Proteome assay of temperature stress and protein stability in extreme environments: groundwork with the heat stress response of the bivalve Mercenaria mercenaria.

ABSTRACT Protein metabolism is an expensive cellular process that can generally account for one third of basal metabolism in animals. Shifts in the stability of proteins under increased environmental temperatures could potentially alter the energy budget of an organism. However, studying the thermal stability kinetics of individual proteins is tedious and ultimately, difficult to relate to changes in the fitness of an organism. Yet understanding how organisms inhabiting extreme environments (polar seas, hydrothermal vents, and deep ocean basins) are able to maintain or limit the rate of protein turnover in the total cellular protein pool is crucial for our understanding of the total metabolic costs associated with survival in these habitats. To assess protein stability in field collected organisms at a proteome scale, we developed a high-throughput assay for protein denaturation profiles of total tissue extracts in bivalves. These profiles are quantitative and reveal unique compositional features of different tissues. Heat stress experiments in the clam Mercenaria mercenaria reveal that the protein pool of mantle and digestive mass tissues are more thermally stable after a short exposure to 35[degrees]C. This increase in stability could have large implications for the energy budget of M. mercenaria when exposed to high summer water temperatures. This methodology could readily be used to assess the stability and/or turnover potential of a variety of organisms experiencing extremes of both temperature and pressure.

KEY WORDS: Mercenaria, protein denaturation, metabolism, physiological ecology 

PROLOGUE

In 1979, an undergraduate without any real career direction in mind took an invertebrate zoology class. During the course, he had to prepare a literature report and chose to look at what was known about snail predators of clams. When the paper was finished, the student could not believe, first of all, that he was able to find so many papers on the topic, and second, that almost all of them had one author in common: Melbourne R. Carriker. Upon expressing his awe to the invertebrate zoology professor after turning in the paper, the professor replied that he was well aware of the depth and detail of the snail predator work because Prof. Carriker had been his M.S. advisor. The impact of the course led this undergraduate into an academic career as a marine biologist. Twenty-one years later when this same student was interviewing for a faculty position at the University of Delaware, he was given a tour of the facilities. Walking down a long hallway of offices following his guide, he suddenly stopped, stared at a neatly lettered name on a door, "Melbourne R. Carriker," and blurted out to his host, "That's the guy!"

INTRODUCTION

Physiological ecology has sometimes been described as simply the study of why animals live where they do. However, the novelty in the development of this research field in the 1960s and '70s was really about adopting an integrative approach to understand how tissue-level biochemistry, organismal-level physiology and population-level ecology were all integrated into a single continuum across hierarchical scales. The intellectual foundation then established by physiological ecologists, now serves as the basis for the current emphasis and development of "systems-level" biological research. In many ways, Carriker's work on the biochemistry, functional morphology, physiology, and ecology of the boring gastropod Urosalpinx cinerea (Say) is a stellar example of this evolution of integrative research that defined the ideals of physiological ecology (Carriker 1951, Carriker 1960, Carriker et al. 1967, Smarsh et al. 1969, Carriker 1971, Carriker 1977, Carriker et al. 1978a, Carriker & Williams 1978, Carriker et al. 1978b, Gordon & Carriker 1978). In this paper, we present a study that develops a new method for assessing the hierarchical integration between protein biochemistry and cellular function in the clam Mercenaria mercenaria (Linnaeus). The focus is essentially to identify a single measurement for the degree of coordination that exists among different levels, to assess when environmental changes (thermal stress) begin to alter the normal, homeostatic level of coordination. Although we are using new tools (qPCR thermal cyclers) and new approaches (proteome homeostasis), it is nonetheless "physiological ecology" in the same way that Carriker exploited his new tool (electron microscopy) and his new approach (calcibiocavitology; (Carriker & Smith 1969).

A large proportion of the energy expended by all eukaryotic cells is dedicated to protein metabolism (~1/3 of basal metabolic rates; Marsh et al. 2001). Most of this energy demand is driven just by the steady turnover of cytoplasmic proteins. We know that the full tertiary structures of most proteins are only marginally stable (Jaenicke 1991) and that there is an inherent rate of protein denaturation that persists in living cells. Most of these cytoplasmic proteins have such marginal stability, because they must maintain sufficient structural flexibility to catalyze reactions. It is this balance between functional flexibility and structural rigidity that is challenging for organisms living in extreme environments (polar seas, hydrothermal vents, deep ocean basins) because the physical conditions of their habitats impose more strenuous constraints on protein structure (temperature extremes, pressure). Thus, subtle shifts in environmental variables may impose more dramatic changes in the protein turnover balance, leading to the expenditure of a larger portion of metabolic energy to protein metabolism. At present, there is no methodological approach for assessing the overall stability of a cellular protein pool (total proteome).

