Comparative Protein Profiles of the Ambrosia Plants

Abstract

Ragweed pollen is primarily responsible for the hay fever allergies of sufferers throughout the world. A proteome study of three ragweed plants (Ambrosia artemisiifolia, Ambrosia trifida, and Ambrosia psilostachya) was undertaken to document and compare their protein profiles. Proteins extracted from the pollen of the three plants were subjected to one dimensional electrophoresis followed by tandem liquid chromatography-mass spectroscopy. Peptide sequence mapping permitted discovery of proteins not previously reported for all three plants and 45% of the identified proteins were shared by all three of them. Application of stringent criteria revealed not only a majority of known allergens for short ragweed but also allergens not previously reported for the other two plants. Additionally, potentially allergy inducing enolases are reported for the three plants. These results suggest that all three ragweed plants could contribute to the allergy malady.

1. INTRODUCTION

According to the CDC, ragweed pollen is largely responsible for the allergic hay fever reaction suffered by approximately 16 percent of the United States population each year. Despite the prevalence of this disease, only 11 proteins of the total of 99 listed by UniProt as arising from Ambrosia artemisiifolia, short ragweed, have well annotated sequence entries. The sequences for most of those 99 proteins were translated from nucleotide sources. The nomenclature committee of the World Health Organization/International Union of Immunological Societies (WHO/IUIS) lists 11 known antigens of short ragweed; six of these sequences were reviewed by UniProt. Although three prominent ragweed plants, Ambrosia artemisiifolia (common/short), Ambrosia trifida (giant), and Ambrosia psilostachya (western), are found in the United States, the majority of the documented proteins are associated with A. artemisiifolia. For A. trifida, the UniProt list of 26 proteins includes 3 reviewed, one of which is the antigen Amb t 5. For A. psilostachya, having the fewest sequence entries, only 2 of 17 are reviewed.

Early work focused on the immunogenic properties of extracts from short ragweed [1]; however, there was some coverage of all three species [2]. Significant work expansion began in the 1960s with the isolation of 2 antigens of 32 kDa and 10 kDa by King and Norman [3]. Multiple antigens were discovered in a glycosylated protein extract of short ragweed [4]. Major allergen E, Amb a 1 in current nomenclature, with four isoforms, was also isolated [5]. Employing IEF, King [6] demonstrated multiple proteins in the short ragweed extract and isolated two acidic proteins, antigen E (Amb a 1) and antigen K (Amb a 2). Later, the size of Amb a 1 at 37.8 kDa and its alpha and beta forms were determined by MALDI/ TOF analyses [7]. Rafner [8] produced protein translates from three short ragweed cDNA clones (Amb a 1.1, 1.2, and 1.3) with 85% nucleotide sequence identity; polymorphism was confirmed for Amb a 1, and a fourth (Amb a 1.4) was identified by Griffith, et al. [9]. They also reported finding Amb a 2 RNA in flowers but not in pollen [9]. The cDNA based amino acid sequence of Amb a 2 was found to share 65% identity with the Amb a 1 group [10]. Subsequently, as reported on the WHO/IUIS page, Amb a 2 was renamed as a subcategory of Amb a 1.

Extensively investigated, the basic short ragweed pollen glycoprotein Ra3 (Amb a 3) reported as being15 kDa size presented differences in amino acid composition and immunological specificity when compared to antigen E [11]. Sequencing of a single isoform of Ra3 yielded 101 amino acids and confirmed glycosylation [12]. In a follow up study, two forms of Ra3 of 12.3 kDa were reported biochemically and immunologically similar [13].

Purified Ra5 (Amb a 5) from short ragweed, comprised of 45 amino acids [14], evinced amino acid sequence polymorphism [15]. Purified antigenic Amb t 5, a 4.4 kDa polypeptide, showed limited cross reactivity with Amb a 5 [16]. Likewise, Goodfriend et al. [17] demonstrated that Ra5G (Amb t 5) was biochemically similar but immunologically dissimilar to Ra5S (Amb a 5). Lowenstein and Marsh [18] detected 52 short ragweed antigens by immuno-electrophoresis and reported isoelectric points for some. In a refined procedure, specific antigens E (Amb a 1), K (Amb a 2), Ra5, and Ra6 (Amb a 6), but not Ra3, were identified amongst the 52 antigens [19], and nine of them were found to be significant allergens [20]. Purified Ra6 was shown to be composed of four 8 kDa (SDSPAGE) isoforms with equivalent immunological properties [21].

Metzler [22, 23] reported composite 3-dimensional structures for Amb t 5 and Amb a 5 with some variation in the beta sheet cores but similar C-terminal helices. Ghosh et al. [24] confirmed the entire protein sequence of Amb t 5 from its cDNA sequence.

