Responding to environmental challenges can have both positive and negative effects on fitness, but the mechanisms and/or biomarkers underlying these links are not well understood. Telomeres, repetitive sections of DNA that form protective caps at chromosome ends, which enhance genome integrity and often predict organismal lifespan, may be important in this context. Some evidence suggests that stress exposure shortens telomeres, but most studies are correlational and/or conducted during early life; thus, how stress exposure influences telomeres in adulthood is not well known. Here we experimentally exposed wild-captured adult house sparrows (Passer domesticus) to a series of standardized protocols of repeated, rotating stressors (including threats, unpredictable food availability, and immune challenges) designed to increase stress levels. Although there were some effects of the treatment on glucocorticoid levels and mass dynamics, there were no effects on telomeres during any of the 6-11 month studies. Taken together, these results suggest that telomeres are robust or resilient to a range of stressors in adulthood. The effects of stress exposures on telomeres are likely to be highly context-dependent and vary depending on the timing, intensity, and duration of the stress exposures as well as an individual’s life stage.

We completed 4 experiments in sequence, which reflected a combination of three agendas. First, space limited sample size for any one experiment, so replicates were needed. The first experiment was conducted inside, whereas the 2nd-4th were conducted in outdoor aviaries in small flocks, which for house sparrows is a more natural context than indoors, and so we thought our treatments would better reflect natural, albeit elevated challenges. Finally, we were also interested in assessing the potential for different types of stressors to have different effects; hence, we varied the potential threats within the first 3 experiments and shifted to different types of stressors in the last experiment (below).

Captive housing conditions

The first experiment was conducted in an indoor aviary using local male house sparrows captured from Fargo, North Dakota, and three others were done in an outdoor aviary facility using local house sparrows of both sexes captured from Lexington, KY. In North Dakota, mist nets were used to capture 30 adult male house sparrows in early 2017. Males were individually housed in wire cages (23” wide x 16” deep x 16” tall) on racks in two rooms at North Dakota State University. All cages were visually and acoustically open to the other birds in the room, and the male cage racks were positioned about 3 feet apart with a centrally located cage stand housing 4 females, which were visible to all males. Food (Kaytee supreme mixed seed for finches), water, and cuttlebones were provided ad libitium, and supplemental food (broccoli, baby spinach, peas and carrots, or Quicko Classic Egg Food) was provided once per week. The bird rooms were kept on a short-day cycle of 8h light: 16 h dark for 27 weeks. After the period of stress exposure, the rooms were shifted to 16 h light: 8 h dark.

In Kentucky, we captured free-living house sparrows using mist nets at several locations between October and December in 2017, 2018, and 2019. Subject birds were then released into aviaries in groups of 4 consisting of 2 males and 2 females. We provided each aviary with a retired Christmas tree for cover with the quality of cover (more needles on the tree) higher in control aviaries than experimentals, and ad libitum food (white millet and chicken starter mix), water, and sand grit. Assignment of birds to aviaries was based on filling aviaries in order by their identity number (matched by eventual treatment) as subjects of each sex were caught producing a random assignment by body size and capture date between treatments.

Details of stress protocols

The experiments were modeled on previous research. For the first three, we were informed by prior work using a standardized series of rotating stressors designed to induce chronic stress in house sparrows and other songbird species. They differed in the exact rotating stressors employed, their schedules, and their duration (Figure 1 of published paper). In the fourth experiment, we shifted to repeatedly stimulating the immune system and to manipulate the predictability of food. In total, these 4 experiments served as replicates of generalized stress exposure, but because they were conducted across some variation in context, we could potentially assess a broader array of stressor intensity and type than could be accomplished in any one experiment. Our specific protocols for each are described in detail below and summarized in Figure 1 of the published paper.

Experiment 1 - We assigned North Dakota birds randomly to either experimental or control treatment upon capture (n = 15 males per treatment). One week after the last bird was captured, we began the stress protocol to all birds in one room. We exposed subjects to one of five stressors daily, applied in a randomly generated order and at a random time each day between the hours of 10am and 10pm. Each stressor was applied for 30 minutes. At the same time as the initiation of the stressor each day, the control room received a nontreatment protocol of opening and immediately closing the door without entry. The five experimental stressors were selected to mimic stressful events that house sparrows might be exposed to in the wild: a live avian predator (common kestrel, Falco tinnunculus), a live feline predator (domestic cat, Felis catus), a mounted avian predator (merlin, Falco columbarius), loud radio (to represent anthropogenic acoustic disturbance), and a human rattling and shaking of the cage racks. We inserted the live predators into the stress order twice, such that the birds received 7 stress events per cycle per week with the two live predators represented twice. This protocol was repeated for 27 weeks.

