pH- and Temperature-Dependent Peptide Binding to the Lactococcus lactis Oligopeptide-Binding Protein A Measured with a Fluorescence Anisotropy Assay.

Bacterial ATP-binding cassette transporters are a superfamily of transport systems involved in the import of various molecules including amino acids, ions, sugars, and peptides. In the lactic acid bacteria Lactococcus lactis, the oligopeptide-binding protein A (OppA) binds peptides for import to support nitrogen metabolism and cell growth. The OppA protein is of great interest because it can bind peptides over a broad variety of lengths and sequences; however, current methods to study peptide binding have employed low throughput, endpoint, or low dynamic range techniques. Therefore, in this study, we developed a fluorescence anisotropy-based peptide-binding assay that can be readily employed to quantify OppA function. To test the utility of our assay, we characterized the pH dependence of oligopeptide binding because L. lactis is commonly used in fermentation and often must survive in low pH environments caused by lactic acid export. We determined that OppA affinity increases as pH or temperature decreases, and circular dichroism spectroscopy further indicated that acidic conditions increase the thermal stability of the protein, increasing the unfolding transition temperature by 10 °C from pH 8 to pH 6. Thus, our fluorescence anisotropy assay provides an easy technique to measure peptide binding, and it can be used to understand molecular aspects of OppA function under stress conditions experienced during fermentation and other biotechnology applications.

Introduction

The lactic acid bacteria Lactococcus lactis is vital to the dairy industry and has become a synthetic biology platform for recombinant protein production, metabolic engineering, and the development of probiotics and vaccines.[1−8] Like many bacteria, L. lactis relies on extracellular peptides as a major source of nutrients to support carbon and nitrogen metabolism, and peptide transport plays an important role in bacterial growth and survival. In general, bacteria can utilize proton-coupled transporters and ATP-binding cassette (ABC)-type transporters to import di-, tri-, and oligopeptides. For L. lactis, which is a branched-chain amino acid auxotroph, the oligopeptide ABC transporter is necessary for growth on natural substrates such as milk casein peptides.[9−14] The oligopeptide ABC transporter is a multi-subunit complex consisting of two transmembrane proteins (OppB and OppC), two ATP-binding proteins (OppD and OppF), and a substrate-binding protein [oligopeptide-binding protein A (OppA)].[12] The L. lactis OppA is part of a superfamily of substrate-binding proteins that share a similar overall tertiary structure and domain organization.[15−17]

OppA is particularly interesting because it has a broad capacity to bind peptides of varying lengths and sequences[18−25] unlike other substrate-binding proteins that have high specificity for ligands such as metal cofactors and sugars.[15−17,26,27] The crystal structures of several OppA proteins from diverse bacteria have been solved, including that of L. lactis OppA.[17,28−31] In general, substrate-binding proteins consist of two domains connected by a hinge region with the substrate-binding pocket located between the two domains. Upon binding substrate, the protein undergoes a conformational change from open to closed as it clamps down on the substrate, which is known as the “Venus flytrap” mechanism.[26,27,30,32] The OppA proteins also undergo the canonical substrate-dependent conformational change (Figure ), but they have an additional third domain that increases the size of the binding pocket, which classifies them as a cluster C substrate-binding protein.[28−30] The L. lactis OppA accepts peptides ranging from 4 to 35 residues in length with a preference for hydrophobic and basic ligands. Overall, it has relatively low specificity for the peptide sequence, which is likely because hydrogen bonds are almost exclusively formed between the protein and the peptide backbone and not with peptide side chains.[18−25] The crystal structures of L. lactis OppA reveal that the peptide-binding site is a large, aqueous cavity with a single hydrophobic pocket that can accommodate a bulky residue. Crystal structures with peptide bound show that the termini of peptides are not in fixed positions and peptides of same length and composition but differing sequences can bind different registers.[29,31] Thus, peptide selection is based on composition rather than exact sequence, which allows OppA to import a wide range of peptides for nutrients.

Figure 1

Fluorescence anisotropy peptide-binding assays used to quantify relative changes in L. lactis OppA binding affinity. (A) Substrate peptide binding causes a “Venus flytrap” conformational change from an open state (PDB 3DRK) to a closed state (PDB 3DRG). (B) Dye-labeled peptide rotates freely in solution with low fluorescence anisotropy. Once bound to OppA, the fluorescent peptide–OppA complex rotates more slowly, causing an increase in fluorescence anisotropy. (C) For competition binding assays, the fluorescent peptide is prebound to OppA. The unlabeled peptide displaces the fluorescent peptide, causing a decrease in fluorescence anisotropy. Green, blue, and orange indicate lobes Ia, Ib, and II in the protein structures, respectively. The unlabeled peptide is pink, and the dye-labeled peptide is shown in red with a star.

