Determination of the Membrane Translocation pK of the pH-Low Insertion Peptide

Abstract

The pH-low insertion peptide (pHLIP) is a leading peptide technology to target the extracellular acidosis that characterizes solid tumors. The pHLIP binds to lipid membranes, and responds to acidification by undergoing a coupled folding/membrane insertion process. In the final transmembrane state, the C terminus of pHLIP gets exposed to the cytoplasm of the target cell, providing a means to translocate membrane-impermeable drug cargoes across the plasma membrane of cancer cells. There exists a need to develop improved pHLIP variants to target tumors with greater efficiency. Characterization of such variants typically relies on determining the pK parameter, the pH midpoint of peptide insertion into the lipid bilayer. Here we report that the value of the pK can be strongly dependent on the method used for its determination. Membrane insertion of pHLIP involves at least four intermediate states, which are believed to be linked to the staggered titration of key acidic residues. We propose that some spectroscopic methods are influenced more heavily by specific membrane folding intermediates, and as a result yield different pK values. To address this potential problem, we have devised an assay to independently monitor the environment of the two termini of pHLIP. This approach provides insights into the conformation pHLIP adopts immediately before the establishment of the transmembrane configuration. Additionally, our data indicate that the membrane translocation of the C terminus of pHLIP, the folding step more directly relevant to drug delivery, occurs at more acidic pH values than previously considered. Consequently, such a pK difference could have substantial ramifications for assessing the translocation of drug cargoes conjugated to pHLIP.

Introduction

The pH-low insertion peptide (pHLIP) holds great promise for diagnostic and therapeutic applications in diseases characterized by acidosis. In particular, pHLIP has been shown to target different solid tumor types, and to translocate membrane-impermeable drugs across the plasma membrane into the cytoplasm of cancer cells (1, 2, 3, 4). Based on environmental factors, pHLIP can assume three different stable states: at neutral pH, pHLIP is unstructured in solution (State I), but in the presence of lipid bilayers it binds to the membrane surface (State II). Finally, a drop in pH causes pHLIP to insert across the membrane and fold into a transmembrane helix (State III) (5). The ability of pHLIP to target cancerous cells and deliver drug cargoes across the membrane is due to the extracellular acidity common to solid tumors (4, 6). Cancerous cells have a local extracellular pH roughly 0.5–1.0 pH units lower than healthy cells (pH 6.4–6.8 vs. pH 7.2–7.4, respectively (6, 7)). At physiological pH conditions pHLIP is believed to remain primarily surface-bound on the outer leaflet of the plasma membrane. Membrane insertion is precluded by the seven negative charges the peptide carries at this pH, consisting mostly of aspartic acid residues (8). However, the extracellular acidity of cancerous cells leads to the protonation of the acidic groups distributed along the sequence of pHLIP. As a result there is a large increase in peptide hydrophobicity that leads to pHLIP finding its energy minimum by inserting into the core of the lipid bilayer as a transmembrane α-helix (9). Membrane insertion is unidirectional, as only the C terminus of the peptide crosses the bilayer and finds its way into the cytoplasm (5).

The key metric used for studying the membrane insertion of pHLIP is the apparent pK (also referred to as pH₅₀ or pH dependence), which has been previously defined as the pH midpoint of peptide insertion into the lipid bilayer (8, 10, 11). The pHLIP contains two tryptophan residues (W9 and W15), which have been commonly used as reporters for the environmental changes occurring during membrane insertion. Such changes in tryptophan fluorescence have been monitored primarily following fluorescence spectral maximum (SM, also referred to as λ_max) changes as a function of pH (5, 8, 12). The pK of insertion of pHLIP in 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC) large unilamellar vesicles obtained by different laboratories using SM is 6.0–6.2 (5, 8, 10). Circular dichroism (CD) and tryptophan fluorescence intensity (FI) have also been used sparsely to obtain the pK of insertion (9, 13). However, a systematic comparison of the apparent pK values measured using different methods in identical conditions is lacking. As a result, it is not known if the different methods yield the same pK value.

Stopped-flow kinetic measurements have shown that the membrane insertion of pHLIP is not a two-state event. On the contrary, it is a complex process involving at least four intermediate conformations that precede the final transmembrane state (14, 15). The multistep membrane insertion of pHLIP is expected to be controlled by the titration of seven acidic groups (4 D and 2 E side chains plus the C_t group). Importantly, a recent solid-state nuclear magnetic resonance (ssNMR) study on pHLIP shows that the four aspartic acid residues (D14, D25, D31, and D33) have clearly differentiated pK_a values as pHLIP inserts into the bilayer (10). Given the complexity of the membrane insertion, we wondered if using a single pK value could adequately describe the process. Furthermore, it has not been proven that this pK faithfully informs exclusively on the final membrane translocation step, when drug cargoes are transferred across the lipid membrane into the cytoplasm.

Recently, we have begun to use different analysis methods to further understand the complex insertion process of pHLIP (13). Here, we use six different approaches to measure the apparent pK of insertion of pHLIP. These include the development of an assay to independently study the environment of the N- and C- terminus via an extrinsic probe. Intriguingly, the different experimental methods did often yield different pK values. We use the recent ssNMR data to interpret the pK results, and propose that the analysis methods can report on different step/s of the membrane insertion pathway of pHLIP.