Heightened interest in assigning the relative antigenicity and allergenicity among Amb 5 isoforms revealed immunological dissimilarities [16, 23, 25, 26] and similarities [16, 26]; Amb p 5 being the most similarity to Amb a 5 [27].

From short ragweed, two new endopeptidases, a chymotrypsin-like [28] and a trypsin-like [29], were purified and characterized. cDNA expressed sequences yielded pan allergens including two profilin isoforms (Amb a 8), two 2 EF-hand calcium binding proteins, (Amb a 9), and one 3 EF-hand calcium binding protein (Amb a 10) [30]. Newly isolated Amb a 4 allergen has a defensin domain and is a homolog of Mugwort Art v 1 [31]. The newest antigen Amb a 11, a 37 kDa cysteine protease, was discovered and characterized by a combination of electrophoretic, immunological, and mass spectroscopy methods [32].

Identification of proteins extracted from various tissue sources by mass spectroscopy has become a prominent proteomics tool increasingly applied to plant tissue, including the pollen of allergy inducers. Initial studies employed peptide mass fingerprinting (PMF) [33–37] that evolved to the more powerful peptide sequence mapping (PSM) [36–41], wherein the spectra of experimentally generated peptides are matched to those produced virtually from all of the database proteins. Bordas-Le Floch et al. [41] created a transcriptome-derived proteome database to determine identities of short ragweed proteins obtained in a shotgun PSM approach. One dimensional SDS PAGE coupled to liquid chromatography and tandem mass spectroscopy (LC-MS/MS) enhanced the powerful PSM method [42–45].

Justification for this project included the paucity of protein-based sequencing and limited-sequence entries for the giant and western plants. In addressing these aims, we undertook proteomic profiling of all three species. Based on our findings, we have increased the number of proteins attributed to each of them. Such knowledge enables profile comparison, reveals similarity in the allergens produced by the three plants, and perhaps suggests a basis for still unaccounted allergic reactions.

For this investigation, all profiles were generated by combining one dimensional SDS gel electrophoresis with liquid chromatography and tandem mass spectroscopy. The study yielded 257 proteins, including 28 orthologs that met stringent criteria. Of the 257 proteins identified, 117 were found in all three species, 21 were in common and giant, 29 in common and western, and 22 in giant and western. There were 68 proteins associated with only one species. Western was by far the most populated with 50; 12 were attributed to common and 6 to giant. To avoid duplicate protein names appearing in the data reported in the results section, it was decided to eliminate the orthologs. Ortholog elimination yielded 229 identified proteins.

2. METHODS

2.1 Materials

The source of defatted pollen for all three Ambrosia species came from Greer Laboratories, Lenoir, NC. The following were purchased from Invitrogen (Life Technologies), Carlsbad, CA: NuPAGE Novex 4–12% Bis-Tris Protein Gels, 1.5 mm, 10-well; NuPAGE® MES SDS Running Buffer; NuPAGE® LDS Sample Buffer; and the BenchMark Protein Ladder 10–220 kDa. Iodoacetamide, Protease Inhibitor Cocktail, Trizma base , and Coomassie Brilliant Blue were obtained from Sigma-Aldrich, Saint Louis, MO. Dithiothreitol came from Bio-Rad, Hercules, CA, and sequencing grade Trypsin (V511) from Promega, Madison, WI. All general chemicals were acquired from Thermo Fisher Scientific, Waltham, MA.

2.2 Pollen Extraction

A Life Technologies procedure was modified to create an extraction process more proteomics suitable [46]. The pollen (50 mg) was suspended in 909 μL of Invitrogen LDS buffer (2% LDS and 2.5 mM Tris, pH 8.5), 3 μL of 1 M Tris Base, 10 μL of Sigma Protease Inhibitor Cocktail, 10 μL of 2 M dithiothretiol (DTT), and 18 μL of ultra-pure water. Placed on ice, it was sonicated at 20% power 5 times for 30 seconds followed by 15 minutes of shaking, while maintaining pH 8.5 by addition of 1 M Tris, if needed. Disulfide bonds were reduced by adding 10 μL of 2 M DTT to the lysate; sulfhydryl groups were alkylated by adding 121 μL of 0.4 M iodoacetamide to the lysate, and the samples were placed on the shaker for 30 minutes. The extracts were centrifuged at 4 °C and 16,000 x g for 30 minutes, with the resultant supernatant transferred to Amicon Ultra 0.5 mL centrifugal filters for concentrating while lowering the LDS concentration level to about 0.5% with water. The protein extracts were stored at −80 °C.

2.3 Electrophoresis

Samples of the protein extracts were subjected to gradient one dimensional electrophoresis on Invitrogen NuPAGE 4–12% Bis-Tris Gels, with 10 1.5 mm wells in an Invitrogen MES (2-(N-morpholino)ethane sufonic acid) SDS buffer (2.5 mM MES, 2.5 mM Tris base, 0.005% SDS, 0.05 mM EDTA, pH 7.3) for 40 minutes at 200 V using an Invitrogen XCell SureLock ® Mini-Cell. After fixation in 50% ethanol and 10% acetic acid, the gels were stained with Coomassie Colloidal Blue stain and destained with water.

