Precursor Reagent Hydrophobicity Affects Membrane Protein Footprinting

Abstract

Membrane proteins (MPs) play a crucial role in cell signaling, molecular transport, and catalysis, and, thus, are at the heart of designing pharmacological targets. Although structural characterization of MPs at the molecular level is essential to elucidate their biological function, it poses a significant challenge for structural biology. Although mass spectrometry-based protein footprinting may be developed into a powerful approach for MPs study, the hydrophobic character of membrane regions makes structural characterization difficult by using water-soluble footprinting reagents. Herein, we evaluated a small series of MS-based photo-activated iodine reagents with different hydrophobicities. We used tip sonication to facilitate diffusion into micelles, thus enhancing reagent accessibility to the hydrophobic core of MPs. Quantification of the modification extent in hydrophilic extracellular and hydrophobic transmembrane domain provides structurally sensitive information at the residue-level as measured by proteolysis and LC–MS/MS for a model MP, Vitamin K epoxide reductase (VKOR). It also reveals a relationship between the reagent hydrophobicity and their preferential labeling sites in the local environment. The outcome should guide future development of chemical probes for MPs and promote a direction for relatively high throughput, information-rich characterization of MPs in biochemistry and drug discovery.

Graphical Abstract

Introduction

Membrane proteins (MP) contain ~60% therapeutic targets and perform vital cellular functions, including signal relay, transport, reaction catalysis, and cell communication.^1-5 The elucidation of MP structures brings understanding of their biological function. Despite the importance of this class of proteins, high-resolution structural information is sparse compared with water soluble proteins. Regularly used methods, such as X-ray crystallography, nuclear magnetic resonance (NMR) spectroscopy, and Cryo-EM deliver high-resolution structural information, but are time-consuming, require relatively large amounts (and large sizes for Cryo-EM) of proteins, and it is challenge to “trap” possible conformational states.^6-10 These difficulties have led to the development of alternative approaches for investigation of MP structural organization and function in the membrane. Advanced mass spectrometry (MS)-based structural proteomics is emerging as a powerful tool that can bypass the limitations associated with the high-resolution approaches while offering moderate-resolution structural data for MPs.⁶

“Footprinting”, an emerging tool in structural proteomics, examines protein structure by using chemical reactions to assess the exposure of the protein backbone or side chain to solution or the lipid environment. Footprinting affords detailed protein spatial and temporal information on solvent accessibility, conformational variations, dynamics, and binding interfaces at the peptide- and sometimes at the residue-levels. The key to success in footprinting is the use of suitable chemical reagents that can access and modify the protein while satisfying the criteria: (1) remain inert with the aqueous buffer media, (2) modify the protein with high reactivity; (3) label the protein over either a broad range of residues or specifically and only target several residues; (4) form stable, non-hydrolysable derivatives; and (5) preserve the protein in near native structure and functional integrity.

MPs have transmembrane domains (TMs) in a hydrophobic lipid environment. The TMs consist predominantly of nonpolar amino-acid residues that often traverse the lipid bilayer several times. Developing a footprinter for MPs and their TMs is challenging because the reagent should at least react in the hydrophobic membrane and ideally also with the extramembrane regions.

Currently, there are two major ways of using reactive species to footprint MPs.^{11, 12} The most popular footprinting reagent is the hydroxyl radical generated by photolysis of hydrogen peroxide (commonly known as fast photochemical oxidation of proteins (FPOP)),¹³ radiolysis of water¹⁴ or other methods. ^{15, 16} This type of reaction modifies a range of protein residues on a short timescale. The hydroxyl radical,^17-21 other FPOP reagents,^22-24 or other footprinting methods^{25, 26} are suitable probes for water soluble proteins and extra-membrane domains of MPs. For the TM domain, however, there is little modification^{18-20, 22} because the hydrophobic core resists penetration of reagents. An example where footprinting works is the work of Chance^{27, 28} who utilized radiolytic labeling coupled with MS to ionize and visualize bound water already located in membrane region. To extend TM footprinting by external reagents, the Gross lab utilized photocatalytic titanium dioxide (TiO₂) nanoparticles and a [2+2] cycloaddition to open membrane and admit externally formed reagents.²⁹

