Simple and fast maximally deuterated control (maxD) preparation for HDX MS experiments

Abstract

During the analysis steps of hydrogen deuterium exchange (HDX) mass spectrometry (MS) there is an unavoidable loss of deuterons, or back-exchange. Understanding back-exchange is necessary to correct for loss during analysis, to calculate the absolute amount of exchange, and to ensure that deuterium recovery is as high as possible during LC-MS. Back-exchange can be measured and corrected for using a maximally deuterated species (here called maxD) in which the protein is deuterated at positions and analyzed with the same buffer components, %D₂O, quenching conditions, and LC-MS parameters used during the analysis of other labeled samples. Here we describe a robust and broadly applicable protocol, using denaturation followed by deuteration, to prepare a maxD control sample in ~40 minutes for non-membrane proteins. The protocol was evaluated with a number of proteins that varied in both size and folded structure. The relative fractional uptake and level of back-exchange with this protocol were both equivalent to those obtained with a previous protocol that requires much more time or one requiring isolation of peptic peptides prior to deuteration. Placing strong denaturation first in the protocol allowed maximum deuteration in a short time (~10 mins) with equal or more deuteration found in other methods. The absence of high temperatures and low pH during the deuteration step limited protein aggregation. This high-performance, fast, and easy to perform protocol should enhance routine preparation of maxD controls.

INTRODUCTION

Hydrogen deuterium exchange (HDX) measured with mass spectrometry (MS) is a widely used technique to probe protein dynamics, conformational changes, and protein interactions ^2–5. In a typical HDX MS experiment, a protein is diluted in D₂O buffer under physiological conditions and the backbone amide hydrogens are allowed to exchange with deuterium. The exchange reaction is carried out for various lengths of time [e.g. 10 seconds, 1 minute, 10 minutes, etc.] before the exchange reaction is quenched (pH 2.5, 0 °C), the protein is digested with an acid-tolerant protease, and the deuterium incorporation of the newly generated peptides is measured using LC-MS. Because HDX depends on structural features such as hydrogen bonding and solvent accessibility, the rate and location of exchange can provide information about conformation, dynamics, interactions, and protein folding.

The loss of deuterons during the post-quench analysis steps, also known as back-exchange, is unavoidable ^3,6. Back-exchange was described in the very first exchange experiments ⁷ and its ultimate outcome is a loss of information ⁶. In most protocols, the operations downstream of the deuteration step, that is from the quenching step onward, are carried out in 100% H₂O solvents and the labile backbone amide positions that became deuterated in the labeling step can exchange back to hydrogen. Although many improvements have been introduced over the years to reduce and deal with back-exchange at all steps along the workflow ^8–14 it is not possible to completely eliminate the loss of label.

The sequence of a polypeptide influences back-exchange ^15,16; therefore, in quench conditions, each peptide experiences different levels of deuterium loss. The observed back-exchange for a peptide is a function of the intrinsic rate constants (k_ch) of each backbone amide hydrogen composing that peptide, which are dictated by the chemistry of amino acids nearby to each backbone amide hydrogen. For example, peptides containing histidine residues may have very high levels of back-exchange (back exchange can be ~50% in standard quench and LC-MS conditions) due to an inductive effect of the imidazole group, as opposed to peptides containing isoleucine with an isobutyl side chain that sterically blocks exchange (back exchange can be ~10% in standard quench and LC-MS conditions) ¹⁷.

As was demonstrated in the first peptide-level experiments by Zhang and Smith ^3,18, to determine the absolute deuteration value of a peptide/protein it is necessary to measure and correct for back-exchange. Conversely, in relative measurements ², it is not strictly necessary to correct for back-exchange when the same protein is studied in two or more states and those deuteration levels are compared (e.g. apo vs holo, or bound vs unbound); when the generated peptides will be the same for both states and the experimental variables in the LC-MS instrumentation remain constant, all samples experience the same level of back-exchange in the post-quench steps. Back-exchange correction requires the preparation of a maximally deuterated sample – here called maxD, but also known by other names including m₁₀₀, FD, allD – in which a protein is deuterated at all exchangeable positions and analyzed with the identical experimental parameters as for all other labeling time points. Unfortunately, it cannot be known if complete deuteration – that is 100% labeling of all exchangeable positions – is achieved (see also below) and for this reason we refer to this maximally deuterated sample as maxD instead of m₁₀₀.