Materials and Methods

Intrinsic tryptophan fluorescence spectroscopy

The pHLIP (sequence: N_t-AAEQNPIYWARYADWLFTTPLLLLDLALLVDADEGTCG-C_t) was synthesized using standard solid phase protocols (P3 Biosystems, Louisville, KY) and purified by reverse-phase high performance liquid chromatography to >95% purity. A lyophilized peptide stock was dissolved in 10 mM sodium phosphate (NaP_i) pH 8.0 buffer. POPC vesicles were prepared via extrusion through a 100-nm pore size membrane using a Mini Extruder (Avanti Polar Lipids, Alabaster, AL) in the same buffer to form large unilamellar vesicles. POPC and pHLIP were then incubated for 1 h at room temperature, for a final lipid/peptide of 200:1. Final peptide concentration was 1 μM. Previous studies have shown that the C-terminal cysteine present in pHLIP does not cause disulfide-mediated dimerization under our experimental conditions (8). To perform a pH insertion titration, the pH of the different samples was adjusted by mixing with 100 mM stocks of sodium acetate, 2-N-morpholino ethanesulfonic acid, or 4-2-hydroxyethyl-1-piperazineethanesulfonic acid buffers (25 μL), to obtain the desired pH values. Final sample volume was 140 μL. To perform a pH exit titration, preincubated pHLIP and POPC was acidified to pH 4 using 100 mM sodium acetate. This acidic sample was then mixed with the same buffer series as the insertion titration to obtain the exit titration. The final pH of each sample was measured using a 2.5 mm-bulb pH-electrode (Microelectrodes, Bedford, NH). Emission spectra were recorded using a Photon Technology International (Edison, NJ) Quanta Master fluorometer at room temperature with the excitation wavelength set to 280 nm, the emission wavelength range from 310 to 400 nm, and excitation and emission slits set to 3 nm. A previous report has shown minimal Tyr contribution in the observed fluorescence signal with pHLIP for an excitation of 280 nm (15). Lipid blanks were subtracted in all cases. Data were analyzed by calculating the spectral center of mass (CM), with the following equation:

$$CM=\sum _1^n{I}_i{\lambda }_i/\sum _1^n{I}_i,$$ (1)

where I_i is the fluorescence intensity measured at a wavelength λ_i. CM is used to monitor the environment of the two W residues, and information is extracted from the entire spectral range of the data (16, 17). The data were also analyzed by following the fluorescent emission intensity change at 335 nm, which is directly proportional to the population of molecular species present (18). For SM determination, curves were fit using a Lorentzian distribution to determine the wavelength of maximum intensity in each curve over the pH titration. CM, SM, and FI pH-titrations were then fitted to determine the pK using

$$\text{Signal}=\left[F_a+F_b{10}^{m\left(pH-pK\right)}\right]/\left[1+{10}^{m\left(pH-pK\right)}\right],$$ (2)

where F_a is the acidic baseline, F_b is the basic baseline, m is the slope of the transition, and pK is the midpoint of the curve, and signal is fluorescence or circular dichroism changes. Here the term pK is not being used in its strict sense, but it merely describes the apparent midpoint of the titration, as discussed elsewhere (13).

Circular dichroism

Measurements were performed on a Jasco (Easton, MD) J-815 spectropolarimeter at 25°C. pHLIP was incubated with POPC vesicles (prepared as described earlier) in 10 mM NaP_i pH 8.0 buffer for 1 h. Afterwards, the pH was adjusted with 100 mM sodium acetate, 2-N-morpholino ethanesulfonic acid, or NaP_i (62.5 μL) to a range of desired final pH values. Final sample volume was 250 μL. The lipid/peptide was 200:1 with a final peptide concentration of 7 μM. Changes in helical content were reported by following the ellipticity at 222 nm. Spectra were collected from 222 to 262 nm in 10-nm steps, and values at 222 nm were subtracted from values at 262 nm to correct for any changes in the baseline of each individual spectrum. By limiting the collected spectral range, the complete pH titration could be conducted in 5 h while acquiring data with high accuracy (each sample was averaged 20 times using a scan rate of 100 nm/min). Calculated molar ellipticity ([θ] = θ/(10 lcN) at 222 nm was plotted against measured pH, and the resulting sigmoidal transition was fitted using Eq. 2 to obtain the pK_CD.

