A major research focus in basic and applied ecology is to understand the origin and maintenance of ecosystem services [1
]. Ecosystem services include regulatory services such as carbon sequestration, climate control, and decomposition. Ecosystems also provide services in the form of plant and animal products that can be sustainably harvested [2
]. Furthermore, plant and animal products are broken down to yield a suite of elemental nutrients and more complex organic molecules, including proteins. However, the generation of even a simple inventory of molecular ecosystem services within an intact food web is a daunting prospect. The physical and biological boundaries of a food web are usually impossible to objectively define, and the spatial extent of a typical natural food web is simply too large for a meaningful comprehensive study of biomolecular services. Until recently, it has not been possible for ecologists to even begin an efficient survey of the full spectrum of molecular products from an intact ecosystem [3
]. As a consequence, most studies of the biochemical throughput of ecosystems have been limited to analyses of a few critical nutrients and stoichiometric ratios [4
]. In this study, we report an initial proteomic characterization of a naturally-intact and spatially-tractable ecosystem [5
]. Our proteomic survey revealed protein- and peptide-based biomarkers that could be used to monitor specific taxa or particular proteins within a defined food web.
An entire aquatic food web () is harbored within the cupped leaves of the carnivorous pitcher plant Sarracenia purpurea
. The leaves of this plant form a simple, open tube that fills with roughly 10ml of rainwater and functions as a trap that catches invertebrate prey (mostly ants and flies [6
]). The dead carcasses of captured prey are the base of a distinctive aquatic food web that includes microbes, protozoa, rotifers, mites, and the larval stages of three dipterans: midges, mosquitoes, and sarcophagid flies. The midge and sarcophagid larvae shred the prey, which is further decomposed by the microbial component of the food web. Protists, mites, and mosquito larvae feed on the bacteria, and late instar sarcophagid larvae feed on protists and mosquito larvae [7
]. Nutrients and biomolecules released by this decomposition process are actively sequestered by the living plant, which in turn releases oxygen and probably other metabolites and organic molecules to the pitcher liquid [8
Figure 1 (A) The Sarracenia food web is a detritus-based “brown web” with a resource input of captured insect prey. The three common dipteran species chosen for proteomic characterization in this study are indicated by shaded boxes. (B-Left Panel) (more ...)
A major goal is to comprehensively identify the protein component of the biomolecular services provided by and exchanged within this “Sarracenia
microecosystem” and to identify quantifiable protein signatures indicative of its biological state. Towards this goal, we first sought to conduct a proteomic survey to determine protein signatures of the aquatic larvae of the three common dipteran species residing within the pitcher plant: a sacrophagid fly (Fletcherimyia fletcheri
), the pitcher plant mosquito (Wyeomyia smithii
), and the pitcher-plant midge (Metriocnemus knabi
). These three species occur at the base, the middle, and the top of this aquatic food web, and they are commonly collected in the leaves of S. purpurea
throughout its extensive geographic range [9
]. We reasoned that each of these species might display a unique protein profile that would help us to identify the origin of particular proteins encountered during more comprehensive surveys of the pitcher plant liquid and which may be indicators of ecosystem state. See Supplementary Materials and Methods
for information on sample collection and preparation.
shows the protein profiles obtained from the isolated resident arthropods (left panel) and three samples of mixed particulates from individual pitcher plants that represent medium, low and high biomass density respectively (right panel). Although the profiles of the isolated species show a number of visual similarities, characteristic differences can be observed in relative band intensities, creating a SDS-PAGE “signature” of each species. In some cases the SDS-PAGE signature of a given isolate suggested a possible diagnostic for determining the dominance of a given larval species within a mixed-species sample (compare the SDS-PAGE signature of W. smithii in lane 1 of the left panel of with mixed particulate 1 (P1) in the right panel). However, as in the case of particulate fraction 3 (P3), when the pitcher plant had an abundance of captured arthropod prey, no one of the larval signatures was readily discernable. Together these data suggest a rough diagnostic for determining protein signatures of the three major pitcher plant dipteran species in isolation or when they dominate the biomass of a given particulate fraction.
Although the SDS-PAGE profiles might prove useful, identifying species-specific, or at least species-predominant, proteins would add an additional level of confidence in establishing protein signatures for future proteomics surveys of the pitcher plant food web. We therefore endeavored to catalogue the major proteins of the three dipteran larvae. Such an analysis would also provide the first proteomic reference database of these organisms. The entire gel lane for each organism shown in (left panel) was divided into 16 (F. fletcheri)
or 17 slices. Major bands were kept common to facilitate comparisons. Gel slices were diced into 1mm cubes and subjected to in-gel tryptic digestion and peptide extraction as described [10
]. MS and data analysis is described in the Supplementary Materials and Methods
More than fifty proteins from each species were identified following Sequest search analysis using the D. melanogaster
database (Supplementary Tables 1-3
). As expected, many proteins were common to all three species (Supplementary Fig. 4 and Supplementary Table 7
) because a search of the D. melanogaster
database would favor the identification of highly conserved proteins. However, a few proteins were uniquely identified in each of the three species ( and Supplementary Table 7
). Of note are the 71 peptides identifying isoforms of larval serum proteins that were found only in F. fletcheri
. These peptides were identified primarily from the major coomassie-stained F. fletcheri
band running at ~78 kDa. Given its predominance, this protein family could potentially be used as a diagnostic indicator protein of F. fletcheri
. Additionally, a number of proteins were identified in one species by more than three-fold the number of peptides found in either one of the other two (), suggesting that additional proteins, or their increased relative abundance, could also be used as species-specific indicators. It is anticipated that once protein databases are established for these organisms many additional proteins will be identified from the data collected in this study.