Experiment 2 – Subjects were captured between 26 September 2017 – 5 December 2017. Birds were assigned treatment by aviary, and aviaries were assigned treatments based on aviary location. Aviaries were arrayed around four separate central chambers accessed through a door adjacent to 1 or 2 of the aviaries. Birds were given food and water from inside the chamber, so this meant the aviaries adjacent to the door were disturbed more than other aviaries each time food and water were provided. These aviaries were assigned to the elevated stress treatment. Aviaries on the back side of each chamber were assigned the control treatment (no stressors), and a few other less disturbed aviaries were randomly assigned with the goal of generating equal sample sizes. We began stress treatments on 5 December 2017. Each aviary was stressed twice per day at a random time in the morning and in the afternoon for 30 min. We chose among six stressor types: a human moving around just outside the aviary, an owl model with moveable head sitting on a platform inside the aviary, a model cat with moveable head and tail on the floor of the aviary, a rubber snake (1 of 2 types) draped realistically in the refuge tree in the aviary, one of 10 different novel objects (determined randomly) placed in the food dish, and a radio-operated streamer placed near the roost tree that periodically jerked back and forth. We assigned stressors in a mixed procedure that included random assignment but with biases toward balancing the number of times we presented each stressor and some constraints (such as the proximity of two aviaries to each other). We employed the stress protocol for 6 days each week until March 30, 2018 for a total of 16.5 weeks.

Experiment 3 – New subjects were captured starting on 18 September 2017 and continued until 11 October 2018. They were assigned to aviaries as in 2017. Treatments were assigned to aviaries similarly as well. We began the experimental treatment on 21 October 2018. The protocol was similar to the 2017 experiment with a few key differences. First, we stressed birds 3 times daily at random times within three 3-hour blocks throughout the day. We also added an additional stressor; a taxidermy mount of a Cooper’s hawk (Accipiter cooperii) placed on a platform inside the aviary. We also added a capture-restraint-sampling protocol, experienced only by the experimental group, twice (on 16-17 November and 20-21 January) in the middle of the experiment. The stressor schedule continued until 26 March 2019 (22 weeks) when all birds were captured again and sampled.

Experiment 4– A third set of Kentucky subjects were captured from 20 October to 29 November 2019. Birds were measured, sampled, and assigned to aviaries as in the previous two years. In this replicate, we altered the nature of the stressors. We divided aviaries into three groups, with one group assigned as control and treated as in the other two years (and in similar aviary positions given the disturbance of providing food and water at the other aviaries). Both other groups were injected intradermally with 0.1 mL of PBS containing 25 ug bacterial lipopolysaccharide (LPS) on 4 occasions, approximately every 3-4 weeks between January 13, 2020 to the end of March 2020. At the end of February, we also held experimental subjects in both groups for 30 min and collected blood. One experimental group was additionally treated to a 6 h food deprivation regime starting 2 days after each injection, starting at a random time in the morning and repeated 3 times in 5 days. Each time birds were caught we also recorded their mass.

In experiments 2-4, subjects were kept after the end of the stress protocol in aviaries under minimal disturbance so that they would breed. This included supplementing their diet with additional protein and sources of calcium and providing nesting material. Some birds did breed and those were provided with insect food to raise their offspring. Non-breeders, regardless of treatment, were removed to other aviaries either in hopes they would breed or to keep until release in August of each year. At the end of the summer, all remaining subjects were blood sampled and weighed again and released back into the wild.

