ABSTRACT
Animals encounter many novel and unpredictable challenges when moving into new areas, including pathogen exposure. Because effective immune defenses against such threats can be costly, plastic immune responses could be particularly advantageous, as such defenses can be engaged only when context warrants activation. DNA methylation is a key regulator of plasticity via its effects on gene expression. In vertebrates, DNA methylation occurs exclusively at CpG dinucleotides and, typically, high DNA methylation decreases gene expression, particularly when it occurs in promoters. The CpG content of gene regulatory regions may therefore represent one form of epigenetic potential (EP), a genomic means to enable gene expression and hence adaptive phenotypic plasticity. Non-native populations of house sparrows (Passer domesticus) – one of the world's most cosmopolitan species – have high EP in the promoter of a key microbial surveillance gene, Toll-like receptor 4 (TLR4), compared with native populations. We previously hypothesized that high EP may enable sparrows to balance the costs and benefits of inflammatory immune responses well, a trait critical to success in novel environments. In the present study, we found support for this hypothesis: house sparrows with high EP in the TLR4 promoter were better able to resist a pathogenic Salmonella enterica infection than sparrows with low EP. These results support the idea that high EP contributes to invasion and perhaps adaptation in novel environments, but the mechanistic details whereby these organismal effects arise remain obscure.
INTRODUCTION
As organisms move into new environments, they can be released from native pathogens and/or be exposed to many novel ones (Keane and Crawley, 2002; Lee et al., 2006, 2005; Liu and Stiling, 2006; Martin et al., 2010; Marzal et al., 2011; Torchin et al., 2001). One defense strategy is therefore unlikely to be amenable to all invasions and range expansions. Indeed, the most adaptive immune response could be the most flexible one (Prüter et al., 2020). Phenotypic plasticity – the ability of the same genome to produce a range of phenotypes – is an important mechanism by which populations can respond rapidly to changing conditions (Pigliucci et al., 2006; Snell-Rood et al., 2018). A better understanding of how selection acts on immune plasticity may therefore be important in understanding the molecular underpinnings of successful range expansions and introductions of individuals outside their native ranges.
Phenotypic plasticity can arise through various mechanisms including epigenetic ones such as DNA methylation (Feinberg, 2007). When DNA methylation occurs within regulatory genomic regions (e.g. gene promoters), it can affect phenotypic plasticity via its effects on the transcriptional regulation of gene expression (Bird, 2002; Lemire et al., 2015; Weaver et al., 2004; Zhi et al., 2013). In vertebrates, DNA methylation almost always occurs at the cytosine residue of CpG sites (i.e. adjacent cytosines and guanines linked by phosphates) on the DNA sequence (Feinberg and Irizarry, 2010). DNA methyltransferase enzymes can directly catalyze the addition and removal of methyl groups at a CpG site (Moore et al., 2013; Shi et al., 2021). Importantly, patterns of DNA methylation can be influenced by environmental factors such as pathogen exposure (Law and Holland, 2019; Qin et al., 2021).
The presence of many CpG sites in the promoter of a particular gene may present more opportunities for the de novo addition and/or removal of methyl groups. Thus, more CpG sites may represent more chances to adjust gene expression via the regulatory effects of DNA methylation on transcription (Kilvitis et al., 2017; Weber et al., 2007). In other words, more CpG sites may enable an individual to fine-tune or update its phenotype rapidly in response to fluctuating challenges, including changes in pathogen exposure (Levis and Pfennig, 2016; West-Eberhard, 2003). CpG content of promoters therefore represents a form of ‘epigenetic potential’, or ‘EP’ for short (Kilvitis et al., 2017).
Previously, we hypothesized that high EP would be favorable in range expansions because it facilitates phenotypic plasticity (Kilvitis et al., 2017). Multiple indirect tests of this hypothesis have supported the role of high EPin the global spread of house sparrows (Passer domesticus), one of the world's most successful, introduced vertebrate species (Hanson et al., 2020b,c). First, in the ongoing range expansion across Kenya, EP across a large fraction of the entire genome of house sparrows increased with distance from the site of initial introduction and was selectively favored towards the range edge (Hanson et al., 2022). Second, across the globe, EP was higher in introduced than in native house sparrow populations for two but not a third Toll-like receptor (TLR) gene (Hanson et al., 2020a). Finally, house sparrows from Tampa, FL, USA, with higher EP in an important microbial surveillance gene (TLR4) had greater inducibility and reversibility of TLR4 expression during an immune challenge in blood, but in spleen and liver in the same birds, TLR4 expression was higher in birds with low EP (Hanson et al., 2021).
