The metabolic response to an immune challenge in a viviparous snake, Sistrurus miliarius

A key tenet of life history theory is the concept that physiological functions compete for available resources, resulting in tradeoffs that may have fitness consequences (Dunham et al., 1989; Zera and Harshman, 2001). Predicting the outcome of physiological tradeoffs requires a clear delineation of the costs of the competing processes involved, the currencies by which costs are paid, and the factors that impact allocation priorities (Zera and Harshman, 2001). Costs associated with immune performance are becoming increasingly appreciated as drivers of energetic tradeoffs (Demas et al., 2012; Downs et al., 2014; Lochmiller and Deerenberg, 2000), and recent studies implicate energetic costs of infection as a proximate mechanism contributing to the impacts of emergent pathogens on individuals and populations (e.g. Agugliaro et al., 2020; Mayack and Naug, 2009; Reeder et al., 2012; Wu et al., 2018). However, for many groups of organisms, immune-associated energetic costs have not been quantified, and the intrinsic and extrinsic factors that impact allocation priorities have not been identified. Broadening studies to enhance our knowledge of the interspecific and intraspecific sources of variation in physiological responses to immune challenges may help shed light on the evolution of energy allocation strategies, as well as inform efforts to conserve wild populations threatened by emergent pathogens.

The metabolic costs of immune system activation have been well described in endothermic vertebrates. In mammals and birds, immune challenge with non-pathogenic antigens elicits a 5–30% increase in resting metabolic rate (RMR; Demas et al., 2012; Hasselquist and Nilsson, 2012; Lochmiller and Deerenberg, 2000; Martin et al., 2003). Studies on endotherms also clearly demonstrate that immune investment trades off against competing physiological functions, such as reproduction (e.g. Bonneaud et al., 2003; French et al., 2013). In contrast, studies on ectothermic vertebrates are less common, are not methodologically consistent and have yielded mixed results. For example, a majority of studies indicate that the metabolic cost of mounting an immune response is negligible in squamate reptiles (Malvin and Kluger, 1979; Meylan et al., 2010; Smith et al., 2017). However, at least one study has indicated that immune costs are potentially high in lizards (Cox et al., 2015) and other studies clearly show that immune activation trades off against reproduction in oviparous lizards in a manner similar to endotherms (e.g. Durso and French, 2018; French et al., 2007; Uller et al., 2006). Furthermore, in vertebrates generally, few studies have compared the energetic cost of mounting an immune response among reproductive states (including during pregnancy; Cutrera et al., 2010; Merlo et al., 2014), which would help elucidate the proximate mechanisms underlying tradeoffs between immunity and reproduction.

The physiological state of pregnancy has been most intensively studied in placental mammals, in which pregnancy is associated with a suite of changes in the maternal immune system (e.g. Drazen et al., 2003; Sacks et al., 1999; Weetman, 2010), including attenuated or divergent physiological responses to immune challenge in pregnant compared with non-reproductive individuals (Christe et al., 2000; Cutrera et al., 2010; French et al., 2013; Martin et al., 1995). Downregulation of immune performance during pregnancy may result from either energetic tradeoffs imposed by the metabolic costs of pregnancy or the need to protect non-self fetal antigens from the maternal immune system (Bainbridge, 2000). While many physiological components of the reproductive process are highly conserved (e.g. Brandley et al., 2012; Callard et al., 1992), both the metabolic costs of pregnancy and the nature of placental transfer vary widely among viviparous vertebrates (Foucart et al., 2014; Schultz et al., 2008; Stewart and Thompson, 2000; Van Dyke and Beaupre, 2011). The immunological mechanisms that facilitate placental exchange may therefore exhibit understudied phylogenetic variation. In squamate reptiles alone, viviparity, with varying degrees of placentation and feto-maternal exchange, has independently evolved over 100 times (Blackburn, 2006). Thus, studies in viviparous squamates provide an opportunity to explore the evolution of the immunological mechanisms that facilitate pregnancy in non-mammalian vertebrates (Van Dyke et al., 2014).

