals have extended lifespans, because they can allocate all available resources to somatic protection (6).

Comparisons of reproductive rate across species, however, are problematical because of the confounding variable of body size. Larger individuals may be constrained in their reproductive output by their ability to dissipate heat (7). Consequently, the only life history combinations that allow viable populations of large animals to survive involve animals also having long lives (8). Slow reproduction and long lives may become associated therefore for reasons unrelated to damage accumulation. Moreover, during calorie restriction in laboratory animals, the ad libitum fed groups generally are not allowed to breed, and hence, these have the greatest available resource for somatic protection, yet they live the shortest (9).

Direct tests of the DST within species are rare because such tests require long-term monitoring of marked populations that have low rates of emigration, and many assays to assess somatic damage require lethal tissue sampling. However, several studies meeting these conditions in the wild have shown that individuals that allocate more resources to reproduction in early life suffer from greater mortality when they are older (10–12). For example, in Seychelles warblers (Acrocephalus seychellensis), those that started to breed later and hence had lower overall investment in reproduction had a delay in the onset of later life mortality (13). In contrast, no cost of reproduction on survival was found in 18 mammal and 12 bird species from zoos (14).

In laboratory mice, a similar lack of effect of reproduction on longevity has been previously reported (15). There are several issues, however, with this previous study. Reproductive females were paired at age 25 wk, meaning that much of the early reproduction was missed. Since early reproduction might mediate much of the trade-off, the absence of this might explain why no trade-off effects were observed. Moreover, housing density was different between the control and reproductive animals potentially confounding any impacts of reproduction.

Studies in humans found a negative association between fecundity and longevity among females of the British aristocracy born between 1600 and 1900 (16). In contrast, however, in 3,500 French-Canadians, there was a decrease in mortality risk by about 1.4% for each additional child (17). Such studies can be criticized because they are observational and do not take into account individual genetic “quality.” Studies which have attempted to control for individual genetic quality were not supportive of the DST. For example, Chereji et al. (2012) looked at 15,622 twins born between 1901 to 1925 and compared the survival of the twins that had children compared with those that were childless (18). They found contrary to the DST prediction the twins with children survived longer and the genetic correlation of fecundity and survival was effectively zero. However, a recent study comprising 270,000 subjects found that reproduction and lifespan were genetically negatively correlated, i.e., genetic mutations promoting reproduction shortened lifespan (19).

Lactation is a highly energetically demanding period for females with the energy demands of reproduction reaching a maximum and plateauing in the final days of lactation (20–22). Females continuously housed with a male undergo a postpartum estrus (23, 24), meaning that periods of pregnancy overlap with lactation. According to the DST, females must strategically allocate resources, and the increased level of demands will result in a trade-off between somatic protection and reproduction. A popular idea is that oxidative stress may drive this trade-off which occurs because of an imbalance between antioxidant protection and production of reactive oxygen species (ROS). Higher energy demands during reproduction, especially when concurrently pregnant and lactating, would therefore theoretically result in increased oxidative damage and hence aging. The oxidative stress theory of aging (OSTA) assumes that ROS generation is proportional to metabolic rate (25). However, this has largely been disproven, as increased metabolic rates can lead to lower mitochondrial inner membrane potential which in turn may lead to less production of ROS (26, 27). To date, evidence for oxidative stress as a cost of reproduction, and as such, a mediator of life history trade-offs, is inconclusive. However, these contradictory data may be because oxidative stress only affects certain tissues. For example, studies based on blood assays (28–31) tend to support the idea that oxidative stress is increased in reproductive females, compared with measurements of tissues (32–38). Studies comparing both plasma and tissue oxidative damage in the same individuals reported reduced or unchanged damage coupled with increased levels of protection in the liver, yet paradoxically elevated damage and reduced protection in the serum/plasma (30, 31, 39). A meta-analysis reported a clear positive relationship between oxidative damage and reproductive effort across all tissues and markers measured among breeding individuals, but breeders had lower levels of oxidative damage compared to nonbreeders (40). An attempt to explain this apparent contradiction was provided by the oxidative shielding hypothesis which proposes that individuals may preemptively invest to reduce oxidative damage to mitigate the consequences of increased damage to themselves and their offspring (40). Few studies have measured oxidative stress across a wide range of tissues extracted from breeding animals, and none from those undergoing concurrent pregnancy (CP) and lactation.

In the present study, we explored the relationship between fecundity and mortality in C57BL/6J mice. This is a strain inbred for over 200 generations, and hence, individuals are effectively genetically identical, greatly minimizing the individual genetic quality issue. We allocated female mice to breed (paired with male) or not breed (paired with another female) and then compared mortality between these groups. As a direct test of the DST, the relationship between individual differences in fecundity and lifespan during or postbreeding was explored. Whether the increased energy demands of CP and lactation resulted in increased oxidative stress was measured across multiple tissues and in relation to lifespan.

Results

Long-Term Study.

Reproductive output.

In total, 672 litters were produced in the lifespan study (n = 80 mothers). The total number of pups born was 3,165, with 64% (2,032 pups) successfully weaned. Reproductive senescence was reached at ~18 mo of age, with the last litter produced from a mother aged 533 d. Data were first analyzed in the mothers which died during the breeding period (n = 30: Fig. 1 A–D and Table 1). Then, the lifespans of mothers alive at the end of the breeding period were analyzed (n = 34; Fig. 1 E–H and Table 1).

Fig. 1.

Reproductive output of C57BL/6J mice continuously housed with male from 12 wk. Data shown from 30 mothers which died within the breeding period (A–D) or the 34 mothers surviving past the period (18 mo; E–H). (A and E) Frequency of different litter sizes. (B and F) Total numbers of pups born and weaned including numbers of males and females. (C and G) Litter size. (D and H) The duration of delay in pregnancy.

