Mass spectrometric analysis of electrophoretically separated allergens and proteases in grass pollen diffusates

Background Pollens are important triggers for allergic asthma and seasonal rhinitis, and proteases released by major allergenic pollens can injure airway epithelial cells in vitro. Disruption of mucosal epithelial integrity by proteases released by inhaled pollens could promote allergic sensitisation. Methods Pollen diffusates from Kentucky blue grass (Poa pratensis), rye grass (Lolium perenne) and Bermuda grass (Cynodon dactylon) were assessed for peptidase activity using a fluorogenic substrate, as well as by gelatin zymography. Following one- or two-dimensional gel electrophoresis, Coomassie-stained individual bands/spots were excised, subjected to tryptic digestion and analysed by mass spectrometry, either MALDI reflectron TOF or microcapillary liquid chromatography MS-MS. Database searches were used to identify allergens and other plant proteins in pollen diffusates. Results All pollen diffusates tested exhibited peptidase activity. Gelatin zymography revealed high Mr proteolytic activity at ~ 95,000 in all diffusates and additional proteolytic bands in rye and Bermuda grass diffusates, which appeared to be serine proteases on the basis of inhibition studies. A proteolytic band at Mr ~ 35,000 in Bermuda grass diffusate, which corresponded to an intense band detected by Western blotting using a monoclonal antibody to the timothy grass (Phleum pratense) group 1 allergen Phl p 1, was identified by mass spectrometric analysis as the group 1 allergen Cyn d 1. Two-dimensional analysis similarly demonstrated proteolytic activity corresponding to protein spots identified as Cyn d 1. Conclusion One- and two-dimensional electrophoretic separation, combined with analysis by mass spectrometry, is useful for rapid determination of the identities of pollen proteins. A component of the proteolytic activity in Bermuda grass diffusate is likely to be related to the allergen Cyn d 1.


Introduction
In most economically-developed countries, at least 20% of the population suffers from IgE-mediated (Type I) allergic diseases, typically manifesting as asthma or seasonal rhinitis/conjunctivitis. In the majority of patients, this is related to sensitisation to airborne allergens [1,2] but the mechanisms by which exposure triggers an allergic response remain incompletely understood. Nor is it clear why the incidence of allergic asthma and rhinitis in economically-developed countries, including Australia, is so high. Although genetic and environmental factors both influence induction of an IgE response to an inhaled antigen [3], environmental exposure to allergens is likely to be particularly important; for example, the prevalence of asthma and rhinitis in immigrants to Australia increases significantly with the length of stay [4].
Tight junctions between airway epithelial cells apparently constitute a physical barrier between inhaled antigens and the immune system. If the access of allergenic proteins to the subepithelial antigen-presenting dendritic cells is facilitated as a consequence of breaching the integrity of the airway epithelial barrier at the time of initial exposure, for example by concurrent infection or exposure to environmental pollutants, an IgE response may be stimulated [5]. In this context, it is noteworthy that several important allergens of the house dust mite [6,7], cat allergen [8] and some fungal allergens [9] have intrinsic protease activity. Exposure to the serine protease major mite allergen Der p 1 disrupts epithelial tight junctions, increases epithelial permeability and may facilitate allergen delivery [10][11][12]. Experimental studies in vivo suggest that the protease activity of Der p 1 promotes induction of an IgE response, to both Der p 1 and bystander antigens [13,14].
Airborne allergens derived from plant pollens are important triggers for asthma and rhinitis. Pollens do contain a variety of enzymes, including proteases. Because 20-25% of the pollen mass is released rapidly on hydration [15], very high concentrations of pollen solutes are likely to be achieved when pollen grains are deposited on mucosal surfaces. However, there is little information about whether allergenic proteins of pollens have intrinsic protease activity [16]. Proteases from a limited number of allergenic pollens have been purified (e.g. [17][18][19]), but to date there is only one set of reports suggesting that a pollen allergen might have protease activity [20,21].
We previously showed that proteases released by a variety of allergenic pollens cause concentration-dependent detachment of airway epithelial cells in culture [22]. Significantly, these proteases were not efficiently blocked by major endogenous antiproteases, including a1-proteinase inhibitor and secretory leucocyte protease inhibitor. Thus, it is plausible that disruption of the epithelial barrier by proteases in pollens might promote sensitisation, not only to the protease but also to simultaneously released pollen proteins with no intrinsic enzymatic activity. Our preliminary characterisation of the proteolytic activity in diffusates of various allergenic pollens revealed that some (e.g. Kentucky blue grass) exhibited high substrate preference for Arg and Lys while others (e.g. ryegrass, Bermuda grass, ragweed) cleaved a Cys-containing substrate most rapidly and were also associated with marked preference for hydrophobic amino acids Leu and Met. These patterns were not mutually exclusive because Acacia pollen diffusate exhibited an overlapping profile of activities [23].
In the present study, we sought to directly test the relationship between pollen allergens and proteases. We employed a micro-plate assay using a fluorescent substrate, as well as gelatin zymography, to assess peptidase and protease activity present in pollen diffusates. Pollen diffusate proteins were separated by one-and two-dimensional SDS/PAGE. Proteolytically active bands/spots were then identified after "in gel" tryptic digestion, mass spectrometry and database searches.