We are interested in measuring subtle changes in protein pool metabolism associated with shifts in the turnover balance and ultimately the consequent changes in energy metabolism that those shifts impose. We hypothesize that any organism experiencing any daily temperature variation will also experience concomitant shifts in protein turnover rates and consequently the total energy demand of both protein synthesis and degradation pathways. Towards this end, we have developed a sensitive assay for assessing the stability/denaturation potential of the cytoplasmic protein pool, using the heat stress response of the hard clam, M. mercenaria, as our experimental organism.

Protein denaturation profiling was developed as a rapid, high-throughput extension of the genome and transcriptome complexity analyses developed in our laboratory (Fielman & Marsh 2005, Marsh & Fielman 2005) but now applied to the proteome. This approach combines the analytical strategy and statistics of nucleic acid reannealing kinetics (Marsh & Fielman 2005) with isothermal protein denaturation kinetics (Epps et al. 2001, Sarver et al. 2002) to rapidly assess potential stability and composition of protein pools in a high-throughput (96-well microtiter plate) assay. This novel method relies on the fluorescent dye NanoOrange (Invitrogen). The fluorescence of NanoOrange increases on association with hydrophobic amino acid residues in proteins and has been used as a quantitative measure of amount of protein in a sample solution (Jones et al. 2003). As native protein structure begins to denature, more hydrophobic residues become exposed to the surrounding aqueous environment. Thus, as a protein transitions from a stable molten globule structure to a random coiled state, more binding sites will be available for the NanoOrange fluorophore. This process could then be used to quantitatively measure denaturation kinetics of proteins, composition of complex protein pools, and thermal sensitivity of protein pools isolated from field collected organisms.

[FIGURE 1 OMITTED]

MATERIALS AND METHODS

Stepwise isothermal protein profiling was typically performed with 10 [micro]g protein. In brief, the protein was added in triplicate to QPCR strip tubes containing NanoOrange dye (diluted 1/650 from stock) and 150 [micro]L of 20 mM Tris (pH 8.0) or 150 [micro]L of the NanoOrange diluent provided by the manufacturer. NanoOrange diluent appeared to yield the best results: this is likely a function of a detergent included in the diluent (as discussed later). An ABI 7700 thermal cycler was used to monitor fluorescence at 570 nm during denaturation profiling. Temperature of the thermal cycler was incremented from 25[degrees]C to 90[degrees]C in 5[degrees]C steps. Each increment lasted from 20 to 45 min. Raw fluorescence data from a typical run is illustrated in Figure 1. Here, the temperature can be seen to vary as a step function. Sample fluorescence increases at the initial point of each temperature step as the temperature change denstabilizes the protein mixture. Essentially the method collects sequential denaturation curves at increasing temperatures. Detail of the increase in relative fluorescence intensity from a single temperature step (50[degrees]C) is shown in Figure 2. The form of a second-rate function describing this process is clearly evident. A single denaturation run generates roughly 20,000 data points for each well, or ~1,500 measurements per temperature step.

Fluorescence data were exported from the ABI as chromatograms using the onboard Sequence Detection software (Applied Biosystems). A PERL script was used to parse the 570 nm wavelengths and generate a single channel data table for each sample/well. A C++ program was written to generate the diversity and kinetic statistics. Denaturation curves for each well at each temperature were iteratively fitted to the following second-order equation:

[F.sub.t] = [alpha] + [beta] x (1 - [e.sup.-[gamma]t]) (1)

where F, is the fluorescence at time t, [alpha] is the y-intercept, [beta] is the maximum asymptote, and -[gamma] is the rate constant.