Once a protein is forced to exchange the maximum exchangeable backbone amide protons, and its back-exchange is measured, a mathematical correction is made (Figure 1) to yield the D_level for each peptide at each labeling time point. The first challenge is to prepare the maximally deuterated version of the protein of interest. Forcing a protein to become deuterated at all backbone amide positions requires that all backbone amide positions find themselves in an exchange competent state in the presence of D₂O and OD^{- 19}, which frequently means the breaking of hydrogen bonds that may involve backbone amide hydrogens. Note that the discussion here and going forward has been limited to non-membrane proteins. The various historical ways of making a maxD sample as described below were nearly all developed and tested on non-membrane proteins. We have elected to optimize a protocol on non-membrane proteins first, then in the future test and optimize ways of preparing maxD versions of membrane proteins.

Figure 1.

MaxD in HDX MS experiments. The classic equation for correcting the measured centroid mass (m_t) to the absolute deuteration level (D_level), at a certain time (t) for a given protein or peptide. m₀ is the measured centroid mass of an undeuterated species, maxD(m₁₀₀) is the measured centroid mass of the maximally-deuterated control [we refer to this as maxD, in the past it has been called m₁₀₀]. On the right is a schematic example of a mass spectrum for m₀, m_t, and maxD(m₁₀₀). At the bottom, in the red box, is a summary of several strategies for preparing maxD(m₁₀₀) with examples and representative references.

Theoretically, one can actively denature a protein using heat, acidic conditions, or chaotropic agents (Figure 1, bottom). Each of these strategies has pros and cons and has been used to varying degrees over the years; however, ultimately the protein may dictate which of these is most effective ²⁰. In early work, Zhang and Smith ³ simply incubated a protein in 100% deuterium buffer at elevated temperature. This method can lead to successful results for some proteins, provided that aggregation of the protein is avoided. Englander and collaborators found that heating their protein to a temperature ~5°C below the melting temperature for 10 min ²¹ worked well, and using only a 10 minute exposure to high temperature reduced the risk of deuterating the hydrogen in C2 of histidine ²². However, the melting temperature of a protein is not always known, some proteins irreversibly aggregate when heated, and some simply do not become totally exchange competent. Another way to accomplish maximal deuteration is to expose the protein to deuterium for a very long time (>24h) ²³ during which natural protein breathing and fluctuations will eventually lead to each backbone amide becoming exchange competent. Remarkably however, waiting even very long periods (> 1 week to 1 month) is sometimes not enough; proteins may have very stable regions that never become deuterated due to natural fluctuations over time. The use of acid as a denaturant is simple and straightforward ¹ but it can require a long incubation time, during which the protein may undergo modification or aggregation. Protein denaturation with urea or guanidinium can reduce the time required to achieve complete deuteration ²⁰, but can lead to chemical modifications (i.e., carbamylation in the presence of urea at high temperature or pH) or alter the final concentration of D₂O during labeling ²⁴. In addition, if there are species in solution capable of interacting with the protein (charge-charge interactions, hydrogen bonds, etc.), this could interfere with the exchange reaction causing a reduction in deuteration levels. Sometimes even a combination of heat, acid, and denaturant is not enough to maximally deuterate some proteins, or it is not possible to do so and still leave the proteins in a native fold that can be digested in a similar manner to the other labeling time point samples. A last strategy, an alternative to denaturing the folded protein and exposing it to deuterium, is to digest the protein under study using the exact conditions that will be used for labeling time points, isolate the resulting peptides, lyophilize them, then deuterate and analyze these isolated peptides with the identical LC-MS conditions that will be used for labeling time points ^6,25. While this strategy is a good way to obtain maxD-labeled peptides, it is not trivial to implement into HDX MS workflows while maintaining the identical experimental conditions that will be used for the actual labeling time points, and simultaneously not lose peptides that have weak signals particularly in online digestion protocol formats.

While making a reliable, maximally deuterated maxD sample can present challenges, sometimes these challenges must be overcome in order to make the best use of HDX MS. The maxD control is essential when studying protein folding or intrinsically disordered proteins. Without a reference deuteration of the denatured state, it is impossible to determine how much a region has actually folded/exchanged at a particular labeling time, thus also precluding the possibility of fitting the data into a kinetic model. Another case in which it is very useful to obtain maxD controls is when one wants to compare different isoforms of the same protein. Since the loss of deuterons during analysis is sequence dependent, if isoforms contain one or more mutations, it is important to determine how much a sequence change(s) affects the back-exchange before doing relative HDX comparisons between proteins with different sequence. Even if it is possible to calculate or simulate the influence of mutation on back-exchange at a given time, temperature and pH, it is still advisable to measure deuteration experimentally; hence a maxD sample is required.