NBD fluorescence assay

We conjugated the environmentally sensitive dye NBD to the C-terminal cysteine in pHLIP using n, n′-dimethyl-n-iodoacetyl-n′-7-nitrobenz-2-oxa-1,3-diazol-4-yl ethylenediamine (IANBD; Thermo Fisher Scientific, Waltham, MA). For the conjugation of NBD to the N-terminus of pHLIP, we used instead succinimidyl 6-n-7-nitrobenz-2-oxa-1,3-diazol-4-yl amino hexanoate (NBD-X SE; Anaspec, Fremont, CA). To this end, a pHLIP variant lacking a cysteine residue (N_t-AAEQNPIYWARYADWLFTTPLLLLDLALLVDADEGT-C_t) was employed. After purification by size exclusion chromatography using a PD-10 column (GE Healthcare Bio-Sciences, Marlborough, MA) in 1 mM NaP_i buffer, pH 7.5, the conjugation was verified by MALDI-TOF and high-performance liquid chromatography, and samples were aliquoted and lyophilized. For the experiment, samples were rehydrated in 10 mM NaP_i pH 8.0 at a final concentration of 0.8 μM (the NBD extinction coefficient was 18,482 M⁻¹ cm⁻¹) and incubated with POPC vesicles (prepared as described earlier) for 1 h. All experimental conditions, including lipid/peptide and final salt concentration (19.3 mM), were identical to W fluorescence and CD experiments. The pH of each row of a 96-well plate was decreased using a series of 100 mM buffers (sodium acetate and NaP_i (14.3 μL)). Final sample volume was 100 μL. Fluorescence spectra were recorded at 25°C with excitation at 470 nm and an emission range of 520–600 nm using a Cytation 5 imaging plate reader (Biotek Instruments, Winooski, VT). The final pH of each well was measured. Fluorescence emission intensity pH-titrations at 540 nm were then fitted to determine the pK_Ct-NBD and pK_Nt-NBD using Eq. 2.

Stopped-flow kinetic measurements

Lyophilized pHLIP and its two NBD conjugates were rehydrated in 10 mM NaP_i pH 8.0 and incubated with POPC vesicles (prepared as described above) for 1 h at a lipid/peptide of 200:1. Peptide concentration in the incubation was 5 μM in the case of pHLIP and 0.2 μM in the case of the NBD conjugates. Final salt concentration was 55 mM for both pHLIP and both NBD conjugates. Spectral measurements were collected on a SX-20 Stopped-Flow instrument with temperature control (Applied Photophysics, Surrey, UK). Preincubated peptide lipid samples (2 mL) at pH 8 were mixed with 100 mM sodium acetate pH 3.9 to obtain State III. In the case of NBD conjugates, preincubated samples were also run at pH 8 by mixing with 100 mM NaP_i pH 8. For pHLIP, peptide-only blanks were collected by mixing pHLIP with 100 mM NaP_i pH 8, and the final pH was measured afterwards. The mixing volume was 120 μL at 1:1 with a dead time of ∼1 ms. NBD conjugates were excited at 488 nm and fluorescence was collected through a 505-nm cutoff filter (excitation at 280 nm with a cutoff filter of 320 nm in the case of pHLIP fluorescence). After mixing, the concentration of the NBD conjugates was 0.1 μM, and the final pHLIP concentration was 2.5 μM. Data were collected over a 50-s interval with 4000 data points, repeated 30 times and averaged. The first 100 ms were not included in the fitting, because a small signal increase was observed in the controls in this time range. Slits were set to 2 nm. Gain was automatically set to the acidic sample containing inserted peptide. The first five injections were always discarded to ensure the previous sample was flushed out of the sample cell. Temperature was held at 25°C. POPC blanks were also collected by mixing POPC vesicles with 100 mM NaP_i pH 8. For NBD conjugates, data for State II were subtracted from data for State III to correct for any potential photobleaching and light scatter from the vesicles. For W intensity experiments, pHLIP in solution was subtracted from State III data. Subtracted data were analyzed as in Karabadzhak et al. (14), using the following equation:

$$\text{Signal}=A1\times \text{exp}\left(-k1\times x\right)+A2\times \text{exp}\left(-k2\times x\right)+A3\times \text{exp}\left(-k3\times x\right)+A4\times \text{exp}\left(-k4\times x\right)+F_o,$$ (3)

where A is the amplitude of the signal (fluorescence intensity), k is the rate constant in s⁻¹, x is time, and F_o is the vertical offset. As described elsewhere (15), a four-exponential model was required to adequately fit the data, due to the inability to find good agreement between the theoretical and experimental curves with fewer than four exponentials. Amplitudes were normalized to a total intensity range by summing the values only for the positive amplitude change in fluorescence.

Statistical analysis

Statistical analysis was performed on the pK values obtained to determine if a statistical difference existed. Analysis was performed using the software SPSSv24 (IBM Analytics, Armonk, NY). Two different multiple comparisons tests were used based on the homoscedasticity (two-sided Dunnett's t-test) or heteroscedasticity (Dunnett T3) of the data. Comparison of all membrane entry pK values to SM were analyzed using a Dunnett T3 test. Comparison of C_t-NBD to FI and N_t-NBD pK was performed using a two-sided Dunnett's t-test. All membrane exit pK values were compared to SM with a two-sided Dunnett’s t-test. Both the two-sided Dunnett and the Dunnett T3 t-tests set one variable as the control to compare to all other treatment groups; p ≤ 0.05 was considered significant for all tests.