Crotaline snakes (pitvipers) have evolved a viviparous reproductive mode independently from other squamate groups (Blackburn, 2006). Although primarily lecithotrophic, viviparous pitvipers have a placenta and can transfer nutrients, including amino acids, to offspring during pregnancy (Van Dyke and Beaupre, 2012). Additionally, the pregnancy process may be metabolically costly and is associated with elevated glucocorticoids (Lind et al., 2020; Smith et al., 2012; Taylor et al., 2004). Energetic tradeoffs, the immunosuppressive effects of glucocorticoids (reviewed by Moore and Hopkins, 2009) and/or modulation of immune factors to protect developing offspring may contribute towards alteration of immune function during pregnancy in snakes. Nevertheless, the direct impact of pregnancy on immune performance is little understood in snakes and in reptiles in general (but see Graham et al., 2011; McCoy et al., 2017; Palacios and Bronikowski, 2017). Pitvipers, therefore, provide a comparative model to investigate immunity-induced tradeoffs that arise during pregnancy. Investigating tradeoffs in field-acclimatized individuals can also provide a mechanistic understanding of why immunocompetence varies in association with seasonal reproductive patterns (Martin et al., 2008), which may be of value in conservation contexts (Willis, 2015).

To quantify the metabolic cost of immune performance and to examine how costs may vary in association with the physiological state of pregnancy, we artificially elicited an immune response in pregnant and non-pregnant pygmy rattlesnakes using lipopolysaccharide (LPS) extracted from the bacterium E. coli. The pygmy rattlesnake (Sistrurus miliarius) is a viviparous pitviper native to the southeastern United States (Conant and Collins, 1998). We hypothesized that LPS injection induces a metabolically costly innate immune response. Additionally, we hypothesized that pregnancy, either through energetic tradeoffs or modulation to protect developing embryos, impairs immune responses. We therefore predicted that we would observe an increase in resting metabolic rate, plasma bactericidal ability, and the relative number of circulating heterophils in response to LPS challenge, and that these immune responses would be attenuated in pregnant compared with non-reproductive individuals. Testing these hypotheses may contribute towards a mechanistic understanding of the impacts of emergent pathogens on non-mammalian vertebrates.

From late July to early August, 2018, male (N=12), non-reproductive female (N=17) and pregnant (N=24) Sistrurus miliarius (Linnaeus 1766) were sampled in the field, collected and housed in 0.45×0.6 m outdoor enclosures (see Lind et al., 2017 for detailed housing information). Snakes were collected from Lake Monroe Conservation Area and Lake Woodruff National Wildlife Refuge in central Florida, USA. Upon capture, a blood sample (0.1–0.3 ml) was taken from the caudal vein using a 1 ml syringe and a 27 gauge needle. Non-reproductive individuals were held in outdoor enclosures and fasted for 8–22 days (mean=15.0 days) prior to baseline respirometry measurements to ensure a post-absorptive state (Zaidan and Beaupre, 2003). Pregnant females were assumed to be aphagic and therefore were not subjected to a similar fasting period (holding time in outdoor enclosures prior to baseline respirometry measurements: 1–8 days, mean=4.2 days). Subsequent to experimentation, pregnant females were maintained in outdoor enclosures and monitored daily until parturition, at which point litter characteristics (number and mass of living and stillborn offspring) were recorded. Pygmy rattlesnake populations in central Florida are afflicted with ophidiomycosis (snake fungal disease, SFD). All snakes were assigned SFD clinical sign scores (see McCoy et al., 2017) and swabbed cutaneously for DNA of the causative agent, Ophidiomyces ophiodiicola (Oo), as part of ongoing field projects on SFD in the population. Any snake with clinical signs of disease and Oo DNA amplification above the lower limit of detectability of PCR tests (Bohuski et al., 2015) at the time of measurement was removed from analyses. One snake diagnosed with pentastomiasis, caused by the invasive lung parasite Raillietiella orientalis (Farrell et al., 2019), was also removed from all analyses. All snakes were released at their site of capture after the experiment. Care and use of vertebrate animals was overseen by the Stetson University IACUC committee (protocols SU-IACUC-140, SU-IACUC-142, SU-IACUC-154).