2.4 Mass Spectroscopy

The protein bands were excised from the gel in roughly 2 x 5 mm slices and were cut into 3 or 4 pieces. A minimum of 20 slices was produced from each gel well. The gel slice pieces were washed twice with 200 μL of washing buffer (0.2 M ammonium bicarbonate and 50% acetonitrile) for 45 minutes at 37 °C in a water bath with gentle agitation. The washing buffer was discarded and the slices were dried with the Eppendorf Vacufuge for 5 minutes at 45 °C. The gel pieces were swelled with 19 μL of digestion buffer A (0.2 M ammonium bicarbonate and 5mM CaCl₂), 1 μL of 0.4 μg/μL trypsin, and sufficient digestion buffer B (0.2 M ammonium bicarbonate and 10% acetonitrile) to bring the total liquid volume to 70 μL. In order to ensure proper mixing, the solution was centrifuged at 12,000 x g for 3 minutes prior to incubating the samples overnight in a water bath at 37 °C with gentle agitation. Twice the samples were acidified to 1% acetic acid, centrifuged at 12,000 x g for 3 minutes, the supernatant was removed, and the volume of combined supernatants was reduced to about 30 μL with the Vacufuge. Any remaining gel pieces were removed by microfiltration using Millipore Ultrafree-MC-HV centrifugal filters.

The digested samples were subjected to liquid chromatography tandem mass spectrometry (LC-MS/MS) at the Analytical Proteomics Laboratory of the University of Kansas. Used was a NanoAcquity chromatographic system (Waters Corp., Milford, MA) with Binary Solvent Manager with mobile phase A and B of 0.1 % Formic acid (Water) 0.1 % Formic acid (ACN), respectively, to produce a gradient with solvents of highest purity (Optima LC-MS grade, Fisher Scientific) and delivering analytes through a reverse-phase column (ThermoScientific Acclaim PepMap300) to an electrospray source of an LTQ-FTICR mass spectrometer (ThermoFinnigan, Bremen, Germany). A linear gradient was developed in 50 minutes with a flow rate of 10 μL/min. The mass spectrometer was operated in a data-dependent acquisition mode.

XCalibur software was used to acquire the data. Mascot software was used for database search and protein mapping. It was setup to search SwissProt Viridiplantae database. Mass tolerance was 20 ppm for precursor ions and 0.60 Da for fragment ions. Carbamidomethyl (+57) was set as a fixed modification of cysteine residues, oxidation of methionine residues and acetylation of protein N-terminus were specified as variable modifications, and trypsin was the digestion enzyme, with a maximum of 2 missed cleavages. The results were exported into Scaffold software for statistical analysis and data sharing.

The limited number of proteins identified in the SwissProt database for these three ragweed species informed our decision to use the green plant database, Viridiplantae, which provided 35480 entries, including those from ragweed. It is not uncommon for researchers performing tandem mass spectroscopy proteomic analysis on plant sources to use the Viridiplantae database, when none exists for their particular species [34, 40, 47]. Sequence comparison is consistent with the long history of evolutionary homology discovered by protein sequence similarity and is the basis for BLASTP.

2.5 Data Validation

The criterion for naming found proteins involved Scaffold (Scaffold_4.3.4, Proteome Software Inc., Portland, OR) use to validate all MS/MS based peptide and protein identification results generated from Mascot, Sequest, and X! Tandem search engines [48, 49]. Scaffold employs the Peptide Prophet algorithm [50] to generate probability scores from the three search engines to enhance the accuracy of identifying peptides. Scaffold employs the resulting peptide probabilities to correctly match peptide sequence and mass to proteins with the Protein Prophet statistical algorithm [51].

2.6 Data Analysis

In Scaffold, the peptide threshold identification filter was established at 50.0% probability, the protein threshold identification filter was established at 95.0% probability, and the minimum number of peptides filter was set at 2. Using the Scaffold outcome, an additional criterion for including proteins in this report was 2 exclusively unique peptides (matched to no other protein in the database). Moreover, the protein had to be found in at least two separately extracted samples for a species. Additional criteria for reporting proteins were the total number of exclusively unique peptides and percentage of sequence coverage. When multiple entries were found for the same protein in a species, we selected those with the greatest number of exclusively unique peptides unless the coverage was greater with fewer. The same criteria were used when orthologs were excluded from the final data.