General
Reagents and chemicals were analytical grade (Sigma, St. Louis, MO, USA; BioRad, Hercules, CA, USA) and solvents were HPLC grade (Mallinckrodt, Clayton South, Vic., Australia). SDS/PAGE/Western blotting were performed using a Mini Protean II apparatus (BioRad, Hercules, CA) with 12 or 15% gels and a Tris/Tricine buffer system [24]. Twodimensional PAGE was performed using Immobiline IPG strips (7 cm) with a pH range of 3-10 (Amersham, Biosciences, UK). The monoclonal antibody known as IG 12, to the allergen Phl p 1 of timothy grass (Phleum pratense), was a gift from Drs Kay Grobe and Arnd Petersen [20]. Protein concentrations were determined using a micro BCA assay (Pierce, Rockford, IL). Amino acid sequences were compared using the best fit program in the Genetics Computer Group Package, Version 8, Madison, WI, USA http://www.angis.org.au.

Two-Dimensional SDS/PAGE and Zymography
To achieve satisfactory results, Bermuda grass pollen diffusate (250 µl, ~ 100 mg/ml) was concentrated and desalted to 50 µl using a Centricon 10 kDa cut-off ultracentrifugation device (Millipore, USA). The concentrated diffusate was dissolved in 75 µl of IPG gel rehydration buffer containing 8 M urea, 2% w/v CHAPS, 2% carrier ampholytes (100 × Bio-lyte, pH range 3-10 Bio-Rad, USA) and traces of Bromophenol blue. Reducing agents such as DTT or tributyl phosphine were excluded from the rehydration buffer as they destroyed the proteolytic activity present in the diffusate. Separation of the proteins in the first phase was followed by a short equilibration step of the IPG strip in SDS equilibration buffer, after which proteins were separated by SDS-PAGE and/or gelatin zymography as described above.

Determination of Diffusate Peptidase Activity for a Peptide Substrate
Diffusates from grass pollens were assayed for peptidase activity with NBAMC, a fluorogenic substrate with tryptic specificity. Profiles typical of fluorescence vs time for Kentucky blue diffusate and trypsin after incubation for 1 hour are shown in Fig. 1A. All diffusates digested the substrate immediately upon addition of the diffusate and gave similar results. Using the same diffusate protein concentration, the relative proteolytic activity of rye grass diffusate was comparable to that of trypsin after 1 hour of incubation, whereas the activity of Kentucky blue grass and Bermuda grass diffusates was approximately 75% less. The fluorescence of each diffusate, reached after 1 hour, was scaled to 100 as were changes in activity observed after addition of an inhibitor (Fig 1B). This approach allows ready comparison of activities and similar results were obtained in repeated experiments (n = 3, not shown). All proteolytic activity was almost completely eliminated by the complete protease inhibitor cocktail or the effective serine protease inhibitor AEBSF. Trypsin-like activity was confirmed by the almost complete inhibition with TLCK, whereas inhibition by TPCK, a chymotrypsinspecific inhibitor, was very modest (Fig 1). Little inhibition was observed with the cysteine protease inhibitor iodoacetamide on Bermuda grass diffusate, whereas the activity of Kentucky and rye grasses was inhibited by 4 0%, suggesting that these diffusates contain cysteine peptidase activity. The cysteine protease papain was completely inhibited by iodoacetamide (not shown). The metalloprotease inhibitor EDTA was inactive.
These results indicate that, under the conditions of assay, the predominant peptidase activity for the substrate NBAMC present in these pollen diffusates is due to serinetype enzymes, but cysteine protease activity was also detected in Kentucky and rye grass diffusates. The total fluorescence levels observed after 1 hour incubation varied by only ~ 75% between the pollen diffusates (not shown). It was not possible to determine relative levels of proteases present in the diffusates from these experiments. These studies indicate that all the pollen diffusates examined here possessed intrinsic activity for Lys/Arg substrates. However, whereas our previous study had suggested that trypsin-like peptidase activity was a minor component of the activity in rye and Bermuda grass and that the activity present in Kentucky blue grass was almost entirely specific for the basic amino acids [23], the results from the current experiments indicate otherwise. We believe this might be a consequence of the relative rates of lysis of different peptide substrates being possibly influenced by the assay conditions or substrate. The previous study employed an assay in which the initial pH of the reaction mixture was markedly alkaline and activity was revealed by an indicator dye sensitive to the progressive fall in pH. Trypsin-like peptidase activity in pollen diffusates is well established. Mesquite (Prosopis velutina) and ragweed (Ambrosia artemisiifolia) pollens contain peptidases with trypsin-like and chymotrypsin-like specificities [18,19]. However, the function of these pollen enzymes remains largely unknown.

