This protocol describes the removal of endogenous lipids from allergens, and their replacement with user-specified ligands through reverse-phase HPLC coupled with thermal annealing. 31P-NMR and circular dichroism allow for the rapid confirmation of ligand removal/loading, and the recovery of native allergen structure.
Many major allergens bind to hydrophobic lipid-like molecules, including Mus m 1, Bet v 1, Der p 2, and Fel d 1. These ligands are strongly retained and have the potential to influence the sensitization process either through directly stimulating the immune system or altering the biophysical properties of the allergenic protein. In order to control for these variables, techniques are required for the removal of endogenously bound ligands and, if necessary, replacement with lipids of known composition. The cockroach allergen Bla g 1 encloses a large hydrophobic cavity which binds a heterogeneous mixture of endogenous lipids when purified using traditional techniques. Here, we describe a method through which these lipids are removed using reverse-phase HPLC followed by thermal annealing to yield Bla g 1 in either its Apo-form or reloaded with a user-defined mixture of fatty acid or phospholipid cargoes. Coupling this protocol with biochemical assays reveal that fatty acid cargoes significantly alter the thermostability and proteolytic resistance of Bla g 1, with downstream implications for the rate of T-cell epitope generation and allergenicity. These results highlight the importance of lipid removal/reloading protocols such as the one described herein when studying allergens from both recombinant and natural sources. The protocol is generalizable to other allergen families including lipocalins (Mus m 1), PR-10 (Bet v 1), MD-2 (Der p 2) and Uteroglobin (Fel d 1), providing a valuable tool to study the role of lipids in the allergic response.
A survey of the allergen database reveals that allergens are found in only 2% of all known protein families, suggesting common functional and biophysical properties contribute to allergenicity1. Of these properties, the ability to bind lipid cargoes appears to be strongly over-represented among allergens, suggesting that these cargoes may influence the sensitization process1. Indeed, it has been shown that the Brazil Nut allergen Ber e 1 requires co-administration with its endogenous lipid to realize its full sensitizing potential2. These lipids could potentially stimulate the immune system directly as illustrated by the mite allergens Der p 2 and Der p 7, both of which share a strong structural homology with LPS-binding proteins3,4,5. Based on this observation it was proposed that Derp 2 and Der p 7 could bind bacterial lipids and directly stimulate the host immune system through TLR4-mediated signaling, facilitating the sensitization process5,6. It is also possible that endogenously bound lipids could alter the biophysical properties of allergenic proteins themselves. For example, the ability of Sin a 2 (mustard) and Ara h 1 (peanuts) to interact with phospholipid vesicles significantly enhanced their resistance to gastric and endosomal degradation7, while ligand binding to the major birch pollen allergen Bet v 1 altered both the rate of endosomal processing and the diversity of the resulting peptides8. This is particularly relevant to allergenicity given the correlation that has been observed between stability, T-cell epitope generation and allergenicity for proteins such as Bet v 1 and Bla g 1; the latter of which will be the subject of this work9,10.
Bla g 1 represents the prototypical member of the insect Major Allergen (MA) protein family, and possesses a unique structure composed of 12 amphipathic alpha helices which enclose an abnormally large hydrophobic cavity9,11. The available X-ray crystal structure of Bla g 1 shows electron density within this cavity consistent with bound phospholipid or fatty acid ligands; a conjecture confirmed by 31P-NMR and mass spectrometry. These cargoes were heterogeneous in nature and their composition was heavily dependent on the allergen source, with different lipid profiles observed for recombinant Bla g 1 expressed in E. coli and P. pastoris. Curiously, Bla g 1 purified from its natural allergen source (cockroach frass) contained predominantly fatty acids within its binding site, with a mixture of palmitate, oleate, and stearate being identified as its “natural” ligands9,11. The ability of Bla g 1 to retain lipids and fatty acids following multiple purification steps hinders efforts to study the protein in isolation. Conversely, it has been suggested that the natural palmitate, stearate, and oleate ligands of Bla g 1 (henceforth referred to as nMix) play a key role in both its allergenicity and native biological function9. However, these ligands are not present in Bla g 1 obtained from recombinant sources, making it difficult to assess this hypothesis. Similar issues have been observed for other lipid binding allergens such as Bet v 112,13. To facilitate the systematic study of lipid-allergen interactions we have developed a protocol through which allergens can be quantitatively stripped of their endogenously bound lipids and reconstituted in either Apo-form or loaded with specific ligands.
