Epidermal cell death in frogs with chytridiomycosis

Study organisms

Litoria verreauxii alpina that were excess to a reintroduction trial (a total of 300 from two different populations) were obtained from Taronga Zoo, Sydney, and were part of a larger experiment (Brannelly et al., 2016b; Brannelly et al., 2016c). Litoria v. alpina are a declining anuran endemic to the Australian Alps in New South Wales and Victoria, Australia. The species is highly susceptible to Bd, which is the primary cause of decline (Bataille et al., 2015; Brannelly et al., 2015b; Brannelly et al., 2016a). The animals sexually mature at 2 years and these animals ranged from two to three years old and were captive raised under strict quarantine protocols and had never been exposed to Bd. Animals were housed individually in 300 × 195 × 205 mm terrarium with gravel substrate, at a room temperature of 18−20 °C. They were fed ad libitum three times weekly with juvenile (10 mm) crickets (Acheta domestica) (dusted with amphibian vitamins and gut-loaded). Animals were misted twice daily for 60 s with reverse osmosis water. Temperature and humidity were monitored daily.

Pseudophryne corroboree were excess to a breeding program at the Amphibian Research Centre, Pearcedale, Victoria, Australia (a total of 120 adult animals from five different populations). These animals were 5–8 years old and part of a larger research experiment (Brannelly et al., 2016b; Brannelly et al., 2016c). Pseudophryne corroboree are functionally extinct in the wild, and highly susceptible to Bd (Brannelly et al., 2015a; Brannelly, Skerratt & Berger, 2015). These animals were housed in the same conditions as above but on a paper towel substrate that was changed once a fortnight.

Inoculation

Animals were allowed to acclimate to their new environment for at least seven days. All animals were tested for Bd infection prior to the start of the experiment (see ‘Testing for Bd’ below) and all were found to be negative. Cultures of Bd were harvested from tryptone, gelatin hydrolysate, lactose (TGhL) agar plates after incubation at 23 °C for 5 days. Three millilitres of artificial spring water was poured onto the plates and incubated for 10 min to allow zoospores to be released from zoosporangia to create the inoculum. Inoculum was poured off the plates and zoospores were counted using a hemocytometer and then diluted with artificial spring water to achieve the appropriate concentration as described below Samples sizes for each inoculation found in Table 1. Note that the inoculation strain used for obtaining infected samples for the TUNEL assay in L. v. alpina differs from the others, and was used to obtain higher infection rates

Data collection

Each week animals were swabbed for Bd infection (see below), weighed to the nearest 0.01 g with a digital scale, and measured snout to venter (SVL) to the nearest 0.1 mm with dial callipers. Animals were euthanized with an overdose of MS-222 when severe clinical signs of chytridiomycosis were displayed (irregular skin slough, leg redness, inappetence, lethargy, loss of righting reflex) and animals appeared moribund in accordance with animal ethics. Clinical signs of disease were seen by L. v. alpina from days 52–70 and by P. corroboree from days 45–83. The experiment ended on day 90, when all control and remaining exposed animals were euthanized.

Testing for Bd

We tested for Bd infection by using qPCR on skin swabs (Boyle et al., 2004). We used a standard protocol that involved 45 strokes per animal with a sterile rayon-tipped swab (MW-113; Medical Wire & Equipment), five on the middle of the venter, five on each side of the venter, five on each thigh, and five on the ventral surface of each hand and foot. The swab was gently rotated during and between strokes to ensure the greatest amount of DNA was gathered on the swab. Genomic DNA was extracted from the swabs using the Prepman Ultra kit and 2 min of bead beating to break apart the fungal cell walls. The extract was analysed using quantitative PCR following Boyle et al. (2004), in singlicate (Kriger, Hero & Ashton, 2006; Skerratt et al., 2011; Brannelly et al., 2015b) with a positive and negative control, and a series of dilution standards (made in house from a local Australian Bd isolate) in order to determine zoospore equivalents (ZE) per sample.