The expression of TLR4 is among the most important elements of a protective immune response against bacterial infections (Coburn et al., 2007; Gou et al., 2012). In rodents and chickens, the expression of many factors such as galectin 8 (Hodges and Hecht, 2013), NOD2, NLRP6 and NLRC4 (Thaiss et al., 2016) can protect against bacterial infections including Salmonella. However, the expression of TLR4 is among the most important elements of a protective immune response against this and other bacterial infections (Coburn et al., 2007). Mouse strains genetically deficient for TLR4 are highly susceptible to Salmonella enterica; likewise, a single nucleotide polymorphism (SNP) in the exon of the ligand-binding domain of TLR4, which varies among chicken lines, can affect resistance to and mortality from the bacteria (Leveque et al., 2003). Most TLR4 expression derives from macrophages and heterophils (Alkie et al., 2019), and these cells as well as lymphocytes are recruited in huge numbers to the gut once TLR4 is activated and inflammation is initiated locally by gut pathogens. Dendritic cells, too, residing just below the gut epithelium, express abundant TLR4, and can further sculpt the local and systemic immune response against S. enterica and other infections (Ijaz et al., 2021).
High TLR4 expression thus probably provides protection to infected hosts through various and dynamic means including enhanced phagocytosis and activation of several cell types, which often change over the course of the infection (Coburn et al., 2007; Gou et al., 2012). However, high TLR4 expression can also cause significant host damage via inflammatory over-exuberance (Klein and Diamond, 2008; Kobasa et al., 2007). Indeed, inflammatory responses mediated by TLR4 are among the most expensive and self-damaging immune responses available to vertebrates (Lee and Klasing, 2004; Martin et al., 2017). Plasticity in TLR4 expression should be advantageous because it would balance the costs and benefits of an immune response against pathogens, especially for organisms colonizing areas where threats are evolutionarily novel. In the present study, we tested whether high EP in the TLR4 promoter of house sparrows was associated with an individual bird's ability to resist (shed fewer bacteria in their feces as a result of a more effective gut immune response) and/or tolerate (maintain body mass while infected) a particularly pathogenic serovar of Salmonella enterica Typhimurium. This particular serovar has caused the deaths of thousands of individual passerines worldwide including British house sparrows (Lawson et al., 2014; Mather et al., 2016). Should high EP in TLR4 in house sparrows be implicated in the control of a S. enterica infection, our study would link the patterns of EP in TLR4 observed in wild native and invasive populations to individual performance and hence support a role for EP in range expansion success.
MATERIALS AND METHODS
House sparrow capture and housing
House sparrows [n=38: females n=11, males n=19, juveniles (sex unknown) n=8] were captured via mist nets from different locations across Tampa Bay (FL, USA) in June 2021. The body mass of each bird was recorded (to 0.1 g) at capture, and thereafter birds were kept in opaque cloth bags until transfer to the Biosafety-level 3 (ABSL-3) facility at the University of South Florida. At the ABSL3, birds were housed in individual cages (33×38×46 cm) surrounded by impervious covers (to reduce seed spillage and fecal transfer among cages) around the lower third of each cage. Cages were then placed next to each other in audial and visual contact inside a secondary containment system (bioBUBBLE, Fort Collins, CO, USA), which further ensured no aerosols or feces could circulate among birds. Food (mixed seeds) and water were provided ad libitum throughout the study. Before transferring birds into cages, aluminium foil was placed on the bottom of each cage to collect a fecal sample to determine Salmonella infection status at the beginning of the experiment. Because of space constraints of the bioBUBBLE system, the experiment was conducted in 4 cohorts of birds caught from the same population (cohort 1: n=10, cohort 2: n=7, cohort 3: n=13, cohort 4: n=8). There were no statistically significant differences in Salmonella burden prior to experimental exposure to the pathogenic serovar among cohorts (linear mixed model, estimate=−0.21, s.d.=0.22, t=−0.98, P=0.34) (for further details, see Figs S1, S2).