[FIGURE 2 OMITTED]

Estimates of the complexity of the protein composition in a sample were calculated at each temperature by scaling fluorescence data between 0 and 100% and separating into binned intervals. Then a modified Shannon-Weaver information statistic was calculated using the following equation:

H' = -1 x [SIGMA][p.sub.i] x [log.sub.2]([p.sub.i]), (2)

where Pi equals the proportion of the ith observation in the total data set for a given temperature. H' is considered an estimate of the complexity of a mixture of proteins in a sample at that temperature. To facilitate comparison of complexity among different temperature intervals (i.e., the abundance of each denaturing protein and the number of unique denaturing proteins affected at a given temperature), H' was scaled by the fluorescence of the asymptote ([beta]) from Eq. 1 rather than -1 as in Eq. 2 to yield the relative complexity as

[H.sub.flr] = -[beta] x [SIGMA][p.sub.i] x [log.sub.2]([p.sub.i]). (3)

The application of Eq. 3 to describing changes in the distribution and abundance of biomolecules in complex mixtures is presented in a review by Hoover et al. (2007).

The preliminary assays for developing these techniques used commercially available (and well characterized) proteins RNAse A and bovine serum albumin. Experimental work with biological protein samples used extracts from oyster heart and clam gill tissues. For the oyster heart extract, a single heart was dissected and homogenized in cold 10 mM Tris (prt 7.8) with 100 mM NaCl using a motor driven pestle homogenizer. The homogenate was cleared by high speed centrifugation and an aliquot of the supernatant used in the denaturation assay (Ulrich & Marsh 2006).

For the heat shock experiment, four quahog clams (M. mercenaria) were collected from the Cape Henlopen sand flats (Lewes, DE; water temperature ~27[degrees]C, 26 ppt salinity) and kept overnight in water at 25[degrees]C and 26 ppt salinity. Clams were randomly assigned (n = 3) to control (22[degrees]C 23[degrees]C, 26 ppt salinity) and heat stress (35.5[degrees]C-37[degrees]C, 26 ppt salinity) treatments. Treatments lasted two hours and clams were allowed to recover for approximately four hours at room temperature. Mantle and digestive tissue (100-170 mg) were homogenized in 50 mM Tris (8.0) with a PowerGen 35 homogenizer (Fisher) and sonicated 15s at 3 to 4 W power (Sonics and Materials VibraCell). Homogenates were clarified by centrifugation at 12k g at 4[degrees]C for 15 min. The supernatant was further clarified at 20k g at 4[degrees]C for 10 min. The supernatant was frozen at -80[degrees]C until used for protein denaturation profiling. Protein was quantified with Coomassie Plus Bradford reagent (Pierce).

RESULTS

To assess the sensitivity to protein amounts, denaturation profiles of bovine pancreatic RNAse A were generated with three different amounts of protein: 10 [micro]g, 1 [micro]g, and 100 ng. RNAse A protein was added in triplicate in a 5-[micro]L volume to 150 [micro]L of 1X NanoOrange diluent containing 1/650 NanoOrange dye. Denaturation was performed on an ABI 7700 from 25[degrees]C to 85[degrees]C with 45 min steps of 5[degrees]C. A final step of 25[degrees]C for 15 min was also included to assess the fluorescence after denaturation profiling. Fluorescence data was analyzed with the software tools developed for this project. In Figure 3, 1 [micro]g and 10 [micro]g RNAse A yielded profiles with peaks at 30[degrees]C and 50[degrees]C. No protein denaturation was detected in the lowest 100 ng sample. The shift in the maximal peak at different quantities of protein indicate a strong potential for protein-protein interactions as hydrophobic residues become exposed and may ionically bind to the amino acid R groups of neighboring proteins in solution. Thus, it appears that with high protein concentrations there is a potential snowball effect of one particular class of proteins denaturing and setting off a chain reaction that would destabilize many other proteins. For this study, all subsequent assays used 10-[micro]g protein.

Given the importance of the surrounding environment for protein stability, we assessed the denaturation profiles for 10 [micro]g RNAse A using 20 mM Tris (prt 7.7) or 1X NanoOrange diluent. The composition of NanoOrange diluent is proprietary but likely contains a detergent such as SDS, which would have three important effects: (1) the presence of the detergent would further sensitize proteins to thermal denaturation, (2) the detergent would likely surround any exposed hydrophobic residues during denaturation and provide a more efficient interaction with the NanoOrange fluorophore, and (3) the detergent would inhibit hydrophobic residues from interacting with other amino acid groups. In Figure 4, these denaturation profiles were performed as mentioned earlier and yield greater fluorescence when using the manufacture's diluent. Though the complexity profiles have similar shapes, the Shannon-Weaver complexity of RNAse A in NanoOrange diluent had substantially higher maximum than that produced by thermal denaturation in a Tris buffer. This is most likely a function of detergent present in the diluent to enhance the efficiency of NanoOrange binding when it is associated with a detergent coat around hydrophobic residues (Jones et al. 2003).

