Simple method to introduce an ester infrared probe into proteins

Abstract

The ester carbonyl stretching vibration has recently been shown to be a sensitive and convenient infrared (IR) probe of protein electrostatics due to the linear dependence of its frequency on local electric field. While an ester moiety can be easily incorporated into peptides via solid‐phase synthesis, currently there is no method available to site‐specifically incorporate it into a large protein. Herein, we show that it is possible to use a cysteine alkylation reaction to achieve this goal and demonstrate the feasibility of this simple method by successfully incorporating a methyl ester group (—CH₂COOCH₃) into a model peptide (YGGCGG), two amyloid‐forming peptides derived from the insulin B chain and Aβ, and bovine serum albumin (BSA). IR results obtained with those peptide and protein systems further confirm the utility of this vibrational probe in monitoring, for example, the structural integrity of amyloid fibrils and ligand binding‐induced changes in protein local hydration status.

Introduction

Infrared (IR) spectroscopy is a well‐established and widely used technique for studying protein structure, dynamics and function.1, 2 This is because a protein's backbone and sidechains can give rise to a multitude of vibrational transitions, some of which are sensitive to factors that depend on either environment, conformation, or both.3 For example, the amide I′ band arising from the backbone carbonyls has been extensively used to assess protein folding dynamics and mechanisms due to its sensitivity to protein secondary structures.4 Despite the proven utility of various intrinsic vibrational modes of proteins, it is found that in many applications naturally occurring protein IR signals either cannot provide any insight or can only provide limited information about the biochemical or biophysical process of interest. Therefore, the past decade has seen a continued effort to develop extrinsic IR probes that can circumvent this limitation.1, 5 For instance, to improve the structural and spatial resolution of IR measurements, non‐natural amino acids consisting of a unique vibrational probe, such as a —CN, —SCN, or —N₃ moiety, have been utilized in various studies, including those concerning protein folding,6 amyloid formation,7 the structure and function of membrane proteins,8, 9 electrostatic field of enzymes,10, 11, 12, 13, 14 and drug binding.15, 16

Recently, Pazos et al.17 have shown that the ester carbonyl stretching vibration is a useful and convenient site‐specific IR probe of protein electrostatics because (1) its frequency is linearly correlated with the local electrostatic field, even when the carbonyl group is engaged in hydrogen‐bonding interactions; (2) its frequency is located in the spectral range of 1700–1800 cm⁻¹, wherein no intrinsic protein vibrations show up at neutral pH; and (3) it has a relatively high extension coefficient. In particular, Pazos et al.17 have tested and demonstrated the utility of two ester‐containing non‐natural amino acids, l‐aspartic acid 4‐methyl ester (hereafter referred to as D_M) and l‐glutamic acid 5‐methyl ester (hereafter referred to as E_M), by incorporating them into various model peptide systems via solid‐phase peptide synthesis. In contrast, other non‐natural amino acid‐based IR probes, such as p‐cyanophenylalanine18, 19 and p‐azido‐phenylalanine,20, 21 have been incorporated into large proteins in vivo via amber codon suppression.22 Currently, there are no in vivo methods available to site‐specifically introduce either D_M or E_M into a protein of interest. Herein, we demonstrate a simple, robust method for selective incorporation of a methyl ester (hereafter referred to as ME), an IR probe analogous to D_M, into proteins.

As shown (Scheme 1), alkylation through an S_N2 reaction between a free thiol (—SH) and a haloalkyl reagent is commonly used to covalently attach a desired functional (R) group, such as an alkyl, alkenyl, or aryl moiety, to the molecule of interest.23 Because such alkylation reactions can be carried out under mild reaction conditions, they have found useful applications in protein post‐translational modifications. For example, cysteine alkylation24 has been used to incorporate various spectroscopic probes.25, 26 In this study, we further expand the utility of this simple reaction by showing that it can be exploited to site‐specifically label a peptide or protein with an ester IR probe. In addition, we verify the applicability of this IR probe by using it to assess the structural heterogeneity of amyloid fibrils and a ligand‐binding induced local hydration status change in bovine serum albumin.

Scheme 1.

Generalized sulfhydryl alkylation reaction with haloalkyl reagents.