How then should one best prepare a maxD sample? Complete deuteration of a particular protein to measure the maxD values for peptides can be challenging and sometimes there can be doubts about the effectiveness of the protocol used. Also, those who are new to HDX MS may not know how to determine the best method amongst the various different strategies to make a maxD control for their specific protein. While each strategy has its weaknesses, depending on the protein under investigation it may be more appropriate to follow one strategy rather than another. Often it becomes an empirical exercise where all methods are tried until one produces a maxD sample that yields similar peptides to those found in the other deuteration timepoints of the experiment. Various protocols are then developed, specific to each protein. Starting from these considerations, we aimed to develop an easy and fast method that could be adapted to most non-membrane proteins. Here, we evaluated different ways to prepare a maxD control, developed a protocol that is fast and easy to perform but widely applicable to many proteins, and serves as an excellent starting point for the preparation of a maxD control for nearly all non-membrane proteins. We demonstrate how to perform the protocol with six commercially available proteins differing in size and folded structure, and then illustrate the application of the protocol on several other proteins.

EXPERIMENTAL SECTION

Reagents

The proteins tested are further described in Supporting Information Figure S1. Myoglobin (Mb, from equine skeletal muscle), hemoglobin (Hb, from bovine blood), cytochrome C (Cyt C, equine heart), carbonic anhydrase (CA, from bovine erythrocytes) and phosphorylase B (Phos B, rabbit muscle) were purchased from Sigma-Aldrich. Immunoglobulin G (IgG) was generously supplied by Biogen. Pepsin (porcine gastric mucosa) was from Sigma-Aldrich, part P6887. Immobilized pepsin resin for digestion was prepared as previously described ²⁶ and packed into empty 2.1 × 50 mm stainless steel columns (Restek). TCEP and DTT were purchased from Thermo Scientific Pierce and RPI, respectively. D₂O (D, 99.96%) was from Cambridge Isotope Laboratories. All other salts, solvents, and reagents were of analytical grade and purchased from Sigma-Aldrich or RPI.

Sample preparation

In addition to the steps described below alongside the main protocol description, we note the following details. For IgG, the quenching buffer was 200 mM sodium phosphate, pH 2.37, 4 M GdnHCl. 0.72 M TCEP, H₂O. For comparison experiments, fully deuterated myoglobin and hemoglobin were also made under acidic conditions (acid induced method), using a previously published protocol ¹ with minor modification. For this method, protein samples (15 μL, 20 μM) in 150 mM potassium phosphate buffer, pH 2.4, were diluted ten times with D₂O DCl (final pH 2.4) and incubated for 48 h at 20 °C. After this time, 30 μL of protein sample were diluted with 30 μL of quenching buffer and immediately analyzed by LC-MS. To prepare peptic peptides for deuteration after digestion (free peptide method ^6,25), proteins were digested using pepsin beads (5 μL of pepsin beads for 50 pmol of protein) in 0.1% FA for 10 min at 15 °C. After digestion, beads were removed with 0.45 μm centrifuge filters (Spin-X, Costar), peptides were SpeedVac concentrated and then labeled using the same procedure described for our maxD protocol.

LC-MS

In addition to the following descriptions below, comprehensive experimental details and parameters are provided in the Supporting Datafile, in the recommended tabular format ²⁷. All HDX MS data have been deposited to the ProteomeXchange Consortium via the PRIDE ²⁸ partner repository with the dataset identifierPXD032924.

Myoglobin, hemoglobin, cytochrome C, and carbonic anhydrase were analyzed using a nanoAcquity UPLC system coupled to a Waters Xevo G2-S QTof, and phosphorylase B and immunoglobulin G using an Acquity M-Class system coupled to a Waters Synapt G2-Si. Both instruments were equipped with a Waters HDX system [based on Wales et al. 2008 ⁹]. Maximally deuterated and undeuterated control samples were digested online in the HDX cooling unit, where the digestion chamber was held at 15 °C, using a porcine pepsin column packed in-house (POROS 20AL beads, 2.1 mm × 50 mm). Peptides were trapped and desalted on a VanGuard Pre-Column trap [2.1 mm × 5 mm, ACQUITY UPLC BEH C18, 1.7 μm, (Waters, 186002346)] for 3 minutes at 100 μL/min. Peptides were then eluted from the trap using a 5%–35% gradient of acetonitrile over 6 minutes at a flow rate of 55 μL/min (nanoAcquity System) or over 10 minutes at a flow rate of 100 μl/min (Acquity M-Class system), and separated using an ACQUITY UPLC HSS T3, 1.8 μm, 1.0 mm × 50 mm column (Waters, 186003535).