A minimum of 20 gel slices generated from each of 12 separately and independently extracted samples, four from each of the three plants, were subjected to LC-MS/MS analysis. To compare the protein profiles of the three plants, the Scaffold data was exported into Excel files and then converted into CSV format. The individual proteins from all 12 experiments were captured by name and accession number from the CSV file and aligned by protein name using a JAVA file designated as ProcessCSVCount, which also counted the incidences for a protein found in each of the ragweed plants. These counts were used to assign proteins among seven categories of All, Common, Giant, Western, Common & Giant, Common & Western, and Giant & Western. Visual Basic (VB) programs were written to accomplish the latter assignment. Additionally, other VB programs were created to manage the data, including eliminating entries that had fewer than 2 exclusively unique peptides and generating the non-exclusive assignment of proteins among the three plants

3. RESULTS and DISSCUSSION

The above criteria allowed protein assignment among the three kinds of ragweed pollen. A total of 257 unique proteins (redundancies removed) were found in the SwissProt database. When orthologs were eliminated, the number of found proteins reduced to the 229 that are displayed in Supplementary Table 1. UniProt lists a total of 142 proteins (99 short, 26 giant, and 17 western) found among the three plants. When the redundant UniProt entries and associated viruses and other pathogens were removed, the number of proteins for the plants became 57 (34 short, 18 giant, and 5 western). This investigation not only expands the protein profile of the ragweed species but also has the distinction of employing proteomics, an alternative to the often used nucleotide tools.

Table 1 displays a selection of the 229 proteins found through this LCMS/MS investigation. The presented 59 proteins are among those not previously reported to be found in the pollen of any of the three ragweed species. Having an average sequence coverage of 28%, these selected proteins also represent a wide range of protein function, and 66% of them were found for all three species. The information presented in the table by column has headings of protein name, the name of the database plant which yielded the protein sequence, the accession number, molecular mass of the protein, the number of exclusively unique peptides, the percent probability of protein identification (PID), total spectra, percentage coverage of the database sequence by the peptides, whether the protein was identified for the common/short, giant, and western plants, and the putative function in the cell. A broad array of protein types was identified, such as, metabolic enzymes for carbohydrate, amino acids, lipids, citric acid cycle, proteins, and nucleic acids. The list includes, anti-oxidant, regulatory, signaling, structural and motion, electron transport and ATP synthases, nucleosomes, proteasomes, folding, stress response, chaperones, translation, transcription, and transport (intra and intercellular) functions. The entire list of 229 proteins can be found in the supplemental material as Table S1.

Table 1.