Table 1.

Reproductive output from C57BL/6J females which died during or after reproductive senescence (534 d)

	Within breeding period (n = 30)	Out with breeding period (n = 34)
Number of litters	7.6 ± 3 (2, 13)	9.1 ± 2.3 (3, 12)
Number of pups born	36.9 ± 16.4 (7, 63)	42 ±10.8 (18, 61)
Number of pups weaned	24.7 ± 13.8 (0, 50)	27.3 ± 9.4 (0, 45)
AvPup mass Litter 1—Day 2—Wean (n = X mums/X pups)	1.37 ± 0.26 (n = 24/116) 6.91 ± 1.08 (n = 18/99)	1.29 ± 0.16 (n = 29/143) 7.41 ± 0.95 (n = 26/122)
AvPup mass Litter 10—Day 2—Wean	1.63 ± 0.00 (n = 2/9) 8.28 ± 1.12 (n = 4/9)	1.5 ± 0.09 (n = 6/13) 6.71 ± 2.79 (n = 4/9)
Time between litters	30.2 ± 8.3 (20, 93)	34.1 ± 9 (23.4, 68)
Time between wean and next litter	12.7 ± 8.2 (3, 38)	16.1 ± 9 (5, 50)
Age at last litter	308 ± 90 (130, 458)	378 ± 79.7 (204, 534)
Time between last litter and death	77 ± 67 (1, 296)	339 ± 134.4 (65, 592)

Data are shown as average ± SD (minimum, maximum).

Individual mothers produced between 2 and 13 litters with the highest frequency of litters between 9 and 11 per mum (Fig. 1 A and E). The females which died during the breeding period produced 228 litters in total and mothers that survived to the postbreeding period had a total of 311 litters (Fig. 1 A and E). The 16 mothers culled at set timepoints produced 133 litters. Similar patterns in reproductive output were noted between the survivors and nonsurvivors of the breeding period. There was no statistically significant difference in the total number of pups born (1,378 vs. 1,106, t₄₉ = −1.47, P = 0.15) or weaned (937 vs. 742, t₅₀ = −0.86, P = 0.39) or % concurrencies (T₆₁ = 1.29, P = 0.20). However, in total, there were more litters produced by mothers surviving after the breeding period compared to those which died during (311 vs. 228; t₅₃ = −2.27, P = 0.03). There were four more mothers in the surviving group. Reproductive output decreased with increased parity (Fig. 1 B and F). A high preweaning mortality was observed for litter 1, with 44% of pups lost. The weaning rate improved over litters 2 to 4, to ~74%, after which a decline was recorded, to ~50% success rate for the remaining litters. There was no statistically significant pup sex bias, and the male to female ratio remained even. Average litter size was highest in the first four litters, 6.5 ± 2.6 (ranging from 1 to 14 pups), declining to 2.9 ± 2.6 over the following litters (Fig. 1 C and G). Average pup mass, measured at Day 2 or weaning Day 18, was similar across all litters for females that survived breeding and those that did not (Data for Litters 1 and 10, Table 1). Mothers may delay implantation to avoid peak pregnancy and lactation coinciding. Here, implantation was delayed up to 9 d. Whether pregnancies were concurrent with lactation was highly variable (from 0 to 100%, average 67% of total pregnancies). No statistically significant difference in the duration of implantation was found between females which died during the breeding period or those that survived (Fig. 1 D and H).

Survival and cause of death.

The DST predicts a higher investment in reproduction decreases somatic maintenance and consequently shortens lifespan. We tested this prediction directly in four separate analyses. 1) The initial analysis, carried out on 64 females (with scheduled culls removed), supported the DST (Fig. 2A). Median survival in nonreproductive females was 715 d, vs. 582 d in reproductive females (Chisq = 17.3, df = 1, P < 0.0000, Fig. 2A). 2) We then investigated whether reproduction had a lasting negative effect on female survival (as predicted by the DST) or whether the negative effects of reproduction were transient and only present during the reproductive period. There was a marked divergence between the gradients of the survival curve for reproductive vs. nonreproductive females in the reproductive period, where the survival curve declined earlier for reproductive females (Fig. 2B). 3) Recognizing that several mice (n = 11) were killed due to specific pregnancy or birthing difficulties, we reran the analysis with those removed (Fig. 2C). Nonreproductive mice again survived significantly longer; 715 vs. 609 d, Chisq = 10.9, df = 1, P = 0.001). 4) When the analysis focused on females that survived until after the breeding period independent of the cause, no statistically significant residual effect of reproduction was detected. Median survival was similar between nonreproductive and reproductive mice, 755 and 725 d, (Chisq = 3.3, df = 1, P = 0.07; Fig. 2D). Survival within the breeding period was similar between the 30 reproductive females and the eight nonbreeding females (385 ± 112 d vs. 388 ± 138 d, respectively; t test, t_9.6 = 0.06, P = 0.95).

Fig. 2.

Survival probability of reproductive (REP) or virgin, nonreproductive (NR) female C57BL/6J mice. (A) Smoothed survival approximation of the survival curves from 64 reproductive and 64 nonreproductive females. (B) First derivative (i.e., slope) of the smoothed survival approximation from panel A. (C) Kaplan–Meier survival curve with 11 females killed due to specific pregnancy or birthing difficulties removed (REP n = 53, NR n = 64). (D) Removal of mice killed within the breeding period [last birth was recorded at 18 mo/534 d of age (REP n = 34, NR = 56)]. (E–L) The effect of reproductive output on lifespan. Analysis was carried out in the 30 females that died within the reproductive period (E–H) and the 34 females from panel D surviving breeding (I–L). (E and I) Number of litters, (F and J) total number of pups born, (G and K) number weaned, and (H and L) % of concurrent pregnancies and lactations.