Allergens are most commonly purified from their natural or recombinant sources using affinity chromatography and/or size-exclusion chromatography. Here, we introduce an additional purification step in the form of high-performance liquid chromatography (HPLC) employing a reverse-phase C18 column from which the allergen is eluted into an organic solvent similar to protocols developed for fatty acid binding proteins14. The resulting protein is then subjected to a thermal annealing step in the absence or presence of fatty acids and/or phospholipids. In addition to recovering the native Bla g 1 fold, the elevated temperatures increase the solubility and accessibility of the lipid cargoes, yielding Bla g 1 in either the Apo-form or uniformly loaded with the desired hydrophobic ligand. 31P-NMR spectra of Bla g 1 purified in this manner confirmed the complete removal of endogenously bound ligands and uniform replacement with the desired compounds, while circular dichroism confirmed the successful recovery of the Bla g 1 fold. The utility of this method is highlighted in a recent work in which cargo binding was found to enhance Bla g 1 thermostability and proteolytic resistance, altering the kinetics of T-cell epitope generation with potential implications for sensitization and allergenicity9.
1. Bla g 1 cloning
2. Initial expression and purification
3. Endogenous lipid removal via reverse-phase HPLC
Time (Min) | Buffer A (%) | Buffer B (%) |
0 | 97 | 3 |
10 | 97 | 3 |
25 | 35 | 65 |
55 | 5 | 95 |
65 | 5 | 95 |
70 | 97 | 3 |
Table 1: Elution protocol for Bla g 1. Table illustrating the elution gradient employed in the isolation of Bla g 1 using a C18 HPLC column.
4. Reconstitution of Apo- and cargo-loaded Bla g 1
5. Confirming phospholipid cargo removal/loading via 31P-NMR
6. Confirming Bla g 1 folding
Using affinity chromatography, recombinant GST-Bla g 1 was readily isolated to a high level of purity (Figure 1A), producing a yield of ~2–4 mg/L of cell culture. Overnight incubation with TEV protease at 4 ˚C is sufficient to remove the GST tag, yielding the final product at ~24 kDa. Note that in this instance there is a significant amount of GST-Bla g 1 in the flow-through and wash fractions, suggesting the Glutathione resin binding capacity was exceeded. The use of more resin or multiple cycles of sample loading and elution could provide remedy for this issue.
Applying the Bla g 1 to a reverse-phase C18 column yields a distinctive elution profile (Figure 1B), with two large peaks at ~50% buffer B, and a second large peak at ~75% buffer B. SDS-PAGE analysis of the resulting fractions suggest that the former correspond to the cleaved GST tag, while the latter corresponds to Bla g 1. Occasionally a third, smaller peak will occur in the middle corresponding to residual, un-cleaved GST-Bla g 1. The presence of this un-cleaved product can be eliminated by increasing the amount of TEV employed in the cleavage reaction or extending the incubation time. While incomplete cleavage will reduce the yield, the separation obtained on the C18 column is sufficient to ensure that the purity of the final Bla g 1 product remains uncompromised. A consequence of reverse-phase HPLC is that the final protein product is eluted into an organic solvent environment. While this facilitates removal of any hydrophobic ligands, removal of this solvent via lyophilization is required, yielding a fluffy white powder (Figure 1C).
Annealing of the protein is required to reconstitute the native Bla g 1 fold and can be carried out either in the absence or presence of a lipid cargo. Addition of DMSO to the dried Bla g 1 and phospholipid cargoes prior to the refolding buffer facilitates the solubilization process, though some longer chain lipid cargoes will not fully dissolve even at elevated temperatures. However, this was not observed to impact the loading efficacy among the lipids tested in our studies (Figure 1C). Similarly, excess lipids will often precipitate out of solution or form large vesicles upon cooling, resulting in a cloudy appearance after annealing (Figure 1C). This was also not observed to effect loading efficiency, and any aggregates are readily removed through the filtration and subsequent buffer exchange steps to yield a clear, transparent solution. Despite the harsh conditions, no thermolysis was observed for Bla g 1.