Another method is exemplified by carbene footprinting whereby the carbene precursor is partitioned into the TM to label regions of trimeric E. coli OmpF.³⁰ The approach achieved efficient labeling of the TM domain, obtaining broad reactivity upon irradiation at 349 nm to form the carbene directly in the membrane. A third approach is to use a relatively slow reacting footprinter that partitions to a lipid phase but needs no activation; an example is provided by footprinting MPs by diethylpyrocarbonate.^{31, 32}

The development of the direct method for labeling transmembrane domains was driven by several factors. Transmembrane domains play vital roles in cellular processes, often characterized by α-helices spanning the lipid bilayer. The interaction between transmembrane helices, lipids, and ligands profoundly shapes the overall structure and functionality of the membrane protein. For instance, G protein-coupled receptors (GPCRs) have ligand-binding sites deeply embedded within their transmembrane α-helical domains;³³ vital membrane protein events, such as protein-protein³⁴ and protein-lipid interactions,³⁵ occur within the transmembrane region. All these considerations necessitate an approach that can directly measure the transmembrane domains.

In this article, we report the development of photo-activated iodine labeling by inserting the stable precursor by gentle tip sonication into the membrane for TM labeling In contrast to hydroxyl radicals and carbenes, photo-activated iodine labelling is more specific labeling approach, which predominantly targets amino acids such as Tyr, Trp, and His.³⁰ Moreover, previous studies reported that even heavy iodine modification does not significantly alter the quaternary structure of a protein.^{36, 37} Labeling of proteins with iodine, as applied in crystallography and radioimmunoassays, is a time-honored strategy that also satisfies footprinting criteria.^37-42 The direct iodination of peptides and proteins,^{38, 43} however, is associated with the detrimental oxidation on sensitive residues such as Met, Trp, and Cys because the reaction was carried out in the presence of an oxidizing agent^{32, 37}. Indirect iodination, as used with the Bolton-Hunter^{44, 45} method and N-succinimidyl 3-iodobenzoate^{46, 47} can avoid oxidative side-reactions, but they are less specific for proteins that do not contain Tyr residues. Both methodologies achieve abundant modification within seconds to several minutes. Photo-activated iodine labelling bypasses the limitations of direct or indirect iodination by avoiding oxidation reagents but maintains iodine selectivity for modification.

As per existing literature,⁴⁸ the photolysis of iodinated molecules under 248 nm UV light results in the formation of an excited species, I(2P1/2) or I*. The formation of these excited states, denoted as X* for other halogenated molecules, can also be achieved through UV light photolysis at different wavelengths. For iodinated molecules, the optimal laser wavelength is typically 248 nm, whereas for brominated molecules, 193 nm is more suitable due to its higher quantum yield. Nevertheless, we choose aryl iodine as our testing reagent as it can be readily photolyzed by 248 nm laser which is compatible to protein. The estimated lifetime of this excited state, as determined for iodobenzene, is approximately 0.5 picoseconds for its excited state. In an early study examining the kinetics of the reaction between iodine atoms and phenols in water, it was found that the rate constant is approximately 1.6 x 10⁷ L mol⁻¹ s^−1,49 which is smaller than that of diffusion control but still allows reactions in the ms range. Consequently, the reactive iodine is expected to predominantly react with nearby residues. Taken together, photo-activated iodine compounds have good light absorption at 248 nm and high quantum yield (~0.25) to give the reactive species, satisfying a fast-labeling criterion.⁵⁰ The iodinated derivatives are stable, and the modification mainly irreversible; thus, iodination yields a straightforward readout of site modification as reported by predictable increases in the MW of the probe.

Recently, our group developed a highly hydrophobic reagent perfluoroalkyl iodide (logP = 3.52) that mainly labels the TM segment of VKOR membrane protein.⁵¹ This approach is another example of partitioning a stable precursor to the lipid environment and activating it there to yield a reactive species. We suggest that hydrophobicity is the major driving force for the partition of the reagent into the lipid environment. To label the intact MP thoroughly in a structural study, we need to determine the appropriate hydrophobicity of MP footprinters, establishing a relationship between reagent hydrophobicity and MPs labeling efficiency. The aim of this study is to evaluate a small series of iodination reagents to characterize reagent hydrophobicity on MPs footprinting. To vary hydrophobicity of a reagent, we use its partition coefficient (P).