For the duration of the experiment, birds were checked twice daily, and any individual showing lethargy or other sickness behaviors was euthanized by isoflurane overdose and rapid decapitation. Four birds were euthanized upon detection of sickness (on days 10, 11, 12 and 13), and two birds were found dead on the morning of day 14, preventing the use of tissues from these birds for gene expression analysis. All remaining birds were euthanized 14 days after pathogenic S. enterica exposure. All procedures were approved by the USF Animal Care and Use Committee prior to the start of the study.
Experimental infection and the quantification of S. enterica burden over time
For infections, cryopreserved S. enterica known to be pathogenic for passerines (Hughes et al., 2008) was defrosted rapidly in a warm water bath and diluted to 107 colony-forming units (cfu) in phosphate-buffered saline (PBS). The particular S. enterica serovar used in this experiment, Typhimurium isolate 244, was isolated from a greenfinch (Carduelis chloris) in northern England in 2006 as part of the Garden Bird Health Initiative investigating ‘die-offs’ of passerine birds in the UK (Hughes et al., 2008). The isolate has a DT56 phage type and a sequence type (ST) 586 that was associated with invasive salmonellosis in several passerine species (including house sparrows) and shared a common genotype and pulsed field gel electrophoresis pattern indicating a specific epidemic strain associated with passerines (Hughes et al., 2010).
To achieve infections, birds were gavaged with disposable gavage needles with 100 μl of 107 cfu S. enterica followed by another 100 μl of PBS to flush the needle and ensure each bird received the full dose. This bacterial dosage was used in our experiments as a pilot study showed (i) that S. enterica could be detected post-infection in fecal samples, and (ii) that birds did not show overt signs of sickness or die quickly post-exposure. In other words, the choice of concentration for experimental infection was a compromise between the dose being infective but not causing extensive mortality while also being able to detect the bacterial burden in the feces with our qPCR method. It is also consistent with dosages used in comparable experiments (Connolly et al., 2006). The experimental exposure took place on the day of capture, as even short durations in captivity can lead to immune dysregulation in this species (Love et al., 2017; Martin et al., 2011).
On days 3, 6, 9, 12 and 14 after exposure to S. enterica, foil was again placed on the bottom of each cage to collect fecal samples over time. This method allowed us to quantify bacterial burden (i.e. the amount of S. enterica DNA detected in the feces) in each bird over the course of infection. After collection, each fecal sample was diluted (1:5 mass to volume) in PBS and kept at −80°C until the extraction of S. enterica DNA. To extract S. enterica DNA from fecal samples, a DNA/RNA-free bead was first added to each microtube with diluted feces, then each sample was agitated for 2 min at 2000 rpm on a Bead Mill 24 homogenizer (Fisherbrand). Then, 50 μl of each homogenate was processed for genomic DNA extraction using a QIAmp Powerfecal pro DNA kit (Qiagen) following the manufacturer’s protocol. Salmonella enterica burden in each fecal sample was then determined via quantitative real-time PCR (qPCR).
For qPCR, DNA from the same strain used to infect birds (isolate 244) was extracted from 100 μl of cultured bacteria (107) using a DNEasy Blood and Tissue kit (Qiagen) and quantified using a Qubit Fluorometer and Quant-iT dsDNA HS assay kit (Invitrogen). Following Park et al. (2008), bacterial DNA was diluted to 4×105, 4×104, 4×103, 4×102, 4×101 and 0 genome equivalents per 5 μl. Genome equivalents were calculated using the following equation: DNA genome equivalent=(A×6.022×1023)(660×B)−1, where A is the DNA concentration and B is the length of genomic DNA (Park et al., 2008). Primers and a FAM-probe (TaqMan) validated by Park et al. (2008) for the detection and quantification of S. enterica were then used on these standards to create a standard curve. All qPCR amplifications were performed in a total volume of 25 μl in duplicate on a Rotor-Gene Q system (Qiagen). Each reaction contained 12.5 μl of TaqMan master mix (TaqMan Universal PCR Master Mix, Applied Biosystems), 1 μl of each primer (10 μmol l−1), 0.5 μl of probe, 5 μl of DNA and 5 μl of nuclease-free water. Thermal cycling conditions were a first-step for 2 min at 50°C followed by 10 min at 95°C, then a second step of 40 cycles of 95°C for 15 s and 60°C for 1 min. Fluorescence signals were measured at the end of the extension step for each cycle, and the Salmonella burden in each sample determined by relating Ct values from samples against the standard curve.