In total, six mice were found dead. All others were killed when agreed humane endpoints for aging animals were reached. From necropsy examination, any pathological conditions were ascertained. As noted above, several reproductive females were killed due to issues directly related to reproduction, e.g., pups stuck in the birth canal during birthing (17%). C57BL/6J mice are susceptible to ulcerative dermatitis (41) which is typically unresponsive to treatments. Despite efforts, the highest cause for killing reproductive females was ulcerative dermatitis (31% vs. nonreproductive 23%). Cancer/neoplasia incidence was, however, higher in nonreproductive vs. reproductive (38% vs. 27%). Of the cancers, liver cancer was predominant, 18% in reproductive females and 42% in nonreproductive. Other deaths were divided into age or non-age related (Table 2).

Table 2.

Postmortem findings from female C57BL/6J mice paired continuously with male mice or paired non-reproductive females

	Direct reproduction issues	UD	Cancer	Age related	Non-age related
Reproductive	11	20	17	11	5
Nonreproductive	N/A	15	24	14	11

The killing of animals was carried out on recommendation of the veterinary surgeon, and necroscopy findings were noted. Scheduled culls were not included (n = 64 for both groups). UD: ulcerative dermatitis.

The association between reproductive output and lifespan.

Analysis of the females culled within the breeding period found positive significant relationships between lifespan and the number of litters (R² = 0.55, F_1,28 = 33.99, P < 0.0001; Fig. 2E), total number of pups born (R² = 0.29, F_1,28 = 11.19, P = 0.002; Fig. 2F), or weaned (R² = 0.13, F_1,28 = 4.13, P = 0.052; Fig. 2G). Although these results oppose the prediction of the DST, we do not believe that these results to be a valid and robust test of the DST. These data were confounded by age, i.e., if a female died when young she was not around to produce litters later in her life, hence the relationships were strongly positive. The negative relationship between lifespan and % of CP and lactations (R² = 0.19, F_1,28 = 6.45, P = 0.017; Fig. 2H) offers support of the DST; however, this was driven by the shorter lifespans of six 100% concurrent mice. Deletion of these females from the analysis removed the negative relationship (R² = 0.05, F_1,22 = 1.27, P = 0.27. There was no residual effect of reproduction on lifespan found in the females surviving the breeding period (number of litters, R² = 0.003, F_1,32 = 0.14, P = 0.75; Fig. 2I, total number of pups born, R² = 0.0005, F_1,32 = 0.17, P = 0.89; Fig. 2J or weaned (R² = 0.002, F_1,32 = 0.08, P = 0.78; Fig. 2K, or % of CP and lactations (R² = 0.04, F_1,32 = 1.38, P = 0.25; Fig. 2L).

Body composition.

Variation in body composition was determined using a general linear model with reproductive group and age as factors (Fig. 3). No significant interaction was found in any of the analyses. Body mass (BM) increased from 12 to 18 mo, with reproductive females heavier than nonreproductive by 3.7 g at 12 m and 5.1 g at 18 m (Group: F_{1, 29} = 12.27, P = 0.001; Age: F_{1, 29} = 7.25, P = 0.012). Increased BM was not reflected by greater fat mass as measured by Dual-energy X-ray absorptiometry (DXA), but by changes in fat-free mass (Group: F = 21.66, P < 0.000; Age: F = 13.52, P = 0.0009). Fat-free mass estimates both bone and lean tissue mass. While bone mineral content was not affected by reproductive status or age, a lowered bone mineral density in the reproductive group was suggestive of osteoporosis (Group: F = 8.66, P = 0.006). In contrast, bone area was higher in reproductive mice (F = 14.75, P = 0.0006). Increased bone area was not related to BM (F = 0.257, P = 0.62).

Fig. 3.

Body composition of female C57BL/6J under continuous reproduction (R) or nonreproductive (NR). Measurements were taken at 12 and 18 mo (n = 8 per group). Points represent individual mice with average weight (g) indicated by a bar ± SD. BM: body mass, FM: fat mass, FFM: fat-free mass, BMD: bone mass density, B Area: Bone area, Sub Cut: subcutaneous fat, Epi: epididymal fat, Retro: retroperitoneal fat, BAT: brown adipose tissue, RepO: reproductive organs, Small Int: small intestines. Points denote individual mice. The bar indicates average per group ± SE.

For mice that were involved in the scheduled culls, the males were removed to ensure that the females were not pregnant or lactating at the time of cull. At 12 m, the average time from previous last weaning was 55 d (ranging from 19 to 107). At 18 mo, the average was 205 d from previous last weaning (ranging from 140 to 282). BM recorded at the time of cull was added as a covariate for the dissected tissue analysis. The heavier BM of reproductive females explained the significantly higher masses of liver, kidneys, epididymal fat, small intestine, colon, carcass, and skin (SI Appendix, Table S1). After adjustment for mass effects, the heart, lungs, and spleen remained significantly larger in the reproductive females (P = 0.03), and a significant impact of age was noted with larger kidneys, pancreas, spleen, brown adipose tissue (BAT), and stomach at 18 mo (P = 0.03).

Daily energy expenditure (DEE).

DEE was increased by both reproduction and age (Group: F_1,25 = 16.43, P = 0.0004, Age: F_1,25 = 37.74, P < 0.0001). Differences were not explained by BM or fat-free mass.

Coordination and neurodegenerative tests.

Aging is the primary risk factor for neurodegenerative diseases. At 23 mo, all surviving mice underwent four tests which hallmark coordination impairment and neurodegeneration. Previous reproduction did not have a detrimental impact on hindlimb clasping (Z = 1.37, P = 0.17), ledge test (Z = 1.19, P = 0.23), gait (Z = 0.82, P = 0.41), or kyphosis (Z = 1.12, P = 0.26) (SI Appendix, Fig. S1).