Comparison of proteolytic activity towards a peptide substrate in grass pollen diffusates
In addition to peptidases, pollen diffusates may contain novel endoproteases with sequence identity to the group 1 allergens [20]. Therefore, we assayed the 3 pollen diffusates for protease activity using gelatin zymography.

SDS/PAGE and Zymography
All pollen diffusates analysed contained relatively high protein concentrations (~ 2-5 mg/ml) and SDS/PAGE analysis showed intensely-stained bands of M r ranging from <10,000 to ~ 100,000 (Fig 2). Previous reports have demonstrated numerous bands within each pollen diffusate, corresponding to allergenic proteins [28]. Diffusates of Kentucky blue (Fig 3, lanes 2-6), rye (Fig 3, lanes 7-11) and Bermuda grasses (Fig 3, lanes 12-16) were tested for proteolytic activity using gelatin and casein zymography. Gelatin zymograms generally revealed more intense activity and were used for most experiments. All zymograms contained several intense negatively-stained bands at M r 9 5,000 (Fig 3, lanes 2, 7 and 12) possibly corresponding to a common protease. A proteolytic band at M r ~ 65,000 was evident in rye grass diffusate and Bermuda grass diffusate also contained an additional proteolytic band at M r 3 5,000 (Fig 3, lane 12). All proteolysis was fully inhibited by the complete protease inhibitor cocktail (Fig 3, lanes 4,  9 and 14), partially inhibited by AEBSF (Fig 3, lanes 5, 10  and 15) whereas little inhibition was observed with iodoacetamide (Fig 3, lanes 6, 11 and 16). These results suggest that these components have serine protease activity.
A monoclonal antibody specific to epitopes on the Phl p 1 allergen from timothy grass (Phleum pratense) was tested using Western blotting. One intense band M r ~ 35,000, which corresponded to the predicted M r of the major group 1 allergen, was readily detected in Kentucky blue and Bermuda grass diffusates. The immunoreactivity indicates common epitopes and suggests common functions of these allergens in pollen. The Phleum pratense group 1 allergen (Phl p 1) has 70%, 90% and 93% identity to homologous group 1 allergens derived from Poa pratensis (Poa p 1), Lolium perenne (Lol p 1) and Cynodon dactylon (Cyn d 1) respectively [29]. Group 1 allergens are pollen glycoproteins of predicted M r ~ 30,000, that are released after hydration. They are a subset of the b-expansin family of genes in plants [30]. Sequence similarities to expansins, and functional studies, showed that the group I allergen from maize pollen (Zea m 1) exhibits expansin activity specifically towards grass cell walls [30]. Expansins are defined by their characteristic function and sequence similarities. Grobe et al [20] have proposed that timothy grass group 1 allergen (Phl p 1) possesses proteolytic activity, demonstrable after pre-incubation under reducing conditions. The authors suggested that the amino acid sequence of Phl p 1 contained a cysteine protease motif similar to that found in papain. However, there are conflicting reports about whether the observed proteolytic activity is due to Phl p 1 itself or a contaminating protease [20,21,31,32]. The M r of the proteolytic band at ~ 35,000 observed in Bermuda grass diffusate was close to the observed M r of the Phl p 1 immuno-reactive band detected in the Western blot (Figs 2 and 3). Because of their high sequence identity, it is possible that the group 1 allergen of Bermuda grass (Cyn d 1) shares epitopes with Phl p 1 and may also possess proteolytic activity. SDS/ PAGE and zymograms were repeated, but it was not possible to accurately align silver-stained gels, zymograms and Western blots to determine if Cyn d 1 was the protein responsible for the proteolytic activity. Therefore, we undertook additional studies using two-dimensional gel electrophoresis and mass spectrometry as described below.