Controlling for pre-existing Salmonella burden
Because our study dealt with wild-caught birds, we expected that some birds could already be infected with Salmonella sp. or have been previously infected and recovered. While it is not possible to determine unequivocally whether birds have been previously infected and have cleared any Salmonella variant, we were able to determine whether birds were currently infected by at least some type of Salmonella sp. at the beginning of the experiment and take this pre-existing infection condition into account in our analyses. To do this, we collected a fecal sample at capture to quantify pre-experimental Salmonella sp. burden using qPCR. We included this pre-existing burden measure in all relevant analyses and treated it as a continuous rather than a binary ‘infected versus uninfected’ variable. Importantly, we used a specific serovar of S. enterica in our experiments. This serovar was isolated from a greenfinch in the UK in 2006. Thus, Tampa Bay house sparrows in our study were unlikely to have been pre-infected with this serovar because of geographic distance.
Quantification of body mass and S. enterica tolerance
The body mass of each bird (to 0.5 g) was also recorded at capture (before experimental infection), 24 h after infection, and 3, 6, 9, 12 and 14 days after infection. This approach allowed us to determine how body mass changed over the course of the infection. Comparing each individual's rate of change in body mass relative to its rate of change in S. enterica burden allowed us to estimate a form of each birds ‘tolerance’ to the infection (i.e. an individual's capacity to maintain body mass while infected) (Burgan et al., 2018). We also recorded the occurrence of sickness and mortalities during the experiment.
Quantification of EP
After euthanasia, the whole gut and a piece of the liver were collected and stored in RNA later at −80°C for less than 1 month. Samples were thawed, and DNA was extracted from ∼0.1 g of house sparrow liver tissue using a DNAEasy Blood and Tissue kit (Qiagen). Kilvitis et al. (2019) designed the primers used in this study to encompass the putative promoter region 726–1228 nucleotides upstream of the TLR4 transcription start site, which includes regulatory regions and CpG sites that affect expression (Table S3). Each PCR reaction contained 12.5 μl of 2× PCR Master Mix (Promega), 1 μl forward primer (10 μmol l−1), 1 μl reverse primer (10 μmol l−1), 8.5 μl of nuclease-free water and 2 μl of DNA; PCR was run on a T100 Thermal Cycler (Bio-Rad). Cycling conditions included an initial denaturation at 95°C for 2 min followed by 35 cycles at 94°C for 40 s, annealing at 62°C for 40 s and extension at 72°C for 150 s, and a final extension at 72°C for 5 min. PCR products were purified using ExoSAP-IT (Affymetrix), and Sanger sequencing using BigDye Terminator technology with forward primers was conducted at the Roy J. Carver Biotechnology Center, University of Illinois at Urbana-Champaign (Urbana, IL, USA), on an Applied Biosystems 3730xl DNA Analyzer. We did not map the sequenced region to chromosomes or align sequences; neither action was necessary as a BLAST search of our regions indicated no homology with other loci. The chromatograms from DNA sequences were then analyzed manually on Unipro UGENE (Okonechnikov et al., 2012). When sequencing was ineffective (i.e. regions of the target sequence could not be analyzed), samples were re-sequenced. This re-sequencing was rare, and ultimately high-quality sequences were obtained for all birds.
All CpG sites in the putative TLR4 promoter were quantified across all individuals, counting CpG sites on each homologous region (Hanson et al., 2021). In our study, across all birds, EP in TLR4 ranged from 6 to 10 CpG sites (EP=6–10; Fig. S3): 76.3% of birds either had EP=7 or EP=8, and the remaining 23.7% of birds had EP=6, EP=9 or EP=10 (Fig. S3). In previous work that also quantified EP in TLR4 in house sparrows from Tampa, we also found that 96.7% (30 out of 31 birds) of birds had either EP=7 or EP=8, and only one bird had EP=9 (Hanson et al., 2021). In this prior study, the binary form of EP was the best predictor of TLR4 expression (Hanson et al., 2021). In addition to treating EP as a continuous variable, we also assessed whether ‘CpG identity’ was related to TLR4 expression, asking whether the specific location of the CpG polymorphism(s) was associated with expression. However, EP as a binary variable was consistently the best predictor for TLR4 expression (Hanson et al., 2021).