Figure 1: Initial purification of Bla g 1. (A) SDS-PAGE showing the soluble protein fraction following initial lysis (S); flow-through (FT), wash (W), and elution from the glutathione-sepharose column (E); and the final Bla g 1 product following TEV cleavage of the GST tag (TEV). The HPLC elution profile of the resulting Bla g 1 product following TEV cleavage is shown in (B). A280 is shown in blue, while the elution gradient (% Buffer B) is shown in green. Fractions corresponding to the cleaved GST tag (H1, H2), residual un-cleaved GST-Bla g 1 (H3), and purified Bla g 1 (H4) are indicated with red arrows at ~50%, ~65%, and ~74% Buffer B respectively. SDS-PAGE analysis of fractions H1- H4 are shown in (A) and labeled accordingly. (C) Representative images showing Bla g 1 at various stages of the annealing process. Note that the precise and extent of precipitate formation as depicted in ii and iii is dependent on the type of lipid cargo employed. Please click here to view a larger version of this figure.
31P-NMR spectra of Apo-Bla g 1 purified in this manner show no detectable phospholipids either by NMR (Figure 2A) or thin layer chromotography (data not shown). By contrast, similar spectra obtained for Bla g 1 loaded with a distearoylphosphatidylcholine (DSPC) phospholipid show a strong peak corresponding to the phosphatidylcholine headgroup. For comparison, a representative 31P-NMR spectrum of Bla g 1 purified from recombinant E. coli without the use of the lipid removal/annealing protocol described herein (ecBla g 1) show a heterogeneous mixture of endogenous lipids extracted from the recombinant expression system (Figure 2B). Taking advantage of the quantitative nature of NMR, a standard curve can be produced using reference samples of known DSPC concentrations (Figure 2C). Comparing the 31P signal intensity obtained from DSPC-Bla g 1 against this standard curve yields a binding stoichiometry of 4.7 ± 0.5 lipids per protein; a value that compares favorably to the predicted full binding stoichiometry obtained from in silico studies and structural analysis9. Note that this technique will only detect ligands which contain a 31P nucleus such as phospholipids, lysophospholipids, lipopolysaccharides etc. However, this protocol can be easily adapted for 13C-NMR analysis. In this case, methyl-13C labeled fatty acids would be recommended due to its favorable NMR relaxation properties. Restricting isotopic labeling to a single site also facilitates spectral interpretation, as only a single peak is expected, while simultaneously reducing the cost relative to uniform 13C-labeled counterparts. An alternative approach would be to employ mass-spec to identify bound ligands, as demonstrated in a previous study which identified a mixture of fatty acids as the natural cargo of Bla g 1 isolated from cockroach frass (nBla g 1)9. However, the limited quantitation capabilities of mass spec precluded an accurate measurement of binding stoichiometry without sufficient standards.
Figure 2: Verifying lipid removal and loading of Bla g 1. (A) 31P-NMR spectra of Apo- (black) or DSPC-loaded Bla g 1 (red) prepared using the annealing protocol described in this work demonstrating the complete removal of lipids in the former, and the homogeneous loading of phosphatidylcholine (PC) lipids achieved in the latter. In contrast, Bla g 1 purified from recombinant E. coli without lipid stripping and annealing (ecBla g 1) shows a heterogeneous mixture of endogenous phosphatidylethanolamine (PE) and phosphatidylglycerol (PG) lipids when analyzed using this method (B). A representative standard curve obtained from DSPC reference samples of known concentrations is shown in (C), from which the Bla g 1 binding stoichiometry can be obtained. Figures adapted from Foo et al. (2019) and presented under the Creative Commons CC BY License9. Please click here to view a larger version of this figure.