Data Analysis

Peptides were identified using PLGS 3.0.1 (Waters, 720001408EN) using replicates of undeuterated control samples. Raw MS data were imported into DynamX 3.0 (Waters, 720005145EN) and filtered with minimum consecutive products of 1 and minimum number of products per amino acid of 0.20. Those peptides meeting the filtering criteria were further processed automatically by DynamX followed by manual inspection of all spectra and data processing. The amount of deuterium in each peptide was determined by subtracting the centroid mass of the undeuterated form of each peptide from the maximally deuterated form. All measured and processed values can be found in the Supporting Information Datafile. maxD measurements were conducted in triplicate, with the exception of fully deuterated free peptides, which were analyzed in duplicate (a detailed replication table is found in the Supporting Information Datafile).

Data are reported as relative fractional uptake (RFU), obtained by calculating the ratio between experimental deuteration level (D) and theoretical maximum uptake, according to the formula: RFU = D / (N-P-1), where N is the number of amino acids in a given peptide and P is the number of prolines. If a proline is present at the N terminus, the equation becomes RFU = D / (N-P). Error values are those reported by the DynamX software, which takes into both account multiple charge states and replicates. The average back-exchange (BE) was calculated considering the percentage of deuterium used (%D, e.g., 90%, 0.9) and applying the following formula: BE = (1 – RFU / 0.9) * 100.

RESULTS AND DISCUSSION

Method development rationale

Many proteins are stable under physiological conditions and deuteration for very long times is often not sufficient method to deuterate backbone amide hydrogens to completeness. Promoting protein denaturation and forcing proteins to become exchange competent with the use of denaturing conditions (such as low pH, chaotropic agents, heat) is a reasonable strategy. Most of the previously published protocols promote the denaturation of the protein and at the same time its deuteration. Any such simultaneous denaturation and labeling strategy must match the labeling conditions applied to the actual labeling time-point samples for the connected HDX experiment (meaning the buffer composition, the % deuterium in the labeling step, temperature, pH etc.) while at the same time not promoting protein aggregation. We found that a divided protocol actually works better: in the first part, the protein is denatured; in the second part, it is deuterated. For protein denaturation, high concentrations of GdnHCl (7 M) coupled with high temperatures (90 °C) were found to be the best, and if the protein contains disulfide bridges, DTT (50 mM) can be added to the denaturing solution. Under these conditions, all proteins we tested (including the ones illustrated below) were denatured. Once denatured, and exchange competent, it remains only to place the protein in the right concentration of deuterium and complete the deuteration step.

Protocol description and evaluation

The maxD preparation protocol was designed to start under the experimental conditions (protein concentration and buffer) of a typical HDX MS experiment and except for a GdnHCl 7M solution, does not require the preparation of any buffers other than the ones used during the regular HDX MS measurements. All proteins tested started in phosphate buffered saline (PBS: 10 mM sodium phosphate, pH 7.5, 150 mM NaCl, H₂O) at a concentration of 20 μM, conditions which represent a typical starting point of an HDX MS experiment (i.e., protein in equilibration buffer). Note that these conditions are intended to illustrate the protocol, and therefore are modifiable according to need.