Results and Discussion

Ester incorporation into short peptides

To demonstrate the feasibility of selective incorporation of an ester moiety into a protein via the aforementioned cysteine alkylation reaction, we first tested this method using a short, C‐terminally amidated peptide, YGGCGG, where the Tyr residue is introduced to help determine the peptide concentration. As indicated (Scheme 2), the alkylation reaction involves the use of an α‐halocarbonyl electrophile, methyl bromoacetate, and is carried out under mild reaction conditions, which converts the thiol group (—SH) of the Cys residue in YGGCGG to —SCH₃COOCH₃ (hereafter referred to as —SME) with a high yield (approximately 90%). The identity of this cysteine alkylation product, YGGC^*GG, where C^* represents Cys‐SME, was verified by mass spectrometry (Supporting information Fig. S1). This reaction also works well in the presence of 8 M urea, which may be needed in cases where the target cysteine residue is buried in the interior of the protein of interest.

Scheme 2.

Methyl ester incorporation into peptides via cysteine alkylation.

As shown (Fig. 1), comparison between the Fourier transform infrared (FTIR) spectra of YGGCGG and YGGC^*GG obtained in D₂O also validates the successful incorporation of the ME group. This is because the YGGC^*GG peptide, but not the YGGCGG peptide, gives rise to an IR band within the region of 1700–1750 cm⁻¹.17 As expected, this ester carbonyl stretching vibrational band is broad and peaked at about 1720 cm⁻¹, which is similar to that observed for a short peptide containing a D_M residue, due to hydrogen‐bonding interactions with the solvent molecules.17

Figure 1.

FTIR spectra of YGGCGG and YGGC^*GG in the amide I′ region, as indicated. The peptide concentration was about 2.5 mM in D₂O

In the second test of the cysteine alkylation strategy, we sought to incorporate the ME IR probe into two peptides that have been shown to form amyloid aggregates. The first peptide corresponds to the 9–23 segment of the insulin B chain, which contains one cysteine residue (sequence: SHLVEALYLVCGERG, hereafter referred to as the IB peptide). The study by Ivanova et al.27 indicated that the LVEALYL segment within the IB peptide can forms fibrils and may play an important role in the fibril formation of the full length insulin B chain. The second peptide correspond to a double mutant of Aβ_16–22 (sequence: KCVK^*FAE, hereafter referred to as the Aβ peptide), with a cysteine at the 17 position and a lys‐Nvoc (Fmoc‐Lys(4,5‐dimethoxy‐2‐nitro‐benzyloxycarbonyl)‐OH, K^*) at the 19 position. Following the study of Measey and Gai,28 the lys‐Nvoc was used in this case to facilitate amyloid fibril formation. We found that the ME group can be incorporated into both peptides using the reaction conditions described in Scheme 2 and in both cases the yield of the resultant peptide product (hereafter referred to as IB‐SME and Aβ‐SME, respectively) was approximately 90% and was verified by mass spectrometry (Supporting information Figs. S2 and S3).

As indicated (Fig. 2), the FTIR spectra of IB‐SME and Aβ‐SME obtained in deuterated phosphate buffer (50 mM, pH 7.0) confirm that both peptides can aggregate as their amide I′ bands consist of a sharp feature at about 1625 cm⁻¹, due to formation of aggregates composing of tightly packed β‐strands.3 In addition, both peptide samples give rise to an IR band characteristic of an ester carbonyl stretching vibration, further validating the successful incorporation of the ME moiety. Interestingly, the ester carbonyl stretching vibrational bands of these two aggregated samples show distinct differences; the one from IB‐SME is broad and centered around 1725 cm⁻¹, whereas the one from Aβ‐SME consists of two major and narrower peaks centered at 1722 and 1743 cm⁻¹. The study of Pazos et al.17 demonstrated that when the carbonyl group of an ester moiety is engaged in a hydrogen‐bonding interaction with a protic solvent, its peak frequency is expected to be lower than 1730 cm⁻¹. In addition, the bandwidth of an IR transition in the condensed phase is typically a manifestation of the degree of heterogeneity of its underlying environment. Therefore, the above IR results suggest that under the current experimental conditions the aggregates formed by IB‐SME are not well‐structured, leaving a large fraction of the —SME groups exposed to D₂O, whereas those formed by Aβ‐SME contain well‐organized fibrils, leading to the burial of a large fraction of the —SME groups in a dehydrated environment and, hence, producing a sharp IR peak at 1743 cm⁻¹. To further corroborate these assessments, we also examined the structure and morphology of these aggregates using atomic force microscopy (AFM). Consistent with our expectations, the AFM images (Fig. 3) clearly show that the Aβ‐SME peptide forms distinct fibrils, whereas the IB‐SME peptide produces mostly globular aggregates under our experimental conditions. Thus, taken together, these results highlight the potential utility of the ester carbonyl stretching vibration as a convenient means to probe the structural integrity of amyloid fibrils formed by peptides and proteins.