Sampling

Timing of sampling for all four experiments is summarized in Figure 1 of the published paper. In experiment 1, we measured mass and took blood samples from all birds before the experimental protocol began on 19 April 2017, and then again at the end of the experiment on 9 October 2017. Birds in control and stress treatments were also weighed on 3 May and 7 June 2017. To measure if stressors elevated corticosterone as intended, birds on the stress protocol were also captured, weighed, and blood sampled on 3 May and 7 June after 30 min exposures to the live cat, again on 25 September when they were sampled at baseline and after a 30 min restraint. On 9 October the treated birds were exposed to the live kestrel mount for 30 min, blood sampled and sperm collected through cloacal massage. All birds were then euthanized and tissue from heart, liver, and pectoralis muscle collected. At each blood sampling event approximately 75 μL of whole blood was collected via brachial venipuncture into heparinized micro-hematocrit tubes. The blood was stored on ice for 5-120 min, centrifuged, separated into plasma and red blood cell fractions, and frozen at -80 °C until measurement.

In experiments 2-4, we collected blood samples from the bird’s brachial vein in heparinized microcapillary tubes from all subjects upon initial capture, at the end of treatment, and upon release. In experiment 2, 5 birds were sampled 4 times in two days in the period between the treatment ending and being released as part of another experiment; we used the first sample from that series in the present analysis. In experiment 3, we also took samples during the capture-restraint protocol. Not all birds in experiments 2-4 were sampled at all time points due to either death, escape or failed DNA extraction. In experiment 3, a set of excess birds were sampled once but not used as subjects. Details on sample sizes in the analysis are reported in Table S1 linked to the published paper.

Telomere measures

DNA was extracted from red blood cells and organs using commercially available extraction kits (Macherey-Nagel NucleoSpin Blood #740951, Macherey-Nagel NucleoSpin Tissue #740952) and following the manufacturer’s protocol. DNA was extracted from whole diluted semen samples using commercially available extraction kits with modifications. After extraction, DNA concentration and purity were assessed with a Nanodrop 8000 spectrophotometer (Thermo Scientific, Waltham, MA, USA) and DNA quality was verified by electrophoresis on a 2% agarose gel.

Relative telomere length was measured using quantitative PCR on a Strategene Mx3000P (Stratagene, San Diego, CA, USA). The relative telomere length of samples was calculated as the ratio (T/S) between telomere repeat copy number (T) and single copy control gene number (S) relative to the reference sample. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was used as the single-copy control gene. The telomere forward and reverse primers used were: Tel 1B (5′-CGGTTTGTTTGGGTTTGGGTTTGGGTTTGGGTTTGGGTT-3′), Tel2b (5′-GGCTTGCCTTACCCTTACCCTTACCCTTACCCTTACCCT-3′). Zebra finch GAPDH primers were used to amplify the control gene: GAPDH-F (5’-AACCAGCCAAGTACGATGACAT-3’), GAPDH-R (5’CCATCAGCAGCAGCCTTCA-3’). The telomere and GAPDH reactions were run in duplicate, and each reaction contained 20 ng of DNA, telomere or GAPDH primers at a concentration of 200 nM forward/200 nM reverse, and 12.5 ml of perfeCTa SYBR green supermix Low ROX (Quantabio) for a total volume of 25 ml. The thermal profiles for the telomere reactions were 10 min at 95°C, followed by 27 cycles of 15 s at 95°C, 30 s at 58°C, and 30 s at 72°C, finishing with 1 min at 95°C, 30 s at 58°C, and 30 s at 95°C. The thermal profiles for the GAPDH reactions were 10 min at 95°C, followed by 40 cycles of 30 s at 60°C, finishing with 1 min at 95°C, 30 s at 55°C, and 30 s at 95°C.

The number of PCR cycles (Ct) required for the products to accumulate enough fluorescence to cross a threshold was determined. Samples from a given experiment were randomized across plates but balanced with respect to treatment. Each plate also included a serially diluted 5 point standard curve (40, 20, 10, 5, 2.5 ng) that was run in triplicate created from a pooled house sparrow reference sample to measure reaction efficiencies. The average reaction efficiencies were 93.30 ± 0.97 (mean ± s.e.m.) for telomere plates and 98.59 ± 0.86 (mean ± s.e.) for GAPDH plates. The 20 ng point from the standard curve was used to as the reference sample to calculate relative telomere lengths for the unknown samples. Relative telomere lengths 2^-ΔΔC^, where ΔΔCt = (Ct _telomere – Ct _GAPDH) focal sample - (Ct _telomere – Ct _GAPDH) reference sample. We previously established that the repeatability of the T/S ratio is high (0.86, p < 0.001, 95% CI[0.72, 0.93) by running samples from 28 adult house sparrows in random well locations across two plates.