Proteomic analysis of pollen diffusates
A proteomic approach, enabling rapid identification and analysis of proteins, was employed to confirm the identities of the bands surrounding the proteolytic activity at M r 35,000. The identities of other intensely staining bands were also examined to verify that mass spectrometry was suitable for analysis of pollen proteins. Bands were excised from diffusates of Kentucky blue grass, rye grass and Bermuda grass that had been separated by SDS/PAGE and stained using CBB (Fig 4a), and were digested with trypsin. Peptides were analysed by MALDI reflectron TOF peptide mass fingerprinting and micro-LC/ESI low energy CID MS/MS. Figures 4b and 4c show MALDI reflectron TOF spectra of tryptic peptides of band 4 from rye grass and band 12 from Bermuda grass diffusates. Intense ions corresponding to protonated tryptic peptides of proteins present in each band were readily detected, with mass errors of < 1 00 ppm. Database searches using peptide masses from each spectrum indicated that the most likely protein in band 4 was the allergen Lol p 1 (rye grass) and that in band 12 was the allergen Cyn d 1 (Bermuda grass). Five additional proteins corresponding to known pollen allergens were readily identified by peptide mass fingerprinting from the 16 excised bands analysed (data not shown). These corresponded to some of the most intensely staining bands observed in the gel. While MALDI peptide mass finger printing is a sensitive, quick and specific approach, bands containing several proteins are more difficult to identify because the peptide masses may correspond to the theoretical digest of an unrelated protein leading to false positives after database searching [33]. To circumvent this problem, diffusates were reanalysed using two-dimensional electrophoresis, followed by extraction of the tryptic digests of the protein spots from the gels, analysis of the sequence tags derived from LC separation and low energy CID MS/MS, and database searching. This approach allows easy identification of peptide mixtures [27,34].

Two-Dimensional SDS/PAGE and Zymography
Two-dimensional electrophoretic separation of the concentrated diffusate yielded a number of protein spots at Mr 30,000-35,000 Da. Isoelectric focusing/gelatin zymography of the diffusate yielded diffuse proteolytically active streaks around the same region (Fig 5). Relatively weaker proteolytic activity in the two-dimensional zymograms might have been a consequence of extreme conditions of denaturation and subsequent renaturation.

Micro-LC/MS of pollen diffusates
The proteolytic digests of the protein spots from the isoelectric focusing SDS/PAGE of Bermuda grass diffusate were separated by micro C18 RP HPLC followed by auto-Gelatin zymography of pollen diffusates and sensitivity to inhibitors mated data-dependent low energy CID MS/MS analysis of the most intense multiply charged-peptide ion identified in the eluting peptides [27]. Peptide sequence tags from the entire chromatogram were used for database searches. The base peak mass chromatogram of tryptic peptides from spot 4 is shown in Figure 6a. Typical MS/MS spectra from multiply-charged peptide ions are shown in Fig 6b. After database searches of all spectra, spot 4 was identified as allergen Cyn d 1. Intense ions corresponding to sequential cleavage of the [M+3H] 3+ ion of Cyn d 1 111-129 (isoform 1) were readily identified and an almost complete series of y-type fragment ions was present (Fig 6b). The identification of multiple peptide sequence tags from the same protein allowed identification with a high probability.
For each of the excised and digested spots, micro-LC/MS followed by semi-automated database searches were per-Micro C18 RP-HPLC electrospray MS/MS analysis of tryptic peptides derived from spot 4 (Cynodon dactylon) Figure 6 Micro C18 RP-HPLC electrospray MS/MS analysis of tryptic peptides derived from spot 4 (Cynodon dactylon). Peptides from tryptic digestion of spot 4 (Fig 5a) were separated using micro C18 RP-HPLC and the eluate analysed by ESI-MS/ MS. A. Base peak mass chromatogram, each peak is labelled with the retention time, mass, charge state and identity determined by low energy CID tandem MS of the multiply-charged precursor ions. B. MS/MS spectrum of a triple charged precursor ion m/z 741.7 corresponding to the sequence ITDKNYEHIAAYHFDLSGK from allergen Cyn d 1 was identified by database searches (Mascot).
formed. Table 1 summarises proteins identified. Searches performed on the amino sequence tags determined after micro-LC separation and MS/MS permitted identification of peptides from protein mixtures, but because the current plant protein sequence databases are incomplete, in some cases the proteins identified were from other plant species, indicating that the derived sequence tags may be common to two or more species.
The spots that separated at M r 30,000-35,000 and comigrated with the observed proteolytic activity were identified by micro-LC ESI-MS/MS as the group 1 allergen Cyn d 1 from Bermuda grass.

Conclusion
SDS/PAGE and mass spectrometry proved quite useful in rapidly determining the identities of pollen proteins after digestion with trypsin. In the future, this approach will allow identification and differentiation of allergen isoforms and of post-translational modifications. Our data strongly suggest that, analogous to house dust mite and other allergens, pollen allergens may also be proteases. The major proteolytically-active band in Bermuda grass, which localised at M r ~ 35,000 in one-dimensional electrophoresis, was the group 1 allergen Cyn d 1. Similarly, the major proteolytically active spots at approximately the same Mr in two-dimensional electrophoresis were also identified as Cyn d 1, thus strongly suggesting that this allergen has intrinsic proteolytic activity. While confirmation of this result by examination of recombinant Cyn d 1 could be useful, the well documented contamination of such preparations by proteases from the expression system (e.g. yeast) [32] may limit the value of such additional experiments. High purity isolates of native Cyn d 1 may be more reliable for such assessment.