Oxidative damage varies in response to bacterial, fungal and viral antigen challenges in bats Free

Pathogens are a major selective force in animals, driving behavioural and life-history evolution, and even causing dramatic population collapses (Lips et al., 2006; McCallum et al., 2009). Bats are unique in terms of their immunological function. They can act as reservoirs for pathogens with zoonotic potential, without showing obvious clinical symptoms (Wibbelt et al., 2010; Baker et al., 2013). Because of their high species richness, bats are natural reservoir hosts for many viruses (Luis et al., 2013; Mollentze and Streicker, 2020). On the one hand, bats infected with a number of viral strains and other intracellular pathogens are often asymptomatic (Baker et al., 2013; Guito et al., 2021). On the other hand, they exhibit mild to extreme pathologies following infection with various extracellular pathogens, such as some bacteria and fungi (Brook and Dobson, 2015). An important question is why and how bats handle (e.g. resist or tolerate) some viral infections that are often lethal in other species, but are still vulnerable to other pathogens. One answer may lie in unique immune properties and functional differences in regulating their innate immune system, such as epithelial defence receptors, the natural killer gene complex and the interferon system (Clayton and Munir, 2020; Moreno Santillán et al., 2021). Additionally, it was hypothesized that bats downregulate their inflammatory pathways and have evolved a unique anti-inflammatory response to deal with flight associated metabolic stress (Gorbunova et al., 2020; Kacprzyk et al., 2017). Another relevant factor could lie with the costs of immune activation that can result in immunopathological consequences (Sheldon and Verhulst, 1996; Ahn et al., 2019). The existence of these costs associated with an immune response would favour the maintenance of immunological variation because they are likely to vary with individual condition, environment, species and diversity of pathogens the hosts are exposed to (e.g. Sheldon and Verhulst, 1996; Klasing, 2004; Maizels and Nussey, 2013). Much research has documented that activation of an immune response is costly for the organism in terms of increased energy expenditure and nutrient consumption (e.g. Lochmiller and Deerenberg, 2000; Hasselquist and Nilsson, 2012; Gombart et al., 2020). The generation of molecular oxidative damage and perturbations in antioxidant levels that occur during infection are also deemed to be relevant proximate mechanisms underlying immunopathological consequences because of their potential detrimental effects on cell homeostasis, fertility or reproduction (Sorci and Faivre, 2009; Costantini, 2019, 2022).

Phagocytes contain a multi-component enzyme complex, specifically nicotinamide adenine dinucleotide phosphate (NADPH) oxidase, which is responsible for the production of reactive oxygen species (ROS) during an immune response (Babior, 2004). During an immune challenge, phagocytes increase their oxygen uptake (the so-called respiratory or oxidative burst) to generate ROS that are cytotoxic to pathogens and act as signalling molecules to orchestrate the inflammatory response (Babior et al., 1973; Segal, 2005). However, ROS can also cause oxidation (i.e. oxidative damage) to important biomolecules such as proteins, lipids or nucleic acids (Halliwell and Gutteridge, 2015). This can lead to pathological consequences for the host (Costantini, 2014). Thus, natural selection has likely acted to optimise the need for an effective ROS-mediated immune response against the potential detrimental side effects of ROS on cellular and whole-body homeostasis. This may have contributed to shaping the way in which organisms interact with their pathogens.

Previous work has shown that free-living female greater mouse-eared bats (Myotis myotis) mount a strong acute phase response to a subcutaneous injection of lipopolysaccharide (LPS), an endotoxin from the cell walls of gram-negative bacteria used experimentally to simulate bacterial infection (Seltmann et al., 2022). Conversely, conspecific individuals mounted a less pronounced acute phase response to injections of polyinosinic:polycytidylic acid (PolyI:C, a synthetic RNA used to experimentally simulate a viral infection) or zymosan (a fungal glucan used to experimentally simulate a fungal infection) antigens (Seltmann et al., 2022). These results suggest that greater mouse-eared bats may better handle viral or fungal infections than bacterial ones. Following up on these results, we tested the hypotheses that: (i) the antigen-specific immunological responsiveness of greater mouse-eared bats leads to an antigen-specific response in oxidative status, and (ii) because of their weak response to viral infections, bats show no to small changes in oxidative status to PolyI:C than to LPS and zymosan injections. If bats reduce their responsiveness to an antigen because of the associated physiological costs, we expected that the reduced immunological responsiveness of bats to zymosan and PolyI:C (Seltmann et al., 2022) would result in lower levels of oxidative damage and stable baseline circulating antioxidant levels. Finally, we assessed correlations between markers of oxidative status and immunity from our previous study (Seltmann et al., 2022), namely, immune cells, haptoglobin levels (the major acute phase protein in bats) and the bacterial killing assay (BKA; a marker of overall innate immune system functionality) (Fritze et al., 2019, 2021; Voigt et al., 2020; Moreno Santillán et al., 2021; Costantini et al., 2022).