To this end, we selected three different iodinated reagents (iodobenzoic acid, iodobenzyl alcohol, iodobenzene), exhibiting a range of logP values but relatively similar molecular weights and structure. Using VKOR MP, we can simultaneously measure the effect of reagent hydrophobicity on footprinting the water-soluble extracellular segment and the hydrophobic membrane-spanning region. Furthermore, circular dichroism (CD) and an enzyme activity-based assay can test whether the structural and functional integrity of VKOR is preserved upon the sonication, addition of the photo-activated iodination reagents, and activation. In contrast to traditional iodination methods,^{38, 39, 43-47, 52} the FPOP platform provides time control, minimizing structural perturbation potentially caused by slow modification. The specific labeling on certain residues (i.e., His, Tyr, and Trp) offers straightforward data analysis and pinpoints the influence of reagent hydrophobicity on those amino acids as a function of their local environment. Further, the modification ratio on each residue will help map the topology and the binding sites of MPs. Evaluation of these characters would provide a benchmark to future development of chemical probes for MPs studies.

Materials and methods

All reagents and solvents were commercially available and were used without further purification. Urea, water, Tris base (>99.9% purity), acetonitrile, isopropanol, ammonium bicarbonate, dimethyl sulfoxide (DMSO), dithiothreitol (DTT), and formic acid were obtained from Sigma-Aldrich Chemical Company (St. Louis, MO, USA). n-Dodecyl-β-D-Maltopyranoside (DDM) was obtained from Anatrace (Maumee, OH, USA). Chymotrypsin and Tris(2-carboxyethyl)phosphine hydrochloride (TCEP-HCl) were from Thermo Fisher Scientific (Waltham, MA, USA). 130-Watt ultrasonic processor (Model No. VCX 1300 was used for tip sonication. Microcon-30kDa Centrifugal Filter Unit was purchased from Millipore Sigma. (St. Louis, MO, USA) The concentrations of all the protein stock solutions were determined by UV absorption using a Thermo Scientific^™ NanoDrop^™. The VKOR membrane protein was from Dr. Weikai Li, DBBS, Washington University in St. Louis.

Reagent preparation:

Iodobenzene (1 M) and 4-iodobenzyl alcohol were prepared in isopropanol, and 1 M 4-iodobenzoic acid in DMSO.

VKOR membrane protein preparation:

DDM (2 mg) was dissolved in 10 g Tris buffer (200 mM, pH 7.5). The resulting 0.02% DDM Tris buffer solution was vortexed to mix thoroughly. A VKOR stock solution (1 mM) was then diluted by 0.02% DDM Tris buffer solution to 100 μM.

Tip sonication:

Samples (50 μL) in 500 μL Eppendorf tube were sonicated in an ice bath. The probe of the ultrasonic processor was immersed into the solution, and the sonication was as follows: amplitude 20% followed by sonication for 2 s, followed by resting for 10 s.

Photoactivated iodination of protein on FPOP platform:

A 50 μL sample (at various protein to reagent ratios) in a syringe was introduced to the syringe pump. A flow rate (20.6 μL/min) was calculated based on the width of the laser spot (2.1 mm) and laser frequency (7.2 Hz) to ensure 20% exclusion volume. The sample solution was passed through capillary window and irradiated by a 248 nm KrF excimer laser (15 mJ/pulse, GAM Laser Inc., Orlando, FL). All samples were collected in 500 μL Eppendorf tube.

Protein digestion:

The triplicate labeled samples were digested by using a modified FASP protocol.⁵³ Chymotrypsin was selected for protein digestion due to its ability to provide excellent sequencing coverage, and it was chosen because iodination does not adversely affect its digestion efficiency, according to our previous studies.⁵¹ Samples were transferred to ultrafiltration units and then 200 μL of denaturing solution containing 8 M urea in 0.1 M ammonium bicarbonate (pH 8.0) was added. The ultrafiltration units were centrifuged for 20-25 min at 10,000 g until 10 μL solution was left in the filter. A denaturing solution (200 μL) was then added to the ultrafiltration units, and the washing step was repeated twice. The solution in the collection tube was discarded, and 100 μL 50 mM TCEP-HCl in 0.1 M ammonium bicarbonate was added to the filter and incubated at 37 °C for 30 min. Iodoacetamide (IAA, 100 μL 100 mM) in 0.1 M ammonium bicarbonate was added, and incubation took place in the dark for 30 min. The ultrafiltration units were centrifuged at 10,000 g for ~20 min. The addition of 150 μL digestion buffer (DB, 0.1 M ammonium bicarbonate) was followed by centrifugation at 10,000 g for 10 min. This step was repeated twice. Then, 60 μL DB and protease (enzyme:protein of 1:20) was added. The ultrafiltration unit was placed in the water bath at 37 °C overnight. After incubation, a new collection tube was replaced to collect the digested peptides. The ultrafiltration unit was centrifuged at 10,000 g until the solution completely passed the filter membrane. The washing step was repeated. Formic acid (1 μL) was finally added to the solution in the collection tube to acidify the solution. The digested peptide solution was used for the following MS analysis. Concentration of the peptides was determined by using a Nanodrop^™ spectrophotometer (Thermo Fisher, Walthan, MS).