While the specificity of OppA peptide binding and its role in nutrient import have been well-established, no studies have been performed to characterize the peptide-binding properties of OppA under stress conditions. Lactic acid bacteria frequently experience stress conditions particularly when used in food fermentation.[33]L. lactis exhibits optimal growth at pH 6.3 and at 30 °C; thus, stress conditions could include both pH and temperature changes that still reside within mesophilic ranges.[34,35] Furthermore, several studies have shown differential expression of genes involved in amino acid metabolism under stress conditions, including and implicating OppA in stress response.[36−38] Thus, in this study, we developed a fluorescence anisotropy assay that can be used to measure the effects of pH and temperature on peptide binding to the L. lactis OppA protein. To quantify relative changes in OppA function and structure, we used a combination of our fluorescence anisotropy peptide-binding assays and circular dichroism (CD). We discovered that increased peptide affinity under acidic conditions correlates with increased structural stability and neutralization of surface charge within the substrate-binding pocket, pointing to a potential molecular adaptation to the optimal growth conditions for L. lactis.

Results

Development of a Fluorescence Anisotropy Assay for OppA Peptide Binding

To quantify relative differences in OppA peptide affinity, we developed fluorescence anisotropy peptide-binding assays because they are nonradioactive solution-state assays that offer high signal over background with a good dynamic range and ease of execution.[39−41] We based our assay on the binding of a sulforhodamine 101 (SR101)-labeled peptide to OppA (Figure ). We chose SR101 because it is commercially available for custom dye-modified peptide synthesis and it is water-soluble, bright, and has a fluorescence lifetime of ∼4.2 ns that makes it sensitive to polarization changes upon binding large molecular weight proteins. As described below, SR101 was highly effective as an anisotropy reporter, and therefore, we did not test other fluorophores. However, in principle, our assay should work with other dyes.

We then chose to conjugate SR101 to the bradykinin peptide (RPPGFSPFR) because it has previously been used to characterize peptide binding to OppA. Bradykinin is a peptide hormone that is vital to endocrine signaling throughout mammalian tissues, including the gut. In theory, it is also an example of an endogenous peptide that could be used as a nutrient, though it is not clear that L. lactis would encounter high concentrations of bradykinin in the gut lumen. For the purpose of this study, the primary reason we chose to use bradykinin is because bradykinin has been used extensively in other functional assays for the Opp-transport system.[23,25,42,43] Furthermore, we also chose to label bradykinin because it is the highest affinity ligand reported for L. lactis OppA (KD = 0.1 μM),[21] which we hypothesized would mitigate any loss of affinity caused by conjugation to SR101. As demonstrated in our results, the SR101-conjugated bradykinin provided excellent anisotropy detection characteristics, and therefore, we did not screen other peptides to label.

We initially tested OppA binding using bradykinin that was labeled on either its N- or C-terminus (SR101-RPPGFSPFR and RPPGFSPFRK-SR101). The C-terminal configuration included a lysine residue that was necessary for coupling to the dye (Figure ) (Figure S3). To determine the affinities of the two different dye-labeled peptide configurations, direct-binding dose response curves were performed with varying concentrations of OppA (Figure ). The affinities of OppA for the N and C-terminally labeled bradykinin were determined to be 45 ± 3 and 13.1 ± 0.7 μM, respectively (mean ± stdev, n = 3). The higher affinity for the C-terminally labeled peptide aligns well with previous studies that showed that OppA tolerates large bulky groups on the C-terminus of a peptide but not the N-terminus.[22] Additionally, the crystal structure of OppA in complex with bradykinin reveals that the entire peptide can be accommodated within the binding pocket, with the N-terminus buried more deeply within the binding pocket compared to the C-terminus.[29] The additional lysine residue on the C-terminally labeled peptide may also contribute to the higher affinity relative to the N-terminally labeled peptide. The higher affinity C-terminally labeled peptide, referred to as “bradykinin-SR101”, was selected for use in subsequent assays because of its greater binding affinity, which provided good sensitivity to binding over a range of submicromolar to millimolar concentrations. Using bradykinin-SR101, our assay exhibited a dynamic range with a maximal increase of 0.201 ± 0.002 in fluorescence anisotropy between the free and OppA-bound states, and the reproducibility was excellent across independent protein preparations (Figure ).

Figure 2

Direct binding to L. lactis OppA measured by fluorescence anisotropy. Bradykinin labeled with sulforhodamine 101 on the N-terminus (SR101-RPPGFSPFR) exhibited lower affinity for OppA compared to bradykinin labeled on the C-terminus (RPPGFSPFRK-SR101). The lines show the average fit for n = 3. Error bars are stdev.