Experimental details were previously described in Seltmann et al. (2022). Briefly, fieldwork was conducted from 15 to 26 August 2015 at the Tabachka Bat Research Station, Bulgaria (permits 54-2532.2-9/10, 55.1-8642G062/10 and 238 55.1-8642-01-17/10). Permits (639/28.05.2015) were obtained from the relevant Bulgarian authorities (MOEW-Sofia and RIOSV-Ruse) to conduct this research. Thirty-eight non-lactating adult female greater mouse-eared bats [Myotis myotis (Borkhausen 1797)] were captured using harp traps. The bats were housed individually in boxes, fed 5 g of mealworms daily and provided with water ad libitum. The experimental injection of immunostimulants started 36 h later to allow the bats to acclimatise to the new environment. Individual bats were randomly assigned to four experimental groups – control (saline), LPS, zymosan and PolyI:C – and they were injected as described previously (Seltmann et al., 2022). A blood sample (∼70 µl) and body mass measurement were taken from each individual bat just before the injection, and then again 24 and 48 h after injection, so that each bat was bled three times during the experiment. A blood smear was prepared for white blood cell counts (Schneeberger et al., 2013) and the tubes containing the blood were immediately centrifuged. This separated the plasma from the red blood cells. Both plasma and RNAlater-fixed red blood cells were stored in liquid nitrogen in the field and at −80°C in the laboratory. All bats were released at the end of the experiment. Plasma samples have been used to measure two markers on innate immunity, haptoglobin and BKA, details of which were published previously (Seltmann et al., 2022).

For this study, we selected a marker of plasma oxidative damage and a marker of plasma non-enzymatic antioxidant protection, which have previously been shown to be sensitive to immune challenge (Costantini, 2022). Using the leftover samples after immune measurements, these two markers were assessed using commercially available kits that are commonly applied to vertebrates, including bats (Costantini and Dell'Omo, 2006; Schneeberger et al., 2013), following the manufacturer's instructions unless otherwise stated. Reactive oxygen metabolites (mainly organic hydroperoxides) were measured using the d-ROMs assay (Diacron International, Grosseto, Italy). The values were expressed as mmol l⁻¹ H₂O₂ equivalents. The OXY-Adsorbent test (Diacron International) was used to quantify the ability of non-enzymatic antioxidant compounds present in plasma to cope with the in vitro oxidative effect of hypochlorous acid (HOCl, an endogenous oxidant produced by immune cells). Values were expressed either as mmol l⁻¹ HOCl neutralised or as mmol l⁻¹ HOCl neutralised per mg protein to estimate the antioxidant potential of micromolecular antioxidants (e.g. vitamins, carotenoids, glutathione) without the contribution of protein (i.e. non-enzymatic micromolecular antioxidant capacity). Quality controls were included in all assays performed. The Bradford protein assay (Bio-Rad Laboratories, Hercules, CA, USA) with albumin as reference standard was used to quantify the concentration of plasma proteins. Quality controls were included in all assays performed.

To assess the correlation between oxidative stress and immunity, we used previously published data on white blood cell counts (lymphocytes, neutrophils, eosinophils, monocytes and basophils), haptoglobin and plasma BKA (Seltmann et al., 2022).