Figure 2.

FTIR spectra of IB‐ME and Aβ‐ME in the amide I′ region, as indicated. The peptide concentration was about 4 mM in deuterated phosphate buffer (50 mM, pH 7.0). The spectrum for Aβ‐ME has been scaled by a factor of 2.4 for easy comparison. Shown in the inset are the spectra in the frequency region of 1700–1760 cm⁻¹

Figure 3.

AFM images of the aggregates formed by IB‐ME (A) and Aβ‐ME (B) peptides. For both cases, the incubation time was 10 days

Ester incorporation into bovine serum albumin

To test whether the aforementioned chemical method can be used to incorporate a ME group into large proteins, we carried out the same cysteine alkylation reaction described above on bovine serum albumin (BSA), a 603‐residue protein that has 35 cysteine residues but only one free cysteine residue (i.e., Cys34). Experimental results obtained by FTIR spectroscopy (Fig. 4) confirm the incorporation of the ME group (hereafter the ester‐labeled BSA is referred to as BSA‐SME). In particular, the ratio between the integrated area of the amide I′ band and that of the ester carbonyl stretching band of the ester‐labeled BSA is approximately 600:1, indicating that the protein is labeled with only one ME group. Using this information and the molar quantities of BSA (reactant) and BSA‐SME (product), we estimated the ME labeling efficiency to be around 90%. Furthermore, since methyl bromoacetate can potentially react with other residues, such as Ser, Lys, Asp, Met, and His, we ruled out this possibility by means of trypsin digestion and mass spectrometry. As shown in Supporting information Fig. S4, comparison between the masses of trypsin‐digested fragments of BSA and BSA‐SME corroborates the exclusive addition of the ME group to Cys34. In addition, the circular dichroism (CD) spectra of the wild‐type and ME‐labeled BSA molecules indicate the reaction conditions used do not cause any significant changes in the structural integrity of the protein (Supporting information Fig. S5). In addition, we have successfully repeated the ester labeling reaction of BSA under denaturing conditions (i.e., in 8 M urea), further demonstrating the general applicability of this method to target a buried cysteine residue in the interior of a large protein.

Figure 4.

FTIR spectra of BSA and BSA‐SME in the amide I′ region, as indicated. The protein concentration was about 1 mM in deuterated phosphate buffer (50 mM, pH 7.0). The spectrum of BSA‐SME has been normalized against that of BSA for easy comparison. Shown in the inset are the spectra in the frequency region of 1700–1780 cm⁻¹

Previous studies29, 30 have shown that BSA can interact with various ligands, such as fatty acids. Crystallographic studies showed that upon complexion with a fatty acids, such as oleic acid31 or myristate,32 human serum albumin (HSA), which is very similar to BSA33 in sequence and structure, shows a substantial conformational change which, in particular, makes Cys34 more solvent accessible. In free BSA, Cys34 is located in a cavity approximately 9 Å from the protein surface, thus being protected from bulk water.34, 35 Since Cys34 in BSA‐SME is labeled with an ester IR probe, we set to use IR spectroscopy to verify whether BSA undergoes a similar binding‐induced conformational change as HSA. As shown (Fig. 5), the ester carbonyl stretching vibrational band of BSA‐SME in buffer is peaked at ∼1742 cm⁻¹. However, upon addition of oleic acid (OA), which is known to bind to BSA with a binding constant (K _d) of 3.96 µM,30 this band is shifted to ∼1737 cm⁻¹. This red‐shift indicates that interaction with OA causes the —SME group to be more solvent exposed, a result that is consistent with the crystallographic studies.31, 32 However, since the ester carbonyl stretching vibrational band of a completely hydrated —SME group is peaked at a frequency of less than 1730 cm⁻¹ (Fig. 1), this result suggests that Cys34 is only partially hydrated even in the OA‐bound state of BSA.