Snakes were randomly assigned to control and LPS treatment groups. RMR (as CO₂ production rate at 32°C over an 18 h sampling period; see below) was measured in a paired design (i.e. before and after treatment for each snake). LPS snakes received an intraperitoneal injection of 20 mg kg⁻¹ LPS (E. coli serotype O128:B12; cat. no. L2755, Sigma-Aldrich, St Louis, MO, USA) dissolved in sterile phosphate-buffered saline (PBS). LPS solutions were adjusted to an injection volume of 1 ml 100 g⁻¹ body mass. Control snakes received a sham injection of 1 ml sterile PBS vehicle 100 g⁻¹ body mass. All injections were administered between 10:00–14:30 h EST. All pregnant females were injected with their randomly assigned treatment 1 day after commencement of baseline RMR measurement. Males and non-reproductive females were injected 1–13 days (mean 5.0 days) after commencement of baseline RMR measurement. Snakes were randomly assigned to blocks of seven individuals for daily respirometry measurements, with the restriction that completion of pregnant female sampling was prioritized first because of the uncertainty in individual timing of parturition. In all cases, post-injection RMR measurement began on the day following injection, such that mean RMR was estimated over an 18 h period during the ∼27–48 h after injection. All snakes were maintained individually in the laboratory at room temperature (∼22–24°C) with water available ad libitum during the time period between pre- and post-injection RMR measurements. After completion of post-injection RMR measurements, snakes were immediately removed from metabolic chambers and blood was sampled via caudal venipuncture using the methods described above.

Samples were run in two separate assays. Any snake that was digesting a meal when sampled in the field (N=2) was removed from the analysis.

The cellular response to LPS injection was quantified from blood smears taken two days post-injection. Blood smears were stained with Wright–Giemsa using a standard protocol. The cellular response to bacterial antigens in reptiles involves an increase in circulating heterophils (Goessling et al., 2017; 2019; Merchant et al., 2006). To examine the magnitude of this response, the ratio of the number of heterophils to the number of lymphocytes (H:L ratio) was quantified from digital photographs taken from 50 microscope fields under oil immersion at 1000× magnification. Leukocyte counts from these 50 images were conducted using ImageJ.

Effects of treatment and reproductive category on RMR were examined using repeated-measures analysis of covariance (rmANCOVA) in SAS PROC MIXED. The model included the between-subjects effects of treatment (LPS or control) and reproductive category (male, non-reproductive female, or pregnant), the within-subjects effect of time (before or after injection), and their interactions on log₁₀ V̇_CO2. Log₁₀ body mass was included as a covariate.

The effects of treatment and reproductive category on BKA score were analyzed within each combination of treatment and reproductive category (non-reproductive or pregnant) by calculating the difference in the BKA score of plasma taken from individuals in the field and the BKA score of plasma taken after injection (ΔBKA). Exclusion of field samples from non-reproductive snakes (see criteria above: N=7) led to a low sample size in the non-reproductive groups. Therefore, males and non-reproductive females were grouped in the analysis of ΔBKA. We therefore refer to the main effect as grouped reproductive category. Because many snakes killed 100% of E. coli CFUs at 32°C, and individual variation above 100% was not possible at the high incubation temperature, only data from the 25°C incubation protocol were analyzed. ΔBKA at 25°C was used as a response variable in a generalized linear model in SAS PROC MIXED that included the fixed effects of grouped reproductive category (non-reproductive or pregnant female), treatment (LPS or control), and their interaction. The assay in which each sample was run was included as a random effect. The effect of female reproductive status (non-reproductive female versus pregnant) on arcsine-transformed BKA score at 25°C was also analyzed in free-ranging, unmanipulated females in a mixed model including assay as a random effect.

Blood smears were not taken in the field and H:L ratios were determined for post-treatment snakes only. To satisfy the normality assumption of parametric statistics, H:L ratios were square root-transformed (Kolmogorov–Smirnov, KS=0.11, P>0.150). The effects of reproductive category (male, non-reproductive female, and pregnant female), treatment (LPS or control), and their interaction on square root-transformed H:L ratio were analyzed in a generalized linear model in SAS PROC GLM.

Pregnant females were dosed based on their total mass measured at the time of respirometry measurements (i.e. maternal mass+mass of embryos), and thus variation in the effective dose administered to each pregnant female could represent an additional source of variation in response variables measured in the present study. To test whether variation in effective dose explained individual variation in the metabolic response to LPS of pregnant females, LPS dose was calculated relative to maternal mass (as mg LPS kg⁻¹ postpartum mass), and a multiple regression was performed with post-injection log₁₀ V̇_CO2 as the response and effective dose and pre-injection log₁₀ V̇_CO2 as independent variables. Additionally, effects of dosage on ΔBKA and square root-transformed H:L ratio were analyzed using linear regression.