Oxidative stress.

The DST predicts higher reproductive effort should decrease somatic maintenance resulting in increased tissue damage. One marker of tissue damage is oxidative stress. Of the two damage assays (protein and DNA) and three antioxidants [catalase, superoxide dismutase (SOD), and glutathione peroxidase (GPx)] measured in the liver, there were no statistically significant relationships (P > 0.05: for detailed P values, see SI Appendix) between these parameters and the total number of litters produced, pups born, weaned, or the % time concurrent (Fig. 4, for statistics, see SI Appendix, Table S2). No indication of oxidative stress was found between the non- or reproductive groups at ages 12, 18, or 24 mo (Group × Age interactions were nonsignificant for all markers; Catalase: t = 0.792, P = 0.432, SOD: t = 0.646, P = 0.521, Gpx: t = −0.695, P = 0.491, protein carbonyls: t = −1.16, P = 0.252, DNA damage: t = −1.648, P = 0.106, SI Appendix, Fig. S2).

Fig. 4.

The effect of reproductive output on oxidative stress measured in the liver of female C57BL/6J mice housed continuously with a male. Measurements were made when females were 12, 18, or 24 mo old. For each female the number of litters produced, total number of pups born, total number of pups weaned, and % of concurrent pregnancies and lactations were related to their levels of Catalase (U/min/mg protein, A–D), SOD (U/mg protein, E–H) and GPx (nmoles NADPH/min/mg protein, I–L), protein damage as measured by protein carbonyls (nmol/mg, M–P) and DNA damage (8-hydroxy-2'-deoxyguanosine ng/mL, Q–T).

Short-Term Study.

Physiological parameters.

During the CP period, there was initially, a small decrease in BM in the reproductive females, followed by a continual increase from day 8 (Fig. 5A). The reproductive females gained significantly more weight daily than the nonreproductive females from day CP11 and remained so throughout the CP (interaction: F_17,204 = 62.98, P < 0.001). Reproductive females ate significantly more than nonreproductive females throughout the concurrent period (interaction: F_14,168 = 10.23, P < 0.001; Fig. 5B). Reproductive females’ daily food intake peaked on day CL14 (9.23 ± 0.72 g) which was significantly higher than days CL4 to CL8 (post hoc P < 0.001 for all days).

Fig. 5.

Physiological parameters of reproductive concurrent (REP; n = 7) and nonreproductive (NR; n = 7) C57BL/6J female mice through a CP and lactation period. (A) Change in BM (g; CP: aligned by pregnancy date), (B) food intake (g; CL: aligned by lactation date), (C) light phase body temperature (°C), (D) dark phase body temperature (°C), (E) light phase activity (counts – summed over 12 h), (F) dark phase activity (counts), (G) number of pups born in litter 1 and 2, (H) number of pups weaned, (I) litter weaning weight (g), and (J) DEE (kJ/d). All data are presented as mean ± SE.

Body temperature (T_b).

T_b during the light phase was significantly affected by reproductive status (F_1,11 = 274.75, P < 0.001) and day (F_17,204 = 2.72, P < 0.001). Light phase T_b was higher (37.5 ± 0.02 °C) in reproductive females compared to nonreproductive females (36.3 ± 0.02 °C) over the concurrent period. There was no significant interaction between reproductive status and day (Fig. 5C).

The pattern of dark phase T_b between reproductive females and nonreproductive females differed over the concurrent period (interaction: F_17,187 = 2.56, P = 0.001; Fig. 5D). For the first 3 d T_b was similar between the groups. The T_b of reproductive females steadily increased to 38.11 ± 0.03 °C on CL11 (which aligns to days 4 to 9 of pregnancy depending on implantation). T_b then decreased continually until the end of the period. From CL4 to CL13, reproductive females’ T_b was significantly higher than nonreproductive T_b.

Physical activity (PA).

During the light phase, PA between the females only differed on the day of birth (reproductive: 7,530 ± 405 counts vs. nonreproductive 4,727 ± 199 counts; interaction: F_19,322 = 9.18, P < 0.001; Fig. 5E).

Dark phase PA, however, differed between reproductive females and nonreproductive females (Fig. 5F; interaction: F_17,187 = 2.67, P = 0.001). Reproductive females’ PA levels initially declined across the concurrent period before plateauing from CL10 (Fig. 5F). From CL6 to CL15, reproductive females were significantly less active compared to nonreproductive females.

Reproductive output.

The number of pups born in litter 1 (7.4 ± 0.69) was significantly higher than in litter 2 (5.1 ± 0.67) from concurrently pregnant and lactating mothers. However, by day 16 when pups were weaned this effect was no longer observed, and concurrency had no statistically significant effect on the litter size or weights at weaning (Fig. 5 G–J).

DEE.

DEE was measured on day 14 of lactation two following the CP and lactation period. BM on the day of the doubly labeled water (DLW) procedure was a significant covariate in the model (F_1,9 = 113.3, P < 0.001). After accounting for BM, DEE was significantly higher in reproductive concurrent females (82.7 ± 3.2 kJ/d) compared to nonreproductive (49.6 ± 1.1 kJ/d) females (F_1,9 = 6.7, P = 0.029; Fig. 5J).

Body composition.