After various steps were optimized, a final, effective maxD protocol was developed (Figure 2). Steps 1–5 are the denaturation (Figure 2, green), and steps 6–9 are the labeling (Figure 2, red). Protein solutions (15 μL, 20 μM) are SpeedVac concentrated (step 1, 15 min) and resuspended in 7 M GdnHCl for 1 minute (step 2). Freshly made 50 mM DTT can be added to the GdnHCl solution if the protein has disulfide bonds (e.g., immunoglobulin G). After a brief vortex and spin (step 3, duration 1 min), protein samples are heated at 90 °C for 5 minutes (step 4) and then cooled to 20 °C over the course of 2 minutes (step 5). The now denatured protein is ready to be labeled with deuterium using the same concentration of D₂O as will be used for actual timepoint samples. 27 μL of labeling buffer (10 mM sodium phosphate, pD 7.5, 150 mM NaCl) are added to 3 μL of denatured protein solution (1:10 dilution, step 6, duration 1 min) and the exchange reaction allowed to proceed at 50 °C for 10 minutes (step 7). The now maximally deuterated protein samples are cooled to 20 °C over 2 minutes (step 8) and then to 0 °C over 2 minutes (ice bath, step 9). We found that two stages of cooling helped prevent aggregation in some proteins. Exchange is quenched with the addition of 30 μL of ice-cold quenching buffer (150 mM potassium phosphate, H₂O, pH 2.49, step 10) and the sample analyzed immediately by LC-MS (step 11a). An alternative to direct injection is to flash freeze protein samples for future analysis (step 11b). This entire procedure takes 40 minutes.

Figure 2.

Schematic representation of maxD protocol. The entire protocol takes about 40 minutes and is divided into denaturation (left, green) and a deuteration (right, red). The starting point (blue tube) is a solution containing the protein of interest at the same concentration and in the same equilibration buffer as actual time-point samples of the connected HDX experiment. In the denaturation step, 15 μL of protein material are concentrated for 15 min with a SpeedVac and then resuspended in an equal volume of 7M GndHCl, H₂O (containing 50 mM DTT if the protein has disulfide bonds). After a brief vortex and spin, the protein material is heated at 90 °C for 5 minutes and then cooled back to room temperature (20 °C). The protein is now denaturated/unfolded (green tube). D₂O buffer is then added (maintaining the same buffer components and D₂O concentration, e.g. 90% D₂O, found in the actual timepoints of the connected HDX experiment) to the denaturated protein. The protein solution is heated at 50 °C for 10 minutes and then cooled back to 20 °C. Finally, the deuterated protein is cooled to 0 °C (ice bath), ready to be acid-quenched and injected into the mass spectrometer or flash frozen for future analysis.

The protocol was initially tested using myoglobin and compared with two other previously published methods for preparing maxD, namely the acid-induced method ¹ and the free peptide method ^6,25, see experimental section. The free peptide method has the disadvantage of generating fewer peptides – and therefore perhaps less sequence coverage – but because deuteration is performed on peptides derived from a protein in the absence of tertiary and quaternary protein structure, it should produce peptides with deuteration levels equal to the concentration of deuterium in the labeling solution, the highest that can be achieved. Figure 3 shows the exchange profiles that were obtained for myoglobin, expressed in relative fractional uptake (RFU) for each of the monitored peptic peptides. A typical, regular time course of deuteration with five labeling time points, identical to those reported in ¹, was performed as a reference (Figure 3, blue lines). The results (Figure 3, red and orange lines) of using our maxD labeling protocol (Figure 2) reveal that the levels of deuterium measured for these peptic peptides was nearly identical to the measured deuteration of maxD myoglobin prepared using the free peptide method or an acid denaturation method. Our method showed good reproducibility (compare biological triplicates: Figure 3, red line versus orange line, performed 3 months apart from each other) and the maxD protocol did not introduce more measurement variability than one would expect from biological replication without a maxD preparation step. The deviation of the mass measurements (as per DynamX output) was ±0.08 Da, in line with the acid induced method (±0.12 Da) and in general in keeping with the average reproducibility achievable through the HDX MS technique (±0.10–0.50 Da, using 90% deuterium) ^29–32.

Figure 3.

Comparison of myoglobin deuteration with different maxD preparation protocols. In the main panel (bottom), relative fractional uptake (RFU, y-axis) is shown for each myoglobin peptic peptide (x-axis). For reference, a labeling time course (1, 10, 30, 60, 120 minutes; uncorrected for back-exchange) is shown using different shades of blue. The red line is the RFU obtained by applying our protocol (Figure 2); the orange line is a biological replicate of the red line, performed 3 months apart; the green line is the RFU using the protocol described in Sowole et al., 2014 ¹; the black line is the RFU obtained using the free peptide method (see Experimental Section). The red, orange, green, and black lines each represent the average of a technical triplicate measurement (the bands around the lines represent the error as calculated by DynamX software). The average back-exchange across all peptides for each protocol was calculated and is shown in the inset. In the top panel, the RFU values from each type of maxD protocol have been normalized against our protocol (red line triplicate only). All raw values and RFU calculations used to make this figure are provided in the Supporting Information Datafile.