We next validated that the fluorescent dye-labeled bradykinin-SR101 binds to the same site as unlabeled peptides using fluorescence anisotropy competition assays (Figure ). After optimizing assay conditions for sensitivity in our system, we used a prebound complex of OppA and bradykinin-SR101 at approximately 50% saturation and a half-max anisotropy in the absence of the competitor peptide. Typically, 50–80% saturation provides an excellent signal window because there is sufficient anisotropy to be sensitive to probe displacement while avoiding experimental uncertainty and signal dampening caused by complete saturation.[44] In this competition assay, the addition of the unlabeled competitor peptide results in a decrease in the anisotropy when the OppA-bound dye-labeled peptide is displaced. As expected, for peptides that bind to the same site, unlabeled bradykinin was able to completely displace prebound bradykinin-SR101 (Figure ). We additionally validated that another reported high-affinity peptide derived from casein was able to completely displace bradykinin-SR101. Furthermore, the low-affinity peptide, neuropeptide S, was able to partially displace bradykinin-SR101 within the tested peptide concentration range, which was limited by peptide solubility (Figure ). Thus, our competition assays show that bradykinin-SR101 exhibits a normal mode of binding to OppA.

Figure 3

Competition binding to OppA measured by fluorescence anisotropy validates that bradykinin-SR101 binds to the canonical substrate binding site. (A) Unlabeled competitor peptides are able to displace bradykinin-SR101. The previously reported high-affinity bradykinin and a casein-derived peptide are able to completely displace bradykinin-SR101. The previously reported low-affinity peptide, neuropeptide S, partially displaces bradykinin-SR101 because the dose–response was limited by our peptide concentration range. (B) Related opioid peptides Leu-enkephalin, dynorphin-A (1–9), and dynorphin-A (1–17) exhibit increasing affinity with increasing length, respectively. The lines show the average fit for n = 3. Error bars are stdev.

Interestingly, we were also able to use our competition assay to determine the relative affinities of leu-enkephalin and dynorphin opioid peptides that have not been previously reported. Leu-enkephalin (YGGFL) has been previously reported to bind OppA with low affinity, which could not be quantified with gel shift or intrinsic protein fluorescence assays.[18,21,29] Using our fluorescence anisotropy assay, we found that leu-enkephalin displaced bradykinin-SR101 with an IC50 ≈ 2 mM (Figure ) (Table ). Furthermore, dynorphin-A (1–9) (YGGFLRRIR) exhibited an IC50 ≈ 34 μM, and dynorphin-A (1–17) (YGGFLRRIRPKLKWDNQ) exhibited an IC50 ≈ 11 μM. The Cheng–Prusoff equation is commonly used to convert the IC50 value to the KD dissociation constant for the unlabeled competitor.[45] However, fluorescence anisotropy competition assays often do not fulfill the assumptions made by the Cheng–Prusoff equation.[44,46,47] Therefore, we directly fitted the competition dose response data to a three-state, two-ligand competition binding model to determine the competitor affinities (Table , Supporting Discussion, and Figures S4–S6).[44,46,47] Our measured affinities for unlabeled bradykinin, the casein-derived peptide, and leu-enkephalin showed excellent agreement with previously reported values,[21,29] validating our competition assay.

Table 1

Binding Characterization of Peptides As Determined by Fluorescence Anisotropy-Based Competitive Binding Assays at pH 6, 30 °C (n = 3, Mean ± SD)

peptide	sequence	IC50 (μM)	KD (μM)a	ref. KD (μM)b
bradykinin	RPPGFSPFR	5.2 ± 0.2	0.13 ± 0.04	0.1 (pH 6, 15 °C)[21] & 0.26 (pH 6, 25 °C)[29]
casein-derived peptide	SLSQSKVLPVPQ	6.33 ± 0.07	0.20 ± 0.01	0.77 (pH 6, 15 °C)[21]
neuropeptide S	SFRNGVGSGVKKTSFRRAKQ	no fit	1800 ± 100	this study
leu-enkephalin	YGGFL	2000 ± 400	66 ± 8	50–100 (pH and temp not given)[29]
dynorphin-A (1–9)	YGGFLRRIR	34.2 ± 0.7	13.3 ± 0.7	this study
dynorphin-A (1–17)	YGGFLRRIRPKLKWDNQ	11.0 ± 0.5	3.3 ± 0.6	this study

Fitted KD values obtained by directly fitting competition dose response data to the competition binding model described in the Supporting Discussion.

Previously reported KD values from the literature.

