Seasonal variation in harbour seal (Phoca vitulina) blubber cortisol - A novel indicator of physiological state?

Cortisol is one of the main glucocorticoid hormones involved in both the mammalian stress response, and in fat metabolism and energy regulation, making it of increasing interest as a biomarker for stress, health and overall physiological state. However, transient stress responses to animal handling and sampling may be important sources of measurement artefact when investigating circulating concentrations of this hormone in wildlife. Here, cortisol concentrations were measured in the plasma and, for the first time, in the blubber of live captured adult harbour seals (Phoca vitulina). Plasma cortisol concentrations were positively correlated with capture time, suggesting that they were largely driven by a stress response to the capture event. In contrast, blubber cortisol concentrations were shown not to be significantly affected by capture time and varied significantly by sex and by season, with higher concentrations during natural fasting periods of their life cycle, particularly during the moult. These results suggest that cortisol may play a key role in increased fat metabolism during highly energetically demanding periods, and that blubber concentrations have the potential to be used as physiological state indicators in phocid seals.


Cortisol extraction from blubber biopsies
Briefly, the blubber biopsies were accurately weighed and homogenized in 1000µl ethanol.
Homogenates were then centrifuged at 3,000 rcf for 10 minutes and the supernatants collected. These were then evaporated under compressed air while incubating at 25 0 C. Two milliliters of ethanol : acetone (4:1) were added to the residue and after vortexing and centrifugation, the solution was again evaporated to leave a new residue. 1ml diethyl ether was added to this residue and after similar evaporation steps, 1ml of acetonitrile was added with 1ml of hexane. The solution was vortexed and centrifuged for 20 minutes. The solvents formed two immiscible layers with hexane on top. The hexane layer was removed, and the acetonitrile layer re-extracted with 1ml hexane, centrifuged for 20 minutes and the final acetonitrile layer aspirated and evaporated. The final residue was centrifuged briefly and then re-dissolved in 500µl phosphate buffered saline (pH 7.5) containing 1% bovine γ globulin [1].

Parallelism Assays
Five plasma samples and six blubber extracts were serially diluted from neat through 1/2, 1/4 and 1/8 with the 0 ng/ml cortisol standard provided in the ELISA kit. In a set of guidelines published for assessing parallelism between bioassay dilution curves derived from ELISAs, logistic-log models were found to describe the data with the greatest accuracy over the widest dilution range [2]. 4 parameter logistic-log models describe, with a high degree of accuracy, standard reference samples and sample curves that display a pronounced sigmoidal shape when plotted on an optical density versus log dilution scale [2]. Here, optical density was modelled against the dilution factor, using a 4 parameter logistic-log model to maintain accuracy across a range of concentrations shown by our data. As the plasma data dilutions ranged across the whole standard curve, between 40 ng/ml to 750 ng/ml, the whole standard curve was used. As the blubber dilution extracts did not range across the whole standard curve, but were concentrated between 5 ng/ml and 75 ng/ml, only the lower end of the standard curve between 0ng/ml and 100ng/ml was used for comparison to assess parallelism with this part of the curve that we are interested in. Statistical comparisons among the standard curve and the extract dilution curves and the plasma dilution curves were carried out by modelling the dilutions as 4-parameter log-logistic models using the 'drm' function in the 'epicalc' library in the statistical package R, version 3.1.2 (R Development Team, 2014).
Examples for these models are shown in Supplementary Fig. 1. Parameter estimates for each model were then compared to assess if the curves were parallel to each other and to the standard curve. There were no significant differences between the parameter estimates of each plasma dilution curve or the standard curve (p values > 0.1), thus providing strong evidence that these dilution curves and standard curve were parallel to each other. The same statistical comparisons were carried out on the blubber extract dilutions and there were no significant differences between the parameter estimates of each dilution curve or the standard curve (p values > 0.1), with the exception of the upper values ie. the asymptote of the curves.
Logistic-log curves can have identical slopes and different asymptotes and do not violate the parallelism rule for accurate quantification of antibody in a sample [2]. Parallelism of these curves supports the assumption that the binding characteristics allow the reliable determination of hormone levels in the diluted blubber extracts and in the plasma samples [3].

Matrix Effect Tests
Successful immunologic assays require an optimal pH and ionic strength that promotes specific antibody-antigen complexes while reducing the nonspecific binding of other proteins in the samples that increase assay interference. As the ELISA kit used here was designed for use with serum or plasma samples, the compatibility of the kit with extracts resuspended in PBS with 1% bovine γ globulin was assessed. Equal volumes of each standard were spiked with PBS with 1% bovine γ globulin and assayed in tandem with the unspiked standard curve. The known and the apparent concentrations in the spiked samples should show a positive linear relationship with a slope of approximately 1.0 [4]. The matrix effect test was successful with a slope of 0.99 ( Supplementary Fig. 2) (a slope of between 0.8-1.2 was considered acceptable). It was concluded that matrix effects are minimal and this sample diluent is therefore compatible with the immunoassay.

Cortisol Recovery from Spiked Samples
As the blubber biopsies were of varying sizes, it was necessary to determine the extraction efficiency of the method across tissue samples of different masses. Pooled biopsy blubber samples were divided into masses of 0.1g, 0.15g, 0.2g, 0.25g, 0.3g, 0.35g, and 0.4g (all ±0.025g), each one in triplicate, such that one sample of each mass was unspiked while the other two were cold spiked with 100ng of cortisol. Cortisol was then extracted and measured as discussed above, and the percent recovery calculated for each sample.
There was a significant negative correlation between extraction efficiency and sample mass (linear model, p = 0.006, Adjusted R 2 = 0.46) with efficiencies ranging between 99.88% for the 0.1g samples down to 63.21% for the 0.4g samples (Supplementary Fig. 3). This linear model ( Supplementary Fig. 3) was used to calculate the expected extraction efficiencies for all the biopsies samples based on their mass (Equation 1). These extraction efficiencies were then used to correct the measured cortisol concentrations in each sample to give a final cortisol concentration used for statistical analysis.  Blubber~ Biopsy+as.factor(Sex)+as.factor(Season)+as.factor(Location) 10 914.7 Blubber~Biopsy+as.factor(Sex)+ as.factor(Location) 8 927.6 Blubber~Biopsy+Plasma+as.factor(Sex)+as.factor(Location) 9 927.8 Blubber~Biopsy+ as.factor(Season)+as.factor(Location) 9 937.7 Supplementary