Organs of the digestive tract are remodeled during reproduction to cope with the increased energy demands of lactation. After correcting for BM, only the liver and small intestine differed in weight between the reproductive concurrent and nonreproductive females. The liver weight of reproductive concurrent females was nearly 2× heavier than that of nonreproductive females (reproductive status: F_1,11 = 23.71, P < 0.001; SI Appendix, Fig. S3). The small intestine of reproductive concurrent females was 1.4× heavier than nonreproductive females (reproductive status: F_1,11 = 9.67, P = 0.01, SI Appendix, Fig. S3). The kidneys of reproductive concurrent females were also significantly greater than the nonreproductive females (reproductive status: F_1,10 = 6.88, P = 0.025; SI Appendix, Fig. S3). Reproductive concurrent females had significantly less retroperitoneal, epididymal, and subcutaneous fat compared to nonreproductive females (F_1,10 = 19.29, 15.40, 29.47, P < 0.02, respectively, which was independent of BM; SI Appendix, Fig. S3).

Oxidative stress.

Of the 51 measures (11 tissues × 5 assays) of oxidative stress, few differences were found between the reproductive concurrent and nonreproductive females when adjusted for multiple testing. The activity of GPx was significantly lower in the small intestine and epididymal fat of reproductive concurrent females (interaction: F_10,118 = 5.27, P < 0.0001, post hoc P = 0.047 and P < 0.01, respectively; Fig. 6). Both SOD and catalase activity were similar in all tissues (Fig. 6). Reproductive concurrent females had lower levels of protein carbonyls, a marker of protein damage, in BAT compared to the nonreproductive females (interaction: F_11,128 = 2.25, P = 0.015; Fig. 6). No statistically significant differences in protein carbonyl levels were found in other tissues. Similarly, levels of 8-hydroxy-2'-deoxyguanosine, a marker of DNA damage, were not different between reproductive concurrent and nonreproductive females in any tissue (interaction between reproductive status and tissue: F_6,70 = 1.81, P = 0.11; Fig. 6).

Fig. 6.

Markers of oxidative stress at peak lactation in reproductive concurrent (REP; n = 7) and nonreproductive (NR; n = 7) C57BL/6J female mice following a CP and lactation period and subsequent lactation. SOD (U/mg protein), catalase (U/min/mg protein), GPx (nmoles NADPH/min/mg protein), protein carbonyls (nmol/mg), and DNA damage (8-hydroxy-2'-deoxyguanosine ng/mL). Data are presented as mean ± SE.

Discussion

The DST predicts that organisms should trade-off investment of limited resources in reproduction and somatic maintenance, which could lead to long-term survival costs (1, 2). This is because reproduction is energetically expensive, incurring numerous direct and indirect physiological costs (42). Support for the predictions of the DST remains mixed (9–11, 18, 43). Here, we showed that reproduction is indeed costly and there is an elevated risk of mortality, but for individuals that survive the breeding period, there is no lasting impact of reproduction on survival. This contradicts the predictions from the DST which proposes that the accumulated somatic damage during the reproductive period would lead to higher mortality after the period of reproduction ends.

Our findings provide empirical support on the lack of an effect of reproduction on female survival previously reported in gray mouse lemurs (Microcebus murinus) in captivity (44), and 18 species of mammals and 12 species of birds held in zoos (14). One possible explanation for these findings is that the exceptionally favorable conditions of captivity and laboratory conditions could mask negative effects of the trade-off between reproduction and survival. For instance, captive and/or laboratory individuals are often given ad libitum access to food, meaning that individuals could in theory increase their food intake to meet the energetic demands of reproduction while maintaining somatic integrity. The conditions we studied are therefore compatible with testing the DST. Clearly, however, in an ecological setting, there may be other nonphysiological factors at play, such as increased predation risk, that could mediate a trade-off between survival and reproduction.

The OSTA postulates that aging results from the accumulation of cellular damage (25). Joint predictions from OSTA and DST predict that a lack of somatic maintenance during reproduction could result in accumulation of oxidative damage and consequently, aging (45). We evaluated markers of oxidative stress in 11 tissues, from females killed at peak lactation following a CP and lactation. Of the 51 oxidative stress measurements, only three measurements differed between the reproductive concurrent and nonreproductive females. A reduction in GPx activity in the small intestine and the epididymal fat of reproductive concurrent females did not correspond to elevation in protein or DNA oxidative damage. Furthermore, protein carbonyl levels were decreased in the BAT in reproductive concurrent females, without a concomitant increase in antioxidant activity that may have protected against this damage. This may be explained by reduced nonshivering thermogenesis during lactation, for which BAT is a major contributor. Thus, lowered ROS generation may be anticipated with reduced cellular activity in BAT (as reviewed refs. 42, 46, and 47). These results provide evidence that reproduction does not incur oxidative stress due to downregulation of protection as predicted by the OSTA and DST.

In life-long reproductive females, there was no statistically significant relationship between the lifetime reproductive output and any measures of oxidative stress in the liver. Thirty-four females survived beyond the cessation of the breeding period and therefore may have had a “recovery period” prior to death. Organ weights of the lifespan females were similar to those of the nonreproductive females, both at 12 and 18 mo, when organs may have “recovered” from the remodeling of reproduction. At peak lactation, the digestive tract organs and adipose tissues were larger in the reproductive concurrent females. Therefore, such a recovery period may account for lack of oxidative stress and any oxidative damage may have been repaired or removed. However, evidence of oxidative stress was not apparent when females were at maximum energy expenditure at peak lactation so we would not expect a recovery period to account for the lack of oxidative stress in the liver in the lifespan study. Overall, results from the short and lifetime studies are complementary and demonstrate the recovery period did not explain the lack of relationship between oxidative stress and reproductive output.