Proteins Newly Assigned to the Ragweed Plant and Their Properties^{^}

PROTEIN NAME	DBSP	Num	Mr	EUP	PID%	TS	COV%	C	G	W	PUTATIVE FUNCTION
14-3-3 protein 6	S. lycopersicum	P93211	28938	2	100	24	41	1	1	1	Metabolic Regulation
17.3 kDa class I heat shock protein	G. max	P02519	17347	5	100	5	31			1	Protein Folding
2,3-bisphosphoglycerate-independent phosphoglycerate mutase 1	A. thaliana	O04499	60580	4	100	9	12	1	1	1	Glucose Metabolism
26S protease regulatory subunit S10B homolog B	A. thaliana	Q9MAK9	44758	4	100	5	16	1		1	Protein Catabolism
40S ribosomal protein S13	G. max	P62302	17141	4	100	7	30	1		1	Translation
5-methyltetrahydropteroyltriglutamate--homocysteine methyltransferase	C. roseus	Q42699	84860	2	100	14	14	1		1	Amino Acid Metabolism
6-phosphogluconate dehydrogenase, decarboxylating 1	S. oleracea	Q94KU1	53247	3	100	9	16	1	1	1	Gluconate Metabolism
60S ribosomal protein L11	O. sativa. ind	A2YDY2	20836	5	100	6	28	1	1	1	Translation
ADP,ATP carrier protein 1, mitochondrial	G. hirsutum	O22342	42095	2	99	8	11	1	1	1	Membrane Transport
ADP-ribosylation factor 1	S. bakko	O48649	20609	8	100	14	58	1	1	1	Protein Transport
ATP synthase subunit alpha, mitochondrial	H. annuus	P18260	55488	5	100	33	52	1	1	1	ATP Synthesis
ATPase 8, plasma membrane-type	A. thaliana	Q9M2A0	104136	5	100	35	18	1	1	1	ATP Synthesis
Aconitate hydratase 1	A. thaliana	Q42560	98153	4	100	18	20	1	1	1	Citrate Metabolism
Adenosylhomocysteinase	T. aestivum	P32112	53438	2	100	10	18	1		1	One Carbon Metabolism
Adenylate kinase A	O. sativa.jap	Q08479	26407	2	100	2	10			1	Nucleotide Metabolism
Alpha-1,4-glucan-protein synthase [UDP-forming]	Z. mays	P80607	41206	3	100	17	35	1	1	1	Cellulose Biosynthesis
Aspartate aminotransferase, cytoplasmic	D. carota	P28734	44175	5	100	5	12		1	1	Amino Acid Metabolism
Calmodulin	T. aestivum	P04464	16847	6	100	20	74	1	1	1	Enzyme Mediator
Calreticulin-2	A. thaliana	Q38858	48159	3	100	8	17			1	Protein Folding
Cell division cycle protein 48 homolog	G. max	P54774	89772	6	100	25	30	1	1	1	Cell Cycle Metabolism
Citrate synthase, mitochondrial	F. ananassa	P83372	52284	3	100	8	17	1	1	1	Citrate Metabolism
Clathrin heavy chain 2	A. thaliana	Q0WLB5	193276	6	100	31	14	1	1	1	Endocytosis
Cytochrome c oxidase subunit 2	A. thaliana	P93285	29675	4	100	5	21	1	1	1	Redox-Electron Transport
Dihydrolipoyl dehydrogenase (Fragment)	S. lycopersicum	P80503	3912	2	100	2	78		1	1	Redox Alpha Keto Acids
Elongation factor 1-alpha	O. sativa.jap	O64937	49294	3	100	25	40	1	1	1	Protein Biosynthesis
Eukaryotic initiation factor 4A-1	O. sativa.jap	P35683	47088	2	100	13	36	1		1	Translation
Fructokinase-2	O. sativa. ind	A2YQL4	35518	2	100	5	12	1	1	1	Starch Biosynthesis
GTP-binding protein SAR1A	A. thaliana	O04834	22032	5	100	12	54	1	1	1	Vesicle-mediated Transport
Glutamate decarboxylase	P. hybrida	Q07346	56726	4	100	13	20	1	1	1	Amino Acid Metabolism
Glyceraldehyde-3-phosphate dehydrogenase 2, cytosolic	Z. mays	Q09054	36542	2	100	10	32	1	1	1	Glycolysis
Guanosine nucleotide diphosphate dissociation inhibitor	A. thaliana	Q9LXC0	49551	6	100	19	21	1	1	1	Regulate Transport
Heat shock protein 81-1	O. sativa. ind	A2YWQ1	80198	4	100	35	36			1	Stress Response
Histone H4	G. max	P0CG89	11410	9	100	13	60	1	1	1	Nucleosome Assembly
Isocitrate dehydrogenase [NADP], chloroplastic (Fragment)	M. sativa	Q40345	48385	4	100	9	21	1	1	1	Glyoxylate Cycle
Luminal-binding protein	S. lycopersicum	P49118	73237	2	100	41	41	1	1	1	ER Chaperone
Malate dehydrogenase 1, mitochondrial	A. thaliana	Q9ZP06	35805	5	100	7	21	1	1		TCA Cycle
Mediator of RNA polymerase II transcription subunit 37f	A. thaliana	Q39043	73563	2	100	48	38	1	1	1	Transcription Regulation
NADH dehydrogenase [ubiquinone] iron-sulfur protein 1, mitochondrial	A. thaliana	Q9FGI6	81526	6	100	6	12	1	1	1	Electron Transport
Nucleoside diphosphate kinase B	F. bidentis	P47920	16200	2	100	10	36	1	1	1	Nucleotide Biosynthesis
Peptidyl-prolyl cis-trans isomerase	C. roseus	Q39613	18285	4	100	8	44	1	1	1	Protein Folding
Phosphoenolpyruvate carboxylase 1	F. trinervia	Q9FV66	110418	2	100	13	15	1	1		Carbon Fixation
Phosphoglucomutase, cytoplasmic 2	Z. mays	P93805	63042	4	100	10	21	1	1	1	Glucose Metabolism
Phosphoglycerate kinase, cytosolic	N. tabacum	Q42962	42365	4	100	17	36	1	1	1	Glycolysis
Phospholipase D alpha 1	C. cardunculus	Q70EW5	91786	5	100	15	16	1	1	1	Lipid Catabolism
Probable mediator of RNA polymerase II transcription subunit 37c	A. thaliana	P22954	71388	3	100	27	31	1	1	1	Transcription Regulation
Proteasome subunit alpha type-5-A	A. thaliana	O81149	25947	4	100	4	23	1	1		Protein Catabolism
Putative vesicle-associated membrane protein 726	A. thaliana	Q9MAS5	24874	4	100	6	17	1	1		Exocytosis
Pyruvate decarboxylase 2	N. tabacum	P51846	67057	2	100	7	10	1	1	1	Aerobic Fermentation
Ras-related protein RABB1c	A. thaliana	P92963	23165	7	100	7	42	1	1	1	Protein Transport
Ribulose bisphosphate carboxylase small chain, chloroplastic	H. annuus	P08705	20212	4	100	7	30			1	Reductive Pentose Phosphate Pathway
Soluble inorganic pyrophosphatase	O. sativa. ind	A2X8Q3	24160	3	100	8	16	1	1		Phosphate Metabolism
Succinate dehydrogenase [ubiquinone] flavoprotein subunit 1, mitochondrial	A. thaliana	O82663	69658	3	100	9	15	1			TCA Cycle
Superoxide dismutase [Cu-Zn]	S. canadensis	O04996	15436	3	100	4	22	1	1	1	Antioxidant Defense
Triosephosphate isomerase, cytosolic (Fragment)	L. sativa	P48493	20540	5	100	9	50	1	1	1	Glucose Metabolism
Tubulin alpha chain	P. dulcis	P33629	49653	11	100	13	26	1	1	1	Microtubule Based Process
UDP-glucose 6-dehydrogenase 4	A. thaliana	Q9FM01	53098	2	100	6	15	1		1	Carbohydrate Metabolism
UTP--glucose-1-phosphate uridylyltransferase	S. tuberosum	P19595	51876	2	100	10	18	1	1	1	UDP-Glucose Metabolism
V-type proton ATPase catalytic subunit A	G. hirsutum	P31405	68524	2	100	33	48	1		1	ATP Metabolism
Vesicle-fusing ATPase	A. thaliana	Q9M0Y8	81490	10	100	13	16	1	1	1	Golgi Vesicle Docking