Previously, it was reported that OppA optimally binds to nonapeptides, but additional residues can contribute to increased affinity.[21,23] Using our fluorescence anisotropy assay, we quantified the trend in relative affinities for a series of closely related leu-enkephalin, dynorphin-A (1–9), and dynorphin-A (1–17) peptides. In particular, it is interesting that dynorphin-A (1–17) has a significantly higher affinity for OppA relative to the shorter dynorphin-A (1–9). This seems to indicate that OppA forms further favorable contacts with the longer peptide sequence, which may be possible due to the particularly voluminous binding cavity of OppA (∼4900 Å3) conferred by the third domain of this substrate-binding protein.[29]

pH and Temperature Dependence of OppA Peptide-Binding

Having established and validated our fluorescence anisotropy assay, we quantified how pH and temperature affect the peptide binding affinity of OppA. We measured the direct binding of bradykinin-SR101 to OppA at pH 5 to 8 at six different temperatures (18, 24, 30, 37, 45, and 55 °C) (Figure ) (Table ), and we validated that unlabeled bradykinin could still bind and displace bradykinin-SR101 under these conditions (Figure ) (Table ). OppA exhibits its highest peptide affinities for the labeled bradykinin-SR101 under low-pH and low-temperature conditions. The highest affinities were measured at pH 6 near its optimal growth pH, and no further increase in affinity was observed at pH 5. In solution at pH 6 and pH 7, increasing temperature causes a decrease in bradykinin-SR101 affinity. Interestingly, at pH 8, the measured peptide affinity did not vary widely from 18 to 30 °C, and the lowest affinities measured across the pH 5–8 range were in a similar concentration range between 20 and 30 μM, suggesting a lower affinity limit within this mesophilic range.

Figure 4

pH and temperature dependence of OppA binding affinity for labeled bradykinin-SR101. Direct binding of bradykinin-SR101 was measured by fluorescence anisotropy at varying temperatures at (A) pH 5, (B) pH 6, (C) pH 6.5, (D) pH 7, and (E) pH 8. The lines show the average fit for n = 3. Error bars are stdev.

Table 2

Characterization of OppA Binding Affinities for Labeled Bradykinin-SR101 at Varying pHs and Temperatures Using Direct Binding Assays (n = 3, Mean ± Stdev)

	direct binding assay KD (μM)
temp (°C)	pH 5	pH 6	pH 6.5	pH 7	pH 8
18	10.0 ± 0.5	8.5 ± 0.4	13.0 ± 0.5	15.4 ± 0.4	20 ± 1
24	12.3 ± 0.5	10.4 ± 0.4	16 ± 1	17.8 ± 0.7	20 ± 1
30	15.4 ± 0.7	13.1 ± 0.7	19 ± 1	19.3 ± 0.8	22 ± 1
37	21.4 ± 0.5	18.0 ± 0.7	26 ± 1	22.9 ± 0.5	15.9 ± 0.5
45	31 ± 1	24 ± 1	30 ± 10	a	a
55	a	a	a	a	a

Not determined because of protein aggregation.

Figure 5

pH and temperature dependence of OppA binding affinity for unlabeled bradykinin. Competitive binding of unlabeled bradykinin was measured by fluorescence anisotropy of displaced, prebound bradykinin-SR101 at varying temperatures at (A) pH 6, (B) pH 7, and (C) pH 8. The lines show the average fit for n = 3. Error bars are stdev.

Table 3

Characterization of OppA Binding Affinities for Unlabeled Bradykinin at Varying pHs and Temperatures Using Competition Binding Assays (n = 3, Mean ± stdev)

	affinities from competition binding assays (μM)
	pH 6		pH 7		pH 8
	IC50	KDb	IC50	KDb	IC50	KDb
18	5.1 ± 0.1	≤0.1c	5.2 ± 0.2	0.2 ± 0.1	5.3 ± 0.4	0.4 ± 0.2
24	5.2 ± 0.3	≤0.1c	5.37 ± 0.03	0.20 ± 0.03	5.6 ± 0.3	0.6 ± 0.3
30	5.3 ± 0.2	0.13 ± 0.04	5.7 ± 0.1	0.36 ± 0.05	6.0 ± 0.5	0.9 ± 0.3
37	5.6 ± 0.1	0.31 ± 0.06	6.3 ± 0.4	0.7 ± 0.1	7.2 ± 0.7	2 ± 1
45	6.8 ± 0.2	0.9 ± 0.1	a	a	a	a
55	a	a	a	a	a	a

Not determined because of protein aggregation.

Fitted KD values obtained by directly fitting competition dose response data to the competition binding model described in the Supporting Discussion.

KD is below the estimation limit for data fitted to the competition binding model.