[FIGURE 3 OMITTED]

[FIGURE 4 OMITTED]

A series of standard denaturation profiles were run on different proteins to assess if the complexity profiles generated were unique for proteins with different amino acid structures. The denaturation profiles of 10 [micro]g RNAse A, BSA, and total oyster heart protein are shown in Figure 5. Denaturation profiles of RNAse A and BSA have a single rounded peak at 45[degrees]C to 50[degrees]C and 55[degrees]C, respectively. The peaks in complexity correspond roughly to published values of thermal denaturation for RNAse A and BSA. Freire and Biltonen (1978) show that the majority of RNAse A molecules in a population are unfolded at temperatures greater than ~42[degrees]C. The oyster heart protein profile is similar to the complexity curve of RNAse A with a single peak at 45[degrees]C. What is important to note is that even between 30[degrees]C and 35[degrees]C we can detect a change in the denaturation state of the total pool extract. Although the peak at 45[degrees]C is beyond physiological relevancy, the shape of the denaturation curve between 30[degrees]C and 45[degrees]C is likely to hold some functional biological information for assessing changes in protein pool stability.

[FIGURE 5 OMITTED]

To determine the field relevance for utilizing this method for detecting subtle changes in the thermal stability of complex protein pools, we assessed the impact of direct thermal stress on mantle and digestive mass proteins of Mercenaria mercenaria. Four quahog clams (M. mercenaria) were collected from the Cape Henlopen sand flats (Lewes, DE; water temperature ~27[degrees]C, 26 ppt salinity) for the heat shock experiments. The first assays were performed on control individuals immediately after collection from the field. Protein denaturation was assayed in triplicate with 10 lag protein over a thermal gradient of 30[degrees]C to 76[degrees]C in 2[degrees]C temperature steps to give greater resolution in the denaturing profiles. Complexity profiles of these field protein samples for M. mercenaria mantle and digestive mass tissues are shown in Figure 6. Digestive mass protein is characterized by a single peak at 36[degrees]C, whereas mantle exhibits peaks in complexity at both 40[degrees]C and 46[degrees]C. Both traces are skewed toward low temperatures, but there are apparent differences between the denaturation kinetics in each. The lower peak temperature in the digestive mass proteins suggests that this pool is not as thermally stable as in the mantle. However the sharp shift in both pools between 30[degrees]C and 35[degrees]C suggests that both tissues have protein pools that would be sensitive to changes in environmental temperatures above 30[degrees]C. In Figure 7, the heat stress results are presented for both tissue types. After a two hour exposure to 36[degrees]C followed by a four hour recovery at 22[degrees]C, the protein pools in both tissues evidence a shift in their thermal stability relative to the control individuals held for six hours at 22[degrees]C. In mantle (Fig. 7A) and digestive mass (Fig. 7B) the general patterns of denaturation are conserved but heat stress results in a reduction in the magnitude of the denaturation rates (lower values of [H.sub.flr]) between 30[degrees]C and 50[degrees]C. This is particularly evident in digestive mass protein, where the individual variance is low and the separation between the control and the heat treatment groups is noticeable. The 42[degrees]C peak that is present in the mantle tissue controls, but absent from the heat stress group suggests that there is a dominant protein in that pool is sensitive to 42[degrees]C. After heat exposure, however, it seems that the thermal stability of this protein(s) is increased because of the absence of this denaturation peak.

[FIGURE 6 OMITTED]

[FIGURE 7 OMITTED]

DISCUSSION

Protein denaturation profiling has the potential to extend methods of assessing molecular complexity and kinetics (Fielman & Marsh 2005, Marsh & Fielman 2005) to assays relevant to understanding dynamics of the proteome in organisms exposed to environmental extremes. The method we have pursued here is an amalgam of established approaches for studying second order rate kinetics of nucleic acids and denaturation kinetics of proteins. The complexity statistic used is one used by many areas of biological research as an integrative index of the distribution and abundance of "individuals" (or elements) within a complex system. The metric here provides an index of thermal stability for individual proteins and complex protein samples. These results suggest that the composition and relative abundances of proteins in a protein mixture can be detected through analysis of thermal profile complexity and denaturation kinetics without having to isolate and characterize the thermal stabilities of individual proteins.