ANCOVA or ANOVA (see Table 2 for model details) was used to test for the effect of treatment (LPS or control) on four litter characteristics: live litter mass, total litter mass, live litter size and mean mass of individual live offspring. Postpartum mass and parturition date were investigated as covariates (and retained in the model when their effects were significant) because prior research on the reproductive ecology of S. miliarius in central Florida has demonstrated that larger females and females that give birth earlier in the season produce litters of larger mass (Farrell et al., 1995). To account for differences in the timing of LPS treatment in relation to parturition date (individual females were injected 9–31 days prior to parturition, mean 15.2 days), we divided the sample of 20 pregnant females that produced litters including viable offspring roughly in half into early (>14 days prepartum; N=9) and late (<14 days prepartum; N=11) pregnancy stages. We then analyzed the impact of LPS injection on litter characteristics in a model including all pregnant females and again in a model restricted to those females injected during early pregnancy.

Statistical significance was assessed at P<0.05 in all analyses. Reported pairwise comparisons were made using post hoc Tukey tests.

There were significant effects of reproductive category (F_2,35.9=3.39, P=0.045), time (F_1,36.8=37.61, P<0.0001), treatment×time (F_1,36.6=31.11, P<0.0001) and log₁₀ body mass (F_1,38.3=95.07, P<0.0001) on least-squares mean (lsmean) log₁₀ V̇_CO2; all other effects were not significant (Table 1, Fig. 1). Regardless of reproductive category, lsmean log₁₀ V̇_CO2 increased significantly in response to LPS injection (adj. P<0.0001) but was unaffected by control injection (adj. P=0.977). Based on comparisons of back-transformed lsmeans, LPS challenge increased mean V̇_CO2 by 25.9%, 18.0% and 12.6% in males, non-reproductive females and pregnant females, respectively. Overall, pregnant females exhibited significantly higher lsmean log₁₀ V̇_CO2 than non-reproductive females (adj. P=0.038). Lsmean log₁₀ V̇_CO2 of males did not differ from that of either female group (adj. P≥0.280).

The estimated effective dose of LPS administered to pregnant females used in the respirometry portion of the study varied from 27.6–37.8 mg kg⁻¹ postpartum mass (mean=33.0 mg kg⁻¹). While accounting for the significant effect of pre-injection log₁₀ V̇_CO2 (P<0.0001), post-injection log₁₀ V̇_CO2 of LPS-treated pregnant females was unaffected by effective LPS dose (P=0.686).

Injection with LPS significantly increased mean H:L ratio (F_1,38=21.70, P<0.001; Fig. 2). There was also a significant effect of reproductive category on H:L ratios (F_2,38=4.04, P=0.026). Post hoc Tukey tests indicated that males had significantly higher H:L ratios compared with pregnant and non-reproductive females, which did not differ from each other. There was no significant interaction between reproductive category and LPS treatment (F_2,38=2.57, P=0.096).

There were significant effects of grouped reproductive category (F_1,32=10.35, P=0.003) and the interaction between treatment and grouped reproductive category (F_1,32.1=5.35, P=0.027) on mean ΔBKA. There was no significant overall treatment effect (F_1,33=0.41, P=0.527). Post hoc Tukey tests indicated that LPS-injected pregnant snakes exhibited a significantly greater decline in BKA (lower ΔBKA) compared with LPS-injected non-reproductive snakes and control non-reproductive snakes (Fig. 3). A reduced analysis of effects of treatment and reproductive category on ΔBKA including only females yielded a significant main effect of female reproductive status (F_1,24=6.27, P=0.020) and a significant treatment×reproductive category interaction (F_1,24=4.65, P=0.041). Neither ΔBKA (P=0.613) or H:L ratio (P=0.829) was affected by variation in effective LPS dose. Pregnant females exhibited higher BKA scores compared with non-reproductive females when sampled in the field prior to experimental manipulation (F_1,29.1=4.95, P=0.030).