Many disparities exist between controlled laboratory studies and those executed in the wild, with different tissues and oxidative stress biomarkers analyzed (48). Laboratory-based studies often measure biomarkers in tissues and generally conclude that oxidative damage levels are either unchanged or decreased in reproductive individuals compared to nonreproductive females or individuals with higher reproductive investment (32–38). Conversely, studies utilizing blood sampling tend to report serum/plasma oxidative damage to be elevated in reproductive females or with increasing litter size (28, 30, 31). However, no relationship between reproductive investment and plasma oxidative damage was noted in Soay sheep (49). It has thus become apparent that oxidative stress may be tissue and marker dependent (40, 48). To account for the disparity in tissues, here we quantified oxidative stress in the widest range of tissues to date at maximum energetic requirement in the same females and overwhelmingly observed a lack of oxidative stress during peak reproduction. We found no evidence to support this as a potential mechanism of the DST.

Limitations

Both non- and reproductive females were fed AIN-93G until 18 m old when switched to AIN-93 M. We recognize that high protein diets may negatively impact lifespan but chose to feed all animals the same diet at the same age. The 5% higher protein content of AIN-93G (19% vs. 14% in AIN-93 M) fed for the whole life would be predicted to shorten lifespan by about 4%, hence feeding this diet for about half of the lifespan, would only have a minor (c. 2%) effect on lifespan of the nonreproductive mice. Our study aimed to test the predictions of the DST. To exclude reproduction, we used paired nonreproductive, virgin females as controls. Female–female pairing, as opposed to female–male, may have impacted on mortality risk.

Our study shows that contrary to the DST, there were no residual costs to lifespan from reproduction. This is evident in the postbreeding period, where mortality was similar between breeders and nonbreeders. In the wild, extrinsic mortality could curtail individuals’ lifespan prior to the end of breeding, and selection does not act on individuals that reach postbreeding age (“selection shadow”). Nevertheless, our results show that DST cannot explain mortality costs associated with long-lasting effects of reproduction in the selection shadow.

The study only considered oxidative stress as a potential marker of “somatic damage” and did not account for alternative mechanisms of tissue damage. In addition, once allocated to “breeding,” the females chose themselves how much to invest and those allocating a lower investment in reproduction may survive longer.

Conclusions

In conclusion, our study shows that reproduction has an immediate effect on female mortality but has no apparent lasting physiological impact. Differences in survival probability between reproductive and nonreproductive females were not observed postbreeding. Furthermore, we show here that oxidative stress is not a cost at either peak lactation or in the lifespan, and therefore, oxidative stress is not a physiological mechanism mediating any trade-off between reproduction and survival. These data do not support the disposable soma hypothesis.

Methods

Animal Procedures.

All procedures, carried out under Home Office License (PPL600/4362), were reviewed and accepted by the University of Aberdeen Animal Welfare and Ethical Review Body. Studies were compliant with the Animals (Scientific Procedures) Act 1986 and in accordance with the ARRIVE Guidelines (50). Mice were housed in transparent polycarbonate cages and provided with sawdust, paper wool bedding with paper/plastic tubes, and houses for enrichment. Photoperiod was controlled on a 12 h:12 h light:dark with lights on 04:30 and off 16:30 with 20-min dimming at either end. Temperature was maintained at 21 ± 2 °C, and mice were provided with ad libitum water and food (AIN-93G or AIN93M: Research Diets, USA).

Lifespan Study.

One hundred and 60 female C57BL/6J mice and 80 males were purchased at 8 wk of age (Charles River, UK). C57BL/6J mice were chosen to minimize genetic variation providing a direct test of the DST independent of any genotype and genotype-by-environment interactions. All mice were initially provided AIN-93G (19.3 protein: 64 carbohydrate: 16.7 fat, %kcal; ~3.8 kcal/g) specifically formulated to assist pregnancy, lactation, and growth. The increased fat content in the AIN-93G diet was mainly due to higher soybean oil which contains essential fatty acids required in reproduction (51). Following the end of the reproductive period, all mice were transferred to AIN-93 M (14.1 protein: 75.9 carbohydrate: 10 fat, %kcal; ~3.6 kcal/g) formulated to support the overall maintenance of the mature mouse.

At 12 wk, mice were randomly allocated to a reproductive group (n = 80 reproductive females) and paired with a male or split into 40 nonreproductive pairs (n = 80 nonreproductive females). Both paired groups were continuously housed together; should one of the pair be killed, a new partner was introduced from the same cohort. Female mice can undergo a postpartum estrus, therefore by housing continuously with a male, periods of pregnancy overlap with the periods of lactation; known as concurrency. In this way, a female will lactate almost continuously and hence operate almost at maximum levels of energy demands, potentially trading off somatic protection and reproduction. C57BL/6J mice have a gestation period of 19 d and lactation of 21 d, whether females were concurrent was calculated by counting days between litters. The first litter was not included in this calculation.

Reproductive output was continuously monitored, with all adult mice and litters checked twice daily throughout the study. Sex of offspring, litter size, and mass were recorded on days 2, 10, and 18 postbirth. On lactation day 18, the pups were weaned and removed from their mother. Mice were monitored until a humane endpoint was reached (see below), or they were selected for scheduled measures and culls at 12 and 18 mo of age (during and after the reproductive period, respectively). Eight pairs of mice from the non- or reproductive groups (n = 32) were randomly selected at each timepoint. The scheduled measures carried out at 12 and 18 mo are described below were body composition using Dual-energy X-ray absorptiometry (DXA), energy expenditure measured using DLW technique. Males were separated from females 4 wk prior to prevent pregnancy.

Ulcerative dermatitis treatment.

A number of mice were affected by ulcerative dermatitis, a condition common to C57BL/6J mice C57BL/6J (41). Ulcerative dermatitis is a progressive disease which may lead to ulcerations. Treatments included use of lanacane cream, hibiscrub, and saline baths, which averted the need for euthanasia in many cases. A scoring system, based on the character and size of the lesion, scratching occurrence, and treatment plan, was agreed with the Veterinary Surgeons. Lesions were scored from 1 to 4, where 4 = humane endpoint, i.e., regular to severe scratching; with larger area of skin broken (≥1 cm²), skin is ulcerated, thicker than normal and/or wet or scabby or bloody.