The denaturation profiles we present here successfully capture many characteristics of a protein sample: (1) the profiles reflect the amount of protein present in a sample (Fig. 3), (2) the profiles detect differences in protein composition (Fig. 5), and (3) the profiles yield tissue-specific "fingerprints" of protein denaturation (Fig. 6). Additionally, the profiles are sensitive to immediate environmental conditions such that changes in the proteome in response to heat stress can be detected by shifts in denaturation at specific temperatures (i.e., 42[degrees]C in M. mercenaria mantle, Fig. 7A). Because the assays are conducted in a 96-well plate format, we now have the capacity to make these measurements at a scale that is relevant for understanding population level processes (i.e., with three duplicate assays per individual, 32 individuals can be sampled at a time). At this sampling density, denaturation profiling could also potentially be used to identify and correlate biomarkers with environmental stress.

Overall, there is likely to be a significant metabolic cost for temperate bivalves experiencing any thermal stress above 30[degrees]C even though this temperature is well within the range of a "normal" exposure. Heat stress has received most of the attention in the literature in the form of heat shock protein expression at near-lethal levels of heat exposure. However, routine exposure to an elevated temperature can have a protracted negative impact because of an increase in metabolic energy expenditure required to deal with a persistent increase in protein turnover rates. Future work should focus on assessing the correlation that may exist between metabolic rate measurements and these protein denaturation profiles before and after thermal exposures. The associate ATP costs with different curve profiles would establish a firm linkage between this molecular level measurement and an organismal level processes. The key advantage to the denaturation profiles used here is the integration of a broad response across the entire protein population of a tissue. On one hand profiles measure a molecular level response (denaturation) but on the other they measure a global process (turnover) that can be directly related to organismal function via energy metabolism.

This upwards integration of a measure of protein stability to potential changes in fitness (metabolic costs) of an organism provides a new tool for assessing the cellular homeostasis of marine organisms inhabiting extreme environments. Ongoing work is focused on relating changes in the thermal stability profiles of polar bivalve veligers to shifts in total individual larval metabolism (oxygen consumption). Microrespiration methods, and now this current protein stability assay, can be performed with such small amounts of tissue (i.e., individual larvae) and with many samples (96-well plate formats) that statistically informative surveys at a population scale can be conducted with sample sizes of ~30 individuals for any habitat site. Obtaining large data sets like this for marine invertebrates in extreme environments will provide a new perspective on the metabolic costs associated with protein metabolism, and ultimately organismal survival, in these habitats.

At present, we are well aware of the necessity of studying organisms at a systems level rather than at just the level of a single discipline. Crossing levels of biological organization requires novel insights into how to apply data collected at one level to processes at another. We are essentially looking for the functional linkages between those levels. Today, we call it functional ecology or functional genomics. In the 1960s, this research frontier was called functional morphology, and it was the rapid increase in the availability of electron microscopy for revealing tissue ultrastructure that opened the door for a new perspective on the structure and function of organisms. Mel Carriker crossed this threshold in marine biology with his work on the accessory boring organ in Urosalpinx. His two publications in the journal Science attest to the novelty and importance of what he did to develop the blossoming field of marine physiological ecology (Carriker et al. 1967, Gordon & Carriker 1978). His research and his academic statesmanship have been inspirational to many.

EPILOGUE:

Shortly after Mel's stroke, the now old-undergraduate visited him in the hospital. Once again, he mentioned to Mel about the undergraduate paper he had written for the invertebrate zoology course taught by one of Mel's former students. The important point being the fact that the paper cited almost 20 papers that Mel had written about Urosalpinx and this body of work had really excited this undergraduate about marine biology as an academic career. With a smile, Mel commented that maybe a better job could have been done by that undergraduate on the literature search, because there were a beck of a lot more than just 20 Carriker papers on Urosalpinx. As always, Mel's wit was quick yet gracious and gentlemanly.

ACKNOWLEDGMENTS

The authors thank Dr. T. Hanson for discussions regarding the development of this technique. This research was supported by a grant from Delaware Sea Grant (RB/43).