Pregnant females in the LPS group produced litters with significantly lower lsmean log₁₀ live litter mass compared with control females (F_1,16=4.60, P=0.048; Table 2), after adjusting for the significant positive effect of log₁₀ postpartum mass (F_1,16=8.66, P=0.010) and the significant negative effect of parturition date (F_1,16=5.71, P=0.030) on log₁₀ live litter mass. Based on comparison of back-transformed lsmeans, the effect of an LPS challenge during pregnancy equated to a mean 5.1 g reduction in live litter mass. Other measures of reproductive output were not significantly affected by LPS challenge in the full analysis including all pregnant females (Table 2). When the analysis was restricted to those females injected during early pregnancy, LPS treatment resulted in significantly lower live litter mass, total litter mass and live litter size, but did not affect the mean mass of individual live offspring (Table 2).

Lipopolysaccharide challenge elicited a metabolically costly immune response in S. miliarius. The metabolic response to LPS injection was similar in magnitude to that of other vertebrates (Demas et al., 2012; Lochmiller and Deerenberg, 2000), but contrasts with several previous studies that did not detect a significant metabolic effect of immune activation in other squamates (Malvin and Kluger, 1979; Meylan et al., 2010; Smith et al., 2017). Immune challenge also elicited a cellular response consistent with reports in other reptiles challenged with LPS (Goessling et al., 2017; 2019; Merchant et al., 2006). Counter to experimental predictions, LPS challenge did not increase plasma bactericidal ability, and BKA scores declined in response to LPS challenge in pregnant females compared with non-reproductive individuals. LPS treatment significantly reduced several litter characteristics, particularly when administered during early pregnancy. Taken together, our results support the hypothesis that immune activation is costly in both non-reproductive and pregnant S. miliarius and that immune activation during pregnancy may negatively impact offspring.

Treatment with LPS elicited a cellular response characterized by a relative increase in circulating heterophils (i.e. increase in H:L ratio). Heterophils are the functional equivalent of the mammalian neutrophil and are a component of the inflammatory response (Zimmerman et al., 2010). The observed shift in circulating leukocytes indicated that the experimental treatment was successful in eliciting an acute immune response across reproductive categories. The leukocyte response was similar to the observed cellular response to LPS treatment in chelonians (Goessling et al., 2017, 2019) and crocodilians (Merchant et al., 2006).

The effect of LPS on plasma bactericidal capacity was not consistent with experimental predictions. Bactericidal assays provide a functional measure of the performance of complement and acute-phase proteins in the blood plasma (Demas et al., 2011). Bacterial LPS acts as a pathogen-associated molecular pattern that interacts with complement proteins and activates an innate immune response across vertebrate groups (Day et al., 1970; Fritz and Girardin, 2005). Little is known regarding the effect of LPS on complement performance in squamates, but immune challenge with LPS elicits an increase in plasma BKA in chelonians (Goessling et al., 2017, 2019). While mean ΔBKA increased in response to LPS treatment in non-reproductive snakes, the result was not statistically significant. In pregnant females, LPS reduced plasma bactericidal ability (i.e. lowered ΔBKA) compared with both control and LPS-treated non-reproductive snakes. Although an attenuated immune response in pregnant snakes was predicted, a decline in bactericidal ability in response to LPS challenge was unexpected.

The decline in the BKA score of pregnant females compared with non-reproductive snakes challenged with LPS suggests that the complement response to LPS is impaired or potentially reversed during pregnancy in S. miliarius. In contrast, the pre-treatment field BKA score of pregnant females was significantly higher compared with non-reproductive females. Previous studies on viviparous snakes report either reduced (Graham et al., 2011) or equivalent (Palacios and Bronikowski, 2017) plasma BKA in pregnant females compared with non-reproductive individuals. However, a previous study on free-ranging central Florida S. miliarius found that pregnant snakes have higher mean BKA scores (not statistically significant), compared with non-reproductive snakes (McCoy et al., 2017). Our results support this finding and demonstrate that pregnancy is associated with increased constitutive complement activity in free-ranging S. miliarius. The physiological mechanisms that may underlie the decline in BKA score in response to LPS treatment in pregnant females were beyond the scope of the current study and are open to interpretation. Complement function during mammalian pregnancy is a balance between protecting the feto–uterine interface from infection and avoiding inflammation in response to non-self fetal antigens (Denny et al., 2013). Research in mouse models indicates that overactive complement activity can lead to adverse pregnancy outcomes, including fetal loss (Girardi et al., 2011). Blockade of specific complement pathways can ameliorate these adverse effects (Gelber et al., 2015). Stated simply, mammalian models indicate that facilitating successful pregnancy may require up- or downregulation of specific complement components. Complement expression during pregnancy may be similarly nuanced in placental squamates. Indeed, uterine expression of complement component 3, a specific protein in the complement cascade, is downregulated during pregnancy in a viviparous lizard (Brandley et al., 2012). Field differences and LPS effects on BKA score in S. miliarius both indicate that complement function is modulated during pregnancy, but that the direction of such modulation during pregnancy depends upon context (i.e. unmanipulated versus immune-challenged). Nevertheless, the conditional factors that impact immune function during pregnancy in non-mammalian vertebrates remain poorly understood.