Results

CD pK as a tool to monitor helical formation as a function of the membrane insertion of pHLIP

Stopped-flow CD kinetic experiments of pHLIP show that upon acidification, helix formation starts at the membrane surface before membrane insertion, as reported by intrinsic tryptophan fluorescence intensity changes (15). The initial helix formation is followed by a multistep insertion process resulting in the establishment of the final transmembrane state (14, 15). Fig. 1 A shows the CD spectrum of pHLIP in the presence of POPC vesicles at pH 8 (State II, solid trace), which corresponds to the largely unstructured conformation (with a minimum at ∼200 nm) found before insertion starts. The conformational change leading to the transmembrane state (State III, shaded trace) can be quantified by following the ellipticity change at 222 nm, the larger of the two minima characteristic of an α-helix (19). We performed a titration between pH 4 and 8 and the change in ellipticity at 222 nm was monitored to determine the pK of the helical changes of pHLIP (Fig. 1 B), which is referred to thereafter as pK_CD, with a value of 6.03 ± 0.05. The use of CD to determine the insertion pK allows us to characterize the ensemble of steps of helix formation occurring at the central part of the sequence. However, based on the previous stopped-flow data, we reasoned that analyzing the pH-induced changes in the local environments of the two W residues could report on different steps of the insertion process.

Figure 1.

CD monitors helical formation during the membrane insertion of pHLIP. (A) Given here are CD spectra showing the membrane surface-bound State II (solid trace) and the inserted transmembrane State III (shaded trace) of pHLIP in POPC vesicles. (B) The difference in molar ellipticity ([θ]) between 222 and 262 nm was plotted against the pH to determine pK_CD, the midpoint of helical formation. (Inset) Given here is titration CD spectra, showing that increasing acidity leads to α-helix formation, and thus more negative molar ellipticity at 222 nm.

Intrinsic tryptophan fluorescence reports changes in the local environment of the W residues during membrane insertion

Tryptophan fluorescence is the method most commonly used to monitor the insertion of pHLIP into the hydrophobic core of membranes. The location of W residues is easily discernable between hydrophobic and hydrophilic environments due to the high sensitivity of the W fluorescence emission dipole to polarity (20). For this reason, environmental information can be extracted from the emission wavelength shift of the combined spectrum of residues W9 and W15 as the membrane insertion of pHLIP occurs. Specifically, analysis of the emission spectrum shift used to obtain an apparent pK of insertion can be performed using a dual approach: calculating the fluorescence CM and the SM, with SM being used to a greater extent (5, 8, 12, 13). The spectral CM provides wavelength-averaged information from the entire spectrum and is sensitive to changes in the shape and width of the emission peak (16, 17). On the other hand, fluorescence SM only reports changes in the spectral maximum, and is determined by fitting the spectral curve with a Lorentzian distribution, to obtain the peak emission wavelength. We also employed to obtain the apparent pK a third analysis method that has seen limited use, which consists of monitoring changes in the W FI (9, 13). A single wavelength, 335 nm, was chosen because it shows a large fluorescence intensity change from State II to State III.

We performed titrations of pHLIP in the presence of POPC vesicles and analyzed the fluorescence data using the three methods described above (Fig. 2) and determined three pK values (Fig. 3; Table S1). Fig. 2 shows the W fluorescence spectra and the sigmoidal titrations obtained with CM, SM, and FI analysis, which were used to determine pK_CM, pK_SM and pK_FI, respectively. Interestingly, whereas CM and SM display a similar pK of insertion (6.07 ± 0.08 and 6.22 ± 0.17, respectively), the pK obtained monitoring the FI was ∼0.5 pH units lower than both CM and SM. Correspondingly, the pH range where the titration occurred also differed when comparing FI to both CM and SM (arrows mark the approximate titration pH range in Fig. 2 B–D), in agreement with the sigmoidal FI curve being shifted to a lower pH range. Determination of the pK_FI was not dependent on the wavelength employed for the analysis (data not shown). Interestingly, the values of pK_CM and pK_SM overlap with pK_CD, with a value of ∼6.1, suggesting that these three parameters report on similar events. The different pK values obtained are compared in Fig. 3.

Figure 2.

Three different analyses of W fluorescence pH titrations. (A) Sample acidification (shaded arrow) results in an increase in W fluorescence intensity and spectral blue shift. A similar pK was obtained by following changes in CM (B) and SM (C) upon acidification. (D) W intensity changes were also analyzed by following the FI at 335 nm. Lines in (B–D) show the fits of Eq. 2 to the data. Solid arrows mark the approximate pH values for the start and end of the titrations.

Figure 3.

The pHLIP does not have a unique pK value. Bars correspond to the pK determined by W fluorescence SM, spectral CM, or FI, by CD, or by N_t-NBD or C_t-NBD of pHLIP. Mean values are shown, for experiments repeated three to six times, and error bars are the SD. Color shading represents groups with no statistical difference. ^∗p ≤ 0.05; ^∗∗p ≤ 0.01; ^∗∗∗p ≤ 0.005; NS; no significance.