We were unable to assess both markers of oxidative status in all females at each sampling time point, either because the plasma sample was haemolysed or because the volume was too small. We had completed data for 22 individuals (4 control, 6 LPS, 7 PolyI:C, 5 zymosan) for reactive oxygen metabolites, and 29 individuals (5 control, 10 LPS, 7 PolyI:C, 7 zymosan) for non-enzymatic antioxidant capacity. Thus, we first ran general linear mixed models in SPSS (version 23) on a reduced dataset including only those individuals for which we had data for each sampling day. Then, we ran general linear mixed models on the full dataset including all individuals. In each model, we included experimental group, sampling day and their interaction as fixed factors. The term individual was included as a random factor. We performed Fisher LSD post hoc tests when the interaction between experimental group and sampling day was significant. We used Pearson's test to quantify the correlations between (i) oxidative status and immune markers, previously published by Seltmann et al. (2022), within each sampling day, and (ii) changes in markers over the experimental period, calculated at the individual level (i.e. value of marker A at 48 h in individual X minus value of marker A in the same individual X before antigen injection). For each sampling day, we pooled all individuals together, irrespective of their experimental group, because otherwise the sample size would not have been adequate to estimate robust correlations.

For both full and reduced datasets, general linear mixed models detected a significant interaction between experimental group and sampling day for plasma reactive oxygen metabolites (P≤0.002), but not for the plasma non-enzymatic antioxidant capacity (P≥0.807) (Fig. 1). Models based on the full dataset showed that pre-injection plasma levels of reactive oxygen metabolites were similar between the four experimental groups (F=2.41, P=0.10; Fig. 1A). By contrast, we found differences between groups in the levels of reactive oxygen metabolites 24 h (F=3.59, P=0.025) and 48 h (F=9.96, P<0.001) post-injection. Post hoc comparisons showed that at 24 h post-injection, control bats had significantly lower plasma levels of reactive oxygen metabolites than those of bats challenged with bacterial (Fisher LSD, P=0.005) or viral (Fisher LSD, P=0.021) antigens (Fig. 1B). At 48 h post-injection, LPS-challenged bats had significantly higher plasma levels of reactive oxygen metabolites than control bats (Fisher LSD, P<0.001), or bats that were injected with PolyI:C (Fisher LSD, P<0.001) or zymosan (Fisher LSD, P=0.023; Fig. 1C); and zymosan-challenged bats had significantly higher plasma levels of reactive oxygen metabolites than control bats (Fisher LSD, P=0. 035) and PolyI:C bats (Fisher LSD, P=0.056). By contrast, control bats and PolyI:C bats had similar levels of reactive oxygen metabolites (Fisher LSD, P=0.69; Fig. 1C).

We found that plasma levels of reactive oxygen metabolites were positively but weakly correlated with plasma levels of haptoglobin before injection, and 24 or 48 h after injection (Fig. 2A–C, Table 1). Two days after the injection, plasma non-enzymatic antioxidant capacity was positively but weakly correlated with eosinophils (Fig. 2D), while plasma non-enzymatic micromolecular antioxidant capacity was negatively and positively correlated with neutrophils (Fig. 2E) and eosinophils (Fig. 2F), respectively (Table 1). All other correlations were not significant (Table 1).

The change in BKA over the experimental period correlated positively with the change in reactive oxygen metabolites (Fig. 2G, Table 1). The changes in levels of eosinophils and of neutrophils were correlated positively and negatively with levels of plasma micro-molecular antioxidant capacity, respectively (Fig. 2H,I, Table 1). All other correlations were not significant (Table 1).

The results of our experiment showed that non-lactating adult female greater mouse-eared bats injected with a bacterial antigen had significantly higher reactive oxygen metabolites (a marker of intermediate oxidative damage products) than conspecifics of the control group or conspecifics injected with either a viral or a fungal antigen 48 h post-injection. Bats injected with the fungal antigen also had significantly higher reactive oxygen metabolites than control bats or bats injected with the viral antigen 48 h after injection, suggesting a slower response to fungal infection. We also found a moderate increase in reactive oxygen metabolites in bats injected with the viral antigen compared with control bats 24 h after injection. This effect was no longer seen 48 h after injection, suggesting that the response to the viral antigen was rather rapid and transient. Finally, we found that no antigen injection had any effect on circulating non-enzymatic antioxidants.