Non-reproductive S. miliarius exhibited an average metabolic increment of 22% (26% in males and 18% in non-reproductive females) in response to LPS injection. The observed increments are within the range of metabolic increments reported for birds challenged with LPS (Burness et al., 2010; Hegemann et al., 2012; King and Swanson, 2013; Marais et al., 2011) and are at the low end of the range of metabolic increments reported in amphibians (Llewellyn et al., 2012; Moretti et al., 2018; Sherman and Stephens, 1998). In brown anole lizards, Anolis sagrei, individual metabolic increments are positively correlated with the magnitude of the immune response and may be high in individuals that exhibit a strong response to phytohaemagglutinin injection (Cox et al., 2015). However, several squamates do not exhibit significant metabolic increments in response to immune activation (e.g. Malvin and Kluger, 1979; Meylan et al., 2010; Smith et al., 2017). Temperature and dose may impact the physiological response to LPS in ectotherms (e.g. Llewellyn et al., 2012; Merchant et al., 2006; Mondal and Rai, 2001; Sherman and Stephens, 1998), and differences in methodology may underlie the conflicting results reported in reptile studies. For example, Moretti et al. (2018) found that the metabolic response to LPS injection was undetectable when Cururu toads, Rhinella icterica, were held at field night-time or febrile temperatures (17°C and 26°C, respectively), but a significant metabolic increment was observed at an intermediate temperature (22°C). In the current study, metabolism was measured at an ecologically relevant body temperature at or above temperatures associated with behavioral fever in snakes (Burns et al., 1996; Tetzlaff et al., 2017) and consistent with field body temperatures taken from summer-active S. miliarius in central Florida (May et al., 1996). In a study on the costs of coping with ophidiomycosis, afflicted S. miliarius exhibited elevated RMR compared with healthy individuals across three ecologically relevant temperatures (17, 25 and 32°C; Agugliaro et al., 2020). Snakes with ophidiomycosis exhibited a mean metabolic increment of 35% (45% at 17°C, 30% at 25°C, 30% at 32°C), demonstrating that coping costs are detectable across a broad range of body temperatures and potentially higher at lower temperatures. It is, therefore, likely that the response to LPS would be detectable across a wide range of temperatures in S. miliarius. Future studies considering the metabolic response to immune challenge in ectotherms should directly examine the effects of dose and temperature, as these factors may be critical in formulating cost estimates that are applicable across ecologically relevant contexts.

Pregnant female S. miliarius exhibited a statistically significant 13% increase in RMR in response to LPS treatment. Although not significantly different in statistical tests, the magnitude of this response was, notably, half that of LPS-challenged males and 70% that of LPS-challenged non-reproductive females. The observed metabolic increment in pregnant females was coincident with a significant decline in plasma BKA compared with non-reproductive snakes and a significant increase in H:L ratio relative to control snakes, indicating that pregnant female S. miliarius are able to mount a metabolically costly immune response to a non-pathogenic antigen. The energetic costs associated with immune responses have the potential to induce tradeoffs that impact competing physiological functions. It should be acknowledged that, because the LPS dose was calculated on total mass (including the litter), pregnant females likely received a higher effective dose compared to non-reproductive animals. Uncertainty regarding the dose-dependence of the metabolic and immune response to LPS challenge in S. miliarius makes it impossible to rule out the potential for dosage effects on immune responses. However, increased LPS dosage increases the in vivo cellular response to immune challenge in American alligators, Alligator mississippiensis (Merchant et al., 2006), and dosage is positively associated with hepatic amino acid allocation in male brown anole lizards, Anolis sagrei (Brace et al., 2015). Assuming that dosage effects are similar in pygmy rattlesnakes compared to other reptiles studied in vivo (i.e. a higher dose elicits a stronger response), it is unlikely that any potential attenuation of the metabolic response and decline in plasma BKA score in pregnant females was the result of a higher effective dose. Indeed, across the range of effective LPS doses administered to pregnant S. miliarius in this study, no response variable was sensitive to variation in effective dose.