^{^}Column abbreviations: DBSP, Num, Mr, EUP, PID%, TS, COV%, C, G, W are: database species, accession number, molecular mass, number of exclusively unique peptides, percent probable protein identification, total spectrum count, percentage coverage, and common/short, giant, and western ragweed plants; numbers in columns C, G, W indicate protein presence.

The PID values presented in Table 1 are calculated protein identification probabilities, rounded to the ones-digit. The calculations done in Scaffold used Bayesian statistics to combine the scores of three search engines (Mascot, Sequest, X! Tandem). The outcome operated on by the Scaffold version of the Protein Prophet yields the PID score.

Certificates of analysis from Greer Labs for each of the three species show less than 1% contamination from other sources. The three plants were harvested from different locales and different environments. There are preferences in natural environment; giant tends to grow in moist areas, common in multiple types of drier soils, and western in dry uplands [52]. The pollen from all sources was processed and stored in the same manner. Results of studies done on a specific plant grown at the same location but subjected to different environmental treatments have found differential proteome effects [47, 53]. To our knowledge, the three species investigated in this research grew under native conditions of the site and were not subjected to environment experiments. Still, it is not impossible the proteome reported here bears witness to some effects by growing at different sites. Whether the differences found are more a reflection of the species native genome expression or the locale environment is not determinable by this work. Protein expression response to an identical stress could be similar or individualistic and perhaps a reflection of the native habitat.

As shown in Figure 1, the molecular masses of database proteins assigned to the ragweed species match very well those of the gel slices from which the data arose. The scatter about calculated lines is quite small with an average R² value of 0.9885, especially considering that the plotted data came from three different gels and species. Such excellent agreement further supports the findings of the presented proteome results.

Fig. 1.

Assigned Protein and Gel Slice Molecular Mass from western, giant, common/short

Considering the results reported in Table S1, assessment of exclusive protein distribution among the species reveals both similarities and distinctions. Of the 229 unique proteins, 104 are assigned to all three plants. Common and western are the most similar, sharing 26 proteins; also shared are 19 and 21 proteins for common and giant and giant and western, respectively. Fifty-nine proteins are assigned exclusively to a single species: 10 to common, 4 to giant, and 45 to western, the only perennial of the three. A non-exclusive assignment yields a more comprehensive protein profile for each plant, with western being the most complex with 196, followed by common with 159, and giant with 148 proteins.

It is well known that the pollen of A. artemisiifolia contains several allergens of which UniProt lists 11 reviewed proteins. Among the reviewed are five pectate lyases better known as Amb a 1.1, 1.2, 1.3, 1.4, and Amb a 2, now Amb a 1.5; all are isoallergens. The remaining six reviewed proteins include Amb a 3, a plastocyanin; Amb a 5, an allergen; Amb a 6, a non-specific lipid transfer protein; and Amb a 8, having three forms of profilin. The UniProt list of non-reviewed proteins includes three calcium-binding proteins that appear to be the equivalent of the allergens, polcalcins Amb a 9, composed of 2 isoforms, and Amb a 10. A cysteine protease, probably Amb a 11, and an Art v 1 homolog, also known as Amb a 4 are among the non-reviewed allergens. In contrast, the known allergens are limited for the other two ragweed plants; UniProt shows Amb t 5 for A. trifida and Amb p 5a and 5b for A. psilostachya

The Allergen Nomenclature site (Allergen.org) lists the following ragweed pollen allergens: 11 for short (Amb a 1- Amb a 11), 1 for giant (Amb t 5), and 2 for western (Amb p 5a and 5b isoforms). Allergen.org manages the allergens somewhat differently from UniProt, subsuming all isoforms of Amb a 1 (the pectate lyases) under the major numeral. The site also contains allergen Amb a 7 a plastocyanin, not found in UniProt.