Our anisotropy competition assays measured a similar trend in which low pH and low temperature favored binding of the unlabeled bradykinin to OppA (Figure ) (Table ). As described above, we directly fitted the competition dose response data to a competition binding model to determine the KD values because the Cheng–Prusoff equation is not valid here (Supporting Discussion, Figures S4–S6). As expected, the measured affinities for unlabeled bradykinin were approximately 10-fold higher relative to the labeled bradykinin-SR101. However, similarities between the behavior of the two peptides indicate that the presence of the SR101 dye itself does not obscure the pH and temperature dependence of the binding equilibrium. We also note that the IC50 values are insensitive to the pH-dependence because of the inherent limitations of the IC50 parameter, and the comparison of the IC50 and Kd values illustrates the importance of our fitting approach. As discussed by Huang,[44] the IC50 value is a nonlinear function of the peptide affinities, protein concentration, and labeled peptide concentration. For any competitor with much higher affinity relative to the labeled peptide affinity, the IC50 approaches a lower bound and no longer correlates with the Kd value. As expected, we observed this effect in our system, which matches well with simulations reported by Huang.[44] Thus, our fitting approach for the determination of Kd values is critical.

Additionally, we observed protein aggregation and precipitation at higher temperatures (Figures and 5) (Tables and 3) (Figure S7). At pH 6, peptide binding was well-behaved at temperatures from 18 to 45 °C, but aggregates were observed at 55 °C that precluded measurement of fluorescence anisotropy. At pH 7 and pH 8, the temperature threshold was lower, and protein aggregation was observed at 45 °C and above. It is also possible that aggregation that was not visible by eye caused the affinity measured at pH 8 and 37 °C to be an outlier relative to lower temperatures at pH 8 (Table ). These results clearly demonstrate that OppA function is disrupted with increasing temperature, but low pH can increase resistance to loss of binding. These observations match well with the lower pH and temperature optimum for L. lactis growth. One possible explanation for the loss of binding at high temperature and high pH is that the OppA protein structure is less stable under these conditions. Therefore, to complement our functional studies using our fluorescence anisotropy assay, we next quantified thermal stability at different pHs more precisely.

Low pH Increases OppA Thermal Stability As Measured by CD

To determine if lower pH confers structural stability and facilitates peptide-binding function at high temperatures, we monitored the unfolding transition and quantified the threshold melting temperature (Tm) by CD spectroscopy. The CD spectrum for OppA shows an α-helical secondary structure signature with minima at 208 and 222 nm, allowing us to monitor unfolding (Figure ). Melting curves were measured from the CD signal at 222 nm in the absence and presence of saturating unlabeled bradykinin at pH 6, 7, and 8 (Figure ). The melting temperature of OppA is higher at lower pH, confirming our hypothesis that lower pH stabilizes the structure of OppA. As expected, binding of bradykinin systematically increases the melting temperature at each pH relative to apo-OppA, and lower pH also increases the thermal stability of the peptide–protein complex (Figure ) (Table ). These data show that the pH-dependent loss of protein stability could explain the decrease in peptide binding to OppA under alkaline conditions and high temperature, though this does not explain the pH-dependent change in OppA affinity for peptides within the folded regime. Thus, we next qualitatively investigated this using electrostatic surface calculations.

Figure 6

Acidic pH increases thermal stability measured by CD. (A) Example of typical OppA CD spectra during a thermal ramp at pH 6. (B) Summary of the pH dependence of the melting temperature (Tm) of OppA with or without unlabeled bradykinin bound. Melting curves determined by CD at 222 nm in the (C) absence and (D) presence of 100 μM unlabeled bradykinin. CD shown as mean residue ellipticity ([θ]mrw) × 10–3. Lines show the average fit for n = 4. Error bars are stdev.

Table 4

OppA Melting Temperature Measured by the CD Signal at 222 nm at pH 6, 7, and 8 in the Presence and Absence of 100 μM Bradykinin (n = 4, Mean ± stdev)

pH	OppA Tm(°C)	OppA + bradykinin Tm(°C)	ΔTm(°C)
6	47.5 ± 0.3	52.2 ± 0.4	4.7 ± 0.5
7	42.8 ± 0.4	47.0 ± 0.5	4.2 ± 0.6
8	37.3 ± 0.1	40.4 ± 0.2	3.1 ± 0.2

Electrostatic Surface Potential at the OppA Peptide-Binding Site

It is possible that a change in the electrostatic environment of the binding site contributes to the pH-dependence of OppA peptide affinity. The electrostatic surface charge of L. lactis OppA[48] was calculated using PROPKA and APBS[49−52] at pH 6, 7, and 8 in the open state (Figure ) and in the closed state (Figure S10). At pH 6, there is a neutralization of the exposed surface area within the peptide-binding site compared to pH 7 and pH 8, which could facilitate binding of hydrophobic and neutral peptides. In contrast, Escherichia coli OppA, which prefers positively charged peptides,[53] exhibits less charge neutralization at pH 6 compared to pH 7 and pH 8, maintaining significant negative surface charge within the peptide-binding site. These observations provide a qualitative rationalization of the pH-dependent effects on peptide binding. In the future, beyond the scope of this technical study, the effects of charge changing mutations of the OppA binding pocket and systematic sequence variation of the peptide ligand could be measured using our fluorescence anisotropy assay.