Figure 5.

FTIR spectra of BSA‐SME in the presence and absence of oleic acid (OA), as indicated. The protein concentration was about 1 mM in deuterated phosphate buffer (50 mM, pH 7.0)

Conclusion

In summary, we demonstrate a simple, post‐translational method to site‐specifically incorporate a methyl ester moiety into proteins via cysteine alkylation. Because the carbonyl stretching frequency of this ester exhibits a linear dependence on local electric field and is sensitive to interactions with water, this method, which requires only mild reaction conditions and can be easily implemented, is expected to be useful in various applications. For example, in this study we show that this IR reporter can be used to characterize the structural heterogeneity of peptide aggregates and to site‐specifically assess the local hydration status of a protein, as well as changes in this status due to ligand binding.

Materials and Methods

Peptide synthesis and sample preparation

All peptides were synthesized using standard 9‐fluorenylmethoxycarbonyl (Fmoc) solid‐phase synthesis protocol with Fmoc‐protected amino acids from either Bachem Americas (Torrence, CA) or AnaSpec (Fremont, CA) on a CEM (Matthews, NC) Liberty Blue automated microwave peptide synthesizer. In each case, the C‐terminus was capped by amidation. All synthesized peptides were cleaved from the corresponding resin using trifluoroacetic acid and the resultant crude sample was purified by reverse‐phase HPLC (Agilent Technologies 1260 Infinity) using a C18 column (Vydac). The mass of every peptide was verified by matrix‐assisted laser desorption/ionization mass spectrometry (MALDI‐MS). Peptide samples used in cysteine ligation reactions were prepared by dissolving lyophilized peptides into 50 mM sodium phosphate buffer (pH 5) with a final peptide concentration of 1–5 mM, determined optically using the absorbance of either tryptophan, tyrosine, or Lys(Nvoc) at 280 nm or 350 nm. Bovine Serum Albumin (BSA) was purchased from Sigma‐Aldrich and used as received.

Ester labeling via cysteine alkylation

For peptide labeling, an appropriate amount of lyophilized peptide solid was first dissolved in sodium phosphate buffer (50 mM, pH ∼5) to reach a final peptide concentration of 1 mM. Then, an aliquot of a concentrated acetonitrile solution of methyl bromoacetate (100 mM) was added to 5 mL of this peptide solution to reach a 10:1 molar ratio between methyl bromoacetate and the peptide. The reaction mixture was left to slowly shake on a table top rocker for at least 8 h at room temperature. Reaction progress was monitored by mass spectrometry until the majority of the starting material was converted to the desired product. The excess methyl bromoacetate from the synthesis was first removed by HPLC, the resultant product was then lyophilized and reconstituted into sodium phosphate buffer (50 mM, pH 7). The procedures used for BSA labeling are identical to those described above with the only differences being (1) the methyl bromoacetate stock solution was prepared in dimethyl sulfoxide (DMSO), (2) a PD‐10 column (GE) was used to remove the excess methyl bromoacetate post reaction, and (3) the protein sample was then dialyzed overnight using a Slide‐A‐Lyzer (Thermo Scientific) dialysis cassette with a 10,000 molecular weight cutoff against a sodium phosphate buffer (50 mM, pH 7) to remove any remaining methyl bromoacetate. For labeling reactions conducted in denaturing conditions, the peptide/protein sample was prepared in 8 M urea sodium phosphate buffer (50 mM, pH ∼5) while keeping the other conditions the same. In addition, the removal of urea and unreacted methyl bromoacetate was achieved through three rounds of dialysis against a sodium phosphate buffer (50 mM, pH 7.0).