Quantification of TLR4 expression
Whole gut tissue from each bird was left to thaw on a dissection board placed in a tray filled with ice. When thawed, each gut was opened along its length and the contents washed out with distilled water. The small intestine was then separated into three sections: proximal, medial and distal. From the middle portion of each section, we collected a transverse fragment (about 1 mm wide) and immediately placed it into a microtube on dry ice. We also collected a section from a cecal segment of the gut and processed it in the same way. All gut samples were then stored at −80°C until RNA extraction. RNA was extracted from each gut sample separately using a TRI-reagent extraction method; each extract was then diluted to 25 ng μl−1 (Hanson et al., 2021). From each extracted RNA sample, we measured TLR4 mRNA abundance using one step RT-qPCR. All RT-qPCR reactions (20 μl) were run in duplicate alongside non-template controls (NTC) and no reverse transcriptase controls (NRT) on a Rotor-Gene Q system (Qiagen). Each reaction contained 10 μl of iTaq Universal SYBR Green One-Step Kit (Bio-Rad), 0.3 μl of forward primer (10 μmol l−1), 0.3 μl of reverse primer (10 μmol l−1), 0.25 μl of reverse transcriptase (or 0.25 μl nuclease-free water for NRTs), 7.15 μl of nuclease-free water and 2 μl of diluted RNA (or 2 μl of nuclease-free water for NTCs).
Thermal cycling conditions were: 10 min at 50°C for the reverse transcription reaction, then 1 min at 95°C for polymerase activation and DNA denaturation, followed by 40 amplification cycles of 15 s at 95°C then 30 s at 60°C. Melt-curve analyses were performed from 65 to 95°C with 0.5°C increments every 3 s. A calibrator (i.e. a mix of RNA from a homogenate of the four different gut samples from four individuals) and an internal reference gene (hydroxymethylbilane synthase, HMBS) (Zimmer et al., 2021) were run on all plates to calculate mRNA abundance using the comparative Ct method (2ΔΔCt) (Livak and Schmittgen, 2001). TLR4 expression could not be estimated in four birds because of RT-qPCR failures.
Statistical analyses
All statistical analyses were conducted in R 4.1.2 (http://www.R-project.org/). Using the R packages ‘olsrr’ and ‘lmtest’, we visually inspected plots of residual versus fitted values and conducted Shapiro–Wilk and non-studentized Breusch–Pagan analyses. These tests indicated that the residuals of our three regression analyses (Tables 1, 2 and 3) were normally distributed and homoscedastic (Razali and Wah, 2011) (Table S1). Because we observed tissue-specific effects in our previous study (Hanson et al., 2021), we examined whether TLR4 expression was affected by gut region, EP in TLR4, or their interaction using a linear mixed model (LMM) (Table 3). In this model (Table 3), TLR4 expression was included as the dependent variable and gut region (i.e. proximal, medial, distal and cecal), EP (i.e. high or low) and their interaction were included as fixed effects.
Association between S. enterica resistance, EP and TLR4 expression
Our first goal was to investigate whether EP in TLR4 and/or TLR4 expression was related to resistance of S. enterica. Here, we quantified resistance as the ability of individuals to limit the absolute amount of and the increase in S. enterica shed in their feces over time. We used the lme4 package in R to conduct our first LMM with S. enterica burden as the dependent variable (Table 1). We included EP in TLR4 (treated as a binary term, i.e. high or low EP), TLR4 expression (averaged across four gut tissues, as EP in TLR4 effects on TLR4 expression were not dependent on gut region; see Results), and their interaction with day of sampling as fixed effects. This approach allowed us to simultaneously detect the effects of EP in TLR4 and TLR4 expression on absolute S. enterica burden (at each day of sampling) and on the change in S. enterica burden over time. In all our models, we also included body mass at capture and pre-existing Salmonella burden as fixed effects. In all models, we also included the status of the bird (male, female or juvenile) as a fixed effect, and bird ID as a random effect to account for within-individual differences.
Association between S. enterica tolerance, EP and TLR4 expression
Our second goal was to assess whether EP in TLR4 and/or TLR4 expression was associated with tolerance of S. enterica. Here, we quantified tolerance as the relationship between the body mass of a bird and its S. enterica burden over the course of the infection (an approach similar to that which we used for West Nile virus responses in this species; Burgan et al., 2018; Kernbach et al., 2019). We characterized individuals better able to maintain body mass while infected as more tolerant of the S. enterica infection. We fitted this second LMM with body mass as the dependent variable (Table 2). We included EP in TLR4, average TLR4 expression, and their interaction with S. enterica burden2 as fixed effects. We fitted a quadratic effect of burden (i.e. S. enterica burden2) as a fixed effect and an interaction term (instead of an untransformed value of S. enterica burden) because S. enterica effects on body mass appeared to be non-linear (results of the untransformed value of S. enterica burden are presented in Table S2). This approach allowed us to simultaneously detect potential effects of EP in TLR4 and TLR4 expression on body mass (at each day of sampling) and the rate of change in body mass with burden over time.