Crystal structures of Bla g 1 reveal a unique fold consisting of 12 amphipathic alpha-helices. Circular dichroism represents a quick and convenient method to assess whether this fold has been successfully reconstituted after the annealing process. CD spectra for Apo- and lipid (nMix)-loaded Bla g 1 show minima ~220 and 210 nm indicative of a predominantly alpha-helical structure (Figure 3A). This spectrum is extremely similar to that obtained for ecBla g 1 and nBla g 1, providing further evidence that the native structure of Bla g 1 is successfully recovered. This was further confirmed through the use of 19F and 1H-15N solution-NMR, a full discussion of which is available elsewhere9. CD-based thermal denaturation assays show a cooperative loss of alpha-helical secondary structure indicative of a folded globular domain (Figure 3B). Analysis of the resulting melting temperatures (Figure 3C) show a significant increase upon nMix ligand binding. This elevated thermostability is in line with that calculated for nBla g 1, indicating that we are able to fully reproduce the natural state of Bla g 1. Note that ecBla g 1 also shows a similar, if not greater enhancement in thermostability, illustrating the potential for residual endogenously bound lipids to interfere with biophysical characterization of allergens purified using traditional FPLC-based approaches. In contrast, the ability to quantitatively remove and reload hydrophobic cargoes from allergens such as Bla g 1 provides a unique avenue to examine the role of lipids in the allergic response. Here, we describe a method to examine the influence of lipid cargoes on the structure, stability, and endosomal processing of the allergenic proteins themselves, though other avenues of study could be considered.
Figure 3: Confirming successful recovery of the Bla g 1 fold. (A) CD spectra of Apo- (black) or nMix-loaded (red) Bla g 1 purified and annealed using the protocol described herein, with minima at ~220 and 210 nm indicative of a predominantly alpha-helical structure consistent with the available X-ray crystal structure. Both Apo- and nMix-loaded Bla g 1 spectra are extremely similar with that obtained for Bla g 1 purified from recombinant E. coli (ecBla g 1, green) or from its natural allergenic source (nBla g 1, blue) without the lipid removal and annealing protocol, further supporting the successful recovery of the native structure in the former. (B) Representative thermal profiles for Apo- (black) and nMix-loaded (red) Bla g 1 showing a sigmoidal curve indicative cooperative unfolding. nBla g 1 (blue) and ecBla g 1 (green) shown as reference. The calculated melting temperatures (MT25) of Bla g 1 are shown in (C). Binding of nMix ligands (red) yields a significant increase in thermostability relative to Apo-Bla g 1 (black). This mirrors the trend observed for nBla g 1 (blue), suggesting that we are able to successfully recover the native state. The even greater stability observed for ecBla g 1 highlights the potential of endogenously bound lipids to interfere with biophysical characterization of allergens. MT25 values presented in C represent the mean value obtained from at least three independent trials. Error bars represent the corresponding standard deviation values. Figures adapted from Foo et al. (2019) and presented under the Creative Commons CC BY License9. Please click here to view a larger version of this figure.
The protocol described in this work has been successfully applied to systematically study the lipid binding properties of Bla g 1. This revealed a correlation between cargo binding, thermostability, and endosomal processing, the latter of which was correlated with decrease in the generation of a known T-cell epitope with potential implications for immunogenicity9,18. In addition to Bla g 1, other allergens such as Pru p 3 and Bet v 1 have been shown to retain their endogenously-bound cargoes when purified using standard affinity and size-exclusion chromatography methods13,19,20,21,22. These unwelcome guests could alter the biophysical and immunological properties of these proteins in a similar manner, highlighting the need for techniques to ensure complete delipidation such as the one presented here.