The membrane insertion of pHLIP is a reversible process (15). We determined next whether the differences observed between the membrane entry pK values obtained with FI and the other methods were also found in the membrane exit. Accordingly, fluorescence experiments were performed to characterize the membrane exit. This process consists of the transition from the transmembrane State III to State II, obtained when pH is increased starting from the acidic end-point of the membrane insertion transition (Fig. 4 A) (13, 15). The membrane exit titrations of pHLIP in the presence of POPC vesicles were analyzed via CM, SM, and FI changes (Fig. S1). We observed that the pK_SM obtained for the membrane insertion and exit were close (6.22 ± 0.17 and 6.29 ± 0.08, respectively). Similar results were also obtained for pK_CM (6.07 ± 0.08 and 6.21 ± 0.06, respectively). However, the entry and exit transitions did not have the same midpoint when quantified using pK_FI. In fact, a clear hysteresis was observed in the FI data, because the membrane entry value was 5.68 ± 0.09, whereas the membrane exit value was 6.15 ± 0.18 (Fig. S1). Fig. 4 B illustrates that, in contrast to the membrane entry, the three analysis methods yielded similar exit pK values, with differences that were not statistically significant (p > 0.05), in contrast to the lower pK_FI observed for membrane insertion (Fig. 3). The implications of this difference will be examined in the Discussion.

Figure 4.

Membrane exit of pHLIP yields similar pK_CM, pK_SM and pK_FI values. (A) The increase in pH (shaded arrow) results in a decrease in the tryptophan fluorescence intensity and a red shift in spectral maximum in the emission spectra. (B) The three membrane exit pK values determined by W fluorescence (CM, SM, and FI) are not statistically significant (p > 0.05). NS; no significance. Error bars are the SD.

There are at least two intrinsic caveats of employing W fluorescence to study pHLIP. One is the location of both W residues near the N-terminus. Because W fluorescence is sensitive to the local environment (20), this assay might have low sensitivity to the membrane translocation of the C terminus of the peptide. The second limitation is that the W fluorescence reports on two local environments, those of W9 and W15, which are different (for example, by having different titratable groups in their vicinities). To overcome these limitations, we decided to employ an extrinsic probe conjugated to either end of pHLIP.

Use of site-specific NBD labeling to monitor each terminus

We next conjugated the NBD dye to either the N-terminal amino group of pHLIP, or to a cysteine residue located close to the C terminus, the end that translocates across the bilayer. Our aim was to obtain pK values that report on the local environment of each terminus of pHLIP. NBD is a pH-insensitive probe that responds to changes in hydration altering its fluorescence emission intensity (21, 22, 23, 24, 25). We performed pH titrations in the presence of POPC vesicles with the NBD conjugated to a C-terminal cysteine residue of pHLIP (pHLIP-NBD) or to the N-terminus (NBD-pHLIP) (Fig. 5). We observed that the insertion pK of NBD-pHLIP (pK_Nt-NBD) was 5.64 ± 0.06. This value was similar to pK_FI, which reports on changes of the W residues located at the N-terminal half of pHLIP. Strikingly, the pK of pHLIP-NBD (pK_Ct-NBD) was 5.24 ± 0.17, significantly more acidic than any other pHLIP pK values reported here (Figs. 3 and 5 C) or elsewhere. As a control, fluorescence spectra of the two NBD conjugates of pHLIP were collected in buffer (Fig. S2). No significant shift in fluorescence SM was observed between the conjugates. This suggests that the different neighboring residues of NBD present at either end of pHLIP do not significantly influence the NBD fluorescence. As a result, we propose that the NBD signal is not influenced by local effects but is primarily reporting on global conformational changes of pHLIP. As a result, our data suggest that the translocation of the C terminus occurs at more acidic pH values than initially considered.

Figure 5.

NBD conjugates of pHLIP display different pK values based on location. (A and B) Representative emission spectra of pHLIP-NBD (A) and NBD-pHLIP (B) show an increase in fluorescence intensity as acidity is increased (shaded arrows). (C and D) Normalized fluorescence intensity at 540 nm is plotted versus pH to determine pK_Ct-NBD (C) and pK_Nt-NBD (D). Lines in (C) and (D) show the fits of Eq. 2 to the data. Solid arrows mark the approximate pH values for the start and end of the titrations.

Next, we performed a stopped-flow kinetic experiment in the presence of POPC vesicles that monitored W, N_t-NBD, and C_t-NBD fluorescence intensity changes. The objective of this experiment was twofold. It served as a control to ensure that the conjugation of the NBD dye to either end of pHLIP did not alter the membrane insertion mechanism. However, more importantly, it allowed gaining new mechanistic understanding of the membrane insertion process. Rapid mixing of acidic buffer with preincubated peptide-lipid samples led to the transition from State II to State III (Fig. 6). As reported previously, W fluorescence initially increased and then saturated (14, 15). Interestingly, although both NBD conjugates also showed an initial signal increase, it was followed by a significant signal decrease (Fig. 6). We do not suspect the decrease to result from photobleaching, as control experiments showed that the signal was essentially flat at pH 8 (Fig. S3). Kinetic traces for W, N_t-NBD, and C_t-NBD changes were fitted with four exponential terms (fitting residuals can be found in Fig. S4). Table 1 shows that our control experiment monitoring W changes yielded results similar to previous reports (14, 15). Importantly, we obtained overall similar rate constant values for both NBD conjugates of pHLIP, suggesting that the presence of the dye does not have a significant effect on the insertion kinetics.