Figure 7

Electrostatic surface models for OppA calculated at different pHs. The surface charge was calculated and represented using PROPKA and APBS for (A) the L. lactis OppA apo-structure in the open conformation (PDB 3FTO) and (B) the E. coli OppA apo-structure in the open conformation (PDB 3TCH). The dashed circle highlights the peptide-binding site.

Discussion

The lactic acid bacteria L. lactis exhibits optimal growth at pH 6,[34,35] and we have discovered that acidic conditions promote the peptide binding function of OppA that is necessary to support the auxotrophic metabolism of this important species.

In this study, we developed and validated a fluorescence anisotropy peptide-binding assay to quantify relative pH-dependent differences in affinity. In previous work, X-ray crystallography and isothermal titration calorimetry have been extensively used to qualitatively and quantitatively characterize binding function for OppA proteins from different species.[28,29,53−57] In addition, radioactivity, gel shift, and intrinsic protein fluorescence methods have been used to measure dissociation constants and relative affinities for a number of peptides,[18,21,23,25,42] and solvatochromic dye-labeled peptides have been used to characterize positional effects on OppA binding.[22] Fluorescence anisotropy has a number of practical advantages for measuring protein–ligand interactions over these other methods,[39−41] and for our study, it offered an excellent signal and dynamic range. For example, we were able to quantify the binding affinity of leu-enkephalin, which could not be quantified by gel shift or intrinsic protein fluorescence because its affinity to OppA is too low. Highlighting the versatility of our assay, in addition to leu-enkephalin, we measured binding of two related but not previously reported peptides, dynorphin-A (1–9) and dynorphin-A (1–17). This series of peptides exhibited relative affinities ranging from 2 mM to 11 μM, and our assay was sensitive over at least 4-orders of magnitude in substrate concentrations over which we could quantify interactions. To demonstrate that our fluorescence anisotropy assay is broadly useful to study other peptide-binding proteins, we also showed it can be used with the Bacillus subtilis AppA protein (Figure S11) (Table S1).

Next, using our fluorescence anisotropy assay, we discovered that the L. lactis OppA peptide affinity increases at acidic pH. We measured the highest affinity interaction with the bradykinin-SR101 peptide at pH 6 and 18 °C, with affinities ranging from 8.5 μM at 18 °C to 18 μM at 37 °C. At the optimal growth temperature of 30 °C, we measured affinities of 13, 19, and 22 μM at pH 6, 7, and 8, respectively, demonstrating a decrease in binding affinity under neutral and alkaline conditions. We also qualitatively observed a lower temperature threshold for protein aggregation at pH 7 and 8 compared to pH 6. The loss of peptide binding at high temperature and high pH suggested that the OppA protein could be more thermally stable at low pH. In fact, CD spectroscopy clearly showed that the melting temperature for the unfolding transition of OppA was also decreased at pH 7 and 8 relative to pH 6. Both the anisotropy and CD results are consistent with one another. Direct binding of the labeled BK-SR101 peptide and competitive binding of the unlabeled bradykinin peptide exhibit a decrease in affinity at higher temperatures and higher pH. Thus, acidic conditions stabilize the OppA structure and promote its ability to bind substrate peptides for import into the bacteria.

The pH-dependence of OppA peptide binding within the folded regime also suggests that acidification could induce a classic change in surface charge near or within the substrate-binding pocket. Supporting this hypothesis, the electrostatic surface[49−52] of the substrate binding pocket showed a clear neutralization of negative surface charge at pH 6 relative to pH 7 and 8. However, we observed a decrease in bradykinin affinity with increasing pH, suggesting that an electrostatic interaction between the negative surface of OppA and the arginine residues of bradykinin is not the main binding determinant. We did observe that at pH 6, there is a small patch of residual negative surface charge from residues D134, D138, D455, E478, and D483. Thus, at all pH values studied, there is some negative charge on OppA that could contribute to an electrostatic interaction with the arginine residues, helping to explain the high affinity for bradykinin. However, other binding factors may contribute to a greater extent, which is consistent with the literature. For example, Berntsson et al. demonstrated that a hydrophobic pocket in OppA is a major binding determinant and in fact, the phenylalanine residue of bradykinin binds to this hydrophobic pocket.[31] Therefore, another interpretation could be that the neutralization of surface charge at low pH promotes the development of a hydrophobic environment to facilitate binding of bradykinin, such as via interactions with the phenylalanine residue of bradykinin. Alternatively, decreasing pH could promote a conformational change that favors peptide binding, and this may be consistent with our CD results that show an increase in global structural stability at lower pH. In the future, a combination of systematic mutagenesis, structural analysis, and functional analysis with our anisotropy assay can be used to gain further insight into the binding mechanism of OppA.