Association between house sparrow mortality, S. enterica infection, EP and TLR4 expression
In an additional analysis, we asked how EP was related to mortality by S. enterica. We used a multivariate Cox proportional hazard regression to assess whether the probability of death was associated with EP, average TLR4 expression across all tissue types in the gut, average S. enterica burden across the entire infection (this model was also re-run with maximum S. enterica burden with analogous results), the status of the bird (male, female or juvenile) and absolute body mass lost over the experiment (Table 4). The interaction between EP and S. enterica burden was also included, as this allowed us to detect potential effects of EP on mortality in relation to S. enterica burden. Additionally, the interaction between absolute body mass loss and S. enterica burden was included so that we could test whether mortality was associated with burden-related body mass changes.
RESULTS
The effects of EP in the TLR4 promoter on TLR4 expression were not dependent on gut region (Table 3). Therefore, to simplify our models, we included average TLR4 expression across all gut regions for each individual bird in our LMMs. Just over half (n=21) of the birds were infected with a detectable form of Salmonella sp. at the time of capture (i.e. prior to experimental infection; Fig. 1). However, pre-existing Salmonella infection was not related to EP in TLR4 (low EP: estimate: −0.29451, s.d.=0.258, t=−1.141, P=0.256), TLR4 expression (TLR4 expression: estimate: 0.139, s.d.=0.218, t=0.43400, P=0.529), resistance or burden in our study (Tables 1 and 2). Pre-existing infection was also not associated with bird body mass at capture (estimate=−0.169, t=0.331, P=0.492).
Association between S. enterica resistance, EP and TLR4 expression
Low EP in TLR4 was positively associated with a faster rate of S. enterica burden increase over time, indicated by the positive effect of the interaction between low EP and day of sampling on S. enterica burden (Table 1, Fig. 1). High EP birds were therefore more resistant than low EP birds. The status of the bird (male, female or juvenile), body mass at capture and pre-existing Salmonella burden were unrelated to S. enterica infection dynamics (Table 1). Bird ID explained 10.5% of variance in S. enterica burden whereas the majority (37.0%) of variance was explained by the fixed effects (Table 1). These patterns were analogous when treating EP as a continuous variable.
Association between S. enterica tolerance, EP and TLR4 expression
House sparrow body mass was affected by exceptionally high S. enterica burden, indicated by the negative quadratic effect of S. enterica bacterial burden on absolute mass loss (Table 2, Fig. 2). However, this effect was not dependent on EP in TLR4 (Table 2). Similarly, body mass also tended to decrease with increasing S. enterica burden when it was not treated as a quadratic effect (i.e. when non-quadratic values of S. enterica burden were used), but this effect was non-significant (Table S2). Body mass at capture had an effect on body mass change over the course of the experiment, with smaller birds tending to lose more mass than large birds (Table 2). Bird ID explained 27.69% of variance in body mass in the model; fixed effects explained 51.30% (Table 2).
Association between EP and TLR4 expression across the gut
TLR4 expression differed among gut regions; expression was higher in the cecum and distal regions than in the proximal and medial regions (Table 3). EP in the TLR4 promoter also affected TLR4 expression; sparrows with high EP expressed lower levels of TLR4 (mean=1.613, range=0.01–9.221, s.d.=1.445) than birds with low EP (mean=2.992, range=0.08–9.57, s.d.=2.546) (Fig. 3, Table 3), but this EP effect did not differ among gut regions (Table 3). Within-individual differences (i.e. bird ID) explained 36.23% of the variance in TLR4 expression whereas fixed effects explained 30.66% of the variance (Table 3).
Association between house sparrow mortality, S. enterica infection, EP and TLR4 expression
Of the 38 birds in our experiment, only six died. A multivariate Cox proportional hazard regression revealed that mortality was not related to EP in TLR4, average (or maximum) S. enterica burden or their interaction. Mortality was also not associated with absolute body mass or its interaction with S. enterica, TLR4 expression or the status of the bird (male, female or juvenile) (Table 4).