LITERATURE CITED

Carriker, M. R. 1951. Observations on the penetration of tightly closing bivalves by busycon and other predators. Ecology 32:73-83.

Carriker, M. R. 1960. Comparative functional morphology of the boring mechanism in boring gastropods. Anat. Rec. 138:340-340.

Carriker, M. R. 1971. Preliminary study by scanning electron microscopy of dissolution of shell of mytilus by accessory boring organ of Urosalpinx. Biol. Bull. 141:380-381.

Carriker, M. R. 1977. Ultrastructural evidence that gastropods swallow shell rasped during hole boring. Biol. Bull. 152:325-336.

Carriker, M. R. & E. H. Smith. 1969. Comparative calcibiocavitology: summary and conclusions. Am. Zool. 9:1011-1020.

Carriker, M. R., D. Vanzandt & G. Charlton. 1967. Gastropod urosalpinx: ph of accessory boring organ while boring. Science 158:920-922.

Carriker, M. R. & L. G. Williams. 1978. Chemical mechanism of shell dissolution by predatory boring gastropods: review and an hypothesis. Malacologia 17:143-156.

Carriker, M. R., D. Vanzandt & T. J. Grant. 1978a. Penetration of molluscan and non-molluscan minerals by the boring gastropod urosalpinx-cinerea. Biol. Bull. 155:511-526.

Carriker, M. R., L. G. Williams & D. Vanzandt. 1978b. Preliminary characterization of secretion of accessory boring organ of shell-penetrating muricid gastropod urosalpinx-cinerea. Malaeologia 17:125-142.

Epps, D. E., R. W. Sarver, J. M. Rogers, J. T. Herberg & P. K. Tomich. 2001. The ligand affinity of proteins measured by isothermal denaturation kinetics. Anal. Biochem. 292:40-50.

Fielman, K. T. & A. G. Marsh. 2005. Genome complexity and repetitive DNA in metazoans from extreme marine environments. Gene 362:98-108.

Freire. E. & R. L. Biltonen. 1978. Statistical mechanical deconvolution of thermal transitions in macromolecules. 1. Theory and application to homogeneous systems. Biopolymers 17:463-479.

Gordon, J. & M. R. Carriker. 1978. Growth lines in a bivalve mollusk: sub-daily patterns and dissolution of shell. Science 202:519-521.

Hoover, C. A., M. Slattery & A. G. Marsh. 2007. A functional approach to transcriptome profiling: linking gene expression patterns to metabolites that matter. Mar. Biotech. 9:411-419.

Jaenicke, R. 1991. Protein stability and molecular adaptation to extreme conditions. Eur. J. Biochem. 202:715-728.

Jones, L. J., R. P. Haugland & V. L. Singer. 2003. Development and characterization of the nanoorange (r) protein quantitation assay: a fluorescence-based assay of proteins in solution. Biotechniques 34:850-861.

Marsh, A. G. & K. T. Fielman. 2005. Transcriptome profiling. Of individual larvae of two different developmental modes in the poecilogonous polychaete streblospio benedieti (spionidae). J. Exp. Zool. 304B:238-249.

Marsh, A. G., R. Maxson & D. M. Manahan. 2001. High macromolecular synthesis with low metabolic cost in antarctic sea urchin embryos. Science 291:1950-1952.

Sarver, R. W., J. M. Rogers & D. E. Epps. 2002. Determination of ligand-murb interactions by isothermal denaturation: Application as a secondary assay to complement high throughput screening. J. Biomol. Screen. 7:21-28.

Smarsh, A., H. H. Chauncey, M. R. Carriker & P. Person. 1969. Carbonic anhydrase in accessory boring organ of gastropod, urosalpinx. Am. Zool. 9:967-982.

Ulrich, P. N. & A. G. Marsh. 2006. Interindividual variation in malate dehydrogenase activity in the oyster Crassostrea virginica. Mar. Freshw. Behav. Physiol. 39:293-306.

PAUL N. ULRICH (1) AND ADAM G. MARSH (2) *

(1) Center for Tropical and Emerging Global Diseases, University of Georgia, Athens, Georgia 30602; (2) College of Marine & Earth Studies, University of Delaware, Lewes, Delaware 19958

* Corresponding author. E-mail: [e-mail]