TUNEL assay

Upon euthanasia, a subset of L. v. alpina animals (n = 2 control and n = 2 exposed), and P. corroboree (n = 5 Bd− control, n = 3 Early Bd+ animals with light Bd infection, and n = 10 Late Bd+ animals at morbidity) (Table 1) were dissected for skin samples (one sample each from the dorsum, venter and thigh, although no thigh skin was taken for the P. corroboree Early Bd+ animals because that skin was used in a transcriptomics project). Skin was fixed in 4% phosphate buffered formaldehyde for 2 h, and tissues were transferred to 80% ethanol prior to embedding in paraffin wax for histological preparation. Routine histological techniques were used to prepare the tissues for light microscopy following standard methods (Woods & Ellis, 1994). Tissues were dehydrated in a graded series of ethanol, cleared with xylene, and embedded in paraffin (all three skin samples in one block). Tissues were serially sectioned at 5 µm, affixed to hydrophilic glass slides with four serial histosections per slide. Three slides were made per tissue section from the three different areas of the skin per animal and stained in the following order: the first slide was stained with hematoxylin followed by eosin counterstaining (H&E). The next two slides were processed with TUNEL assay following manufacturer’s instructions (ApopTag^® Red In Situ Apoptosis Detection Kit, Merck Millipore), followed with a DAPI counterstain. The first of the slides used in TUNEL were the assay slides (where we assessed number of apoptotic cells) and the second slide was a quality control test (with two histosections used as a positive control and the last two as a negative control). For the TUNEL assay slides, cells were counted using fluorescent microscopy at 200× magnification under DAPI fluorescence, then the same section counted for TUNEL positive staining, indicating apoptotic cells. At least 100 cells per animal were counted per skin section. To reach 100 cells, all cells were counted within a field of view. If a field of view contained less than 100 cells, another field of view was selected and counted entirely. The H&E stained histosections were used to ensure the site of infection were the areas used in the TUNEL assay to count apoptotic cells.

Caspase 3/7 assay

Frozen toe clip samples were extracted in 100 µL Buffer A (25 mM HEPES pH 7, 5 mM MgCl₂) with two 3.2 mm stainless steel beads in a 1.5 mL microtube. Samples were lysed by four cycles of 1 min bead beating followed by 3 min on ice. Samples were then centrifuged at 4 °C for 5 min at 12,000 × g. Supernatant was collected and used in the assay. As toe samples were very small, reaction volumes were kept to a minimum. Protein concentration of each toe was determined to standardise the sample sizes for the caspase assay. Concentration was quantified using the Bradford assay, with 10 µL Coomassie Bradford reagent (Pierce) and 10 µL of protein extract mixed and incubated at room temperature for 2 min. The samples and BSA standards were run in duplicate or triplicate in a 384 well plate, and the absorbance was measured at 595 nm (POLARstar Omega; BMG Labtech). Caspase 3/7 assay (Caspase Glo^® 3/7; Promega) was performed in triplicate in a 384 well plate with 10 µL Caspase Glo reagent and 10 µL protein extract. After mixing, the reactions were incubated in the dark at room temperature for 30 min, after which luminescence was measured (POLARstar Omega; BMG Labtech) (caspase activity). All samples of animals that had severe clinical signs of chytridiomycosis (n = 4), all control samples (n = 8), and a randomised subset of animals that cleared Bd infection (n = 11) were analysed. Only samples through week 7 post inoculation were run, which is the last sampling point where all animals with severe infection were alive, in order to ensure statistical validity of the Bd+ animals.

Statistical analysis

Animal ethics

Animal ethics was approved by James Cook University in applications A1897 and A2171 for L. v. alpina and A1875 for P. corroboree.

Bd Infection

Infection progressed as expected for the TUNEL assay individuals: all P. corroboree animals inoculated with Bd developed clinical disease, except for the three individuals euthanized at day 21 and represented as Early Bd+. Animals with clinical disease (Late Bd+) survived between 22 and 83 days post inoculation (average of 59.6 days) and their average infection load at date of death was a near to 1,000 fold higher infection load (793.3 time increase in ZE) than the infection load of animals with light infection loads (Early Bd+), which were euthanized on day 21 post inoculation (Late Bd+ = 5.19 Log₁₀(ZE) ± 0.34; Early Bd+ = 2.29 Log₁₀(ZE) ± 0.89; d = 3.04). In Bd inoculated L. v alpina animals (n = 6) all developed clinical chytridiomycosis and survived 39–63 days post inoculation (average of 52.7 days) and their average infection load at date of death was 3.57 Log₁₀(ZE) ± 0.75.

Of the 27 L. v. alpina inoculated in the caspase infection experiment, all became infected, but only four developed severe chytridiomycosis. The first animal died on day 58 after inoculation and the last on day 71 (average of 63 days). All animals that did not develop severe chytridiomycosis had cleared infection by week 12 post-inoculation. The factors that influenced infection load were week, infection status and week*status (linear mixed effects model: week, F₁₁ = 5.425, p < 0.01; infection status, F₁ = 23.763, p < 0.01; week*status, F₉ = 3.071, p < 0.01) (Fig. 1). Litoria v. alpina animals that developed severe chytridiomycosis had, on average of all time points, a near 200× higher infection load than animals that cleared infection (194 time increase in ZE) (Died = 3.09 ± 1.22 Log₁₀(ZE); Cleared = 0.80 ± 1.28 Log₁₀(ZE)) (Cohen’s d statistic = 1.30).