Figure 6.

Intrinsic W, N_t-NBD, and C_t-NBD fluorescence display similar kinetics during pHLIP membrane insertion. Given here are average spectra of W (dark shaded), N_t-NBD (dashed light shaded), and C_t-NBD (solid) in POPC.

Table 1.

Rate Constants and Normalized Amplitudes of Membrane Insertion Kinetics

	k1−1, s A1, %	k2−1, s A2, %	k3−1, s A3, %	k4−1, s A4, %
Nt-NBD	0.13 ± 0.04	0.50 ± 0.21	1.55 ± 0.74	34.47 ± 16.8
	12.41 ± 6.29	51.97 ± 8.30	35.62 ± 14.24	−22.74 ± 3.80
Ct-NBD	0.27 ± 0.12	0.86 ± 0.24	2.27 ± 1.17	55.80 ± 20.78
	11.26 ± 11.47	58.10 ± 10.28	30.64 ± 17.98	−43.91 ± 5.83
Trp	0.20 ± 0.06	1.03 ± 1.08	3.47 ± 2.43	24.04 ± 18.72
	31.32 ± 6.24	42.95 ± 14.70	11.90 ± 4.90	13.84 ± 8.67

Discussion

The pK of insertion is often the only parameter used to describe the pH-responsiveness of peptides developed for targeting acidosis (26, 27, 28, 29, 30). Thus, progress in the field can be hindered if the molecular event(s) the pK reports on is (are) not solidly defined. Here we perform a systematic analysis that directly compares the pH dependence of pHLIP obtained employing several methods in identical conditions, including, to our knowledge, a new approach that employs an extrinsic fluorescent probe. Our data suggest that there is not a single pK value for pHLIP. On the contrary, significantly different values were obtained when employing an array of common measurements. Furthermore, we report that the use of extrinsic probes can be used to determine the pH range where specific regions of pHLIP undergo an environmental change during the membrane insertion process.

Fig. 3 shows the different pK values obtained employing the six different spectroscopic approaches. The pK values can be classified in three groups, shown with different color shading. The highest value corresponds to pK_CM, pK_SM, and pK_CD. These three approaches yield values whose differences are not statistically significant (p > 0.05), with an average pK of ∼6.1. Next, pK_FI and pK_Nt-NBD group for an average pK value of ∼5.6. Interestingly, these two parameters employ probes (W and NBD) that are located near or at the N-terminus of pHLIP. Finally, pK_Ct-NBD reports on the most acidic pK value, ∼5.2. Strikingly, the pK_Ct-NBD value is ∼0.9 units lower than pK_SM, the typical value used in the literature (5). This is a large pH difference, which roughly corresponds to the entire pH difference existing between the extracellular environment of healthy (pH 7.2–7.4) and cancer cells (pH 6.4–6.8) (6, 7).

NBD has been used to study membrane protein insertion as a function of pH (24), but the use of NBD as a tool to determine the pK of pHLIP insertion is, to our knowledge, a novel approach. Previously, NBD-pHLIP conjugates were used to determine the directionality of the membrane insertion of pHLIP and to test the ability of pHLIP to translocate cargoes (5, 31). The translocation of different cargo molecules by pHLIP WT and variants has been previously studied. These reports indicate that cargo molecules do not affect the pH-dependent insertion, as determined by monitoring W fluorescence changes (14, 31, 32). We attempted to collect W fluorescence data with our pHLIP-NBD conjugate but the signal was reduced beyond reliable detection in the presence of NBD. Significant W signal loss in the presence of NBD has been observed previously, and was suggested to result from FRET from W to NBD occurring in specific conditions (33, 34). Previous work has shown that pHLIP translocation is not affected by the conjugation of NBD at either terminus (31).

The method commonly used to determine the membrane insertion pK of pHLIP has been the SM analysis of W fluorescence (5, 8, 12, 13). Here, we find that the pK obtained by analyzing SM and CM is significantly higher than the one obtained monitoring FI changes. The pK is shifted ∼0.5 units and the differences are statistically significant (p ≤ 0.005) (Fig. 3). However, caution must be exercised when rationalizing the differences. The shift in pK could result from the methods reporting on different steps of the membrane insertion process. However, an alternative possibility is that the shift is a consequence of the intrinsic differences of each analysis method. Ladokhin et al. (18) and Moon and Fleming (35) have discussed the problems associated with relying on tryptophan spectral blue-shift analysis, because spectral center of mass and spectral maximum changes do not linearly depend on the amount of folded protein or peptide (18, 35). This deviation from linearity can cause one to erroneously follow membrane insertion and not accurately report on the acquisition of final folded state (18). Conversely, W fluorescence intensity at a fixed wavelength shows a linear relationship with the amount of folded species. The linear nature of FI changes thus allows a more accurate determination of the amount of peptide in the final folded state (18).