Our protocol gave results comparable to the other two methods, but in much shorter time. The entire procedure took ~40 minutes, clearly much faster than the acid-induced method, which required a prolonged deuteration of 48 hours. Although the free peptide method can be completed in a relatively short time compared to the acid-induced method, it is more complicated to implement as it requires offline digestion, capture of the digestion products, deuteration, and injection to LC-MS with no online digestion column. The additional manipulation steps have the potential to produce a different final yield of peptides that arrive at the mass spectrometer (with respect to online digestion) ^6,25. We prepared a common peptide map (Supporting Information Datafile ‘sequence coverage’) using peptides we identified with the free peptide method and that were also found in the other methods. It might be possible that the maxD protocol, which involves an initial strong denaturation, could produce a slight difference in identified peptides with respect to the undeuterated control samples or the labeling time points, especially as the expected efficiency of proteolysis for a denatured protein (as in the maxD protocol) is higher compared to the non-denatured undeuterated control samples and labeling time points. We did not find major differences in identified peptides for myoglobin using multiple protocols, or in the other proteins tested comparing maxD protocol and the free peptide method (see Supporting Information Datafile ‘sequence coverage, map comparison’ and Figure S2).

The average back-exchange level for the myoglobin peptic peptides (inset to main panel of Figure 3) varied from 33.4±0.69% with the free peptide method (average uptake of 8.47±0.08 Da) to 37.3±0.96% with the acid induced method (average uptake of 8.01±0.12 Da). Our protocol (average of both triplicates, red and orange in Figure 3) showed intermediate levels of back-exchange 35.4±0.89% (average uptake of 8.24±0.12). The RFU in some peptides in the free peptide method deviate by as much as 10% from the RFU found in other methods, although this was not so evident in the other proteins tested (below). The reason for this deviation is unclear. The peptides with large deviations between the free peptide method and the maxD protocol in the free peptide method were generally more deuterated in the free peptide method than those of the maxD protocol, which implies that they were more deuterated to begin with. It is possible that some residual structure might remain in parts of myoglobin even in the denaturing conditions of the maxD protocol, and that the free peptide method eliminates this structure during the digestion before deuteration.

The influence of sequence on back-exchange in myoglobin can be seen in the exchange profiles. The myoglobin peptide showing the lowest loss of deuterons covered residues 14–20 (¹⁴WGKVEAD²⁰) with a ~20% (acid induced: 22.8%; our prot_1: 22.5%; our prot_2: 21.1%; free peptides: 17.2%) back-exchange, as opposed to the peptide covering residues 110–126 (¹¹⁰AIIHVLHSKHPGDFGAD¹²⁶) which with its three histidine residues showed the highest (~55%; acid induced: 58.2%; our prot_1: 56.4%; our prot_2: 53.6%; free peptides: 58.6%) amount of back-exchange.

Another very obvious aspect from Figure 3 is that without correction for back-exchange, it is impossible to address if the differences between nearby peptides, even those with partially overlapping sequences, are due to a difference in deuterium loss or are rather due to differences in labeling from structural differences in the folded protein. That is, do overlapping peptides with different levels of deuteration back-exchange differently or become deuterated differently? Two examples of this idea are found in two distinct regions: residues 30–55 and 84–134 (indicated with dotted boxes in Figure 3). Comparing the profile of a labeling time point (e.g., 120 min) with the maxD profile, one can see that in the region 30–55, peptides covering primarily residues near the N-terminal end of this sequence (residue 30, left side of dotted box) are far from maximally deuterated while those covering residues nearer the C-terminal end (residue 55, right side of dotted box) are very close to the maximally deuterated level. The deuterium level across the peptides, left to right in the box, is fairly constant in the maxD controls (sitting at around 55% RFU). In contrast, in the region 84–134 (dotted box on the right of Figure 3) the difference between the deuterium level at the labeling time point (e.g., 120 min) and the maximum deuterated form is relatively constant as one moves from peptides primarily covering the N-terminal part of the region to peptides covering primarily the C-terminal part of the region. We conclude that the increased deuteration in the folded protein nearer the C-terminal parts of region 30–55 is due to structural differences in these peptides within the folded protein. In contrast, in residues 84–134 the level of deuteration remains almost identical for all peptides, suggesting that the peptides in region 84–134 have similar protection due to folding in the protein. Without knowing that the maxD levels remain constant for the peptides in the region 30–55 but vary according to peptide in region 84–134, it could be hard to understand the variability of the measured deuterium RFU for the time course data. These two examples illustrate why it is not easy to study the folding and dynamics of a protein alone, meaning without comparison to another form of the protein as in relative HDX MS experiments, without performing maximally deuterated controls.