DISCUSSION
The central aim of our experiment was to investigate whether one form of EP, the number of CpG sites in the promoter region of TLR4, affected the capacity of house sparrows to resist, tolerate or survive a pathogenic Salmonella infection. We found that birds with high EP in TLR4 shed fewer bacteria than low EP birds. As bacteria shed in feces likely relates to the burden in the gut at or near the time of fecal sampling, high EP in TLR4 was thus related to higher host resistance to this pathogen. In showing that high EP is associated with host resistance, our results support the hypothesis that EP in TLR4 could be an important target for pathogen-driven selection. That said, tolerance and mortality were not associated with EP in TLR4 in our study; indeed, body mass was generally unaffected by S. enterica until the shed bacteria reached very high levels (>3 log) and mortality overall was very low for what was a pathogen to English house sparrows. As we did not measure methylation because our study design was not conducive to capturing the expected dynamic interplay among EP, methylation, gene expression and S. enterica resistance, we cannot be sure that EP operates in the manner our hypothesis portrays it to do. Nevertheless, our study provides empirical and conceptual support for the idea that one form of EP contributes to the efficacy of control of an important pathogen, a finding that supports our hypothesis about why EP is important in range expansions of this species (Hanson et al., 2021, 2022; Kilvitis et al., 2017). Below, we discuss the potential ecological and immunological ramifications of these results for this and other range-expanding animals.
TLR4 expression is a key element of the vertebrate immune response to S. enterica
Salmonella enterica serotype Typhimurium is a Gram-negative bacterial taxon that has a major impact on human, wildlife and livestock health (Mahmoud, 2012; Malik et al., 2021). It can infect and be transmitted by many host species, typically after exposure in food or water, and it is most often transmitted directly via excretion in feces but also via other routes (i.e. persistence in the soil or on surfaces for weeks to months) (Hilbert et al., 2012; Tizard, 2004). Significant to this study, the S. enterica serovar studied here has been responsible for large die-offs of wild birds in the past (Hughes et al., 2008, 2010). In rodents and chickens, many factors (Thaiss et al., 2016) can protect against bacterial infections including Salmonella, but the expression of TLR4 is among the most important (Coburn et al., 2007). Exactly which of the above mechanisms described in the Introduction (i.e. the types of leukocytes conferring protection against Salmonella, methylation of CpGs among and within those cell types, or both) was potentiated by high EP in TLR4 is unknown. Nevertheless, EP played an important role in the control of this pathogen. More work will be necessary to elucidate the complex molecular and cellular mechanisms whereby EP provides protection.
Could high EP defend against Salmonella via the dynamic regulation of TLR4?
Whereas our study showed that high EP in TLR4 was linked to high S. enterica resistance, it also showed that high EP in TLR4 was linked to low TLR4 expression in the gut tissue measured at the end of our experiment. This result conflicts with observations we have made in blood, but it partly resembles the patterns observed in spleen and liver (Hanson et al., 2021). The current results are also intriguing because high TLR4 expression is more commonly associated with bacterial resistance in chickens (Gou et al., 2012). However, our hypothesis did not predict that high EP in TLR4 would be protective because it imbues high, constitutive expression of TLR. Rather, we expected that it would facilitate TLR4 expression plasticity (perhaps including reversibility) by increasing the potential for DNA methylation modification. In other words, we predicted that high EP in TLR4 is associated with Salmonella resistance through a greater propensity to tune TLR4 expression (increase and decrease) over the course of an infection, rather than enduringly elevating it. Indeed, in the previous study on blood (Hanson et al., 2021), this higher reversibility is exactly what we observed and what we would expect here too had our study design allowed us to describe TLR4 expression dynamics (and associated methylation) in the gut and other relevant lymphoid tissues or cells. However, the heterogeneity of effects of EP across cell types suggests a new but not altogether incompatible mechanistic possibility for EP: EP might enable cell types to have different and perhaps more suitable levels of gene expression for pathogen control.