LPS administration reduced live litter mass compared to PBS-injected controls, but did not impact other litter characteristics when analyses included all pregnant females. However, when administered during early pregnancy, LPS significantly reduced all litter characteristics other than mean mass of individual live offspring, despite the limited sample size (Table 2). Likewise, immune challenge during vitellogenesis reduces follicle and egg size in oviparous squamates (French et al., 2007; Uller et al., 2006), and activation of the immune system via administration of sheep red blood cells in European common lizards, Zootoca vivipara, during early pregnancy elicits a significant reduction in total litter mass (Meylan et al., 2013). Viviparous pitvipers are thought to be primarily lecithotrophic (Van Dyke and Beaupre, 2011), and the mechanism by which immune activation during pregnancy may impair offspring growth and development is unclear. A significant metabolic increment and increased H:L ratio in pregnant snakes challenged with LPS, coupled with a decrease in live litter mass, is consistent with a tradeoff between immune function and reproduction. However, the nature of feto–placental transfer of both nutrients and immune factors, and how these processes may trade off during the process of pregnancy in non-mammalian vertebrates, is little understood. In rodent models, intraperitoneal administration of an LPS challenge during pregnancy evokes both maternal and fetal immune responses, and is associated with various dose- and gestational age-dependent negative pregnancy outcomes, including reduced pregnancy success, increased preterm birth rate, reduced litter mass, and/or pathological changes in placental and fetal tissues (e.g. French et al., 2013; Fricke et al., 2018; Hudalla et al., 2018). In those studies, LPS-induced negative effects on offspring have been considered sequelae of the maternal immune response (e.g. resulting from effects of maternal pro-inflammatory cytokines that undergo placental transfer), rather than associated with changes in placental blood flow or direct effects of LPS on developing offspring (Fricke et al., 2018; Gayle et al., 2004; Salminen et al., 2008). However, given the relative paucity of knowledge regarding the nature and extent of placental transfer in snakes, as well as apparent differences in LPS sensitivity in mammals compared with reptiles (e.g. a dose of 2 mg kg⁻¹ LPS, an order of magnitude lower than that used in the present study, is lethal to pregnant mice; Fricke et al., 2018), it is difficult to extrapolate mechanisms from rodent models to viviparous reptiles.

Our results provide an experimental demonstration of the costs of immune responses in a viviparous squamate. The metabolic increments observed in non-reproductive snakes were similar to the metabolic increments associated with an emergent disease (ophidiomycosis) in S. miliarius. Experimental estimates of immune costs may, therefore, be valuable in understanding impacts of emergent diseases and may have predictive power in conservation contexts. The altered responses observed in pregnant females may be the result of tradeoffs between the competing energetic demands of pregnancy and immune function. However, the disparity between field-active BKA and ΔBKA in response to immune challenge in pregnant females suggests context-dependent modulation of immune components during pregnancy. While our study provides an important demonstration of the physiological response to immune challenge in a viviparous ectotherm, our results emphasize the need for more targeted comparative studies of the interaction between specific components of the immune system and the reproductive process in viviparous vertebrates. Immune activation and associated tradeoffs during pregnancy may negatively impact offspring and may contribute towards the sublethal impacts of emergent pathogens. However, the proximate mechanism(s) by which maternal immune challenge impairs fetal growth or development in squamates remains a relatively unexplored frontier.

We thank Priya Gandhi and Jamie Loughlin for assistance with leukocyte counts. Kathryn McLaughlin and Nilanjana Das assisted with bactericidal assays. Samantha McPherson and Erynn Brisson provided field assistance. Candice Stephenson facilitated access to Lake Woodruff National Wildlife Refuge. Jeffrey Lorch and the National Wildlife Health Center assisted with establishing the disease status of individual snakes.