Trypsin digestion of BSA and BSA‐SME

Trypsin was purchased from Promega. The protein to be digested was first denatured by incubating it in a pH 8 buffer containing 8 M urea, 50 mM Tris‐HCl and 5 mM dithiothreitol (DTT) at 37°C for 1 h. Then the protein sample was diluted with a pH 7.8 buffer containing 50 mM Tris‐HCl and 2 mM CaCl₂ until the urea concentration reached 0.5 mM. Trypsin was then added to this solution at a ratio of 1:10 (trpsin:protein). The digestion process was allowed to proceed for 24 h and then the trypsin activity was halted by lowering the pH to 4 with acetic acid. The digested protein sample was then passed through a Pierce® C‐18 spin column (Thermo) and subsequently analyzed by MALDI‐MS.

FTIR and CD measurements

FTIR spectra were collected on a Nicolet 6700 FTIR spectrometer using 1 cm⁻¹ resolution and a CaF₂ sample cell with a Teflon spacer (52 µm). Sample concentrations were in the range of 1–5 mM. For all reported spectra, a solvent background has been subtracted. The oleic acid‐bound BSA‐SME samples were prepared using the protocol described by Spector et al.30 Specifically, a 1 mM BSA‐ME solution (in 50 mM deuterated phosphate buffer, pH 7.0) was first prepared and then an appropriate aliquot of oleic acid in heptane was added to this solution to reach a final oleic acid concentration of 10 mM. The mixture was slowly shaken continuously and incubated at 37°C for 24 h in a water bath. The aqueous layer of this mixture was used for further FTIR measurements. CD spectra were collected on an AVIV 410 spectrometer using a 1 mm quartz cell. Sample concentration was ∼40 µM.

Atomic force microscopy (AFM) measurements

AFM experiments were performed at room temperature, using an atomic force microscope (Bruker Dimension Icon AFM, Billerica, MA). To ensure optimal surface coverage, the aggregated peptide sample under investigation was first diluted to ∼1 mM (monomer concentration) with deionized water and then a 20 µL of this diluted peptide sample was deposited onto the mica surface. This sample was allowed to settle on the mica surface for 1 min before washing with 0.5 ml deionized water and drying under a stream of N₂ gas. AFM images were processed and analyzed using Gwyddion, a free program for visualizing and analyzing scanning probe microscopy data.

Supporting information

Figure S1. MADLI mass spectra for the unlabeled model peptide YGGCGG (black) and labeled peptide YGGC^*GG (red). The expected mass for YGGCGG is 511.2 Da and a value of 534.33 Da was observed, which has an adduct of 23 Da from a sodium molecule. The expected mass for YGGC^*GG is 583.2 Da and a value of 606.33 Da was observed, which has an adduct of 23 Da from a sodium molecule. This increase in 72 Da corresponds to the successful addition of the methyl ester (ME) moiety to YGGCGG

Figure S2. MADLI mass spectra for wildtype insulin B (IB) (black) and ME‐labeled IB‐SME (red). The expected mass for IB is 1588.85 Da and a value of 1588.29 Da was observed. The expected mass for IB‐SME is 1660.85 Da and a value of is 1660.32 Da was observed. This increase in 72 Da corresponds to the successful addition of the ME moiety to IB

Figure S3. MADLI mass spectra for an Aβ peptide (black) and Aβ‐SME (red). The expected mass for Aβ is 1062.17 Da and a value of 1061.99 Da was observed. The expected mass for Aβ‐SME is 1134.17 Da and a value of 1134.26 Da was observed. This increase in 72 Da corresponds to the successful addition of the ME moiety to Aβ

Figure S4. MADLI mass spectra for trypsin digested BSA (black) and BSA‐SME (red). The expected mass for BSA fragment containing Cys34 is 2435.24 Da and a value of 2458.47 Da was observed (Na⁺ ion adduct). The expected mass for BSA‐SME fragment containing Cys34 is 2507.24 Da and a value of 2530.15 Da was observed (Na⁺ ion adduct). This mass increase of 72 Da verifies the successful addition of a ME moiety to BSA at Cys34. There are no additional mass adducts that deviate from the expected masses of BSA fragments, indicating that ME is exclusively added to Cys34

Figure S5. Circular Dichroism (CD) spectra of BSA (red) and BSA‐SME (blue) which shows no major difference between the secondary structures of wild type and ME‐labeled BSA, indicating that the reaction conditions used do not cause any significant changes in the structural integrity of the protein