Enzyme activity:

A vitamin K (1 mM) stock solution was prepared and diluted to 100 μM by isopropanol, and a 1 M DTT stock solution was diluted to 12.5 mM by buffer A (0.1% DDM in 150 mM NaCl and 20 mM Tris). Buffer B was prepared as 12.5 mM DTT (37.5 μL), 100 μM vitamin K (300 μL), buffer A (2.66 mL). Protein samples (40 μL) were added to a microplate, and 40 μL of buffer B was added to each suspension. The decrease in absorbance at 430 nm was measured every 30 s for a total of 10 min by a Spectra Max M5 multimode plate reader (Radnor, PA). As a control, VKOR membrane protein without either sonication treatment or the addition the reagents was similarly monitored.

Circular dichroism:

Protein samples (0.25 mg/mL) were incubated in the presence or absence of iodination reagents in 1×PBS buffer. Circular dichroism (CD) spectra were measured at room temperature over the wavelength range of 195-340 nm at 0.5 nm intervals by using a JASCOJ815CD spectrometer (JASCO Analytical Instruments, Tokyo, Japan).

Tandem MS/MS:

The digested peptide solution (20 μL) was diluted by water with 0.1% FA to 50 μL. A sample (5 μL) was loaded onto a custom-built silica capillary column packed with C18 reversed-phase material (Waters Symmetry, 5 μm, 100 Å, 75 μm×30 cm) with an integrated emitter. The HPLC gradient was as follows: from 2.0% solvent B (80 % acetonitrile,0.1 % formic acid) to 65% solvent B over 80 min, then to 98% solvent B over 5 min at a flow rate of 0.5 μL/min, followed by a 10 min re-equilibration step. The Q Exactive Plus hybrid quadrupole orbitrap mass spectrometer coupled with a Nanospray Flex ion source (Thermo Fisher, Santa Clara, CA) was utilized for analysis. The spray voltage was set as 3.0 kV, and capillary temperature was 250 °C. The 15 most abundant molecular ions were automatically chosen for fragmentation (DDA) by using the mass spectrometer scanned from m/z 300–2200 with a resolving power (RP) of 70,000 for MS1 and 17,500 for MS2 at m/z 200 throughout the chromatography. Precursor ions were with higher energy collisional dissociation (HCD) with a normalized collision energy (NCE) of 32%. The automatic gain control (AGC) targets were 5 × 10⁵ for MS and 5 × 10⁴ for MS/MS acquisitions. Maximum injection times (maxIT) were 200 ms for MS and 100 ms for MS/MS².

Data analysis:

The LC–MS/MS raw data were searched for unmodified and modified VKOR membrane chymotryptic peptides by using Byonic Software (Protein Metrics, San Carlos, CA). Iodination modifications on Tyr, Trp, and His were added as modifications to the database. The parameters were 10 ppm precursor mass tolerance, 60 ppm fragment mass tolerance, and CID/HCD fragmentation. Modification-sites were validated by manually checking the product-ion spectra and were double-checked with Thermo Xcalibur (a custom program from Thermo Fisher). Quantification of modification including miscleavages was with the peptide giving the largest MS intensity rather than quantifying over all peptides that contain the residue of interest. Including other peptides when they are of low abundance introduced larger deviations.