Presently, we favor the possibility that TLR4 expression reversibility is the adaptive/functional mechanism because the ability to flexibly regulate TLR4 expression should better balance the costs and benefits of an immune response against a pathogen. However, we were only able to measure TLR4 expression in the gut once, 14 days after the onset of an infection (because of the destructive nature of sampling the gut tissue), so we cannot really test this idea in this study. Indeed, further experimentation – for example, sampling the gut from different birds at different stages of an infection – will be critical to revealing the specific pathway by which EP may have affected S. enterica resistance. Such a study could also help elucidate the specific manner by which methylation mediates the effect of EP on Salmonella resistance. As many types of leukocytes will be integral to successful control of the bacteria, it will be important to consider both the timing post-infection and cell/tissue type in any measurements of methylation. It will be useful to test directly whether EP enables a given cell to express an optimal amount of TLR4 as it encounters or receives signals about Salmonella from other host cell types and/or if cell types vary in their propensity to upregulate and downregulate gene expression via DNA methylation.
Ecological implications of EP in TLR4
Our results indicate that EP in TLR4 protects Tampa house sparrows from S. enterica by enhancing resistance, but not to the point of sterilizing the gut of this pathogen. High levels of EP among range-expanding/non-native house sparrows (Hanson et al., 2021) may also afford certain sparrow populations with an indirect means to outcompete resident host species, exposing their competitors to microbes that cause no pathogenic effects to themselves but potentially harm these other host species (Coon and Martin, 2014; Martin et al., 2010, 2014, 2017, 2015). Indeed, house sparrows have already been implicated as potential progenitors of Salmonella epidemics that have drastically reduced the population sizes of other birds (e.g. greenfinches) (Hernandez et al., 2016; Tizard, 2004). Somewhat surprisingly, too, most sparrows in this study were able to tolerate and survive infection with the same microbe that was lethal to other songbirds including European house sparrows. Perhaps this outcome is a vestige of past selection for TLR4 EP in North American birds, which occurred at the time of their introduction from ancestral Europe.
Conclusions and further work
Besides investigating EP in other genes and other invading or range-expanding species, it would be beneficial to resolve how TLR4 expression and symbiosis in the gut are interrelated. For example, gut microbiota-derived metabolites (e.g. short chain fatty acids) can modulate gut transcriptional output by affecting CpG methylation in TLR4 (Takahashi et al., 2011). High EP in TLR4 could thus potentially facilitate microbiota-induced epigenetic changes in sparrows by providing more genomic substrate for modification of methylation profiles. Host cells largely regulate their sensitivity to commensal microbes via DNA methylation and its subsequent effects on defensive gene expression (Thaiss et al., 2016). As host intestinal epithelial cells (IECs) form a physical barrier with, sense signals from, and secrete peptides directed at microbes (Alenghat and Artis, 2014), the potential epigenetic mechanisms whereby IECs and other host cells cope with resident Gram-negative bacteria might have been exploited by some individual birds to affect how they combat gut pathogens (Takahashi et al., 2011). Although the mechanisms whereby EP fostered resistance and the selective value of EP remain unresolved, here we have shown that EP in TLR4 was positively associated with Salmonella resistance in the house sparrow. Our findings support the hypothesis that variation in EP could be adaptive for hosts encountering novel and dynamic pathogen risk scenarios (Hanson et al., 2022, 2021), but much more mechanistic work, especially focused on methylation among various lymphoid tissues, is crucial.
Acknowledgements
We thank the Martin and Schrey labs and Jeb Owen and Olivia Smith for constructive feedback on the results and manuscript. We thank the staff of Roy J. Carver Biotechnology Center, University of Illinois at Urbana-Champaign (Urbana, IL, USA) for sample sequencing.
Footnotes
Author contributions
Conceptualization: C.Z., H.H., B.K., L.B.M.; Methodology: C.Z., H.H., B.K., L.B.M.; Validation: C.Z., H.H., B.K., L.B.M.; Formal analysis: E.S.; Investigation: C.Z., H.H., B.K., A.S., D.R., P.W., A.L.W., L.B.M.; Resources: P.W., L.B.M.; Data curation: E.S., C.Z.; Writing - original draft: E.S., C.Z.; Writing - review & editing: E.S., C.Z., H.H., A.S., D.R., P.W., A.L.W., L.B.M.; Visualization: E.S.; Supervision: C.Z., L.B.M.; Project administration: C.Z., L.B.M.; Funding acquisition: L.B.M.
Funding
L.B.M. recognizes the National Science Foundation (NSF-IOS grant 2027040) for support.
Data availability
Data are available from figshare at https://doi.org/10.6084/m9.figshare.23596692.
References
Competing interests
The authors declare no competing or financial interests.