For the three plants, our results of found allergens, shown in a Table 2, are somewhat broader. All five of the pectate lyases, Amb a 3, Amb a 5, Amb a 6, and profilin (Hel a 2) and profilin-3 (Amb a 8), were found for short ragweed. If a less restrictive criterion were applied, we also found profilin-2 once in all three species. For giant ragweed Amb t 5 was found, but also were some of the pectate lyases (Amb a 1.2, 1.4, and Amb a 2), profilin and profilin-3. Pectate lyases Amb a 1.2, 1.3, and Amb a 2 were found for western ragweed as well as Amb a 6 and profilin-3. These findings imply that although short ragweed may be the primary source of allergens, giant and western might also contribute to the symptoms of hay fever.

Table 2.

Pollen Allergen Species Distribution

Amb^	IUIS%	Protein Name	Common*	Giant*	Western*
1.1	1.0101	Pectate Lyase 5	2
1.2	1.0201	Pectate Lyase 1	2	2	2
1.3	1.0301	Pectate Lyase 2	2		2
1.4	1.0401	Pectate Lyase 3	2	2	1
2	1.0501	Pectate Lyase 4	2	2	2
3	3.0101	Plastocyanin	3
5	5.0101	Pollen Allergen Amb a 5	3
5 t	5.0101	Pollen Allergen Amb t 5		2
6	6.0101	NS Lipid Transfer Protein	4		4
Hel a 2	2.0101	Profilin#	2	2
8	&	Profilin-3	4	4	4
Hev b 9	9.0101	Enolase-1$	1	4

^{^}Amb a is the database species unless otherwise specified; 5 t is giant; NS is non-specific;

^%International Union of Immunological Societies;

^#from Helianthus annuus;

^$from Hevea brasiliensis;

^*the numbers correspond to occurrences of the proteins for each of the ragweed plants; & not defined in IUIS

We also found a probable calcium-binding protein (CBP) with a 3 EF-hand, which may be related to the polcalcin allergens of short ragweed. The evidence for the presence of the CBP is very compelling, having been found for all three ragweed plants in all twelve experiments. An NCBI Blastp search with this 3 EF-hand protein of 148 residues as the query sequence revealed a sequence identity of 83% with an Artemisia vulgaris polcalcin, a protein with similar length of 149 residues. This striking similarity of the 3 EF-hand protein with the Artemisia vulgaris polcalcin suggests the discovered probable calcium- binding protein might be a member of the polcalcin family. The CBP (148 residues) might be viewed Amb a 10-like (160 residues) as the two proteins show 24% identity in a CLUSTAL Omega alignment done at UniProt.org. Interestingly, an additional alignment query revealed the polcalcin-like Amb a 10 allergen with a 4 EF-hand and 160 residues shared 23% identity with the polcalcin allergenic protein Amb a 9, a 2 EF-hand protein with 83 residues. Consideration of these sequence similarities might suggest a possible allergenic character for the probable CBP, even though its sequence shows just 15% identity to the much shorter Amb a 9 protein. Whether CBP has allergenic properties will require immunological evidence. Both Amb a 9 and Amb a 10 are listed on Allergen.org as separate short ragweed allergens.

Our results show the well-known latex Hevea brasiliensis allergen enolase-1 was assigned to giant ragweed multiple times but only once in the short plant. We also found in all three ragweed plants, the Zea mays enolase-1 having 89% identity to that of H. b. Additionally, the enolase of Ricinus communis was found multiple times in the short and western plants but only once in giant; it displayed a 95% identity to the H. b. enolase-1. This high sequence identity is compelling evidence to consider these two enolases as possible allergens, worthy of further investigation. Our results are consistent with those of Bordas-Le Floch et al. [41] who demonstrated an enolase allergen in short ragweed. Babbitt et al. [54] defined an enolase fingerprint motif, which can be found in H. b. [55]. In a UniProt alignment of all three enolases, we found the fingerprint motif runs from S380 to D388 for H. b. and R. c. and includes active site residue H381; the motif is shifted one residue toward the C-terminal residue for Z. m., having a sequence one residue longer. The significant identity R. c. and Z. m. demonstrate with the H. b. enolase-1 could imply an enolase contribution to allergic reactions inducible by all three ragweed plants.