Method applicability to many types of proteins

To evaluate the broader applicability of our maxD protocol, we measured the deuterium levels in peptic peptides from 5 additional proteins of various sizes and complexity: hemoglobin (Hb), cytochrome c (Cyt C), carbonic anhydrase (CA), phosphorylase b (phos B) and immunoglobulin G (Ig G), see Supporting Information, Figure S1. Each of these proteins was treated with both our maxD protocol and the free peptide method (Hb was also tested with the acid-induced unfolding method) and the levels of deuterium incorporated were compared (Figure 4). As is clearly evident in Figure 4, all proteins responded well to the application of our protocol, and the deuterium level measurements from maxD samples made using our protocol were very similar to those obtained with the free peptide method (the peptide maps were also very similar, see Supporting Information Datafile). Depending on the peptides produced and followed, the average back-exchange values varied from protein to protein and spanned from 25 to 35%. This remarkable difference also gives rise to questions about the significance of the average back-exchange values provided in publications in that without making explicit which peptides were considered to calculate those values, they can vary widely. In this regard, and to promote reproducibility among various laboratories around the world, it is advisable ³³ to establish a known mixture of peptides from a standard protein to determine the back-exchange levels of any experimental setup such that those values could be compared more easily between systems and laboratories.

Figure 4.

Measured deuterium levels in various proteins subjected to the MaxD protocol. Each plot compares the relative fractional uptake (RFU, y-axis) of peptic peptides (x-axis) from test proteins after either our maxD protocol (red lines) or prepared using the free peptide method (black lines). Each line is an average of triplicate measurements. RFU in hemoglobin was also measured using the Sowole et al 2014 ¹ protocol. At the bottom left of each panel, the average back-exchange across all peptides is reported. For phosphorylase b, step 7 of our protocol was modified slightly [from 50 °C to room temperature (RT)] to avoid aggregation, see main text. The raw values and RFU calculations used to make this figure, along with the identification of each peptide with residue numbers, are provided in the Supporting Information Datafile.

It is well known that the use of high concentrations of guanidine hydrochloride is an effective method to denature proteins and at the same time prevent protein aggregation. There are some potential drawbacks to this choice, however. When the solution is diluted in deuterated water (step 6 in Figure 2), the denaturant is also diluted (from 7M down to ~0.35–0.7 M depending on the labeling dilution, e.g. 1:10 or 1:15, etc.) and this may lead to 1) interference with deuteration, leading to a distortion of the maximum achievable deuteration levels, and/or 2) partial renaturation of the protein, thus impeding its complete deuteration. Regarding the first point, it has been reported ²⁴ that guanidine hydrochloride increases the rate of HDX, but it is not known if it influences the maximum deuteration level. As shown empirically in this work and elsewhere ²⁴, even if this effect is theoretically possible, it does not seem to significantly influence the final result since the maximum deuteration values measured with guanidinium are similar to those obtained with other methods that do not use guanidinium (i.e. acid induced method). To the second point, each protein has its own propensity to renature and therefore there may be some variability introduced at this step where denaturant is diluted. For the proteins we studied, all peptic peptides had a level of deuteration very similar to that of free peptides and therefore we conclude that even if there is a certain degree of refolding during denaturant dilution, this seems to occur at a lower rate than the chemical rate of deuteration, k_ch. To partially ensure that proteins remained at least partially denatured during deuteration, we introduced a higher-temperature deuteration step (step 7 in Figure 2), thereby maintaining the protein at 50 °C during deuteration. However, as we found for phosphorylase b, the protocol had to be modified at step 7 to avoid aggregation. When incubated at 50 °C, phosphorylase b aggregates, perhaps because of hydrophobic surfaces that become exposed once denatured. Therefore, for phosphorylase b, we modified the protocol to remove elevated temperature during deuteration, that is step 7 was set to 20 °C, Figure 2. We found that 20 °C in step 7 gave the same results as the free peptide method for phosphorylase b. In Supporting Information, Figure S3 we conducted a related study showing that for myoglobin, the level of maximum deuteration does not change if step 7 is altered from 50 °C to 20 °C, even after reducing labeling time from 10 minutes to 1 minute. There is therefore a certain degree of freedom in step 7, in which the temperature or the incubation time could be modified as needed depending on the protein in question. Our first recommendation is to start with the standard conditions of this protocol to ensure maximum deuteration – meaning perform step 7 at 50 °C, especially if the protein is known to have a strong tendency to refold quickly from denaturant or has a T_m well above 50 °C. If protein aggregation is found during this step, eliminate the heating component and perform step 7 at 20 °C instead, potentially for longer than 10 minutes (up to 30 minutes perhaps) to ensure total deuteration. Again, how step 7 is performed will likely be protein specific and although we strived to test the protocol on a wide variety of proteins, it remains unknown know how well all proteins will behave at this point of the protocol.