Killing of animals/pathology.

Mice were checked twice daily, and any signs of pain or stress or detraction from normal behavior were noted. Humane endpoints for aging animals were agreed with the University's named veterinary surgeon at the onset of the study (52, 53). Body condition was scored between 1 to 5, with three being the optimal condition, one emaciated, and five obese (53). A humane endpoint was reached when the body condition score was <2. While the possible cause of death was identified at necroscopy, samples were not further analyzed for malignancies. Disease states were categorized as cancer/ neoplasia, UD, age related (as age increases the risk also increases), or non-age related (risk does not increase with age) (53). Lifespan was recorded as the age at which a mouse was killed.

Short-Term study.

Forty-two virgin C57BL/6J mice (30 females and 12 males; Charles River, Ormiston, UK) were purchased at 7 wk of age. Water and food (AIN3G diet, Research Diets, NJ, USA) were available ad libitum. Body temperature (T_b) and PA were recorded using the VitalView™ telemetry and data acquisition system (MiniMitter, OR, USA). Twenty virgin females underwent surgery at 9 wk old for implantation of a VitalView™ transmitter in the abdominal cavity. Each transmitter signals minute-by-minute T_b and PA via two unique frequencies to a data-receiving platform situated under the cage. To ensure no interference between cages, receiving platforms were placed 25 cm apart and only one female with a transmitter was housed in each cage.

At 14 wk of age, females were randomly allocated to a reproductive group and housed with a male (n = 7), or a nonreproductive group (n = 14) which comprised seven pairs of females, one with and one without a transmitter. BM, food intake, T_b, and PA were recorded over an 8-d baseline period, after which reproductive females were paired with a male. BM and food intake were recorded daily throughout the study except on the day of parturition. Reproductive females underwent two successful reproductive bouts (one while concurrently pregnant and lactating). The male remained housed with the female until 3 d following the birth of litter 1 to allow for mating during postpartum estrus. The date of pregnancy was back-calculated 19 d from parturition (1st day of pregnancy denoted as P-19 and parturition day 0). Litter size and mass were recorded from day 1 and pups from litter 1 were weaned on day 18.

Reproductive females were culled at the peak of lactation two (day 16) with a terminal overdose of CO₂. A control nonreproductive female, weight-matched at baseline, was killed at the same timepoint.

Scheduled Measures.

DEE.

DEE (kJ/d) was measured using the DLW technique previously validated by comparison to indirect calorimetry in small mammals (54). For the long-term study, DLW was undertaken at 12 and 18 mo. For the short-term study, DLW was undertaken on day 14 of lactation 2. Briefly, the animals were weighed and a 100 μL blood sample was taken by tail tipping to estimate the background isotope enrichments of ²H and ¹⁸O (method D). Further blood samples were taken after 1 h of isotope equilibration to estimate initial isotope enrichments (55). A final blood sample was taken 24 h later. Analysis of the isotopic enrichment of blood was performed blind, using a Liquid Isotope Water Analyzer (Los Gatos Research, USA). For full details of the method, please refer ref. 56.

Body composition.

Fat mass and fat-free mass were measured at 12 and 18 mo using Dual-energy X-ray absorptiometry (DXA), GE PIXImus2 Series Densitometer (GE Medical Systems Ultrasound and BMD, UK). Mice were anesthetized with isoflurane for the duration of the scan. Data were corrected using an equation specific for the DXA machine (57).

Coordination and neurodegenerative tests.

Tests were performed in triplicate in weeks 88, 89, and 90 to establish the impact of cerebellar ataxia, which manifests with age in mice. A scoring system of performance, where 0 is least or not affected and three severely affected, in four tests was used to demonstrate the severity of disease (58). All female mice were subjected to testing. Ledge test The mice were placed on a ledge around 15 cm high and observed as they walked along it. A score of 0 was given to mice that showed coordination and did not lose their footing. A score of 1 was given when footing was lost occasionally but overall coordination was present. A score of 2 showed lack of coordination and possible falling, and a three was scored when the animal refused to move or fell off completely. Hind limb clasping Mice were held by the base of the tail just off the bench for around 30 s, and the hind limbs were observed. A score of 0 showed that both legs continuously splayed outward. A score of 1 showed one leg retracting toward the body for less than half of the time, 2 both legs retracting for less than half of the time, and 3 both legs retracted continuously. Gait The mice were placed on the bench and the position of their underside and ability to walk was observed. A score of 0 was given if the underside was elevated from the bench, and no limping or tremor was seen. A score of 1 showed minor limping or tremor, 2 was given if the underside dragged along the bench with movement, or if there was substantial limping or tremor, and 3 was given if the underside and back legs dragged with movement and extreme limping or tremor were observed. Kyphosis Curvature or hunching of the spine is often observed in aging mice (59). A score of 0 was given if the mice showed no kyphosis. A score of 1 showed slight kyphosis but the ability to straighten the spine, 2 showed substantial kyphosis, but the movement was not impaired, and 3 showed extreme kyphosis and difficulty moving. Scores for the triplicate measures were summed.

Scheduled culls.