Table 2 contains a list of some Ambrosia allergens found in Allergen.org and revealed through our LCMS/MS study of the pollen from all three ragweed species. Most are widely considered known allergens of short/common ragweed; however, as observed in this table some also appear in the data for the other two species. Enolase-1, a known allergen in Hevea brasiliensis, shown as Hev b 9, is included in Table 2 based on the data providing evidence for its presence in these ragweed plants as well. Profilin, Amb a 8, is represented in the table by Helianthus annuus, in the same family as Ambrosia, and appears as allergen Hel a 2. The table also includes Profilin-3, known allergen of A. artemisiifolia, and is assigned to all three species. Interestingly, Profilin-3 occurs in the UniProt and NCBI protein databases, but is not found at Allergen.org. The numbers given in the last three columns of Table 2 are the number of occurrences of the protein for each species.

The Bordas-Le Floch paper reporting shotgun LC MS/MS of short ragweed pollen found 573 proteins using the criterion of 1 unique peptide. Their list, somewhat different from ours, seems to include identical UniProt homolog entries as well as homologs of the same name but different UniProt identification entries. Our list eliminates all duplicates by name whether found in the ragweed plants or homologs in other species, as well as name identical UniProt entries. When the authors used the more stringent criterion of the total number of sequenced peptides being equal to or greater than 5 their list reduced to 328 proteins [41].

Employing the 1 exclusively unique peptide criterion on our short ragweed data we report 681 homologs, which represent 378 distinct proteins. It is difficult to compare these two sets of data as we seem to be using somewhat different criteria. Although it is possible these two ways of reporting data are the same, the Bordas-Le Floch paper appears to be using total unique sequenced peptides, whereas our results contain exclusively unique peptides. The outcome of applying exclusivity and elimination of homologs is our reported 229 distinct proteins.

Although The Bordas-Le Floch paper and ours report many of the same proteins, missing from both lists are both RNA and DNA polymerases; however, we do report finding transcription factors. Both papers, theirs and ours, report initiation and translation proteins. They report the inability to find Amb a 3 by either transcriptome or proteome analysis, but, instead, an Amb a 3-like protein was identified. Our proteomic analysis revealed the presence of Amb a 3 in short pollen, but not giant or western.

We used Gene Ontology (GO) Annotations from NCBI to display broad categories of biological processes associated with greater than 98% of the distinct 229 proteins. Slightly under 50% of GO designations fell into a broad metabolism category including biosynthesis of cell wall, proteins, and nucleotides among others but not including regulation. Regulation (metabolic, cell cycle, enzyme chromatin silencing, transport, transcription, and so forth), transport (protein, ion, membrane, electron, cytosis, and so forth) protein folding (heat shock, chaperone, isomerase, and others) and translation (ribosomes and initiation factors) constituted another 50% with additional contributions from microtubule processes and nucleosome assembly. There were only three entries -Amb a 5, Amb t 5, and a protein fragment- for which no GO assignment could be found.

The proteome for all three ragweed species shows important similarities and distinctions. Similarities are demonstrated by 45% of the found proteins shared by all. Notable are the differences whether one considers what is the unique for each or the number of proteins in the non-exclusive distribution category among the plants. The finding of a protein only in one of the three species could have multiple explanations. It is possible the other two plants may not express the protein; alternately, there could be differences in expression level; also harvesting of pollen granules might be time and environment dependent; although the pollen extraction protocol was common and samples of all species were simultaneously extracted on four different occasions, variations are not impossible. The somewhat higher similarity shown between short and western ragweed may reflect their likeness of appearance. It is difficult to distinguish them on the basis of foliage, unless one is expert in the field; however, it is possible to do so by the root system. Being an annual short has a tap root not shared by the perennial west.

We found seven categories of allergens distributed among the three plants. The categories include all five pectate lyases, Amb a 3, Amb a 5, Amb t 5, Amb a 6, profilin and profilin-3 (Amb a 8). Although we were unable to find Amb p 5 at the two peptide criterion, we did find it in two experiments at the one peptide level. Evidence for the enolase-1 of H.b. or its equivalent was found for giant and common ragweed. Sequence similarity evidence for the presence of possible new allergens, enolases and polcalcin-like, was found for all three species. The identity information for the enolases is quite strong; that for the CBP polcalcin-like protein is less so.

4. CONCLUSION

Overall, in comparison to UniProt, this work has added to the protein knowledge for all three plants: short 159, giant 148, and western 196. The contribution to the protein knowledge base of the giant and western plants is most notable. Not only did we report a new antigen enolase-1 for some ragweed plants but also we found not previously reported antigens for giant and western; both were found to have some pectate lyases, Amb a 2, and profilin; Amb a 6 was found in western. Although short ragweed may still be the most important source of allergies, there is now evidence for a greater role of giant and western plants in the hay fever malady.

Highlights.

• Total of 229 proteins representing multiple cellular protein functions identified

• Found a minimum of 59 proteins not previously reported for any ragweed plant

• Known allergens of short ragweed identified for giant and western species

• Newly found enolase-1 allergen for some ragweed species

• Potential allergens proposed based on sequence similarity