Corticosterone measures

To measure corticosterone levels in our samples, we used a commercially available CORT enzyme immunoassay kit (Enzo Life Sciences ADI-901-097) following a protocol that was previously validated in house sparrows. Briefly, we added 5 mL of 10% steroid displacement buffer (SDB) to our 5 mL of plasma samples. After 5 minutes, we then added 240 mL of Assay Buffer 15 to each sample, thoroughly vortexed and then aliquoted 100 mL duplicates into a 96 well plate. A standard curve ranging from 32 – 200,000 pg/ml was aliquoted in triplicate on each plate. All wells were incubated with conjugated CORT and antibody on a shaker; wells were then washed, and substrate was added before a second incubation. Stop solution was added, and the plate was read at 405 nm (corrected at 580 nm per manufacturer’s recommendation). Individuals were randomized and treatments were balanced across plates. We ran a total of 4 plates and the inter- and intra-assay variation were 8.07% and 2.43% respectively.

Statistical Analysis

We constructed two datasets. The first contained telomere measures from the five tissues at the end of the North Dakota experiment. The second dataset contained the repeated measures of blood sample telomeres, corticosterone, and mass from all four experiments. This included 60 telomere measures from 30 subjects in experiment 1, 153 measures of telomeres from 54 subjects in experiment 2, 206 measures from 64 subjects in experiment 3, and 173 measures from 58 subjects in experiment 4 (see Table S1 linked to published paper for the distribution of subjects with different numbers of measures). Also included in this dataset was the elapsed time since the start of the experiment in fraction of a year, the bird’s mass at each capture, and the treatment they experienced. We subset this dataset to analyze individual experiments but used the whole dataset for some measures of power and to facilitate generating some figures.

All univariate analyses involved using a mixed model in the R package lme4 with subject identity as a random effect. Corticosterone measures following threat presentations were natural log transformed and compared with baseline measures within a mixed model, and we also compared the corticosterone following the two cat presentations separated by 35 days to assess habituation. We calculated a separate raw repeatability for telomeres using the R program rptR for each experiment as the proportion of total variance in telomere length that was among-subjects. For experiment 1, we calculated repeatability for the two blood telomeres measures alone and also across all 6 tissue types. We adjusted this repeatability by including assay unit (the pair of plates containing a subset of samples) in all experiments as a fixed effect, which removed all potential variance due to methodological differences between assays. We retained this base model structure in all subsequent analyses. We then investigated the impact of elapsed time from initial capture on telomere length regardless of treatment; we treated this variable as a fixed effect to obtain the mean change in telomeres over time, and for experiments 2-4, included a subject by time random effect to estimate the variance in the effect of time across subjects. We then added treatment and a treatment by time interaction to assess the impact of treatment on telomere dynamics. Similar models were constructed for mass, except we also included a quadratic of time for all experiments except experiment 2 which did not have enough repeated observations of mass.

We assessed power to detect different effect sizes reported in the literature which we could standardize and, ideally assess as a rate per year. For instance, one result (Hau et al. 2015), obtained after 1 year, is reported only in a figure, so we measured the observed effect size from the published figure, standardized it by the estimated standard deviation, calculated from standard errors given in the figure and sample sizes given in original publication. We then assessed our ability to detect an effect that size on a standardized scale. For our data, we obtained standardized effect sizes (by standardizing telomere measures and rerunning relevant models) and then retrieved standard deviations for the slope of telomere length by time for controls in each experiment separately and overall. We used an online calculator for power (statskingdom.com) of a two-sample t-test with the observed standard deviation and sample sizes of our control and experimental treatments. We also used values from the meta-analysis of Chatelain et al. (2020) which were already in standardized units.

References

Chatelain M., Drobniak S.M., Szulkin M. 2020 The association between stressors and telomeres in non-human vertebrates: a meta-analysis. Ecology Letters 23(2), 381-398. (doi:10.1111/ele.13426).

Hau M., Haussmann M.F., Greives T.J., Matlack C., Costantini D., Quetting M., Adelman J.S., Miranda A.C., Partecke J. 2015 Repeated stressors in adulthood increase the rate of biological ageing. Frontiers in Zoology 12. (doi:10.1186/s12983-015-0095-z)