Mice were culled with a terminal overdose of CO₂. Tissues were rapidly dissected, weighed, and snap-frozen in liquid nitrogen and stored at −80 °C until required for assays. Tissues included six adipose tissues (SubCut: subcutaneous, Epi: epididymal, Retro-retroperitoneal, omental, mesenteric, and BAT), vital organs (liver, kidneys, heart, lungs, pancreas, and spleen), digestive organs (small intestine, stomach, cecum, and colon), and structural (carcass, skin, tail, and skeletal muscle) and reproductive organs. The liver was split into six to avoid freeze–thaw cycles. Brains were frozen on dry ice over isopentane, to preserve the brain structure allowing the hypothalamus to be dissected. Blood samples were collected via cardiac puncture in ethylenediaminetetraacetic acid-treated tubes, kept on ice, and later centrifuged for plasma collection at 4,000G for 12 min at 4 °C.

Oxidative stress (antioxidants, DNA damage, and protein carbonyls).

Markers of oxidative stress (antioxidant activity, protein, and DNA damage) were measured in the liver of females [12, 18, and 24 mo (n = 8)] in the lifetime study and in the liver, kidney, heart, lung, stomach, small intestine, colon, BAT, hypothalamus, skeletal tissue, and plasma of females in the short-term study. Tissue (~50 mg) was homogenized in 50 mM potassium phosphate buffer (pH 7.4) to a final concentration of 50 mg/mL. Homogenized samples were centrifuged at 4,000G for 20 min at 4 °C, supernatant removed, and recentrifuged. The resulting supernatant was used for analysis.

Catalase activity was measured on the day of homogenization due to its lability. Measurements were based on a spectrophotometric method (60, 61). Measurement of SOD activity was also based on a spectrophotometric methodology which measured the inhibition of auto-oxidation of pyrogallol by SOD (62). GPx activity was measured using modified protocols which monitored the oxidation of β-Nicotinamide adenine dinucleotide 2’-phosphate reduced tetrasodium salt hydrate (NADPH) to Nicotinamide adenine dinucleotide phosphate (63, 64). The total protein content of samples was measured using the Bradford method (65), modified for different tissue types. Oxidative phosphorylation of proteins was measured using BioCell Protein Carbonyl Enzyme Immuno-assays according to the manufacturer’s instructions (BioCell Corporation Ltd., New Zealand; intra–inter assay variation: 5%). Oxidative damage of DNA was measured using the JAICA Highly Sensitive 8-hydroxy-2'-deoxyguanosine ELISA kit according to the manufacturer’s instructions (Japan Institute for the Control of Aging, Japan; intra–inter assay variation: 1 to 8.7%). DNA was extracted and digested using sodium iodide, an iron chelator, to control for artificial oxidation of samples during the extraction process (66, 67).

Statistical Analysis.

All statistical tests were conducted in RStudio 2022.02.2 + 485 using R Statistical Software (v4.1.1;43) and Minitab (version 18). The ggplot2 R package was used for data analysis and visualization (68). Reproductive output measures included the total number of litters produced per mother, total number of pups born and weaned, ratio male: female, % of CP/lactations. Whether these measures affected the lifespan of reproductive females was determined using least-square linear regressions (α = 0.05). Survival analysis Survival analyses were conducted in R using the “survival” package (69) and plotted using the “ggplot2” package. The effect of reproduction on survival was analyzed using the Kaplan–Meier method and Cox proportional hazards regression model. Survival was first analyzed with the scheduled cull animals removed (n = 64 each group). Noting that 11 females were killed due to specific pregnancy or birthing difficulties, these were also removed (n = 53 reproductive and 64 nonreproductive). Finally, survival after the breeding period was calculated. The last birth was recorded at 18 mo of age (533 d), and final analysis was carried out on nonreproductive (n = 56) and reproductive (n = 34). We approximated the Kaplan–Meier survival curve using a thin-plate spline generalized additive model from the “mgcv” package, which allowed us to construct differentiable smoothed approximation of the survival curves for reproductive and nonreproductive females. We used the derivative function of the “gratia” package to obtain the first derivative of the smoothed approximating function (i.e., their slopes). Negative slopes are found when the proportion of living individuals decreases and is therefore a proxy for the instantaneous mortality rate of the treatment groups. Statistical significance was assessed by 95% CI provided by the function. Slopes whose CI did not overlap zero were considered statistically significant.

Whether the number of litters, total number of pups born or weaned, or the number of concurrent pregnancies/lactations affects the lifespan of reproductive females was determined using least-square linear regressions (α = 0.05). Body composition, tissue weights, and DLW were all analyzed using two-way ANOVA with Group and Age. The weights of two livers and a spleen were removed from analysis as outliers due to tumors. Poisson regression was used to analyze the summed scores from the four neurodegenerative tests. A P value < 0.05 was considered statistically significant.

Short-term study.

Pregnancy had a greater effect on BM than lactation, hence for the concurrent period, data were aligned by pregnancy date for BM change. As lactation has a greater effect on food intake, T_b, and PA than pregnancy, these measures were aligned by the birth of litter 1 (70). T_b was averaged, and PA counts were summed over the 12-h light and dark periods. BM change (daily change compared to the start of pregnancy), food intake, T_b, and PA were analyzed over the concurrent period using a mixed effects linear model with post hoc comparisons corrected with the Bonferroni method. Reproductive status, day, and the interaction were explanatory variables, and ID was used as a random factor. Model selection was completed using backward elimination to remove nonsignificant interactions (P > 0.05). Model residuals were checked for appropriate assumptions and raw data loge transformed when necessary. DEE was analyzed by two-way ANOVA with reproductive status and the BM at the time of the DLW procedure as fixed effects. Tissue weights were also analyzed by two-way ANOVA with reproductive status as a fixed effect and the BM at the time of cull as a covariate. The number of pups born and litter weight at weaning were analyzed by the t test. The number of pups weaned was analyzed by the Mann–Whitney U test.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

Data for the project are available (https://osf.io/ksrwb/?view_only=e55b07b3ea81495589b974ca8282e8c2) (71).