Results and discussion

Selecting new footprinting probes:

A good MP footprinter should fulfill three criteria: (1) diffuse efficiently into the TM domain; (2) achieve extensive labeling; and (3) give results that correlate with the native protein structure. To develop an MP footprinter and study the effect of footprinter hydrophobicity on MP labeling, we selected three photo-activated reagents, iodobenzoic acid, iodobenzyl alcohol, iodobenzene with a range of logP values from −1.56 to 3.02 (Figure 1). A negative logP (for iodobenzoic acid) means the compound has a higher affinity for the aqueous phase, whereas a positive logP (iodobenzyl alcohol) and iodobenzene) should allow a higher partition into the lipid phase. The hydrophobicity gradient of the reagents allows us to evaluate the relationship between reagent hydrophobicity and TM labeling. The incorporation of an aromatic ring in the precursor is to add lipophilicity to allow diffusion of the reagent into the membrane-spanning region. The attached functional groups at the para position (carboxyl group, -CH₂OH group, and H) adjust the probes’ relative hydrophobicity. The photochemical iodination by iodobenzoic acid has been reported for water soluble protein modification, the rate constant of I· is similar to hydroxyl radical (3×10¹⁰ to 3×10¹¹ M⁻¹s⁻¹), making photochemical iodination reagents promising for fast protein labeling. ⁴⁷

Figure 1.

Structures of 4-iodobenzoic acid, 4-iodobenzyl alcohol, and iodobenzene.

We envisage the mechanism as photolysis to produce a phenyl radical (or substituted phenyl radical) and an iodide radical, as discussed above. The more reactive phenyl radical readily abstracts •H from OH or NH on aromatic rings to produce a resonance-stabilized protein radical. This latter radical is “capped” with •I followed by tautomerization. Verification of the mechanism awaits future studies.⁵⁴

Testing maintenance of high-order structure by an enzyme activity-based assay.

An important need for footprinting is to elucidate the protein structure and protein interactions without affecting protein native structure. Another issue, as mentioned above, is transporting the reagent to the protein site. In our previous work, we used sonication and found it to facilitate transfer to the MP and give sufficient footprinting⁵¹. Because we also utilized sonication in this study to facilitate reagent incorporation into the micelle, we need to test if sonication disturbs the protein high-order structure by checking VKOR enzymatic activities in the presence or absence of the sonication and reagents. The enzyme activity of the VKOR is sensitive to the tertiary structure because it contains a catalytic core surrounded by a four transmembrane helix bundle connected via a linker TM segment with the extracellular domain.⁵⁵ We consider there is no significant structural perturbation if the enzyme maintains more than two thirds of its activity although that criterion may be too lenient.

No significant change in enzyme activity was detected upon sonication, indicating that the VKOR’s structure is reasonably maintained upon sonication. We next mixed VKOR and the iodination reagents at several concentrations under sonication conditions to screen the optimum concentration. All three iodination reagents at 10 mM concentration inactivated the enzyme activity dramatically. No deterioration of the VKOR membrane protein activity was seen with the addition of up to 1 mM iodobenzene, 1 mM iodobenzyl alcohol, and 10 μM iodobenzoic acid, illustrating the MPs can preserve their native protein structures under these conditions (Figure 2). The loss of function, however, does not necessarily result in significant structural changes if the disruptions only affect substrate binding or catalysis.⁵⁶ To obtain more detailed structural perturbation information, we characterized the high-order structure of VKOR MPs under several conditions by circular dichroism spectroscopy.

Figure 2.

Activity of VKOR after treatment with sonication and several concentrations of iodobenzene, iodobenzyl alcohol, and iodobenzoic acid. (The (s) in the x-axis labels represents use of sonication.)

Testing maintenance of high-order structure by circular dichroism (CD) spectroscopy.

CD is essential for characterizing the secondary structure of proteins in solution and assessing their structural integrity. The CD signals are dependent on the absorption of radiation, to give spectral bands associated with distinct structural features.⁵⁷ The spectra for VKOR exhibit a typical pattern of α-helix content with two negative maxima at 222 and 208 nm, showing that the MP keep their α-helical structure upon sonication (Figure 3). VKOR in admixture with iodobenzene, iodobenzyl alcohol (1 mM) with sonication treatment also maintained the α-helix content well. The far UV CD curve, however, did not preserve its shape with the addition of iodobenzoic acid (1 mM), showing a condition where secondary structure was disrupted. When we further decreased the concentration of iodobenzoic acid to 100 μM and found that VKOR nearly maintains its secondary structure at that concentration of the acid. (Figure S2)

Figure 3.