We decided to perform additional experiments to investigate if the pK differences observed between FI and spectral shift methods (SM and CM) could be due to the latter improperly following the folding of pHLIP as membrane insertion occurs. Because the membrane insertion of pHLIP is a reversible process (13, 15), we sought to investigate if the fluorescence pK difference observed for the membrane insertion was also observed for the membrane exit. We rationalized that if the difference between pK_CM and pK_SM and pK_FI was due to intrinsic properties of the analysis methods, we might expect to see a similar pK difference in the membrane entry and exit. Conversely, Fig. 4 B shows that there is not a significant difference in pK exit values of SM, CM, and FI (p > 0.05). However, the membrane exit of pHLIP follows a different pathway with less kinetic intermediates than the membrane insertion (15), which complicates comparing the two processes. As a result, we cannot rule out that the values of pK_CM and pK_SM might have an artifactual contribution. Thus, we suggest that FI is the most accurate intrinsic fluorescence parameter to determine the pK value. Conversely, CD also shows a linear response during peptide folding (36), and thus pK_CD would not be affected by the potential problems of the other members of the high pK group (pK_CM and pK_SM). Because pK_CD (helical formation) was significantly higher than pK_FI (W burial), we suggest that the two parameters are reporting on different aspects of the multistep membrane insertion/folding of pHLIP. Others have reported deviations between linear spectroscopic methods when determining folding midpoints (37, 38, 39). This difference often indicates that the folding/unfolding is a multistep process, as is the case for pHLIP (37, 39). Conversely, perfect overlap of the midpoint region generally suggests a two-step process (37).

pK_CD as the midpoint of helical formation

We wanted to investigate the relationship between the different pK values (Fig. 3) and the separate steps of the folding/membrane insertion of pHLIP. Interestingly, the pK_a values of D14, D25, D31, and D33 of pHLIP have been recently determined by ssNMR in POPC vesicles (10). Comparison among each of the three pK groups and the individual ssNMR pK_a titrations is highly informative (Fig. 7). As a representative of the group with a higher pK we selected SM, because it is the parameter most commonly employed to obtain the pK of pHLIP. We observed overlap between the SM sigmoidal curve and the ssNMR curves for the titrations of D25, D31, and D33. As a result, there is a good agreement between pK_SM (6.22 ± 0.17) and the value obtained by averaging these three independent pK_a values (6.30 ± 0.21) (10). On the other hand, the overlap with D14 was not clear (Fig. 7 A). Interestingly, the curve of pK_SM nicely overlaps with the pK_a of D33 (6.34 ± 0.21). The agreement led us to speculate that CD (and potentially also SM and CM) might report on the ensemble of the membrane insertion intermediates resulting from the titration of residues D33, D31, and D25. Our data thus suggests that there is correlation among the protonation of D33, D31, and D25 and transmembrane helical formation, as reported by pK_CD (10, 19). Fig. 8 displays the proposed steps required for the insertion of pHLIP. The cartoon also shows at what point each aspartate group titrates into the protonated state, and highlights the insertion step/s that each pK value reports on.

Figure 7.

Qualitative comparison of fluorescence pK titrations with ssNMR pK_a titrations for the individual aspartate residues. The individual titrations for D14 (magenta), D25 (blue), D31 (green), and D34 (red) obtained by Hanz et al. (10) are overlaid with the fluorescence data obtained following tryptophan SM (A) and FI (B), and pHLIP-NBD (C). The ssNMR data were represented by the fitting to the Boltzmann equation (10). Normalized chemical shifts are shown for a better comparison, and values were inverted in (B) and (C) to align with the fluorescence data changes.

Figure 8.

Tentative model showing the transition from State II to State III during the membrane insertion of pHLIP, highlighting the insertion step/s that each pK value reports on. The solid arrows represent the transitions between each step in the membrane insertion process, with the corresponding pH midpoints. The change of the D residues from shaded to solid represents the protonation. CD, and potentially SM and CM, reports on the insertion process encompassing D33, D31, and D25, and are then shown labeling the first three transitions. FI and N_t-NBD follow the protonation of D14, and to a lesser extent D25 and D33 (not shown in figure). Finally, C_t-NBD reports on the last insertion step, where the C-terminal end translocates across the bilayer.

pK_FI as the midpoint of N_t conformational changes

We observed a less complete overlap between the ssNMR curves and the FI titration, as a result of the acidic shift in the pK value for FI (and NBD-pHLIP) (Fig. 7 B). In this case, there is no clear overlap with the titration of D31, and only partial overlap with the D33 and D25 titrations. However, the pK_FI of 5.68 ± 0.09 is within the error to the pK_a of D14, 5.82 ± 0.08 (10). We hypothesize that pK_FI reports primarily on the partial local burial of the N-terminus of pHLIP, which is expected to be controlled by the protonation of D14 (Fig. 8) (10). This is supported by the similar values of pK_FI, which reports on the environments of W9 and W15, and pK_Nt-NBD, which is informed by the N-terminus (Fig. 3). Interestingly, we observed hysteresis when FI was used to compare the two directions of the process: membrane entry (insertion) and exit. Hysteresis was not observed in the case of SM or CM (Fig. S2). The existence of hysteresis might not be surprising, because the intermediate steps populated in both directions of the folding pathway are different. During the membrane exit of pHLIP, the N-terminal region might change environment concomitantly with helix unraveling, and explain why the three pK values in Fig. 4 are similar. This is supported by stopped-flow CD data indicating that the pHLIP helix starts to unfold within the membrane before membrane exit occurs (15).