Overall, this surface charge pattern also correlates well with the previously reported partial preference of L. lactis OppA for hydrophobic and basic peptides and decreased use of negatively charged acidic peptide substrates.[18−25] In contrast, Klepsch et al.[53] have shown that the E. coli OppA prefers positively charged peptide substrates, and they showed that there is an extensive negative surface around the binding site at pH 7. We additionally calculated the surface charge of E. coli OppA at pH 6 and 8 and found that substantial negative charge is maintained over the entire pH range unlike L. lactis OppA. This observation might suggest that binding to E. coli OppA is pH-independent, which would be interesting to study in the future along with other OppA proteins of interest such as OppA from the Lyme disease culprit Borrelia burgdorfei.[58,59]

There continues to be a strong interest in understanding the fundamental determinants of optimal growth and metabolism of L. lactis because of its significant industrial and biomedical potential.[1,4,6−8,38,60,61] We focused our efforts studying the purified OppA protein because it is the main determinant for peptide specificity of the Opp transport complex, and in the future, it may be possible to adapt our assay to study binding to the whole transport complex reconstituted in liposomes.[25] Our current study suggests that the L. lactis OppA protein is well-tuned to solution conditions experienced during optimal growth, such as acidic pH and lowered temperature. It might also suggest that direct molecular manipulation of OppA or other substrate-binding proteins could be employed in strain engineering to adapt L. lactis to different metabolic conditions.[42] Furthermore, future studies of the pH-dependent structure and function of OppA from other species could provide similar insights into the sporulation of B. subtilis or the virulence of B. burgdorfei, for example, in which an oligopeptide-binding protein is also a critical aspect of auxotrophy.[58,59,62,63] Overall, the fluorescence anisotropy assay that we developed provides an easy and versatile method to quantify the function of oligopeptide binding proteins in general.

Materials and Methods

Reagents and Materials

Chemicals and media were purchased from Fisher Scientific, Sigma, and Formedium. High-fidelity master mix for Gibson Assembly was from New England Biolabs (Cat# M0492). Bradykinin peptide (RPPGFSPFR) was purchased from and HPLC-purified (98% purity) by Bachem. All other unlabeled peptides were custom-synthesized and HPLC-purified (>95% purity) by GenScript. The sulforhodamine 101 labeled bradykinin peptides (SR101-RPPGFSPFR and RPPGFSPFRK-SR101) were synthesized and HPLC-purified (>90% purity) by Anaspec.

Plasmid Construct

The L. lactis OppA amino acid sequence is from subspecies cremoris MG1363 (GenBank accession AAO63470.1). The N-terminal signal peptide for palmitoylation and surface tethering was removed similar to previous studies (Figure S1).[21,22,29] The nucleotide sequence was optimized to minimize hairpins for cloning and was synthesized as a gBlock by Integrated DNA Technologies (IDT). The OppA gBlock was cloned into a pRSETB vector by Gibson Assembly for bacterial expression and purification (Figure S2).

OppA Expression and Purification

The polyhistidine-tagged protein was expressed in BL21(DE3) E. coli in Auto Induction Media (AIM) purchased from ForMedium (Cat# AIMLB0205). Heterologous expression was used to avoid copurification with endogenous L. lactis peptides, and there was no evidence that endogenous peptides interfered with any assays reported here. First, single colonies were picked and used to inoculate 4 mL of Luria Broth (LB) starter cultures. Starter cultures were grown overnight with continuous shaking at 37 °C and then adjusted to an OD600 of 0.6 with LB. Large AIM cultures (250 mL) were inoculated with 2.5 mL of 0.6 OD600 LB starter cultures and grown at room temperature (RT) for ∼65 h with continuous shaking at 160 rpm. Cultures were pelleted at 1000g for 15 min and lyzed by sonication. First, pellets were frozen at −80 °C and thawed at RT twice. The pellets were resuspended in Tris buffer (50 mM Tris HCl, 300 mM NaCl, 10% v/v glycerol, 15 mM imidazole, pH 8.0) with 0.2 mg/mL lysozyme, 0.1% v/v Triton X, 1 mM phenylmethylsulfonyl fluoride, and 1 mM dithiothreitol (DTT) and rotated end-over-end at RT for 20 min. The lysate was sonicated (QSonica, LLC model# Q125) on ice for 1 min (pulse 2 s on with 2 s off, 80% amplitude) with 2 min rest, and sonication was repeated two more times. The lysate was pelleted at 30 000g for 30 min, and the supernatant was filtered through a 0.45 μm low protein-binding Millex syringe filter, Durapore (Cat# SLHV033NS). The protein was purified in Tris buffer by nickel affinity chromatography using a GE Healthcare HiTrap IMAC HP column (Cat# 45-000-163) loaded with 100 mM Ni2+ and eluted by gradient with 15–500 mM imidazole on a GE Healthcare AKTA purifier. The eluted protein was collected and dialyzed in sodium phosphate buffer (25 mM sodium phosphate, 150 mM NaCl, 10% glycerol, pH 7.0). The dialyzed protein was concentrated to 1 mL with an EMD Millipore Amicon Ultra centrifugal filter with 10 000 MWCO and purified by size exclusion chromatography (SEC) on tandem GE Superdex 200 Increase 10/300 GL columns (product # 28990944, total bed volume = ∼48 mL) in sodium phosphate buffer. The concentration of the purified protein was determined by absorbance at 280 nm using an extinction coefficient of 99 140 M–1 cm–1 calculated using the Northwestern University Peptide Properties Calculator (http://biotools.nubic.northwestern.edu/proteincalc.html#helpexco).