Far UV CD spectra of VKOR upon sonication and addition of iodobenzene, iodobenzyl alcohol, and iodobenzoic acid at 1 mM concentration.

Evaluating the relationship between reagent hydrophobicity and MP labeling efficiency.

We first labeled the protein with the three photo-activated reagents at the same concentration of 100 μM where the VKOR structure remains native-like. To locate modifications at the residue level, we performed LC-MS/MS analysis. We relied on the accurate masses of chymotryptic peptide and their modified counterparts to obtain extracted ion chromatograms (EICs) for quantification. The modification ratio is calculated by using equation 1. For peptide 272-277, containing one Tyr and undergoing mono-iodination for example (Figure 4), we can establish: (1) the monoiodinated peptides are more hydrophobic and elute ~ 14 min later than their unmodified counterparts (Figure 4a). (2) Despite the fact that the photolysis of iodinated reagent could generate two different radicals, only iodination that result in a mass shift +125.90 Da was observed, with no reaction with the aryl radical was observed. The MS/MS spectra locate modification at residue Y277 by identifying that all y ions shift by +125.90 Da (Figure 4b, and 4c). Summarizing the bottom-up analysis, we found that all detectable photochemical iodination occurs on Tyr and Trp residues. (Figure 5).

Figure 4.

Example LC-MS/MS analysis for VKOR membrane protein labeled by iodobenzyl alcohol (100 μM). a) EIC of unmodified (red) and mono-iodinated (blue) peptide 272-277. MS/MS of b) mono-iodinated and c) unmodified peptide 272-277.

Figure 5.

Comparison of the iodine modification extents on various residues of VKOR at a reagent concentration of 100 μM.

$$Modificationratio\%=\frac{modifiedpeptides}{modifiedpeptides+unmodifiedpeptides}\times 100\%$$ (1)

To view the location of the labeled residues clearly, we next mapped the photolabeling result on the protein 3D structure. (PDB 3KP9) (Figure 5 and 6). The hydrophilic reagent 4-iodobenzoic acid labels residues at the extra-cellular domain: Y4 with 1.6% modification, Y277 with ~ 0.2% modification, and other residues (Y207, Y228, Y252, Y262, Y277) with less than 0.1% modification. It can also label residues at the interface between the outer and membrane-embedded domain (W64 with ~ 0.15% modification), whereas no modification was found for residues that are buried deep within the hydrophobic TMs (Figure 6a, 6b). For 4-iodobenzyl alcohol, with intermediate hydrophobicity, labelling covers residues from both extracellular and TM domains (Figure 6c, 6d). In the outer membrane domain, Y4 shows ~ 5.6% labeling, Y277 undergoes ~ 1.1% modification, and other residues, including Y204, Y207, Y228, Y262, exhibit < 0.5% modification. In the TM segment, Y39 and W64 undergo ~ 0.5% and 0.1% labeling extent, respectively, whereas W99 undergoes 6% labeling.

Figure 6.

Iodination footprinting of VKOR with 100 μM a) and b) 4-iodobenzoic acid, c) and d) 4-iodobenzyl alcohol, e) and f) iodobenzene. Footprinted residues as mapped on crystal structure (PDB file 3KP9). The extracellular domain is color-coded in green, while the membrane-embedded domain is depicted in tan.

Iodobenzene, with the highest logP value among three reagents, not surprisingly can target two residues buried inside the TM domain (Y39 with 0.1% modification and W99 with 3% modification). Noteworthy is that we also observe four residues modified at the outer membrane domain (Y4, Y207, Y228, Y277), but at lower extent (<0.05%). (Figure 6e, 6f)

Comparing the number, the modification ratio, and the location of each labeled residue, we note that 4-iodobenzoic acid mostly labels residues in hydrophilic regions of the outer membrane domain, indicating that a monolayer of micelle prevents 4-iodobenzoic acid from accessing the TM domain. In addition, the higher modifications on Y4 and Y277 demonstrate that these two residues are located in more solvent-accessible or dynamic regions, which correlates well with their N-term and C-term characters. With iodobenzene, we observe minimal labeling of residues in the outer membrane region, but there is notable footprinting on Y39 and W99 within the transmembrane (TM) domain. This indicates that, at this lower concentration, the iodobenzene reagent begins to partition into the membrane, resulting in the labeling of the membrane-spanning region. Interestingly, iodobenzyl alcohol, which is less hydrophobic than iodobenzene , labels Y39 and W99 at 0.5% and 6%, respectively , and it labels these two residues relatively more extensively. This indicates the modification ratio is not solely determined by reagent hydrophobicity. These two reagents have different substitutes on the aromatic ring. The intrinsic chemical reactivity, which can be tuned by the substitute, can also play a significant role in determining the modification ratio.