The results of our experimental study are consistent with the hypothesis that bats generally suffer less damage during viral infections than bacterial or fungal infections. However, our experiment does not enable to infer whether this difference in damage is due to variation in tolerance or resistance to the different types of pathogens. Our results are partially consistent with those we found on the response of immunological markers analysed in the same experiment (Seltmann et al., 2022). On the one hand, we observed similar patterns in the two sets of markers with respect to the response to bacterial antigens. In particular, bats challenged with LPS showed a stronger acute phase response (an increase in the neutrophil-to-lymphocyte ratio and haptoglobin concentration) than bats challenged with the viral or fungal antigens. This effect lasted for the entire duration of the experiment. On the other hand, the immune response to viral and fungal antigens was different: although there was an increase in the neutrophil-to-lymphocyte ratio to the viral antigen, we did not detect any immunological response to the fungal antigen (Seltmann et al., 2022).

We found that plasma levels of reactive oxygen metabolites were positively but weakly correlated with either haptoglobin or BKA. These results indicate that the generation of oxidative damage is weakly associated with these two inflammatory traits, which is consistent with previous studies in bats (Fritze et al., 2019; Costantini et al., 2022). We also found that reactive oxygen metabolites were not correlated with the relative numbers of white cell subtypes known to release ROS into the bloodstream. However, the amount of ROS produced by phagocytes depends on the respiratory activity of immune cells, which we did not measure. It is possible that the production of oxidative damage after an immune challenge depends more on the physiological changes that occur within the white blood cells than on relative changes in their abundance. Prior work on great tits (Parus major) showed that plasma levels of reactive oxygen metabolites were positively correlated with production of the anion superoxide in red blood cells (Delhaye et al., 2016), suggesting that any increase in superoxide production in the bloodstream may in turn increase circulating reactive oxygen metabolites. Previous work in pigeons (Columba livia) also confirmed that LPS injection resulted in increased oxygen consumption at the organism level and that such an increase in oxygen consumption was positively correlated with circulating reactive oxygen metabolites (van de Crommenacker et al., 2010). Thus, an alteration in the metabolism of bats may also have contributed to the increased production of reactive oxygen metabolites.

Another explanation for the contrasting response patterns between markers may lie in the mechanisms governing the interaction between cellular oxidative status and immune function. The cytotoxic effect of ROS released by immune cells is not specific to pathogens, meaning that they can also cause damage to local cells and tissues. For example, prior to immune challenge, anion superoxide levels were negatively correlated with the strength of the subsequent immune response to phytohaemagglutinin in male painted dragon lizards (Ctenophorus pictus) (Tobler et al., 2011). Similarly, blood cells from LPS-challenged painted dragon lizards exhibited a lower oxidative burst when their pre-challenge mitochondrial anion superoxide levels (measured 5 days before LPS challenge) were higher, suggesting an immunoregulatory or limiting effect of this pro-oxidant (Tobler et al., 2015). It may be that the partial or complete lack of leukocyte immune response to the viral or fungal antigens in our bats was to some extent due to a cytotoxic effect of ROS. ROS can also act as immunomodulatory signalling molecules, coordinating the migration of immune cells to the site of infection and promoting the retention of immune cells at that site (Nathan and Cunningham-Bussel, 2013). They can facilitate the adherence of phagocytes to the endothelium or stimulate the synthesis of cytokines (Sorci and Faivre, 2009; Halliwell and Gutteridge, 2015), which in turn can inhibit the generation of ROS in macrophages (Dokka et al., 2001). Strategic suppression of certain immune and inflammatory pathways may be required to mitigate cellular damage. However, this may come at the expense of protection against pathogens. The specific responses of oxidative damage to bacterial or fungal but not viral antigens might explain why bats can be asymptomatically infected with a high number of virulent viral strains but suffer from bacterial or fungal infections. However, it is important to keep in mind that there are other aspects of the oxidative and immunological statuses causative for such differences, and we have not measured these in our work. Thus, further experiments will be needed to explore in more detail the patho-physiological mechanisms underlying the responsiveness of bats to different pathogens.