While the use of reverse-phase HPLC in the purification of allergens has been described previously2, coupling it with a thermal annealing protocol provides the rather unusual opportunity to reconstitute allergens with a range of natural and un-natural ligands, allowing users to probe lipid-allergen interactions. This thermal denaturation step was found to be essential for two main purposes. First, thermal denaturation is required to facilitate ligand access to their binding cavities which, due to their hydrophobic nature, are often buried away from the aqueous solvent9,22. Secondly, hydrophobic ligands such as fatty acids and phospholipids often form larger supramolecular structures such as micelles or vesicles when placed in an aqueous environment. The concentration of monomeric, or “free” ligands available for protein binding can be approximated using the critical micelle concentration (CMC). DSPC and other long-chain phospholipids have CMC values in the nM range, indicating that there are virtually no free ligands available for Bla g 1 binding. Even short chain lipids and fatty acids have CMC’s in the low µm to mM range, indicating that a large proportion of these ligands remain in the micellar or bilayer phase23. However, the high temperatures employed in our denaturation protocol disperses these larger aggregates, facilitating binding. Previous studies have typically employed prolonged incubation periods to facilitate this process. However, the lack of a thermal denaturation/annealing process raises doubts to the efficacy of loading. For example, incubating the mite allergen Der p 5 with the fluorescent fatty acid analogue 11-(Dansylamino)undecanoic acid (DAUDA) yielded a binding stoichiometry of 0.66 despite possessing a large hydrophobic cavity on par with Bla g 124. Likewise, the binding specificity and stoichiometry of plant nsLTPs were found to vary greatly depending on whether the lipids and protein are first solubilized in methanol prior to the addition of aqueous buffer, indicating that ligand and/or binding site accessibility was a limiting factor25.
In addition to Bla g 1, we have successfully applied the same strategy to several other MA domain proteins from cockroaches and mosquito (A. aegypti), as well as Der p 2 (data not shown). We noted that both the Bla g 1 homologues and Der p 2 eluted at a different time than Bla g 1 from the C18 column (step 3.3). The elution gradients in this step may need to be optimized for other proteins. Alternatively, HPLC columns with a less hydrophobic stationary phase (e.g., C8) may be employed, though in the case of Bla g 1 the increased hydrophobicity of the C18 column was necessary to completely remove diacyl phospholipid contaminants from ecBla g 1. Despite the differences in biophysical and biochemical properties, we have found this protocol to be extremely robust and could be easily applied to other allergenic proteins. While the harsh conditions employed may present a potential limitation, the increased resilience observed for many allergens reduces its impact26,27. Indeed, several food and inhalation allergens such as Der p 2, Ber e 1, Ara a 6 and Lep w 1 have been observed to recover their structure and immunogenicity following thermal denaturation, though optimization of buffer conditions may be required28,29,30,31,32,33; for example reversible denaturation of nsLTP’s (Cor a 8) and thaumatins (Mal d 2 and Act d 2) is only observed under acidic (pH <4) conditions28,30,31. Additionally, it should be noted that the authors did not attempt to optimize either the timing or temperatures employed in the annealing protocol. It is possible that ligand solubilization and protein folding/unfolding may be achieved using a lower maximum temperature as seen with Ber e 1 for which reversible denaturation is achieved at 82 ˚C29. The use of such measures is expected to expand the range of allergens to which this protocol can be applied.
Another important consideration when adapting this protocol to other allergen systems is the concentration of ligands required during the annealing process. In the case of Bla g 1 the expected yield is ~0.25–0.4 μmol of protein per 1 L cell culture. Given the demonstrated binding stoichiometry of 8 fatty acids or 4 diacyl chain lipids per allergen, a 20–40 fold molar excess of cargo (5–10 μmol) was employed. It should be noted that the lipid binding ability of Bla g 1 and its homologues is unique; for example nsLTP’s are generally accepted to bind at most two lipid ligands25 while lipocalins have less than 1 stoichiometry34. As such, complete loading of these types of allergens may be accomplished with a smaller excess of ligands. A final consideration when adapting this protocol to other allergen systems is the presence of disulfide bonds, which can be problematic if not properly formed prior to denaturing. One possible approach would be to carry out the annealing process in the presence of a reducing agent such as 2 mM DTT. The native disulfide bonds could be subsequently re-formed through the addition of reduced and oxidized glutathione as described for the peanut allergen fragment studied by Aalberse et al.35. In this case, recovery of the correct disulfide bonding should be empirically assessed by mass spectrometry35.