Among the three reagents, activation of 4-iodobenzyl alcohol modifies the largest number of residues on the intact MP, and the modification extent also stands out. This may be attributed to its appropriate hydrophobicity that assists it to diffuse into the hydrophobic TM region and its higher chemical reactivity of the aryl radical that initiates the labeling. The rank of modification ratio on residues in TM (W99 > Y39 > W64) matches well with the trend of “accessible surface area in the membrane” of these residues as discussed later). In addition, the relatively high modification on Y4, Y207 and Y277 in extra-cellular region associated with high accessible surface area in the membrane and dynamic character in that region.

Summarizing the labeling results at the low labeling reagent concentration, we found that the location of labeled residues depends on footprinter hydrophobicity. With increases in hydrophobicity, the reagent can enter the membrane more easily, leading to modification of more residues in TM domain. It is not true, however, that the most hydrophobic reagent (iodobenzene) gives the best coverage in the hydrophobic region. That reagent labeled fewer residues in hydrophobic region than did 4-iodobenzyl alcohol with medium hydrophobicity. Thus, we need to consider both the hydrophobicity and reactivity, a “sweet spot”, for MP labeling. The 4-iodobenzyl alcohol presumably shows higher modification extent because this reagent possesses a balance of hydrophobicity and reactivity of the aryl radical that initiates the footprinting.

Footprinting VKOR MP:

Though several residues underwent modification at the reagent concentration of 100 μM, the extent of modification is too low for accurate footprinting. Thus, we further increased the concentration of 4-iodobenzyl alcohol and iodobenzene to 1 mM for footprinting. With increases in concentration, mono iodination of Tyr and Trp increases. (Figure 7) The di-iodination also becomes obvious (Figure 7), manifesting a major peak shift of +251.79 Da. MS/MS analysis confirmed the di iodination (Figure S3a-S3d). The unmodified peptide34-37, mono-iodinated, and di-iodinated eluted at 37.95, 51.08, and 58.22 min. The di-iodinated peptide (di-iodination on same residue) is even more hydrophobic and eluted ~7 min compared with its mono-iodinated counterpart. By integrating the EIC, we calculated that peptide 34-37 undergoes 1.90% mono-iodination and 0.40% di-iodination. In the MS/MS spectrum, the shift of the y2 ion by 251.79 Da confirms the modification occurs on Y36.

Figure 7.

Comparison of the iodine modification extents on various residues at reagents’ concentration of 1mM.

We then analyzed labeling result of iodobenzene and 4-iodobenzyl alcohol at 1 mM concentration. Iodobenzene can label 15 residues in total. It labeled more residues that were not labeled at the low reagent concentration including residues in TM domain (i.e., Y36, W64, Y120, Y132, Y163, Y178) and those in outer membrane domain (i.e., Y252 and Y262) (Figures 5 and Figure 7). The extensive modifications on residues in TM domain occur on W99 and Y163. W99 and Y163 have relatively high accessible surface area in the membrane, which contributes to the higher modification. We did not detect any modification on Y117 with low accessible surface area. In comparison, using 4-iodobenzyl alcohol leads to modification of 16 residues by covering an additional residue Y117. In this case, higher reagent concentrations not only enhance the modification ratio but also increase the number of modified residues. Specifically, at a concentration of 100 uM reagent (Figure 5), 9 out of 14 Tyr residues and 2 out of 7 Trp residues were modified. However, when the concentration is increased to 1 mM, all Tyr residues and 2 out of 7 Trp residues were modified.