The non-enzymatic antioxidant defences in plasma were not affected by any immune treatment. This might indicate that (i) the oxidation processes were not strong enough to deplete the circulating antioxidants, (ii) antioxidants were replenished by mobilization from other tissues, or (iii) antioxidants reacted with a time delay. Past studies on birds showed that the antioxidant response may vary between different antigen challenges, and may thus be difficult to predict (Costantini, 2022). For example, the non-enzymatic antioxidant capacity decreased (Cohen et al., 2007) or increased (Marri and Richner, 2015) after LPS challenge. Further, Pašková et al. (2008) found that glutathione and glutathione peroxidase co-varied negatively after immuno-stimulation with cyanobacterial cells. This was confirmed in two other studies (Koinarski et al., 2005; Georgieva et al., 2006), which found a negative covariation between two antioxidant enzymes (catalase and superoxide dismutase) after immune challenge with Eimeria oocysts. Although the immune challenges did not influence levels of circulating antioxidants, our correlation analyses showed that the antioxidant capacity was positively and negatively correlated with eosinophils and neutrophils, respectively. Past studies on pigeons found that circulating non-enzymatic antioxidants were positively correlated with heterophils, the functionally avian equivalent to neutrophils (van de Crommenacker, 2011). Yet, it remains to be demonstrated what the exact mechanisms are that link circulating non-enzymatic antioxidants to white blood cells.

Our results were unlikely to have been influenced by the sampling regime because the acute phase inflammatory response occurs rapidly after organismal exposure to an antigen. Furthermore, we relied on single doses for each antigen and the doses also differed between antigens. Although this might be an issue considering that these antigens trigger different signalling pathways and because we do not know whether pattern recognition receptors ligands were delivered to the cells in comparable amounts between antigens, our approach was similar to other studies on both laboratory and wildlife species, comparing the effect of different antigens [e.g. in laboratory mice in vivo (Arsenault et al., 2014) and in vitro (He et al., 2021), in house sparrows Passer domesticus (Coon et al., 2011), or in red swamp crayfish Procambarus cclarkii (Wu et al., 2017)]. Using the same dosages for each antigen might also lead to erroneous conclusions, as the infection characteristics (e.g. mode of transmission, infectious dosage) of pathogens we mimicked the responses against also vary in a biologically realistic context (real bacterial/viral/fungal infection). This is why we choose dosages that are known to trigger a physiological response in bats (including this species) or are higher than the previously used ones. Although the dosage of LPS (1 mg kg⁻¹ body mass) and zymosan (0.7 mg kg⁻¹) already triggered an immune response in bats (Fritze et al., 2019; Seltmann et al., 2022), the PolyI:C was only used in two other bat species, the great fruit-eating bat (Artibeus lituratus) (Triana-Llanos et al., 2019) and black flying fox (Pteropus alecto) (Periasamy et al., 2019). In both cases, the dosage used was 2 mg kg⁻¹, which is significantly lower than the dose we applied (25 mg kg⁻¹). In case of the black flying fox, as expected, there was no effect on B cell populations (Periasamy et al., 2019), whereas in the great fruit-eating bat, such a relatively low dosage not only increased their resting metabolic rate 6 h post-injection, but also resulted in experimental animals losing more body mass compared with the control animals and showing an increase in body temperature (Triana-Llanos et al., 2019).

In conclusion, our experimental work showed that bats experienced increased plasma oxidative damage only after being challenged with antigens simulating infections with either bacteria or fungi, which are known pathogens with clear pathology for bats. Even if the immunological response to the fungal antigen was similar to that mounted against a viral antigen (Seltmann et al., 2022), the exposure to this viral antigen did not influence the oxidative status of bats. The capacity of bats to control viral infections while mitigating side-effects such as molecular oxidative damage indicates that bats may have evolved specific endogenous response mechanisms that need to be explored further. The evolution of these mechanisms may have also contributed to the longevity of bats (Huang et al., 2019; Wilkinson et al., 2021). Oxidative stress is one cellular mechanism implicated in senescence, and bats are renowned for their propensity to live well beyond expectations given their body mass. Our results point to oxidative stress as one potential currency to quantify the physiological costs associated with the immune responsiveness of bats to different antigens.

We are very grateful to the Directorate of the Rusenski Lom Nature Park (Director Tsonka Hristova) for cooperation and support. We thank Sara Troxell for help in conducting the experiments, Stefan Greif for his support in capturing bats in Bulgaria, Anne Seltmann, Katja Pohle and Marcus Fritze with their help with the immune measurements, and two anonymous reviewers for providing constructive comments on our work.