TUNEL assay

There was more cell death in infected animals compared with uninfected animals in both L. v. alpina and P. corroboree. The location of TUNEL positive cells in situ differed in Bd+ and control animals of both species. In control animals, low levels of background TUNEL positive cells were evenly distributed throughout the epidermal and dermal layers of the skin (Fig. 2A), but in the Bd+ animals, the TUNEL positive cells appeared more frequently in the epidermis (Fig. 2B). On microscopic examination of the H&E stained sections, clumped Bd sporangia were observed to be scattered through the ventral and thigh areas of skin, but none were seen in the dorsum of either species (Late Bd+), or in the P. corroboree with light infections (Early Bd+). TUNEL positive cells appeared to be more concentrated at infection foci (Fig. 2C), but were also more widespread over the epidermis and in deeper epidermal layers than Bd.

In P. corroboree, all three skin types showed an increase in TUNEL positive cells when infected with Bd in both early and late infection (Fig. 3A). In the thigh skin, Late Bd+ animals had 12.01 (95% CI [4.92–26.30]; Odds Ratio: Z = 5.46, p < 0.01) times more TUNEL positive cells than control animals (Pearson’s chi-squared: ${\chi }_1^2=44.30$, p < 0.01). In the venter skin, Late Bd+ animals have 22.31 (95% CI [5.25–94.82]) times more TUNEL positive cells than control animals (Odds Ratio: Z = 4.21, p < 0.01) and 2.16 (95% CI [1.15–4.03]) times more TUNEL positive cells than Early Bd+ animals (Odds Ratio: Z = 4.21, p < 0.01). The Early Bd+ animals had 10.33 (95% CI [2.37–45.067]) times more TUNEL positive cells than control animals in the venter skin (Odds Ratio: Z = 3.11, p < 0.01; Pearson’s chi-squared: ${\chi }_2^2=33.45$, p < 0.01). In the dorsal skin, the Late Bd+ animals had 14.38 (95% CI [3.32–62.24]) times more TUNEL positive cells than control animals (Odds Ratio: Z = 3.57, p < 0.01) and Early Bd+ animals had 19.88 (95% CI [4.67–84.20]) times more TUNEL positive cells than control animals (Odds Ratio: Z = 4.05, p < 0.01; Pearson’s chi-squared: ${\chi }_2^2=29.45$, p < 0.01) but there was no difference observed in TUNEL positive cells of the dorsum in Early and Late Bd+ animals (Fig. 3A).

Due to the small sample sizes for infected and control L. v. alpina (n = 2 for each group), only trends could be determined. But there was a higher proportion of TUNEL positive cells in the venter and the thigh skin with no observable difference in the dorsal skin (Fig. 3B) (Pearson’s chi-squared: Venter skin, ${\chi }_1^2=5.38$, p = 0.02; Thigh skin, ${\chi }_1^2=9.198$, p < 0.01, Dorsal skin: ${\chi }_1^2=1.694$, p = 0.19).

Caspase 3/7 assay

Caspase activity was positively correlated with infection load in inoculated L. v. alpina animals (Pearson’s correlation: R₆₄ = 0.463, p < 0.01; Linear Regression: F_1,62 = 16.943, p < 0.01) (Fig. 4). However, there was no overall difference between the animals that cleared infection (Bd cleared) and those that developed severe chytridiomycosis (Bd+) (GLM: F₁ = 0.079, p = 0.685), or between weeks (GLM: F₄ = 0.226, p = 0.717).

There was a difference in total caspase activity over time among the three groups (control, Bd + and Bd cleared). The three groups differed over week and week*disease status (Linear mixed effects model: week, F₄ = 11.974, p < 0.01; week*status, F₈ = 2.139, p = 0.037). There was no effect of week on the control group (Mixed model: F₄ = 2.463, p = 0.069) (Fig. 5A). At week 3, there was 48.36% less log₁₀ caspase activity in the Bd+ compared with the control animals, and 41.63% less activity in the Bd cleared animals compared with the control animals (ANOVA: F_2,18 = 5.512, p = 0.014; Bonferroni Post-Hoc: control v Bd+, p = 0.046, d = 1.408; control v Bd cleared, p = 0.028, d = 0.923).

When investigating the change in caspase activity each week over the first seven weeks post inoculation, there was a difference among the three groups, with week and week*disease status as important factors (Mixed Model: week, F₃ = 5.764, p < 0.01; week*status, F₆ = 3.044, p = 0.01), but there was no effect of week on the control group (Mixed model: F₃ = 20.004, p = 0.371) (Fig. 5B). The change in caspase activity between weeks 3 and 5 differed significantly among the three groups, with the Bd+ animals increasing in caspase 3/7 activity 15.35 times the change in control animals, and 2.162 times the change in Bd cleared animals (ANOVA: F_2,25 = 10.65, p < 0.01; Bonferroni Post-Hoc: control v Bd+, p < 0.01, d = 2.519; Bd+ v Bd cleared, p < 0.01, d = 1.241).