Next, we employed an online tool to calculate the VKOR accessible surface area in the membrane for iodine radicals (similar to calculating SASA but substituting the radius of an iodine atom instead of that of water). The results, along with their corresponding modification ratios induced by 4-iodobenzyl alcohol, are summarized in Table S1. A dual Y-axis chart was generated to evaluate the correlation between these datasets. This chart, along with its discussion, is also provided in the supplementary information as Figure S4. Overall, there is a noticeable correlation between surface area and modification extent, except for three outliers (Y120, Y132, Y163), where the relationship between modification ratio and “SASA” contradicts the general trend (see Fig.S4). The phenomena observed in these three outliers may be associated with the microenvironment surrounding the residues;⁵⁸ contributing are as ionization state, electrostatic interaction, steric hindrance, and H-bonding. In principle, using the modification ratio as a reporter, we can determine the accessibility in the buried core of the MPs. Conclusions, however, based on modification ratio to predict the “SASA” of residues should be made with caution, because the intrinsic reactivities of side chains and the reagent, the steric hindrance of nearby residues, and the primary and higher order structure of protein may also affect the labeling ratio⁵⁸ Importantly, residue Y4, located near the N-terminus, displays the highest level of labeling, with a remarkable 96% modification. This aligns with the requirement that the N-terminus had to be removed for successful VKOR crystallization. Furthermore, the need to truncate underscores the utility of footprinting in exploring highly dynamic regions where obtaining structural information through high resolution structural methods proves challenging.

Comparing 4-iodobenzyl alcohol and DEPC labeling from our previous work (Table 1),³¹ we see (1) unlike DEPC’s less-specific labeling, use of 4-iodobenzyl alcohol specifically target Y and W residues, (2) 4-iodobenzyl alcohol gives higher labeling percentage on Y residues, (3) 4-iodobenzyl alcohol can label more Y and W residues in the TM domain. All these observations reflect the different hydrophobicity and labeling mechanisms between 4-iodobenzyl alcohol and DEPC. Of importance is that DEPC labeling was previously utilized in another IMP system.⁵⁹ However, peptide sequencing didn’t cover the critical transmembrane domains (as depicted in Figure S2 from the reference).⁶⁰ This limitation likely stems from the scarcity of tryptic residues within the transmembrane domain. Whereas transmembrane regions can potentially be labeled by DEPC, the bottom-up analysis approach through tryptic digestion might challenge identification of the modification owing to some reversibility of the labeling reaction.^61-63

Taken together, the data illustrate that the VKOR is progressively iodinated, starting from the most accessible residues. Higher concentration footprinting leads to more structural information by labeling more residues in TM region with higher labeling extent. Thus, the optimized concentration (1 mM) will work for further differential experiments designed to report the structural difference between two states. Moreover, the better labeling performance of 4-iodobenzyl alcohol than iodobenzene helps confirm that a reagent with greater hydrophobicity does not guarantee more labeling of MP. Overall, 4-iodobenzyl alcohol is the best MP footprinter among the three, satisfying three criteria of a good MP footprinter: (1) efficiently diffuses into the MP TM segment; (2) affords extensive labeling on the intact MP; (3) labels consistent with the high-order structure of the MP.

Conclusion

Methods that permit efficient mapping of both hydrophilic outer membrane domain and hydrophobic core of the MP with improved throughput and sensitivity are highly desirable. To test the relationship between reagent hydrophobicity and labeling efficiency on MP, we selected three photo-activatable iodine reagents with varying hydrophobicity for footprinting the VKOR MP. Using the reactive photo-iodination labeling helps achieve fast, efficient, irreversible, and specific labeling. Facilitated by sonication and combined with MS-based analysis, the method provides reliable data for locating the labeled residues. A hydrophilic reagent preferentially labels in the hydrophilic extracellular domain, whereas a hydrophobic reagent comparably modifies residues within the hydrophobic membrane-spanning region and the outer membrane region; the reagent with an intermediate logP can label residues in both regions. The LC-MS/MS data provide an approximate map of the SASA of the VKOR MP. Extensive iodination occurs on the most accessible residues of the protein. Further, the enzyme activity assay and CD characterization confirm the functional and structural integrity of the protein upon sonication and addition of the iodination reagent.

In summary, we established a rough relationship between hydrophobicity of the reagent and the preferred labeling location on the MP. This information should allow customization of reagents for MP footprinting. The efficient labeling of intact MP by photo-activated iodine reagents provides insight on their potential as a tool in MP modification in structural biology. These encouraging results should lead to further developments and application to challenging transmembrane proteins.