Major proteins and signature proteins identified from arthropod associates of S. purpurea
Because of the high overlap in identified proteins among species, most of the identified proteins would not serve as species-specific indicators. However, we asked whether the same protein identified from all three species might nonetheless harbor species-specific peptide indicators. If sufficient numbers of such peptides were identified from the relatively invariant residents and prey constituents of the S. purpurea
food web, these peptides conceivably could serve to trace species and specific proteins even from a mixed protein sample. Such methods would be particularly useful if coupled with quantitative proteomic approaches such as AQUA analyses [11
]. It is important to note, however, that the relative abundance of a given protein may only be a rough indicator of the relative species abundance as some proteins are highly regulated at the level of the individual organism.
To search for such indicator peptides, we took the acquired data for the most identified protein in the entire dataset, muscle myosin heavy chain. Mass spectral data came from the peptides derived from the dominant band at approximately 220 kDa and its flanking regions. These spectral data were then subjected to a SEQUEST search against a composite database of the most readily identified D. melanogaster
myosin heavy chain protein sequence concatenated to the next 99 most similar protein sequences identified by BLAST searching. Concatenated to the end of this database were the reversed sequences of all 100 proteins. This entire database is provided as Supplementary Database 1
. The search was conducted as described in the Supplementary Material and Methods
, except trypticity was not required until after the search was complete for increased stringency. Tryptic peptides from this search were filtered to a less than 0.1% false discovery rate using stringent Xcorr, ΔCorr and ppm values [12
]. Peptides were then compiled and sorted to identify species-specific peptides ( and Supplementary Table 4
). A number of sets of homologous muscle myosin heavy chain peptides were identified (Supplementary Tables 5 and 6
) where a peptide from one species was unique, typically by a difference in only one amino acid (). Furthermore, definitive mass spectra were obtained for these unique peptides (examples are shown in supplementary Figures 1-3
). However, there remains a formal possibility that these peptides are not species-specific for the following reasons. First, although data for each of the arthropod associates were from a mixture of several individual isolates mixed together, the MS data were generated from a single gel () and as such will need to be further established in future proteomics analyses. Second, if unlike Drosophila, these insects harbor two or more muscle myosin heavy chain genes then we cannot exclude the possibility that the sequence of a second muscle myosin protein, perhaps in low abundance in our samples, negates a perceived uniqueness. However, our data do not support this as individual arthropods did not show multiple homologous peptides covering the same regions of myosin heavy chain. Third, although these peptides may distinguish between the three insects examined here, other insects not examined may contribute to the heterogeneous biomass of the pitcher plant community and these insects may have protein sequences that also negate uniqueness with the pitcher plant ecosystem. This last possibility is likely to be true to a degree. However, if multiple peptides are monitored, stoichiometric measurements could be used to identify sets of peptides belonging uniquely to one organism. Furthermore, for a state analysis of a dynamic ecosystem it is not a requirement to know the species origin of a given peptide biomarker. It would be sufficient to know the relative abundance of biomarker peptides if they provided predictive power regarding the ecosystem state and dynamics. Ultimately we plan to identify peptides that serve these goals and our initial characterization here suggests that significant identification and precision can be achieved.
Myosin heavy chain signature peptides for major arthropod associates of S. purpurea
In conclusion, in spite of the lack of database representation for these three dipterans, our analyses identified apparent species-specific peptide biomarkers from a database of related proteins. Our approach should be applicable toward the identification of numerous species-specific peptide biomarkers in this specific food web as well as in other systems for which protein databases do not yet exist. It is important to note that while simple species identification can be achieved readily by inexpensive methods such as light microscopy, an analysis of the individual components of the ecosystem biomass will provide molecular information that may be more acutely regulated than whole organisms and may be more sensitive indicators of ecosystem state. Once specific peptide biomarkers are confirmed, quantitative MS can facilitate their absolute quantification and determine how these proteins or peptides change in abundance when the ecosystem is in different states. Furthermore, specific markers may facilitate the evaluation of the protein constituent of ecosystem services, which is still an unexplored area of study. All MS data from this study are freely available in the PRIDE database (http://www.ebi.ac.uk/pride
) under the accession numbers 16271-16273 and can be viewed using the freely available PRIDE Inspector program (http://code.google.com/p/pride-toolsuite/wiki/PRIDEInspector