We applied our maxD protocol to several other proteins under investigation in our laboratory and use two examples to make several additional points about why it can be important to prepare a maxD control. The results of these experiments are found in the Supporting Information, Figures S4-S5. In Supporting Information Figure S4, our protocol was applied to 16 different IgG light chain constructs ³⁴. The sequence of these light chains was different, making direct comparison of deuteration without back-exchange correction problematic. However, with the maxD protocol, and correction for losses during analysis, we were able to compare and highlight the differences between pathogenic isoforms (leading to amyloidosis or multiple myeloma) with the germline isoform, and to prepare and analyze each of the 16 maxD control samples quickly ³⁴. In Supporting Information Figure S5, we applied the maxD protocol to histone proteins H3 and H4. H3H4 are packed into the nucleosome, on which DNA is wrapped, as a consequence of a monomer-dimer-tetramer shift. The maxD protocol in this context was useful because it helped differentiate, in the absence of a relative HDX MS experiment, which regions in H3 and H4 (e.g., N terminal tails) are completely unfolded and nearly totally deuterated even at the shortest labeling times from those that are protected or undergo conformational changes upon binding to protein partners (see Figure S5).

Finally, we offer three additional suggestions related to maxD experiments. The first is to be sure to examine the actual mass spectra carefully to understand if there is any evidence of EX1 ³⁵, which can be a sign of carryover, aggregation, or unusual protein-related phenomena ³⁶. Not only will understanding EX1 kinetic signatures in a maxD sample help reveal if there is aggregation or carryover, it will ensure that the centroid mass value for maxD is not a function of finding the centroid value for a multi-modal isotopic distribution. Second, we suggest that should there appear to be lower intensity MS signals for the maxD sample, this may indicate aggregation (see previous points) or loss of material during one or more steps of the protocol. Be on the lookout for reduced signals and endeavor to understand what might be causing that. Finally, third, because the speed-vac step (step 1) is a delicate step where there may be a loss of material, it is advisable to start with “workable” volumes (>15 μL) and, if necessary, increase the volume taking into account a correction factor of at least 1.2 (e.g. speed-vac 18 μL instead of 15 μL, and then resuspend the concentrated material in 15 μL of GdnHCl). Be aware that not all proteins will recover from being dried completely – although the use of a denaturing protocol increases the chances that they will – and therefore be prepared to leave some liquid in the sample in step 1 should it appear that protein cannot be resolubilized.

CONCLUSIONS

A fast, reliable, widely applicable protocol for preparing a maxD sample is desirable. There are several situations where one truly needs to perform back-exchange correction and therefore a maximally deuterated form of the protein must be prepared ^27,37. A maxD control sample permits measurements of the level of back-exchange for each protein/peptide analyzed and then application of a correction factor for all samples analyzed under the same experimental conditions.

Here we have described a protocol for preparing a maximally deuterated sample that involves protein denaturation and then deuteration. This two-step approach is generally different from other protocols present in the literature, and it allows the use of harsher conditions for denaturation and shorter times to obtain complete deuteration. As we have shown, our protocol worked equally well for many different types of non-membrane proteins, and the results were equivalent to those obtained using previously published protocols which took much longer or were more difficult to execute. We believe that this protocol is an excellent starting point – and in our hands often the ending point – to obtain a maxD control sample. Being able to reliably and quickly prepare a maxD control should be valuable for HDX MS beginners and a valid alternative for those who have problems in maximally deuterating their protein.