In this work we describe a technique through which allergens can be delipidated and re-annealed with various phospholipid and fatty acid cargoes. However, there are many other classes of potentially immunogenic or adjuventising ligands present within common allergen reservoirs. For example, cat, dog, and mite allergens have been proposed to bind lipopolysaccharides (LPS) and other bacterial lipids from house dust36, while the Bet v 1 has been shown to extract complex flavonoids from the pollen matrix13. The protocol described in this work can be easily adapted to explore the role of these lipids in a more detailed manner. As a proof of concept we have been able to demonstrate that the hydrophobic cavity of Bla g 1 is capable of binding lipoteichoic acid (LTA) from the cell walls of gram positive bacteria, but excludes LPS from gram negative species, potentially reflecting the greater number of acyl chains in the latter9. Taking this one step further, one could utilize the thermal denaturation/annealing protocol to incorporate fluorescent probes and other non-natural fatty acid analogues into allergen proteins. Indeed, we were able to load the hydrophobic cavity of the mosquito homologue of Bla g 1 with DAUDA, opening additional avenues to examine the effects of lipid ligands on allergenic disease.
The authors have nothing to disclose.
We would like to thank Dr. Tom Kirby, Scott Gabel, and Dr. Robert London for their help and assistance throughout this work, along with Dr. Bob Petrovich and Lori Edwards for the use of their instrumentation and their assistance in generating the Bla g 1 constructs employed in this study. We thank Andrea Adams for assistance with the mass spectrometry, and Dr. Eugene DeRose for assistance with the NMR instrumentation. This research was supported by the Intramural Research Program of the NIH, National Institute of Environmental Health Sciences, Z01-ES102906 (GAM). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institute of Environmental Health Sciences.
Bla g 1 Gene | Genescript | N/a | Custom gene synthesis service. GenBank Accession no AF072219 Residues 34-216 |
Affinity purified natural Bla g 1 (nBla g 1) | Indoor biotechnologies | N/a | Custom order |
Agilent 1100 Series HPLC System | Agilent | G1315B, G1311A, G1322A | UV Detector, Pump, and Degasser |
Agilent DD2 600 MHz spectrometer | Agilent | N/a | |
Amicon Ultra-15 Centrifugal Filter Unit | Amicon | UFC-1008 | |
Ampicillin | Fisher Scientific | BP1760-5 | |
Benzonase | Sigma-Aldrich | E1014-5KU | |
Broad- band 5 mm Z-gradient probe | Varian | N/a | |
ChemStation for LC (Software) | Agilent | N/a | |
cOmplete Mini Protease Inhibitor Cocktail | Roche | 11836153001 | |
Distearoylphosphatidylcholine (18:0 PC) | Avanti Polar Lipids | 850365C | |
E. Coli BL21 DE3 Cells | New England Biolabs | C2530H | |
Freezone 4.5 Freeze Dry System | Labconco | 7750000 | |
Glutathione Resin | Genescript | L00206 | |
Glutathione, Reduced | Fisher Scientific | BP25211 | |
Isopropyl-β-D-thiogalactopyranoside (IPTG) | Fisher Scientific | 34060 | |
Jasco CD spectropolarimeter | Jasco | J-815 | |
Millex Syringe Filter Unit | EMD Millipore | SLGS033SS | |
NMRPipe (Software) | Delaglio et al. | N/a | Delaglio, F. et al. Nmrpipe – a Multidimensional Spectral Processing System Based On Unix Pipes. J. Biomol. NMR 6, 277–293 (1995). |
NMRViewJ (Software) | Johnson et al. | N/a | Johnson, B. A. & Blevins, R. A. NMR View: A computer program for the visualization and analysis of NMR data. J. Biomol. NMR 4, 603–614 (1994). |
Oleic acid | Sigma-Aldrich | O1008 | |
Pierce BCA Protein Assay | Sigma-Aldrich | BCA1-1KT | |
Polaris 5 C18-A 250×10.0 mm HPLC Column | Agilent | SKU: A2000250X100 | |
SD-200 Vacuum Pump | Varian | VP-195 | |
Sodium Cholate Hydrate | Sigma-Aldrich | C6445 | |
Sodium Palmitate | Sigma-Aldrich | P9767 | |
Sodium Stearate | Sigma-Aldrich | S3381 | |
VnmrJ (Software) | Varian | N/a |