Fluorescence Anisotropy Assay

For direct dose response binding, 100 μL of 1.5 μM dye-labeled bradykinin (SR101-RPPGFSPFR or RPPGFSPFRK-SR101) was added to each well in a 96-well nonbinding microplate (VWR, 89089-582). Purified OppA (50 μL) was added to each well at final concentrations varying from 0.01 to 100 μM and incubated at the desired temperature for at least 30 min to reach equilibrium. Unless otherwise noted, all anisotropy assays were performed at 30 °C in assay buffer (25 mM sodium phosphate, 150 mM NaCl, 10% glycerol, 1 mM DTT, 1× SIGMAFAST protease inhibitor cocktail, 0.05% v/v TWEEN, pH 6.0). The assay proved robust and adaptable to different assay conditions, and protease inhibitors and detergent did not interfere for example. For all protein assays, at least two separately expressed and purified batches of OppA were used. Fluorescence anisotropy was measured using a BioTek Synergy H4 microplate reader with a Chroma 585 nm single-edge dichroic beamsplitter (T585Ipxr), 575/15 nm excitation filter, and 620/15 nm emission filter. For assays performed at 18 and 24 °C, the microplate reader was moved into a 4 °C cold room for constant cooling, while the instrument was set to the desired temperature for heating.

For competition dose response binding, a mixture of 1 μM bradykinin-SR101 and 15 μM OppA was prepared, and 150 μL of the mixture was added to each well in a 96-well nonbinding microplate. The unlabeled peptide (50 μL) was added to each well at final concentrations varying from 0.1 to 2500 μM (1 to 10 000 μM for leu-enkephalin) and incubated at the desired temperature for at least 1 h to reach equilibrium. Fluorescence anisotropy was measured as described above.

Anisotropy (r) was calculated using the intensity of parallel (I∥) and perpendicular (I⊥) light emitted as follows: . Anisotropy was plotted against the log concentration of protein for the direct dose response binding and log concentration of the unlabeled peptide for competition dose response binding. A dose response curve fit was performed in OriginPro to determine the dissociation constant (KD) for direct dose response data or the half maximal inhibitory concentration (IC50) for competition data. The KD values for competition data were determined as described in the Supporting Information.

CD Spectroscopy

Pure OppA was diluted to 2 μM in 25 mM sodium phosphate buffer (no NaCl or glycerol due to CD absorbance) at the desired pH with or without 100 μM bradykinin. Diluted protein (600 μL) was added to a quartz cuvette with a 2 mm path length (Starna Cells, 18F-Q-10). CD absorbance was measured from 190 to 260 nm on a Jasco J-1500 CD spectrophotometer using a 50 nm/min scan speed, 1.0 nm data pitch, 1.0 nm bandwidth, and 1 s digital integration time.

Mean residue ellipticity ([θ]mrw) was calculated as described in Kelly et al’s work.[64] First, the mean residue weight (MRW) of OppA was calculated using the following equation: , where M is the molar mass of the protein in Da, N is the number of amino acids in the protein, and the number of peptide bonds is N – 1. Here, our L. lactis OppA molar mass is 66 419.5 Da and is composed of 601 amino acids (Figure S2), which gives an MRW of 110.7 Da. Then, [θ]mrw was calculated using the following equation: where θλ is the observed ellipticity in degrees, l is the pathlength in cm, and c is the concentration in g/mL.

To measure thermal denaturation, the temperature was ramped from 4 to 74 °C (to 79 °C for OppA + bradykinin at pH 6), increasing 1°/min measuring the full spectrum every 5 °C with a 2 min wait time before measurement for equilibration. The melting temperature (Tm) was determined by plotting the mean residue ellipticity at 222 nm against the temperature and fitting to a Boltzmann distribution in OriginPro.