pK_Ct-NBD as the translocation midpoint

The rate-limiting step in the formation of the transmembrane state of pHLIP is the translocation of the C terminus (10, 15). For therapeutic applications, drug cargoes are conjugated to the C terminus. Its unidirectional translocation enables pHLIP to be an effective drug delivery agent to acidic tissues such as solid tumors (2, 5). It has been recently suggested that tryptophan fluorescence changes are not capable of detecting the final translocation of the C terminus of pHLIP across the bilayer (10). Here we address this potential shortcoming and study the membrane insertion of pHLIP using a C-terminal NBD probe. Interestingly, the pK obtained using this probe, pK_Ct-NBD, is more acidic than any of the individual aspartic acid pK_a values determined via ssNMR (Fig. 7 C) (10). Due to the location of NBD at position 37, it is interesting to compare the pHLIP-NBD curve with the titration of the closest residue studied by ssNMR, D33. Although both are essentially C-terminally located, pK_Ct-NBD is ∼1.0 pH unit lower than the pK_a of D33 (10). Additionally, D33 displays a much broader titration compared to aspartates 14, 25, and 31, encompassing a larger pH range in the titration event, similarly to pK_Ct-NBD. We speculate that the differences in slope might result from long-range effects of the yet-unresolved titrations of E34 and/or the C_t group affecting the cooperativity of the D33 titration (40). It would not be surprising if E34 and the C_t had the most acidic pK_a values, because they are located in the most polar region of the sequence. As a result, water accessibility would be less hindered, and their pK_a values might be closer to those typically found in solution (∼3.7 and ∼4.3 for C_t and E, respectively (41)), but it is unknown what the effect of titration of E34 and the C_t is on each pK value. Additional ssNMR determination of the pK_a of E34 and the C_t group is needed to shed light on this hypothesis. We suggest that the C-terminal NBD group might be sensitive to the titration of such neighboring acidic groups, to report on the translocation of the C-terminal end of pHLIP that results in the acquisition of the final transmembrane state (Fig. 8). We suggest that a new pK term might be needed to adequately describe this step, and we refer to it here as the pK of translocation, which is significantly more acidic than the well-established pK of insertion.

Fig. 8 also shows the proposed location of the N- and C terminus at each step, based on the stopped-flow data (Fig. 6). As expected for a kinetic study of the full transition from State II to State III, the three parameters employed here reported on all the intermediates. However, the differences observed for NBD conjugates compared to W provided clues on the location of the peptide termini. As reported previously, the W intensity increased in all steps, as W9 and W15 became progressively buried in the membrane (Fig. 6) (14, 15). During the first insertion steps, intensity increased for the two NBD conjugates, indicating that both the N- and C termini were buried in the membrane (Fig. 8). Strikingly, the NBD intensity later decreased in both cases. This suggests that as the transmembrane configuration is being acquired, both the N- and C termini move from a membrane-buried state to a more hydrated final location. Interestingly, comparing the relative magnitude of the NBD decreases can inform on the conformational changes needed for formation of the final transmembrane state. In particular, our data suggest that both peptide termini are transiently buried in the membrane. The larger C_t-NBD decrease is expected to result from the NBD leaving the core of the membrane and gaining hydration as translocation is completed. However, the N_t-NBD signal decrease was unexpected, and it suggests that the N_t of pHLIP transiently explores a hydrophobic position of the membrane. This situation, expected to be energetically unfavorable, is reversed when pHLIP insertion is completed and the final transmembrane state is established.

Our results show that significantly different pH dependence values are obtained when employing different methods to study the membrane insertion of pHLIP. As a result, caution must be exercised in interpreting pK data. This is particularly important when discussing pHLIP and other peptides to translocate a drug cargo linked to the inserting end (27). We propose that the common pK_SM parameter is primarily influenced by intermediate steps that precede translocation of the C_t. As a result, we suggest that SM does not accurately report on translocation of the C_t, and the pK_SM value would be higher than the pH midpoint of cargo translocation. More precise cargo translocation information might be obtained instead employing pK_Ct-NBD, which we propose informs on the pK of translocation occurring at more acidic pH values. The use of this parameter might be used as a guide in the design of pHLIP variants that efficiently target the extracellular pH of tumors. Understanding which step in the folding/insertion transition a particular analysis method reports on is critical in evaluating the cancer targeting ability of peptides that undergo acidosis-dependent transmembrane translocation.

Author Contributions

H.L.S. and J.M.W. performed the research. H.L.S., J.M.W., and F.N.B. designed the research, analyzed the data, and wrote the manuscript.

Editor: Charles Deber.